Abstract
In recent years, the number and type of treatment options in advanced bladder cancer (BC) have been rapidly evolving. To select an effective therapy and spare unnecessary side effects, predictive biomarkers are urgently needed. As the host's anti-cancer immune response is by far the most effective system to impede malignant tumor growth, immune system-based biomarkers are promising. We have recently described altered proteasomal epitope processing as an effective immune escape mechanism to impair cytotoxic T-cell activity. By altering the neoantigens' characteristics through different proteasomal peptide cleavage induced by non-synonymous somatic mutations, the ability for T-cell activation was decreased ("processing escapes"). In the present study, we analyzed primary chemo-naïve tissue samples of 26 adjuvant platinum-treated urothelial BC patients using a targeted next-generation sequencing panel followed by the epitope determination of affected genes, a machine-learning based prediction of epitope processing and proteasomal cleavage and of HLA-affinity as well as immune activation. Immune infiltration (immunohistochemistries for CD8, granzyme B, CD45/LCA) was digitally quantified by a pathologist and clinico-pathological and survival data were collected. We detected 145 epitopes with characteristics of a processing escape associated with a higher number of CD8-positive but lower number of granzyme B-positive cells and no association with PD-L1-expression. In addition, a high prevalence of processing escapes was associated with unfavorable overall survival. Our data indicate the presence of processing escapes in advanced BC, potentially creating a tumor-promoting pro-inflammatory environment with lowered anti-cancerous activity and independence from PD-L1-expression. The data also need to be prospectively validated in BC treated with immune therapy.