Omic characterization of melanoma for the identification of potential biomarkers in response to immune checkpoint inhibitors

Abstract

Cutaneous Malignant Melanoma (CMM) is a rapidly increasing malignancy for which immune check-point inhibition has proved efficient in inducing an anti-cancer immune response. However, not all pa-tients are able to benefit from these therapies and finding biomarkers able to appropriately predict response stands as a big challenge to overcome. Elucidating the molecular and immune landscape through omics technologies may unveil new potential predictors. For this purpose, computational anal-ysis of Whole-exome sequencing (WES) and RNA-sequencing (RNAseq) data was performed in a 24-patient melanoma cohort with responders and non-responders. Mutational landscape pointed to a po-tential implication of NRAS mutations in response while dismissing Tumor Mutation Burden (TMB) role. MCPCounter deconvolution of immune cell populations showed no differences in quantities of infiltra-tion in pre-treatment samples, while suggested infiltration after treatment. However, differential expres-sion analysis through DESeq2 implied a more enabling immunogenic landscape in pre-treatment re-sponders. In that sense, integration of expression and Copy Number Alteration (CNA) data through DIABLO allowed discrimination of responders and non-responders in pre-treatment samples through cytokine-related genes. Finally, 10-fold cross validation and selection modeling of differentially ex-pressed genes with generalized linear models evoked two new potential biomarkers: CD58 and PMEL.