The test marks one of the first clinical uses of polygenic scores and stems from Broad research that began four years ago.
The risk for some diseases can increase significantly with mutations in a single gene, such as the BRCA1 gene for certain forms of breast and ovarian cancer. But these and many other diseases also have far more complex genetic roots, where the risk is influenced by thousands of changes with smaller effects scattered across the genome. Over the last four years, researchers have increasingly used an approach called polygenic scoring to assess the risk for some of these diseases by calculating the cumulative effect of each of these small genetic changes across up to 6.6 million locations in the genome.
This technology has mostly been used as a research tool, and is now increasingly moving into clinical practice. Cardiologist Amit V. Khera — associate director of the Program in Medical and Population Genetics at the Broad Institute of MIT and Harvard, group leader in Massachusetts General Hospital’s Center for Genomic Medicine, and a key developer of polygenic scores as a Merkin Institute Fellow at Broad — and his team recently introduced a clinical test that includes polygenic scores to estimate a patient’s risk for one of the world’s leading causes of death — heart attack. Mass General’s new Preventive Genomics Clinic, which Khera co-founded in late 2019, began offering the test to patients in December.
When Khera joined the Mass General faculty in 2019, one of his main goals was to further develop ways that Broad discoveries can help patients. He says assessing both small contributions to disease risk from millions of genetic variants (polygenic risk) and larger contributions from single genes (monogenic risk) — which the new test does for heart attack risk — could empower patients and healthcare providers to better prevent disease.
“My ultimate goal is for patients everywhere to know their inherited risk for a range of important diseases like heart attack from a routine genetic test with their primary care provider,” Khera said. “It’ll be up to us, as a clinical community, to tailor recommendation and screening guidelines to help people overcome inherited risk, based on that genetic susceptibility report card.”
The pilot test is a collaboration between Mass General and Color, a health technology company, which will run the test and generate results. The test is not yet covered by insurance and is available to any patient, including someone with a strong family history of heart disease.
During the pilot phase, Khera and his team will collect data on what patients want to learn from the test, their scores, and how the test results may affect their care over time. This data will help Khera’s group further show the clinical benefits of polygenic scores, establish clinical guidelines for its use, and eventually get the test covered by insurance. Khera hopes that increased awareness and acceptance of genetic risk assessment will help shift medicine toward earlier disease prevention based on predicted risk — still a relatively new idea in clinical care. (The Broad is funding the pilot phase and providing the test for free to the first 100 patients.)
Khera emphasizes that someone’s genetic makeup doesn’t determine their health outcomes, but rather is a wake-up call for patients and their providers to implement medical and lifestyle changes to lower disease risk.
Once the pilot is complete, Khera and his collaborators plan to expand the genome interpretation assessment to other institutions across the country.
“We’re hoping to eventually make the test broadly available. For those who are high-risk but otherwise healthy, the test might provide a decades-long head start for prevention,” Khera said. “But some people who have had heart attacks are also interested in understanding why, if they didn't smoke and their cholesterol was normal, they had this terrible thing happen to them.”
Journey to the clinic
The new test was developed based on findings published in August in Nature Communications. The study, which Khera co-led, showed that in patients with a mutation in a single high-risk disease gene, including polygenic risk in their test can significantly modify the overall genetic risk. While polygenic and monogenic risk have traditionally been considered separate concepts, Khera says integrating the two into a single test gives a more complete and accurate picture of common disease risk than just looking at a single high-risk disease gene.
The study was an important step in moving genetic risk assessment into the clinic, and also the culmination of more than four years of work that began at the Broad. Khera was then a postdoc in the lab of Sekar Kathiresan — formerly a Mass General cardiologist and Broad institute member, now co-founder and CEO of Verve Therapeutics. In 2016, the lab started looking across all genetic variants to predict disease risk.
The original paper co-authored by Khera and others, published in 2018 in Nature Genetics, showed how polygenic scores can accurately estimate genetic risk for five common diseases and identify high-risk individuals who would otherwise go undetected. The researchers then expanded the concept to obesity in a follow-up paper in Cell in 2019, noting that inborn risk begins to affect health as early as four years of age. In both papers, the researchers showed that an algorithm they developed could estimate a person’s inborn disease risk by comparing a patient’s set of genetic variants against established variants that years of genome-wide association studies have most strongly linked with various common diseases.
To pave the way for more widespread adoption of polygenic scores in the clinic, Khera and his team showed in 2019, through a collaboration with Color, that a method called “low-coverage whole genome sequencing” could be used to comprehensively and accurately look for changes across the entire genome.
In a flurry of papers in 2020, Khera and his group further demonstrated the potential value of polygenic scores. In September in Arteriosclerosis, Thrombosis, and Vascular Biology, they showed in a study of nearly 29,000 people that the tool can better predict lifetime risk of heart disease than traditional risk factors such as high blood pressure or cholesterol — and can assess risk decades before traditional risk factors emerge.
In a Journal of the American College of Cardiology paper in August, Khera’s team extended the concept to heart disease in more than 7,000 people of South Asian ancestry — a population with a higher-than-average risk of the disease. And in the same journal in June, Khera and a team of researchers led by the Broad’s and Mass General’s Krishna Aragam and Pradeep Natarajan showed in a study of 47,000 individuals that most patients with high polygenic scores were not being identified reliably and prescribed standard preventive measures such as cholesterol-lowering medications – even at Mass General and other top hospitals. The findings suggest that traditional methods of identifying high-risk patients could be improved by incorporating genetic risk assessments.
An important step in getting polygenic scores into the clinic was the team’s combination of polygenic and monogenic risk into a single test to improve clinical accuracy. The hope is that the new test will also boost the acceptance of polygenic scores by the medical community, which already assesses disease risk in some patients by looking at mutations in single high-risk genes such as BRCA1.
“The motivation and momentum to offer polygenic scores has been very low because the predictive nature of them had been limited,” said Heidi Rehm, institute member at the Broad and chief genomics officer at Mass General. “Now, offering them combined with monogenic risk assessment to a population that’s already attuned to risk will surely accelerate progress.”
Although this research was key to showing the benefits of better genetic risk assessment, Khera says what really enabled the translation of the technology into an actual clinical test for patients is the behind-the-scenes work since 2018 to find and work with an industry collaborator, Color, to make sure genetic risk could be accurately calculated and reported to patients.
“This represents a new, refined approach to screening for genomic conditions,” said Alicia Zhou, Color’s chief science officer. “We hope that better information about the risk of genomic diseases in a clinical setting will improve screening and treatment for people, and hopefully improve patient outcomes.”
Diversifying the model
So far, polygenic scores have been developed mostly using genetic data from people of European ancestry, which limits the accuracy of the scores for other populations. Researchers at Broad, including Khera’s team, are building bigger genetic datasets representing greater diversity of ancestries to ensure that polygenic scores will benefit more people. Researchers from the Broad’s Program in Medical and Population Genetics, the Broad’s Stanley Center for Psychiatric Research, the National Institute of Health’s All of Us Reseach Program, and elsewhere are linking genetic data to the health records of many underrepresented groups both in the United States and around the world. “There's still a gap, but we’ve started to make a ton of progress, internationally, in collecting more genetic data across diverse ancestries,” said Khera, who has built a team to tackle this issue.
For the new genetic test for heart attack at Mass General, Khera says accounting for a patient’s genetic ancestry is critical for accurate results. “We make sure we compare your polygenic score to people of similar ancestry,” he said. His team also recently compared the performance of polygenic scores for heart disease across different ancestries, gaining insight that will help in the interpretation of test results for patients.
As the new test is rolled out, Khera and other Broad researchers continue to improve their polygenic score algorithms. For example, Khera and colleagues from the Broad’s Cardiovascular Disease Initiative are collaborating with IBM Research to further refine reporting of genetic risk and develop machine learning-based models that predict risk of heart attack, sudden cardiac death, and other important heart disease endpoints. The team is also developing an online tool for patients to enter their polygenic scores and other health measures and estimate their heart disease risk over the next decade — similar to a commonly used tool that includes only clinical factors.
Khera and his team, including Deanna Brockman, a genetic counselor with Mass General’s Preventive Genomics Clinic, hope to help convince the medical community of the value of such genetic risk tests. “We hope that our experience returning polygenic scores for heart disease to patients will serve as a blueprint for the clinical community, enabling rapid expansion to additional sites and diseases to help inform future guidelines for how genome interpretation can improve care,” Brockman said.
If you’re interested in being seen in the MGH Preventive Genomics Clinic for possible genetic testing to determine risk for heart attack, please email MGHPreventiveGenomics@mgh.harvard.edu