Prognostic significance of B cell senescence-associated genes as risk markers in prostate adenocarcinoma

Introduction

Prostate adenocarcinoma (PRAD) is a common malignant tumor in men and has gradually become the second leading cause of cancer death in men (1). In recent years, with the development of cancer diagnosis and treatment strategies, the progress of PRAD has been alleviated, and the mortality rate of patients has shown a certain degree of decline (2). In particular, the wide application and development of prostate-specific antigen detection has enabled most PRAD patients to be diagnosed early in the development of the disease, which has also improved the cure rate of cancer to a certain extent (3). However, the accuracy of antigen-specific detection strategies depends on the representativeness and reliability of PRAD-related markers, so it is increasingly urgent to find safer and more accurate PRAD biomarkers in clinical practice.

Cancer progression is closely related to the regulation of tumor microenvironment (TME). The interaction of immune cells, mesenchymal cells, inflammatory factors and extracellular matrix molecules in TME often regulates a variety of cancer progression including PRAD through complex mechanisms (4). In particular, B cells in TME have the functions of producing antibodies, presenting antigens, secreting cytokines and activating T cells. They are usually used as positive immune regulatory factors and are directly related to the clinical outcomes of cancer patients (5,6). A number of studies have explored cancer diagnosis and prognostic markers based on the relevant characteristics of B cells. Wang et al. found that the expression level of B cell regulatory factor in the serum of patients with multiple myeloma was significantly higher than that in the healthy control group. The expression abundance of this factor was positively correlated with cancer progression, plasma cell infiltration in TME, and the expression of poor prognostic markers such as TNF-α and IL-6 (7). A clinical retrospective study identified a feature set consisting of B cell-related genes such as IFNγ, CD8, and CRMA, and successfully revealed the reliability of this set for predicting the efficacy of clinical immunotherapy in patients with triple-negative breast cancer, lung cancer, colorectal cancer, and melanoma (8). In view of the anti-cancer effect of B cells and their important regulatory role in TME, their aging characteristics usually mean poor prognosis of cancer and anti-tumor immunodeficiency (9). A study has shown that the immunosenescence of B cells leads to the susceptibility of the human body, especially the elderly population, to infectious diseases and the decrease of the response rate to vaccines (10). In the field of cancer, it is manifested as the high incidence of inflammatory response and the defect of anti-cancer immune response function (10). However, there is no relevant research on the mining of cancer prognostic markers based on B cell senescence, and the related mechanism of promoting cancer progression has not been elucidated, which is also the focus of this study.

In this study, genes related to B cell senescence characteristics were screened from differentially expressed genes in PRAD samples from the public database for univariate, least absolute shrinkage and selection operator (LASSO) and multivariate Cox regression analyses, and the characteristic genes related to PRAD progression and B immune senescence characteristics were identified to construct a prognostic model. Then, the accuracy of the model for evaluating the prognostic survival rate of PRAD patients was clarified by survival analysis. Through the analysis of immune microenvironment cell infiltration and immune score, the reliability of the model to predict the immune microenvironment characteristics of PRAD patients was evaluated. The correlation between risk score and PRAD mutation characteristics was revealed by mutation analysis of samples in high and low risk score groups. Finally, the potential chemotherapeutic drugs acting on PRAD treatment were elucidated based on drug sensitivity analysis. The molecular markers and prognostic risk assessment models screened in this study are expected to provide guidance for clinical diagnosis and targeted treatment of PRAD patients. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-724/rc).

Methods

Data download

The messenger RNA (mRNA) expression data of PRAD were downloaded from The Cancer Genome Atlas (TCGA) Program database (https://portal.gdc.cancer.gov/), including 52 healthy samples and 499 tumor samples, and the corresponding clinical information was used as the training set. Single nucleotide polymorphism (SNP) data were downloaded from the TCGA database. The GSE116918 gene expression data and corresponding clinical information of PRAD samples were downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/gds) as the validation set. ‘B cell senescence’ was retrieved from Genecards (https://www.genecards.org/) database to screen B cell senescence-related gene sets. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Prognostic model construction based on B cell senescence-associated genes

R package ‘edgeR’ was used to analyze the difference of mRNA data in TCGA dataset [|logfold change (FC)| >0.585, false discovery rate (FDR) <0.05]. The ‘upsetplot’ package was used to intersect differentially expressed genes with B cell senescence-related genes as candidate genes affecting PRAD progression and B cell senescence. The R package ‘survival’ was used to conduct univariate Cox regression analysis on candidate genes in order to identify genes that were substantially related to prognosis (P<0.05). The R package ‘glmnet’ was used to conduct LASSO Cox regression analysis on the genes that were screened using univariate Cox regression analysis to identify candidate feature genes. This was done to avoid the model from being over-fit. The candidate characteristic genes were screened using the two-way stepwise regression approach and a multivariate Cox regression model was constructed using the R package ‘survival’. The forest map was drawn using the R package ‘survminer’.

Prognostic model validation

The samples were grouped according to the median risk score of the prognostic model, and the R package “pheatmap” was used to draw the heat map of the model characteristics, as well as the distribution of risk scores and survival status. The R package ‘survival’ was used to conduct the Kaplan-Meier (K-M) analysis, and the R package ‘survminer’ was used to generate the survival curve. Receiver operating characteristic (ROC) analysis was performed using the R package ‘timeROC’, and curve was drawn to verify the 3-year survival rate of the patient’s prognosis, and the corresponding area under the curve (AUC) values were obtained and verified in the validation set.

Immune infiltration analysis

The 29 immune characteristics of TCGA-PRAD high and low risk score groups were analyzed by single sample gene set enrichment analysis (ssGSEA), using R package ‘GSVA’. R package ‘estimate’ was used to score the samples, and the ‘pheatmap’ package was used to draw the immune characteristics and score difference heat maps of high and low risk score groups. The difference of leukocyte antigen gene expression between high and low risk groups was analyzed by t-test, and the box plot of differential expression of leukocyte antigen between groups was drawn by R package ‘ggpubr’. In the follow-up study, Pearson correlation analysis was performed on CTLA4 checkpoint gene expression and model risk score to reveal the regulatory relationship between the two. On the last step, the ‘CIBERSORT’ program was employed to compare the immune cell infiltration rates of high-risk and low-risk groups.

Comparison of high-risk and low-risk groups based on mutational characteristics

The risScore scores were obtained in the construction of a multifactorial cox regression model, for which the sample was divided into high and low risk groups based on median scores. With the help of the R package ‘maftools’, we were able to visually inspect the mutation loci of high-frequency genetic mutations in both the high-risk and low-risk groups of TCGA-PRAD, allowing us to investigate the possibility of co-mutation or mutual exclusion.

Analysis of small molecule drug sensitivity in high and low risk groups

For the purpose of predicting the 50% inhibitory concentration (IC₅₀) in patients with drug-induced PRAD, the “oncoPredict” R package was utilized to pull down the cancer drug sensitivity Genomics (GDSC) gene expression profile along with the associated drug response data. Medications were categorized into high-risk and low-risk categories using a t-test.

Prognostic risk model independence verification and nomogram construction

We used the R package “survival” to conduct a one-way Cox regression analysis with clinical characteristics and risk scores in the training set. Then, we used the R package “forestplot” to draw the corresponding forest plot and see if the risk scores could be used as a distinct prognostic factor for patients with colorectal cancer. In order to find out whether the risk score may be a separate factor for CRC patients’ prognoses, the R program “forestplot” was used to plot the forest plot. The R package “rms” was used to construct a column chart combining clinical characteristics and risk scores, and the accuracy of the column chart was verified by plotting a calibration curve for assessing the 3-year survival rate of patients’ prognosis.

Results

Prognostic model construction of PRAD based on B cell senescence-related genes

In this study, 5,758 differentially expressed genes were obtained by differential analysis of TCGA-PRAD samples and adjacent samples (Figure 1A). In order to obtain the genes related to B cell senescence characteristics, 290 known B cell senescence-related genes were intersected with the above 5,911 genes, and a total of 56 genes related to PRAD progression and B cell senescence were obtained for subsequent analysis (Figure 1B). In the subsequent analysis, TCGA-PRAD was used as the training set. Through single factor cox regression analysis and LASSO Cox regression analysis, 9 candidate feature genes (Figure 1C,1D) were screened by the best penalty parameter λ. To build a prognostic risk assessment model for PRAD, 9 potential characteristic genes were subjected to multivariate Cox regression analysis. Ultimately, three characteristic genes were found (Figure 1E). The risk assessment model was constructed according to the following formula: Riskscore=−0.705∗WNT16+0.299∗INS+0.840∗BMP2.

Figure 1 Construction of prognostic risk model. (A) Volcano map of differentially expressed genes between TCGA samples and adjacent samples. The red dots indicate genes that are highly expressed, the dots that are green represent genes that are low-expressed, and the dots that are black indicate genes that are not significantly differentiated. (B) Screening of B cell senescence-related genes in differentially expressed genes of TCGA-PRAD samples relative to adjacent samples. The abscissa represents the number of differential genes. (C) Screening of the optimal penalty parameter (λ) of LASSO Cox regression model. (D) LASSO Cox regression analysis. (E) Multivariate Cox regression analysis. **, P<0.01; ***, P<0.0001. FC, fold change; FDR, false discovery rate; CI, confidence interval; WNT16, wnt family member 16; INS, insulin; BMP2, bone morphogenetic protein 2; AIC, Akaike Information Criterion; TCGA, The Cancer Genome Atlas; TCGA-PRAD, The Cancer Genome Atlas of Prostate adenocarcinoma; LASSO, least absolute shrinkage and selection operator.

Validation of PRAD prognostic risk assessment model

The model characteristic gene heat map showed that among the three PRAD prognostic characteristic genes, two characteristic genes INS and BMP2 were highly expressed in the high-risk group, while the characteristic gene WNT16 was lowly expressed in the high-risk group (Figure 2A). The high-risk group has a higher RiskScore, according to the risk score distribution map, and a higher death sample count, according to the survival status distribution map (Figure 2B,2C). Survival analysis showed worse prognostic survival in the high-risk group (Figure 2D). ROC analysis showed that the AUC of the model in the TCGA-PRAD training set to predict the 3-year survival rate of patients was 0.87 (Figure 2E), while the AUC of the model in the training set GSE116918 to predict the 3-year survival rate of patients was 0.73 (Figure 2F). Based on these findings, it seems that the three-feature gene model is a reliable predictor of PRAD patients’ outcomes.

Figure 2 Reliability verification of PRAD risk assessment model. (A) The expression level heat map of the characteristic genes in the high and low risk group model. (B) RiskScore distribution map of patients. Red indicates high-risk groups and green indicates low-risk groups. (C) Distribution map of survival status. (D) Survival analysis of high and low risk group samples. (E) ROC curve of training set samples. (F) ROC curve of validation set samples. INS, insulin; WNT16, wnt family member 16; BMP2, bone morphogenetic protein 2; TCGA, The Cancer Genome Atlas; ROC, receiver operating characteristic; AUC, area under the curve; PRAD, prostate adenocarcinoma.

Analysis of immune infiltration in high and low risk score groups

In order to reveal the association between the three characteristic genes used to construct the risk assessment model in this study and the immune regulation of patients, this study first based on ssGSEA analysis showed that 29 immune-related features were more abundant in the high-risk group (Figure 3A). At the same time, the high-risk group showed significantly higher immune scores than the low-risk group (Figure 3B). In terms of leukocyte antigen expression level, most leukocyte antigen genes such as HLA-DOB, HLA-DRB1, HLA-DMA, etc. showed higher differential expression levels in low-risk group samples (Figure 3C). In terms of differences in immune cell infiltration, most immune cells such as NK cells and M1 macrophages have higher infiltration abundance levels in the high-risk group than in the low-risk group (Figure 3D). The above results indicate that the high-risk group is in line with the hot-type immune characteristics of anti-tumor immunity, and is more likely to respond to immunotherapy.

Figure 3 Evaluation of tumor immune characteristics in high and low risk groups. (A) 29 immune characteristic immune characteristic heat maps in high and low risk score groups. (B) Violin plot of expression difference of immune score in high and low risk groups. (C) Boxplot of distribution difference of leukocyte antigen in high and low risk score groups. (D) Boxplot of infiltration differences of different immune cells in high and low risk groups. *, P<0.05; **, P<0.01; ***, P<0.0001. ns, the difference is not significant. aDC, activated DC; APC, antigen-presenting cell; CCR, chemokine receptor; DC, dendritic cell; HLA, human leukocyte antigen; MHC, major histocompatibility complex; pDC, plasmacytoid DC; IFN, interferon; NK, natural killer.

Analysis of mutation characteristics of high and low risk groups

By calculating the total number of mutations per million bases, the tumor mutation burden (TMB) score for each sample was determined. We looked at the top 20 genes for mutation frequency in both the high-risk and low-risk categories. Both the high-risk and low-risk groups had distinct mutation frequencies and different top 20 mutation genes. Within the low-risk category, the five most common genes with mutations were TTN, SPOP, KMT2D, FOXA1, and SYNE1. In the high-risk group, the five most common genes with mutations were TP53, TTN, SPOP, LRP1B, and MUC16. Compared to the low-risk group, the high-risk group had a much greater mutation frequency of TP53 and TTN (Figure 4A-4D). The low-risk group had a higher frequency of co-mutations and mutually exclusive mutations, according to the gene co-mutation and mutually exclusive mutation map (Figure 4E,4F).

Figure 4 Analysis of gene mutation frequency in high and low risk groups. (A,B) Statistical graph of high-frequency mutant genes, mutation sites, and mutation types in low-risk group (A) and high-risk group (B). (C,D) Waterfall map of the top 20 mutation frequency genes in the low-risk (C) and high-risk (D) groups. (E,F) Map of co-mutation and mutually exclusive mutation of top 20 genes with mutation frequency in low-risk group (E) and high-risk group (F). SNV, single nucleotide variants; TMB, tumor mutation burden.

Prediction of small molecule drugs acting on PRAD

To further understand the disparity in drug sensitivity between the model’s high-risk and low-risk groups, a t-test revealed that the high-risk group samples had higher IC₅₀ values for the chemotherapeutic medicines Camptothecin, Cisplatin, and WIKI4 (Figure 5A-5C). The results revealed that three chemotherapeutic drugs may be potential therapeutic drugs for PRAD samples in the low-risk group.

Figure 5 Analysis of differences in drug sensitivity between high-risk and low-risk groups. (A) IC₅₀ differences in the treatment of CC by Camptothecin between high-risk and low-risk groups. (B) IC₅₀ differences in the treatment of CC with Cisplatin between high-risk and low-risk groups. (C) IC₅₀ differences between high-risk and low-risk drug WIKI4 treatment CC. IC₅₀, half maximal inhibitory concentration; CC, colon cancer.

Prognostic risk model independence verification and nomogram construction

T, N, M, stage, and RiskScore were important in assessing patients’ prognoses, according to univariate cox regression analysis that included risk score and clinical information (Figure 6A). Multivariate Cox regression analysis also showed significant clinical implication of RiskScore (Figure 6B). The results demonstrate that the RiskScore developed in this research may be used as a standalone indicator of prognosis. The nomogram constructed by clinical information combined with risk score and the corresponding correction curve verified the good performance of the model in evaluating the 3-year survival rate of patients (Figure 6C,6D).

Figure 6 Construction and evaluation of nomogram. (A) Univariate cox regression analysis was performed for risk scores and clinical information. (B) Multivariate Cox regression scores were performed for risk scores and clinical information. (C) Nomogram of 3-year OS prediction in CC patients combined with 8-characteristic genetic risk scores and other clinical factors. (D) Calibration curve of the column chart for predicting 3-year survival. N, node; T, tumor; CI, confidence interval; OS, overall survival; CC, colon cancer.

Discussion

PRAD is a common male urinary system disease with a high mortality rate. According to statistics, there are about 1.4 million new PRAD patients and about 370,000 deaths worldwide each year (11). At present, the traditional therapy combined with surgery and radiotherapy and chemotherapy is commonly used in the clinical treatment of PRAD. The therapeutic effect is acceptable, but it will cause great body damage to patients, and the recurrence rate is high (12). The development of targeted therapy has improved this situation, but the development of this therapy is inseparable from safe and effective target screening (13). Therefore, it is urgent to explore the potential mechanism of PRAD progression and find new therapeutic targets. B cells are increasingly considered to be an important cell population in TME. However, there are inconsistent and even contradictory reports on the role of tumor-infiltrating B cells in cancer prognosis. This study is based on the immune aging characteristics of B cells as the starting point, based on the differential expression genes of PRAD relative to adjacent samples and the characteristic gene screening of known B cell aging-related genes, to construct a prognostic model, and to predict the immune characteristics, genomic mutation landscape and drug sensitivity of PRAD patients. The effectiveness of the reaction is verified to provide guidance for the diagnosis and treatment of PRAD diseases.

In this study, a three-gene construct PRAD prognostic model characterized by B-cell senescence was established by Cox regression analysis. Two characteristic genes INS and BMP2 were highly expressed in the high-risk group as prognostic risk factors, while the characteristic gene WNT16 was used as a prognostic protective factor. INS is usually present as an insulin-coding gene, and its expressed protein is a peptide hormone that plays a key role in regulating carbohydrate and lipid metabolism (14). Although the role of this gene in cancer has not been reported, cellular metabolic regulation may be the key mechanism for its role in PRAD in this study. Given that the important role of lipid metabolism in cancer is to regulate the signaling of T cells, B cells, and other signaling pathways in the immune microenvironment, which in turn affects cancer progression through the suppression of cellular immune functions (15,16). This is also consistent with the results of this study revealing the heterogeneity of immune microenvironment cell infiltration in the high and low risk score groups of PRAD. The regulation of bone morphogenetic protein-encoding gene BMP2 on cancer progression is related to the type of cancer. For example, it inhibits cancer cell proliferation by down-regulating the expression of oncogene EZH2 in gastric cancer (17). However, in endometrial cancer, non-small cell lung cancer and other cancers, the up-regulation of this gene promotes the migration phenotype of cancer cells through the activation of cancer-related signaling pathways (18) and the up-regulation of malignant markers such as c-kit. WNT16 is expressed in a variety of cells in the immune microenvironment, especially in cancer-associated fibroblasts, which is positively correlated with the malignant progression and chemotherapy resistance of gynecological cancers such as breast cancer (19). In addition, the gene can be used as a poor prognostic marker for colorectal cancer in a bioinformatics study, which can be used to predict the difference of macrophage infiltration in the immune microenvironment and the immune characteristics of the immunosuppressive microenvironment (20). However, at variance with the present study, which reveals WNT16 as a prognostic protective factor for PRAD, perhaps the differential regulation of cancer prognosis by this gene correlates with cancer type, which may be due to the heterogeneity of the immune infiltration microenvironment in patients with different cancer types. In summary, we identified genes in PRAD that were closely related to B cell senescence characteristics and prognosis to construct a risk assessment model, and verified its good predictive performance. The subsequent exploration of the molecular mechanism of these genes may provide new ideas for the diagnosis and treatment of PRAD.

In order to explore the mechanism of risk assessment model associated with the prognosis of PRAD patients, this study established the relationship between RiskScore and PRAD immune microenvironment. We found that most leukocyte antigen genes such as HLA-DOB, HLA-DRB1, HLA-DMA, etc. showed higher differential expression levels in low-risk group samples. HLA is the expression product of human major histocompatibility complex, which is the most complex polymorphic system in human body. In view of the abnormal expression of this family in the tissues of patients receiving immunotherapy, recent studies have focused on the immune regulation of this family (21). HLA-DOB is polymorphic in B cell-related cancers such as leukemia, which directly affects cancer progression (22). Mechanistically, the expression of HLA-DOB cooperates with a variety of cytokines in the anti-tumor immune response, especially the anti-tumor killing effect of activated cytotoxic T lymphocytes, which also explains the higher expression level of this factor in the low-risk group of this study (23,24). The presence of HLA-DRB1 can regulate the effect of anti-tumor immunotherapy mediated by CD4⁺ T cells (25). The allele of this protein is associated with programmed death 1/programmed cell death 1 ligand 1 (PD-1/PD-L1) immune checkpoint blocking reaction, and is associated with better prognosis in patients with non-small cell lung cancer, hepatocellular carcinoma, cervical cancer and other cancers (26-28). Similarly, the expression of HLA-DMA is also associated with T cell activation and immune checkpoint blockade in the immune microenvironment (29). The immunomodulatory effect of the HLA family verified the effectiveness of the construction of an immune-related gene prognosis model, and also confirmed the rationality of the risk assessment model in this study to predict tumor immune patterns.

With the goal to delve deeper into the connection between the prognostic model and PRAD’s drug sensitivity, this study explored the link between the model’s RiskScore and the IC₅₀ of genomic mutations and anticancer drugs. It also showed that the high-risk group had a significantly higher frequency of TTN mutations and resistance to chemotherapeutic drugs like Cisplatin, WIKI4, and Camptothecin. The high frequency of TTN mutation is associated with a good prognosis of many cancers, and the mutation can predict the immune response and inflammatory response of cancer (30,31). Maybe it is the regulatory effect of TTN mutation on immune microenvironment. The difference of drug sensitivity between high and low risk groups in this study is consistent with the conclusion that we reveal the heterogeneity of immune characteristics between samples. At the same time, the chemotherapeutic drugs excavated in this study are also closely related to gene interference and immune regulation. The combination of Camptothecin and immunotherapy can effectively overcome the limitations of single immunotherapy by promoting immunogenic cell death to stimulate a strong immune system, inhibiting tumor growth and improving the immunosuppressive tumor microenvironment (30,31). Cisplatin is a well-known cancer chemotherapy drug. It causes DNA damage by interfering with DNA repair, which is associated with genomic mutations in cancer cells and induces apoptosis of cancer cells (32). Related studies have shown that Cisplatin helps to increase the expression of PD-L1 in cancer tissues, increase the percentage of infiltration of CD8⁺ T cells and dendritic cells in the immune microenvironment, enhance the immune checkpoint blocking response, and optimize the therapeutic effect of cancer (33,34). WIKI4 is an inhibitor of the Wnt/β-catenin signaling pathway (35). The disorder of this pathway is related to the inhibition of T cytokines and the pro-inflammatory phenotype of T cells, which is associated with the malignant progression of various cancers (36). It can be seen that the B cell senescence-related genes excavated in this study can be used to evaluate the drug response of PRAD patients, which is potentially related to the difference of PRAD genome mutation and the heterogeneity of immune microenvironment, and provides guidance for clinical treatment of patients.

Conclusions

In summary, we identified three genes with prognostic value and B cell senescence characteristics by screening out differentially expressed genes between PRAD and paracancerous samples. The prognostic model was followed up and its clinical value was verified. In addition, immune characteristics analysis, gene mutation and drug sensitivity analysis found that high-risk patients showed differential TTN mutations, high immune microenvironment and high drug resistance. The risk model of this study is expected to provide guidance for clinical PRAD diagnosis and treatment strategies. But inevitably, there are some limitations in this study. This study mainly analyzed the retrospective data of the public database, and the prognosis information of the patients may not be comprehensive. Subsequently, clinical samples need to be collected for prospective studies to determine the clinical value of the model.

Acknowledgments

Funding: None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-724/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-724/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-724/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Siegel RL, Miller KD, Fuchs HE, et al. Cancer statistics, 2022. CA Cancer J Clin 2022;72:7-33. [Crossref] [PubMed]
2. Teo MY, Rathkopf DE, Kantoff P. Treatment of Advanced Prostate Cancer. Annu Rev Med 2019;70:479-99. [Crossref] [PubMed]
3. Negoita S, Mariotto A, Benard V, et al. Reply to annual report to the nation on the status of cancer, part II: Recent changes in prostate cancer trends and disease characteristics. Cancer 2019;125:318-9. [Crossref] [PubMed]
4. Kang HJ, Oh JH, Chun SM, et al. Immunogenomic landscape of hepatocellular carcinoma with immune cell stroma and EBV-positive tumor-infiltrating lymphocytes. J Hepatol 2019;71:91-103. [Crossref] [PubMed]
5. Wang Y, Liu J, Burrows PD, et al. B Cell Development and Maturation. Adv Exp Med Biol 2020;1254:1-22. [Crossref] [PubMed]
6. Franchina DG, Grusdat M, Brenner D. B-Cell Metabolic Remodeling and Cancer. Trends Cancer 2018;4:138-50. [Crossref] [PubMed]
7. Wang P, Qian L, Yuan X, et al. BlyS: a potential hallmark of multiple myeloma. Front Biosci (Landmark Ed) 2013;18:324-31. [Crossref] [PubMed]
8. Lundberg A, Li B, Li R. B cell-related gene signature and cancer immunotherapy response. Br J Cancer 2022;126:899-906. [Crossref] [PubMed]
9. Frasca D, Diaz A, Romero M, et al. Age effects on B cells and humoral immunity in humans. Ageing Res Rev 2011;10:330-5. [Crossref] [PubMed]
10. Frasca D, Diaz A, Romero M, et al. B Cell Immunosenescence. Annu Rev Cell Dev Biol 2020;36:551-74. [Crossref] [PubMed]
11. Sung H, Ferlay J, Siegel RL, 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. [Crossref] [PubMed]
12. Zhou J, Ding J, Qi J. Comparison of Typical Prostate Adenocarcinoma and Rare Histological Variant Prostate Cancer Showed Different Characteristics and Prognosis: A Surveillance, Epidemiology, and End Results Database Analysis. Eur Urol 2022;82:152-5. [Crossref] [PubMed]
13. Korsen JA, Kalidindi TM, Khitrov S, et al. Molecular Imaging of Neuroendocrine Prostate Cancer by Targeting Delta-Like Ligand 3. J Nucl Med 2022;63:1401-7. [Crossref] [PubMed]
14. Niswender KD. Basal insulin: physiology, pharmacology, and clinical implications. Postgrad Med 2011;123:17-26. [Crossref] [PubMed]
15. Lim SA, Su W, Chapman NM, et al. Lipid metabolism in T cell signaling and function. Nat Chem Biol 2022;18:470-81. [Crossref] [PubMed]
16. Zechner R, Zimmermann R, Eichmann TO, et al. FAT SIGNALS--lipases and lipolysis in lipid metabolism and signaling. Cell Metab 2012;15:279-91. [Crossref] [PubMed]
17. Chen Z, Yuan L, Li X, et al. BMP2 inhibits cell proliferation by downregulating EZH2 in gastric cancer. Cell Cycle 2022;21:2298-308. [Crossref] [PubMed]
18. Huang F, Cao Y, Wu G, et al. BMP2 signalling activation enhances bone metastases of non-small cell lung cancer. J Cell Mol Med 2020;24:10768-84. [Crossref] [PubMed]
19. Xiang L, Song Z, Rong G. Taxotere-induced WNT16 Expression in Carcinoma-Associated Fibroblasts Might Associate with Progression and Chemoresistance of Breast Cancer. Ann Clin Lab Sci 2020;50:205-12. [PubMed]
20. Yu S, Chen M, Xu L, et al. A senescence-based prognostic gene signature for colorectal cancer and identification of the role of SPP1-positive macrophages in tumor senescence. Front Immunol 2023;14:1175490. [Crossref] [PubMed]
21. Seliger B. Molecular mechanisms of HLA class I-mediated immune evasion of human tumors and their role in resistance to immunotherapies. HLA 2016;88:213-20. [Crossref] [PubMed]
22. van Lith M, van Ham M, Neefjes J. Novel polymorphisms in HLA-DOA and HLA-DOB in B-cell malignancies. Immunogenetics 2002;54:591-5. [Crossref] [PubMed]
23. Kang YJ, Zeng W, Song W, et al. Identification of human leucocyte antigen (HLA)-A*0201-restricted cytotoxic T lymphocyte epitopes derived from HLA-DOβ as a novel target for multiple myeloma. Br J Haematol 2013;163:343-51. [Crossref] [PubMed]
24. Wu G, Xiao G, Yan Y, et al. Bioinformatics analysis of the clinical significance of HLA class II in breast cancer. Medicine (Baltimore) 2022;101:e31071. [Crossref] [PubMed]
25. Lester DK, Burton C, Gardner A, et al. Fucosylation of HLA-DRB1 regulates CD4(+) T cell-mediated anti-melanoma immunity and enhances immunotherapy efficacy. Nat Cancer 2023;4:222-39. [Crossref] [PubMed]
26. Correale P, Saladino RE, Giannarelli D, et al. Distinctive germline expression of class I human leukocyte antigen (HLA) alleles and DRB1 heterozygosis predict the outcome of patients with non-small cell lung cancer receiving PD-1/PD-L1 immune checkpoint blockade. J Immunother Cancer 2020;8:e000733. [Crossref] [PubMed]
27. Repáraz D, Ruiz M, Llopiz D, et al. Neoantigens as potential vaccines in hepatocellular carcinoma. J Immunother Cancer 2022;10:e003978. [Crossref] [PubMed]
28. Zhang S, Wang H, Liu J, et al. RGS1 and related genes as potential targets for immunotherapy in cervical cancer: computational biology and experimental validation. J Transl Med 2022;20:334. [Crossref] [PubMed]
29. Liu D, Hofman P. Expression of NOTCH1, NOTCH4, HLA-DMA and HLA-DRA is synergistically associated with T cell exclusion, immune checkpoint blockade efficacy and recurrence risk in ER-negative breast cancer. Cell Oncol (Dordr) 2022;45:463-77. [Crossref] [PubMed]
30. Wang X, Duanmu J, Fu X, et al. Analyzing and validating the prognostic value and mechanism of colon cancer immune microenvironment. J Transl Med 2020;18:324. [Crossref] [PubMed]
31. Zhao L, Fan W, Luo K, et al. Construction of a TTN Mutation-Based Prognostic Model for Evaluating Immune Microenvironment, Cancer Stemness, and Outcomes of Colorectal Cancer Patients. Stem Cells Int 2023;2023:6079957. [Crossref] [PubMed]
32. Tchounwou PB, Dasari S, Noubissi FK, et al. Advances in Our Understanding of the Molecular Mechanisms of Action of Cisplatin in Cancer Therapy. J Exp Pharmacol 2021;13:303-28. [Crossref] [PubMed]
33. Fournel L, Wu Z, Stadler N, et al. Cisplatin increases PD-L1 expression and optimizes immune check-point blockade in non-small cell lung cancer. Cancer Lett 2019;464:5-14. [Crossref] [PubMed]
34. Fu G, Wu Y, Zhao G, et al. Activation of cGAS-STING Signal to Inhibit the Proliferation of Bladder Cancer: The Immune Effect of Cisplatin. Cells 2022;11:3011. [Crossref] [PubMed]
35. James RG, Davidson KC, Bosch KA, et al. WIKI4, a novel inhibitor of tankyrase and Wnt/ß-catenin signaling. PLoS One 2012;7:e50457. [Crossref] [PubMed]
36. Pai SG, Carneiro BA, Mota JM, et al. Wnt/beta-catenin pathway: modulating anticancer immune response. J Hematol Oncol 2017;10:101. [Crossref] [PubMed]