BackgroundBladder cancer has a poor clinical outcome because of its high aggressiveness. Basement membrane plays vital functions in tumor invasion and migration. Invasion and distant metastasis of cancer are facilitated by degradation of the basement membrane and extracellular matrix.MethodsTen machine learning methods were utilized to develop the basement membrane-related signature (MRS) using datasets from TCGA, GSE13507, GSE31684, GSE32984 and GSE48276. Three anti-PD1 or anti-CTLA4 datasets and several predicting scores were used to investigate the performance of MRS in predicting the immunotherapy benefits.ResultsA predicting model based on the Enet algorithm (alpha = 0.1) was chosen as the optimal MRS since it had the highest average C-index being 0.72. According to TCGA data, the MRS showed good performance in predicting bladder cancer patients' clinical outcomes, with area under curves of 0.744, 0.766 and 0.817 for 1, 3, and 5-year receiver operating characteristic curve, respectively. PD1 and CTLA4 immunophenoscopes were associated with a low MRS score, as well as a lower tumor immune dysfunction and exclusion score. As MRS score increased, immune-activated cells levels decreased, tumor immune dysfunction and exclusion score decreased, immune escape score decreased, intratumor heterogeneity score decreased, PD1&CTLA4 immunophenoscore increased, and tumor mutational burden score increased in bladder cancer, suggesting better immunotherapy benefits. Bladder cancer cases with high MRS score was correlated with higher cancer related hallmark scores, including NOTCH and glycolysis signaling.ConclusionA new MRS has been developed for bladder cancer, which could be used to predict prognosis and the success of immunotherapy.
Read full abstract