Mendelian randomizations with spatial gene networks reveal shared and distinct drivers of risk in major skin cancer types

Abstract

Skin cancer is the most common malignancy worldwide, comprising three major types: melanoma, basal cell carcinoma (BCC), and squamous cell carcinoma (SCC). A critical goal in advancing our understanding of skin cancer is to identify the shared and distinct biological mechanisms that drive the risk profiles for each skin cancer type. In this study, we integrated tissue-specific chromatin conformation and expression quantitative trait loci data (spatial eQTL) to construct gene regulatory networks (GRNs) for melanocytes, sun-exposed skin, not sun-exposed skin, and blood. These GRNs, which capture spatially regulated gene expressions, were used as instrumental variables along with GWAS summary statistics for melanoma, BCC and SCC, to infer causal relationships between specific gene expression changes and each type of skin cancer. These Mendelian randomization analyses identified 82, 62, and 125 causal genes for melanoma, BCC, and SCC, respectively, with many of these genes not evident from the GWAS data alone. Our analyses revealed distinct mechanisms for each skin cancer type: telomere maintenance and nevus pathways in melanoma, inherited immune traits in BCC, and p53 dysfunction in SCC. Notably, apoptosis and pigmentation emerged as shared biological processes across skin cancers. These findings provide new insights into the genetic drivers of three distinct skin cancers, highlighting new gene targets which can be used to increase diagnostic precision, as well as potential therapeutic targets.