Evaluating and validating immune effector dysfunction as a biomarker in bladder cancer.

Abstract

e16601 Background: Checkpoint inhibitors (CPI) have transformed bladder cancer treatment, yet response rates remain limited. Molecular subtypes provide prognostic information, but response to immunotherapy varies within subtypes. T-cell exhaustion and senescence represent biologically distinct mechanisms of immune effector dysfunction (IED) that may limit CPI efficacy. A 172-gene transcriptomic IED score (IED172) was previously derived and validated in acute myeloid leukemia and melanoma. We hypothesized that IED172 would demonstrate cross-tumor validity as a prognostic biomarker in bladder cancer. Methods: We analyzed bulk-RNA sequencing data from three independent cohorts: TCGA-BLCA, IMvigor210 (phase 2 single arm trial of atezolizumab in metastatic urothelial carcinoma), and an institutional Baylor College of Medicine (BCM) cohort with matched proteomic profiling. Patient-specific transcriptomic and proteomic IED172 scores were derived from the mean expression of 172 genes and 122 constituent proteins, respectively. An IED Prognostic Index (IED-PI), a weighted combination of LASSO-selected genes, was used to stratify patients into high and low groups based on median IED-PI for survival analysis. Statistical analyses included Kruskal-Wallis, Kaplan-Meier, univariate and multivariate Cox regression. Results: We analyzed 406 patients from TCGA-BLCA, 348 from IMvigor210, and 60 from BCM. Transcriptomic IED172 scores varied significantly across molecular subtypes in TCGA-BLCA (p = 7.97x10⁻²²), BCM cohort (p = 0.014) and IMvigor210 (p = 3.2x10 -15 ). In TCGA-BLCA, luminal infiltrated exhibited the highest median IED172 score (7.19), while luminal papillary tumors had the lowest (5.9). This pattern was reproduced in both the BCM cohort and IMvigor210. In BCM cohort, transcriptomic and proteomic IED172 scores were strongly correlated (R = 0.75; p = 1.4x10⁻⁶). In TCGA-BLCA, high IED-PI was associated with inferior median progression-free survival (18.2 vs 55.2 months; p < 0.0001) and overall survival (OS) (20.2 vs 92.9 months; p < 0.0001) compared to low IED-PI. In multivariate Cox regression adjusting for stage, grade, and molecular subtype, IED-PI remained independently prognostic (HR 4.93; 95% CI 3.19-7.63; p < 0.001). In IMvigor210, high IED-PI was associated with inferior median OS compared to low IED-PI (6.7 vs 16.0 months, respectively; p < 0.0001). Conclusions: IED172 demonstrates cross-tumor validity in bladder cancer, stratifying tumors by molecular subtype with orthogonal transcriptomic-proteomic validation. The consistent prognostic significance across both CPI-treated (IMvigor210) and non-CPI-treated (TCGA-BLCA) cohorts suggests IED-PI identifies patients with inherently aggressive disease. Establishing predictive value for CPI response would require demonstrating a treatment-by-biomarker interaction in randomized data comparing CPI to non-CPI treatment.