Thursday, February 25, 2010

In Search of the Best Medicine for Depression - University of Pennsylvania

New research explores why different patients respond better to different treatments.

Scientific meta-analysis may not be a typical way to make headlines. But psychology doctoral student Jay Fournier found himself a media magnet last month following the publication of a paper in theJournal of the American Medical Association (JAMA) that examined antidepressant-drug treatment data from studies conducted in each of the last three decades. Fournier and his coauthors, including his mentor, Samuel H. Preston Term Chair in the Social Sciences Robert DeRubeis, reported that for people with mild to moderate depression, medication failed to outperform placebos. The story received wide coverage, nationally and internationally, including a debate about interpretation and methodology that played out in the science pages of the New York Times.

While the JAMA paper received a surprising amount of attention, what is perhaps more remarkable is how little research has been done to date on the effectiveness of antidepressants for people with mild to moderate depression. Thousands of studies have demonstrated that antidepressants work well in treating patients with severe depression. But evidence from clinical trials involving people with less severe forms of depression was, according to Fournier, hard to come by. By including this subset in his analysis, Fournier and his coauthors have provided evidence that not all depression is the same, and that medication may not always be the best treatment.

Fournier cautions that their results do not indicate that medication is ineffective, but they do suggest just how effective placebos are for people with mild to moderate depression. "There are a series of steps," Fournier notes, "that someone needs to take in order to get a placebo in a clinical trial, and these steps alone seem to have a powerful effect. We don't fully understand how placebos work, but it seems that part of the benefit comes from patients taking their depressive symptoms seriously and acting on their concern about their own mental health."

Fournier's dissertation explores this notion—that the decision to take active steps is, in itself, powerful therapy for many people—from a slightly different angle. Rather than comparing medication to placebos, Fournier has compared medication to cognitive therapy. His work builds upon an authoritative 2005 clinical study by DeRubeis and a team of researchers at Penn and Vanderbilt University which found that for moderate to severe depression, cognitive therapy works as well as antidepressants. The focus of Fournier's work is whether there are patient-specific factors that determine which individuals fare better with medication, and which respond to cognitive therapy.

To assess this, Fournier carried out three studies examining the relationship of treatment outcomes with patient characteristics, types of depressive symptoms, and the presence or absence of personality pathology. "Cognitive therapy," Fournier found, "was more effective for patients who are married or cohabit, who are unemployed, who experienced a large number of life events, and with no diagnosis of personality disorder." Further, he found that cognitive therapy was better at treating "atypical vegetative symptoms" of depression, such as sleeping or eating too much. In contrast, medication seems to be the better treatment for patients who in addition to depression have a personality disorder.  Patients with chronic depression, older patients, and patients with lower intelligence experienced poor outcomes with both cognitive therapy and medication.

In demonstrating that different treatments are appropriate for different patients, Fournier's research is consistent with the larger movement toward personalized medicine—a revolution being fueled in many medical specialty areas by the availability of molecular and genetic data. But at this point, according to Fournier, his research might best be considered "informative." To get to the point where it's possible to custom-match treatments to patients with depression, he notes that the first step will be for another study to replicate his findings. Beyond that, Fournier notes, "Whenever we see a subgroup of patients that responds differently to one treatment over another, it tells us that there might be differences in mechanisms of response, and the variables give us clues about the mechanisms." Pursuing these clues will remain Fournier's priority after getting his doctoral degree this May.