This article was edited to add the role of the NCI as co-funder of this consortium.
NEW YORK – An international consortium founded to develop and evaluate methods to improve the use of polygenic risk scores (PRS) for predicting disease and health outcomes in populations of diverse ancestry has recently found some benefit in applying them to cardiovascular disease and type 2 diabetes.
The Polygenic Risk Methods in Diverse Populations (PRIMED) Consortium brings together 284 researchers from 12 countries and approximately 50 institutions. Most of its funding comes from the National Human Genome Research Institute (NHGRI) and the National Cancer Institute (NCI) in the US, with additional grants from a variety of sources provided by consortium investigators.
Research groups within the consortium primarily study the most effective use of existing genetic data for PRS. "The consortium has a little bit of funding to do its own genotyping, but overwhelmingly, we're using existing genetic data, drawn from as diverse a group as we can get," said Ken Rice, principal investigator of the consortium's coordinating center at the University of Washington.
Robb Rowley, NHGRI Program Director, said the lack of genealogical diversity in genomic datasets provided considerable momentum in forming the consortium. "The goals of the PRIMED network are twofold," he said. "One is to aggregate data to come up with more robust data for diverse groups, and the second is to come up with new methods to use PRS in [people with] diverse ancestry."
Data gathered by the investigators is largely aggregated within the NHGRI's Analytics Visualization and Informatics Lab-space (AnVIL) portal, a cloud computing platform that facilitates data sharing and analysis. "It provides a place to store data and also develop pipelines to process data in a distributed and a consistent way," Rice said.
Some of the consortium's progress can be seen in a series of recent studies that found strong correlations between PRS and clinical outcomes for type 2 diabetes (T2D) and cardiovascular disease (CVD), along with some evidence for the clinical utility of these scores, although the authors were careful to note that much work remains to be done before PRS could be included in clinical guidelines.
"There's a lot of interest in polygenic risk scores for the clinical setting, but there's not a lot of data on their clinical utility," said Iftikhar Kullo, professor of medicine at the Mayo Clinic and an investigator for the consortium.
Kullo is the senior author of an observational study –– currently a preprint on MedRxiv –– of how polygenic risk scores and social determinants of health (SDoH), as assessed by a survey, impact the risk of coronary heart disease (CHD) in the US, using data from 67,256 participants in the NIH-sponsored All of Us research study. "Social determinants of health are being increasingly recognized as a major risk factor for common diseases," Kullo said.
Kullo's group found that its ability to accurately risk-stratify patients was heavily influenced by SDoH, suggesting that they should be included along with PRS in any clinical risk prediction models. Nonetheless, Kullo cautioned that these results must be studied in more diverse populations, since the All of Us cohort, despite including people from diverse backgrounds, was still nearly 40 percent White.
In another consortium-led study, an international group led by researchers from South Korea's Seoul National University College of Medicine calculated a PRS predictive of which women with gestational diabetes are at risk of later developing T2D.
In a group of five cohorts representing 1,895 women from four ancestry groups (European, East Asian, South Asian, and Hispanic), all with histories of gestational diabetes, the study found that a PRS for T2D associated significantly with development of the disease later in life.
While the predictive power of PRS generally remains lower than strong clinical indicators such as blood glucose levels, technological advances in genetic data production and PRS calculation methods have already increased the predictive power of PRS within just the past decade, Jaewon Choi, the study's first author, said in an email, adding that PRS continue to improve.
Choi said that T2D is a good indication for investigating the utility of PRS because the traditional clinical risk factors used to predict development of the disorder among women with gestational diabetes are limited. This is because these women are relatively young and often don't exhibit the phenotypes associated with diabetes as prominently as other at-risk individuals.
"We hypothesized that the integration of genetic information could enhance the accuracy of prediction," Choi said. "These results suggest that the T2D PRS could be effective in stratifying women with gestational diabetes and identifying those at particularly high risk of developing T2D."
Soo-Heon Kwak, Choi's collaborator and the senior author of the study, added that their group is initiating a new cohort study of women diagnosed with gestational diabetes to investigate other risk factors and biomarkers, including lifestyle and omics data, as well as strategies to prevent them from developing T2D. "This contemporary cohort will also allow us to evaluate the utility of genetic information, such as polygenic risk scores, in predicting outcomes and guiding prevention strategies," he said.
Bridging the prior studies on T2D and CVD, James Meigs, professor of medicine at Harvard University and another PRIMED investigator, and his colleagues recently published a study showing that individuals with T2D shared many genetic risk factors for CVD with the general population.
"[The findings] confirm that many genetic CVD risk factors found in the general population also act in individuals with diabetes," said Meigs. "If and when genetics are added to CVD risk scores, then the findings may have clinical implications."
For now, however, his group's findings are better seen as a biological discovery than as something to be applied clinically, he said. He used every dataset he could find globally to assemble a diverse cohort, he explained, leaving no additional datasets for validation and replication. "However, new datasets are becoming available and further validation studies may be possible in the future," he added.
The PRIMED Consortium is not the only group assessing the clinical utility of PRS. The NHGRI-funded Electronic Medical Records and Genomics (eMERGE) study, for example, is also assessing the clinical use of PRS in diverse populations.
The consortium running that study recently published cross-ancestry PRS for 10 chronic conditions and plans to publish clinical utility data for those scores next year. "Controlled studies of PRS in clinical care are lacking but we may have information when the eMERGE study is completed," said Meigs, who is also an investigator with the eMERGE consortium.
PRIMED investigators envision that PRS will eventually find their way into the clinic but caution that more and better evidence is still needed. "I don't anticipate PRS being included in guidelines in the near future," said Meigs.
"I believe the key question is not whether or not to use genetic information or PRS," Kwak said, "but how to best utilize this information in clinical practice."
Other groups working to implement PRS in the clinical setting appear to share Kwak's view. UK firm Genomics plc, for instance, recently published the results of a study evaluating the clinical implementation of cardiovascular PRS, part of an effort to eventually gain acceptance of PRS as a clinical tool within the UK National Health Service.
In addition to developing and testing PRS, several groups within the PRIMED Consortium focus on the social and ethical implications of incorporating them into routine clinical care.
Eimear Kenny, professor of medicine and genetics at the Icahn School of Medicine at Mount Sinai, said that while bioethics working groups had not been a high priority for the consortium initially, "these working groups … really evolved as experts did big landscape audits of what was out there in the field and what was really needed."
Such audits led the consortium to merge its social, ethical, and population descriptors working groups in order to more comprehensively address concerns of how populations are described when considering how to integrate various population-specific PRS, said Sarah Nelson, senior research scientist at the PRIMED Coordinating Center at UW.
"As a consortium using diverse studies from around the globe," she said, "it's been key to be thoughtful about how study participants are grouped, labeled, and analyzed, [in order] to move away from coarse continental categories that can falsely reify race as a biological concept and incorrectly mis- or over-attribute health disparities to genetic rather than social causes/determinants."
Rice pointed to the number of publications coming out of the consortium as one of its key successes so far. PRIMED gives a strong voice to the PRS community and helps build support for establishing better methods to derive cross-ancestry PRS and change the way that race is viewed within the scientific community, he said. "How do we think of individuals' ancestries with diversity in mind as more of a continuum of ancestry rather than discrete categories?"