Scientists Discover the Fountain of Youth! Or Not.

They say getting old is better than the alternative, and it's much better if you manage to do it like Florrie Baldwin, who died in May after 114 years of great health. (She was still climbing ladders at 104 and almost never took any medication.) Baldwin attributed her long, hale, and hearty life to a daily fried-egg breakfast, but her true advantage was probably in her DNA. Scientists have long suspected that she and others who grow very old have genetic variants that protect against the molecular ravages of age.

The trick, though, is finding those genes. Much of the research so far looks more clear-cut in mice or worms or fruit flies than it does in humans. And because the topic of old age and genes in general inspires so much excitement, it's often hard to tell where any given study falls on the continuum between brilliant (last year, the Nobel Prize for medicine went to three scientists who studied the relationship between DNA and aging) and ridiculous (this year, the cosmetics company Lancôme launched an eye cream that purports to "boost genes' activity and stimulate the production of youth proteins," which is about as believable as Baldwin's fried-egg theory).

What, then, to make of the new headline-grabbing paper that identifies somewhere between 33 and 70 genes (depending on how definitive you like your results) associated with extreme longevity—and also introduces a model that claims to predict with 77 percent accuracy whether you're going to be one of those ripe old folks? If the study's findings are correct, they are a very big deal, with the potential to overhaul how scientists think about aging and genetics. Indeed, they're so striking that some of the world's top geneticists think they can't possibly be right. More on that in a bit, but first, a point that even the study's authors have to concede: the research doesn't actually describe normal aging. It concerns only genes that may govern the process in people who make it to 100 or more. "The question is, of course, do the findings apply to the general population? Can we apply your model and predict the average lifespan?" says Paola Sebastiani, the Italian researcher who led the team. "And the answer is no, we can't." So what exactly can we learn from the new study?

The paper, published today in Science, has two basic parts. The first is what's called a "genome-wide association study," or GWAS. Researchers obtained genetic data (300,000 variants) from about 1,000 very old people (those over 100 years of age) enrolled in the New England Centenarian Study, and then compared the readouts to results from a same-size group of average people often used as a standard control in genetic studies. They found 70 genes that were more common in the centenarians. Then they repeated their study with smaller groups and confirmed 33 of those. They also looked at known disease-causing genes in both centenarians and regular subjects. It turned out that the centenarians had just as many dangerous variants as everyone else, which suggests that the 70 longevity genes (or 33, if you prefer the confirmed ones) were actively protecting them against illness. In other words, very long life isn't just about not having genes that make you sick—it's also about having genes that keep you well.

Next, the researchers employed some unusual math to build a model that described the combined effects of 150 variants they had found in the centenarians. (Some of those variants were related to the same genes, so the final number of suspected genes was still 70.) Then they applied the model to each of their study subjects, blinding themselves as to whether an individual was a centenarian or a member of the control group. Seventy-seven percent of the time, the model rightly predicted which group a given person belonged to—a success rate that is not only statistically significant but unprecedentedly high for a model that predicts a complex trait. It also pointed to 19 different genetic "signatures," or combinations, that seemed to confer long and healthy life in the centenarians.

Here's the weird thing: 15 percent of the control group had those signatures, too. That means that, genetically speaking, 15 percent of us should be living to 100 or more, when in reality only about 1 in 6,000 people do. What's happening to the rest of the would-be centenarians? Thomas Perls, a co-author of the paper and a geriatrics researcher at Boston University, says that maybe they're getting bumped off by things no gene can prevent. "Remember, this generation [in the New England Centenarian Study] lost a quarter of its population to childhood infectious diseases," he says. "And just because you've been handed the genetic blueprint for long life doesn't mean you're going to get there if you smoke a lot or you get killed in World War I or you're hit by a bus."

These are provocative ideas, and that may explain the wide variety of reactions that scientists had upon seeing the paper. Some said it was groundbreaking and could lead to drugs that mimic the protective genes in those who aren't blessed with them naturally. "What this paper answers, which I think is very important, is how many variants are there that can assure longevity," says Nir Barzilai, director of the Institute for Aging Research at the Albert Einstein College of Medicine in New York City. "Now scientists can start tracking those genes down and leading the findings toward drug development." Barzilai, who has collaborated with the study's authors on previous projects, has done some of that work already, probing a gene that influences how the body processes a hormone called IGF-1.

But other researchers were concerned about the new study's methods, calling the results everything from "somewhat surprising" to "preposterous." The problems start, they say, with the size of the group the researchers examined. The New England Centenarian Study is the largest cohort of very well-seasoned people in the world, but compared to the numbers typically used in GWAS-style research, it's actually quite small. Modern GWAS techniques usually involve groups of tens of thousands or hundreds of thousands of people. To attain statistical significance in a GWAS as small as the one in the Science paper, any gene would have to have a hugely strong effect in the body. It's odd that the Science paper finds not just one strong gene but a whole raft of them, especially since common diseases are usually caused not by strong genes but by weak genes acting together. (Why is a bit of a mystery, though it's possible that many strong genes have been weeded out over time by natural selection.) Aging, of course, is humanity's most common ailment. "I am very surprised that in a cohort of this size they have found 33 variants of genome-wide significance in extreme longevity," says Kári Stefánsson, the Icelandic researcher whose company, deCODE Genetics, has led many GWAS efforts. "We haven't seen any of them in our work with a different kind of cohort, but [a] much larger [one]." Other work at deCODE has also suggested that extreme old age is influenced by just a few genes—certainly not as many as 33 or 70.

The study's authors have a response to that. Yes, they admit, the sample is smaller than you'd need to do a GWAS of a common disease (which isn't really their fault, since there aren't many centenarians available for study in the first place). But common diseases, with their panoply of weak genes, aren't the right comparison to make, they say. Centenarian status isn't just an extreme form of the common condition known as aging, they argue; it's a rare condition of its own, and rare conditions are often caused by genes with powerful effects.

There are other potential problems with the new study. David Altshuler, a leading geneticist at the Broad Institute (a collaboration between MIT and Harvard), says that "one has to be cautious in interpretation, because the cases and controls were drawn from different times and places, and the DNA from cases and controls was measured using different technologies, which could lead to false apparent relationships." Duke University's David Goldstein, also a prominent geneticist, echoed those concerns. (And Altshuler and Goldstein are renowned for rarely agreeing on anything.) The control data in the Science study came from a standard set of numbers that Goldstein's lab has used, too. "We have found when we compare samples run at Duke to the [standard] control panel, there are a lot of [variants] that appear significant just because the samples were run in different ways," he says. Until the data has been replicated using identical technologies for the centenarians and controls, he adds, "I think we've got to hold judgment on this."

The authors of the Science paper are respectable researchers, and they're not trying to sell anyone genetically enhanced snake oil. (Tom Perls, in fact, has been such a vocal critic of anti-aging hype that he was once sued by the American Academy of Anti-Aging Medicine for defamation. They settled.) Sebastiani, Perls, et al. will be doing lots of follow-up research, including, they hope, a whole-genome sequencing project that will shed more light on their findings.

But for now, their research isn't ready to be translated from the lab to the clinic. You don't need to rush out and get tested for all the genes found in the Science study (although someone somewhere is surely making plans right now to sell you such a test). You can probably get a good idea of your risk just by looking through your family scrapbooks. "Using this technology might be helpful," says Robert Marion, a clinical geneticist at Montefiore Medical Center in New York. "But I would think that by just asking how old the person's parents and grandparents were at their time of death, the accuracy would probably be higher." So, if you want to know your chances of making it to 100, you should learn your family history and eat right and exercise while you're at it. Surely you didn't need a highly technical Science paper to tell you that.