"Long before it's in the papers"
August 03, 2010

RETURN TO THE WORLD SCIENCE HOME PAGE

Longevity findings in question

July 9, 2010
World Science staff

Some ge­neti­cists are doubt­ing the val­id­ity of a new study claim­ing that a group of genes can pre­dict who will live the longest with 77 per­cent ac­cu­ra­cy, The New York Times re­ported Thurs­day.

Mean­while, the sen­ior sci­ent­ist on the study ac­knowl­edged there was a tech­ni­cal er­ror in the work, but said it did not af­fect the con­clu­sions.

“I think it is very un­likely in­deed that the find­ings… are cor­rect, or even mostly cor­rect,” Da­vid B. Gold­stein, a ge­net­i­cist at Duke Uni­vers­ity in North Car­o­li­na, wrote last week in an e-mail mes­sage, ac­cord­ing to the Times.

The publication quot­ed ge­net­i­cist Kari Stef­ans­son of De­code Ge­net­ics, an Ice­land­ic com­pa­ny, as say­ing that the stu­dy’s key weak­ness in­volved one of the de­vices, called gene chips, used to scan the ge­nomes of study par­ti­ci­pants.

The de­vice, called the Il­lu­mi­na 610, at­tributes an un­com­mon form of a gene as hav­ing come from both par­ents in­stead of just one, Stef­ans­son told the Times. The re­sult, she added, would be that the chip would throw off re­sults if just 10 per­cent of par­ti­ci­pants were tested us­ing it. 

One of the sci­ent­ists in­volved in the study told the news­pa­per that about 10 per­cent of the par­ti­ci­pants were in­deed tested us­ing that chip, but that the prob­lem would af­fect only two of the 33 gene vari­ants that the study fo­cused on. Stef­ans­son dis­a­greed and said that the prob­lem af­fects all the re­sults, ac­cord­ing to the Times.

The stu­dy, from Bos­ton Uni­vers­ity Schools of Pub­lic Health and Med­i­cine and Bos­ton Med­i­cal Cen­ter, was pub­lished July 1 on­line by the re­search jour­nal Sci­ence.

In a state­ment this week, ed­i­tors of the jour­nal’s said they were hav­ing the find­ings re­an­a­lyzed, but that its re­view­ers had “de­ter­mined that the sta­tis­tics and the de­sign of the study were sound.”

* * *

Some geneticists are doubting the validity of a new study claiming that a group of genes can predict who will live the longest with 77 percent accuracy, The New York Times reported Thursday. Meanwhile, the senior scientist on the study acknowledged there was a technical error in the work, but said it did not affect the conclusions. “I think it is very unlikely indeed that the findings… are correct, or even mostly correct,” David B. Goldstein, a geneticist at Duke University in North Carolina, wrote last week in an e-mail message, according to the Times. The Times quoted geneticist Kari Stefansson of Decode Genetics, an Icelandic company, as saying that the study’s key weakness involved one of the devices, called gene chips, used to scan the genome of study participants. The device, called the Illumina 610, attributes an uncommon form of a gene as having come from both parents instead of just one, Stefansson told the Times. The result, she added, would be that the chip would throw off results if just 10 percent of participants were tested using it. One of the scientists involved in the study told the newspaper that about 10 percent of the participants were indeed tested using that chip, but that the problem would affect only two of the 33 gene variants that the study focused on. Stefansson disagreed and said that the problem affects all the results, according to the Times. The study, from Boston University Schools of Public Health and Medicine and Boston Medical Center, was published July 1 online by the research journal Science. In a statement on Wednesday, editors of the journal’s said they were having the findings reanalyzed, but that its reviewers had “determined that the statistics and the design of the study were sound.”