Gene's reported role in depression questioned by subsequent studies: combined analysis of 14 papers fails to support connection.

The last thing depression investigators need is another dead-end research downer. Efforts to find genes that directly contribute to depression have come up empty. And a research team now concludes, after a closer inspection of accumulated research, that a gene variant initially tagged as a depression promoter when accompanied by stressful experiences actually has no such effect.

By showing that follow-up studies collectively don't support the initial study that launched this line of research, a two-part analysis debunks the proposed pathway to depression. The chances of becoming depressed rise as stressful events mount, regardless of genetic makeup, report statistical geneticist Neil Risch of the University of California, San Francisco and his colleagues.

The new analysis, published in the June 17 Journal of the American Medical Association, also demonstrates the difficulty of replicating reports of any gene variants that appear to work with environmental triggers to foster psychiatric disorders. Individual studies typically lack the statistical power to detect gene-by-environment interactions correctly, because most candidate genes and stressful events exert modest effects on mental ailments at best, the scientists say.

"I'm supportive of looking for gene-by-environment risk factors, but we'll need much larger samples to find interactions that can be independently replicated," Risch says. In his view, statistically rigorous studies will need tens of thousands of participants.

In 2003, scientists led by Avshalom Caspi and Terrie Moffitt, both psychologists now at Duke University in Durham, N.C., studied 847 New Zealand volunteers who had been followed since age 3. Between ages 21 and 26, those who encountered several stressful events--such as health crises, money woes and relationship breakups--and who had inherited one or two copies of a short version of the serotonin transporter gene exhibited high depression rates.

The serotonin transporter gene makes a protein that reduces transmission of serotonin, a mood-related chemical messenger in the brain. Many depression medications block the serotonin transporter gene's protein.

Caspi and Moffitt's report elicited excitement among researchers who had been unable to link any genes directly to psychiatric conditions.

But the new meta-analysis directed by Risch challenges the Caspi-Moffitt findings. In the first part, Risch's team combined and reanalyzed data on 14,250 participants in 14 studies published through March 2009. A second part included unpublished data on 10,943 of the volunteers from 10 of the 14 studies. With that data, the researchers could identify any interactions between the key gene and stressful events for men and women separately. The team transformed data into a format that could be compared with the 2003 study.

Meta-analyses have their own problems, though. Chief among them is the difficulty of mixing studies with different sample sizes, participant characteristics and measurements of varying quality into a mathematically meaningful concoction.

Although the new meta-analysis could not totally avoid such problems, it raises a valid concern that Caspi and Moffitt's original results have yet to be replicated, says psychiatrist Kenneth Kendler of Virginia Commonwealth University School of Medicine in Richmond.

Alleged replications have measured stressful events in a variety of ways and have usually included only male or only female volunteers, he says. Some reports have defined genetic risk as the presence of two copies of the gene variant, whereas others have required only one. A study Kendler directed found a link between the gene variant and mild stress, but Caspi and Moffitt looked at high stress.

In a joint comment, Caspi and Moffitt say the new meta-analysis underscores the need not for larger samples but for "more research of better quality" into gene-by-environment interactions.

The new meta-analysis, like meta-analyses in general, gave more mathematical weight to studies with larger samples, Caspi and Moffitt note. But in this case, larger studies--containing as many as 4,060 participants--assessed stressful life events and depression symptoms via phone or questionnaires, rather than by comprehensive interviews. "Not surprisingly, these big studies with weak measures did not find positive results, and this tilted the meta-analysis toward a null finding," the scientists say.

Also, the meta-analysis didn't include recent human and animal studies that have linked the gene variant to pronounced hormone and mental responses in stressful lab situations, they say.

Epidemiologist Myrna Weissman of Columbia University notes that the 2003 study set the methodological bar high by identifying stressful life events that occurred before rather than after first episodes of depression. Studies that have failed to replicate those findings fall short of that methodological rigor, Weissman asserts.