cancer research article types

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

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

Classifying Ten Types of Major Cancers Based on Reverse Phase Protein Array Profiles

Pei-wei zhang.

2 The Key Laboratory of Stem Cell Biology, Institute of Health Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, P.R. China

4 College of Information Engineering, Shanghai Maritime University, Shanghai, P.R. China

3 Department of Biomedical Engineering, Tianjin Key Lab of BME Measurement, Tianjin University, Tianjin, P.R. China

Xiang-Yin Kong

Yu-dong cai.

1 College of Life Science, Shanghai University, Shanghai, P.R. China

Conceived and designed the experiments: TH XYK YDC. Performed the experiments: PWZ TH. Analyzed the data: PWZ LC TH. Contributed reagents/materials/analysis tools: YDC. Wrote the paper: PWZ TH NZ LC.

Associated Data

All relevant data are within the paper and its Supporting Information files.

Gathering vast data sets of cancer genomes requires more efficient and autonomous procedures to classify cancer types and to discover a few essential genes to distinguish different cancers. Because protein expression is more stable than gene expression, we chose reverse phase protein array (RPPA) data, a powerful and robust antibody-based high-throughput approach for targeted proteomics, to perform our research. In this study, we proposed a computational framework to classify the patient samples into ten major cancer types based on the RPPA data using the SMO (Sequential minimal optimization) method. A careful feature selection procedure was employed to select 23 important proteins from the total of 187 proteins by mRMR (minimum Redundancy Maximum Relevance Feature Selection) and IFS (Incremental Feature Selection) on the training set. By using the 23 proteins, we successfully classified the ten cancer types with an MCC (Matthews Correlation Coefficient) of 0.904 on the training set, evaluated by 10-fold cross-validation, and an MCC of 0.936 on an independent test set. Further analysis of these 23 proteins was performed. Most of these proteins can present the hallmarks of cancer; Chk2, for example, plays an important role in the proliferation of cancer cells. Our analysis of these 23 proteins lends credence to the importance of these genes as indicators of cancer classification. We also believe our methods and findings may shed light on the discoveries of specific biomarkers of different types of cancers.

Introduction

Identifying cancer-specific genes involved in tumorigenesis and cancer progression is one of the major ways to understand the pathophysiologic mechanisms of cancers and to find therapeutic drug targets. Many efforts have been made to identify cancer biomarkers by using gene expression profiles [ 1 ]. However, the robustness of microarray-derived biomarkers is very poor [ 2 ]; this is in part because the robustness can be easily influenced in gene expression levels by small environmental changes. Without the evaluation of protein expression levels, there would be no way to illustrate causes of tumor proliferation and differentiation. Therefore, better understanding of the translational states of these genomes will bring us a step closer to finding potential drug targets and to illustrating off-target effects in cancer medicine.

Reverse phase protein array (RPPA) is a powerful and robust antibody-based high-throughput approach for targeted proteomics that allows us to quantitatively assess target protein expression in large sample sets [ 3 ]. In this process, sample analytes are immobilized in the solid phase, and analyte-specific antibodies are used in the solution phase. Through using secondary tagging and signal amplification to detect bound antibodies, proteins may be measured. Compared with conventional protein quantify methods, such as western blotting or ELISA, the advantages of RPPA include: large-scale quantification of the protein, high sensitivity, and small sample volume requirements [ 4 ]. While mass spectrometry, usually used to quantify the numbers of phosphorylation sites or phosphopeptides, requires further protein digestion, peptide fractionation and phosphopeptide enrichment after protein extraction, RPPA can directly quantify the extracted protein [ 5 ]. The application of RPPA has been extensively validated for both cell lines and patient samples [ 6 ], and it illustrates mechanistic insights behind diseases.

Currently, cancer types are classified by anatomical positions where they are found, such as lung cancer, breast cancer, etc. Whether these names could present their proteomic feature has not been determined until now. Although there have been some methods to find biomarker signatures for specific cancer types, there is still little research being done that considers different types of cancer as a whole in order to identify their similar or distinct proteomic expression patterns and classification features.

In this study, we proposed a computational workflow to successfully use 23 proteins to classify patient samples into ten main cancer types. First, we randomly divided the 3467 samples from ten types of cancers into a training set with 2775 samples and an independent test set with 692 samples. The proportions of each cancer type were similar in the training set and the independent test set. Then, with the training set, all features for distinguishing groups were ranked by the mRMR (minimum Redundancy Maximum Relevance Feature Selection) criteria. With 10-fold cross-validation on the training set, the SMO (Sequential minimal optimization) and the IFS (Incremental Feature Selection) [ 7 ] methods were used to choose an optimal feature set. A total of 23 proteins were selected from the training set. Their MCC (Matthews Correlation Coefficient) for the training set was 0.904 evaluated by 10-fold cross validation and their MCC on the independent test set was 0.936. Our methods could provide clinicians with knowledge of key distinct biochemical features of cancer types and could shed some new light on the discoveries of specific biomarkers of different types of cancers.

Materials and Methods

The RPPA data were downloaded from TCPA (The Cancer Proteome Atlas) database [ 8 ] ( http://app1.bioinformatics.mdanderson.org/tcpa/_design/basic/download.html under Pan-Cancer 11 RBN), which contained proteomic expression of 3467 cancer patients in 11 cancer types ( Table 1 ). Because COAD (Colon adenocarcinoma) and READ (Rectum adenocarcinoma) share similar pathologies and were analyzed together in the TCGA (The Cancer Genome Atlas) colon and rectal cancer study [ 9 ], we combined the COAD and READ samples together as 'Colon adenocarcinoma and Rectum adenocarcinoma' samples. Therefore, ten cancer types were analyzed in following steps.

Because we did not have a different cohort to do multi-center validation, we randomly divided the 3467 samples into a training set with 2775 samples and an independent test set with 692 samples. The ratio of training samples over test samples was approximately 4:1 and we kept the proportion of each cancer type roughly the same in the training set and the independent test set. The description of the ten cancer types and their sample sizes in are given in Table 1 . The training and test data sets are provided in S1 File .

Each sample contained 187 proteins whose expression levels were measured with reverse phase protein array (RPPA). RPPA is a protein array that allows measurement of protein expression levels in a large number of samples simultaneously in a quantitative manner when high-quality antibodies are available [ 4 ]. The 187 protein expression levels were considered as 187 features to be used for the cancer type classifications in this study.

Feature selection

The expression levels of 187 proteins may not all contribute equally to the classification. The maximum relevance minimum redundancy (mRMR) method [ 10 – 13 ] was employed to rank the importance of the 187 features in the training set. The 187 features can be ordered by using this method according to each feature’s relevance to the target and according to the redundancy among the features themselves.

Let Ω denotes the whole set of 187 features, while Ω s denotes the already-selected feature set which includes m features and Ω t denotes the to-be-selected feature set which includes n features. The relevance D of the feature f in Ω t with the cancer classes c can be calculated by:

And the redundancy R of the feature f in Ω t with the already-selected features in Ω s can be calculated by:

To obtain the feature f j in Ω t with maximum relevance with cancer classes c and minimum redundancy with the already-selected features Ω s , Equation ( 1 ) and Equation ( 2 ) are combined as the mRMR function:

The feature evaluation will continue 187 rounds. After these evaluations, a ranked feature list S by mRMR method can be obtained:

The feature index h indicates the importance of feature. A feature with a smaller index h indicated that it had a better trade-off between the maximum relevance and the minimum redundancy, and it may contribute more in the classification.

Based on the ranked feature list in the mRMR table, we adopted the Incremental Feature Selection (IFS) method [ 14 , 15 ] to determine the optimal feature set, or one that achieves the best classification performance. To perform this method, features in the mRMR table were added one by one from higher to lower rank.

When another feature had been added, a new feature set was generated. And we get 187 feature sets, and the i-th feature set is:

Based on each of the 187 feature sets, the classifiers were built and tested on the training set with 10-fold cross validation. With Matthews Correlation Coefficient (MCC) of 10-fold cross validation calculated on training set, we obtain an IFS table with the number of features and the performance of them. S optimal is the optimal feature set that achieves the highest MCC on training set. At last, the model was build with features from S optimal on training set and elevated on the test set.

Prediction methods

We randomly divided the whole data set into a training set and an independent test set. The training set was further partitioned into 10 equally sized partitions. The 10-fold cross-validation on the training set was applied to select the features and build the prediction model. The constructed prediction model was tested on the independent test set. The framework of model construction and evaluation was shown in Fig 1 .

First, we randomly divided the whole data set into a training set and an independent test set. Then, the training set was further partitioned into 10 equally sized partitions to perform 10-fold cross validation. Based on the training set, the features were selected and the prediction model was built. At last, the constructed prediction model was tested on the independent test set.

We tried the following four machine learning algorithms: SMO (Sequential minimal optimization), IB1 (Nearest Neighbor Algorithm), Dagging, RandomForest (Random Forest), and selected the optimal one to construct the classifier. The brief description of these algorithms was as below.

The SMO method is one of the popular algorithms for training support vector machines (SVM) [ 16 ]. It breaks the optimization problem of a SVM into a series of the smallest possible sub-problems, which are then solved analytically [ 16 ]. To tackle multi-class problems, pairwise coupling [ 17 ] is applied to build the multi-class classifier.

IB1 is a nearest neighbor classifier, in which the normalized Euclidean distance is used to measure the distance of two samples. For a query test sample, the class of a training sample with minimum distance is assigned to the test sample as the predicted result. For more information, please refer to Aha and Kibler’s study [ 18 ].

Dagging is a meta classifier that combines multiple models derived from a single learning algorithm using disjoint samples from the training dataset and integrates the results of these models by majority voting [ 19 ]. Suppose there is a training dataset ℑ containing n samples. k subsets are constructed by randomly taking samples in ℑ without replacement such that each of them contain n ′ samples, where kn ′ ≤ n . A selected basic learning algorithm is trained on these k subsets, thereby inducing k classification models M 1 , M 2 ,…, M k . For a query sample, M i (1≤ i ≤ k ) provides a predict result and the final predicted result of Dagging is the class with most votes.

Random Forest algorithm was first proposed by Loe Breiman [ 20 ]. It is an ensemble predictor consisting of multiply decision trees. Suppose there are n samples in the training set and each sample was represented by M features. Each tree is constructed by randomly selecting N , with replacement, from the training set. At each node, randomly select m features and select the optimized split to grow the tree. After constructing multiply decision trees, the predicted result of a given sample is the class that receives the most votes from these trees.

Matthews Correlation Coefficient (MCC)

MCC [ 21 ], a balanced measure even if the classes are of very different sizes, is often used to evaluate the performance of prediction methods on a two-class classification problem. To calculate the MCC, one must count four values: true positives (TP), false positive (FP), true negative (TN) and false negative (FN) [ 22 , 23 ]. Then, the MCC can be computed by

However, many problems involve more than two classes, say N classes encoded by 1,2,…, N ( N > 2). In this case, we can calculate the MCC for class i to partly measure the performance of prediction methods by counting TP, FP, TN and FN as following manners:

TP i : the number of samples such that class i is their predicted class and true class;

FP i : the number of samples such that class i is their predicted class and class i is not their true class;

TN i : the number of samples such that class i is neither their predicted class nor their true class;

FN i : the number of samples such that class i is not their predicted class and class i is their true class.

Accordingly, MCC for class i , denoted by MCC i , can be computed by

However, these values can’t completely measure the performance of prediction methods, the overall MCC in multiclass case is still necessary. Fortunately, Gorodkin [ 24 ] has reported the MCC in multiclass case, which was used to evaluate the performance of the prediction methods mentioned in Section “Prediction methods”. In parallel, The MCC for each class will also be given as references. Here, we gave the brief description of the overall MCC in multiclass case as below.

Suppose there is a classification problem on n samples, say s 1 , s 2 ,…, s n , and N classes encoded by 1,2,…, N . Define a matrix Y with n rows and N columns, where Y ij = 1 if the i -th sample belongs to class j and Y ij = 0 otherwise. For a classification model, its predicted results on the problem can be represented by two matrices X and C , where X has n rows and N columns,

and C has N rows and N columns, C ij is the number of samples in class i that have been predicted to be class j .

For Matrices X and Y , their covariance function can be calculated by

where X k and Y k are the k -th column of matrices X and Y , respectively, X ¯ k and Y ¯ k are mean value of numbers in X k and Y k , respectively. Then, the MCC in multiclass case can be computed by the following formulation [ 25 ]:

Like the MCC in two-class case, the MCC in multiclass case ranges between -1 and 1, where 1 indicates the perfect classification, -1 the extreme misclassification.

Results and Discussion

The mrmr and ifs results.

By using the maximum relevance minimum redundancy (mRMR) method, the 187 features were ranked by importance in the training set. The result of the mRMR table can be found in S2 File .

During the IFS approach, each protein feature was added one by one. The classification MCCs which were obtained by four prediction methods, on the training set evaluated by 10-fold cross validation are presented in S3 File . We depicted the classification MCCs as Fig 2 from the data in S3 File . It can be observed that the highest MCCs for SMO, IB1, Dagging and RandomForest were 0.985, 0.937, 0.969 and 0.925, indicating SMO can be used to construct an optimal classifier. By carefully checking the predicted results of SMO, it can be seen that by using the top 23 proteins, the MCC reached 0.904 which was the first reach above 0.900. With more proteins, the MCC did not increase by much. Therefore, in this study, we considered the 23 proteins as the optimal feature set and these 23 proteins were regarded as the most important proteins in classifying these ten types of cancers. We evaluated their prediction performance on the independent test set and the MCC was 0.936. The MCC for each cancer type can be found in S3 File .

Plot to show the MCCs of the different classifiers constructed by different number of protein features selected from the mRMR table during the IFS process on training set. When the first 23 proteins were selected, the MCC reached 0.904, which was the first reach above 0.900 and with more protein features, the MCC did not increase much. We considered the 23 proteins as the most significant proteins for the classification.

The selected top 23 proteins for distinguishing cancer types

The selected top 23 proteins are summarized in Table 2 . These proteins may play important roles in classifying the ten different cancer types. Most of these proteins have been reported to be related to certain tumors. For example, Claudin-7 has been reported to be over-expressed in breast tumors [ 26 ] and down-regulated in head and neck carcinomas [ 27 ]. TIGAR is up-regulated in colon tumors [ 28 ]. Gene amplification of ESR1 occurs frequently with breast cancer [ 29 ]. PREX1 is highly expressed in prostate cancer [ 30 ]. Thus, our findings are further corroborated by these previous results. Below, we will discuss the biological significance of the 23 proteins in detail based on gene function, cell pathways and biological functions, which may shed some light on the differences of different cancers in protein expression levels. We mainly discuss these genes in sections according to Robert A. Weinburg’s [ 31 ]. For some genes that do not apply to cancer’s hallmarks, we try to put these genes with similar functions together for discussion (see Fig 3 ).

The 23 selected proteins are ascribed to seven sections mainly based on hallmarks of cancer. For those that are not associated with cancer-related pathways, we put genes with similar functions together to discuss.

Preventing cell death is crucial for cancer development because cancer cells are often resistant to apoptotic signaling caused by DNA damage and other factors. In our results, we found one gene that is related to apoptotic machinery and could be used to distinguish different cancers. Here, we discuss NDRG1, as well as previous findings showing its relationship to cancer. NDRG1 (N-myc downstream regulated gene 1) is a phosphorylated protein [ 32 ] that could be activated by the tumor suppressor gene p53 and required for the induction of p53-mediated apoptosis in the colon cancer cell line [ 33 ]. Because the NDRG1 protein has a crucial role in inhibiting primary tumor growth, it is well-known as a metastasis suppressor in a number of cancers including colon, prostate and breast cancers [ 34 ].

Replicative immortality is an important hallmark of cancer, which is commonly recognized as deregulated cell proliferation. Our findings on several important cell cycle-related genes in selected proteins not only illustrate their importance to the development of cancer, but are also first used as indicators of cancer classification. These cell cycle-related genes are discussed below: Cyclin B1 has a role in the regulation of cell cycle: before entering mitosis, cells flip between G2 and mitosis until there is sufficient accumulation of cyclin B to support CDK1 activity [ 35 ]. Misexpressed cyclin B1 in the nucleus has been found in a huge proportion of cells of some neoplasms, and cyclin B1 has been regarded as a potent prognostic factor in human breast carcinoma and squamous cell carcinoma [ 36 ]. Cyclin E1, encoded by CCNE1, is one of the members of the cyclin family, which controls cell cycle processes by dramatic periodicity of abundance. Recently, a genome-wide association study found that rs8102137 within the CCNE1 gene is associated with bladder cancer [ 37 ]. Meta-analysis also indicates that there is over-expression of this protein with breast cancer [ 38 ]. Chk2 (checkpoint kinase 2), as a serine/ threonine protein kinase, could respond to DNA damage in order to maintain genomic integrity [ 39 ]. It has been shown that Chk2 plays an important role in the proliferation of cancer cells [ 40 ], attracting much attention to make it a possible anti-cancer drug design target [ 41 ].

It is clear that invasion is a hallmark of cancer, even if its underlying mechanisms are still an enigma. Until now, the gain and loss of cell-cell attachment proteins are the main reasons of invasion, especially the loss of E-cadherin [ 31 ]. In our results, E-cadherin and some polarity-related proteins are found that could be used to distinguish different cancer types. These proteins are discussed below: E-Cadherin, as the type-1 classical cadherin, mediates cell interactions. Tumor progression is often linked with the loss of E-cadherin function, leading to a more motile and invasive phenotype [ 42 ]. PREX1 (phosphatidylinositol-3,4,5-trisphosphate-dependent Rac exchange factor) is highly expressed in prostate cancer, indicating a relationship between the cell invasion and its expression [ 30 ]. In melanomas, PREX1 over-expression was connected to the activation of ERK-MAPK signaling and required for efficient melanoblast metastasis as well as for migration [ 43 ]. Claudin-7, a common transmembrane protein, plays a vital role in the formation and maintenance of the permeability in polarized epithelial cells [ 44 ]. The aberrant Claudin-7 expression profile has been found in various tumors, such as highly induced Claudin-7 expression in both primary and metastatic breast tumors, [ 26 ] yet it is down-regulated in head and neck carcinomas [ 27 ]. These previous studies further supported our findings that Claudin-7 could be used as a biomarker for the differentiation and classification of various tumors. Rab-25, as a member of the Rab family of GTPases, Rab-25 is a constitutively active Rab GTPase that plays a crucial role in apical recycling and transcytosis pathways in polarized epithelial cells. Because loss of cell polarity is an essential hallmark of cancer, Rab-25 related trafficking has an important impact on epithelial cell polarity program in cancer progression [ 45 ].

Anomalous cancer cell energy metabolism was first observed by Otto Warbugy in 1930 and has been accepted as a hallmark of cancer. Abnormal fatty-acid synthesis as one type of energy metabolism is found in many cancer cells [ 46 ]. Here, several important fatty acid and glycolytic metabolism-related genes are found in the selected 23 proteins: FASN is a key enzyme which is required for de novo synthesis of fatty acid. It has been found that the FASN expression and activity are abnormally elevated in many types of human cancers, which may contribute to cellular resistance to drug- and radiation-induced apoptosis [ 46 ]. ACC1 is a rate-limiting enzyme in de novo fatty acids synthesis. It seems to be the limiting enzyme in proliferating cancer cells. ACC1 has been found to be up-regulated in proliferating cancer cell lines such as prostate, breast and liver. Indeed, it has been shown that knock-down of ACC1 by siRNA promotes apoptosis in prostate cancer and breast tumor cells but not in control noncancerous cells, underlining cancer cells' higher reliance on this enzyme than normal tissue [ 47 ]. AMPK (AMP-activated protein kinase, encoded by the gene PRKAA1/2) plays a crucial role in sensing available energy and coordinating external growth signals with cellular metabolism [ 48 ]. A decrease of AMPK signaling, mostly caused by the loss of function gene STK11, could lead to increased activation of mTOR and a shift toward glycolytic metabolism, which is found in a variety of cancers, including NSCLC [ 49 ] and cervical cancer [ 50 ].

Abnormal expression of hormone receptors are often shown in sex-related cancers, such as breast cancer and prostate cancer. Three hormone receptors are also reported in the selected proteins: Progestin receptor (PR), as a nuclear steroid receptor, has a high specificity for binding progesterone [ 51 ]. It has been shown in literature that PR inhibits the transition from G1 to S in the cell cycle and promote apoptosis in endometrial cancer cells [ 52 ]. In the GOG119 phase II trial, an estrogen surrogate named tamoxifen could enhance progestin activity in order to induce PR and cure endometrial patients [ 53 ]. Estrogen receptor (ER, activated by the hormone estrogen) is one of the most important therapeutic targets in breast cancers, given that the correlation between ER expression and cellular response to estrogen [ 54 ]. It has been reported that gene amplification of ESR1 frequently occur with breast cancer [ 29 ]. Androgen receptor (AR; NR3C4) is believed to solely mediate all the biological actions of endogenous, functioning mainly in regulating male development. Due to the strong connection between ARs and prostate cancer, androgen antagonists or androgen deprivation therapy has been applied to impede cancer cell proliferation of patients with androgen-dependent prostate cancer in clinical treatment [ 55 ].

Surprisingly, among these 23 selected proteins that are used to distinguish different cancers, α-tubulin and GAPDH are often used as controls in western blot analysis. In the following part, we will discuss known findings about α-tubulin and GAPDH that lend credence to the validity of our findings for their importance to distinguish cancers. For example, both α- and β- tubulin proteins are responsible for assembling microtubules (MTs, cytoskeletal polymeric structures), and certain posttranslational modifications. The acetylation of α-tubulin (Lys-40) [ 56 ] could alter dynamic behavior of MTs, which may lead to changes in biological functions that MTs perform during cell division, migration, and intracellular trafficking. Taking the dynamic parameters into account, MTs provide an attractive target for chemotherapy against rapidly growing tumor cells such as in lymphoma and leukemia, metastatic cancers, and slow growing tumors of the breast, ovary, and lung [ 57 , 58 ]. Over the last decade, GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was considered a housekeeping gene and was as a control for equal loading during the experimental process. However, it has been shown that GAPDH expression varies different types of tissues. Moreover，GAPDH expression varies due to oxygen tension [ 59 ], and the expression levels of GAPDH vary in fallopian tube cancers and ovarian cancers [ 60 ]. On the basis of GAPDH’s predilection for AU-rich elements, it has been shown that GAPDH can bind to the CSF-1 3'UTR that stabilize the mRNA [ 60 ]. To summarize, combining all the evidence, tubulin proteins and GAPDH may bring a new perspective on cancer studies, and it is suggested that they are not used as controls in western blot analysis of different types of cancer.

Other selected proteins include phosphatases, transcriptional activators, linker proteins and transferrin receptors: GATA3 is a transcriptional activator with high expression levels [ 61 ] and the third most frequently mutated gene in breast cancer [ 62 ]. Thus, GATA3 has proved to be a useful immunohistochemical marker to predict tumor recurrence early in the progression of breast cancer. PEA15, as a multifunctional linker protein predominantly expressed in the cells of the nervous system, such as astrocytes [ 63 ], controls a variety of cellular processes, such as cell survival, proliferation, migration and adhesion [ 64 ]. PEA15 functions in various cancers, concluding glioblastoma, astrocytoma, and mammary, as well as skin cancers. PEA15 can have both anti- (in ovarian carcinoma [ 65 ]) and pro- (glioblastoma [ 66 ]) tumorigenic functions, depending on its interactions. TFRC is a transferrin receptor. It is a major iron importer in most mammalian cells. It has been shown that TFRC proteins increase in breast, malignant pancreatic cancer, and other cancers [ 67 , 68 ]. PKCα is encoded by PRKCA gene and is a serine- and threonine- specific kinase. This gene is highly expressed in multiple cancers, and the high activation of PKCα has been identified to promote the genesis of breast cancer [ 69 ]. The high abundance in serum makes this protein to be a good diagnostic biomarker of lung cancer [ 70 ] and gastric carcinoma [ 71 ]. TIGAR is a fructose-2-6-bisphosphatase that promotes the production of antioxidant (NADPH) and nucleotide synthesis material (ribose-5-phosphate) and seems to be important for tissue renewal and intestinal tumorigenesis. Up-regulated expression of TIGAR in human colon tumors along with other evidence suggest its importance in the development of cancer and metabolism regulation and may be used as a therapeutic target in diseases such as intestinal cancer [ 28 ]. CD20 (Membrane-Spanning 4-Domains, Subfamily A, Member 1, MS4A1) encodes a surface molecule B-lymphocyte during the differentiation of B-cells into plasma cells. Currently, a CD20 monoclonal antibody has been utilized in the treatment of cancer, even though its dosage is still under discussion [ 72 ]. GAB2 (GRB2-associated-binding protein 2) is a docking protein, which mainly interacts with signaling molecules. Research has shown that the oncogenesis of many cancers including gastric, colon, ovarian and breast cancer is related to GAB2 [ 73 , 74 ]. For example, GAB2 can amplify the signal of receptor tyrosine kinases (RTKs), which plays roles in breast cancer development and progression [ 75 ].

As shown above, all of the top 23 proteins are closely related to certain types of cancers. Researchers have focused on common features of different cancer types for decades [ 31 ]. Admittedly, in theory, the hallmarks of cancer would help us develop drugs to treat all types of cancers as a whole. However, this “one size fits all” cancer treatment has disappointed us due to its treatment-related toxicity and inefficiency. Despite the fact that personalized treatments have been proposed, the theory still stays at a conceptual phase. Thus, having a better understanding of the potential values and the applied ranges of cancer drugs based on different biomarkers may be a more realistic way to treat different types of cancers.

Potential values of our findings

Previous experimental studies in the literature could consolidate our results showing that the selected 23 proteins could be used as biomarkers for certain cancers. They also can explain partially why the combination of these proteins could be used to accurately classify different cancer types. However, to our knowledge, reasons behind the varying expression patterns in different types of cancers have not been found. At least, by using our computational method, one could gain a better understanding of the similarities and differences among different cancers. This could help us identify proteins that could promote the development of cancers and proteins that might not be indispensable for cancer development. Further studies should be performed to determine whether the differential expression patterns of proteins in various cancers are influenced by their original tissues. Those proteins specifically expressed in certain types of cancers could be considered as potential specific cancer targets, which could be used to improve the target efficiency. Therefore, our results may help drug designers obtain a better understanding of the potential targets of drugs by shedding some light on the cancer type-specific biomarker discoveries.

Supporting Information

There were 3467 cancer patient samples in 10 cancer types, with 187 proteins for each sample. The 3467 samples were randomly divided into 2775 training samples and 692 independent test samples. The first column is the sample ID, the second column is the cancer types whose description can be found in Table 1 . The third to the 189th columns were proteins.

All the 187 protein features were ranked from the most important to the least by using the mRMR method on training set. The top 23 proteins were regarded as composing the optimal feature set because by using the 23 protein features, the MCC on the training set evaluated by 10-fold cross validation reached 0.904 which was the first reach above 0.900, and with more protein features, the MCC did not increase much.

Funding Statement

This work was supported by grants from National Basic Research Program of China (2011CB510102, 2011CB510101), and National Natural Science Foundation of China (31371335, 81171342, 81201148, 61401302), Innovation Program of Shanghai Municipal Education Commission (12ZZ087), the grant of “The First-class Discipline of Universities in Shanghai”, Tianjin Research Program of Application Foundation and Advanced Technology(14JCQNJC09500), the National Research Foundation for the Doctoral Program of Higher Education of China (20130032120070, 20120032120073) and the Independent Innovation Foundation of Tianjin University (60302064, 60302069). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability

• History, Facts & Figures
• YSM Dean & Deputy Deans
• YSM Administration
• Department Chairs
• YSM Executive Group
• YSM Board of Permanent Officers
• FAC Documents
• Current FAC Members
• Appointments & Promotions Committees
• Ad Hoc Committees and Working Groups
• Chair Searches
• Leadership Searches
• Organization Charts
• Faculty Demographic Data
• Professionalism Reporting Data
• 2022 Diversity Engagement Survey
• State of the School Archive
• Faculty Climate Survey: YSM Results
• Strategic Planning
• Mission Statement & Process
• Beyond Sterling Hall
• COVID-19 Series Workshops
• Previous Workshops
• Departments & Centers
• Find People
• Biomedical Data Science
• Health Equity
• Inflammation
• Neuroscience
• Global Health
• Diabetes and Metabolism
• Policies & Procedures
• Media Relations
• A to Z YSM Lab Websites
• A-Z Faculty List
• A-Z Staff List
• A to Z Abbreviations
• Dept. Diversity Vice Chairs & Champions
• Dean’s Advisory Council on Lesbian, Gay, Bisexual, Transgender, Queer and Intersex Affairs Website
• Minority Organization for Retention and Expansion Website
• Office for Women in Medicine and Science
• Committee on the Status of Women in Medicine Website
• Director of Scientist Diversity and Inclusion
• Diversity Supplements
• Frequently Asked Questions
• Recruitment
• By Department & Program
• News & Events
• Executive Committee
• Aperture: Women in Medicine
• Self-Reflection
• Portraits of Strength
• Mindful: Mental Health Through Art
• Event Photo Galleries
• Additional Support
• MD-PhD Program
• PA Online Program
• Joint MD Programs
• How to Apply
• Advanced Health Sciences Research
• Clinical Informatics & Data Science
• Clinical Investigation
• Medical Education
• Visiting Student Programs
• Special Programs & Student Opportunities
• Residency & Fellowship Programs
• Center for Med Ed
• Organizational Chart
• Leadership & Staff
• Committee Procedural Info (Login Required)
• Faculty Affairs Department Teams
• Recent Appointments & Promotions
• Academic Clinician Track
• Clinician Educator-Scholar Track
• Clinican-Scientist Track
• Investigator Track
• Traditional Track
• Research Ranks
• Instructor/Lecturer
• Social Work Ranks
• Voluntary Ranks
• Adjunct Ranks
• Other Appt Types
• Appointments
• Reappointments
• Transfer of Track
• Term Extensions
• Timeline for A&P Processes
• Interfolio Faculty Search
• Interfolio A&P Processes
• Yale CV Part 1 (CV1)
• Yale CV Part 2 (CV2)
• Samples of Scholarship
• Teaching Evaluations
• Letters of Evaluation
• Dept A&P Narrative
• A&P Voting
• Faculty Affairs Staff Pages
• OAPD Faculty Workshops
• Leadership & Development Seminars
• List of Faculty Mentors
• Incoming Faculty Orientation
• Faculty Onboarding
• Past YSM Award Recipients
• Past PA Award Recipients
• Past YM Award Recipients
• International Award Recipients
• Nominations Calendar
• OAPD Newsletter
• Fostering a Shared Vision of Professionalism
• Academic Integrity
• Addressing Professionalism Concerns
• Consultation Support for Chairs & Section Chiefs
• Policies & Codes of Conduct
• First Fridays
• Fund for Physician-Scientist Mentorship
• Grant Library
• Grant Writing Course
• Mock Study Section
• Research Paper Writing
• Establishing a Thriving Research Program
• Funding Opportunities
• Join Our Voluntary Faculty
• Child Mental Health: Fostering Wellness in Children
• Faculty Resources
• Research by Keyword
• Research by Department
• Research by Global Location
• Translational Research
• Research Cores & Services
• Program for the Promotion of Interdisciplinary Team Science (POINTS)
• CEnR Steering Committee
• Experiential Learning Subcommittee
• Goals & Objectives
• Issues List
• Print Magazine PDFs
• Print Newsletter PDFs
• YSM Events Newsletter
• Social Media
• Patient Care

INFORMATION FOR

• Residents & Fellows
• Researchers

Investigating the Link Between Metabolism and Cancer

Rachel perry, phd.

When Rachel Perry, PhD , assistant professor of medicine (endocrinology) and of cellular and molecular physiology, started her lab in 2018, she set out to understand how the tools used to study systemic metabolism could be applied in the context of cancer.

“We’re looking at how can we change the nutrients that we take in and the energy that we put out,” Perry explained. “For instance, we want to find out how can we alter exercise or other behavior to improve our body's metabolic or nutritional response to cancer.”

In an interview, Perry discusses the link between insulin and cancer, a surprising finding in her research, and the future of precision medicine for metabolism-related cancers.

What metabolic factors drive tumor growth?

Cancer cells divide quickly, and to grow, tumors require many nutrients. One of the major nutrients for tumors is glucose. Research has shown that insulin, a hormone produced by the pancreas, is key to regulating glucose uptake into tumors. Insulin tells various cells in the body to take up glucose in response to a meal to use for energy.

Individuals with obesity tend to be insulin resistant, meaning their cells don’t take up glucose in response to insulin. When that occurs, our bodies produce more insulin to allow us to overcome that insulin resistance. Insulin is both a growth factor and a metabolic factor in cells. Insulin tells tumor cells to take up glucose like it tells liver and muscle cells to take up glucose.

The problem is tumor cells do not become insulin resistant, whereas other cells in the body do. In an individual with obesity and insulin resistance, tumor cells continue to take up more glucose in response to insulin, but the other cells in the body do not. When insulin levels are high, and other cells aren't responding, more glucose is funneled into tumor cells.

Are there ways to reduce the risk of developing this type of cancer?

Numerous studies have shown that low-carbohydrate diets and exercise can be beneficial.

Have you found anything surprising in your research?

We did a study in preclinical models in which we combined a drug called dichloroacetate, which can activate glucose oxidation, with immunotherapy. We thought that if you activate glucose metabolism, it could improve the response to immunotherapy. The drug didn't slow tumor growth, but it did reduce cancer-related fatigue.

We weren’t expecting to go in the direction of cancer-related fatigue, but my philosophy in the lab is we follow the science. Ninety percent of patients with cancer report substantial fatigue that impairs their quality of life and reduces the probability that they will complete cancer treatment. Currently, there are no effective medications for this debilitating condition. We hope that this drug can potentially help these patients.

What is the future of treatment for metabolism-related cancers?

There are 13 tumor types that the CDC has associated with obesity. But there are also people with a BMI under 30 who have metabolic dysfunction, and there are people with a BMI over 30 who do not have metabolic dysfunction. One key thing that we need to do is nuance our understanding of what obesity means and link tumor factors to metabolic dysfunction rather than BMI.

We are working to learn more about how systemic metabolism affects the immune response to cancer and the tumor's metabolic response to cancer. We hope to repurpose metabolic drugs used for diabetes or obesity to potentially improve cancer treatment.

We showed in a paper published a couple of years ago that in various breast cancer preclinical models, depending on the tumor genetics, different models respond better to insulin-lowering therapies. That research opens the possibility of developing precision medicine-type approaches to use metabolic therapy for cancer. That’s our goal moving on.

Y ale School of Medicine’s Section of Endocrinology and Metabolism works to improve the health of individuals with endocrine and metabolic diseases by advancing scientific knowledge, applying new information to patient care, and training the next generation of physicians and scientists to become leaders in the field. To learn more, visit Endocrinology and Metabolism .

• Internal Medicine
• Breast Cancer
• Endocrinology

Featured in this article

• Rachel Jamison Perry, PhD Assistant Professor of Medicine (Endocrinology) and of Cellular and Molecular Physiology

• Systematic Review
• Open access
• Published: 20 May 2024

Bioelectrical impedance analysis—derived phase angle (PhA) in lung cancer patients: a systematic review

• Melania Prete 1 ,
• Giada Ballarin 2 ,
• Giuseppe Porciello 3 ,
• Aniello Arianna 4 ,
• Assunta Luongo 3 ,
• Valentina Belli 5 ,
• Luca Scalfi 4 &
• Egidio Celentano 3  

BMC Cancer volume  24 , Article number:  608 ( 2024 ) Cite this article

244 Accesses

1 Altmetric

Metrics details

Lung cancer is the second most diagnosed cancer in the world. Up to 84% of diagnosed patients have malnutrition, which can negatively affect quality of life and survival and may worsen with neoadjuvant treatment. Bioelectrical Impedance Analysis-Derived Phase Angle (PhA) in these patients could be a valid tool to assess the nutritional status in order to improve their condition.

This review provides an update on PhA assessment in lung cancer patients over the past twenty years. We searched PubMed, Embase, Scopus, Web of Science, and Cochrane, for articles regarding the PhA obtained from Bioelectrical Impedance Analysis in lung cancer patients. The authors independently performed a literature search: sample size, patient population, study type, study dates, survival and interventions were evaluated. The final review included 11 studies from different countries.

Eight studies only considered patients with lung cancer, while three studies considered patients with different kind of cancer, including lung. Correlation data between PhA and age are conflicting. In patients undergoing clinical treatment and patients undergoing surgical treatment lower PhA was observed. A lower PhA is associated with a shorter survival. In three studies emerged a relationship between Karnofski Performance Status and Handgrip Strenght with PhA. From one study, univariate logistic regression analysis showed that higher PhA values represent a protective factor for sarcopenia.

Our research underlined interesting, but not conclusive, results on this topic; however more researches are needed to understand the clinical meaning of PhA.

Peer Review reports

Introduction

Lung Cancer (LC) is the second most diagnosed cancer worldwide, especially in males. Most recent data have shown an incidence of 2.2 million of new cases (11.4%) and 1,8 million of deaths (18.0%) occurred in 2020. It represents leading cause of cancer death in 93 countries [ 1 ]. Following diagnosis, 5-year survival rates ranges from 10 to 20% in most countries, with higher rates in Japan (33%), Israel (27%), and Korea (25%) [ 1 ]. In Italy, LC showed a 5-year survival of 16% in men and 23% in women [ 2 ]. LC aetiology is multifactorial and complex. In addition to a family history of LC, tobacco smoke currently represents the leading risk factor [ 2 , 3 ]. Second-hand tobacco use may also increase LC risk, causing more than 3.000 deaths each year [ 4 ]. Other lung carcinogens include inhaled chemicals such as arsenic, cadmium and asbestos [ 5 ].

LCs are traditionally classified in small cell lung carcinoma (SCLC) and non-small cell lung carcinoma (NSCLC) divided into four major classes (adenocarcinoma, squamous cell carcinoma and large cell carcinoma) [ 6 ]. Conventional LC therapies include surgical intervention for resectable diseases and, in selected cases, a combination of radiotherapy (RT) and chemotherapy (CHT) for locally advanced or metastatic disease. Advancements in the understanding of LC molecular pathogenesis has led to the development of targeted strategies like immune checkpoint inhibitor (ICI) in first and later lines of treatment [ 7 ].

In addition to cancer-related symptoms, including chronic cough, dyspnoea, pain and adverse effects from anti-neoplastic treatments, LC patients’ may experience fatigue, weight loss or nutritional status alterations, such as malnutrition [ 8 , 9 ]. Cancer patients are more likely to become malnourished, with a prevalence ranging from 20.0% to 80.0% [ 10 ]. Recent studies indicate that dietary nutrient deficiency in cancer patients may induce unintentional body weight loss to sarcopenia, up to cachexia [ 11 ]. It is known that malnutrition was prevalent in advanced LC patients [ 12 ]: up to 84% of them showed malnutrition status during illness, which can be worsened by ongoing neoadjuvant treatment [ 12 , 13 , 14 , 15 , 16 , 17 ]. This condition has been associated with poorer prognosis, decreased treatment response, poorer tolerance to treatment, lower quality of life (QoL) and increased healthcare costs [ 12 , 18 ]. Additionally, sarcopenia, defined as progressive loss of muscle mass and functioning, is highly prevalent among LC patients ranging from 42.8% to 45.0%, in association with increased postoperative complications and increased risk of mortality, regardless of cancer stage and treatment [ 19 ]. Furthermore, LC is more commonly linked to cancer cachexia than other types of cancer [ 16 ], characterized mainly by a decrease in muscle strength, due to the loss of adipose tissue and skeletal mass [ 20 , 21 ].

Body Composition (BC) is a crucial requirement for the overall body assessment of cancer patients: it can reflect the nutritional status of patients and predict clinical outcomes and prognosis [ 22 ]. Bioelectrical Impedance Analysis (BIA) is a simple, cost-effective and non-invasive method that measures electrical characteristics of human body, i.e. impedance (Z), through application of four electrodes and an applied alternate current, using single (SF-BIA) or multiple (MF-BIA) current frequencies. Z derives from resistive component (Resistance, R) and capacitive component (Reactance, X c ), by equation Z = R 2  + Xc 2 ​. R indicates how much a substance opposes the flow of electric current: greater resistance indicates greater difficulty of passage. It can be affected by factors such as tissue density or hydration and cell membrane permeability. Reactance reflects the ease with which electricity can flow through tissues: high reactance means that there is more resistance of tissues and less conductivity. It can therefore indicate cellular and membrane integrity.

BIA-derived Phase Angle (PhA) is obtained as [arctangent (X c /R) × 180°/π]. PhA represent an indicator of cellular health, cell membrane integrity, and better cell function: low values are indicators of apoptosis and cell matrix alteration [ 23 , 24 ]. Therefore, BIA analysis and the use of the PhA have a good consistency in the application in cancer patients to evaluate nutritional and hydration status [ 25 , 26 ]. Despite importance of nutritional status and BC for the clinical evaluation of cancer patients, these conditions remain in part an undertreated issue [ 27 ]. Thus, nutritional status assessment of these patients is essential for adequate nutritional support, and may also improve QoL and consequently survival post-diagnosis [ 28 , 29 ]. In the literature, a large number of publications on BIA and in particular of PhA assessment, confirm its prognostic role in different types of cancer (e.g., breast, pancreatic and colon) [ 30 , 31 , 32 , 33 ]. Although PhA is a useful prognosis tool even in patients with LC [ 34 ], this has not been discussed in detail in scientific literature. So, our review aims to critically report and discuss available clinical data relating to PhA in LC patients, to provide a broad and clear picture of topic.

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed for performing the present review, considering the possibility of including both cross-sectional and longitudinal studies. Further details about PRISMA checklist and study protocol were provided in Supplementary File. Due to the study type, ethical approval was not required. This systematic revision is not currently registered in any database.

Data sources

Authors independently performed a systematic literature research between 2000 and 2023 of the electronic databases PubMed, Embase, Scopus, Web of Science and Cochrane. The following terms were used as search strategy string on full texts: phase angle" AND (lung or pulmonary) AND (bioelectrical OR impedance OR bia) AND cancer”, "phase angle" AND (bia OR bioelectrical OR bioimpedance) AND cancer”. Zotero and EndNote X7 citation management software were used to manage citations.

Criteria for analysis

In this review, cross-sectional, case–control, and longitudinal studies were included. The measurement of PhA values by BIA was a necessary and indispensable condition for an article to be included. The studies had to be present in the literature in English form and had to be published no earlier than 2000 to include the most recent evidence. Outcome of interest included associations between PhA and survival, mortality, or other variables related to LC patients. Furthermore, it was necessary that studies must have been conducted in the health field. Articles that did not meet these requirements were excluded from the review. All studies include a phase-sensitive BC measurement tool. In all selected studies PhA is calculated by the ratio of R to X c equal to [arctangent (X c /R) × 180°/π. BC parameters were not obtained by predictive regression models.

Data extraction and analysis

To assess the suitability of the articles obtained from the literature search, authors carefully and meticulously examined all the titles and abstracts. Subsequently, the authors independently extracted the data from the papers and reported in an excel file. The data included: first author, year of publication, country of origin, design of study, sample size, age, sex, presence of control groups, type of tumour, methods used (BIA, BIVA), focusing on PhA.

Selected studies

One hundred and sixty-five papers were identified from the systematic search, 92 from Pubmed, 32 from Embase, 6 from Scopus and 38 from Web of Science. After removing duplicates ( n  = 38), 120 articles were excluded, including 2 reviews, 1 symposium and 117 that did not concern LC. A total of 11 full-text articles were selected for eligibility, as shown in Fig.  1 .

General characteristics of included studies

This systematic review includes several types of studies: five prospective studies [ 35 , 36 , 37 , 38 , 39 ], three observational studies [ 40 , 41 , 42 ], two retrospective studies [ 34 , 43 ] and one cross-sectional study [ 44 ]. Shi et al. showed the highest number of patients involved (804). The remaining selected scientific papers included 30 to 204 patients. Five studies included both males and females, six studies included males only [ 36 , 38 , 40 , 41 , 42 , 44 ]. However, in two studies [ 36 , 37 ], we do not know if the percentages of women and men present in the study referred to LC patients or patients with other kind of cancers. General characteristics of studies included in this review are shown completely in Table  1 .

Relationship between PhA, general characteristics and cancer features

Four studies investigated the potential relationship between age and PhA. In the study by Ji et al. [ 36 ], Pearson’s analysis showed an inverse correlation for age ( r  = -0.238 p  =  < 0.001). Shi et al. [ 39 ], showed both for men and women, higher PhA in younger (for both, p  < 0.001); Spearman’s correlation analysis also showed that PhA was significantly correlated with age (men, r  = -0.46, p  < 0.001; women, r  = -0.24, p  < 0.001). In Castanho et al. [ 44 ], Pearson’s correlation showed no significant results between PhA and age ( r  = -0.32). Suzuki et al. showed a negative correlation between age and PhA ( r  = -0.51; p  < 0.001); Spearman’s correlation was used to assess the correlation between age and PhA.

Only 8 of 11, considered data on patients with LC, while 3 studies considered patients with different types of cancer, including LC. Three other studies investigate PhA in different types of cancer, including LC: they respectively include 8, 244 and 26 patients with LC [ 36 , 37 , 45 ]. All studies showed patients with confirmed cancer diagnosis: five studies evaluated NSCLC patients with stage III and IV [ 34 , 38 , 40 , 41 , 42 ]. Shi et al. included adenocarcinoma, squamous cell carcinoma, SCLC and other type of LC patients in different stage of disease (from I to IV) [ 39 ]. Suzuki et al. and Hui et al. showed data on LC patients but did not specify cancer type or stage [ 43 , 45 ]. Cancer stage was not specified in two studies [ 36 , 37 ]. Castanho et al. [ 44 ] considered LC patients presenting from stage IB to IIIB. Results showed, by multifactorial analysis of variance, correlations between PhA and tumour size ( r  = -0.55; p  < 0.001) or Karnofski Performance Status (KPS) ( r  = 0.44; P  < 0.05). Three studies included patients not undergoing CHT, RT and without specific treatment information [ 37 , 39 , 45 ] while newly diagnosed patients and/or patients with ongoing therapies were evaluated in remaining studies [ 34 , 36 , 38 , 42 ]. Suzuki et al. [ 43 ] evaluated LC patients after surgery while patients were evaluated after cancer treatment in two studies [ 40 , 41 ]. Castanho et al. [ 44 ] evaluated patients after surgery and after cancer treatment (CHT, RT).

Relationship between PhA and body composition

The main associations identified in selected studies, between PhA and different variables, are shown in Table  2 . In seven studies parameters related to BC have been evaluated [ 36 , 39 , 40 , 41 , 43 , 44 ]. Castanho et al. [ 44 ] correlates PhA with arm circumference and weight loss over time. Authors also indicate that between patients undergoing surgery, those with lowest survival (52 days) showed a lower PhA (3.8°) and a High Extracellular Mass/Body Cell Mass (ECM:BCM) ratio (1:5), independent from tumour size. Whereas, in those treated medically, patients with a lower survival also had lowest PhA (3.9°) and highest ECM/BCM ratio (1:6), this were related to tumour volume (849 ml).

Shi et al. [ 39 ] shows that male patients < 65 years, with a lower Body Mass Index (BMI) and lower cancer stage have higher PhA values, but the differences were not statistically significant. In Suzuki et al. [ 43 ], Spearman’s correlation showed a positive correlation between PhA and BMI (rho = 0.29; p  < 0.001); no correlation with body Fat Mass (FM) was found. In Hui et al. [ 45 ] study, PhA was associated with several known prognostic variables, including Fat-Free Mass (FFM) and FFM index (FFMI). However, the Spearman correlation was weak (rho < 0.4; p  < 0.001). Two articles assessed different aspect within the same LC patients' population (Stage IV, male patients), including PhA. In Detopoulou et al. [ 40 ] were found significant correlations between PhA and FFM (rho = 0.247; p  = 0.02), but no significant correlation for waist and hip circumference (cm), waist-hip ratio, body fat (%), BCM (Kg), total body water (TBW), extra-cellular and intra-cellular water (ECW, ICW). In the other study, Spearman’s correlation shows no significance between PhA, anthropometric and BC variables [ 41 ]. In Wei Ji et al. [ 36 ], in addition to PhA, other variables were also examined, such as appendicular skeletal muscle mass (ASMM), BMI and skeletal muscle mass index (SMI). Pearson’s correlations showed a moderate correlation between PhA values and variables considered (ASMM r  = 0.301, p  < 0.001; BMI r  = 0.450, p  < 0.001; SMI r  = 0.463, p  < 0.001.

Relationship between PhA, nutritional status and nutritional risk

PhA relationships with nutritional status and malnutrition screening scores were evaluated. Shi et al. indicated a significant correlation between PhA and some nutritional index: results of Spearman’s rank correlation test showed correlation with Nutritional Risk Score-2002 (NRS-2002) (men, r  =  − 0.25, p  < 0.001; women, r  =  − 0.15, p  < 0.001). No correlation between PhA and Patient-Generated Subjective Global Assessment (PG-SGA) score was found. Then, logistic regression analysis showed significant correlation between PhA, NRS-2002 score (men, p  < 0.001; women, p  < 0.001) and PG-SGA score (men, p  < 0.001; women, p  < 0.001) in both men and women, indicating an association with secondary clinical outcomes such as nutrition and well-being [ 39 ]. In Suzuki et al. [ 43 ], Spearman’s correlation showed a positive correlation between PhA and albumin (rho = 0.33; p  < 0.001), considered a useful biochemical markers for nutrition assessment, and Prognostic Nutritional Index (PNI) (rho = 0.32; p  < 0.001), a simple index obtained from serum albumin concentration and total peripheral blood lymphocyte count, used to assess the immune-nutritional status of patients who undergo surgery. In Hui et al. [ 45 ] study, PhA was associated with several known prognostic variables, including serum albumin, but correlation was weak (γ < 0.4, p  < 0.001; Spearman correlation test). Detopoulou et al. [ 40 ] found significant correlations between PhA and dietary pattern (Food pattern 2) rich in potatoes, meat and poultry (rho = 0.254, p  = 0.02). No significant results with PhA and other dietary patterns (food pattern 1: whole grains, fruits, vegetables; food pattern 3: high olive oil, low alcohol; food pattern 4: legumes, fish). Finally, in the same patient’s sample, Mediterranean Diet Score (MedDiet Score) was positively related to PhA changes (rho = 0.251; p  = 0.02).

Relationship between PhA, prognostic indices, quality of life and survival

Some of the selected studies evaluated the correlation between PhA and some prognostic indices, QoL scores and survival in patients with LC. Five out of eleven studies indicate patient survival data in relation to the PhA [ 34 , 35 , 38 , 39 , 42 ]; in two studies, indicators associated with survival were evaluated [ 34 , 35 , 38 , 39 , 42 , 43 , 44 ]. Sanchez-Lara et al. and Shi et al. have evaluated QoL in relation to PhA.

Multifactorial analysis of variance showed correlations between PhA and KPS ( r  = 0.44; P  < 0.05) in cross-sectional study by Castanho et al. [ 44 ].

In Hui et al. [ 45 ] study, PhA was associated with several known prognostic variables, including the Palliative Performance Scale (γ = 0.18; p  = 0.007), KPS (γ = 0.18; p  = 0.007), Palliative Prognostic Score (γ = -0.21; p  = 0.002), and Palliative Prognostic Index (γ = -0.22; p  = 0.001). Unadjusted PhA ( P  = 0.001) was found to be significantly associated with overall survival, as indicated by Kaplan-Meyer curves analysis: a lower value is associated with poor survival (PhA < 3°, median 35 days; 95% CI, 29–41 days).

Sanchez-Lara et al. [ 38 ] evaluated the association of PhA, QoL’s dimensions EORTC QLQ C30 (QLQ-C30) and survival in patients with advanced NSCLC. No significant association between PhA and QoL scores were found. The bivariate survival analysis shows that PhA ≤ 5.8° was significantly associated with low overall survival; multivariate analysis indicate for highest PhA values a higher survival rate (HR = 3.02; 95% CI, 1.2–7.11; p  = 0.011).

The results of Spearman’s rank correlation test in Shi et al. [ 39 ] showed correlation between PhA and QoL. It was found a L-shaped association between PhA and LC survival in both sexes (men p  = 0.019 and women p  = 0.121); an association between higher PhA and better survival resulted for men and women ( p  = 0.007 and p  < 0.001, respectively). Kaplan–Meier survival curves for patients with high and low PhA values in different cancer stages showed longer OS in patients with high PhA than patients with low PhA, without taking account stage. Univariate Cox regression analysis showed that continuous PhA was significantly associated with mortality in men with LC ( p  = 0.015); also in women, PhA was significantly associated with survival ( p  = 0.029). After adjusting for several covariates, in a multivariate-adjusted Cox regression analysis PhA was identified as an independent risk factor for mortality in men (HR = 0.79, 95% CI = 0.65–0.95, p  = 0.015), but not in women ( p  = 0.105) [ 39 ].

In Gupta et al., univariate Kaplan–Meier survival analysis showed statistically significant differences ( p  = 0.02) between patients with PhA <  = 5.3 (median survival = 7.6 months; 95% CI: 4.7 to 9.5; n  = 81) and those with > 5.3 (12.4 months; 95% CI: 10.5 to 18.7; n  = 84) [ 34 ].

No correlation with Charlson Comorbidity Index was found in Suzuki et al. (rho = -0.09; p  = 0.16). Also, in this study, multivariate logistic analysis reveals that PhA (OR = 0.51, 95% CI: 0.29–0.90, p  = 0.018) was an independent predictor of Clavien-Dindo grade ≥ II, index used for surgical complications [ 43 ].

Data from univariate survival analysis of Toso et al., stratified by the cancer stage, indicated that LC patients with a PhA ≤ 4.5° had significantly shorter survival compared to those who have a higher PhA ( p  = 0.01) (median of 3.7 vs 12.1 months in patients with a PhA ≤ 4.5° vs > 4.5°, respectively, from stage IIIB, and 1.4 vs 5.0 months in in patients with a PhA ≤ 4.5° vs > 4.5, respectively from stage IV) [ 42 ].

Relationship between PhA, muscle strength and physical efficiency

Navigante et al. [ 37 ] evaluated weakness assessed with Hand-grip strength (HGS). In patients with LC only statistically significant result was linear correlation between grip work and PhA ( p  = 0.007), which was considered very significant (95% CI: 0.3843 to 0.9717).

In Hui et al., PhA was also associated with HGS, but correlation was not very strong (Spearman’s correlation γ < 0.4; p  > 0.001) [ 35 ].

Wei Ji et al. have evaluated muscle strength and ASMM to define sarcopenia diagnosis: PhA had strongest correlation with SMI ( r  = 0.463) and HGS ( r  = 0.354). Logistic regression analysis adjusted for potential confounders showed that higher PhA values represent a protective factor for sarcopenia (OR 0.309, 95% CI, 0.246 0.617; p  < 0.001) [ 36 ].

Comparison between different groups and PhA

In the study of Toso et al. were reported differences between healthy subjects, patients with IIIB and IV stages.

Comparing patients with healthy controls was found a reduction in PhA value (resulting in a reduction in capacity, but not R. No significant differences between two groups of patients with IIIB and IV stages were found. However, a significant difference between patients with different stages (statistically lower in patients with higher stages) was observed for survival [ 42 ].

In Navigante et al. was carried out a comparison between different groups (healthy volunteers vs patients), but in reference to muscle strength (maximal muscle strength, mean muscle strength, median muscle strength) and not to PhA [ 37 ].

In the Hui et al. patients’ cohort ( n  = 204) with different types of cancer (including breast, gastrointestinal, head and neck and gynaecological) two different groups have been distinguished: patients with edema and without edema. Univariate analysis showed a reduced survival for PhA ≤ 3° vs PhA > 3° for total patients ( p  = 0.045) and no edema patients ( p  < 0.001). PhA ≤ 3° was associated with shorter survival in the non-oedematous cohort (HR 4.42, 95% CI 2.09–9.36, p  < 0.001), but this association did not occur in the whole cohort (HR 1.44, 95% CI 0.99–2.09, p  = 0.054) and in the cohort with edema (HR 1.04, 95% CI 0.67–1, 62, p  = 0.85) (Cox regression analysis) [ 35 ].

Wei Ji et al. evaluated the association between PhA in older male patients with different types of malignancies, with sarcopenia (22.0%) and without. PhA in patients without sarcopenia was 5.02° (SD ± 0.72°), while in sarcopenic was 4.18° (SD ± 0.85°); this difference was statistically significative ( p  < 0.001) [ 36 ].

The present review aims to investigate the current data regarding PhA in LC patients. We did not find a large number of studies focused on the assessment of PhA, which made it difficult to reach a comprehensive conclusion on this topic. We found 11 studies evaluating the PhA value obtained from BIA in patients with LC. In order to choose the right cancer treatment and plan carefully, survival prediction is crucial. In any case, new tools are necessary to be applied in daily clinical practice.

In the latest years, a growing body of studies have evaluated the prognostic role of PhA not only in patients with LC, but also in patients with respiratory disease. Indeed, De Benedetto et al. reported that lower PhA is associated with a decreased muscle mass, muscle strength and exercise capacity in patients with idiopathic pulmonary fibrosis, regardless of body weight. Moreover, patients with Chronic Obstructive Pulmonary Disease (COPD) and lower PhA have reduced cell mass, evident skeletal muscle depletion, worsening gas exchange and an increased risk of all-cause mortality [ 46 ]. Similarly, a lower PhA has been associated with an increased risk of malnutrition, sarcopenia, fluid retention, systemic inflammation, symptoms, and poorer QoLin patients with cancer. Moreover, a lower PhA may be a novel prognostic factor of poorer overall survival and higher risk of postoperative complications in cancer patients [ 47 ].

Overall, lower PhA levels have been associated with poorer physical condition and shorter survival in patients with LC: in Sanchez-Lara et al. [ 38 ], Gupta et al. [ 30 , 31 , 32 , 34 , 48 ] and Toso et al. [ 42 ] LC patients with lower PhA had a shorter survival than patients with higher PhA. In addition, in Shi et al. [ 39 ] patients with a higher PhA had a better survival and PhA was an independent risk factor for mortality in men with LC. Navigante et al. [ 37 ] in cancer patients without edema, PhA values ≤ 3° were associated with mortality within three days of BIA analysis, while in sarcopenic patients the PhA value was reduced compared to non-sarcopenic patients.

Patients with a lower PhA also had a higher risk of complications after surgical procedures [ 36 ]. In the prospective observational study of Uccella et al., it was observed that PhA was an independent prognostic factor of optimal cytoreduction and postoperative complications among patients with primary diagnosis of advanced ovarian cancer [ 49 ]. Similarly, in the prospective observational study of Inci et al. (The Risk-Gin trial) the authors observed that patients undergoing surgery for gynecological cancer with PhA < 4.75° and HGS < 44 kg in both hands had a three-fold increased risk of 30 days severe postoperative complications [ 50 ].

Thus, in addition to being a marker of cellular function, muscle mass and nutritional status, PhA may be a predictive factor of acute catastrophic complications risk. Interestingly, PhA was weakly but significantly associated with other prognostic variables, suggesting that it captures some additional information compared to existing prognostic factors. Further studies are needed to examine physiological and cellular changes associated with PhA.

Gupta et al. have evaluated role of PhA in the prognosis of 52 patients with advanced colorectal cancer: patients with PhA ≤ 5.7º had an 8-months average survival rate (Kaplan–Meier method), while those with PhA > 5.57º had a 40-month average survival rate [ 32 ]. Bosy-Westphal et al. [ 51 ] showed that patients with PhA < the fifth percentile had a deterioration in nutritional and functional status, decreased QoL and increased morbidity and mortality. The fifth percentile was a clinically relevant indicator of cancer cachexia. In this context, Hui et al. [ 35 ]investigated the association between PhA and survival in individuals with terminal cancer, where the increment of 1 degree in PhA was associated with higher survival rates.

In the prospective observational study by Paiva et al. PhA is reported as an independent prognostic factor, and Norman et al. [ 52 ]showed that in cancer patients, PhA values (stratified by age, sex, and BMI) below the fifth percentile of reference corresponded to a significant deterioration in nutritional status. In addition, these patients showed decreased HGS, increased incidence of weight loss, dyspnoea, fatigue, and increased risk of mortality. In three articles positive correlation has emerged between HGS and PhA, so weakness is related to the reduction in PhA in cancer patients. In other patient’s populations, it was shown that impaired muscle strength was associated with a poorer prognosis [ 53 , 54 ]. In addition, correlations between HGS, PhA and other BIA variables have emerged in adolescents and young adults [ 55 ].

In Navigante et al. [ 37 ], univariate logistic regression analysis showed that higher PhA values represent a protective factor for sarcopenia. In Pérez-Camargo et al. [ 56 ] palliative care patients with cancer (the most frequently were gastric cancer, gynecological cancer, LC, and haematological malignancies) and severe sarcopenia had a lower mean PhA (3.9°) compared to patients without sarcopenia (mean PhA was 4.1°) showing that PhA is an independent measurement that can be associated to detect sarcopenia. Moreover, authors found that sarcopenic patients had a shorter overall survival and an increased risk of death compared to patients without sarcopenia. A recent review [ 57 ] indicated that PhA and sarcopenia are related to LC prognosis through different mechanisms including inflammation and oxidative stress. Detection of sarcopenia and the evaluation of BC and PhA can be a valuable tool for identifying and timely intervention of the state of malnutrition of the cancer patient. Timely analysis of patient’s nutritional status is essential, as it allows to avoid significant loss of cell mass and lean mass, but ensures a proper nutritional approach in order to avoid aggravation of the general condition of the patient. Moreover, to obtain an accurate clinical interpretation of PhA, simultaneous assessment of hydration and BCM status is required [ 58 ]. These informations could be derived from the analysis of vector length on the R/Xc plot using Bioelectrical Impedance Vectorial Analysis (BIVA) (a scatterplot that represents R in X-axis and Xc in Y-axis divided by height in meters). In healthy subjects, a balance is observed between BCM and hydration status of FFM: malnutrition, sarcopenia and cachexia can lead to a loss of BCM and cell membrane surface area, provoking cell damage and a reduction in PhA [ 58 ].

According to the American Cancer Society (ACS), one of the first hallmarks of several types of cancer, including LC, is unexplained weight loss. In addition, treatments such as RT and CHT could generate side effects that cause inappetence to patients (e.g., mouth ulcers, nausea, vomiting). Tumour growth leading to extreme loss of appetite and weight in association with systemic signs of inflammation may be breeding grounds for cachexia [ 59 ]. Cachexia leads to a considerable loss of skeletal muscle mass (SMM) that cannot be completely compensated with traditional nutritional supports. Furthermore, it may be an underlying condition in patients with sarcopenia. We know that malnutrition and cachexia are often present in cancer patients [ 13 ]: this state can worsen the effects of anticancer therapies, with premature discontinuation of treatment, patient's QoL decreased and higher risk of mortality [ 60 , 61 ]. Nutritional status in cancer patients, especially the elderly, should be evaluated before and during CHT [ 61 ]. However, evaluation and detection of malnutrition status is not simple and instantaneous in patients. Therefore, it is preferable and useful to use the different variables defined by the BIA, to evaluate the changes in cellular membrane and body water levels. Future research on the use of PhA in clinical practice will be valuable in establishing cut-off values to better categorize oncology patients [ 58 ].

Limitations

To our knowledge, this is the most recent review focusing on the assessment of BIA-derived PhA in LC patients, however, we had to consider several limitations. Although several search engines outside PubMed were used, the review included only 11 studies of which only 8 were aimed at analysing PhA exclusively on patients with LC. Only four studies related to this topic were published in 2022, the other articles we included in our work were published between 2000 and 2013, so we do not have a large number of recent evidence.

Finally, accurate data on PhA relationships and other variables such as anthropometric values, HGS test, ECM/BCM ratio, tumour stage, tumour volume, etc., are absolutely necessary for a broader understanding of how PhA can be beneficial for this population. No information about PhA timing measurement by bioimpedance were provided in selected studies Monitoring the health of these patients is very important in order to be able to act promptly with targeted integrated therapies involving nutrition and physical activity, so further studies are needed. Patients with LC are at greater risk of malnutrition, sarcopenia and cachexia, with implications for physical function and overall QoL. Therefore, before making an assessment of the BC and PhA, it would be useful to consider the stage of the disease, cancer therapies carried out and the side effects, comorbidities and possible states of inflammation. All measurement should be performed in standard conditions and, in these specific cases, in days that do not correspond with cancer therapy. Patients with chronic respiratory disease could experience dysfunction in skeletal muscle mass (SMM) and BC changes as consequence of smoking, alcohol abuse, systemic inflammation, systemic oxidative stress, hormonal deficiencies, comorbidities, aging, and inappropriate diet [ 46 , 47 ]. Moreover, fat-free mass depletion and decreased muscle strength are common features of these patients. In this context, usefulness of PhA as a health status marker was investigated by a growing body of study. Similarly, patients with cancer patients have multiple comorbidities (chronic kidney disease) which may impact BC and cellular function. Moreover, patients with cancer frequently have fluid retention including edema, ascites, and pleural effusions. These factors may complicate the interpretation of PhA because this parameter is affected by altered ECW/ICW, or fluid disruption. Therefore, given the few studies currently available and the high number of factors that can affect BC measurements (hydration status, concomitant intake of food, alcohol use, physical activity, menstrual cycle, use of specific drugs that increase cell retention and catabolism, etc.) it should maintain accurate standardization in measurement.

Conclusions

The evaluation of PhA by BIA analysis in LC patients is not widely discussed in scientific literature. However, early identification in nutritional status changes in cancer patients represents a crucial aspect to improve patient’s quality of life both post-diagnosis and during and after anticancer therapies, avoiding possible states of malnutrition and sarcopenia, which can aggravate patient’s status. This systematic review shows an association between a very low PhA and an increased risk of a deficent physical condition, linked to reduced survival in lung cancer patients. In the selected studies, various cut-offs point for PhA have been reported that need to be interpreted with caution: to date, it is not possible to define a single threshold or cut off point for PhA due to technical differences in commercial BIA devices (single-, multiple-frequency and BIS). Given the high incidence of this cancer and the low number of studies on this issue, it would be important and necessary to make greater use of this screening method in clinical practice.

Availability of data and materials

All the data analysed for this review are present in the tables; there are no additional files.

Abbreviations

American Cancer Society

Appendicular Skeletal Muscle Mass

• Body composition

Body Cellular Mass

Bioelectrical Impedance Analysis

Bioelectrical Impedance Vectorial Analysis

Body Mass Index

Calf Circumference

Chemotherapy

Chronic Obstructive Pulmonary Disease

Extra-cellular Mass

Extra-cellular Water

Fat-Free Mass

Fat-Free Mass index

Hand-grip strength

Health Related Quality of Life

Immune checkpoint inhibitor

Intra-Cellular Water

Karnofski Performance Status

Lung Cancer

Mediterranean Dietary Score

Multiple-frequency Bioelectrical Impedance Analysis

Nutritional Risk Score-2002

Non-small cell lung carcinoma

Overall Survival

Phase Angle

Patient-Generated Subjective Global Assessment

Prognostic Nutritional Index

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Quality of Life Questionnaire – Cancer 30

Quality of Life

Radiotherapy

Small cell lung carcinoma

Single-frequency Bioelectrical Impedance Analysis

Skeletal Muscle Mass Index

Skeletal Muscle Mass

Total Lymphocyte Count

Total Body Water

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209–49.

Article   PubMed   Google Scholar  

I Numeri del Cancro in Italia 2022 [ https://www.aiom.it/i-numeri-del-cancro-in-italia/ ].

Zhang P, Chen PL, Li ZH, Zhang A, Zhang XR, Zhang YJ, Liu D, Mao C. Association of smoking and polygenic risk with the incidence of lung cancer: a prospective cohort study. Br J Cancer. 2022;126(11):1637–46.

Article   PubMed   PubMed Central   Google Scholar  

Kim CH, Lee YC, Hung RJ, McNallan SR, Cote ML, Lim WY, Chang SC, Kim JH, Ugolini D, Chen Y, et al. Exposure to secondhand tobacco smoke and lung cancer by histological type: a pooled analysis of the International Lung Cancer Consortium (ILCCO). Int J Cancer. 2014;135(8):1918–30.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Luo J, Hendryx M, Ducatman A. Association between six environmental chemicals and lung cancer incidence in the United States. J Environ Public Health. 2011;2011: 463701.

Zheng M. Classification and Pathology of Lung Cancer. Surg Oncol Clin N Am. 2016;25(3):447–68.

Mele MC, Rinninella E, Cintoni M, Pulcini G, Di Donato A, Grassi F, Trestini I, Pozzo C, Tortora G, Gasbarrini A, et al. Nutritional support in lung cancer patients: the state of the art. Clin Lung Cancer. 2021;22(4):e584–94.

Beckles MA, Spiro SG, Colice GL, Rudd RM. Initial evaluation of the patient with lung cancer: symptoms, signs, laboratory tests, and paraneoplastic syndromes. Chest. 2003;123(1 Suppl):97S–104S.

Polanski J, Jankowska-Polanska B, Rosinczuk J, Chabowski M, Szymanska-Chabowska A. Quality of life of patients with lung cancer. Onco Targets Ther. 2016;9:1023–8.

PubMed   PubMed Central   Google Scholar  

Arends J, Baracos V, Bertz H, Bozzetti F, Calder PC, Deutz NEP, Erickson N, Laviano A, Lisanti MP, Lobo DN, et al. ESPEN expert group recommendations for action against cancer-related malnutrition. Clin Nutr. 2017;36(5):1187–96.

Article   CAS   PubMed   Google Scholar  

Porro C, La Torre ME, Tartaglia N, Benameur T, Santini M, Ambrosi A, Messina G, Cibelli G, Fiorelli A, Polito R, et al. The potential role of nutrition in lung cancer establishment and progression. Life (Basel). 2022;12(2):270.

CAS   PubMed   Google Scholar  

Pilikidou M, Palyvou F, Papadopoulou SK, Tsiouda T, Tsekitsidi E, Arvaniti K, Miziou A, Tsingerlioti Z, Apostolidis G, Ntiloudis R, et al. Lung cancer, treatment and nutritional status. Mol Clin Oncol. 2021;15(6):248.

Dewys WD, Begg C, Lavin PT, Band PR, Bennett JM, Bertino JR, Cohen MH, Douglass HO, Jr., Engstrom PF, Ezdinli EZ et al.: Prognostic effect of weight loss prior to chemotherapy in cancer patients. Eastern Cooperative Oncology Group. Am J Med 1980, 69(4):491–497.

Ravasco P, Monteiro-Grillo I, Vidal PM, Camilo ME. Nutritional deterioration in cancer: the role of disease and diet. Clin Oncol (R Coll Radiol). 2003;15(8):443–50.

Prado CM, Lieffers JR, Bowthorpe L, Baracos VE, Mourtzakis M, McCargar LJ. Sarcopenia and physical function in overweight patients with advanced cancer. Can J Diet Pract Res. 2013;74(2):69–74.

Fortunati N, Manti R, Birocco N, Pugliese M, Brignardello E, Ciuffreda L, Catalano MG, Aragno M, Boccuzzi G. Pro-inflammatory cytokines and oxidative stress/antioxidant parameters characterize the bio-humoral profile of early cachexia in lung cancer patients. Oncol Rep. 2007;18(6):1521–7.

Duguet A, Bachmann P, Lallemand Y, Blanc-Vincent MP. Good clinical practice in nutritional management in cancer patients: malnutrition and nutritional assessment. Bull Cancer. 1999;86(12):997–1016.

Polanski J, Chabowski M, Swiatoniowska-Lonc N, Dudek K, Jankowska-Polanska B, Zabierowski J, Mazur G: Relationship between Nutritional Status and Clinical Outcome in Patients Treated for Lung Cancer. Nutrients 2021;13(10):3332.

Lin TY, Chen YF, Wu WT, Han DS, Tsai IC, Chang KV, Ozcakar L. Impact of sarcopenia on the prognosis and treatment of lung cancer: an umbrella review. Discov Oncol. 2022;13(1):115.

Bapuji SB, Sawatzky JA. Understanding weight loss in patients with colorectal cancer: a human response to illness. Oncol Nurs Forum. 2010;37(3):303–10.

Argiles JM, Lopez-Soriano FJ, Busquets S. Mechanisms and treatment of cancer cachexia. Nutr Metab Cardiovasc Dis. 2013;23(Suppl 1):S19–24.

Wang J, Tan S, Gianotti L, Wu G. Evaluation and management of body composition changes in cancer patients. Nutrition. 2023;114:112132.

Yamada Y, Buehring B, Krueger D, Anderson RM, Schoeller DA, Binkley N. Electrical Properties Assessed by Bioelectrical Impedance Spectroscopy as Biomarkers of Age-related Loss of Skeletal Muscle Quantity and Quality. J Gerontol A Biol Sci Med Sci. 2017;72(9):1180–6.

PubMed   Google Scholar  

Barbosa-Silva MC, Barros AJ. Bioelectrical impedance analysis in clinical practice: a new perspective on its use beyond body composition equations. Curr Opin Clin Nutr Metab Care. 2005;8(3):311–7.

Almada-Correia I, Neves PM, Makitie A, Ravasco P. Body Composition Evaluation in Head and Neck Cancer Patients: A Review. Front Oncol. 2019;9:1112.

Branco MG, Mateus C, Capelas ML, Pimenta N, Santos T, Makitie A, Ganhao-Arranhado S, Trabulo C, Ravasco P: Bioelectrical Impedance Analysis (BIA) for the Assessment of Body Composition in Oncology: A Scoping Review. Nutrients 2023, 15(22):792.

Article   Google Scholar  

Caccialanza R, Cotogni P, Cereda E, Bossi P, Aprile G, Delrio P, Gnagnarella P, Mascheroni A, Monge T, Corradi E, et al. Nutritional support in cancer patients: update of the Italian Intersociety Working Group practical recommendations. J Cancer. 2022;13(9):2705–16.

Shintani Y, Ikeda N, Matsumoto T, Kadota Y, Okumura M, Ohno Y, Ohta M. Nutritional status of patients undergoing chemoradiotherapy for lung cancer. Asian Cardiovasc Thorac Ann. 2012;20(2):172–6.

Marin Caro MM, Laviano A, Pichard C. Nutritional intervention and quality of life in adult oncology patients. Clin Nutr. 2007;26(3):289–301.

Gupta D, Lis CG, Dahlk SL, Vashi PG, Grutsch JF, Lammersfeld CA. Bioelectrical impedance phase angle as a prognostic indicator in advanced pancreatic cancer. Br J Nutr. 2004;92(6):957–62.

Gupta D, Lammersfeld CA, Vashi PG, King J, Dahlk SL, Grutsch JF, Lis CG. Bioelectrical impedance phase angle as a prognostic indicator in breast cancer. BMC Cancer. 2008;8:249.

Gupta D, Lammersfeld CA, Burrows JL, Dahlk SL, Vashi PG, Grutsch JF, Hoffman S, Lis CG. Bioelectrical impedance phase angle in clinical practice: implications for prognosis in advanced colorectal cancer. Am J Clin Nutr. 2004;80(6):1634–8.

Grundmann O, Yoon SL, Williams JJ. The value of bioelectrical impedance analysis and phase angle in the evaluation of malnutrition and quality of life in cancer patients–a comprehensive review. Eur J Clin Nutr. 2015;69(12):1290–7.

Gupta D, Lammersfeld CA, Vashi PG, King J, Dahlk SL, Grutsch JF, Lis CG. Bioelectrical impedance phase angle in clinical practice: implications for prognosis in stage IIIB and IV non-small cell lung cancer. BMC Cancer. 2009;9:37.

Hui D, Bansal S, Morgado M, Dev R, Chisholm G, Bruera E. Phase angle for prognostication of survival in patients with advanced cancer: preliminary findings. Cancer. 2014;120(14):2207–14.

Ji W, Liu X, Zheng K, Yang H, Cui J, Li W. Correlation of phase angle with sarcopenia and its diagnostic value in elderly men with cancer. Nutrition. 2021;84:111110.

Navigante A, Morgado PC, Casbarien O, Delgado NL, Giglio R, Perman M. Relationship between weakness and phase angle in advanced cancer patients with fatigue. Support Care Cancer. 2013;21(6):1685–90.

Sanchez-Lara K, Turcott JG, Juarez E, Guevara P, Nunez-Valencia C, Onate-Ocana LF, Flores D, Arrieta O. Association of nutrition parameters including bioelectrical impedance and systemic inflammatory response with quality of life and prognosis in patients with advanced non-small-cell lung cancer: a prospective study. Nutr Cancer. 2012;64(4):526–34.

Shi J, Xie H, Ruan G, Ge Y, Lin S, Zhang H, Zheng X, Liu C, Song M, Liu T, et al. Sex differences in the association of phase angle and lung cancer mortality. Front Nutr. 2022;9:1061996.

Detopoulou P, Tsiouda T, Pilikidou M, Palyvou F, Mantzorou M, Perzirkianidou P, Kyrka K, Methenitis S, Kondyli FS, Voulgaridou G, et al. Dietary habits are related to phase angle in male patients with non-small-cell lung cancer. Curr Oncol. 2022;29(11):8074–83.

Detopoulou P, Tsiouda T, Pilikidou M, Palyvou F, Tsekitsidi E, Mantzorou M, Pezirkianidou P, Kyrka K, Methenitis S, Voulgaridou G et al: Changes in Body Weight, Body Composition, Phase Angle, and Resting Metabolic Rate in Male Patients with Stage IV Non-Small-Cell Lung Cancer Undergoing Therapy. Medicina (Kaunas) 2022;58(12):1779.

Toso S, Piccoli A, Gusella M, Menon D, Bononi A, Crepaldi G, Ferrazzi E. Altered tissue electric properties in lung cancer patients as detected by bioelectric impedance vector analysis. Nutrition. 2000;16(2):120–4.

Suzuki Y, Kushimoto Y, Ishizawa H, Kawai H, Ito A, Matsuda Y, Hoshikawa Y. The phase angle as a predictor of postoperative complications in patients undergoing lung cancer surgery. Surg Today. 2023;53(3):332–7.

Castanho IA, Lopes AJ, Koury JC, Tessarollo B, Silva AC, Nunes RA. Relationship between the phase angle and volume of tumours in patients with lung cancer. Ann Nutr Metab. 2013;62(1):68–74.

Hui D, Moore J, Park M, Liu D, Bruera E. Phase Angle and the Diagnosis of Impending Death in Patients with Advanced Cancer: Preliminary Findings. Oncologist. 2019;24(6):e365–73.

De Benedetto F, Marinari S, De Blasio F. Phase angle in assessment and monitoring treatment of individuals with respiratory disease. Rev Endocr Metab Disord. 2023;24(3):491–502.

Amano K, Bruera E, Hui D. Diagnostic and prognostic utility of phase angle in patients with cancer. Rev Endocr Metab Disord. 2023;24(3):479–89.

Naughton MJ, Weaver KE. Physical and mental health among cancer survivors: considerations for long-term care and quality of life. N C Med J. 2014;75(4):283–6.

Uccella S, Mele MC, Quagliozzi L, Rinninella E, Nero C, Cappuccio S, Cintoni M, Gasbarrini A, Scambia G, Fagotti A. Assessment of preoperative nutritional status using BIA-derived phase angle (PhA) in patients with advanced ovarian cancer: Correlation with the extent of cytoreduction and complications. Gynecol Oncol. 2018;149(2):263–9.

Inci MG, Richter R, Woopen H, Rasch J, Heise K, Anders L, Mueller K, Nasser S, Siepmann T, Sehouli J. Role of predictive markers for severe postoperative complications in gynecological cancer surgery: a prospective study (RISC-Gyn Trial). Int J Gynecol Cancer. 2020;30(12):1975–82.

Bosy-Westphal A, Danielzik S, Dorhofer RP, Later W, Wiese S, Muller MJ. Phase angle from bioelectrical impedance analysis: population reference values by age, sex, and body mass index. JPEN J Parenter Enteral Nutr. 2006;30(4):309–16.

Norman K, Stobaus N, Pirlich M, Bosy-Westphal A. Bioelectrical phase angle and impedance vector analysis–clinical relevance and applicability of impedance parameters. Clin Nutr. 2012;31(6):854–61.

Hodgev VA, Kostianev SS. Maximal inspiratory pressure predicts mortality in patients with chronic obstructive pulmonary disease in a five-year follow-up. Folia Med (Plovdiv). 2006;48(3–4):36–41.

Cooper R, Kuh D, Hardy R, Mortality Review G. Falcon, Teams HAS: Objectively measured physical capability levels and mortality: systematic review and meta-analysis. BMJ. 2010;341: c4467.

Sacco AM, Valerio G, Alicante P, Di Gregorio A, Spera R, Ballarin G, Scalfi L. Raw bioelectrical impedance analysis variables (phase angle and impedance ratio) are significant predictors of hand grip strength in adolescents and young adults. Nutrition. 2021;91–92: 111445.

Google Scholar  

Perez Camargo DA, Allende Perez SR, Verastegui Aviles E, Rivera Franco MM, Meneses Garcia A, Herrera Gomez A, Urbalejo Ceniceros VI. Assessment and impact of phase angle and sarcopenia in palliative cancer patients. Nutr Cancer. 2017;69(8):1227–33.

Detopoulou P, Voulgaridou G, Papadopoulou S. Cancer, Phase Angle and Sarcopenia: The Role of Diet in Connection with Lung Cancer Prognosis. Lung. 2022;200(3):347–79.

Bellido D, Garcia-Garcia C, Talluri A, Lukaski HC, Garcia-Almeida JM. Future lines of research on phase angle: Strengths and limitations. Rev Endocr Metab Disord. 2023;24(3):563–83.

Blauwhoff-Buskermolen S, Versteeg KS, de van der Schueren MA, den Braver NR, Berkhof J, Langius JA, Verheul HM. Loss of Muscle Mass During Chemotherapy Is Predictive for Poor Survival of Patients With Metastatic Colorectal Cancer. J Clin Oncol 2016;34(12):1339-1344.

Aapro M, Arends J, Bozzetti F, Fearon K, Grunberg SM, Herrstedt J, Hopkinson J, Jacquelin-Ravel N, Jatoi A, Kaasa S, et al. Early recognition of malnutrition and cachexia in the cancer patient: a position paper of a European School of Oncology Task Force. Ann Oncol. 2014;25(8):1492–9.

Caillet P, Liuu E, Raynaud Simon A, Bonnefoy M, Guerin O, Berrut G, Lesourd B, Jeandel C, Ferry M, Rolland Y, et al. Association between cachexia, chemotherapy and outcomes in older cancer patients: A systematic review. Clin Nutr. 2017;36(6):1473–82.

Download references

This research was funded by Ministry of Health (5X1000_2021_2).

Author information

Authors and affiliations.

Division of Radiotherapy, Istituto Nazionale Tumori IRCCS Fondazione G. Pascale, Naples, 80131, Italy

Melania Prete

Department of Medical, Movement Sciences and Wellbeing, University of Naples “Parthenope”, Naples, 80133, Italy

Giada Ballarin

Epidemiology and Biostatistics Unit, Istituto Nazionale Tumori, IRCCS Fondazione G. Pascale, Naples, 80131, Italy

Giuseppe Porciello, Assunta Luongo & Egidio Celentano

Department of Public Health, Federico II University Hospital, Via Pansini 5, Naples, 80131, Italy

Aniello Arianna & Luca Scalfi

Scientific Direction, Istituto Nazionale Tumori, IRCCS Fondazione G. Pascale, Naples, 80131, Italy

Valentina Belli

You can also search for this author in PubMed   Google Scholar

Contributions

M.P., G.B. and L.S. have proposed, designed and edited data analysis of this systematic review; M.P. and G.P. wrote the main manuscript text and prepared Fig.  1 and Tables 1 and 2 . All work was supervised by E.C and L.S. All authors have read and revised the manuscript.

Corresponding author

Correspondence to Giuseppe Porciello .

Ethics declarations

Ethics approval and consent to participate.

Not applicable.

Consent for publication

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Prete, M., Ballarin, G., Porciello, G. et al. Bioelectrical impedance analysis—derived phase angle (PhA) in lung cancer patients: a systematic review. BMC Cancer 24 , 608 (2024). https://doi.org/10.1186/s12885-024-12378-4

Download citation

Received : 21 December 2023

Accepted : 13 May 2024

Published : 20 May 2024

DOI : https://doi.org/10.1186/s12885-024-12378-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Lung cancer
• Phase angle

ISSN: 1471-2407

• Submission enquiries: [email protected]
• General enquiries: [email protected]

• Open access
• Published: 27 May 2024

Integrated analysis of tumor-associated macrophages and M2 macrophages in CRC: unraveling molecular heterogeneity and developing a novel risk signature

• Lujing Shi 1 ,
• Hongtun Mao 1 &

BMC Medical Genomics volume  17 , Article number:  145 ( 2024 ) Cite this article

Metrics details

Emerging investigations have increasingly highlighted the critical role of tumor-associated macrophages (TAMs) and M2 macrophages in cancer development, progression, and metastasis, marking them as potential targets in various cancer types. The main objective of this research is to discover new biomarkers associated with TAM-M2 macrophages in colorectal cancer (CRC) and to dissect the molecular heterogeneity of CRC by combining single-cell RNA sequencing and bulk RNA-seq data.

By utilizing weighted gene co-expression network analysis (WGCNA), we acquired TAM-M2-associated genes by intersecting TAM marker genes obtained from scRNA-seq data with module genes of M2 macrophages derived from bulk RNA-seq data. We employed least absolute shrinkage and selection operator (LASSO) Cox analysis to select predictive biomarkers from these TAM-M2-related genes. Quantitative polymerase chain reaction (qPCR) was employed to validate the mRNA expression levels of the genes identified in the screening. This led to the development of the TAM-M2-related signature (TAMM2RS). We also conducted functional and immune landscape analyses of different risk groups.

The combination of scRNA-seq and bulk RNA-seq analyses yielded 377 TAM-M2-related genes. DAPK1, NAGK, and TRAF1 emerged as key prognostic genes in CRC, which were identified through LASSO Cox analysis. Utilizing these genes, we constructed and validated the TAMM2RS, demonstrating its effectiveness in predicting survival in CRC patients.

Our research offers a thorough investigation into the molecular mechanisms associated with TAM-M2 macrophages in CRC and unveils potential therapeutic targets, offering new insights for treatment strategies in colorectal cancer.

Peer Review reports

Colorectal cancer (CRC) remains a major global health concern, with an estimated 2 million new diagnoses and approximately 900,000 deaths in 2020 [ 1 ]. Moreover, CRC’s diverse clinical and molecular profiles exhibit marked differences in tumor progression and therapeutic responsiveness [ 2 ]. However, the pathogenic pathways driving CRC, though intricate, remain only partially elucidated. Hence, this situation highlights an exigent need for comprehensive investigation endeavors and the development of novel signatures to refine our predictive capabilities for CRC patient outcomes.

Myeloid cells emerge as a dominant immune subset within TME, involved in a spectrum of roles from immunosuppressive to immunostimulatory activities [ 3 ]. Notably, tumor-associated macrophages (TAMs) delineate a dynamic subpopulation, displaying a plasticity that enables phenotypic transitions contingent on TME cues [ 3 ]. The traditional dichotomy of macrophages into pro-inflammatory M1 and pro-tumoral M2 subsets has undergone a foundational shift [ 4 , 5 , 6 ]. Advances in single-cell RNA sequencing (scRNA-Seq) technologies have illuminated a more delicate macrophage spectrum, revealing overlapping transcriptional gene expression profiles between M1 and M2 entities [ 7 , 8 , 9 ]. While TAMs are conspicuously absent under normal physiological conditions, their presence in various tumors has prompted reconsideration of their classification. Intriguingly, while TAMs exhibit characteristics reminiscent of both M1 and M2 polarization, their operational functionalities mirror M2 macrophages [ 10 ]. Their pivotal roles in modulating TME immune landscapes, predominantly through immune suppression and facilitation of tumor immune evasion, accentuate their significance [ 11 ]. Furthermore, the paramountcy of TAMs in the TME crystallizes their potential as therapeutic targets, underscoring the imperative for in-depth insights into TAM-M2-mediated CRC pathogenesis and the consequent development of associated prognostic signatures.

Here, we collaboratively employ the scRNA-seq and bulk RNA-seq datasets to delineate the molecular heterogeneity of CRC based on marker genes of M2-TAMs. Then, we introduced a TAM-M2-related signature (TAMM2RS) for CRC. The integrative approach is illustrated in Fig.  1 .

Workflow of the study

Data curation

The single-cell data (GSE132465) were obtained from the Gene Expression Omnibus (GEO) database. The bulk RNA-seq data (including clinical data) were retrieved from The Cancer Genome Atlas (TCGA-CRC, n  = 612) and GEO (GSE39582, n  = 585) repositories.

Analyses of macrophage infiltration

Utilizing the CIBERSORT method, we evaluated the infiltration of M2 macrophages across TCGA-CRC samples. The optimal threshold for distinguishing high and low infiltration of M2 macrophages was established by the survminer package. Employing the survival package, we analyzed survival differences between sub-groups specified by high and low M1/M2 macrophage infiltration. By stratifying CRC specimens into categories based on their M2 macrophage infiltration (high or low), we conducted the Weighted Gene Co-expression Network Analysis (WGCNA) [ 12 ]. The objective of this analysis was to identify genes closely associated with M2 macrophage infiltration. We performed a clustering of the samples to evaluate their collective significance within the dataset, while excluding any outliers. Guided by the point where the scale-free topology fit index showed a substantial value, the selection of the soft-thresholding power β was determined at the minimum power. We set the minimum threshold for genes per module at 60.

ScRNA-seq data processing

We conducted the processing of the scRNA-seq dataset using Seurat package (version 4.3.0) [ 13 ]. The quality control criteria: preserving cells that have more than 300 identified genes and genes expressed in over three cells; excluding cells with > 20% mitochondrial gene expression. After cell filtration, normalization of the high-quality cellular data was performed, identifying highly variable genes pivotal for subsequent steps. Principal component analysis (PCA) was then applied to these genes to determine key principal components (PCs). For the visualization of cell clusters, we employed the Uniform Manifold Approximation and Projection (UMAP) method. Following this, the FindAllMarkers function was instrumental in identifying marker genes specific to each cluster. Each cell type was subsequently annotated, drawing references from the CellMarker 2.0 database ( http://bio-bigdata.hrbmu.edu.cn/CellMarker/ ). The FeatureHeatmap function was employed to illustrate the distinctiveness of each cell type and its corresponding biological processes.

Intersection of TAM and M2 macrophage associated genes

We executed correlation assessments between modules and specific traits to identify modules that significantly correlate with high M2 macrophage infiltration. Subsequently, the genes within these identified modules were cross-referenced with TAM marker genes, which were derived from scRNA-seq data analyses.

Consensus clustering

Initially, we conducted a Cox regression analysis to identify TAM-M2 genes with potential prognostic value, which were then used for consensus clustering analysis utilizing the ConsensusClusterPlus package [ 14 ]. The optimal cluster count was ascertained by examining the cumulative distribution function (CDF) and its delta area. The prognostic relevance of our clustering was corroborated by constructing K-M survival curves, using the survminer package. Furthermore, we employed the limma package to identify differentially expressed genes (DEGs) across clusters, focusing on those with an absolute log fold change (|logFC|) > 1 and an adjusted P-value < 0.05 [ 15 ].

Functional analyses

The DEGs identified were subsequently incorporated into a Gene Ontology (GO) functional analysis. Following this, we presented a comprehensive heatmap to depict the expression patterns and clinicopathological features of screened genes across the various clusters. Moreover, to elucidate the distinct biological profiles, we employed the GSVA (Gene Set Variation Analysis) package [ 16 ]. This approach facilitated a precise assessment of the unique biological attributes inherent to each identified cluster.

Immune landscape

To illustrate the abundance of 23 different types of immune cells across distinct clusters and explore the TME, we performed CIBERSORT and then employed the ggplot2 package for visualization [ 17 ]. Furthermore, to uncover potential targets for CRC immunotherapy, we examined the variability in expression of human leukocyte antigen (HLA) and immune checkpoint inhibitor (ICI) genes across the clusters.

Construction of TAMM2RS

We employed Cox regression analyses using the glmnet package on genes that were common to both TAM from scRNA-seq data and M2 macrophages from bulk RNA-seq data, with the aim of identifying prognostic genes. Following this, we employed LASSO Cox regression analysis to ascertain the coefficients of each gene that held predictive value. Based on these coefficients, risk scores were computed:

Based on median risk score, CRC samples were categorized into either a low-risk group (LRG) or a high-risk group (HRG). This categorization enabled the generation of K-M curves, which were used to underscore the survival differences between the risk groups. To further corroborate the robustness of this risk signature, the GSE39582 dataset was included to form an external validation cohort.

Moreover, the variation in clinicopathological features was delineated using the ggplot2 package. Furthermore, the survival disparities between the HRG and LRG were stratified and analyzed in the context of age, gender, and clinical stage.

Utilizing the clusterProfiler package, we conducted gene set enrichment analysis (GSEA) to investigate the pathways that are predominantly enriched in HRG [ 18 , 19 ]. These pathways were ordered based on their normalized enrichment scores, and the most significant pathways were selected for detailed visualization.

Immune infiltration analysis

We employed the Estimation of Stromal and Immune cells in Malignant Tumor tissues using Expression data (ESTIMATE) analysis to assess immune cell infiltration. This approach quantified stromal and immune scores, as well as the overall ESTIMATE scores, within the tumor microenvironment. Moreover, we utilized several algorithms to evaluate the correlation between risk score and macrophage infiltration.

Prediction of immunotherapy response

We sourced immunophenotype data from The Cancer Immunome Atlas to predict how CRC samples would respond to immunotherapy treatments. This data was utilized to compute the immunophenoscore (IPS) for each sample.

Cell culture

Human CRC cell line HCT116 (CL-0096) was obtained from Procell Life Science & Technology (China), while the NCM460 (iCell-h373) cell line, representing normal colonic epithelial cells, was acquired from Cellverse (China). HCT116 cells were propagated in DMEM (C11995500BT, Gibco, USA), and NCM460 cells were cultivated in RPMI 1640 medium (10-040-CV, Corning, USA). Both media were fortified with 10% fetal bovine serum (FBS; C0235, Gibco, USA) and a combination of 100 U/ml penicillin and 100 µg/ml streptomycin (C0222, Beyotime, China). Cell cultures were incubated at 37 °C in a humidified incubator with a 5% CO2 atmosphere.

Quantitative polymerase chain reaction (qPCR)

qPCR was employed to quantify mRNA expression levels utilizing the EZBioscience™ PCR array (EZBioscience, Roseville, CA, USA). Gene expression quantification was conducted applying the comparative 2^(-ΔΔCt) method, normalizing to GAPDH as the endogenous control. This analysis was independently replicated on three separate occasions. Primer sequences utilized for amplification are detailed in Table  1 .

Immunohistochemistry (IHC) analysis

We verified the protein expression profiles of both normal and CRC samples through the Human Protein Atlas (HPA) database.

Anti-cancer drugs prediction

Utilizing the oncoPredict package, our study assessed the anticancer effectiveness by gauging drug sensitivity in CRC patients [ 20 ].

Statistical analyses

R software (version 4.3.1) was employed to conduct the statistical analyses. Survival time distributions were estimated via the K-M method. The Wilcoxon test was utilized to compare two cohorts, while the Kruskal-Wallis test was carried out to examine differences among multiple groups. Statistical significance was indicated by establishing a P-value threshold below 0.05.

Acquisition of genes related to M2 macrophages

Our investigation sought to elucidate the prognostic impact of macrophages in CRC. Patients from the TCGA-CRC dataset were stratified into groups with high or low M2 macrophage infiltration via CIBERSORT method. Survival analysis using the K-M approach revealed that patients characterized by a high infiltration of M2 macrophages exhibited reduced survival rates (Fig.  2 A). This finding implicates M2 macrophages as significant prognostic factors in CRC. Subsequent WGCNA identified gene modules associated with M2 macrophage levels in CRC, with 11 modules emerging from the analysis (Fig.  2 B, Figure S1 ). From these, the black module, containing 1,158 genes, was chosen for further analysis (Table S1 ).

Identification of M2 macrophage-related genes. ( A ) The group with high infiltration of M2 macrophages exhibited a worse prognosis. ( B ) WGCNA was utilized to identify M2 macrophage-related modular genes. WGCNA, weighted gene co-expression network analysis

Identifying TAM marker genes

We analyzed the scRNA-seq data to identify TAM marker genes in CRC, with the objective of mapping the composition of the TME. After rigorous quality control and data normalization, a cohort of 63,252 cells from 23 CRC specimens was curated for subsequent analysis. Utilizing the FindNeighbors and FindClusters functions, we classified the cells into 39 distinct clusters (Figure S2 ). These clusters were subsequently annotated with cell type identities using the CellMarker 2.0 database, identifying diverse cellular populations including Epithelial cells, Plasma cells, T cells, Endothelial cells, Stromal cells, B cells, Mast cells, and Myeloids (Fig.  3 A). Further refinement of the myeloid population allowed us to re-cluster these cells into 9 distinct clusters, which were categorized as TAMs and macrophages based on extant literature (Fig.  3 B). 3,083 marker genes characteristic of TAMs were thus elucidated (Fig.  3 C, Table S2 ). Additionally, the distribution and expression of marker genes for each cell type were illustrated in Fig.  3 D.

ScRNA-seq processing. ( A ) Cell annotation. ( B ) Myeloids was re-clustered and annotated. ( C ) Identification of TAM marker genes. ( D ) Distribution and expression of marker genes. TAM, tumor-associated macrophage

Furthermore, we employed a heatmap for the validation of our annotation outcomes, which delineates the array of enriched biological processes alongside each cellular annotation, as illustrated in the rightmost column (Fig.  4 ).

Heatmap displaying marker genes and biological processes for each cell type

Determination of TAM-M2-mediated clusters

Upon merging the gene sets pertaining to M2 macrophages with TAM marker genes, we retrieved a total of 377 genes associated with TAM-M2 macrophages (Fig.  5 A, Table S3 ). To elucidate the TAM-M2-mediated heterogeneity within CRC, we applied consensus clustering to stratify the TCGA-CRC dataset into two principal clusters based on the expression profiles of TAM-M2-related genes (Fig.  5 B–D). Notably, patients classified under cluster C2 were observed to have an adverse prognosis relative to those in cluster C1 (Fig.  5 E).

Additionally, we incorporated the DEGs from clusters C1 and C2 into GO enrichment analysis. The findings revealed that these DEGs were predominantly enriched in biological processes related to ossification, regulation of vascular development, angiogenesis, extracellular matrix organization, and macrophage activation (Fig.  5 F). Furthermore, we generated a heatmap to portray the distribution of significant genes identified by univariate Cox analysis and the differential clinicopathological features between clusters C1 and C2 (Fig.  5 G).

Clustering. ( A ) Intersection of TAM marker genes and M2 macrophage-related modular genes. ( B-D ) Unsupervised consensus clustering. ( E ) K-M survival analysis between C1 and C2. ( F ) GO analysis for DEGs between C1 and C2. ( G ) Distribution of clinical features and expression of prognostic DEGs. GO, Gene Ontology; DEGs, differentially expressed genes

The findings from GSVA revealed a pronounced enrichment of C2 in several key biological processes, including natural killer cell chemotaxis, Toll-like receptor 7 signaling pathway, regulation of monocyte chemotaxis, macrophage cytokine production, macrophage activation and response to macrophage colony-stimulating factor (Fig.  6 A).

Immune analysis

The analysis using CIBERSORT demonstrated a notable infiltration of a majority of the 23 immune cell types in C2 (Fig.  6 B). Following this, the expression patterns of ICI- and HLA-related genes were comparatively analyzed across the two clusters. The expression of genes related to ICI and HLA exhibited distinctive patterns across the two clusters. (Figs.  6 C, D).

Functional and immune analyses between C1 and C2. ( A ) Heatmap illustrating GSVA differences between C1 and C2. ( B ) Immune infiltration analysis of 23 immune cells. ( C-D ) Difference in major ICI- and HLA-related genes between clusters. GSVA, gene set variation analysis; ICI, immune checkpoint inhibitor; HLA, human leukocyte antigen. * P  < 0.05; ** P  < 0.01; *** P  < 0.001

Utilizing the intersecting genes, we first applied univariate Cox regression analysis to screened genes of significant prognostic importance (Fig.  7 A). Subsequently, these genes were incorporated into a LASSO Cox regression analysis. Through this process, we identified three key genes - DAPK1, NAGK, and TRAF1 - for the construction of the TAMM2RS (Fig.  7 B, C, Table S4 ). The TCGA-CRC dataset was utilized as the training set, while the GSE39582 served as the validation set for external verification. K-M survival analysis across these cohorts indicated a less favorable outcome for the HRG (Fig.  7 D, E). Additionally, Fig.  7 F-K illustrated the variance in gene expression, risk scores, and survival statuses for both the training and external validation sets.

Construction of TAMM2RS. ( A ) Univariate Cox analysis for screened genes. ( B - C ) LASSO Cox regression analysis. K-M survival analyses for TCGA-CRC cohort ( D ) and GSE39582 cohort ( F ). Distribution of gene expressions, risk scores and survival statuses for TCGA-CRC cohort ( F-H ) and GSE39582 cohort ( I-K ). TAMM2RS, tumor-associated macrophages and M2 macrophages related signature; LASSO, least absolute shrinkage and selection operator; TCGA, The Cancer Genome Atlas; CRC, colorectal cancer

Additionally, we performed a sub-stratification analysis comparing the HRG and LRG. This analysis revealed a slight predominance of older age and advanced tumor stages within the HRG (Fig.  8 A-E). K-M survival analysis further elucidated that patients in the HRG had a less favorable prognosis across both age groups (≤ 65 years and > 65 years) and in both genders (male and female), as shown in Fig.  8 F and G. Notably, there was a significant prognostic difference in stages III-IV between HRG and LRG, whereas no such difference was apparent in stages I-II (Fig.  8 H).

Distribution of clinical features and survival analyses. ( A-E ) Distribution of different clinical features between HRG and LRG. K-M survival analyses for stratified by age ( F ), gender ( G ) and stage ( H ). HRG, high-risk group; LRG, low-risk group

Functional analyses between HRG and LRG

GSVA disclosed that the HRG exhibited enrichment in several biological processes, notably T helper 1 cell differentiation, positive regulation of monocyte differentiation, response to macrophage colony-stimulating factor and regulation of monocyte differentiation (Fig.  9 A). Furthermore, GSEA highlighted that the HRG predominantly showed enrichment in pathways related to cell adhesion molecules (CAMs), T cell receptor signaling pathway, cytokine-cytokine receptor interaction and chemokine signaling pathway (Fig.  9 B).

Immune investigations between HRG and LRG

The findings from ESTIMATE analyses revealed that, compared to the LRG, the HRG consistently showed higher scores in immune, stromal, and overall ESTIMATE scores, suggesting significant distinctions in the TME (Fig.  9 C). Moreover, the HRG was characterized by a marginally higher immune infiltration relative to the LRG (Fig.  9 D). To further investigate the relationship between risk score and M2 macrophage infiltration, we applied four distinct algorithms – XCELL, QUANTISEQ, CIBERSORT-ABS, and CIBERSORT. These analyses confirmed a positive correlation between the infiltration of M2 macrophages and the risk score (Fig.  9 E).

Functional and immune analyses between HRG and LRG. ( A ) Heatmap depicting GSVA differences between C1 and C2. ( B ) GSEA for HRG. ( C ) ESTIMATE analysis. ( D ) Immune infiltration analysis of 23 immune cells. ( E ) Correlation between risk score and M2 macrophage infiltration. HRG, high-risk group; LRG, low-risk group; GSVA, gene set variation analysis; GSEA, gene set enrichment analysis; ESTIMATE, Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data. * P  < 0.05; ** P  < 0.01; *** P  < 0.001

Further investigation revealed that C2 presented with a higher risk score compared to C1 (Fig.  10 A). In the search of identifying prospective treatment targets, we found that the HRG was characterized by a reduced expression of genes related to both ICI and HLA (Fig.  10 B, C). By contrast, the LRG patients demonstrated a heightened IPS, suggesting potentially greater sensitivity to immunotherapies (Fig.  10 D).

Prediction of response to immunotherapy. ( A ) In comparison to C1, C2 showed a higher risk score. ( B-C ) Difference in ICI- and HLA-related genes between clusters. ( D ) IPS score predicting immunotherapy response. ICI, immune checkpoint inhibitor; HLA, human leukocyte antigen; IPS, immunophenoscore. * P  < 0.05; ** P  < 0.01; *** P  < 0.001

Using qPCR, we were able to assess the relative expression levels of mRNA in CRC cells. The findings of this study revealed a notable upregulation of TRAF1 and DAPK1 expression in CRC cells (Fig.  11 A).

For further validation, we consulted the HPA database, which corroborated the transcriptional patterns and supplemented them with protein expression data derived from IHC staining (Fig.  11 B). Employing the oncoPredict algorithm, our analysis identified ten anticancer drugs (bortezomib, cediranib, gemcitabine, ibrutinib, irinotecan, mitoxantrone, rapamycin, vincristine, vinorelbine, and zoledronate) with reduced IC50 levels in the HRG (Fig.  11 C-L). This suggests a heightened likelihood of therapeutic efficacy for these drugs in patients classified within the HRG.

Experimental validation and anti-cancer drugs prediction. ( A ) qPCR. ( B ) IHC analysis from HPA database. ( C-L ) Anti-cancer drugs prediction. qPCR, Quantitative Polymerase Chain Reaction; IHC, immunohistochemistry; HPA, human protein atlas. * P  < 0.05; ** P  < 0.01; *** P  < 0.001

CRC exhibits significant heterogeneity in clinicopathological and molecular profiles, influencing tumor progression and treatment responses [ 21 ]. TME interactions facilitate CRC progression via multiple pathways. TAMs play a pivotal role in tumor progression by mediating immunosuppression, extracellular matrix remodeling, and releasing growth factors [ 22 ]. TAMs also interact with various immune cells within the TME, aiding in immune suppression and facilitating tumor immune evasion [ 11 ]. The critical role of TAMs in CRC progression has sparked interest in TAM-targeted therapeutic strategies [ 23 , 24 ]. Advances in immunotherapy have led to significant progress in cancer treatment [ 25 ]. However, not all patients respond to immunotherapy, largely due to TME characteristics [ 26 ]. PD-1 inhibitors have been shown to target TAMs directly, enhancing their phagocytic abilities [ 27 ]. Specific TAM markers, like CD40, show promise for use in adoptive cell immunotherapy [ 28 ]. Additionally, studies have demonstrated that CSF1R inhibitors can decrease TAM levels in the TME and induce macrophage repolarization towards the M1 phenotype, offering considerable clinical potential [ 29 ]. Eliminating SPP1 + TAMs could improve the efficacy of myeloid-targeted immunotherapy or enhance outcomes when combined with ICI therapies [ 9 ]. Despite these advancements, research in CRC remains insufficient, and further detailed studies on TAMs’ role in CRC prognosis are urgently needed.

In this study, we utilized bulk RNA-seq data to explore the prognostic implications of M2 macrophage infiltration in TCGA-CRC samples. Our findings indicate that higher levels of M2 macrophage infiltration are linked to poorer outcomes in CRC, underscoring the pivotal role of M2 macrophages in the prognosis. Acknowledging the dual roles of M2 macrophages in immunosuppression and tumor promotion, our study involved an intersectional analysis of M2 macrophage-associated genes from the TCGA-CRC dataset with TAM marker genes derived from scRNA-seq datasets [ 5 , 6 ]. This integrative approach led to the discovery of 377 genes that are related to both M2 macrophages and TAMs. Subsequent univariate and LASSO Cox regression analyses facilitated the selection of three prognostic genes (DAPK1, NAGK, and TRAF1) for the construction of the TAMM2RS. DAPK1 is noteworthy for its specificity in anal squamous cell carcinoma and potential as a molecular biomarker [ 30 ]. Methylation of DAPK1 correlates with nodal metastasis and is considered a significant risk factor in CRC plasma [ 31 , 32 ]. The DAPK1-ERK1 signaling axis is implicated in CRC metastatic progression, positioning DAPK1 as a key anti-metastatic factor and a prospective predictive biomarker [ 33 ]. Furthermore, inhibiting DAPK1 enhances cancer stem cell (CSC) stemness and the epithelial-mesenchymal transition (EMT) process, with the DAPK1-ZEB1 axis potentially intersecting the TGF-β and WNT pathways and influencing both CSCs and EMT processes [ 34 ]. TRAF1, on the other hand, is targeted by miR-483, a suppressor in colorectal cancer that hampers cell proliferation and migration [ 35 ]. Moreover, TRAF1 plays a role in the mobility and M1 polarization of macrophages, a process mediated by TNFSF9/TRAF1/p-AKT/IL-1β signaling in response to F. nucleatum AI-2 [ 36 ]. This comprehensive analysis elucidates the multifaceted roles of these genes in CRC, providing valuable insights for future research and potential therapeutic strategies.

In the field of CRC research, there has been a growing interest in developing various risk assessment models. Zhang et al. focused on crucial lysosome-related genes integral to CRC, thereby establishing a corresponding risk signature [ 37 ]. Similarly, Han and colleagues delved into adipogenesis-associated genes, creating a prognostic model while illuminating the immunogenomic landscape of CRC [ 38 ]. In another vein, Huang et al. explored genes with prognostic significance from the angle of fatty acid metabolism, suggesting their potential relevance in immunotherapy strategies [ 39 ]. These studies collectively enhance the accuracy of CRC prognosis predictions through diverse methodologies, demonstrating their models’ effectiveness. Conventional bulk RNA-seq methods provide an overall gene expression profile at the tissue level, yet fail to discriminate the transcriptomic diversity of various cell types and their proportions within these tissues. In a novel approach, our study integrates scRNA-seq data, which offers precise cell type identification and high-resolution expression profiles, with bulk RNA-seq data. This integration allowed us to pinpoint specific TAM-M2 prognostic biomarkers for CRC. To our knowledge, this is the first study to integrate these two data types for the purpose of identifying TAM-M2-related genes and developing a risk signature in CRC. The identification of these signature genes opens new avenues for deeper comprehension and exploration in CRC research. Moreover, our discovered prognostic signature holds promise for enhancing the clinical management of CRC patients.

Despite the encouraging outcomes of our research, it is important to acknowledge some inherent limitations. Firstly, the reliance on data sourced from public databases could potentially limit the representativeness of our findings across the broader patient demographic. Secondly, our conclusions are predominantly based on bioinformatics analyses, underscoring the need for further validation through detailed studies of molecular mechanisms. Consequently, there is a compelling necessity for more comprehensive investigations to elucidate the intricate roles of TAM-M2-related genes in CRC.

In our study, we combined scRNA-seq and bulk RNA-seq analyses to unveil the diverse landscape of CRC at both the individual cell and tissue levels, culminating in the development of the TAMM2RS. This approach sheds new light on the multifaceted nature of M2 macrophages and TAMs, contributing to a deeper understanding of the TME complexities. Furthermore, our research identifies potential therapeutic targets for CRC, offering novel avenues for treatment strategies.

Data availability

The datasets employed in this research are available in the GEO (GSE132465, https://www.ncbi.nlm.nih.gov/geo/) and TCGA (TCGA-CRC, https://portal.gdc.cancer.gov/) repositories. For additional information, correspondences can be addressed to the authors responsible.

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer statistics 2020: GLOBOCAN estimates of incidence and Mortality Worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209–49.

Article   PubMed   Google Scholar  

Punt CJA, Koopman M, Vermeulen L. From tumour heterogeneity to advances in precision treatment of colorectal cancer. Nat Rev Clin Oncol. 2017;14:235–46.

Article   CAS   PubMed   Google Scholar  

Goswami S, Anandhan S, Raychaudhuri D, Sharma P. Myeloid cell-targeted therapies for solid tumours. Nat Rev Immunol. 2023;23:106–20.

Murray PJ, Wynn TA. Protective and pathogenic functions of macrophage subsets. Nat Rev Immunol. 2011;11:723–37.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Sica A, Mantovani A. Macrophage plasticity and polarization: in vivo veritas. J Clin Invest. 2012;122:787–95.

Mosser DM, Edwards JP. Exploring the full spectrum of macrophage activation. Nat Rev Immunol. 2008;8:958–69.

Hwang B, Lee JH, Bang D. Single-cell RNA sequencing technologies and bioinformatics pipelines. Exp Mol Med. 2018;50:1–14.

Cheng S, Li Z, Gao R, Xing B, Gao Y, Yang Y, et al. A pan-cancer single-cell transcriptional atlas of tumor infiltrating myeloid cells. Cell. 2021;184:792–e80923.

Zhang L, Li Z, Skrzypczynska KM, Fang Q, Zhang W, O’Brien SA, et al. Single-cell analyses inform mechanisms of myeloid-targeted therapies in Colon cancer. Cell. 2020;181:442–e45929.

Ml LC-G. O, D V, I G. Much more than M1 and M2 macrophages, there are also CD169(+) and TCR(+) macrophages. Front Immunol. 2015;6.

Yin Y, Liu B, Cao Y, Yao S, Liu Y, Jin G, et al. Colorectal Cancer-derived small extracellular vesicles promote Tumor Immune Evasion by upregulating PD-L1 expression in Tumor-Associated macrophages. Adv Sci Weinh Baden-Wurtt Ger. 2022;9:2102620.

CAS   Google Scholar  

Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.

Article   PubMed   PubMed Central   Google Scholar  

Hao Y, Hao S, Andersen-Nissen E, Mauck WM, Zheng S, Butler A, et al. Integrated analysis of multimodal single-cell data. Cell. 2021;184:3573–e358729.

Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010;26:1572–3.

Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47.

Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinformatics. 2013;14:7.

Ito K, Murphy D. Application of ggplot2 to Pharmacometric Graphics. CPT Pharmacomet Syst Pharmacol. 2013;2:79.

Article   Google Scholar  

Yu G, Wang L-G, Han Y, He Q-Y. clusterProfiler: an R Package for comparing Biological themes among Gene clusters. OMICS J Integr Biol. 2012;16:284–7.

Article   CAS   Google Scholar  

Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102:15545–50.

Maeser D, Gruener RF, Huang RS. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform. 2021;22:bbab260.

Gao H, Zou Q, Ma L, Cai K, Sun Y, Lu L, et al. Unveiling mitophagy-mediated molecular heterogeneity and development of a risk signature model for colorectal cancer by integrated scRNA-seq and bulk RNA-seq analysis. Gastroenterol Rep. 2023;11:goad066.

DeNardo DG, Ruffell B. Macrophages as regulators of tumour immunity and immunotherapy. Nat Rev Immunol. 2019;19:369–82.

Sharma P, Hu-Lieskovan S, Wargo JA, Ribas A. Primary, adaptive, and Acquired Resistance to Cancer Immunotherapy. Cell. 2017;168:707–23.

Engblom C, Pfirschke C, Pittet MJ. The role of myeloid cells in cancer therapies. Nat Rev Cancer. 2016;16:447–62.

Zhang Y, Zhang Z. The history and advances in cancer immunotherapy: understanding the characteristics of tumor-infiltrating immune cells and their therapeutic implications. Cell Mol Immunol. 2020;17:807–21.

Gasser S, Lim LHK, Cheung FSG. The role of the tumour microenvironment in immunotherapy. Endocr Relat Cancer. 2017;24:T283–95.

Li Y, Chen Z, Han J, Ma X, Zheng X, Chen J. Functional and therapeutic significance of Tumor-Associated macrophages in Colorectal Cancer. Front Oncol. 2022;12:781233.

Rodriguez-Garcia A, Lynn RC, Poussin M, Eiva MA, Shaw LC, O’Connor RS, et al. CAR-T cell-mediated depletion of immunosuppressive tumor-associated macrophages promotes endogenous antitumor immunity and augments adoptive immunotherapy. Nat Commun. 2021;12:877.

Pyonteck SM, Akkari L, Schuhmacher AJ, Bowman RL, Sevenich L, Quail DF, et al. CSF-1R inhibition alters macrophage polarization and blocks glioma progression. Nat Med. 2013;19:1264–72.

Zhang J, Martins CR, Fansler ZB, Roemer KL, Kincaid EA, Gustafson KS, et al. DNA methylation in anal intraepithelial lesions and anal squamous cell carcinoma. Clin Cancer Res off J Am Assoc Cancer Res. 2005;11:6544–9.

Pack S-C, Kim H-R, Lim S-W, Kim H-Y, Ko J-Y, Lee K-S, et al. Usefulness of plasma epigenetic changes of five major genes involved in the pathogenesis of colorectal cancer. Int J Colorectal Dis. 2013;28:139–47.

Leong KJ, Beggs A, James J, Morton DG, Matthews GM, Bach SP. Biomarker-based treatment selection in early-stage rectal cancer to promote organ preservation. Br J Surg. 2014;101:1299–309.

Steinmann S, Kunze P, Hampel C, Eckstein M, Bertram Bramsen J, Muenzner JK, et al. DAPK1 loss triggers tumor invasion in colorectal tumor cells. Cell Death Dis. 2019;10:895.

Yuan W, Ji J, Shu Y, Chen J, Liu S, Wu L, et al. Downregulation of DAPK1 promotes the stemness of cancer stem cells and EMT process by activating ZEB1 in colorectal cancer. J Mol Med Berl Ger. 2019;97:89–102.

Niu Z-Y, Li W-L, Jiang D-L, Li Y-S, Xie X-J. Mir-483 inhibits colon cancer cell proliferation and migration by targeting TRAF1. Kaohsiung J Med Sci. 2018;34:479–86.

Wu J, Li K, Peng W, Li H, Li Q, Wang X, et al. Autoinducer-2 of Fusobacterium nucleatum promotes macrophage M1 polarization via TNFSF9/IL-1β signaling. Int Immunopharmacol. 2019;74:105724.

Zhang L, Yang J, Deng Y, Deng W, Li L. Computational profiling and prognostic modeling based on lysosome-related genes in colorectal cancer. Front Genet. 2023;14:1203035.

Han J, Li S, Zhan Q, Hu Y, Zhong C, Yang J, et al. Comprehensive characterization of adipogenesis-related genes in colorectal cancer for clinical significance and immunogenomic landscape analyses. Lipids Health Dis. 2023;22:217.

Huang X, Sun Y, Song J, Huang Y, Shi H, Qian A, et al. Prognostic value of fatty acid metabolism-related genes in colorectal cancer and their potential implications for immunotherapy. Front Immunol. 2023;14:1301452.

Download references

No funding was received for conducting this study.

Author information

Authors and affiliations.

Department of Gastroenterology Surgery, Shengli Oilfield Central Hospital, Dgongying, Shandong, P. R. China

Lujing Shi, Hongtun Mao & Jie Ma

You can also search for this author in PubMed   Google Scholar

Contributions

L.S. and J.M. conceived the study. L.S. and H.M. analyzed the data and wrote the initial manuscript. J.M. revised the manuscript. Every author has reviewed and given their approval to the final version of the manuscript.

Corresponding author

Correspondence to Jie Ma .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Conflict of interest.

The authors have no conflicts of interest to declare that are relevant to the content of this article.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary material 2, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Shi, L., Mao, H. & Ma, J. Integrated analysis of tumor-associated macrophages and M2 macrophages in CRC: unraveling molecular heterogeneity and developing a novel risk signature. BMC Med Genomics 17 , 145 (2024). https://doi.org/10.1186/s12920-024-01881-z

Download citation

Received : 22 January 2024

Accepted : 16 April 2024

Published : 27 May 2024

DOI : https://doi.org/10.1186/s12920-024-01881-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Tumor-associated macrophages
• M2 macrophages
• Colorectal cancer
• Risk signature

BMC Medical Genomics

ISSN: 1755-8794

• Submission enquiries: [email protected]
• General enquiries: [email protected]

Brain and Spinal Cord Tumor Research Results and Study Updates

See Advances in Brain and Spinal Cord Tumor Research for an overview of recent findings and progress, plus ongoing projects supported by NCI.

FDA has granted an accelerated approval to tovorafenib (Ojemda) for kids and teens who have low-grade glioma with changes in the BRAF gene. In a small clinical trial, the drug shrank or completely eliminated tumors in nearly 70% of patients.

The activity of 34 genes can accurately predict the aggressiveness of meningiomas, a new study shows. This gene expression signature may help oncologists select the best treatments for people with this common type of brain cancer than they can with current methods.

An NCI-supported study called OPTIMUM, part of the Cancer Moonshot, was launched to improve the care of people with brain tumors called low-grade glioma in part by bringing them into glioma-related research.

Treating craniopharyngioma often requires surgery, radiation therapy, or both. But results of a study suggest that, for many, combining the targeted therapies vemurafenib (Zelboraf) and cobimetinib (Cotellic) may substantially delay, or even eliminate, the need for these treatments.

In a large clinical trial, vorasidenib slowed the growth of low-grade gliomas that had mutations in the IDH1 or IDH2 genes. Vorasidenib is the first targeted drug developed specifically to treat brain tumors.

Researchers have found that the aggressive brain cancer glioblastoma can co-opt the formation of new synapses to fuel its own growth. This neural redirection also appears to play a role in the devastating cognitive decline seen in many people with glioblastoma.

Two companion studies have found different forms of some brain tumors, diffuse midline glioma and IDH-mutant glioma, become dependent for their survival on the production of chemicals called pyrimidines. Clinical trials are planned to test a drug that blocks pyrimidine synthesis in patients with gliomas.

The combination of dabrafenib (Tafinlar) and trametinib (Mekinist) shrank more brain tumors, kept the tumors at bay for longer, and caused fewer side effects than chemotherapy, trial results showed. The children all had glioma with a BRAF V600 mutation that could not be surgically removed or came back after surgery.

Two separate but complementary studies have identified a new way to classify meningioma, the most common type of brain tumor. The grouping system may help predict whether a patient’s tumor will grow back after treatment and identify new treatments.

A nanoparticle coating may help cancer drugs reach medulloblastoma tumors in the brain and make the treatment less toxic. Mice treated with nanoparticles containing palbociclib (Ibrance) and sapanisertib lived substantially longer than those treated with either drug alone.

A new test could potentially be used to identify children treated for medulloblastoma who are at high risk of their cancer returning. The test detects evidence of remaining cancer in DNA shed from medulloblastoma tumor cells into cerebrospinal fluid.

Standard radiation for medulloblastoma can cause long-term damage to a child’s developing brain. A new clinical trial suggests that the volume and dose of radiation could be safely tailored based on genetic features in the patient’s tumor.

In people with glioblastoma and other brain cancers, steroids appear to limit the effectiveness of immunotherapy drugs, a new study shows. The findings should influence how steroids are used to manage brain tumor symptoms, researchers said.

Results from two studies show that a liquid biopsy that analyzes DNA in blood accurately detected kidney cancer at early and more advanced stages and identified and classified different types of brain tumors.

A method that combines artificial intelligence with an advanced imaging technology can accurately diagnose brain tumors in fewer than 3 minutes during surgery, a new study shows. The approach can also accurately distinguish tumor from healthy tissue.

Glioblastoma cells sneak many copies of a key oncogene into circular pieces of DNA. In a new NCI-funded study, scientists found that the cells also slip several different genetic “on switches” into these DNA circles, helping to fuel the cancer’s growth.

Men and women with glioblastoma appear to respond differently to standard treatment. A new study identifies biological factors that might contribute to this sex difference.

A liquid biopsy blood test can detect DNA from brain tumors called diffuse midline gliomas, researchers have found. This minimally invasive test could be used to identify and follow molecular changes in children with these highly lethal brain tumors.

Despite continued efforts to develop new therapies for glioblastoma, none have been able to improve how long patients live appreciably. Despite some setbacks, researchers are hopeful that immunotherapy might be able to succeed where other therapies have not.

Progress against the brain cancer glioblastoma has been slow. Drs. Mark Gilbert and Terri Armstrong of NCI’s Neuro-Oncology Branch discuss why and what’s being done to change that.

Studies presented at the 2017 AACR annual meeting suggest that therapies which take advantage of the mutations in the IDH gene may be more effective than drugs that block it.

• Search Please fill out this field.
• Manage Your Subscription
• Give a Gift Subscription
• Newsletters
• Sweepstakes

Tattoos Increase Risk of Developing Lymphoma by 21%, New Study Finds

A new study revealed that getting a tattoo — regardless of size — increases the risk of developing lymphoma, a type of blood cancer

Getting a tattoo, regardless of size, increases the risk of developing lymphoma by 21%, according to a new study.

Researchers from Lund University in Sweden analyzed 11,905 participants — 2,938 of whom had lymphoma, a type of blood cancer, between ages 20 and 60. Participants who had lymphoma and participants who were part of the control group completed a questionnaire about tattoos.

The study — published May 21 in eClinicalMedicine — found that the risk of developing lymphoma was 21% higher among those who were tattooed. The risk of lymphoma was also highest in individuals who had their first tattoo less than two years prior. 

Additionally, researchers found that there is no evidence of an increased risk with larger sized tattoos.

The study found that the most common types of cancer were diffuse large B-cell lymphoma and follicular lymphoma.

“After taking into account other relevant factors, such as smoking and age, we found that the risk of developing lymphoma was 21 percent higher among those who were tattooed,” Dr. Christel Nielsen, study author and Lund University professor, said in a statement . “It is important to remember that lymphoma is a rare disease and that our results apply at the group level. The results now need to be verified and investigated further in other studies and such research is ongoing.”

Getty Images

Never miss a story — sign up for PEOPLE's free daily newsletter to stay up-to-date on the best of what PEOPLE has to offer, from celebrity news to compelling human interest stories.

“We already know that when the tattoo ink is injected into the skin, the body interprets this as something foreign that should not be there and the immune system is activated,” she continued. “A large part of the ink is transported away from the skin, to the lymph nodes where it is deposited.”

Researchers said they will now examine whether or not there is a link between tattoos and other forms of cancer and inflammatory diseases.

“People will likely want to continue to express their identity through tattoos, and therefore it is very important that we as a society can make sure that it is safe,” said Nielsen. “For the individual, it is good to know that tattoos can affect your health, and that you should turn to your health care provider if you experience symptoms that you believe could be related to your tattoo.”

According to the Mayo Clinic , tattoos can also make skin prone to infections, since they breach the skin's barrier. Oftentimes, people may have allergic reactions to the tattoo dyes, which can cause rashes. While not common, tattoos can also lead to MRI complications, because the pigments can interrupt the image's quality.

In August 2023, the Pew Research Center released data showing that 32% of adults have a tattoo. Among those, 22% have more than one tattoo on their body.

Related Articles

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts

Review Articles

A guide to artificial intelligence for cancer researchers

This Review provides an introductory guide to artificial intelligence (AI)-based tools for non-computational cancer researchers. Here, Perez-Lopez et al. outline the general principles of AI for image analysis, natural language processing and drug discovery, as well as how researchers can get started with each of them.

• Raquel Perez-Lopez
• Narmin Ghaffari Laleh
• Jakob Nikolas Kather

Cancer therapy with antibodies

In this Review, Paul et al. provide an overview of therapeutic antibodies as an important modality in cancer therapy today. They summarize the different approaches used by antibodies to target cancer cells including those of immune checkpoint inhibitors, bispecific antibodies and antibody–drug conjugates, as well as describing current strategies aimed at improving their efficacy and reducing toxicities.

• Maximilian F. Konig
• Shibin Zhou

Developmental origins shape the paediatric cancer genome

In this Review, Zhang and colleagues provide an overview of the molecular characteristics of paediatric cancer and highlight how these malignancies arise from developmental aberrations resulting in paediatric-specific cancer genomes that influence both the initiation and progression of cancer. Additionally, they discuss genetic vulnerabilities within these cancer genomes that present opportunities for therapeutic interventions.

• Xiaolong Chen
• Wentao Yang
• Jinghui Zhang

Combinatorial strategies to target RAS-driven cancers

In this Review, Cichowski and colleagues provide an overview of combinatorial strategies designed to treat RAS-driven cancers that are based on four concepts that include vertical pathway inhibition, co-targeting RAS and adaptive survival pathways, co-targeting downstream or converging pathways and capitalizing on other cancer-associated vulnerabilities.

• Naiara Perurena
• Karen Cichowski

Lymphatic vessels in the age of cancer immunotherapy

Tumour-associated lymphatic growth and remodelling were once viewed as a passive means by which cancer cells could regionally spread to lymph nodes. However, recent data point to an active and contrasting role for lymphatic vessels and their transport in antitumour immune surveillance. In this Review, Karakousi et al. provide a working framework to define this role for the lymphatic system in tumour progression and present avenues for its therapeutic manipulation to improve cancer immunotherapy.

• Triantafyllia Karakousi
• Tenny Mudianto
• Amanda W. Lund

Immunogenic cell death in cancer: targeting necroptosis to induce antitumour immunity

In this Review, Meier et al. discuss the molecular mechanisms of necroptosis, delineate how this form of immunogenic cell death activates antitumour immune responses and explore the opportunities and limitations of targeting necroptosis for anticancer therapy.

• Pascal Meier
• Arnaud J. Legrand

Protein lipidation in cancer: mechanisms, dysregulation and emerging drug targets

This Review provides an overview of the complexity and significance of protein lipidation in cancer, outlines how targeting protein lipidation pathways offer promising avenues for developing cancer treatments, and discusses the current state of drugs targeting these pathways.

• Edward W. Tate

Extrachromosomal DNA in cancer

Extrachromosomal DNA (ecDNA) is now accepted as a major contributor to cancer pathogenesis. In this Review, Yan, Mischel and Chang highlight the recent advancements in ecDNA research, providing new insights into the biogenesis and maintenance of ecDNA, as well as its role in altering gene expression and promoting tumour heterogeneity.

• Xiaowei Yan
• Paul Mischel
• Howard Chang

Beyond genetics: driving cancer with the tumour microenvironment behind the wheel

In their Review article, Fuchs and colleagues discuss how a single or a few mutations in adult cells can lead to invasive cancers without a high mutational burden, demonstrating that non-genetic factors induce the epigenetic changes necessary for tumorigenesis.

• Shaopeng Yuan
• Jorge Almagro
• Elaine Fuchs

Multiplex protein imaging in tumour biology

In this Review, de Souza et al. discuss how advances in the ability to image protein markers at high-plex, at single-cell and even subcellular resolution, are expanding our understanding of tumour biology and clinical outcomes, and outline the future promise of combining such multiplex protein imaging methods with other forms of spatial omics.

• Natalie de Souza
• Bernd Bodenmiller

Translating p53-based therapies for cancer into the clinic

Although p53 was once considered undruggable, in this Review, Peuget et al. discuss the progress made in targeting p53 as a form of cancer therapy with approaches ranging from restoration of mutant p53 function to inhibition of the negative regulator of p53, MDM2, as well as newer strategies, including p53-based mRNA vaccines and antibodies.

• Sylvain Peuget
• Xiaolei Zhou
• Galina Selivanova

Towards targeting transposable elements for cancer therapy

Transposable elements, also known as junk DNA, constitute nearly half of the human genome. This Review by Liang et al. discusses how tumours exploit these transposable elements during their evolution but also how they represent a vulnerability that could be targeted through immunotherapeutic approaches.

• Yonghao Liang

The role of CRAF in cancer progression: from molecular mechanisms to precision therapies

Although hyperactivation of BRAF has been well-established to drive tumour progression and drug resistance, the role of CRAF in cancer is becoming increasingly relevant. Here, Riaud et al. summarize the various oncogenic roles of CRAF and the potential for CRAF-targeted therapies to improve the clinical outcome for RAF1 altered tumours.

• Melody Riaud
• Jennifer Maxwell
• April A. N. Rose

Decoding the basis of histological variation in human cancer

In this Review, Sato and colleagues provide an overview of the clinicopathological and phenotypical impact of lineage commitment and plasticity during tumorigenesis and progression of human epithelial cancer and discuss the molecular mechanisms that underlie histological lineage transition.

• Masayuki Fujii
• Shigeki Sekine
• Toshiro Sato

How chemokines organize the tumour microenvironment

In this Review, Mempel et al. use our understanding of the physiological response programmes of the immune system to the more commonplace challenges it encounters as a framework to interpret observations of chemokine function in tumours. When viewed in this way, the design of more effective therapeutic interventions leveraging the chemokine system to recalibrate response patterns to cancer might be possible.

• Thorsten R. Mempel
• Julia K. Lill
• Lukas M. Altenburger

Fcγ receptors and immunomodulatory antibodies in cancer

Numerous immunomodulatory antibodies for cancer treatment have been developed following the discovery of negative regulators of antitumour immunity such as programmed cell death protein 1 (PD1) and cytotoxic T lymphocyte-associated antigen 4 (CTLA4). The efficacy of these antibodies is determined not only by their ability to block or engage their target but also by their interactions with Fcγ receptors (FcγRs). This Review outlines our current knowledge of these interactions and discusses how we can use this knowledge to generate more effective cancer immunotherapies in the future.

• Felipe Galvez-Cancino
• Alexander P. Simpson
• Sergio A. Quezada

The dynamic role of platelets in cancer progression and their therapeutic implications

Targeting platelets represents a promising approach to improve the therapeutic efficacy of chemotherapy and cancer immunotherapy. Here, Li and colleagues highlight the dynamic role of platelets in tumour development, progression, and response to therapy, and underscore the utility of tumour-educated platelets for precise tumour diagnosis and treatment.

• Guangjun Nie

Eph receptors and ephrins in cancer progression

This Review by Elena B. Pasquale outlines the current understanding of Eph receptor–ephrin signalling mechanisms in cancer progression and therapy resistance, and also details therapeutic strategies for targeting the Eph system as a novel cancer therapy and for improving the efficacy of conventional cancer therapies.

• Elena B. Pasquale

Deubiquitinases in cancer

Deubiquitinating enzymes (DUBs) function in opposition to E3 ubiquitin ligases by removing ubiquitin from substrates to control protein and organelle homeostasis and responses to cellular stimuli. In this Review, Dewson et al. describe the many associations of DUBs with the hallmarks of cancer, with a view to identifying those DUBs most likely to impact cancer-associated phenotypes if targeted with selective inhibition.

• Grant Dewson
• Pieter J. A. Eichhorn
• David Komander

Metabolic pathway analysis using stable isotopes in patients with cancer

Although tumour metabolism is well recognized as a key feature in cancer initiation and progression, little is known about metabolic reprogramming in patients. In this Review, Bartman et al. discuss stable-isotope tracing as a means to probe tumour metabolism in vivo and provide an overview of isotope labelling studies performed in patients with cancer.

• Caroline R. Bartman
• Brandon Faubert
• Ralph J. DeBerardinis

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies