Conference 2021 Pre-Recorded Video
Project title
Distinct tumor microenvironments derived from aggregated genomic data suggests personalized treatment strategies for cutaneous melanoma
Authors and Affiliations
Michael Balas1, Alexander Landry, Zsolt Zador1
1. Division of Neurosurgery, Department of Surgery, University of Toronto, Toronto, Ontario, Canada
Abstract
Background
Immune checkpoint inhibitors (ICIs) represent the most recent development and are now a standard of care, however the treatment response is variable raising the problem of patient selection. Tissue microenvironment has been shown critical to shaping tumor biology and treatment response in multiple tumor types, thereby representing the next frontier in diagnostics, treatment selection and prognostics. With evolving data analytics and expanding molecular databases we can define tumor subgroups with homogenous molecular features translating to shared outcomes and potentially, treatment response. We therefore hypothesize that aggregating multimodal genomic data can identify subgroups for cutaneous melanoma, the most aggressive skin cancer, with distinct microenvironments and markers of treatment response.
Methods
We applied the well-established technique of similarity network fusion, which enables the aggregation of high dimensional datasets into a single matrix. We aggregated gene expression and matching DNA methylation data of 471 untreated melanomas from The Cancer Genome Atlas (TCGA) and applied spectral clustering to identify homogenous subgroups. Abundance of tumor microenvironment components (e.g. stroma), in addition to population specific gene expression and programs predictive of treatment response to ICIs, were derived from previously established gene expression signatures.
Results
We discovered two subgroups (consisting of 325 and 146 samples) from the aggregated molecular data, which were mainly dictated by methylation profile. Tumors in the smaller, “stroma-rich” subgroup had lower purity with more stromal and immune components compared to the larger, “high-purity” subgroup. Immune fractions were higher in the “stroma-rich” subgroup for M2 macrophages, CD8+ and regulatory T cells. We further analyzed CD8+ T cell biology/interactions via correlations with treatment response signatures and abundance of cellular components in the tumor microenvironment.
Conclusions
We identify two states in melanoma microenvironments that are mostly determined by methylation profiles. The “stroma-rich” subgroup carries multiple features of a suppressed immune microenvironment. Our results may be translatable into early stratification of melanoma treatment response based on methylation profile and enable personalized implementation of immune modulation therapy.