Skip to main content
Premium Trial:

Request an Annual Quote

Broad Institute, Mass General Team Develops Polygenic Risk Scores for Five Common Diseases

NEW YORK (GenomeWeb) – Researchers from Massachusetts General Hospital, the Broad Institute, and Harvard Medical School have identified genome-wide polygenic scores (GPS) for finding individuals at enhanced risk for one of five common diseases ranging from cardiovascular conditions to breast cancer.

"We envision polygenic risk scores as a way to identify people at high or low risk for a disease, perhaps as early as birth, and then use that information to target interventions — either lifestyle modifications or treatments — to prevent disease," senior author Sekar Kathiresan, director of Massachusetts General Hospital's Center for Genomic Medicine and the Broad Institute's Cardiovascular Disease Initiative, said in a statement.

The researchers relied on summary statistic data from several large genome-wide association studies done on individuals of European ancestry to find polygenic risk scores that distinguished individuals at three times or more the typical population risk for coronary artery disease (CAD), atrial fibrillation, type 2 diabetes, inflammatory bowel disease (IBD), and breast cancer. As they reported today in Nature Genetics, they subsequently validated these risk scores using data from the UK Biobank project.

Based on the results so far, the team is considering possibilities and potential partnerships aimed at commercializing the GPS tests in the coming year. Within the next six months or so, it plans to establish a web service that can provide risk scores and interpretive information on a research basis to individuals who upload their existing direct-to-consumer genotyping profiles.

"If you don't have genotypes already, it would have to be offered," Kathiresan told GenomeWeb. "We're in discussions with several companies as to how to get this available to the general population."

The researchers initially looked at the possibility of establishing a GPS for heart attack risk, he explained, since his team has been studying the genetics of that condition for more than a decade. They were specifically interested in determining whether polygenic disease risk could be as informative as the risk attributed to single gene mutations linked to familial hypercholesterolemia, which are typically detected in less than 0.5 percent of the population and increase CAD risk by an estimated three-fold.

Using a three-step strategy, the team converted results from GWAS files to potential polygenic risk scores for CAD and the other conditions, using several different algorithms. They ultimately settled on an algorithm that re-weights SNPs based on the correlation structure between SNPs, their statistical significance, and so on, Kathiresan explained.

These candidate risk scores were then tested with directly genotyped and imputed SNP sets for the first 120,280 individuals enrolled in the UK Biobank project before being tested in an independent set of 288,978 UK Biobank participants.

Following the familial hypercholesterolemia mutation risk model, the researchers looked at the proportion of individuals with at least three times the typical risk of each condition based on the polygenic risk score distributions. In the case of the CAD score, which encompassed some 6.6 million common SNPs, they noted that 8 percent of the population had SNP profiles coinciding with three-fold or higher risk.

"The familial hypercholesterolemia mutation is present in about one in 250 in the population and increases risk by about three-fold. Here, we have genetic factors present in more than 5 percent of the population … giving you the same level of risk," Kathiresan said.

"The number of patients at risk here, based on the new polygenic test, is millions in Canada or the US," he added, noting that these individuals are "currently unaware of their risk" since they often do not have elevated cholesterol, blood pressure, or other commonly measured risk factors for heart disease.

On the atrial fibrillation curve, more than 6 percent of individuals had GPS profiles that put them at three times or more the typical population risk for that form of arrhythmia, the team reported, while 3.5 percent of the population reached that risk level for type 2 diabetes based on their common variant profiles. More than 3 percent of individuals had polygenic scores for IBD that corresponded to more than three-fold risk, and 1.5 percent of the population was at pronounced breast cancer risk.

Such information may be important for disease screening and prevention, Kathiresan argued. For example, his team's prior studies have suggested that genetic risk for CAD can be mitigated with lifestyle modifications and/or statin use.

The researchers plan to pursue GPS for additional conditions, particularly Alzheimer's disease and obesity. They also emphasized the importance of garnering more extensive genetic association data for individuals from non-European populations.

"The polygenic risk scores described here were derived and tested in individuals of primarily European ancestry — the group in which most genetic studies have been undertaken to date," the authors noted. "Because allele frequencies, linkage disequilibrium patterns, and effect sizes of common polymorphisms vary with ancestry, the specific GPS here will not have optimal predictive power for other ethnic groups."

Kathiresan also noted that the team has gotten promising results applying its GPS to individuals from other ethnic backgrounds, though those findings have not yet been published.