Intratumour heterogeneity and genomic instability as a marker of patient prognosis

▸ Andor N, Graham TA, Jansen M, et al. Pan-cancer analysis of the extent and consequences of tumour heterogeneity. Nat Med 2016;22:105–13.

Intratumour heterogeneity is one of the current cancer research hot topics. The observation that tumours are composed of a large number of diverse clones has provided a potential explanation as to why cancers are hard to treat—the likelihood that a therapeutic resistant clone can evolve is high. However, there is no consensus on the consequence of tumour heterogeneity on the prognosis of patients because previous studies have relied on single sample sequencing of moderate depth. In this article, Andor and colleagues use mixture-separation algorithms such as EXPANDS (‘Expanding Ploidy and Allele frequency on NesteD populationS’), to model the cellular prevalence of each single nucleotide variant as a copy number-dependent probability distribution. Another algorithm (PyClone) was also used to validate EXPANDS data through the statistical inference of clonal population structure in cancers. The authors obtained whole exome sequencing data from the cancer genome atlas database from 11 165 tumour/normal paired samples across 12 cancer types. These included thyroid, prostrate, lung, gastric and melanoma cancers among others. The data revealed that 86% of cancers had two or more detectable clones and correlated with patient age at diagnosis. They reported a range in the number of clones between cancers (thyroid cancer containing the fewest, melanoma the greatest). Single clones were only detected in a subset of cancers (14% overall). Furthermore, the role of driver genes in clonal expansion within cancers was evaluated and found to be the expected size and revealed some cancer driver genes associated with specific cancers. However, driver gene mutations that appeared in smaller clones were associated with poor survival. Interestingly, poor survival was associated with cancers that exhibited >2 clones, particularly in patients with copy number variants detected in …