Cancers evolve through a process of Darwinian evolution driven by heritable variation at the cellular level. A plethora of somatic alterations occurring within the DNA of a cell leads to an aberrant phenotype characterized by increased proliferation and growth, and the ability to evade cell death, senescence, and immune surveillance. Characterizing the mutational landscape of cancer has been a major aim of the past 50 years of research. Initial efforts were able to identify numerous mutations occurring in specific genes. With the development of high-throughput sequencing technologies, it has been possible to perform whole-genome sequencing (WGS) to catalogue changes occurring within any given cancer genome. The Pan-Cancer Analysis of Whole Genomes (PCAWG) Consortium was created with the goal of identifying somatic events driving common cancer development across all tumor types. A team of researchers, counting more than 700 groups, is divided across 16 working groups analyzing 2,658 genomes from 38 tumor types. Their discoveries are illustrated in 22 articles published across Nature journals.
The findings presented in their flagship article, supervised by Drs. Peter Campbell and Gad Getz (among others), highlight shared abnormalities between cancer types and the importance of using WGS data, compared to targeted approaches, to analyze cancer genomes. After carefully benchmarking the pipeline to call single-nucleotide variants (SNVs), indels, copy number alterations (CNAs), and structural variants (SVs), the researchers show that 95 percent of cases had a driver mutation. On average there are 4.6 driver events per tumor, for coding point mutations the average was 2.6 per tumor. Non–coding point mutations were less frequent. A coding driver mutation and/or a CNA was identified in 76 percent and 73 percent of tumors, respectively. In B-cell lymphomas, there was a higher prevalence of SNV driving events compared to SVs. Not surprisingly, the guardian of the genome, the tumor suppressor TP53, was the most frequently altered gene, with both alleles mutated in 77 percent of TP53-mutant cases (usually a somatic point mutation on one allele and a somatic deletion of the other allele).
Across samples initially identified not to have any driving mutations, the authors identified numerous technical reasons for missing driver mutations. For example, failure of bioinformatic algorithms affected 35 myeloproliferative neoplasms where JAK2 V617F point mutation was missed, likely owing to contamination by JAK2 V617F clones in matched normal or “panels of normal” samples. This highlights the challenge of applying WGS to study certain blood cancers. Nonetheless, even after accounting for technical shortfalls, 5.3 percent of cases still had no identifiable somatically altered driver. The authors conclude that, despite the scale and sophistication of the analyses, cancer driver discovery is not yet complete.
The advantage of using WGS compared to simply assessing the coding genome (i.e., targeted exome sequencing) is the opportunity, firstly, to identify SVs (CNA and chromosomal rearrangement) and clusters of mutations (hotspots), and secondly, to investigate whether they are early or late events in cancer development. The authors show that chromoplexy (a double-strand break–induced chromosomal rearrangement) was found in 17.8 percent of cases, while kataegis (generation of a hypermutated hotspot with locally clustered nucleotide substitutions) was found in 60.5 percent, and chromothripsis (chromosomal shuttering randomly stitched back together) was present in 22.3 percent. The latter was also shown to be associated with TP53 mutation and an early event occurring at the clonal level, suggesting a prominent driving role in those tumors. This finding is significant given that in the clinical setting, a mutational panel assessing a few hundred genes is often used instead of WGS, which will miss such chromosomal clustering of mutations and SVs. For instance, extreme kataegis burden (>30 foci) was identified in B-cell non-Hodgkin lymphoma. WGS also allows the analysis of telomeric sequences to gain further insights into telomere maintenance — a well-known mechanism to escape senescence and gain replicative potential. The authors found that 16 percent of the tumors have somatic mutations altering either the function or the expression of either TERT, ATRX, or DAXX genes, which are all known to be responsible for telomere maintenance. However, an association with alterations in the retinoblastoma 1 (RB1) gene was also observed, possibly representing a novel pathway for telomere length preservation. Finally, the study also describes many novel germline genetic variants that determine rates and patterns of somatic mutation.
In Brief
The PCAWG consortium reported the integrated genetic analysis of more than 2,600 genomes, describing common genetic features across 38 tumor types. This is a truly remarkable achievement, requiring computational analysis at a staggering scale, achieved through cloud computing. Only a relatively small number of somatic driver events are typically necessary to trigger cancer development, and the majority of these are either SNVs within the coding genome or CNAs, with a smaller number of somatic driver mutations occurring in non-coding regions of the genome and frequent occurrence of clustered mutations and SVs. These heterogeneous somatic mutation events can only be comprehensively characterized by using WGS. While undoubtedly some driver mutations remain to be discovered, the occurrence of certain tumors without any detectable somatic mutations despite WGS also raises the possibility of other mechanisms of tumor development that may play a role. For instance, the involvement of tumor microenvironment is also a common feature among cancer types, but it may not be driven by somatic mutation in tumor cells, as demonstrated in numerous blood cancer model systems. The next herculean effort, as highlighted by the authors, will be to focus on translating WGS-based cancer genomics into clinical practice. The same goal is at the heart of other projects such as the 100,000 Genome England initiative. The long-term ambition is for WGS to become a routine clinical assessment on cancer biopsies in order to apply precision medicine. As the authors acknowledge, the dramatic patient-to-patient variability and even cell-to-cell heterogeneity within a tumor will necessitate “knowledge banks” comprising tens of thousands of patients with integration of detailed clinical data, which will require large international consortia. This PCAWG resource article, despite its impressive scale, represents only the first early steps in this process.
Competing Interests
Dr. Orlando and Dr. Mead indicated no relevant conflicts of interest.
