Differentiating bladder cancer (BLCA) tumor immune microenvironment (TME) profiles based on predicted intrinsic radiosensitivity (RS).

Abstract

e16622 Background: Radiation is an essential part of organ preservation therapy for muscle-invasive BLCA patients. There is increasing interest in use of immunotherapy in this setting and defining biomarkers of RS and immunity will be crucial to guide patient selection. We investigated whether predicted RS of BLCA patients can differentiate TME profiles based on RNA-seq. Methods: Twenty genomically-verified BLCA cell lines were profiled with log(TPM+1) normalized RNA-seq counts. Colony forming assays (CFA) for each cell line were done in triplicate with RS indices measured via area-under-the-cell-survival curve (AUC) and surviving fraction (SF) at varying doses (0-8Gy). The RS signature (RSS) was derived in two steps. First, NMF-decomposition of the RNA-seq data was done via CoGAPS v.3.6 run at k = 13 gene sets for 10,000 iterations then gene sets significantly correlated (p ≤ 0.05) with RS indices defined as an absolute Spearman’s rho >0.3 in at least 4/5 indices between AUC, SF2, SF4, SF6, and SF8 were combined. Secondly, the top 100-genes overexpressed by top-third most sensitive cell lines were identified by ComparativeMarkerSelection. The RSS contained genes shared between approaches and the RS-scores for the 408-patient TCGA-BLCA cohort were defined by averaging z-score normalized gene expression in the RSS. Patients were grouped as RS-high (top-quartile RS scores) and -low (bottom-quartile RS scores. Enrichment of the TME was assessed via EcoTyper algorithm and significant differences were defined a priori as: 1) Student’s t-test p ≤ 0.1 with Benjamini-Hochberg correction with an FDR of 0.2; 2) absolute correlation of r ≥ 0.4 between cell-state and RS-score; 3) ≥30% difference in enrichment between RS-high and RS-low. Results: Satisfactory CFA were obtained for 18 cell lines (range of RS indices: AUC = 3.40-5.94; SF2 = 0.32-0.75; SF4 = 0.09-0.42; SF6 = 0.02-0.19; SF8 = 0.002-0.09). CoGAPS analysis revealed 4 gene sets that were significantly correlated with RS totaling 1,997 genes. The RSS ultimately contained 78 genes and TCGA-BLCA RS scores ranged from -0.3 to 5.2. RS-high patients had significant enrichment for polymorphonuclear neutrophils (PMNs S02 state; 46-fold) and mast cells (S01 state; 77-fold). RS-low patients had significant enrichment for multiple cell-states including exhausted CD4 T-cells (S01 state; 1.31-fold; associated with CXCR6 & CTLA-4) and cancer-associated fibroblasts (S03 state, 1.33-fold; associated with POSTN & COL10A1). Conclusions: We demonstrate that predicted RS of BLCA patients is associated with a differential TME profile characterized by PMN inflammation and CD4 T-cell mediated immunosuppression among other cellular states. This potential link between RS and immunity may have broader implications for biomarker-guided immunotherapy as part of organ-preservation for BLCA patients although further validation is needed.