Part Ⅰ Identification And Validation Of An Immune-related Gene Pairs Signature For Three Urologic Cancers

Abstract

Reliable biomarkers are needed to recognize urologic cancer patients at high risk for recurrence. In this study, we built a novel immune-related gene pairs signature to simultaneously predict recurrence for three urologic cancers. We gathered 14 publicly available gene expression profiles including bladder, prostate, and kidney cancer. A total of 2,700 samples were classified into the training set (n = 1,622) and validation set (n = 1,078). The 25 immune-related gene pairs signature consisting of 41 unique genes was developed by the least absolute shrinkage and selection operator regression analysis and Cox regression model. The signature stratified patients into high- and low-risk groups with significantly different relapse-free survival in the meta-training set and its subpopulations, and was an independent prognostic factor of urologic cancers. This signature showed a robust ability in the meta-validation and multiple independent validation cohorts. Immune and inflammatory response, chemotaxis, and cytokine activity were enriched with genes relevant to the signature. A significantly higher infiltration level of M1 macrophages was found in the high-risk group versus the low-risk group. In conclusion, our signature is a promising prognostic biomarker for predicting relapse-free survival in patients with urologic cancer.

Keywords

Urologic cancers; Biomarkers; Immune-related gene pairs; Recurrence-free survival; Prognostic biomarker; Cistanche's benefits.

Click here to gain more information about what are the effects of Cistanche

Introduction

Bladder cancer, prostate cancer, and kidney cancer are the main tumors in the urinary system, and nearly 2.4 million new cases are diagnosed each year [1]. The advance of targeted therapy and neoadjuvant therapy has prolonged the survival of patients [2–4]. However, numerous patients suffer relapses even after complete surgical resection [5–7], and their prognoses are still not optimistic. A reliable prognostic biomarker that could identify patients with a higher risk for relapse and select patients who have responses to therapies would be valuable for the management of urologic cancers.

Gene-expression signatures have been identified for survival stratification of bladder cancer [8, 9], prostate cancer [10, 11], and kidney cancer [12, 13]. However, most biomarkers have not been translated to clinical practice due to the over-fitting of training datasets or lack of sufficient validation [14]. A chance to develop more reliable prognostic biomarkers has been brought by sufficient large-scale public gene expression datasets [15, 16]. However, it is a challenge to integrate data derived from different platforms. The traditional method has made it difficult to normalize different datasets, given the technical biases and biological heterogeneity of multiple platforms [17, 18]. New methods based on the relative ranking of gene expression levels have been used to eliminate the requirement for data preprocessing, and have attained robust results in many applications [19–21].

Increasing evidence has indicated that the tumor immune microenvironment is correlated with the formation and progression of the three main urologic tumors [22–24]. The immune checkpoint molecules, such as programmed cell death 1 (PD-1), PD-1 ligand 1 (PD-L1), and cytotoxic T-lymphocyte associated antigen 4 (CTLA-4), have demonstrated a remarkable, durable response in bladder cancer [25, 26], prostate cancer [27, 28] and kidney cancer [29, 30]. The biomarkers related to the tumor immune microenvironment may thus have the potential as prognostic markers of urologic cancers.

As is well-known, bladder cancer, prostate cancer, and kidney cancer are closely related anatomically and result from similar insults that promote tumor formation [31–33]. Therefore, we have developed in this study a signature based on immune-related gene pairs (IRGPs) to simultaneously predict the recurrence of bladder cancer, prostate cancer, and kidney cancer.

Cistanche tubulosa

Materials and methods

1. Study design and datasets

We comprehensively analyzed 14 gene expression profiles in three urologic tumors of the bladder, prostate, and kidney cancer, including seven microarray datasets and seven RNA-Seq datasets (Supplementary Figure 1). The accession numbers, platforms, and sample sizes of these gene expression profiles are shown in Supplementary Table 1. RNA-Seq data were downloaded from UCSC Xena and the International Cancer Genome Consortium (ICGC). Microarray data were downloaded from the Gene Expression Omnibus (GEO). Cohorts from The Cancer Genome Atlas (TCGA) were used as the training set, and other datasets were used as the validation set. Only patients with complete survival information were included. We also excluded patients who had received radiation therapy, neoadjuvant therapy, and targeted molecular therapy in all independent training cohorts. In total, 2,700 cases were included in our study. Our project was approved by Chongqing Medical University’s Ethical Review Committee.

2. Gene expression data processing

The publicly available datasets from GEO were firstly normalized using the normalizeBetweenArrays function as implemented in the ‘limma’ package and then were further log-transformed. Normalization methods were not used in TCGA and ICGC cohorts.

3. Identification of specific IRGPs for prognosis prediction

We downloaded immune-related genes (IRGs) from the ImmPort database accessed on 3/3/2021. 2,483 unique IRGs, constituting 17 categories, including cytokines, cytokine receptors, antigen processing, presentation pathways, interleukins, natural killer cell cytotoxicity, TGFb, and TNF family members. Only IRGs measured by all platforms with a median absolute deviation > 0.5 in all independent training sets were chosen. The score for each IRGP was generated by pairwise comparisons of the gene expression level in a certain sample of profiles. The IRGPs score was defined as 1 if the expression level of IRG 1 was larger than IRG 2; otherwise, the IRGPs score was set as 0 [34]. After removing IRGPs with constant values in any individual dataset, the remaining IRGPs were further analyzed.

4. Construction of the immune-related gene pairs index (IRGPI) for prognosis prediction

Prognostic IRGPs were selected based on the following steps. Firstly, the predictive ability of each IRGP to predict patients’ relapse-free survival (RFS) was evaluated by using the Cox regression model in the meta-training dataset, and the IRGPs with a p-value < 0.05 were selected as initial candidate markers. Secondly, the least absolute shrinkage and selection operator (LASSO) analysis was utilized to further filter out some less informative IRGPs. The tuning parameter was determined by the expected generalization error estimated from 10-fold cross-validation. To improve robustness, we randomly split the full meta-training dataset into new training and testing datasets with a 2:1 ratio and repeated the random split scheme 30 times to stabilize the IRGPs selection procedure. The LASSO model was then applied to the 30 training sets, and those IRGPs with a frequency > 15 were selected. Finally, the multivariate Cox regression model was used to build the IRGPs-based prediction model and generate the IRGPI for all samples. The patients were classified into low and high-immune risk groups using the median IRGPI value.

Cistanche extract

5. Evaluation and validation of the IRGPI

The prognostic value of the IRGPI was evaluated in the meta-training and independent training sets and was further verified in the meta-validation and multiple independent validation sets. The log-rank test and time-dependent receiver operating characteristic (ROC) curves were adopted to assess the prognostic accuracy of the IRGPI. We combined IRGPI with clinical factors of age, gender, and tumor stage in multivariate Cox analyses. Age (>60) was transformed into 1, and age (<60) was transformed into 0. Stages III and IV were transformed into 1, and stage I and II were transformed into 0.

6. Profiling of infiltrating immune cells

CIBERSORT characterizes immune cell composition by using bulk-tumor gene expression profiles [35]. It inferred the relative proportions of 22 types of infiltrating immune cells based on the reference gene expression values (LM22) [35]. In this study, the proportions of 22 infiltrating immune cells were determined by using the R package ‘CIBERSORT’. The perm was set at 1,000, and cases with a CIBERSORT output p-value < 0.05 were selected for further analysis. The Wilcoxon rank sum test was utilized to compare differences in immune cell subtypes between the high and low IRGPI groups.

7. Gene ontology (GO) analysis

The R package ‘clusterProfiler’ was utilized to conduct GO enrichment analysis of the genes related to the IRGPI in the meta-training cohort. The BenjaminiHochberg-adjusted p-value < 0.05 (false discovery rate, FDR) was used as the threshold to determine significance.

8. Construction and evaluation of the nomogram

A nomogram was constructed to establish a quantitative approach for RFS prediction in the meta-training cohort based on the IRGPI and clinical factors, which was further verified in the meta-validation cohorts. A point was calculated for each factor, and the total points of all factors were then obtained for the estimation of RFS rates at 1, 3, 5, and 10 years. The calibration plots were then drawn to evaluate the reliability of the nomogram.

Standardized Cistanche

9. Statistical analysis

The R package ‘survival’, ‘glmnet’, ‘sure miner’, ‘timeROC’, ‘rms’, ‘CIBERSORT’, and ‘clusterProfiler’ were used to construct the Cox regression model, LASSO model, Kaplan-Meier curve, time-dependent ROC curve, nomogram, immune cell composition computation and GO analysis. The association of IRGPI score with RFS was analyzed by log-rank test. The Cox regression model was adopted to perform a multivariate analysis of the association of IRGPI with RFS. A two-sided p-value < 0.05 was considered statistically significant for all tests. All statistical analyses were conducted using R (version 4.0.2).

References

1. 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; 71:209–49.

2. Seiler R, Oo HZ, Tortora D, Clausen TM, Wang CK, Kumar G, Pereira MA, Ørum-Madsen MS, Agerbæk MØ, Gustavsson T, Nordmaj MA, Rich JR, Lallous N, et al. An Oncofetal Glycosaminoglycan Modification Provides Therapeutic Access to Cisplatin-resistant Bladder Cancer. Eur Urol. 2017; 72:142–50.

3. Maroto P, Solsona E, Gallardo E, Mellado B, Morote J, Arranz JÁ, Gómez-Veiga F, Unda M, Clement MÁ, Alcaraz A. Expert opinion on first-line therapy in the treatment of castration-resistant prostate cancer. Crit Rev Oncol Hematol. 2016; 100:127–36.

4. Fu Q, Xu L, Wang Y, Jiang Q, Liu Z, Zhang J, Zhou Q, Zeng H, Tong S, Wang T, Qi Y, Hu B, Fu H, et al. Tumor-associated Macrophage-derived Interleukin-23 Interlinks Kidney Cancer Glutamine Addiction with Immune Evasion. Eur Urol. 2019; 75:752–63.

5. Zhao Z, Wang Y, Zhang JJ, Huang XY. Fascin Inhibitors Decrease Cell Migration and Adhesion While Increase Overall Survival of Mice Bearing Bladder Cancers. Cancers (Basel). 2021; 13:2698.

6. Peltomaa AI, Raittinen P, Talala K, Taari K, Tammela TL, Auvinen A, Murtola TJ. Prostate cancer prognosis after initiation of androgen deprivation therapy among statin users. A population-based cohort study. Prostate Cancer Prostatic Dis. 2021; 24:917–24.

7. Tian M, Wang T, Wang P. Development and Clinical Validation of a Seven-Gene Prognostic Signature Based on Multiple Machine Learning Algorithms in Kidney Cancer. Cell Transplant. 2021; 30:963689720969176.

8. Zhang C, Gou X, He W, Yang H, Yin H. A glycolysis-based 4-mRNA signature correlates with the prognosis and cell cycle process in patients with bladder cancer. Cancer Cell Int. 2020; 20:177.

9. Chen S, Zhang N, Shao J, Wang T, Wang X. A novel gene signature combination improves the prediction of overall survival in urinary bladder cancer. J Cancer. 2019; 10:5744–53.

10. Xu Z, Xu L, Liu L, Li H, Jin J, Peng M, Huang Y, Xiao H, Li Y, Guan H. A Glycolysis-Related Five-Gene Signature Predicts Biochemical Recurrence-Free Survival in Patients With Prostate Adenocarcinoma. Front Oncol. 2021; 11:625452.

11. Nguyen HT, Xue H, Firlej V, Ponty Y, Gallopin M, Gauthier D. Reference-free transcriptome signatures for prostate cancer prognosis. BMC Cancer. 2021; 21:394.

12. Lu J, Qian C, Ji Y, Bao Q, Lu B. Gene Signature Associated With Bromodomain Genes Predicts the Prognosis of Kidney Renal Clear Cell Carcinoma. Front Genet. 2021; 12:643935.

13. Wu Y, Wei X, Feng H, Hu B, Liu B, Luan Y, Ruan Y, Liu X, Liu Z, Liu J, Wang T. Integrated Analysis to Identify a Redox-Related Prognostic Signature for Clear Cell Renal Cell Carcinoma. Oxid Med Cell Longev. 2021; 2021:6648093.

14. Shi R, Bao X, Unger K, Sun J, Lu S, Manapov F, Wang X, Belka C, Li M. Identification and validation of hypoxia derived gene signatures to predict clinical outcomes and therapeutic responses in stage I lung adenocarcinoma patients. Theranostics. 2021; 11:5061–76.

15. Brendel M, Barthel H, van Eimeren T, Marek K, Beyer L, Song M, Palleis C, Gehmeyr M, Fietzek U, Respondek G, Sauerbeck J, Nitschmann A, Zach C, et al. Assessment of 18F-PI-2620 as a Biomarker in Progressive Supranuclear Palsy. JAMA Neurol. 2020; 77:1408–19.

16. Lee ES, Jung SY, Kim JY, Kim JJ, Yoo TK, Kim YG, Lee KS, Lee ES, Kim EK, Min JW, Han W, Noh DY, Moon HG. Identifying the potential long-term survivors among breast cancer patients with distant metastasis. Ann Oncol. 2016; 27:828–33.

17. Franks JM, Cai G, Whitfield ML. Feature-specific quantile normalization enables cross-platform classification of molecular subtypes using gene expression data. Bioinformatics. 2018; 34:1868–74.

18. Shen Y, Rahman M, Piccolo SR, Gusenleitner D, ElChaar NN, Cheng L, Monti S, Bild AH, Johnson WE. ASSIGN context-specific genomic profiling of multiple heterogeneous biological pathways. Bioinformatics. 2015; 31:1745–53.

19. Wu J, Zhao Y, Zhang J, Wu Q, Wang W. Development and validation of an immune-related gene pairs signature in colorectal cancer. Oncoimmunology. 2019; 8:1596715.

20. Xu F, Zhan X, Zheng X, Xu H, Li Y, Huang X, Lin L, Chen Y. A signature of immune-related gene pairs predicts oncologic outcomes and response to immunotherapy in lung adenocarcinoma. Genomics. 2020; 112:4675–83.

21. Wang S, Xu X. An Immune-Related Gene Pairs Signature for Predicting Survival in Glioblastoma. Front Oncol. 2021; 11:564960.

22. Pfannstiel C, Strissel PL, Chiappinelli KB, Sikic D, Wach S, Wirtz RM, Wullweber A, Taubert H, Breyer J, Otto W, Worst T, Burger M, Wullich B, et al, and BRIDGE Consortium, Germany. The Tumor Immune Microenvironment Drives a Prognostic Relevance That Correlates with Bladder Cancer Subtypes. Cancer Immunol Res. 2019; 7:923–38.

23. Long X, Hou H, Wang X, Liu S, Diao T, Lai S, Hu M, Zhang S, Liu M, Zhang H. Immune signature driven by ADT-induced immune microenvironment remodeling in prostate cancer is correlated with recurrence-free survival and immune infiltration. Cell Death Dis. 2020; 11:779.

24. Vuong L, Kotecha RR, Voss MH, Hakimi AA. Tumor Microenvironment Dynamics in Clear-Cell Renal Cell Carcinoma. Cancer Discov. 2019; 9:1349–57.

25. Maio M, Blank C, Necchi A, Di Giacomo AM, Ibrahim R, Lahn M, Fox BA, Bell RB, Tortora G, Eggermont AM. Neoadjuvant immunotherapy is reshaping cancer management across multiple tumor types: The future is now! Eur J Cancer. 2021; 152:155–64.

26. Kates M, Nirschl TR, Baras AS, Sopko NA, Hahn NM, Su X, Zhang J, Kochel CM, Choi W, McConkey DJ, Drake CG, Bivalacqua TJ. Combined Next-generation Sequencing and Flow Cytometry Analysis for an AntiPD-L1 Partial Responder over Time: An Exploration of Mechanisms of PD-L1 Activity and Resistance in Bladder Cancer. Eur Urol Oncol. 2021; 4:117–20.

27. Sharma P, Pachynski RK, Narayan V, Fléchon A, Gravis G, Galsky MD, Mohammed H, Patnaik A, Subudhi SK, Ciprotti M, Simsek B, Saci A, Hu Y, et al. Nivolumab Plus Ipilimumab for Metastatic Castration-Resistant Prostate Cancer: Preliminary Analysis of Patients in the CheckMate 650 Trial. Cancer Cell. 2020; 38:489–99.e3.

28. Ruiz de Porras V, Pardo JC, Notario L, Etxaniz O, Font A. Immune Checkpoint Inhibitors: A Promising Treatment Option for Metastatic Castration-Resistant Prostate Cancer? Int J Mol Sci. 2021; 22:4712.

29. Kamai T, Kijima T, Tsuzuki T, Nukui A, Abe H, Arai K, Yoshida KI. Increased expression of adenosine 2A receptors in metastatic renal cell carcinoma is associated with poorer response to anti-vascular endothelial growth factor agents and anti-PD-1/AntiCTLA4 antibodies and shorter survival. Cancer Immunol Immunother. 2021; 70:2009–21.

30. Chu G, Xu T, Zhu G, Liu S, Niu H, Zhang M. Identification of a Novel Protein-Based Signature to Improve Prognosis Prediction in Renal Clear Cell Carcinoma. Front Mol Biosci. 2021; 8:623120.

31. Gorin MA, Verdone JE, van der Toom E, Bivalacqua TJ, Allaf ME, Pienta KJ. Circulating tumor cells as biomarkers of prostate, bladder, and kidney cancer. Nat Rev Urol. 2017; 14:90–7.

32. Tölle A, Suhail S, Jung M, Jung K, Stephan C. Fatty acid binding proteins (FABPs) in prostate, bladder, and kidney cancer cell lines and the use of IL-FABP as survival predictor in patients with renal cell carcinoma. BMC Cancer. 2011; 11:302.

33. Jungwirth N, Haeberle L, Schrott KM, Wullich B, Krause FS. Serotonin is used as a prognostic marker of urological tumors. World J Urol. 2008; 26:499–504.

34. Li B, Cui Y, Diehn M, Li R. Development and Validation of an Individualized Immune Prognostic Signature in Early-Stage Nonsquamous Non-Small Cell Lung Cancer. JAMA Oncol. 2017; 3:1529–37.

35. Newman AM, Liu CL, Green MR, Gentles AJ, Feng W, Xu Y, Hoang CD, Diehn M, Alizadeh AA. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods. 2015; 12:453–7.

Biao Xie 1, Kangjie Li 1, Hong Zhang 1, Guichuan Lai 1, Dapeng Li 2, Xiaoni Zhong 1.

1. Department of Biostatistics, School of Public Health and Management, Chongqing Medical University, Chongqing, China

2. Institute of Hepatology, National Clinical Research Center for Infectious Disease, Shenzhen Third People’s Hospital, Shenzhen, China