HONOLULU— In the past decade, childhood depression, long under-diagnosed and under-treated, has been increasingly recognized as a significant public health problem. Left untreated, depression can develop into a chronic and often lifelong affliction; even if the illness remits, its occurrence during childhood or adolescence may interfere with normal biological and psychosocial development.

As with many psychiatric disorders, treatment strategies that are effective for adults cannot necessarily be extrapolated to children. For example, at least one class of antidepressants seems to be largely ineffective in young patients. Unfortunately, clinicians seeking evidence-based approaches to treating childhood depression will find few large clinical trials to guide them. The National Institute of Mental Health (NIMH), spurred by this dearth of data, has launched several important studies in recent years; various pharmaceutical companies, spurred by the lure of patent extensions, have done the same. Still, only a handful of multicenter trials have been completed thus far.

At the annual meeting of the American Academy of Child and Adolescent Psychiatry (AACAP), Graham Emslie, MD, Professor of Psychiatry at the University of Texas Southwestern Medical Center in Dallas, reviewed the key antidepressant trials in pediatric patients and discussed some of the clinical issues that physicians should keep in mind when diagnosing and treating childhood depression. For example, one important issue, which applies equally to adults, is that the goal of treatment should be remission rather than response. There is “increasing evidence that if people have residual symptoms of depression at the end of the treatment period, they will be much more likely to relapse and to have continuing dysfunction,” Dr. Emslie emphasized. Unfortunately, most antidepressant studies, whether involving adults or children, have not reported remission data.

THE STORY SO FAR

At this point, the best studied of the effective antidepressants in children is, more or less by default, fluoxetine, which has been the focus of two large placebo-controlled trials and many smaller, open ones. The first of the two multicenter trials, published in 1997 by Dr. Emslie and colleagues, was funded by the NIMH and featured an unusually extensive pre-randomization period. Each of the 96 subjects, ages 8 to 18, underwent three evaluation visits (two of them involving a structured diagnostic interview) and then began a weeklong placebo run-in period. If subjects still met criteria for depression, they were randomized to eight weeks of treatment with either fluoxetine (20 mg/d) or placebo; they were seen weekly throughout the treatment period.

The results were generally encouraging: Fluoxetine recipients were significantly more likely than placebo recipients to be rated “much” or “very much” improved on the Clinical Global Impressions scale (56% vs 33%). That difference in response rates is roughly the same as the 20% drug–placebo differential typically reported in adult trials, Dr. Emslie noted. However, remission rates were fairly low in both groups (31% and 23%, respectively).

A second large fluoxetine trial, reported at the 2000 AACAP meeting but not yet published, paralleled the first trial in most respects. The 219 children and adolescents in the study received either fluoxetine (10 mg/d for one week, then 20 mg/d for eight weeks) or placebo and were evaluated weekly. Here, statistically significant differences in Children’s Depression Rating Scale (CDRS) scores were already apparent at week 1 and continued throughout the study. Treatment response, defined as a 50% reduction in a patient’s CDRS score, was observed in 57% of fluoxetine recipients and 40% of placebo recipients; the corresponding rates for remission (defined as a final CDRS score of 28 or less) were 41% and 19%, respectively. Further improvement was apparent at a follow-up evaluation several weeks after the trial’s end. “Eight weeks is still short in terms of getting remission of symptoms,” Dr. Emslie noted.

A third multicenter trial, published last year, was the first to directly compare a selective serotonin reuptake inhibitor (SSRI) with a tricyclic. In this trial, which, like the previous one, was industry sponsored, 275 adolescents were treated with either paroxetine (mean dosage, 28 mg/d), imipramine (mean dosage, 205 mg/d), or placebo for eight weeks following a run-in period of seven to 10 days. All patients also received supportive therapy.

Using a final Hamilton Depression Rating Scale score of 8 or less as a benchmark for good outcome, the investigators found no difference in response rates between imipramine (50%) and placebo (46%). Paroxetine, however, was superior to both, with a 63% response rate. Similarly, on the Clinical Global Impressions scale, patients were more likely to have been much or very much improved if they had received paroxetine (66%) rather than imipramine (52%) or placebo (48%). This poor showing for the tricyclic antidepressant has proven to be a recurring theme in the childhood depression literature.

MIRTAZAPINE: A NEGATIVE TRIAL—OR JUST A FAILED ONE?

To date there have been no published controlled pediatric trials of the new dual-mechanism antidepressants (eg, venlafaxine, nefazodone). However, Dr. Emslie shared some data from a recent unpublished industry-sponsored mirtazapine trial that, unfortunately, failed to show that the drug is superior to placebo.

The study actually consisted of two identical but separate 17-site trials, each involving different patients and treatment locales. The 250 moderately depressed participants (ages 7 to 17) who participated in either of the trials were randomized to mirtazapine or placebo at a 2:1 ratio; those in the active treatment group were started on 15 mg/d but were allowed dosages of up to 45 mg/d, at the treating doctor’s discretion, during the second half of the eight-week studies. In the first of the two trials, the proportion of patients showing improvement on the Clinical Global Impressions scale was nearly identical in the mirtazapine (59%) and placebo (56%) groups. The response rates diverged a bit more in the second study (53% vs 41%) but still fell short of statistical significance.

Given the relatively high placebo response rates, particularly in the first trial, it is possible that these findings reflect a failure to find positive effects (a “failed trial”) rather than an actual lack of efficacy (a “negative trial”), according to Dr. Emslie. Inclusion in the trials of a second drug known to be effective in this setting, such as an SSRI, would have shed light on the issue. “It’s not likely that fluoxetine or paroxetine would have beaten a 56% placebo response rate either,” Dr. Emslie noted. (As an aside, he complimented Organon, the drug’s manufacturer, for giving permission to present the mirtazapine data, noting that many industry-sponsored studies that fail to show benefits “never see the light of day.”)

Although the study did not provide evidence of efficacy, its safety data may be a source of reassurance to clinicians who wish to try a course of mirtazapine in appropriate patients, Dr. Emslie said. The investigators found that mirtazapine is well tolerated by children and adolescents: The proportion of participants who dropped out due to adverse events was similar in the two treatment groups.

TRICYCLICS: NOT A FIRST CHOICE—OR A SECOND, EITHER

For adults, the tricyclic antidepressants have fallen to second- or third-line status because their side effect profile is worse than that of newer medications. The same is true for the treatment of children, although here the drawback of adverse events is compounded by less-than-impressive treatment outcomes. For example, a 1995 meta-analysis of 12 small tricyclic trials (only one involved more than 50 patients) found no difference between tricyclics and placebo. “That really hasn’t changed with new data since 1995,” Dr. Emslie noted; indeed, a small desipramine trial reported at the AACAP meeting did not show any benefits.

One of the best-designed and largest tricyclic trials, again involving desipramine, was published by Kutcher and colleagues in 1994. Despite appropriate dosing, the response rate for desipramine recipients was only 48%, not significantly better than that of the placebo group (35%).

TREATMENT–RESISTANT CASES

Treatment resistance is a common phenomenon in pediatric depression, as illustrated by the fact that 30% to 40% of patients failed to show a response even in the successful trials cited above. Predictors of poor outcome include severe depression, comorbidity, family discord, and high levels of impairment. The presence of parental depression is not an important predictor of poor response for the child unless the parental depression is not being treated, according to Dr. Emslie.

One step that clinicians can take to improve outcome is to schedule the patient for weekly or biweekly office visits during the first month of treatment. “We figure that if children are severely depressed enough to require medical treatment, they should be seen reasonably frequently,” Dr. Emslie stated. Substantial clinical improvement is unlikely at this early stage of treatment, but office visits during this period may allow the clinician to “catch” patients who would otherwise drop out of treatment due to adverse effects or other problems.

At present there is “no good empirical evidence for what you should do if the first line of treatment fails,” Dr. Emslie noted. However, “one thing that we do is reassess the original diagnosis” to see if the patient’s symptoms are due to bipolar disorder or unrecognized substance abuse. He also checks for medical illness, poor family functioning, and comorbid psychopathology and investigates whether the patient has been adhering to treatment.

When an adequate dose of appropriate medication fails to yield a response, the key clinical decision, of course, is “switch or add?” Dr. Emslie noted that “there is very little data on augmentation strategies at the moment, but what little there is probably supports the addition of bupropion.” Lithium has also shown some benefits in studies. The potential advantages of switching include the avoidance of drug interactions, lower cost, better adherence, and the possibility that a drug that has a different mechanism of action may be more successful than the current medication. The disadvantages of switching include the loss of any partial response, a possible lag in response time, and the loss of drug synergy. The NIMH-sponsored TORDIA trial (Treatment of Resistant Depression in Adolescents), involving more than 400 adolescents who failed to respond to treatment with an SSRI, should offer guidance on the switch-or-add issue, but the study was just launched in 2001 and presumably won’t yield results for several years.

Another important option in cases of treatment failure is to add or switch to cognitive behavioral therapy (CBT). Although CBT has proven effective in adults, it is not appropriate for young children; for adolescents, who often do well with CBT, it remains unclear whether this therapy is as effective as medication. Once again, a multisite NIMH study that is currently being conducted should shed light on the issue. Participants are randomized to 12 weeks of treatment with either fluoxetine or CBT, both treatments, or placebo; follow-ups are conducted at one year. Dr. Emslie said that although he doesn’t expect to see substantial differences between the treatment groups after 12 weeks, “what I would hope is that the combination treatment will show less relapse and better social function at the end of the one-year follow-up period.”

A RECURRING PROBLEM

Depression is highly recurrent in pediatric patients—perhaps even more so than in adults, according to Dr. Emslie. Although the scope of the problem has not been well delineated, existing naturalistic data “is a bit worrying,” he said. In one study, 72% of recovered patients had another episode within five years. In the first fluoxetine trial described above, follow-up of 87 participants (91% of the sample) revealed that 85% had recovered within a year—and that 39% of these had gone on to have another episode after at least two months of being well, in some cases despite continued treatment.

The second fluoxetine trial has also yielded some discouraging findings on this issue. A subsample of 40 children who had responded to acute treatment with fluoxetine were randomized either to stay on the drug or to switch to placebo. Follow-up evaluations revealed that within six months, 60% of placebo recipients and a third of fluoxetine recipients had relapsed (defined as a CDRS score of at least 40 with two consecutive weeks of clinical deterioration). Mean time to relapse in the placebo group was 71 days. “So [depression] is, at least in these samples, a highly recurrent disorder,” Dr. Emslie reiterated.

Not surprisingly, there is little published data on the value of continuation therapy and no data at all on maintenance treatment, Dr. Emslie said. When relapse does occur, however, his first approach would be to increase the dose of the medication that the patient is currently taking or had last responded to. His next step would be to switch medications or to add CBT.

ON THE HORIZON

Pediatric trials of most of the commonly used antidepressants, including sertraline, venlafaxine, nefazodone, and citalopram, have recently been completed or are well under way, so data from these trials will be emerging during the next few years. Another paroxetine trial has also been conducted.

But beyond the value of specific medications, other clinical issues that affect outcome urgently need further study, Dr. Emslie said. Improving compliance remains an important goal; approaches such as weekly dosing may help. The relative benefits of pharmacological and psychosocial interventions (as well as the combination of the two) need to be determined, as does the optimal time course of these treatments. Identifying shorter, “watered down” psychotherapeutic regimens that are effective will also be important. Finally, Dr. Emslie said, we need more data on remission (as opposed to response) as a predictor of treatment outcome, and better information on the long-term effects of both antidepressant use and the illness itself.

—Peter Doskoch

Suggested Reading
1. Emslie GJ, Rush AJ, Weinberg WA, et al. A double-blind, randomized, placebo-controlled trial of fluoxetine in children and adolescents with depression. Arch Gen Psychiatry. 1997;54:1031-1037.
2. Hazell P, O’Connell D, Heathcote D, et al. Efficacy of tricyclic drugs in treating child and adolescent depression: a meta-analysis. BMJ. 1995;310:897-901.
3. Keller MB, Ryan ND, Strober M, et al. Efficacy of paroxetine in the treatment of adolescent major depression: a randomized, controlled trial. J Am Acad Child Adolesc Psychiatry. 2001;40:762-772.

Return to table of contents