Soothsaying was once a fringe pursuit — the purview of psychics, subway cranks, and speed-addicted sci-fi novelists. Today, it’s big business. Abetted by new technology, thousands of salaried STEM types are now engaged in figuring out the future.
Among this crowd, Steve Horvath, a biostatistician at the University of California, Los Angeles, stands out. He isn’t trying to predict what you’ll buy next or whether you’ll commit a violent crime after being paroled from prison. His project, almost magisterially bleak, is to figure out how much of the future you’ll actually get to see.
In a paper published this week in Aging, Horvath and his colleague Ake T. Lu formally announced a project they’ve been teasing for a couple months now: a “time to death” clock called DNAm GrimAge that they claim can predict, better than any other tool, when a given person might die. It was announced in tandem with AgeAccelGrim, which provides a countdown to the year you’ll develop cancer or coronary heart disease. Horvath said he can estimate the number of cigarettes someone has smoked in their lifetime and predict when they’ll go through menopause.
There’s a whiff of Theranos to Horvath’s promise that “a spoonful of blood is more than enough.”
The research has already captured the attention of the life insurance industry. After all, a solid death date could mean real savings when it comes to pricing policies.
Horvath has been tinkering with his clocks for nearly a decade now. Some of them sound even less inviting than GrimAge, which directly references the Grim Reaper. There’s last year’s “skin and blood clock,” for example, and the “pan-tissue” clock, which measured a person’s age. Its results were fairly impressive, with a median error of just 3.6 years, but the broader scientific community was nonplussed. There are, after all, less cumbersome ways to figure out how old a patient is. (Asking them is a good start.) And so Horvath set out to demonstrate that his method had more significant applications.
Central to that method is a process called methylation, which Horvath compares to rust. Over time, certain parts of our DNA become more methylated, or “rusted,” hastening the aging process along with our inevitable death. Horvath and Lu measured methylation spots from each of their DNA samples, which were taken from the blood of dead people. From there, GrimAge crunched the numbers and popped out the corresponding death dates. More often than not, these aligned fairly closely with the dates on record.
There’s a whiff of Theranos to Horvath’s promise that “a spoonful of blood is more than enough” for GrimAge. But Lu and Horvath aren’t isolated cranks: Researchers all over the world are examining the connection between DNA and physical decline. Yan Zhang, a researcher at the German Cancer Research Center, came out with his “mortality risk score” in 2017, and just last week, researchers at the University of Edinburgh announced their own scoring system for predicting mortality based on DNA. Lu and Horvath’s project may sound implausibly futuristic, but they’re not advancing any radical claims on its behalf. Horvath likens GrimAge to a more powerful microscope, a refinement of something that already exists.
The 2,356 blood samples that Lu and Horvath used came from the Framingham Heart Study, a well-known cardiovascular cohort study with data that has been used in many medical papers. That study includes medical information for its participants, which gave GrimAge something to measure against. The paper is amply sourced, and its co-authors are affiliated with a range of major research centers, like Northwestern, Columbia, and Stanford.
Still, the only living person to get a GrimAge reading so far is Horvath himself. He had fared pretty poorly on his pan-tissue clock — “I want to say I was four or five years older than expected; I didn’t like that” — but according to GrimAge, he’s on track to die more or less when he expected.
He does not offer his service to anyone else — and people have asked — for the reason that, from a medical perspective, it’s effectively useless. There is not much value in knowing you only have 10 years (or five years, or six months) to live, because “clinicians only order tests if they have a treatment strategy.”
But the hope is that if and when legitimate anti-aging drugs are developed, GrimAge could be used to test their effectiveness. In a world with functional anti-aging drugs, “doctors could test [your GrimAge number] and say, ‘You know what, you’re aging too quickly. Take this,’” Horvath said.
This scenario would be the culmination of a lifelong dream for Horvath. As a teenager in Germany, more than 30 years ago, he and his friends actually made a pact to solve the problem of death. “At some point, we realized that this was one of the central problems of human existence: our short lives,” he said. They called their initiative the Gilgamesh project, after the Sumerian king who hoped to conquer mortality.
But there are other, more immediate uses for Horvath’s invention.
The Affordable Care Act’s preexisting conditions clause and the Genetic Information Nondiscrimination Act (GINA) would prevent health insurance companies from considering a person’s GrimAge when determining a person’s rate. But GINA, according to Seth J. Chandler, a law professor at the University of Houston specializing in insurance, “quite deliberately excluded other forms of insurance.”
Companies like Reinsurance Group of America are already looking into Horvath’s methylation method as a way to price life insurance policies. Horvath frames this as a positive development: A bad candidate for life insurance — say, a diabetic smoker — might have their case bolstered by a better-than-expected GrimAge score. But the reverse, he concedes, is true as well, meaning otherwise healthy-seeming people might find themselves saddled with a worse policy or denied one outright.
“Life insurance kind of depends on people not having private information about how likely they are to die.”
There is also the possibility that life insurance as we know it would simply cease to exist.
“Life insurance kind of depends on people not having private information about how likely they are to die,” Chandler said. If the only people buying life insurance are those who know they’re going to die soon, the insurance industry would have to “raise its rates dramatically.”
Chandler envisions a post-GrimAge dystopia in which “someone who doesn’t get tested ‘voluntarily’ is now likely to be assumed by the insurance company to have gotten a bad result.” He figures that if GrimAge works and eventually takes off, life insurance would largely be supplanted by accidental death insurance — the sort of policy that was once widely popular but today is best known as a plot point in Double Indemnity. (For the time being, at least, GrimAge cannot measure your likelihood of dying via free-falling air conditioner, swerving convertible, or gunshot wound.)
Mark Kuczewski, director of the Neiswanger Institute for Bioethics at Loyola University Chicago, has a less apocalyptic view. “It’s only going to make a difference in marginal cases,” he said. “Would you use this instead of whether someone’s 100 pounds overweight? Instead of whether someone has a terrible history of genetic illness? Would you use this instead of whether they’re a three-pack-a-day smoker?”
While health insurance is safe for now, an accurate death estimator could only embolden those who don’t see the sense in shelling out for the sick. As Allison Hoffman, a law professor at the University of Pennsylvania specializing in health insurance, points out, health care is an inherently redistributive system and is able to function that way primarily because we can’t know what will happen to someone.
“The closer we come to perfect knowledge, the more we have to reckon with the question of who should pay and how much?” Hoffman said. “I think it heightens the seriousness of those conversations.”
It is important to note that Horvath has no control over how GrimAge will eventually be used. His inventions are the property of UCLA, which excludes its researchers from the licensing process. And as Horvath is quick to emphasize, DNAm GrimAge’s results are, at present, “not very accurate.” A given GrimAge might be off by four or five years in either direction, meaning a given GrimAge might only reveal the decade you’ll die.
But at the rate Horvath and his colleagues are working, a more precise death clock can’t be far off. At that point, whatever use it’s eventually put to, we’ll have to reckon with the fact that it exists at all. How will we behave when we know exactly when death is coming for us?
Horvath, for his part, seems to be managing just fine, though he’s an admittedly unusual case.
“I love these tests,” he said. “I use my own private funds for measuring various clinical markers — I’m just interested.”
Horvath understands on an intellectual level why some people find his invention terrifying — he said his parents lived in fear that he’d covertly draw their blood — but he sounds almost cheerful when talking about his own death date.
“Some people believe information is power,” he explained. “Others don’t.”
Update: An earlier version of this piece misidentified a law professor from the University of Pennsylvania. Her name is Allison Hoffman.