Atypical Depression, Panic Attacks, and Response to Imipramine and Phenelzine A
Replication
Frederic M. Quitkin, MD; Patrick J. McGrath, MD; Jonathan W. Stewart, MD; Wilma Harrison, MD; Elaine Tricamo, RN; Steven G. Wager, MD; Katja Ocepek-Welikson, MPhil; Edward Nunes, MD; Judith G. Rabkin, PhD; Donald F. Klein, MD
\s=b\ In an initial study with 120 patients with reactive mood and associated atypical symptoms, phenelzine sulfate was superior to imipramine hydrochloride and placebo. Since their response to phenelzine appears to be unique, this suggests that atypical depression may be a distinct subgroup of unipolar depressive illness. Unexpectedly, the benefit of antidepressants was limited to patients who also had spontaneous panic attacks. To help establish the validity of this syndrome, a new sample of 90 atypical depressives was studied. The clinical and demographic characteristics of the original and replication sample were virtually identical at baseline. In addition, the treatment response with either placebo, imipramine, or phenelzine was also indistinguishable in the two patient groups. The outcome in the replication study supports the hypothesis that this may be a distinct unipolar depressive subgroup. In the replication sample, a history of panic attacks did not appear to be a relevant predictor. We discuss the explanations for this discrepancy in the two patient
samples. (Arch Gen Psychiatry. 1990;47:935-941) of an the group from monoamine oxidase in¬ who hibitors more than 20 years ago, was first validation has proved to be elusive." Patients responsive to MAOIs, described by early investigators, appear to have
concept atypical depressive (AD) Although selectively benefited discussed (MAOIs) .
heterogeneous symptom profiles, including phobic anxiety (panic disorders) and vegetative atypical depression characterized by hypersomnia, lethargy, and hyperphagia.6 The inclusion of patients with panic disorder, a disorder clear¬ ly responsive to tricyclic antidepressants (TCAs), might have obscured the identification of a superior MAO response in a vegetative atypical group.6"8 We studied depressed patients with a nonautonomous mood who met DSM-III for a major or dysthymic mood to deter¬ mine if distinct syndromes could be identified in this heteroge¬ neous population.9'11 Approximately half of the patients had
included
Accepted for publication December 7,1989. From the Department of Psychiatry, Columbia University, New York, NY, and the New York (NY) State Psychiatric Institute. Reprint requests to New York State Psychiatric Institute, 722 W 168th St, Box 12, New York, NY 10032 (Dr Quitkin).
some spontaneous panic attacks. It was not clear whether those with spontaneous panic attacks, as well as AD symp¬ toms, would have the same treatment response as those with only atypical depression. Our concern was that patients with both atypical depression and panic attacks might benefit from imipramine hydrochloride, which would obscure selective responsivity to MAOIs in others. Patients were stratified by the number of atypical symp¬ toms, presence of panic attacks, and evidence of hysteroid dysphoria (Table 1). The associated atypical symptoms used to classify patients included hyperphagia, hypersomnolence, a leaden feeling, and rejection sensitivity (defined below). Pa¬ tients with at least two associated symptoms were considered definite atypical depressives (DADs), and those with one associated symptom were considered probable atypical depressives (PADs). Patients with only a reactive mood were called simple mood reactive depressives. Hysteroid dysphoria is a proposed depressive subtype characterized by a history of depression associated with a disappointment in a romantic context, histrionic personality, and excessive need for atten¬ tion.12,13 Depressive exacerbations of hysteroids are character¬ ized by vegetative atypical symptoms. Two main points emerge from our first study of 120 DADs9: (1) phenelzine sulfate was superior to imipramine and placebo, and (2) unexpectedly, the superiority of both imipramine and phenelzine to placebo was limited to patients with spontane¬ ous panic attacks and/or hysteroid dysphoria. Unreplicated clinical trials, especially those with unantici¬ pated results, should be greeted with skepticism. The promi¬ nent role of panic attacks in identifying drug-responsive disor¬ ders was not predicted and required cross-validation. Therefore, a replication with an independent sample of DADs
undertaken. The following questions were asked: (1) Will the clinical picture of patients referred to as ADs be the same in two independent samples? (2) Will ADs treated with phenelzine have a superior response compared with those treated with imipramine or placebo? These data were collected during approximately an 8-year period. If the two samples had simi¬ lar baseline characteristics and treatment response, it sug¬ gests this syndrome had a predictable phenomenology and treatment response.14 (3) Will a history of panic attack continwas
Downloaded From: http://archpsyc.jamanetwork.com/ by a Cornell University User on 09/01/2016
Table 1 .—Stratification of Patients With Nonautonomous Mood* Stratification Definite atypical depressives Mood reactive Has at least 2 of 4 associated
symptoms (hyperphagia, hypersomnolence, rejection sensitivity, and leaden paralysis* Subdivided into those with and without any spontaneous panic attacks
Probable atypical depressives Mood reactive Has at least 1 of associated symptoms Subdivided into groups with and without any spontaneous panic attacks Mood reactive depressives Mood reactive No associated symptoms Subdivided into groups with and without spontaneous panic attacks
Hysteroid dysphorics—a proposed depressive subtype characterized byt
Pathologic rejection sensitivity in a romantic context
When
rejected, develops depression characterized by reactive atypical symptoms *See text for definition of associated symptoms. mood and associated
fAII hysteroid dysphorics met criteria for definite atypical depression.
to be associated with a greater drug-placebo difference? (4) Are anxious depressives (with or without panic attacks) more likely to benefit from MAOIs than depressed patients without anxiety? (5) Does the drug response add to the relevance of the can hysteroid dysphoric subtype? (6) With a larger sample, more narrow confidence limits of the responsivity of ADs to each of the treatments be developed? ue
NATURE OF THIS REPORT
We report on a clinical trial with an independent sample of 90 ADs and will contrast outcome in this sample with the original sample of 120 ADs.9 The only inconsistency found between the two samples of DADs was the relevance of panic as a predictor of drug effect. In the original study, a history of panic appeared to be a relevant predictor, but in the replication study, it was not. We recently published an article that suggested that PADs (char¬ acterized by one atypical symptom) are indistinguishable from DADs.10 We therefore considered reactive mood and presence of any associated atypical feature sufficient criteria for atypical depression. The data from PADs will be used to help clarify the discrepant effect of panic in the two samples of DADs. To summarize, the following data are presented and ana¬ lyzed: 90 DADs not previously described, 120 DADs with two or more symptoms previously reported,9 and data for 60 PADs with one associated symptom previously reported.10 METHOD The study procedures, including the dose of drugs, rating scales, biochemical measures, and reliability assessment were virtually identical to those described in the study of patients with definite atypical depression.9 Therefore, only a brief review of the method is given. The minor change in the dose schedule is described below. Patients included in the study all attended an outpatient, universityaffiliated clinic and gave written informed consent. The following criteria were required. 1. Met Research Diagnostic Criteria for major, minor, or intermit¬ tent depression.15 2. Maintained mood reactivity while they were depressed, defined as present if mood could be lifted at least 50% of when they were not depressed (scale and anchors available from authors). 3. Exhibited at least two of the following: increased appetite or weightgain while they were depressed (defined as present if appetite
markedly increased or there was at least a 4.5-kg weight gain), oversleeping (sleeps at least 10 h/d), severe fatigue that created a sensation of leaden paralysis or extreme heaviness of arms or legs
(defined as present if a marked decrease in energy or heaviness of limbs was a prominent symptom), and sensitivity to rejection as a trait throughout adulthood (defined as present if rejection resulted in depression with functional impairment, eg, the subject missed work or
school).
4. Were between 18 and 65 years of age. 5. Had no other Axis I diagnoses other than panic disorder. 6. Had no treatment with 45 mg/d or more of phenelzine sulfate (or an equivalent MAOI) or more than 150 mg/d of imipramine hydrochlo¬ ride (or an equivalent TCA) for 2 or more weeks during the current
episode.
Those subjects with a lifetime history of at least one spontaneous panic attack (by Research Diagnostic Criteria) were judged to be panic positive. The study consisted of two double-blind 6-week phases preceded by a 10-day single-blind placebo period. Patients who responded (see definition below) to the single-blind placebo were removed from the study. At the end of the first 6-week phase, nonresponders were removed from the study. Responders continued in the study for the second double-blind 6-week phase. The rationale for the second 6week phase stemmed from our concern that in a population character¬ ized by a reactive mood, some improvement might be transient. It was anticipated that the second 6-week phase would identify persis¬ tent, clinically relevant improvement. The double-blind procedure and dose of medicine were identical to that previously reported. The imipramine hydrochloride dose was raised on a fixed schedule to 200 mg/d, and phenelzine sulfate was raised on a fixed schedule to 60 mg/d; in nonresponders, the doses were raised to 300 mg/d and 90 mg/d, respectively, in the fifth and sixth weeks. The Clinical Global Impression (CGI) Scale, Hamilton Depression Rating Scale (HAM-D) (21 items), and the Hopkins Symptom Check¬ list (SCL-90) were used. A scale with definitions and specific anchors especially designed to assess reactive mood and atypical symptoms was used. Analyses of reliability are in preparation. In the original study, criteria for being included as a completer included a minimum of 28 consecutive days (out of 42 days) and a dose of 60 mg of phenelzine sulfate or 200 mg of imipramine hydrochloride (four pills of either per day) or the equivalent number of placebo pills.9 In the original study, the conditions of nine patients who continued in the study for 6 weeks, three receiving imipramine and six receiving phenelzine, improved but could not tolerate four pills per day. They were all receiving three pills per day (45 mg of phenelzine sulfate or 150 mg of imipramine hydrochloride). These nine patients were rated double blindly. There were no patients receiving placebo who could not tolerate four pills per day. In the replication study, the dose schedule was modified so that patients who tolerated only three pills per day (45 mg of phenelzine sulfate or 150 mg of imipramine hydro¬ chloride) were considered completers. Therefore, we will adjust the 6-week improvement rate of the original study to match the replica¬ tion study outcome by counting the three patients receiving imipra¬ mine and the six receiving phenelzine who improved with three pills per day as responders. To assess the role of drug side effects in contrast to other reasons for noncompliance, dropouts were divided into two types: (1) with¬ drawal attributable to medication side effects and (2) those patients who did not complete the study for reasons other than an adverse drug reaction, eg, noncompliance, substance abuse (ie, administra¬ tive removals). These judgments were made by a rater who was blind to the type of treatment. Patients rated 1 (very much improved) or 2 (much improved) on the 7-point improvement scale of the CGI were considered to be respond¬
These ratings were done conservatively; only patients with a marked diminution of symptoms were rated as much improved. Analyses that involved categorical variables were done by using the uncorrected 2 statistic or, when appropriate, Fisher's Exact Test. Continuous variables were analyzed by using the t test, F statistic, and analysis of covariance (to correct for baseline differ¬ ences). Tests without a priori hypotheses were two tailed; all others were one tailed. Specifically, we hypothesized that both drugs would be better than placebo and that phenelzine would be superior to ers.
imipramine. The evaluations at week 12 were done by using an end-point analysis, with the last observation carried forward. Patients who
Downloaded From: http://archpsyc.jamanetwork.com/ by a Cornell University User on 09/01/2016
Table 2.—Contrast of Baseline Characteristics of Original and
Replication Sample: Multiple Analysis Covariance* Main Effects
Adjusted Means
Study
Imipramine
Placebo
Hydrochloride
Phenelzine Sulfate
Overall
3.7
3.8
3.9
3.8
3.7
3.8
4.0
3.8
3.8
3.8
3.8
3.8
3.9
3.8
3.8
3.9
13.9
15.2
14.6
14.5
13.6
14.3
15.7
14.6
Original Replication Original Replication Original
21.9
21.4
20.7
21.4
20.9
20.2
19.5
20.2
3.3
3.2
3.2
3.2
3.1
3.0
3.0
3.1
2.5
2.5
2.2
2.4
Replication Original Replication
2.2
2.3 1.8
2.2
2.2
1.8
1.8
1.8
Original Replication
2.8
2.8
2.7 2.7
Original Replication Original
2.9 2.7
Replication Original Replication
Original Replication Original Replication Original Replication
Original Replication
Sample Type
Scale CGI
Severity Improvement HAM-D
Original Replication Original Replication Original Replication
Drug
Interaction
0.1
NS
2.2
NS
0.4
NS
0.6
NS
0.1
NS
0.3
NS
0.0
NS
2.2
NS
1.1
NS
2.0
NS
0.9
NS
0.0
NS
1.5
NS
0.28
NS
0.0
NS
2.6
NS
1.0
NS
0.8
NS
1.9
1.6
NS
0.1
1.6
NS NS
1.1
1.9
1.1
NS
0.1
NS NS
2.7
2.7
0.7
NS
1.8
NS
0.6
NS
2.4
2.6
2.7
2.8
2.8
NS
0.3
NS
0.2
NS
2.7
2.7
2.7
2.0
2.1
2.1
2.0
5.5
.02
0.4
NS
0.1
NS
1.8 2.1
1.8
1.8
2.2 2.1
2.0 2.1
2.1
0.1
NS
0.4
NS
0.3
NS
2.5 2.4
2.3
2.1
2.3
0.7
NS
3.1
.05
0.1
NS
2.1
2.0
2.2
2.0
1.9
2.0
2.0
2.8
NS
0.6
NS
0.3
NS
1.9
1.7
38.2
36.3
37.2
3.3
NS
1.3
NS
3.9
.02
30.1
34.2
20
17
1.8 36.7 38.9 23
7.5
.01
2.6
NS
2.3
NS
14
17
1Í
SCL-90
Summary score
Depression Anxiety Phobia
Interpersonal sensitivity
2.0 2.0
Obsessive-
compulsive Somatization
Hostility Paranoia
Psychoticism Age
Age at onset
2.2
*CGI Indicates Clinical Global Impression; NS, not
Table 3.—Proportion of
significant; HAM-D,
Hamilton
Responders: Definite Atypical Sample, % (No.)
Original Placebo
Week 6 28
Imipramine hydrochloride
54
Phenelzine sulfate
75
Placebo
(13/47) (22/41) (30/40)
Week 12 19 (9/47)
Imipramine
44
Phenelzine
70
(18/41) (28/40)
Replication 19 50 83 15
47 80
(5/26) (17/34) (25/30) (4/26) (16/34) (24/30)
completed the 12 weeks and maintained their improvement, as well as patients who were judged to be responders and who withdrew from the study at 6 weeks, were considered to be 12-week responders.
2.1
34.7 20
17
Depression Rating Scale; and SCL-90, Hopkins Symptom Checklist-90.
Patients who were nonresponders at 6 weeks and those patients who failed to maintain their improvement from weeks 6 to 12 were consid¬ ered to be nonresponders. From weeks 6 to 12, six patients who were receiving phenelzine, three who were receiving imipramine, and one who was receiving placebo dropped out. If the proportion of respond¬ ers was calculated by removing these patients from the analysis (instead of a last observation carried forward), contrasts across groups were not changed.
RESULTS In the replication study, patients received a single-blind placebo and 10% (11/115) were judged to be responders and were removed, and 3 dropped out. One hundred four patients were randomized to double-blind medication. Eighty-seven percent (90/104) completed the trial. Thirty-four were assigned to a placebo, 37 to imipramine, and 33 to phenelzine. Seventy-six percent (26/34) who were receiving placebo, 92% (34/37) who were receiving imipramine, and 91% (30/33) who were receiving phenelzine completed the study. With placebo, there were two side effect and six administrative removals; with imipramine, there were no side effects and three administrative removals; and with phenelzine, there was one side effect and two
Downloaded From: http://archpsyc.jamanetwork.com/ by a Cornell University User on 09/01/2016
Table
4.—Replication Sample Outcome by Treatment Group* Comparisons^
Adjusted Meansf Scale
Placebo
Imipramine Hydrochloride
Phenelzine Sulfate
Overall
(n 30)
F=13.3,
Imipramine Placebo
vs
Phenelzine Placebo
vs
Phenelzine
T 3.9, .000
T 2.6, P=.011
T 5.5, .000
T 2.9, .004
vs
Imipramine
CGI
Severity
3.4
(n 26) =
2.8
(n 34) =
2.0
=
=
Change
3.3
(n 26) =
2.6
(n 34) =
1.8
F=14.5,
(n 30) =
=
HAM-D
12.0
(n 25) =
10.0
(n 33) =
5.8
.000 .000
F =10.0, .000
(n 30) =
=
SCL-90 Somatization
1.8
Obsessive-
2.6
compulsive Interpersonal sensitivity Depression
2.5
(
=
(
=
(
=
23)
1.6
23)
2.3
23)
2.1
(
( (
=
=
=
29)
1.6
29)
2.3
29)
1.9
(
=
(
=
(
=
27)
F
27)
F
27)
F
=
(
2.1
(
=
23)
2.4
(
=
29)
2.0
(
=
23)
2.0
(
=
28)
1.6
(
=
Anxiety
=
1.6
(
Paranoia
2.1
(
=
23)
1.6
(
23)
1.9
(
=
28)
1.4
(
28)
1.7
(
=
.13
T 2.8, .007 =
=
T=1.4, P
=
=
=
=
=
=
=
=
=
.16
3.5,
=
2.3,
=03
=001
1.3,
NS =
5.4, 9.2,
27)
F
27)
F 6.8, .002
=
=1.9,
.006
=
27)
F 2.53, .086
27)
F
=
=
=
.07
=2.0,
.000
=
=
Phobia
=
1.0,
=
=
P
NS
=
2.8
T=1.5,
=
=
.05
0.03,
NS
=.04
.000
3.8, =
=
-2.1,
3.7,
=
.000
2.0, =
=
.05
2.1, =04 2.5, =
.01
=
=
=
=
=
=
4.49, =
Psychoticism
1.7
(
=
23)
1.6
(
=
28)
1.5
(
=
27)
F-3.1,
19.2
(
=
23)
17.4
(
=
28)
15.3
(
=
27)
F 7.7, =.001
=
Summary score
.014 .053
=
=1.6, =
.13
=
3.7,
1.6,
=2.6,
=1.4,
=3.4,
=
=.12 =.17
=
2.3,
=.02
=000
—1.1, NS
=.01
=
2.2,
=.03
=.001
*By analysis of covariance (regression model). CGI indicates Clinical Global Impression; HAM-D, Hamilton Depression Rating Scale; and SCL-90, Symptom Checklist-90. For Hostility subscale of SCL-90, analysis of covariance was not applicable, with heterogeneity of slopes. fMeans are adjusted for scores at randomization (analysis of covariance). tOne-way analysis of variance.
administrative removals. There were no statistically significant dif¬ ferences in the proportion failing to complete the trial with each treatment. Hereafter, all data refer to the 90 study completers. The replication sample (n 90) was compared with the original sample of DADs (n 120) on a variety of baseline variables (Table 2). In addition, the proportion of each sample meeting Research Diag¬ nostic Criteria subtypes was contrasted. There were no significant differences, with roughly 70% in each sample meeting a major depres¬ sive disorder, 50% meeting an intermittent disorder, and 13% meet¬ ing a bipolar II disorder. Chronicity was determined by considering the proportion of time as an adult that the patient was depressed. Patients who reported that they had depressive symptoms for the majority of their adult life were considered to be mostly depressed. Those who reported that they never had had a 3-month period of well-being when they were primarily free of depressive symptoms were considered to be always depressed. Seventy-nine percent of the original sample (59% mostly, 20% always depressed as an adult) and 90% of the replication sample had chronic depression (35% mostly, 55% always). The replication sample had more patients who were always depressed (55% vs 20%). The number of patients who were considered to have chronic depres¬ sion by either definition was similar in the two samples (79% vs 90%). The difference on this item was of little clinical significance. Baseline data appropriate for parametric tests were contrasted for the two samples, by treatment groups using analysis of variance (Table 2). The total HAM-D and CGI scores, the SCL-90 factors, the age at study entry, and the age at onset were compared. Four significant differences were found: patients receiving a placebo had higher scores on the SCL-90 paranoia factor, the original sample had a higher somatization factor score, the placebo group in the original sample was older than the replication placebo group, and the age at onset of the replication sample was younger (17 vs 20 years). There were 40 contrasts in this analysis. Four differences approximated chance expectation, and no meaningful pattern was apparent. =
=
Hopkins
Pattern of Associated Features
The most frequently occurring symptom was rejection sensitivity (present in 81% of the patients; 83% of original sample; and 79% of the replication sample); less frequent were leaden paralysis (64% of the patients; 65% of the original sample; and 69% of the replication sample), overeating (64% of the patients; 66% of the original sample; and 57% of the replication sample), and oversleeping (49% of each sample). Of 11 patterns observed, there was only a trend for the proportion ofthe two samples to differ on one pattern, ie, overeating, rejection sensitivity, and leaden paralysis (20% vs 10%, 2 3.3, 1 df .07). The most frequent patterns involved a combination of rejec¬ tion sensitivity and overeating and leaden paralysis. The combination of these three symptoms, ie, (1) rejection sensitive, eating, and leaden paralysis, (2) rejection sensitive and eating, and (3) rejection sensitive and leaden paralysis accounted for 47% of the patients; the other 8 patterns accounted for 53%. While the original sample had more patients (20% vs 10%) with all three, the replication sample had a greater proportion with only two of the symptoms combination, ie, rejection eating and rejection leaden. Thus, roughly equal propor¬ tions of each sample had some combination of these symptoms, ie, 47% (57/119) of the original sample and 46% (41/90) of the replication sample. Thus, rejection sensitivity and some combination of overeat¬ ing and leaden paralysis were prominent characteristics of these patients. Outcome by Treatment Group: Replication Sample of DADs Nineteen percent (5/26) of the placebo-, 50% (17/34) of the imipra¬ mine-, and 83% (25/30) of the phenelzine-treated patients were judged to be responders at 6 weeks ( 2 23.0, P
Replication
Frederic M. Quitkin, MD; Patrick J. McGrath, MD; Jonathan W. Stewart, MD; Wilma Harrison, MD; Elaine Tricamo, RN; Steven G. Wager, MD; Katja Ocepek-Welikson, MPhil; Edward Nunes, MD; Judith G. Rabkin, PhD; Donald F. Klein, MD
\s=b\ In an initial study with 120 patients with reactive mood and associated atypical symptoms, phenelzine sulfate was superior to imipramine hydrochloride and placebo. Since their response to phenelzine appears to be unique, this suggests that atypical depression may be a distinct subgroup of unipolar depressive illness. Unexpectedly, the benefit of antidepressants was limited to patients who also had spontaneous panic attacks. To help establish the validity of this syndrome, a new sample of 90 atypical depressives was studied. The clinical and demographic characteristics of the original and replication sample were virtually identical at baseline. In addition, the treatment response with either placebo, imipramine, or phenelzine was also indistinguishable in the two patient groups. The outcome in the replication study supports the hypothesis that this may be a distinct unipolar depressive subgroup. In the replication sample, a history of panic attacks did not appear to be a relevant predictor. We discuss the explanations for this discrepancy in the two patient
samples. (Arch Gen Psychiatry. 1990;47:935-941) of an the group from monoamine oxidase in¬ who hibitors more than 20 years ago, was first validation has proved to be elusive." Patients responsive to MAOIs, described by early investigators, appear to have
concept atypical depressive (AD) Although selectively benefited discussed (MAOIs) .
heterogeneous symptom profiles, including phobic anxiety (panic disorders) and vegetative atypical depression characterized by hypersomnia, lethargy, and hyperphagia.6 The inclusion of patients with panic disorder, a disorder clear¬ ly responsive to tricyclic antidepressants (TCAs), might have obscured the identification of a superior MAO response in a vegetative atypical group.6"8 We studied depressed patients with a nonautonomous mood who met DSM-III for a major or dysthymic mood to deter¬ mine if distinct syndromes could be identified in this heteroge¬ neous population.9'11 Approximately half of the patients had
included
Accepted for publication December 7,1989. From the Department of Psychiatry, Columbia University, New York, NY, and the New York (NY) State Psychiatric Institute. Reprint requests to New York State Psychiatric Institute, 722 W 168th St, Box 12, New York, NY 10032 (Dr Quitkin).
some spontaneous panic attacks. It was not clear whether those with spontaneous panic attacks, as well as AD symp¬ toms, would have the same treatment response as those with only atypical depression. Our concern was that patients with both atypical depression and panic attacks might benefit from imipramine hydrochloride, which would obscure selective responsivity to MAOIs in others. Patients were stratified by the number of atypical symp¬ toms, presence of panic attacks, and evidence of hysteroid dysphoria (Table 1). The associated atypical symptoms used to classify patients included hyperphagia, hypersomnolence, a leaden feeling, and rejection sensitivity (defined below). Pa¬ tients with at least two associated symptoms were considered definite atypical depressives (DADs), and those with one associated symptom were considered probable atypical depressives (PADs). Patients with only a reactive mood were called simple mood reactive depressives. Hysteroid dysphoria is a proposed depressive subtype characterized by a history of depression associated with a disappointment in a romantic context, histrionic personality, and excessive need for atten¬ tion.12,13 Depressive exacerbations of hysteroids are character¬ ized by vegetative atypical symptoms. Two main points emerge from our first study of 120 DADs9: (1) phenelzine sulfate was superior to imipramine and placebo, and (2) unexpectedly, the superiority of both imipramine and phenelzine to placebo was limited to patients with spontane¬ ous panic attacks and/or hysteroid dysphoria. Unreplicated clinical trials, especially those with unantici¬ pated results, should be greeted with skepticism. The promi¬ nent role of panic attacks in identifying drug-responsive disor¬ ders was not predicted and required cross-validation. Therefore, a replication with an independent sample of DADs
undertaken. The following questions were asked: (1) Will the clinical picture of patients referred to as ADs be the same in two independent samples? (2) Will ADs treated with phenelzine have a superior response compared with those treated with imipramine or placebo? These data were collected during approximately an 8-year period. If the two samples had simi¬ lar baseline characteristics and treatment response, it sug¬ gests this syndrome had a predictable phenomenology and treatment response.14 (3) Will a history of panic attack continwas
Downloaded From: http://archpsyc.jamanetwork.com/ by a Cornell University User on 09/01/2016
Table 1 .—Stratification of Patients With Nonautonomous Mood* Stratification Definite atypical depressives Mood reactive Has at least 2 of 4 associated
symptoms (hyperphagia, hypersomnolence, rejection sensitivity, and leaden paralysis* Subdivided into those with and without any spontaneous panic attacks
Probable atypical depressives Mood reactive Has at least 1 of associated symptoms Subdivided into groups with and without any spontaneous panic attacks Mood reactive depressives Mood reactive No associated symptoms Subdivided into groups with and without spontaneous panic attacks
Hysteroid dysphorics—a proposed depressive subtype characterized byt
Pathologic rejection sensitivity in a romantic context
When
rejected, develops depression characterized by reactive atypical symptoms *See text for definition of associated symptoms. mood and associated
fAII hysteroid dysphorics met criteria for definite atypical depression.
to be associated with a greater drug-placebo difference? (4) Are anxious depressives (with or without panic attacks) more likely to benefit from MAOIs than depressed patients without anxiety? (5) Does the drug response add to the relevance of the can hysteroid dysphoric subtype? (6) With a larger sample, more narrow confidence limits of the responsivity of ADs to each of the treatments be developed? ue
NATURE OF THIS REPORT
We report on a clinical trial with an independent sample of 90 ADs and will contrast outcome in this sample with the original sample of 120 ADs.9 The only inconsistency found between the two samples of DADs was the relevance of panic as a predictor of drug effect. In the original study, a history of panic appeared to be a relevant predictor, but in the replication study, it was not. We recently published an article that suggested that PADs (char¬ acterized by one atypical symptom) are indistinguishable from DADs.10 We therefore considered reactive mood and presence of any associated atypical feature sufficient criteria for atypical depression. The data from PADs will be used to help clarify the discrepant effect of panic in the two samples of DADs. To summarize, the following data are presented and ana¬ lyzed: 90 DADs not previously described, 120 DADs with two or more symptoms previously reported,9 and data for 60 PADs with one associated symptom previously reported.10 METHOD The study procedures, including the dose of drugs, rating scales, biochemical measures, and reliability assessment were virtually identical to those described in the study of patients with definite atypical depression.9 Therefore, only a brief review of the method is given. The minor change in the dose schedule is described below. Patients included in the study all attended an outpatient, universityaffiliated clinic and gave written informed consent. The following criteria were required. 1. Met Research Diagnostic Criteria for major, minor, or intermit¬ tent depression.15 2. Maintained mood reactivity while they were depressed, defined as present if mood could be lifted at least 50% of when they were not depressed (scale and anchors available from authors). 3. Exhibited at least two of the following: increased appetite or weightgain while they were depressed (defined as present if appetite
markedly increased or there was at least a 4.5-kg weight gain), oversleeping (sleeps at least 10 h/d), severe fatigue that created a sensation of leaden paralysis or extreme heaviness of arms or legs
(defined as present if a marked decrease in energy or heaviness of limbs was a prominent symptom), and sensitivity to rejection as a trait throughout adulthood (defined as present if rejection resulted in depression with functional impairment, eg, the subject missed work or
school).
4. Were between 18 and 65 years of age. 5. Had no other Axis I diagnoses other than panic disorder. 6. Had no treatment with 45 mg/d or more of phenelzine sulfate (or an equivalent MAOI) or more than 150 mg/d of imipramine hydrochlo¬ ride (or an equivalent TCA) for 2 or more weeks during the current
episode.
Those subjects with a lifetime history of at least one spontaneous panic attack (by Research Diagnostic Criteria) were judged to be panic positive. The study consisted of two double-blind 6-week phases preceded by a 10-day single-blind placebo period. Patients who responded (see definition below) to the single-blind placebo were removed from the study. At the end of the first 6-week phase, nonresponders were removed from the study. Responders continued in the study for the second double-blind 6-week phase. The rationale for the second 6week phase stemmed from our concern that in a population character¬ ized by a reactive mood, some improvement might be transient. It was anticipated that the second 6-week phase would identify persis¬ tent, clinically relevant improvement. The double-blind procedure and dose of medicine were identical to that previously reported. The imipramine hydrochloride dose was raised on a fixed schedule to 200 mg/d, and phenelzine sulfate was raised on a fixed schedule to 60 mg/d; in nonresponders, the doses were raised to 300 mg/d and 90 mg/d, respectively, in the fifth and sixth weeks. The Clinical Global Impression (CGI) Scale, Hamilton Depression Rating Scale (HAM-D) (21 items), and the Hopkins Symptom Check¬ list (SCL-90) were used. A scale with definitions and specific anchors especially designed to assess reactive mood and atypical symptoms was used. Analyses of reliability are in preparation. In the original study, criteria for being included as a completer included a minimum of 28 consecutive days (out of 42 days) and a dose of 60 mg of phenelzine sulfate or 200 mg of imipramine hydrochloride (four pills of either per day) or the equivalent number of placebo pills.9 In the original study, the conditions of nine patients who continued in the study for 6 weeks, three receiving imipramine and six receiving phenelzine, improved but could not tolerate four pills per day. They were all receiving three pills per day (45 mg of phenelzine sulfate or 150 mg of imipramine hydrochloride). These nine patients were rated double blindly. There were no patients receiving placebo who could not tolerate four pills per day. In the replication study, the dose schedule was modified so that patients who tolerated only three pills per day (45 mg of phenelzine sulfate or 150 mg of imipramine hydro¬ chloride) were considered completers. Therefore, we will adjust the 6-week improvement rate of the original study to match the replica¬ tion study outcome by counting the three patients receiving imipra¬ mine and the six receiving phenelzine who improved with three pills per day as responders. To assess the role of drug side effects in contrast to other reasons for noncompliance, dropouts were divided into two types: (1) with¬ drawal attributable to medication side effects and (2) those patients who did not complete the study for reasons other than an adverse drug reaction, eg, noncompliance, substance abuse (ie, administra¬ tive removals). These judgments were made by a rater who was blind to the type of treatment. Patients rated 1 (very much improved) or 2 (much improved) on the 7-point improvement scale of the CGI were considered to be respond¬
These ratings were done conservatively; only patients with a marked diminution of symptoms were rated as much improved. Analyses that involved categorical variables were done by using the uncorrected 2 statistic or, when appropriate, Fisher's Exact Test. Continuous variables were analyzed by using the t test, F statistic, and analysis of covariance (to correct for baseline differ¬ ences). Tests without a priori hypotheses were two tailed; all others were one tailed. Specifically, we hypothesized that both drugs would be better than placebo and that phenelzine would be superior to ers.
imipramine. The evaluations at week 12 were done by using an end-point analysis, with the last observation carried forward. Patients who
Downloaded From: http://archpsyc.jamanetwork.com/ by a Cornell University User on 09/01/2016
Table 2.—Contrast of Baseline Characteristics of Original and
Replication Sample: Multiple Analysis Covariance* Main Effects
Adjusted Means
Study
Imipramine
Placebo
Hydrochloride
Phenelzine Sulfate
Overall
3.7
3.8
3.9
3.8
3.7
3.8
4.0
3.8
3.8
3.8
3.8
3.8
3.9
3.8
3.8
3.9
13.9
15.2
14.6
14.5
13.6
14.3
15.7
14.6
Original Replication Original Replication Original
21.9
21.4
20.7
21.4
20.9
20.2
19.5
20.2
3.3
3.2
3.2
3.2
3.1
3.0
3.0
3.1
2.5
2.5
2.2
2.4
Replication Original Replication
2.2
2.3 1.8
2.2
2.2
1.8
1.8
1.8
Original Replication
2.8
2.8
2.7 2.7
Original Replication Original
2.9 2.7
Replication Original Replication
Original Replication Original Replication Original Replication
Original Replication
Sample Type
Scale CGI
Severity Improvement HAM-D
Original Replication Original Replication Original Replication
Drug
Interaction
0.1
NS
2.2
NS
0.4
NS
0.6
NS
0.1
NS
0.3
NS
0.0
NS
2.2
NS
1.1
NS
2.0
NS
0.9
NS
0.0
NS
1.5
NS
0.28
NS
0.0
NS
2.6
NS
1.0
NS
0.8
NS
1.9
1.6
NS
0.1
1.6
NS NS
1.1
1.9
1.1
NS
0.1
NS NS
2.7
2.7
0.7
NS
1.8
NS
0.6
NS
2.4
2.6
2.7
2.8
2.8
NS
0.3
NS
0.2
NS
2.7
2.7
2.7
2.0
2.1
2.1
2.0
5.5
.02
0.4
NS
0.1
NS
1.8 2.1
1.8
1.8
2.2 2.1
2.0 2.1
2.1
0.1
NS
0.4
NS
0.3
NS
2.5 2.4
2.3
2.1
2.3
0.7
NS
3.1
.05
0.1
NS
2.1
2.0
2.2
2.0
1.9
2.0
2.0
2.8
NS
0.6
NS
0.3
NS
1.9
1.7
38.2
36.3
37.2
3.3
NS
1.3
NS
3.9
.02
30.1
34.2
20
17
1.8 36.7 38.9 23
7.5
.01
2.6
NS
2.3
NS
14
17
1Í
SCL-90
Summary score
Depression Anxiety Phobia
Interpersonal sensitivity
2.0 2.0
Obsessive-
compulsive Somatization
Hostility Paranoia
Psychoticism Age
Age at onset
2.2
*CGI Indicates Clinical Global Impression; NS, not
Table 3.—Proportion of
significant; HAM-D,
Hamilton
Responders: Definite Atypical Sample, % (No.)
Original Placebo
Week 6 28
Imipramine hydrochloride
54
Phenelzine sulfate
75
Placebo
(13/47) (22/41) (30/40)
Week 12 19 (9/47)
Imipramine
44
Phenelzine
70
(18/41) (28/40)
Replication 19 50 83 15
47 80
(5/26) (17/34) (25/30) (4/26) (16/34) (24/30)
completed the 12 weeks and maintained their improvement, as well as patients who were judged to be responders and who withdrew from the study at 6 weeks, were considered to be 12-week responders.
2.1
34.7 20
17
Depression Rating Scale; and SCL-90, Hopkins Symptom Checklist-90.
Patients who were nonresponders at 6 weeks and those patients who failed to maintain their improvement from weeks 6 to 12 were consid¬ ered to be nonresponders. From weeks 6 to 12, six patients who were receiving phenelzine, three who were receiving imipramine, and one who was receiving placebo dropped out. If the proportion of respond¬ ers was calculated by removing these patients from the analysis (instead of a last observation carried forward), contrasts across groups were not changed.
RESULTS In the replication study, patients received a single-blind placebo and 10% (11/115) were judged to be responders and were removed, and 3 dropped out. One hundred four patients were randomized to double-blind medication. Eighty-seven percent (90/104) completed the trial. Thirty-four were assigned to a placebo, 37 to imipramine, and 33 to phenelzine. Seventy-six percent (26/34) who were receiving placebo, 92% (34/37) who were receiving imipramine, and 91% (30/33) who were receiving phenelzine completed the study. With placebo, there were two side effect and six administrative removals; with imipramine, there were no side effects and three administrative removals; and with phenelzine, there was one side effect and two
Downloaded From: http://archpsyc.jamanetwork.com/ by a Cornell University User on 09/01/2016
Table
4.—Replication Sample Outcome by Treatment Group* Comparisons^
Adjusted Meansf Scale
Placebo
Imipramine Hydrochloride
Phenelzine Sulfate
Overall
(n 30)
F=13.3,
Imipramine Placebo
vs
Phenelzine Placebo
vs
Phenelzine
T 3.9, .000
T 2.6, P=.011
T 5.5, .000
T 2.9, .004
vs
Imipramine
CGI
Severity
3.4
(n 26) =
2.8
(n 34) =
2.0
=
=
Change
3.3
(n 26) =
2.6
(n 34) =
1.8
F=14.5,
(n 30) =
=
HAM-D
12.0
(n 25) =
10.0
(n 33) =
5.8
.000 .000
F =10.0, .000
(n 30) =
=
SCL-90 Somatization
1.8
Obsessive-
2.6
compulsive Interpersonal sensitivity Depression
2.5
(
=
(
=
(
=
23)
1.6
23)
2.3
23)
2.1
(
( (
=
=
=
29)
1.6
29)
2.3
29)
1.9
(
=
(
=
(
=
27)
F
27)
F
27)
F
=
(
2.1
(
=
23)
2.4
(
=
29)
2.0
(
=
23)
2.0
(
=
28)
1.6
(
=
Anxiety
=
1.6
(
Paranoia
2.1
(
=
23)
1.6
(
23)
1.9
(
=
28)
1.4
(
28)
1.7
(
=
.13
T 2.8, .007 =
=
T=1.4, P
=
=
=
=
=
=
=
=
=
.16
3.5,
=
2.3,
=03
=001
1.3,
NS =
5.4, 9.2,
27)
F
27)
F 6.8, .002
=
=1.9,
.006
=
27)
F 2.53, .086
27)
F
=
=
=
.07
=2.0,
.000
=
=
Phobia
=
1.0,
=
=
P
NS
=
2.8
T=1.5,
=
=
.05
0.03,
NS
=.04
.000
3.8, =
=
-2.1,
3.7,
=
.000
2.0, =
=
.05
2.1, =04 2.5, =
.01
=
=
=
=
=
=
4.49, =
Psychoticism
1.7
(
=
23)
1.6
(
=
28)
1.5
(
=
27)
F-3.1,
19.2
(
=
23)
17.4
(
=
28)
15.3
(
=
27)
F 7.7, =.001
=
Summary score
.014 .053
=
=1.6, =
.13
=
3.7,
1.6,
=2.6,
=1.4,
=3.4,
=
=.12 =.17
=
2.3,
=.02
=000
—1.1, NS
=.01
=
2.2,
=.03
=.001
*By analysis of covariance (regression model). CGI indicates Clinical Global Impression; HAM-D, Hamilton Depression Rating Scale; and SCL-90, Symptom Checklist-90. For Hostility subscale of SCL-90, analysis of covariance was not applicable, with heterogeneity of slopes. fMeans are adjusted for scores at randomization (analysis of covariance). tOne-way analysis of variance.
administrative removals. There were no statistically significant dif¬ ferences in the proportion failing to complete the trial with each treatment. Hereafter, all data refer to the 90 study completers. The replication sample (n 90) was compared with the original sample of DADs (n 120) on a variety of baseline variables (Table 2). In addition, the proportion of each sample meeting Research Diag¬ nostic Criteria subtypes was contrasted. There were no significant differences, with roughly 70% in each sample meeting a major depres¬ sive disorder, 50% meeting an intermittent disorder, and 13% meet¬ ing a bipolar II disorder. Chronicity was determined by considering the proportion of time as an adult that the patient was depressed. Patients who reported that they had depressive symptoms for the majority of their adult life were considered to be mostly depressed. Those who reported that they never had had a 3-month period of well-being when they were primarily free of depressive symptoms were considered to be always depressed. Seventy-nine percent of the original sample (59% mostly, 20% always depressed as an adult) and 90% of the replication sample had chronic depression (35% mostly, 55% always). The replication sample had more patients who were always depressed (55% vs 20%). The number of patients who were considered to have chronic depres¬ sion by either definition was similar in the two samples (79% vs 90%). The difference on this item was of little clinical significance. Baseline data appropriate for parametric tests were contrasted for the two samples, by treatment groups using analysis of variance (Table 2). The total HAM-D and CGI scores, the SCL-90 factors, the age at study entry, and the age at onset were compared. Four significant differences were found: patients receiving a placebo had higher scores on the SCL-90 paranoia factor, the original sample had a higher somatization factor score, the placebo group in the original sample was older than the replication placebo group, and the age at onset of the replication sample was younger (17 vs 20 years). There were 40 contrasts in this analysis. Four differences approximated chance expectation, and no meaningful pattern was apparent. =
=
Hopkins
Pattern of Associated Features
The most frequently occurring symptom was rejection sensitivity (present in 81% of the patients; 83% of original sample; and 79% of the replication sample); less frequent were leaden paralysis (64% of the patients; 65% of the original sample; and 69% of the replication sample), overeating (64% of the patients; 66% of the original sample; and 57% of the replication sample), and oversleeping (49% of each sample). Of 11 patterns observed, there was only a trend for the proportion ofthe two samples to differ on one pattern, ie, overeating, rejection sensitivity, and leaden paralysis (20% vs 10%, 2 3.3, 1 df .07). The most frequent patterns involved a combination of rejec¬ tion sensitivity and overeating and leaden paralysis. The combination of these three symptoms, ie, (1) rejection sensitive, eating, and leaden paralysis, (2) rejection sensitive and eating, and (3) rejection sensitive and leaden paralysis accounted for 47% of the patients; the other 8 patterns accounted for 53%. While the original sample had more patients (20% vs 10%) with all three, the replication sample had a greater proportion with only two of the symptoms combination, ie, rejection eating and rejection leaden. Thus, roughly equal propor¬ tions of each sample had some combination of these symptoms, ie, 47% (57/119) of the original sample and 46% (41/90) of the replication sample. Thus, rejection sensitivity and some combination of overeat¬ ing and leaden paralysis were prominent characteristics of these patients. Outcome by Treatment Group: Replication Sample of DADs Nineteen percent (5/26) of the placebo-, 50% (17/34) of the imipra¬ mine-, and 83% (25/30) of the phenelzine-treated patients were judged to be responders at 6 weeks ( 2 23.0, P
