CHICAGO—The potential benefits of being able to predict a depressed patient's response to treatment—either early in the course of therapy or before treatment even begins—are substantial. Patients could be matched to the treatment or combination of treatments most likely to improve their depression; therapies destined to fail could be avoided, shortening the time to recovery. The cost of treatment would be reduced, as would the likelihood of adverse effects.

Although the science of predicting treatment response is still in its infancy, over the past decade researchers have made great strides in identifying physiologic and clinical variables associated with eventual remission. As several studies reported at the annual meeting of the American Psychiatric Association (APA) illustrate, this line of research is not only allowing investigators to predict outcome but may also provide clues regarding the physiologic origins of depression.

For example, a series of recent studies from researchers at Weill Medical College of Cornell University has demonstrated the importance of anterior cingulate dysfunction—and associated electrophysiologic markers—as a predictor of treatment response in depression.

In one study, published in the August 1999 issue of the Archives of General Psychiatry,Balu Kalayam, MD, and George S. Alexopoulos, MD, reported that elderly depressed patients who exhibited dysfunction on a constellation of clinical and electrophysiologic measures were unlikely to respond to antidepressants. Specifically, patients who had psychomotor retardation, abnormal initiation/perseveration scores, and long P300 evoked potential latencies—all of which are associated with prefrontal dysfunction—were far less likely than subjects with normal function to respond to nortriptyline (30 patients) or other antidepressants (19 patients). Remarkably, a discriminant function analysis based on these variables correctly predicted 23 of the 24 remitters, giving a 96% sensitivity rate; although the analysis yielded several false negatives, the specificity was still an impressive 66%.

NEW AND IMPROVED

More recently, Dr. Kalayam and colleagues have refined this approach. In an ongoing study described at the APA meeting, the researchers have replaced the P300 measure, which they noted is "rather nonspecific," with a response interference task designed to assess striatofrontal activation. "It is more or less a signature of anterior cingulate functioning," Dr. Kalayam said in an interview. During the task, subjects perform a computerized version of the Stroop Color Interference test; the key predictive variable is the error negative wave (ERN), a component of the evoked response that occurs when subjects make Stroop errors.

Thus far, Dr. Kalayam and colleagues have evaluated 22 depressed geriatric patients without dementia or psychosis and seven normal controls of similar age, sex, education, and cognitive function. Patients underwent clinical, neuropsychologic, and electrophysiologic testing at baseline and then received six weeks of treatment with citalopram, sertraline, or nortriptyline.

The study, like its predecessor, indicated that treatment response could be predicted quite accurately. Once again, nonresponders differed from responders on a measure of initiation/perseveration; no differences were observed on measures of attention, memory, construction, or conceptualization. In addition, the ERN latencies and amplitudes were significantly greater in the left frontal regions of nonresponders than in the right frontal areas; the reverse was true in responders.

"A MAJOR IMPROVEMENT"

From a predictive standpoint, the new data "look strong," according to Dr. Kalayam. The APA presentation did not include a discriminant analysis, but "we anticipate that the [new technique] will be a major improvement" over the P300 approach.

Will clinicians in an office setting be able to assess anterior cingulate function sufficiently well to predict outcome? Although it seems unlikely that ERN testing will become a routine part of treating depression, it may be that the standard Stroop test can serve as a clinical proxy for anterior cingulate dysfunction, allowing it to be used in concert with initiation/perseveration measures to predict treatment response. "These are all instruments that can be administered in an office setting without hardware or capital cost," Dr. Kalayam noted.

SAME REGION, DIFFERENT APPROACH

Findings from a new study conducted at the University of California, Los Angeles, also point to anterior cingulate physiology as a key factor in the pathogenesis and recovery of depression. William F. Stubbeman, MD, and colleagues reported at the APA meeting that pretreatment quantitative EEG of the region overlying the cingulate can accurately predict which patients will respond to electroconvulsive therapy (ECT).

The study included 10 patients with unipolar or bipolar treatment-resistant depression and a mean baseline Hamilton Depression Rating Scale (Ham-D) score of 26.9; depression scores were monitored throughout the study. Patients underwent quantitative EEG the day before their first ECT treatment and again about 24 hours after the initial session.

Dr. Stubbeman, who is a Research Fellow funded by the National Institute of Mental Health and is currently working with Andrew Leuchter, MD, at UCLA, used quantitative EEG cordance—defined as the Z-transformed sum of absolute and relative power—as a measure of cingulate activity; recent studies have shown that cordance is strongly associated with cortical perfusion. When the researchers measured cordance in three apical electrodes overlying the cingulate cortex, they found that all three patients who eventually responded to ECT had positive cordance values before treatment, whereas the seven patients who failed to respond all had negative cordance values.

Equally striking, the three responders had decreased cordance following their first ECT session, indicating a reduction in neurophysiologic activity in the area over the cingulate. This suggests that "responders may have had [ECT-induced] seizure activity in that region, which points to that region being very important in ECT response," Dr. Stubbeman said in an interview. In contrast, six of the seven nonresponders had increased activity in the region following ECT; the only exception had an extremely low decrease. In effect, then, the combination of pre-ECT cordance and post-ECT cordance change was almost 100% predictive of treatment response.

SHOULD ECT TARGET THE CINGULATE?

In implicating the anterior cingulate as a key region in depression physiology and treatment response, the findings are consistent not only with Dr. Kalayam's recent work but with a 1997 study by Helen Mayberg, MD, and colleagues, who found that cingulate hypometabolism was associated with poor response to antidepressant medications. Taken together, these studies suggest that it may be possible to "tell a priori which patients should try ECT and which should try a more complex psychopharmacologic regimen," according to Dr. Stubbeman.

The observation that treatment responders had decreased cingulate cordance following ECT—suggestive of a seizure in the region—has intriguing implications as well. For example, perhaps ECT treatment should focus on inducing a seizure in or near the cingulate. "If that area is the key area, then it should be possible to put ictal electrodes there and increase the response rate for treatment-resistant patients," Dr. Stubbeman suggested.

Importantly, cordance appears to predict response not only to ECT but to antidepressant medications. In a separate study from the UCLA group, also reported at the APA meeting, the researchers observed decreased prefrontal cordance in patients who responded to fluoxetine or venlafaxine, but not in medication nonresponders, placebo responders, or placebo nonresponders. The cordance changes were detectable as early as 48 hours after antidepressant therapy began and occurred with both the seroternergic (fluoxetine) and mixed mechanism (venlafaxine) medications.

THE IMPACT OF UNTREATED ADHD

Finally, two studies presented at the meeting examined the impact of various clinical variables on treatment response. One factor that can greatly reduce the likelihood of successful treatment is comorbidity. In a retrospective analysis of data from an earlier antidepressant trial, Boris Birmaher, MD, and colleagues found that depressed adolescents with untreated attention-deficit/hyperactivity disorder (ADHD) were significantly less likely to respond to treatment than were patients with disruptive disorders or those without comorbidities.

The researchers conducted post hoc analyses of data on 271 adolescents with major depression who participated in a trial comparing the effects of paroxetine, imipramine, and placebo. Thirty-two patients had comorbid ADHD, while 48 had a disruptive disorder, defined as either conduct disorder, antisocial personality disorder, or oppositional defiant disorder.

Among subjects who did not have ADHD, antidepressant response rates were quite high, ranging from 59% in the placebo group to 71% for paroxetine recipients; these rates were higher than those typically seen in medication trials, perhaps because subjects were permitted to receive concurrent psychotherapy. However, even with this option available, relatively few of the adolescents with comorbid ADHD improved with antidepressant therapy. Only two of eight paroxetine recipients (25%), five of 16 imiprimine recipients (31%), and one of eight placebo recipients (12%) achieved a treatment response. In contrast, subjects with disruptive disorder had response rates similar to subjects without such disorders or ADHD.

The findings may stem from the fact that children with comorbid ADHD were required to be off medication for the condition during the trial. Such children may "continue to be impulsive and to do poorly at school and at home," perhaps developing persistent depression in response to these problems, suggested Dr. Birmaher, who is Associate Professor of Child Psychiatry at the University of Pittsburgh School of Medicine. While the study should be interpreted cautiously in light of its post hoc nature, it does suggest that "ADHD is an important consideration in the treatment of adolescents with major depression." Moreover, Dr. Birmaher added, the findings are consistent with his own clinical experience treating children who have both depression and ADHD. "If you don't treat the ADHD, the response to antidepressants is very partial."

PREDICTIVE PERSONALITIES?

Since 1962, at least 26 studies have found that one or more personality factors predict treatment response in depression, according to Andres Heerlein, MD, of the University of Chile. In a new study reported at the APA meeting, Dr. Heerlein and colleagues found two broad sets of personality features that are associated with either good or poor outcome in patients receiving combined therapy.

In the prospective study, Dr. Heerlein and colleagues examined 51 outpatients who met DSM-IVcriteria for a major depressive episode. The patients received six weeks of combination treatment incorporating interpersonal therapy and a selective serotonin reuptake inhibitor, at which point treatment response was assessed using the Ham-D and the Clinical Global Impressions scale. Premorbid personality traits were measured using a clinical checklist.

Dr. Heerlein and colleagues identified numerous premorbid personality traits that were associated with successful outcome at six weeks: stability, perfectionism, extraversion, calmness, rigidity, and lack of fantasy. In contrast, premorbid neuroticism, impulsivity, negativism, anxiety, hostility, infantilism, irritability, and a tendency to blame others were associated with poorer outcomes.

These findings "show very clearly that we have two different clusters [of personality traits] that can predict treatment response to fluoxetine," Dr. Heerlein stated. "This leads us to the conclusion that there is strong evidence that premorbid personality traits are predictors of treatment response in major depression." It further suggests, he said, the possibility that a nosological differentiation can be made between "two main subtypes of depressive patients."

—Peter Doskoch

Suggested Reading
1. Kalayam B, Alexopoulos GS. Prefrontal dysfunction and treatment response in geriatric depression. Arch Gen Psychiatry.1999;56:713-718.
2. Mayberg HS, Brannan SK, Mahurin RK, et al. Cingulate function in depression: a potential predictor of treatment response. Neuroreport.1997;8:1057-1061.

Return to table of contents