Big "oops" forces retraction of longevity study

WASHINGTON - The authors of a widely reported study that offered an early glimpse into factors leading to long life are withdrawing the paper because of problems with some of the data they used.

The study, published last July in the journal Science, said that by looking at genetic markers the researchers were able to determine with 77 percent accuracy which gene groups came from people over 100.

The team looked at the genomes of 1,055 Caucasians born between 1890 and 1910 and compared them with 1,267 people born later. They called the results an early step to understanding the pathways that lead to surviving into old age.

The research team, led by Paola Sebastiani and Thomas T. Perls of Boston University, says in Thursday's edition of Science that a quality control problem with one of the instruments they used to study genes caused some false positive findings, and they have reanalyzed the results without that information.

"We feel the main scientific findings remain supported by the available data," the researchers said.

However, they added: "Details of the new analysis change substantially from those originally published online to the point of becoming a new report. Therefore, we retract the original manuscript and will pursue alternative publication of the new findings."

In an accompanying statement, the journal said that while the researchers remain confident about their findings: "Science has concluded on the basis of peer-review that a paper built on the corrected data would not meet the journal's standards for genome-wide association studies."

In an interview with the Boston Globe, Jeffrey Barrett, group leader of statistical genetics at the Wellcome Trust Sanger Institute, a genomic research center in the United Kingdom, described the episode as "a teaching moment."

"There was a pretty broad consensus among people who do these genome-wide association studies all the time, that the [problems in the] original version of this paper should have been caught in peer review," he said. "It is a teaching moment -- I use this longevity paper in lectures, showing if you drill down to some of the supplementary material and some of the data that isn't captured in the headline, if you look carefully -- you can smell something isn't right here."

The editors of Science stressed that "there was no misconduct by Sebastiani and colleagues. The researchers worked exhaustively to correct the errors in the original paper and we regret that the outcome of the extensive revision and re-review process was not more favorable."

Science publishes about 800 articles a year of which three to five are eventually retracted for one reason or another.