NEW YORK – More than one-fifth of cancer cases appear to involve mutations that affect both the maternal and paternal allele at the same genome site, new research suggests. The findings, published in Nature Genetics on Thursday, argue against an "infinite sites model" that assumes mutations do not arise at the same site twice.
"Until now, these double mutations have been overlooked or were considered 'technical errors.' That is the reason why they were not included in genetic analyses of cancer," first author Jonas Demeulemeester, a postdoctoral researcher in cancer genomics and human genetics at the Francis Crick Institute and KU Leuven, said in a statement. "Our study shows that they are indeed important and that there is a need for methods capable of detecting them and to take their occurrence into account."
Using an analytical approach that assumed "uniform mutability" at diploid genome sites that can reliably be called, researchers at the Francis Crick Institute, KU Leuven, the European Bioinformatics Institute, and the Wellcome Sanger Institute first searched for biallelic mutation events across whole-genome sequences for 2,658 tumors from the Pan-Cancer Analysis of Whole Genomes, or PCAWG, effort.
"We focused on biallelic mutations, which become problematic when artificially treating genomes as haploid, hypothesizing these may be observed directly in bulk tumor genome sequencing data," the authors explained.
After flagging potential infinite site violations that may stem from biallelic mutations in a small subset of the tumors, the team went on to do similar analyses focused on sets of tumors from the same cancer types, focusing on tumors with mutational signatures that resembled one another.
Across the PCAWG tumor set, the investigators unearthed nearly 18,300 biallelic mutations in 559, or 21 percent, of the samples, including more than 12,900 biallelic changes classified as "parallel" mutations involving the same nucleotide on both copies of the altered site and 5,330 divergent mutations, in which mutations to each copy of a gene or surrounding sequence resulted in different nucleotide changes at a given allele.
"Although the probability that the same letter will mutate in both the maternal and paternal copy is statistically very small, it does occur in one in five tumors studied," Demeulemeester said, calling that prevalence "too large to be a coincidence."
Digging into the biallelic mutation profiles further, the researchers found ties to the mutation history of the tumors and their cancer type. They included simultaneous mutations affecting both copies of a gene at sites prone to damage by ultraviolet light in melanoma skin cancer. Those alterations resulted in distinct biallelic mutation patterns compared to those detected in other cancer types such as esophageal adenocarcinoma or colorectal adenocarcinoma, they reported.
These and other findings "highlight the need for accurate models of mutation rates and tumor evolution, as well as their inference from sequencing data," the authors suggested. Potential biallelic mutations need to be taken into account, they added, when calling variants and modeling tumor evolution.
"If not correctly identified, biallelic mutations confound variant interpretation, ranging from driver inference to subclonal clustering and timing analyses, as well as phylogenetic inference," the authors wrote, adding that "at-scale detection of biallelic mutations affords an intimate look at the mutational processes operative in cells, such as hotspots, hypermutable motifs, and the molecular mechanisms of DNA damage and repair."