Tumor Immunity Microenvironment-based classifications of bladder cancer for enhancing cancer immunotherapy

Abstract

Background: Bladder cancer is composed by a mass of heterogenetic characteristics, immunotherapy is a potential way to save the life of bladder cancer patients, but only benefit to about 20% patients. Methods and materials: A total of 4003 bladder cancer patients from 19 cohorts was enrolled in this study, collecting the clinical information and mRNA expression profile. The unsupervised non-negative matrix factorization (NMF) and nearest template prediction (NTP) algorithm was used to divide the patients to immune activated, immune exhausted and non-immune class. Verified gene sets of signatures were used to illustrate the characteristic of immunophenotypes. Clinical and genetic features were compared in different immunophenotypes. Results: We identified the immune class and non-immune classes in from TCGA-BLCA cohort. The 150 top different expression genes between these two classes was extracted as the input profile for the reappearing of the classification in the other 19 cohorts. As to the activated and exhausted subgroups, a stromal activation signature was conducted by NTP algorithm. Patients in the immune classes shown the highly enriched signatures of immunocytes, while the exhausted subgroup also shown an increased signature of TITR, WNT/TGF-β, TGF-β1 activated, and C-ECM signatures. Patients in the immune activated shown a lower CNA burden, better overall survival, and favorable response to anti-PD-1 therapy. Conclusion: We defined and validated a novel classifier among the 4003 bladder cancer patients. Anti-PD-1 immunotherapy could benefit more for the patients belong to immune activated subgroup, while ICB therapy plus TGF-β inhibitor or EP300 inhibitor might be more effectiveness for patients in immune exhausted subgroup.

Multi-omics consensus ensemble refines the classification of muscle-invasive bladder cancer with stratified prognosis, tumour microenvironment and distinct sensitivity to frontline therapies

“…The treatment effect of immune checkpoint inhibitors was measured by two nonproportional hazards statistical approaches, namely, restricted mean survival (RMS) and long-term survival inference, by using the R packages "survRM2" and "ComparisonSurv", respectively [37]. A random forest (RF) predictive model was developed by the R package "varSelRF", technical details of which have been described previously [38]. The receiver operating characteristic (ROC) curves were used to assess the model predictive performance.…”