Health predictions based on polygenic risk scores: reasonable? or too controversial?

As these GEITP pages have often discussed, genome-wide association studies (GWAS) have been used for well over a decade now –– in an attempt to determine which genes contribute to a complex trait (e.g. schizophrenia, type-2 diabetes, obesity cancer), or to a quantitative trait (e.g. height, body-mass index, level of academic achievement –– which GEITP presented just recently). Large research consortia collate tens or hundreds of thousands of individuals for their experimental (affected), and their control (unaffected), cohorts for complex diseases; large gradients of tens or hundreds of thousands of individuals are collected for quantitative traits. And then DNA from these individuals is screened for as many as six or seven million single-nucleotide variants (SNVs) across virtually the entire genome. These SNVs are sometimes located within a gene’s coding region, but more often located in noncoding regions, near a gene or far away from the nearest gene.

GWAS are “fishing expeditions,” i.e. there is no hypothesis needed; scientists are simply looking for genes that might provide insight into the etiology of a disease, a new drug target for a disease, or to explain drug efficacy or toxicity. In the case of risk of coronary artery disease (CAD), many dozens or several hundred SNV locations are found to be “associated with” this disorder. But can any test be used to predict whether someone will die from CAD? One approach is to stratify people into clear trajectories for heart attack, based on specific SNVs.

The poly­genic risk score represents one of the latest approaches in the hunt for the genetic contributors that will predict common diseases [see the attached editorial]. Researchers have been struggling to account for the degree of heritability of conditions — including heart disease, dementia, type-2 diabetes and schizophrenia. Polygenic scores add together the small contributions of tens, to millions, of SNV locations in the genome, to create some of the most powerful genetic diagnostics to date. Recent studies have analyzed more than a million participants by combining information from several different sources, increasing scientists’ ability to detect tiny effects [For anyone interested, the 11 Oct 2o18 issue of Nature (vol. 562, pp 194, 203 and 210) provides three recent examples of these large-cohort studies.]

The attached editorial illustrates in more detail these polygenic risk scores, which could become the next great stride in genomic medicine. However, this approach has also generated considerable debate. Some research presents ethical quanda­ries as to how the scores might be used (e.g. in predicting academic performance). Critics also worry about how people will interpret the complicated, and ofttimes equivocal, information that emerges from the tests. Because leading biobanks lack ethnic and geo­graphic diversity, for example, the current crop of genetic-screening tools might have predictive power only for the populations represented in those particular databases. 🙁


Nature 11 Oct 2o18; 562: 181–183

This entry was posted in Center for Environmental Genetics. Bookmark the permalink.