Chromosome arm aneuploidies shape tumour evolution and drug response

Abstract

Chromosome arm aneuploidies (CAAs) are pervasive in cancers. However, how they affect cancer development, prognosis and treatment remains largely unknown. Here, we analyse CAA profiles of 23,427 tumours, identifying aspects of tumour evolution including probable orders in which CAAs occur and CAAs predicting tissue-specific metastasis. Both haematological and solid cancers initially gain chromosome arms, while only solid cancers subsequently preferentially lose multiple arms. 72 CAAs and 88 synergistically co-occurring CAA pairs multivariately predict good or poor survival for 58% of 6977 patients, with negligible impact of whole-genome doubling. Additionally, machine learning identifies 31 CAAs that robustly alter response to 56 chemotherapeutic drugs across cell lines representing 17 cancer types. We also uncover 1024 potential synthetic lethal pharmacogenomic interactions. Notably, in predicting drug response, CAAs substantially outperform mutations and focal deletions/amplifications combined. Thus, CAAs predict cancer prognosis, shape tumour evolution, metastasis and drug response, and may advance precision oncology.

Complexities of pharmacogenomic interactions in cancer

“…In fact, we demonstrated that CAAs considerably outperform mutations and focal CNAs in predicting drug response. 8 Additionally, this demonstrates the complexity of some pharmacogenomic interactions in cancer cells.…”

“…Recently, we utilized an expanded version of the extensive pharmacogenomic cell line resource referred to above 5 (cancerrxgene.org) to assess whether additional complex forms of cancer pharmacogenomic interactions exist. 8 Chromosome arm aneuploidies (CAAs) are common in human tumors and on average affect 25% of the cancer genome, compared to focal CNAs affecting 10%. 9 Thus, we reasoned that simultaneous copy number gains or losses of genes encoded on chromosome arms, due to CAAs, could comprise a novel form of complex pharmacogenomic interactions.…”

“…Indeed, an unbiased machine learning approach that included both well-established cancer functional events (CFEs, i.e., common cancer gene mutations and focal CNAs; n = 710) and CAAs (n = 78), as well as IC 50 values (half of the maximum inhibitory concentration of a drug) pertaining to 453 drugs and 988 cell lines, identified 365 robust CFE-or CAA-drug interactions. 8 This involved approximately equal numbers of drug sensitivity (n = 181) and resistance interactions (n = 184) ( Figure 1a). However, the number of events involving copy number loss is considerably higher for those involved in drug resistance.…”

“…For instance, of all 64 identified CAA-drug interactions, only two can be explained by focal CNAs or any combination of two focal CNAs affecting the same chromosome arm. 8 Thus, CAAs represent a new form of complex cancer pharmacogenomic interactions.…”

“…Altogether, we identified 1024 and 89 of such interactions, respectively. 8 Meta-analysis of these shows that focal copy number loss and chromosome arm gains dominate involvement in such sensitivity and resistance interactions, respectively (Figure 1b). Notably, the ratios between events involving copy number gain and loss are radically different from single-event interactions (compare the gain:loss ratios 'G:L' in Figure 1a,b).…”