It takes a village to interpret multigene panels

see Improving performance of multigene panels for genomic analysis of cancer predisposition

The use of multiple information sources and expert reviewers can greatly reduce the uncertainty involved in interpretation of multigene panels, according to a new study. A clinical team from the University of Washington evaluated nearly 1,500 consecutive patients referred for multigene panels that assess breast or colorectal cancer risk. More than 80% of the study subjects had a personal history of cancer. Following International Agency for Research on Cancer (IARC) guidelines, the research team used evolutionary conservation, known and predicted variant consequences, and personal and family cancer history to classify each variant. Among breast and colorectal cancer patients, 13% had variants considered actionable, as did 4% of cancer-free subjects. In contrast to previously reported results of multigene testing that included rates of variables of unknown significance (VUS) up to 88%, this study reduced that number to 7.5%. Investigators attribute the low rate of VUS to a two-step process in which multiple experts first independently evaluated primary calls for all variants, flagging any that were potentially pathogenic. In the second step, reviewers met to discuss all flagged variants and developed a consensus classification for each one. The results indicate that multigene testing need not be overwhelmed by reports of VUS. The authors conclude that for the foreseeable future, medical judgment by experts will be necessary to minimize the number of VUS reported to patients and caregivers. —Karyn Hede, News Editor

Diving into evolutionary relationships to improve mutation prediction

see page Establishing the precise evolutionary history of a gene improves prediction of disease-causing missense mutations

The sifting process of evolution can help predict the effect of a given mutation, but only if used judiciously. Not all families of related genes help predict deleterious effects. In this issue, Adebali et al. show that, by carefully analyzing the evolutionary history of genes involved in Niemann-Pick disease type C (NP-C), they could improve the predictive ability of algorithms that categorize single amino acid substitutions. The proof-of-concept study used a computational approach to analyze the evolutionary history of the NPC1 gene and pinpoint evolutionary events that most likely affected its function. An algorithm built on the evolutionary understanding of the gene was then able to more accurately distinguish between potentially damaging versus probably benign single amino acid substitutions. The key, according to the researchers, was limiting their analysis to genes in other species whose function remains similar to NPC1, that is, orthologous genes. Genes that resulted from an evolutionary duplication event and then diverged in their function (paralogous genes) were not helpful in predicting effects of mutations. Yet the current tools that use evolutionary conservation information do not discriminate between orthologous and paralogous proteins. The investigators suggest that eliminating paralogous genes from analyses would improve predictive capabilities. However, they also note that doing so is labor-intensive, involving the manual work of building high-quality data sets, alignments, and trees and defining orthologs and paralogs. They suggest that trying their approach on other well-defined Mendelian diseases could lead to better predictive methods applicable in clinical practice. —Karyn Hede, News Editor