Pan-cancer copy number analysis identifies optimized size thresholds and co-occurrence models for individualized risk stratification

Abstract

Abstract Chromosome instability leading to aneuploidy and accumulation of copy number gains or losses is a hallmark of cancer. Copy number alteration (CNA) signatures are increasingly used for cancer risk stratification, but size thresholds for defining CNAs across cancers are variable and the biological and clinical implications of CNA size heterogeneity and co-occurrence are incompletely understood. Here we analyze CNA and clinical data from 691 meningiomas and 10,383 tumors from The Cancer Genome Atlas to develop cancer- and chromosome-specific size-dependent CNA and CNA co-occurrence models to predict tumor control and overall survival. Our results shed light on technical considerations for biomarker development and reveal prognostic CNAs with optimized size thresholds and co-occurrence patterns that refine risk stratification across a diversity of cancer types. These data suggest that consideration of CNA size, focality, number, and co-occurrence can be used to identify biomarkers of aggressive tumor behavior that may be useful for individualized risk stratification.