Cancer chromosomal instability (CIN) results from dynamic changes to chromosome number and structure. The resulting diversity in somatic copy number alterations (SCNA) may provide the variation necessary for cancer evolution. Multi-sample phasing and SCNA analysis of 1421 samples from 394 tumours across 24 cancer types revealed ongoing CIN resulting in pervasive SCNA heterogeneity. Parallel evolutionary events, causing disruption to the same genes, such as BCL9, ARNT/HIF1B, TERT and MYC, within separate subclones were present in 35% of tumours. Most recurrent losses occurred prior to whole genome doubling (WGD), a clonal event in 48% of tumours. However, loss of heterozygosity at the human leukocyte antigen locus and loss of 8p to a single haploid copy recurred at significant subclonal frequencies, even in WGD tumours, likely reflecting ongoing karyotype remodeling. Focal amplifications affecting 1q21 (BCL9, ARNT), 5p15.33 (TERT), 11q13.3 (CCND1), 19q12 (CCNE1) and 8q24.1 (MYC) were frequently subclonal and exhibited an illusion of clonality within single samples. Analysis of an independent series of 1024 metastatic samples revealed enrichment for 14 focal SCNAs in metastatic samples, including late gains of 8q24.1 (MYC) in clear cell renal carcinoma and 11q13.3 (CCND1) in HER2-positive breast cancer. CIN may enable ongoing selection of SCNAs, manifested as ordered events, often occurring in parallel, throughout tumour evolution.

Whole genome doubling (WGD) is a prevalent event in cancer, involving a doubling of the entire chromosome complement. However, despite its prevalence and prognostic relevance, the evolutionary selection pressures for WGD have not been investigated. Here, we combine evolutionary simulations with an analysis of cancer sequencing data to explore WGD during cancer evolution. Simulations suggest WGD can be selected to mitigate the irreversible, ratchet-like, accumulation of deleterious somatic alterations, provided they occur at a sufficiently high rate. Consistent with this, we observe an enrichment for WGD in tumor types with extensive loss of heterozygosity (LOH), including lung and triple negative breast cancers, and we find evidence for negative selection against homozygous loss of essential genes prior to, but not after, WGD. Finally, we demonstrate that LOH and temporal dissection of mutations can be exploited to identify novel tumor suppressor genes and to obtain a deeper characterization of known cancer genes.