"Even if there are beneficial results, the standard should be that you need to document that knowing the genetic information is clinically useful," says Sekar Kathiresan, director of preventive cardiology at Massachusetts General Hospital.

Coronary heart disease caused one of every five deaths in the United States in 2006, so scientists have for quite some time been on the hunt for genes linked to heart attacks.

Rapid advances in technology have made that task much easier. At the same time, many of the genetic links to heart disease identified so far haven't held up on further analysis. At present, the only credible link is to a variant of the gene 9p21, identified last year by the Icelandic company deCODE Genetics, says Kathiresan. DeCODE offers a $200 diagnostic test for the 9p21 variant. (See "Gene Variant Linked to Heart Disease.")

A second gene, PCSK9, also looks promising, Kathiresan adds. "Nearly everything else is in the realm of 'possible but not definite.'"

It's good that KIF6 has been identified as a potential risk factor in several different studies, Kathiresan says. In each of the studies, he notes, there is less than a one-in-20 probability that the finding is a result of chance, which is generally considered an acceptable threshold for statistical significance.

But because of the high possibility of false positives, the threshold for genome-wide association studies should be much higher, on the order of one in 20 million, Kathiresan says. Both the 9p21 and the PCSK9 pass that test, he says.

"The key issue here is we don't know if these [KIF6 studies] are real results," Kathiresan says. "You need to show that it is clinically useful, and they have not crossed that threshold."