by Matthew Herper
A Harvard scientist thinks he’s reached a new milestone: a genetic test that helps identify people who are at high risk of having a heart attack. Can he convince doctors to use it?
“I think–in a few years, I think everybody will know this number, similar to the way we know our cholesterol right now,” muses Sekar Kathiresan, director of the Cardiovascular Disease Initiative at the Broad institute and a professor at Harvard Medical School.
Not everyone else is so sure. “I think it’s a brilliant approach,” says Harlan Krumholz, the Harold H. Hines Jr. professor of cardiology at Yale University and one of Kathiresan’s collaborators. But he worries about whether Kathiresan’s tests are ready to compete with the plethora of diagnostic tests, from AI-boosted CT scans to new types of “bad” cholesterol proteins, that are on offer. And he worries about cost. There is no commercial version of the gene test. But the very idea that such a test is not only available, but also near, is the result of a cresting wave of new genetic science, the result of large efforts to gather genetic information from millions of volunteers.
The number in question is what is called a polygenic risk score. Instead of looking for one miswritten gene that causes heart attacks, or, for that matter, other health problems, geneticists are increasingly looking at thousands of genetic alterations without even being sure what each does. In the case of Kathiresan’s polygenic score, the test looks for 6.6 million single-letter genetic changes that are more prevalent in people who have had early heart attacks.
Our genetic inheritances, the current thinking goes, are not so much a set of declarative orders as a cacophony of noise. There are big genetic changes that can have a big effect, but most diseases are the result of lots of tiny changes that add up. In Kathiresan’s words, it’s mostly a gemish (Yiddish for “a mixture”). And it’s not clear which changes are biologically important – Kathiresan says only 6,000 or so of the 6.6 million genetic changes are probably actually causing heart attacks. But finding those specific changes will take a long time. The risk score could be used now.
The effect of this genetic cacophony can be huge. The most common single mutation that increases the risk of heart disease is a gene that causes a disease called heterozygous familial hypercholesterolemia (literally: inherited high cholesterol) that occurs in one person in 250 and triple’s a person’s risk of having a heart attack. But today, in a paper in Nature Genetics, Kathiresan and his colleagues present data that 5% to 8% have a polygenic score that also at least triples their risk of having a heart attack. That’s about 20 times as many people, Kathiresan says.
“These patients are currently unaware of their risk because the polygenic patients don’t have higher levels of the usual risk factors,” Kathiresan says. “Their cholesterol is not high. Their blood pressure is not that high. They are hidden from the current risk assessment tools.”
In the Nature Genetics paper, Kathiresan’s team tested the 6-million-variant polygenic score in two groups of patients numbering, respectively, 120,280 and 288,978 people, from the U.K. BioBank, a government-backed effort in the United Kingdom to collect genetic data. For some patients, the risk was even higher, with the genetic changes predicting a fivefold increase in heart attack risk. The paper also argues that polygenic risk scores could be used to predict risk of conditions such as type 2 diabetes and breast cancer.
Another study, yet to be published, looked at the prevalence of both familial hypercholesterolemia and the polygenic score in a population of people who had heart attacks in their 40s and 50s, Katherisan says. Only 2% had familial hypercholesterolemia, but 20% had a high polygenic risk score. Knowing one’s polygenic risk score might matter. A 2016 paper in the New England Journal of Medicine showed that people with high polygenic scores had fewer heart attacks if they had healthier lifestyles, and a 2017 paper in the medical journal Circulation showed that patients with high polygenic risk scores got an outsize benefit from cholesterol-lowering statin drugs. Those papers, both by Kathiresan’s group, used a score that included only a few dozen gene variants.
Doctors should be skeptical of such a test. There’s a long history of tests in medicine that have done more harm than good by leading to people to take drugs they do not need. Cardiologists have gotten used to even higher standards for data. For instance, many might want to see if the test can show a benefit in a large study in which people are tested at random. Many will want more evidence that the test can identify people at high risk they’d otherwise miss, as Kathiresan says, and that it doesn’t lead to treatment in those who don’t need it. Kathiresan says he hopes to do a study in the highest-risk individuals to prove that statin drugs can lower their risk. If the test becomes a commercial prospect, more studies will drive up the eventual cost.
Kathiresan is hoping to follow a less expensive path. He notes that 17 million people have already used genotyping services like 23andMe and Ancestry. He hopes that people who use those services (23andMe costs $99, Ancestry $59) will submit their data to a portal he’ll build for free. He also says he’s in discussions with commercial providers, but he’s hoping that people will be able to get their polygenic scores for about as much as the cost of a cholesterol test. For the people at the highest risk, he argues, this is information that could be important. For others, he argues, why deny people information that has been scientifically validated?
Whether Kathiresan can really pull off a low-cost version in a medical system that is optimized to make money is as big a question as whether the test is ready for prime time. Krumholz worried about the cost of the test until a reporter told him of Kathiresan’s planned website. “If you say it’s free, I’m going, ‘Why not?'” Krumholz says. “It’s a better family history,” he says, comparing the test to asking whether a relative has had a heart attack. But that may be the biggest ‘if’. If anything is more puzzling than genetics, it is the economics of healthcare in the U.S.A.