In a separate study to be published in the American Journal of Psychiatry, McMahon and his colleagues assessed whether 68 genes involved in neural signaling were linked to suicidal thinking induced by citalopram. Preliminary evidence suggests that markers in two genes involved in chemical signaling by the neurotransmitter glutamate may play a role. Psychiatrists caution that both these and earlier studies assess suicidal thinking, which is much more common than suicide itself. As with all genetic studies of this type, all three candidates must be confirmed in other populations.

While the findings are promising, none of the variations uncovered to date can predict suicidality well enough to make them the basis for a clinical test. For example, according to McMahon, the two variations his group identified could predict about 60 percent of people at risk. "But we would miss 40 percent," he says. Scientists hope that a combination of several such variants could be much more accurate, and McMahon and others are now searching through the entire genome, rather than just the candidate genes previously studied, for additional markers.

Both McMahon and Perlis are also searching for genetic variants that predict who will respond to particular drugs. "It would be best to be able to pair the two tests so you can weigh the benefits and the risks," says David Brent, a psychiatrist at the University of Pittsburgh not involved in the study. He adds that people who are identified as at risk may not need to skip treatment altogether--they might need additional monitoring by a psychiatrist, for example.

Scientists aren't sure why drugs that usually reduce suicidal thoughts seem to trigger them in a small subset of people. While antidepressants can take weeks to exert their mood-lifting effects, biochemical changes occur right away. "Perhaps suddenly elevating serotonin in the brain causes imbalances in certain people that evoke symptoms like nervousness, sleeplessness, and maybe suicidal thinking," says McMahon.