Changes in Weight During Treatment for Depression DAVID J. KUPFER, MD, PATRICIA A. COBLE, RN, AND DEBRA RUBINSTEIN, MS Changes in appetite and weight were examined in a group of 47 carefully diagnosed primary depressives who were treated in a random design with either placebo (N = 17) oramitriptyline [N = 30) over a 35-day protocol. While the amitriptyline treated group as a whole showed a greater gain in weight than did the placebo group (4.5 vs. 0.5 lb, p < 0.05), no differential effects could be demonstrated between drug responders and nonresponders. Likewise, while a consistent relationship between the self-report of decreased appetite and final weight change was noted in the placebo group, no simple relationship between final weight change and self-reported changes in appetite were apparent in the drug-treated patients. There was, however, a relationship between the report of decreased appetite and clinical severity of depression in the drug nonresponder subgroup despite significant weight gain during the protocol. Thus, weight change during this study period did not appear to show a simple relationship to corresponding clinical change. The clinical lore that has supported the notion that increased appetite and weight gain in patients being treated with tricyclic antidepressants are "good" signs cannot be confirmed by our findings.
INTRODUCTION
Over the last decade, increased attention has been devoted to the development of improved diagnostic criteria for the affective syndromes. The nosological advances described in the Washington University (St. Louis) criteria have been elaborated more fully in the Research Diagnostic Criteria and have pointed out that among the required symptoms for a major depressive syndrome are psychobiologic changes in sleep, motor activity, appetite, and weight (1, 2). With respect to these areas of psychobiology, only sleep has received sufficient invesFrom the Department of Psychiatry, Western Psychiatric Institute and Clinic, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261 Address requests for reprints to: David J. Kupfer, M.D., Western Psychiatric Institute and Clinic, 3811 O'Hara Street, Pittsburgh, PA 15261. Received for publication April 9, 1979; revision received August 7, 1979.
tigative attention; for example, objective measures of EEG sleep have now been correlated with clinical course and as predictors of outcome (3). While other psychobiologic areas have not been actively pursued to date because of technological problems, such relative lack of interest for weight and appetite changes in affective states cannot be ascribed to technological drawbacks. Several investigators have examined psychobiologic variables as a group, assessing the magnitude or duration of change in appetite, weight, and libido seen in depression. Pollitt and Young (in 1971) found that depressed patients who exhibited what they termed "typical" symptoms (i.e., decrease in appetite, weight, and libido) were older than those patients who had more variable symptoms (4). In a 1972 study by Detre et al., weight gain was typical during the depressive phase of bipolar illness while weight loss was typical of unipolar illness (5). Regard-
Psychosomatic Medicine Vol. 4 1 , No. 7 (November 1979) Copyright c 1979 by the American Psychosomatic Society, Inc. Published by Elsevier North Holland, Inc.
535 0033-31 74/79/07053S10S0175
DAVID J. KUPFER et al.
less of diagnostic subtype, all of the depressed patients in this study reported decreased libido. A study by Polivy and Herman on depressed outpatients hypothesized that those patients who gained weight (6.3 lb) in conjunction with their depression would be individuals they identified as "restrained eaters," categorized as such by being weight conscious and dieters, while "unrestrained eaters" lost weight (5.3 lb) (6). In examining the ascending (recovery) phase of depression, Kraines (7) found that 70% of depressed patients being treated with ECT or a psychotropic medication gained weight. The greatest amount of weight gain occurred in women between ages 30 and 50 and seemed to precede the lifting of the depressed mood. In a single case study by Hawkins et al. (8), a 51-year-old male treated with a course of ECT was found to have a continuous weight loss during the first two weeks of observation that was sharply reversed coincident with an abrupt improvement in appetite one week after the beginning of ECT. Changes in the patient's behavior and mood occurred subsequently. In a recent study on over 200 untreated depressed patients by Paykel (9), 66% of the patients had a decreased appetite, 20% showed no change, and 14% reported an increased appetite. Not surprisingly, greater appetite change was associated with greater severity of illness. The depressives with increased appetite were distinguishable from those with anorexia by being female and more mildly depressed, with neurotic rather than psychotic illnesses but with a greater reduction in sexual interest. A review of these studies suggested to us a need to examine appetite and weight changes in a group of carefully diagnosed depressed patients 536
who were treated in a random design either with placebo or amitriptyline for a major depressive syndrome of the primary type. METHODS All forty-seven patients were hospitalized on the Clinical Research Unit (CRU) at Western Psychiatric Institute and Clinic (WPIC). At the time of admission, all patients had a traditional psychiatric interview and a physical examination. In addition, collateral information was obtained from their families and from case records of previous hospitalizations. During a two-week drug-free period, they underwent a series of routine laboratory tests, including thyroid function tests, an electroencephalogram, and any other tests that, based on their history or physical examination, were indicated. All patients thus underwent an "entrainment period" during this time with respect to their sleep-wake cycle, meal schedule, etc. Following the two-week drug-free period, the Schedule for Affective Disorders and Schizophrenia (SADS) was filled out by their psychiatrist. The SADS, a structured research interview, which collects data necessary to make diagnoses using the Research Diagnostic Criteria (RDC) (2), was completed using information obtained from the initial interview, the case record, collateral history from relatives, observation on the CRU, and, if necessary, a second interview with the patient. After diagnoses were made using the RDC information obtained in the SADS, all cases were reviewed to obviate any problems regarding reliability among interviewers. If the level of severity of depression remained sufficiently high at the end of the drug-free period (a minimum score of 30 on the 17-item Hamilton Rating Scale using the sum of two raters), patients then entered the actual protocol and were subsequently evaluated twice weekly for severity of depression using the HRS throughout the investigation. Patients also completed self-rating forms (KDS) three times a week. (10, 11). All patients received four identical "study capsules" daily during the 35-day protocol investigation, which was divided into time segments on the basis of increasing drug dosage. The first seven days represent a placebo period for all subjects. During the second period, 50 mg of amitriptyline was administered at 9 PM for three consecutive nights. During the next four days, patients received 100 mg daily (50 mg at 5 PM and 50 mg at 9 PM); then followed a 7-day period of 150 mg daily
Psychosomatic Medicine Vol. 41, No. 7 (November 1979)
WEIGHT CHANCES DURING DEPRESSION TREATMENT Amitriptyline and nortriptyline levels were mea(50 mg at 1 PM, 5 PM, and 9 PM); finally there was a 14-day period at a dose level of 200 mg/day (50 mg sured in plasma utilizing a modification of the gasQID). Throughout the study period, patients re- chromatography-mass-spectrometry (GC/MS) techceived a constant number of capsules that appeared nique described by Biggs and associates (12). Instead identical so that both the patients and the staff re- of using electron-impact ionization mass specmained "blind" to the medication. trometry, however, as in the studies of Biggs and Thirty patients received active amitriptyline co-investigators, the more specific and sensitive treatment and had a mean age of 42.1 ± 2.5 years. As technique of chemical ionization mass spectrometry shown in Table 1, 21 of these patients were female was used in this work. This technique utilizes a Finand 9 were male. Their mean Hamilton Rating score nigan 3200 GC/MS unit equipped with chemical was 40.3 ± 2.3. Nine of these patients were consid- ionization capabilities and a six-channel proered psychotic and 21 were nonpsychotic. The drug grammed multiple ion monitor (13). groups were further divided into a responder group The major statistical techniques used in this inves(n = 13) and a group of partial and nonresponders (n tigation were Pearson product-moment correlations, = 17) on the basis' of the HRS score at the end of the Student's t-tests, and paired t-tests. ANOVA was also protocol period (an HRS score of 12 was used as the used when appropriate. cutoff score for responders). The 17 patients who continued to receive placebo throughout the investigation had a mean age of 40.0 + 3.9 years. As shown in Table 1, 11 of these patients were female and 6 RESULTS were male. Their mean HRS score at the beginning of the protocol was 45.1 ± 4.1. Five of these patients Data analysis for this investigation adwere considered psychotic and 12 were nonpsychodressed essentially three issues: weight tic. Throughout the protocol, patients completed KDS changes during the protocol period, forms 1 and 2 (10, 11) three times a week. These weight and appetite change in relation to short but comprehensive self-administered ques- clinical outcome and, finally, the relationtionnaires have been used in both inpatient and outpatient populations to elicit a wide range of psycho- ship of changes in weight and appetite to logical and somatic symptoms. In this study, queries tricyclic blood levels. As shown in Table regarding the presence or absence of an increased or 2, there were no significant differences in decreased appetite were examined in order to assess weight at the beginning of the protocol changes in appetite across the 35-day protocol between the drug-treated and the placebo period. Only YES-NO responses are possible. A YES response was assigned a score of 1 and a NO groups. However, when weight change response, a score of 2. Since each time period in- was calculated as the difference between volved multiple ratings, the mean of the scores for weight on the last day of the protocol and each patient was calculated for each time period and weight on the eighth day of the protocol the overall perception of appetite for that period determined as the difference between decreased and (first drug administration), a significant increased appetite. Thus, negative totals reflected ah difference in the final weight as measured overall decreased appetite arid positive totals, an by change was demonstrated, with the overall increased appetite. TABLE 2. TABLE 1.
Selected Characteristics of Depressed Patients Drug Group
Age(X±SEM) Sex Hamilton Rating Scale(X±SEM) Psychotic depressives
Placebo (N=17) (Ib)
Placebo Group
42.1 ±2.5 21 F 9 M
40.0 ± 3.9 11 F 6 M
40.3 ± 2.3 N= 9
45.5 + 4.1 N=5
Weight Change in Protocol Patients Drug (N = 30) (Ib)
Prob.
Initial Weight 143.9 ± 8 . 4 150.2 ± 7.0 NS Day 8 Weight 144.2 ± 8.0 149.8 ± 7 . 0 NS Final Weight 144.8 ± 8 . 3 154.4 ± 6 . 9 NS Change in Weight 0.5 ± 1 . 6 4.6 ± 0.8 p
INTRODUCTION
Over the last decade, increased attention has been devoted to the development of improved diagnostic criteria for the affective syndromes. The nosological advances described in the Washington University (St. Louis) criteria have been elaborated more fully in the Research Diagnostic Criteria and have pointed out that among the required symptoms for a major depressive syndrome are psychobiologic changes in sleep, motor activity, appetite, and weight (1, 2). With respect to these areas of psychobiology, only sleep has received sufficient invesFrom the Department of Psychiatry, Western Psychiatric Institute and Clinic, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261 Address requests for reprints to: David J. Kupfer, M.D., Western Psychiatric Institute and Clinic, 3811 O'Hara Street, Pittsburgh, PA 15261. Received for publication April 9, 1979; revision received August 7, 1979.
tigative attention; for example, objective measures of EEG sleep have now been correlated with clinical course and as predictors of outcome (3). While other psychobiologic areas have not been actively pursued to date because of technological problems, such relative lack of interest for weight and appetite changes in affective states cannot be ascribed to technological drawbacks. Several investigators have examined psychobiologic variables as a group, assessing the magnitude or duration of change in appetite, weight, and libido seen in depression. Pollitt and Young (in 1971) found that depressed patients who exhibited what they termed "typical" symptoms (i.e., decrease in appetite, weight, and libido) were older than those patients who had more variable symptoms (4). In a 1972 study by Detre et al., weight gain was typical during the depressive phase of bipolar illness while weight loss was typical of unipolar illness (5). Regard-
Psychosomatic Medicine Vol. 4 1 , No. 7 (November 1979) Copyright c 1979 by the American Psychosomatic Society, Inc. Published by Elsevier North Holland, Inc.
535 0033-31 74/79/07053S10S0175
DAVID J. KUPFER et al.
less of diagnostic subtype, all of the depressed patients in this study reported decreased libido. A study by Polivy and Herman on depressed outpatients hypothesized that those patients who gained weight (6.3 lb) in conjunction with their depression would be individuals they identified as "restrained eaters," categorized as such by being weight conscious and dieters, while "unrestrained eaters" lost weight (5.3 lb) (6). In examining the ascending (recovery) phase of depression, Kraines (7) found that 70% of depressed patients being treated with ECT or a psychotropic medication gained weight. The greatest amount of weight gain occurred in women between ages 30 and 50 and seemed to precede the lifting of the depressed mood. In a single case study by Hawkins et al. (8), a 51-year-old male treated with a course of ECT was found to have a continuous weight loss during the first two weeks of observation that was sharply reversed coincident with an abrupt improvement in appetite one week after the beginning of ECT. Changes in the patient's behavior and mood occurred subsequently. In a recent study on over 200 untreated depressed patients by Paykel (9), 66% of the patients had a decreased appetite, 20% showed no change, and 14% reported an increased appetite. Not surprisingly, greater appetite change was associated with greater severity of illness. The depressives with increased appetite were distinguishable from those with anorexia by being female and more mildly depressed, with neurotic rather than psychotic illnesses but with a greater reduction in sexual interest. A review of these studies suggested to us a need to examine appetite and weight changes in a group of carefully diagnosed depressed patients 536
who were treated in a random design either with placebo or amitriptyline for a major depressive syndrome of the primary type. METHODS All forty-seven patients were hospitalized on the Clinical Research Unit (CRU) at Western Psychiatric Institute and Clinic (WPIC). At the time of admission, all patients had a traditional psychiatric interview and a physical examination. In addition, collateral information was obtained from their families and from case records of previous hospitalizations. During a two-week drug-free period, they underwent a series of routine laboratory tests, including thyroid function tests, an electroencephalogram, and any other tests that, based on their history or physical examination, were indicated. All patients thus underwent an "entrainment period" during this time with respect to their sleep-wake cycle, meal schedule, etc. Following the two-week drug-free period, the Schedule for Affective Disorders and Schizophrenia (SADS) was filled out by their psychiatrist. The SADS, a structured research interview, which collects data necessary to make diagnoses using the Research Diagnostic Criteria (RDC) (2), was completed using information obtained from the initial interview, the case record, collateral history from relatives, observation on the CRU, and, if necessary, a second interview with the patient. After diagnoses were made using the RDC information obtained in the SADS, all cases were reviewed to obviate any problems regarding reliability among interviewers. If the level of severity of depression remained sufficiently high at the end of the drug-free period (a minimum score of 30 on the 17-item Hamilton Rating Scale using the sum of two raters), patients then entered the actual protocol and were subsequently evaluated twice weekly for severity of depression using the HRS throughout the investigation. Patients also completed self-rating forms (KDS) three times a week. (10, 11). All patients received four identical "study capsules" daily during the 35-day protocol investigation, which was divided into time segments on the basis of increasing drug dosage. The first seven days represent a placebo period for all subjects. During the second period, 50 mg of amitriptyline was administered at 9 PM for three consecutive nights. During the next four days, patients received 100 mg daily (50 mg at 5 PM and 50 mg at 9 PM); then followed a 7-day period of 150 mg daily
Psychosomatic Medicine Vol. 41, No. 7 (November 1979)
WEIGHT CHANCES DURING DEPRESSION TREATMENT Amitriptyline and nortriptyline levels were mea(50 mg at 1 PM, 5 PM, and 9 PM); finally there was a 14-day period at a dose level of 200 mg/day (50 mg sured in plasma utilizing a modification of the gasQID). Throughout the study period, patients re- chromatography-mass-spectrometry (GC/MS) techceived a constant number of capsules that appeared nique described by Biggs and associates (12). Instead identical so that both the patients and the staff re- of using electron-impact ionization mass specmained "blind" to the medication. trometry, however, as in the studies of Biggs and Thirty patients received active amitriptyline co-investigators, the more specific and sensitive treatment and had a mean age of 42.1 ± 2.5 years. As technique of chemical ionization mass spectrometry shown in Table 1, 21 of these patients were female was used in this work. This technique utilizes a Finand 9 were male. Their mean Hamilton Rating score nigan 3200 GC/MS unit equipped with chemical was 40.3 ± 2.3. Nine of these patients were consid- ionization capabilities and a six-channel proered psychotic and 21 were nonpsychotic. The drug grammed multiple ion monitor (13). groups were further divided into a responder group The major statistical techniques used in this inves(n = 13) and a group of partial and nonresponders (n tigation were Pearson product-moment correlations, = 17) on the basis' of the HRS score at the end of the Student's t-tests, and paired t-tests. ANOVA was also protocol period (an HRS score of 12 was used as the used when appropriate. cutoff score for responders). The 17 patients who continued to receive placebo throughout the investigation had a mean age of 40.0 + 3.9 years. As shown in Table 1, 11 of these patients were female and 6 RESULTS were male. Their mean HRS score at the beginning of the protocol was 45.1 ± 4.1. Five of these patients Data analysis for this investigation adwere considered psychotic and 12 were nonpsychodressed essentially three issues: weight tic. Throughout the protocol, patients completed KDS changes during the protocol period, forms 1 and 2 (10, 11) three times a week. These weight and appetite change in relation to short but comprehensive self-administered ques- clinical outcome and, finally, the relationtionnaires have been used in both inpatient and outpatient populations to elicit a wide range of psycho- ship of changes in weight and appetite to logical and somatic symptoms. In this study, queries tricyclic blood levels. As shown in Table regarding the presence or absence of an increased or 2, there were no significant differences in decreased appetite were examined in order to assess weight at the beginning of the protocol changes in appetite across the 35-day protocol between the drug-treated and the placebo period. Only YES-NO responses are possible. A YES response was assigned a score of 1 and a NO groups. However, when weight change response, a score of 2. Since each time period in- was calculated as the difference between volved multiple ratings, the mean of the scores for weight on the last day of the protocol and each patient was calculated for each time period and weight on the eighth day of the protocol the overall perception of appetite for that period determined as the difference between decreased and (first drug administration), a significant increased appetite. Thus, negative totals reflected ah difference in the final weight as measured overall decreased appetite arid positive totals, an by change was demonstrated, with the overall increased appetite. TABLE 2. TABLE 1.
Selected Characteristics of Depressed Patients Drug Group
Age(X±SEM) Sex Hamilton Rating Scale(X±SEM) Psychotic depressives
Placebo (N=17) (Ib)
Placebo Group
42.1 ±2.5 21 F 9 M
40.0 ± 3.9 11 F 6 M
40.3 ± 2.3 N= 9
45.5 + 4.1 N=5
Weight Change in Protocol Patients Drug (N = 30) (Ib)
Prob.
Initial Weight 143.9 ± 8 . 4 150.2 ± 7.0 NS Day 8 Weight 144.2 ± 8.0 149.8 ± 7 . 0 NS Final Weight 144.8 ± 8 . 3 154.4 ± 6 . 9 NS Change in Weight 0.5 ± 1 . 6 4.6 ± 0.8 p
