Still Blaming Your Genes?

You can pay good money to find out if you carry the form of a gene that well-respected scientists, in well-regarded journals, reported was linked to an increased risk of acute coronary syndromes, which include heart attack and a form of angina. But geneticist Thomas Morgan of Washington University suggests you save your cash.

In the current issue of the Journal of the American Medical Association, he and colleagues describe their efforts to confirm previously-reported links between genes and heart disease, and the results are not encouraging for those who believe their medical fate lies in their DNA.

Over the years, biologists have identified 85 variants in 70 genes as cardiovascular risk factors. A "variant" means that the gene, a sequence of chemicals designated A, T, C and G, has a misspelling, so that an A in the healthy form of the gene is a T, for instance, in the increased-risk form. After the initial report of a linkage between variant and disease, other scientists would, ideally, try to replicate the finding in larger populations as well as in people of different ethnicities from those in the original study, to rule out the possibility that the association was a random fluke. But there is little glory in reproducing someone else's discovery, so that is rarely done. Morgan thought that was a mistake: profit-making companies offer tests for some of the variants linked to coronary syndromes even though "to date, no such comprehensive validation of genetic variants potentially associated with ACS or atherosclerosis has been reported."

To fill in the gap, he and colleagues recruited 811 patients with acute coronary syndromes and 650 healthy people, and tested them for the 85 variants in 70 genes linked to the syndromes. Of the 85, none was more frequent in the heart patients than was expected by chance; only 41 were even slightly more frequent in the heart patients. "None [of the 85 variants] were unequivocally validated," the scientists report. As a result, clinical use of the variants--genetic testing to determine if someone is at elevated risk for heart disease--is, they say diplomatically, "premature."

What is striking about this study is not that it is the first to throw cold water on gene-disease association studies--the usual reason journalists pay attention to something--but that it is not the first. In other words, a few intrepid researchers before Morgan have performed the thankless task of combing through old studies and seeing whether the claims hold up. In one of the most exhaustive, Joel Hirschhorn of the Broad Institute (a collaboration of MIT and Harvard) led a study which found that of more than 600 reported associations between a common gene variant and a disease, only 166 had even been studied at least three times. Only six had "been consistently replicated," they reported in 2002 in the journal Genetics in Medicine.

Why do these studies so often go awry? Part of the explanation is how easy it is to analyze genes; what used to take a graduate student a year can now be done in a day. That allows scientists to scan thousands of bits of DNA looking for a hit. By chance alone, some of those hits--a higher prevalence of some gene variant in people with some disease than in people without it--will be statistically significant, leading to the claim of a gene "for" the disease. But that link might still be a coincidence; when another group replicates the study, the association disappears. It's an important caution to keep in mind whenever scientists announce the discovery of yet another gene that you can supposedly blame for what ails you.