Researchers Closer to Matching Best Treatments to Depressed Teens

Researchers at the National Institute of Mental Health (NIMH) are seeing some intriguing patterns as they continue to analyze data from the NIMH-funded Treatment of Adolescents with Depression Study (TADS).

Although the analyses are preliminary, NIMH scientists are hopeful that the data may give clinicians evidence-based indications of which types of treatments will most benefit specific patients.

Some of the preliminary analyses were presented in May at APA's annual meeting in Atlanta by Benedetto Vitiello, M.D., chief of NIMH's Child and Adolescent Psychiatry Branch.

“We have been conducting secondary analyses of the [initial 12 weeks of data from the] TADS database, looking at two research questions,” explained Vitiello. “The aims of the subanalyses were to identify a subgroup of patients who are more likely to improve during treatment for depression, regardless of which type of treatment they receive, and a subgroup or groups of patients who tend to respond to a specific type of treatment, compared with other available treatments.”

The new secondary analyses—which have yet to be published—are aimed at identifying what Vitiello termed “mediators,” which he defined as “factors that may explain which direction to pursue in treatment.” However, the mediator analysis was not completed by the annual meeting presentation.

Vitiello instead discussed what NIMH researchers have discovered with respect to “predictors” and “moderators” of treatment in adolescents with depression. He defined a predictor as a variable that is measured prior to starting treatment that identifies a group of patients whose depression will improve, regardless of the treatment to which they are randomly assigned. A moderator of treatment, Vitiello said, is a variable that is measured prior to the start of treatment, but identifies a group of patients who tend to improve preferentially to one treatment compared with another.

New Hypotheses Generated

Being able to predict which specific treatment for adolescents with depression is most likely to achieve the best response, Vitiello noted,“ is of course akin to finding the Holy Grail.” But, he added,“ we are not going to be able to issue any clinical recommendations based upon” the secondary analyses of the TADS data. He pointed out that these analyses are not hypotheses testing; rather, they are hypotheses generating and may lead researchers to questions that need to be addressed in future research.

TADS researchers looked at a large group of potential predictors, moderators, and possible mediators, Vitiello said. “We looked at demographic variables, severity and duration of illness, presence of comorbidity, parental psychopathology, socioeconomic variables, and treatment expectancy of the patient and [his or her] parent.”

Predictors Identified

The TADS researchers found that predictors of improvement included age: those younger than 16 were more likely to be improved at 12 weeks, regardless of the type of treatment they received. Patients who were less functionally impaired—with a Children's Global Assessment of Function Scale score of greater than 50—were more likely to improve with treatment. Also, the shorter the duration of the current depression episode, the more likely patients were to improve.

Patients with fewer melancholic features were more likely to improve than patients with more melancholic features or higher levels of hopelessness. Patients without an anxiety component to their depression and a lower level of suicidal ideation were also more apt to improve regardless of treatment type.

“So far,” Vitiello said, “everything seems to make sense. [These are good] data that confirm the experience of a lot of us in clinical practice.”

He also noted that several variables that researchers thought would affect likelihood of improvement did not show any statistically significant effect. Neither the patients' gender nor their race/ethnicity showed a statistically significant effect on likelihood for improvement. They also found that a patient's method of recruitment into the TADS trial (physician referral compared with recruitment through advertising) showed no statistically significant effect.

“I.Q. was also not a variable that influenced outcome,” Vitiello noted, “and this was a bit remarkable to us, because CBT certainly requires the cognitive abilities to understand and apply the principles underlying CBT and then do the follow through.”

Comorbidity for dysthymia, ADHD, and conduct disorder also did not predict outcome.

Medication Indicated for Severely Ill

But perhaps the most intriguing part of the analysis, Vitiello said, was that depression severity was moderating the response to CBT—yet not in the way the researchers had hypothesized.

“The interpretation of this is just a bit counterintuitive,” Vitiello explained. “We had postulated a priori that the sickest patients would be more likely to have the greater response to combined treatment. But that hypothesis was not confirmed.”

Instead, “It looks like for patients who [have severe depression], the advantage of the CBT component is not that apparent. It may be that a patient who is that sick is not able to follow CBT particularly well, and what is needed at that point is medication. As the depression gets less severe, then the advantage of using CBT in addition to medication may become apparent.”

For patients with severe depression, CBT alone appeared to be no better than placebo, while fluoxetine alone and in combination with CBT were both significantly better treatments than placebo. CBT and the combination, however, were not statistically different.

Simply being randomized to the CBT group was associated with significantly higher odds of improvement for those patients with higher levels of cognitive distortion, for example, extreme or inappropriate negativism or guilt. For patients with more severe cognitive distortion, however, the combination treatment was again the most effective, followed by fluoxetine alone and then CBT alone—all three of which were significantly more effective than placebo.

Income-Associated Data Unexpected

Finally, Vitiello said, “another variable unexpectedly turned out to be a significant moderator, and this may require further analysis: family income.” Income was the lone variable the researchers used to indicate socioeconomic class, Vitiello added, “but is probably only a proxy for other variables, like education, for example.”

The researchers found that adolescents whose annual family income was above $75,000 responded equally to the combination treatment, CBT alone, or fluoxetine alone. Intriguingly, he said, “in the group with family income below $75,000, there was a statistically significant advantage of using medication, either alone or in combination with CBT.”

Taken together, Vitiello concluded, “these findings offer the opportunity to generate new hypotheses about which kids will benefit more from which treatment modality. Much of the analysis made intuitive clinically relevant sense, and the findings may lead to means of early detection and [allow] directed treatment—before the adolescent's depression becomes chronic, ingrained, and therefore more difficult to treat.”

More information on TADS is posted online at<www.nimh.nih.gov/healthinformation/tads.cfm>.▪