Abstract

Immune checkpoint inhibition (ICI) has become a standard of care in the treatment of advanced melanoma, and is increasingly used in treatment-naïve patients. While ICI is particularly successful in some patients, more than half will not obtain a durable response. The identification of prognostic markers and the elucidation of resistance mechanisms remain key challenges in immune-oncology today. We initiated a single-center, non-interventional study to profile the transcriptomic landscape of melanoma and its tumor environment at single cell resolution, before and early during treatment with anti-PD1 based ICI.    

Drug-naive patients with advanced melanoma were prospectively recruited to undergo serial biopsy before initiation of therapy and right before administration of the second ICI cycle. Tissue was processed for single-cell RNA, VDJ and ATAC sequencing, using the Chromium 10X platform. In addition, embedded material was kept for multiplex immune histochemistry and where possible, spatial transcriptomics. Baseline demographic, clinical, histopathological and genetic information was collected and patients were stratified as responders (complete remission, partial remission) and non-responders (stable disease, progressive disease) according to RECIST1.1.

Using single cell RNA-sequencing we analyzed approximately 60K cells from 46 samples obtained from 22 patients. Extensive characterization of the tumor immune micro-environment (TIME) allowed the identification of 20 major immune cell types and study their association with response. In addition, we found extensive intratumoral heterogeneity (ITH) in the malignant subset. Leveraging transcriptomic profiling at single cell level, we identified several novel melanoma cell states, in addition to the previously described melanocytic  (MITFhigh)  and  mesenchymal-like  (MITFlow)  states. Some of these cell states were significantly associated with response to therapy.

Click here to access data