Psychiafry Research. 37: 119-193 Elsevier
Growth Hormone and Other Hormonal Responses to Clonidine in Melancholic and Nonmelancholic Depressed Subjects and Controls Philip Mitchell, George Smythe, Gordon Brodaty, Philip Boyce, and Ian Hickie Received 1990.
June 4. 1990; revised version received September
10, 1990; accepted
Abstract. To study putative differences in central neurotransmitter function in depressive subtypes, growth hormone, adrenocorticotropic hormone (ACTH), cortisol, and prolactin responses to the cy,-noradrenergic receptor agonist clonidine (1.3 pg/ kg i.v.) were examined in 26 subjects with major depression, 13 of whom had melancholia. The responses of 10 of these endogenousi melancholic subjects were compared with those of 10 controls who were matched to the patients on age, sex, and menopausal status. In 15 of the depressed subjects, prolactin and cortisol responses to the putative serotonergic agonist fenfluramine were also examined to test for associations between these challenges. There were no significant differences in any of the responses between melancholic and nonmelancholic depressive subgroups after controlling for age and sex. With the exception of a greater reduction in ACTH in the endogenousimelancholic subjects, there were also no significant differences in hormonal responses between these patients and controls. There was, however, a significantly greater reduction in systolic blood pressure in the control subjects. There were no significant correlations between the responses to clonidine and fenfluramine. The findings suggest that clonidine at a dosage of 1.3 pg/ kg is neither able to differentiate reliably between depressive subtypes nor to differentiate reliably between depressed and control subjects. Key Words. Clonidine, sion.
Since at least the 1920’s (Gillespie, 1929), the subclassification of depression has been extensively debated (Farmer and McGuffin, 1989). As noted recently, “an assumption of psychiatric research is that there is at least one form of depression that constitutes a disease with an underlying pathophysiological although it remains controversial as to whether
dysfunction” (Hsaio et al., the endogenousimelancholic
Phihp Mitchell, M.D.. M.R.C.Psych., F.R.A.N.Z.C.P., is Senior Lecturer, School of Psychiatry, Universlty of New South Wales, Sydney. Austraha. George Smythe, A.S.T.C.. B.Sc., Ph.D., is Head, C.E. Heath Neuroscience Laboratory, Garvan Institute of Medical Research, Sydney, Australia. Gordon Parker, M.D., Ph.D.. F.R.A.N.Z.C.P., is Professor, School of Psychiatry, University of New South Wales. Svdnev. Australia. Kay Wilhelm is Senior Staff Specialist, Divislon of Psychiatry, Prince Henry Hospital, Sidney: Australia. Ian B. Hickle, M.D., F.R.A.N.Z.C.P.. 1s Staff Specialist. Divislon of Psvchiatrv. Prince Henrv Hospital, Sydney, Australia. Henry Brodaty, M.D.. F.R. A.C.P., F.R.A.N.Z.C.P., is Seni& Staff Spe&ist, Division of Psychiatry, Prince Henry Hospital, Sydney, Austraha. Philip Boyce, M.D., F.R.A.N.Z.C.P., is Senior Lecturer, School of Psychiatry, University of New South Wales, Sydney, Australia. (Reprint requests to Dr. P.B. Mitchell, Mood Disorders Unit, School of Psychiatry, Prince Henry Hospital, Little Bay, NSW. 2036, Australia.) 01651781/91/%03.50
@ 1991 Elsevier Scientific
ease” differs from residual disorders as a distinct type of depression or by severity. To demonstrate the validity of subtypes, a number of approaches have been adopted: (1) A search for necessary and sufficient clinical features (Foulds, 1973); (2) the use of multivariate statistical techniques such as factor analysis, discriminant analysis, cluster analysis, and latent class analysis; and (3) examination of biological markers such as the dexamethasone suppression test. Despite early enthusiastic claims (e.g., Carroll et al., 198 I). there is still no clinically or statistically derived system, or biological marker, yet accepted as a “gold standard.” More recently, hormonal responses to pharmacological challenges have been used to examine central neurotransmitter function indirectly in depression and so indicate possible subtypes (Checkley, 1980). Differences between clinically diagnosed melancholic and nonmelancholic depressive patients have been investigated with various putative challenges of central noradrenergic neurotransmitter systems (Matussek et al., 1988). Although findings have been inconsistent, significant differences between these depressive subtypes have been reported for the insulin tolerance test (Czernik, 1982), the growth hormone (GH) response to amphetamine (Langer et al., 1976) the cortisol response to methylamphetamine (Checkley, 1979) and the GH response to desipramine (Laakmann et al., 1986). There has been, however, considerable uncertainty about the specificity of these challenges for central noradrenergic systems. For this reason there has been particular interest in neuroendocrine responses to the specific a,-noradrenergic receptor agonist clonidine. The GH response to clonidine has been demonstrated to be reduced in depressed subjects compared to controls in most (Matussek et al., 1980; Checkley et al., 1981; Charney et al., 1982; Siever et al., 1982; Boyeretal., 1986; Amsterdamet al., 1989) though not all (Horton et al., 1986; Schittecatte et al., 1989) studies. This blunted GH response may also represent a trait abnormality in subjects with endogenous depression (Mitchell et al., 1988). The negative study of Horton et al. (1986) is of interest, being the only study to use a clonidine dosage of I .3 pgi kg body weight (i.v.), whereas the other studies used either 150 1.18or 2 pg/ kg i.v., or 5 pg/ kg oral. Horton et al. also used an inadequate drug-free interval of only 7 days before testing. (Periods of at least 2 or 3 weeks without medications are usually considered necessary in such studies.) As outlined in Table I, there have been seven published studies comparing the GH response to clonidine in endogenous and nonendogenous subjects, including the original report by Matussek et al. (1980). The major limitations of that original study were: (1) the lack of a standardized assessment interview and operationalized diagnostic criteria; and (2) a fatlure to match or control for either age or sex--factors that significantly affect the GH response to this challenge (Checkley, 1980). Of the subsequent studies, only two have accounted for these covariates in their comparisons (Checkley et al., 1984; Ansseau et al., 1988). Two other studies (Boyer et al., 1986; Horton et al., 1986) used inadequate drug-free intervals (< 2 weeks) before testing. Because of the significance of the methodologically strict study of Checkley et al. (1984) which found a blunted GH response in endogenous compared with reactive depressed subjects, we aimed to extend that study by: ( I) confirming that the dosage of clonidine used by that group (1.3 pgi kg) was also able to demonstrate hormonal differences between closely matched melancholic and control subjects; and (2) investi-
depression Diagnostic criteria for ED
Matussek et al 119801
NonED ED 0~) 0-0 10
Drugfree interval (days) 28
Age, sex, matching/ control No
1.3 pg/kg i.v.
5 pgfkg p.0.
Clonidine dose/ route
156 Pg i.v.
156 fig i.v.
119881 et al.
Note. Newcastle = Newcastle Scale (16 = endogenous, 55 = nonendogenous!. NS = no signiffcant difference in reduced growth hormone response In endogenous subjects growth hormone iGW response. * = signlflcantly 1. “Major depression” (classification system unspeclfled 1. 2. Minor depression by Research Dlagnostlc Criteria ~RDCI 3. Slgniflcant difference for mean cumulative response only
gating other hormonal responses to clonidine (adrenocorticotropic hormone [ACTH], cortisol, and prolactin) in addition to GH in the melancholic and nonmelancholic subtypes. In an associated study, we examined relationships between the GH response to clonidine and hormonal responses to a putative serotonergic agonist, fenfluramine. Although abnormal responses to pharmacological challenges of both noradrenergic (Matussek, 1988) and serotonergic systems (Meltzer and Nash, 1988) have been identified in depressed subjects, there has not yet been a published investigation that has used putative challenges of both neurotransmitter systems in the same subjects. Both the prolactin (Siever et al., 1984~; Coccaro et al., 1989) and cortisol (Weizman et al., 1988) responses to fenfluramine have been reported to be blunted in depressed subjects as compared to controls, though there has been one negative report (Asnis et al., 1988). We have recently reported a study of 27 depressed and 14 control subjects, in which we found a significantly diminished prolactin response in endogenous subjects as compared with controls, though this difference was partially dependent on reduced baseline prolactin levels and increased baseline cortisol levels in the depressed subjects (Mitchell and Smythe, 1990). We also demonstrated a reduced prolactin response to fenfluramine in endogenous compared to nonendogenous subjects (Mitchell et al., 1990).
182 Methods Subjects. Study 1. Hormonal responses to clonidine in melancholic and nonmelancholic depressed subjects. Twenty-six depressed subjects were tested with clonidine. All subjects
met DSM-III (American Psychiatric Association, 1980) criteria and Research Diagnostic Criteria (RDC; Spitzer et al., 1978) for major depression, derived from a semistructured interview (Brodaty et al., 1987). Because of recent evidence indicating that the number of patients allocated to depressive subtypes varies significantly depending on the classification system (Davidson et al., 1984) subjects were classified as having melancholic (endogenous) or nonmelancholic (nonendogenous) depression according to each of the following diagnostic systems: DSM-III, RDC, and ICD-9(World Health Organization, 1978). For the RDC system, subjects were divided into endogenous (definite endogenous) and nonendogenous (probable endogenous + nonendogenous) subtypes to examine a more clearly identified endogenous group. Likewise for [CD-9, subjects were allocated to endogenous (296.1,296.3) and nonendogenous (300.4) subgroups. No patient over the age of 65 was included in the study. Subjects with significant medical disorders or regular alcohol intake > 30 g daily were excluded. All subjects were required to have been free of antidepressant or other psychotropic medications for at least 2 weeks, although benzodiazepine hypnotics were allowed. Depression severity was assessed by both the observer-rated 2l-item Hamilton Rating Scale for Depression (Hamilton, 1967) and the Zung Self-rating Scale (Zung, 1986). Study 2: Hormonal responses to clonidine in endogenous/melancholic subjects and matched controls. Ten of the depressed subjects from study 1 who fulfilled DSM-III, RDC, and ICD-9 criteria forendogenousimelancholicdepression wereable to be matched with IO controls for age (* 5 years), sex, and menopausal status. Control subjects were assessed by clinical interview to exclude previous or current psychiatric disorders. Controls with significant medical disorders (as assessed by interview and physical examination) or regular alcohol intake > 30 g daily were excluded. They were also required to have been free of antidepressant or other psychotropic medications for at least 2 weeks, and were also excluded if they were taking other medications that might affect the neuroendocrine responses to clonidine. such as steroids or thyroid hormone replacement. Study 3: Hormonal responses to clonidine subjects. Fifteen of the 26 depressed subjects
in the samedepressed
from study 1 consented to testing with both fenfluramine and clonidine. For 12 of the 15 subjects, the fenfluramine test was performed first. The average interval (* SD) between tests was 4.3 (* 3.7) days (range 1-14 days). Table 2 provides details of the depressed samples for each of the three studies. Procedures. Clonidine
test. Patients were fasted overnight. and at 8-8:30 a.m.. a catheter was inserted into an antecubital vein, after which they rested for I hour. Three baseline samples were then taken at l5-min intervals. The clonidine (1.3 wg, kg body weight, as used by Checkley et al. [ 19841) was diluted in 10 ml of normal saline and injected slowly over IO min. For 90 min after the injection began, blood was withdrawn at 15min intervals for hormone estimation, Throughout the procedure, pulse, blood pressure, and observer ratings of sedation were made every 5 min. Fenfluramine test. This test was carried out on 15 of the same patients on a different test day. The same initial procedure was carried out as for the clonidine test. However. after the three baseline samples had been taken, n.t.-fenfluramine. 60 mg, was administered orally. Blood samples were then drawn hourly for 5 hours while the subject remained supine. A light meal was provided 4 hours after fenfluramine administration
Assays. Growth hormone levels were measured using polyclonal antiserum with an intra-assay coefficient of variation (CL’)of 5.80/c and an interassay cv of 10.99: at 3.95 mum I. Serum cortisol
Table 2. Details of subjects in each study Study 2
Female Total Postmenopausal
46.6 11 .o
40.7 11 .o
Bipolar disorder, depressed phase
RDC endogenous (definite)
Hamilton score Mean SD
Zung score Mean SD
Age Mean SD
Note. RDC = Research Self-rating Scale.
Rating Scale for Depression.
Zung = Zung
levels were measured by radioimmunoassay (RIA), with an mtra-assay cv of 6.8% and an interassay cv of 7.9% at 375 nmol, I. Plasma ACTH levels were measured by RIA, with an intra-assay cv of 13.2% at 26.6 ngil and an interassay cv of 23.5% at 25.4 ng/ I. Serum prolactin levels were also measured using a double antibody RIA wtth an interassay cv of 5% and an intra-assay cv of 6.5% at 658 mlUi 1. Cortisol assays were performed for 25 of the 26 depressed subjects and all of the matched controls, while ACTH levels were assayed for 17 of the depressed subjects and all 10 of the controls. However, only six of the matched pairs had complete ACTH assays for both subjects.
Data Analysis. Baseline concentrations of the various hormones were taken as the average of the three prechallenge levels. Hormonal responses were analyzed as the absolute mcrease to peak above baseline (A). The frequency distributions of the examined variables were assessed by means of the Shapiro and Wilk test for normality of data (as set out in Madansky, 1988). Apart
184 from subjects’ age and blood pressure
responses, all data were nonnormally distributed. For nonnormally distributed data, variables were log,,,-transformed before use of I tests. Student’s I test was used to compare group means. Repeated measures analysis of variance (ANOVA). testing for both linear (FIUI) and quadrattc (F ~UUI)trends, was used to look for any significant change in hormonal levels after the clonidine challenge. Analysis of covariance (ANCOVA) was used to determine the effect of covartates such as age and sex. All dependent variable data were log,,,-transformed before ANOVA or ANCOVA. Spearman’s rank order correlations (r~) were used to assess interrelationships between variables.
Results Study 1: Hormonal Responses to Clonidine in Melancholic and Nonmelancholic Depressed Patients. Baseline levels. As indicated in Table 3, there were no significant differences m baseline levels between DSM-/I/ melancholic and nonmelancholic subjects for any of
the four hormones. Similarly, there were no differences between subgroups defined according to either the RDC or ICD-Y diagnostic systems. Responses to clonidine. Repeated measures ANOVA of the combined depressed groups demonstrated a significant increase In GH ( Fqrru,/ = 20.4; l/f= I. 25; /I < 0.001) and significant reductions in both ACTH (F/J,, = 5.0; @= I, 15; />< 0.05) and prolactin (F/,,, = 8.0; &= I, 25; p < 0.0 1). There was no significant change in cortisol ( F/!jg=0.55; ~/f I, 24; p = 0.45). q
Table 3. Hormonal baseline levels and responses and nonmelancholic major depression Melancholic
Baseline levels GH mu/I, ACTH
2.2 401.1 82 3 210.5
26 178.9 83.7 173.1
nmol/l,2 rig/l 13
Note GH growth according to DSM-III 1 p 2 n 3n
(n = 13)
Mean Age yr
(n = 13)
Raw means are given
I 0 05 t test 11.14 8. 9
As indicated rn Table 3 and Fig. 1, there were no stgnificant differences in the responses of any of the hormones between DSM-III melancholic and nonmelancholic subjects (e.g., for A GH: I = -0.03, #= 24,~: 0.98). ANCOVAs covarying for either age (F = 0.00 I ; L//‘= 1, 25; p 0.97) or sex (F = 0.002; df= I, 25; p = 0.97) also indicated no q
185 Fig. 1. Growth hormone response melancholic depressed subjects
Time (minutes) Mean + SEM. DSM-//I
n = 13 [solid Ilne,; DSM-III
n = 13 #dotted line
significant differences m A GH between groups. Even after excluding the one nonmelancholic subject with an elevated baseline level (10.0 mUil), there was still no significant difference between groups for A G H (f = 0.3 1, u”= 23, p = 0.76). There were likewise no differences between groups defmed by RDC or /CD-9. To examine whether prior fenfluramine testing had an effect on responses to clonidine, the I2 subjects given prior fenfluramme were compared with the other 14. There were no significant differences between responses in these two groups. Within the sample of 12 subjects given prior fenfluramine, there was no significant correlation between the interval between the challenges and (GH or other hormonal) responses. Blood pressure. Effects on blood pressure were used as an indicator of the biologtcal activity of clonidine at this dose. Clonidine significantly reduced both systolic(l= -12.7,df= 25,p