Characterization of Intrinsic Radiosensitivity and Immune Cell Infiltrates in Bladder Cancer

Abstract

Bladder cancer is managed by surgery, systemic therapy and radiation (RT) depending on clinicopathologic features and physician/patient choice. Various RT doses are used in bladder preservation, which are guided by normal tissue tolerance, not intrinsic biology. Emerging data supports the use of immunotherapy (IO) in bladder cancer, but optimal patient selection and therapy sequencing are not defined. This study characterizes the immune cell infiltrate (ICI) and intrinsic RT-sensitivity of bladder cancer in an institutional cohort and 4 public datasets with gene expression data to understand the relationship between RT-sensitivity and the ICI. With an IRB-approved prospective tissue collection protocol, 223 bladder cancer patients profiled with Affymetrix Hu-RSTA microarray chips (Affymetrix; Santa Clara, CA) were identified. Also, 4 public bladder cancer datasets from GEO with various clinicopathologic features were analyzed: GSE48075 (n = 142) and GSE32984 (n = 308) without neoadjuvant chemotherapy (NAC); GSE87304 (n = 305) with NAC; GSE120736 (primary/recurrent tumors; n = 145). In each dataset, tumors were grouped as non-invasive (NMIBC) or invasive (MIBC) by T category. Intrinsic tumor RT-sensitivity was inferred by the radiosensitivity index (RSI), a cancer-type agnostic gene expression algorithm. RSI values were split at the median and grouped as RSIlo (i.e. RT-sensitive) or RSIhi. Tumor ICI composition was inferred by 2 algorithms: CIBERSORT (22 distinct ICI types) and XCell (64 distinct immune and stromal cell types). Groups were compared with the Mann-Whitney U test. Multiple correction testing was performed. The institutional cohort median RSI is 0.42 (range 0.02-0.86). The median RSI values for the public datasets are: [GSE48075] 0.50 (range 0.22-0.69), [GSE32894] 0.23 (range -0.40-0.45), [GSE87304] 0.33 (range -0.07-0.48) and [GSE120736] 0.40 (range 0.06-0.72). No difference was seen in the median RSI between primary and recurrent bladder tumors (p = 0.38). Only one dataset had a difference in the median RSI between NMIBC and MIBC (GSE32894, p <0.001), with a higher RSI in MIBC. Next, using CIBERSORT and XCell, we found that resting or naïve ICIs (e.g. NK, CD8+ T, CD4+ T, B, CD4+ memory T cells) were higher in NMIBC in 3 out of 4 datasets (q <0.05). Several ICIs (e.g. activated NK, M1/M2 macrophages, CD4+ memory T cells) were higher in MIBC in all datasets (q <0.05). In 3 out of 4 datasets and the institutional cohort, RSIlo vs RSIhi tumors had higher CD4+ T, M1 macrophages, activated (NK, dendritic, CD4+ memory T cells) and CD8+ T cell types (q <0.05). Bladder tumors have a range of RT-sensitivities in this analysis, implying uniform RT dosing may lead to over and under treatment. Also, NMIBC has higher resting or naïve ICIs compared to MIBC and RT-sensitive tumors are enriched with specific effector ICIs. Integration of ICI and RT-sensitivity data may assist with optimizing patient selection for certain IO and RT combinations.