There is much to like about the Treatment of Resistant Depression in Adolescence (TORDIA) study (1) and the 24-week follow-up article in this issue of the Journal (2). Foremost, TORDIA is a tribute to hard work and perseverance. The authors, a terrific group of investigators, took on an extremely challenging clinical issue at a time not conducive to antidepressant research. In the middle of the study, the Food and Drug Administration reported on the association of youth suicidal behavior with antidepressant use (3). Consequently, the study was shut down until the investigators, the institutional review boards, and the National Institute of Mental Health (NIMH) could find a way to safely continue the study. Once the study was restarted, the TORDIA team had to figure out how to resume recruitment during a time of great uncertainty and patients' reluctance to use selective serotonin reuptake inhibitors (SSRIs). The delays in the study would have spelled failure for most investigative groups, but this group of investigators got the study up and running again, then applied for additional NIMH funding, convincing the peer reviewers and NIMH to see the project through to its conclusion. The fact that such a large study received additional funding is a testimony to the credibility of the research group and the importance of the topic. If TORDIA was not done then and by this group, it was likely never to be done.

The TORDIA 24-week outcomes report the rates of remission/recovery, relapse, and, importantly, the characteristics of subjects that predicted outcome. To summarize, TORDIA enrolled teens ages 12—18 years (N=344) who had failed a previous trial of an SSRI and randomly assigned them to a medication switch only (another SSRI or venlafaxine) or a medication switch plus cognitive behavioral therapy (CBT) (another SSRI plus CBT or venlafaxine plus CBT). Subjects were treated for 12 weeks, and then week-12 responders were continued in their assigned arms and followed until week 24. Roughly 40% of those who failed a previous trial of an SSRI reached remission by week 24, regardless of treatment group. Responders at week 12 reached remission more quickly and at higher rates than those who responded after 12 weeks. Lower baseline depression, hopelessness, and anxiety predicted higher rates of remission, as did lower week-12 ratings of depression, hopelessness, anxiety, suicidal ideation, and family problems as well as the lack of dysthymia, anxiety, and drug use.

With TORDIA, the Treatment for Adolescents with Depression Study (TADS) (4), and the Adolescent Depression Antidepressant and Psychotherapy Trial (ADAPT) (5), we have a pretty clear picture of what to expect from our best strategies for treating teen depression. TADS suggests that upward of 60%—70% of teens with moderate-severe depression will respond to medication or medication and CBT. TADS also suggests that younger, less impaired, and less comorbid patients do better with treatment generally and that combined treatment was robust to any moderating factor. Importantly, patients who had severe and persistent depression benefited equally from medication alone or combined CBT and medication (6). TORDIA suggests that of those who fail that first test of medication, approximately 40% will remit to the next antidepressant trial. In the ADAPT trial, nonresponders to a brief intervention (N=128) and those too ill for the brief intervention or already on medication (N=85) were evaluated, and those appropriate for entry (N=208) were randomly assigned 1:1 to an SSRI or an SSRI plus CBT. Response rates at 12 weeks were 41.6% in combined treatment and 43.6% in SSRI only treatment. The lower 12-week response rates relative to TADS may reflect the more severe baseline status of ADAPT subjects or may reflect the exclusion of brief intervention responders, which may have reduced the overall number of responders in the main trial (N=34).

TORDIA, TADS, and ADAPT all followed subjects past week 12. Maximum response rate in TADS was 80%—85% at week 36, and ADAPT estimated an 80% response rate over the course of the trial, with 10% persistently refractory.

What is the role of psychotherapy in the treatment of teen depression? A number of trials have demonstrated the efficacy of CBT relative to a psychosocial control (7). However, in treatment trials involving medication, the data are less clear. In the TADS acute phase, CBT alone was not significantly more effective than medication management with placebo, except in those who had milder symptoms and shorter duration of illness and in those whose family had higher incomes. CBT and medication was better than medication alone on some outcomes but not for the more severely affected, where the addition of CBT to medication did not offer additional benefit. TADS did identify that the addition of CBT to medication may have a protective effect on the risk for suicidality observed in the medication alone group. In the ADAPT trial, the addition of CBT to medication did not significantly improve outcome and did not identify either a risk for increased suicidality in those on medication or a protective effect of CBT on suicidality. In the TORDIA acute phase, the groups getting combined treatment had an approximately 10% greater response rate, but this between-group difference did not persist to week 24. TORDIA, like ADAPT, did not find a signal for SSRI-associated suicidality or for the protective effects of CBT.

There is probably a role for CBT alone for milder and shorter duration depressive illness and in those who might be considered ideal candidates for psychotherapy. However, it is very difficult to argue that CBT is not helpful at all for those with more severe depression, but the data do not support either the use of CBT as first-line treatment or the utility of CBT as an adjunct to medication for severely ill patients. The data from the ADAPT trial is particularly clear on this point.

Is there a problem with CBT in general, or do manualized CBTs lack the ability to address issues often present with severe depression (anxiety symptoms, drug use, suicidality, family problems), or is there simply a problem of measurement? Most manualized CBTs focus on depression symptoms and do not include evidenced-based modules for the moderators of outcome including anxiety, drug abuse, suicidality, or substantial family problems. The lack of these components in the CBT used in teen depression trials may have limited the usefulness of CBT. Also, current depression rating scales focus on the symptoms of depression but not necessarily on the targets of CBT, so our measures of treatment outcome may not capture CBT's effects.

What should we expect from medication treatment? There is a group of depressed patients who have not been exposed to antidepressant medication who respond rapidly to treatment. Even among those who have failed one antidepressant, there is a group of patients who respond quickly to a switch in medication, even as early as week 6. There is probably little reason why these teens should not routinely be identified and successfully treated to remission and recovery.

Based on these data, how do we assure that these teens with depression benefit from treatment? In all of these clinical trials, the clinic visits were frequent and dose adjustments brisk. In my clinical experience, the pace of research care is much faster and more assertive than what I observe in standard clinical care, where "start low and go slow" is often the mantra. Teens with depression might have better outcomes in clinical practice if their prescribers mimicked the pacing and dosing of medication treatment in these clinical trials. For those who are more complex and who may take longer to remit, it is probably more important to adjust dosing quickly and to use adequate doses to either establish the capacity to respond or to take the next step, a switch in antidepressant treatment. How long to wait before switching antidepressants is not fully established, but remitters usually demonstrate improvement by 8—10 weeks. Minimal response or failure to respond by 8—10 weeks does not preclude later improvement, but clinicians and patients should not let grass grow under their feet and should be prepared for the management of resistant depression, a la TORDIA, early in treatment.

What about those who failed to respond in these studies? We still have much to learn. Improving CBT response seems a reasonable goal. Including modules for specific depression issues, as is done in interpersonal psychotherapy (8), may be more helpful than generic CBT, but as these complex patients have other issues, implementing evidence-based modules for anxiety (exposure and response prevention), drug abuse, suicidality, family problems, etc., may go a long way to reducing depression in teens. Although adult studies would suggest that augmentation with lithium or a thyroid hormone or adding a second antidepressant of a different class might be helpful (9), those strategies have not been evaluated in teens.

Perhaps the most important step in improving outcomes for teen depression is to make sure that teens get to the clinic and get there early in their course of illness. There has been a lot of public chatter about how antidepressants are not effective or are harmful for teens that may be keeping teens and their families away from treatment. Investigator-initiated studies such as TADS, APADT, and TORDIA (as opposed to industry-sponsored studies, which are likely fundamentally flawed (10)—a topic for another editorial) are unequivocally clear that treatment for teen depression that includes medication is effective and can be implemented safely. Hopefully, broadly disseminating the results of TORDIA, TADS, and ADAPT can improve outcomes for depressed teens.