A study on genetic variants of extreme longevity that was reported in Science in 2010 (see ARF related news story) and later retracted by the authors due to technical concerns (see ARF related news story), now appears in revised form in PLoS ONE. In a paper posted online January 18, Thomas Perls of Boston University School of Medicine and colleagues report data that they say support, albeit more conservatively, the initial conclusion that genetics have a big say in whether a person will live to 100 years. “The major scientific findings of the paper are no different than what we had in Science,” Perls told ARF. The methods are novel and complex, though, causing some scientists to question how the data are interpreted.

In the initial analysis—a genomewide association study of 1,055 centenarians—first author Paola Sebastiani and colleagues identified 150 single nucleotide polymorphisms (SNPs) that predicted with 77 percent accuracy whether someone in a separate cohort would survive to extreme old age. The team retracted the findings after problems with the analysis were pointed out.

Realizing “there was a cloud over the data,” Perls said, his team approached researchers in an independent lab—coauthor Josephine Hoh’s at Yale University School of Medicine in New Haven, Connecticut—and asked them to validate the data using stricter quality control measures. With these restrictions, “you miss some true positives but get rid of a lot of false positives,” Perls said. The revamped analysis identified 281 SNPs that, when applied to an independent sample (median age 100 years), could distinguish centenarians and genetically matched controls with 58 percent specificity and 60 percent sensitivity. In other words, using no information besides that provided by those 281 variants, scientists correctly identified who lived to 100 about 60 percent of the time. For the subset of subjects 102 years and older, the 281-SNP model made the right prediction 71 percent of the time, and for the very oldest individuals (105 years and up), the model was 85 percent accurate. “This is consistent with the notion that at older and older ages, the subjects are more and more [genetically] exceptional from a survival point of view,” Perls said. For further validation, the Boston University scientists applied their model to an additional sample of older centenarians (median age 107 years) and found it distinguished them from controls with 61 percent specificity and 78 percent sensitivity.

Confirming previous research (Schächter et al., 1994), the analysis found that apolipoprotein E4 (ApoE4), the top genetic risk factor for sporadic Alzheimer’s disease, is virtually absent in centenarians. On the whole, though, extremely old people have about as many disease-associated variants as the general population, suggesting they must have longevity-associated variants that trump or counteract the bad variants, Perls said.

While some lay media reports tout the findings as a “blood test that predicts your chances of living to 100” (see stories by Fox News and the U.K.’s Daily Mail), “that’s not what this is,” Perls said. “This is a means to get a much better understanding of the genetic basis of exceptional longevity and to understand what genes may be playing a role.”

Other scientists still have concerns about the paper’s statistical methods, which consider the effects of multiple variants simultaneously instead of one gene at a time like conventional models. David Hinds of 23andMe, a personal genomics company in the San Francisco Bay Area, said he and coworkers could not replicate the 281-SNP model in their 23andMe participant cohort. “Longevity is a difficult trait for testing genetic risk models because we know so little about its genetics,” Hinds wrote in an e-mail to ARF. “We think the authors’ methodology might be better tested on a trait where we already know more about genetic risk factors, so other approaches could be more easily compared.” (See full comment below.)

John Hardy of University College London, U.K., finds the new study “reliable but underpowered.” The data are now “high quality and could be added to future studies,” he noted in an e-mail to ARF. But the “only certain finding (of a longevity variant) is ApoE, which was reported 15 years ago.”—Esther Landhuis


  1. This new paper addresses some of the issues raised by the authors' retracted study, but some methodological issues remain that call into question the conclusions the authors offer. The finding of an association between longevity and a variant in ApoE is convincing, but we were not able to replicate their more complex 281-SNP model in the 23andMe participant cohort. Longevity is a difficult trait for testing genetic risk models because we know so little about its genetics. We think the authors' methodology might be better tested on a trait where we already know more about genetic risk factors, so other approaches could be more easily compared.

    View all comments by David Hinds

Make a Comment

To make a comment you must login or register.


News Citations

  1. Studying the Oldest Old—Chances for Longevity Spelled Out in Genes?
  2. Research Brief: Genes and Longevity Study Retracted

Paper Citations

  1. . Genetic associations with human longevity at the APOE and ACE loci. Nat Genet. 1994 Jan;6(1):29-32. PubMed.

External Citations

  1. Fox News
  2. Daily Mail

Further Reading


  1. . Genetic signatures of exceptional longevity in humans. Science. 2010 Jul 1;2010 PubMed. RETRACTED

Primary Papers

  1. . Genetic signatures of exceptional longevity in humans. PLoS One. 2012;7(1):e29848. PubMed.