a mab case study

OR WAIT null SECS

Editorial Information
Do Not Sell My Personal Information
Privacy Policy
Terms and Conditions

Quality by Design A-MAb Case Study Challenges Conventional Thinking

When the CMC Biotech Working Group began developing a case study on applying Quality by Design (QbD) to biotech products, the goal was to challenge conventional thinking on the subject.

When the CMC Biotech Working Group began developing a case study on applying Quality by Design (QbD) to biotech products, the goal was to challenge conventional thinking on the subject. “If the regulatory authorities read our final document and said ‘yes, this is all fine,’ we will have failed,” said Ken Seamon, PhD, one of the project’s facilitators. Now that the project has been completed, John Berridge, PhD, another one of the working group’s three facilitators, thinks they achieved that goal. “When people read it, they will find areas that they agree with, and almost certainly find other areas where they don’t,” he said. “It’s an aspirational document.” The case study, which discusses a fictional monoclonal antibody product, referred to as “A-MAb,” was developed over the course of a year by participants of the working group representing Abbott, Amgen, Genentech, GlaxoSmithKline, Eli Lilly, MedImmune, and Pfizer. Two examples of where the case study challenged conventional thinking, Berridge said, related to scale and change management. Scale-Down Models The limits of the design space for the 15,000-L production bioreactor step were largely based on data derived using a 2-L scale-down model, built on extensive development and manufacturing data (at both the 2-L and 15,000-L scales) for a previous approved product, “X-MAb”. To demonstrate that the 2-L scale-down model was representative and predictive of large-scale manufacturing performance, the team developed a principal component analysis (PCA) model. PCA transforms a large number of possibly correlated variables into a smaller number of uncorrelated variables, called principal components, and analyzes the variability in the data. The analysis included 13 variables, such as peak viable cell density, final viability, pH, glucose, lactate, and peak lactate, and a high degree of correlation was seen in the data sets for the different scales. Also, a complete comparability analysis for the product was made at the 500, 1,000-, 5,000- and 15,000-L scales, and an “engineering design space” was developed for the bioreactor design. The result, the report says, supported the use of the scale-down model. Lifecycle Approach to Process Validation The report also recommends a validation strategy that “relies more on the continuous process verification rather than a minimum number of ‘validation batches’ typically practised.” For the bioreactor operation, the validation strategy involved just two batches at the 15,000-L scale to confirm that the process performance at the 15,000-L scale was within the model predictions. These two batches are seen as “the start of the continuous process verification process” and a lifecycle approach to validation. To provide continued assurance that the process would remain in a state of control throughout the life of commercial manufacturing, the team would create a multivariate statistical partial least squares (PLS) model, which would ensure that internal correlations among variables is also considered. “For example, if at any given time the titer is lower than expected for the measured variable cell concentration, the PCA model will be able to detect this as a potential out of normal signal, even if both parameters are within their respective univariate ranges,” the report says. In this way, the PCA model can detect a large number of potential shifts, trends, and excursions that would not be detected by univariate monitoring tools. Scale-up From 15K to 25K Considered Within the Design Space The report also anticipated that the A-MAb bioreactor process would be scaled up further, to the 25,000-L scale. For the case study, it was assumed that the 25,000-L plant had an extensive and proven commercial manufacturing record of cGMP compliance and MAb production. Based on bioreactor design and engineering parameter characterization, the 25,000-L bioreactors were considered to be within the engineering design space, thus providing a very high degree of assurance that operation at this scale will result in comparable process performance and product quality. “Thus, the scale-up to the 25,000-L bioreactor is considered a movement within the engineering design space,” the report says. If a change to a different bioreactor (e.g., one with a different impeller design or geometry) were considered, the report says that an assessment would be made to determine if the bioreactor characteristics fell within the engineering design space. If they did not, then equipment modifications or changes in operational parameters would be considered to bring the bioreactor operation within the approved engineering design space. Highlights and Workshops The best way to read the 278-page report, Berridge says, is to start with the introduction and chapter 2, then move to the highlighted information that appears in blue boxes throughout the report. “If the information in a particular blue box is of interest, you can read the surrounding details,” he says, “If not, you can go to the next chapter.” In addition, members of the working group will participate in various public workshops hosted by ISPE and CASSS, who have been designated as the primary facilitators of education related to the case study. Sessions will be held during three 2010 CASSS events, including the Well Characterized Biological Products (WCBP) conference in Washington, DC, in January, the CMC Strategy Forum Europe in Vienna, in April, and the CMC Strategy Forum in Bethesda, MD, in July. Workshops also will be incorporated into already scheduled ISPE 2010 meetings, in Milan in March, Tokyo in April, Washington, DC, in June, and Brussels in September. The entire case study can be found on the CASSS (www.casss.org) and ISPE (www.ispe.org) web sites. Previous coverage: QbD Case Study Will Push Limits http://biopharminternational.findpharma.com/biopharm/News/QbD-Case-Study-Will-Push-Limits/ArticleStandard/Article/detail/592288?ref=25

A Mab Case Study Version

Related Papers

Biotechnology and Bioengineering

dietmar lang

Modupe Oluwasegun

Rouag Abdelkarim

Trends in Biotechnology

rahul bhambure

Brian Kelley

Analytical and Bioanalytical Chemistry

The “Pharmaceutical Current Good Manufacturing Practices (CGMPs) for the 21st Century—A Risk Based Approach” initiative announced by the FDA in August 2002 to improve and modernize pharmaceutical manufacturing facilitated adoption of process analytical technology (PAT) by the pharmaceutical industry. The potential for improved operational control and compliance resulting from continuous real-time quality assurance was highlighted as a likely benefit that would result from PAT implementation. A considerable amount of work has been done on this topic by academic and industrial contributors in the last decade. In this paper, we will start with a brief overview of evolution of PAT concepts and a review of their application in the wider pharmaceutical industry. The rest of the paper focuses on PAT applications for biotech processes with emphasis on developments in the last five years. It is our observation that while significant advances have been accomplished with regard to our ability to analyze/monitor key process and quality attributes in the biotech industry, much more needs to be done with regard to utilizing the collected data for subsequent control of the process, to achieve optimum yield and product quality. The latter is necessary to achieve the benefits that will result from PAT implementation. Figure Ease of PAT implementation for some of the commonly used unit operations in biotech processes

Chemical Engineering Science

Michel Eppink , Beckley Nfor , Emile J.a.x. Sandt

Satish Singh

All protein therapeutics have the potential to be immunogenic. Several factors, including patient characteristics, disease state, and the therapy itself, influence the generation of an immune response. Product-related factors such as the molecule design, the expression system, post-translational modifications, impurities, contaminants, formulation and excipients, container, closure, as well as degradation products are all implicated. However, a critical examination of the available data shows that clear unequivocal evidence for the impact of these latter factors on clinical immunogenicity is lacking. No report could be found that clearly de-convolutes the clinical impact of the product attributes on patient susceptibility. Aggregation carries the greatest concern as a risk factor for immunogenicity, but the impact of aggregates is likely to depend on their structure as well as on the functionality (e.g., immunostimulatory or immunomodulatory) of the therapeutic. Preclinical studies are not yet capable of assessing the clinically relevant immunogenicity potential of these product-related factors. Simply addressing these risk factors as part of product development will not eliminate immunogenicity. Minimization of immunogenicity has to begin at the molecule design stage by reducing or eliminating antigenic epitopes and building in favorable physical and chemical properties.

Andrea Gaggioli , Francesca Luciani

Loading Preview

Sorry, preview is currently unavailable. You can download the paper by clicking the button above.

Quality by Design risk assessments supporting approved antibody products

Brian kelley.

Genentech, South San Francisco, CA, USA

Quality by Design (QbD) is a global regulatory initiative that enhances pharmaceutical development through the design of the manufacturing process and controls to consistently deliver a product that performs as intended. The principles of pharmaceutical development relevant to QbD are described in the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guidance documents ICHQ8-11. 1–4 In 2008, the Food and Drug Administration (FDA) initiated a QbD pilot program for biological products wherein companies could contribute either with full biological license applications or supplements. Many biopharmaceutical firms participated in the pilot program, and progress was made on establishing the basis for pre- and post-approval filings. 5 One outcome of the pilot was an approval, granted in 2010, for an expanded change protocol for multiproduct/multisite transfer of drug substance processes for production of monoclonal antibodies at Roche/Genentech. 6

The A-mAb case study provided another substantial contribution to the field. 7 It described a variety of approaches to the major elements of QbD used by 7 companies (Pfizer, GlaxoSmithKline, Genentech, Abbott, Amgen, Lilly, MedImmune) with experience in the development and commercialization of biologics. This publication presented a diverse set of solutions to common problems, but, because many companies were involved, it lacked a self-consistent formalism.

Advances in refining the applications of QbD principles have continued in recent years, although progress is slower than hoped. 8-11 Roche/Genentech has licensed 2 therapeutic recombinant monoclonal antibody products, obinutuzumab (Gazyva®) and atezolizumab (Tencentriq®), in the US using QbD principals. 12-13 We believe these represent the first approvals for biologics that were comprehensively based on QbD information, including approved design space claims as well as a post-approval lifecycle management plans, contained in the license application.

Roche/Genentech recently published 7 articles describing the application of the principles of QbD for development and licensure of therapeutic monoclonal antibodies. 14-20 These articles present a self-consistent set of risk assessments and logical elements developed over the last decade, based on refinement through the FDA and European Medicines Agency pilot QbD programs and approvals of obinutuzumab and atezolizumab. They provide a standardized basis for the global health authorities to assess new product license applications from Roche/Genentech, and seek to establish transparent communication of the links between the manufacturing process and product storage, product quality and impact to patients, the commercial control strategy, and post-licensure change management. The articles cover all key elements of QbD, including establishment of critical quality attributes, definition of a design space, identification of critical process parameters, assembly of the commercial control system, description of the post-approval life cycle management, considerations of an overarching risk assessment that addresses elective decisions taken at the time of license application, and analysis of how the full set of assessments manage the residual risk to product quality. The refinement of these tools benefitted from the substantial experience the company has gained in over 25 years of biological drug development, which included licensure and production of 9 commercial monoclonal antibodies and the use of applicable process and product platform knowledge. 21

Roche/Genentech view these articles as an open-source sharing of the results of a decade-long internal investment. We hope that the tools will be used by other companies (adapted to their particular product requirements, if needed), which could advance the adoption of improved methodologies in support of license applications for products with enhanced product and process knowledge. This formalism will be used for all biologics in the Roche/Genentech pipeline, and, when combined with the use of process and product platform knowledge, should result in significant efficiencies and resource savings during late-stage development. We also hope that this will enable post-licensure flexibility for production and the appropriate regulatory oversight for process parameter changes within the design space. The large commercial and clinical biologics portfolio at Roche/Genentech will build on this QbD foundation for both life-cycle management and the commercialization phase.

Although discussed in the context of antibody products, the tools we describe in the articles should be applicable to other protein therapeutics, and perhaps to small molecules and other pharmaceutical modalities as well. We encourage publication of future refinements of these tools and advances in their applications for the benefit of the larger industrial and regulatory community. Sharing of such information could result in superior approaches to product license applications in the future, as well as streamlining of late-stage product development, and the review and approval phases of the biologics product lifecycle.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

ISPE Announces Availability of A-Mab Case Study at its Annual Meeting*

ISPE, a global not-for-profit association of 24,000 technical professionals working in the manufacturing and drug development sectors of the pharmaceutical industry, announced today a major extension of its Product Quality Lifecycle Implementation initiative (PQLI) further into biotechnology.

Firm forecasts global pharmaceutical growth markets

Honeyman Group launches revolutionary pure water system into the Pharmaceutical Sector*

New healthcare business head offers a global packaging perspective

Manage your validation missions with GxP Manager*

EXPO PACK México 2024 Surpasses Expectations, Sets New Records

Episciences was able to integrate ACSIS’ Serialization Management system without the hassle and cost of modifying existing secondary packaging artwork or labeling procedures to prioritize barcode placement.

Cosmeceuticals Co. Completes First Phase of Serialization

Newer workers may resonate with video content for training, some industry executives suggest.

How Today’s Technology Can Enhance Training

L’Occitane is using PET bottles made entirely from enzymatically recycled PET waste for its Amande shower oil range.

L’Occitane Cosmetics Bottle Uses 100% Enzymatically Recycled Content

EXPO PACK México 2024: Latin America's Largest Packaging and Processing Trade Show Opens Today

Modular Bioprocessing Platform (MBP)

Child-resistant wallet card for blisters, daikyo crystal zenith (cz) ready-to-use nested vials in tub.

Spring 2024 Healthcare Packaging

Winter Healthcare Packaging 2023

Fall 2023 Healthcare Packaging

Spring 2023 Healthcare Packaging

Cobot Palletizers

347 Hcp Sustainable Healthcare&medical Device Packaging Suppliers

Researched List: Sustainable Healthcare & Medical Device Packaging Suppliers

Healthcare Packaging's Most Popular Articles of 2023

Preparing For Your Next FDA Inspection

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Publications
Account settings
My Bibliography
Collections
Citation manager

Save citation to file

Email citation, add to collections.

Create a new collection
Add to an existing collection

Add to My Bibliography

Your saved search, create a file for external citation management software, your rss feed.

Search in PubMed
Search in NLM Catalog
Add to Search

Micro scale self-interaction chromatography of proteins: A mAb case-study

Affiliations.

1 Surfaces and Particle Engineering Laboratory, Department of Chemical Engineering, Imperial College London, London, United Kingdom.
2 Surfaces and Particle Engineering Laboratory, Department of Chemical Engineering, Imperial College London, London, United Kingdom. Electronic address: [email protected].
3 R&D Group, FUJIFILM Diosynth Biotechnologies, Billingham, United Kingdom.
PMID: 26810801
DOI: 10.1016/j.chroma.2015.12.034

Self-interaction chromatography is known to be a fast, automated and promising experimental technique for determination of B22, but with the primary disadvantage of needing a significant amount of protein (>50 mg). This requirement compromises its usage as a technique for the early screening of new biotherapeutic candidates. A new scaled down SIC method has been evaluated here using a number of micro LC columns of different diameters and lengths, using typically 10 times less stationary phase than traditional SIC. Scale-down was successfully accomplished using these micro-columns, where the SIC results for a range of differing columns sizes were in agreement, as reflected by k', B22 and column volumes data. The results reported here demonstrate that a scaled down version of SIC can be easily implemented using conventional liquid chromatography system where the final amount of mAbs used was 10 times less than required by conventional SIC methodologies.

Keywords: Formulation; Monoclonal antibodies; Protein stability; Scale-down; Second virial coefficient; Self-interaction chromatography.

PubMed Disclaimer

Publication types

Search in MeSH

Related information

PubChem Compound (MeSH Keyword)

Grants and funding

Biotechnology and Biological Sciences Research Council/United Kingdom

LinkOut - more resources

Full text sources.

Elsevier Science

Other Literature Sources

scite Smart Citations
Citation Manager

NCBI Literature Resources

MeSH PMC Bookshelf Disclaimer

The PubMed wordmark and PubMed logo are registered trademarks of the U.S. Department of Health and Human Services (HHS). Unauthorized use of these marks is strictly prohibited.

A Case Study-Based Approach to Integrating Qbd Principles in Gene Therapy CMC Programs

Project A-Gene A case study-based approach to integrating QbD principles in Gene Therapy CMC programs

Presented by Contributors

CHAPTER LEADS / CONTRIBUTING MEMBERS

Greg MacMichael Hone Megan Furcolo Patrick Jeffrey Hung CMC Bioservices Pfizer Pfizer Vigene Biosciences Saroj Ramdas Baldus Phoebe Parker Joann Alexandra Beumer-Sassi Amicus Pfizer Pfizer Voisin Consulting Life Sciences Mo Heidaran Banbula Agnieszka Dawn Henke Luis Maranga Parexcel Pfizer Standards Coordinating Body Voyager Therapeutics Paul McCormac Lavoie Janelle (SCB) Shamik Sharma Pfizer Pfizer James Warren Voyager Therapeutics Iryna Sanders McEnroe Janet Ultragenyx Nripen Singh Pfizer Pfizer Jessie Sun Voyager Therapeutics Aili Cheng Micklewright Althea Ultragenyx Pfizer Pfizer Jim Richardson Cirelli David Alphonse Ignatius Arun United State Pharmacopia Pfizer Pfizer (USP)

REVIEWERS Note: All chapters were reviewed, but individual reviewers did not review every chapter. Reviewers volunteered to provide editorial review for select chapters according to their expertise and interest.

Tayler Renshaw Nadine Sandhöfer Jean Stanton Mike Winkler Audentes Therapeutics Cevec Johnson & Johnson (J&J) REGENEXBIO Malou Gemeniano Tim Farries Kelvin Lee Justin Horvath Audentes Therapeutics ERA Consulting National Institute for REGENEXBIO Innovation in Manufacturing William Werner Raj Puri Rob McCombie Biopharmaceuticals (NIIMBL) Audentes Therapeutics (Regulatory chapter only) Sangamo Therapeutics FDA Hardeep Samra David Gray Robert Shaw Neurogene Audentes Therapeutics Tim Miller SCB Forge Biologics Tina Parikh John Grunkemeier Sven Kili Neurogene Audentes Therapeutics Michele Meyers Sven Kili Consulting GSK David Litwack Dominic DeMuro Khandan Baradaran Prevail Therapeutics Audentes Therapeutics Steven Weisser Ultragenyx GSK Michael Boychyn Isabella Palazzolo Jeffrey Hung REGENEXBIO Biogen Michael Mercaldi Vigene Homology Medicines Mike Byrne Stuart Beattie Erica Giordano REGENEXBIO Biogen Maria Lobikin Voisin Consulting Life Sciences Homology Medicines Tristan Marshall Clare Blue Emmanuelle Sabbah- REGENEXBIO Biogen James McGivney IV Petrover Homology Medicines Don Startt Voisin Consulting Life Sciences Nicole Faust REGENEXBIO Cevec Eric Faulkner

Homology Medicines

Medical/Technical Writers Project Management Editor Graphic Design (select figures) Darshana Gupte Michael Lehmicke Marie Daghlian Michaela Flatly Cecelia Wall Adam Roose Overall Graphic Design and Layout Amr Eissa Carol Collier Julia Grant

ii Project A-Gene Introduction

The effort that has ultimately resulted in the document below began with a short, yet complex question posed to an audience of cell and gene therapy (CGTx) CMC experts. The Alliance for Regenerative Medicine (ARM) convened its first of a now annual series of CMC Summits in December of 2017, inviting individuals from across the industry to join in a conversation on manufac- turing. As part of this discussion, the question of “What is the biggest challenge to cell and gene CMC? How can ARM help to address it?” was raised to the audience and, across the numerous responses and specific examples given, there emerged a common complaint related to the heterogeneity of the process. After exploring this further, there was general agreement that many of the issues driving the challenges to streamlined, cost effective manufacture of CGTx products derive from a lack of standardized methodologies and training around CMC programs. It was further suggested that this phenomenon was not new, and that we should look to the past for inspiration on how to address this challenge for the future. Therefore, it was decided to embark on a mission to recreate ‘A-Mab’ for the CGTx industry. In previous years, the broader life sciences industry has encountered systemic barriers to the continued development of promising technologies. In both the mono- clonal antibody industry and the vaccine development world, the whirl of energy around scientific discovery was stalled by the realities of manufacturing. While a small team of well-trained experts can produce high quality batches of drug product for use in process development and early clinical trials, it is inevitable that this process will need to be dramatically increased in scale, and the process transferred to other parties for commercial production. To help lower the barrier to this nec- essary tech transfer, and to better prepare new entrants to the industry, the leading developers of monoclonal antibodies and vaccines have produced a consolidated set

iii of recommendations for implementation of Quality by Design (QbD). A-Mab and A-Vax, respective to each industry, have played a significant role in elevating best practices within their industries, and have been effective in continued workforce development efforts. Borrowing from this model, the members of the Alliance for Regenerative Medicine have worked to produce a similar document for use by the burgeoning gene therapy sector. In order to make A-Gene an effective resource, and reflecting the ongoing inno- vation in the sector, we sought to:

1) draw from as wide a set of expertise as possible, 2) specify our area of focus to direct gene therapy, and 3) focus on AAV as the case study.

This effort to catalogue expertise in gene therapy development occurred in parallel to approval of the first cohort of human gene therapies, which have shown the value and clinical relevance of such programs. As the field continues to develop it has been recognized that the future of cost effective gene therapy relies on implementation of common practices, development of specialized technologies, and above all else, standardization of methods. Given the wide scope of innovation underway in each of these areas of focus, ARM and the A-Gene team sought to bring in as much thought leadership as possible to ensure that what was recorded was truly a representation of best practices in the industry. Reflecting on this, A-Gene is a truly collaborative effort that has been crafted with contributions from more than 50 industry experts from more than 20 leading therapeutics developers. To further maximize the utility of A-Gene, the drafting team decided to focus on in vivo gene therapy. Current FDA language around Gene Therapy encompasses in vivo gene therapy, gene editing/manipulation, and gene modified cell therapy. Each application carries a wide array of specific manufacturing steps and consid- erations, and so in order to draft a cohesive and informative narrative, the A-Gene team decided to focus on in vivo gene therapy. Ex vivo gene-modified cell therapies will be the subject of a future case study. Finally, the A-Gene team deliberated on what the underlying case study would be for this document. As each chapter is meant to operate as both an independent re- source, as well as part of a comprehensive narrative, we felt it was necessary to focus on one specific use case to facilitate the utility of A-Gene. Lentiviral and AAV vectors are the two most frequently used viral vector platforms used in gene therapies, and the approach to producing these two vectors are similar. While lentiviral vectors are predominantly used for ex vivo cell modification for gene modified cell therapies, AAV is the major vector type for in vivo applications. Given our previous decision

iv to draw from the widest set of expertise possible on direct gene therapies and the focus on in vivo gene therapy approaches, it naturally made sense to focus our case study on human rAAV therapeutic development. This is not to imply either a relative value in rAAV vs other viral platforms for direct gene therapy, but is, in general, recognition of the number of companies pursuing rAAV applications. We deliberated on inclusion of parallel tracks throughout the document, comparing and contrasting rAAV to lentivirus, but in the end decided this would be too cumbersome. Beyond this, there is a great deal of overlap in thought process and methodology between the two approaches, and so in lieu of a running comparison we elected to spend some time in the introduction addressing differences in CMC methods between the two viral platforms. A-Gene is not intended to represent a standard to be rigidly applied. It is a hy- pothetical case study representing an archetype of an AAV vector for gene therapy. Therefore, it is a snapshot in time of current best principles in a rapidly evolving field. The data cited in the document are non-proprietary, and are intended to be for illustrative purposes only. Where appropriate the authors have borrowed formatting and structure from the A-Mab case study. While we have attempted to be as comprehensive as possible, and have subjected the document to rigorous review, it is not a “recipe book” for AAV manufacture. Some aspects of process development (e.g., facility design), were deliberately omitted for the sake of brevity. Importantly, A-Gene is not an example of a mock regulatory submission, nor should it be interpreted as regulatory advice, or cited as regulatory guidance. As a final point, we wish to thank those who contributed to this effort. The Alliance for Regenerative Medicine is grateful to the innumerable thought leaders, subject matter experts, and researchers who have helped to make this project a reality. We also wish to extend our deep appreciation to the National Institute for Innovation in Manufacturing Biopharmaceuticals for their support and contribu- tions to this effort, and for working with ARM to make this project a reality. We look forward to continuing our work with our members, key opinion leaders, and the numerous innovators who have made gene therapies a reality for the thousands of patients who rely on biotherapeutic developments to improve their quality of life. We intend to continue our work with these groups to maintain the relevance and accuracy of this document as the industry advances.

This work was performed under financial assistance award 70NANB17H002 from the U. S. Department of Commerce, National Institute of Standards and Technology.

v Chapter 1 Regulatory Considerations Chapter 1 | Contents

Introduction...... 3 Regulatory Framework in the U.S., EU, and Japan...... 3 U.S...... 3 European Union...... 3 Japan...... 4 Designations to Expedite Development...... 5 Health Authority Innovation Teams...... 7 Health Authorities and Sponsor Meetings...... 8 FDA...... 10 Preclinical Meetings...... 10 Clinical Meetings...... 12 EMA...... 12 ATMP Classification...... 13 EMA Scientific Advice...... 13 National Competent Authority...... 14 Parallel Consultation...... 15 EMA-FDA Programs...... 15 Parallel Scientific Advice...... 15 Parallel Consultation...... 15 PMDA...... 16 Preclinical and Clinical Meetings...... 16 Registration Meetings...... 16 FDA: Pre-BLA...... 17 EMA: Presubmission Rapporteur/Co-rapporteur Meetings...... 17 Topics to be covered in HA/sponsor meetings...... 18 Submission Content...... 18 Electronic Common Technical Document (eCTD)...... 18 IND submissions to FDA...... 18 Clinical Trial Application (CTA)...... 23 Module 3 Content...... 23 Lifecycle Management...... 23 Established Conditions...... 24 Post Approval Change Management Protocols (PACMP)...... 24 Product Lifecycle Management Document (PLCM)...... 24 Conclusion...... 24 Appendix...... 25 Abbreviations...... 25 Endnotes...... 26

CHAPTER 1 Regulatory Considerations 2 Introduction Figure 1-1. CAT role in ATMP review process in EU Gene therapies, a subset of regenerative medicine ther- apies, include plasmid DNA and RNA, viral vectors, Certifies quality and Assists scientifically bacterial vectors, products incorporating human gene non-clinical data for in developing editing technology, and patient-derived cellular gene small and medium documentation enterprises related to regulation therapy products. These are transformative therapies developing of ATMPs addressing conditions such as cancer and genetic and ATMPs infectious diseases.1 Few commercial assets have been approved for patient use globally, reflecting the infan- cy of this modality. Currently, there are pathways for CAT regulatory review and approval in three major markets: United States (U.S.), European Union (EU), and Japan. Contributes This chapter summarizes (1) the regulatory frameworks to scientific in these three major markets, (2) expedited regulatory advice and supports pathways, (3) Health Authority (HA) innovation teams, the Scientific Advice Trains assessors and Working Party organizes scientific (4) HA and sponsor meetings, (5) gene therapy specific (SAWP) workshops information to be included in the electronic Common Technical Document (eCTD) format for the license applications, and (6) lifecycle change management. • Testing of retroviral vector-based human gene therapy products3 Regulatory Framework in the U.S., EU, • Devices used with regenerative medicines advanced and Japan therapies4 • Microbial vectors used for gene therapy5 The novel and diverse nature of gene therapies has re- • Potency tests for cellular and gene therapy products6 sulted in evolving regulatory frameworks specified to support these products in markets such as the U.S., EU, To provide additional support to developers, CBER has and Japan. With increased experience with this broad established OTAT Learn as an educational resource for family of products, regulatory agencies will have the industry and includes several course listings led by OTAT opportunity to further define guidance that will facilitate staff.7 their development to address patients’ unmet needs. EUROPEAN UNION UNITED STATES In the EU, gene therapy products are included under In the U.S., regenerative medicine therapies are regulated the umbrella of Advanced Therapy Medicinal Products by the Food and Drug Administration’s (FDA’s) Office (ATMPs) and are regulated by the European Medicines of Tissues and Advanced Therapies (OTAT) within the Agency’s (EMA’s) Committee for Advanced Therapies Center for Biologics Evaluation and Research (CBER). (CAT), which is tasked to assess the quality, safety, and OTAT oversight ensures the safety, purity, potency, efficacy of medicinal products. There may be instances and effectiveness of gene therapy products. CBER has where classification of medicinal products as ATMPs may released guidance documents addressing the following be borderline with respect to other areas (e.g., medical gene therapy CMC topics: devices); thus, EMA has established an ATMP classifica- tion process8 further described in ATMP classification. • Chemistry, Manufacturing, and Controls (CMC) CATs primary responsibility in ATMP regulation is Information for Investigational New Drug to provide draft opinions regarding approval to The Applications (INDs)2 Committee for Medicinal Products for Human Use

CHAPTER 1 Regulatory Considerations 3 Figure 1-2. Japan Regulatory Framework for Regenerative Medicines 12

MHLW SAFETY ACT (Ministry of Health, Labor, and Welfare )

Health Policy Bureau Health and Welfare Branches

Certified Pharmaceutical and PMD ACT Committees Environmental Health Bureau

Approval Decision, License Execution

Hospitals PMDA (Pharmaceuticals and Medical Devices Agency )

Manufacturing MAHs CPFs Office of Cellular and Tissue-Based Sites Products Product Review Office of Manufacturing Quality and Compliance

Office of Safety I Site Inspections

(CHMP). In addition to drafting recommendations, for ATMPs on the following topics:10 CAT is integral in the following activities summarized in Figure 1.9 • Procedural advice • Dossier requirements and submission dates The EMA has released guidance documents address- • Guidelines for risk-based approaches ing the following gene therapy CMC topics: JAPAN • Quality, nonclinical and clinical aspects of gene Japan has two main regulatory authorities that are in- therapy medicinal products dependent agencies with distinct roles during the drug • Scientific requirements for the environmental risk approval process: (1) The Ministry of Health, Labor, and assessment of gene therapy medicinal products Welfare (MHLW), which is responsible for publishing • Quality, nonclinical and clinical aspects of medicinal regulatory guidelines, managing advisory committees, products containing genetically modified cells and providing final authorizations for applications and (2) the Pharmaceuticals and Medical Devices Agency In addition to guidelines related to several aspects of (PMDA), which is responsible for regulatory and sci- gene therapy development, the EMA has established entific review (i.e., Office of Cellular and Tissue based guidance regarding marketing authorization procedures Products), and Good Manufacturing Practice (GMP)

CHAPTER 1 Regulatory Considerations 4 compliance and inspections (i.e., Office of Manufacturing Designations to Expedite Quality and Compliance). Japan’s regulatory framework for regenerative medicines, including production of Development therapeutic products by industry, is based on “The Act The U.S., EU, and Japan have established expedited path- on Pharmaceuticals and Medical Devices (PMD Act),” ways to support accelerated development and regulatory which underpins activities within the PMDA regarding approval for medicinal products that have the potential gene therapy regulatory evaluation (see Figure 211). to address unmet medical needs. These pathways provide opportunities for developers to engage with regulators

Table 1-1. Summary of Expedited Pathways (U.S.)

Breakthrough Regenerative Therapy (BTD) Medicine Fast Track (FT) Advanced Priority Review Accelerated Approval Therapy (RMAT)

Date 1997 2012 2017 1992 1992 Established Must treat serious Chemical, Regenerative Must treat a serious Must treat a serious Qualifying condition. Biological and Medicines. condition. condition. Criteria Clinical or Regenerative Must treat Must provide Provides a meaningful nonclinical data Medicines. serious or life- a significant advantage over demonstrates Must treat threatening improvement available therapies. that the therapy serious disease/ in safety or Demonstrates an has the potential condition. condition. effectiveness. effect on a surrogate to address unmet Preliminary Preliminary Any supplement endpoint that is medical needs clinical evidence clinical evidence that proposes a reasonably likely for such disease indicates that indicates that labeling change to predict clinical or condition. the therapy may the therapy has pursuant to a report benefit or on a clinical Must be demonstrate the potential to on a pediatric study endpoint that can be designated substantial address unmet under 505A. measured earlier than as a qualified improvement medical needs An application for a irreversible morbidity infectious on a clinically for such disease drug that has been or mortality (IMM) disease product significant or condition designated as a that is reasonably endpoint(s) qualified infectious likely to predict an over available disease product. effect on IMM or other therapies Any application or clinical benefit (i.e., an supplement for a intermediate clinical drug submitted with endpoint) a priority review voucher Frequent written Same as FT, plus: Same as BTD, Shorter review Approval based on the Key communication. Early and plus: of marketing effect on a surrogate Program Actions to more frequent Early discussion application (6 endpoint or an Features expedite communications of potential months compared intermediate clinical development and with FDA during surrogate or with the 10-month endpoint review. development. intermediate standard review) Rolling review Rolling clinical endpoint submission and review. Priority Review

CHAPTER 1 Regulatory Considerations 5 Table 1-2. Summary of Expedited Pathways EU

Conditional Marketing Exceptional Accelerated Assessment PRIME Authorization 20,21 Circumstances Date 2004 2004 2004 2016 Established

Request should be Filling an unmet medical Applicants are not Address unmet Qualifying made at least two to need. able to provide clinical medical need. Criteria three months before Pertaining to life- data comprehensively Provide a major submitting a makreting- threatening, serious, or because of rarity of the therapeutic advantage authorisation application. emergency disease, or disease for example. over existing Important in terms orphan products. Applicable to life- treatments. of public health and Company must be able threatening or serious Based on early clinical innovation. to provide clinical data disease. data. Fulfills an unmet need. comprehensively. Strong evidence. Positive benefit/risk balance.

Reduce the timeframe for Active for one year only Applicants do not Enhanced Key marketing authorisation with an annual renewal need to submit interaction and early Program to 150 days. of the approval until comprehensive data. communication with Features the EMA converts the sponsors. approval to standard Accelerated authorization. assessment and Enables early approval scientific advice. while confirmatory.

during the development process and participate in ac- (AA), Conditional Marketing Authorization, and celerated review programs within each agency. Authorization under Exceptional Circumstances. In In the U.S., the FDA has developed five designations addition, there is also the Priority Medicine (PRIME) for expedited pathways that are relevant for gene ther- scheme. The Accelerated Assessment reduces the time apies: Fast Track designation, Breakthrough Therapy of assessment by the EU from the 210-day maximum designation, Regenerative Medicine Advanced Therapy to 150-day maximum. The Conditional Marketing designation, priority review designation, and accelerated Authorization (CMA) is a temporary authorization approval. The fast track designation provides advantages for medications filling an unmet medical need. The for facilitating development and expediting review of Authorization under Exceptional Circumstances (AEC) the product. The Breakthrough Therapy designation is a temporary authorization awarded for medications (BTD) is an expedited pathway available for all treatment dealing with very rare diseases. The PRIME scheme was modalities, including gene therapies, and incorporates introduced in 2016 to support accelerated development all the benefits of fast track designation and more. This of clinical programs to facilitate earlier patient access pathway was followed by a regenerative medicine path- for unmet, serious medical needs. The PRIME scheme way known as Regenerative Medicine Advanced Therapy leverages on existing procedures and tools provided by (RMAT) designation in 2017. Gene therapy products, the EMA with a commitment to engage more closely. Of including those that received fast track designation, BTD, the requests submitted, 81 products have been granted or RMAT designation, may also be eligible for priority the PRIME scheme, while 239 have been denied. review designation and accelerated approval. In Japan, expedited pathways that are relevant for In the EU, there are three expedited pathways that gene therapies are: Priority Review, Conditional and are relevant for gene therapies: Accelerated Assessment Term-Limited Approval, Conditional Approval, and

CHAPTER 1 Regulatory Considerations 6 Table 1-3. Summary of Expedited Pathways (Japan)

Conditional and Term- Priority Review Conditional Approval 25 Sakigake 26 Limited Approval 24 Date 2004 2004 2004 2016 Established No standard Promising results No standard therapy exists or Products for diseases Qualifying existing therapy of early-phase I/II superior clinical usefulness is in dire need of Criteria or superior clinical registration trials in demonstrated as compared innovative therapy usefulness as terms of efficacy and with the existing products Applied for compared with the safety in terms of quality of life of approval firstly or existing products Sponsors must conduct patients, efficacy, or safety simultaneously in terms of quality postmarketing clinical Applicable to serious disease (defined as of life of patients, studies and so on to It is difficult or would take submissions within 30 efficacy, or safety confirm the efficacy too long to conduct a days of each other) in Applicable to and safety and resubmit confirmatory study Japan serious disease applications for regular Exploratory clinical studies Prominent approval within a show efficacy and safety effectiveness can be predetermined period Surveillance or clinical studies expected based on Only for regenerative must be conducted as a post- nonclinical and early medicines marketing requirement phase trials

Target total review Valid for no more than Conditional approval for drugs Prioritized consultation Key time is nine months seven years Priority Review Prioritized review Program Review partner Features Substantial post- marketing safety measures Rolling submission and review

Sakigake. Priority Review lessens the target review date Health Authority Innovation Teams to nine months and is available for medications that ful- fill an unmet need. The Conditional and Term-Limited Health Authorities (HAs) such as the FDA and EMA are Approval pathway is for regenerative medicines that eager to support innovation as gene therapies mature and show promising early phase results. The Conditional aim to ensure patient safety and efficacy while increasing Approval pathway is targeted for highly useful and patient access. However, the biopharmaceutical industry effective drugs treating serious diseases. The Sakigake has been slow in adopting innovative manufacturing early access scheme was introduced in 2014 to expedite technologies due to concerns regarding regulatory ac- innovative assets and was implemented in 2015. ceptance and the impact on global supply chains due to Table 1, Table 2, and Table 3 summarize the criteria the difficulty in post-approval change processes. Thus, for expedited pathways in the U.S., EU, and Japan that engagement with HAs via innovation teams within the can be utilized for gene therapy medicinal products.13 U.S. and EU agencies can facilitate dialogue between a In addition to Sakigake, the PMDA Act also provid- sponsor and the HA, and support progression of inno- ed a new scheme for expediting regenerative medical vative approaches. products. Differences between the traditional approval The FDA’s CBER Advanced Technologies Team process and the new scheme for regenerative medical (CATT) was recently established to promote engagement products are presented in Figure 3. with prospective innovators and developers and sponsors regarding advanced manufacturing technologies. This

CHAPTER 1 Regulatory Considerations 7 Figure 1-3. Expedited Approval System under PMDA Act. Sato

Traditional Approval Process

Clinical Phased clinical trials Marketing Marketing study (confirmation of efficacy and safety) authorization

Re-application New scheme for regenerative medical products (within 7 years)

Clinical trials (Likely to Conditional/ Marketing Marketing Clinical Clinical Marketing predict efficacy, term-limited (Further confirmation authorization study study continues confirming authorization of efficacy and safety) or revocation safety)

Post-marketing safety measures must be taken, including prior informed consent of risk to patients team serves as a resource that provides early engagement technologies and (2) ensure EMA readiness for eval- during development of an innovative technology through uation of developments in innovative medicines and opportunities for feedback from CBER regarding issues technologies. The ITF is a multidisciplinary team that related to implementation of advanced manufacturing includes scientific, regulatory, and legal competencies.28 and testing technologies (in addition to facilitating To apply to the ITF, sponsors should complete the ITF logistics to support the discussion). Industry appli- briefing meeting request form (link to the form found cants should submit requests electronically to Industry. here: https://www.ema.europa.eu/en/human-regulatory/ [email protected] and include the following:27 research-development/innovation-medicines) and submit via email to [email protected] . In addition to • Brief description of the technology the centralized EU ITF, national innovation offices are • Explanation of why the technology is novel and another resource for engagement (e.g. PEI Innovation unique Office for ATMPs in Germany29). • Description of the impact of the technology in terms of improved product manufacturing, product safety Health Authorities and Sponsor and efficacy • Summary of the development plan and questions to Meetings be addressed Development of innovative investigational products, such as gene therapy products, can introduce unique In the EU, the Innovation Task Force (ITF) in the challenges due to unknown safety profiles, complex EMA provides an opportunity for sponsor and HA manufacturing technologies, incorporation of innovative engagement regarding emerging therapies and technol- devices, and the use of cutting-edge testing method- ogies, which include gene therapies. In contrast to the ologies. In recognition of the complex nature of gene FDA’s CATT, the ITF helps EMA to (1) clarify questions therapy products, HAs in the United States and Europe regarding the pathway for emerging therapies and have introduced preliminary informal consultations to

CHAPTER 1 Regulatory Considerations 8 allow sponsors to obtain feedback from the HAs to assist guaranteed, and the agency might respond to a sponsor’s on the product development and clinical planning. These inquiry for advice in writing or via teleconference. early meetings are in addition to the conventional HA/ sponsor meetings. • Type A meetings: necessary for an otherwise stalled As each HA has its own pathways, so does each have product development program to proceed or to its own expectations on engagement by sponsors. During address an important safety issue. It is important to the life cycle of drug development, sponsors may seek point out that Type A meetings are only granted for advice from the FDA regarding several topics, including, stalled product development due to an action taken but not limited to, the following: regulatory, clinical phar- by the FDA, and not for an issue from the developer macology, safety, product quality, and nonclinical matters. side. Topics that are often covered by Type A meetings include dispute resolutions as described in 21 CFR FDA 10.75, 312.48, and 314.103, clinical holds, receipt of an FDA Nonagreement Special Protocol Assessment Meetings between FDA and sponsors occur at critical letter, FDA regulatory action other than an approval, junctures during the life cycle of product development and FDA issuance of a refuse-to-file letter. and are aimed at minimizing wasteful expenditures of • Type B meetings: cover pre-investigational new drug time and resources. In addition to INTERACT meetings, applications (pre-INDs), pre-BLAs, pre-emergency other particularly important milestone meetings under use authorization, FDA regulatory actions other Prescription Drug User Fee Act (PDUFA) include: pre- than approval, risk evaluation and mitigation IND, end-of-phase 1 (EOP1), EOP2, and pre-Biologics strategies, post-marketing requirements outside the License Application (pre-BLA) meetings. Additional context of the review of a marketing application, and details on the available meetings between HAs and spon- development programs for products granted BTD sors are described below and summarized in Figure 4. and/or RMAT designation status. The FDA offers four types of meetings related to the • Type B (EOP) meetings: include certain end-of-phase development and review of investigational new drugs 1 meetings for products considered for marketing and biologics under the PDUFA: Type A, Type B, Type approval under 21 CFR part 312, subpart E, or 21 CFR B (end of phase (EOP)), and Type C, as further described part 314, subpart H, or similar products and end-of- below.30 During the preclinical and early clinical stages, phase 2 or pre-phase 3 meetings (21 CFR 312.47). most of the in-person meetings with the agency are not

Figure 1-4. Interactions with FDA31 for Regenerative Medicine

IND submitted BLA submitted BLA Milestone meetings (PMC/PMR) RMAT/BT Meetings Interactive Review

pre- Research Phase I Phase II Phase III Marketing Authorization Post-Marketing clinical

Interact Clinical Trials BLA Review: 10 months from filing date Meetings Related to Pre-IND Priority Review: 6 months Supplements Meeting IND Milestone Meetings from filing date (EOP1, EOP2, EOP3, Pre-BLA) PAS: 4 Months All Others: 6 months

CHAPTER 1 Regulatory Considerations 9 Table 1-4. Dos and Don’ts for INTERACT Meeting Package

INTERACT Meeting Package Content Dos Don’ts

Description of the product and the Submit the package Include questions regarding candidate disease or condition being treated or together with the meeting selection. prevented. request. Request an INTERACT meeting if the Summary of information about Package should be no more sponsor has already requested and the product development to date than 50 pages. obtained formal regulatory advice about and future development plans, if Identify the specific a similar product/indication from the appropriate. investigational product to FDA. Brief statement summarizing the be evaluated in a clinical Include questions regarding the purpose of the meeting. study. adequacy and design of toxicology List of questions for discussion, Define key strategic studies that have been completed–these grouped by topic. product development should be submitted as pre-IND. Summary of the data to support a activities such as Include requests for pre-review of discussion organized by topic and manufacturing process, completed proof-of-concept–these question. starting materials, should be submitted as pre-IND. cell sources, critical List of all participants, with their Include review of clinical study designs components, and use titles and affiliations, who will attend or protocols–these should be submitted of devices prior to the the meeting from the sponsor’s as pre-IND. INTERACT meeting. organization, including consultants Ask questions that are not necessarily and interpreters. Ask specific, targeted product-specific, such as those about questions. Suggested dates and times for the novel technologies that can significantly meeting. impact on a product class.

• Type C meetings: any meeting other than a Type INTERACT meeting is not mandatory, but may be highly A, Type B, or Type B (EOP) meeting regarding the valuable to developers. This meeting is non-binding in development and review of a product, including nature, which means that a sponsor is not bound to pursue meetings to discuss adequacy of facility design a particular regulatory pathway. This also means that the and establishment issues, and to facilitate early FDA feedback can change depending on information/ consultations on the use of a biomarker as a new updates the sponsor provides in the future. surrogate endpoint that has never been previously Sponsors can obtain non-binding advice regarding used as the primary basis for product approval in the different aspects of the development process, such as: proposed context of use. • Planning initial clinical development strategies Preclinical Meetings: (INTERACT) • Chemistry, manufacturing, and controls Sponsors applying to the FDA can obtain a preliminary • Pharmacology/Toxicology development informal non-binding consultation with the Agency • Clinical aspects of the product development program through the INTERACT meeting32,33 prior to a pre-IND meeting. Some sponsors are already familiar with this type Identifying the optimal time of the meeting relative to of early meeting, as it replaces the pre-pre-IND meeting product development might be the sponsor’s greatest chal- that was in place until 2018. It is important to note that the lenge when seeking an INTERACT meeting. The meeting INTERACT meeting is available for innovative investiga- might be declined if it is requested too early in the process tional products at an early stage of development on issues at a point when a clear design has not been established, or that are not yet at the pre-IND meeting phase, validating when it is considered too late, after a clinical protocol has FDA’s recognition of the complexity of such products. The already been developed. At the same time, sponsors are

CHAPTER 1 Regulatory Considerations 10 advised to apply for the INTERACT meeting early rather product not previously approved or licensed; a new active than late, as this meeting is the only opportunity to engage pharmaceutical ingredient (API) with a novel pharmaco- with the FDA during the pre-IND process. CBER strives to logic mechanism; products for which it is critical to public schedule INTERACT meetings within 21 calendar days and health to have an effective and efficient drug development hold the meeting within 90 calendar days of receipt of the plan (drugs to treat life-threatening or severely debilitating request. INTERACT meetings are held via teleconference illnesses); drugs with substantial early development outside only, and generally last for one hour. the United States; and drugs with adequate and well-con- The INTERACT meeting requests and packages are trolled trials to support a new indication.35 submitted to CBER by email to INTERACT-CBER@fda. The broad range of topics that can be discussed hhs.gov. The INTERACT meeting package should not during a pre-IND meeting may appear overwhelming exceed 50 pages. As with other FDA/sponsor meetings, at first. Instead, it should be seen as an opportunity to it is expected that the sponsor will provide the scientific obtain feedback from the Agency on several topics that rationale to support each question in the meeting pack- can propel the clinical development of a product.36 age. Table 4 provides tips on the Dos and Don’ts for the Once the meeting request is granted, the sponsor must meeting package content. provide the meeting package at least 30 days prior to the meeting date. Preparation for the pre-IND meeting, and Pre-IND other FDA/sponsor meetings, is critical for achieving a Pre-IND meetings are Type B meetings and are meant to productive discussion. The meeting package should pro- initiate or continue the dialogue regarding product devel- vide information relevant to the discussion topics and opment in its early stages, with the aim of understanding enable the FDA to prepare adequately for the meeting.37 the mechanism of action of the drug and possible study It is highly recommended that sponsors initiate the designs. These meetings are valuable to anticipate and briefing package draft at the same time as the meeting potentially prevent clinical hold issues from arising and request. This strategy can facilitate the definition of the aid sponsors in developing a complete IND.34 questions in the meeting request and help the sponsor to FDA encourages sponsors to request a pre-IND meet- avoid delays in getting the final meeting package ready. ing for gene therapy products for the following topics: a A failure to deliver the meeting package 30 days prior to

Table 1-5. Dos and Don’ts for Pre-IND meeting package

Pre-IND Meeting Dos Don’ts Package Content

Description of product Submit the package 30 days prior to the Ask questions for answers that are manufacturing and scheduled meeting. already available in FDA guidance testing. There is no page limit, but it is documents. Completed and planned recommended to be around 100-150 Ask open-ended questions. preclinical study pages. Exceed more than 12 questions summaries. Include relevant CMC information. (including sub-questions). Phase 1 clinical study At a minimum, include a description of the Present new data/information design or protocol. manufacturing scheme for drug substance or alternate approaches during (DS) and drug product (DP), quality of the the meeting in response to the starting materials, release specifications, preliminary FDA feedback. and a stability plan. Include information for device and if the product will be a combination product.

CHAPTER 1 Regulatory Considerations 11 Table 1-6. Summary of Meeting Management Procedural Goals

Requester FDA FDA Response Scheduled FDA Meeting FDA Receipt Preliminary FDA Response to FDA Meeting Date Minutes to Meeting Type of Meeting Responses to to Request Preliminary (days from Requester (if Package Requester (if Responses (if receipt of applicable†) applicable†) applicable†) request)

A 14 days With meeting No later than -- Within 30 days 30 days after request 2 days before meeting meeting

B 21 days No later than No later than -- Within 60 days 30 days after 30 days before 2 days before meeting meeting meeting

B (EOP) 14 days No later than No later than No later than Within 70 days 30 days after 50 days before 5 days before 3 days after meeting meeting meeting receipt of preliminary responses

C 21 days No later than No later than No later than Within 75 days 30 days after 47 days before 5 days before 3 days after meeting meeting meeting receipt of preliminary responses

† Not applicable to written response only.

the meeting can result in the FDA cancelling the meeting. of any additional information necessary to support a While these meetings are free of charge, sponsors marketing application for the uses under investigation. should approach them with deliberate purpose. Typically, All EOP meetings are Type B meetings and subject sponsors should ask no more than 10 to 12 questions, to different timelines as summarized in Table 6.38 which should, naturally, be specific to the sponsor’s product From a CMC perspective, by the time of the end of the and process. Table 5 summarizes the “Dos” and “Don’ts” phase 2 clinical studies, the sponsor should have a very for the pre-IND meeting package, with a focus on CMC. robust knowledge of the manufacturing process and have started preparing for the phase 3 clinical materials Clinical Meetings — End of Phase 1, 2, and 3 that will be representative of the commercial product. End of Phase (EOP) meetings serve to evaluate the For gene therapy products, this timeline is not straight- next clinical phase plan and protocols, the adequacy forward. As previously discussed, the clinical results of current studies and plans to assess safety and effi- in gene therapy products is often ahead of the CMC cacy, and the adequacy of manufacturing and testing development. Clinical phase 2 and phase 3 timelines plans to support the next clinical phase studies. In are condensed and, as a result, the CMC development particular, EOP2 meetings allow for preparation for must be expedited. Therefore, these meetings can commercial manufacturing, evaluation of the human often be almost overlapping with pre-BLA meetings factors validation plan if a device is used for adminis- and preparation is key for obtaining the right feedback tration of the gene therapy product, and identification from the Agency.

CHAPTER 1 Regulatory Considerations 12 In preparation for the EOP2 meeting, the sponsor (including marketing application) for gene (and cell) should take the opportunity to seek advice from the therapies, even though formal recommendation is still FDA that the current data package, in addition to the issued by the CHMP. potential planned studies, will be sufficient for a BLA submission.39 It is recommended that sponsors request ATMP classification a CMC-focused EOP2 meeting to ensure that there is In Europe, ATMPs are governed under the ATMP sufficient time dedicated to CMC discussions. Typical Regulation (Directive 2001/83/EC, as amended by topics discussed during the EOP2 meetings include, but Regulation [EC] 1394/2007). In case of “borderline” are not limited to: release specifications and justifications, product, if the developer is unsure if its product falls in overall control strategy with definitions of critical quality the ATMP category, the developer can request a formal attributes (CQAs) and critical process parameters (CPPs), classification to the EMA. Although the scientific rec- manufacturing process and analytical assay validation ommendation on classification of ATMPs is an optional plans, and stability data to support product storage procedure, there are advantages of requesting one.42 The and shelf-life. At such a meeting, sponsors should also purpose of this request is to allow sponsors to “clarify the discuss readiness/plans for the device (used for product classification whether a given product based on genes, administration) and/or companion diagnostics that will cells, or tissues meets the scientific criteria that define be part of the marketed product. ATMPs, in order to address, as early as possible, ques- tions of borderline with other areas such as cosmetics EMA or medical devices, which may arise as science devel- Early engagement and scientific advice with the EMA are ops.”43 Though it is advised to request this classification key drivers of faster, and, more often, successful registra- before submission of other requests, including scientific tion. Similar to the FDA, the EMA offers several oppor- advice, Pediatric Investigation Plan (PIP) evaluation, tunities for a sponsor to start an early conversation with certification of quality and nonclinical data for Small and the agency to seek scientific, technical, and regulatory Medium-Sized Enterprises (SMEs) developing ATMPs, feedback. Sponsors can request meetings with the EMA orphan drug designation, and Marketing Authorization for overall advice and there is no limit to the number of Application (MAA), it can be submitted at any time scientific advice meetings (or protocol assistance as it during the product development. The CAT, after consul- is called for products with orphan drug status) that can tation with the European Commission (EC), delivers the be requested during the development of a given gene ATMP classification recommendation within 60 calendar therapy. However, clinical trials are still in the remit of days following receipt of the request. national competent authorities. Developers of ATMPs are mandated to seek mar- EMA scientific advice keting application authorization under the centralized Scientific advice can be requested to the EMA at any procedure, along with specific therapeutics of certain stage of the product’s development, although information modalities and for certain indications.40 on the target indication is preferred for the Agency to In the centralized procedure, the CHMP plays a vital provide advice as accurately as possible. There is no limit role in the authorization of medicines in the EU. The to the number of advices that can be requested; nor the CHMP also evaluate medicines authorized at a national number of the questions that can be on CMC/quality as level in a harmonized procedure. In addition, the CHMP well as nonclinical and clinical topics. For products that and its working parties contribute to the development have been granted the orphan drug status, the scientific of medicines and medicine regulations by providing advice procedure is called “protocol assistance” and can scientific advice to develop new medicines, prepare guid- also include questions related to the “significant benefit” ance, and cooperate on harmonization of international and/or “clinical superiority” of the product. regulatory requirements.41 As mentioned previously, Questions in the briefing package should be detailed the CAT is the “central” committee for all procedures and precise and should, in all cases, be followed by a

CHAPTER 1 Regulatory Considerations 13 justification of the company’s planned strategy with regard Following completion of the procedure and receipt to the question, all relevant information about the topic, of the final advice letter (FAL), the requester may ask and cross-references to any relevant annexes. It is highly for clarification if it disagrees with, suspects a misun- recommended that the sponsors provide a clear strategy, derstanding, or spots contradictions or imprecisions in compelling argument, and well-rehearsed preparation, as the advice given by CHMP. This will not lead to a new these are key to having a successful EMA meeting.44 discussion and is purely in writing, but may help clarify The briefing package should contain the following: some wording in the FAL, for example. Submission deadlines for scientific advice/protocol • Background information about the product and its assistance are published every year for the following mechanism of action calendar year. • CMC and quality data • Preclinical data National Competent Authorities • Clinical data In the European regulatory ecosystem, sponsors can • Intended indications interact not only with the centralized EMA, but also • Regulatory status with national agencies, which will give specific recom- • Stage of program development mendations on the clinical trials. National Competent Authorities (NCAs) often offer significant contributions • Stage of clinical study development to the product’s development plans, including critical • Questions for the reviewers and applicant’s points on how to define the starting materials, and how justification (“position”) to define drug substance (DS) and drug product (DP), The level of information included will vary depending on as these can be not well-defined during development in the stage of development of the product and the topics of a continuous manufacturing process. The EMA works the questions. Annexes can be included in the package, closely with the NCAs of the Member States of the EU if deemed appropriate to provide further information. and the European Economic Area (EEA) responsible for Scientific advice/protocol assistance can be a written human medicines.46 procedure only (40 days) or include a discussion meeting For gene therapy products specifically, it is recom- with the Agency (70 days procedure). Of note, the deci- mended to engage discussions with NCAs early in the sion of a discussion meeting is at the Agency’s discretion. development process, especially if the developer plans to Additionally, the applicant can request a “preparatory conduct the clinical development (i.e., clinical trials) in meeting” that will take place before the submission of one or several EU member states. Though this does not the final package and will help to refine the format of rule out validation or questions following clinical trial the document. No assessment or in-depth review will applications (CTAs), it will enable the NCAs to be aware take place; however, it might be relevant for an applicant of the development of the product and upcoming CTA. submitting for the first time to the EMA that wants to maximize the chance of getting appropriate advice. Parallel Consultation Applicants should give a presentation at the meeting In addition, early discussion with health technology assess- and submit a draft version of the briefing document a ment (HTA) bodies and other stakeholders can be critical few days ahead that includes questions based upon what towards early patient access and commercial success in EMA staff will review, including, but not limited to the Europe. Some initiatives have been implemented over the following: overall compliance of the intended submission last years to facilitate such dialogue, via, for instance, the package with applicable regulatory requirements, pos- parallel consultation where EMA and HTA provide simul- sible gaps in knowledge that could be useful to discuss, taneous advice to a developer. The importance of seeking documentation against relevant scientific and regulatory parallel HTA-EMA advice and being well-prepared for guidelines about products in the same class, and relevant the meeting are critical to ensure successful development, guidance.45 followed with registration and commercial success.47

CHAPTER 1 Regulatory Considerations 14 The main benefits of the parallel consultation proce- guidelines or for those indications for which existing dure include: EMA and FDA guidelines differ significantly. Sponsors wishing to nominate a product for PSA • Streamlined procedure should address a single “Request for PSA” letter to both • Increased mutual understanding and problem- [email protected] and OC-OIPEurope@ solving ability between EMA and HTA bodies fda.hhs.gov. In this letter, the sponsor should provide • Improved coordination with HTA bodies and greater information about the following: the product in de- participation of HTA bodies in parallel consultations velopment; why a discussion with EMA and FDA staff through EUnetHTA’s Early Dialogue Working Party would be beneficial to the product’s development; spe- (EDWP) and the EUnetHTA early dialogue (ED) cific questions requiring clarification; the desired goals secretariat.48 for the meeting; and an explicit authorization for the agencies’ comprehensive exchange of all information EMA-FDA PROGRAMS relevant to the product, including trade secret informa- tion. Any fees applicable for scientific advice at either Parallel Scientific Advice agency are unaffected by PSA status. If both agencies In addition to the separate interactions with the EMA grant the PSA request, the sponsor will receive an email and FDA, these two major agencies offer the Parallel from each agency acknowledging the agreement and Scientific Advice (PSA) program in order to provide a indicating the primary contact person at each agency. mechanism for staff from both EMA and FDA to con- The PSA process generally corresponds to the 70-day currently convey to sponsors their views on scientific timeline of SAWP at EMA and the timeline for a Type issues during the development phase of new medicinal B meeting at FDA. The designated primary contact for products. These interactions are meant to increase dia- each agency will coordinate with the sponsor regarding logue between the two agencies and sponsors from the final meeting logistics, including timelines for submis- beginning of the lifecycle of a new product, provide a sion of pre-meeting background information to both deeper understanding of the basis of regulatory deci- agencies. The two agencies will conduct a pre-sponsor sions, optimize product development, and avoid unnec- meeting tele- or video conference (usually around day essary testing replication or unnecessary diverse testing 60 of the 70-day timeframe) to discuss the sponsor’s methodologies. The agencies conduct PSA procedures questions prior to the meeting. The two agencies may according to the confidentiality arrangement between also conduct a post-sponsor tele- or video conference the European Commission, EMA, and FDA.49 if needed.50 PSA procedures usually occur at the request of the If a sponsor’s request for PSA is not granted, the sponsor, though in special circumstances, EMA or FDA sponsor is free to pursue a scientific advice procedure may also initiate the PSA process in full cooperation with with each agency individually, following each agency’s the sponsor. PSA requests should focus on specific ques- normal procedural rules. Both agencies may also engage tions or issues involving the development of a medicinal in a Consultative Advice procedure, as described below. product for which the sponsor desires to gain further scientific input from both EMA and FDA. The PSA Consultative Advice procedures should focus on sharing information and The Consultative Advice procedure allows sponsors to perspectives. Following PSA meetings, sponsors should request scientific advice from one regulatory agency have a clearer understanding of the agencies’ respective and concurrently notify the other regulatory agency of requirements and perspectives regarding the develop- the request. At the invitation of the first agency, the ment program discussed, and if divergent, the reasons second will participate in the sponsor meetings or tele- for the divergence. FDA and EMA consider the best conferences, as able. Unlike the PSA process, the second candidates for PSA to be important medicinal products agency will be expected to only engage on top level being developed for indications lacking development issues. The review and sponsor meeting will follow the

CHAPTER 1 Regulatory Considerations 15 timelines of the regulatory agency with whom the sponsor prior to the meeting. Meeting minutes from the PMDA initially seeks scientific advice. Only the initially contacted are provided 30 days after the meeting. regulatory agency will provide written scientific advice in In addition to the pre-phase 1 consultation, the PMDA accordance with standard agency meeting procedures. has implemented new consultations as of 2011 to promote the practical application of innovative drugs, medical PMDA devices, and regenerative medical products originating in In Japan, the PMDA provides opportunities for meetings Japan. These consultations provide significant benefit for between sponsors and the Agency to allow for feedback universities, research institutions, and venture companies and guidance during clinical development.51 In clinical that are involved in the discovery of promising “seed- trial consultations for new drugs, PMDA checks whether stage” technologies. These early consultations have been a proposed clinical trial complies with the requirements very beneficial to developers of gene therapy products. for regulatory submission, taking into consideration More recently, in April 2018, PMDA started to provide the ethical and scientific aspects of the development the Collaborative Consultation on Practical Application program, the reliability of the clinical trial, as well as the of Innovative Products while sharing information with safety of trial subjects. The PMDA also gives advice to the Medical Innovation Support. The consultation ser- facilitate the improvement of the clinical trial. vice also provides guidance and advice on the quality and Since 2009, PMDA started providing prior assessment safety of regenerative medical products (including gene consultations, which adds value to the development pro- therapy products intended for transgene expression in cess through feedback from reviewers on CMC data, in the human body and used to prevent diseases (e.g., live addition to efficacy and safety feedback on the product. recombinant vaccines)) at an early development stage.52 This consultation process constitutes part of the review Additional consultations with the PMDA for times of the product once the application is submitted. in the development process applicable to CMC develop- For sponsors that are located outside of Japan, it ment are available and include:53 is recommended to appoint a Japanese Marketing • Before start of early phase 2 study Authorization Holder (MAH). The sponsor can request • Before start of late phase 2 study meetings with the PMDA through the MAH, who can • After completion of phase 2 study also assist with translation and interpretation, since all communications and submission forms are in Japanese. • Prior Assessment Quality consultation

Preclinical and clinical meetings REGISTRATION MEETINGS The pre-phase 1 study consultation is an opportunity for sponsors to obtain guidance from the PMDA prior to ini- FDA: Pre-BLA tiation of the clinical study in Japan. The goal of the early Pre-BLA meetings are meant for FDA reviewers to phase consultation is to solve potential issues in clinical provide advice to the sponsor regarding the format and development, identify tests that will be needed in the early content of the planned marketing application, including product development stage, and shorten the time before labeling and risk management activities, presentation the application, saving time and costs by avoiding critical and organization of data, dataset structure, acceptability issues during development, as well as during NDA review. of data for submission, and the projected submission The process for obtaining a meeting with the PMDA date of the marketing application. They are also intended can take eight weeks from the acceptance of the meeting to uncover major issues, identify studies intended to es- request to the face to face or online meeting. Five weeks tablish the drug’s safety and efficacy, discuss the status of prior to the meeting, the briefing package is submitted. pediatric studies, and discuss statistical analysis methods During the review of the briefing package, PMDA may and results. FDA encourages sponsors to request pre- ask questions to which the sponsor must respond in a BLA meetings for all planned marketing applications.54 timely manner. PDMA provides the opinion four days Sponsors should plan for a single-multidisciplinary

CHAPTER 1 Regulatory Considerations 16 pre-BLA meeting because the FDA only grants one pre- to permit marketing in the different EU countries. The BLA meeting. EC is the authorizing body for all centrally authorized The timelines for meeting request and submission of products, and makes a legally binding decision based on the meeting packages follow the Type B meeting require- EMA’s recommendations. This decision is issued within ments. Once the meeting request is granted, the sponsor 67 days of receipt of EMA’s recommendations. Once must provide the meeting package at least 30 days prior granted by the European Commission, the centralized to the meeting date. The meeting package should provide marketing authorization is valid in all EU Member information relevant to the discussion topics and enable States, as well as in the EEA countries of Iceland, the FDA to adequately prepare for the meeting. A fail- Liechtenstein and Norway. ure to deliver the meeting package 30 days prior to the Commission decisions are published in meeting can result in the FDA cancelling the meeting. the Community Register of medicinal products for hu- The briefing package for a pre-BLA meeting, or any man use. The national competent authorities are primarily other pre-marketing authorization meeting, should be responsible for the authorization of medicines available adequately prepared to ensure that there are no surprises in the EU that do not pass through the centralized pro- during the review of the BLA. Sponsors may provide cedure. They also supply thousands of European experts summaries of their stability and process validation who serve as members of the Agency’s scientific commit- studies for comment on adequacy, but the FDA will tees, working parties, or in assessment teams supporting not comment on final product specifications, shelf their members. life, or actual sufficiency of process validation in a pre- Once the eligibility to the centralized procedure at BLA meeting. Sometimes, the timing for the pre-BLA the EMA has been confirmed (a mandatory step even for meeting needs to align with the timing of availability of gene therapies) and the rapporteur and co-rapporteur of data to be presented. If the meeting is done too early, the procedure have been appointed, the future marketing the FDA will not be able to provide full feedback on authorization holder (MAH) can request two different questions and will postpone decisions to the review of meetings prior to the submission of the MAA dossier. the BLA. If the meeting is done too late, and too close to the BLA submission, there is not sufficient time to EMA: presubmission meeting generate more data prior to submission. To maximize Presubmission meetings offered by the EMA are in- the chance for success, requests for pre-BLA meetings tended to address product-specific legal, regulatory, and should be planned together with the overall regulatory scientific issues, to facilitate the validation of the MAA, strategy and timelines for product development. This and to support applicants in submitting applications for approach is applicable not only to meetings with the smooth evaluation. Sponsors may discuss final practical FDA, but also other HAs. and regulatory aspects of their upcoming application and clarify application-specific issues not addressed on the EMA: presubmission meeting and rapporteur/ EMA website. co-rapporteur meeting In addition to the presubmission meeting held with Under the centralized authorization procedure, pharma- the EMA, the applicant of an MAA can request a meeting ceutical companies submit a single MAA to the EMA. with the appointed rapporteur and co-rapporteur, the This allows the MAH to market the medicine and make key experts who will drive the assessment. it available to patients and healthcare professionals This meeting is not meant to discuss procedural throughout the EU on the basis of a single marketing aspects but to present the product and key informa- authorization. tion, as well as discuss questions and potential issues. CHMP carries out a scientific assessment of the ap- Importantly, the advice given during this meeting is plication and gives a recommendation on whether the non-binding and informal; it does not preclude the medicine should be marketed or not. outcome of the MAA but will provide some insights on However, under EU law, the EMA has no authority potential “weaknesses.”

CHAPTER 1 Regulatory Considerations 17 POTENTIAL CMC TOPICS TO BE COVERED IN HA/ Submission Content SPONSOR MEETINGS Different elements in gene therapy development that are ELECTRONIC COMMON TECHNICAL DOCUMENT considered essential and typical topics of interest are usu- (ECTD) ally raised during these interactions with HAs that start The eCTD, and electronic submission structure devel- with early product development. The types of potential oped by the ICH, provides the backbone for providing questions asked during these meetings between HAs and information regarding Chemistry, Manufacturing, sponsors may or may not be evident to sponsors when and Controls (CMC) in Module 3, with a summary in developing their gene therapy products. Module 2.3 (Quality Overall Summary).55 These two A key challenge is the selection and the quality of the modules include sections for DS and DP. Information to raw materials. Often, the early development process of be provided about the DS includes the proper identifica- gene therapy products will utilize research grade mate- tion, quality, purity, and strength of the active ingredient, rials that could compromise the quality of the product with an emphasis on the identification and control of when progressing into clinical manufacturing. Another raw materials and the new drug substance. Information challenge regarding starting materials is the utilization to be provided about the DP is similar to that required of materials of human or animal origin that can be of for the DS section, with information about the assays insufficient quality for clinical studies. Engaging in early and acceptable results for assessing identity, strength, discussions with the Agency can positively impact deci- quality, and purity. Additionally, information about sions made in the product development process. stability for at least the duration of the clinical trial, with Another key challenge is the proper development of a the purpose of establishing the drug product shelf-life potency assay. Although the requirements for a potency and recommended storage conditions, are expected.56 assay in the U.S. and EU differ during early stage develop- It should be noted that these data are expected to ment, it is highly encouraged that sponsors start the devel- evolve over time as the sponsor optimizes production opment of one or more potency assays that can successfully processes, analytical methods, and formulation of the demonstrate the mechanism of action of the product. Even drug product. It can be challenging to fit information if the method is still undergoing optimization, early discus- about gene therapy products into the eCTD structure sions with the Agency can help with the proper design and due to the lack of clear delineation between DS and DP, phase-appropriate implementation of the assay. as well as the specific need to control and release critical A few questions that are commonly asked during starting materials. early interactions with the three agencies are listed below. Both IND and clinical trial application (CTA) As expected, these questions should not be followed as submission content directly relating to CMC is to be a recipe, but are intended to provide some guidance on submitted in documentation structured according to the points to consider prior to requesting a meeting. heading of the corresponding sections of Module 3 of the eCTD. • Does the Agency agree on the proposed quality and selection criteria (including testing scheme) for starting IND SUBMISSIONS TO FDA materials and raw materials used in DS and subsequent Sponsors who wish to conduct a clinical trial in the U.S. DP manufacturing? must submit an IND. The FDA’s review of the IND takes • Does the Agency agree on the suitability of the 30 days. From the perspective of CMC, FDA will focus characterization studies and proposed specification for on determining if there are any reasons to believe the DS and DP release testing? manufacturing or controls for the clinical trial product • Does the Agency agree on the stability plan for present unreasonable health risks to the subjects in the DS and DP? initial IND trials; as always, safety is the first priority. • Does the Agency agree on the comparability plan When filing an initial IND submission, details proposed for the nonclinical and clinical batches? about the following CMC information are presented

CHAPTER 1 Regulatory Considerations 18 Table 1-7. Summary of CMC information for Module 3 (Drug Substance)

Main Sections Gene Therapy Recommended Content

3.2.S.2.2 Manufacturing process description and process controls should include the following, as Description of applicable: cell culture; transduction; cell expansion; harvest(s); purification; filling; and storage Manufacturing and shipping conditions. The description of your manufacturing process should include a process Process flow diagram(s) and a detailed narrative. and Process A description of how you define each manufacturing run (i.e., batch, lot, other) should be Controls submitted with an explanation of the batch numbering system. Indicate whether any pooling of harvests or intermediates occurs during manufacturing. Pooling may be needed as some gene therapy batches are made at a low scale. Any reprocessing during manufacture of the active substance should be described and justified. Your description should clearly identify any process controls and in-process testing (e.g., titer, bioburden, viability, impurities) as well as acceptable operating parameters (e.g., process times, temperature ranges, cell passage number, pH, CO2, dissolved O2, glucose level). Even in early stages, monitoring process performance parameters is recommended. Describe any controls for cleaning and changeover as well as tracking and segregation procedures that are in place to prevent cross-contamination. For the IMPD, in addition to the items listed above, the guidelines specify that the rationale for the use of a particular cell substrate should be provided. Additionally, a purification process should be in place to reduce impurities. This is also expected by the FDA. Impurities include hybrid viruses in the case of virus vector production, host cell-DNA and protein, residual plasmid DNA, lipids and polysaccharides in the case of production systems which involve bacterial fermentations, and RNA and chromosomal DNA in the case of plasmid purification. For the IMPD, for non-replication competent viral vectors and conditionally replicating viral vectors, information should be provided on process parameters, and controls conducted to prevent contamination of the packaging cell line by wild-type, helper, or hybrid viruses that might lead to the formation of replication-competent recombinant viruses during production. For the FDA, this is expected in the IND submission, as this is a safety issue.

3.2.S.2.3 In this section, you must provide a list of all materials used in manufacturing and a description of Control of the quality or grade of these materials. Typically, this is presented in a tabular format, including Materials the identity of the material, the supplier, the quality, the source of material, and the stage at which each material is used in the manufacturing process. It is recommended that for all materials used, the acceptance criteria be included, or at a minimum, the Certificate of Analysis (CoA) for each material be included. This includes information on components, such as cells, cell and viral banking systems, and reagents, and also includes raw materials and equipment that come into contact with the product, such as culture bags, culture flasks, chromatography matrices, and tubing. Sponsor should note equipment that is single use, product-dedicated or intended for multiproduct use. It is important to note that the terminology used in the U.S. and EU for materials is slightly different. Therefore, when preparing a global dossier to support submission for the IND and IMPD, these differences should be taken into consideration. According to the FDA, materials used for manufacturing (e.g., cell growth, differentiation, selection, purification, or other critical manufacturing steps) that are not intended to be part of the final product are called reagents or ancillary materials, while according to the EMA, these are called raw materials. For biologically sourced reagents, the FDA guidance for industry “Characterization and Qualification of Cell Substrates and Other Biological Materials Used in the Production of Viral Vaccines for Infectious Disease Indications” is applicable. Specific recommendations for each type of biological material is presented in the following: https://www.fda.gov/media/113760/download. For the IMPD, specific guidance is provided for all raw materials of biological origin in Ph.Eur. (5.2.12) “Raw Materials for the Production of Cell based and Gene Therapy Medicinal Products.” continued on next page

CHAPTER 1 Regulatory Considerations 19 Table 1-7. Summary of CMC information for Module 3 (Drug Substance) continued from previous page

Information for starting materials should be organized to include description of the cell source, collection procedure, and any related handling, culturing, processing, storage, shipping, and testing performed prior to use in manufacture. When using allogeneic cells or tissues, you must perform donor screening and testing, as required in 21 CFR Part 1271, Subpart C, except for those cells and tissues that meet the exceptions in 21 CFR 1271.90(a) for the IND. For the IMPD, donation, procurement and testing of human cell-based products need to comply with the requirements of Directive 2004/23/EC or, where applicable, Directive 2002/98/EC. The cell banking systems should be presented at a minimum for the Master Cell Bank (MCB) and, if applicable, the Working Cell Bank (WCB). Information is expected on the cell banking and characterization and testing of the established cell banks, as well as available information on cell substrate stability. For both the IND and IMPD, the safety assessment for adventitious viruses should be presented in 3.2.A.2. Additional sources of information regarding qualification of cell substrates can be found in ICH Q5D.

You should provide a description and discussion of the developmental history of the 3.2.S.2.6 manufacturing process as described in 3.2.S.2.2. For early stage INDs, there may be differences Manufacturing between the manufacturing and testing of the nonclinical and clinical batches. It is crucial Process to clearly present, typically in a tabular format, the changes made in the process with an Development assessment of the impact of these changes on the quality of the product, as well as a rationale for the change. This provides a clear view for the reviewer and allows for a better understanding of the process development. If you make significant manufacturing changes, then comparability studies may be necessary to determine the impact of these changes on the identity, purity, potency, and safety of the product. Comparability reports including information on the changes put in place, the risk assessments performed, the testing strategy, and the supporting data collected to demonstrate comparability should be submitted.

3.2.S.3.1 Characterization studies will provide a comprehensive picture and knowledge of the gene Elucidation therapy product. The knowledge will evolve as the product development progresses. In this of Structure section, include annotated sequence analysis for your vector in the original IND submission and and other any additional sequence information gathered during the course of product development in Characteristics subsequent submissions. Provide any further information confirming the primary, secondary, or higher order structure; post-translational modifications; and/or distribution of cell types for the DS if it has not already been described in 3.2.S.1.2. For the IMPD, it is clearly stated that reference to the literature data alone is not acceptable. Basically, the characterization of a gene therapy active substance is necessary to allow relevant specifications to be established. Tests should be included to show integrity and homogeneity of the recombinant viral genome or plasmid and the genetic stability of the vector and therapeutic sequence. 3.2.S.3.2 Information on process-related and product-related impurities should be provided. Impurities Process-related impurities (e.g., media residues, growth factors, host cell proteins, host cell DNA, column leachables) and product-related impurities (e.g., cell types not linked to the therapeutic effect, cell fragments or non-viable cells, precursors, degradation products, and aggregates) should be kept to the minimum or a risk assessment should be provided. Based on the risks identified, consideration should be given to the maximum amount for the highest clinical dose and an estimation of the clearance should also be provided. Overall, the manufacturing process should be designed to remove the impurities to levels that are acceptable and justifiable.

continued on next page

CHAPTER 1 Regulatory Considerations 20 Table 1-7. Summary of CMC information for Module 3 (Drug Substance) continued from previous page

3.2.S.4.1 You should list DS specifications in your original IND/IMPD submission. Specifications are Specifications defined as a list of tests, references to analytical procedures, and appropriate acceptance criteria used to assess safety and quality. Since the acceptance criteria are normally based on a limited number of development and nonclinical batches, it is understandable that they are preliminary and need to be optimized during development. It is important to emphasize that it is better to present a wider range for acceptance criteria than to not present one at all. Even at early stages, presenting acceptance criteria as “report results” are often questioned by the Health Authorities, as these specifications do not demonstrate that you will have control of the quality of your product.

3.2.S.7 Stability Describe in your original IND submission the types of stability studies (either conducted or planned) to demonstrate that the DS is within acceptable limits. The protocol should describe the storage container, formulation, storage conditions, testing frequency, and specifications (i.e., test methodologies and acceptance criteria). As is the case for several gene therapy products, if the DS is immediately processed into a DP, long term DS stability data may not be needed. In the stability protocol, it is often helpful to demonstrate that at least one or more of the test methods in your stability analysis are stability-indicating. Although this is not required for early clinical trials, you can demonstrate a test is stability-indicating by using forced degradation studies, accelerated stability studies, or another type of experimental system that demonstrates product deterioration. When sufficient stability data is not available for the clinical batch, stability data for at least one batch representative of the manufacturing process of the clinical trial material should be included. Any other stability data relevant from development and nonclinical batches can be provided as supportive data. For the IMPD, vector integrity, biological activity (including transduction capacity) and strength are critical product attributes that should always be included in stability studies. In addition, if a shelf-life extension is planned, the applicant should commit to performing the proposed stability program according to the presented protocol and inform the Competent Authorities in the event of unexpected issues.

in an eCTD structure and should include: (1) DS; (2) prevent the clinical hold with the sponsor before issuing DP; (3) placebo formulation, if applicable; (4) labeling the clinical hold. information for the labeled products relevant to the FDA also acknowledges there can be specific chal- investigational drug; and (5) an environmental analysis lenges for applications that have received (or are likely for assessment of the effects of the investigational new to receive) expedited designations. These challenges drug or biological product on the environment (though include possible difficulties in aligning CMC and clinical many gene therapy products qualify for an exemption development and possible difficulties in making risk/ from this assessment, some may not).57,58 benefit assessments (with particular regard to patient If the FDA identifies an unresolved safety issue benefit) in situations in which there may be a relative (CMC, clinical, or nonclinical) in the IND, or if FDA lack of CMC information. These situations are consid- identifies such an issue arising during development, ered on a case-by-case basis. the Agency will issue a clinical hold on the application. Once an IND has been deemed safe to proceed by the Regulations require the FDA to attempt to discuss FDA, multiple studies can be conducted under the same and satisfactorily resolve any resolvable issue that may IND, as per the FDA’s legal requirements described in 21

CHAPTER 1 Regulatory Considerations 21 Table 1-8. Summary of CMC information for Module 3 (Drug Product)

3.2.P.1 Drug Product You should provide a description of the DP and its composition (21 CFR 312.23(a) Description and (7)(iv)(b)). This includes a description of the dosage form and a list of all of its Composition components (active and inactive), the amount on a per unit basis, the function, and a reference to quality standards for each component (e.g., compendial monograph or manufacturers’ specifications). If a placebo treatment is used in the clinical trial, a separate DP section should be provided for the placebo. In addition, you should provide a description of any accompanying reconstitution diluents and a description of the container and closure used for the dosage form and accompanying reconstitution diluent in a separate DP section (3.2.P. Diluent), if applicable. The 3.2.P. Diluent section should contain all the information for DP diluent manufacturing, testing, and stability.

3.2.P.2 The Pharmaceutical Development section should contain information on Pharmaceutical the development studies conducted to establish that product formulation, Development manufacturing process, container closure system, microbiological attributes, and instructions for use are appropriate for the stage of clinical development. In early stages of development, it is acceptable that limited information is available. Most importantly, and similarly to 3.2.S.2.6, any changes in the process and formulation of the drug product from the nonclinical to the clinical batches should be clearly identified. Compatibility studies (or in-use stability studies) should be included to support recommended hold times and conditions outlined in the clinical protocol for patient administration. It should be demonstrated that the specified reconstitution or preparation process is sufficiently robust and consistent to ensure that the product fulfils the specifications and can be administrated without negative impact on quality/safety/clinical properties.

3.2.P.5.1 DP specifications should be listed. Your testing plan should be adequate to describe Specifications the physical, chemical, or biological characteristics of the DP necessary to ensure that the DP meets acceptable limits for identity, strength (potency), quality, and purity (21 CFR 312.23(a)(7)(iv)(b)). For IND and IMPD, tests for contents, identity and purity are mandatory. Tests for sterility and endotoxin are mandatory for sterile products. For the IMPD: a potency test should be included unless otherwise justified. For the IND, some measure of potency is required but a biologically relevant potency test is not mandatory. Although typically not required, the need for a biologically relevant potency test is highly dependent on the patient population and type of clinical trial proposed.

3.2.P.8 Stability You should summarize the types of studies conducted, protocols used, and the results of the studies. Your summary should include, for example, conclusions regarding storage conditions and shelf-life, as well as in-use and in-device storage conditions. If a short-term clinical investigation is proposed, or if a DP manufacturing process has limited product hold times, stability data submitted may be correspondingly limited. Early in development, stability data for the gene therapy product may not be available to support the entire duration of the proposed clinical investigation. Therefore, we recommend that you submit a prospective plan to collect stability information and update this information to the IND in a timely manner (e.g., in an annual IND update).

CHAPTER 1 Regulatory Considerations 22 Figure 1-5. ICH Guideline Development Process

STEP 5 Implementation

STEP 4 Adoption of an ICH Harmonized Guideline

STEP 3 Regulatory Consultation and Discussion

STEP 2 a. ICH Parties Concensus on Technical Document b. Draft Guidelines Adoption by Regulators

STEP 1 Consensus Building — Technical Document

CFR 312.22. These studies must use the same investi- MODULE 3 CONTENT gational drug but do not require the same indication. The CMC content for the Module 3-structured IND After the initial clearance, subsequent protocols can and IMPD will highly depend on the specificities of the be initiated immediately after submission of the IND gene therapy product in terms of the level of information for agency review without a statutory waiting period provided. Table 7 and Table 8 summarize recommenda- as long as appropriate supporting documents are also tions for the main Module 3 sections for Drug Substance submitted. and Drug Product, respectively,59,60 on how to provide sufficient CMC information required to assure product CLINICAL TRIAL APPLICATION (CTA) safety, identity, quality, purity, and strength of the inves- Sponsors who wish to conduct clinical trials in the EU tigational product for IND and IMPD submissions. The must do so by submitting a CTA. The authorization and tables combine the recommendations from EMA and oversight of clinical trials remains the responsibility of FDA. Recommendations applicable to one HA only are Member States until full implementation of regulation clearly identified as such. Overall recommendations on N°536/2014 and the Clinical Trials Information System manufacturing process information to be included in the (CTIS). sections for DP are similar to the ones recommended for The central document required for the CTA is the the DS and therefore have not been repeated. investigational medicinal product dossier (IMPD), which contains the quality information. The quality Lifecycle Management data are presented according to the heading structure of the eCTD. It should be noted that CMC information Novel gene therapies currently being commercialized in the IMPD is subject to specifications not only issued represent a new genre of medicinal products. As noted by EMA, but also to European Pharmacopoeia (Ph Eur) earlier, Health Authorities are interested in advancing monographs and the European Directorate of the Quality manufacturing technologies among other innovations. of Medicines (EDQM) standard terms database. One of the challenges to innovation is the burden of Data requirements are known to evolve as develop- global change management. Thus, the International ment progresses from exploratory to confirmatory clini- Council for Harmonisation of Technical Requirements cal trials. As such, quality data compiled in the IMPD are for Pharmaceuticals for Human Use (ICH) developed the to reflect increasing knowledge and experience during ICH Q12 Lifecycle Management Guidance: “Technical product development. and Regulatory Considerations for Pharmaceutical Unlike INDs, CTAs are not subject to clinical holds; Product Lifecycle Management,”63 which provides a the CTA is either approved (perhaps with mandatory framework to facilitate management of post-approval changes) or rejected. changes in CMC. The guidance has achieved Step 4 in

CHAPTER 1 Regulatory Considerations 23 the ICH process and is pending adoption and publication countries, this tool will lead to global simplification of in various regulatory regions. The guidance development management of changes and acceptance. process is depicted in Figure 5. Once adopted, this guidance will provide a global PRODUCT LIFECYCLE MANAGEMENT DOCUMENT framework to further innovation and implement changes (PLCM) during the lifecycle of the product. Product Lifecycle Management (PLCM) is a new con- The scope of the guidance includes biologics, even if it cept for industry and Health Authorities. The PLCM does not specifically reference gene therapies. Tools within document is a living document that will be a central this guidance can be used to support post-approval changes repository, preferably in tabular format, for Established for gene therapy medicinal products. Thesetools include: Conditions, PACMPs and Post Approval Commitments in each region. This document is to be submitted in the • Established Conditions regional section of Module 3. • Post Approval Change Management Protocols (PACMPs) Conclusion • Product Lifecycle Management Strategy (PLCM) The regulatory framework governing gene therapy ESTABLISHED CONDITIONS products can pose a large degree of complexity for Established Conditions are legally binding information developers. In addition, in comparison to more tradi- also referenced as “registered details” considered neces- tional biopharmaceutical products, the field is relatively sary to assure process performance and desired quality in immature. Therefore, both development efforts and the product. Changes to Established Conditions require regulatory guidelines are evolving. The FDA, EMA, and regulatory action and the classification of the regulatory PMDA provide a range of opportunities for developers change is driven by risk-to-product quality and control. operating in the U.S., EU, and Japan markets to meet, Regulatory action is defined by two categories: (1) Prior: discuss, and gain clarification on various aspects of the approval that requires regulatory authority review and gene therapy product development process, including approval prior to implementation of the change; and (2) topics relating to early phases of development, CMC, Notification High/Low: does not require prior approval and clinical trials. In addition, references specific to gene for implementation. Non-established conditions do not therapy products are starting to be addressed within the require regulatory action and are managed internally ICH. Materials developed and released by the ICH will within the sponsor’s Product Quality System (PQS). provide guidance on a more general level and can be referenced as development progresses. Furthermore, POST APPROVAL CHANGE MANAGEMENT the FDA, EMA and PMDA offer a number of expedited PROTOCOLS (PACMP) regulatory pathways to support and facilitate the inno- A Post-Approval Change Management Protocol vation and therapeutic value promised by gene therapy (PACMP), as defined in ICH Q12, is a two-step process products. As the gene therapy field continues to mature, that allows (1) the sponsor to define the specific change it is expected that the corresponding regulatory struc- they would like to implement, assessment of the change, tures and systems will continue to gain knowledge and and how the change will be managed, reviewed, and ap- experience from accumulated data, which will, in turn, proved by HA prior to execution; and (2) a proposed re- allow both developers and regulators to move forward duced reporting category for the change when all criteria with ensuring that patients are given access to the safest are met. This process can accelerate implementation of and most efficacious products possible. a change by leading to approval for a reduced reporting category. Currently, this tool is acceptable in the U.S. and EU; however, upon adoption of Q12 among all ICH

CHAPTER 1 Regulatory Considerations 24 Appendix Abbreviations

API active pharmaceutical ingredient GMP Good Manufacturing Practice ATMP Advanced Therapy Medicinal Products HA Health Authority BTD Breakthrough Therapy ICH International Council for Harmonisation CAT Committee for Advanced Therapies IMPD investigational medicinal product dossier CATT CBER Advanced Technologies Team IND New Drug Application CBER Center for Biologics Evaluation and INTERACT Initial Targeted Engagement for Research Regulatory Advice CHMP Committee for Medicinal Products for ITF Innovation Task Force Human Use MAA Marketing Authorization Application CMC Chemistry, Manufacturing, and Controls MAH Marketing Authorization Holder CPP critical process parameters MHLW Ministry of Health, Labor, and Welfare CQAs critical quality attributes OTAT Office of Tissues and Advanced Therapies CTA clinical trial application PACMP Post Approval Change Management CTIS Clinical Trials Information System Protocol DNA Deoxyribonucleic acid PDUFA Prescription Drug User Fee Act DP drug product Ph Eur European Pharmacopoeia DS drug substance PIP Pediatric Investigation Plan EC European Commission PLCM Product Lifecycle Management Strategy eCTD electronic Common Technical Document PMD Act The Act on Pharmaceuticals and Medical ED early dialogue Devices PMDA EDQM European Directorate of the Quality of Pharmaceuticals and Medical Devices Medicines Agency pre-BLA EDWP Early Dialogue Working Party pre-Biologics License Application pre-IND EMA European Medicines Agency pre-investigational new drug application PRIME EOP End of Phase Priority Medicines PSA EOP1 end-of-phase 1 Parallel Scientific Advice RMAT EOP2 end-of-phase 2 Regenerative Medicine Advanced Therapy RNA EOP3 end-of-phase 3 Ribonucleic Acid SME EU European Union Small and Medium-Sized Enterprises U.S. United States FAL final advice letter FDA Food and Drug Administration

CHAPTER 1 Regulatory Considerations 25 Endnotes

1. U.S. Food and Drug Administration. What is Gene Therapy? https:// Programs for Regenerative Medicine Therapies for Serious Conditions. www.fda.gov/vaccines-blood-biologics/cellular-gene-therapy-products/ https://www.fda.gov/regulatory-information/search-fda-guidance-docu- what-gene-therapy. July 2018. ments/expedited-programs-regenerative-medicine-therapies-serious-condi- 2. U.S. Food and Drug Administration. Guidance for Industry: Chemistry, tions. February 2019. Manufacturing, and Control (CMC) Information for Human Gene Therapy 17. U.S. Food and Drug Administration. Priority Review. Investigational New Drug Applications (INDs). https://www.fda.gov/ https://www.fda.gov/patients/fast-track-break- regulatory-information/search-fda-guidance-documents/chemistry-manu- through-therapy-accelerated-approval-priority-review/ facturing-and-control-cmc-information-human-gene-therapy-investigation- priority-review. al-new-drug. January 2020. 18. U.S. Food and Drug Administration. Accelerated 3. U.S. Food and Drug Administration. Guidance for Industry: Testing of Approval. https://www.fda.gov/patients/fast-track-break- Retroviral Vector-Based Human Gene Therapy Products for Replication through-therapy-accelerated-approval-priority-review/ Competent Retrovirus During Product Manufacture and Patient Follow-up. accelerated-approval. https://www.fda.gov/regulatory-information/search-fda-guidance-docu- 19. European Medicines Agency. Accelerated assessment. https://www. ments/testing-retroviral-vector-based-human-gene-therapy-products-repli- ema.europa.eu/en/human-regulatory/marketing-authorisation/ cation-competent-retrovirus-during. January 2020. accelerated-assessment. . 4 U.S. Food and Drug Administration. Guidance for Industry: Evaluation 20. European Medicines Agency. Accelerated assessment. https://www. of Devices Used with Regenerative Medicine Advanced Therapies. https:// ema.europa.eu/en/human-regulatory/marketing-authorisation/ www.fda.gov/regulatory-information/search-fda-guidance-documents/eval- accelerated-assessment. uation-devices-used-regenerative-medicine-advanced-therapies. February . 2019. 21 Ministry of Health, Labour and Welfare. Strategy of SAKIGAKE. https://www.mhlw.go.jp/english/policy/health-medical/pharmaceuti- . 5 U.S. Food and Drug Administration. Guidance for Industry: cals/140729-01.html. Recommendations for Microbial Vectors Used for Gene Therapy. https:// . www.fda.gov/regulatory-information/search-fda-guidance-documents/ 22 European Medicines Agency. Accelerated assessment. https://www. recommendations-microbial-vectors-used-gene-therapy. September 2016. ema.europa.eu/en/human-regulatory/marketing-authorisation/ accelerated-assessment. 6. U.S. Food and Drug Administration. Guidance for Industry: Potency . Tests for Cellular and Gene Therapy Products. https://www.fda.gov/ 23 Nagai S. Flexible and Expedited Regulatory Review Processes for Innovative regulatory-information/search-fda-guidance-documents/potency-tests-cel- Medicines and Regenerative Medical Products in the U.S., the EU, and lular-and-gene-therapy-products. January 2011. Japan. Int J Mol Sci. 2019 Aug 3;20(15):3801. doi: 10.3390/ijms20153801. PMID: 31382625; PMCID: PMC6696404. 7. U.S. Food and Drug Administration. OTAT Learn. https://www.fda.gov/ . vaccines-blood-biologics/news-events-biologics/otat-learn. March 2018. 24 Nagai S. Flexible and Expedited Regulatory Review Processes for Innovative Medicines and Regenerative Medical Products in the U.S., the EU, and . 8 European Medicines Agency. Advanced therapy classification. https:// Japan. Int J Mol Sci. 2019 Aug 3;20(15):3801. doi: 10.3390/ijms20153801. www.ema.europa.eu/en/human-regulatory/marketing-authorisation/ PMID: 31382625; PMCID: PMC6696404. advanced-therapies/advanced-therapy-classification. 25. Nagai S. Flexible and Expedited Regulatory Review Processes for Innovative . 9 European Medicines Agency. Committee for Advanced Medicines and Regenerative Medical Products in the U.S., the EU, and Therapies (CAT). https://www.ema.europa.eu/en/committees/ Japan. Int J Mol Sci. 2019 Aug 3;20(15):3801. doi: 10.3390/ijms20153801. committee-advanced-therapies-cat. PMID: 31382625; PMCID: PMC6696404. . 10 European Medicines Agency.Marketing-authorisation procedures for 26. Ministry of Health, Labour and Welfare. Strategy of SAKIGAKE. advanced-therapy medicinal products. https://www.ema.europa.eu/ https://www.mhlw.go.jp/english/policy/health-medical/pharmaceuti- en/human-regulatory/marketing-authorisation/advanced-therapies/ cals/140729-01.html. marketing-authorisation-procedures-advanced-therapy-medicinal-products. 27. U.S. Food and Drug Administration. CBER Advanced Technologies Team . 11 Pharmaceuticals and Medical Devices Agency. PMDA presentation on (CATT). https://www.fda.gov/vaccines-blood-biologics/industry-biologics/ Regulation of Regenerative Medicine in Japan. https://www.pmda.go.jp/ cber-advanced-technologies-team-catt. files/000219466.pdf. May 2017. 28. European Medicines Agency. Innovation in medicines. https:// . 12 Pharmaceuticals and Medical Devices Agency. PMDA presentation on www.ema.europa.eu/en/human-regulatory/research-development/ Regulation of Regenerative Medicine in Japan. https://www.pmda.go.jp/ innovation-medicines. files/000219466.pdf. May 2017. 29. Paul-Ehrlich-Institut, Agency of the German Federal Ministry of Health. . 13 Detela G, Lodge A. EU Regulatory Pathways for ATMPs: Standard, Innovation Office. https://www.pei.de/EN/regulation/advice/innovation-of- Accelerated and Adaptive Pathways to Marketing Authorisation. fice/innovation-office-node.html. Mol Ther Methods Clin Dev. 2019 Jan 29;13:205-232. doi: 10.1016/j. . omtm.2019.01.010. PMID: 30815512; PMCID: PMC6378853. https:// 30 U.S. Food and Drug Administration. Guidance for Industry: Formal www.sciencedirect.com/science/article/pii/S2329050119300130#fig2. Meetings Between the FDA and Sponsors or Applicants of PDUFA Products. https://www.fda.gov/regulatory-information/search-fda-guid- . 14 U.S. Food and Drug Administration. Fast Track. ance-documents/formal-meetings-between-fda-and-sponsors-or-appli- https://www.fda.gov/patients/fast-track-break- cants-pdufa-products-guidance-industry. December 2017. through-therapy-accelerated-approval-priority-review/ . fast-track. 31 U.S. Food and Drug Administration. FDA presentation: Facilitating Expedited Development of Advanced Therapy Products. https://ipq.org/ . 15 U.S. Food and Drug Administration. Breakthrough wp-content/uploads/2019/09/Facilitating-Expedited-Development_Oh.pdf. Therapy. https://www.fda.gov/patients/fast-track-break- June 2019. through-therapy-accelerated-approval-priority-review/ . breakthrough-therapy. 32 U.S. Food and Drug Administration. CBER SOPP 8214: INTERACT Meetings with Sponsors for Drugs and Biological Products. https://www.fda. 16. U.S. Food and Drug Administration. Guidance for Industry: Expedited

CHAPTER 1 Regulatory Considerations 26 Endnotes, continued

gov/media/124044/download. June 2020. 49. European Medicines Agency. EMA/309801/2017: General Principles EMA- FDA Parallel Scientific Advice (Human Medicinal Products). https://www. ema.europa.eu/en/documents/other/general-principles-european-medi- 33. U.S. Food and Drug Administration. INTERACT Meetings. https://www. cines-agency-food-drug-administration-parallel-scientific-advice_en.pdf. fda.gov/vaccines-blood-biologics/industry-biologics/interact-meetings. April 2017. 34. U.S. Food and Drug Administration. Investigational New Drug 50. U.S. Food and Drug Administration. CDER SOPP 8001.6: Procedures for (IND) Application. https://www.fda.gov/drugs/types-applications/ Parallel Scientific Advice with European Medicines Agency (EMA). https:// investigational-new-drug-ind-application#preIND. www.fda.gov/media/87478/download. October 2019. 35. U.S. Food and Drug Administration. Guidance for Industry: 51. Pharmaceuticals and Medical Devices Agency. Consultations. https://www. Considerations for the Design of Early-Phase Clinical Trials of Cellular pmda.go.jp/english/review-services/consultations/0002.html. and Gene Therapy Products. https://www.fda.gov/regulatory-information/ search-fda-guidance-documents/considerations-design-early-phase-clini- 52. Pharmaceuticals and Medical Devices Agency. Profile of Services. https:// cal-trials-cellular-and-gene-therapy-products. June 2015. www.pmda.go.jp/files/000221139.pdf. 36. U.S. Food and Drug Administration.End of Phase 2 Meeting. https://www. 53. Pharmaceuticals and Medical Devices Agency. PMDA presentation: accessdata.fda.gov/cder/sb-navigate/topic3/topic3/da_01_03_0186.htm. Clinical trial notifications and scientific consultation system in Japan. https://www.pmda.go.jp/files/000152326.pdf. 37. U.S. Food and Drug Administration. FDA presentation: Meetings with CDER. https://www.fda.gov/media/89217/download. 54. U.S. Food and Drug Administration. Draft Guidance for Industry: Formal Meetings Between the FDA and Sponsors or Applicants of PDUFA 38. U.S. Food and Drug Administration. End of Phase 2 Meeting. https://www. Products. https://www.fda.gov/media/109951/download. December 2017. accessdata.fda.gov/cder/sb-navigate/topic3/topic3/da_01_03_0186.htm. 55. International Conference on Harmonisation of Technical Requirements for 39. U.S. Food and Drug Administration. Questions to Ask FDA during Formal Registration of Pharmaceuticals for Human Use. ICH M2 EWG Electronic Meetings. https://www.accessdata.fda.gov/cder/sb-navigate/topic3/topic3/ Common Technical Document Specification. https://admin.ich.org/sites/ da_01_03_0180.htm. default/files/inline-files/eCTD_Specification_v3_2_2_0.pdf. July 2008. 40. European Medicines Agency. Authorisation of medicines. 56. U.S. Food and Drug Administration. Guidance for Industry: M4Q: The https://www.ema.europa.eu/en/about-us/what-we-do/ CTD — Quality. https://www.fda.gov/media/71581/download. August authorisation-medicines#scope-of-the-centralised-procedure-section. 2001. 41. European Medicines Agency. EMEA/45110/2007: Committee for Medicinal 57. U.S. Food and Drug Administration. IND Applications for Clinical Product for Human Use Rules of Procedure. https://www.ema.europa.eu/ Investigations: Chemistry, Manufacturing, and Control (CMC) Information. en/documents/other/chmp-rules-procedure_en.pdf. March 2020. https://www.fda.gov/drugs/investigational-new-drug-ind-application/ 42. European Medicines Agency. EMA/CAT/600280/2010: Committee for ind-applications-clinical-investigations-chemistry-manufacturing-and-con- Advanced Therapies (CAT) Reflection paper on classification of advanced trol-cmc-information. therapy medicinal products. https://www.ema.europa.eu/en/documents/ 58. U.S. Food and Drug Administration. Guidance for Industry: Determining scientific-guideline/reflection-paper-classification-advanced-therapy-medic- the Need for and Content of Environmental Assessments for Gene inal-products_en-0.pdf. May 2015. Therapies, Vectored Vaccines, and Related Recombinant Viral or Microbial 43. European Medicines Agency. EMA/CAT/99623/2009: Committee for Products. https://www.fda.gov/regulatory-information/search-fda-guid- Advanced Therapies (CAT) Procedural advice on the provision of scientific ance-documents/determining-need-and-content-environmental-assess- recommendation on classification of advanced therapy medicinal products ments-gene-therapies-vectored-vaccines-and-related. in accordance with article 17 of regulation (EC) no 1394/2007. https:// 59. U.S. Food and Drug Administration. Guidance for Industry: Chemistry, www.ema.europa.eu/en/documents/regulatory-procedural-guideline/ Manufacturing, and Control (CMC) Information for Human Gene Therapy procedural-advice-provision-scientific-recommendation-classification-ad- Investigational New Drug Applications (INDs). https://www.fda.gov/ vanced-therapy-medicinal/2007_en.pdf. London, December 2013. media/113760/download. January 2020. 44. Voisin Life Sciences Consulting. Unlocking The EMA 60. European Medicines Agency. EMA/CAT/852602/2018: Guideline on Meetings: The Process And How To Prepare Adequately. quality, non-clinical and clinical requirements for investigational advanced https://voisinconsulting.com/resources/webinars/ therapy medicinal products in clinical trials. https://www.ema.europa.eu/ unlocking-the-ema-meetings-the-process-and-how-to-prepare-adequately/. en/documents/scientific-guideline/draft-guideline-quality-non-clinical-clini- 45. European Medicines Agency. EMA/4260/2001: European Medicines cal-requirements-investigational-advanced-therapy_en.pdf. January 2019. Agency guidance for applicants seeking scientific advice and protocol 61. European Medicines Agency. EMEA CPMP/ICH/294/95: ICH Topic Q 5 assistance. https://www.ema.europa.eu/en/documents/regulatory-procedur- D Quality of Biotechnological Products: Derivation and Characterisation al-guideline/european-medicines-agency-guidance-applicants-seeking-sci- of Cell Substrates Used for Production of Biotechnological/Biological entific-advice-protocol-assistance_en.pdf. October 2020. Products. https://www.ema.europa.eu/en/documents/scientific-guideline/ 46. European Medicines Agency. National competent authorities (human). ich-q-5-d-derivation-characterisation-cell-substrates-used-production-bio- https://www.ema.europa.eu/en/partners-networks/eu-partners/ technological/biological-products-step-5_en.pdf. March 1998. eu-member-states/national-competent-authorities-human. 62. U.S. Food and Drug Administration. Federal Register / Vol. 63, No. 182. 47. Voison Life Sciences Consulting. Opportunities For Early Dialogue With https://www.fda.gov/media/71463/download. September 1998. The EMA And HTA Bodies. Optimizing Product Development To Ensure 63. International Council for Harmonisation of Technical Requirements for Commercial Success. https://voisinconsulting.com/resources/webinars/ Pharmaceuticals for Human Use. ICH Quality Guidelines. https://www.ich. opportunities-for-early-dialogue-with-the-ema-and-hta-bodies-optimiz- org/page/quality-guidelines. ing-product-development-to-ensure-commercial-success/. 48. European Medicines Agency. Parallel consultation with regulators and health technology assessment bodies. https://www.ema.europa.eu/en/hu- man-regulatory/research-development/scientific-advice-protocol-assistance/ parallel-consultation-regulators-health-technology-assessment-bodies.

CHAPTER 1 Regulatory Considerations 27 Chapter 2 Standards in Gene Therapy Chapter 2 | Standards in Gene Therapy

Introduction...... 30 Benefits of Voluntary Consensus Standards in Gene Therapy...... 30 Documentary Standards...... 30 What are documentary standards...... 30 How are documentary standards, regulations, guidance, and best practices different?...... 31 How documentary standards can be used in gene therapy...... 31 How documentary standards are developed...... 31 The standards development process...... 31 Documentary standards case studies...... 34 Reference Material Standards ...... 34 What are reference material standards?...... 34 Reference material case studies...... 35 Current Landscape of Standards in Gene Therapy...... 36 Current SDOs involved with standards in gene therapy...... 36 SCB...... 39 Current documentary standards in gene therapy...... 39 Documentary standards intended for preclinical setting...... 40 Documentary standards intended for clinical settings...... 40 Five gene therapy standards currently in development...... 40 Developing supportive standards in gene therapy...... 41 Non-documentary standards in gene therapy...... 41 Needs and gaps for standards in gene therapy by functional area...... 42 Implementing Standards in Gene Therapy Manufacturing...... 42 How to implement standards based on intended use...... 42 Gene therapy bioprocessing and production standards for manufacturing ...... 43 Standards in Gene Therapy and Regulatory Approval...... 44 FDA’s informal recognition of voluntary consensus standards...... 44 FDA’s formal recognition of standards...... 44 Coordination, Community Engagement, and Education: The Standards Coordinating Body...... 45

CHAPTER 2: Standards in Gene Therapy 29 Introduction transition from the clinical development phase to being commercially available for the benefit of patients. The ap- Gene therapy is an emerging ﬁeld of medicine, with propriately targeted, field-wide development of standards many promising products in development that could help will help to address the previously tolerated shortcomings. manage and potentially cure conditions and diseases that Specifically, standardizing equipment, methodol- are intractable, chronic, and even terminal. Given that the ogies, and testing methods will result in a number of gene therapy ﬁeld is currently at a key tipping point, with benefits for industry stakeholders. Standards establish disruptive innovation pushing the boundaries of science a base of legitimacy on which patients, regulators, and and a number of products poised for commercialization investors can rely—standards development will instill the (with a few already on the market), robust standards that public with confidence that development of gene therapy can support developers in ensuring the safety, efficacy, and is adequately informed by thoroughly researched best quality of products must be established. This chapter will practices and is, therefore, a safe and effective option for provide an overview of why standardization in the field of treating a wide range of diseases. In addition, from an op- gene therapy is needed, as well as an examination of the cur- erational standpoint, standardization can help industry rent landscape of standards development in gene therapy. stakeholders streamline business practices by allowing for more efficient coordination of efforts throughout the Benefits of Voluntary Consensus entire supply chain, as well as improving the predictabil- ity of costs and resource management. In effect, barriers Standards in Gene Therapy to entry into the clinical development space may be low- Standards are considered to be voluntary rules, condi- ered for smaller companies or academic researchers, thus tions, characteristics, or physical materials that an orga- facilitating the delivery of therapies to patients. Standards nization can adopt to make a process safer, more efficient, can assure regulators that the fundamental processes or better aligned with the practices of other organizations underlying the development of a new therapy are sound. in their industry. In general, standards can be considered This assurance, in turn, allows regulators to more rapidly to be either documentary or non-documentary. review a product. A smoother, less uncertain regulatory Gene therapy is one of the key modalities of newly review process increases industry stability, lowers per- emerging regenerative therapeutics. The number and di- ceived risk to investors, and accelerates market availabil- versity of the product platforms in current use has explod- ity of products, all of which serve to increase access and ed over the last ten years. However, given that relatively options for patients. Lastly, standards can greatly aid the little experience has accumulated in the years that gene regenerative medicine therapy community as a whole by therapy has been a part of the therapeutic landscape, there enhancing the ability of developers to collaborate and are relatively few ways for developers across the research share knowledge with others. This can reduce the po- and industrial sectors to achieve consistency in areas such tential for redundant efforts to be undertaken and serve as protocol development, process infrastructure, and prod- to patch up the relatively fragmented state of knowledge uct quality testing and assurance. This lack of consistency that characterizes the emerging state of this field. and experience has made it challenging for stakeholders involved in gene therapy development and manufacturing Documentary Standards to operate with a sense of certainty and provide patients in need of these essential and groundbreaking therapies with WHAT ARE DOCUMENTARY STANDARDS? the confidence that products are of the maximum possible Documentary standards set guidelines, protocols, quality. The ongoing absence of standards to support safe procedures, methods, technical speciﬁcations, and ter- and efficient practices that reduce the burden on compa- minology in the form of consensus-based documents nies seeking regulatory approval for their products may that developers can use to ensure a high level of product result in promising therapies being unable to successfully quality and safety. These standards can apply to any step

CHAPTER 2: Standards in Gene Therapy 30 • Increasing product safety by deﬁning testing and process- of development, from the evaluation of source materials, ing parameters throughout the product life cycle the manufacturing of products, and assessment of safety • Speeding review processes by eliminating the need to and quality attributes of final products, to the transport, re-evaluate common operational steps for each new storage, and commercial clinical use of products. product Documentary standards are developed by standards • Decreasing costs of therapies by increasing testing and developing organizations (SDOs), accreditation bodies, process efficiency and professional societies, including (but not limited to): • Decreasing time to market by minimizing time required for implementation of common operational • ASTM International steps and validation of unique manufacturing processes • The British Standards Institution (BSI) • Increasing quality of raw materials and ﬁnal products • The Clinical and Laboratory Standards Institute by standardizing reporting requirements and quality (CLSI) assays. • The Foundation for the Accreditation of Cellular Therapy (FACT) HOW DOCUMENTARY STANDARDS ARE DEVELOPED • The International Organization for Standardization Consensus-based standards are standards developed (ISO) following a consensus-based process, which means an • The United States Pharmacopeia (USP) organization uses practices that are fair, open, balanced, equitable, accessible, and responsive to stakeholder HOW ARE DOCUMENTARY STANDARDS, needs. The American National Standards Institute REGULATIONS, GUIDANCE, AND BEST PRACTICES (ANSI) accredits U.S. standards developing organizations DIFFERENT? (SDOs) that follow consensus-based processes. Documentary standards can be contrasted with regu- Non-consensus-based standards are standards created lations, guidance, and best practices. Regulations have by organizations that do not follow consensus-based the force and effect of law and are usually mandatory, processes. With respect to biopharmaceutical products, setting out speciﬁc requirements, which products and some of the most important of these standards are pub- organizations must meet. In the United States, regula- lished in compendiums known as pharmacopoeia. tions are written in the Code of Federal Regulations and published in the Federal Register. Guidances are formal THE STANDARDS DEVELOPMENT PROCESS documents issued by a government agency to clarify the The documentary standards development process: regulatory body’s thinking on existing laws or regula- • Brings together experts throughout the community tions and offer guidelines for how industry can comply to share pre-competitive knowledge with these regulations. Best practices are the informal • Makes research results more readily available to the methods most people in a ﬁeld agree are the best way of public to drive the entire ﬁeld forward accomplishing a goal. These are sometimes published by • Gives stakeholders a voice in deﬁning the standards professional societies in academic journals. Guidances that will best support their work and best practices, for example, may include expectations that may be considered to be documentary standards. Figure 1 illustrates the development of documenta- ry standards from the perspective of one particular HOW DOCUMENTARY STANDARDS CAN BE USED non-profit organization, the Standards Coordinating IN GENE THERAPY Body for Gene, Cell, and Regenerative Medicines and Documentary standards development can benefit the Cell-Based Drug Discovery (SCB), which complements gene therapy community by: current SDO processes for standards development by en- gaging regenerative medicine stakeholders to ensure that • Replacing costly time-intensive trial-and-error new or revised standards provide the greatest benefits to processes with proven best practices the broad regenerative medicine community.

CHAPTER 2: Standards in Gene Therapy 31 Figure 2-1. Standards Coordinating Body—Process for Development of Documentary Standards Documentary Standard Development Process with SCB Support

STEP STEP Stakeholders discuss or identify an area for standards Working groups solicit input from internal and external 4 experts on the feasibility, scope, content, and definitions 1 development and determine if the standard meets the Final draft criteria for SDO development: Consensus in the standard and develop a final draft; as needed, Identify of the • Do existing standards address this need? on needed Obtain they also conduct or review results from data validation standards standard for • Is the proposed standard market-relevant? standard and additional and round robin studies need and review and • Are stakeholders committed to advancing the standard? identification of stakeholder assess priority publication a relevant SDO, inputs and and feasibility Opportunities for Community Engagement Supported by SCB: professional conduct Opportunities for Community Engagement Supported by SCB: or scientific scientific Work with SDO technical committees to coordinate standard Define the standards landscape: organization, studies reviews, revisions, and scientific studies • Participate in surveys and interviews to identify needed certification/ standards accreditation body, • Attend facilitated sessions at meetings and conferences to pharmacopoeia, STEP discuss both existing and in-development standards or government SDOs publish the final standard for sale online or in Published agency 5 print; some offer parts of the standard for free (e.g., Prioritize needed standards: introductions, references, definitions) standard • Participate in community surveys to prioritize needed standards Publish • Attend facilitated meetings with sector-specific working groups to finalized prioritize needed standards by their urgency and potential impact standard SCB Outputs: Conduct feasibility assessment: Educational materials (e.g., case studies, blog posts, fact sheets, • Join project working groups to evaluate the feasibility of web content, newsletters, webinars) to encourage the standard’s developing and adopting high-priority standards adoption and inform stakeholders on how to implement it SCB Outputs Implementation assessment of the new standard to understand the extent of its adoption, synthesize lessons learned during its • Needed Standards in Regenerative Medicine report implementation, and communicate this information to the broad • Feasibility reports regenerative medicine community

STEP STEP The community introduces the concept to an SDO, Concept SDOs revise or withdraw standards if they are no longer 2 accreditation body, or professional organization/society for introduced 6 relevant, become outdated, or were under revision and Revisions to consideration (which varies in terms of formality required) Introduce to an SDO, consensus was not reached on the changes; withdrawn published professional Revise standards remain available for reference purposes standard, or concept standard, to SDO Opportunities for Community Engagement Supported by SCB: or scientific withdrawn organization, as needed standard Work with SDO technical committees to develop plans for standard certification/ SCB Outputs: advancement (e.g., schedules, resources, critical milestones) accreditation Updated Regenerative Medicine Landscape Report with SCB Output body, information on active and withdrawn standards pharmacopoeia, • Quarterly newsletter communicating substantive updates such as Updated feasibility reports or government new, revised, or withdrawn standards • Standard advancement plans agency

STEP SDO technical committees review the idea and determine 3 if the standard should advance, developing an initial Initial working standard for review if warranted draft of the Review new standard concept, initiate Opportunities for Community Engagement Supported by SCB: discussion, and Work with SDO technical committees to coordinate standard draft initial drafting, including by assembling experts and gathering additional standard information and data

CHAPTER 2: Standards in Gene Therapy 32 Figure 2-1. Standards Coordinating Body—Process for Development of Documentary Standards continued from previous page

Documentary Standard Development Process with SCB Support

STEP SDO technical committees review the idea and determine 3 if the standard should advance, developing an initial Initial working standard for review if warranted draft of the Review new standard concept, initiate Opportunities for Community Engagement Supported by SCB: discussion, and Work with SDO technical committees to coordinate standard source: Standards Coordinating Body; Documentary Standards for Regenerative Medicine draft initial drafting, including by assembling experts and gathering additional standard information and data

CHAPTER 2: Standards in Gene Therapy 33 DOCUMENTARY STANDARDS CASE STUDIES quality controls and analysis, and method validation. The standard can be purchased here: https://www.iso. Pre-existing AAV Immunity org/standard/67893.html Patients who have pre-existing immunity to the vectors used in gene therapy products such as AAVs may ex- Reference Material Standards perience suboptimal treatment outcomes. However, the gene therapy community currently lacks standards for WHAT ARE REFERENCE MATERIAL STANDARDS? evaluating pre-existing AAV immunity. Reference materials are considered to be non-documen- SCB is coordinating the development of a consen- tary standards and are typically highly characterized sus-based documentary standard that defines common reagents or substances that are distributed to enable language for evaluating pre-existing immunity to AAVs. researchers and developers to assure consistency and Defining basic language and concepts in the field of quality in measurement processes intended to report pre-existing immunity is needed to inform the develop- on the safety and potency of a manufactured product. ment of standards in this area and facilitate pre-compet- These materials can be linked to speciﬁc manufacturing itive dialogue among stakeholders. This documentary processes and can be used for product development standard is anticipated to be available sometime between purposes such as the calibration of analytical equipment 2023 and 2025. or the comparison of therapeutic products or product components to similar reference materials with known Nucleic Acid Target Sequence Quantification quality attributes. Reliable reference materials are of Accurate quantification of nucleic acid target sequences particular importance to the gene therapy community allows developers and manufacturers of regenerative since many current testing methods and equipment are therapy products to monitor factors such as process not yet standardized. controls, efficiency, and quality attributes. However, Reference materials development beneﬁts the regen- performing the required assays at a high enough degree erative medicine community as a whole by: of quality so as to effectively demonstrate product safety and/or efficacy can be challenging. ISO has developed the • Enabling analytical equipment calibration to ensure ISO DNA Quantitation Standard (ISO 20395) that can valid results help manufacturers achieve high quality measurements • Deﬁning safe, reliable baseline materials for the quantification of nucleic acid sequences. • Speeding review processes for researchers using The document provides requirements and guidelines non-standard testing equipment and test methods for ensuring the quality of methods used for the quan- • Decreasing time and money lost to repetitive, incon- tification of specific nucleic acid sequences. The main sistent tests methods covered are based on digital (dPCR) and quan- • Improving coordination, community engagement, titative real-time PCR (qPCR) amplification technolo- and education gies, which can be applied to target sequences present • Increasing consistency through standardization of in nucleic acid molecules including double-stranded testing results and reporting DNA (dsDNA) such as genomic DNA (gDNA) and plasmid DNA, single stranded DNA (ssDNA), com- In 2015, FDA (Gavin, HGTM 26:3) and NIBSC/Medicines plementary DNA (cDNA), and single stranded RNA and Healthcare Products Regulatory Agency (Werling et (ssRNA) including ribosomal RNA (rRNA), messenger al. HGTM 26:82) encouraged the use of reference standard RNA (mRNA), and long and short non-coding RNA materials as benchmarking tools for qualifying in-house [microRNAs (miRNAs) and short interfering RNAs reference materials and controls, and for demonstrating (siRNAs)], as well as double-stranded RNA (dsRNA). that assay methods are appropriately controlled. Specific topics covered include (but are not limited to) For additional information and definitions, see: analytical design, assay design and optimization, data https://www.nist.gov/srm/srm-definitions

CHAPTER 2: Standards in Gene Therapy 34 REFERENCE MATERIAL CASE STUDIES FOR GENE THERAPY — CASE STUDY #1 FOR MORE INFORMATION Gene therapy involves modifying the expression of a Further information about AAV2 and AAV8 can be patient’s genes and/or repairing abnormal genes. Since found in the following references (provided links genes within cells cannot be changed without a way to are from the ATCC website): administer the genetic changes to the cell itself, scientists AAV2 produce delivery tools known as vectors that are capable Production: of administering speciﬁc nucleic acids (DNA or RNA) Flotte, TR, Burd, P and Snyder, RO. Utility of a Re- into the cell for expression and replication. Vectors are combinant Adeno-Associated Viral Vector Refer- frequently of viral origin and insertion of the desired ence Standard. BioProcessing J. 2002;1: 75. nucleic acid into cells can occur either ex vivo or in vivo. Potter, M., Phillipsberg, G., Phillipsberg, T., et al. Following the establishment of the Adenovirus Type 5 Manufacture and stability study of the recombi- reference material (ATCC VR-1516) close to 20 years ago, nant adeno-associated virus serotype 2 vector discussions were held to facilitate the preparation of ref- reference standard. Bioprocess J. 2008;7: 8-14 erence standard materials for AAV vectors. A major goal was to be able to use the resulting reference standards to Characterization: validate each laboratory’s internal standards and methods. Lock M, McGorray S, Auricchio A, et al. Charac- terization of a recombinant adeno-associated This would enable comparisons among studies conducted virus type 2 Reference Standard Material. Hum by different laboratories and aid in the manufacturing of Gene Ther. 2010;21(10):1273-1285. doi:10.1089/ higher quality vectors. hum.2009.223 Since it was the primary serotype in use, the initial focus was on AAV2. The AAV Reference Standards AAV8 Working Group (AAVRSWG) was formed to include https://pubmed.ncbi.nlm.nih.gov/18574495/ members from industry, academia, government, and 2016 Paper: regulatory bodies. Early discussions finalized the profile https://www.ncbi.nlm.nih.gov/pmc/articles/ of the material with regard to concentration and testing PMC4813604/ requirements. The reference material was produced, pu- rified, formulated, and filled by a group of organizations that donated substantial time and materials to the effort. A few years after the AAV2 effort, another working using similar methods. Even when the protocols and group was formed to facilitate the production of a ref- associated reagents were provided, there were sometimes erence standard for AAV8. Information regarding the large variances observed. For the AAV2 material, the production and characterization, as well as ordering mean titer was 3.28 x 1010 vector genomes/ml and the information, can be found here: 95% confidence intervals were 2.70 x 1010 to 4.75 x 1010 https://www.atcc.org/en/Standards/Standards_Programs/ vector genomes/ml. For a dose-determining assay, such as ATCC_Virus_Reference_Materials.aspx the assessment of genome titer, such variances can cause These efforts were successful in highlighting the lack difficulties for developers and regulatory bodies with of standardization within the gene therapy community. regard to being able to reliably cross-compare data across To this day, organizations developing gene therapy prod- the field. In 2018, the FDA hosted a workshop entitled: ucts continue to work in relative isolation with respect to “Quantitation of AAV-based gene therapy products.” In solving issues with variability in the analytical methods this workshop, presentations covered both the biology of used to characterize AAV and other gene therapy vectors. AAV, as well as different analytical approaches for quan- The laboratory studies conducted using the original AAV tifying AAV-based products. Information and recordings reference standard materials resulted in a large data set of the workshop can be found here: https://www.fda. that showed a high degree of inter-laboratory variability gov/vaccines-blood-biologics/workshops-meetings-con-

CHAPTER 2: Standards in Gene Therapy 35 ferences-biologics/quantitation-aav-based-gene-thera- Table 2-1. Existing Standards by SDO py-products-12072018-12072018 The existing AAV reference materials provide useful 6 AABB tools for developers to use in benchmarking equipment, 58 ASTM International methods, and analyst performance. Although the PCR 8 American Type Culture Collection (ATCC) targets used for these materials are different from those used in product-specific methods, they are still extremely 3 British Standards Institutions (BSI) valuable as tools for interrogating the variability of meth- 7 Clinical & Laboratory Standards Institutes (CLSI) ods. The results obtained from the studies associated with 9 European Directorate for the Quality of these efforts were also extremely valuable in highlighting Medicines and Healthcare (EDQM) the variability of data obtained in different laboratories, 8 Foundation for the Accreditation of Cellular Therapy (FACT) which has led to investigations into potential sources of 6 International Conference on Harmonization of variability. In 2016, another study was published on the Technical Requirements for Pharmaceuticals for use of the AAV reference materials, titled: “Practical uti- Human Use (ICH) lization of recombinant AAV vector reference standards: 1 International Council for Commonality in Blood Focus on vector genomes titration by free ITR qPCR.” Banking Automation (ICCBBA) The authors used the reference materials to investigate 55 International Organization for Standardization differences in titers obtained using slightly different (ISO) methods and concluded that free inverted terminal 1 International Society for Advancement of repeat (ITR) qPCR eliminated the differences. Cytometry (ISAC) 3 International Society for Cellular Therapy (ISCT) 1 International Society for Stem Cell Research Current Landscape of Standards (ISSCR) in Gene Therapy 22 National Institute for Biological standards and Currently, additional research is needed to assess the Control (NIBSC) safety and efficacy of gene therapies for commercial 1 ONE Study Consortium use, which is becoming more widespread with the 6 Parenteral Drug Association (PDA) regulatory approval of several products in recent years. 6 Pharmaceuticals and Medical Devices Agency Furthermore, factors such as variations in manufac- (PMDA), Japan turing, measurement, and analytical techniques across 14 United States Pharmacopeia (USP) developers of experimental gene therapy products make it difficult to evaluate product quality and safety, and to assess the impact of manufacturing changes or innova- are discussed in greater detail below. tions intended to improve product safety and efficacy. A common set of standards in gene therapy will advance ASTM International development of treatments beyond the realm of clinical ASTM International (formerly known as American trials to approved treatments for a number of diseases Society for Testing and Materials) is a global developer and syndromes subject to genetic control. of voluntary standards using an open and transpar- ent process. ASTM is a broad SDO with more than CURRENT SDO’S INVOLVED WITH STANDARDS 12,000 standards globally and follows the World Trade IN GENE THERAPY Organization’s principles for standards development Table 1 describes the number of important gene thera- of transparency, openness, impartiality and consensus, py-related standards that have been developed by sev- effectiveness and relevance, coherence, and development eral established SDOs. Some of the more active SDOs dimension. ASTM develops standards for regenera- involved currently with gene therapy-related standards tive medicine primarily through Committee F04. In

CHAPTER 2: Standards in Gene Therapy 36 particular, Subcommittees F04.41 (Classification and as the ANSI-appointed Secretariat for ISO Technical Terminology for TEMPs), F04.42 (Biomaterials and Committee 212 (ISO/TC 212), CLSI provides ISO/TC Biomolecules for TEMPs), F04.43 (Cells and Tissue 212 with both draft and approved standards related to Engineered Constructs for TEMPs), F04.44 (Assessment laboratory testing and in vitro diagnostic test systems, for TEMPs), and F04.45 (Adventitious Agents Safety) are with specific standards topics including quality man- most relevant to production of regenerative medicines. agement, pre- and post-analytical procedures, analytical performance, laboratory safety, reference systems, and ATCC quality assurance. CLSI also serves as the administrator ATCC is an ISO accredited non-profit organization that for the ANSI-accredited US Technical Advisory Group seeks to support the advancement and application of (TAG) for ISO/TC 212. With regard to gene therapy, scientific knowledge by facilitating the development of many of the nucleic acid quantification assays common- standards related to biological materials and informa- ly used during gene therapy product manufacturing tion. With respect to supporting regenerative therapies, are implemented using standards established by CLSI. ATCC is responsible for distributing standard doses Developers may also reference a number of standards of recombinant adeno-associated virus 2 (ATCC VR- covering a wide range of molecular-level testing used 1616) and 8 (ATCC VR-1816) vector reference standard in clinical development settings. stock (RSS). Adeno-associated virus (AAV) -based recombinant vectors are important tools used for the EDQM manufacturing clinical gene therapy (in addition to cell The EDQM supports patient access to quality medi- therapy). ATCC-distributed RSSs establish reference cines and healthcare in the EU. Through its European points that define standard particle, vector genome, Pharmacopoeia (Ph. Eur.) arm, the EDQM determines and infectious unit specifications for AAVs. In line with standards for the quality of medicines, including those recommendations issued by FDA’s CBER, OCTGT, and for gene therapy products, and codifies these standards DCGT, the global availability of AAV reference standard into pharmacopoeial texts. The EDQM engages with the materials aids in standardization across organizations field of gene therapy as part of its role as the technical involved in gene therapy manufacturing and distribu- secretariat for the network of Official Medicines Control tion by allowing for the accurate calibration on internal Laboratories (OMCLs) of Europe. The OMCLs are a reference materials and assays for products that make network of more than 70 laboratories spanning both use of AAV viral gene transfer-based technologies and the public and industry healthcare sectors that support techniques. In turn, data from different preclinical and European regulatory authorities in ensuring the quality clinical studies can be more easily and meaningfully of gene therapy products by assessing the landscape compared. of marketing licensure applications with respect to preclinical data and data gathered from monitoring mar- CLSI keted medicines, as well as current legal requirements. Clinical and Laboratory Standards Institute (CLSI) is The interaction of the OMCLs with regulators allows an international SDO comprised of more than 1,400 manufacturers of gene therapy products to take into member organizations that seeks to promote the devel- account all relevant regulatory considerations starting opment and use of voluntary consensus standards and from the early stages of product development. In order guidelines within the health care community at large. to organize and direct the OMCLs for their role in the In order to develop standards, CLSI uses a unique con- surveillance of gene therapy products, a gene therapy sensus-based process that includes authorization of a Working Group has been established within the General project, development and open review of documents, European OMCL Network. The gene therapy Working revision of documents in response to users’ comments, Group aims specifically to define the most appropriate and finally, acceptance of a document as a consensus analytical methods for gene therapy products and rel- standard or guideline. Importantly, through its role evant reference standards. The gene therapy Working

CHAPTER 2: Standards in Gene Therapy 37 Group meets once a year to review field-wide activities ISO in the past year, define the goals for the following year, The International Organization for Standardization and discuss any relevant topics as needed. The areas of (ISO) is a worldwide federation of national standards greatest focus include adeno-associated viral (AAV) vec- bodies (ISO member bodies). ISO develops standards tors, retroviral/lentiviral (RV/LV) vectors, and plasmids. for regenerative therapy production via ISO Technical Committee 276 Biotechnology. ISO Technical Committee FACT 276 Biotechnology operates through four main Working FACT (Foundation for the Accreditation of Cellular Groups corresponding to the following areas: Therapy) at the University of Nebraska Medical Center is a non-profit organization that establishes standards • Terminology for medical and laboratory practice for developers of • Biobanks and bioresources cellular therapies. FACT aims to promote the adoption of • Analytical methods voluntary inspection and accreditation within the field of • Bioprocessing cellular therapy, which is an emerging and evolving field intimately tied to gene therapy (particularly with respect Additional topics covered in due measure by the to the use of viral vector products in the manufacturing Working Groups include data processing, validation, and process). The standards developed by FACT are intended comparability. In addition, ISO Technical Committee 276 to represent the fundamentals of cellular therapy that can Biotechnology works closely with both governmental and be applied to a broad range of cell sources and therapeu- non-governmental agencies such as International Society tic applications, and are to be used throughout product for Cell and Gene Therapy (ISCT) to identify priorities development and clinical trials. with regard to standardization efforts worldwide. Such The overarching goal of the FACT Standards is to collaboration not only streamlines standardization into promote improvement and progress in cellular therapy a comprehensively coordinated process, but also serves and regenerative medicine across all aspects of man- to allow organizations to avoid duplications and overlap- ufacturing and administration that are relevant to the ping standardization activities. quality of products and therapeutic care. To this end, FACT Standards are evidence-based (where possible, USP if published data are not available, accepted scientific The United States Pharmacopeia (USP) was created near- theory is referenced) and decided upon within commit- ly 200 years ago and is dedicated to instilling trust where tees comprised of world-renowned clinicians, scientists, it matters most: in the medicines, supplements, and foods technologists, and quality experts that span the entire people rely on for their health. The quality standards continuum of cell manufacturing. Additionally, input developed by USP help manufacturers deliver on their from both consumers and regulatory bodies are sought promises of safe products, while building confidence wherever possible. Before final approval by the FACT among healthcare practitioners, patients, and consumers. Board of Directors, the standards development process In the field of regenerative medicine, USP currently has undergoes both public review and legal review. multiple documentary and physical standards to aid As the only set of standards that are clearly and defin- developers in bringing safe and effective novel therapies itively focused on promoting the use of cellular therapy to patients. Under the guidance of the scientific experts products manufactured under rigorous controls, the on the recently established Advanced Therapies Expert FACT Standards form the basis of the FACT accreditation Committee, USP is active in developing standards for program. By gaining FACT accreditation, developers and AAV and lentiviral therapies, as well as materials such manufacturers of cell therapy products can demonstrate as plasmid DNA used in manufacturing. a commitment to maximizing the quality of products and therapeutic care, thereby instilling confidence in patients.

CHAPTER 2: Standards in Gene Therapy 38 STANDARDS COORDINATING BODY variables related to delivery mechanisms, and dosing. FOR REGENERATIVE MEDICINE, CELL Delivery of appropriate doses of vectors to patients is a AND GENE THERAPY (SCB) foundational challenge that could be addressed through SCB is an organization that was created in 2017 to help vector quantiﬁcation standards. Additionally, factors that accelerate the standards process. SDOs play a critical role cause adverse reactions in gene therapy patients (e.g., im- in the publishing of consensus standards that are univer- mune response or complications from replication-com- sally recognized. It was recognized that an organization petent retroviruses [RCRs]) must be better understood. was needed to help focus on the standards development process in order to accelerate this process by facilitating Specific gene therapy areas to standardize include: the use and development of standards in response to • Vector genome titration demonstrated needs expressed by a range of stakeholders. • Protocols for patient monitoring after infusion SCB functions as a 501(3)c organization with no vested • Factors for selection of biodistribution methods interest in a particular scientiﬁc, commercial, clinical, • Methodology for screening patients for immunity or policy approach. This is critical to its function and to adeno-associated virus (AAV) vectors success in addressing the complex challenges related to scientiﬁc protocols, product testing, and product quality Most current standards that are applicable to gene and performance speciﬁcations. The ﬁeld of regenerative therapy have been compiled in SCB’s The Regenerative medicine faces challenges common to emerging indus- Medicine Standards Landscape: tries, including fragmentation of knowledge, insufficient https://www.standardscoordinatingbody.org/landscape communication and coordination, and rapid advance- Of note, a number of documentary standards related ment of innovation. to quantification and sequencing have been published For the ﬁeld to thrive, it is recognized that develop- which may be referenced by developers of gene therapy ment and implementation of standards and best practices products. The main areas of focus relate to DNA diag- will help accelerate regulatory reviews of therapeutic nostic sequencing and molecular diagnostic testing, developers’ CMC documentation. SCB’s processes for DNA extraction methodology, general best practices for development of standards and best practices address the manufacturing, testing, and administration of gene ther- diverse needs of stakeholder groups, including govern- apy products, reference materials (e.g., genomic DNA, ment and regulatory agencies, researchers, providers of vector plasmids, cell lines, reference panels and reagents) raw materials, product developers, equipment manufac- virology standards, and testing for acceptable levels of turers, and clinicians and healthcare professionals. Since residual host-cell proteins in gene therapy products. 2017 SCB has been involved in 23 standards projects with Some current standards of note include: four advanced to SDOs, greatly shortening the historical times it has taken (up to 10 years in the past) for stan- International scope dards to progress. • ISO 20395:2019 Biotechnology — Requirements for evaluating the performance of quantification CURRENT DOCUMENTARY STANDARDS methods for nucleic acid target sequences — qPCR IN GENE THERAPY and dPCR Gene therapy involves modifying the expression of a • ISO 20688-1:2020 Nucleic acid synthesis — Part 1: patient’s genes and/or repairing abnormal genes using Requirements for the production and quality control recombinant DNA technology. Gene therapy products of synthesized oligonucleotides are delivered using viral or nonviral vectors that admin- • ISO 20391-1:2018 Biotechnology — Cell counting — ister speciﬁc nucleic acids (DNA or RNA) into the cell for Part 1: General guidance on cell counting methods expression and replication. This sector’s key challenge is • ISO 20391-2:2019 Biotechnology — Cell counting — to better understand and control how products interact Part 2: Experimental design and statistical analysis to with the human body. This can be achieved by deﬁning quantify counting method performance

CHAPTER 2: Standards in Gene Therapy 39 Europe (Seventh Edition). These documentary standards are • EP 5.14 EDQM: Gene Transfer Medicinal Products regularly updated with new editions. for human use • EP 2.6.35 EDQM: Quantification and FIVE IMPORTANT GENE THERAPY STANDARDS Characterization of residual host-cell DNA CURRENTLY IN DEVELOPMENT • EP 2.6.34 EDQM: Host-cell protein assays 1. NIST Gene Editing Consortium United States A prominent standards development effort is being led • USP Ancillary Materials for Cell, Gene and by NIST through the NIST Gene Editing Consortium Tissue-Engineered Products (https://www.nist.gov/programs-projects/nist-ge- • USP Cell and Gene Therapy Products nome-editing-consortium). The goal of this effort is to • USP Gene Therapy Products advance the rapidly developing field of genome editing by involving experts in an ongoing consortium to devel- Documentary standards intended for op standards for terminology, specificity measurements, preclinical settings and data and metadata. Though much of the current Pre-clinical studies are intended to test a drug, proce- effort is only indirectly related to the gene therapy field, dure, or other medical treatment in animals, and are one directly related project is the “Feasibility study for required to take place before clinical trials in humans standards for pre-existing immunity to AAV vectors.” can be started. Because the preclinical phase of research is critical for optimal decision-making about a possible 2. NIBSC/WHO Lentiviral Vector Copy Number future therapy, experiments done at this stage should Standard be conducted using best practice-based methods (e.g., Gene therapy is a rapidly evolving field. A prerequisite choosing the most appropriate animal model and ensur- for producing gene therapy products is ensuring their ing that experiments are comparable and reproducible quality and safety. This requires appropriately controlled across different labs). and standardized production and testing procedures that Currently, there are no known gene therapy standards result in consistent safety and efficacy. Assuring the qual- related to preclinical studies. ity and safety of lentiviral-based gene therapy products in particular presents a substantial challenge because Documentary standards intended for clinical settings they are cell-based multi-gene products that include Clinical trials are research studies intended to determine viral and therapeutic proteins as well as modified cells. whether a treatment or device is safe and efficacious for In addition to the continuous refinement of a product, human use. These studies must follow strict scientiﬁc changes in production sites and manufacturing processes research standards (e.g., indication-speciﬁc endpoints, have become increasingly common, posing challenges data collection, analytics) to ensure patients are protected to developers regarding reproducibility and compara- and results are reliable. bility of results. The paper describing the NIBSC/WHO In 2018, FACT published four standards (two in Lentiviral Vector Copy Number Standard discusses the collaboration with JAICE) that gene therapy developers concept of developing a first World Health Organization can reference in the context of clinical trials. These stan- International Standard, suitable for the standardization dards are: FACT Standards for Immune Effector Cells of assays and enabling comparison of cross-trial and (First Edition, Version 1.1); FACT Immune Effector cross-manufacturing results for this important vector Cells Accreditation Manual (First Edition, Version 1.1); platform. The standard will be expected to optimize the FACT-JACIE International Standards for Hematopoietic development of gene therapy medicinal products, which Cellular Therapy Product Collection, Processing, and is especially important given the usually orphan nature Administration (Seventh Edition); and FACT-JACIE of the diseases to be treated, which naturally hamper Hematopoietic Cellular Therapy Accreditation Manual reproducibility and comparability of results.

CHAPTER 2: Standards in Gene Therapy 40 Links to further information: • Existing reference materials—new collaborative NIBSC Gene Therapy page: study in kit form with primers/probes, standards and with accompanying methods for quantitation and https://www.nibsc.org/science_and_research/advanced_ infectious titer (supply cells) therapies/gene_therapy.aspx • AAV9 as a new reference material—could be pro- duced for a platform program Publication on development of the standard: • Empty and full AAV capsid preparation https://www.liebertpub.com/doi/full/10.1089/ • AAV plasmid standards with multiple PCR targets hgtb.2017.078 • Raw materials standards (plasmid DNA), possibly Collaborative Study report: https://www.who.int/ both best practices and reference standards for biologicals/expert_committee/BS.2019.2373_Lentiviral_ quality of these materials. vector_IS_Study_Report_final.pdf 5. Plasmid DNA Expert Panel 3. ISBio Lentiviral Vector Reference Material Project USP is also forming an expert panel, under the guidance The International Society for BioProcess Technology of the new Advanced Therapies Expert Committee, to (ISBio) has organized a consortium of stakeholders develop a documentary standard for the use of plasmid with the goal of producing a reference material for use in DNA in the manufacturing of advanced therapies. analytical methods used to characterize lentiviral prod- The landing page for information relating to USP’s de- ucts. This effort is currently underway and information velopment activities for advanced therapies can be found relating to the effort can be found here: https://isbiotech. here: https://www.usp.org/biologics/cell-tissue-standards org/ReferenceMaterials/lentivirus-home.html DEVELOPING SUPPORTIVE STANDARDS 4. USP Standard Development for Gene Therapy IN GENE THERAPY Over the last few years, USP has engaged with stake- Supportive standards are not developed speciﬁcally for the holders, regulatory authorities, and government scien- gene therapy sector, but are applicable to one or more of tists to explore the development of new standards for these sectors. They often occur as part of new standard the growing field of gene therapy. In early 2019, USP development and cover the same application areas as sec- held a roundtable with representatives from industry tor-speciﬁc standards. As a result, they can be used directly and regulatory agencies to explore opportunities for in each sector or as foundations for creating sector-speciﬁc the development of standards to support advancement standards. As the efforts to develop standards for regener- of AAV-based gene therapies. The goal of the meeting ative medicine continue, more supportive standards will was to facilitate a robust discussion on possible ways that be identiﬁed and categorized in the aforementioned The documentary and performance standards could help Regenerative Medicine Standards Landscape. standardize methods for assessing the quality of both raw materials and AAV drug substance. The roundtable NON-DOCUMENTARY STANDARDS IN GENE THERAPY participants shared challenges and discussed the need In addition to documentary standards, multiple non-doc- for best practices and methods, in addition to physical umentary reference standards have been, or are being, reference standards that could be made available through developed that can be useful for developers of gene ther- USP. The outcome of this and other more recent inter- apy. Groups such as ATCC are working to identify and actions is a list of AAV standards that USP is looking to distribute the most impactful and critical reference stan- develop, either internally or with collaborators, to aid dards needed by the gene therapy community. Examples developers of AAV products in harmonizing and stan- include ATCC Recombinant Adeno-Associated Virus 2 dardizing their practices, methods, and materials. The reference standard material (AAV2-RSM) and ATCC list, as of early 2020, is as follows: Recombinant Adeno-Associated Virus 8 reference stan- dard material (AAV8-RSM).

CHAPTER 2: Standards in Gene Therapy 41 In addition, the reference material development pro- data and identify ways for studies to consistently validate cess is being supported by a number of organizations. It is their methodology to speed up approvals. critical to bring the community of stakeholders together to • Clinical trial standards needs identify and discuss needed reference material for industry An overview of standards that are needed is provid- use (e.g., vectors, genomic DNA, cell cultures, serums) and ed for each of the regenerative medicine sectors in the then to determine if a given material meets the criteria for Needed Standards Report, which is a component of the development. Key questions often asked include: Regenerative Medicine Landscape Report. This report is • Do existing standards or reference materials address a useful reference for SDOs, researchers, industry mem- the need at hand? bers, and other stakeholders as they work to bring for- • Is the proposed material market-relevant? ward therapies from bench to bedside. Regular updates • Are stakeholders committed to advancing the to the report will help to ensure that the report remains reference material? The community introduces the comprehensive and will enable its use in identifying concept to a reference material development orga- emerging standards needs in the ﬁeld. nization or group for consideration (which varies in terms of formality required). Implementing Standards in Gene NEEDS AND GAPS FOR STANDARDS IN GENE Therapy Manufacturing THERAPY BY FUNCTIONAL AREA Variations in manufacturing, measurement, and an- STANDARDS BASED ON INTENDED USE alytical techniques across developers of experimental Ways to implement standards based on intended use include: gene therapy products cause difficulties for evaluating product quality and safety and addressing the impact of • Deﬁning the standards landscape manufacturing changes or innovations. A common set • Allowing stakeholders to more easily identify gaps of standards in gene therapy will advance development and ways to move the ﬁeld forward Coordinate and of treatments beyond the realm of clinical trials, to safe support standard development approved treatments for genetic diseases and syndromes. • Driving efficiency and allow stakeholders from across The gene therapy sector’s overall needs for standards the regenerative medicine community to make their development are summarized below by functional area: voices heard • Educating and build awareness of standard • Reference materials for analytical testing • Encouraging adoption of the standard and help • Analytical and testing methodologies needs — identi- stakeholders understand the beneﬁts the standard can fy more consistent methods for cell counting in gene bring their organization therapy to reduce disparities in evaluation, standard- • Prioritizing needed standards ized methods for evaluating endogenous chimeric • Allowing energy to be focused on the standards that antigen receptor T-cells (CAR-T), and consistent will have the greatest impact reporting methods and requirements to use for vector • Conducting feasibility assessment genome quantiﬁcation • Ensuring that the standards selected are scientiﬁcally • Product quality and characterization needs — revisit ready for development and likely to be adopted by the existing standards for replication competent retrovi- regenerative medicine community rus/lentivirus testing, standardize methods to assess • Outlining Development Process, Post-Development product activity or comparability of gene therapy Process, Pre-Development Process products, and reﬁne and clarify release criteria to • Distributing educational materials that convey ensure that products are effective and reliable beneﬁts of a speciﬁc standard, stakeholders impacted, • Preclinical study needs—identify and standardize the meth- relevant regenerative medicine sectors, and product odologies used for collecting and evaluating biodistribution development processes, meetings and conferences.

CHAPTER 2: Standards in Gene Therapy 42 Table 2-2. Gene Therapy Standards Addressing Bioprocessing and Production

Standard Publication SDO* Standard Name Number Status

EP 5.2.12* EDQM Raw materials of biological origin for the production of cell-based and Published 2017 gene therapy medicinal products

EP 5.14* EDQM Gene transfer medicinal products for human use Published 2008

N/A* FACT FACT Standards for Immune Effector Cells (First Edition, Version 1.1) Published 2018

N/A FACT FACT Immune Effector Cells Accreditation Manual (First Edition, Published 2018 Version 1.1)

N/A* FACT FACT-JACIE International Standards for Hematopoietic Cellular Therapy Published 2018 Product Collection, Processing, and Administration (Seventh Edition)

N/A* FACT FACT-JACIE Hematopoietic Cellular Therapy Accreditation Manual Published 2018 (Seventh Edition)

ISO/PWI 20389* ISO Collection, processing, conserving, and transportation technology In development criteria for human genetic resources

ISO/DI 20688-1* ISO Biotechnology—Nucleic acid synthesis—Part 1: General definitions and In development requirements for the production and quality control of synthesized oligonucleotides

ISO Biotechnology—Nucleic acid synthesis—Part 1: General definitions and In development requirements for the production and quality control of synthesized gene fragments, genes, and genomes

USP ** USP Gene Therapy Products Published 2011, revised 2020

*Standards Development Organization **Indications standards that apply to multiple standards application areas or sectors

GENE THERAPY BIOPROCESSING AND industry stakeholders, to “coordinate and prioritize the PRODUCTION STANDARDS FOR MANUFACTURING development of standards and consensus deﬁnition of Bioprocessing involves the design and development of terms... [that] support, through regulatory predictability, processes, materials, and equipment for manufacturing the development, evaluation, and review of regenerative products from raw/ancillary biological materials (with ap- medicine therapies and regenerative advanced therapies.” propriate procedures for characterization or starting mate- Gene therapies present complex challenges related to rials such as cells, gene therapy vectors, and biomaterials). product testing, scientiﬁc protocols, product quality and Table 2 lists gene therapy standards related to biopro- speciﬁcations, performance characteristics, and compli- cessing and production. ance criteria. The FDA assigned Nexight and SCB the role The 21st Century Cures Act—signed into law in of developing the landscape of existing standards relevant December 2016—directs the U.S. Food and Drug to regenerative medicine, a task not previously undertaken Administration (FDA), in consultation with the National for this field because of its novelty and the relative lack Institute of Standards and Technology (NIST) and standards speciﬁc to regenerative medicine therapies.

CHAPTER 2: Standards in Gene Therapy 43 Standards in Gene Therapy and stakeholders and CDER staff proposing voluntary con- sensus standards related to pharmaceutical quality for Regulatory Approval informal recognition. CDER believes that this informal program, which is different than the formal recognition FDA’S INFORMAL RECOGNITION OF VOLUNTARY standards program in FDA’s Center for Devices and CONSENSUS STANDARDS Radiological Health, will help promote innovation in Voluntary consensus standards can be defined as stan- pharmaceutical development and manufacturing and dards developed by voluntary consensus standards bodies. streamline the compilation and assessment of marketing The Food and Drug Administration Modernization Act applications for products regulated by CDER. It should of 1997 (FDAMA) (Pub. L. No. 105-115) and the 21st be noted, however, that even if an applicant decides to Century Cures Act of 2016 (Pub. L. No. 114-255) amended use one of CDER’s informally recognized voluntary section 514(c) of the Federal Food, Drug, and Cosmetic standards, CDER may request that the applicant provide Act (FDC Act) to require FDA recognition of voluntary additional information to support an Investigational New consensus standards. FDA has used such standards to Drug (IND) application or a marketing application. In develop and/or evaluate performance characteristics of addition, the applicant’s use of an informally recognized dosage forms, testing methodologies, manufacturing consensus standard will be strictly voluntary. CDER has practices, product standards, scientiﬁc protocols, compli- issued this draft guidance to obtain public comments on ance criteria, ingredient speciﬁcations, labeling of drug the proposed program. After CDER considers submit- products, and other technical or policy criteria. ted comments, CDER will establish this program and In addition, FDA’s Center for Drug Evaluation and describe it by publishing a ﬁnal guidance. Thus, though Research (CDER) has drafted a document titled CDER’s not yet formalized, this draft guidance, when ﬁnalized, Program for the Recognition of Voluntary Consensus will comprehensively represent the current thinking of Standards Related to Pharmaceutical Quality Guidance the Food and Drug Administration. for Industry. This guidance describes a proposed pro- gram at CDER to make public a comprehensive listing FDA’S FORMAL RECOGNITION OF STANDARDS of informally recognized voluntary consensus stan- In addition to informal recognition, FDA also issues dards related to pharmaceutical quality. The program, formal recognition of standards. FDA recognizes con- once established, will facilitate submissions by external sensus standards are standards that FDA has vetted

Figure 2-2. Gene Therapy Standards Needs by Functional Area

Bioprocessing and Production 3

Analytics and Testing Methods 6

Product Quality and Characterization 9

Logistics and Compliance 7

Preclinical Study 3

Clinical Trial 4

source: Standards Coordinating Body; Needed Standards Report; December 2020

CHAPTER 2: Standards in Gene Therapy 44 Want a say in the development of standards for regenerative medicine therapies? and determined are appropriate to support clearance Standards Coordinating Body or approval of a device. The purpose of FDA’s formal recognition of consensus standards is to streamline the —Coordination, Community premarketSCB review process. CAN This formal recognition HELP. allows Engagement, and Education companies to submit a declaration of conformity with a recognizedThe Standards standard Coordinating in a premarket Body application, (SCB) is an rather unbiased The rapid development of gene therapy products – 373 thannon-profit submit complete organization data and dedicated test reports to acceleratingdemonstrat- theproducts in clinicalTHE trials, STANDARDS and 536 developers worldwide advancement of standards that address the needs ing conformity with a standard. (ARM SectorCOORDINATING data Feb 2021)—presents BODY challenges and of the global regenerative medicine community. FDA maintains a formal database of recognized opportunitiesENGAGES for standards the broader development. regenerative The goal of consensus standards. This database consists of national SCB was tomedicine establish community processes inthat the create an effective SCB’s coordination ensures: identification, prioritization, and international standards recognized by FDA to network of regenerative medicine community stakehold- • Standards that impact the field have broad and advancement of potential standards to incorporate a range of which community manufacturers input can declare conformity and is a ers to coordinate and complement current standards component of the information that regulators can use developmentperspectives processes. and It is expertise widely accepted that broad to• make Time an and appropriate resources decision are focused regarding on the the clearance stakeholderCOORDINATES involvement is and necessary communicates to ensure that the about standards activities across the or approvalstandards of a thatsubmission. could yield the greatest establishment of any new standards and/or reference benefits to the field regenerative medicine community to In the Jan 2020 FDA Chemistry, Manufacturing, and materials provideaccelerate the standards greatest benefits advancement to the gene ther- Control• Those (CMC) who Informationare championing for Human standards Gene work Therapy apy community.EDUCATES SCB theapproaches community these about challenges by Investigational together instead New ofDrug duplicating Applications efforts (INDs)— focusing onavailable Coordination, standards Engagement, and their benefits, and Education. standards development processes, Guidance• Hurdles for areIndustry, reduced FDA for recognizessubject matter 3 types experts of ref - Coordination:and standards A critical implementation first step was to identify and erence providing standards: input 1) Certified into standards reference standards (e.g., document the standards needed in regenerative medi- USP compendial standards); 2) Commercially supplied cine. SCB has accomplished documenting the needs in its reference standards obtained from a reputable commer- Needed Standards Report updated in December of 2020. cialSCB source; connects and/or 3) Other the materialsregenerative of documented medicine In that community report, 32 areas were identified as important purity,to the custom-synthesized standards bydevelopment an analytical laboratory process. for Gene Therapy (Figure 2). orStandards other noncommercial development establishment. for regenerative medicine therapies is being encouraged by the 21st Century Cures Act and will continue to ramp up as the field matures. With buy-in from a range of stakeholder Figureorganizations 2-4. Standards across the Coordinating regenerative medicine Body’s Approach community, to our Accelerating goal is to address Standards your needs Development and support the advancement of standards that enhance the entire community and accelerate innovation.

IDENTIFY CONDUCT EDUCATE DEVELOP STANDARDS PRIORITIZATION AND BUILD Industry STANDARDS NEEDS AND FEASIBILITY AWARENESS Regenerative Care Providers Medicine Patients Community Researchers Gov’t Agencies

Standards Development Organizations www.standardscoordinatingbody.org source: Standards Coordinating Body

CHAPTER 2: Standards in Gene Therapy 45 Executive Summary

COMMUNITY PRIORITIZATION OF STANDARDS NEEDS These prioritization results are based on input received from approximately 60 stakeholders from various groups within the regenerative medicine community, including industry, public-private partnerships, government agencies, standards developing organizations (SDOs), academia, and healthcare providers. These results have not been peer reviewed but are intended to provide a snapshot of perspectives from the community. Figure 2-3. Community Prioritization of Standards Needs → HIGH URGENCY W ← LO

← LOW IMPACT HIGH →

Figure from The Standards Coordinating Body Needed Standards Report; December 2020. The areas for standards development are identified and then ranked for both urgency and impact. (G = Gene Therapy Specific, C = Cell Therapy Specific, T = Tissue Engineering Specific).

The identified areas for standards development are Operational Feasibility: identified and then ranked for both urgency and impact. • Are there sufficient interested parties willing to (Figure 3). commit individuals with appropriate training and resources to provide scientific and experiential Engagement: SCB uses a process (Figure 3) to then knowledge to the project? both prioritize and perform a Feasibility Assessment • Does SCB have sufficient staffing to perform the to determine if a standard is ready to progress to formal program management function for the project? development. • Are sufficient funds available immediately (or likely InCommunity general, operational Perspectives: and technicalNeeded Standards feasibilities in are Regenerative to be obtainable) Medicine to December adequately 2020 support the standards5 key to standards development. These can be defined as: development? • Does an existing SDO(s) have a committee project that is a good fit for the standards development?

CHAPTER 2: Standards in Gene Therapy 46 Technical Feasibility: After the feasibility assessment concludes that a stan- • Documentary Standard dard is ready for development, formal working groups • Are there applicable scientific/engineering knowl- are established to address that standard’s needs. As stan- edge in the field to provide sufficient information dards development can take months to years to occur to inform development of the standard(s) (i.e., the depending on the complexity and nature of the project, concept is not completely science fiction)? SCB’s facilitated process ensures that development occurs • Does the documentary standard require round-robin as efficiently and quickly as possible. testing? • Standards Reference Material Education and Implementation—SCB recognizes that access to information and help with implementation • Do potential materials that could easily be adapted of standards are important for the industry. To help the for a standards reference material (SRM) exist, or is industry, SCB has developed an online portal for search- discovery science needed to get there? ing for standards and uses its website for timely and up • Do manufacturing methods exist to create sufficient to date information. quantities of the SRM to generate a stockpile? It is through the processes of: Coordination—com- • Do testing methods that could be directly applied to munity and stakeholder engagement (identification of the SRM already exist? standards), Engagement—facilitated Working Groups • Does a documentary standard for testing the SRM (feasibility and development), and Education (outreach need to be developed concurrent with development and implementation) that SCB works to accelerate stan- of the SRM? dards development.

CHAPTER 2: Standards in Gene Therapy 47 Chapter 3 Generation of a Quality Target Product Profile Chapter 3 | Contents

Introduction...... 50 Key Terminology and Definitions...... 50 Unique Challenges ...... 52 Manufacturing and Product Quality Considerations...... 54 QTPP and QbD...... 54 Developing a Target Product Profile...... 54 Developing a QTPP...... 55 Identification of CQAs...... 55 Analytical Methods...... 55 Strength and Dose...... 58 Vector Genome Titration Methods...... 58 Infectious Genome Titration: Biological Activity...... 58 Product-Related Impurities...... 59 Particle Quantification: Percentage of Empty Capsids...... 60 Illegitimate Encapsidated DNA...... 60 Conclusion...... 60 Endnotes...... 61

CHAPTER 3 Generation of a Quality Target Product Profile 49 Introduction of gene therapy products will follow CMC (Chemistry, Manufacturing, and Control) aspects already established Cell and gene therapies cover a broad spectrum of for therapeutic proteins. therapies, from stem cell transplants and lymphocyte Conventional biologics and gene therapies are similar transplants to ex vivo and in vivo gene therapies and with respect to replacing missing proteins or increasing gene editing. Many of these therapies are moving from the level of a protein for disorders where insufficient academic settings to biopharmaceutical-focused compa- amounts of a given protein are produced. The primary nies, and as these therapies enter late-phase clinical trials difference between biologics and gene therapies is that a and commercialization, their manufacturing processes biologic is repeatedly given to the patient as a parenteral are being improved to increase reproducibility and ro- product. In the case of conventional biologics, a produc- bustness and to reduce the cost of manufacturing. These tion cell line, such as Chinese Hamster Ovarian (CHO) technologies have the capacity to revolutionize how we or SP2/0 myeloma, is genetically modified to secret treat and potentially cure disease. However, because the protein of interest (e.g., monoclonal antibodies or they are complex biologics, much of the development therapeutic proteins). The majority of these proteins are

Key Terminology and Definitions

Quality Target Product Profile (QTPP) Critical Quality Attribute (CQA) A prospective summary of the quality characteris- The QTPP facilitates the identification of CQAs, tics of a drug product that ideally will be achieved which are physical, chemical, biological, or micro- to ensure the desired quality, taking into account biological properties or characteristics that should safety and efficacy.1 Because the QTPP defines be within an appropriate limit, range, or distribu- essential criteria that relate to the development of tion to ensure the desired product quality (ICH a new product, it is critical that the development Q8.1 Examples of CQAs of AAV products include team clearly deﬁnes the quality attributes of the physical viral titer (i.e., viral genomes or full AAV product. In the pharmaceutical industry, these particles per unit volume), capsid content (i.e., per- product attributes are referred to as the target centage of empty capsids), potency, purity, and product proﬁle (TPP). Ideally, the TPP describes product stability. Preliminary CQAs are defined how the end user will utilize the product, and early in phase 1 of the drug development lifecy- includes the clinical delivery of the drug product. cle using risk assessments. Preliminary CQAs are For the company, the TPP will help to identify further investigated using design of experiments project goals and potential risks.2 The terms TPP (DOEs)/process experience and continue to be and QTPP are sometimes used interchangeably. In refined during the early phase of the development practice, the TPP is broader in scope and typi- lifecycle based on enhanced product knowledge cally includes some items that are absent from and early clinical experience. CQAs serve as the the QTPP, including marketing inputs (ie, desired basis to identify critical process parameters (CPPs) claims) and clinical inputs. The QTPP is meant to and facilitate development of the design space.3 identify, define, and justify quality characteristics so as to ensure safety and efficacy expectations established by the TPP. The focus is on the QTPP for the purposes of the A-Gene document. continued on next page

CHAPTER 3 Generation of a Quality Target Product Profile 50 Key Terminology and Definitions continued from previous page

Critical Process Parameter (CPP) Risk Assessment A process parameter whose variability has an Risk assessment is a valuable science-based impact on a CQA and therefore should be moni- process that can aid in identifying which material tored or controlled to ensure the process produc- attributes and process parameters potentially es the desired quality (ICH Q8). Examples of CPPs have an effect on product CQAs. Risk assessment include temperature, pH, cooling rate, rotation is typically performed early in the pharmaceutical speed, etc. Because CPPs impact the CQA, they development process and is repeated as more must be monitored or controlled via a well-de- information becomes available and greater knowl- signed process to enable early and accurate edge is obtained. It can be defined as “an assess- detection of deviations outside acceptable limits ment of the ability of the process to reliably pro- that will impact product quality. Not all process duce a product of the intended quality (e.g., the parameters have the same impact on CQAs; performance of the manufacturing process under some may have a greater impact than others. As a different operating conditions, at different scales result, it is important to prioritize CPPs over other or with different equipment). An understanding process parameters as they will have the most of process robustness, or process capability, can impact. Of all process parameters, CPPs must be be useful in risk assessment and risk reduction the most rigorously controlled. Critical material and to support future manufacturing and process attributes (CMAs) are often used when determin- improvement, especially in conjunction with the ing CPPs and their impact on CQAs. A non-CPP use of risk management tools.” Examples of com- is a process parameter whose variability has no mon risk assessment tools include parameter risk significant impact on a CQA and therefore does assessments (PRA) and failure mode and effects not have to be controlled to ensure the process analysis (FMEA), which are covered in detail in produces the desired quality. Chapter 4 (see also ICH Q9 Quality Risk Man- agement).1 Chapter 4 contains a more in-depth Critical Material Attribute (CMA) discussion of risk assessment approaches. A physical, chemical, biological, or microbiological property or characteristic of an input material that Quality by Design (QbD) should be within an appropriate limit, range, or A systematic approach to development that be- distribution to ensure the desired quality of output gins with predefined objectives and emphasizes material.1 product and process understanding and process Key Process Parameters (KPPs) control, based on sound science and quality risk Parameters of the manufacturing process that management.1 may not be directly linked to critical product qual- ity attributes but need to be tightly controlled to ensure process consistency as it relates to prod- uct quality.1

CHAPTER 3 Generation of a Quality Target Product Profile 51 Figure 3-1. Overview of Gene Therapy Manufacturing

Drug Cell Plasmid Virus Virus Purification product Expansion Transfection Production Recovery production Master Transfection Virus Virus 1- or 2-step Filtration cell bank with 2-4 production extraction, purification and vial and cell plasmid phase post- clarification, process filling expansion vectors transfection and concen- ahead of DS adherent in tration formulation cell culture

For additional details on upstream and downstream manufacturing, please refer to Chapter 5.

produced, purified, and formulated with well-defined Unique Challenges and robust processes. Many of these processes have been optimized through the application of QbD and require Gene therapies are complex biologics that pose a chal- well-defined processes to ensure equivalent quality attri- lenge to the currently available analytic techniques for butes between lots and manufacturing sites. Similarly, the full and comprehensive characterization when compared FDA requires that gene therapy manufacturing processes to a monoclonal antibody. Rather than the repeated ad- be well characterized.4 ministration of a well-characterized therapeutic protein The manufacturing of AAV often differs from that to the patient, the patient’s own cells produce the miss- of conventional therapeutic proteins; additional critical ing protein or express the appropriate receptors for the manufacturing steps are required for the expression and desired biological function with gene therapy products. assembly of viral vector components (Figure 1). In par- Plasmids, viral vectors, or mRNA are used to introduce ticular, AAV therapeutics are generally produced from a the genetic material required to modify these cells. transient transfection of cells and do not derive from a Plasmids can be introduced directly into cells using DNA stably transfected cell line. This requires the design and nanoparticles or used as the precursor genetic material production of plasmids that encode the viral vector cap- required for the production of viral vectors. In either sid, enzymes required for replication, and the transgene case, QbD principles and process optimization can be (sequence encoding the protein of interest). Once the applied equally to manufacturing of plasmids and viral appropriate plasmids have been synthesized, they are vectors and to modified cells produced for theex vivo used to produce bacterial cell banks that will produce autologous therapies (e.g., CAR-T therapies and ther- the plasmids required for producing the viral vector. The apies for hematological disorders). Additionally, QbD production of the plasmids can be considered as precur- can be applied to in vivo or in situ gene therapies using sor starting materials required for the production of the lentiviral and AAV vectors, as well as nonviral vectors. viral vector. Viral vectors can be made using transient The following discussion will focus on the application of production by transfecting HEK293 cells with the appro- QbD principles to AAV vectors. priate plasmids. It is feasible that HEK293 cells could be Ensuring the process is capable and in-control is es- engineered to stably express the capsid and adenovirus sential. Many of these complex processes are difficult to helper components, thus enabling a single-plasmid define and lack the forward planning typically deployed transfection approach. Alternative production systems in the development and registration of a commercial include insect cells/baculovirus systems, producer cell product. Though the industry has achieved significant lines in combination with adeno- or HSV-helper viruses, milestones, the initial products required additional or upcoming helper virus–free producer cell systems. efforts, resources, and time to retrospectively address

CHAPTER 3 Generation of a Quality Target Product Profile 52 concerns around the quality of the material inputs, scaled to meet the market demand with minimal major testing and characterization, quality systems, and the changes during clinical development, especially during development of robust processes. As the gene therapy the pivotal and post-pivotal stages. industry matures, the speed to clinical commercialization Often during DOE studies, following a process and must be tempered with the appropriate development of product risk assessment, some attributes may be des- these products. ignated as CQAs but are adequately controlled by the It should be noted that though the FDA is flexible process. Thus, it is not appropriate to have tests for these with respect to early-stage cell and gene therapies for CQAs on the control system. However, agencies may unmet medical needs, prior to initiating pivotal clinical expect characterization data to show that the applicant studies the sponsor must demonstrate that the product in has evaluated these CQAs and understands the risk. An development includes a solid commercial plan for meet- example in AAV-derived gene therapy is certain capsid ing demands. While a product may be developed under post-translational modifications such as deamidation accelerated timelines, the sponsor is expected to apply that can severely impact AAV infectivity in hepatocytes. best practices in order to ensure safety and efficacy. The For an in-depth discussion of risk assessment, please refer manufacturing process should be robust and sufficiently to Chapter.4

Table 3-1. Comparison of Empirical vs QbD Approach6

Aspect Minimal Approaches (Empirical) Enhanced QbD Approaches

Overall • Mainly empirical • Systematic, relating mechanistic understanding of pharmaceutical • Developmental research often material attributes and process parameters to drug development conducted with one variable at a product CQAs time • Multivariate experiments to understand product and process • Utilization of process analytical tools (PATs)

Manufacturing • Fixed • Lifecycle approach to validation and ideally continu- process • Validation primarily based on initial ous process verification full-scale batches • Focus on control strategy and robustness • Focus on optimization and • Use of statistical process control methods reproducibility

Process controls • In-process tests primarily for go/ • PATs utilized with appropriate feedback and feed-for- no-go decisions ward controls • Off-line analysis • Process operations tracked and trended to support continual improvement efforts after approval

Product • Primary means of control • Part of the overall quality control strategy specifications • Based on batch data available at • Based on desired product performance with relevant time of registration supportive data

Control strategy • Drug product quality controlled pri- • Drug product quality ensured by risk-based control marily by intermediates (in-process strategy for well understood product and process materials) and end product testing controls • Quality controls shifted upstream, with the possibility of real-time release testing or reduced end-product testing

Lifecycle • Reactive (i.e., problem solving and • Preventive action management corrective action) • Continual improvement facilitated

CHAPTER 3 Generation of a Quality Target Product Profile 53 MANUFACTURING AND PRODUCT QUALITY defining the QTPP, which defines the product’s mode CONSIDERATIONS of action, patient population, and dosage and product The sponsor of a product that receives an expedited drug format, as well as the preliminary CQAs (pCQAs) that development designation may need to pursue a more ensure safety and efficacy. Developers of gene therapies rapid manufacturing development program to accom- leverage historic processes and use QTPP, CQAs, and modate the accelerated pace of the clinical program. The process risk assessments to develop the preliminary pro- sponsor’s product quality and CMC teams should initiate cess control strategy (pPCS). The pPCS is further refined early communication with the FDA to ensure that the through additional studies that use the DOE approach, manufacturing development programs and timing of which permits the systematic reduction of the degree submissions meet the Agency’s expectations for licen- of experimentation necessary to define and optimize sure or marketing approval. When sponsors receive an the process control parameters, KPPs, and non-KPPs. expedited drug development designation, they should Through DOE, control limits can be precisely defined be prepared to propose a commercial manufacturing with respect to a parameter’s impact on quality, produc- program that will ensure availability of quality product tivity, and yield. This increases the degree of freedom or at the time of approval. The proposal should consider “design space” and permits continuous refinement within estimated market demand and the commercial manu- the product’s lifecycle.1 facturing development plan, as well as manufacturing A product’s development lifecycle is based on various facilities and a lifecycle approach to process validation. elements, including, but not limited to, the following.1 Additionally, the proposal should include a timeline for development of manufacturing capabilities with goals • In-process controls aligned to the clinical development program.5 • Process design A challenge for developing gene therapies for orphan • Environmental controls products is that the small patient population requires • Process and analytical capability only a limited number of batches, impacting the ability • Raw material controls to generate sufficient data to develop acceptance crite- • Drug substance and drug product specifications ria for CQAs and control strategies. The small patient • Demonstrated product stability population impacts the justification to expend significant • Process monitoring controls resources for process definition and optimization. • Product comparability studies • Process validation QTPP and QbD DEVELOPING A TARGET PRODUCT PROFILE QbD ensures quality by defining the critical process con- The TPP is the foundation for the strategy that in- trols and testing that guarantee that the drug substance corporates scientific, clinical, and market information and drug product will meet the attribute specifications necessary for an effective development plan. The TPP that ensure safety and efficacy. Through the application of is a living document that is updated continually during DOE techniques, defined upper and lower control limits the drug development process.7 The initial version of the for CPPs can be set to ensure lot-to-lot comparability. In TPP should be created at the Pre-IND stage. contrast to the empirical approach of manufacturing a A typical biopharmaceutical TPP includes the follow- product and performing release testing, the quality of the ing sections:7 product is considered at the earliest possible stage rather than simply testing the product for quality towards to the • General product information: brief description of end of the process (Table 1). Although it is evolving, most the genetic construct, product name (or designa- of the gene therapy industry is at the minimal approach tion), general information stage due to the challenges mentioned here. • Mechanism of action: mechanism by which the The principles of QbD are built on the foundation of product produces an effect on a living organism

CHAPTER 3 Generation of a Quality Target Product Profile 54 • Clinical pharmacology: pharmacokinetic informa- information on how the process impacts the identified tion, distribution, and pathways for transformation CQAs and the detectability of the CQA to determine a and safety control strategy.6 Reassessment of the criticality will con- • Indication for use: target disease and population, tinue as part of life cycle management as more data are dosage, and any relevant concerns with special available or as related to changes of process or methods populations and the ability to remove or detect an attribute. Quality • Target manufacturing profile: formulation, shelf life, risk management (QRM, as described in ICH Q9) can be storage conditions, and delivery system used to assess the risk and criticality of variability in the • Primary efficacy endpoints: primary clinical out- identified product quality attributes during manufacture, come measures; endpoints are usually proposed as and the resulting analyses form the basis of setting man- three different scenarios: minimal, base, optimal ufacturing processing parameter controls. Knowledge • Secondary efficacy endpoints: additional endpoints management (as described in ICH Q10) is key to that are not required to be met in a clinical trial capturing and applying prior knowledge of biological, • Expected safety outcomes: primary safety outcome chemical, and manufacturing principles and experience measures to the establishment of the QTPP and also assessing the • Contraindications: known or expected criticality of quality attributes and the degree to which a contraindications control strategy is needed.7 Table 2 shows an example of • Commercial landscape: description of the competi- a QTPP for a generic AAV gene therapy product. tive landscape at expected launch time • Regulatory: expected BLA/approval date Analytical Methods DEVELOPING A QTPP QbD requires the use of robust and comprehensive ana- The FDA states that the QTPP should consider the safe- lytical methods to confirm product identity and measure ty and efficacy of the product, giving consideration to the impact of process-related impurities. Analytical dosage strength, delivery system, dosage form, container methods cover a wide array of attributes and can be system, purity, stability, and sterility, and more (Table 2). tested for by a number of mechanisms. In the generation The QTPP describes the design criteria for the product of an AAV gene therapy, a developer needs to focus on and should therefore form the basis for development of product characterization, such as strength or genomic the CQAs, CPPs, and control strategy.8 identity, as well as process related characterization, such as empty/full capsid ratio. Safety tests may evaluate puri- IDENTIFICATION OF CQAS ty, toxicity, and stability, as well as the physical, chemical, The second foundational step to the QbD process is and biological characteristics to the product in order to defining the CQAs that impart safety and efficacy. CQAs ensure that the product falls within the predetermined derived from the QTPP and/or prior knowledge are used acceptable limits of identity, potency, quality, and purity. to guide product and process development. These can Analytical procedures are similar to the method be modified throughout the product development as lifecycle approach to process validation in that the pro- knowledge and product understanding increase through cedure should be developed as early as possible during experimentation and risk assessment.9 the lifecycle to minimize risk. Consistent methods should CQAs are defined during development and locked be used throughout the process development and man- for the marketing application. Using risk assessment ufacturing stages. However, as process knowledge and (see Chapter 4), the QbD process is used to identify understanding increase, analytical assays may change CQAs with each parameter or attribute given a rating of or be replaced. A risk-based approach must be used criticality with respect to its impact on safety and efficacy. to determine whether manufacturing process changes The criticality rating of each attribute will subsequently necessitate revalidation of the analytical procedures that be assessed together with other knowledge, such as are used. In addition, methods should be evaluated to

CHAPTER 3 Generation of a Quality Target Product Profile 55 Table 3-2. QTPP for a Generic AAV Gene Therapy Product

ASSUMPTIONS

• This is a quality target product profile (QTPP) for a generic AAV-based gene therapy product produced by trans- fection of HEK293 cells and purified via column chromatography. • This QTPP is written to be adaptable for various routes of administration such as intravenous, ocular, and via the central nervous system. • The targets for general properties and product attributes based on these assumptions are listed below. However, this approach can be adapted for other products based on dose, target tissue, or route of administration. • Targets for product quality attributes are listed below but specifications will be assessed throughout the product lifecycle based on clinical, manufacturing, and assay experience.

GENERAL PROPERTIES

Property Target Geographic Scope Global: test to USP, EP, JP compendia where possible Shelf life ≥1 to 2 years at intended storage conditions Storage conditions Formulation stable as a liquid at 5°C or alternately frozen at -20°C or ≤-65°C

Container Selected to support product compatibility and stability and to ensure sterility Delivery volume Variable based on patient weight

BULK DRUG SUBSTANCE (DS) AND FINAL DRUG PRODUCT (DP) ATTRIBUTES

Attribute Category Attribute Bulk DS Target Final DP Target Safety Bioburden Set limit to provide confirmation of NA process and facility controls

Safety Sterility NA Sterile

Safety Endotoxin Set limit to provide confirmation of Below USP body weight- process and facility controls with based limit for the route additional consideration to USP limit of administration specific to route of administration

Content/strength Appearance and Clear to slightly cloudy, colorless solution practically free of foreign particulates particulates

Content/strength pH Appropriate for the formulation to support product stability and compatible with the route of administration

Content/strength Osmolality Appropriate for the formulation to support product stability and compatible with the route of administration

Content/strength Vector genome titer Stable concentration that balances Dose-specific volumetric impact of sampling and FDP concentration for manufacturing requirements ddPCR methods 10 Content/strength Potency Relative to reference standard by in vitro transduction (disease relevant cell line preferred)

Content/strength Potency/infectious Infectivity (e.g., TCID50), cell-based assay for GOI function, animal genome titer model demonstrating clinical benefit after transduction for early phase trials Identity Capsid identity Capsid serotype confirmed

CHAPTER 3 Generation of a Quality Target Product Profile 56 Table 3-2. QTPP for a Generic AAV Gene Therapy Product continued from previous page

Identity Genome identity Genome identity confirmed via Genome identity combination of transgene-specific ddPCR confirmed via transgene- method and sequence analysis specific ddPCR method Complete viral genome confirmation by sequencing recommended for DS only

Process impurities Residual host-cell Base limit on amount dosed in relevant NA protein toxicology studies

Process impurities Residual host-cell free Base limit on amount dosed in relevant NA DNA toxicology studies11, 12

Process impurities Residual host-cell Base limit on amount dosed in relevant packaged non-target toxicology studies11, 12 DNA Process impurities Residual plasmid DNA Base limit on amount dosed in relevant NA toxicology studies

Process impurities Residual cell culture Base limit on supplier safety data or other NA media components available literature

Process impurities Residual transfection Base limit on supplier safety data or other NA reagent available literature

Process impurities Residual chromatogra- Base limit on supplier safety data or other NA phy ligand available literature

Process impurities Replication-competent ≤1 replication competent AAV/108 genome NA AAV copies

Purity Capsid protein purity ≥90%

Purity Capsid protein ratio Consistent across product lifecycle NA

Purity % full capsids Specific target set for general alignment NA with relevant toxicology studies

Purity Capsid protein modifi- Within set limits to ensure functional NA cation (deamidation, consistency in manufactured products oxidation) Purity Total capsids When considered with % full capsids, NA ensure the total number of viral particles delivered does not exceed amount delivered in relevant toxicology studies

Purity Subvisible particles NA Meets relevant USP chapter, USP , or USP

Purity Aggregates Acceptable level so as not to affect dose Acceptable level so as or concentration not to affect dose or concentration

For additional information on the final DP target, please refer to Chapter 6.

CHAPTER 3 Generation of a Quality Target Product Profile 57 determine their robustness and applicability to late-phase Table 3-3. Impurities Encountered in AAV Vector clinical trials. Here, we discuss a number of attributes Manufacturing 15 a developer must consider in their development, and have included examples of analytical methods which Attribute Class Quality Attribute may be used to measure these attributes. Additional at- Microbiological Sterility tributes and assays are discussed throughout this A-Gene quality document. Bacterial endotoxin

STRENGTH AND DOSE Adventitious agents Measurement of strength/dose of purified AAV vectors Bacteriostasis, fungistasis, no typically includes assays to quantify the genome concen- inhibition of growth tration, infectious concentration, and functional activity Mycoplasma, mycoplasmastasis of the transgene (gene of interest).13 Replication-competent AAV Vector Genome Titration Methods Product-related Viral capsid ratio and purity Because vector genome (VG) is the key component impurities involved in rendering the therapeutic effect in gene Empty capsid to full capsid ratio therapy, focus has been on developing fast, reliable, Aggregation and robust methods for its titration. ddPCR is the gold standard method to titer VG AAV. It is important to note Process-related Residual host cell DNA that attention must be paid when designing the region impurities Residual plasmid DNA dedicated to performing the VG titer. qPCR standards preparation and stability are key factors to consider when Residual affinity column ligand attempting to increase inter-assay precision.13 Host-cell protein

Infectious Genome Titration: Biological Activity Transfection agent (PEI) Methods for quantifying rAAV infectious particles that can be applied to any vector, independently of the trans- Lysis agent (detergent) gene product, and rely on the detection of rAAV genome Product quality Appearance replication in the presence of AAV rep-cap genes and characteristics adenovirus. In particular, two methods are used most Osmolality frequently: replication center assay (RCA) and the 50% pH tissue culture infective dose (TCID50) assay, which in- volve inoculation of serial dilutions of the rAAV vector Biological activity: Gene copy number content made on HeLa rep-cap-trans-complementing cells (i.e., Viral particle number HeLaRC32 or C12 cells) co-infected with adenovirus type 5. Vectors carrying reporter genes such as green Biological activity: In vitro infectivity fluorescent protein (GFP) can be easily titrated by flow strength cytometry in transduction units (TU/mL). For therapeu- Biological Activity: Transgene expression tic AAV-based transgenes, infectious genome titers (IG) Potency are expressed as infectious units (IU)13 The development Transgene product functional activity of relative infectivity assays is on the rise leading to more precise measurements of infectivity.14 Identity Sequence identity

Capsid identity

CHAPTER 3 Generation of a Quality Target Product Profile 58 Table 3-4. Example of release testing of AAV vector product

Attribute Assay Method

Strength/Dose VG titer Spectroscopy/fluorimetry, qPCR/ddPCR Infectious Genome Relative infectivity assay, TCID50

Total vector particles ELISA

Activity (expression assay) Cell-based assay

Potency (functional activity) Cell-based or in vivo assay

Identity Genome DNA Sequencing VP proteins Western blot, mass spectroscopy

Host-cell DNA qPCR/ddPCR

Helper plasmids or virus DNA qPCR/ddPCR

Residual reagents and raw materials (antibiotic ELISA, HPLC resistance genes, detergent, benzonase, BSA, column leachables, etc)

Percentage of empty capsid qPCR/ELISA, HPLC, electron microscopy, analytical ultracentrifugation

Safety Sterility EP 2.6.1 USP Bacterial endotoxins EP 2.6.14, USP

Mycoplasma EP 2.6.7

Adventitious viruses EP 2.6.16

rcAAV Cell-based assay

Bioburden Direct inoculation of sample into various media; confirmed by a bacteriostasis/ fungistasis test

Vector aggregates Dynamic light scattering, SEC-HPLC

PRODUCT-RELATED IMPURITIES for each attribute due to safety/impact and uncertainty. Product-related impurities remaining after purification Table 3 is an example of product-related impurities in include, but are not limited to, nucleic acids (encap- AAV product. sidated DNA, such as host-cell DNA, helper plasmid, Process-related impurities may include proteins or helper virus DNA, or host-cell DNA or plasmid and nucleic acids derived from the production cells DNA that is not encapsidated) and vector aggregates. and viruses used to manufacture the therapeutic vec- Particles that are empty or have encapsidated nucleic tors; unpackaged viral vector genomes, and empty or acids fragments other than the therapeutic genome partially filled capsids. In addition, residual cell culture cassette are considered product-related impurities. components (e.g., antibiotics, supplements, inducers) These impurities are inactive and may represent safety may be present, as well as residual purification buffers, concerns. Therefore, reliable methods for their measure- chromatography media ligands, centrifugation media, ment and characterization have been suggested.13 CQAs detergents, enzymes, inorganic salts, etc. are derived from QTPP considering the risk assessment

CHAPTER 3 Generation of a Quality Target Product Profile 59 PARTICLE QUANTIFICATION: PERCENTAGE OF Conclusion EMPTY CAPSIDS Empty particles are undesired AAV products that are The compilation of all analytical methods covers the produced at a significant level during the biosynthesis end-to end of a developer’s quality control program in of AAV vectors. Empty particles may represent up to drug substance (DS) generation and will inform the 90% of vector preparations. In order to determine the parameters for release testing of the drug product (DP). ratio of full to empty particles in a single experiment, Table 4 provides an example of release testing born from techniques such as spectrometry, HPLC, and electron the combination of a number of product- and process-re- microscopy have been proposed. For example, one lated analytical methods. spectrometric approach uses the absorbance ratio at Assay validation is an important consideration. 260 nm/ 280 nm to quantify the number of empty and Assays that measure identity, strength, quality, purity, and full particles of AAV2; however, this approach requires potency of DS and DP must be validated to demonstrate purified material and is sensitive to impurities and buffer that the analytical procedures and assays are adequate formulation. Some laboratories have already used HPLC and meet standards of accuracy, sensitivity, specificity, as an analytical tool.13 Currently, employing analytical and reproducibility. Further, it is important to consider ultracentrifugation (AUC) has proven one of the most phase-appropriate development of assays and note that consistent methods for full versus empty capsid analysis.6 assays may change throughout the product lifecycle as process and product knowledge improves. Finally, as ILLEGITIMATE ENCAPSIDATED DNA technology advances, real-time release testing and PAT During recombinant AAV production, viral capsids are will improve the monitoring of product quality, efficacy, known to package not only their genomes flanked by 2 and safety while simultaneously shortening the manu- ITR but also various DNA fragments. Several types of facturing time and decreasing costs. illegitimate DNA encapsidation, helper virus sequences An enhanced, carefully designed approach to gene including rep/cap sequences, DNA fragments from therapy development allows for the reduction of release plasmids and cellular genome have been identified in tests. QbD allows for testing at the appropriate control purified AAV vector preparations.11, 12 However, vector point, which may be better as a critical in-process control design and a good manufacturing process can reduce with appropriate rejection or acceptance limits, rather illegitimate encapsidated DNA, and this should be con- than release testing. sidered when discussing QTTP.

CHAPTER 3 Generation of a Quality Target Product Profile 60 Endnotes

1. International Conference on Harmonisation of Technical 9. US Department of Health and Human Services, Requirements for Registration of Pharmaceuticals for Food and Drug Administration, Center for Biologics Human Use. Pharmaceutical development Q8(R2): step Evaluation and Research. Chemistry, manufacturing, 4 version. International Conference on Harmonisation and control (CMC) information for human gene therapy of Technical Requirements for Registration of investigational new drug applications (INDs): guidance Pharmaceuticals for Human Use website. https:// for industry. https://www.fda.gov/media/113760/ database.ich.org/sites/default/files/Q8_R2_Guideline.pdf. download. Published January 2020. Accessed November Published August 2009. Accessed November 15, 2020. 15, 2020. 2. Lambert WJ. Considerations in developing a target 10. Košir AB, Divieto C, Pavšič J, et al. Droplet volume product profile for parenteral pharmaceutical products. variability as a critical factor for accuracy of absolute AAPS PharmSciTech. 2010;11(3):1476-1481. quantification using droplet digital PCR. Anal Bioanal Chem. 2017;409(28):6689-6697. 3. Kelly M. Implementing the principles of quality by design for early stage gene therapy products. Presented at: 11. World Health Organization. Guidelines on the quality, WCBO 2012, January 23-25, 2012. safety, and efficacy of biotherapeutic protein products 4. Wright JF. Manufacturing and characterization of prepared by recombinant DNA technology. World Health AAV-based vectors for use in clinical studies. Gene Ther. Organization website. https://www.who.int/biologicals/ 20008;15(11):840-848. biotherapeutics/rDNA_DB_final_19_Nov_2013.pdf. Published 2013. Accessed November 15, 2020. 5. US Department of Health and Human Services, Food and Drug Administration, Center for Biologics 12. CMC Biotech Working Group. A-Mab: a case study in Evaluation and Research. Guidance for industry: expedit- bioprocess development (v2.1). https://cdn.ymaws.com/ ed programs for serious conditions – drugs and biologics. www.casss.org/resource/resmgr/cmc_no_amer/cmc_ US FDA website. https://www.fda.gov/files/drugs/ amab_case_study/A-Mab_Case_Study_Version_2-1.pdf. published/Expedited-Programs-for-Serious-Conditions- CASSS website. Published October 30, 2009. Accessed Drugs-and-Biologics.pdf. Published May 2014. Accessed November 15, 2020. November 15, 2020. 6. Burnham B, Nass S, Kong E, et al. Analytical ultracen- 13. Dorange F, Le Bec C. Analytical approaches to trifugation as an approach to characterize recombinant characterize AAV vector production and purification: adeno-associated viral vectors. Hum Gene Ther Methods. Advances and challenges. Cell Gene Ther Insights. 2015;26(6):228-242. 2018;4(2):119-129. 7. US Department of Health and Human Services, 14. Li Z, Wu Z, Maekawa T, et al. Analytical technology used Food and Drug Administration, Center for Biologics in the latest development of gene therapy candidates. Cell Evaluation and Research. Guidance for industry: target Gene Ther Insights. 2019;5(4):537-547. product profile — a strategic development process tool 15. Fraser Wright J. Product-related impurities in clini- (draft guidance). US FDA website. https://www.fda.gov/ cal-grade recombinant AAV vectors: characterization and media/72566/download. Published March 2007. Accessed risk assessment. Biomedicines. 2014;2(1):80-97. November 15, 2020. 8. European Biopharmaceutical Enterprises. Considerations in setting specifications. European Biopharmaceutical Enterprises website. https://www.ebe-biopharma. eu/wp-content/uploads/2017/04/ebe-concept-pa- per-%E2%80%93-considerations-in-setting-specifications. pdf. Published March 28, 2013. Accessed November 15, 2020.

CHAPTER 3 Generation of a Quality Target Product Profile 61 Chapter 4 Process Development Using Quality by Design (QbD) Principles

The National Institute for Innovation in Manufacturing Biopharmaceuticals 1.0 2 .0 3.0 Brand Overview Logo Stationery 4 .0 5.0 6.0 Collateral Media Legal 7.0 8.0 Frequently Used Contact Page Terms

3 | The National Institute for Innovation in Manufacturing Biopharmaceuticals Chapter 4 | Contents

Chapter Summary...... 64 Key Points...... 64 Introduction...... 65 The QbD Process...... 65 Similarities and Differences Between AAV Products and Biologics/Vaccines...... 65 Overview of Risk-Based Strategy Timeline...... 67 Stage I: Process Design...... 68 Stage II: Process Qualification...... 69 Stage III: Continued Process Verification...... 69 Development and Manufacturing Activities...... 70 Risk Assessment...... 70 Target Product Profile and QTPP...... 70 Overview of CQAs...... 70 CQAs as Related to Regulatory Timing and Expedited Pathways...... 71 CQA Considerations Relevant to Gene Therapy...... 73 Quality Attributes Risk Assessment...... 73 Severity Scoring and Filtering...... 74 Preliminary Hazard Analysis (PHA)...... 76 Case Study...... 76 Process Development Considerations for Product Quality...... 78 Unencapsidated DNA...... 78 Cell Culture–Related Impurities...... 78 Downstream-Related Impurities...... 78 Helper Viruses...... 78 Parameter and Material Risk Assessment...... 78 Upstream Examples...... 79 Downstream Examples...... 81 Scale-down Models...... 82 Process Characterization...... 84 Example 1: OFAT Study...... 85 Example 2: Screening Study...... 85 Example 3: Response Surface Methodology (RSM)...... 86 Example 4: Definitive Screening Design (DSD)...... 86 Output for PC: Proven Acceptance Ranges and Normal Operating Ranges...... 86 Parameter Criticality Assessment...... 88 Parameter Classification Assessment (PCA) or FMEA...... 90 Control Strategy for Critical Process Parameters...... 92 Potency...... 92 Viral Clearance...... 93 HCDNA...... 93 Process Analytical Technologies (PAT)...... 93 Capacitance Probe...... 94 Differential Digital Holographic Microscopy (DDHM)...... 94 Spectroscopy Technology...... 95 Conclusion...... 96 CMC Readiness Checklist...... 97 Endnotes...... 98

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 63 Chapter Summary

Process development aims to establish and characterize a manufacturing process that can be scaled up to commercial-size batches while continuing to yield a consistent, quality product. Various risk-based approaches are utilized to ensure that the process stays within appropriate limits and meets all safety and quality benchmarks. In this chapter, we discuss var- ious aspects of QbD that are used to evaluate the process and mitigate risks, quality attributes risk assessment from the quality target product profile (QTPP), and parameter risk assessment (PRA), among others. Other important pieces embedded into QbD include process development considerations, scale-down models, process characterization, parameter criticality assessments, parameter classification assessments (PCA), associ- ated control strategy for critical process parameters, and process analytical technologies. Process analytical technologies (PAT) have been supported by agencies to gain better process control, as PAT can provide real-time monitoring for important indicators for the process that could potentially impact the quality.

Key Points • As many gene therapy products er-stage development focuses on process characterization could help proceed toward late-stage devel- scale, optimization (productivity to define the control of process opment and BLA-enabling activi- and yield), and final facility. parameters and the control of raw ties, it is going to be increasingly • Establishment of process perfor- materials. Failure mode and effects important to take lessons from mance requirements within the analysis (FMEA) can then identify biologics and vaccines in applying manufacturing process requires areas that require improvement to a systematic, risk-based approach knowledge of a product’s critical further reduce risk. This ensures for late-stage development. quality attributes (CQAs) and ulti- that the process delivers a quality mately allows the robust, reproduc- product in a reproducible process. • The goal of early process develop- ible production of a quality product. ment is to establish a platform-rel- • The implementation of PAT into evant process for scalable and • Various risk-based approaches process development and GMP transferable process that allows are used throughout late-stage manufacturing is relatively recent characterization and proof-of-clin- development to understand and and aims to obtain real-time infor- ical-concept to be obtained within summarize risks for parameters in mation for process control. a reasonable amount of time. Lat- each unit operation. The follow-up

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 64 Introduction • Parameter risk assessment (PRA): A PRA is conducted to determine which parameters and The goal of process development is to optimize a associated materials have potential to affect the high-titer process and maintain product quality at an CQAs and thus provide a defined set of parameters early stage, as well as to characterize a manufacturing that may pose critical risk to quality and process process to yield a consistent, quality product at late- performance. A PRA is carried out using a cause stage development. Quality-by-design (QbD) is one and effect methodology to assess each parameter approach that places emphasis on the use of the product for its effect on each CQA. Scientific understanding and understanding of the process to develop a robust is used to score the severity of each effect. During process control strategy that is rooted in sound science a PRA, it is important to assess the criticality of and quality risk management. For a deeper discussion of process materials to ensure that the parameter QbD, please refer to Chapter 3. Establishment of process studies consider material variability. Failure to performance requirements within the manufacturing consider material variability during process char- process requires knowledge of a product’s critical qual- acterization may result in process characterization ity attributes (CQAs) and ultimately allows the robust, that is specific to material lots and not necessarily reproducible production of a quality product. In this reflective of the manufacturing process. way, quality is not maintained through product testing; instead, quality is built into the entire process though • Process characterization (PC): PRA can be used rational, intentional design.1 Figure 1 illustrates the as part of input for PC. Certain low-risk parameters overall flow of the QbD approach.2 may need to be characterized for compliance. PC defines the list of high-risk parameters and THE QBD PROCESS materials that should be investigated further based Various risk-based approaches are used throughout the on high severity scores determined during the PRA. manufacturing process to establish processing operations Essentially, the PC results in a prospective plan of and define the control of critical quality attributes, the all experiments to be conducted during process control of raw materials, and the areas for further investi- characterization. gation. Such assessments ensure that the process delivers a quality product in a reproducible process. A cross-func- • Parameter criticality assessment: The PCA is a tool tional team of subject matter experts conducts the risk that is used after process characterization studies assessments using scientific understanding and product to define which process parameters are critical, key, knowledge to identify areas that pose a risk of failure or noncritical based on study data and additional to deliver quality product and product that satisfies the operational information. quality target product profile (QTPP), as described in Chapter 3. Such risk-based approaches include:4 SIMILARITIES AND DIFFERENCES BETWEEN AAV PRODUCTS AND BIOLOGICS/VACCINES • Critical quality attribute assessment (CQAA): A Some of the approaches in developing AAV products are CQAA is conducted to determine which quality similar to those already in place for biologics and vaccine attributes identified in the QTTP are potentially development. For example, regulatory requirements critical to product safety and efficacy. It forms the include process characterization, process performance initial list of potential analytical method require- qualification (PPQ) for the understanding of links be- ments and the starting point for process assessment. tween process and product quality, defining acceptable/ Over time, the CQA list may be updated based normal operating ranges, and defining critical/key on clinical history, product characterization, and parameters to enable PPQ. In addition, making sure increased scientific understanding. the processes to define process and analytical control strategies to ensure product quality meets the release

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 65 Figure 4-1. Overview of the QbD Process3

Quality Target A prospective summary of the quality characteristics of a drug product that ideally Product Profile will be achieved to ensure the desired quality, taking into account safety and (QTTP) efficacy of the drug product. (ICH Q8 (R2))

Critical Quality The physical, chemical, biological, or microbiological property or characteristic Attribute (CQA) that should be within an appropriate limit, range, or distribution to ensure the desired product quality. (ICH Q8 (R2))

Parameter Risk A tool used by a cross-functional team of subject matter experts to step through Assessment (PRA) each process unit operation, assessing the parameters and the material used with the purpose of identifying potential risks and scoring the risk

Process Process characterization is the step to find out the proven acceptance range (PAR) Characterization for the studied parameters. A prospective plan derived from the high severity scores determined during the PRA, as well as any specific functional studies area owners determine are needed. The high-risk parameters and materials are identified in the PC for further study to determine the effect their variability may have on the ability to control the process and produce a quality product. Qualified scale-down model is prerequsite for process characterization and could also be included in PC.

Parameter A tool used by a cross-functional team of subject matter experts after process Classification characterization studies are completed to define, based on study data and Assessment (PCA) additional operational information, which process parameters are critical, key, or noncritical.

Process and Analytical A planned set of controls, derived from current product and process Control Strategy understanding, that assures process performance and product quality. (ICH Q10). (P/ACS) The controls can include parameters as inputs and attributes or indicators as outputs related to drug substance and drug product materials and components, facility and equipment operating conditions, in-process controls, finished product specifications, the associated methods, and frequency of monitoring and control.

Process Performance A series of process runs that combines the actual facility, utilities, equipment (each Qualification (PPQ) now qualified), and the trained personnel with the commercial manufacturing process, control procedures, and components to produce the commercial batches. A successful PPQ will confirm the process design and demonstrate that the commercial manufacturing process performs as expected.

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 66 criteria. Throughout the development of these three types Opportunities for a risk-based approach in the devel- of products, it is important to maintain a vigilant eye opment of AAV products include the following: for comparability when implementing process changes. As in biologics and other manufacturing processes, it is • Process characterization: risk assessments must be critical to establish controls using good manufacturing carried out to target the most critical parameters practice to ensure the quality and safety of investigational and limit the scope of experimentation. 5 drug products. • Viral clearance: robust risk assessment must be car- However, challenges are associated with expedited ried out for non-helper virus processes (Ph1-Ph3). development, and abbreviated approval pathways differ from those of biologics/vaccine development. For exam- • Starting materials (e.g., plasmids, helper viruses, ple, multiple manufacturing platforms (e.g., HEK293, etc.): it is important to define a streamlined path to qualifying plasmid processes. It is also critical to HeLa, SF9) are available for AAV products, whereas understand the link between plasmid quality and the approach is more standardized for biologics (e.g., vector process performance. mainly on CHO for monoclonal antibodies) and vac- cines. Thus, manufacturing processes are insufficiently Importantly, in the development of assessment to only defined compared to other well-established methods in target most critical parameters, a greater risk approach biologics. This variance leads to the need for different can be used. A risk-based approach such as this could control strategies that must be implemented for AAV help to identify and therefore focus on the most critical compared to biologics and vaccines. Further, the full parameters, while logging and monitoring uncertain risks structure-function relationship of AAV products is not or parameters. For example, a robust process that does not completely characterized, and the analytical methods are involve a helper virus may be used so that a risk assess- complex and potentially variable. Issues associated with ment can be conducted in lieu of formal viral clearance AAV manufacturing include low productivity/limited studies during early stages of clinical development. While material, complicated production systems, low volumes a comprehensive characterization of the plasmid processes and concentrations, accelerated timelines, and limited is not necessary, it is important to streamline the path understanding of mechanisms of action. towards process characterization. It is perhaps less critical Many gene therapies target conditions for which to characterize detailed operating ranges of the plasmid there are high, urgent unmet medical needs. Such con- manufacturing process, but certainly more critical to con- ditions may allow for expedited development pathways. duct the development work to understand the link between QbD approaches, including process development and variability in the plasmid product quality attributes and the optimization, are critical to incorporate early during vector process performance and product quality. process development if a product is expected to achieve breakthrough designation from the U.S. Food and Drug Overview of Risk-Based Strategy Administration (FDA). Without early adoption of QbD principles, the process knowledge and design informa- Timeline tion that are needed to support BLA filing may not be Based on FDA guidance, process validation (defined available to support the aggressive regulatory timelines. as the collection and evaluation of data) involves three A breakthrough therapy, fast track, priority review, and/ stages: process design, in which the commercial manu- or accelerated approval designation may allow BLA filing facturing process is defined based on knowledge gained based on phase 2 data or a single phase 3 trial; however, through development and scale-up activities; process a full Chemistry, Manufacturing, and Controls (CMC) qualification, in which the process design is evaluated in package is still required, including process development order to determine whether the process allows reproduc- and validation. This CMC package may be needed at least ible commercial manufacturing; and continued process 3 to 5 years prior to when it would be for a traditional verification, in which routine production produces filing with such designations. ongoing assurance that the process remains in a state of

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 67 Figure 4-2. Risk-based Strategy for Late-Stage Development8

IND = investigational new drug MRA = material risk assessment; QTTP = quality target product profile; PCP = process characterization plan; attribute; pCQA = potential critical quality FMEA = failure mode effect analysis CQAA = critical quality attribute assessment; p/ACS = process and analytical control strategy CQA = critical quality attribute SDM = scale-down model PRA = parameter risk assessment; PPQ = process performance qualification

Pre-IND Early Stage Late Stage Commercial

Stage I: Stage II: Stage III: Process Design Process Continued Qualification Verification Process Process Validation Stages

CQAA PRA FMEA

QTPP PCP Final Parameter Process Criticality P/ACS Lock Assessment pCQA MRA Draft P/ACS CQA Risk Assessments and Milestone Documents

Platform Fit Process SDM PPQ Assessment Optimization

Process Characterization

Phase 1/2 MFG Process Potential Additional Campaign Characterization Phase 3 MFG DEV/MFG Campaigns DEV/MFG

Risk assessment Milestone document Activity

control. These stages allow the establishment of scientific variation in a manner appropriate to the level of risk that evidence to show that a process is capable of delivering a it carries to the process and the product.7 An overview of consistent, quality product.6 risk-based strategy for late-stage development for AAV Ultimately, a successful validation program requires products is shown in Figure 2. Note that some activities information and knowledge from product and process may occur in multiple stages. development. In order to attain this, manufacturers must understand sources of variation, detect the presence and STAGE I: PROCESS DESIGN degree of variation, understand the impact of variation Process design defines the commercial manufacturing on the process and product attributes, and control the process that will be implemented in planned master

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 68 production and control records. The ultimate goal of this is critical during this stage of process validation. The stage is to design a process that is suitable for routine protocol may include:14 commercial manufacturing and allows the consistent • Manufacturing conditions (e.g., operating parame- delivery of a product that meets its quality attributes. ters, processing limits, and raw material inputs) Steps involved in process design include building and • Data to be collected and when/how it will be capturing process knowledge and understanding and evaluated establishing a strategy for process control.9 • Tests to be performed and acceptance criteria for each step STAGE II: PROCESS QUALIFICATION • The sampling plan (including sampling points, During process qualification, the process design is evalu- number of samples, and frequency of sampling for ated in order to determine whether the process is capable each unit operation/attribute) of reproducible commercial manufacture. Completion of • Criteria and process performance indicators that stage II is required prior to commercial distribution, and allow a science- and risk-based view of whether the acceptable products that are manufactured during this stage process is capable of consistently producing quality may be released for distribution. This stage involves two el- products ements: design of a facility and qualification of utilities and • Design of facilities and qualifications of utilities, equipment and process performance qualification (PPQ).10 equipment, and personnel training and qualifica- Qualification of utilities and equipment may include tion, and verification of material sources (such as selecting utilities and equipment construction materials, components and container/closure) operating principles, and performance characteristics; • Status of validation of analytical methods that were verifying that utility systems and equipment are built/ used during the process, in-process materials, and installed in compliance with design specifications; and the product verifying that utility systems and equipment operate • Review and approval of the protocol by appropriate within the anticipated operating ranges.11 boards The PPQ combines the now qualified facility, utilities, equipment, and trained personnel with the commercial The output of process characterization is the establish- manufacturing process, control procedures, and compo- ment of acceptable operating ranges for those parameters nents in order to produce commercial batches. The goal that have been fully characterized. Process performance of PPQ is to confirm the process design and demonstrate qualification begins by the drafting of a protocol that that the commercial manufacturing process performs as defines the process by unit operations, set points, and it is expected. Successful PPQ is required prior to com- acceptable parameter ranges. The protocol prospectively mercial distribution, and data from commercial-scale specifies a number of full scale manufacturing lots that batches should be used as support to begin commercial will be run in the commercial manufacturing facility, and distribution, supported by data from laboratory and pilot product released for commercial disposition, in order studies. During the PPQ process, previous credible ex- to demonstrate that the process operates as expected perience with sufficiently similar products and processes within predefined ranges and thereby confirms that it is may be helpful, and meaningful objective measures, such operating under a satisfactory state of control. as statistical metrics, should be used whenever feasible.12 While PPQ will often have a higher level of sampling, STAGE III: CONTINUED PROCESS VERIFICATION additional testing, and greater scrutiny than would be The control strategy after stage II (or PPQ) is maintained expected in routine commercial processes, it should be and updated throughout stage III based on scientific sufficient to confirm uniform product quality throughout understanding, product knowledge, and gained infor- the batch.13 mation. In continued process verification, the goal is con- A written PPQ protocol to specify manufacturing tinued assurance that the process remains in a validated conditions, controls, testing, and expected outcomes state of control during commercial manufacture. In this

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 69 stage, it is critical to have one or more systems to detect abbreviated version of the process for evaluation (e.g., unplanned departures from the designed process.15 bioreactor). It could help to evaluate potential technology For this step, an ongoing program must collect and transfer problems, understand the scales and ranges against analyze product and process data related to product qual- which the product should be compared, and determine ity, including relevant process trends and the quality of which attributes demand focus. incoming materials or components, in-process material, After the process is locked and when the project moves and finished products. The data should be statistically to late-stage development, PRA and MRA commence. The evaluated and reviewed by appropriately trained per- outputs of PRA and MRA are included in the PCP, which sonnel. Information gathered during this stage should guides process characterization. The criticality assessment verify that the quality attributes are being controlled is performed based on the results of characterization and appropriately throughout the process. the updated PRA/MRA. Manufacturing, MSAT, and/or Although a thorough process design and develop- development teams carry out FMEA. ment should anticipate significant sources of variability and establish appropriate detection/control/mitigation Target Product Profile and QTPP strategies, it is likely that a process will encounter varia- tions that were not previously detected or to which the The target product profile (TPP) is a potential set of process was not exposed. Data gathered during this stage product label concepts that is refined during clinical allows the determination of ways to improve or optimize development. The quality target product profile (QTPP) the process by altering certain aspects (e.g., operating is developed from the TPP in order to provide a set of conditions, process controls, component, in-process ma- requirements for the product quality to ensure safety/ terial characteristics). However, well-justified rationale efficacy. These concepts define the product design- cri for the change, an implementation plan, and quality unit teria and form the basis for the development of CQAs, approval should be documented prior to implementing identification of CPPs, and the overall control strategy.16 such changes. Importantly, early definition of QTPP forms boundaries around quality product attributes. Once the QTPP is DEVELOPMENT AND MANUFACTURING ACTIVITIES defined, it is possible to move between manufactur- Development and manufacturing activities are nor- ing platforms so long as the QTPP is maintained and mally designed to be aligned with FDA-guided process comparability is demonstrated in the respective drug validation. Early-stage development involves testing on substance and drug product quality attributes. For more the platform if there is one or optimization to have a information on QTPP, please refer to Chapter 3. high titer while maintaining acceptable product quality. It is important to characterize vectors for clinical Scale-down model development/qualification and pro- studies and verify that the purified clinical vectors main- cess characterization occur around the time for phase tain purity, potency, safety, and stability over the course 3 clinical campaigns, and the output of the late-stage of their potential use in investigational studies. Safety development contributes to the PPQ. concerns may include sterility, mycoplasma, endotoxin, general safety, and adventitious viral agents. Potency may RISK ASSESSMENT be measured using assays to measure the concentration For risk assessment prior to IND filing, the QTPP should and functional activity of purified AAV vectors. Of note, be defined, including dosage, administration, and other stability studies can be run concurrent with clinical use product profile information. Potential CQA can also be during early-phase studies. developed based on QTPP so that the CQA assessment could be done before entering early stage development. OVERVIEW OF CQAS Starting during late-stage development, it is import- Manufacturers of gene therapy products must take ap- ant to determine the CQAs for characterization. During propriate measures to ensure that products being manu- the early-stage process, it may be possible to create an factured meet the required standards of quality from the

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 70 Figure 4-3. CQA Considerations During Development

Identify quality attributes that impact potency, pharmacokinetics/pharmacodynamics, Understanding CQA'a helps define an appropriate process immunogenicity, or safety (i.e., those that affect the drug recipients).

Control Strategy Process • Release testing CQAs Characterization • Stability testing • In-process control testing • Control of material attributes • Control of process parameters

Understanding functional relationships between process and CQA provides for an enhanced approach

perspective of both patients and regulatory authorities. parameters and monitoring, CQAs serve as the bench- In the context of biopharmaceutical manufacturing and marks that enable the properly informed selection of as stipulated in ICH Q6A, quality refers to the suitability operational ranges. Most importantly, CQAs help ensure of a drug substance or drug product for the intended that the final product will provide patients with the safest eventual use, and specifications that are associated with and most efficacious therapy possible. Figure 3 summa- this suitability are known as quality attributes. Product rizes the purpose and involvement of considerations quality attributes are selected for their ability to help related to CQA considerations with respect to the overall indicate the suitability of the product for its intended use. development process. They are important to demonstrate lot-to-lot consistency, generate solid clinical data, determine relationships be- CQAS AS RELATED TO REGULATORY TIMING AND EXPEDITED PATHWAYS tween product quality attributes and safety and efficacy, support establishing meaningful specifications, and show As gene therapy products provide patients with treatment comparability after manufacturing changes. options for conditions with few or no alternative therapies, All quality attributes must be assessed for consistently the timing for securing regulatory approval is of utmost measurable and quantifiable impacts on the safety, importance in allowing patients to benefit from the ad- efficacy, or other aspects of quality of the product. In vancements that gene therapy products can offer. For this turn, developers can define a profile of CQAs and CPPs. reason, developers often aim to achieve expedited pathway For further, more detailed discussion of CPPs and their designation(s) for gene therapy products. As an example relationship to CQAs, please refer to Chapter 5. of expedited regulatory pathway development, the FDA Understanding product CQAs is perhaps the most is putting an effort into various expedited pathways to critical aspect of establishing a suitable manufacturing accelerate the process product approval while still ensuring process, as well as establishing controls for assuring safety of products. For specific information about expe- product quality and consistency. Because variations in dited regulatory pathways with respect to gene therapy CQAs indicate the importance and impact of process products, please refer to Chapter 1.

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 71 CMC readiness, of which identification of CQAs is success of the product, leading to negative outcomes for a primary component, remains one of the major chal- patient access to needed therapies. To address these po- lenges in the expedited development of gene therapy tential pitfalls, manufacturers are strongly encouraged products. Moreover, unlike traditional small molecule by the FDA and other health authorities to devise a plan pharmaceuticals and well-characterized proteins, manu- of action to understand the CQAs that could potentially facturing processes for many gene therapy products must correlate with product quality and the clinical outcomes, consider commercial scalability and viability at a very and to implement this plan at all stages of the develop- early stage. As a result, insufficient readiness could poten- ment process at which CQAs could potentially be iden- tially lead to unnecessary delay tified. Accomplishing this goal and may further complicate and requires not only an in-depth CMC Readiness convolute interpretation of very understanding of the product costly clinical trial studies. This Developers, manufacturers, and and its associated analytics, but underscores the importance of sponsors should be aware of the also a systematic approach to CMC-focused communications following regarding CMC readiness correlate key product attributes with regulatory agencies early and regulatory timing: to various clinical outcomes. and often. Manufacturers should • Safety is the main focus allowing IND Besides the changes in- lay out a sound development to proceed. tentionally introduced by the and facility plan covering the • Pivotal or innovative trials cannot manufacturer, it is often chal- product’s lifecycle, including be initiated without sufficient lenging to establish and main- post-approval considerations. phase-based appropriate product tain consistency of the product Furthermore, manufacturers manufacturing control. throughout the development should conduct comprehensive • Expedited programs are designed to stages to the final commercial and detailed CMC readiness accelerate clinical development. process. The ability to produce exercises prior to the initiation • An accelerated clinical development a consistent product will depend of pivotal or licensing trials or program will allow less time for on controlling CPPs, which are before asking agencies for one CMC-related activities. established through the mon- of many expedited designations. • CMC readiness for an expedited itoring of CQAs, along with Sponsors should make sure that program requires additional other factors that define the the treatment is adequately well evidence of manufacturing control. overall quality of the product. understood, safe, and effective Collection of robust characteri- for marketing approval. zation data early in development The identification of CQAs can help better define CQAs and must be appropriate with respect to the context of the demonstrate comparability for major changes later in the timing required for the regulatory pathway(s) being pur- development process. sued. Changes to a process/product late in development At a minimum, regulators expect that manufacturing (such as during clinical trials) could potentially change development should include “identification of potential the product’s critical characteristics as they relate to CQAs associated with the drug substance so that those safety, efficacy, and other aspects of quality, which in characteristics having an impact on drug product quality turn could have great influence on decisions made by can be studied and controlled” (ICH Q11). Additional regulatory reviewers. However, for a variety of reasons, information is expected to be identified after commence- manufacturers are often compelled to introduce major ment of clinical trials, with ongoing refinement of knowl- manufacturing changes very late in the product’s devel- edge throughout the entirety of the development process. opment life cycle. Thus, failure to detect the potential Assay development should start at the preclinical impact of these changes at critical times during the devel- phase in order to promote better decision-making opment process may detrimentally affect the commercial throughout later phases of the development process

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 72 Figure 4-4. Systematic and Iterative Approach to Identifying Clinically Relevant CQAs

Candidate CQAs

Correlate CQA Clinically relevant Develop CQAs data with clinical outcomes assays

Correlate CQA with Define product quality specifications

and provide a high level of confidence that observations and clinically relevant CQAs is extremely complicated. during clinical phases are explainable and addressable. Accordingly, the agency encourages a systematic ap- Assessments based on assays should consider what is proach involving several steps: known about the impact (or lack thereof) of a particular attribute on the quality attributes listed in the QTPP • Start with the identification of several candidate (refer to tables in Chapter 3). These assessments can be CQAs for each product and the development of designed based on factors such as prior knowledge and qualified assays to measure such candidate CQAs. experience (e.g., platform knowledge, published informa- The knowledge gathered during the product devel- tion), non-clinical data, and clinical data. In some cases, opment cycle forms a scientific basis for establishing additional studies can help confirm attribute criticality meaningful specifications. or address gaps in knowledge regarding CQA impact on • In addition, there should be a systematic approach attributes such as potency or safety. to correlate CQAs with product quality, and clinical outcomes form the basis for establishing the biolog- CQA CONSIDERATIONS RELEVANT TO GENE THERAPY ic and clinical relevance of each candidate CQA. The process of acquiring in-depth product knowledge for well-characterized biologics and small molecules is well Through a highly systematic and iterative process, it is established and includes a development approach yield- possible to identify potential clinically relevant CQAs ing in-depth understanding of the CQAs of the product (Figure 4). and the need to better define critical manufacturing steps and CPPs. Quality Attributes Risk Assessment However, this approach has, for a variety of reasons, yet to be fully adopted by the gene therapy industry, Risk can be defined as a metric for the amount of danger partially due to challenging technical issues and inherent posed by a given situation or variable. With respect to biological limitations: biological products are complex, quality-related risk in cell therapy product manufactur- often heterogeneous mixtures, with complex mecha- ing, as is the case for other biopharmaceutical manufac- nism(s) of action (MOA). turing contexts, risk should always be assessed with the The FDA recognizes that identifying product-specific potential for diminishment of safety to the patient as the

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 73 primary concern. However, it should be noted that as far as business- and operations-related matters are con- SOURCES FOR GUIDANCE cerned, the attribution of risk to any given case may not The following sources provide guidance related to be straightforward, given that a multitude of stakeholders ensuring the quality of products and implementing are often involved in materializing the operations of the an effective risk management system: manufacturing program, and subjectivity and variation • FDA, Guidance for Industry: Quality Systems in the exact levels of factors such as uncertainty and/or Approach to Pharmaceutical cGMP Regulations severity are possible. If risk is inaccurately or dispropor- (September 2006) tionately estimated, patient access to efficacious therapies • ICH Q8(R2), Pharmaceutical Development may be reduced if risk is assessed too conservatively; (August 2009) similarly, patients may be exposed unnecessarily to side • ICH Q9, Quality Risk Management effects if risk is not assessed conservatively enough. Therefore, the goal of risk assessment is to minimize • ICH Q11, Development and Manufacture of Drug the potential for harm to patients while maximizing Substances, November 2012 therapeutic benefit. • Parenteral Drug Association, Technical Quality risk management is a systematic process to Report No. 54-4. Implementation of Quality assess and control risks to quality across the product’s Risk Management for Pharmaceutical and lifecycle. The level of formal documentation associated Biotechnology Manufacturing Operations, 2014 with the quality risk management process should be commensurate with the level of risk as determined through risk assessment. Risk assessments—a system- During risk evaluation, the results of the identification atic approach to support risk decisions using accurate, and analysis steps are compared against set criteria in analyzable, and well-organized information—serves as order to place risks in proper context with respect to the basis for establishing control measures and making the manufacturing program as a whole. Risk assessment informed decisions when managing risk. Risk assessment should be considered to be an evolving process that starts may be part of a formalized, integrated risk management at the bench-based research phase, develops through strategy, but can also benefit an organization when used the clinical stages, and continues up through product less formally to increase the scope and accuracy of an approval and commercial manufacturing. In order to organization’s institutional knowledge of the risks and properly guide this evolution, as well as to ensure that hazards intrinsic to its current processes, systems, and risk assessment results in the best possible outcomes operational business models. for patients, ongoing communication involving indus- The first step of risk assessment is the identification try, regulators, and where possible, patients, is vital. of possible sources of harm that are present in and/or Communication may relate to the existence, probability, inherent to the manufacturing process. Identification can severity, acceptability, and/or detectability, of risks, as be based on factors such as historical data, theoretical well as other aspects. analysis, and the informed opinions of experts. The next step of risk assessment is risk analysis, in which the SEVERITY SCORING AND FILTERING risks associated with the identified sources of harm are Severity scoring is an approach in which multiple factors estimated. Through qualitative and quantitative means, associated with each quality attribute identified to be a associations are established between the likelihood of a potential source of risk are evaluated regarding the se- harmful event occurring, the severity of harm should verity (also referred to as “impact”) of each factor with the event occur, and in many cases, the ease of detecting respect to potential effects on the safety (including im- the harmful event. munogenicity), efficacy (determined from either clinical The last step of risk assessment is risk evaluation. experience or potency assays), and/or pharmacokinetics

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 74 Table 4-1. Overview of Severity Scoring

Value Severity Severity of Effect 1 Low Variability in attribute has minor or negligible potential for decreased safety/efficacy. Negligible or minor transient adverse effects are expected based on historical experience.

3 Medium Variability in attribute may have moderate potential for decreased safety or efficacy within the clinical history of the product. Attribute may result in manageable adverse effect seen historically but no new adverse effects.

10 High Variability in attribute may have potential for severe effect on patient. Potential significant change in safety/efficacy or risk/benefit profiles. May result in a serious (reversible or irreversible) adverse effect.

and pharmacodynamics of the product, along with the Severity scoring is particularly helpful in situations in uncertainty about the information used to assess the which the range of risks and the potential consequences severity. In this way, scoring matrices for each factor are to be managed are diverse and difficult to compare using developed, after which the individual scores for each fac- other methods, and has the advantage of allowing for tor are multiplied together to generate a composite risk both quantitative and qualitative assessment of risks score. The risk score can then be compared with respect within the same organizational framework. If severity to the range of scores established by the scale generated scoring is properly applied at key points throughout by calculating the lowest and highest risk scores possible the product lifecycle, starting at the pre-IND phase and according to the systems of measurement being em- through to licensure and post-approval, developers will ployed. Finally, “filters,” in the form of weighting factors be able to identify the attributes that pose the highest or cut-offs for risk scores, can be used to scale or fit the levels of risk, and therefore will be able to implement ef- severity scoring to management or policy objectives. fective precautionary measures and mitigation strategies.

Table 4-2. Overview of Uncertainty Scoring

Value Severity Prior Knowledge Pre-Clinical Studies Clinical Studies

1 Low Well characterized effect based on Demonstrated relevance Significant extensive data (in vitro, in vivo, or of animal model results. clinical clinical). Large body of knowledge in Extensive in vitro and experience with the literature. in vivo studies for this this product. product.

2 Medium External published literature available. Only moderate in vitro and/ Only limited Well characterized effect known. or in vivo data available for clinical Internal data (in vitro, in vivo, or clinical) this product. experience with from this or similar class products. this product.

3 High Limited or no published external No data available for this No data scientific literature and no internal data product. available for this from this or similar class products. product.

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 75 Table 4-3. Critical Quality Attribute Classification Outcome

Uncertainty

1 (Low) 2 (Medium) 3 (High)

10 (High) 10 (Critical) 20 (Critical) 30 (Critical)

3 (Medium) 3 (Potential) 6 (Potential) 9 (Potential) Severity 1 (Low) 1 (Non-critical) 2 (Non-critical) 3 (Potential)

Product- and process-related impurities, as well as refers to the probability that, should a quality attribute microbiological quality/safety, biological activity, and stray outside of accepted ranges based on the most re- product identity can be ranked using the system described cent understanding and knowledge about the attribute in Table 1 and Table 2. By multiplying the severity score (drawn from literature, clinical, and non-clinical studies and uncertainty score, it is possible to rank and classify relevant to the product in question or similar products), the critical quality attributes, as shown in Table 3.17 the occurrence will affect the safety and/or efficacy of the For example, bioburden is a critical quality attribute. product. When limited clinical data are available for a Because the impact/severity of bioburden is high (10) particular quality attribute, likelihood is to be assessed and uncertainty is low (1), the overall ranking will be conservatively. In a similar way to how risk ranking critical (10 x 1 = 10). Another example might be AAV scores are determined, the risk priority number (RPN) impurities by SDS PAGE. The severity could be medium of PHA is calculated by multiplying the severity score (3) because protein impurities may have moderate or and the likelihood score. The direness of the risk posed potential immunogenic response. The uncertainty is low by the attribute in question can then be judged based on (1) due to the potential for preclinical or clinical data. its relative placement along the priority scale compared Thus, the overall ranking of AAV impurities would be to the other attributes being assessed. Because the assess- potential (1 x 3 = 3). ment of likelihood depends on prior knowledge, PHA is particularly useful when performing risk assessment PRELIMINARY HAZARD ANALYSIS (PHA) in existing systems. Preliminary hazards assessment (PHA) is a risk assess- PHA is most commonly used early in the development ment approach based on applying prior experience or of a project, at a time when there is little information on knowledge of a hazard or failure to identify future haz- design details or operating procedures. Thus, results of ards, hazardous situations, and events that might cause PHA can inform process and facility design, as well as harm, as well as to estimate their probability of occur- serve as a pointer for further study of quality attributes rence for a given activity, facility, product, or system. The using other risk management tools. tool consists of: 1) identifying the possibility that the risk event will happen, 2) the qualitative evaluation of the CASE STUDY extent of possible injury or damage to health, 3) a relative In this section, a risk assessment is performed to deter- ranking of the hazard using a combination of severity mine which quality attributes are critical to guide process and likelihood of occurrence, and 4) the identification validation and process characterization experiments. The of possible remedial measures. purpose of this risk assessment is to identify and summa- Similar to risk ranking, PHA is based in part on rize the CQAs for the generic AAV-based gene therapy severity, but unlike risk ranking, uses likelihood as product introduced in Chapter 3. CQAs are physical, the other parameter instead of uncertainty. Likelihood chemical, biological, or microbiological properties or

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 76 characteristics that should be within an appropriate limit, A risk-based approach has been adopted for identifi- range, or distribution in order to ensure the desired prod- cation and assignment of CQAs, similar to the principles uct quality. International Conference on Harmonization outlined in the A-Mab and A-Vax case studies published (ICH) Guideline Q8(R2): Pharmaceutical Development by CASSS, ISPE, and the PDA for applying Quality by requires the following: Design (QbD) principles to process development. The CQAs are derived from the Quality Target Product Identification of potential CQAs, including those re- Profile (Chapter 3), which forms the basis of design for lated to drug substance, drug product, and excipients, the development of the product as well as any knowledge so that any characteristics that may have an impact gained during the product and process development on the desired product quality can be studied and activities, in order to assign a risk ranking score for each controlled. quality attribute. Each quality attribute is evaluated for

Table 4-4. Quality Attribute Risk Assessment

Quality Severity Uncertainty Overall Attribute Quality Attribute Criticality Score Score Ranking Category Safety Bioburden 10 1 10 CQA Endotoxin 10 1 10 CQA Sterility 10 1 10 CQA Content/ Appearance/particulates 10 1 10 CQA strength pH 10 1 10 CQA Osmolality 10 1 10 CQA Vector genome titer 10 2 20 CQA Potency (protein expression) 10 2 20 CQA Potency/infectious genome titer 10 3 30 CQA Identity Capsid identity 10 1 10 CQA Genome identity 10 1 10 CQA Process Residual cell culture media 3 1 3 Potential CQA impurities components Residual host cell protein 1 1 1 Non-critical Residual plasmid DNA 3 1 3 Potential CQA Residual host cell DNA 3 1 3 Potential CQA Residual transfection reagent 3 1 1 Potential CQA Residual chromatography ligand 3 1 3 Potential CQA Replication-competent AAV 10 1 10 CQA Purity Capsid protein purity 10 2 20 CQA Capsid protein ratio 10 1 10 CQA % full capsids 3 3 9 Potential CQA Total capsids 3 1 3 Potential CQA Aggregates/subvisible particles 10 1 10 CQA

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 77 criticality by assessing its potential impact and uncertain- Downstream-Related Impurities ty as it relates to the efficacy and safety of the product. During product purification, downstream processes can During stage I activities during process validation, the introduce buffer, resin, and other impurities. In order to CQAs will be used to identify the CPPs for the proposed mitigate these risks, understanding and evaluation of the commercial manufacturing process via a risk assessment, tests are needed. For example, a leachable and extractable observations from historical experience, and findings study might be needed to demonstrate the effectiveness from process characterization experiments. of impurity removal during the downstream process. The identification and justification of CQAs are being performed in accordance with the QbD principles and Helper Viruses recommendations made in ICH Q8(R2), Pharmaceutical Helper viruses may be used in the upstream production Development and ICH Q9, Quality Risk Management, of AAV, and the viruses must be absent from the drug as well as the generally accepted approach described in product. Additional viral inactivation steps following guidance documents related to process validation. affinity purification may be needed, such as heat or acid inactivation after affinity step. A viral filtration step is Process Development Considerations recommended for AAV products going into late-phase production, even if viruses are not used in the system. for Product Quality Health authorities will require a clearance study when An overview of various process development con- approaching the registration/late phase stage. siderations can be found in Wright JF. Biomedicines 2014;2:80-97, including vector-related impurities (e.g., Parameter and Material Risk empty AAV particles), residual host nucleic acids, re- sidual helper DNA sequences, replication-competent Assessment AAV species, and noninfectious AAV vector particles.18 The PRA determines which parameters and associated Highlights and additional considerations are provided materials may affect CQAs using cause and effect meth- below. For additional information on processes used odology to assess each parameter for its effect on each to remove impurities, please refer to Chapter 5 of this CQA. The purpose of PRA is to define a set of parameters document. that may pose critical risk to quality and process perfor- mance. Such an assessment of the criticality of process Unencapsidated DNA materials is needed in order to ensure that material Foreign DNA can be introduced through a few ways: variability is considered during parameter studies.19 plasmid DNA, host cell DNA and intermediate species Gene therapy involves complicated analytics that of the gene of interest. The capsidated foreign DNA is would consume a considerable amount of time if each considered to be a product-related impurity. To minimize parameter were evaluated. A PRA allows a company to the amount of foreign DNA in the process, nucleases, reduce the number of parameters to a manageable num- such as benzonase, could be introduced. In addition, ber of high- or medium-risk parameters to be evaluated depth filtration and other downstream processes can within a reasonable amount of time. For example, hun- remove unencapsidated DNA. dreds of parameters may be identified, which following PRA may be reduced to ~30, allowing laboratory work Cell Culture–Related Impurities to be done within 6 months and analytics and reports Depending on the media used in cell culture, potential within an additional 3 months. Prioritizing parameters impurities can be introduced to the system. To mitigate that require further study help to reserve resources; the risks, nonanimal-derived or even chemical-defined analytics may require several samples of drug product, media can be used for production. For the working cell which is challenging in gene therapy development when bank (WCB), extensive tests and growth characterization product is extremely valuable and limited. are required before release for GMP production. It is important to analyze all parameters involved

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 78 Table 4-5. PRA Template

Line Unit Process Parameter Development Target PI PI PI CQA CQA CQA CQA CQA CQA Max of Sum of Justification/ Comments no. Operation Step Range Rating Rating Rationale 1 2 3 4 5 6 7 8 9 10 during the upstream-to-downstream process. Table • Study purpose 5 shows a PRA template to rank the impact of CQAs, • Materials and methods to be used which is then incorporated in future studies to allow • Study design faster product development. While process performance • Study acceptance criteria indicators are mainly business-related, critical quality attributes pertain to efficacy and safety and thus carry UPSTREAM EXAMPLES more weight in the risk assessment (Table 6). To provide more direction on how to perform the risk as- The output of PRA is a process characterization plan sessments, three assessment examples for upstream process (PCP), which defines the list of high-risk parameters and parameters and one raw material assessment are presented materials for further investigation based on high severity (see Table 7). The ratings and development ranges in the scores in the PRA. Ultimately, the PCP is a prospective examples are only for demonstration and are not intended plan including experiments that should be performed to represent the rating or ranges for any processes. during process characterization. The PCP should also The examples illustrate how to rate, use ranges, and define aspects of the scale-down model that need to be interpret the sum of the ratings and the maximum rat- considered.20 ing. If the rating is ≥10, further investigation is needed The output of PCP is a detailed characterization to show understanding and control of the process and protocol for each parameter study that will be conduct- product quality. Depending on the program, the sum of ed. Such characterization protocols should include the ratings requires team input to reflect both the impact and following:21 uncertainty for the evaluated raw material or parameter.

Table 4-6. Risk Ranking Criteria for Operating Parameters and Material Attributes

Parameter Risk Score Process Performance Indicator Critical Quality Attributes

1 No effect No effect 4 Minimal to moderate effect Minimal effect

7 Moderate to severe effect NA

10 NA Moderate to severe effect

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 79 Table 4-7. Upstream PRA Examples. Line Number Operation Unit Step Process Parameter Range Development Target Units VCD Viability Titer E:F Ratio Potency of Rating Max of Rating Sum

Incubator 1 Thaw Set-up 36-38 37 ºC 4 4 1 1 1 4 11 temp

Production Temp set 2 Set-up 35-59 37 ºC 7 7 7 4 4 7 29 (N) point

3 Transfection Mixing DNA ratioa N/A 1.0:1.0:0.2 N/A 4 4 7 10 4 10 29

PEI 4 Transfection Transfection N/A N/A N/A 4 4 7 4 4 7 23 quality

a(Ad Helper:Trans:Cis(GOI))

Example 1: Temperature for Thaw Stage control the final productivity. For cell growth, previous The set point for the temperature at this stage is 37°C. experience with higher temperature (i.e., 40°C) resulted Due to equipment capacity, 36-38°C was selected to in slower growth and low titer. Therefore, all process cover the normal operating range. As the temperature indicators (titer, VCD, and viability) were rated at 7. is controlled within a relatively narrow range and it is There was limited knowledge on the effect of temperature expected that the effect on cell growth will be minimal, variation on the quality of the product, so the CQAs were the rating for viable cell density (VCD) and viability for each rated a 4. Finally, the maximum rating for tempera- that stage are both 4. This is the first stage of the passage, ture for N stage was 7. The sum of the ratings was 29 due and cells could resume normal growth at a later stage. to the high effect on cell growth and titer, as well as the Therefore, the effect for the production stage for titer and uncertainty of the effect on the CQAs. the CQAs is negligible, and the titer and CQAs were all rated as 1. This results in the maximum rating of 4, and Example 3: Plasmid Ratio for Transfection the sum of the ratings is 11. Both the maximum and the The set point for the plasmid ratio was 1:1:0.2, and the sum indicate that this parameter has a low risk to affect developmental range has not been defined due to the the process performance or product quality. complexity of three variables. However, from previous experiments, it was known that plasmid ratio had mini- Example 2: Temperature for N Stage mal effect on cell growth and a moderate effect (20% to The set point for the temperature at this stage is 37°C. 30%) on titer. Therefore, VCD and viability were rated at The developmental range was selected to cover both the 4, and titer was rated at 7. Variation in the plasmid ratio normal operating range, as well as the potential charac- affects the empty-to-full ratio (E:F) with no observed terization range. Understanding the effect of temperature effect on potency. Therefore, E:F was rated at 10, and variation on the N-production stage could help to better potency and the remaining CQAs were rated at 4. The

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 80 Table 4-8. Downstream PRA Examples Line Number Operation Unit Step Process Parameter Range Development Target Units Yield Aggregation Purity Page SDS E:F Ratio of Rating Max of Rating Sum

1 Affinity Load Residence 4-10 7 min 4 1 1 1 4 7 time

2 CIM Q Column Capacity 5-11 10 CV 4 1 4 10 10 19 load

3 Buffer Formulation Pluronic N/A 10 ppm 1 4 1 1 4 7 F68

maximum rating was 10, indicating significant effect on presented (see Table 8). The ratings and development rang- at least 1 CQA or performance indicator. This will lead to es in the examples are only for demonstration and are not further characterization of the parameter to understand intended to represent the rating or ranges for any processes. the effect on the CQA (i.e., E:F). The sum of the ratings was 29. A high sum of the rating could reflect both high Example 1: Residence Time on the AAV8 Column impact on some CQAs, as well as an indication of re- The proposed residence time (4 to 10 min) for the AAV8 quiring more process knowledge around this parameter chromatography was expected to moderately affect the to sufficiently determine a control strategy to minimize performance indicator (PI) step yield. Poros resins are risk to the process and product. specifically engineered for the purification of large bio- molecules, and the rated pore size of the AAV8 resin is Example 4: Raw Material – Polyethyleneimine (PEI) ~0.2 µm. Therefore, the effect of residence time at the PEI quality is known from past experience to influence lower end of the proposed range was rated as 4 because titer. Depending on the amount used, as well as lot-to-lot it is expected to only moderately affect the step yield. variation, there could be minimal to moderate effect on Because AAV8 is operated in a bind-and-elute mode cell growth. Quantity of PEI will be evaluated separately where the impurities will be flowing through and only as a process parameter. For the purposes of this exam- the product of interest will bind, the lower residence ple, only the lot-to-lot variation will be considered. The times are unlikely to affect impurity clearance. Therefore, effect on cell growth was rated as 4. The CQAs were each the effect on CQAs was rated at 1 (no impact). rated 4 due to uncertainty of effect of PEI quality on each attribute. The maximum rating for PEI quality effect was Example 2: Total Load on CIM Q Column 7, indicating influence on productivity or uncertainty of The CIM Q step is designed to enrich the fraction of full significant effect on CQAs. The sum of the rating was 23, particles. The exact mechanism of separation is unclear. reflecting uncertainty of effect on CQAs. However, the empty particles are relatively weakly bound and elute at a slightly lower ionic strength. In general, the DOWNSTREAM EXAMPLES chromatographic resolution is strongly dependent on the To provide more direction on how to perform the risk capacity of the column. Therefore, the effect of column assessments, two assessment examples for downstream load was rated at 10. Given the limited capacity of the process parameters and one raw material assessment are column, the higher end of the proposed range is also

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 81 Figure 4-5. Example of Outcome Using Pareto Chart

Max Sum Parameters Materials Total 10 NA 4 0 High Risk 13 Not 10 >40 5 4 Medium Risk Not 10 24-40 11 1 12 Low Risk Not 10 <24 229 7 233 1 6 11 11 71 21 51 31 81 41 16 61 91 76 26 56 36 86 66 46 96 171 211 121 151 131 141 181 116 191 161 101 221 176 241 251 231 126 216 156 186 136 146 166 196 201 106 226 246 256 236 206 Count of Parameters Assessed

Max of Rankings Sum of Rankings High Risk Criterion Medium Risk Criterion

Figure 5 demonstrates an example of the risk analysis outcome. The X-axis contains each parameter that is assessed. Any parameter with a maximum of rankings of 10 is automatically considered a high-risk pa- rameter and may require further study. Parameters with a sum of rankings >40, but no maximum ranking of 10, are also considered high-risk. Medium-risk parameters are those with a sum of rankings of 24 to 40, whereas low-risk parameters are those with a sum of rankings <24. As shown in the example, such an analysis allows the ranking of ~260 parameters into high-, medium-, and low-risk categories. Rather than performing analytics on 260 parameters, focus can be applied to the 13 high-risk and 12 medium-risk parameters, thus conserving time and resources while still producing a consistent, high-quality product.

expected to lead to yield loss. Consequently, the effect commercial processes. Developing SDMs requires con- on yield was rated at 4. sideration of scale-dependent effects.22 SDMs are associated with several challenges. For Example 3: Raw Material – Pluronic F68 example, CQAs that are affected by upstream process Pluronic F68 is used in the formulation buffer at a parameters cannot be measured directly with upstream concentration of 10 ppm. Pluronic is a known stabiliz- materials due to the high level of impurities found within ing agent. There is conflicting literature on the use of the product. Products often require sample treatment Pluronic F68 to prevent aggregation. Therefore, a rating with downstream purification steps, which introduces of 4 is assigned. Because this is a raw material risk assess- variability to the data analysis. Further, the downstream ment and not a PRA, the yield rating is left at 1. process is tied to the upstream process for a consistent supply of materials. Small-scale unit operations may not Scale-down Models represent the actual manufacturing process scale.23 Therefore, SDMs require rational design in order to In order to conduct complex process characterization produce data that are sufficiently predictive of and rele- studies, key areas can be scaled down. However, such a vant to full-scale manufacturing (i.e., size of bioreactor scale-down model (SDM) must represent the proposed or column). Sometimes, there is no perfect scale-down

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 82 Table 4-9. Design of SDMQ Examples.

Unit operation Purpose SDM At-scale PI or CQA selection for comparison

Seed train/Shake Biomass NA 125 mL to 2 L • NA flasks accumulation • The same or modified seed train • Widely accepted that shake flask sizes can be interchangeable

Production stage Product 3 L 500 L • Cell growth • Titer • Product-related impurities • Identity

Affinity Yield 1-cm ID 20-cm ID • HCP • GC titer • Other upstream related impurities that can be cleared out at affinity step

model by design (i.e., continuous centrifugation for performance with that of the large-scale process. This is biologics at manufacturing stage vs bench-scale filter), generally done using a statistical approach that is selected but a study can be designed to determine the equivalence based on the availability of the data and the number of of material quality after that step. Furthermore, it is im- variables being evaluated in the model. The goal of the portant to understand the degree of differences between qualification is to demonstrate equivalency of the SDM the model and the commercial process because this can to the manufacturing scale. Each unit operation SDM impact the relevance of information derived from scale- down models.24 Examples of SDM at each unit operation are provided in Table 9. Figure 4-6. Overview of Planning, Execution, In general, it is recommended to have a minimum and Report of three runs at manufacturing scale, although more is better to enable a stronger analysis. If the model requires • Input: parameters from PRA for each in-process samples to assay specific criteria such as titer, unit ops full capsid, or other specific CQAs, then an in-process Planning • Design (ie DOE, OFAT, RSM, excursion) sampling plan should be provided to manufacturing • Process Characterization Protocol prior to running the process in order to maximize the collection of information from each batch. In some cases, there may not be enough manufactur- • Individual PC (may depend on previous ing-scale batches to provide statistically significant data PC results) for SDM generation. A risk-based decision may be made Execution • Linkage study (Cross-functional to move forward with an SDM at risk to allow process collaboration) • Analytical testing characterization studies to be conducted. The justification of the decisions based on characterization studies using a model that has not been fully qualified will need to be docu- • Parameter classification (KPP, CPP) mented. In some instances, studies may need to be repeated • Proven acceptance range (PAR) with a modified model if further manufacturing scale batch Report • Updated parameter risk assessment data are collected that demonstrate that the SDM did not (PRA) suitably represent the manufacturing-scale process. An SDM is qualified by demonstrating its equivalent

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 83 will have a plan predefining the criteria for qualifying Process Characterization the SDM. When nonequivalency is observed for a particular Process characterization is a key part of the QbD process, data point, justification for the nonequivalency is in- as well as a key portion of the regulatory file. Process vestigated first to identify potential root cause(s) before characterization seeks to not only document but also establishing that the data point truly is not comparable understand the impact of predetermined, deliberate (i.e., too small of a sample size, too large/small analyt- variations in the process parameters and raw material ical variation). If the incomparability is confirmed, the attributes as a first step. Process characterization results understanding of the offset either from theory or experi- in the identification of potential sources of variability in ment is needed to interpret the correlation between input product quality. In turn, this allows the determination parameters versus output (process performance indica- of how to control such sources of variability. When tors and CQAs). Unjustifiable offset or nonequivalence performing process characterization, the classification will lead to failure of SDMQ. Failure of a qualification and level of control over process parameters should of an SDM will require investigation as to the cause of consider the corresponding risk to product quality.25 the failure to determine whether it is due to execution Essentially, process characterization provides insight issues or that the model truly does not provide sufficient into which parameters are critical to both attain and relevance to the manufacturing scale. The appropriate maintain product quality and process performance. process development leadership will develop a strategy Process characterization is repeated and modified as the for how to manage the situation. results of process knowledge from multiple iterations, preclinical data, and clinical data become available.26 In

Table 4-10. Design of Studies

Design Description Example

OFAT Main effect only • Upstream: Harvest timing • Downstream: Affinity wash buffer volume

Screening Main effect with some interaction • Upstream: pH, Temp, DO, Seeding Density Low resolution • Downstream: 6 factors in AEX step (i.e., loading CV, loading pH, conductivity, elution pH, elution pH and elution CV)

RSM Main effect, interactions, and • Upstream: ≤3 factors from above row (Response Surface quadratic terms • Downstream: 3 factors in AEX step Model) High resolution

DSD (Definitive Main effect and interactions; note • Upstream: initial optimization with 6 factors Screening Design) that DSD is not appropriate in all situations

Excursion Short time excursion • Upstream: DO or temperature excursion

Linkage Study Between unit operations • Upstream: N-1 and N stage • Downstream/Upstream: Harvest

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 84 gene therapy approaches, characterization approaches for Example 1: OFAT Study plasmid products should take a minimalistic approach. Sometimes, OFAT can be used to define the study range, Starting materials, vector, and plasmid process especially when there is a wide range to study. For ex- characterizationChapter 4: Process attempt Developmentto understand how Using risk Qualitypro- ample,by Design instead (QbD) of having Principles many levels per factor prior files differ through a series of experiments that allow to DOE, OFAT can be used to define a narrow range. the definition of the protocol by which the process is This helps to reduce levels in the subsequent DOE. For qualified. Some parameters are critical by default due example pH ± 0.4 could potentially have up to five levels: toExample 2: Screening Study compliance requirements. Experiments may include -0.4, -0.2, set point, +0.2, and +0.4. These five levels in one-factor-at-a-timeScreening studies (OFAT) are typicallyand design used of experiments when multiple later factors DOE would must either be screened lead to lower for mainresolution effec (i.e.,ts and (DOE)partial approaches, interactions. with DOE The movingexample from shows screen 6DOE factors screening) with 2 levels or too eachmany runsfor testing. to study Thefor the outcome main effect from approachesthe screening to response study surface is usually DOEs (Tablewhether 10). there is aand main interactions. effect from Confirming the factor the tested.ranges to Using have onlyDOE, minimal runs could be tested (ie, 2 x 6 =12 runs for each OFAT, but onlytwo 8levels runs (withinare needed a range in the FigureFigure 4-7. 7 forExample the main of Settings effects andonly). Output Another for Screeningoption is 2Studies-factor interactionthat would in the not design impact iftiter there or is interest and resources are not limited. quality) and a center point is more effective for later studies. Another scenario could be one with no potential interac- tions with other factors and likely no need to study inter- actions by DOE. An example of an OFAT study would be investigating the effect on via- bility and viable cell density by varying the media hold/media storage time for a thawed cell bank vial passage step prior to transferring to shake flask production. In addition, har- vest time is typically tested at control conditions so that no extra runs are needed for different harvest timing.

Example 2: Screening Study Screening studies are typically used when multiple factors must be screened for main effects and partial interactions. The example shows six factors with two levels each for testing. The outcome from the screening study is usually whether there is a main effect from the factor tested. Using DOE, minimal runs could be tested (i.e., 2 x 6 Figure 7. Example of Settings and Output for Screening Studies.

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 85 Example 3: Response Surface Methodology (RSM) RSM provides high resolution of both main effects and interactions between factors. The example below shows 3 factors, each with 2 levels. Central composite design (CCD) was used without hybrid points. CCD in a DOE design allows the determination of the settings of factors that would result in the optimum response. This experimental design builds a second-order model for the response variable without needing to complete a three-level factorial experiment. In the example here, a rotatable axial point (with 1.68 by default) was selected to test the extreme condition for 1 factor while keeping the rest of factors at target level. CCD can test main effects and interactions and also reflect any quadratic effect for factors. In addition, CCD can be used to test more levels Chapter 4: Process Development Using Quality by Design (QbD) Principles

(ie, axial points for extreme conditions) or to build a model for prediction. However, a drawback for 3 factors, as in this example, is that more runs are required (ie, 16). Therefore, CCD would be =12 runs for each OFAT, but only eight runs are needed in recommendedFigure 4-8. Response after Surface first confirming Methodology the impact for quality or titer from factors. the Figure 7 for the main effects only). Another option is two-factor interaction in the design if there is interest and resources are not limited. Factorial points

Example 3: Response Surface Methodology (RSM) Axial points

RSM provides high resolution of both main effects Center points and interactions between factors. Figure 8 shows three factors, each with two levels. Central composite Hybrid points design (CCD) was used without hybrid points. CCD in a DOE design allows the determination of the settings of factors that would result in the optimum response. This experimental design builds a second-order model for the response variable without needing to complete a three-level factorial experiment. In the example here, a Figure 8. Illustration of Response Surface Methodology. rotatable axial point (with 1.68 by default) was selected (and thus have three levels), which allows a curve rather to test the extreme condition for 1 factor while keeping than a straight line for each continuous factor.27, 28 the rest of factors at target level. CCD can test main DSDs are not appropriate in some circumstances. effects and interactions and also reflect any quadratic Although DSDs are efficient, in systems with many sig- effect for factors. In addition, CCD can be used to test nificant factors and interactions DSDs can only screen more levels (i.e., axial points for extreme conditions) or for main effects. Further, DSDs should not be used when to build a model for prediction. However, a drawback there are constraints on the design region because an for three factors, as in this example, is that more runs implicit assumption behind the use of DSDs is that it are required (i.e., 16). Therefore, CCD would be rec- is possible to set up levels of a factor independently of ommended after first confirming the impact for quality the level of any other factor; when this is not the case, or titer from factors. certain factor combinations are not feasible and thus cannot be evaluated in a DSD. In addition, DSD should Example 4: Definitive Screening Design (DSD) not be used when some of the factors are ingredients At times, it is necessary to do both screening and opti- within a mixture. For example, if the percentage of one mization simultaneously. One way to do this is through ingredient (e.g., media) is increased, the percentage of definitive screening design (DSD). DSD is a DOE another ingredient (e.g., additives) must decrease, so methodology first published in 2011 that uses correla- these factors cannot vary independently. Another sce- tion-optimized designs to screen several factors for both nario in which to avoid the use of DSD is when there main effects and interactions. DSD requires fewer runs are categorical factors with ≥2 levels. Although DSDs than similar fractional-factorial designs and allows for can run with a few categorical factors at two levels, a the unambiguous identification of the main effects and DSD with too many categorical factors with ≥2 levels is interactions. Thus, DSD is an efficient, one-step approach inefficient. DSDs run as split-plot designs also should to process characterization. be avoided. Lastly, DSDs should not be used when the a DSD is most appropriately used during the earliest priori model of interest has higher order effects because stages of experimentation when there is a large number cubic terms are confounded in DSDs.29,30 of factors that are potentially important and which may affect a response of interest. In particular, DSD is best Output for PC: Proven Acceptance Ranges and used when the goal is to identify a much smaller number Normal Operating Ranges of factors that are highly influential. DSDs are best suited After completion of the PC study and analysis of data, the for situations in which most of the factors are continuous results are utilized to establish process control limits and

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 86 Figure 9. Response Surface Methodology Responses.

Figure 4-9. Response Surface Methodology Responses

ranges and to understand criticality. Normal operating Some NORs are carried through from the beginning range (NOR), in which the process is allowed to vary of development with no changes, whereas others are around a set point target value with no negative effect on adjusted and honed to improve productivity or quality the process, occurs for all parameters in a process. NORs during process development. NORs are not used to should be wider than the ability to control the parameter define design space but can be adjusted based on char- at target. For example, the bioreactor agitation NOR may acterization studies. NORs can be widened if there is be 180 rpm to 220 rpm, but the system can control this a need to provide more range around the ability of the parameter within 10% of the target. Thus, NORs are equipment to control the parameter at the set point and considered ranges that are practically achievable. Most there is no negative effect on the process. Alternatively, two-sided NORs have the target point within one or two NORs can be narrowed if there is need to more tightly equipment capacities. Once additional statistical analysis control the parameter to prevent failure, for example is available following characterization, statistical model- if the characterization study identified a NOR was too ing may be used to determine the NOR. close to the failure point. In this case, the range may be

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 87 FigureChapter 4-10. 4: Process Example Development of Response SurfaceUsing Quality Methodology by Design Output (QbD) Principles

Figure 10. Example of Response Surface Methodology Output. narrowed if it can be practically controlled. Alternatively, the parameters should be included in a Design Space. theExample 4: Definitive Screening Design (DSD) entire range and target may shift away from the fail- Alternatively, a PAR can be defined for only one of the ureAt times, point if it there is necessary is no risk ofto moving do both close screening to another and optimizationparameters in simultaneously.the process description, One and way other to do process this failureis through point. definitive For example, screening the affinity design column (DSD). residence DSD parameters is a DOE will methodology be limited to firsttarget published operating conditionin 2011 timethat characterization uses correlation study-optimized using a range designs of 1.5 to to 5.5 screen or NOR. several factors for both main effects and mininteractions. demonstrated DSD that requires 1.5 min was fewer too short runs and than 5.5 hadsimilar fractionalFor some -parameters,factorial designs the characterization and allows range for and/the nounambiguous negative effect identification on the process of and the CQAs. main Operatingeffects andor interactions.knowledge range Thus, may beDSD outside is anof the efficient, PAR. This one can- withstep aapproach NOR of 2 to 4process min with characterization. a target of 3 min may be due to findings in development or if the characteriza- be considered too close to the failure point. It could be tion studies identified failure points in the range. Not all decidedDSD is to most narrow appropriately the range to 2.5 used to 4 minduring or shift the theearliest parameters stages willof experimentationhave a knowledge range when and there not all is will a targetlarge tonumber 3.5 and of shift factors the rangethat areto 2.5potentially to 4.5 min. important For have and a PAR. which Figure may 12 affect shows athe response nesting that of may interest. occur eachIn particular, parameter, DSD the NOR is best will used need whento be determined the goal is toif identify a parameter a much has all smaller three levels. number of factors that followingare highly the PC. influential. DSDs are best suited for situations in which most of the factors are continuousA PAR allows (and deliberate thus have change 3 levels), in one parameter which allows a curve rather than a straight line for each 27,28 Parameter Criticality Assessment withoutcontinuous changing factor. the others outside their NOR/tar-

get. PARs may be presented in the description of the Parameter criticality assessment is done to assess the

manufacturing process of the drug substance as ranges. overall criticality of parameters (e.g., as key or non-key).

PARs for single parameters are proposed in the licensing Following PC, every process parameter and material is

application and are subject to regulatory assessment and classified based on the effect it has on the CQAs and the approval. The PAR should be adequately justified regard- process performance (PIs). This is done quantitatively less of whether the process parameter is considered a for parameters studied in PC and qualitatively for the critical process parameter or not (ICH Q8 R2). remaining parameters. There are 3 levels of classification Where interaction effects between different param- used in a common process. The classification of the pa- eters exist and the acceptable range for one process rameters is used to build the P/ACS (process/analytical parameter depends on the setting of another parameter, control strategy).

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 88 Figure 4-11. Example of DSD output

Factor Factor Factor Factor Factor Factor Factor Factor Factor Factor 1 2 3 4 5 6 7 8 9 10 1 13 0 1 1 1 1 1 1 1 1 1 2 16 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 3 21 1 0 -1 -1 -1 -1 1 1 1 1 4 17 -1 0 1 1 1 1 -1 -1 -1 -1 5 8 1 -1 0 -1 1 1 -1 -1 1 1 6 5 -1 1 0 1 -1 -1 1 1 -1 -1 7 7 1 -1 -1 0 1 1 1 1 -1 -1 8 4 -1 1 1 0 -1 -1 -1 -1 1 1 9 15 1 -1 1 1 0 -1 -1 1 -1 1 10 10 -1 1 -1 -1 0 1 1 -1 1 -1 11 6 1 -1 1 1 -1 0 1 -1 1 -1 12 14 -1 1 -1 -1 1 0 -1 1 -1 1 13 18 1 1 -1 1 -1 1 0 -1 -1 1 14 2 -1 -1 1 -1 1 -1 0 1 1 -1 15 9 1 1 -1 1 1 -1 -1 0 1 -1 16 12 -1 -1 1 -1 -1 1 1 0 -1 1 17 20 1 1 1 -1 -1 1 -1 1 0 -1 18 11 -1 -1 -1 1 1 -1 1 -1 0 1 19 3 1 1 1 -1 1 -1 1 -1 -1 0 20 1 -1 -1 -1 1 -1 1 -1 1 1 0 21 19 0 0 0 0 0 0 0 0 0 0 22 22 0 0 0 0 0 0 0 0 0 0 23 23 0 0 0 0 0 0 0 0 0 0 24 24 0 0 0 0 0 0 0 0 0 0 Best -1 0 0 0 1 0 -1 -0.4 0 0 Worst 1 0 0 0 0 0 1 0 0 0

A process parameter is an input variable of the man- Figure 13 and Figure 14. Figure 13 covers parameter ufacturing process that can be directly controlled such classification for parameters that were characterized and as temperature, time, pH, flow rate, etc. Based on their which have quantitative data supporting decision-mak- impact to process performance and product quality, ing. Figure 14 covers parameters that fell below the water process parameters are divided into the following three line and were not selected for characterization; therefore, classes (definitions of which can be found in Chapter 3): classification is made qualitatively based on scientific critical process parameter (CPP), key process parameter knowledge, literature, or previous experience. (KPP), and non-critical process parameter (non-CPP). For a parameter that was characterized, and data Two classification decision trees are provided in reflects the potential to affect a CQA with both statistical

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 89 Figure 4-12. Nesting of Ranges Around the Target

Knowledge Range

Proven Acceptable Range

Normal Operating Range

Parameter Scale Target and practical significance, it is deemed a CPP. Statistical Parameter Classification Assessment significance relates to whether an effect exists. Practical significance refers to the magnitude of the effect. A CPP (PCA) or FMEA will have a specified range in which the operation must Failure mode and effects analysis (FMEA) is a step-by- be maintained. Falling outside of the range will result in step approach to identify all possible failures in the man- an investigation and will likely lead to a rejected batch ufacturing process. In FMEA, failures are categorized due to potential effect on a CQA. based on how serious the consequences of the failure If the effect was statistically significant but is not are, how frequently the failure occurs, and how easily practically significant (relating to efficacy or safety), they can be detected. Ultimately, the goal of FMEA is or if there was no statistical effect on a CQA, then it to eliminate or reduce failures starting with those that will be considered further. If the parameter affected have the greatest effect on the process. The use of FMEA the process performance, then it will be classified as in QbD processes allows the documentation of current a KPP. A KPP will have a control range, and falling knowledge and actions about the risk of failures to be outside of the control range will trigger a quality action used in continuous process improvement. FMEA should to determine the outcome of the batch. If the parameter begin during the earliest stages of process design and does not affect CQAs or process performance, it is continue throughout the entire process. classified as a non-key parameter. Non-key parameters For gene therapy or biologics, prior to the genera- have control ranges and are monitored in continued tion of the P/ACS (process/analytical control strategy) process verification. For all other parameters that were for PPQ, a process control FMEA will be conducted by not selected for PC, a science-based decision is made to development and manufacturing teams to ensure any classify the parameters as CPP, KPP or non-key. Each failures with potential impact to product quality have a classification of a parameter will require documenting documented mitigation and control strategy. the rationale for the classification assignment, as this Process parameter classification is governed by FMEA will be part of the licensing application. For CPPs, the and is also assessed by results of the process parameter CQAs that are affected will need to be identified. For risk assessment and using data from completed PC KPPs, the effect on the performance and the action studies. The purpose of process parameter classification limits will need to be defined. The P/ACS will be used is to use the results of these completed PC studies to to define the controls used to reduce risk to the quality classify process parameters and input material attributes and robustness of the process. based on their likelihood of having an impact on process

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 90 Figure 4-13. Decision Tree for Parameter Classification for Characterized Parameters Based on Quantitative Data

CQA Statistically yes Practically yes PC Significant Effect Significant? CPP on CQA?

Affects yes Performance? KPP (PI)

Non- Key performance or product quality. Criticality is assigned have potential for severe effect on safety and efficacy, may based on the impact to drug substance and drug product result in a serious adverse event, or may result in loss of quality, by establishing a link between CPPs and CQAs. product and adversely affect subsequent unit operations. Each process parameter is classified as either a CPP or In the highest occurrence scores, the excursion has been non-CPP based on its potential impact on quality. A pro- documented frequently with the historical platform and cess parameter is classified as a KPP based on its potential potential for excursion is generally expected to be signifi- impact to critical process performance attributes. This clas- cant in a high percentage of runs. In the highest detection sification is used to develop a P/ACS that ensures that CPPs scores, no known controls are available to detect the are adequately monitored and/or controlled. In addition, particular failure mode and as such excursion may occur the FMEA should be used to designate certain materials without real-time knowledge. Based on individual SOD as critical raw materials and to develop an appropriate scores for each process parameter, a risk priority number strategy to ensure that raw materials, starting materials, (RPN) is calculated, which is essentially the product of reagents, solvents, intermediates, and process components individual parameter SOD scores. do not have a negative impact on product quality. Whereas the initial SOD stage of the FMEA defines The process control FMEA builds upon the risk as- high-risk parameters, the second stage of the process sessment performed during the initial PRA phase and control FMEA assesses multiple aspects of the manu- begins with a formal risk assessment of each process facturing process (equipment capacity, normal operat- parameter based upon the “SOD” model of severity ing ranges, process characterization range, parameter (potential magnitude of impact of parameter excursion setpoint, potential failure modes/causes) along with to the quality of the product), occurrence (predicted like- various controls in place to prevent failure (engineering lihood or frequency of an excursion, based on historical controls, process controls, facility controls, personnel process knowledge), and detection (ability of the manu- controls, testing, PAT, etc.). The purpose of the second facturing team to observe the excursion and take steps to stage of the FMEA is to compile all potential for risk mitigate). In the highest severity scores, excursion may mitigation on a parameter-specific basis with the intent of

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 91 Figure 4-14. Decision Tree for Parameter Classification for Uncharacterized Parameters Based on Science-Based Knowledge

Early Development/ Platform Knowledge yes Non-PC Affects CQA? CPP Risk Assess/ Lit. Review no

Affects yes Performance? KPP

modifying the RPN to reflect a mitigation-adjusted RPN. to transfection, elution volume of the affinity column, As such, knowledge and experience compiled during PC step-yields at key unit operations, and IPC. IPC, process and prior manufacturing runs can be applied to some of monitoring, material control, product control through the higher-risk parameters to modify the criticality based quality testing, and stability are all parts of the control on process control. The end result of the process control strategy, which will be discussed thoroughly in Chapter FMEA is a shorter list of parameters that are deemed critical. 6. In this chapter, we will focus on points that are gene therapy–specific through the next few examples. Control Strategy for Critical Process POTENCY Parameters The FDA published guidance for potency testing of gene In addition to establishing parameter control ranges as therapy in 2011. Potency is defined as “the specific ability the previous section discussed, certain steps within a or capacity of the product, as indicated by appropriate process may require additional controls to ensure that laboratory tests or by adequately controlled clinical data the process is performing as expected. In these instanc- obtained through the administration of the product in es, in-process controls (IPC) are established. These are the manner intended, to effect a given result.” Regulations determined by the process development team using require potency testing through in vitro and/or in vivo the combination of process development, scale-up, and tests that have been specifically designed for each prod- characterization study data to provide expected ranges uct. Potency measurements are used to demonstrate at specific steps. Some examples of IPC that may be filed that product lots meet the predefined specifications or in the licensing application as part of the control strat- acceptance criteria not only during all phases of develop- egy include: viability and viable cell density of the cells ment, but also following market approval. Often, a single following transfer of the cell bank vial contents to the biological or analytical assay will not provide sufficient shake flask, viable cell density and cultivation time prior measurement of potency, so multiple complementary

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 92 assays, referred to as an assay matrix, can be developed such as viable cell concentration (VCC) and nutrient that measure potency through measures of quality, and metabolite concentrations that may impact product consistency, and stability. An assay matrix may include quality attributes.36 assays that provide both quantitative readouts (e.g., units PAT uses a risk assessment template for parameter of activity) and qualitative readouts (e.g., pass/fail).31 criticality to mitigate some of the risk. The PAT used affects the number of critical process parameters in VIRAL CLEARANCE the process. Relative to monoclonal antibodies, viral Some AAV platforms require the addition of a virus, such products are notoriously uncharacterized, but some as adenovirus. In the downstream process design, virus considerations may include monitoring particle size of inactivation (i.e., heat, acid) and viral clearance need to transfection complexes (lentiviral vectors and AAV vec- be in place before filing. Extensive viral clearance using tors), and cell size/growth and metabolites (i.e., Raman, another model virus is also required to demonstrate the capacitance, microscopic based instruments). process capability. Normally, viral clearance is a prod- Whereas much monitoring has been “retrospective” in uct-specific practice and required for the production of the past, it is important to obtain real-time information every product. For additional information, please refer to to inform the process strategy and control using PAT. Chapter 5 on Upstream/Downstream Processing. For example, bioreactors typically have probes that allow for the monitoring of the culture environment, such as HCDNA physicochemical factors (e.g., temperature, pH, dissolved AAV can package a large amount of nonvector DNA oxygen); however, it is more difficult to measure other (e.g., plasmid DNA, helper virus sequences, host DNA), cell culture parameters, such as glucose/lactate concen- and it may be challenging to remove this DNA from the tration, cell density, or cell population characterization. product to the agency-requested level to ensure safety.32 It is vital that characteristics such as these be monitored Therefore, cell lines and helper vectors must be designed to ensure efficiency and safety of the product and reduce and selected carefully to reduce product risks. Quality batch-to-batch variability. Online monitoring technology data, risk assessments, and/or details of their process, and is important for automatic feedback control of the cul- product control strategies should be in place to address ture, increasing knowledge of the process and facilitating and mitigate potential risks using the selected system. QbD approaches. The following sections discuss online monitoring technology that can be used to improve batch Process Analytical Technologies (PAT) consistency and efficiency.37 Compared to the process for monoclonal antibodies, Another way of ensuring that quality is built into the unique considerations exist for PAT in gene therapy. process is to mitigate some key process risks by utilizing process analytical technology (PAT). PAT is a framework CELL GROWTH/MORPHOLOGY MONITORING: used to design, analyze, and control “manufacturing Even for capacitance probe or Raman, which is com- through timely measurements (i.e., during processing) monly used for CHO cell and monoclonal antibody of critical quality and performance attributes of raw production, multiple platforms are used within the and in-process materials and processes, with the goal same company for gene therapy. Therefore, tools may be of ensuring final product quality.33,34 Through the use either process- or platform-specific (i.e., HEK vs Sf9). of PAT, better process control is gained by identifying In addition, cells will experience either transfection or and managing the sources of variability throughout the infection along the process, and morphology or growth process and proactive decision-making throughout the could be impacted for that unique step, which introduces process. It also results in reduced cost due to optimized additional challenges. On the other hand, if the changes use of raw materials and minimization of product cycle can be captured, the infection/transfection step can also times.35 Ultimately, the main objective of PAT is to mon- be monitored. itor in real-time the values of some process parameters,

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 93 TRANSFECTION/INFECTION MONITORING: process development and better mitigation of process The majority of biologics production utilizes stable cell risks compared to older technology. Thus, capacitance lines. Although some protein production processes also sensors likely represent a viable method for monitoring use transient transfection, normally only a single plas- VCC in gene therapy applications, but it is not without mid is involved. In contrast, for AAV production, either unique challenges, such as limitations of measurement multiple plasmid transfections or virus infections are during the stationary growth phase and death phase due included in the upstream step. This is a critical step that to cell diameter changes of apoptotic cells.39,40 requires careful monitoring. For transfection specifically, transfection efficiency may be monitored through prod- DIFFERENTIAL DIGITAL HOLOGRAPHIC uct production or cell morphology changes or via the MICROSCOPY (DDHM) transfection complex (the size of the complex is related Few methods are available to monitor viral particle to the transfection efficiency).38 Some DLS methods production during cell culture. Existing processes use mentioned below could be used to monitor the kinetic chemometrics approaches by measuring process vari- change for transfection complex size. ables related to viral production kinetics or changes in cell morphology or physiology. Ideally, these processes PRODUCT MONITORING: should be monitored via label-free methodologies to For downstream and formulation steps, most of the avoid the addition of compounds that may influence protein monitoring system can be used to track AAV. cellular behavior. While most label-free cell culture However, the product is a protein capsid with DNA in- monitoring methods use spectroscopic techniques, side. Tracking both protein and DNA can reflect not only image-based cell monitoring is also an option. Because the yield but also quality of the product (i.e., whether cells are mostly transparent, systems must be in place to empty capsid has been removed). generate the needed image contrast. One such method is digital holographic microscopy. A study published CAPACITANCE PROBE in 2020 found that DDHM was successfully used to VCC is a key performance indicator during upstream monitor cell concentration and viability and also assess technologies. VCC is often measured through offline AAV production kinetics in an insect cell system. While methods (e.g., staining dead cells with Trypan Blue and most attributes that are calculated via DDHM have no counting cells microscopically). Online monitoring of biologic meaning per se, they can be used collectively biomass remains challenging due to complex calibration to characterize a dynamic phenotype that is indica- and integration of analytics. However, one such online tive of cell adaptation to various biological situations. technique is radio frequency impedance, which can However, some attributes calculated with DDHM may be measured in the cell broth via capacitance probes be more directly relevant, such as phase correlation (e.g., to monitor cell concentration online. This principle time-specific characteristics that are similar to the culture is based on the polarization of the cells by applying a viability profiles). Even more so, attributes more directly periodic electric field to the system. Only viable cells related to viable cell concentration rely on light intensi- are polarized and thus can be evaluated to correlate ty (due to light dispersion caused by suspension cells, VCC. In the past, capacitance sensors have been used to which is analogous to turbidity-based measurements). monitor growth and infectious status. A study published In AAV in particular, DDHM appears to create “phase in 2020 found that capacitance sensors are also able to skewness” (which refers to a lack of symmetry for the successfully facilitate the scale-up of bioreactor processes phase histogram of the cell) due to the molecular density from 50 L up to 2000 L. Using such an approach can of certain organelles (e.g., nucleus, nucleolus) compared help to preserve resources and reduce failures by keep- to surrounding regions, and the location of AAV capsid ing the batch within the approved trajectory for VCC. assembly (nucleolus). DDHM appears to be a valuable In addition, capacitance probe technology allows faster tool to support online monitoring to determine time of

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 94 Table 4-11. Summary of Spectroscopic Techniques43

Method Measured Attributes Application Reference Notes

Near-infrared Vibrational Identification of Rüdt (2017)44 Low sensitivity and overtones of analytes selectivity peptide backbone

Raman Vibrational; peptide Glucose, glutamine, Rüdt (2017)45 Generally low spectroscopy backbone glutamate, lactate, sensitivity but high and ammonium selectivity concentrations; VCC; product concentration Fluorescence Intracellular rAAV production; Pais (2019)46 and Broad measurement spectroscopy fluorophores aggregation; amino Pais (2020)47 ranges possible; acid concentration difficult calibration; low-cost; high selectivity Dynamic light Diffusion behavior of Protein folding Rüdt (2017)48 Based on time scattering (DLS) macromolecules Particle size correlation

Dielectric Dielectric potential VCC, cell biovolume Pais (2019)49 and Based on changes spectroscopy of cells in an Pais (2020)50 in cell physiology; alternating electrical continuous field monitoring; high sensitivity Adapted from Rüdt M, et al. J Chromatogr A. 2017;1490:2-9.

harvest and to establish controlled feeding strategies, and information about cell phenotype.51 Markers, which may allows a simpler workflow with real-time monitoring include glucose, glutamine, glutamate, lactate, ammoni- compared to other methods.41 um, VCC, and product concentrations, among others, can be evaluated in real-time with Raman spectroscopy to SPECTROSCOPY TECHNOLOGY inform adaptive manufacturing and decision-making by Glucose or lactate concentrations can be measured online providing immediate feedback on process performance.52 in bioreactor cultures via spectroscopic analysis, such as However, the success for Raman models depends on Raman spectroscopy, near infrared (NIR) spectroscopy, calibration techniques. Traditionally, Raman spectro or fluorescence techniques.42 An overview of spectrosco- scopy calibration generates highly specific models, but py techniques is shown in Table 11. these models are only reliable in the exact conditions in Raman spectroscopy is a technique that can be used which they are calibrated. Thus, it is expected that model to observe molecular vibrations to identify and quantify performance would degrade over time due to changes molecules by measuring changes in the wavelength of in the process (recipe changes, raw material variability, laser light to identify which molecules are present within process drifts). To circumvent this issue, a real-time the cell culture media. It can be used to noninvasively just-in-time learning (RT-JITL) framework or other measure time-dependent molecular properties of cells methods can be used to automatically calibrate, assess, (without labels) during bioreactor growth by providing and maintain Raman models. The RT-JITL framework

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 95 allows the calibration of generic models that can be used VCC and recombinant protein titers in mammalian cell in cell culture experiments with various conditions. The culture systems, as well as CQAs such as aggregation. A use of generic calibration models allows the real-time recent study found that in situ fluorescence spectroscopy prediction of cell culture performance parameters with- with recombinant AAV can predict relevant process out jeopardizing the calibration component of this highly variables, such as viability and product titer, to enable important process.53 PAT. However, complications associated with fluores- NIR spectroscopy is an online, time-efficient, nonin- cence spectroscopy in gene therapy–related applications vasive technique that measures the interaction of near-in- include difficult interpretation of results due to the rAAV frared light with a sample to facilitate the identification production profile, which increases for some time before of analytes, especially glucose, within the bioreactor. NIR decreasing and then plateauing throughout the remain- spectroscopy can also facilitate the scale-up process by ing culture time. In addition, it may take more time for developing models. However, NIR use may be limited spectra acquisition than other methods. 57,58 for glucose concentration monitoring due to deviations For biologics, many techniques are already included in accuracy during some phases of cell culture, such as in GMP. Most of the techniques used in biologics may when feeds are added to the culture.54 also be used in AAV-specific technologies, although It is impossible for one sensor to measure all prod- some technologies are gene therapy–specific. For AAV, uct quality attributes during production. In fact, even especially for an SF9 system, multiple PAT tools apply, but one attribute may require multiple sensors. Therefore, most techniques are still in the early stage (at bench scale). multimodal spectroscopy may be needed. Multimodal spectroscopy may include various types of spectroscopy, Conclusion including UV spectroscopy and DLS. When using multi- ple sensors in a process stream, it is important to account In closing, as many gene therapy products proceed to- for dispersion between the detectors. Therefore, accurate ward late-stage development and BLA-enabling activities, data analysis is necessary to extract correct conclusions.55 it is going to be increasingly important to take lessons DLS has been used for particle size and concentration for from biologics and vaccines in applying a systematic, AAV gene product. The potential application for AAV risk-based approach for late-stage development. The production would be product formulation and purifica- individual PRAs described in this chapter allow a large tion as well as to monitor the transfection mix. pool of hundreds of parameters to be whittled down to Fluorescence spectroscopy uses electronic and vi- a smaller number of parameters deemed important for brational states and is based on the excitation of species further characterization during late-stage development. from the ground electronic state to a vibrational state This allows a greater amount of focus on aspects of in its excited electronic state. It can be used to monitor the process that present the highest amount of risk to cell and product formation, as well as metabolite con- the overall quality of the product. Risk continues to be sumption and production in various biological systems. mitigated throughout the process by identifying critical For example, fluorescence spectroscopy can help to process parameters and using a combination of risk as- assess the concentrations of aromatic amino acids in the sessments and process analytical technology to ensure bioreactor (e.g., tryptophan) and has been proposed as that the commercial manufacturing process is robust and an in-line PAT tool for a chromatography purification reproducible. step of a fusion protein.56 It is also possible to monitor

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 96 CMC Readiness Checklist/Considerations For Expedited Pathways

Because the FDA does not have guidance in this area, a helpful list of questions to be con- sidered by manufacturers of cell and gene therapy products is found here.

 Have you performed a careful review of your  What is your plan for manufacturing of the manufacturing process to ensure that you are final drug product? Do you anticipate needing entering phase III trials with a product that is to make a change to your existing facility? Do optimal? you plan for automation, scale-out, or scale-up  Have you introduced major manufacturing post approval or prior to initiation of phase III changes that may require conducting compa- study? rability studies and if so, what is your plan for  Have you made a final determination of conducting such comparability studies? whether the current release specifications are  What is the status of your analytical method adequate for ensuring safety and potency of development? Have you qualified or preferably your final drug product? validated your assays prior to initiation of your  Have you conducted shipping validation for pivotal trial? source materials and the final drug product  Do you have appropriate potency assays in under worst-case scenarios or conditions of place for the final drug product? transport?  Do you have knowledge of CQAs, CPPs, and  Have you reviewed the quality of ancillary KPPs? materials as well as the reliability and sus- tainability of your supply chain, and do you  Have you determined the shelf life of the have a plan to review your quality agreements final drug product by conducting stability and SOPs that are in place for material assays using assays that are appropriate and qualification and vendor qualification? Have qualified/validated? you developed an identity test for your critical  Do you have a well-defined plan to collect ancillary materials? materials and reserve samples for in-process  Have you finalized your choice of the final and the final drug product? container and have a plan for how to affix the  What is your plan of action for conducting label on the final drug product? process validation to demonstrate that the  What is your plan for testing of the source final drug product can be successfully manu- material, in-process materials, or the final factured consistently? drug product? Do you plan to outsource your  Have you defined standard operating proce- testing, or will it be conducted in-house? dures (SOPs), protocols, and/or instructions  Do you need to develop any in-house stan- for use in outlining any additional manufactur- dards (physical or performance standards) for ing, processing, formulation, or thaw/dilution your assays? Do you know what standards are of the final drug product at clinical sites? needed for your product development and  Do you plan to gain a better understanding of release testing? the requirements for conducting leachable  Have you had an End of Phase 2 (EOP2) and extractable studies for materials that are meeting with the agency to assess your CMC in direct contact with your product? readiness?

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 97 Endnotes

1. Reference for information in this paragraph: Section 6.1 of 13. Food and Drug Administration. Guidance for industry. VV-05118 Process validation: general principles and practices. Food 2. Reference for information in this paragraph: Section 2 of VV- and Drug Administration website. https://www.fda.gov/ 05118 files/drugs/published/Process-Validation--General-Princi- ples-and-Practices.pdf. Published January 2011. Accessed 3. Reference for information in this paragraph: Section 2 of VV- May 4, 2020. 05118 14. Food and Drug Administration. Guidance for industry. 4. Reference for information in this paragraph: Section 2 of VV- Process validation: general principles and practices. Food 05118 and Drug Administration website. https://www.fda.gov/ 5. Wright JF. Gene Therapy. 2008;15:840-848. files/drugs/published/Process-Validation--General-Princi- 6. Food and Drug Administration. Guidance for industry. ples-and-Practices.pdf. Published January 2011. Accessed Process validation: general principles and practices. Food May 4, 2020. and Drug Administration website. https://www.fda.gov/ 15. Food and Drug Administration. Guidance for industry. files/drugs/published/Process-Validation--General-Princi- Process validation: general principles and practices. Food ples-and-Practices.pdf. Published January 2011. Accessed and Drug Administration website. https://www.fda.gov/ May 4, 2020. files/drugs/published/Process-Validation--General-Princi- 7. Food and Drug Administration. Guidance for industry. ples-and-Practices.pdf. Published January 2011. Accessed Process validation: general principles and practices. Food May 4, 2020. and Drug Administration website. https://www.fda.gov/ 16. Section 6.3 of VV-05118 files/drugs/published/Process-Validation--General-Princi- 17. Parenteral Drug Association. PDA Technical Report No. 60 ples-and-Practices.pdf. Published January 2011. Accessed (TR 60): Process Validation: A Lifecycle Approach. February May 4, 2020. 2013; Bethesda, MD: Parenteral Drug Association. 8. Food and Drug Administration. Guidance for industry. 18. Wright JF. Biomedicines 2014;2:80-97. Process validation: general principles and practices. Food and Drug Administration website. https://www.fda.gov/ 19. Section 7.6 of VV-05118 files/drugs/published/Process-Validation--General-Princi- 20. Section 7.7 of VV-05118 ples-and-Practices.pdf. Published January 2011. Accessed 21. Section 7.7 of VV-05118 May 4, 2020. 22. Section 8 of VV-05118 9. Food and Drug Administration. Guidance for industry. 23. Section 8 of VV-05118 Process validation: general principles and practices. Food 24. Section 8 of VV-05118 and Drug Administration website. https://www.fda.gov/ files/drugs/published/Process-Validation--General-Princi- 25. Section 6.2 of VV-05118 ples-and-Practices.pdf. Published January 2011. Accessed 26. Section 6.2 of VV-05118 May 4, 2020. 27. Jones B, Nachtsheim CJ. A class of three-level designs for 10. Food and Drug Administration. Guidance for industry. definitive screening in the presence of second-order effects. J Process validation: general principles and practices. Food Qualt Technol. 2011;43(1):1-15. and Drug Administration website. https://www.fda.gov/ 28. Tai M, Ly A, Leung I, Nayar G. Efficient high-throughput files/drugs/published/Process-Validation--General-Princi- biological process characterization: definitive screening ples-and-Practices.pdf. Published January 2011. Accessed design with the Ambr250 Bioreactor System. Biotechnol Prog. May 4, 2020. 2015;31(5):1388-1395. 11. Food and Drug Administration. Guidance for industry. 29. Jones B, Nachtsheim CJ. A class of three-level designs for Process validation: general principles and practices. Food definitive screening in the presence of second-order effects. J and Drug Administration website. https://www.fda.gov/ Qualt Technol. 2011;43(1):1-15. files/drugs/published/Process-Validation--General-Princi- 30. Tai M, Ly A, Leung I, Nayar G. Efficient high-throughput ples-and-Practices.pdf. Published January 2011. Accessed biological process characterization: definitive screening May 4, 2020. design with the Ambr250 Bioreactor System. Biotechnol Prog. 12. Food and Drug Administration. Guidance for industry. 2015;31(5):1388-1395. Process validation: general principles and practices. Food 31. US Department of Health and Human Services, Food and and Drug Administration website. https://www.fda.gov/ Drug Administration. Guidance for industry: potency tests files/drugs/published/Process-Validation--General-Princi- for cellular and gene therapy products. https://www.fda.gov/ ples-and-Practices.pdf. Published January 2011. Accessed media/79856/download. Published January 2011. Accessed May 4, 2020. June 26, 2020.

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 98 Endnotes, continued

32. US Department of Health and Human Services, Food and Drug 44. Rüdt M, Briskot T, Hubbuch J. Advances in downstream Administration. Guidance for industry: chemistry, manufactur- processing of biologics – spectroscopy: an emerging process ing, and control (CMC) information for human gene therapy analytical technology. J Chromatogr A. 2017;1490:2-9. investigational new drug applications (INDs). https://www. 45. Pais DAM, Portela RMC, Carrondo MJT, Isidro IA, Alves fda.gov/media/113760/download. Published January 2020. PM. Biotech Bioeng. 219;116:2803-2814. Accessed July 8, 2020. 46. Pais DAM. Development of advanced monitoring and control 33. US Department of Health and Human Services, Food and tools for rAAV production in the insect cell system [thesis]. Drug Administration. Guidance for industry: PAT – a frame- https://run.unl.pt/bitstream/10362/97879/1/Daniel%20 work for innovative pharmaceutical development, manufac- Pais%20PhD%20Thesis%20Final.pdf. Published April, 2020. turing, and quality assurance. US Food and Drug Adminis- 47. Rüdt M, Briskot T, Hubbuch J. Advances in downstream tration website. https://www.fda.gov/regulatory-information/ processing of biologics – spectroscopy: an emerging process search-fda-guidance-documents/pat-framework-innova- analytical technology. J Chromatogr A. 2017;1490:2-9. tive-pharmaceutical-development-manufacturing-and-quali- 48. Pais DAM, Portela RMC, Carrondo MJT, Isidro IA, Alves ty-assurance. Published 2004. Accessed June 26, 2020. PM. Biotech Bioeng. 219;116:2803-2814. 34. Baradez M, Biziato D, Hassan E, Marshall D. Application 49. Pais DAM. Development of advanced monitoring and control of Raman spectroscopy and univariate modeling as a process tools for rAAV production in the insect cell system [thesis]. analytical technology for cell therapy bioprocessing. Front https://run.unl.pt/bitstream/10362/97879/1/Daniel%20 Med. 2018;5:47. Pais%20PhD%20Thesis%20Final.pdf. Published April, 2020. 35. Marshall D, Ward S, Baradez M. Requirement for smart 50. Rodrigues CAV, Nogueira DES, Cabral JMS. Next-generation in-process control systems to deliver cell therapy processes fit stem cell expansion technologies. http://insights.bio/wp-con- for the 21st century. Cell and Gene Therapy Insights. http:// tent/uploads/sites/2/2018/11/0408_Bioprocessing_Spotlight. insights.bio/cell-and-gene-therapy-insights/wp-content/up- pdf. Published October 31, 2018. Accessed July 8, 2020. loads/2016/12/Marshall-et-al.pdf. Accessed June 26, 2020. 51. Baradez MO, Biziato D, HassanE, Marshall D. Application 36. Zavala-Ortiz DA, Ebel B, Li M, et al. Interest of locally of Raman spectroscopy and univariate modelling as a process weighted regression to overcome nonlinear effects during in analytical technology for cell therapy bioprocessing. Front situ NIR monitoring of CHO cell culture parameters and Med. 2018;5:47. antibody glycosylation Biotechnol Process. 2020;36:e2924. 52. Tulsyan A, Wang T, Schorner G, Khodabandehlou H, 37. Pezzoli D, Giupponi E, Mantovani D, Candiani G. Sci Rep. Coufal M, Undey C. Automatic real‐time calibration, 2017;7:44134. assessment, and maintenance of generic Raman models for 38. Joshi PRH, Cervera L, Ahmed I, et al. Achieving high-yield online monitoring of cell culture processes. Biotech Bioengin. production of functional AAV5 gene delivery vectors via 2019;117(2):406-411. fedbatch in an insect cell-one baculovirus system. Mol Ther 53. Koxma B, Salgó A, Gergely S. On-line glucose monitoring Methods Clin Dev. 2019;13:279-289. by near infrared spectroscopy during the scale up steps of 39. Metze S, Ruhl S, Greller G, Grimm C, Scholz J. Monitoring mammalian cell cultivation process development. Bioprocess online biomass with a capacitance sensor during scale-up of Biosystems Engineer. 2019;42:921-932. industrially relevant CHO cell culture fed-batch process in 54. Rolinger L, Rüdt M, Hubbuch J. A critical review of recent single-use bioreactors. Bioprocess Biosystems Engineering. trends, and a future perspective of optical spectroscopy as 2020;43:193-205. PAT in biopharmaceutical downstream processing. Analytic 40. Pais DAM, Galrão PRS, Kryzhanska A, Barbau J, Isidro IA, Bioanalytic Chem. 2020;312:2047-2064. Alves PM. Holographic imaging of insect cell cultures: online 55. Rüdt M, Briskot T, Hubbuch J. Advances in downstream non-invasive monitoring of adeno-associated virus produc- processing of biologics – spectroscopy: an emerging process tion and cell concentration. Processes. 2020;8:487. analytical technology. J Chromatogr A. 2017;1490:2-9. 41. Rodrigues CAV, Nogueira DES, Cabral JMS. Next-generation 56. Pais DAM, Portela RMC, Carrondo MJT, Isidro IA, Alves stem cell expansion technologies. http://insights.bio/wp-con- PM. Biotech Bioeng. 219;116:2803-2814. tent/uploads/sites/2/2018/11/0408_Bioprocessing_Spotlight. pdf. Published October 31, 2018. Accessed July 8, 2020. 57. Pais DAM. Development of advanced monitoring and control tools for rAAV production in the insect cell system [thesis]. 42. Rüdt M, Briskot T, Hubbuch J. Advances in downstream https://run.unl.pt/bitstream/10362/97879/1/Daniel%20 processing of biologics – spectroscopy: an emerging process Pais%20PhD%20Thesis%20Final.pdf. Published April, 2020. analytical technology. J Chromatogr A. 2017;1490:2-9. 43. Rüdt M, Briskot T, Hubbuch J. Advances in downstream processing of biologics – spectroscopy: an emerging process analytical technology. J Chromatogr A. 2017;1490:2-9.

CHAPTER 4 Process Development Using Quality by Design (QbD) Principles 99 Chapter 5 Upstream and Downstream Processing Chapter 5 | Contents UPSTREAM MANUFACTURING PROCESS DEVELOPMENT Upstream Process Overview...... 104 Overview of Process Steps...... 104 Plasmids...... 104 Quality Attributes of the AAV Vector Influenced by Cell Culture...... 105 Process Description...... 105 Master Cell Bank (MCB) and WCB Generation, Characterization, and Testing....106 Seed Expansion in Shake Flasks and Cell Bags...... 109 Seed Expansion in N-1 Bioreactor...... 109 Production Bioreactor...... 111 Seed Expansion in N-1 Bioreactor...... 110 Batch History...... 111 Process Understanding...... 113 Critical Raw Materials...... 114 HEK293 Cell Bank...... 114 Cell Culture Media...... 115 PEI...... 116 Plasmids...... 118 Vial Thaw...... 119 Development History...... 119 Seed Expansion in Shake Flasks and Cell Bags...... 119 Development History...... 119 Seed Expansion in Disposable Bioreactors...... 121 Development History...... 121 Production Bioreactor (Transfection/Infection)...... 121 Development History...... 121 Process Characterization...... 125 Risk Assessment 1...... 126 Risk Assessment 2...... 126 Process Characterization Studies...... 127 Risk Assessment 3 and Control Strategy...... 130 Applicability of Design Space to Multiple Operational Scales and Bioreactor Configurations: Engineering Design Space...... 132 Development of a Scale Down Model...... 132 Establishing a Scale Down Model and Design Space Applicability to Multiple Operational Scales...... 133 Qualification of Scale-down Model for Production Bioreactor and Engineering Design Space...... 133

CHAPTER 5 Upstream and Downstream Processing 101 Chapter 5 contents, continued

DOWNSTREAM PROCESS DESCRIPTION AND CHARACTERIZATION Summary...... 137 Process Understanding Based on Prior Knowledge...... 138 Viral Clearance...... 139 Batch History...... 139 Downstream Process Characterization...... 140 Lysis and Clarification...... 140 Step Description...... 142 Depth Filtration and Bioburden Filtration...... 142 Prior Knowledge...... 142 Scale-Down Model...... 143 Characterization Studies to Assess Impact to Product Quality...... 143 Hold Time Study...... 143 Characterization Studies to Assess Viral Inactivation...... 143 Summary of Process Parameter Classification and Ranges...... 144 Affinity Capture Chromatography...... 144 Step Description...... 144 Scale-Down Model...... 146 Risk Assessment to Plan Process Characterization Studies...... 146 Multivariate DOE Studies...... 146 Chromatography...... 146 Univariate Studies...... 146 Process Ranges Based on Platform Knowledge...... 146 Summary of Process Parameter Classification and Ranges...... 146 Reuse/Lifetime Resin Studies...... 147 Anion Exchange Chromatography...... 147 Step Description...... 149 Scale-Down Model...... 150 Risk Assessment to Define Process Characterization Studies...... 150 Multivariate DOE Studies...... 150 Univariate Studies...... 151 Process Ranges Based on Platform Knowledge...... 151 Summary of Process Parameter Classification and Ranges...... 151 Reuse/Lifetime Resin Studies...... 151 Ultrafiltration/Diafiltration...... 152 Step Description...... 152

CHAPTER 5 Upstream and Downstream Processing 102 Chapter 5 contents, continued

Prior Knowledge...... 152 Scale-Down Model...... 153 Risk Assessment to Define Process Characterization Studies...... 153 Process Characterization Studies...... 154 Tangential Flow Filtration Univariate Studies...... 154 TFF Diafiltration Buffer Conductivity...... 155 Flow Rate and Transmembrane Pressure Optimization...... 155 Poloxamer Sieving Rates...... 155 Tangential Flow Filtration Multivariate Studies...... 156 Summary of Parameter Classifications and Ranges...... 156 Summary of Downstream Process Design Space...... 157 Control Strategy for Downstream Process...... 157 Endnotes...... 158

NOTE: Due to inherent similarities in the processes for the production of therapeutic mono- clonal antibodies and gene therapies, some of the overall content and flow of this chapter was primarily based, but adapted from when applicable, A-Mab: a Case Study in Bioprocess Development, a document in the public domain. The authors of this A-Gene chapter acknowledge the work of the authors and editors in constructing the A-Mab case study.

For further details on the A-Mab process, please review: CMC Biotech Working Group. A-Mab: A Case Study in Bioprocess Development (chap- ter 3). CASSS website.

https://cdn.ymaws.com/www.casss.org/resource/resmgr/imported/a-mab_case_study_ version_2-1.pdf. Updated October 30, 2009. Accessed February 16, 2021.

CHAPTER 5 Upstream and Downstream Processing 103 Upstream Manufacturing Figure 5-1. A-Gene Upstream Process Overview. Process Development Thaw Vial

Upstream Process Overview Seed Expansion In Shake Flasks and Cell Bags OVERVIEW OF PROCESS STEPS A number of different systems may be used to produce AAV, including: the HEK293 – plasmid transfection Seed Expansion in N-1 Bioreactor system, the baculovirus system, and the producer cell line/helper virus system that could include HSV or adenovirus. As discussed earlier in this document, for Cultivation to Transfection Density the purposes of the A-Gene case study we will describe (in 200-L production bioreactor) a manufacturing process utilizing a HEK293 suspension cell line with transient plasmid transfection to produce adeno-associated virus (AAV) at a scale of 200 L (Figure Triple Transfectlon (in 200-L production bioreactor) 1). The upstream process is conducted entirely using disposable raw materials to maintain flexibility during manufacturing. Major steps in the upstream process include: Purification

Step 1. Vial thaw from cell bank Step 2. Seed expansion plasmids (gene of interest, rep/cap, and helper plasmids) Step 3. Production bioreactor and a transfection reagent. The production bioreactor is a. Cultivation to transfection density harvested approximately 2-4 days post transfection. At b. Triple transfection with gene of interest the time of harvest, detergent is added to the production (transgene), rep/cap, helper, plasmids bioreactor to lyse cells and release AAV. A nuclease is The A-Gene upstream process uses a commercially also added to digest endogenous DNA. Following lysis, available medium for the seed train and the production the contents of the bioreactor are clarified by filtration, bioreactor steps. During the seed expansion steps (Steps and further downstream purification operations are 1 and 2), one or more vials of the working cell bank commenced. (WCB) are expanded through a series of passages of Note: for the purpose of simplicity, the risk assess- increasing volume, generating sufficient biomass to inoc- ments presented in the upstream section of the case ulate the production bioreactor at the target inoculation study do not include extensive raw material and medi- cell density (Step 3). To reduce time between successive um composition considerations. In a real-life scenario, production batches, the seed train can be maintained upstream process risk analysis would require a thorough as a continuous or ‘rolling inoculum’ so that multiple understanding of the impact of medium and raw material production bioreactor batches can be inoculated from variability on process performance and product quality. the same vial thaw. Cells are cultivated in the production bioreactor until PLASMIDS a target cell density is reached, at which point the cells are The A-Gene process uses 3 plasmids Figure( 2). One plas- transfected. The transfection involves addition of three mid encodes the rep and cap genes of AAV (pRepCap)

CHAPTER 5 Upstream and Downstream Processing 104 using their endogenous promoters; the helper plasmid Figure 5-2. Plasmids Used to Manufacture (pHelper) encodes three additional adenoviral helper A-Gene Using a HEK293 Cell Line genes (E2a, E4, and VA RNAs) not present in HEK293 cells; and the final plasmid (pAAV-GOI) contains an ex- pression cassette with the gene of interest (GOI) flanked ITR P GOI pA ITR by two inverted terminal repeat (ITR) sequences.1 pAAV-GOI Of note, the helper plasmid may be a common com- ponent across multiple programs. X-Gene and Y-Gene (hypothetical) therapies also utilize the same helper plas- rep cap mid with the same HEK293 host to manufacture AAV vectors. It should be noted that the helper plasmid can pRep-Cap be considered a universal component across programs and be procured in greater quantities to reduce costs Adeno VA E4 E2A and provide operation efficiencies. However, the AAV serotype is chosen based on biodistribution in the target pHelper tissue; hence, the plasmid expressing the AAV capsid may also vary between programs.

AR, antibiotic resistance gene; ITR, inverted terminal repeat; QUALITY ATTRIBUTES OF THE AAV VECTOR GOI, gene of interest; P, promoter; pA, polyadenylation signal. INFLUENCED BY CELL CULTURE During development of the commercial process for

A-Gene, cross functional assessments were conducted atmosphere of 8% CO2 in air on an orbital shaker plat- to understand the impact of each step on product quality form. After the generation of sufficient cell mass, the attributes, as well as process attributes of the next step cell bank was created by aseptically harvesting cells by in the A-Gene manufacturing process. A summary of centrifugation and resuspending in cryopreservative product quality attributes influenced by the upstream medium (90% SuperExpress Medium and 10% dimethyl process is shown in Table 1. A detailed assessment to sulfoxide [DMSO]). SuperExpress Medium is a fictitious, understand the impact of each process parameter on chemically defined, serum-free and animal origin-free process performance and product quality is presented medium containing no proteins, hydrolysates, or compo- in the Process Understanding section. nents of undefined composition. The cells were frozen at a final density of 1x107 cells /mL using a controlled rate PROCESS DESCRIPTION freezing apparatus and stored in the vapor phase of liquid nitrogen. The MCB was approximately 9 generations Master Cell Bank (MCB) and Working Cell Bank (doublings) from the cells acquired from ATCC. Testing (WCB) Generation, Characterization, and Testing for the MCB is outlined in Table 3. Cell banks used in the manufacture of A-Gene are shown In addition to the testing strategy provided in Table 3, in Table 2. The commercial process uses the WCB, testing for AAV serotypes, JC/BK polyoma viruses, and which is established from the MCB. HSV I and II may also be applicable. The A-Gene HEK293 MCB was generated from The A-Gene HEK293 WCB was generated from the HEK293 cells acquired from a commercial source. The A-Gene HEK293 MCB using a similar procedure as de- cells were expanded per standard aseptic cell culture scribed for the MCB. Briefly, a vial of A-Gene HEK293 methods in a dedicated clean-room environment. MCB was expanded by passaging using standard Inoculation density for each passage was approximately aseptic cell culture methods in a dedicated clean-room 0.3x106 cells/mL with harvest densities between 1-3x106 environment. Inoculation density for each passage was cells/mL. The cells were cultured at 37°C in a humidified approximately 0.3x106 cells/mL with harvest densities

CHAPTER 5 Upstream and Downstream Processing 105 Table 5-1. Summary of Quality Attributes of the AAV Vector Influenced by Cell Culture

Attribute Description Analytical Method Notes % Full Capsids Capsids containing AUC, qPCR/ELISA, HPLC, electron transgene microscopy Capsid protein identity Capsid Serotype Mass Spectrometry Capsid protein purity >90% Capillary Electrophoresis Viral protein ratio Relative abundance of viral CE-SDS proteins VP1, VP2, and VP3 in the capsid Vector genome sequence Integrity of the transgene Sanger sequencing or NGS (Inclusive of ITRs and all sequence; may also be other components and controlled at the drug substance level gene of interest itself) Vector genome species DNA size distribution (GOI or Capillary Electrophoresis any fragments) Potency In vitro measure of activity ELISA or RT-qPCR RT-qPCR may be used to read out transgene mRNA expression as a measure of potency Infectious titer Concentration of viral particles TCID50 that can transduce cells Replication-competent ≤1 replication competent Cell based assay Serotype specific AAV AAV/108 genome copies positive control PTMs Intact mass Liquid chromatography, MS/MS methods Adventitious virus From cell line or reagents; not EP 2.6.16 Controlled at the cell line an issue for defined medium Mycoplasma From cell line or reagents EP 2.6.7 Controlled at the cell line Aggregation Propensity of the capsid to Acceptable level so as not to affect aggregate loss in concentration or potency Residual host-cell DNA Residual host-cell packaged and Base limit on amount dosed in non-target DNA and free DNA relevant toxicology studies Residual Host Cell Protein Base limit on amount dosed in relevant toxicology studies Residual plasmid DNA Base limit on amount dosed in relevant toxicology studies Capsid degradation and Capsid protein modification Within set limits to ensure functional modification (deamidation, oxidation) consistency in manufactured products

AUC, analytical ultracentrifugation; CE-SDS, capillary electrophoresis sodium dodecyl sulfate; ELISA, enzyme-linked immunosorbent assay; GOI, gene of interest; HPLC, high-performance liquid chromatography; ITR, internal terminal repeat; MS, mass spectrometry; NGS, next-generation sequencing; PTM, post-translational modification; TCID50, median tissue culture infectious dose (signifies concentration at which 50% of the cells are infected).

Table 5-2. Cell Banks Used to Manufacture A-Gene

Cell Bank Description A-Gene HEK MCB1 Master cell bank (MCB) A-Gene HEK WCB1 Working cell bank (WCB)

CHAPTER 5 Upstream and Downstream Processing 106 Table 5-3. A-Gene MCB Testing

Test Performed Method Description Specification

Cell line species identity Isoenzyme analysis to determine cell line identity Cells confirmed to be of human origin

Cell line identity STR DNA Profiling Analysis ≥80% match between the cell line and its original source

Mycoplasma detection FDA “Methods to Consider” – Inoculation of Free of detectable Mycoplasma Indicator Cell Line and Direct Cultivation contamination

In vitro assay to detect Direct inoculation into MRC-5, Vero, and A549 No CPE or hemadsorption observed of adventitious viral cell lines followed by an extended incubation; cell contaminants lines are observed for changes in morphology attributable to viral agents as well as testing for hemadsorption

In vivo assay for viral In-vivo assay utilizing guinea pigs, adult mice, Free of detectable adventitious contaminants and suckling mice; after injection with the test viruses substance, animals are observed for survival and good health

Detection of HIV-1 DNA Detection of HIV-1 DNA by qPCR Negative: HIV-1 DNA sequences not detected

Detection of HIV-2 DNA Detection of HIV-2 DNA by qPCR Negative: HIV-2 DNA sequences not detected

Detection of HTLV-1 DNA Detection of HTLV-1 DNA by qPCR Negative: HTLV-1 DNA sequences not detected

Detection of HTLV-2 DNA Detection of HTLV-2 DNA by qPCR Negative: HTLV-2 DNA sequences not detected

Detection of HCV RNA Detection of human HCV DNA by RT-qPCR Negative: HCV not detected

Detection of HBV DNA Detection of HBV DNA by qPCR Negative: HBV DNA sequences not detected

Detection of CMV DNA Detection of CMV DNA by qPCR Negative: CMV not detected

Detection of EBV DNA Detection of human EBV DNA by qPCR Negative: EBV sequences not detected

Detection of parvovirus Detection of human parvovirus B-19 DNA by qPCR Negative: parvovirus B-19 DNA B-19 DNA sequences not detected

Detection of HHV-6 Detection of HHV-6 Variant A & B DNA by qPCR Negative: HHV-6 A & B DNA variant A & B DNA sequences not detected

Detection of HHV-7 DNA Detection of HHV-7 DNA by qPCR Negative: HHV-7 DNA sequences not detected

Detection of HHV-8 DNA Detection of Human Herpes Virus 8 DNA by qPCR Negative: HHV-8 DNA sequences not detected

Detection of HAV RNA Detection of HAV DNA by RT-qPCR Negative: HAV RNA sequences not detected

CHAPTER 5 Upstream and Downstream Processing 107 Table 5-3. A-Gene MCB Testing continued from previous page

Detection of human Ad5 Detection of human Ad5 by qPCR Negative: Ad5 DNA sequences not (hexon) DNA detected

Cell morphology Electron microscopy for retrovirus-like particles No viral particles detected and virus detection/ tabulation Reverse transcription Fluorescence PCR–based reverse transcriptase Free of detectable reverse assay assay transcriptase activity Detection of Direct inoculation of vero and bovine turbinate cells Free of detectable adventitious adventitious bovine followed by an extended incubation. The cells are viruses viruses stained to detect changes in morphology as well as antibody testing against the following viruses: bovine viral diarrhea virus, bovine parvovirus, bovine adenovirus, bovine respiratory syncytial virus, blue tongue virus, reovirus, and rabies virus Detection of Direct inoculation of vero and swine testis cells Free of detectable adventitious adventitious porcine followed by an extended incubation. The cells are viruses viruses stained to detect changes in morphology as well as antibody testing against the following viruses: bovine viral diarrhea virus, porcine parvovirus, porcine adenovirus, reovirus, transmissible gastroenteritis virus, hemagglutinating encephalomyelitis virus, and rabies virus Sterility USP Immersion Free of viable microbial contamination Bacteriostasis/ USP Immersion Free of detectable microbial fungistasis inhibitors Endotoxin Kinetic chromogenic LAL 70% is considered percentage of viable of cells by trypan blue staining suitable

between 1-3x106 cells/mL. The cells were cultured at performed once on each WCB, a risk-based approach can

37°C in a humidified atmosphere of 8% CO2 in air on an be adopted to establish additional WCB testing needs. orbital shaker platform. After generation of sufficient cell Specifically testing for the presence of the adventitious mass, the cell bank was created by aseptically harvesting agents that may have been introduced during the manu- cells by centrifugation and resuspending in cryopreser- facture of the WCB from the MCB should be conducted. vative medium (90% SuperExpress Medium and 10% For example, testing for specific viruses may be executed DMSO). The cells were frozen at a final density of 1x107 on the WCB if not tested on the MCB. cells/mL using a controlled rate freezing apparatus and An important consideration of cell line choice is post stored in the vapor phase of liquid nitrogen. The WCB bank viability assessment, characterization, and tumori- was approximately 9 generations from the MCB. Testing genicity. Depending on the specific cell line being used, for the WCB is outlined in Table 4. The information to tumorigenicity may need to be evaluated through risk document qualification and characterization for a WCB assessments though the data may be leveraged across proj- is generally less extensive than that for the MCB. While ects that utilize the same host cell line. Animal studies may tests of purity and limited tests of identity should be be conducted if deemed necessary to mitigate the risk.

CHAPTER 5 Upstream and Downstream Processing 108 Table 5-4. A-Gene WCB Testing

Test Performed Method Description Specification Cell line species identity Isoenzyme analysis to determine cell Cells confirmed to be of human origin line identification Mycoplasma detection Ensuring lack of Mycoplasma Free of detectable Mycoplasma contamination In vitro assay for detection of Direct inoculation into MRC-5, vero, No CPE or hemadsorption observed adventitious viral contaminants and A549 cell lines followed by an extended incubation. Cell lines are observed for changes in morphology attributable to viral agents as well as testing for hemadsorption In vivo assay for viral contaminants In vivo assay utilizing guinea pigs, Free of detectable adventitious adult mice, and suckling mice. After viruses injection with the test substance, animals are observed for survival and good health Cell morphology and virus detection/ Electron microscopy to inspect for No viral particles detected tabulation retrovirus-like particles Reverse transcription assay Fluorescence PCR-based reverse Free of detectable reverse transcriptase assay transcriptase activity Sterility USP Immersion Free of viable microbial contamination Bacteriostasis/fungistasis USP Immersion Free of detectable microbial inhibitors Endotoxin Kinetic chromogenic LAL 5 EU/ml Cell viability assay Thaw cells after freezing to determine Generally >70% is considered suitable the percentage of viable of cells by trypan blue staining

Seed Expansion in Shake Flasks and Cell Bags and growth rate (doubling time) were monitored during The seed culture expansion stage is performed using the seed train expansion stages. SuperExpress Medium supplemented with L-glutamine In order to generate sufficient biomass for inoculation to a final concentration of 4 mM prior to use. The seed of the N-1 bioreactor, a passage at the 25-L scale in cell expansion stage involves cultivation in shake flasks of in- bags was required. One cell bag with a working volume of creasing volume, followed by cultivation in cell bags. The 20 L each was used. The cell bag was agitated by rocking medium is supplemented with 0.2% v/v Pluronic-F68 or at 25 rpm with a rock angle of 10°. The cells were cultured other shear protectant for cultivation in cell bags. at 37°C while maintaining a continuous supply of air A single vial of the A-Gene HEK293 WCB was thawed through the cell bag using dissolved oxygen at 40% of in SuperExpress Medium and expanded in shake flasks. equilibrium with air. The pH was kept at 7.2 using CO2. A seed density of 0.3x106 was used for inoculation. For Culture duration was generally between two and three cultivation in shake flasks, one or more flasks of suc- days. Table 3 outlines seed expansion process parameters. cessively larger volume (250 mL to 2.5 L) were cultured Seed Expansion in N-1 Bioreactor at 37°C in a humidified atmosphere of 8% CO2 in air on an orbital shaker platform rotating at 80 to 135 rpm In order to generate sufficient biomass for inoculation of the depending on flask volume. The culture density, viability, 200-L production bioreactor, a passage at the 50-L stirred

CHAPTER 5 Upstream and Downstream Processing 109 Table 5-5. Seed Expansion Process Parameters

Process Step In-Process Monitoring and Control Parameter Range Vial thaw Thaw duration 6 minutes Final thaw temperature 37ºC Duration at final thaw temperature 5-15 minutes Viability after thaw ≥70% Seed expansion Viable cell density Inoculation target: 0.3×106 cell/mL (shake flasks) At passage target: 3×106 cell/mL Viability ≥85% viability at passage Passage duration 3 ± 1 days Incubator temperature 37°C Incubator relative humidity 80% Incubator rpm/throw radius 2.5 cm

Incubator CO2 % 8% Flask nominal volume/working volume 250 mL/100 mL 500 mL/250 mL 1000 mL/500 mL 2500 mL/1000 mL Seed expansion Viable cell density Inoculation target: 0.3×106 cell/mL (cell bags) At use target: 3×106 cell/mL Viability ≥85% final viability Passage duration 3±1 days Temperature 37ºC pH 7.2 Dissolved oxygen 40% Rock rate 25 rpm Rock angle 10º Gas flow Air: 0.5 lpm max Oxygen: 0.25 lpm max Nominal volume/working volume 25 L/20 L bag bioreactor was required (Table 6). The N-1 bioreactor culture was maintained for two to three days until it reached operates with a 50-L working volume. Once sufficient the transfection density. The culture density, viability, and biomass was obtained through the seed train, the N-1 biore- growth rate (doubling time) were monitored during the actor was inoculated at a seed density of 0.3 x 106 cells/mL. seed train expansion stages. SuperExpress Medium supplemented with L-glutamine to A rolling inoculum may be established as needed a final concentration of 4 mM and Pluronic-F68 (0.2% v/v) to support a multibatch manufacturing campaign for or other shear protectant. An agitation rate of 70 rpm was A-Gene. Based on development data, the seed train may used during cultivation. Dissolved oxygen was maintained extend 50 generations from the MCB to production bio- at 40% of equilibrium with air through a continuous air reactor inoculation or 40 generations from the WCB to sparge and was further supplemented with pure oxygen production bioreactor inoculation without significant loss on demand. The pH was controlled at 7.2 with CO2. The in productivity or impact on product quality attributes.

CHAPTER 5 Upstream and Downstream Processing 110 Table 5-6. Seed Expansion in N-1 Bioreactor Process Parameters

Process Step In-Process Monitoring and Control Parameter Range Maximum working volume 50L Temperature 37°C N-1 Bioreactor pH 7.2 Dissolved oxygen 40% Agitation 70 rpm Gas flow Headspace air: 1.2 lpm Air sparge: 1 lpm max Oxygen sparge: 0.25 lpm max Sparger type 0.5 mm ring sparger Viable cell density Inoculation target: 0.3 x 106 cell/mL

Viability ≥85% final viability at passage Passage duration 3 ± 1 days Antifoam 50 ppm maximum

Production Bioreactor was used to dilute the plasmid DNA and transfection The production bioreactor operates at a 200-L scale and reagent to a volume of 10 L. After 10 to 15 minutes of utilizes SuperExpress Medium. To ensure that sufficient incubation at room temperature, the transfection cocktail volume was available for the addition of transfection was pumped into the production bioreactor. At 3 hours reagents, the culture was started at approximately 190 L. post-transfection, HEK293 media was pumped into the Once sufficient biomass was obtained through the seed production bioreactor at a volume of 10% of the final vol- train, the production bioreactor was inoculated at a seed ume of cell culture. Cell culture was harvested from the density of 0.5x106 cells/mL. SuperExpress Medium was bioreactor bag at least 72 hours post-transfection. Table supplemented with L-glutamine to a final concentration 7 outlines production bioreactor process parameters. of 4 mM and Pluronic-F68 (0.2% v/v), or another shear protectant. An agitation rate of 120 rpm was used during Batch History cultivation. Dissolved oxygen was maintained at 40% of equilibrium with air through a continuous air sparge To-date, eight at-scale batches have been produced in and was further supplemented with pure oxygen on this hypothetical case study. Manufacturing history for demand. The pH was controlled at 7.2 with CO2. The A-Gene is shown in Table 8. culture was maintained for 2 to 3 days until it reached Three manufacturing processes were used for the the transfection density. The culture density, viability, and manufacture of A-Gene. Process 1 was used to generate growth rate (doubling time) were monitored during the material to support toxicology studies to enable the phase seed train expansion stages. 1 IND. This process was scaled up to meet clinical de- Once transfection density was reached, the transfec- mand in the form of Process 2. This process was further tion mixture was added to the production bioreactor. The optimized to support late-stage clinical studies as Process transfection mixture included sufficient plasmid DNA to 3, which was then validated for commercial manufacture. ensure 2.0 μg of DNA/mL of cell culture (0.5 pg pDNA/ Updates made between Process 1, Process 2, and Process cell). The plasmid DNA in the transfection mixture was 3 to accommodate facility fit requirements and increase at a ratio of 2:1:1 for pHelper:pRepCap:pAAV-GOI. The process productivity and robustness are listed in Table transfection reagent polyethyleneimine was added at a 9. Pilot batches (PLT-001 and PLT-002) support toxicol- ratio of 2:1 relative to plasmid DNA. SuperExpress media ogy studies. ENG-001 supports process development to

CHAPTER 5 Upstream and Downstream Processing 111 Table 5-7. Production Bioreactor Process Parameters

Process Step In-Process Monitoring and Control Parameter Range Maximum working volume 200 L

Temperature 37°C Production Bioreactor pH 7.2 Dissolved oxygen 40%

Agitation 120 rpm

Gas flow Headspace air: 1.2 lpm Air sparge: 2 lpm max Oxygen sparge: 0.5 lpm max

Sparger type 0.5 mm ring sparger

Viable cell density Inoculation target: 0.3x106 cell/mL Transfection target: 4x106 cell/mL

Viability ≥85% final viability at transfection

Passage duration 3±1 days

Antifoam 50 ppm maximum

A product yield of 3x1010 vector genomes/mL is expected from the production bioreactor at harvest.

Table 5-8. A-Gene Manufacturing History

Process Lot # Bioreactor Scale Purpose Process 1 PLT-001 50 L GLP toxicology study PLT-002 50 L GLP toxicology study Reference Material 1 PLT-002 50 L Process consistency Process 2 ENG-001 200 L Process development and development stability Reference Material 2 GMP-001 200 L FIH clinical studies and stability studies GMP-002 200 L FIH clinical studies Process 3 ENG-002 200 L Process development and development stability Reference Material 3 GMP-003 200 L Phase 2/3 clinical studies and stability studies GMP-004 200 L Phase 2/3 clinical studies and stability studies EN-003 200 L Process development and development stability Reference Material 4 PPQ-001 200 L Launch supplies Confirm design space and control strategy PPQ-002 200 L Launch supplies Confirm design space and control strategy PPQ-003 200 L Launch supplies Confirm design space and control strategy

CHAPTER 5 Upstream and Downstream Processing 112 Table 5-9. Process Steps to Accommodate Facility Fit Requirements and Increase Process Productivity and Robustness

Process Step Process 1 Process 2 Process 3 Vial thaw Controlled rate thaw using Controlled rate thaw using Controlled rate thaw using thaw device thaw device thaw device Seed expansion N-1 at 25-L scale N-1 stage at 50-L N-1 50-L scale medium 1 medium 1 medium 2 Cultivation to 50 L medium 1 200 L medium 1 200 L medium 2 transfection density

Triple transfection • PEI 40K transfection • PEI 40K transfection • PEI MAX 40K transfection reagent reagent reagent • Platform transfection and • Platform transfection and • Optimized feed addition at harvest parameters harvest parameters transfection • Optimized transfection density • Optimized plasmid ratios • Optimized harvest time

facilitate technology transfer. GMP-001 and GMP-002 manufacturing. The overall strategy that guided process generate materials for FIH studies. ENG-002 supports development for the upstream process is presented in process development to facilitate technology transfer of Figure 4. This section only includes discussion through the updated upstream process. GMP-003 and GMP-003 the establishment of a design space and a draft control generate materials for phase 2/3 clinical studies. The strategy. Process performance qualification, control process is continuously changed and updated to improve strategy, and life cycle management are not included. robustness and to ensure facility fit. Early-stage process development was conducted Details regarding development conducted prior to mainly to support early clinical development and used implementation of process changes are described in the readily available raw materials to meet immediate de- Process Understanding Section. mand. As commercial process development was initiated, a high-level assessment considered the impact of raw ma- Process Understanding terial and process parameters on productivity and prod- uct quality. Preliminary identification of critical quality The following section describes the development attributes (CQAs) was important for this assessment. The history and summarizes process characterization that crucial role of raw materials, specifically plasmids, cell enabled incorporation of a QbD approach for A-Gene culture media, and transfection reagents, was considered

Table 5-10. Raw Material Risk Assessment

Risk of Impact to Product Risk of Impact to Key Raw Material Quality Attributes Process Attributes

Plasmids High High

PEI Low High

Cell culture medium High High

Post-transfection medium High High

Antifoam Low Medium

CHAPTER 5 Upstream and Downstream Processing 113 Figure 5-4. Process Characterization Strategy

Life cycle PROCESS Experimental management PARAMETERS lab work

Draft control Prior knowledge strategy Process Final Process Process performance control Process understanding characterization characterization verification strategy BLA Product understanding Design space

QUALITY ATTRIBUTES

Risk assessment 1 Risk assessment 2 Risk assessment 3 PPQ Risk assessment 4

High level risk Low level risk Develop control assessment CQA FEMA assessment strategy determination experiment design Initial control strategy

(Table 10). A template for a high-level risk assessment is upstream process unit operations is described in this shown in Table 11. Commercial process development was section, starting with raw materials. Please note that the conducted to optimize AAV productivity while ensuring raw materials section does not cover all raw materials for suitable product quality attributes were achieved. simplicity. A similar treatment would be applied to other Upon completion of clinical stage process develop- key raw materials as well. ment, detailed risk assessments were conducted to assess the impact of process parameters on product quality at- CRITICAL RAW MATERIALS tributes. The risk assessment was based on knowledge de- veloped during early- and late-stage process development HEK293 Cell Bank and also identified gaps in knowledge. On the basis of this The host cell line is a critical material. Maintaining a risk assessment, laboratory process qualification/charac- sufficiently characterized cell bank is essential to manu- terization was conducted. Statistically designed experi- facturing success. The HEK293 cell bank establishment ments were planned and executed to enable development and characterization has been described in preceding of models to link process parameters to productivity and sections. The cell bank should be periodically assessed product quality. Through process characterization and for stability. This may include basic assessment of growth statistical modeling of results, process parameters having and viability for three to five passages out of thaw. In significant impact on the product quality and productiv- certain circumstances full production of AAV may also ity were identified, as well as potential ranges for these be considered. For A-Gene, the HEK293 cell bank was es- parameters that would ensure desired product quality tablished prior to phase 1 GMP manufacturing (Process and productivity. This represented the design space for 1) and stored in the vapor phase of liquid nitrogen. A A-Gene. Finally, optimized process parameters were vial of the cell bank was thawed and passaged three times established at a large scale through process validation. every 5 years to assess stability. If a vial of the cell bank Implementation of this development approach for was used for GMP manufacturing within 5 years from

CHAPTER 5 Upstream and Downstream Processing 114 Table 5-11. High-Level Risk Assessment Template

Expansion in Vial Shake Flasks/ N-1 Production Category Process Step Thaw Bags Bioreactor Bioreactor

Vector genome concentration % full capsids Capsid protein identity Capsid protein purity VP ratio Vector genome sequence Vector genome species Potency Product Quality Infectious titer Attributes Replication-competent AAV Capsid PTMs (Deamidation, oxidation) Adventitious virus Aggregation Residual host cell DNA Residual HCP Residual plasmid DNA Residual E1A oncogene DNA

Capsid degradation and modification Pre-transfection doubling time Viability at transfection Process Attributes Cell density at harvest Viability at harvest AAV titer at harvest the prior stability pull point, data from initial passaging Ensuring that the supply and performance of the medi- were used to support stability. um can be maintained across multiple lots during late stage clinical and commercial manufacturing is essential. Cell Culture Media To this end, the hypothetical project team conducted a Cell culture media is of critical importance to the up- high-level risk assessment for A-Gene prior to initiating stream process because it impacts not just cell growth late-stage development to ensure that the choice of media but also productivity and product quality. The process was appropriate for late-stage and commercial manufac- for A-Gene uses SuperExpress medium supplemented turing from technical and business perspectives. with glutamine in the cell expansion and production A high-level risk summary template has been provid- bioreactor. Additionally, HEK293 media is used to add ed for illustrative purposes (Table 12). The level of detail supplementary nutrients to the culture after transfection. in these categories as well as the classification of risks

CHAPTER 5 Upstream and Downstream Processing 115 Table 5-12. High-Level Risk Assessment Template for Media Selection

Risk Category Associated risks Justification/Mitigation Transfection efficiency Process performance Productivity Product quality Consistency across lots Potential for future improvement Media stability Media characteristics Shipping/handling considerations Testing requirements beyond CoA Cost Business and quality risk Supply continuity

BSE/TSE risk Regulatory and safety Animal-derived components Recombinant factors into categories is dependent upon the team conducting atom. This attracts and holds onto positive charges due the assessment and the details of the medium itself. For to low pKa values (a phenomenon commonly referred example, in the case of internally developed medium to as the proton sponge effect) and enables complexing with in-house clinical/commercial manufacturing ca- with negatively charged pDNA molecules. Another key pabilities, the nature of such an assessment will be dif- attribute is the length of the PEI molecule and heteroge- ferent. Similarly, contract manufacturing facilities using neity of the PEI molecules with respect to the length. PEI commercially available media may require consideration with an optimal molecular weight (length) manages cell of different factors. For A-Gene, the initially chosen com- toxicity attributable to long fragments and has reduced mercially available media was changed after Process 3 complexation potential attributed to shorter fragments. development. The risk assessment was conducted and Heterogeneity in the length of PEI molecules in the final the commercial process was updated. product leads to heterogenous complex size formation and poor reliability in transfection efficiency. When PEI using PEI, an addition and complexing protocol should Polyethylenimine (PEI) is a synthetic polycation that has be optimized for the specific plasmid constructs. relatively high transfection efficiency and is commer- Several commercially available PEI products are cially produced for GMP manufacturing. PEI adheres used for transient transfection. As a critical raw mate- to and condenses plasmid DNA to form a complex that rial, PEI products for GMP manufacturing require a is endocytosed by the host cell. Upon entry into the cell, validated manufacturing process and quality control the endosome swells from osmotic pressure and lyses, with quality attribute testing. Manufacturing of material releasing plasmid DNA into the cytoplasm. The plasmid for toxicology also requires high-quality material with DNA then migrates into the nucleus and the host cell be- appropriate quality documentation. However, it is not gins replication of the transgene as well as transcription necessary to follow GMP standards. We used hypothet- of the capsid protein genes and helper genes. ical SuperTransfect PEI transfection reagent for A-Gene. In summary, efficacy of PEI is dependent on many Key quality attributes used to assess quality of the PEI factors, including molecular weight, branching, cationic include identity, potency, purity, quality, and safety (Table charge density, genetic material load, and buffer capaci- 13).4 Molecular weight and heterogeneity are generally ty.3 PEI can be linear or branched with added functional measured using SEC-based methods. Free nitrogen groups to improve transfection efficiency. The key char- available for complex formation may be measured using acteristic of PEI that determines its ability to transfect analytical techniques such as NMR, but performance cells is a repeated protonatable nitrogen at every third assays are commonly used as a surrogate.

CHAPTER 5 Upstream and Downstream Processing 116 Table 5-13. PEI Quality Attributes

Attribute Description Assay Method Identity Polymer structure Fourier-transform infrared spectroscopy or similar Molecular weight Size Exclusion Chromatography Polydispersity index Appearance Visual inspection for color and clarity pH Transfection efficiency through activity test Osmolality Potency Performance of material Safety Endotoxin Applicable USP Methods Sterility Mycoplasma Purity Heavy metals Applicable USP Methods

Table 5-14. Testing for Plasmid Used in the Manufacturing Process for A-Gene

Assay Method Absorbance 260/280 ratio purity UV spectrophotometry Appearance Visual inspection Concentration UV spectrophotometry DNA homogeneity Densitometry analysis of EtBr-stained agarose gel electrophoresis Endotoxin Kinetic Chromogenic LAL Identity EtBr-stained agarose gel electrophoresis Plasmid identity Double-stranded primer walking sequencing Residual host genomic DNA qPCR Residual host protein Micro BCA Residual host RNA SYBR gold-stained agarose gel electrophoresis Restriction digest EtBr-stained agarose gel electrophoresis Sterility USP Direct Inoculation Sterility validation (bacteriostasis/ USP Direct Inoculation fungistasis) Mycoplasma contamination qPCR Osmolality USP pH USP Bioburden Testing for total aerobes, anaerobes, spore-formers, and fungi Conductivity Conductivity meter Detection of Kanamycin ELISA

ELISA, enzyme-linked immunosorbent assay; EtBr, ethidium bromide; qPCR, quantitative polymerase chain reaction; UV, ultraviolet

CHAPTER 5 Upstream and Downstream Processing 117 Table 5-15. Testing for Cell Bank Used to Manufacture Plasmid for A-Gene Manufacturing

Assay Method Final product appearance testing Visual testing

Host cell identity Bacterial colony morphology

Lytic phage contamination Plate bacterial cells on media without antibiotics

Host cell identity Gram stain analysis

Antibiotic resistance CFU isolation on antibiotic-containing and antibiotic-free plates

DNA homogeneity Densitometry analysis of EtBr-stained agarose gel electrophoresis

Identity EtBr-stained agarose gel electrophoresis

Restriction digest EtBr-stained agarose gel electrophoresis

Plasmid identity Double-stranded primer walking sequencing

Cell bank viability CFU/mL plate count analysis

Host cell purity TSA and SDA

Detection of lysogenic bacteriophage Plated in the presence of Mitomycin C

Plasmid retention Antibiotic typing

CFU, colony-forming units; EtBr, ethidium bromide; SDA, Sabouraud dextrose agar; TSA, trypticase soy agar.

Plasmids lysate is loaded onto an anion exchange column. After Plasmids are critical raw materials for the manufacture elution, the plasmid is diafiltered into the final buffer. of AAV, and high-quality plasmids are required for the The plasmid DNA concentration in the final formula- manufacturing process. For the A-Gene case study, it tion is assayed and may be further diluted to reach the is assumed that no changes are made to the plasmid or desired concentration. The plasmid preparation is then the plasmid manufacturing process during the course of sterile-filtered using a 0.22-µm membrane and dispensed process development. However, if changes are made to a into final vials in a Class 100 laminar flow hood. The plasmid or its manufacturing process, a risk assessment final vials are labeled and visually inspected prior to should be conducted to thoroughly document the impact frozen storage. Table 14 describes qualities that must be on the process, such as a change in transfection efficien- monitored and methods by which to do so, and Table 15 cy or impurity profile of the final product. Any gaps in provides an overview of cell bank testing. knowledge to mitigate risk should be addressed through The generation ofE. coli cell banks to facilitate plas- experimental work to ensure safety and comparability of mid manufacturing is highly recommended for process the AAV material generated. reproducibility.5 A significant secondary structure of Plasmids used in the manufacture of A-Gene are ITR sequences can result in deletion of these sequences manufactured under good manufacturing practice during plasmid propagation in E. coli.6,7 Plasmids that (GMP). The plasmids are produced in Escherichia coli lose the ITRs have a replication advantage in transformed (generic strain) in animal product-free medium with the cells. Because intact terminal repeats must be maintained appropriate antibiotic. The cultures are harvested by cen- for efficient replication and packaging of the transgene, trifugation, and the biomass is suspended in Tris/EDTA various strategies are employed to maintain ITR integrity. buffer containing RNase A (sourced from Australia or This includes use of specific cell lines, maintenance of New Zealand due to lower risk of BSE and TSE) and selection pressure through appropriate antibiotics (after subjected to alkaline lysis. The crude lysate is clarified assessing safety risks), and limited propagation time after by centrifugation followed by filtration. The clarified thaw of the cell bank. Stability of these banks should be

CHAPTER 5 Upstream and Downstream Processing 118 assessed periodically for cell growth, plasmid retention, SEED EXPANSION IN SHAKE FLASKS AND CELL BAGS and integrity. Additional characteristics of the purified plasmids (among those listed in Table 14) may be as- Development History sessed as part of the stability plan. The purpose of the seed stage is to build biomass while maintaining the health of the cells in suspension. The VIAL THAW seed expansion process for A-Gene corresponds to a well-established platform process, and the same host cell Development History line has been used across multiple toxicology and phase The vial thaw is a well-established manufacturing step I and phase II manufacturing campaigns in different and no development was undertaken for Process 1, 2 media. These data were compiled and reviewed as part or 3. The vial thaw and initial expansion have been ex- of the initial assessment prior to Process 1. Performance ecuted similarly for X-Gene and Y-Gene for pilot scale of the seed train was measured by assessing growth rate and GMP runs with the exception of media into which (doubling time) of the culture during each passage and the thawed vial is inoculated. Before starting Process viability at the end of each passage. Similar and consis- 3 development, a risk assessment was undertaken to tent performance was observed for A-Gene, X-Gene document the impact of the vial thaw on key process and Y-Gene despite the different cell culture medium. and product quality attributes. Since no product accu- Considering the absence of product accumulation during mulates at this stage, the impact of this step on the final seed expansion and extensive experience with routine productivity and product quality is through impact to passaging of the host cell line in shake flasks, rocking bag, cell health. The risk of such impact has been demon- and stirred bag bioreactors, the impact of this step on strated to be low based on prior experience. final product quality was deemed low. Note that the only This risk assessment assumes that the seed expan- difference for this step across these 3 processes was scale sion process is operated following well established and of operation. For Process 1, the production bioreactor successful process control strategies to ensure that seed was 50 L, so a rocking cell bag served as the N-1 stage. culture performance is robust and reproducible. Batch Process 2 and Process 3 were executed at the 200-L scale, record procedures, SOPs, process descriptions, and pro- so the rocking bag reactor served as the N-2 stage. cess controls ensure that the seed expansion steps are This initial assessment assumed that the seed expan- monitored and operated within established limits. This sion process is operated following well established and would include limits for parameters and attributes such successful process control strategies to ensure that seed as inoculation seeding density, culture duration, viability, culture performance is robust and reproducible. Batch pH, temperature, and CO2. record procedures, SOPs, process descriptions, and pro- This risk analysis has been simplified by not including cess controls ensure that the seed expansion steps are medium and raw material considerations along with this monitored and operated within established limits. This step. It could be assumed that such sources of variability would include limits for parameters and attributes such have been identified and that the appropriate raw ma- as inoculation seeding density, culture duration, viability, terial control strategies are in place based on platform temperature, and CO₂. process knowledge and prior experience with other gene Given these considerations an initial assessment of therapy products. If such knowledge and controls are not the parameters assessed during development of the seed available, the risk assessments would be used to guide stage included: a comprehensive evaluation of the impact of medium and raw material variability on process performance • Culture media: While a preferred media formulation and product quality. The results of such studies would was adopted for early-stage manufacturing, different then serve as a basis to establish appropriate testing and commercially available media formulations were ex- control strategies to ensure that raw materials and media amined prior to late-stage manufacturing. One of the meet their respective quality acceptance criteria. goals of the media screening was to enable use of the

CHAPTER 5 Upstream and Downstream Processing 119 Table 5-16. Impact of Process Parameters and Risk Assessment for Seed Expansion in Shake Flasks

Product Quality Process Parameter Impact to Process Risk

Inoculation viable cell density May introduce lag (low density) or exceed the at- Low passage cell density/viability criteria (high)

Final (“at passage”) cell density May introduce lag for next passage Low

Viability at passage May introduce lag for next passage or indicate other Low issues with the culture

Passage duration Manage through inoculation density Low

Incubator temperature Impacts cell growth rate (generally easy to control) Low

Incubator relative humidity Impacts evaporation rate (generally easy to control) Low

Incubator rpm/throw radius Impacts oxygenation (not significant at low cell Low densities during seed train)

Incubator CO2 % Enables maintenance of pH during the culture Low Flask nominal volume/working volume Impacts oxygenation (not significant at low cell Low densities during seed train)

Table 5-17. Impact of Process Parameters and Risk Assessment for Seed Expansion in Cell Bags

Parameter Impact to Process Product Quality Risk Inoculation viable cell density May introduce lag (low density) or exceed the at Low passage cell density/viability criteria (high); leverage shake flask data

Temperature Impacts cell growth rate (generally easy to control) Low

pH Impacts cell growth rate (generally easy to control) Low

Dissolved oxygen Impacts cell growth rate (generally easy to control) Low

Rock rate Impacts oxygenation and hence cell growth rate Low (generally easy to control)

Rock angle Impacts oxygenation and hence cell growth rate Low (generally easy to control)

Gas flow Impacts oxygenation and hence cell growth rate Low (generally easy to control)

Nominal volume/working Impacts oxygenation (not significant at low cell Low volume densities during seed train)

CHAPTER 5 Upstream and Downstream Processing 120 same medium for seed train and production bioreactor a well-established platform process where process un- to streamline raw material sourcing. Consideration of derstanding is derived from prior knowledge with other a different medium for the production bioreactor also AAV products. necessitated examination of the same medium for the To establish operating parameters for Process 2, per- seed stages. Cost, availability for GMP manufacturing, formance of the production bioreactor step for Process and ability to support sufficiently high cell densities 1 prior to transfection was considered. These data were were used as criteria for assessment. Productivity was applicable since the goal of cultivation in the N-1 step is considered in case the same medium was tested for the the same as cultivation in the production bioreactor prior production bioreactor stage as well. to transfection: to build biomass while maintaining cell • Inoculation density: Inoculation density must be health. Process 3 development resulted in selection of a assessed at each passage to minimize any lag that may different media than that used for Process 2, but the op- occur from the introduction of cells at a relatively low erating parameters were unchanged. Appropriate perfor- concentration in a nutrient-rich environment. mance, as measured by culture growth rate and viability, • Maximum attainable cell density: It is important to was observed using these parameters even with the new understand the maximum attainable cell density at the cell culture media. This information has demonstrated end of each passage that enables consistent growth for that the N-1 expansion step is robust and reproducible. the successive passage. Maximizing the usable biomass at the end of each step can reduce the passages required PRODUCTION BIOREACTOR to obtain the target cell number to inoculate the pro- (TRANSFECTION/INFECTION) duction bioreactor. Exceeding the maximum ‘at use’ density may introduce a lag (decline in growth rate) for Development History the next expansion step due to depletion of nutrients The development history through implementation of the or accumulation of toxic metabolic byproducts that commercial process is presented in Figure 5. This section adversely impact cell health. For early-stage manufac- does not cover process validation. turing, media vendor recommendations regarding the Process 1: Process history for A-Gene is shown in maximum recommended cell density were followed, Figure 5. Early-stage process development focused but further characterization of media capabilities was on ensuring adequate clinical supply within program undertaken in preparation for late-stage manufactur- timelines with appropriate quality, and late-stage opti- ing prior to process characterization. mization focused on robustness and productivity (with • Seed train development: Seed train development in- productivity being mainly a business target). The initial volved assessment of the impact of cell age on produc- process developed for A-Gene (Process 1) was used for tivity (i.e., the number of generations from the WCB manufacture of regulatory toxicology supplies. that supported good performance in the production The Process 1 development strategy involved the stage) to enable establishment of a rolling inoculum so following considerations: that multiple production batches could be inoculated from the same seed train. • Material demand for toxicology and clinical material was combined with downstream yield projections to SEED EXPANSION IN DISPOSABLE BIOREACTORS assess the range of productivity appropriate for the upstream process and whether the demand could be Development History met with the available bioreactor (50 L for toxicology Seed expansion in disposable stirred bag bioreactors production and 200 L for GMP manufacturing). was implemented for Process 2 and Process 3 only, • Commercially available cell culture medium that had as required by the larger scale of operation. Similar been used previously to generate material for nonclinical to culture expansion in shake flasks and cell bags, the and clinical studies was used for Process 1. Extensive A-Gene seed expansion step in stirred bag reactors uses screenings of medium and feeding were not conducted.

CHAPTER 5 Upstream and Downstream Processing 121 Figure 5-5. Development History. DOE, Design of Experiments

Process 1 Process 2 Optimization Optimization Process 3 Prior Toxicology Phase I/III DOE 1 DOE 2 Confirmation knowledge 50 L 200 L 2 L 2 L 2 L

Process 3 Process 3 Process Process Process Commercial Scale up Phase III Character- Confirmation Validation Launch 200 L 200 L ization 2 L 200 L 200L

• The transfection process used at the laboratory scale Process knowledge accumulated across programs was and for prior programs was tested without significant considered when determining the development plans development to assess. (Table 19). • Suitability of the process to scale up to 200 L was Statistically designed experiments were executed, with considered early to ensure that the process could the initial goal of selecting the medium and feed added quickly be transferred to GMP manufacturing as post transfection. The statistical design allowed combi- needed after toxicology manufacturing at 50-L scale. nations of the factors to be tested while also detecting interactions (Table 20). Vendor recommendations were Process 2: In order to manufacture GMP supplies, followed for use of the media and reagents as applicable, Process 1 was modified to manufacture A-Gene at the with some variability introduced to understand response 200-L scale instead of the 50-L scale used for Process 1. of the process. Experiments assessed cell growth, ability Process 2 was largely the same as Process 1 except for the to produce AAV, and product quality. A standard trans- seed train and N-1 stage, which was conducted in slightly fection protocol used for Process 1 and 2 was employed. larger volumes to ensure sufficient biomass to operate All transfection related factors were held constant for at the larger scale. No additional process development this evaluation. was undertaken. Similarity of the material produced for The primary criteria to evaluate performance was toxicology and clinical studies was assessed on the basis productivity (vg/mL). The proportion of empty capsids of product quality attributes used for release and charac- and potency were evaluated, but these product quality terization of the material. attributes can be sensitive to the sample preparation Process 3: As the A-Gene program approached late- method. Thus, the experiment was conducted in bio- stage, a more comprehensive assessment of the process was reactors (2-L working volume) to ensure that sufficient conducted to determine how the process may be optimized material was available for in-depth, multi-step sample to improve the productivity of the process. It was intended preparations that reduce yield but are more representa- that the resulting process could then be characterized and tive of the larger scale purification process for A-Gene. ideally commercialized, so a thorough consideration of all Medium 2 (SuperExpress Medium) was chosen as the process parameters was needed. At this time, a preliminary basal medium and dilution medium for the transfection list of CQAs was also available for the drug substance, rep- mixture. The SuperTransfect transfection reagent was resenting a conservative assessment of quality attributes chosen with Nutrient Mix 1 post transfection to boost that may be critical to safety and efficacy of A-Gene(Table productivity. 18). Additionally, important process attributes that are Following the selection of medium other aspects of linked to CQA’s had been characterized(outputs). the process were assessed, including:

CHAPTER 5 Upstream and Downstream Processing 122 Table 5-18. Process Attribute Impact and Mitigation Strategy Template

Attribute Mitigation Severity Probability Detectability Risk Score Productivity Proportion of full capsids Capsid protein purity VP ratio Vector genome sequence (inclusive of ITRs and all other components and the GOI itself) Vector genome species Potency Infectious titer Residual plasmid DNA/fragments Residual E1A oncogene DNA Capsid degradation and modification Capsid PTMs Pre-transfection doubling time Viability at transfection Cell density at harvest Viability at harvest AAV titer at harvest

GOI, gene of interest; ITR, inverted terminal repeat; PTM, post-translational modification; VP, viral protein.

• Total plasmid DNA per cell of Process 2. Performance of the optimized process at • Ratio of the Rep/Cap to transgene to helper plasmid varying generations from vial thaw were also assessed at • Plasmid to PEI ratio this time (Table 22). • Transfection cell density Culture productivity, reported as vg/mL at the time of • Pre-transfection additives harvest, was a key performance parameter. While batch • Complexation volume yield is an important business objective, this was consid- • Time of harvest ered in the context of scaling up or scaling out. Process development assessed whether scaling up from the 200-L The pH, temperature, and dissolved oxygen level of scale to a larger scale was warranted post approval as the culture were not included in this study because pre- demand increases or whether extending manufacturing vious experience has shown little benefit in varying these campaigns at the 200-L scale in a given year would meet relative to established values. A statistically designed projected peak demand. experiment was executed to understand the impact of Scaling up presents challenges associated with the factors as well as interactions (Table 21). engineering aspects of the process (mass and energy Following the completion of the experiment, the most transfer) as well as cell biology (ensure transfection productive conditions were repeated alongside Process efficiency does not decline at larger volume). Facility- 1 to generate more data and confirm the performance related operational challenges associated with handling

CHAPTER 5 Upstream and Downstream Processing 123 Table 5-19. Parameters Optimized to Define the Commercial Production Bioreactor Step

Product quality Parameter Process impact impact and risk Media Basal medium and media additives impact productivity (e.g., compensate for any noted nutrient deficiencies or facilitate cell maintenance) Pre transfection Temperature can impact cell metabolism and growth rate temperature Pre transfection pH pH impacts metabolism and growth rate Dissolved Oxygen While maintaining a minimal level of dissolved oxygen in the bioreactor is essential for cell growth, the level of excess dissolved oxygen maintained has been shown to affect the productivity of the culture Agitation Agitation impacts the mass transfer of oxygen to meet cellular demand and can also impact cell clumping. At lower cell densities, mass transfer is not a significant issue, but the shear stress to which the cells are exposed needs to be considered. The level of shear protectant (e.g., Pluronic® F-68) may be addressed to mitigate stress due to mechanical shear

Gas flow Gas flows impacts oxygen mass transfer and CO2 stripping. Excessive sparging can also lead to bubble-related shear stress in the cell Post transfection pH pH impacts growth rate andmay impact productivity after transfection Post transfection Temperature impacts growth rate and may impact productivity after temperature transfection Transfection Selection of transfection reagents to maximize transfection efficiency is a reagents key aspect of maximizing AAV productivity. Commonly used transfection agents include calcium phosphate, PEI, and commercial reagents (e.g., Lipofectamine, ViaFect). Rather than selecting a platform transfection agent, a platform approach to selecting the transfection reagent may be used. In addition to the transfection reagent used, the ratio of this reagent to the total plasmid present as well as the bioreactor conditions during transfection should be considered Plasmid ratios The ratio of the three plasmids utilized (pAAV-GOI, pHelper and pRepCap) were optimized. Plasmid ratio may impact the percentage of full capsids Total plasmid DNA Total plasmid DNA may impact the percentage of full capsids Inoculation density Inoculation density can impact cell metabolism and growth rate Transfection cell Cell density and cell health at transfection play important roles in density productivity optimization. Allowing cells to achieve transfection cell density or to grow past the target transfection density and diluting down to the target density with fresh media were considered Dilution to Allowing cells to grow past transfection cell density in order to then dilute transfection density to transfection density allows the provision of additional nutrients at the time of transfection Cell age/ passage Generations from vial thaw number Cell thaw time Cell thaw time may affect viability Anti-clumping agent Anti-clumping agents were considered to improve transfection efficiency. These agents may be incompatible with the transfection reagent Anti-foam addition Anti-foam agents were considered to improve transfection efficiency. These agents may be incompatible with the transfection reagent Harvest time The time of harvest was selected to maximize productivity of the culture by maximizing the synthesis of fully formed capsids with packaged GOI, while also considering the potential degradation of capsids post transfection

CHAPTER 5 Upstream and Downstream Processing 124 Table 5-20. Factors and Conditions for Experiments

Factor Conditions Medium Medium 1 Medium 2 Medium 3 Transfection reagent SuperTransfect Complexation medium Medium 1 Medium 2 Medium 3 Post transfection medium addition Nutrient Mix 1 Nutrient Mix 2

Table 5-21. Experimental Parameters

Parameter Low Mid High Transfection density (x106 cells/mL) 2 3 4 Plasmid ratio 1:1:1 Total plasmid per cell pg/cell 0.5 1 2 PEI to plasmid ratio (mg:mg) 2:1 3:1 4:1 Additive Sodium butyrate Feed addition post transfection (v/v) 1% 2.5% 5% Time of harvest post transfection (h) 48 72 96 larger volumes while minimizing hold time may also be are pooled, and determining pooling criteria for lots are a consideration. However, experience from large-scale important. It is important to understand the definition recombinant protein manufacturing can be applied. of a batch from a quality perspective and regulatory Scaling out presents different operational challenges considerations. related to maintaining increased batch frequency through the year. Additionally, if a high dose is required, multiple Process Characterization bioreactor batches may need to be pooled to generate a The goal of process characterization for the production single batch with an appreciable number of doses. This is bioreactor were: desirable because it enables purification of a batch with multiple doses as opposed to a single dose through a • Identify process parameters that impact onto product single downstream train. Under these circumstances, quality and yield; understanding the consistency of batches that are pooled, • Justify manufacturing operating ranges and accep- establishing the point in the process at which batches tance criteria;

Table 5-22. Process Parameter Conditions

Parameter Level 1 Level 2 Level 3 Process Conditions Process 2 (Ctrl) Optimum 1 Optimum 2 Generations from MCB thaw 10 20 40

CHAPTER 5 Upstream and Downstream Processing 125 Table 5-23. Process Attribute Template During the Production Bioreactor Unit Operation

Unit Operation Attribute Mitigation Severity Probability Detectability Risk Score PRODUCT QUALITY ATTRIBUTES Proportion of full capsids Capsid protein purity Vector genome sequence (inclusive of ITRs and all other components and the GOI itself) Vector genome species Potency Production bioreactor Infectious titer Residual plasmid DNA Residual E1A oncogene DNA Capsid degradation and modification Capsid PTMs PROCESS ATTRIBUTES Pre-transfection doubling time Viability at transfection Productivity Viability at transfection Cell density at harvest Viability at harvest AAV titer at harvest

• Identify interactions between process parameters potential critical attribute was risk ranked based on and critical quality attributes; and severity, probability, and detectability. The outcome of • Ensure that the process delivers a product with the first risk assessment included the unit operations reproducible yields and purity. to be evaluated for the process as presented earlier and that resulted in the process used for manufacturing Risk Assessment 1: Prior upstream knowledge, process material for phase 3 clinical studies. Table 23 lists the understanding, and quality attributes that may be im- process attributes that can be influenced during the pacted by the upstream process have been described in production bioreactor unit operation. earlier sections of this chapter. These led to the initial risk assessment conducted prior to late-stage process Risk Assessment 2: A second risk assessment was development, as discussed during the description of completed following late-stage process development Process 3 development. Critical quality attribute as- activities for Process 3. Initial assessment of process sessment was also conducted at this time (as addressed parameters that may influence quality attributes com- in Chapter 4 and Chapter 5). Hence, the appropriate pleted during process development of Process 3 was subset of critical attributes was selected for assess- used as input. The risk assessment is focused on identi- ment during process development for Process 3. Each fying bioreactor equipment design, control parameters,

CHAPTER 5 Upstream and Downstream Processing 126 Table 5-24. Process Parameter Template

Risk Ranking Experiment Setpoints Baseline Process Parameter Full/ AAV Parameter Empty Potency Totals Low Mid High Titer (%) Temperature setpoint DO setpoint VCD at infection Inoculation VCD Agitation (RPM) Hours post transfection Air sparge setpoint Air overlay setpoint Media warming duration Antifoam – daily (mL) Media lot Transfection ratio processing conditions, and starting materials that may on process characterization studies conducted using a have a significant influence over quality attributes of the qualified scale-down model of the production bioreactor. product (Table 24). A risk ranking score is assigned to Multivariate and univariate studies (DOE) designed as each process parameter with respect to their potential to an output of the second risk assessment were executed affect a particular process attribute. as part of process characterization experiments. These Risk mitigation activities are designed to include studies helped quantify the relationship between pro- process parameters with a high risk assessment score. cess parameters and critical quality attributes. Process These activities include some or all the following: parameters ranked either high or medium in the above risk analysis were defined as factors in D-optimal study • DOE: multivariate studies to establish relationships design. The ranges for each factor were low and high ex- between parameters and CQAs perimental setpoints recorded in the risk analysis above. • DOE indirect: parameters that were indirectly varied Critical quality attributes and measures of productivity during DOE studies influenced by the production bioreactor step are defined • EOPC: end of production cell studies to establish as responses for the study. The study design was suffi- limits of in vitro cell age ciently powered to resolve any main effects, effects from • Medium hold studies: studies performed to justify interaction between factors, and quadratic effects to assess medium and feed hold times curvature in the responses. • Not required: indicates that no special risk mitiga- For each response, model fitting is performed using tion was performed; parameters were controlled and data collected from study execution. Outliers are iden- recorded, and data were retrospectively analyzed for tified using studentized residuals or adjusted jackknife correlations. distances. Data points that score >1.96 absolute value are identified as potential outliers with 95% confidence Process Characterization Studies: Operation design and are removed from model fitting.Figure 6 contains a space and process control strategy were defined based matrix of plots indicating each of the effects found in the

CHAPTER 5 Upstream and Downstream Processing 127 Table 5-25. Parameter Estimates for Fitted Models to Responses

CQA 1 CQA 2 CQA 3 Term Estimate P value Estimate P value Estimate P value Factor 1 305.24 <0.001 -0.5 <0.001

Factor 2 * Factor 4 67.17 <0.001 2.34 <0.001

Factor 1 * Factor 1 15.32 0.0104 0.045 <0.001

Factor 3 7.63 <0.001 0.11 <0.001 -1.98 0.0116

Factor 3 * Factor 4 5.0 <0.001 -0.09 0.0190 1.50 <0.001

DOE. This matrix plot is made by combining statistical The design space for the production bioreactor unit models generated for individual responses. Table 25 operation was defined by combining models for each summarizes the process parameters and factor interac- studied response and the levels of CQAs. Acceptable tions found to significantly affect CQAs (i.e., P≤0.05). CQA levels were decided during critical quality attribute Scaled estimates provide a measure of how much a given assessment as described in the CQA section above. The response changes as a function of an input parameter, or ranges for CQA values are based on safety and efficacy combination in the case of interactions. These estimates data and account for the analytical method used. For form the coefficients of the transfer function for each example, cell-based assays are often more variable than response. The models are suitable for predicting mean capillary- or PCR-based quantification methods. The levels of the CQAs over the ranges of the process param- design space defines a multivariate space within the eters that were included in the experiments. defined ranges of process parameters that provides high

Figure 5-6. Combined Profiler

CHAPTER 5 Upstream and Downstream Processing 128 Table 5-26. Levels of CQAs Used to Define Design Space

Critical quality attribute Lower limit Higher limit

CQA 1 70 140

CQA 2 50 100

CQA 3 90 100

confidence that all CQAs will be within acceptable levels acceptable limits. as indicated in Table 26. Once a design space is defined for the production Figure 7 shows the graphical representation of the de- bioreactor unit operation, edge of failure analysis can be sign space for the production bioreactor unit operation. performed by using the combined model and stimulating The shaded region in these 2D graphs represent process CQA values based off variation in process parameter parameter values that would lead to mean levels of CQAs inputs. This analysis uses the fitted model, not the ex- that will be outside the acceptable limits or specifications. perimental data to stimulate response values. It uses a The colors represent different CQAs that may fall outside Gaussian process statistical model with the failure rate

Figure 5-7. Design Space

CHAPTER 5 Upstream and Downstream Processing 129 Figure 5-8. Graphical Representation of Simulated Response Values for the Production Bioreactor Unit Operation

from the computer experiment as an interpolator to pre- to changes in CQAs by increasing the standard deviation dict and visualize the results of the experiment. A data by a factor of 1.5 to 2. While performing margin analysis table is created with the design and the estimated overall by varying the factors, representative critical factors can failure rate is reported in PPM. Figure 8 is a graphical be evaluated by increasing their standard deviation and representation of the simulated response values for the examining impact on CQAs. Proven acceptable ranges production bioreactor unit operation. Green dots repre- (PAR) are identified by the range in standard deviation sent acceptable CQA values and red dots represent CQAs of critical factors that limits failure PPM limits to <100. that are outside the acceptable range (Table 26). This ap- proach satisfies the requirement to demonstrate assurance Risk Assessment 3 and Control Strategy: A final risk of quality per the ICH definition of design space. assessment was conducted after the completion of the The design space for the production bioreactor shown process characterization studies to define the control in Figure 9 is multidimensional. To define limits of the de- strategy for the commercial manufacturing process. sign space that can be translated to clear manufacturing in- Commercial-scale batch performance was used to verify structions, predictive models for each response were used process performance and demonstrate the control strate- to set limits. This is achieved by examining the sensitivity gy at scale. The risk assessment identified unit operations

CHAPTER 5 Upstream and Downstream Processing 130 Figure 5-9. Design Space for Production Bioreactor

that had an impact on product quality and process per- important to ensure successful and reliable commercial formance. Unit operations that were identified to impact manufacturing operations. product quality were then included in defining the design space for the upstream process. Equation 1. Tolerance. Scaled estimates as noted in Table 27 are used to identify process parameters linked to product quality and process performance. Scaled estimates are scaled based on the ranges tested in the DOEs so that they The proposed control strategy for the production bio- measure change in the response value by half-range. reactor unit operation ensures that the process delivers a The full effect of each factor or interaction is calculated product that meets its specifications, and a consistent and by doubling its scaled estimate. Scaled estimates for robust commercial manufacturing process. Product quality quadratic terms measure change over the full range is ensured by operating the process within the limits of and do not need to be doubled to calculate their full the design space (i.e., all quality-linked process parameters effect. Classification of parameters was based on their [CPPs] must operate within the defined PARs). Process potential impact to product quality, the likelihood consistency is ensured by controlling PPs within estab- of a parameter to exceed acceptable limits, and the lished limits and by monitoring relevant process attributes. ability to detect and/or correct a failure if it occurred. A summary of the control strategy for the production Tolerance (%) was calculated using the following bioreactor is presented in Figure 10. Here, quality-linked equations. For CQAs, terms with tolerance values process parameters must be controlled within the design ≥20% were identified as CPPs, as shown in Table 27. space and in-process quality attributes must be within For productivity measures, terms with tolerance values specified limits to ensure drug safety and efficacy. The ≥20% were identified as PPs (Process Parameters, not control of PPs ensures that commercial success criteria shown here). PPs do not impact product quality but are and yield are met.

CHAPTER 5 Upstream and Downstream Processing 131 Table 5-27. Scaled Estimates

Term Estimate Prob>Itl Multiplier Full Effect % Tolerance CPP Speed(100,200) -0.017007585 <.0001 * 2 -0.034015169 56.7 CPP Temp(250,300) 0.0654243417 <.0001 * 2 0 .1308486833 218.1 CPP Time(5,10) 0.0176018576 <.0001 * 2 0.0352037153 58.7 CPP Pressure(15,30) .0143259292 <.0001 * 2 0.0286518585 47.8 CPP Speed*Temp -0.001554387 0.0190 * 2 -0.003108773 5.2 non-critical Temp*Temp 0.0159445836 <.0001 * 1 0.0159445836 26.6 CPP Temp*Time -0.037915132 <.0001 * 2 -0.075830263 126.4 CPP Time*Time 0.0138420999 <.0001 * 1 0.0138420999 23.1 CPP Temp*Pressure -0.025754387 <.0001 * 2 -0.051508773 85.8 CPP Time*Pressure 0.0346654026 <.0001 * 2 0.0693308051 115.6 CPP

Applicability of Design Space predictive of large-scale manufacturing. This ensures that to Multiple Operational Scales the process characterization executed using the scale- down model is applicable to large-scale manufacturing. and Bioreactor Configurations: The following section describes methodology for the Engineering Design Space selection of parameters used to scale up and scale down the manufacturing process for A-Gene and qualification The design space previously described is based on the of the scale-down model. quality-linked process parameters. These parameters are considered scale-independent variables and thus apply DEVELOPMENT OF A SCALE-DOWN MODEL to all operational scales. However, other scale-dependent The levels of productivity and product quality that are parameters must be considered for successful and con- achievable depend primarily on bulk mixing, oxygen sistent process performance when operating at various mass transfer, and hydrodynamic conditions, which in scales. The engineering design space includes bioreactor turn are affected by bioreactor design, impeller type, and design characteristics and engineering parameters to process operation. To accomplish successful scale-up, ensure robust and consistent bioreactor performance to reasonable similarity between these conditions must be meet product quality targets. maintained across scales. It follows that creating a suc- The engineering design space includes bioreactor de- cessful scale-down model requires that any limitations sign characteristics and engineering parameters for small observed at the large scale are also reflected at the small scale bioreactors used for process development and large- scale, even if these need to be deliberately imposed. scale bioreactors used for manufacturing. During early For example, a limitation in the maximum permissible development, scale-up of the process from small-scale oxygen transfer at large scale may be deliberately im- bioreactors used for development to larger-scale biore- posed at the small scale to reflect equipment capability actors used for manufacturing is of concern. However, at the large scale. In some cases, the bioreactor scale or as the scale for commercial manufacturing is established, configuration utilized for manufacture of toxicology, the small-scale bioreactor model is also adjusted to clinical, and commercial material is not known early in ensure it is representative of the performance observed development. In this case, the small-scale model for early at large scale. Additionally, as a scale-down model is development may use prior knowledge from other pro- established it is qualified by demonstrating that process grams or vendor recommendations regarding operating performance at the small scale is representative and parameters. As information regarding manufacturing

CHAPTER 5 Upstream and Downstream Processing 132 Figure 5-10. Overview of Control Strategy for Production Bioreactor

Critical Process Parameters BIOREACTOR UNIT OPERATION Critical Performance (CPPs) Attributes (CPAs) Process Performance CPP parameters attributes CPA Factor 1 (PPs) PAs) Assay 1 Factor 2 Factor 3 Titer Viability Assay 2 Factor4 Factor 6 Assay 3

Controlled for ability to impact Controlled within specified ranges Outputs corresponding to CPAs and/pr COAs to confirm consistent process a property or characteristic performance linked to ≥ 1 COAs scale and bioreactor configuration becomes available, the Establishing the scale-down model involves matching scale-down model can be adapted to ensure it remains mass transfer and energy dissipation across scales, as well representative and predictive of large-scale operation. as the impact of engineering design parameters. The Table 28 describes primary and secondary reactor maximum mass transfer may be matched by matching design features that should be evaluated to assess the the maximum achievable mass transfer coefficient (kLa) bioreactor capability to support a high-density mam- across scales. Since the kLa depends on mechanical en- malian cell culture for recombinant protein expression. ergy input as well as sparging rates, matching the kLa This matrix was developed based on published literature, essentially involves selecting the appropriate agitation bioreactor engineering industry best practices, and rate, gas flow, and sparger type. extensive prior experience with cell culture operations As mentioned earlier, initial small-scale bioreactors at multiple scales. Secondary parameters have a lesser used for development may not be based on scale-down impact on bioreactor performance capability, whereas models of larger-scale bioreactors and use of vendor or primary design parameters have a direct impact. literature references to select appropriate agitation and Because the A-Gene process does not involve high sparging parameters. However, as the specific knowledge density cultivation, this affords us more flexibility in the regarding the large-scale bioreactor system becomes selection of engineering parameters since the range of available, the small-scale model may be revised to im- parameters can achieve favorable conditions for growth pose limitations observed at large scale. For A-Gene, the and productivity. small-scale bioreactor parameters were adjusted to lower A thorough discussion of scale-up and scale-down the maximum achievable oxygen mass transfer. This was criteria and related bioreactor design is available in the achieved by appropriately lowering the maximum gas A-Mab case study.8 flow rates. At the same time the agitation rate was in- creased to mimic the increased hydrodynamic shear (any ESTABLISHING A SCALE-DOWN MODEL AND increase in oxygen transfer was offset by lower gas flow DESIGN SPACE APPLICABLE TO MULTIPLE rates). A template of bioreactor characteristics for various OPERATIONAL SCALES scales of operation for A-Gene is shown in Table 29. This Since clinical and commercial demand for A-Gene was represents the engineering design space for A-Gene. expected to be manageable with production at 200-L scale, disposable bioreactors were chosen as the hard- QUALIFICATION OF SCALE-DOWN MODEL FOR ware platform for commercial manufacturing. This was PRODUCTION BIOREACTOR AND ENGINEERING advantageous because most configuration details for DESIGN SPACE preassembled disposable bioreactors that impact scale-up To demonstrate the applicability of the scale-down model were readily available from vendors and could be used to predict large-scale production bioreactor performance, to establish a scale-down model early in development. process and product quality attributes are assessed. The

CHAPTER 5 Upstream and Downstream Processing 133 Table 5-28. Impact of Production Bioreactor Engineering and Process Parameters on the Manufacturing Process

BIOREACTOR PERFORMANCE INDICATORS 2 stripping time 2

DESIGN PARAMETERS shear Hydrodynamic Bubble shear per volume Power Mixing time heterogeneity Culture (kLa) Gas transfer Superficial gas velocity Gas holdup volume CO control Bioburden pH DO pCO Temperature Osmolarity Bioreactor aspect ratio P P S P P P S P P Baffles P P S Impeller design/size P P P P P P S Number of impellers P P P P P P S Agitation rate P P P P P P S S S S S S S S Gas composition, flow rates, control P P P P P P P P Sparger design and location S P P P P P P P S P P Location of addition ports/tubes P S Feed addition rates P S S S Vessel pressure P S S S Probe locations S P P S P S DO control loop S S P S S pH control loop S P S S S Temperature control loop S S S P Foam control P P S

P=Primary design consideration expected to impact bioreactor capability. Impact assessment based on prior knowledge, engineering fundamentals, and/or modeling studies (e.g., Computational Fluid Dynamics). S=Secondary design indirectly impacting bioreactor capability based on prior knowledge and engineering standard design. initial scale-down model used for process development compared by examination of growth profiles (viable was qualified based on performance data collected from cell density, viability) as well as metabolites (lactate, prior programs (X-Gene and Y-Gene). Both processes ammonia). The primary options available to achieve this used the same cell host along with a similar cell culture assessment included: and transfection process. For the qualification studies, scale-independent variables (pH, temperature, iVCC, • Examination of process performance profiles one DO, culture duration, etc) in the scale-down bioreactors parameter at a time with simple statistical measures were operated at the proposed target process values of to assess comparability of the data. This may involve commercial operations. For scale-dependent parameters simple comparisons of datapoints across batches at the (agitation, gas flow rates, pressure, volume, etc), operat- same normalized timepoint. Alternatively, the pa- ing conditions at small scale were established to match rameter profile may be fit to a model (curve) and the process performance at full-scale as described in this similarity of the curves for different batches may be section. compared using appropriate statistical methods. This Performance of the processes across scales was mitigates the issue of not having data at corresponding

CHAPTER 5 Upstream and Downstream Processing 134 Table 5-29. Template for Bioreactor Design and Engineering Characterization at Various Scales of Operation for A-Gene

2L Standard 2L Scale-down Model (Process Development Model 200 L (Process 3) Characterization) (Process 3) Nominal volume (L)

Working volume (L)

Aspect ratio (H:D)

Impeller design

Number of impellers

Sparger design

Average P/V (W/m3)

Max local P/V (W/m3)

Vs (x10-3 m/s)

Mixing time (s)

CO2 stripping time

Gas hold-up volume (L)

culture times across scales. While not as sophisticated of the data in a way that best explains the variance as multivariate methods, this approach may be in the data. Thus, this multivariate approach rep- sufficient depending on the number of parameters and resents a powerful means to assess if the correlation quality attributes being compared. structure between key performance attributes and quality attributes in the scale-down model data is • Multivariate analysis using principal component comparable to results from full-scale bioreactors. analysis (PCA) modeling. PCA transforms a large The resulting model is more sensitive than com- number of possibly correlated variables into a monly used univariate comparisons (e.g., t tests) smaller number of uncorrelated variables called because it can detect observations that do not fit the principal components that are formed with different predicted response patterns while resulting in fewer loadings of the original variables. The first principal false-positive signals. component accounts for as much of the variabil- ity in the data as possible, and each succeeding The A-Gene process has been successfully run in bio- component accounts for as much of the remaining reactors from 2-L to 200-L working volumes and various variability as possible. The strength of this approach design configurations. The primary design parameters can be viewed as revealing the internal structure described earlier (aspect ratio, impeller design and

CHAPTER 5 Upstream and Downstream Processing 135 Table 5-30. Template of Process Performance and Product Quality for Various Scales of Operation for A-Gene

2L Development Model 22L scale-down model (Process 3) (process chacterization) 200L (Process 3)

Proportion of full capsids

Capsid protein purity Vector genome sequence (inclusive of ITRs and all other components and the GOI itself) Vector genome species

Infectious titer

Residual plasmid DNA

Residual E1A oncogene DNA

Capsid degradation and modification

Capsid PTMs number, baffles, addition port location, and sparger -de notion that wide ranges of engineering parameters can be sign) were verified during A-Gene process scale-up from used and have been demonstrated to be acceptable, both 2-L to 200-L bioreactors to ensure that the bioreactors in terms of process performance and product quality. could support the A-Gene process. Similarly, process en- Table 30 allows comparison of process performance gineering parameters (P/V, superficial gas velocity, kLa, across all scales and determination of whether product mixing time, and gas holdup volume) were measured quality is within the predicted design space. These results and confirmed to meet A-Gene process requirements. should demonstrate whether the design space defined Information from 2-L scale-down characterization stud- using scale-down data accurately predicts performance ies (using Process 3) provides additional support to the at various operational scales.

CHAPTER 5 Upstream and Downstream Processing 136 Downstream Process Figure 5-11. AAV Purification Process Flow Diagram Description and Cell Culture Process Characterization Harvest/Lysis Summary The commercial manufacturing process for downstream Depth Filtration purification steps are designed to remove and/or control process-related impurities such as host cell proteins, residual unencapsulated DNA, process additives (en- Affinity Chromatography donucleases), and product-related impurities such as aggregated vectors and empty capsids that are generated during the upstream process. The scalable AAV puri- fication process contains orthogonal steps designed to Anion Exchange Chromatography remove impurities and concentrate the viral vectors before final formulation. Formulation focuses on the identification of suitable buffers and excipients to allow Ultrafiltration/Diafiltration for long-term storage of AAV product as well as suitable conditions for delivery to patients. AAV process development activities often rely heav- ily on empirical determination of process raw materials Sterile Filtration and set points. Historically, process development often emphasized the definition of setpoints and conditions for the process through well controlled single variable exper- iments. However, these experiments provided limited or Drug Substance no information as to the robustness of the bioprocess to deliver the specified product. Additionally, the interactions of separate input variables are poorly understood and/or the likelihood that high‐risk issues escape attention the impact of perturbations to the system are unknown. during downstream development. QbD is a more systematic, goals-focused approach For example, when mapping the parameters that im- that leverages both historical knowledge and results pact the quantity of empty capsids in the final dose, the through experimental design (DOE) and utilizes qual- pH of the AEX elution step, column loading, and starting ity risk management during the development cycle. A empty capsid ratio could all be considered. As such, an QbD approach can increase process robustness through experimental design that explores these conditions can knowledge of both what quality attributes are critical identify the impact that the parameter has on the CQA and what process parameters are the most relevant to and aid in the further development of ranges in which those attributes; facilitate process transfer across facilities the step could be performed and produce material that and scale, and thus decrease regulatory burden; facilitate meets the predefined QTPP within normal variation control strategies to enable more consistent products; from upstream processes. All steps are performed streamline lifecycle management; and decrease the like- within a risk assessment and management system that lihood of failure across all stages of processes by reducing uses previous process knowledge to define the stages. As

CHAPTER 5 Upstream and Downstream Processing 137 Table 5-31. Overview of Downstream Process Steps

Downstream Step Purpose of Step

Lysis Nucleic acid degradation Depth Filtration Remove cell debris, host cell proteins, large aggregates Prevent fouling of subsequent downstream processes Affinity exchange chromatography Remove impurities (e.g., host cell proteins [HCPs], unpackaged DNA, protein aggregates, viruses) Anion exchange chromatography Remove impurities (e.g. HCP and hcDNA) Remove empty capsids Enrich viral vector Tangential flow filtration (ultrafiltration/ Exchange in the final formulation buffer diafiltration) Sterile filtration Bioburden control with any risk‐based approach, CQAs or CPPs should be Process Understanding Based on continuously updated or monitored to ensure all process knowledge is captured. For additional information on Prior Knowledge QbD principles, please refer to Chapter 4. Utilizing extensive prior knowledge, an initial risk The downstream process entails the clarification and assessment for our hypothetical A-Gene product was purification of A-Gene with a goal to generate the final conducted to identify which downstream process steps clinical product with high potency, purity, and titer. For potentially impact product quality. For details of how to this case study, assumptions are a yield from the 200-L conduct a risk assessment, refer to Chapter 4. suspension cell culture of ~1x1011 vg/mL, with a final Based on the results of the risk analysis, for the pur- concentration of ~1x1013 vg/mL. poses of this case study in this chapter, only a subset of The downstream manufacturing process for A-Gene quality attributes are considered: comprises several steps that are presented in the flow di- agram Figure 11. The purpose of each step and the scope • Aggregate of information included in the case study are summa- • % Full particles rized in Table 31. Detailed step descriptions and process • Host cell DNA performance analyses are presented in the sections that • Residual HCP describe each step. The purification process consists of: • Residual unpackaged DNA (e.g. rep/cap, helper) • Capsid protein purity • Harvest (in upstream processing) • Nucleic acid degradation by enzymatic treatment By contrast, extensive prior knowledge has demonstrat- • Capture by affinity-based purification ed that the distribution of viral proteins (VP1, VP2, and • Polishing by anion exchange chromatography VP3) is minimally impacted by downstream processing • Concentration and is mainly influenced by the upstream process condi- • Formulation (for more information, please refer to tions. Based on this assessment, viral protein ratios were Chapter 6) not included in the testing for characterization studies of • Fill-Finish (refer to Chapter 6) the downstream process steps. Viral clearance and process residuals (e.g., affinity ligands, Benzonase) were also -in Throughout these steps, output parameters include cluded in the downstream process discussion. In an actual yield, empty/full capsid ratio, aggregates, and other study, the examples and approaches described here would impurities (e.g., HCPs, host cell DNA). include all relevant product quality and material attributes.

CHAPTER 5 Upstream and Downstream Processing 138 VIRAL CLEARANCE For early phase clinical studies viral clearance is generally Terms in the Downstream section not required for HEK293-transient plasmid transfection Process inputs are measurements that can be process platforms. Viral clearance studies are required for directly manipulated or controlled and are classified AAV manufacturing platforms that do have a relevant based on their impact on process performance and process virus, such as baculovirus, HSV or adenovirus product quality. Process inputs are classified as based systems, and should include at least viral clear- described below: ance studies on process relevant virus for phase 1/2 and • Critical Process Parameters (CPPs) are inputs additional model viruses for a BLA. The focus of this controlled for their ability to impact in-process chapter is on manufacturing requirements for earlier CPAs and/or drug substance CQAs. CPPs must be controlled within specified acceptance stage studies. Viral clearance studies may be required for criteria to ensure that drug substance CQAs are marketing applications. In general three model viruses achieved. Confirmed excursions are investigated are sufficient. As noted in the introductory section of this for product quality impact and could lead to lot document, the reccomendations made here should not rejection. • Process Parameters (PPs) are inputs controlled be taken as regulatory guidance. for their ability to impact in-process PAs. PPs must be controlled within specified ranges to Viral Safety Risk Assessment confirm consistent process performance. PPs An outline of the risk assessment conducted to assure have either designated action ranges or alert ranges. Excursions are investigated according to viral safety is summarized here: established procedures. • A-Gene is produced in HEK293 cells using animal component–free (ACF) growth medium, nutrient Monitored Parameters (MPs) are input parameters that are unrelated to product quality, do not have feeds, and supplements. In addition, HEK293 is a established ranges, and are used to monitor the well characterized cell line used for the production of manufacturing process. MP results are reported, other clinical monoclonal antibody products. trended, and monitored. • The A-Gene master cell bank (MCB) and working Process outputs are measurements that cannot be directly manipulated or controlled, such as cell bank (WCB) were characterized and shown to be in-process measurements, and are indicators of free from adventitious virus contaminants. process performance and product quality. Process outputs are classified as described below: Measures are in place to ensure the safety of raw • Critical Performance Attributes (CPAs) are outputs materials used in the manufacturing process. Any ani- corresponding to a property or characteristic mal-derived components used in the medium preparation linked to one or more drug substance CQAs. of the research cell bank, the MCB, and the media for cell CPAs are in-process results and have designated acceptance criteria. Confirmed excursions are cultivation are sourced from low BSE-risk countries that investigated for product quality impact and could have bans in place against ruminant-to-ruminant feeding. lead to lot rejection. • Performance Attributes (PAs) are outputs BATCH HISTORY monitored to confirm process performance and consistency. PAs are in-process results and The downstream platform process did not require any have either designated action ranges or alert significant changes to accommodate the increased pro- ranges. Excursions are investigated according to ductivity of the cell culture process or facility changes established procedures. made through the development life cycle. The only Monitored Attributes (MAs) are outputs unrelated changes made to the downstream process represent to product quality that do not have established scale increases to match the upstream process scales. The ranges and are used to monitor the manufacturing A-Gene batch history is summarized in the upstream process. MA results are reported, trended, and monitored. process section.

CHAPTER 5 Upstream and Downstream Processing 139 Downstream Process LYSIS AND CLARIFICATION The lysis and clarification unit operation, consisting of Characterization a non ionic surfactant treatment followed by filtration, bridges the production bioreactor and affinity capture The following sections describe the approaches used to (AC) chromatography steps. The production bioreac- identify parameters linked to product quality and pro- tor step provides a consistent product pool containing cess performance that serve as the basis for defining the ≤1x1013 viral genomes/mL. The lysis and clarification design space for each process step. The classification of operation consistently provides a process stream at pH process parameters used in this section is based on the 5.0±0.2 to the affinity capture chromatography unit decision logic presented in the Control Strategy section. operation.

Table 5-34. Lysis and Clarification Step Linkages

Input: Eluate from Production Bioreactor Output to Affinity Capture (AC) Chromatography Viral genome concentration ≤1x1012 vg/mL Viral genome concentration ≤1x1012 vg/mL pH >4.0 pH=7.0±0.2 Aggregate 106 ng/mg 105-106 ng/mg % full AAV capsid=15% % full AAV capsid=15% Capsid protein purity >90% Capsid protein purity >90%

Table 5-35. A-Gene Lysis Parameters

Parameter/Step Description Value Range Units Comments Pre-addition volume spike basis (e.g., if 100 L of bulk Surfactant 10% (w/v) 2.86 N/A % (w/v) harvest, add 2.86 L of Triton addition surfactant stock for final volume of 102.86 L) Nuclease Endonuclease, Grade I 10 N/A U/mL ≥99% purity addition pH 2 M tris base, Target 6.8-7.2 pH - adjustment 0.001% (w/v) poloxamer 188 7.0 Target Recommend lysis hold within Temperature 35-39 °C 37 the SUB for better control

Target Hold time starts when the Time 4-6 hr 4 following conditions are met: Hold conditions (1) all lysis components are added; (2) pH adjustment Target P/V Agitation rate 5-7 target is reached; and 6 (W/m3) (3) temperature reaches 37°C±2°C Above addition values should give a final lysis condition of 0.25% (w/v) non ionic surfactant, Other comments 10 U/mL Endonuclease, and pH 7.0±0.2

CHAPTER 5 Upstream and Downstream Processing 140 Table 5-36. A-Gene Depth Filtration Process Parameters

Parameter/step Description Value Range Units Comments + charge, cellulose fiber Larger sizes (≥0.1 m2) require a Filter and diatomasceous N/A N/A N/A POD holder earth media

Target 40-60 L/m2 If running the depth filter and Water for injection 50 sterile filters in-line, the media Pre-use flush (WFI) Target pre-use flush must be done prior 250-350 LMH 300 to connecting the sterile filter

Challenge ≤500 N/A L/m2

Load Target Flux 100-200 LMH 150

50 mM sodium Target 8-12 L/m2 phosphate, 350 mM 10 Recovery flush to be collected Recovery flush sodium chloride, 0.001% Target with the filtered pool (w/v) poloxamer 188, pH 100-200 LMH 7.4 ± 0.3 150

Differential pressure ≤30 N/A psid Pressure limits Inlet pressure ≤50 N/A psig

Measure both back pressure and differential pressure during operation. Ensure that neither Other comments limit for the filter unit is exceeded. It is recommended to run the depth and sterile filters in- line at the same time.

Table 5-37. A-Gene 0.2 µm Filtration Process Parameters

Parameter/step Description Value Range Units Comments

Filter N/A N/A N/A 0.5/0.2 μm

≤250 N/A L/m2 Operational flowrate is applicable if depth and sterile filtration operations are Challenge Load challenge Target 200- performed separately. If both operations LMH 250 400 are performed simultaneously, use depth filter operational flowrate.

Target Parameters are applicable if depth and 50 mM sodium phosphate, 8-12 L/m2 10 sterile filtration operations are performed 350 mM sodium chloride, Recovery flush separately. If both operations are 0.001% (w/v) poloxamer Target 200- performed simultaneously, use depth 188 pH 7.4±0.3 LMH 300 400 filter parameters.

Differential pressure ≤30 N/A psid Pressure limits Inlet pressure ≤50 N/A psi

Measure both back pressure and differential pressure during operation. Ensure that neither limit for Other the filter unit is exceeded. It is recommended to run the depth and sterile filters in-line at the same comments time. No pre-use flush is required for the sterile filter.

CHAPTER 5 Upstream and Downstream Processing 141 Table 5-38. Comparison of Lysis Step Performance at Various Scales

Scale Viral Genome Titer Aggregate Process Scale Yield (%) Purity (%) Factor (vg/mL) (%) Scale-down model (n=6) 100 mL 1 1.0±0.4×1011 95±2 2.6±0.1 95.3±1.2 GLP Toxicology (n=3)a 50 L 500 1.0±0.4×1011 92±3 2.7±0.3 96.5±1.0 Clinical and commercial 200 L 2000 1.0±0.4×1011 94±3 2.5±0.2 97.1±1.5 (n=3)b aAgitation rate is based on bioreactor scale-down model mixing ranges. The USP process uses a mass transfer model and growth performance to scale the mixing setpoints in rpm for the bioreactors. The rpm ranges established for the USP process along with their corresponding power by volume (W/m3) values is assessed across the lysis step. bSame as the small-scale and toxicity process.

Step Description of AAV process intermediates produced in bioreactor The lysis step is a chemically induced, cell membrane detergent treatment is traditionally used for viral clear- disruption process enabling the release of AAV particles ance in gene therapy manufacturing. from the host cell into the media for further capture and purification. Lysis operation is initiated by the addition Prior Knowledge of a pre-determined amount of a stock solution of a Lysis by surfactant treatment has been used extensively surfactant and generally between 10-100 units/mL of to manufacture gene therapies as well as many other an endonuclease to the production bioreactor, and a therapeutic agents. Moreover, the process conditions pH adjustment to pH 7.0 with 2M Tris, 0.001% (w/v) have remained essentially unchanged for these products Poloxamer 188. Conductivity is an important parameter and throughout the A-Gene development process. Thus, that should be controlled. The total bioreactor content of experience gained from the characterization of past A-Gene product and lysis chemicals should be incubated studies constitutes prior product knowledge and may be at 37°C for ≥4 hours. Agitation of the bioreactor content applied directly to the A-Gene process. Because this is must be maintained during the incubation period. The not a purification step, worst-case conditions have been resulting material post-incubation is considered crude identified to assess the stability of the AAV vector during lysate. Table 35 lists all components and volumes re- the lysis process. These worst-case operating conditions quired for execution of the lysis step. involve holding the AAV vector at a higher concentra- tion, at the highest surfactant concentration, and at a lon- Depth Filtration and Bioburden Filtration ger time and higher temperature than routinely specified Depth filtration step uses a size exclusion method for in manufacturing. Following the worst-case surfactant separating lysed cell debris from rAAV particles prior treatment, the product was tested for aggregation by to rAAV chromatographic capture. The depth filtration SE-HPLC and viral genome titer by ddPCR. step is followed by filtration across a dual layer (0.5/0.2 µm) filter for the removal of any fine particulates and Table 5-39. Depth Filtration Scale Comparison bioburden/endotoxin reduction. Table 36 and Table 37 list the operational details required for the execution Parameter Scale-down 200 L of the depth filtration and bioburden reduction steps, model respectively. Total filter area (m²) 0.0023 0.77 Viral safety is required for gene therapy products. The Scale-down factor 335 1 risk of viral contamination can be mitigated through the screening of raw materials, testing process intermediates, Flow-rate (LMH) ≤150 ≤150 and/or evaluating the effectiveness of viral removal/inac- Feed pressure (psig) ≤18 ≤18 tivation during manufacturing processes. The exposure Load factor (L/m²) ≤250 ≤250

CHAPTER 5 Upstream and Downstream Processing 142 Table 5-40. Lysis: Impact on Product Quality Study Design Rationale

Normal Worst Process manufacturing case study parameter target or range conditions Scientific rationale Surfactant 2.86±0.1 2.76 Lowest concentration is expected to result in insufficient lysis, Concentration so lowest concentration was chosen to ensure conditions for sufficient lysis during the study.

A-Gene ≤1×1011 vg/mL 1×1011 vg/mL The maximum gene concentration in the study was set at 1×1011 concentration vg/mL to assess potential aggregation

Time 60-240 240 Longer hold times are expected to result in greater aggregation. minutes minutes Temperature 35-39°C 40C Higher temperatures may result in capsid instability, greater aggregation, and changes to the purity. We studied a temperature above the normal operating condition to ensure that the AAV vector was stable under normal operating conditions. Agitation Rate 5-7 P/V 5 P/V Lower agitation rates may result in insufficient lysis (W/m3) (W/m3)

Scale-Down Model summarized in Table 41. Over time, there was a steady Scale-down models have been used to characterize the decrease in monomer content while the percentage of process performance of the lysis and clarification step. aggregates increased, consistent with results obtained Table 38 compares the scale factors, yield, aggregate, in with other gene therapies under comparable process lab-scale experiments and large-scale manufacturing. To conditions. Based on product quality considerations, no qualify the model and ensure proper performance at full critical process parameters were identified for this step. scale, mixing studies were executed across all scales to ensure efficient mixing within the established time limits. Hold Time Study The lysis for this process is performed in the production Acceptable hold time stability for the detergent-treated bioreactor so that the bioreactor SDM qualification solution after completion of the inactivation, adjust- performed during upstream process development can ment to pH 5.0 and depth filtration was performed. be used to support the use of 2-L bioreactors as a SDM for the 200-L scale bioreactors for the lysis step. The data indicate that the process is consistent and comparable Table 5-41. Product Quality Results Template across all manufacturing scales and that the laboratory for Worst-Case Scenario Studies model is representative of full-scale manufacturing operations. Process flow, operating pressure, and load Process Titer (vg/ml) Aggregate Purity by Hold Time (%) CE-SDS capacity are key scale-down parameters for this unit (minutes) (%) operation. Table 39 contains a comparison of the SDM 0 1×1011 99 filtration process with the 200 L scale process. 30 1×1011 99 Characterization Studies to Assess Impact to 60 1×1011 99 Product Quality 90 1×1011 99 Table 40 summarizes the process parameters that were 120 1×1011 99 used in the study and the rationale for the selection of 150 1×1011 99 these parameters. The experiment was executed in triplicate in a 100-mL 180 1×1011 99 laboratory model (results not shown) and the results are 240 1×1011 99

CHAPTER 5 Upstream and Downstream Processing 143 A solution containing the maximum A-Gene concen- Table 5-42. In-Process Hold Study Results tration of 1x1011 vg/mL was held at the maximum hold Template temperature of 25°C for 7 days (data not shown) and demonstrated that this process intermediate can be held Time (days) Aggregate (%) Purity (%) at 25°C for 7 days without significant degradation or 0 98 impact to product quality. 1 97 3 98 Summary of Process Parameter Classification and Ranges 5 96 7 96 Depth filtration range was established based on a worst case scenario to support lysis studies. The worst case scenario included high load challenge and flowrate impurities such as HCP, DNA, and small molecules are developed from prior process history. removed in the flow through or wash. A low pH buffer elutes the viral vector and sets up the subsequent an- AFFINITY CAPTURE CHROMATOGRAPHY ion exchange step. AAV capsids are then eluted into a The affinity step is linked to the performance of the AEX single-use bag containing a predetermined amount of a chromatography steps. The affinity capture (AC) chro- neutralizing base such as 2M Tris. matography step is linked to the harvest step as shown Input parameters for affinity capture chromatography in Table 44. include the AAV vector load concentration and load challenge to the resin. The input load concentration Step Description would be roughly 1×1011 vg/mL and the loading is gen- The purpose of AC chromatography is to remove im- erally in the range of 1×1013 vg/mL resin to 1×1014 vg/ purities while capturing the product (maximize product mL resin, depending on the resin and its flow properties. yield). This step uses an immobilized resin that binds the Interactions that may impact quality attributes include viral vector from the harvested cell culture fluid (clari- flow rate, elution pH, elution buffer conductivity, and fied harvest) to maximize yield and remove impurities collection UV criteria. To analyze these variables, uni- (e.g., HCPs, unpackaged AAV DNA, aggregates, viral variate and/or multivariate studies are needed. contaminants). AC chromatography is an inherently It is important to consider that the separation of robust processing step with a rich platform performance AAV particles from viral impurities must be tied to history that supports the proposed design space. Process each specific AAV serotype. Some serotypes work well

Table 5-43. Acceptable Ranges and Criticality Assessment for Lysis and Depth Filtration Step

Operating Acceptable Classification Rationale Control strategy Parameter Range Surfactant 2.76-2.96 CPP Triton X-100 concentrations <2.76% (w/v) Batch record concentration may result in insufficient clearance procedure Time 60-180 min Longer times lead to aggregation, Batch record PP shorter times may result in incomplete procedure inactivation A-Gene ≤1×1011 vg/mL MP No effect seen on stability or inactivation Batch record concentration procedure Higher temperatures may result in capsid Temperature Temperature 35°C to 40°C PP instability, greater aggregation, and control changes to the purity

CHAPTER 5 Upstream and Downstream Processing 144 Table 5-44. AC Chromatography Step Linkages

Input from Lysis and Clarification Output to Affinity Capture

Vector concentration ≤1×1011 vg/mL Vector concentration ~1×1012 vg/mL

pH 7.5±0.5 pH 7.0-8.5

Aggregate <3.1% Aggregate <3.1%

Purity ~95% Purity ~95%

HCP ~105-106 ng/mL ~200 ng/mL but may range from 100 to 1200 ng/mL

Unpackaged DNA ~1,300,000-2,000,000 ng/mL may range from 100 to 1,000 ng/mL

with commercially available resins, but this step must Table 5-45. Parameters Included in the be optimized for each gene therapy product because Multivariate Study for AC Chromatography the interaction of the resin with the serotype and the impurities are complicated. This will entail optimization Parameter Potential Interactions Range of the appropriate wash and elution strategy to maximize AAV vector load • Flow rate 1013-1014 removal of viral impurities. • Elution pH vg/mL For development and manufacturing purposes, the • End collection resin columns are packed to a bed height of 10 to 20 cm. The • Bed height (univariate) column is purged of storage buffer and equilibrated. The Flow rate • Vector load 100-500 clarified harvest is then loaded. After loading to 1×1013 to • Temperature cm/hr 1×1014 vg/mL, the column is washed with equilibration • Bed height (univariate) buffer. The viral vector is eluted from the column with a Elution buffer • Vector load pH 2-3 low pH elution buffer. The start of collection is based on pH/conductivity • End collection absorbance at 280 nm and 260 nm and is ended by the End collection • Elution buffer 1-3 CVs absorbance at 280 nm and 260 nm or based on specified • pH column volumes. The elution occurs into a container or • Vector load bag that has a predetermined amount of neutralizing buffer such as 2M Tris to ensure that the affinity elution pool is neutralized immediately. The column is then cleaned, regenerated, and re-equilibrated prior to starting Table 5-46. Linkages from Clarified Harvest to the next load cycle. Upon completion of the processing of Affinity Chromatography the entire harvest, the column is washed with and stored Input From Clarified in storage buffer until the next use. Output to AC Harvest Outputs for affinity capture chromatography include process-related impurities (e.g., HCP, unpackaged DNA), Vector concentration 1×1013 to 1×1014 vg/mL 14product-related impurities (aggregates, full-empty pH 7.0-8.5 ratio), and process attributes (yield). More in-depth Conductivity 17±0.5 mS/cm characterization of post-translational modifications or additional degradation by low pH exposure may be Aggregate percentage ≤3% expected at this stage, and it is important to assess ad- HCP concentration ≤200 ng/mL ditional quality attributes by more advanced techniques such as mass spectrometry to determine conformational Unpackaged DNA ≤1000 ng/mL concentration changes in the AAV vector.

CHAPTER 5 Upstream and Downstream Processing 145 Table 5-47. Design and Results

Testing Potential Parameter Range Interactions Scientific Rationale • We expect no impact or interaction • It is possible that an extended load volume due to a decrease in titer could cause displacement of impurities during the load phase, Vector load 1×1010 to None which would result in lower pool impurities concentration 1×1011 vg/mL • We further evaluated the 2 feedstocks used in the multivariate study via spiking and dilution studies over a range of titers (1x1010 vg/mL to 1x1012 vg/mL)

Scale-Down Model with any other process parameter of this step, thereby Typical scale-down models are columns that have a enabling it to be studied as a univariate process variable bed height of 10 cm and a column diameter of 1 cm. (Table 47). Results of the study showed that no impact Additional scale-down experiments with different was observed on step performance or product quality column heights (10 cm, 20 cm and 30 cm) to ensure over the range tested (data not shown). Thus, this pa- pressure flow properties scale appropriately and consis- rameter was classified as a MP and included along with tency of product quality attributes were performed. This multivariate study results to fully describe the knowledge allows flexibility to transfer the technology to different space for the affinity step. manufacturing locations and contract manufacturing organizations. Process Ranges Based on Platform Knowledge Based on the risk assessment, the process parameters that Risk Assessment to Plan Process were considered as not requiring further investigation Characterization Studies are listed in Table 48. For these parameters, the extensive A risk assessment tool that is used to identify parameters process knowledge and modular process performance that must be included in the design of process charac- claims justify the proposed acceptable ranges. terization studies, which include DOEs and univariate approaches. Details of how to conduct a process risk Summary of Process Parameter assessment (PRA), are shown in chapter 4. In this chapter Classification and Ranges for the sake of brevity, only the results are shown. Results of affinity step characterization studies demon- strated that this step does not impact the distribution of Multivariate DOE Studies product variant CQAs (e.g., VP ratios). Moreover, this Based on the results of a risk assessment (details not step was shown to have robust process performance even shown here) parameters included in the A-Gene multi- when challenged with a wide range of feed stream inputs variate study for affinity capture chromatography include (HCP, DNA, and titer). the AAV vector load, bed height, flow rate, elution buffer The affinity operating conditions influence the HCP pH, and end collection (Table 45). Linkages to the next and unpackaged DNA levels in the resulting product step are shown in Table 46. For additional details on how pool. Because subsequent steps (AEX) can reduce HCP to conduct a risk assessment, refer to chapter 4. and DNA to safe and consistent levels, the acceptable output levels from the affinity step are linked to the op- CHROMATOGRAPHY erating conditions of these subsequent steps. A model defining this linkage is given in the Linkage of Unit Univariate Studies Operations section. Based on the risk assessment results and prior knowl- Risk analysis, process characterization studies. and edge, load concentration was not expected to interact process performance history demonstrated that the

CHAPTER 5 Upstream and Downstream Processing 146 Table 5-48. Process Parameter Ranges Supported by Prior Knowledge and Module Process Performance Claims

Prior Acceptable Parameter knowledge range Scientific Rationale

X-Gene 10-20 cm Platform knowledge shows no significant effect on product quality or process performance. There is potential at low bed Bed height Y-Gene 9-11 cm height, high vector load, and high flow rate to decrease yield and increase product pool impurities. Acceptable range is 8-20 cm. Z-Gene 12-18 cm

X-Gene Platform knowledge shows no significant effect on product quality or process performance. Therefore, the proposed buffer Eq/wash pH Y-Gene pH 6.6-7.6 pH range should not affect the performance of this affinity resin. Acceptable range is pH 6.6-7.6. Z-Gene

X-Gene Platform knowledge shows no significant effect on product quality or process performance. Therefore, the proposed Eq/wash (70-140%) Tris, Y-Gene composition ranges should not affect the performance of composition NaCl this affinity resin. Acceptable range is 60-140% Tris, NaCl concentration. Z-Gene

Start 0.1-1.0 Platform knowledge shows no significant effect on product X-Gene OD quality or process performance. The elution phase elutes the product and does not separate the product from impurities. Start 0.3-1.0 Start collection Y-Gene Therefore, the only potential impact to the process is OD parameter decreased yield or collection of additional equilibration buffer in the product pool, but due to the steepness of the starting Start 0.05-0.5 Z-Gene part of the elution peak neither of these outcomes will occur. OD Acceptable range is 0.05-1.0 OD.

X-Gene (90%-110%)

Platform knowledge shows no significant effect on product Eq/wash volumes Y-Gene (90%-110%) quality or process performance. Acceptable range is 60-140%.

Z-Gene (60%-140%) affinity step does not have any critical process parame- affinity resin is expected to be at least 20 cycles. The data ters (CPPs). Only three parameters were linked to CQAs supporting this conclusion are shown in Table 50. (vector load challenge (vg/mL resin), elution buffer pH, and elution buffer conductivity) and were classified as Anion Exchange Chromatography PPs based on control capabilities to operate within the Ion exchange chromatography is used to reduce the proposed design space. The classification of process residual impurities and enrich viral vectors. IEX requires parameters is summarized in Table 49. process development and optimization that depends on the initial feed stream, serotype, and the target needed Reuse/Lifetime Resin Studies for product quality. Targeted yield percent may vary de- Column lifetime studies using the scale-down model pending on the capsid serotype and the indication, but for A-Gene established that the useful lifetime of the higher yield percentages may be more difficult to achieve.

CHAPTER 5 Upstream and Downstream Processing 147 Table 5-49. Variables, Ranges, Controls, and Parameter Classification

Parameter Range Studied Justification Control Classification 1×1013 to 1×1014 vg/ Batch procedures, Vector load Multivariate PP mL resin skid control Elution buffer pH 2.5-3.5 Multivariate Batch procedures PP Flow rate 100-300 cm/hr Multivariate Skid control PP End collection 2.0-3.2 CV Multivariate Skid control MP Environmental Temperature 15-30°C Multivariate MP control Resin lifetime (Resin A) <30 cycles Univariate Column use log MP Load concentration 1×1010 - 1×1011 vg/mL Univariate Titer analysis MP Bed height 8-20 Modular Column use log MP Batch record Eq/wash pH 6.6-7.6 Modular MP procedure Batch record Eq/wash composition 60-140% of target Modular MP procedure Start collection parameter 0.05-1.0 OD Modular Skid control MP Eq/wash volumes 60-140% of target Modular Skid control MP

Notably, there is a triangular association between cost, increase the percentage of full particles. However, careful speed, and quality. While two of these may be attainable, evaluation of the IEX step can still lead to a successful the third aspect may require additional development. empty/full separation step that provides high yield, ac- Depending on the target, one may need to use ul- ceptable enrichment, and manufacturability. tracentrifugation rather than IEX for enriching empty The design space for the AEX step is linked to the capsids from full capsids in particular. For example, if performance of the affinity capture chromatography there is a very low percentage of full capsids, IEX may step. This section describes the use of prior knowledge not be sufficient for purification because the peak will be to design A-Gene process characterization studies and buried within a larger peak of empty capsids. In that case, support a modular approach to impurities clearance. ultracentrifugation may provide a better alternative to Input parameters to be included in the multivariate

Table 5-50. Hypothetical Affinity Resin Lifetime Study

Reuse cycle number Yield (%) HCP (ng/mL) DNA (ng/mL) Aggregate (%) 2 75 220 1000 2.2 4 70 180 1000 2.4 6 75 300 1000 2.0 8 70 150 1,100 1.9 10 70 190 900 2.5 15 70 240 1,100 2.1 20 68 250 900 2.2

CHAPTER 5 Upstream and Downstream Processing 148 Table 5-51. Anion Exchange Chromatography Step Linkages

Input from affinity capture chromatography Output to small virus retentive filtration Vector concentration 5x1012 to 5x1013 vg/mL Vector concentration 1x1012 to 1x1013 vg/mL pH ~7.0-8.5 pH 7.0-8.5 Aggregate <3% Aggregate <2% HCP ~200 ng/mL HCP < 10 ng/mL Unpackaged DNA 100-1000 ng/mL DNA <50 ng/mL Full capsids ~20-30% Full capsids ~40-50% study for AEX include AAV vector load, % full capsids, pool with 60% full AAV capsids. The bind-elute mode is bed height, flow rate, elution buffer pH, and end collec- preferable for affinity pools with <20% full AAV capsids. tion. Outputs for AEX include process-related impurities The full capsid will bind to the resin while the empty (e.g., HCP, DNA) and product-related impurities (e.g., capsids, HCP, and DNA impurities will flow through. aggregate, full-empty ratio), and process attributes (e.g., The full capsid can then be eluted using a salt step or yield) (Table 51). Other outputs that may need to be linear gradient. considered depending on serotype include deamidation, The column is packed with AEX resin to a height oxidation, and glycosylation. of approximately 10 cm. Prior to loading, the affinity capture chromatography product pool is adjusted to the Step Description appropriate pH and conductivity. Following equilibration The final purification step in the A-Gene process is AEX and loading, the column is washed with equilibration chromatography, which is operated in the flow-through buffer to collect the A-Gene product based on absor- mode to bind impurities such as HCP, DNA, empty cap- bance of 280 nm. The entire batch is typically processed sids, and endotoxins to the resin while the AAV flows in one cycle, but multiple cycles are acceptable where through the column. The AEX step can be operated the AEX product pools are combined for subsequent in the bind-elute mode or the flow-through mode de- processing. If multiple cycles are required, the column pending on the target for impurity clearance. Operating is regenerated and re-equilibrated prior to subsequent AEX in the flow-through mode may be useful to further cycles. After the final cycle, the column is regenerated, enhance impurity clearance (HCP, DNA) with an affinity cleaned, and stored.

Table 5-52. Scale-up Parameters for AEX Chromatography Step

Pilot scale Manufacturing Column parameters Laboratory scale (200 L) scale (200 L) Bed volume (mL) 9.5 3142 3142

Bed height (cm) 9-11 9-11 9-11

Diameter (cm) 1.1 20 cm 20

Linear flow rate (cm/hr) 150 150 300

Residence time (min) 4 4 4

Vector load concentration (vg/mL of resin) 1×1013 1×1013 1×1013

Scale-up factor 1 83 83

CHAPTER 5 Upstream and Downstream Processing 149 Table 5-53. Process Performance for the AEX Chromatography Step at Different Scales

Product quality attributes Laboratory scale Commercial scale (200 L) Minimum Maximum Minimum Maximum HCP (ng/ml) 10 25.6 8 15.1 DNA (ng/ml) 50 110 10 63.0 Yield (%) 85 95 95 100 Aggregate (%) 0.8 1.0 0.2 0.8

% Full Capsids 40% 50% 40% 50%

Table 5-54. Parameters Included in the Multivariate Study for AEX

Parameter Potential interactions Range AAV vector load • Flow rate 1012 to 1013 vg/mL resin • Load pH • Load conductivity • Bed height (univariate) Flow rate • Vector load 75-225 cm/hr • Temperature • Bed height (univariate)

Load pH/conductivity • Load conductivity pH 7.6-8.6 • Vector load Conductivity 17±3 ms/cm

Scale-Down Model Risk Assessment to Define Process A scale-down model for the AEX step was established Characterization Studies following standard scale-down/up considerations for A risk assessment approach was used to categorize all chromatography: the column size was scaled based on AEX chromatography step process parameters into three column diameter, with constant bed height, linear veloc- groups: parameters warranting multivariate evaluation, ity, protein load, and load volume/column volume ratio secondary parameters whose ranges could be supported across the scales. This scale-up approach ensures that by univariate studies, and parameters that did not require residence time and mass transport are constant across new studies, but instead would employ ranges based on scales. Volumes of the equilibration, wash, and other knowledge space or modular claims established from prior buffers are based on column volume, thereby ensuring knowledge. For details on how to conduct a risk analysis, the same amounts are used proportionally at laboratory refer to chapter 4. and production scales. A summary of the scale-up pa- Each process parameter was assessed based on the po- rameters is presented in Table 52. tential impact on quality attributes or process attributes As shown in Table 53, the laboratory-scale AEX chro- and impact assessed. Platform process development and matography step performance is comparable to the full- process characterization knowledge from other gene scale manufacturing (200 L) scale process, including the therapies, manufacturing history, and scientific knowl- quality attributes of the AEX product. The residence time edge were used to rank each process variable in the initial of the product on the columns and the elution profiles risk assessment and set the ranges for evaluation. were comparable in both the laboratory- and full-scale production processes. Furthermore, by visual inspection Multivariate DOE Studies the chromatograms were consistent and comparable for Based on this risk assessment, parameters included in the individual small-scale purification runs. the A-Gene multivariate study for AEX chromatography

CHAPTER 5 Upstream and Downstream Processing 150 include the AAV vector load, load flow rate, elution pH, Table 5-55. Parameters Included in the and conductivity (Table 54). Linkages to the next step are Multivariate Study for AEX shown in 55. Output to ultrafiltration/ Input from affinity step Univariate Studies diafiltration Based on the risk assessment results and prior knowledge, Vector concentration 1×1012 to 1×1013 vg/mL vector load concentration was not expected to interact with Load pH 8.1±0.5 any other process parameter of this step, thereby enabling it to be studied as a univariate process variable (Table 56). % Full Capsids 20-30% Results of the study showed that no impact was observed Load conductivity 17±3 mS/cm on step performance or product quality over the range Aggregate percentage ≤2% tested (data not shown). Thus, this parameter was classified as a MP and included along with multivariate study results HCP concentration ≤10 ng/mL to fully describe the knowledge space for the affinity step. DNA concentration ≤100 ng/mL

Process Ranges Based on Platform Knowledge safe and consistent levels, the acceptable output levels Based on the risk assessment, the process parameters that from the AEX steps are linked to the operating condi- were considered as not requiring further investigation tions of these subsequent steps. are listed in Table 57. For these parameters, the extensive Risk analysis, process characterization studies, and pro- process knowledge and modular process performance cess performance history demonstrate that the affinity step claims justify the proposed acceptable ranges. does not have any CPPs. Four parameters were linked to CQAs (vector load, flow rate, load pH, and conductivity) Summary of Process Parameter Classification and were classified as PPs based on control capabilities and Ranges to operate within the proposed design space. The classifi- Results of AEX chromatography step characterization cation of process parameters is summarized in Table 58. studies demonstrated that this step does not impact the distribution of product variant CQAs (e.g., VP ratios). Reuse/Lifetime Resin Studies Moreover, this step was shown to have robust process Column lifetime studies using the scale-down model for performance even when challenged with a wide range of A-Gene established that the useful lifetime of the affinity feed stream inputs (HCP, DNA, titer, and full capsids). resin is expected to be at least 10 cycles. The resin lifetime The AEX operating conditions influence the HCP and study (Table 59) showed no yield loss with extended use DNA levels in the resulting product pool. Since subse- and is also consistent with change in process-related quent steps cannot reduce process-related impurities to impurities such as HCP and DNA.

Table 5-56. Design and Results

Testing Severity Potential Parameter range rating interactions Scientific rationale

No impact or interaction is expected. An extended load volume due to a decrease in titer would only 1×1012 to potentially cause displacement of impurities during Vector load 1×1013 vg/ None the load phase (resulting in lower pool impurities). concentration mL The two feedstocks that were used in the multivariate study were further evaluated by spiking and dilution studies to cover 1×1012 to 1×1013 vg/mL titers.

CHAPTER 5 Upstream and Downstream Processing 151 Table 5-57. Process Parameter Ranges Supported by Prior Knowledge and Module Process Performance Claims

Prior Acceptable Parameter knowledge range Scientific rationale X-Gene 9-11 cm Platform knowledge shows no significant effect on product quality or process performance. There is potential at low bed height, high Bed height Y-Gene 9-11 cm vector load, and high flow rate to decrease yield and increase product pool impurities. Acceptable range is 8-18 cm with process Z-Gene 10 -18 cm control of vector load and flow rate within specified ranges.

X-Gene pH 7.8-8.8 Platform knowledge shows significant effect on product quality or process performance. Therefore, the proposed buffer pH range Load pH Y-Gene should be carefully evaluated for anion-exchange resin between pH 7.8-8.8. Z-Gene X-Gene 20 ± 6 mS/cm Platform knowledge shows significant effect on product quality or process performance. Therefore, the proposed composition ranges Load Y-Gene should be carefully evaluated for load conductivity. conductivity Z-Gene

X-Gene Start 0.1-1.0 OD Platform knowledge shows no significant effect on product quality or process performance. The elution phase elutes the product and does not separate the product from impurities. Therefore, the only Start collection Y-Gene Start 0.3-1.0 OD potential impact to the process is decreased yield or collection of parameter additional equilibration buffer in the product pool, but due to the steepness of the starting part of the elution peak neither of these Z-Gene Start 0.05-0.5 outcomes will occur. Acceptable range is 0.05-1.0 OD. OD

X-Gene (80%-120%) Platform knowledge shows no significant effect on product quality or process performance. Acceptable range is 60-140%. Y-Gene (80%-120%) Flush volumes Z-Gene (60%-140%)

ULTRAFILTRATION/DIAFILTRATION buffer. Upon the completion of diafiltration, the product retentate pool is recovered from the system and the sys- Step Description tem is flushed with an appropriate volume of formulation Ultrafiltration/diafiltration (UF/DF) is the dedicated step buffer. This flush is used to dilute the product retentate to concentrate the viral vector to the target concentration pool to the specified concentration. The TFF membrane and place it in the final formulation buffer. The UF/DF is used for a single cycle and then discarded. step is accomplished using an ultrafiltration membrane that is retentive to the viral vector and permeable to Prior Knowledge buffer species. A poloxamer excipient is added to the From past process knowledge of UF/DF, it is known that diafiltration buffer. A parameter to consider within the AAV vectors are available at fairly low concentrations on risk analysis is the loading capacity, which would be mea- a total protein basis. Therefore, the viscosity of the AAV sured in vg/m2. The AEX pool is loaded onto a 30-kDa solutions is similar to the buffers they are formulated membrane with a screened channel and concentrated in (1-1.5 cP) with little propensity of intermolecular to a specified concentration of 5×1012 vg/mL. At this interactions that affect the filtrate flux. It is important concentration, the product is diafiltered into formulation to identify optimal operating conditions that minimize

CHAPTER 5 Upstream and Downstream Processing 152 Table 5-58. Variables, Ranges, Controls, and Parameter Classification

Parameter Range studied Justification Control Classification Vector load 1×1012 to 1×1013 vg/mL Multivariate Batch procedures, skid PP resin control Load pH 8.2-9.6 Multivariate Batch procedures CPP

Load conductivity 16-18 Multivariate Batch procedures CPP

Flow rate 75-225 cm/hr Multivariate Skid control PP

Temperature 15-30°C Multivariate Environmental control MP

Resin lifetime (AEX) <10 cycles Univariate Column use log MP

Load concentration 1×1012 to 1×1013 vg/mL Univariate Titer analysis MP

Bed height 9-20 Modular Column use log MP

Eq/wash pH 6.6-7.6 Modular Batch record procedure MP

Eq/wash composition 60% to 140% of target Modular Batch record procedure MP

Start collection parameter 0.05-1.0 OD Modular Skid control MP

Eq/wash volumes 60% to 140% of target Modular Skid control MP

Table 5-59. AEX Resin Lifetime Study

Reuse Cycle Number Yield (%) HCP (ng/mL) DNA (ng/mL) Aggregate (%) 2 85 20 120 1.2 4 82 10 100 1.4 6 86 50 80 1.9 8 85 13 110 1.9 10 88 19 100 1.5 15 80 41 110 2.0 20 82 26 120 1.2

the number of pump passes through the membrane. rate, membrane load, bulk concentration, transmem- In general, a flux vs TMP excursion that provides the brane pressure (TMP), and diafiltration volumes (DV). right balance of filtrate flux and TMP is chosen so that Table 61 contains a comparison of the scale-down model channel-induced shear effects are minimized. for the tangential flow filtration (TFF) process.

Scale-Down Model Risk Assessment to Define Process The TFF operation consists of an ultrafiltration step, Characterization Studies which concentrates the product, followed by a diafiltra- A risk assessment approach was used to categorize all tion step, which is used to exchange the product into Tangential Flow Filtration (TFF) step process parameters the appropriate formulation buffer. The key scale-down into three groups: 1) parameters warranting multivariate parameters for TFF are the membrane pore size (spec- evaluation, 2) secondary parameters whose ranges could ified in terms of a molecular weight cut-off), cross flow be supported by univariate studies, and 3) parameters

CHAPTER 5 Upstream and Downstream Processing 153 Table 5-60. TFF Process Parameters

Parameter Value Comments UF/DF membrane molecular weight cut-off 30 kD

Material of construction Composite regenerated cellulose

Temperature 15°C to 25°C

Equilibration pH 8.5±0.2

Equilibration conductivity 17±2 mS/cm

Membrane Load ≤1×1017 vg/m2

Transmembrane pressure (TMP) 20±5 psig

Cross flow rate 4±2 LMM

Cbulk (5±1)×1012 vg/mL Concentrate and diafilter at this concentration

UF/DF pool concentration (3.5 ± 0.5)× 1012 vg/mL

Diafiltration volumes (DV) ≥6 DV Manufacturing: 6±1 DV

Retentate pool pH 7.3±0.2 Buffer exchanged product

which did not require new studies, but instead would Table 5-61. Tangential Flow Filtration Scale employ ranges based on knowledge space or modular Comparison claims established from prior knowledge. Scale-down Each process parameter was assessed based on the po- Parameter 200 L model tential impact on quality attributes or process attributes and impact assessed. Platform process development and Total membrane area (m²) 0.0088 0.44 process characterization knowledge from other gene Scale-down factor 50 1 therapies, manufacturing history, and scientific knowl- edge were used to rank each process variable in the initial Membrane pore size (kDa) 30 30 risk assessment and set the ranges for evaluation. The Membrane load (vg/m²) ≤1×1017 ≤1×1017 TFF ranking results are summarized in Table 62. Cbulk (vg/mL) 5×1012 5×1012 Process Characterization Studies TMP (psig) 20±5 20±5

Input parameters for TFF include loading (vg/m2), Cross flow rate (LMM) 4±2 4±2 transmembrane pressure (psig), and virus concen- tration (1x1012 to 1x1013 vg/mL). The tangential flow Diafiltration volume 6 DV 6 DV filtration step parameters requiring characterization based on the risk assessment are summarized in Table 62. The DF buffer conductivity will be evaluated Tangential Flow Filtration Univariate Studies using a univariate study, while membrane load, DF Three parameters were evaluated individually in one buffer pH, flow rate, TMP, UF concentration, and DV factor at a time (OFAT) studies. These include DF buffer will be evaluated using DOE study designs. The study conductivity, flow rate, and TMP. The flow rate and TMP outputs used to measure the performance of these were evaluated in the multivariate studies as well, but the studies are listed in Table 63. goal was to optimize the ranges for these two parameters

CHAPTER 5 Upstream and Downstream Processing 154 Table 5-62. Summary of Tangential Flow Filtration Risk Assessment Results

Study Parameter Plans Set-Point 1 Level 2 Level 3 Level 4 Level 5 Level 6 Level 7 Level 8 Level

Membrane load (vg/m2) DOE ≤1.2×1015 2.0×1015 1.0×1016 2.6×1016 4.22×1016 1.0×1017 - - -

TFF DF buffer pH DOE 7.3 6.9 7.1 7.3 7.5 7.7 - - -

TFF DF buffer DOE 19.7 15 18 21 - - - - - conductivity (mS/cm)

Flow rate (LMM) DOE 5 2 3 4 5 6 - - -

TMP (psig) DOE 25 15 20 25 30 35 - - -

End of UF product DOE 5e+12 4e+12 4e+12 5e+12 6e+12 7e+12 - - - concentration (vg/mL)

DV DOE 8 5 6 8 10 11 - - -

Hold time TFF pool Hold ≤12 hrs 0 6 9 12 24 36 48 - (15-25°C) study

Temperature during Hold 15-25°C 25 ------hold time TFF pool study (15-25°C)

Hold time TFF pool Hold ≤30 days 0 1 4 7 14 21 30 40 (2-8°C) study

prior to executing the DOE studies. In addition, studies outline of this study is provided in Table 64. were performed to evaluate poloxamer clearance across Ideally, this study is performed during process devel- the TFF membrane. opment. Initial process development to identify flux vs TMP must be the first study performed. However, this TFF Diafiltration Buffer Conductivity study must be repeated during process characterization The impact of DF buffer conductivity was assessed by using different membrane lots at different permeabilities targeting selected NaCl concentrations and comparing to ensure that the range chosen during development the performance of the DF step using DF buffer with stands valid as a function of membrane lot variation. concentrations of 150 mM, 180 mM, and 210 mM NaCl. This study is important to establish bounds on Poloxamer Sieving Rates the diafiltration buffer conductivity without reducing The sieving rates of poloxamer across the TFF process the resolution of the characterization study. will be evaluated in order to model the co-concentration of poloxamer throughout this unit operation. For this Flow Rate and Transmembrane Pressure Optimization purpose, the sieving coefficients of poloxamer at different Flow rate and TMP will be evaluated through an op- feed flow rates is measured using formulation buffer with timization study examining 5 levels of each parameter different concentrations of poloxamer. Another measure- and comparing resulting flux curves to identify the TMP- ment is made using the AEX pool (load material to UF/ independent operation ranges. These results will help to DF step) to ensure that measurements with the buffer are define the ranges to be evaluated in the DOE studies. The comparable to the measurement in the presence of the

CHAPTER 5 Upstream and Downstream Processing 155 Table 5-63. Tangential Flow Filtration Study Responses

Process Performance Responses Analytical Responses Product pool turbidity (NTU) Residual affinity ligand Product pool pH Residual antifoam Product pool conductivity Residual Surfactant Osmolality Aggregates; submicron Product pool volume Aggregates; subvisible Process time Vector genome titer Appearance Poloxamer-188 concentration Sodium phosphate concentration (pH and conductivity) Sodium chloride concentration (pH and conductivity) Vector genome recovery

AAV. The poloxamer sieving rate study design is shown in process performance history demonstrate that the crit- Table 65. Experiments were performed under conditions ical process parameter was directly related to membrane of total recycle to measure sieving coefficients of poloxam- load challenge. Given the fairly high fluxes of the mem- er. All experiments were conducted at a TMP of 20 psig. brane at different feed flow rates and TMP, parameters were linked to CQAs (vector load, flow rate, load chal- Tangential Flow Filtration Multivariate Studies lenge, TMP, and conductivity) and were classified as PP Due to the number of factors to be studied, the TFF step based control capabilities to operate within the proposed was evaluated in two stages. First, a randomized frac- design space. The classification of process parameters is tional factorial design (FFD) screening study examining summarized in Table 67. all six parameters, followed by a randomized 1.5 axial central composite design (CCD) focused on the four Table 5-64. Flow Rate and TMP Optimization factors demonstrating the most significant estimates of Study Design effect. These parameters include membrane load, DF buffer pH, flow rate, TMP, UF concentration, and DV. Level Level Level Level Level Table 66 summarizes the design for the FFD screening 1 2 3 4 5 study. The full study design is not shown here. Flow Rate (LMM) 2 3 4 5 6 This study was designed allowing estimation of all

2-factor interactions and includes eight center-point Transmembrane 15 20 25 30 35 runs, resulting in a total of 40 runs. Pressure (psig)

Summary of Parameter Classifications and Ranges Results of TFF step characterization studies demon- Table 5-65. Poloxamer Sieving Rate Study Design strated that this step does not impact the distribution of product variant CQAs (e.g., VP ratios). Moreover, this step was shown to have robust process performance Flow Flow Flow Load material rate rate rate even when challenged with a wide range of feed stream (LMM) (LMM) (LMM) inputs (listed in table 64) or a range of diafiltration Formulation buffer with 2 4 6 buffers between 150 and 210 mM, indicating that this 0.001% poloxamer step is unlikely to go out of control within normal manufacturing ranges of pH and conductivity. The TFF Formulation buffer with 2 4 6 0.01% poloxamer conditions influence the aggregate levels in the resulting product pool. AEX pool with 0.001% 2 4 6 poloxamer Risk analysis, process characterization studies and

CHAPTER 5 Upstream and Downstream Processing 156 Table 5-66. TTF DOE Screening Study Design: FFD

Level Membrane Diafiltration Flow rate Transmembrane UF Concentration DV load (vg/m²) buffer pH (LMM) Pressure (psig) (vg/mL)

- 5E+15 7 2 10 3E+12 4

0 5E+16 7.3 4 15 6E+12 8

+ 9.5E+16 7.6 6 20 9E+12 12

Table 5-67. Variables, Ranges, Controls, and Parameter Classification for TFF step

Parameter Range studied Justification Control Classification

Batch procedures, skid Membrane Loading 1×1015 to 1×1017 vg/m2 Multivariate DOE CPP control

Final Concentration Batch procedures, skid 3×1012 to 7×1012 Multivariate DOE CPP (vg/mL) control

Feed Flow Rate (LMM) 2 – 6 Multivariate DOE Batch procedures PP

Transmembrane Pressure 10-20 Multivariate DOE Batch procedures PP (psig)

Multivariate Temperature 15-30°C Environmental control MP DOE

Load concentration 1×1012 to 2×1013 vg/mL Univariate Titer analysis MP

Diafiltration buffer pH 7.3 – 7.6 Multivariate DOE Batch procedures MP

SUMMARY OF DOWNSTREAM PROCESS DESIGN SPACE is possible to perform worst-case linkage studies on a Worst-case linkage studies are part of the BLA-enabling smaller (i.e., not commercial) scale for cost consider- studies with the intent to run each unit operation within ations and to reduce product usage. A detailed descrip- the scope of the DOE at the worst-case scenario (includ- tion of process characterization studies may be reviewed ing hold time) concurrently to determine whether the in A-Mab.10 quality attributes are impacted. Hold times, not only at each step but also cumulatively, are important to con- Control Strategy for Downstream Process: sider in the worst-case linkage studies. Importantly, it For details on the control strategy, refer to chapter 7.

CHAPTER 5 Upstream and Downstream Processing 157 Endnotes 1. Robert MA, Chahal PS, Audy A, Kamen A, Gilbert R, 7. Boissy R, Astell CR. An Escherichia coli recBCsbcBrecF Gaillet B. Manufacturing of recombinant adeno-asso- host permits the deletion-resistant propagation of ciated viruses using mammalian expression platforms. plasmid clones containing the 5’-terminal palindrome Biotechnol J. 2017;12(3). of minute virus of mice. Gene. 1985;35(1-2):179-185. 2. Merten OW, Hebben M, Bovolenta C. Production 8. CMC Biotech Working Group. A-Mab: a case study of lentiviral vectors. Mol Ther Methods Clin Dev. in bioprocess development (chapter 3). CASSS web- 2016;3:16017. site. https://cdn.ymaws.com/www.casss.org/resource/ resmgr/imported/a-mab_case_study_version_2-1. 3. Nouri F, Sadeghpour H, Heidari R, Dehshahri A. pdf. Updated October 30, 2009. Accessed February 16, Preparation, characterization, and transfection effi- 2021. ciency of low molecular weight polyethylenimine-based nanoparticles for delivery of the plasmid encoding 9. ICH. ICH Q5A (R1) Quality of biotechnological CD200 gene. Int J Nanomedicine. 2017;12:5557-5569. products: viral safety evaluation of biotechnology products derived from cell lines of human or animal 4. van der Loo JC, Wright JF. Progress and challeng- origin. ICH website. https://www.ema.europa.eu/ es in viral vector manufacturing. Hum Mol Genet. en/ich-q5a-r1-quality-biotechnological-products-vi- 2016;25(R1):R42-52. ral-safety-evaluation-biotechnology-products-derived. 5. Gray SJ, Choi VW, Asokan A, Haberman RA, McCown Updated January 1997. Accessed February 9, 2021. TJ, Samulski RJ. Production of recombinant adeno-as- 10. CMC Biotech Working Group. A-Mab: a case study sociated viral vectors and use in in vitro and in vivo in bioprocess development (chapter 4). CASSS web- administration. Curr Protoc Neurosci. 2011;Chapter site. https://cdn.ymaws.com/www.casss.org/resource/ 4:Unit 4 17. resmgr/imported/a-mab_case_study_version_2-1. 6. Bi X, Liu LF. A replicational model for DNA re- pdf. Updated October 30, 2009. Accessed February 16, combination between direct repeats. J Mol Biol. 2021. 1996;256(5):849-858.

CHAPTER 5 Upstream and Downstream Processing 158 Chapter 6 Drug Product Chapter 6 | Contents

Chapter Summary...... 161 Key points...... 161 Introduction...... 162 Fill-Finish...... 162 Concentration and Vector Aggregation...... 162 Gene Therapy Delivery...... 162 Aseptic Processing and Sterile Fill-Finish...... 163 Choice of Container...... 163 Formulation...... 163 Characterization...... 164 Safety...... 166 Purity...... 167 Identity...... 168 Potency...... 168 Container Closure Integrity Testing (CCIT)...... 169 Storage and Stability...... 170 Conclusion...... 171 Endnotes...... 172

CHAPTER 6 Drug Product 160 Chapter Summary

The vector manufacturing process culminates contamination, or improper packaging. with the formulation and vialing of the purified It is of the utmost importance to begin the vector product. While this stage may come tem- gene therapy development process while consid- porally at the end of the process, it is important to ering the “end goal” (as outlined in the TPP and consider the desired drug product characteristics, QTPP). In this chapter, we have outlined the best including fill-finish, formulation, characteriza- practices for fill-finish, formulation, characteriza- tion, container closure integrity testing (CCIT), tion, CCIT, and long-term storage and stability and long-term storage and stability, at the com- that can be used to develop a robust final gene mencement of the gene therapy development pro- therapy product; however, the information here cess. It is important to note that the final product is in no way exhaustive. Each process will differ has the greatest value, and thus mistakes at this based on the specific product, its characteristics, stage may significantly impact the gene thera- administration route, titer selection, final formu- py’s success regarding product misformulation, lation buffer, and container choice, among others.

The final stage of the vector manufacturing process important CQA during the formulation stage. At a is the formulation and vialing of the purified vector clinical stage of development, CQAs should cover the product. It is critical that formulation and vialing be four Food and Drug Administration defined catego- considered at the beginning of the gene therapy ries of tests: safety, purity, identity, and potency. With- development process in order to develop a robust in these categories, specific CQAs include but are not final product. limited to physical titer, genetic identity, aggregation • After all the effort that goes into GMP bulk gene state, infectivity, full to empty particles, biological therapy production, the final product has the great- activity or potency, and/or immunological activity. est value. Mistakes at this point are the costliest and • CQAs should be monitored in a longitudinal manner can lead to product aggregation, contamination, or to ensure stability of CQAs over time during long- instability, among other adverse consequences. term stability studies. • Fill-finish refers to the immediate outcome from the • Several factors may impact stability during long-term upstream and downstream processes. Because the storage, including temperature fluctuations, diluents, product container is going to be used for the long- container constituents, and other environmental term storage of the product, it must be robust and be considerations. able to maintain product integrity at the defined stor- • When choosing an appropriate container and con- age conditions. The most important critical quality tainer closure, it is imperative to consider the con- attributes (CQAs) during this stage include aggrega- tainer’s compatibility with the product, the intended tion, potency, identity, sterility, and CCIT. route of administration, and long-term storage condi- • In the preclinical stage, potency is generally the most tions (e.g., ability to withstand cryopreservation).

CHAPTER 6 Drug Product 161 Introduction and increased immunogenicity after clinical administra- tion. The typical concentrations of purified AAV vectors The final stage of the vector manufacturing process is the (1011 to 1013 vg/mL) correspond to dilute solutions of formulation and vialing of the purified vector product. drug substance (~1-100 µg/mL), and nonspecific ad- It is important to ensure the use of excipients that allow sorption of vectors to plastics, glass, metal, and other vector stability under the anticipated storage conditions surfaces during storage/handling of the vector may as well as excipients and resultant formulations that are occur. However, inclusion of a surfactant may help to compatible with the expected route of administration.1 prevent vector losses.1 After the considerable amount of effort that goes into Certain clinical programs may require high doses to GMP bulk gene therapy production, the final product be administered in a relatively small volume.2 It has been has the greatest value (and thus has the most to lose as a reported that recombinant AAV2 undergoes aggregation product failure). Mistakes at this point can lead to prod- in a concentration-dependent manner for titers >1x1013 uct aggregation, contamination, or instability, among vg/mL when formulated in physiologic ionic strength other adverse consequences. buffers. Elevated ionic strength can prevent aggregation There is a holding step between the bulk drug sub- to titers up to 5x1013 vg/mL.2 However, slightly elevat- stance (DS) and drug product (DP) stages. During this ed ionic strength formulations compatible with direct time, there will be GMP measurement to ensure that the parenteral injection can limit the AAV2 vector titers to DP can meet the requirements as specified by regulatory approximately 2x1013 vg/mL. agencies. Analytics related to the safety profile, potency, Serotype appears to affect vector aggregation. AAV2 and titer are performed prior to formulation and fill-fin- vectors are prone to aggregation at concentrations ish (i.e., product vialing). >1x1013 to 2x1013 vg/mL in parenteral-compatible buf- fers, but recombinant AAV8- and AAV9-based vectors Fill-Finish can be prepared in neutral physiologic buffers at much higher concentrations. Thus, use of AAV8- and AAV9- Fill-finish refers to the immediate outcome from the up- based vectors can facilitate gene therapy applications that stream and downstream processes. Because the product require a limited volume but a high dosage.3 container is going to be used for the long-term storage of the product, it must be robust and be able to maintain GENE THERAPY DELIVERY product integrity at the defined storage conditions. During Specific gene therapies may be delivered through differ- the fill-finish stage, CQAs that are typically at the forefront ent routes of clinical administration, such as intravenous, include aggregation, potency, identity, sterility, and CCIT. intraocular, subretinal, intramuscular, or cell therapy This section will discuss these CQAs in more detail. (e.g., T-cells). Different formulations are developed to The main differences between fill and finish for gene suit the clinical administration and preserve the potency therapy products compared to more traditional biologics and stability of the product. include the batch size, which is relatively small for gene For example, for gene therapies targeting the central therapy, and the time to process the batch. Typically, nervous system (CNS), various routes of administration gene therapy products should be processed quickly, are possible (e.g., intraparenchymal, intracerebroventric- between 4 and 5 hours, due to potential instability at ular, cisternal intrathecal, or lumbar intrathecal). Each of room temperature. these routes has various advantages and disadvantages. Intraparenchymal delivery requires a relatively low CONCENTRATION AND VECTOR AGGREGATION dose compared with systemic or cerebrospinal fluid Vector aggregation that may occur during vector concen- (CSF) administration. In addition, delivery into the im- tration (depending on AAV serotype) can contribute to mune-privileged site of the brain reduces the impact of unexpected loss during handling, altered biodistribution, potential preexisting immunity to AAV serotypes. Mouse

CHAPTER 6 Drug Product 162 models have shown that intraparenchymal delivery into Because AAV is charged and will adhere to certain sub- various areas of the brain lead to widespread enzyme stances, thereby causing drug product loss, appropriate distribution and biochemical and histological correction vial material is crucial. Vial types may include glass, in large areas of the brain. Delivery into the CSF (in- cryo-vial plastic, and various polymers (e.g., Crystal tralumbar or cisternal) requires a somewhat larger dose Zenith). For some vectors, moving away from glass vials and may result in effects to non-CNS areas. Intravenous to polymeric vials can decrease potential safety issues delivery of CNS-targeting gene therapies appears to be by ensuring safe containment of the vector.5 Bags are an possible, although the efficiency appears to be limited in important consideration for AAV and are typically used older mice. Further, accumulation of sialic acid within in the context of cell therapy to ensure aseptic connection the CNS, which is an inhibitor of AAV9 transduction, and workflow integration. may limit efficacy of this approach.4 Formulation ASEPTIC PROCESSING AND STERILE FILL-FINISH Facilities, equipment, procedures, and personnel must In the preclinical stage of formulation, potency is gen- be appropriate to ensure aseptic processing and sterile erally the most important CQA. At a clinical stage of fill-finish. In addition, it is important that filling process- development, CQAs may include physical titer, genetic es be qualified prior to the actual product fill.1 identity, aggregation state, infectivity, full to empty parti- Sterilization of each component of a drug product, cles, biological activity or potency, and/or immunological regardless of form, must be carried out prior to aseptic activity.5 Agencies generally look at a combination of processing/fill-finish. Various methods may be used, such route of administration and CQAs in a longitudinal as heat sterilization in an autoclave, radiation sterilization manner to ensure stability of CQAs over time. Therefore, (especially useful if the component is heat-sensitive), it is very important that formulation is considered early filter sterilization, and ethylene oxide gas sterilization during the development process and often so that drug for heat- and moisture-sensitive components. stability is not sacrificed. Formulation is a critical component of the process CHOICE OF CONTAINER because it includes stability, potency, purity, and safety. Ideally, the container should be selected during phase However, it is often overlooked. Formulation should be 1 or 2 of clinical development while considering com- considered early during the process, as early as the pro- mercialization factors, such as scalability. Failure to cess development stage. It is largely determined by two select a container without considering commercializa- factors: the clinical administration route and the process tion may lead to problems and delays in later stages of development stage stability analysis, including for AAV- development. specific gene therapy aggregation, stability, and potency. Several options exist for gene therapy containers, and Aggregation can be measured by dynamic light pros and cons of various vials and bags must be consid- scattering (DLS). Light scatters from the moving mac- ered while keeping in mind the particular characteristics, romolecules. This motion imparts a randomness to the including mode of administration/delivery, for the DP. It phase of the scattered light, such that when the scattered is also important to consider and mitigate shear effects light from two or more particles is added together, there when dispensing. For example, open filling from a needle will be a changing destructive or constructive interfer- is a break in the closed system handling of a BSL-2 virus. ence. This will cause time-dependent fluctuations in the Thus, this would create a greater demand for facility con- intensity of scattered light, which are measured by a fast tainment capabilities than what would be needed when photon counter and are directly related to the rate of dispensing a protein product. diffusion of the molecule that is correlated with particles’ Titer of a gene therapy product must be measured hydrodynamic area (AAV viral particle or aggregate size). so that the fill-finish can be done in a titer-appropriate Expansion of the CQA list beyond infectivity is manner with consideration of both material and size. important to determine the route of viral vector

CHAPTER 6 Drug Product 163 degradation in different buffers and under different refrigerated. Zolgensma is stable for 14 days after receipt extrinsic conditions, and eventually, to enact improve- when stored between 2ºC and 8ºC.7,8 ments in formulation. Some viruses have a temperature threshold, above which the virion structure is abruptly affected. Temperatures for the storage and shipment of Characterization viral vectors could be -40°C, -70°C, or below, which can During preclinical stages of development, characterization complicate the cold supply chain. Bulk material or drug focuses mainly on potency. During clinical development, product may require shipment on dry ice, and there are several CQAs must be considered. It is important that gene additional factors that must be considered, such as the therapy design is “smart” (e.g., considering how large the need for containers with low permeability to carbon di- payload is), and various elements of design will impact the oxide vapor due to the typical instability of viruses at low final drug quality. Specific CQAs during characterization pH. Therefore, temperature studies are required in order include safety and potency considerations. Endotoxin to determine the threshold effect as well as the impact of testing is recommended on the final container product to cumulative excursions near a threshold.5 ensure an appropriate level of endotoxin, which is defined Formulation buffers are also important considerations. as five endotoxin units (EU)/kg/bodyweight/hr) according Luxturna, a currently available AAV vector-based gene to FDA guidance. The ratio of empty-full capsids should therapy for the treatment of patients with confirmed also be considered at this stage, and although the FDA biallelic RPE65 mutation-associated retinal dystrophy, does not provide specific guidance because it is related to is administered via subretinal injection. A single-dose, the biology of the product, it is important for this CQA 5-mL vial of Luxturna contains 5 x 1012 vector genomes to maintain consistency in a longitudinal manner. Titer (vg) per mL, 180 mM sodium chloride, 10 mM sodium measurement is also important to consider due to lethality phosphate, and 0.001% Polaxmer 188 (pH 7.3). Following from liver toxicity related to high doses. dilution, each 0.3-mL Luxturna dose contains 1.5 x 1011 Drug product release is the ultimate goal for gene vg. The diluent, which is supplied in a 1.7-mL extractable therapy manufacturing and gene therapy development. volume per vial in 2-mL vials, contains sterile water In order to do this, AAV vectors must be thoroughly containing 180 mM sodium chloride, 10 mM sodium characterized to ensure that they meet the predeter- phosphate, and 0.001% Poloxamer 188 (pH 7.3).6 mined specifications for vector identity, safety, purity, The Luxturna active substance (bulk drug product) is potency, and stability for every lot. The Biologics License formulated and shipped frozen on dry ice (to maintain Application (BLA) requires that all quality control a temperature of ≤-65ºC) to a filling site, where it is pro- assays be finalized and validated.1 Biological products cessed into the final product by filtration and filling into are complex and are often heterogeneous with complex Crystal Zenith vials. The finished product is then shipped mechanisms of action. Product characterization allows at ≤-65ºC to the secondary packaging and labelling site the manufacturer to determine the relationship between in insulated shipping containers in semi-finished vials product quality attributes and safety and efficacy. with primary labels applied. One challenge involved in the characterization phase Zolgensma, an AAV vector-based gene therapy for is the cost of analytics.5 The product yield from current pediatric patients <2 years of age with spinal muscular manufacturing and purification processes is low, so the atrophy with biallelic mutations in the SMN1 gene, is amount of product needed for complete in-process test- available in 5.5- and 8.3-mL vials with a concentration ing, product characterization, lot-release, and stability of 2.0 x 1013 vg/mL. Zolgensma also contains 20 mM tris testing can consume a significant portion of clinical lots. (pH 8.0), 1 mM magnesium chloride, 200 mM sodium In addition, current practices for qualifying reference chloride, and 0.005% Poloxamer 188. The IV dosage is standards rely on labor-intensive and variable analytics, determined by body weight, with a recommended dose and additionally are not uniform across the industry. It of 1.1 x 1014 vg/kg. The Zolgensma product is shipped is important to note that reference standards for a given and delivered frozen (≤-60ºC) in clear vials and is stored gene therapy are generated from designated clinical lots.

CHAPTER 6 Drug Product 164 Table 6-1. Example of Release Testing for a Clinical AAV Vector Product

Crude Cell Harvest Bulk Drug Substance Drug Product

GENERAL • Appearance • Appearance • pH • pH • Osmolality • Osmolality • vg identity IDENTITY • Capsid • Capsid • Payload sequencing • Payload sequencing POTENCY • vg titer • vg titer • Infectivity • Infectivity • In vitro expression • In vitro expression PURITY • SDS-PAGE silver stain • SDS-PAGE silver stain • OD260/OD280 • Aggregates • Residual host-cell DNA • Residual plasmid DNA • Residual BSA • Residual HEK293 • Residual benzonase • Residual cesium • Aggregates SAFETY • Adventitious viral testing • Endotoxin • Sterility • Mycoplasmas • Sterility • Endotoxin • Bioburden/sterility • rcAAV (replication- competent AAV)

The poor yield, along with the high variability inherent a clinical AAV vector product are shown in Table 1.1 in the manufacturing of gene therapies and the required For some gene therapy products, the empty-full capsid amount of analytical testing, can require the frequent ratio must be considered prior to the product being placed generation of new clinical lots and reference standards. into its vial/container. Some methods that may be used Thus, a significant portion of the product yield is easily to analyze empty-full capsid ratio include spectrometry, consumed through required and necessary analytical ELISA, qPCR, analytical ultracentrifugation, ion exchange testing. The use of QbD and DOE approaches can help chromatography, and transmission electron microscopy. to ensure that the analytics employed in this capacity The U.S. Food and Drug Administration defines are robust and meet the acceptable levels in order to characterization in four categories of tests, as shown in confidently determine product characterization and Table 1: safety, purity, identity, and potency. quality assessment data. The following sections take a closer look at tests con- Examples of tests that may be used for release testing of tained within these categories.

CHAPTER 6 Drug Product 165 Table 6-2. Overview of Safety Testing.1

Assay Purpose of Assay Time of Assay

Adventitious viral Demonstrate absence of AVA Crude cell harvest agents (AVA)

Mycoplasma Demonstrate absence of mycoplasma Crude cell harvest

Bioburden • Ensure aseptic conditions throughout the Following each purification step to manufacturing process ensure • Ensure product conforms to recommendations of USP and 21 CFR 610.12

Endotoxin Demonstrate absence of endotoxin in a manner Final product appropriate for the intended route of administration

Sterility Ensures product safety and aseptic, sterile product Final product without detectable microbial contamination rcAAV Demonstrate absence of potential pathogenic Bulk drug substance derivatives of recombinant AAV

SAFETY endotoxins over a range of 1.0 to 0.1 EU/mL. The product must be tested for the presence of potentially Recombinant AAV replication requires the presence unsafe impurities, which is done by assessing the quality of of a helper adenovirus, in addition to wild-type AAV raw materials as well as process contaminants (e.g., column genes that are involved in virion construction and media, antibiotics, or other agents used). Safety testing packaging. Therefore, rcAAV requires only the pres- should encompass assays to assess sterility, mycoplasma, ence of a helper adenovirus for AAV replication within adventitious viral agents, bioburden, replication-compe- a permissive cell line. Replication-competent AAV tent AAV (rcAAV), and endotoxin. For licensure, general should be assessed using an infectious center assay. safety should also be assessed. Table 2 shows an overview This may be challenging for some AAV serotypes due of safety testing. Sterility testing must be performed on the to the low infectivity of specific serotypes in cultured final product according to 21CFR 610.12.1 cells. When developing new serotypes for clinical stud- Bioburden assays should be performed at product ies, assay development efforts should focus on address- stages, for example following each purification step, in ing this challenge. Replication-competent AAV can be order to ensure aseptic conditions throughout the entire assessed through cell culture and qPCR. manufacturing process. These assays must be performed In addition, commercially relevant production in order to ensure the product conforms to the recommen- methods may use Herpes simplex virus (HSV) or bacu- dations of the USP and 21 CFR 610.12. These assays lovirus to manufacture AAV vectors. It is important to may be performed through the direct inoculation of the ensure the removal of these viruses because residual bulk or final test sample into two different types of media. viral particles could elicit toxic or immune reactions. The samples are incubated for 14 days at two different Methods to purify AAV from HSV-containing matri- temperatures. In addition, results should be confirmed ces generally use detergents such as Triton X-100 at by a bacteriostasis/fungistasis test to demonstrate that the concentrations as high as 1% w/v during harvest or low samples do not interfere with the growth of six organisms pH-induced flocculation of cellular and HSV proteins of varying classes of mold and bacteria. from denatured viral capsids. Although it appears that Bacterial endotoxin can be assayed via a LAL/chromogenic residual HSV proteins are detectable in final rAAV method. In this test, precise amounts of FDA-licensed LAL stocks, preclinical toxicology studies have shown that reagents, chromogenic substrates, and controls are required. these levels were tolerated in animals, and no reactions A variety of commercial products are available to measure have been reported in humans. Like HSV inactivation,

CHAPTER 6 Drug Product 166 Table 6-3. Overview of Purity Testing During GMP

Assay Purpose of Assay Time of Assay Residual host-cell protein Ensure host cell protein is at an acceptable level Purified bulk (DS) Residual host-cell DNA Ensure host cell protein is at an acceptable level Purified bulk (DS) OD260/OD280 Determine protein concentration Purified bulk (DS) Residual plasmid DNA Demonstrate absence of residual plasmid DNA Purified bulk (DS) Residual BSA Demonstrate absence of residual BSA in product Purified bulk (DS) Residual benzonase Demonstrate absence of residual benzonase in Purified bulk (DS) product Residual substances from Ensure removal of purification substances Purified bulk (DS) purification (e.g., cesium) Aggregates Ensure aggregates are at an acceptable level so as Formulated, vialed (DP) to not affect dose or concentration SDS-PAGE silver stain Visualize the VP1, VP2, and VP3 bands and ensure Formulated, vialed (DP) that there are no other proteins chemical lysis by Triton X-100 (0.5% w/v) at harvest antibodies) as well as product isoforms (such as empty results in disruption of the cell membrane and bac- capsids). Table 3 describes assays used for purity testing ulovirus envelope. Detergent inactivation results in during GMP runs. However, note that other factors may inactivation of most baculovirus particles but does not need to be considered during engineering runs. disrupt the integrity of the nucleocapsids. Baculovirus- A visual test should be conducted via S0026 based es are known to transduce mammalian cells but are not on USP and to ensure that a product’s able to replicate, and there is no evidence that baculo- visual liquid appearance is clear and the liquid is free viruses harm humans unless injected intravenously at from visible particulates using both white and blue back- very high doses.20 grounds and appropriate liquid. Vials should be assessed for imperfections, such as cracks and loose caps. PURITY Residual host cell protein (HCPs) can be evaluated Purity is related to levels of cell-culture process impuri- via polyacrylamide gel electrophoresis and protein ties, such as residual host cell proteins and DNA, helper staining, reverse-phase high-pressure liquid chromatog- virus protein and nucleic acids, and helper plasmid DNA. raphy, host-cell protein ELISA, and spectrophotometry. During the early development process, tests for contam- Residual host cell DNA may be assessed using qPCR with inants in the final product should be tested for. As devel- appropriately designed primers and probes, and helper opment progresses, the manufacturing process should be plasmid DNA may be assessed through qPCR designed validated to remove and to not introduce process-related to analyze residual plasmid-derived DNA fragments.1 contaminants. Identified contaminants should be removed, Residual benzonase should be analyzed through or appropriate limits should be set based on data from lots specific and sensitive methods as part of clinical testing. that were shown to be safe in preclinical and/or clinical The data must demonstrate assay sensitivity, linearity, studies. Initial specifications, including acceptance limits, accuracy, precision, and specificity. Residual polyeth- may need to be refined based on manufacturing process yleneimine (PEI) may be assayed via UHPLC-CAD experience gained throughout development. assays that use both linear and branched forms of PEI Purity testing aims to characterize the capacity over a range of molecular weights. Residual iodixanol of the purification process to remove manufacturing should be measured in AAV viral vector products puri- components (such as cell culture media, helper virus, fied by iodixanol gradient ultracentrifugation. Iodixanol

CHAPTER 6 Drug Product 167 Table 6-4. Overview of Identity Testing

Capsid identity Assess heterogeneity and confirm capsid identity • Unpurified bulk (e.g., ELISA, western blot, empty-full capsid • Purified bulk (DS) analysis, silver staining) • Formulated, vialed (DP)

Payload identity Ensure that the identity of the product • Unpurified bulk completely matches that of the designed product • Purified bulk (DS) • Formulated, vialed (DP)

concentration can be assessed through semi-quantitative be time-consuming due to the product-specific consid- HPLC size-exclusion chromatography analysis. erations that must be taken into account during assay Nuclease-resistant, AAV-encapsidated DNA impuri- development. Potency assays should be in place during ties can be assessed by qPCR using primers and probes early product development to demonstrate product designed for relevant sequences in helper plasmids or activity, quality, and consistency throughout product de- high copy number genomic sequences.1 The sensitivity to velopment, generate data to support specifications for lot nuclease treatment performed prior to qPCR allows one release, provide a basis to assess manufacturing changes, to make a distinction between nuclease-sensitive residual assess product stability, recognize technical problems, DNA impurities and nuclease-insensitive–encapsidated and collect sufficient data to support correlation studies residual DNA impurities. The total AAV capsids can be (to link potency to functional activity).10 measured using capsid-specific ELISA assays, with the Formally, potency is defined as “the specific ability or amount of empty capsid determined by comparing to capacity of the product, as indicated by the appropriate capsid particle titer and vg titer. Monoclonal antibodies laboratory tests or by adequately controlled clinical specific for a conformational epitope on an assembled data obtained through the administration of the prod- AAV capsid that is coated onto strips of a microtiter plate uct in the manner intended, to effect a given result” can be used to capture AAV particles from the sample. (21 CFR 600.3(s)). Potency tests are performed to measure product IDENTITY attributes that are associated with product quality and Identity refers to characterization of molecular integrity manufacturing controls throughout all phases of clinical of the gene therapy per design. It should include quanti- study (Table 5). Measurements of potency are used to tative testing by phenotypic and/or biochemical assays to show that product lots meet defined specifications or confirm cell identity and assess heterogeneity (Table 4). acceptance criteria during all phases of clinical devel- Methods to characterize identity may include capsid con- opment as well as following market approval. Although firmation with ELISA or western blot, and/or empty-full no single test can adequately measure product attributes capsid analysis. To confirm payload identity, sequencing, that predict clinical efficacy, data from well-controlled either next-generation sequencing or Sanger sequencing, clinical investigations can provide evidence to show that of AAV may be carried out to ensure that it completely a product has biological activity and is potent.10 matches with the designed product. Appropriate referenc- The exactin vitro potency assay that is used depends es should be included in quality control assays to ensure on the gene therapy product and is highly centric around that the results obtained are valid and reproducible.9 the biological process that the therapy is supposed to interefere with. Assays may range from TCID50, flow cy- POTENCY tometry, protein expression, gene expression, enzymatic Potency is a biology- and product-specific characteristic. assays, etc. Particularly in earlier stages, nuclease-resis- Potency tests primarily focus on determining vg concen- tant genome concentration may be assessed by dot blot tration as well as functional activity. This step tends to hybridization, ddPCR, or qPCR. It is important to note

CHAPTER 6 Drug Product 168 Table 6-5. Overview of Potency Testing

Assay Purpose of Assay Assay Performed on

TCID50/Infectivity Determine concentration at which • Unpurified bulk 50% of cells are infected • Purified bulk (DS) • Formulated, vialed (DP)

Protein expression Ensure protein levels • Purified bulk (DS) • Formulated, vialed (DP)

In vivo animal testing Ensure appropriate dosing and • Formulated, vialed (DP) for reference only (immunostaining) potency

that when using PCR, a linearized DNA template is nec- Container Closure Integrity Testing essary in order to generate a standard curve because the (CCIT) use of supercoiled DNA standards will result in a signifi- cant overestimate of the vg concentration. Determination Although many containers are available, it is important of AAV titer is an important consideration that has prov- to consider the container’s compatibility with the intend- en to be somewhat difficult in practice. Further, sample ed route of administration as well as cryopreservation. preparation is a key consideration in qPCR. It may be prudent to use viral material as the control, which can • Compatibility with route of administration: serve both as a standard curve as well as a control for Depending on the route of administration, it may be sample preparation. important to be able to plug into a system aseptically. In addition to genome concentration, assays must Drug products and containers must be specifically be done to assess the functional activity of the vectors. designed to allow such aseptic removal of drug. In earlier stages of development, ELISA may be used to In addition, extractability studies may be required demonstrate transgene protein expression in a dose-de- to ensure that the dose can be extracted from the pendent manner, whereas later stages of development container, both related to physical volume and (e.g., phase 3 clinical trial) require the establishment of maintenance of potency. a bioassay that quantifies the functional activity of the • Container closure cryopreservation compatibility: transgene product. Additionally, quantification of infec- (-80ºC is typical storage temp, container and closure tivity may be used as a supplemental approach to assess may contract at a different ratio, which may compro- the functional activity of AAV vectors. mise volume, aseptic conditions) Potency assays should take into account the product’s MOAs. However, this tends to be a large hurdle for gene CCIT evaluates the adequacy of the container clo- therapies due to the multifaceted nature of MOAs.5 sure systems and the ability of the container closure Many gene therapies have complex MOAs that rely on to maintain a sterile barrier against potential contam- multiple biological activities (transfection/infection, gene inants (e.g., microorganisms, reactive gases, or other transcription, translation, action of translated protein), substances). While CCIT may not be performed during and the MOAs of many are not fully characterized. Thus, the early-stage development of biologics or vaccines, it several stages of MOA must be captured within a single is important to consider early during development for a potency assay. We currently lack an in vitro to in vivo gene therapy due to the accelerated timelines involved translation for biological assays to determine whether in in bringing a gene therapy for a rare disease to market. vitro measurements (e.g., of potency, efficacy, cytotoxic- It is important that container closure systems maintain ity, and immunogenicity) are meaningful and predictive the sterility and product quality throughout the shelf life of physiological action. (until the expiration date) of the product. Regulatory

CHAPTER 6 Drug Product 169 requirements mandate that the design of the closure Storage and stability must be evaluated in conjunction system be qualified, which can be done in multiple with the container and closure systems; storage and sta- ways. The selection of the appropriate method is based bility should be evaluated early in the process, and the on the contents of the container as well as the container stability of the container system should be known in ad- closure system itself.11 Establishing a proper container vance (Table 6). Assays are required to verify stability and closure system is vital for product and consumer safety. ensure that purified clinical vectors maintain their purity, It is ideal to test it on the final product, but because of potency, and safety profiles during storage and over the the requirement of a large number of vials for testing, it course of their potential use.1 During early-phase clinical is challenging in gene therapy when every vial counts. In studies, stability studies can be performed concurrently some cases, it may be possible to do placebo-based CCIT with clinical use. Accelerated tests performed at 25ºC and in order to preserve the gene therapy product. However, 37ºC may be helpful to accelerate the stability studies and the ultimate proof of suitability of a container closure derive long-term stability data to inform shelf life and system and packaging is a full shelf life stability study.12 storage conditions.15 Container closure systems consist of both primary Many biologics are sensitive to pH, and pH may affect packaging components (those that come into direct the structure of the AAV capsid. pH must be studied early contact with the product such as a vial or syringe) and during the pre-formulation/early formulation phase in secondary packaging components (those that are vital to order to determine the optimal pH range for product ensure correct package assembly, such as aluminum caps stability. pH must be analyzed for compatibility with both over stoppers). Packaging materials must not interact the container material, diluent, and vector. Note that pH physically or chemically with the product to avoid effects may change during freezing due to crystallization of the on safety, identity, strength, quality, and purity.13,14 buffer components, temperature dependence of pH, and CCIT testing will vary based on the specific containers change in the apparent acid dissociation constant as a result and materials used, for example vials vs bags. Methods of the decrease in the polarity of the liquid phase due to that may be used to assess container closure can include: freeze-concentration.15 pH can affect stability based on se- rotype; whereas higher pH may result in increased stability • Electrical conductivity and capacitance test in some serotypes, it may result in lower stability in others. • Laser-based gas headspace analysis • Mass extraction AAV DP are generally stored as a frozen solution; • Pressure decay however, AAV formulations (and other aqueous biophar- • Tracer gas (vacuum mode) maceuticals) are not completely frozen at -15ºC to -25ºC • Vacuum decay and consist of ice and a freeze-concentrated solution. The freeze-concentrated solution (which is liquid) contains all Storage and Stability of the active ingredient and unfrozen water between ice crystals. Storage at -20ºC may lead to destabilization via Several factors may impact long-term storage and sta- various routes, such as aggregation, pH changes, increased bility. During stability testing, temperature fluctuation oxygen concentration, extensive ice/solution interface, and (stable temperature storage and freeze-thaw cycles), crystallization of cryoprotectors. When possible, storage diluents (including serum and solutes with various pH), at lower temperatures (below -65ºC) should be used.15 container constituents (glass, plastics, and steel), and Although storage at low temperatures may be other environmental considerations should be taken into practical, shipping and site storage of frozen biologics account. AAV is relatively stable at -80ºC, but this can becomes more difficult. Therefore, it is desirable to have translate into problems related to clinical administration. AAV formulations that are stable above 0ºC. However, Maintaining the cold chain at -80ºC with minimal ex- results of AAV stability at higher temperatures have been cursions is difficult, particularly when “public-friendly” conflicting. Whereas some report stability of AAV vector -80ºC is not available. It is important to note that stability in neutral phosphate-buffered saline with 5% sorbitol and may differ between AAV serotypes. 0.1% polysorbate 80 without losing transduction activity

CHAPTER 6 Drug Product 170 Table 6-6. Recommended Stability Parameters

Parameter Assay Recommendation Time of Assay Aggregation • DLS 0, 3, 6, 12, 18, 24 and months Potency • Protein expression 0, 3, 6, 12, 18, 24 and months • TCID50 Identity • Gene-specific physical titer (qPCR or ddPCR) 0, 3, 6, 12, 18, 24 and months Sterility • Refer to 0, 3, 6, 12, 18, 24 and months CCIT • Appropriate test for product/materials (refer to CCIT section) 0, 3, 6, 12, 18, 24 and months after 1 year at 2ºC to 8ºC, others have reported up to 40% is a framework that can be used in the development of a loss in transgene expression after 7 weeks at 4ºC for AAV1 robust final product. 15-17 in PBS with 0.5 mM MgCl2. A study of conditions com- CQAs typically considered during the fill-finish stage monly encountered in human gene therapy trials found include aggregation, potency, identity, sterility, and CCIT. that over a range of temperatures, pH, and environmental During formulation, major focus may shift to potency conditions, rAAV was found to be remarkably stable (4°C as the most important aspect, while still considering to 55°C), pH (5.5 to 8.5) in various container materials for stability, purity, and safety. The characterization stage is up to >1 year. The exceptions included heating to 72°C and broad, and the major focus continues to be on potency, exposure to UV for 10 minutes.18 although safety, identity, and purity are still considered. The Reference Standard Materials working group gen- CCIT must ensure compatibility of the DP with route of erated standards for AAV serotype 8. Three independent administration and container closure cryopreservation laboratories provided stability data on the AAV8 serotype compatibility. Finally, several variables must be evaluated (AAV8RSM) deposited to the American Type Culture in order to minimize factors that may impact long-term Collection (ATCC), which required relabeling and con- storage and stability, including temperature fluctuations, tainer cap tightening 2 years after the original deposition freeze-thaw cycles, diluents, container constituents, and due to labels detaching too easily and container caps that other environmental considerations. were not tightly closed. Following these corrective ac- Currently, gene therapy is limited in supply due to tions in 2016, AAV8RSM demonstrated consistent titers a manufacturing bottleneck. However, progress in the using qPCR, TCID50, and ELISA analyses compared to gene therapy field is rapid, and once the manufacturing titers at deposition into ATCC in 2014, and capsid pro- bottleneck is overcome, it will be prudent to focus on tein integrity (SDS-PAGE) was equivalent at 2 years after solving possible rate-limiting steps during the fill-finish deposition at appropriate storage conditions (≤70°C).19 stage. For example, automated filling technology utilizing engineered tubing to eliminate product loss, research Conclusion on polymers used in vials, lyophilization of product, and increased temperature resilience of the DS are im- Starting from the beginning of gene therapy product provements that we look forward to in the next decade development, it is critical to consider the “end goal” with of AAV-based gene therapies. Such industrial improve- regards to desired drug product characteristics and how ments will ensure that the fill-finish process does not to achieve that using standardized, informed design pro- become a rate-limiting step in getting vital gene therapies cesses. In this chapter, we have outlined the best practices to the patients who need them. for fill-finish, formulation, characterization, and CCIT; however, it is in no way exhaustive. Each process will differ based on the specific product, its characteristics, administration route, titer selection, final formulation buffer, and container choice, among others. This chapter

CHAPTER 6 Drug Product 171 Endnotes 1. Wright JF. Manufacturing and characterizing AAV- 11. Ewan S, Jiang M, Stevenson C, et al. White paper: based vectors for use in clinical studies. Gene Ther. Container closure integrity control versus integrity 2008;15(11):840-848. testing during routine manufacturing. PDA J Pharm Sci Technol. 2015;69(3):461-465. 2. Wright JF, Le T, Prado J, et al. Identification of factors that contribute to recombinant AAV2 particle 12. US Food and Drug Administration. Guidance for aggregation and methods to prevent its occurrence industry: container closure systems for packaging hu- during vector purification and formulation. Mol man drugs and biologics. US Food and Drug Admin- Ther. 2005;12(1):171-178. istration website. https://www.fda.gov/media/70788/ download. Updated 1998. Accessed June 18, 2020. 3. Wright JF, Hauck B, Zhou S, Zelenaia O, Sumaroka M, High KA. Concentration of AAV8 and 9 vector 13. USP. Packaging and storage requirements. USP to titers ranging from 1-6 x 1014 vg/mL: feasibility website. https://www.uspnf.com/sites/default/files/ assessment for volume-limited dosing (abstract 547). usp_pdf/EN/notices/2018/c659-proposed.pdf. . Ac- Mol Ther. 2013;21(Suppl 1):S210-211. cessed June 18, 2020. 4. Hocquemiller M, Giersch L, Audrain M, Park- 14. ICH. Pharmaceutical development Q8(R2). ICH er S, Cartier N. Adeno-associated virus-based website. https://database.ich.org/sites/default/files/ gene therapy for CNS diseases. Hum Gene Ther. Q8_R2_Guideline.pdf. Updated August 2009. Ac- 2016;27(7):478-496. cessed June 18, 2020. 5. The National Institute for Innovation in Manufac- 15. Rodrigues GA, Shalaev E, Karami TK, Cunningham turing Biopharmaceuticals. Gene therapy roadmap. J, Slater NKH, Rivers HM. Pharmaceutical devel- National Institute for Innovation in Manufacturing opment of AAV-based gene therapy products for the Biopharmaceuticals website. https://niimbl.force. eye. Pharm Res. 2018;36(2):29. com/resource/1541788844000/NIIMBLGeneThera- pyRoadmap. Accessed May 19, 2020. 16. Wright JF, Qu G, Tang C, Sommer JM. Recombinant adeno-associated virus: formulation challenges and 6. Luxturna [prescribing information]. Philadelphia, strategies for a gene therapy vector. Curr Opin Drug PA: Spark Therapeutics, Inc.; 2017. Discov Devel. 2003;6(2):174-178. 7. Zolgensma [prescribing information]. Bannockburn, 17. Howard DB, Harvey BK. Assaying the stability and IL: AveXis, Inc.; 2017. inactivation of AAV serotype 1 vectors. Hum Gene Ther Methods. 2017;28(1):39-48. 8. US Food and Drug Administration. Summary basis for regulatory action. US Food and Drug Adminis- 18. Gruntman AM, Su L, Su Q, Gao G, Mueller C, tration website. https://www.fda.gov/media/127961/ Flotte TR. Stability and compatibility of recombinant download. Updated May 24, 2019. Accessed June 30, adeno-associated virus under conditions commonly 2020. encountered in human gene therapy trials. Hum Gene Ther Methods. 2015;26(2):71-76. 9. US Food and Drug Administration. Guidance for industry: FDA guidance for human somatic therapy 19. Penaud-Budloo M, Broucque F, Harrouet K, et al. and gene therapy. US Food and Drug Administration Stability of the adeno-associated virus 8 reference website. https://www.fda.gov/media/72402/down- standard material. Gene Ther. 2019;26(5):211-215. load. Updated 1998. Accessed June 14, 2020. 20. Penaud-Budloo M, François A, Clément N, Ayosu 10. US Food and Drug Administration. Guidance for E. Pharmacology of recombinant adeno-associat- industry: potency tests for cellular and gene therapy ed virus production. Mol Ther Methods Clin Dev. products. . US Food and Drug Administration web- 2018;8:166-180. site. https://www.fda.gov/media/79856/download. Updated January 2011. Accessed June 14, 2020.

CHAPTER 6 Drug Product 172 Chapter 7 Process Control Strategy Chapter 7 | Contents

Chapter Summary...... 175 Key Points...... 175 Elements of Control Strategy...... 176 Assessment of Product Quality Attributes...... 177 Element 1: In-Process Control of Product Quality Attributes...... 178 Element 2: Control of Process Parameters...... 179 Element 3: Control of Process Performance...... 180 Element 4: Non-Routine Product Characterization Testing...... 180 Elements 5 and 6: Product Control Through Release and Stability Testing...... 180 Element 7: Control of Materials...... 181 Material Risk Assessment...... 181 Starting Materials...... 181 Excipients...... 183 Ancillary Materials...... 183 Tier 1...... 183 Tier 2...... 183 Tier 3...... 183 Tier 4...... 184 Specification Development...... 184 Retain Strategy and Stability...... 184 Use Periods...... 184 Lifecycle Approach: Ancillary Materials Control Strategy...... 185 Process Risk Assessment...... 186 Clinical Supply Chain Strategies...... 188 Shipping and Distribution...... 188 Packaging and Labeling...... 189 Dose Preparation and Administration...... 189 Recommended Readings...... 189

CHAPTER 7: Process Control Strategy 174 Chapter Summary A comprehensive control strategy for a using principles of pharmaceutical de- pharmaceutical product (including gene velopment, quality risk management, therapy products) is the key to achieving and quality systems, is used for quality process consistency, product quality, product development, robust dossier safety, and efficacy. The International submission and review, inspection, and Council for Harmonization (ICH) Q10 post-approval changes to ensure both defines control strategy as “[a] planned set product quality and process consistency. of controls, derived from current product A comprehensive control strategy enables and process understanding, which assures the design of the appropriate process vali- process performance and product quali- dation and continued process verification ty. The controls can include parameters (CPV) program for implementation and attributes related to drug substance during the product life cycle. CPV mon- (DS) and drug product (DP) materials itoring and trending may identify areas and components, facility and equipment of improvement in the process, leading operating conditions, in-process controls, to the evolution of the control strategy. finished product specifications, and the The GMP control strategy spans from cell associated methods and frequency of bank manufacturing to final dosage form monitoring and control.” delivered to a clinical site. The elements that contribute to control This section describes the approaches strategy are outlined in ICH guidelines to developing a comprehensive control ICH Q8 (R2), Q9, Q10, and Q11. These strategy and a detailed overview of the key guidelines describe a quality paradigm in elements of a control strategy for a gene which a scientific and risk-based approach, therapy product.

• The control strategy determines where in and critical process parameters (CPPs) of the process to place appropriate controls the process. to consistently ensure product quality, • Adjustments to the control strategy (e.g., safety, and efficacy. providing additional controls to further • This is accomplished through an iterative reduce risk, moving control points to risk assessment process that considers optimum location within the process, evolving product and process knowledge or removing controls determined to be that lead to the identification of the critical redundant or ineffective) can be made quality attributes (CQAs) of the product during this iterative process, as needed.

CHAPTER 7: Process Control Strategy 175 assessed, and the critical process controls that are linked to control of CQAs are categorized as established conditions. A more detailed description of types of control that could ElementsElements of Control of StrategyControl Strategy riskbe assessed,considered and for the each critical control process element controls is that provided are in The overall control of the process and product should linkedTable to 1. control of CQAs are categorized as established be examinedThe overall holistically control of andthe processsystematically. and product There should are conditions. eightbe elements examined of holistically control that and the systematically. product team There should are InA the more regulatory detailed descriptiondossier (eg, of BLA), types keyof control control that strategy considereight forelements each product of control or that process the product attribute team to shouldensure couldelements be considered focused foron eachthe control control elementof CQAs, is providedare filed in the a robustconsider control for eachstrategy. product These or process elements attribute are depicted to ensure in inrespective Table 1. DS and DP leaflets. Figurea robust 1. control strategy. These elements are depicted in In the regulatory dossier (e.g., BLA), key control Figure 1. strategyAssessment elements of Product focused onQuality the control Attributes of CQAs, are All product and process inputs and outputs are filed in the respective DS and DP leaflets. All product and process inputs and outputs are risk Iterative assessments of QA criticality are conducted

FigureFigure 7-1. Eight 7-1. ElementsElements of of Control Control for for Product Product Quality Quality and Processand Process Performance Performance

In-process control of 8 product quality 2 attributes Control of Facility and process equipment parameters controls

Control of Control Strategy for Control of 7 materials Product Quality and process 3 Process Performance performance

Product control Non-routine through stability product testing characterization testing Product control 6 through quality attribute testing 4

CHAPTERCHAPTER X: xxxx 7: Process Control Strategy 176 4 Assessment of Product Quality reviews of available information in the public domain, including literature and publications. Attributes Scoring of QAs for A-Gene DS and DP is performed Iterative assessments of QA criticality are conducted taking into consideration the QA’s relation to the QTPP during product development to identify CQAs. This pro- and the potential impact of the DS attributes on safety and cess includes a review of prior knowledge and periodic efficacy of the drug product. Based on prior knowledge and

Table 7-1. Elements of Control for Product Quality and Process Performance

Control Element Definition Type of Control PROCESS CONTROL Element 1 In-process control • Direct in-process tests, or surrogates, of product quality attributes and of product quality their control limits attributes • Product quality attribute in-process control demonstrated through process validation Element 2 Control of process • Control implicit in the design of manufacturing process or unit parameters operations • Manufacturing process controls • Process parameters that impact product quality or process performance attributes and control limits/acceptable ranges • Manufacturing hold time control limits • Manufacturing process development and history for understanding and application of acceptable ranges Element 3 Control of process • In-process tests of process performance and their control limits performance PRODUCT CONTROL Element 4 Non-routine product • Elucidation of structure and other characteristics characterization • Non-routine tests for characterization and demonstration of product test-ing comparability • Characterization included in reference standard or reference material qualification Element 5 Product control • Routine release test and acceptance criteria in product specification through quality • Justification of specification attribute testing • Analytical procedure and its validation • Product quality attribute control demonstrated through process validation

Element 6 Product control through • Routine stability test and acceptance criteria in stability protocols stability testing • Stability data and conclusions

FACILITY AND MATERIAL CONTROL Element 7 Control of materials • Specifications for raw materials • Manufacture and testing of cell banks; cell bank controls • Characteristics of incoming materials (such as raw materials, starting material, intermediates, primary packaging materials) that impact product quality attributes and their acceptable ranges • Compatibility with container closure system; container closure controls Element 8 Facility and equipment • cGMP and pharmaceutical quality systems for the manufacturing facilities controls • Environmental and equipment controls directly impacting product quality attributes

CHAPTER 7: Process Control Strategy 177 Table 7-2. Description of CQA Severity Scores

Score Severity Score Definitions

10 Major Potentially serious impact to patient, may be life threatening or irreversible

7 Moderate Moderate impact on patient – treatable AEa, no permanent harm

5 Minor Low impact on patient – temporary inconvenience/impairment

1 Negligible No patient harm

aNote: impact to patient includes both safety and efficacy; lack of efficacy is considered an adverse event (AE). compendial or regulatory expectations, certain attributes Table 7-3. Description of CQA Uncertainty Scores are not formally scored but are identified as obligatory CQAs. Control of these CQAs through routine test- Score Uncertaintya ing with predefined acceptance criteria is considered 10 Low confidence or no information mandatory. 6 Medium confidence Other QA categories (e.g., purity) are individually evaluated. When applicable, the data from relevant 4 High confidence literature, prior knowledge, structure-function studies, 2 Prior knowledge (well established stress studies, stability studies, and clinical data are used understanding) in the determination of QA criticality. Each category is aCould be based on relevant literature, prior knowledge, in vitro scored separately based on the potential impact (sever- or in vitro S/F study, clinical data, etc. ity) of the attribute and with respect to the uncertainty Table 7-4. Criticality Assignment Matrix associated with the impact. In general, less experience (Potential CQAs are in red) or knowledge with a particular QA leads to a higher uncertainty score. A quality attribute is determined to Severity be a CQA if, in the evaluation of the two categories, any of the scores exceeds the predefined criticality 10 7 5 1 threshold that accounts for the combination of severity 10 and uncertainty scores. The severity threshold matrix is provided in Table 4 below, where a red cell indicates 6 a potential CQA. 4 Severity scores may be defined as provided in Table 2. Uncertainty The definitions are based on the potential impact of an -at 2 tribute being outside of its acceptable range. Uncertainty scores may be defined as provided in Table 3. The crit- icality assignment matrix and criticality determination Table 7-5. Criticality Determination are shown in Table 4 and Table 5, respectively. Designation Severity Uncertainty ELEMENT 1: IN-PROCESS CONTROL OF PRODUCT Score Score QUALITY ATTRIBUTES Potential CQA ≥7 Any value For certain quality attributes, in-process testing can be Potential CQA 5 ≥6 selected as a means of control. The use of in-process Non-CQA 5 <6 testing provides opportunities for real-time monitoring of quality attributes. Table 6 and Table 7 show the in-pro- Potential CQA 1 10 cess tests for A-Gene DS and DP, respectively. Non CQA 1 <10

CHAPTER 7: Process Control Strategy 178 ELEMENT 2: CONTROL OF PROCESS PARAMETERS Although in many cases, the relationships between On a high level, the A-Gene manufacturing unit operations process parameters and CQAs for gene therapy processes are similar to the biologics manufacturing processes and are complex and not fully understood, the control of include: process parameters itself can be directly leveraged from • Upstream unit operations focused on production of well-known and established process control strategies. recombinant viral particles The most unique aspect of gene therapy manufac- • Downstream unit operations consisting of purifica- turing is a transient transfection step. The traditional tion and polishing steps transfection methods for rAAV rely on the transient • Formulation and fill-finish steps transfection of HEK-293 cells facilitated by a transfection

Table 7-6. In-process Monitoring and Controls for A-Gene Drug Substance

Process Step In-process Test Acceptance Criteria or Rationale for Designation Action Limit AAV vector Bioburden (in compliance , Ph. Eur. criteria to demonstrate that the cell (unprocessed 2.6.12, JP 4.05) culture process is free of contam- bulk material) inants, including mycoplasma, Mycoplasma (by direct Not detected bacteria, and adventitious viruses inoculation agar and broth assays, use of indicator cells and fluorochrome stain in compliance with USP and Ph. Eur. 2.6.7) Adventitious virus (by in Not detected vitro virus assay using indi- cator cell lines (MRC-5, Vero, HeLa); 28-day duration) Harvest ddPCR or qPCR In-process titer monitoring for forward processing to affinity step Bioburden Assess if exceeded: 10 cfu/mL Standard assays to help ensure product safety and control during Endotoxin Assess if exceeded: 5 EU/mL manufacture of DS Affinity qPCR In-process titer monitoring for Purification forward processing to AEX step Bioburden Assess if exceeded: 10 cfu/mL Standard assays to help ensure product safety and control during Endotoxin Assess if exceeded: 5 EU/mL manufacture of DS Ion exchange Bioburden Assess if exceeded: 10 cfu/mL Standard assays to help ensure Chromatography product safety and control during Endotoxin Assess if exceeded: 5 EU/mL manufacture of DS Nanofiltration qPCR In-process titer monitoring for pool forward processing to UF/DF step Bioburden Assess if exceeded: 10 cfu/mL Standard assays to help ensure product safety and control during Endotoxin Assess if exceeded: 5 EU/mL manufacture of DS UF/DF Bioburden Assess if exceeded: 10 cfu/mL Standard assays to help ensure product safety and control during Assess if exceeded: 5 EU/mL manufacture of DS

CHAPTER 7: Process Control Strategy 179 Table 7-7. In-Process Tests for A-Gene Drug Product

Process Step In-process Controls Acceptance Criteria

Pre-sterile filtration Bioburden (in compliance with USP , <1 cfu/mL Ph. Eur. 2.6.12, JP 4.05)

Sterile filtration Pre- and post-use filter integrity Pass

Sterile filling Fill weight check 1% to 2% of the target (filler capability) agent. This method presents the biggest challenge for Similarly, the drug product sterile filtration and filling process control, as the transfection efficiency is highly process performance are established by in-process moni- dependent on multiple factors/process parameters, in- toring and through batch instructions. Some of these may cluding pH, concentration of solution components, and be CPPs depending on impact to product quality, including: kinetics of the transfection reaction. In instances where clinical doses are high and the • Bulk DS thaw temperature (controlled vs uncontrolled) production process yields are not high enough, supplies • Filtration pressure/flow rate going to the clinic need to be maximized, which may • Mixing speed require pooling of multiple DS sub-lots into one DS • Mixing time lot prior to converting into drug product. This allows • Hold time(s) for a significant reduction in material that is used for • Filling speed release testing, stability, and retained samples, a critical • Capping pressure consideration in gene therapy given the value of the DP. ELEMENT 4: NON-ROUTINE PRODUCT ELEMENT 3: CONTROL OF PROCESS CHARACTERIZATION TESTING PERFORMANCE Some quality attributes are best evaluated via height- Control of process performance (robustness and con- ened characterization testing rather than by cGMP sistency) is established through defined batch instruc- release testing. These tests are not necessary for rou- tions, in-process monitoring, and in-process testing to tine testing to ensure product control and some are ensure that the selected unit operations are performing not GMP-compatible, but they do provide supportive adequately to achieve the intended product quality. For information during product development and can con- example, in-process monitoring and batch instructions tribute to the overall control strategy. For this relatively during cell expansion can include the following: new modality, product characterization testing will be utilized to further define the additional elements of • Incubator CO2 concentration (%) control strategy (e.g., release testing and stability). Table • Incubator temperature (ºC) 8 highlights several analytical tools used for in-depth • Incubator shaker speed (RPM) 1” throw characterization of A-Gene. • Media equilibration temperature (ºC) • Media equilibration duration (hours) ELEMENTS 5 AND 6: PRODUCT CONTROL • Media volume (mL) THROUGH RELEASE AND STABILITY TESTING • Inoculum volume (mL) In addition to process controls (elements 1 and 2), CQAs • Initial cell density (x106 viable cells/mL) may be controlled with cGMP release and stability test- • Post-inoculation working volume (mL) ing. Table 9 includes a panel of release and stability tests • Batch duration (hours) that may be appropriate for an AAV gene therapy. • Final cell density (x106 viable cells/mL) • Final cell viability (%)

CHAPTER 7: Process Control Strategy 180 Table 7-8. Heightened Characterization Testing Panel

Quality Attribute Analytical Test(s) Rationale

Capsid size SEC-MALS Properly formed capsids are expected to have a consistent DLS size

Capsid proteins rCGE Properly formed viral particles are expected to have a RP-HPLC consistent ratio of VP1, VP2, and VP3 proteins SDS-PAGE

Molecular mass ESI-MS Properly formed viral particles are expected to have a consistent mass

Primary capsid sequence Peptide mapping by Evaluating the primary capsid structure ensures capsid identity LC/MS and enables characterization of posttranslational modifications

Particle content AUC The ratio of empty, intermediate, and full viral particles can be assessed as part of manufacturing consistency. Full particles are the API while empty particles are process-related impurities

Vector genome sequence NGS Confirmation of the ITR and transgene sequence of the viral vector ensures identity; as bioinformatics tools advance, NGS may also be useful for sequence variant analysis and characterization of impurities

Residual impurities (e.g., PEI, Varied (e.g., RP-HPLC) Confirmation of the removal of residual process-related anti-foam) impurities from the product may be used to support process development and process validation

ELEMENT 7: CONTROL OF MATERIALS Starting materials, ancillary materials, and excipients Raw materials for gene therapy are classified as any com- can be available as commercial off-the-shelf (COS) or as ponent or reagent intended for use in the production custom materials. of ATMPs, including those that may or may not appear in the finished product. A formal risk assessment is re- • Custom materials are materials that are part of quired for all raw materials used in the manufacturing the final product that are genetically modified and process that establishes the relationship between raw custom manufactured per contracted specifications material attributes, process performance, and product (e.g., plasmids that are not off-the-shelf or commer- quality attributes. All raw materials can be divided into cially available). the following categories: • Commercial off-the-shelfrefers to materials that are commercially available. • Starting materials are materials that are the starting point at which cGMP process manufacturing principles are applied. Material Risk Assessment • Ancillary materials are components, reagents, or materials used during the manufacture of a gene STARTING MATERIALS therapy product that are not intended to be part cGMP starting material should be used wherever pos- of the final product. These are materials used as sible, but the use of materials that are classified as High processing and purification aids or agents that exert Quality is acceptable, taking into account the clinical their effect on the therapeutic substance. trial phase of the final product, manufacturing process • Excipients are components or reagents used in the controls, and QC testing. The rationale for their use formulation of the final gene therapy product. and qualification strategy must be documented in the

CHAPTER 7: Process Control Strategy 181 Table 7-9. Drug Substance and Drug Product Release Testing Panel

Quality Attribute Analytical Test(s) Rationale Used for DS, DP and/or S CHARACTERISTIC Clarity Appearance Compendial DS, DP, S for both Coloration Appearance Compendial DS, DP, S for both Visible particles Appearance Compendial DP, S Sub-visible particles Sub-visible particles Compendial DP, S pH pH Compendial DS, DP, S for both Osmolality Osmolality Compendial DS, DP Extractable volume Extractable volume Compendial DP Viral particle titer SEC-HPLC; ELISA Measures total viral particles DS, DP, S for both IDENTITY Capsid identity Peptide map by RP- Ensures intended capsid is present DS, DP HPLC; ELISA Vector genome qPCR, restriction map, Ensures intended vector genome is present DS, DP identity sequencing POTENCY Vector genome titer qPCR, ddPCR Vector genome concentration used for dose DS, DP, S for both determination Potency - infectivity Infectious virus titer Infectious virus titer for lot-to-lot comparison DS, DP, S for both (TCID50) May not be needed if quantitative expression or functional assay is in place Relative infectivity Alternative to TCID50 Measure delivered DNA DS, DP, S for both by ddPCR, relative to a reference standard Potency - expression Cell-based assay with Demonstrates that product can infect cells DS, DP, S for both mRNA or immunoas- and express protein of interest say readout (RT-qPCR, May not be needed if a quantitative functional ELISA, Western blot, assay is in place etc.) Potency - activity Cell-based assay with Best expressed as relative potency when DS, DP, S for both readout relevant to compared to an assay standard the therapeutic MOA May be suitable to replace individual infectivity (e.g., enzymatic activ- and expression assays as it encompasses both ity assay) and adds a functional readout PURITY Capsid protein purity rCGE; RP-HPLC; SDS- Provides a measurement of capsid protein DS, DP, S for both PAGE purity Particle content UV260/UV280 ratio Provides an indirect measurement of the DS, DP percentage of full viral particles Particle aggregation SEC-HPLC Provides a measurement of capsid DS, DP, S for both aggregation

CHAPTER 7: Process Control Strategy 182 Table 7-9. Drug Substance and Drug Product Release Testing Panel continued from previous page

SAFETY Endotoxin Endotoxin Compendial DS, DP, S for both rcAAV rcAAV cell-based assay Ensures the product does not contain DS rcAAV Bioburden Bioburden Compendial DS Sterility Sterility (CCITa can be Compendial DP, S used as a surrogate on stability) PROCESS-RELATED IMPURITIES

Residual HCP ELISA Ensure control of impurities DS

Residual affinity ligand ELISA Ensure control of impurities DS

Residual benzonase ELISA Ensure control of impurities DS

Residual BSA ELISA Ensure control of impurities DS

Residual host cell DNA qPCR Ensure control of impurities DS

Residual plasmid DNA qPCR Ensure control of impurities DS risk assessment along with the controls and mitigations Tier 1 required that ensure the quality, safety, efficacy, and • Tier 1 materials are low-risk, highly qualified traceability of the material. ancillary materials suitable for use in manufacturing that are either a licensed biologic, an approved drug, EXCIPIENTS or an approved or cleared medical device. In general, excipients should comply with the guidance in • Examples include injectable monoclonal antibodies, USP-NF General Chapter Good Manufacturing cytokines, vitamins. Practices for Bulk Pharmaceutical Excipients. In March 2015, the Official Journal of the European Union adopted Tier 2 “Guidelines on the formalized risk assessment for ascer- • Tier 2 materials are low-risk, well-characterized taining the appropriate good manufacturing practice for ancillary materials suitable for use in manufacturing excipients of medicinal products for human use.” that are produced under relevant cGMPs. • Examples include recombinant growth factors, cyto- ANCILLARY MATERIALS kines, sterile process buffers, USP-grade chemicals. The following risk classifications were based on the ancillary material qualification risk classifications per Tier 3 USP 1043, “Ancillary Materials for Cell, Gene, and • Tier 3 materials are moderate-risk ancillary materials Tissue-engineered Products.” Both the USP and the EMA that require a higher level of qualification. “Guideline on Good Manufacturing Practice Specific to • These may include custom materials that are defined Advanced Therapy Medicinal Products” provide a gen- as high quality (or similar terminology) and have eral overview of the materials used in the manufacture of been manufactured under a quality system following ATMP and include the guidance on establishing a risk- the principles of GMP with regards to manufactur- based approach for materials used in ATMPs throughout ing, QA oversight, and QC testing. the product lifecycle. • The rationale for their use and qualification strategy

CHAPTER 7: Process Control Strategy 183 Table 7-10. Examples of Materials

Starting Materials Ancillary Materials Excipients

• Bacterial, insect, or mammalian cell • Sterile process buffers • Nonionic surfactant (Pluronic F-68, line carrying the gene to produce • Media components (growth factors, Polysorbate 80, Polysorbate 20) the therapeutic material cytokines, vitamins) • Polyols (glycerol, sorbitol, mannitol, • Host or packaging cell lines of • Animal-derived (including human) polyethyleneglycol) bacterial, insect, or mammalian extracts • Amino acids origin • FBS • Mono- and divalent salts • Virus seed stocks • Virus master banks • Transgene plasmids • Plasmids • Cell lines

must be documented in the risk assessment along • Identity (e.g., visual inspection of material labels, with the controls and mitigations required that analytical methods [sequencing, mass spectrometry, ensure the quality, safety, efficacy, and traceability of ELISA, chromatography, restriction mapping, gel the material. electrophoresis, etc., that identify material active • This tier classification may also be materials pro- ingredients) duced for in vitro diagnostic use or reagent grade • Purity (e.g., HPLC technologies, gel electrophoresis, that were not intended for use in the production of UV260/280, bioanalyzer) gene therapy products. • Safety (e.g., sterility or qualified RMM, mycoplasma • Upgrade of the manufacturing processes may be or qualified RMM, bioburden, endotoxin, adventi- necessary in order to use the material in manufactur- tious agent testing including specific viral testing or ing gene therapy products. qualified RMM) • Examples include diagnostic-grade chemicals, novel • Functionality (where appropriate) polymers, and process buffers. RETAIN STRATEGY AND STABILITY Tier 4 The retain strategy for in vivo gene therapy product • Tier 4 materials are highest-risk ancillary materials materials is performed using a risk-based approach that that require extensive qualification prior to use in factors in both business and quality risk assessments. manufacturing that are not produced in compliance Materials used in the manufacture of a GTMP product with cGMPs. may be placed in a stability program depending on the • Developers in the early stages of development should stage of development. evaluate the necessity of these materials and explore alternative substances or sources. USE PERIODS • Examples include FBS, animal-derived (including If a supplier has provided a use period or expiry date, the human) extracts, animal-derived polymers, scaffolds, supplier’s date may be used if determined to be suitable. and hydrogels. Suppliers may have different review dates for the same material based on their manufacturing process and SPECIFICATION DEVELOPMENT stability profile. If a supplier review date is not provided Specification development refers to testing for starting for a material, a documented risk assessment should be materials, ancillary materials, and excipients, and may used to establish an appropriate review or expiry date. include testing for the following quality attributes:

CHAPTER 7: Process Control Strategy 184 LIFECYCLE APPROACH: ANCILLARY MATERIALS In some cases, the initial qualification of an ancil- CONTROL STRATEGY lary material for use in manufacturing should be the In cases when ancillary materials are chosen for their abil- investigation of the effect of the amount of the ancillary ity to provide a particular biological function in producing material on the desired response (increased yield, purity, the therapeutic product, performance testing becomes an or potency of the therapeutic product). The amount of essential component of their overall qualification. This is the ancillary material used in manufacturing should especially true when the ancillary material plays a critical be chosen to consistently yield the desired effect while role in modulating a complex biochemical effect and has minimizing issues by removing the ancillary material an impact on product manufacturing yield, purity, or final in subsequent processing steps. Such testing frequently product potency. These ancillary materials tend to be com- assesses the important functional attribute expected of plex substances or mixtures, are frequently biologically the ancillary material in a scaled-down or simulated sourced, and can exhibit significant lot-to-lot variability. manufacturing process. Examples include: As a result, these ancillary materials usually have no simple identity test, nor can they be easily characterized by phys- • If an ancillary material is added to the culture media ical or chemical tests. The development of well-defined because it promotes cellular proliferation or the performance assays for complex ancillary materials will secretion of a critical therapeutic agent, the assay could not only ensure process reproducibility and final product demonstrate that each lot of ancillary material produc- quality, but in many cases will satisfy the identity testing es the expected rate and amount of cellular prolifera- criteria in accordance with 21 CFR 211.84(d)(11). tion or the expected level of secreted therapeutic agent.

Figure 7-2. Risk Assessment Milestones

Process Performance Qualification (PPQ)

Understand impact of Revise C&E scoring based Revise C&E and FMEA process parameters on on experimental output risk assessments to QA based on product understand process knowledge through cause FMEA risk assessment parameter effect on CQA and effect matrix scoring to understand process based on PPQ and other parameter effect on CQA batch experience

OUTCOME: OUTCOME: OUTCOME: • Identify and prioritize • List of CPP and non-CPP • Final list of CPP and non-CPP RA1 list of lab studies and RA2 with potential to affect CQA RA3 with potential to affect CQA experimental strategy • PPQ control strategy • Commercial control strategy • PPQcriteria

PPQ and Lab studies: additional Small-scale batch and at-scale experience

Figure content courtesy of Iryna Sanders.

CHAPTER 7: Process Control Strategy 185 • If a deoxyribonuclease is used to degrade cellular Table 7-11. Example of Typical Scoring Criteria for DNA, new lots could be tested for the ability of the Process Parameters deoxyribonuclease to degrade DNA. • If a particular type of density gradient material is 10 Strong relationship known based on the data used to purify a vector, new lots of the material used in hand or experience to make the gradient could be shown to purify the 9 Do not know but expect there is a strong vector to an acceptable level. relationship • If a plasmid or viral vector is used in the production of a gene therapy vector (e.g., helper function), new 6 Known medium to low relationship lots of the helper vector could be shown to produce 4 Do not know but expect there is a medium to the expected amounts of the gene therapy vector. low relationship • The requirement to assess lot-to-lot effect on process 1 Know there is not a relationship performance for ancillary materials may require mitigation for Tier 3 and Tier 4 materials.

Table 7-12. Example of Typical Scoring Criteria for Process Risk Assessment Attributes

To develop a manufacturing process that consistently and 10 Direct impact on product safety and/or reproducibly operates and delivers product that meets efficacy is expected or established the desired quality, the approach used is based on un- derstanding the relationships between process inputs and 7 Moderate or indirect impact on safety and/or efficacy; direct impact on process efficiency output attributes in each unit’s operation over the entire manufacturing process. For this, multiple systematic risk 5 Low or unlikely impact on product safety and/ assessments are conducted throughout the development or efficacy; moderate or indirect impact on process efficiency lifecycle to identify process steps, material attributes, equipment design, and operation parameters that would 1 No impact on product safety and/or efficacy; be most likely to impact DS and/or DP QAs. The three low or unlikely impact on process efficiency major risk assessment milestones are depicted in Figure 2, but additional risk assessments can be completed based on program needs. Tools such as C&E tables and FMEA Process Parameter (PP) = Process Input, “x” tables can help facilitate risk assessment discussions. Process parameters refer to machines, materials, measure- Risk Assessment 1 is recommended to occur in parallel ments, processes, people, and environments. This broad with the decision to commence late-stage development definition is appropriate for first-round risk assessments in and the initiation of moving the project to the commercial which the team may choose to look at a variety of inputs, manufacturing scale or site. Many risk assessment tools some of which may be non-numerical (e.g., the same raw can be used for teams to evaluate the current process and materials from two vendors). Table 11 shows an example product understanding to prioritize process characteri- of typical scoring criteria for process parameters. zation activities. The tool shown here, as an example, is the use of C&E tables. These tables have all process inputs Attribute = Process Output, “y” listed as rows and all process outputs (including product Attributes refer to physical, chemical, or microbiological quality attributes) listed as columns. One relationship score properties or characteristics of a material (or process). is given for each input and each output. The scores can be Attributes include both product quality attributes (QA, ranked with numerical values representing high, medium CQA) and process performance attributes (PPA). and low relationships. A high score is neither good nor Table 12 shows an example of typical scoring criteria bad; it only defines the strength of the relationship. for attributes.

CHAPTER 7: Process Control Strategy 186 Table 7-13. Extract of C&E Risk Assessment for the Gene Therapy AAV Affinity Chromatography Process Unit of Unit Measure pH Effluent Process Time Resin Capacity Resin Lifetime Poros Step AAVX Recovery Elution Pool Volume Total

Attribute Score 5 7 7 7 7 7 Bed Height cm 1 10 10 1 4 10 250 Column Reuse # Cycle # 1 1 9 10 10 9 278 Integrity - Asymmetry Not Specified 1 1 1 1 1 4 61 Integrity - HETP plates/m 1 1 1 1 1 4 61 Mass Loading on Resin VG/mL 1 10 1 9 10 9 278 Number of Cycles Per Not specified 1 10 1 4 1 1 124 Batch Process Temperature °C 1 1 1 1 1 1 40 Differential Pressure psi 1 1 1 1 1 1 40 Volumetric Challenge L/m² 1 1 1 1 1 1 40 Equilibrium Buffer pH pH 10 1 1 1 1 1 85 Equilibrium mS/cm 4 1 1 1 1 1 55 Conductivity

The C&E matrix involves a team-based assessment of the 1 serves as the basis for Risk Assessment 2, which occurs relationship between process parameters and attributes. The prior to PPQ. During this risk assessment, the C&E ta- process attributes are scored first and then the relationship bles are revised to capture increased product and process between each given process parameter is scored, enabling understanding. FMEA tables are also used during this a total relationship score for each process parameter. Team risk assessment to understand control of process pa- members will typically discuss the nuances of score, but this rameters as it relates to product quality and to develop process enables a prioritization of parameters to be included a strategy for PPQ (e.g., defining parameters that must in further process characterization studies. be demonstrated to be in control during production). The output of Risk Assessment 1 is used to guide process These assessments contribute to a robust product and development activities and includes prioritization of process process understanding that ensures an appropriate steps and parameters for optimization and characterization. control strategy and is validated during PPQ. The fol- Experimentation is carried out with univariate and multi- lowing risk categories are defined in the FMEA tables, variate studies (e.g., DOE) as appropriate using scale-down and scores relative to the risk are shown in Table 14: models to establish parameter-attribute relationships and identify robust operating conditions and acceptable process • Severity (S): scored based on the potential impact of ranges. Results from the DOE studies provide an under- the step on the CQA in context of the overall process. standing of the multidimensional relationships between • Occurrence (O): scored based on the probability of the input process parameters and output quality attributes. critical process parameters exceeding the acceptable Additionally, clinical manufacturing experience provides range. understanding of process performance and process control • Detection (D): how well can the failure of the critical at various operational scales. process parameters be detected prior to completion of The information gathered following Risk Assessment the step.

CHAPTER 7: Process Control Strategy 187 Table 7-14. Example Scoring Rubric for FMEA

Score Severity Occurrence Detection 9 Failure to meet DS or DP specifications or >20% No way to detect excursion; not quality target leading to lot rejection. Very frequent tracked and not alarmed. Complete failure of process step to meet intended purpose 7 Potential for variation outside specified ~5% to 20% Difficult to detect excursion, and ranges or limits for product quality and/or Frequent not until after it has impacted the consistency, or variation outside historical process. ranges where specifications or quality targets are not defined. Investigation needed prior to product release. 5 Potential variation within specified ranges ~1% to 5% Excursion can be detected, but or limits for product quality and/or process Occasional not until after it has impacted the performance attributes, or variation within process. historical ranges where specifications or quality targets are not defined. 3 No impact on product quality. Potential for <1% Excursion is usually detected and minor variation in process performance Rare corrected prior to impacting the attributes (e.g., yield). process. 1 No impact to process performance 0% Excursion is obvious and always attributes or product quality. Never observed detected prior to impacting the process.

The acceptable output range for a QA at each step and FMEA tables are updated with the current process corresponds to the range that has been proven to work and product understanding. During this risk assessment, based on DOE studies and prior knowledge. Where link- the final commercial control strategy is holistically pieced ages between steps exist, the acceptable output of a step is together using the outputs of the iterative risk assess- based on what the downstream process steps can handle. ments and the 8 Element Wheel shown in Figure 1. During Risk Assessment 2, the relationships between process parameters and CQAs are better defined, which Clinical Supply Chain Strategies enables the categorization of process parameters. This categorization is based on the potential impact on CQAs Control strategies for the entire clinical supply chain (all and is reevaluated throughout the development lifecycle manufacturing, packaging, labeling, and distribution via subsequent risk assessments that build on cumulative steps) should be the sponsor’s responsibility up to and process and product understanding. The final determi- including the delivery of the investigational product nation of criticality for process parameters is based on to the clinical site where patient dosing will occur. At the ICH Q8 definition as a process parameter whose the point where it is delivered to the site, accountability variability has an impact on a CQA and, therefore, should should shift to the responsibility of the investigator and be monitored or controlled to ensure that the process follow current GCPs. produces the desired quality. Some process parameters that do not significantly impact product quality but are SHIPPING AND DISTRIBUTION important to ensure consistent process performance In addition to DP stability and storage under cGMP, can be identified at this step and considered for further DP shipping and handling studies are typically per- monitoring or control during manufacture. formed using appropriate temperature ranges to ensure Following PPQ and prior to regulatory submission, product quality is maintained. Shipping studies may Risk Assessment 3 is performed, in which both the C&E involve shipping representative cryopreserved buffers

CHAPTER 7: Process Control Strategy 188 or product in shipping container(s) and packaging. For due to limitations that may exist based on labeling of the example, a shipping study may involve shipment of the primary container at the time of DP manufacture. representative product to and from a potential clinical site for confirmation of shipment temperatures and also DOSE PREPARATION AND ADMINISTRATION for testing and analysis, if necessary. A shipping study Dose administration of a gene therapy product involves may also be conducted to ensure distribution plans for loading the prepared dose into a delivery system such as movement of the bulk DP from the site of manufacture a syringe or bag/infusion set and injecting or infusing to the distribution vendor are acceptable. the prepared dose into a subject. Specific development Shipping studies should ensure that all processes studies should be done on representative DP material have been tested in advance of an actual vendor transfer (example: low-strength samples in the exact clinical for- and clinical shipment to ensure the temperature can be mulation) to support the various steps in the dose prepa- properly maintained and monitored allowing for docu- ration and administration process and to demonstrate mentation of shipment data. that these steps are sufficiently robust and consistent so that the product can be administered without negative PACKAGING AND LABELING impact on the quality of the product. These studies might The term labeling designates all labels and other written, include activities such as: printed, or graphic matter on an article’s immediate con- tainer, or on or in, any package or wrapper in which it is • DP storage and handling conditions enclosed, except any outer shipping container. The term • DP vial thaw, in-use shelf life, and storage conditions label designates that part of the labeling on the immedi- • Dose preparation procedure, component compatibil- ate container. In this section, the primary label refers to a ity, and in-use stability of dosing solutions physical label that is affixed to the primary DP container. Secondary label (also referred to broadly as labeling) Recommended Readings refers to the label affixed to the secondary container in which the primary labeled container is placed. US Department of Health and Human Services, Food and Packaging and labeling of gene therapy will require a Drug Administration, Center for Drug Evaluation and different approach from typical product labeling as the Research, Center for Biologics Evaluation and Research. Draft product must be placed in required cryogenic storage soon Guidance — established conditions: reportable CMC changes after manufacture and cannot be thawed to be labeled at for approved drug and biologic products. Food and Drug a later date due to limitations on freeze-thaw cycling of Administration website. https://www.fda.gov/media/92242/ these products. For this reason, in most cases, a minimal download. Published May 2015. Accessed January 26, 2021. text primary label will be affixed to the product at the time ICH. Pharmaceutical development Q8(R2). ICH website. of DP manufacture, after visual inspection but prior to https://database.ich.org/sites/default/files/Q8_R2_Guideline. placement into frozen storage. The primary label text strat- pdf. Published August 2009. Accessed January 26, 2021. egy will need to define label content based on regulatory requirements and also based on the strategy to label the ICH. Pharmaceutical quality system Q10. ICH website. primary container at the time of DP manufacture. https://database.ich.org/sites/default/files/Q10%20Guideline. Due to limited supply for gene therapy, an on-demand pdf. Published June 4, 2008. Accessed January 26, 2021. clinical supply model strategy should be considered. This ICH. Development and manufacture of drug substances model should include plans for labeling of the secondary (chemical entities and biotechnological/biologic entities) container. A trigger for shipment of supply should be Q11. ICH website. https://database.ich.org/sites/default/ determined to ensure distribution controls for supply files/Q11%20Guideline.pdf. Published May 1, 2012. Accessed management. January 26, 2021. This overall label strategy may require regulatory approval as part of the regulatory submission, especially

CHAPTER 7: Process Control Strategy 189 Chapter 8 Comparability Chapter 8 | Contents

Introduction...... 193 Regulation...... 194 Reporting Requirements for Post-Approval Changes...... 195 Reporting Requirements of Manufacturing Changes During the IND Phase...... 195 Understanding Critical Quality Attributes of Gene Therapy Products During IND Phase...... 196 Importance of Establishing Comparability...... 196 Risk Factors That Affect Product Comparability...... 197 Challenges Associated with Comparability Studies in AAV Product Manufacturing...... 197 Product Complexity...... 197 Incomplete Product Knowledge...... 197 Inadequate CQA Assays...... 198 Small Batch Size...... 198 Side-by-Side Comparisons of Product Quality...... 198 Establishing Statistical Significance...... 198 Compressed Timeline...... 198 Lack of Reference Material and Standards...... 199 Tools to Establish Product Comparability...... 199 Comparable Products...... 199 Requirements for Nonclinical and/or Clinical studies...... 200 Essential Elements of a Comparability Study...... 201 Introduction and Background...... 201 Description of Change and Rationale for Introducing Change...... 201 Categorization of Changes...... 201 Comparison of CQAs...... 202 Potency...... 202 Purity...... 202 Strength...... 202 Identity...... 203 Safety...... 203 Predefined Approach to Establishing Product Comparability...... 203 Well-Defined Acceptance Criteria to Establish Analytical Comparability...... 203

CHAPTER 8 Comparability 191 Contents, continued

Detailed Analytical Procedure...... 203 Sampling Plan and Statistical Analysis...... 204 Statistical Strategy for Comparability Assessment...... 204 Overview of Commonly Used Statistical Approaches and Applicability to Process Comparability Assessment...... 205 Applying SPI as the Predetermined Acceptance Criteria...... 209 Summary of Statistical Test Approaches...... 210 Process Validation for New Processes...... 211 Stability of Products Manufactured Using a New Process...... 212 Stability Data Requirements Associated with Process Changes...... 212 Comparability Study Conclusion...... 213 Case Study...... 213 Possible Scenarios...... 214 Conclusion...... 215 Appendix...... 216 Illustration of the Different Statistical Approaches for Comparability...... 216 Visual Assessment...... 216 Statistical Assessments...... 216 Abbreviations...... 218 Endnotes...... 219

CHAPTER 8 Comparability 192 Introduction understand the most relevant critical quality attri- butes (CQA) for AAV products, which may include This chapter provides guidance to sponsors on the best empty-full ratios, capsid post-translational modifica- practices and recommendations in managing manufac- tions, multiplicity of infection (MOI), infectivity, residual turing changes during early development by establishing impurities such as protein, and potency assays. The most basic concepts in comparability appropriate for gene challenging aspect of establishing comparability for AAV therapy products, with a focus on adeno-associated products is the lack of suitable potency assays, which are virus (AAV)–based products. Making changes to the useful in measuring the biological activity of the product, manufacturing process/product is an inevitable part of and measurement tools that are used to quantify the process development with the end goal of improving the virus dose and strength with sensitivity and accuracy. product, and therapeutic developers should be prepared At first glance, a typical approach for establishing to institute robust comparability plans to minimize de- comparability of AAV products includes measurement of lays in commercialization. identity, purity based on residual DNA and protein, and Sponsors and developers are encouraged to begin the potency. Although a straightforward exercise on the sur- process of constructing a comparability plan as early as face, it is extremely challenging to demonstrate analytical possible in product development. Taking into consideration comparability based on existing quality attributes that have both existing guidance documents on well-characterized questionable correlation with in vivo safety and efficacy. In biologics and other regulatory and guidance documents addition, it may be challenging to establish comparability that are informative for implementing post-approval based on analytical methods that exhibit a high degree changes, sponsors should put an early and specific focus of variability and are neither qualified nor validated. on gathering development and clinical data to support the For example, when measuring total residual DNA, PCR rationale and support comparability testing associated may provide different results depending on which target with any manufacturing change. This chapter explores sequence is selected or the type of assay used (e.g., PCR vs comparability as an end-to-end concept, provides best ddPCR). Therefore, it is important that analytical compa- practice recommendations on managing minor and major rability studies are conducted side-by-side with a set of the manufacturing changes, and delineates circumstances that following: 1) well-defined quality attributes that are correl- may require a formal comparability study. ative to quality and efficacy, and 2) measurement tools that Gene therapy products, including AAV products are quantitative and suitably sensitive. Additionally, robust manufactured using manufacturing platforms, have seen method-bridging studies for significant method changes a tremendous growth in recent years. However, the most and enhancements are useful to support development, as promising products are often generated in small scale well as future comparability exercises. or a scale not suitable for commercial manufacturing. Many gene therapy products are discovered and de- As a result, some manufacturers may be challenged veloped in academic laboratories or small biotechnology to identify suitable commercial-scale manufacturing companies with expertise in science and innovation but locations and processes prior to initiating their pivotal limited experience in bringing these products to market. study. This sometimes involves technology transfer from Indeed, scale-up to commercial size and quality control is an academic environment to contract manufacturing or- a laborious and complex process. As a result, the process ganizations (CMOs) and/or the introduction of multiple may include avoidable risks that lead to unnecessary costs process changes that require comparability studies. For and delays later in development. Like most biotherapeutics, example, a major manufacturing change might include gene therapies need to be produced in a living system. The introduction of a new manufacturing platform using parallels with recombinant antibody production during the suspension cells rather than adherent cells, the use of 1990s and 2000s, with regard to the upstream challenges of new cell lines, or a manufacturing site transfer. robust production levels, are important to understand where A key element of establishing comparability is to the industry currently is, and where it needs to strive to be.

CHAPTER 8 Comparability 193 Scaling up introduces challenges into the gene ther- Regulation apy development process. For example, while early on the highest titers were achieved with adherent cells in Developers and sponsors are encouraged to consult ex- either roller bottles or cell stacks, similar results are isting regulations and guidance documents and engage now achievable in suspension adapted HEK293 cells. regulatory authorities early and often as they lay out While this was sufficient to support early clinical trials long-term plans for regulatory filing and clinical eval- and could supply market production for small patient uation. In the United States, the applicable regulations population indications, the deficiencies in scalability are more explicit and defined for post-approval changes. with this platform are a significant limitation. The These applicable regulations and guidance documents delivery of three plasmids to one cell is a relatively in- are not necessarily applicable to how manufacturers are efficient process. For larger-scale manufacturing efforts, expected to manage manufacturing changes during IND transient delivery of plasmid requires excess quantities phases, but are considered to be good practice. of DNA, adding to the overall cost of production and purification. Moreover, transient delivery of rep/cap REPORTING REQUIREMENTS FOR genes in the presence of helper genes can also contrib- POST-APPROVAL CHANGES ute to product heterogeneity, including vector capsids Regulatory requirements for managing manufacturing lacking a transgene. These empty capsids represent a changes are described explicitly in 21 CFR 601.12, includ- significant proportion of virus produced in transient ing the implementation of minor, moderate, and major transfection assays or other manufacturing platforms changes and reporting requirements for these changes to used for AAV production. licensed products. Availability of current Good Manufacturing Practice All post-approval process changes should be monitored (cGMP)-compliant manufacturing facilities, which are and tracked by the manufacturer through a quality man- characterized as multi product facilities, also influence agement system. Gaps in reporting may occur when the the process of scale-up manufacturing and tech transfer sponsor’s quality system does not trigger a regulatory filing from academia to CMO. Currently, CMOs have limited or identify the change; in some cases, the manufacturer capacity for manufacturing and typically serve several may not be aware that the change occurred. Definitions competing clients simultaneously, which creates logis- of major, moderate, and minor changes according to the tical complexity in conducting comparability studies. ANDA Submissions – Prior Approval Supplements Under This chapter provides researchers and early devel- GDUFA Guidance for Industry are shown here.1 opers information on potential risks and insights into how to minimize these risks, and outlines a pathway for Major change: a change that has a substantial easier translation of research into later-stage product potential to have a major effect on the identity, development and commercialization. Common but strength, quality, purity, or potency of a drug avoidable problems related to manufacturing control product as these factors may relate to the safety and comparability of pre-post changes in products will or efficacy of the drug product. A major change be addressed. To this end, the workshop “Comparability requires the submission of a Prior Approval in Cell & Gene Therapies,” organized by ARM and Supplement (PAS) and approval by the FDA before USP in 2019, gathered more than 120 experts actively distribution of the drug product made. engaging in debates relating to different aspects of comparability for cell and gene therapeutic products Moderate change: a change that has a moderate to highlight significant challenges, identify different potential to have an adverse effect on the identity, CQAs, and discuss processes to evaluate CQAs. strength, quality, purity, or potency of a drug product as these factors may relate to the safety or effectiveness of the drug product. Depending on the nature of the change, either a Changes Being

CHAPTER 8 Comparability 194 Table 8-1. Examples of Changes and Associated Risk Categories

Example Risk Category Changes to tubing, bags, or plastic culture dishes Low Changes in critical raw materials, reagents, and ancillary materials Moderate to high Changes to production cell substrate (in vivo gene therapy) Moderate to high Changes to cell differentiation, selection, transfection/transduction steps, or allogeneic bank qualification Moderate to high Overall manufacturing change (e.g., vector sequence change including Moderate to high in the gene of interest or regulatory sequences)

Effected in 30 Days (CBE-30) or Changes Being manufacturing, and controls (CMC) information to an Effected (CBE-0) supplement must be submitted approved biologics license application (BLA) as specified to the FDA for a moderate change. in 21 CFR 601.12 (i.e., post-approval changes). Examples of post-approval manufacturing changes Minor change: a change that has minimal and recommended reporting categories are described potential to have an adverse effect on the identity, in the Appendix of this guidance document, including strength, quality, purity, or potency of a drug a table of frequent manufacturing changes and recom- product as these factors may relate to the safety or mended reporting categories. It is meant to serve as a effectiveness of the drug product. The applicant guide to assist applicants and the FDA to identify report- must describe minor changes in its next annual able post-approval changes and determine appropriate report. reporting categories.3 Categorization of minor, moderate, and major chang- While minor changes can be reported in annual es depends on many factors, but must be determined reports, the manufacturers cannot implement major based on the available product knowledge and potential changes requiring PAS until the change is reviewed and risk to product quality. Changes that have very low risk approved by the agency. The current timeline for review of impacting product quality are considered minor, while of PAS involving manufacturing changes is 4 months. changes that could potentially impact product quality Moderate changes may be implemented under a CBE- are categorized as moderate or high risk. In practice, it 30 supplement or a CBE-0 supplement. The timeline for is extremely challenging to define appropriate reporting CBE-30 and CBE-0 review is currently 6 months. categories for major and moderate changes in cell and The categorization of major, moderate, and minor gene therapy product manufacturing due to the diffi- changes and requirements for comparability assessment culty in assessing the potential impact of these changes for gene therapy products is the major topic of discussion on product quality. For this reason, manufacturers are in several guidance documents, summarized below. encouraged to consult appropriate offices before imple- Recently published guidance entitled “Chemistry, menting moderate or major manufacturing changes. Manufacturing, and Controls Changes to an Approved Table 1 provides examples of changes and potential risk Application: Certain Biological Products,” which is cur- categories. rently in draft form, is arguably the most informative resource about what constitutes major, moderate, and REPORTING REQUIREMENTS OF MANUFACTURING minor changes of biological products, including cell CHANGES DURING THE IND PHASE and gene therapy products.2 This guidance is intended Although the reporting requirement after post-ap- to assist applicants and manufacturers of certain licensed proval changes is well defined by regulatory agencies, biological products in determining which reporting the regulations for reporting changes during the IND category is appropriate for a change in chemistry, phase are less well defined. Generally, manufacturers

CHAPTER 8 Comparability 195 should report major changes in amendments and minor knowledge increases. Defining product characteristics changes in annual reports based on current FDA policies. that are relevant to the clinical performance of the gene This recommendation is summarized in a recent draft therapy may be challenging during early stages of prod- guidance covering the gene therapy products. “The CMC uct development when product safety and quality are information submitted in an IND is a commitment to not sufficiently understood. Therefore, manufacturers perform manufacturing and testing of the investigational should evaluate many product characteristics during product, as stated. We acknowledge that manufacturing early clinical development to aid in the identification changes may be necessary as product development and understanding of CQAs and ensure the ability to proceeds, and you should submit information amend- assess manufacturing process controls, consistency, and ments to supplement the initial information submitted stability as development advances. This is especially for the CMC processes (21 CFR 312.23(a)(7)(iii)). The important for sponsors of gene therapy products who CMC information submitted in the original IND for a are pursuing expedited development programs. CQAs phase 1 study may be limited, and therefore, the effect of may be used to specify key characteristics of the drug manufacturing changes, even minor changes, on product substance (DS) and drug product (DP) including, but safety and quality may not be known. Thus, if a man- not limited to, specifications for a later-phase clinical ufacturing change could affect product safety, identity, study or BLA, and are required to demonstrate product quality, purity, potency, or stability, you should submit comparability by analytical methods. the manufacturing change prior to implementation (21 CFR 312.23(a)(7)(iii)).”4 IMPORTANCE OF ESTABLISHING COMPARABILITY In some cases, FDA reviewers may require additional In all life science industries, the initial product envi- information in support of minor changes reported in the sioned by the inventor undergoes substantial revision annual report if it is deemed to be major and could po- and evolution as it translates from the scientific bench to tentially impact the product quality. For complex changes, the patient. Most manufacturers of cell and gene therapy IND holders are encouraged to have early consultation products make changes at some point during develop- with the agency to determine the category of change. If ment through the post-approval phase. Changes made it is categorized as major based on manufacturer assess- to the manufacturing process may potentially impact the ment and consultation with the FDA, then a compara- product’s critical characteristics and therefore its clinical bility study may be required. It is advisable to develop a outcomes. In addition, minor changes in growth condi- comparability study design with the agency prior to the tions of the common producer cell lines for gene therapy implementation of a major manufacturing change. products (e.g., environmental cues such as extracellular matrix components and spatial organization of signaling UNDERSTANDING CRITICAL QUALITY ATTRIBUTES molecules) may have profound impact on the cellular OF GENE THERAPY PRODUCTS DURING IND PHASE machinery that facilitates viral production. Thus, failure Central to establishing product comparability is sponsor to detect the potential impact of these changes during knowledge of product-specific CQAs that are relevant late-phase clinical trials or post-approval could poten- to the safety and biological activity of the product, as tially affect product quality, effectiveness, and ultimately they are understood at the time of submission. CQAs commercial success of the product. form the backbone and primary reference point for a To address these potential pitfalls, manufacturers comparability plan, and the associated metrics will are strongly encouraged by the FDA and other health evolve over the course of development. For example, authorities to define and implement a plan of action tolerance limits of a product CQA may be broad during to understand the CQAs that could potentially affect early development when manufacturers are still gaining product quality and the clinical outcomes early during information about their product, and will narrow as the product development cycle. However, for a variety information increases and reproducibility improves. of business and logistical reasons, product developers In addition, the list of CQAs may be revised as product often introduce major manufacturing changes late in the

CHAPTER 8 Comparability 196 product development life cycle. One example is when the Challenges Associated with productivity of the initial manufacturing process (func- tion of titer and downstream yield) is sufficient for early Comparability Studies in phase trials, but not for later phase trials and/or the com- AAV Product Manufacturing5 mercial phase (cost of goods). As such, manufacturers are encouraged to introduce major manufacturing changes PRODUCT COMPLEXITY early during the product development life cycle when Gene therapy products represent a novel and complex possible, and to demonstrate that the product is compa- class of biological products and are often heterogeneous rable before and after the implementation of changes. The mixtures. Furthermore, these therapies encompass a wide fundamental reason for this expectation of product com- spectrum of products, each with unique mechanisms of parability comes from the sponsor’s reliance on clinical action, material qualifications, challenges in establish- data generated with the product manufactured prior to ing specifications, manufacturing facilities, product the proposed change to demonstrate product safety and shipping/handling procedures, and storage conditions. effectiveness. Thus, the clinical data may originate from Quality analytics methods and product understanding a product manufactured using a different platform, at enable manufacturing changes with minimal impact on another manufacturing site, or even in another country. product quality, and it is important to note that the min- Although regulatory authorities have established imum level of testing for phase 1 INDs is not sufficient somewhat-defined expectations on how to demonstrate to understand complex biologic products. comparability, the risk is carried by manufacturers, who are responsible for changes made to product man- INCOMPLETE PRODUCT KNOWLEDGE ufacturing processes that adversely impact the clinical Performing comparability in early phases is difficult effectiveness of the product. because establishing comparability requires product knowledge and understanding of CQAs and critical RISK FACTORS THAT AFFECT PRODUCT process parameters (CPPs). In most cases in early de- COMPARABILITY velopment, developers have not established their CPPs Risk factors that affect product comparability are de- because their process has not been characterized yet, pendent on the manufacturing change(s), its impact on and process characterization is not undertaken until product quality attributes, and the timing during the process lock is ready. If CQAs, CPPs, and key process product development life cycle. The complexity of chang- parameters (KPPs) are well known and correlations es introduced during manufacturing also poses a risk. between CQAs and product quality, safety, and efficacy For example, if a developer or manufacturer introduces can be demonstrated, then testing of the product CQAs multiple changes simultaneously, there is an increased pre- and post-change and comparing the results using risk of impact to product quality. It is well understood an acceptable statistical method may be sufficient. The that the relative risk associated with process and product ability to produce a consistent product is dependent on changes is substantially increased during later phases CPP control and CQA monitoring, along with other of clinical trials, as well as when product knowledge is factors that define the overall quality of the product. not comprehensive, particularly in the case of lack of In the context of identifiable CQAs, CPPs and KPPs understanding of how a given CQA relates to product are limited for gene therapy products. In some cases, the safety and efficacy. manufacturing process is optimized based on a limited number of variables using manufacturing scales that are not representative of commercial product manufacturing. Because the establishment of comparability studies relies heavily on the analytical similarity of critical attributes that are informative in assessing product quality, safety, and efficacy, the limited knowledge of CQAs, CPPs, and

CHAPTER 8 Comparability 197 KPP for gene therapy product manufacturing further studies. In general, a side-by-side comparison is valuable complicates the exercise of establishing comparability. to remove inter assay variability and to better home in on the true differences of the product. Methods that involve INADEQUATE CQA ASSAYS separations, CGE, HPLC, and other purity-based meth- In some cases, the analytical methods used to measure ods are best suited for this analysis as well as potency critical attributes that reflect product safety and efficacy type methods. However, a side-by-side comparison for evolve during the course of the development process and gene therapy products may be impractical. As a result, the development of product understanding. Establishing manufacturers may rely upon historical data to establish analytical comparability relies heavily on the availability product comparability. Unfortunately, historical data are of suitable methods that are qualified and/or validated. often not collected by measuring applicable attributes, or Additional requirements for assay qualification and val- using qualified or validated methods. The lack of infor- idation are discussed later in the chapter in the Detailed mation for historical lots thus necessitates a side-by-side Analytical Procedure section. Further, assays used for comparison of attributes for the product manufactured measurement of product quality changes may change before and after changes using materials manufactured during the product development cycle. These changes using the old process via a common analytical method. introduce additional challenges in taking advantage of his- torical data that are often collected using different assays, ESTABLISHING STATISTICAL SIGNIFICANCE and methods should aim to analyze samples side-by-side. Establishing comparability using a small number of batches via an appropriate statistical tool is challenging. SMALL BATCH SIZE Due to a number of factors, such as the timeline for In contrast to biotechnology products for which it is implementation of the proposed changes, cost of man- practical to manufacture a reasonable number of batches, ufacturing, and limited capacity at CMOs, it is a com- at-scale production of AAV products is challenging due to mon practice to establish comparability based on two limited resources and manufacturing capacity at CMOs. to three (or fewer) lots or batches. The limited number The lack of sufficient manufacturing experience and rep- of samples poses a significant challenge to establishing resentative clinical and or commercial products, and the comparability based on well-defined statistical methods small batch size of gene therapy products complicates the and predefined comparability criteria. As a result, the establishment of analytical similarities based on available application of conventional methodology used to estab- statistical tools. Further, there tends to be a major imbal- lish comparability or even similarity for products is not ance between the number of batches manufactured before necessarily applicable to gene therapy products with the and after process changes. In some cases, more data may current state of technology. For additional information be available prior to the change, while for others, more about the appropriate use of statistics to establish com- data may be available post-change. It is important to note parability for gene therapy products, please refer to the that analytical results may not be available for materials Statistical Strategy for Comparability Assessment section for which the initial emphasis was on gaining clinical of this chapter. experience and establishing clinical efficacy, rather than on planning for future analytic comparability. COMPRESSED TIMELINE In gene therapy development, it is common for many SIDE-BY-SIDE COMPARISONS OF PRODUCT QUALITY products to receive expedited program designation (e.g., A well-defined comparability study should rely upon regenerative medicine advanced therapy or breakthrough comparisons of key quality attributes of the product therapy designation). Therefore, manufacturers often before and after major changes have been introduced. have aggressive timelines to implement major manufac- Heightened characterization methods may utilize side- turing changes as part of establishing readiness to initiate by-side comparison, whereas other times release data pivotal or licensing trials. Though aggressive timelines may be used without side-by-side comparisons for other make thoughtful product development changes and

CHAPTER 8 Comparability 198 establishing subsequent comparability challenging, these Tools to Establish Product aggressive timelines do not relax the requirements for comparability for manufacturing changes. Comparability

LACK OF REFERENCE MATERIAL AND STANDARDS COMPARABLE PRODUCTS The lack of commercially available reference material re- Comparability is an essential part of the evolving process mains a challenge in establishing manufacturing compa- to ensure that data gathered is valid through development rability. Although reference material is traditionally used for marketing authorization and beyond. Comparability for assay validation, internal reference materials for gene has become a routine exercise throughout the life cycle therapy products could also be used to establish compa- of biotechnological products. Currently, ICH Q5E is rability for changes introduced to the analytical methods the most comprehensive guidance/guideline document before and after any proposed changes. Manufacturers that is available for establishing product comparability are encouraged to develop internal standards that can between gene therapy products.6 In ICH Q5E, product serve as benchmarks for establishing manufacturing comparability is defined as a conclusion that products control. Standards should be stored under the proper are highly similar before and after manufacturing process conditions to ensure their stability over time. changes with no predicted adverse impact on the quality, The challenges associated with comparability studies safety, or efficacy of the drug product. This conclusion is in AAV product manufacturing include (but are not most often based on an analysis of product quality attri- limited to) product complexity, incomplete product butes. In some cases, in which subtle analytical changes knowledge, inadequate CQA assays, small batch size, are seen, nonclinical or even clinical/immunogenicity establishing statistical significance, compressed timeline, data may be indicated. The demonstration of compa- and lack of reference material and standards. Considering rability does not necessarily stipulate that the quality that manufacturers risk adversely impacting the clinical attributes of the pre-change and post-change product effectiveness, safety, or quality of the product whenever are identical, but that they are highly similar, and that changes are introduced to the manufacturing process, the existing knowledge is sufficiently predictive to ensure they require tools and guidelines for facing challenges that any differences in quality attributes have no adverse related to establishing product comparability. impact upon safety or efficacy of the drug product. In some cases, the regulatory agency may ask the IND sponsor or applicant to submit the comparability study for assessment and review prior to the data collection

Figure 1-1. Schematic of Product Comparison Two processes that differ in the number and type of unit operations can be demonstrated to be comparable

Final Final Product Product Test

Starting materials

CHAPTER 8 Comparability 199 and analysis. In most cases, product comparability of release specifications for the product before and after cannot be established by demonstrating that a product a change may not be sufficient due to lack of CQA/CPP manufactured by a new process meets a predetermined knowledge. Matrix-based examples include: release specification. Fortunately, the minimal elements of a good comparability study for this emerging product • Analytical testing of product attributes, including class have been defined in several public presentations release tests for the impacted product or product by FDA’s CBER.7 Importantly, meeting predetermined intermediate release specification is not sufficient to establish product • Additional in-process testing for the impacted comparability, and determinations of product compa- product or product intermediates rability can be based solely on quality considerations if • Side-by-side heightened characterization of the DS/ the manufacturer can provide assurance of comparability DP/drug substance intermediate (DSI) through robust analytical studies. • Modulation of KPPs to ensure product manufactur- To perform comparability studies, statistical analysis ing control and the generation of sufficient and robust data during • Application of QbD if possible both preclinical development and clinical trials play • Product yield measurement at different stages of critical roles to demonstrate equivalence or superiority in manufacturing a post-change product. However, where the relationship between specific quality attributes and safety and effi- REQUIREMENTS FOR NONCLINICAL AND/OR cacy has not been established, and differences between CLINICAL STUDIES quality attributes of the pre- and post-change product are Comparability studies based solely on in vitro studies observed, it might be appropriate to include a combina- is possible based on the strength of the data, extensive tion of quality, nonclinical, and/or clinical studies in the product knowledge, and measurement of attributes comparability exercise. The comparability study depends that are informative to assess both product quality and on the extent of the change and the stage in the product’s efficacy, and clinical outcome. This scenario is mostly development when the change takes place (for example applicable to changes introduced early during clinical pre- vs post-pivotal clinical trials). It is important to de- studies, which permits manufacturers to collect addi- velop a comparability plan as early as possible in product tional clinical data using the new process. development, preferably before a phase 1 trial. In some scenarios, establishing analytical compa- If the knowledge of CQAs, CPPs, and KPPs is rability may not be possible based on physiochemical complete, the exercise of establishing comparability is and biological assays alone. In those cases, additional straightforward, provided that there is substantial evi- preclinical and/or clinical studies may be needed prior dence that certain CQAs are linked to product efficacy to the licensure of the product in order to be approved at- and clinical outcome. In this case, it may be possible scale and using a process different from what was previ- to establish product comparability based on a limited ously used to generate clinical results. Accordingly, IND set of highly relevant attributes by comparing pre- and holders may be required to conduct additional preclinical post-attributes using common analytical techniques with animal studies and/or bridging clinical studies prior to or predefined comparability criteria for product compara- after licensure. Reliance on in vitro analytical studies may bility and a well-defined statistical method. In particular, be possible for AAV products, particularly if significant predefined comparability criteria may be used for late- clinical data will be generated using the approved new stage/high-risk programs. process. However, when major changes occur very late in If knowledge of CQAs, CPPs, and KPPs is incomplete, the product development cycle, establishment of product then a matrix-based approach is recommended. For comparability based on preclinical animal models or a products at the in-process and final release stages, all rel- bridging clinical study may be unavoidable. evant attributes before and after the change (full/extend- ed characterization) should be compared. Comparison

CHAPTER 8 Comparability 200 Table 8-2. Examples of Manufacturing Changes

Manufacturing Step Process A Process B Rationale

Upstream Adherent cell line Suspension Improvement in yield

Downstream Purification Chromatography Chromatography Improvement in purity Method 1 Method 2 and yield

Essential Elements of DESCRIPTION OF CHANGE AND RATIONALE FOR a Comparability Study INTRODUCING CHANGE This section of the report should include a detailed Based on available information, certain components description of changes reported in both text and tab- are considered by the FDA to be essential elements of ular format. The changes should be reported in the a good, prospective comparability study for gene ther- context of major manufacturing steps (e.g., upstream apy products. In the United States, the comparability or downstream purification steps). Examples of poten- protocol prospectively describes the planned change to tial changes, along with the rationale for a change, are the manufacturing process in the form of a PAS, which described in Table 2. It is critical to include all changes when reviewed by the FDA will determine whether in the process and provide sufficient rationale for the the planned change can be reported in a category that introduced changes. does not require a full comparability protocol. Typically, In some cases, manufacturers also report changes to comparability protocols are submitted to the agency as analytical procedures or changes in the manufacturing PAS in support of lowering the reporting category from facility. Each change should be described in detail under PAS to CBE-30 or CBE-0. The comparability study used a separate heading. to support manufacturing changes during the IND phase is not necessarily identical to comparability protocol that CATEGORIZATION OF CHANGES can be used to lower the required reporting category, but Manufacturing changes for gene therapy products can in principle it contains the essential elements of a good be categorized as minor, moderate, or major changes. comparability study protocol. Minor changes are defined as changes that do not have A comparability study is defined as a prospective potential impact on product quality, while moderate and document that is submitted to the agency in support of major changes could potentially have adverse impact on the proposed manufacturing changes. The comparability product quality and may require submission of a new study should include discrete essential elements as iden- IND or IND amendment (Table 3). The determination tified in detail below. of minor vs major requires not only product knowledge but also an understanding of the relationship between INTRODUCTION AND BACKGROUND CQAs/CPPs with product safety and efficacy. In the The introductory section provides an overview of the situation when the relationship between CQAs/CPPs product, current regulatory status, and manufacturing and product quality is not understood fully, the FDA steps. The background section provides information on encourages the application of risk assessment principles. why a comparability study is being submitted and details It is very important that the categorization of changes is previous regulatory submissions related to the manufac- conducted in consultation with regulatory authorities. turing changes. This section should contain a summary Manufacturers may use risk assessment approaches, of the overall approach used to establish product compa- such as those described in Chapter 4, to prioritize CQAs rability and what is reported in the comparability study. for comparability studies.

CHAPTER 8 Comparability 201 Table 8-3. Examples of Common Changes and Associated Risk Categories

Changes Description Category Notes

Buffer (like for like) Supplier change Minor Manufacturing platform Completely different Major Potentially requires a new IND; consult platform (e.g., change in agency cell line) Formulation Final titer, buffers, Major Affects dose excipients Manufacturing scale Same process, scaling up Moderate or out Manufacturing site Site change for drug Moderate substance (same process) Manufacturing site Site change for drug Minor product (no changes to DS or process) Change of purification Change from centrifugation Moderate process to tangential flow filtration Change of vector Completely different Major Potentially requires a new IND; consult backbone vectors and promoters agency Change of transgene For example, different Major New product requires new IND; consult portion of same gene is agency used as transgene

COMPARISON OF CQAS product, which in AAV products may include residual host cell DNA and proteins, empty AAV capsids, or AAV Potency capsids containing helper virus DNA. Additionally, there Potency, defined in 21 CFR 600.3(s), is interpreted to mean could be present AAV particles containing the targeted the specific ability or capacity of the product, as indicated genetic material that has undergone post-translational by appropriate laboratory tests or by adequately controlled modifications such as (but not limited to) deamidation, clinical data obtained through the administration of the phosphorylation, oxidation, all of which could have an product in the manner intended, to effect a given result. effect on the transduction pathway. A commercial-scale Early development of multiple assays of potentially rel- product with an impurity profile that differs from those evant product activities will not only facilitate develop- of previous noncommercial pilot batches could result in ment of a potency assay, but will also help ensure that the significant challenges in the establishment of product product is consistent and that early results are relevant for comparability and may require further process improve- designing later studies and for licensure. Further, a suitable ment to reach the predefined standards. potency assay provides valuable information concerning overall product stability and is useful for establishing Strength comparability of post-manufacturing process changes. Strength generally refers to the number of AAV particles Countries and/or regions may have different requirements administered to the patient and the potency of the vector to establish the potency testing strategy. product. For products to be considered comparable, the overall strength of the product must be measured. Assays Purity used to establish comparability in strength may measure, Purity, as defined in 21 CFR 600.3 and 21 CFR 610.13, for example, vector infectivity, in vitro RNA and protein refers to relative freedom from extraneous matter in the expression, and in vivo bioactivity.

CHAPTER 8 Comparability 202 Identity WELL-DEFINED ACCEPTANCE CRITERIA TO For gene therapy, identity is defined in 21 CFR 610.14 as ESTABLISH ANALYTICAL COMPARABILITY a test that distinguishes one product from other products A risk-based approach can be used to determine com- manufactured in the same facility, relies heavily on the parability criteria for analytical comparability during sequence information derived from the transgene and process changes. However, selection of analytical meth- its fidelity using different sequencing platforms. Identity ods and acceptance criteria may be the most challenging is not generally considered to be critical aspect of estab- step in a comparability study. Predetermined acceptance lishing product comparability but is important to verify criteria to establish comparability should rely heavily on product quality prior to release/distribution. historical product knowledge, manufacturing capacity, and robustness of the selected analytical methods. For Safety example, analytical methods with a high degree of vari- Safety is defined in 21 CFR 600.3 as the relative freedom ability when measuring identical samples at different from harmful effect to persons affected directly or indi- times, by different operators, or with a different facility rectly by a product. For gene therapy products, safety is or equipment, should be avoided. determined by testing the product for sterility, endotoxin, In the comparability study submitted to regulatory mycoplasma, and the presence of any adventitious agents authorities, the manufacturers propose predetermined that could be derived from the biological material/cell acceptance criteria that is justified based on historical lines used. In addition, the absence of replication-com- data, manufacturing capability, and assay variability. petent virus must be established. Although some of these Historical data may be limited at early stages of de- parameters may be included in comparability studies, velopment, for example, and deriving predetermined measurements of safety may not be critical to establish acceptance criteria from a limited number of lots may product comparability but rather are used in establishing be acceptable. However, acceptance criteria during phase manufacturing control. 3 studies may require additional justification based on a larger body of information collected during the product PREDEFINED APPROACH TO ESTABLISHING development life cycle. PRODUCT COMPARABILITY The approach used to establish product comparability DETAILED ANALYTICAL PROCEDURE for genetically modified cells may involve side-by-side Analytical comparability for gene therapy products is comparison of analytical data obtained from different solely dependent on the robustness of the analytical manufacturing processes using the same starting ma- methods. Depending on the stage at which analytical terials. However, side-by-side comparisons may not comparability is performed, the analytical method must be feasible in AAV-based gene therapy manufacturing. be in a state of control commensurate to the phase of The common approach to establishing comparability the study. Assay qualification may be sufficient for -ear after a major process change relies on comparison of ly-phase analytics, but assay validation is recommended critical attributes, with the highest relevance to product for changes in later phases. However, as a scientific matter bioactivity, with a previously optimized process. This and as good laboratory and manufacturing practice, it is approach requires and mandates the use of analytical highly recommended that manufacturers use validated methods that are deemed comparable and appropriately assays whenever possible to measure CQAs, especially qualified. Importantly, samples collected from reference if they are used to collect information during clinical lots (pilot or non-GMP lots) should be tested using the studies that determine product efficacy (e.g., a pivotal or same assay under identical or similar conditions. licensing trial). The reason for this is simple: validation involves providing assurance that a given process or test can be performed reproducibly and accurately with a high degree of sensitivity, precision, and linearity, even in a worst-case scenario. Thus, the assay yields equivalent

CHAPTER 8 Comparability 203 results using the same sample when it is used by dif- multiple lots, trend analysis should be applied, including ferent operators in different lab environments, or using comparisons to historical lot releases. It is important to different instruments, so long as the assay parameters are provide justification for how lots were chosen for the controlled as specified in a protocol. Qualification sets a comparability exercise and to avoid “cherry picking” (i.e., lower bar and requires that the assay can be performed using certain pre-change lots that are more comparable with some reasonable degree of reproducibility by the to your post-change lots). manufacturers, under very controlled conditions, such as it being performed by a designated operator and using SAMPLING PLAN AND STATISTICAL ANALYSIS a specific instrument or a reagent lot. Manufacturers should provide sufficient justification for As a result, manufacturers who choose to use a qual- the proposed sampling plan, including the number of ified (not validated) assay to collect critical information batches tested, types of batches used for manufacturing during phase 3 or pivotal studies could potentially collect comparability, and sample collection methods for the data sets that are not fully representative of their product comparability runs. The type of batches used to estab- quality. This could potentially impact the usefulness of lish comparability should be a major point of discussion assay results that are relied upon to define meaningful in the comparability study design. Due to a number of specification/acceptance criteria to assess and verify practical considerations, manufacturers should plan to product quality. establish comparability using a limited number of runs A typical approach to qualify and validate critical as- and batches that are manufactured using a process rep- says is as follows. For qualification, regulatory authorities resentative of the at-scale manufacturing (which may be generally expect demonstration of a reasonable degree performed under non-GMP conditions). This approach of assay sensitivity, linearity, precision, and accuracy. is not necessarily encouraged, but it could be potentially For validation, these parameters must be complemented acceptable provided that appropriate scientific justifica- by assay ruggedness and robustness. For the purpose of tion is provided. this discussion, assay ruggedness is the reproducibility of Selection of a statistical method to establish compa- the assay under a variety of variable test conditions that rability is an important consideration when defining the include different instruments, operators, and reagent lots. required number of samples. It is extremely challenging Robustness provides an indication of the assay’s ability to to test a sufficient number of batches for cell and gene perform under normal usage and without being impacted therapy products (e.g., due to lack of sufficient raw ma- by changes in various factors (e.g., incubation time, tem- terials and starting materials, compressed timelines, and perature, sample preparation, buffer, or pH) or parameters limited production capacity). Accordingly, the agency that can be controlled and specified in the assay protocol. has shown a great deal of flexibility in accepting compa- Thus, timing is critical when determining whether rability based on a limited number of runs. to use a qualified or validated assay in the development of gene therapies, and sponsors must make thoughtful STATISTICAL STRATEGY FOR COMPARABILITY decisions so that data used to support development ASSESSMENT and subsequent BLA submission are reliable and utilize Process comparability exercises for gene therapy pose the appropriate assays during a given stage of development. following challenges for the application of proper statis- If new analytical methods are implemented during tical tools, which are discussed in greater detail below. development, older methods may be included in the comparability exercise. It is important to retain samples • Limited sample size: Frequent process changes that can be retested when new and improved methods are during early-stage development often results in implemented. Such samples will identify whether a newly only 1 or 2 batches per process. Also, depending identified item is actually new or whether it was present on the indication and dose, few batches may be in the clinical material but not detectable by the old needed clinically, limiting the batches available for method. When the manufacturing experience contains comparability.

CHAPTER 8 Comparability 204 Figure 8-2. The side-by-side comparison of the pre-change and post-change lots against the mean+/-3SD (74.3, 123.1) and the SPI with 99% confidence level (64.9, 132.4)

140 Pre-change Post-change 132.4

1 2 3 4 5 6 7 8 9 10 11 12 13 Lot number

• Unbalanced sample size: Unbalanced sample intervals, tolerance intervals, and equivalence testing.8 sizes are common in process comparability data. Although each of these statistical methods has an intend- The number of the pre-change batches could be ed use, some may result in similar or identical limits by large, but the number of post-change batches is adjusting confidence and/or coverage levels. The choice usually no more than 4 at the time of comparison. of statistical methods depends on many factors, includ- This unique unbalanced data structure makes it ing 1) knowledge of the product CQA (e.g., is the CQA a challenging to properly apply commonly used numerical variable and amenable to statistical analysis), statistical methods. 2) relevance of the CQA to product efficacy and safety (e.g., CQA range considered to be safe and efficacious for • Lack of knowledge about the clinically meaning- the CQA), 3) associated analytical testing and process ful difference between the pre- and post-change variability, 4) development stage of the product, and 5) processes: inability to assign a numerical value the available number of batches for statistical assessment, to represent the clinically meaningful difference which is generally very limited (e.g., <10). has been a common challenge and poses a direct Not all CQA are amenable to statistics. For instance, challenge to application of an equivalence test. qualitative CQAs and quantitative CQAs with unreason- ably large method variability are generally not suitable Overview of Commonly Used Statistical for statistical analysis. However, the stage of development Approaches and Applicability to Process may make a difference as the available knowledge and Comparability Assessment historical data are typically limited at the early stage but Various statistical approaches have been developed and more plentiful in later stages. It could be misleading to applied in comparability assessments. Commonly used blindly apply statistical analysis methods when data are statistical approaches include visual comparisons, min- limited. imum and maximum, confidence intervals, prediction

CHAPTER 8 Comparability 205 Visual Assessment Figure 1-3. Plot of the deviation of SD estimated Visual assessment is always helpful, regardless of the from the true SD amount of data. Although visual assessment does not Using methods from Burnett (1975). The inset box shows provide a direct answer of “pass/fail,” it provides a basic the sample size from 5 to 30 understanding of the data and is generally recommend- ed as the first step of data analysis. Visual assessment 70 is probably the only assessment that is applicable with 70 only 2 or 3 batches to compare per process. For instance, 50 a side-by-side scatter plot (Figure 2) is a simple way to examine data trends. 50 30 Visual assessment generally does not follow explicit rules. However, when the number of batches is large, 10 some of the trending rules commonly used in the typ- 30

ical statistical process control (SPC) field may apply. (%) Deviation 5 10 15 20 25 30 When data are limited, as they often are, it is difficult to identify trends and SPC trending rules are generally 10 not suitable. Instead, it is tempting to use the data range of the pre-change batches to assess the performance of 0 100 200 300 400 the post-change batches. However, when pre-change data Sample Size are limited, the risk of post-change batches exceeding the pre-change data range is significant, even if the two processes are the same for the attribute of interest. is that the CQA values of all post-change batches fall Therefore, it is not appropriate to simply focus on the within some expected limits instead of, for example, the range of the pre-change and post-change batches. When mean of post-change batches. Therefore, CI is generally data are limited, method variability can be used as an not recommended for process comparability assessment approximate gauge to assess the data spread. In other unless the interest is in the population parameters such words, comparability may be concluded if post-change as mean or standard deviation, which can be reliably batches are within the expected method variability. estimated from a decent number of historical batches.

Minimum/Maximum Mean ± 3 Standard Deviations (SD) Minimum and maximum are two commonly used de- Mean ± 3 SD is one of most well known statistical in- scriptive statistics that represent the range of current data. tervals probably because it is simple and easy to apply. Although simple to use, minimum and maximum have no For an attribute that follows normal distribution, the inference capability in that the range of the current data mean ± 3 SD represents the interval that 99.73% of the does not necessarily represent the range of the future data. population (i.e., all values generated before and in the Furthermore, a minimum/maximum based on a limited future under the same condition) will fall within if the sample size is subject to large uncertainty and is often nar- true mean and the true SD are known. However, it is only rower than the true range of the data.9 The EMA reflection recommended for the two following situations assuming paper on comparative assessment also criticizes it.10 normal distribution: 1) the true mean and the true SD of the pre-change process are known or 2) the pre-change Confidence Interval (CI) process has a long history, thus permitting the true mean CI is a widely used statistical measure defined as the and the true SD to be estimated with high certainty based interval that contains the true value of the population on a large number of pre-change batches (N). The SD parameters, such as the mean or standard deviation. In a requires a larger sample size than the mean to calculate typical process comparability setting, the ideal outcome a reliable estimate. The estimated SD can still differ from

CHAPTER 8 Comparability 206 the true SD by ~10%, even when N=200.11 Figure 3 interval, which represents the desired coverage probabil- shows how much the SD estimate could deviate from ity to include all m post-change batches simultaneously. the true value with increasing sample size. The deviation This approach fits the process comparability data drops significantly when sample size increases from 5 to structure better in that 1) there is no requirement on the 10. The speed of the drop slows after N=10 and the devi- data balance (i.e., N can be much larger than m) and 2) it ation curve begins to flatten out after N=200. Therefore, is the interval for individual future values and therefore mean ± 3 SD is often used for control chart setting for can be used as pre-determined acceptance criteria for situations in which a large sample size is available. It is the assessment of the post-process changes once the generally not suitable for limit or criterion setting, such pre-change data are available and the number of post- as in process comparability, unless a very large sample change batches to produce, m, is known. In most process size (N) is achieved. comparability cases, m>1. Therefore, SPI is more often used than prediction interval. As shown in the equation Prediction Interval above, both the number of pre-change batches (N) and Unlike the confidence interval, which is useful for popula- the number of post-change batches (m) affect the width tion parameters, prediction interval is designed to predict of the interval. Larger N leads to narrower SPI, while the range of future individual values if the future values larger m leads to wider SPI. Having few batches and will Predictioncome from Interval the same population as the historical val- variable methods could result in very wide intervals that ues.12Unlike It is usually the confidence called prediction interval, interval which if is there useful is onlyfor population are not parameters,particularly informativeprediction interval for comparability. is designed one tofuture predict value the to range predict of and future simultaneous individual valuesprediction if the future values will come from the same population intervalas the (SPI) historical if ≥2 future values. values12 It needis usually to be called predicted. prediction intervalTolerance if there Interval is only (TI) one future value to predict In general, to predict m future values based on N his- The tolerance interval (TI) is calculated asx ± k σ for and simultaneous prediction interval (SPI) if ≥2 future values need to be predicted. ∗ torical data values, the interval can be calculated as the normally distributed data and covers at least (100-α)% following:In general, x ± ksto wherepredict x mand future s, respectively, values based represent on N historical of the data measurement values, the populationinterval can withbe calculated (100-γ)% as confi - the samplethe following: mean and sample where standard and deviation , respectively, of the representdence. theHere, sample x and mean σ are andthe samplesample meanstandard and sample N pre-change batches, and k is the multiplier factor. The standard deviation, respectively. The multiplierk directly deviation of the N pre-change batches, and is the multiplier factor. The multiplier is written as multiplier k is written 𝑥𝑥 ± as𝑘𝑘𝑘𝑘 𝑥𝑥 𝑠𝑠 affects the width of the TI and it is determined by the 𝑘𝑘 prespecified confidence level 𝑘𝑘(100-γ)%, the coverage level (100-α)%, and the sample size. α represents the specified ! ! target proportion of the population that is not covered 𝑘𝑘 = 𝑡𝑡!""#,, (&"!) ∗ )1 + & for 2-sided prediction interval and in the interval, and γ represents the specified error rate in the calculated interval.13 TI is generally calculated as the starting point for specification limit development

because it covers almost all future data from the same ! ! 𝑘𝑘 = 𝑡𝑡!"#,, (&"!) ∗ )1 + & population.14 However, the typical TI with high confi- for 1-sided prediction interval, where dence and high coverage level (e.g., >90%) tends to be too wide with a limited sample size.15 Therefore, TI is is the percentile not a recommended approach for biosimilar analytical ! ( similarity assessment.16 It is generally not an appropriate 𝑡𝑡!"#,, (&"!) ,1 − ). 𝑡𝑡ℎ of theof centralthe central student student t distribution t distribution with withN−1 degrees degrees of approachof freedom, for the typical process comparability setting for freedom, and two reasons. First, similar to the situation with biosimi- N − 1 lars, the TI tends to be too wide given that the number of is the percentile ! ( pre-change batches is usually limited (e.g., 1. Therefore, SPI is more often used than prediction interval. As shown in the equation above, both the number of pre-change batches (N) and the number of post-change batches (m) affect the width of the interval. Larger N leads to narrower SPI, while larger m leads to wider SPI. Having few batches and variable methods could result in very wide intervals that are not particularly informative for comparability.

Figure 8-4. Multiplier Comparison Between 95/99 TI and 99% SPI for 3 or 5 Post-Change Batches

7.0 99/99TI 6.5 PI (99%, m=5) 6.0 PI(99%, m=3) 5.5 mean+/-3SD 5.0 4.5 4.0 Multiplier 3.5 Multiplier 3.0 2.5 2.0 8 10 12 14 16 18 20 22 24 NumberNumber of of Prepre-change-change lots Lots multiplier increases with the number of post-change T-test batches. With the same number of pre-change batches, The T-test is a hypothesis test that is commonly used by the 99/99 TI multiplier is larger than 99% SPI for 3 or 5 researchers to establish whether data collected under two post-change batches. Second, the purpose of the compa- different conditions are significantly different. However, rability exercise is to determine whether the post-change the T-test methodology sometimes is mistakenly used batches generated at the time of comparability (usually to establish comparability. In this situation, establishing no more than 4 or 5) are within the expected range based similarity is based on not rejecting the null hypothesis on the pre-change data trend. Therefore, it might be ex- that the two populations are the same. Although it seems cessive to apply TI in the comparability setting with few like a logical approach, it is not. Failure to reject the hy- post-change batches (sometimes only one). pothesis of sameness does not necessarily imply compa- rability. When the sample size (i.e., number of batches) is Quality Range limited, there may be failure to reject the null hypothesis The term “quality range” was first proposed by FDA of sameness even with a relatively large difference. In statisticians to assess the analytical similarity between contrast, the null hypothesis of sameness may be rejected biosimilar and reference products for Tier 2 quality at- with no clinically or practically meaningful difference tributes.17 It was adopted as the recommended statistical if the sample size is large enough or the data variability approach for both Tier 1 and Tier 2 quality attributes in happens to very small. Therefore, the T-test should not be 2019.18 Quality range is defined as the mean ± k∗SD, used to demonstrate comparability of two populations. which is the same form as the other statistical intervals described above. Conceptually speaking, quality range Equivalence Test could be any of the above statistical intervals (CI, pre- In a situation in which additional batches and more diction interval, or TI). There is no detailed guidance on information about the CQAs of interest are available, how to determine the multiplier k in practice, but mean rigorous statistical methods can be applied. One of the ±3 SD is commonly used for biosimilar filings because of most rigorous statistical methods is the equivalence the relatively large sample size of the reference product.19 test. Like the T-test, it is also a hypothesis test. However, However, SPI is more suitable for process comparability. unlike the T-test, the equivalence test is more suitable for comparability because it leads to a direct conclusion of “equivalent” or “not equivalent” based on a pre-set

CHAPTER 8 Comparability 208 acceptance criterion. A proper equivalence test requires Table 8-4. The Multiplier k for Two-Sided and a decent sample size and a quantitative scientific un- One-Sided SPI Based on N Pre-Change Batches derstanding of the clinical relevance. For instance, a for 3 Post-Change Batches minimum of 10 batches per group was suggested when this test was required for biosimilar analytical similarity N One-sided Two-sided in Tier 1 QA assessment.20,21 In the process comparabil- 3 14.07 19.95 ity setting, it can be used for testing several pre-change 4 7.54 9.59 batches while there are few post-change batches available 5 5.66 6.85 at the time of the comparability exercise. In addition, the 6 4.81 5.67 acceptance criterion needs to be set prior to analyzing the data and ideally should be based on 1) the clinical 7 4.34 5.02 relevance or clinically meaningful difference of the CQA 8 4.04 4.62 and 2) the associated analytical and process variability. 9 3.83 4.35 In practice, the first element is generally unknown or 10 3.68 4.15 hard to quantify due to limited data or understanding. 11 3.56 4.00 Without proper quantitative acceptance criterion, the 12 3.47 3.88 equivalence test is practically meaningless, which is 13 3.40 3.79 part of the reason that equivalence tests are no longer 14 3.34 3.71 required for biosimilar similarity assessment by the 15 3.29 3.65 FDA.22 Generally, the equivalence test is not a practical 16 3.24 3.59 approach for process comparability. 17 3.21 3.54 The appendix provides an example that illustrates the approaches described above. 18 3.17 3.50 19 3.14 3.47 Applying SPI as the Predetermined 20 3.12 3.44 Acceptance Criteria 30 2.97 3.25 SPI has been proposed as a predetermined acceptance 50 2.86 3.12 criterion to assess post-change batches. Although it is 100 2.79 3.02 preferred to set (1−α) to a high number such as 99.7% (i.e., the same ideal coverage that mean ± 3 SD would provide with a large sample size), the width of SPI gets pre-change sample size (N) gets larger, the two-sided 99% wider with higher (1−α) values. Due to the limited num- SPI approaches the mean ± 3 SD. The multipliers shown ber of pre-change batches, (1−α)=99% is recommended in Table 4 are for two-sided 99% SPI. In some cases, only for N≥10 (i.e., at least 10 pre-change batches). Figure 2 an upper or lower limit of the SPI is needed for certain provides an SPI with a 99% confidence level (99% SPI) CQAs, such as purity or impurity. In that situation, the compared with mean ± 3 SD based on 10 pre-change user could choose to use one limit (either upper or lower batches and 3 post-change batches. Although SPI can limit) of the two-sided SPI, but the actual coverage of the be calculated based on as few as 2 pre-change batches, limit is increased to 99.5%. Alternatively, the one-sided it is recommended to apply SPI for N≥10. As shown in 99% SPI can be calculated using the one-sided k in Table Figure 3, the SD estimate is subject to a much greater 4. However, doing so could lead to multipliers <3 when uncertainty for N<10. N reaches 30, resulting in a narrower limit than mean As shown in Figure 2, a 99% SPI (orange line) is ± 3 SD. generally wider than mean ± 3 SD (i.e., the blue line) Note that the SPI here is used to represent the his- a realistic sample size (N). Table 4 lists the k values for torical manufacturing range of the pre-change process both one-sided and two-sided limits for m=3. As the given a decent number of pre-change batches. It does

CHAPTER 8 Comparability 209 not necessarily guarantee that SPI is always narrower Figure 8-5. Decision Process Using SPI to Assess than the release specification, which ideally reflects drug the Comparability of Post-Change Batches efficacy and safety. The post-change batches are expected to meet both specification and SPI criteria. Knowledge about the post-change process is often limited by the Are the post-change number of post-change batches, which sometimes can batches within the 99% SPI as be as small as m=1. SPI with 99% confidence simply the predetermined acceptance criteria? means that there is a 99% chance that the m post-change batches will fall within the SPI if there is no change between pre-change and post-change processes for the CQA of interest. However, the m post-change batches all falling within the SPI does not necessarily guarantee that there is absolutely no change between the pre-change and post-change processes. Rather, it indicates the lack yes no of clear evidence or signal to claim that the post-change process is not comparable to the pre-change process. Further investigation However, if one or more of the m post-change batches fall on the batches that fall outside SPI, it is a signal to raise the alarm and consider outside the SPI required. an investigation. Although statistical assessment is an If no specific reasons Claim process can be identified, more important input to decision making, the comparability comparable decision should not be solely based on statistical analysis batches from the post- change process need results. If one or more post-change batches fall outside to be collected before the SPI, an investigation of the batches outside SPI is drawing conclusions. recommended. If no assignable cause is identified and there are few post-change batches (e.g., 2 or 3), more batches may be required prior to drawing conclusions It is also worth emphasizing that the SPI is proposed (Figure 2). to deal with the imbalanced data structure in the typical Like most statistical methods, the SPI approach process comparability exercise. If there are a decent assumes that the CQA of interest follows a normal dis- number of pre-change batches and the number of future tribution. If it does not (e.g., some impurity measures), post-change batches is known, the SPI can be calculated data may need to be transformed (a log transformation as the pre-determined acceptance criteria. The data can is often used to get data closer to normal distribution) be derived directly from the release test or from side- prior to applying the SPI approach. Statisticians should by-side testing. However, if the number of post-change be consulted in these situations. batches is close to the number of pre-change batches or The 99% SPI proposed here may be unrealistically larger (e.g., N=10, m=10), the 99% SPI could become too wide if the number of pre-change batches is <10. wide due to the increase of the number of post-change However, it is very common to have frequent process batches. In such cases, there are two options to consider: changes and only 2 or 3 batches per process in the gene lowering the confidence level of SPI; or using mean ± 3 therapy field. One way to deal with this situation is to SD if the number of pre-change batches is relatively large. start with visual comparison for early process changes. If the first few processes (i.e., process 1 and process 2) are Summary of Statistical Test Approaches considered comparable, the data from these processes As previously mentioned, not all quality attributes are could be combined and treated together as from one relevant to the comparability assessment. Among rele- historical process to meet the sample size requirement vant quality attributes, some may not be amenable for for the SPI calculation. statistical assessment (e.g., qualitative quality attributes).

CHAPTER 8 Comparability 210 Assuming the quality attributes are relevant and amena- require production of the final drug product from patient ble to statistical analysis, it is good practice to first apply materials. For gene therapy products, one can argue that visual assessment and descriptive statistics (e.g., mean, production of product consistently at full scale post- standard deviation, minimum, maximum, N) to the data. change is also required. This is particularly applicable If ≥10 pre-batches are available, SPI is recommended to if the product manufactured at full scale is sufficient to represent the historical range of the pre-change process. treat a small patient population. In this scenario, it is This approach assumes no mean shift and no variability important for manufacturers to demonstrate not only change between pre-change and post-change processes. comparability at scale that could be considered repre- If there is a justifiable mean shift known in advance, the sentative of full scale but also to show that the product mean shift can be added to the sample average x( ) of the at full scale can be manufactured consistently. The need pre-change batches in the formula. for conducting process validation after a manufacturing Like most comparability exercises, the available num- change should be discussed with the agency prior to ini- ber of batches at the time of assessment remains one of the tiation of comparability studies because it could impact biggest challenges. However, as more information about the design of the comparability study. the product, related process, and CQAs is acquired, the The 2011 Guidance for Industry on Process Validation: knowledge may be summarized quantitatively to form an General Principles and Practices provides the most up- “informative prior” as one of the key inputs for a Bayesian dated guidance on the principles of process validation.24 approach.23 With reliable “informative prior” in place, the Process validation is documented evidence that a process required number of pre-change batches decreases slightly will consistently produce product meeting pre-deter- without losing the reliability of the prediction interval. mined specifications. Validation takes time and can be Although no such application has been observed, better expensive, but over the long-term it is good business and more use of historical data are becoming more feasible because it diminishes the likelihood of batch failures. with the rapid development of digital solution transforma- tions in data collection and accessibility. Process validation = Consistent batches = In general, appropriately chosen statistical tools play Conforming batches a critical role in the process comparability assessment. However, the statistical conclusion can only be as good Process validation studies are generally not required as the data at hand, and it should not serve as the only for early-stage manufacturing, so most original IND input for final regulatory decision-making; knowledge submissions do not include a process performance and information that are not contained in the data also qualification. It is recommended to use early-stage man- need to be considered. ufacturing experience to evaluate the need for process improvements and support future process validation PROCESS VALIDATION FOR NEW PROCESSES studies. In addition, establishing process validation Manufacturers may argue that production of several after making a manufacturing change is recommended. batches at scale for the purpose of comparability should To show similarity between products pre- and post- be sufficient evidence that the product can be manufac- change, three consecutive lots must be produced. All tured with some degree of consistency. This argument is facility equipment qualification support systems, product not necessarily adequate justification for not performing specifications, and the process being validated must pass process validation studies after major manufacturing at all steps. changes, particularly when the breadth of clinical data Gene therapy products may be produced in small would be very limited. quantities. Specific issues of concern in the validation For example, it may be sufficient to perform compara- of gene therapy viral vector manufacturing processes bility studies for genetically modified cells using starting include the quality of raw materials, safety testing of material collected from healthy donors with justification, cell and viral banks, production and purification of but establishment of manufacturing consistency may the vector, in-process and final-product testing, and

CHAPTER 8 Comparability 211 Table 8-5. Stability Study Design

Stability-Indicating Assays Conditions 15 days 30 days 3 months 6 months 1 year 2 year -80ºC -60ºC -20ºC validation of analytical methods. Although risk can be CQAs of DS and DP are influenced under specific con- reduced by proper facility and process design (e.g., use of ditions of temperature, relative humidity, light, storage, single use systems), because most vectors are produced pH, and other factors. These studies are required to be in multi-product facilities, cleaning validation may be a conducted following the guidelines issued by the ICH, major concern due to potential product-to-product cross WHO, and/or other agencies. contamination. Viral clearance also presents a major Stability assessment for comparability has a specific validation challenge due to the nature of the product. purpose, which differs from the typical studies in the As development proceeds, analytical methodology and stability program. The product must be demonstrated specifications should be evaluated and refined based on to be stable for the period of time while stored at the development data, engineering runs, and clinical batch storage site after manufacturing, and at the clinical site. analysis. Additional assays may also be added to monitor For products formulated with carrier or support ma- and control critical attributes of the product. ICH Q2R1 terials, the stability of the complex formed with the DS should be followed for late-phase assay validations. As the should be studied. Where relevant, the in-use stability of clinical program reaches pivotal trials, analytical methods the DP (after reconstitution or after thawing) should be and specifications should be based on a sufficient dataset investigated, including its compatibility with any diluents to tighten the release and stability specifications in support used in reconstitution and if appropriate, devices used for of late-phase comparability and process validation activi- administration. The recommended in-use time period ties. Health authority expectations increase considerably at should be justified. The impact of the transport conditions this stage, and pharmaceutical quality and manufacturing on the stability of DS or DP with a short-term shelf life control becomes a critical focal point of product develop- should be considered. Stability protocols, stability data, ment. This component is analogous with the term CMC justifications for the container-closure system used, and used for small molecules. proposed shelf-lives and storage conditions, should be For many gene therapy products, including AAV presented for the DS, DP, and DPI (i.e., intermediates products, the application of principles for process vali- for which a holding time is scheduled on the production dation outlined in the guidance document25 is possible process scheme). The need, extent, and type of stability but may require innovative approaches such as rolling studies depend on product development stage, product process validation (which is performed in very discrete and process knowledge, extent of change, potential impact steps) or unit operation or concurrent process validation of the change on product CQAs, safety, and efficacy, and (which is performed concurrently with the release of the availability and capability of analytical methods (Table 5). final drug product). STABILITY DATA REQUIREMENTS ASSOCIATED STABILITY OF PRODUCTS MANUFACTURED USING WITH PROCESS CHANGES A NEW PROCESS The requirement for conducting stability studies requires Stability testing is a vital part of product development careful analysis of how changes in manufacturing could and is conducted throughout a product’s life cycle. It is impact short- and long-term stability of the DP. In gen- also a part of a biotherapeutic’s quality target product eral, for a gene therapy product with a long shelf life, profile (QTPP) and increases the understanding of how batches manufactured using the new manufacturing

CHAPTER 8 Comparability 212 Table 8-6. AAV Product Attributes and Predetermined Acceptance Criteria

Predetermined Attributes Test Method Procedure Acceptance Priority Criteria Purity (Vp1/Vp2/Vp3) Measure capsid proteins SDS-PAGE X High Residual host DNA Measure total DNA qPCR X Medium Measure replication- Replication-competent AAV qPCR X Medium competent virus DNA identity Transgene sequence NGS X Medium Vector concentration Measure vg/mL ddPCR X High Vector infectivity Measure infectivity TCID50 X Medium Measure of microbial Sterility Culture based Mandatory contamination Endotoxin Measure of endotoxin LAL X Mandatory Aggregation Measure of aggregation Drug Product X High Full/empty capsid ratio Measure of percent full IEX-HPLC X High Potency Measure RNA, protein Cell-based assay or X High expression, or biological preclinical model activity

process should be evaluated in short- and long-term conclusion. The conclusion drawn by the manufacturers stability studies sufficient to initiate the clinical study. should be articulated in a succinct and comprehensive The shelf life of batches released post change may be fur- format. In cases for which the data are not complete or ther extended. Stability-indicating assays include vector the information collected is not supportive of estab- concentration, vector infectivity, potency, and measure- lishment of comparability based on the predetermined ment of other characteristics that could potentially be acceptance criteria, the manufacturers are encouraged to impacted during storage conditions. include a section on risk assessment to better define the The stability studies can also be performed in real-time, risk to product quality and define next steps/future plans. accelerated, or under stress conditions. Accelerated and stressed stability studies are not always possible and/or recommended. However, accelerated stability studies Case Study (e.g., at elevated temperatures or under other stress con- Let’s consider a real-life case of an AAV product for ditions relevant for the product of interest) may provide which a significant manufacturing change is introduced. complementary supporting evidence for the stability For the purpose of this case study, the AAV manufac- of the product and help to establish the stability profile turing platform will be changed from an adherent to a during temperature excursions. They are often useful tools suspension culture to be conducted in a new facility. This to establish degradation rates and/or pathways, identify change is being introduced very late during the product stability-indicating tests, and provide a direct comparison development cycle such that the IND holder cannot of pre-change and post-change product. generate sufficient clinical data to assess the product effectiveness prior to licensure. COMPARABILITY STUDY CONCLUSION First, the most relevant attributes to the study should The conclusion of a comparability study should contain a be determined, as well as the methods and procedures detailed summary of the results collected with a bulleted that should be used to measure these attributes. A

CHAPTER 8 Comparability 213 Table 8-7. AAV Product Attributes, Number of Batches Tested, Predetermined Acceptance Criteria

Predetermined Batch* Batch* Results Method Batch Batch Batch Attributes Acceptance Before Before (pass or Procedure 1 2 3 Criteria Change Change fail) Purity (VP1/Vp2/Vp3) SDS-PAGE Min-max range

Residual host DNA/ protein qPCR X

Replication-competent AAV qPCR X

DNA identity NGS X

Vector concentration dd-PCR X

Vector infectivity TCID50 X

Culture- Sterility based

Endotoxin LAL X

Drug X Aggregation Product

Full/empty capsid ratio IEX-HPLC X

Cell-based X Potency assay

*Batches represent product used in clinical studies prior to implementing major manufacturing changes. The testing of different attributes should be performed if possible, using sample retained from old batches and compared to the new batches using the same method under identical conditions.

risk- and science-based approach allows prioritization acceptance criteria and statistical methods, the manufac- of the relevant attributes, which may include biological turer could argue that the product quality is not affected activity, potency, identity, and purity. These attributes by the process change. However, if the product is not are key elements of comparability studies along with shown to be similar in one or two out of 10 parameters standard safety tests such as sterility and endotoxin. measured, for example, manufacturers may consider Based on the risk assessment and assay considerations, conducting a risk assessment to determine the impact of it is possible to select a subset of release tests to evaluate such excursions on product quality. The risk assessment comparability with the highest relevance to product must take into account several factors, for example, attri- quality and effectiveness. These tests must be performed bute criticality and product efficacy, when predetermined on the final DS/DP, but it is also important to monitor acceptance criteria are not met. Further, it is important to the process by conducting additional in-process testing determine the frequency failure observed in the test and (which could be a subset of the tests described here). whether there is a known reason for the observed failure Table 6 summarizes an example of a prioritized list of that could be related to operator training or excursions attributes to be measured, testing methods, procedures, in test methods and procedures. and predetermined acceptance criteria. Table 7 provides a summary of the results collected for each attribute for POSSIBLE SCENARIOS several batches and indicates whether the results met the If the comparability study results are not sufficient to predetermined acceptance criteria. establish comparability based on the in vitro results, it is If the pre- and post-change products exhibit char- possible that the agency would ask for additional infor- acteristics that are similar based on the pre-determined mation. Because a major change is introduced late during

CHAPTER 8 Comparability 214 Table 8-8. Possible Outcomes for Introducing Major Manufacturing Changes After Completion of Pivotal Study

Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6 X X X X X X In vitro analytical similarity

X X X X X Preclinical analytical similarity

Bridging clinical study before X X licensure X Post-marketing clinical requirement

Additional pivotal study before X licensure the product life cycle in this hypothetical case, the agency prior to licensure. The regulatory scenarios discussed may ask for additional in vitro studies, establishment of here are purely hypothetical, and manufacturers are comparability in relevant animal models, and estab- encouraged to have early discussions with the agency lishment of comparability based on additional clinical and obtain clarification from the agency regarding their studies prior to or after licensure. Agreement with the expectations of comparability results and possible sce- agency on the comparability approach and study design narios (see Table 8 for possible scenarios). does not bind the agency to accept the results as adequate Table 8 summarizes possible outcomes of introducing if the results do not meet the quality standards. For gene major manufacturing changes using worst-case scenar- therapy products, the use of animal models to establish ios. The requirement for establishing comparability is product comparability is generally highly recommended shown for a hypothetical gene therapy product. In this if available and should be included in the proposed study. scenario, the product is undergoing a major change after In some cases, the preclinical establishment of compara- the pivotal or licensing trial has been completed. bility may not be sufficient, and the agency may require additional clinical data to establish product comparability Conclusion (e.g., small bridging study). In general, additional small bridging clinical studies are not meant to demonstrate To date, a significant amount of critical insights into efficacy using the newly manufactured commercial prod- AAV vectors as tools for gene delivery have been uct but are focused on establishing safety and perhaps gathered through preclinical and clinical studies. We limited effectiveness based on patient outcome and/ now know that the AAV therapeutic platform has the or surrogate biomarkers. It should be noted that if the ability to express a given transgene at therapeutic levels treatment effects are high and easily detectable, then it for multiple years, potentially representing a cure for may be possible to demonstrate effectiveness by treating chronic genetic diseases. Although safety must remain a small number of patients. However, if the treatment ef- the overarching goal for the field, achieving therapeutic fects are not as apparent, then additional clinical studies efficacy in a consistent manner in adults and pediatric may need to be performed prior to or after licensure as patients will likely be essential for gene therapies to part of post-marketing requirements to verify the final become competitive with other approaches that are drug product effectiveness using the newly manufactured emerging as treatment modalities for genetic diseases. commercial product. In some scenarios, a small bridg- To meet the challenge, many aspects of AAV biological ing study prior to licensure or post-marketing studies properties in the context of the human host, such as may not be sufficient or informative to establish clinical AAV vector immunogenicity, therapeutic potency, comparability, thus requiring a stand-alone clinical study persistence, and potential genotoxicity, will have to be

CHAPTER 8 Comparability 215 further elucidated. Even though preclinical animal mod- Table 8-9. Simulated Potency Data from els cannot be used to accurately predict the outcome Five Processes of gene transfer in humans, they will continue to be essential for the development of highly optimized gene Batch Process Potency (%) Average therapy drugs. 1 1 97 97 In this chapter, we have discussed that major changes in the manufacturing of gene therapy products are an 2 2 120 inevitable part of process improvement. As such, man- 3 2 88 104 ufacturers should have a well-defined plan to introduce 4 2 103 manufacturing changes and establish product compara- 5 3 98 bility. A comparability plan should be developed as early 97 as possible in product development, preferably before a 6 3 96 phase 1 trial. The acceptability of comparability depends 7 4 102 on a large number of factors, including but not limited 8 4 93 103 to knowledge of the product, availability of well-defined 9 4 107 CQAs that are informative for product safety and efficacy, 10 4 108 and timing of when the major change is introduced in the 11 5 83 product development cycle. If new methods are imple- mented during development, retaining samples for retest- 12 5 111 96 ing is important for identifying any important changes to 13 5 95 product quality or efficacy, as well as to determine whether new analytical methods will be required to detect changes not accounted for by old analytical methods. Ultimately, the potency values are randomly generated by computer the more predictive or informative analytical parameters from the same normal distribution with mean 100% that are included when measuring product quality and and standard deviation of 10%. Therefore, 99.73% of the efficacy, the less likely it is that bridging clinical studies will potency values are expected to fall within (70%, 130%). be required. For this reason, it is strongly recommended that manufacturers introduce major manufacturing Visual Assessment changes early during the development cycle. If this is The potency values from 5 processes that overlap with not possible, manufactures should anticipate the need each other quite well (Figure 6) indicate that these for well-qualified assays and highly relevantin vitro and processes appear to be comparable in potency. Note preclinical measurement tools to establish product com- that if the comparison were made prior to Process 3, parability during later phases of clinical study. the highest value from Process 2 (120) would likely cause unnecessary concern. This type of situation often Appendix occurs when the sample size (i.e., number of batches per process) is limited. ILLUSTRATION OF THE DIFFERENT STATISTICAL Statistical Assessments APPROACHES FOR COMPARABILITY As discussed previously, formal statistical approaches Process changes often occur during gene therapy product should be applied only when the number of batches development. In the following hypothetical example, reaches ≥10. In this case, the total number of the first the process has changed four times and results in five 4 processes is 10. Therefore, the first 10 batches can processes from P1 to P5 (Table 9). The number of batches be treated as pre-change batches to compare with the per process ranges from 1 to 4. The potency is not expect- batches from Process 5, the post-change process. Two ed to be affected by these process changes. Therefore, all types of statistical approaches were introduced in the

CHAPTER 8 Comparability 216 statistical section: 1) statistical intervals (e.g., min/max, Figure 8-6. The Potency (%) Value for Batches confidence interval [CI], mean ± 3 SD, simultaneous from Different Processes prediction interval (SPI), and tolerance interval (TI)) and P1 P2 P3 P4 P5 2) hypothesis testing (e.g., T-test and equivalence test). 120 Min/max, mean ± 3 SD, SPI, and TI are calculated based on the first 10 batches and the number of post- change batches (3) and are presented in Table 10 and Figure 7. Note that 99% represents the coverage level. If 110 these intervals are used as the comparability criteria, all batch values from Process 5 are expected to fall within these intervals. However, Batch #11 from Process 5 fails the min/max criteria. It is worth noting that even 100 though Process 5 passes the mean ± 3 SD criteria, there is a fair chance that it could also fail the mean ± 3 Relative potency (%) potency Relative SD criteria because this interval is narrower than the 90 true mean ± 3 SD, which is (70%, 130%). Even though it is wider than the true mean ± 3 SD, SPI is a good compromise between the often overly wide TI and the often overly narrow min/max and mean ± 3 SD based 1 2 3 4 5 6 7 8 9 11 12 on the limited sample size. Lot number The focus of the CI approach is based on the com- parison between the means of the pre- and post-change process. In this case, the CI of the mean difference Figure 8-7. Potency for Each Batch with the between the pre- and post-change processes can be Different Statistical Intervals calculated using the following formula assuming that Based on the Pre-change Processes (P1 to P4) and the variability is not affected by the process change: Number of Batches in the Post-change Process (P5) 160 P1 P2 P3 P4 P5 1 1 !𝑚𝑚!"#$ − 𝑚𝑚!%&$ ± 𝑡𝑡 × 𝑠𝑠 × ) + 𝑛𝑛!%& 𝑛𝑛!"#$ 140 where and are means of the post- and pre-change process and and are the where mpost and mpre are means of the post- and pre- corresponding sample sizes. Inchange this case, process and n and n are the. correspondingt is the multiplier based on central 𝑚𝑚!"#$ 𝑚𝑚!%& pre post 𝑛𝑛!%& 𝑛𝑛!"#$ s 120 t-distribution and sample sizesample and corresponding sizes. !%&In this confidence case, npre!"#$=10 level and. nispost the=3. pooled t is the standard deviation of the two processes. The 99% CImultiplier of the mean based𝑛𝑛 difference on= 10central is ( 𝑛𝑛-t-distribution26%, 16%),= 3 which and indicatessample the range that the true mean difference betweensize the and processes corresponding is very likelyconfidence fall within. level. It s isis theup topooled the expert to defend or justify whether the potential standarddifference deviation is small enough of the twoto claim processes. comparability. The 99% CI 100 of the mean difference is (-26%, 16%), which indicates Relative potency (%) potency Relative Table 10: Statistical Intervals theBased range on the that First the 10 true Batches mean and difference Number betweenof Post-Change the Batches processes is very likely fall within. It is up to the expert 80 Do Process 5 Batches Fall Statistical Intervals to defend or justifyResult whether (%) the potential difference is small enough to claim comparability. Within the Interval? Min/max The T-test and (88,equivalence 120) tests are both hypoth- No 60 Mean ± 3 SD esis tests that focus(74, on 128) mean comparison, but have Yes 1 2 3 4 5 6 7 8 9 10 11 12 13 99% SPI opposite goals. The(64, T-test 139) evaluates whether pre- and Yes TI with 99% coverage and 95% Batch number (61, 142) Yes confidence Note: only SPI requires both the potencyCHAPTER values ofX: the xxxx first 10 batches and the number of post-change batches (i.e., the number 28 CHAPTER 8 of Process 5 batches) as input. All other intervalsComparability are solely based on the potency values of the first 10 batches. 217

The T-test and equivalence tests are both hypothesis tests that focus on mean comparison, but have opposite goals. The T-test evaluates whether pre- and post-change processes are significantly different. The equivalence test evaluates whether the difference between the two processes is within a preset acceptable limit, often called equivalence margin. The equivalence margin must be set prior to data collection. Assuming equal variance, the p-value of the T-test is 0.4807, indicating that the difference is not statistically significant.

The equivalence test is equivalent to comparing the CI of the mean difference with the equivalence margin. If the confidence interval (usually with 90% confidence level) is fully within the equivalence margin, then a claim of “equivalence” can be made. If the equivalence margin is set as ±10%, then the 90% CI of the mean difference is (-17%, 7%) and it falls outside the equivalence margin. Therefore, Process 5 does not pass this equivalence criterion, which conflicts from the truth that all the batches are from the same distribution.

Table 8-10. Statistical Intervals Based on the First 10 Batches and Number of Post-Change Batches

Do Process 5 Batches Fall Within Statistical Intervals Result (%) the Interval? Min/max (88, 120) No Mean ± 3 SD (74, 128) Yes 99% SPI (64, 139) Yes TI with 99% coverage and 95% confidence (61, 142) Yes post-change processes are significantly different. The If the confidence interval (usually with 90% confidence equivalence test evaluates whether the difference be- level) is fully within the equivalence margin, then a claim tween the two processes is within a preset acceptable of “equivalence” can be made. If the equivalence margin limit, often called equivalence margin. The equivalence is set as ±10%, then the 90% CI of the mean difference is margin must be set prior to data collection. Assuming (-17%, 7%) and it falls outside the equivalence margin. equal variance, the p-value of the T-test is 0.4807, indi- Therefore, Process 5 does not pass this equivalence cri- cating that the difference is not statistically significant. terion, which conflicts from the truth that all the batches The equivalence test is equivalent to comparing the are from the same distribution. CI of the mean difference with the equivalence margin.

Abbreviations

AAV Adeno Associated Virus GDUFA Generic Drug User Fee Act BLA Biologics License Application GMP Good Manufacturing Practice CBE Changes Being Effected ICH International Council for Harmonisation CBER Center for Biologics Evaluation and Research IND Investigational New Drug CFR Code of Federal Regulations KPP Key Process Parameters cGMP Current GMP MOI Multiplicity of Infection CI Confidence Interval PAS Prior Approval Supplement CMC Chemistry Manufacturing and Controls PCR Polymerase Chain Reaction CMO Contract Manufacturing Organization QA Quality Assurance CPP Critical Process Parameters QbD Quality by Design CQA Critical Quality Attributes QTPP Quality Target Product Profile ddPCR Droplet Digital PCR RNA Ribonucleic Acid DNA Deoxyribonucleic Acid SD Standard Deviation DP Drug Product SPC Statistical Process Control DPI Drug Product Intermediate SPI Simultaneous Prediction Interval DS Drug Substance TI Tolerance Interval DSI Drug Substance Intermediate USP United States Pharmacopeia EMA European Medicines Agency WHO World Health Organization FDA Food and Drug Administration

CHAPTER 8 Comparability 218 Endnotes 1. U.S. Department of Health and Human Services, Food 7. Gavin DK. Advanced topics: successful development of and Drug Administration, Center for Drug Evaluation quality cell and gene therapy products. U.S. Food and and Research (CDER), Center for Biologics Evaluation Drug Administration website. https://www.fda.gov/me- and Research (CBER). ANDA submissions – prior ap- dia/80404/download. Accessed September 22, 2020. proval supplements under GDUFA: guidance for industry. U.S. Food and Drug Administration website. https://www. 8. Burdick RK, LeBlond DJ, Pfahler LB, et al. Statistical fda.gov/media/89263/download. Published October 2017. Applications for Chemistry, Manufacturing and Controls Accessed September 18, 2020. (CMC) in the Pharmaceutical Industry. Basel, Switzer- land: Springer International Publishing AG; 2017. 2. U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation 9. Dong X, Tsong Y, Shen M. Statistical considerations and Research (CDER), Center for Biologics Evaluation in setting product specifications. J Biopharm Stat. and Research (CBER). Chemistry, manufacturing, and 2015;25(2):280-294. controls changes to an approved application: certain 10. European Medicines Agency. Reflection paper on statisti- biological products: draft guidance for industry. U.S. Food cal methodology for the comparative assessment of quality and Drug Administration website. https://www.fda.gov/ attributes in drug development. EMA website. https:// media/109615/download. Published December 2017. www.ema.europa.eu/en/documents/scientific-guideline/ Accessed September 18, 2020. draft-reflection-paper-statistical-methodology-compara- 3. U.S. Department of Health and Human Services, Food tive-assessment-quality-attributes-drug_en.pdf. Published and Drug Administration, Center for Drug Evaluation March 23, 2017. Accessed September 22, 2020. and Research (CDER), Center for Biologics Evaluation 11. Burnett R. Accurate estimation of standard deviations and Research (CBER). Chemistry, manufacturing, and for quantitative methods used in clinical chemistry. Clin controls changes to an approved application: certain Chem. 1975;21(13):1935-1938. biological products: draft guidance for industry. U.S. Food and Drug Administration website. https://www.fda.gov/ 12. Hahn GJ, Meeker WQ. Statistical Intervals: A Guide for media/109615/download. Published December 2017. Practitioners. New York, New York: John Wiley & Sons; Accessed September 18, 2020. 1991. 4. U.S. Department of Health and Human Services, Food 13. Hahn GJ, Meeker WQ. Statistical Intervals: A Guide for and Drug Administration, Center for Drug Evaluation Practitioners. New York, New York: John Wiley & Sons; and Research (CDER), Center for Biologics Evaluation 1991. and Research (CBER). Chemistry, manufacturing, and 14. Burdick RK, O’Neill JC. The Goldilocks challenge–con- control (CMC) information for human gene therapy in- trolling uncertainty when setting product specifications. vestigational new drug applications (INDs). U.S. Food and PDA J Pharm Sci Technol. 2020;74(4):439-445. Drug Administration website. https://www.fda.gov/me- dia/109615/download. Published January 2020. Accessed 15. Dong X, Tsong Y, Shen M. Statistical considerations September 18, 2020. in setting product specifications. J Biopharm Stat. 2015;25(2):280-294. 5. Alliance for Regenerative Medicine. ARM-USP Workshop Comparability in Cell & Gene Therapies Final Report & 16. U.S. Department of Health and Human Services, Food Summary. Rockville, MD: May 31, 2019. and Drug Administration, Center for Drug Evaluation and Research (CDER), Center for Biologics Evaluation 6. U.S. Department of Health and Human Services, Food and Research (CBER). Development of therapeutic and Drug Administration, Center for Drug Evaluation and biosimilars: comparative analytical assessment and other Research (CDER), Center for Biologics Evaluation and Re- quality-related concerns: guidance for industry. U.S. Food search (CBER). Guidance for industry Q5E comparability and Drug Administration website. https://www.fda.gov/ of biotechnological/biological products subject to changes in media/125484/download. Published May 2019. Accessed their manufacturing process. U.S. Food and Drug Adminis- September 22, 2020. tration website. https://www.fda.gov/media/71489/down- load. Published June 2005. Accessed September 22, 2020.

CHAPTER 8 Comparability 219 Endnotes, continued

17. Tsong Y, Shen M, Dong X. Development of statistical ap- 25. U.S. Department of Health and Human Services, Food proaches for analytical biosimilarity evaluation. Presented and Drug Administration, Center for Drug Evaluation at DIA/FDA Statistics Forum 2015. North Bethesda, MD: and Research (CDER), Center for Biologics Evaluation April 20-22, 2015. and Research (CBER), Center for Veterinary Medicine (CVM). Guidance for industry – process validation: 18. U.S. Department of Health and Human Services, Food general principles and practices. U.S. Food and Drug Ad- and Drug Administration, Center for Drug Evaluation ministration website. https://www.fda.gov/media/71021/ and Research (CDER), Center for Biologics Evaluation download. Published January 2011. Accessed September and Research (CBER). Development of therapeutic 22, 2020. biosimilars: comparative analytical assessment and other quality-related concerns: guidance for industry. U.S. Food and Drug Administration website. https://www.fda.gov/ media/125484/download. Published May 2019. Accessed September 22, 2020. 19. Gutka HJ, Yang , Shefali K. Statistical considerations for demonstration of analytical similarity. In: Biosimilars: regulatory clinical and biopharmaceutical development. Berlin, Germany; Springer: 2018. 20. Tsong Y, Shen M, Dong X. Development of statistical ap- proaches for analytical biosimilarity evaluation. Presented at DIA/FDA Statistics Forum 2015. North Bethesda, MD: April 20-22, 2015. 21. Tsong Y, Shen M, Dong X. Equivalence margin determi- nation for analytical biosimilarity assessment. Presented at IABS Workshop. Rockville, MD: June 2015. 22. U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER). Development of therapeutic biosimilars: comparative analytical assessment and other quality-related concerns: guidance for industry. U.S. Food and Drug Administration website. https://www.fda.gov/ media/125484/download. Published May 2019. Accessed September 22, 2020. 23. Cheng A, Wang K. Process comparability: setting appro- priate acceptance criteria at late stage/post approval for biological products. Presented at: Midwest Biopharmaceu- tical Statistics Workshop; May 20-22, 2019: Carmel, IN. 24. U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER), Center for Veterinary Medicine (CVM). Guidance for industry – process validation: general principles and practices. U.S. Food and Drug Ad- ministration website. https://www.fda.gov/media/71021/ download. Published January 2011. Accessed September 22, 2020.

CHAPTER 8 Comparability 220

A-Mab Case Study v5 (No Backup)

Date post:	22-Oct-2015
Category:
Upload:
View:	235 times
Download:	18 times

The 2nd QbD Conference

QbD in Large Molecules:

Th A M b t d b th CMC Bi t h

The A-Mab case study by the CMC Biotech Working groupDan Kenett

May 5 2010 Jerusalem

Background: Therapeutic Monoclonal Antibodies

Selected A-Mab chapters:

Design of Molecule and Quality Attributes Assessment

Upstream Manufacturing Process Development

Control Strategy

Therapeutic Monoclonal Antibodies

A total of 21 mAb products are approved in the US, with additional products marketed outside the US

mAb therapeutics are now being developed and marketed by most mAb therapeutics are now being developed and marketed by most major pharmaceutical firms

7 mAbs have global sales of over US$1 billion

In 2006 global market = US$20 billion ,

in 2010 expected to reach US$40 billion

over 200 mAb candidates are currently undergoing clinical study

Reichert J.M. Current Pharmaceutical Biotechnology 2008, 9, 423-430

Monoclonal antibody structure

Schiesti M. (Sandoz) PMDA symposium 2009

Possible natural modes of action of mAbs versus cells cells

6 Schiesti M. (Sandoz) PMDA symposium 2009

Cell fusion with cancerous cells recombinant DNA technology

1975 1986 1994 1997 2002

Schematics of a manufacturing process of a monoclonal antibody

Kozlowski and Swann , Quality by Design for Biopharmaceuticals 2009

Upstream Downstream

FDA Initiatives and QbD Timeline

Small molecule ACEmock case study by

Large molecule A-Mab mock case study by CMC

Biotechnology Working Group

QbD Mock Case studies

Small Molecules (Pharmaceutical)

1. Examplain (P2 module of CTD format) – 2006

EFPIA http://www.efpia.org/Content/Default.asp?PageID=450

2. Acetriptan (ACE) - 2008

cet pta ( C ) 008

Conformia software collaboration with Abbot, AstraZeneca, Eli Lilly, GlaxoSmithKline

3. Illustrain Hcl (S2 module of CTD format)- 2009 target

Monoclonal Antibody (Biopharmaceuticals)

1. A-Mab - 2009

CMC working group of 7 leading Biotech companies:Amgen,

Genentech, Abbot Bio MedImmune, GlaxoSmithKline Bio, Eli

Lilly , Pfizer Biohttp://www.casss.org/displaycommon.cfm?an=1&subarticlenbr=286

2. Mockestuzumab (S2 and P2 module of CTD format) –2010 target

What is A-Mab ?

Humanised IgG1 against Lymph-1 (a surface antigen on CD20 B cells) which is expressed at high levels at surface of B cells from NHL (Lymphoma) patients

A-Mab stimulates CD20 B cells killing primarily through ADCC and possibly also by CDC

It is produced by recombinant DNA technology in CHO cells It is produced by recombinant DNA technology in CHO cells

Delivered by IV administration at a weekly dose of 10mg/kg for 6 weeks

12http://images.google.com/imgres?imgurl=http://rss.xinhuanet.com/newsc/english/2008-03/11/xin_142030511140539054143.jpg&imgrefurl=http://rss.xinhuanet.com/newsc/english/2008-03/11/content_7765049.htm&usg=__oTBauuUxMAYhD-9QSKw8J0uKqs4=&h=442&w=450&sz=21&hl=

The A-Mab chapters

1. Design of Molecule and Quality Attributes Assessment

2. Upstream Manufacturing Process Development

3. A-Mab Downstream Process Description and Characterization3. A Mab Downstream Process Description and Characterization

4. Drug Product

5. Control Strategy

6. Regulatory Section

Initial Target Product Profile

TPP=Prospective and dynamic summary of the quality characteristics of a drug product that id ll ill b hi d t th tideally will be achieved to ensure that the desired quality, and thus the safety and efficacy, of a drug product is realized

Quality attributes of Drug Product not yet defined

Design Features of A-Mab

The design strategy for A-Mab was:

To maximize clinical performance

To minimize potential impact on quality

To mitigate risk from the following product attributes:

By planning the DNA sequence

Unpaired cysteine residues (reduced risk of undesirable disulfide bond formation) Potential deamidation sites in the CDRs (reduced risk of deamidation) O-linked glycosylation sites (reduced risk of heterogeneity and impact on bioactivity)

N-linked glycosylation sites in the CDRs (reduced risk of heterogeneity and impact on bioactivity)

Acid labile (DP) sequences (reduced risk of fragmentation) Oxidation sites in the CDR

Quality Attributes of monoclonal antibodies

Quality attributes that can vary quantitatively and qualitatively in a process

Systematic Identification of critical Quality Attributes

Page 10: A-Mab Case Study v5 (No Backup)

Overview of A-Mab Product Realization Process

Page 11: A-Mab Case Study v5 (No Backup)

Platform Knowledge

Quality Attribute Risk Assessment Tools

Three types of tools for assessing criticality of quality attributes are presented as examples:

Risk ranking (Tool #1)

Criticality = Impact x uncertainty Criticality = Impact x uncertainty

Preliminary hazards analysis (PHA) (Tool #2)

Criticality = Severity (safety,efficacy) x Likelihood (probability of AE due to out of range)

A safety assessment decision tree for evaluating process-related impurities that do not have biological activity based on ISF(Tool #3)

The impurity safety factor (ISF) = LD50 ÷ Level in Product Dose

Page 12: A-Mab Case Study v5 (No Backup)

Quality Attribute Assessment Tool #1

Impact: The impact ranking of an attribute assesses either the known or potential consequences on safety and efficacy. The impact ranking considers the attribute‘s effect on:considers the attribute s effect on:

1. efficacy, either through clinical experience or results using the most relevant potency assay(s),

2. pharmacokinetics/pharmacodynamics (PK/PD), 3. Immunogenicity4. safety.

Impact : Definition and Scale for Tool #1

Page 13: A-Mab Case Study v5 (No Backup)

Uncertainty: Definition and scale for Tool #1

Criticality: Definition

Criticality (Risk Score) = Impact × Uncertainty

All quality attributes are assigned a degree of criticality (criticality continuum) based on their respective risk score. Risk scores range between a low of 2 to a high of 140.

Page 14: A-Mab Case Study v5 (No Backup)

Examples of Quality Attribute Risk Assessment with Tool#1

A-Mab examples of Tool #1

Aggregation

DNAModerate high

The individual impact category with the highest ranking determines

the overall impact ranking for an attribute

None Moderate

Page 15: A-Mab Case Study v5 (No Backup)

Summary of Quality attribute Risk assessments

Vary according to mode of action

Critical quality attribute includes H and VH

Critical Quality attributes and target ranges

Page 16: A-Mab Case Study v5 (No Backup)

Design Space for Bioreactor Production with Reliability dimension Control Space Based on the Overall Reliability of

the Process (Predictive Bayesian Reliability approach)Design space for culture duration 15 days

Regions in dark-red possess > 99%reliability to satisfy all the CQA limitsand determine the Control Space

To achieve a response surface model by DOE suitable to define a Design spaceN=40 bioreactor runs (4 blocks of 10, ~12 weeks) X-Mab data

Design space for Production Bioreactor

Page 18: A-Mab Case Study v5 (No Backup)

Design space examples in A-Mab

Upstream production bioreactor

Low pH viral inactivation

Drug product compounding

Drug product sterile filtration

Bioreactor engineering

Page 19: A-Mab Case Study v5 (No Backup)

QbD vs Traditional approach for the Control Strategy

Risk Assessment Approach used through Development Lifecycle and for setting Control Strategy

Prior knowledge and early development experience used to identify parameters and attributes that must be considered for process characterization studies

Page 20: A-Mab Case Study v5 (No Backup)

The cumulative process understanding serves as the basis for the late-phase risk assessments used to finalize selection of Critical Process Parameters (CPPs) that underpin the proposed design spaces and control strategy.

Final Categorization of Process Parameters for A-Mab Control Strategy

Page 21: A-Mab Case Study v5 (No Backup)

Example: Risk Assessment Results for Process Parameters in the N-1 and Production Bioreactor

Categorization of Process parameters examples in A-Mab

Protein chromatography

Low ph viral inactivation

Cation exchange chromatography

Anion exchange chromatography

Small virus retention filtration

Page 22: A-Mab Case Study v5 (No Backup)

Control Strategy for commercial manufacturing

Output of Risk assesment Tool #1

Each CQA is evaluated independently to ensure that the proposed control strategy will deliver each CQA within its acceptable ranges established for safety and efficacy. A Failure Mode and Effects Analysis (FMEA) approach was used

Example: Final Risk Assessment Results for

Process Parameters in the Production Bioreactor

Initial Process parameters categorization …And after Design space (Control space)fine tuning

Glycosylation: 8 WC-CPP, 3 CPP Glycosylation: 5 WC-CPP,

Page 23: A-Mab Case Study v5 (No Backup)

Process Capability Scales for Severity, Occurrence and Detection

The overall score (RPN=SxOxD) was calculated for each unit operations

Scoring for Process Capability Risk Assessment of Unit Operations

Page 24: A-Mab Case Study v5 (No Backup)

Examples of Process Capability Risk Assessment

(Risk assessment #3)

A-Mab Control Strategy

Product Understanding

Page 25: A-Mab Case Study v5 (No Backup)

Elements of the Control Strategy

Page 26: A-Mab Case Study v5 (No Backup)

Specifications Tests

Based on the enhanced product and process understanding, specification tests are significantly reducedcompared to traditional approaches.

Some specification testing has been moved to in-processtests (including PAT) while other tests were eliminated because operation within the process design space provides a high degree of assurance that the process will deliver consistent product quality

deliver consistent product quality

Page 27: A-Mab Case Study v5 (No Backup)

Example: Control elements for Oligosaccharide Profile

Process capability risk assesment; RPN=30 Low

Example: Integrated Control elements for Aggregation

Process capability risk

Page 28: A-Mab Case Study v5 (No Backup)

Control elements for Aggregate1 and Oligosaccharide2 profile in A-Mab

CQA Process capability Risk

Input material Testing

Procedural control

Process parameter control

In-process tests

Specification Process Monitoring

Stability Testing

Characterization Testing

Aggregate High Y Y N N Y Y Y Y

Oligosaccharide profile

Low Y Y Y N N Y N Y

1 summary of control elements assigned to various downstream process steps

2 control elements assigned to bioreactor production

Arrow indicates differences

Control strategy examples in A-Mab

Drug product filling, stopping, capping

Page 29: A-Mab Case Study v5 (No Backup)

For the purposes of this case study, only a selected number of quality attributes were considered to define the Control Strategy. Normally, the approaches presented here would be expanded to include all critical quality attributes

Quality Attribute Ranges forA-Mab Process

LRV=log reduction value

A-Mab is a document for public consumption and ultimately a backbone for further discussion between industry and agencies across 2009-2010 and beyond

First comprehensive exemplification of a Mock Biotech product developped with QbD principles

Essential in order to overcome “conceptual hurdles” facing those involved in applying QbD

Science based driven all the way

Risk assessment tools used for identification of CQA and categorization of Process parameters

of Process parameters

Illustration of a complex Design Space for the Bioreactor production stage and of the innovative Engineering Design Space concept

The Control Strategy provides a high degree of assurance that the product quality specifications are met.

Page 30: A-Mab Case Study v5 (No Backup)

Take away from A-Mab

Traditional Product Development

Quality by Testing (reactive)

Product=Process

Enhanced Product Development

Quality by Design (proactive)

Targetted Product=Characterised and controlled Process

Targetted Product Characterised and controlled Process

And thanks to the Biotech CMC Working Group for this inspiring masterpiece !!!

MAB Program Overview

Membership Activities Board (MAB) Committee Makoto Kaneko VP for MAB.

FAGOR AUTOMATION MAB

September October 2008 MAB

Veeam Backup & Replication Standard and Enterprise Editions · 2015. 8. 19. · Veeam Software offers 2 editions of its award-winning Veeam Backup & Replication (v5 and above) solution.

Met mab nov 2010

Member Activities Board (MAB) Committee Makoto Kaneko VP for MAB.

MAb Quantitation: Protein A HPLC vs. Protein A Bio-layer ... · MAb purification 3 Mab Process development Consensus Mab purification Majority of MAb purification processes employ

BackupAssist V3 vs V5 Comparison - Windows Server Backup and

Queen Mab Speech

5 3 1 - mmcontrol.com · PS531 V5 May 2002 5 3 1 5 3 1 PLC, DCS, PC CONTROL BACKUP STATION The Interface Solution Experts Auto/Manual Station Hot PID Backup Station

Queen Mab Monologue

MAb to Rubella (E1) MAb to Rubella C Protein MAb to ... · 10190 MAb to Rubella (E1) 1ml 717-Eur Meridian Life Science monocl onal ... -Lp(a) 100ug 468-Eur Meridian Life Science human

I-MAB Presentation

MAB Dep Guide

Veeam Backup & Replicaton v5 with vPower and …...2011/10/04 · • Veeam Backup and Replication v5 vee.am/backupreplication • Veeam Monitor vee.am/vmwmntr • Veeam Reporter

MANUAL MAB 102.pdf

The N-mAb case study brought together over 60 industry and government stakeholders from over 20 organizations to develop shared expectations and vocabulary around a control strategy for an integrated continuous bioprocess for a hypothetical monoclonal antibody. This document presents a starting place for conversations within organizations around implementing continuous integrated bioprocesses in commercial manufacturing.

N-mAb Case Study

N-mAb Poster Presentation

How to cite n-mab.

National Institute for Innovation in Manufacturing Biopharmaceuticals. 2022. N-mAb: A case study to support development and adoption of integrated continuous bioprocesses for monoclonal antibodies. Version 1. N-mab.org. Download Date.

Project A-Gene and Project A-Cell

NIIMBL collaborated with the Alliance for Regenerative Medicine (ARM) on two case studies related to integrating Quality by Design (QbD) principles into cell and gene therapy programs:

Project A-Gene

Case study-based approach to integrating QbD principles into gene therapy Chemistry, Manufacturing, and Controls (CMC) programs

Project A-Cell

Project highlights, workforce expansion in biomanufacturing emerging technologies, modularized pat online training platform to accelerate the workforce innovation in biopharmaceuticals manufacturing, worde: a biopharmaceutical training center to promote development of the workforce, outreach, research, diversity, and education, international assessment of pspt in mice to replace the intracerebral-challenge mouse protection test (mpt) for whole-cell pertussis (wp), a multivariate in-line optochemical sensor platform for continuous monitoring of cross-category process parameters and product attributes in bioreactors, niimbl-biophorum buffer stock blending system, become a member.

We offer a variety of membership options that give you the flexibility to choose your organization’s level of engagement based on technology interests and priorities.

View / Download N-mAb Case Study

View / download n-mab poster presentation test.

IMAGES

A–Mab: A Case Study in Bioprocess Development
QbD Model Case Study of MONOCLONAL ANTIBODY : A-Mab
QbD Model Case Study of MONOCLONAL ANTIBODY : A-Mab
Biosimilar
A-Mab Case Study Version 2-1
QbD Model Case Study of MONOCLONAL ANTIBODY : A-Mab

VIDEO

Nintendo DS Longplay [027] Mega Man ZX (Part 3 of 4)
Sports Card Breaks! NHL #478!
Collapse of Submarine Pressure Hulls
Khairy: Najib's trial complicates things
SBS訪問Richard Wong律師拆解澳洲學生簽證及畢業生簽證最新變動 07-03-2024 #澳洲移民 #澳洲移民法 #學生簽證 #畢業生簽證
Preliminary hearing delayed after State asks for continuance in Monument Advisor case

COMMENTS

A-Mab: A Case Study in Bioprocess Development
A-Mab: A Case Study in Bioprocess Development. 30 October 2009. CMC Biotech Working Group. Version 2.1. This is a detailed case study to stimulate discussion around how the core principles contained in Q8(R2), Q9 and Q10 guidelines could be applied to product realisation programs for a biotechnology-derived monoclonal antibody.
PDF CMC Biotech Working Group
Product Development and Realisation Case Study A-Mab The CMC Biotech Working Group Page 5 of 278 4.6.2.5 Characterization Studies to Assess Impact to Product Quality
PDF A-mAb Study Guide
Study Guide. Version 2.1. 28 th October 2009. A-Mab Study Guide. Study Guide Introduction and Objectives. The Study Guide was developed as a tool to help drive discussions related to the A-Mab Case Study. We recognize that within the Case Study itself, there are areas where the concepts behind QbD were interpreted and subsequently, implemented ...
A-Mab: A Case Study in Bioprocess Development Study Guide
A-Mab: A Case Study in Bioprocess Development Study Guide. Related Files. A-Mab: A Case Study in Bioprocess Development Study Guide; Visit us on. 1111 Exposition Blvd, Building 100 Sacramento, CA 95815 P 510.428.0740 [email protected]. CASSS. About; Leadership; Staff; Mission and Vision; Meetings and Events. Upcoming Events;
Quality by Design A-MAb Case Study Challenges Conventional Thinking
Scale-up From 15K to 25K Considered Within the Design Space. The report also anticipated that the A-MAb bioreactor process would be scaled up further, to the 25,000-L scale. For the case study, it was assumed that the 25,000-L plant had an extensive and proven commercial manufacturing record of cGMP compliance and MAb production.
(PDF) A Mab Case Study Version
A Mab Case Study Version. Vivek Srivastava. See Full PDF Download PDF. See Full PDF Download PDF. Related Papers. Biotechnology and Bioengineering. Aggregates in monoclonal antibody manufacturing processes. 2011 • dietmar lang. Download Free PDF View PDF. Process Design for production of 2.4 kgannum of M72 Antigen from E. coli.
A-Mab Case Study Version 2-1
A-Mab Case Study Version 2-1 - Free ebook download as PDF File (.pdf), Text File (.txt) or view presentation slides online. l
PDF The Process Development of Therapeutic Monoclonal Antibody Products by
0-2% In vitro studies with A-Mab. High Mannose Literature data show afucosylated forms impact ADCC NA NA 3-10%; 3-10% Clinical Experience with A-Mab. Non-Glycosylated Heavy Chain Literature data show that non-glycosylated forms impact ADCC NA NA 0-3% 0-3% Clinical Experience with A-Mab. • A-Mab: a Case Study in Bioprocess Development, Version ...
Quality by Design risk assessments supporting approved antibody
The A-mAb case study provided another substantial contribution to the field. 7 It described a variety of approaches to the major elements of QbD used by 7 companies (Pfizer, GlaxoSmithKline, Genentech, Abbott, Amgen, Lilly, MedImmune) with experience in the development and commercialization of biologics. This publication presented a diverse set ...
ISPE Announces Availability of A-Mab Case Study at its Annual Meeting
The A-Mab case study discusses the development of a monoclonal antibody and incorporates many advanced and aspirational QbD concepts. "The CMC-BWG team has created an amazing and unique case study that is generating intense interest and excitement amongst the Industry and Regulatory Agencies around the world," said ISPE PQLI Project Manager ...
PDF Quality by Design for Biotech, Pharmaceutical and Medical Devices
Quality by Design for Biotech, Pharmaceutical and Medical Devices ...
A-Mab Case Study v5 (No Backup)
A-Mab Case Study v5 (No Backup) - Free download as PDF File (.pdf), Text File (.txt) or read online for free. The document summarizes a case study of a monoclonal antibody called A-Mab. It discusses: 1) How quality attributes were identified and assessed for criticality using risk assessment tools. Critical attributes included heterogeneity, aggregation, and DNA content.
PDF Biologics QbD case study: Characterizing a final sterile filtration step
The A-Mab case study published by CASSS & ISPE proposed that each quality attribute can be ranked on a continuum of criticality rather than a binary critical/non-critical classification. Once identified, CQAs can be monitored and controlled through the process to ensure a consistent product of the required quality. During the development of each
Micro scale self-interaction chromatography of proteins: A mAb case-study
Abstract. Self-interaction chromatography is known to be a fast, automated and promising experimental technique for determination of B22, but with the primary disadvantage of needing a significant amount of protein (>50 mg). This requirement compromises its usage as a technique for the early screening of new biotherapeutic candidates.
A Case Study-Based Approach to Integrating Qbd Principles in Gene
Where appropriate the authors have borrowed formatting and structure from the A-Mab case study. While we have attempted to be as comprehensive as possible, and have subjected the document to rigorous review, it is not a "recipe book" for AAV manufacture. Some aspects of process development (e.g., facility design), were deliberately omitted ...
N-mab: a Case Study Supporting Adoption of Integrated Continuous
The A-mAb case study (2009) was created to stimulate discussion around how the core principles of QbD could be applied to the process development of a monoclonal antibody with examples of a multitude of real-world scenarios, as opposed to a singular approach. A-mAb was very successful in generating that discussion and
PDF QbD for Biologics
collaborations: the A-MAb Case Study and the FDA OBP Pilot Program. The pilot program is still in its early stages but nonetheless provides concrete examples of the types of exchange of ideas between sponsors and regulators. The case study on applying QbD principles in development of a monoclonal antibody represents the culmination of a two-
PDF The Formidable Challenge of Controlling High Mannose-Type N-Glycans in
a-mab-case-study-in-bioprocess-development). Therefore, when developing new mAb manufacturing processes, it is desirable to avoid excessive levels of HM glycans in order to closer mimic the physiological properties of the product.
Safety risk management for low molecular weight process‐related
Subsequently, the CMC Biotech Working Group, consisting of industry experts, adopted this ISF approach in a white paper entitled "A Mab: A Case Study in Bioprocess Development" 6; and the PhRMA working group included the ISF approach in its advice on applying "quality by design for biotechnology products." 7 To measure safety risk of ...
N-mAb
The A-mAb case study (2009) created value in the biopharmaceutical industry by bringing industry and regulatory stakeholders together to develop shared expectations and vocabulary for quality by design. Participation included key thought leaders in industry and from relevant federal agencies. NIIMBL is committed to supporting and ...
A-Mab Case Study v5 (No Backup)
The A-Mab case study by the CMC Biotech Working groupDan Kenett. May 5 2010 Jerusalem. Background: Therapeutic Monoclonal Antibodies. Selected A-Mab chapters: OUTLINE. Design of Molecule and Quality Attributes Assessment. Upstream Manufacturing Process Development. Control Strategy. 2. Page 2. 2.
QbD Model Case Study of MONOCLONAL ANTIBODY : A-Mab.
The A-Mab Case Study involved the efforts of many individuals from CMC-BWG and would not have been made possible if it were not for the countless number of hours spent by the 5 participating companies (GlaxosmithKline, MedImmune, Merck, Pfizer, PWC and sanofi pasteur). Read more. 1 of 278. Download now. Download to read offline. The CMC Biotech.
N-mAb
The N-mAb case study brought together over 60 industry and government stakeholders from over 20 organizations to develop shared expectations and vocabulary around a control strategy for an integrated continuous bioprocess for a hypothetical monoclonal antibody. This document presents a starting place for conversations within organizations ...
The use of cetuximab for oral cancer and adverse events with different
The study comprised 53 patients diagnosed with oral cancer who underwent chemotherapy with C-mab between January 2013 and December 2022 at the Oral Cancer Center, Tokyo Dental College. In total, 68 regimens were analyzed. The cumulative rate was determined by analyzing the timing and incidence of adverse events (AEs) with each regimen using a ...
A rising layered Boride Family for Energy and Catalysis Applications
Finally, we discuss future challenges and prospects for the study of h-MAB and h-MBenes. Accepted Articles. Accepted, unedited articles published online and citable. The final edited and typeset version of record will appear in the future. e202400229. Related; Information; Close Figure Viewer. Return to Figure.