TRH stimulation test as a predictor of acute and long-term antidepressant response in major depression

JOURNAL

OF

AFFECTIV DISORDER

ELSEVIER

Journal of Affective

Disorders

38

(I 996) 16S- 172

Research report

TRH stimulation test as a predictor of acute and long-term antidepressant response in major depression a3*, Maurizio Fava b, Greg Maislin ‘, Jerrold Rosenbaum Mady Hornig-Rohan a

Jay D. Amsterdam a Depression

Research

Unit, University

h Psychopharmacology

Received

17 October

I$ Pennsylumiu

Unif. Mussachusetts

School

of Medicine.

Genenrl

1995; revised 30 November

Hospital,

Philudelphiu,

‘,

PA 19104, USA

Boston. MA, USA

1995; accepted 2 January

1996

Abstract We assessed the prognostic utility of the TRH stimulation test by examining (a) the relationship between pre-treatment ATSH and acute response to fluoxetine treatment, and (b) the relationship between the change in ATSH (AATSH value) after repeated TRH testing at 6 weeks of fluoxetine treatment and long-term outcome during maintenance fluoxetine or placebo therapy. 43 MDD patients were studied with sequential TRH tests at 6-week intervals. Fluoxetine ‘responders’ were defined as patients with a Hamilton Depression Rating Scale score I 7 by week 9 of treatment and who remained in remission at least 3 additional weeks. These subjects were then randomized to one of four fluoxetine/placebo treatment groups and long-term outcome assessed. Overall, there was no difference in the mean pre-treatment ATSH values between acute fluoxetine responders and nonresponders. Moreover, we observed similar AATSH values in patients who maintained long-term remission compared to those who relapsed during maintenance with either fluoxetine or placebo. In contrast to prior reports of an higher AATSH value in long-term remitters, the present observation of similar mean AATSH values patients with long-term remission compared to those who relapsed suggest a limited prognostic utility for the TRH stimulation test in MDD. Keywords:

TRH Stimulation

test; Thyrotropin

response; Depression; Fluoxetine

1. Introduction Studies

of

thyrotropin

(TSH)

response

to thy-

(TRH) stimulation have demonstrated a ‘blunted’ TSH response in about 25% of depressed patients @range et al., 1972: Asnis et al., 1981; Kirkegaard, 1981; Winokur et al., 1983). rotropin-releasing

hormone

* Corresponding author. Depression Research Unit, University Science Center - 8th floor, 3600 Market Street, Philadelphia, PA 19104, USA. 0165-0327/96/$15.00 PII

0 1996 Elsevier Science B.V. All rights reserved

SOISS-0327(96)00010-9

In addition, some investigators have reported an increase in TSH response to TRH stimulation during antidepressant treatment (Kirkegaard et al., 1975, 1977; Kirkegaard, 1981; Langer et al., 1981, 1984; Extein et al., 1980; Linkowski et al., 1981; Targum, 1983; Krog-Meyer et al., 1985; Winokur et al., 19891, and that this increase may predict long-term remission of depressive symptoms. In this context, the change in TRH-induced TSH response during treatment may represent an unique ‘biological marker’ for predicting long-term outcome. Thus, Langer et al. (1986) observed a relationship between

166

J.D. Amsterdum

et ul. /Journal

of Affectioe Disorders 38 (I 996) 165-I 72

the change in TSH response to TRH (or ATSH) during treatment and long-term clinical outcome, with relapse occurring within 1 year in 36% of patients showing a persistently blunted TSH response. Similarly, Itoh et al. (1987) observed a relationship between TSH response to TRH after 4 weeks of clomipramine (n = 11) or nortriptyline (n = 12) therapy, with a mean AATSH value of 3.6 k 4.1 @U/ml in treatment responders and 1.1 f 1.9 +J/ml in drug nonresponders (P < 0.01). In an extensive series of studies, Kirkegaard and coworkers used an empirically derived AATSH value of 2.0 pIU/ml which appeared to predict long-term remission in 27/29 (93%) and relapse in 33/37 (89%) depressed patients who initially responded to antidepressant treatment (predictive value = 91%) (Kirkegaard, 1981; Krog-Meyer et al., 1985). More recently, Winokur et al. (1989) conducted a prospective study using a placebo substitution design and observed a significant relationship between AATSH and long-term response to desipramine ther-

2. Methods

episode. 63% had an initial onset of depression 5 20 years of age. Eleven subjects (28%) were experiencing their initial depressive episode, while 35% reported up to three prior episodes, and the remaining patients reported up to 25 prior episodes. 37% had experienced a poor inter-episode recovery (i.e., incomplete remission), while 30% of the patients had a chronic (> 2 year) episode. At the initial screen visit, all subjects had a minimum Hamilton Depression Rating Scale (HDRS) (Hamilton, 1960) score > 16 on a 17-item scale. Following a 1 week drug-free elimination period, the mean HDRS was 22 f 3 (with a range from 16 to 29). All patients had a complete physical examination and were free of any acute medical illness. Those who had any chronic or concurrent medical illness (e.g., essential hypertension) were stable on appropriate medical therapy. No patient had any evidence of acute renal, hepatic, endocrine or neurological disease, and none had any meaningful laboratory abnormalities. Patients with subclinical hypothyroidism (i.e., serum thyrotropin level > 4.5 pIU/ml) were excluded from the study. While the occasional use of lorazepam or chloral hydrate was permitted for severe insomnia, no concomitant psychotropic medication was prescribed beyond week 6 of fluoxetine treatment. All subjects discontinued their previous psychotropic medication at least 14 days prior to the study, and many were drug-free for longer periods. Finally, all subjects provided written informed consent in accordance with the Institutional Review Boards of the University of Pennsylvania and Harvard University.

2.1. Patients

2.2. Clinical procedures

A total of 43 outpatients were evaluated: 19 were from the Depression Research Unit at the University of Pennsylvania Medical Center and 24 were from the Psychopharmacology Research Unit, Massachusetts General Hospital. Overall, 31 women (mean + SD age 38 f 12 years) and 12 men (39 + 8 years) (range 19 to 62 years) were studied. All patients were screened by a research psychiatrist using the SCID format interview technique (Spitzer et al., 1988) and all satisfied DSM-III-R criteria (American Psychiatric Association, 1988) for major depressive disorder (MDD), single or recurrent

After a l-week drug-free elimination period, all patients received fluoxetine 20 mg daily in an openlabel fashion for at least 9 weeks. Patients were rated by a research psychiatrist on a weekly basis for 5 weeks, then bi-weekly for 1 month, then weekly until week 13 of treatment for efficacy using the HDRS (Hamilton, 1960). Clinical response to acute fluoxetine treatment was defined as a 2 50% reduction in the baseline HDRS score and/or a final HDRS score < 7 which was sustained for a minimum of 3 additional weeks. Responders were then randomly assigned to one

apy. In the present multi-center study, we assessed the relationship between basal and ATSH values and acute response to fluoxetine treatment. We further examined the relationship between the change in ATSH value after 6 weeks of fluoxetine therapy (i.e., AATSH value) as a predictor of sustained remission or clinical relapse during long-term (1 year) fluoxetine or placebo treatment.

J D. Amsterdum

er al. /Journul

of Affective

of four long-term (1 year) treatment groups: (a> fluoxetine 13 weeks followed by placebo for 39 weeks; (b) fluoxetine for 27 weeks followed by placebo for 25 weeks; (c) placebo for 52 weeks; or (d) fluoxetine for 52 weeks. Patients were initially examined weekly for 5 weeks, then bi-weekly for 4 weeks, then weekly for 3 weeks and finally monthly with bi-weekly phone contacts until the end of the study (at week 65). Determination of clinical relapse was based upon (a) a rise in HDRS score to 2 12 for at least 2 consecutive weeks plus (b) once again meeting DSM-III-R criteria for MDD. 2.3. TRH stimulation test procedure All tests were conducted on the Clinical Research Centers at University of Pennsylvania Medical Center and the Massachusetts General Hospital. The initial TRH stimulation test was performed during the drug-free, placebo lead-in week, and the second TRH test was repeated during week 6 of fluoxetine therapy. The TRH stimulation test procedure was as follows: After an overnight fast, at 08 : 30 h, subjects were placed recumbent in bed with an intravenous infusion of normal saline started in an anticubital vein. After obtaining baseline serum TSH samples at - 15 and 0 min, synthetic TRH (protiriline) 400 pg was administered as a slow intravenous bolus through a 3-way stopcock device and additional blood samples were then obtained at + 10, 20, 30, 45, 60 and 90 min for TSH determination. Thyroxine (T,), triiodothyronine (T,), and T, resin uptake determinations were also obtained prior to each TRH stimulation test, and had to be within normal limits. For a given subject, all TSH samples were assayed in duplicate in individual assay procedures as described below. 2.4. Assay procedures TSH concentrations were measured by double-antibody RIA assay technique plus ammonium sulfate precipitation. The TSH reference standard was calibrated against the First International Preparation of Human TSH for immunoassay. Intra-assay and interassay coefficients of variation were 6.5% and 7.6%, respectively. The sensitivity was 1.5 ,uIU/ml.

Disorders

38 (1996) 165-172

167

2.5. Statistical procedures Four parameters for analyzing the TSH response curve were examined. These included (a) baseline (time 0) TSH concentration, (b) peak TSH concentration after TRH, (c) maximum increase in TSH ( ATSH) after TRH, and (d) the net area under the TRH response curve (AUC). Pearson (and Spear-man rank order) correlations between pre-treatment ATSH and the peak TSH and AUC were r = 0.997 and r = 0.98, respectively. Similarly, high correlations were also seen between the ATSH and the peak and AUC TSH values after the second TRH test. Moreover, the Pearson and Spearman correlations between the ATSH values after each TRH test were r = 0.43 (P = 0.004) and s = 0.56 (P < O.OOl), respectively, indicating that the ATSH value represented a stable patient trait. Therefore, we computed the change in ATSH values (or the AATSH value) between the two TRH stimulation tests. Mean AATSH values, as well as individual ATSH values were computed for fluoxetine responders (n = 35) and nonresponders (n = 8). Comparisons between groups were then made using t-tests (followed by Wilcoxon rank sum test to insure that statistical conclusions were not dependent upon an assumption of normal sampling distributions). t-Tests for groups with unequal variance (Welch’s test) were also used wherever variances significantly differed, using an F-test at the alpha level of 0.10. Multivariable logistic regression (Hosmer and Lemeshow, 1989) was also used to assess the relationship of the three TSH response variables and treatment response, while controlling for gender, age and pre-treatment illness severity. Similar analyses were independently performed among the fluoxetine responders (n = 351, examining the associations of the TSH variables with the likelihood of relapse on long-term fluoxetine (or placebo). First, mean basal TSH and ATSH values were compared between patients who relapsed and those who did not. Then, multivariable logistic regression analyses were used to assess the relationship between ATSH and the likelihood of relapse, controlling for gender, age and length of remission (13 weeks vs. 27 weeks vs. 51 weeks vs. 64 weeks). All analyses were implemented using SAS/PC version 6.08 for Windows (SAS Institute, Cary, NC).

168

J.D. Amsterdum et ul. /Journal

of Af/eectiue Disorders

3. Results 3.1. Basal and ATSH responses The mean (+ SD) basal TSH values and the ATSH values after each TRH stimulation test are displayed in Table 1. The mean pre-treatment ATSH values were similar for fluoxetine responders and nonresponders (P = 0.19, Wilcoxon test). Similarly, there was no difference in the mean ATSH values at the 6-week TRH test in acute fluoxetine responders vs. nonresponders (P = 0.94). The mean change in ATSH values between TRH tests (AATSH value) did not significantly differ between acute responders and nonresponders (P = 0.23). Because several prior reports suggested that relative TSH blunting after TRH stimulation might have prognostic utility (Kirkegaard et al., 1975; Kirkegaard et al., 1977; Kirkegaard, 1981; Extein et al., 1980, 1982; Langer et al., 1984; Winokur et al., 1989; Fava et al., 1992), we compared patients with a pre-treatment ATSH value I 7 ~IU/ml to those with a ATSH value > 7 ~IU/ml. Two out of the 4 (50%) with a pre-treatment ATSH value < 7 ~IU/ml and 33 out of 39 (85%) patients with ATSH > 7 ~IU/ml responded to acute fluoxetine treatment (P = 0.15, two-sided Fisher’s Exact test).

38 (1996) 165-I

72

0.004) at pre-treatment. Interestingly, pre-treatment, basal TSH values were also significantly different between fluoxetine responders and nonresponders (P = 0.03) (see Table 2). While the mean AATSH values did not demonstrate a gender difference, there was a smaller proportion of men who responded to fluoxetine, and this factor may have hidden a difference in ATSH between fluoxetine responders and nonresponders during the second TRH test. Thus, 8 of the 12 men (67%) and 27 of the 31 women (87%) responded to treatment ( x 2 = 0.12). Additionally, there was no correlation between age and any of the ATSH variables, although there was a significant correlation between age and pretreatment basal TSH values (r = 0.32, P = 0.03) and a negative correlation between pre-treatment HDRS scores and AATSH values (s = 0.33, P = 0.03). Multivariable logistic regression models were used to produce adjusted odds ratios for the effect of a 1 unit increase in pre-treatment ATSH or AATSH, while controlling for age, gender and illness severity. The adjusted odds ratio for the pre-treatment ATSH was 0.99 (95% CI = 0.92 to 1.06) (P = 0.66). The adjusted odds ratio for AATSH was 0.94 (CI = 0.82 to 1.09) (P = 0.40). Thus, controlling for age, gender and illness severity did not reveal a significant association. Moreover, none of the other variables in the models were statistically significant.

3.2. Gender differences 3.3. ATSH and long-term outcome Mean basal TSH and ATSH values for men and women are displayed in Table 2. Women had a slightly lower pre-treatment basal TSH value compared to men (P = 0.19) and a significantly higher mean ATSH value after TRH stimulation (P =

Table 1 Mean ( f SD) basal and ATSH ( pIU/ml)

values in fluoxetine responders

Pretreatment

Responders (n = 35) Non-responders (n = 8)

and nonresponders During treatment

Difference

Basal TSH

ATSH

Basal TSH

ATSH

AATSH

2.3 f 1.2 1.7 f 0.6

16.1 f 10.8 15.7 f 16.9

2.2 f 1.6 2.1 f 1.0

14.0 f 8.3 16.6 f 15.2

- 2.2 f 12.7 1.0 f 4.1

_

0.82

0.07

0.94

-

Significance tests Pooled t-test

Welch’s t-test

Relapse rates for patients randomized to receive fluoxetine for a total of (a) 13 weeks, (b) 27 weeks, (c) 52 weeks or (d) 64 weeks were 4/6 (67%), 9/l 1 (82%), 3/10 (30%) and 2/6 .(33%), respectively.

_

0.64

0.23

J.D. Amsterdam er al./Journul

c$Affecrirre Disorders 38 (1996)

169

165-172

Table 2 Mean ( f SD) basal and ATSH ( pIU/ml)

values in fluoxetine

patients by gender During treatment

Pre-treatment

Men (n = 12) Women (n = 3 I) Significance

Difference

Basal TSH

ATSH

Basal TSH

ATSH

2.2 f 0.8 2.1 f 1.2

10.4 * 3.2 18.2 + 3.2

3.0 f 1.88 1.9 f 1.3

9.2 f 5.2 16.5 f 10.4

I 8 + 13.7

_ 0.19

0.004

_ 0.03

0.004 _

0.82 _

AATSH 1.2 f 3.1

tests

Welch’s t-test Pooled t-test

1.07 (95% CI = 0.9 to 1.3) (P = 0.49). Within this model, men appeared less likely to relapse with an odds ratio of 0.03 (95% CI = 0.001 to 0.58). Moreover, an increased age appeared to be associated with an increased likelihood of relapse (odds ratio for 5-year increments = 1.6; 95% CI = 0.9 to 3.0). The adjusted odds ratio for the AATSH was 0.97 (95% CI = 0.9- 1. I), with similar results being observed for age and gender in the AATSH model. Finally, in all of the analyses described above, no substantive differences were observed when Wilcoxon Rank Sum tests were used in place of f-tests.

The linear trend in the relapse rate was statistically significant ( x2 = 4.9, df = 1, P = 0.03). Of the 18 patients who relapsed, one-half relapsed during double-blind randomization to fluoxetine and the other half during randomization to placebo. The mean pre-treatment basal and ATSH values for patients who did or did not relapse during longterm therapy is displayed in Table 3. Values did not differ significantly between these groups (P = 0.56). Similarly, there were no significant differences in TSH values during the second TRH stimulation test (P = 0.51). The mean AATSH value for patients with sustained remission were not significantly different from those who relapsed (P = 0.35). When relapsers vs. nonrelapsers were compared based upon a pretreatment ATSH value < 7 pIU/ml, 2/2 (100%) of patients relapsed compared to 16/33 (49%) with ATSH values > 7 ~IU/ml. A multivariable logistic regression model was used to examine the relationship between pre-treatment ATSH and AATSH and the likelihood of relapse, while controlling for differences in treatment status for groups (a> through (d), age and gender. The adjusted odds ratio for pre-treatment ATSH was

Table 3 Mean ( + SD) ATSH ( pIU/ml) values in long-term Pre-treatment

ATSH

4. Discussion The present observations suggest that changes in the TSH response to TRH stimulation do not consistently predict either acute treatment response or long-term remission from depression during fluoxetine (or placebo) therapy. Prange et al. (1972) initially reported a diminished or ‘blunted’ TSH response to TRH stimulation in some depressed patients, and many studies have

remitters and relapsers During treatment

ATSH

Difference

Remitters (n = 17) Relapsers (n = 18) Signifance tests

15.0 f 6.5 17.2 f 13.8

14.9 + 8.5 13.0 f 8.2

-0.1 -4.2

Welch’s t-test Pooled I-test

0.56 _

_ 0.51

_

0.35

AATSH

+ 7.8 + 16.2

170

J.D. Amsterdam

et al. /Journal

qf Arectiue

subsequently reported a blunted TSH response as well (see Winokur et al., 1983; Itoh et al., 1987). Some investigators have hypothesized that a blunted TSH response to TRH stimulation may also represent a ‘biological marker’, and might predict acute and/or long-term treatment outcome during antidepressant therapy (Kirkegaard et al., 1977; Kirkegaard, 1981; Extein et al., 1980, 1982; Targum, 1983; Langer et al., 1984; Krog-Meyer et al., 1985; Winokur et al., 1989). Recently, Fava et al. (1992) found that the pre-treatment TSH responses to TRH might predict acute antidepressant response in a double-blind study where active drug did not differentiate from placebo, while others have reported that an increase in TSH response to repeated TRH stimulation tests might predict long-term clinical remission or relapse with a variety of antidepressant treatments (Kirkegaard et al., 1977; Kirkegaard, 1981; Extein et al., 1980, 1982; Krog-Meyer et al., 198.5; Winokur et al., 1989). Early studies examining the relationship between long-term clinical outcome and AATSH after repeated TRH testing have been limited in scope and methodology. Most of these studies, while prospective in design, did not utilize any controlled treatment methodology for assessing true, long-term outcome status. In this context, few studies used a single antidepressant intervention tested against placebo. Kirkegaard et al. (1975) assessed the prognostic utility of the change in ATSH during electroconvulsive therapy in 15 patients, and found longterm remission associated with an increase in ATSH value. Winokur et al. (1989) also assessed the longterm prognostic utility of repeat TRH testing under double-blind, placebo-controlled treatment in 45 patients with MDD initially treated for 6 weeks with desipramine at I 300 mg daily (with therapeutic plasma levels), and then desipramine (or placebo) for an additional 6 months. 36 patients were categorized as responders and 9 as nonresponders (as defined by a > 50% reduction in HDRS scores plus a final score I 9). The change in ATSH after repeated TRH testing was significantly increased in the responders (3.1 + 5.8 @U/ml) (P = 0.003) and decreased in the nonresponders ( - 1.2 + 1.6 QJ/m 1) (P = 0.052) with a between-group difference of P = 0.001. Moreover, when the criterion of a AATSH 2 2.0 pIU/ml (Kirkegaard, 198 1) was applied, 47%

Disorders

38 (1996) 165-I

72

of desipramine responders and 0% of nonresponders had a AATSH value 2 2.0 pIU/ml (P = 0.01) (Winokur et al., 1989). In a subsequent study of similar design in 19 patients treated with sertraline I 200 mg daily for 6 weeks then randomized to sertraline or placebo for an additional 3 months, a significant increase in the mean AATSH value was observed in patients with sustained (n = 5) (6.2 f 4.8 pIU/ml) and partial remission (n = 8) (7.6 f 8.4 pIU/ml) compared to relapsers (n = 6) (0.1 f 5.6 $U/ml) (P = 0.05) (Winokur and Amsterdam, unpublished data). Several factors might explain the contrasting results observed in the present study. For example, our investigation utilized fluoxetine as the sole antidepressant agent, and this may have influenced the TSH response to TRH stimulation. In this regard, studies of the direct application of serotonin into the third ventricle of experimental rats have shown an increase in serum TSH concentration which is prevented by pretreatment with cyproheptadine (a serotonin antagonist) (Jordan et al., 1978). Langer et al. (1981) also observed an increase in the TSH response to TRH after chronic treatment with chlorimipramine (a potent serotonin reuptake site blocker) in 24 unipolar depressed women, while Widerlov et al. (1978) did not note any differential effect of chlorimipramine or nortriptyhne on TSH response in 7 healthy control subjects after a l-week administration. Similarly, Kirkegaard et al. (1977) did not see any effect of 5-hydroxytryptophan on TSH response to TRH stimulation in 7 depressed patients, while Shelton et al. (1993) observed no differences in the effects of chronic fluoxetine or desipramine treatment on thyroid axis function. Finally, studies examining thyroxine (T,) concentrations have found a lower than normal serum T4 in depressed patients undergoing tricyclic antidepressant treatment 1988; Joffe and Singer, 1990) (Baumgartner et al., while similar changes in TJ were not observed in the cerebrospinal fluid of animals (Joffe et al., 1993). These conflicting observations suggest that complex influences on neurotransmitter systems are normally involved in the regulation of TSH secretion and thyroid axis function, and that these factors become deranged during an affective episode. Moreover, the mechanisms underlying antidepressant action differ substantially among drug classes, and it is possible

J.D. Amsrerdum

et al. /Journal

c$Affectiue

that the TRH stimulation test more consistently predicts the action of treatments that alter multi-transmitter systems (e.g., TCAs and ECT) compared to drugs that largely affect only a single neurotransmitter system. Other caveats should be considered in the interpretation of the present results. For example, we studied a disproportionate number of women who demonstrated a higher pre-treatment ATSH value (Table 2). In this context, Snyder and Utiger (1972) reported a normally lower TSH response to TRH in men (especially men > 40 years of age). Although this factor may have contributed to our negative findings, we observed only a modest gender effect on the ATSH values in the present study. Additional factors which might have influenced the present results include that of subject age (Snyder and Utiger, 1972; Baumgartner et al., 1986), diagnostic subtype (Amsterdam et al., 1979; Kirkegaard et al., 1978; Linkowski et al., 1981; Targum et al., 1982; Baumgartner et al., 1986), severity of illness (Linkowski et al., 19841, in- or outpatient status (Winokur et al., 1983), as well as TRH dose used (see Loosen et al., 1977), among other clinical factors. In conclusion, we administered repeated TRH stimulation tests before and during 6 weeks of fluoxetine treatment in 43 depressed patients, and examined the relationship between ATSH and acute antidepressant response, as well as the relationship between AATSH and long-term remission or relapse. Overall, we observed no significant relationship between TSH response valu.es and acute or chronic treatment outcome, nor did we demonstrate a prognostic utility for the AATSH value after repeated

TRH

stimulation

tests.

Acknowledgements This study was supported by grants from Lilly Research Laboratories, Clinical Research Center Grant NIH 5-MO]-RR0040 (University of PA), The Jack Warsaw Fund for Research in Biological Psychiatry (University of PA), and the Clinical Research Center at Massachusetts General Hospital. We extend our deep appreciation to the Nursing Staffs of the Clinical Research Centers at the Hospital of the

Disorders

38 C19961 165- I72

171

University of Pennsylvania and Massachusetts General Hospital for their assistance in performing the TRH test procedures.

References

American Psychiatric Association Committee on Nomenclature and Statistics: Diagnostic and Statistical Manual on Mental Disorders, 3rd Edn., revised. American Psychiatnc Press, Washington, DC, 1988. Amsterdam, J.D., Winokur, A., Mendels, J. and Snyder. P. (1979) Distinguishing depression subtyped by thyrotropin response to TRH test. Lancet ii, 904. Asnis, G.M., Sachar, E.J., Halbreicht, U., Nathan, R.S., Ostrow, L., Soloman, M. and Halpem, F.S. (1981) Endocrine responses to thyrotropin-releasing hormone in major depressive disorders. Psychiatry Res. 5, 205-215. Baumgartner, A.K., Graff, A.K., Kurtin, I. and Meinhold, H. (1988) The hypothalamic-pituitary-thyroid axis in psychiatric patients and healthy subjects: Parts 1-4. Psychiatr. Res. 24, 271-332. Baumgartner, A., Hahnenkamp, L. and Meinhold, H. (1986) Effect of age and diagnosis on thyrotropin response to thyrotropin-releasing hormone in psychiatric patients. Psychiatry Res. 17, 285-294. Extein, I., Pottash, A.L.C., Gold, M., Cadet, J., Sweeney, D.R., Davies, R.K. and Martin, D.M. (1980) The thyroid-stimulating hormone response to thyrotropin-releasing hormone in mania and bipolar depression. Psychiatry Resr 2, 199-204. Extein, I., Pottash, A.L.C., Gold, M. and Silver, J.M. (1982) Thyroid-stimulating hormone response to thlrotropin-releasing hormone in unipolar depression tefore and after clinical improvement. Psychiatry Res. 6, 16 I - 169. Fava, M., Rosenbaum, J.F., Bimbaum, R., Kelly, K.. Otto, M.W. and MacLaughlin, R. (1992) The thyrotropin response to thyrotropin-releasing hormone as a predictor of response to treatment in depressed outpatients. Acta Psychiatr. Stand. 86. 42-45. Hamilton, M. (1960) A raring scale for depression. J Neural. Neurosurg. Psychiatry 23, 56-62. Hosmer, J.W. and Lemeshow, S. (1989) Applied Logistic Regression. Wiley, New York, NY. pp. 135-145. Itoh, N., Matsui, N., Fuwano. S., Yaginuma, H., Miyashita, 0. and Sakai. M. (1987) Serial DST, TRH tests and TRH-like immunoreactivity measurements in major affective disorders. Biol. Psychiatry 22, 559-572. Joffe. R., Nobrcga, J., Kish, S., Calva. R., Dixon, L. and Wilson. J. (1993) Desipramine reduces plasma but not brain thyroxine levels. Biol. Psychiatry 33, 293-294. Joffe, R.T. and Singer, W. (1990) The effects of tricyclic antidepressants on basal thyroid hormone levels in depressed patients. Pharmacopsychiatry 23, 67-69. Jordan, D.. Poncet. C.. Momex. R. and Ponsin, G. (1978) Partici-

172

J.D. Amsterdam et al. /Journal

ofAffective

pation of serotonin in thyrotropin release. I. Evidence for the action of serotonin on thyrotropin releasing hormone release. Endocrinology 103. 414-419. Kirkegaard, C. ( I98 I > The thyrotropin response to thyrotropin-releasing hormone in endogenous depression. Psychoneuroendocrinology 6, 189-212. Kirkegaard, C., Bjorum, N., Cohn, D., Faber, J., Lauridsen, U.B. and Nerup, J. (1977) Studies of the influence of biogenic amines and psychoactive drugs on the prognostic value of the TRH stimulation test in endogenous depression. Psychoneuroendocrinology 2, 13 I - 136. Kirkegaard, C., Bjorum, M., Cohn, D. and Lauridsen, U.B. (1978) Thyrotropin-releasing hormone (TRH) stimulation test in major depressive illness. Arch. Gen. Psychiatry 35, 1017-1021. Kirkegaard, C., Norlem, N., Lauridsen, U.B., Bjorum, N. and Christiansen, C. (1975) Protireline stimulation test and thyroid function during treatment of depression. Arch. Gen. Psychiatry 32, 1115-1118. Krog-Meyer, I., Kirkegaard, C., Kijne, B., Lumholtz, B., Smith, E., Lykke-Olesen, L. and Bjorum, N. (1985) Effects of amitriptyline on the thyrotropin response to TRH in endogenous depression. Psychiatry Res. 15, 145-151. Langer, G., Karobath, M., Sieghart, W., Aschauer, H., Placheta, P. and Spona, J. (1981) Effects of antidepressant treatment with clomipramine on hormonal responses to thyrotropin-releasing hormone and insulin-induced hypoglycemia: Implications for the ‘monoamine hypothesis’. Pharmacopsychiatry 14, 100-106. Langer, Cl., Konig, G., Hatzinger, R., Schonbeck, G., Resch, F., Aschauer, H., Keshavan, M.S. and Siegart, W. (1986) Response of thyrotropin to TRH as a predictor of treatment outcome: prediction of recovery and relapse in treatment with antidepressants. Arch. Gen. Psychiatry 43, 861-868. Langer, G., Resch, F., Aschauer, H., Keshayan, MS., Koinig, G.. Schonbeck, G. and Dittrich, R. (1984) TSH-response patterns to TRH stimulation may indicate therapeutic mechanisms of antidepressant and neuroleptic drugs. Neuropsychobiology I 1, 213-218. Linkowski, P., Beauman, S. and Mendekwicz, J. (1981) Thyrotropin response to thyrotropin-releasing hormone in unipolar and bipolar affective illness. J. Affect. Dyord. 3, 9-16.

Disorders 38 (1996) 165-I

72

Linkowski, P., Van Wettere, J.-P., Kerkofs, M., Gregoire, F., Brauman, H. and Mendelwicz, J. (1984) Violent suicidal behavior and the thyrotropin-releasing hormone-thyroidstimulating hormone test: a clinical outcome study. Neuropsychobiology 12, 19-22. Loosen, P.T., Prange Jr., A.J., Wilson, I.C., Lara, P.P. and Pettus, C. (1977) Thyrotropin stimulating hormone response after thyrotropin-releasing hormone in depressed, schizophrenic and normal women. Psychoneuroendocrinology 2, 137-148. Prange Jr, A.J., Wilson, I.C., Lara, P.P., Allsop, L.P. and Bruse. G.R. (1972) Effect of thyrotropin-releasing hormone in depression. Lancet ii, 999-1002. Shelton, R.C., Winn, S., Ekhatore, N. and Loosen, P. (1993) The effects of antidepressants on the thyroid axis in depression. Biol. Psychiatry 33, 120-126. Snyder, P.J. and Utiger, R.T. (1972) Response to thyrotropin-releasing hormone (TRH) in normal men. J. Clin. Endocrinol. Metab. 34, 380-385. Spitzer, R.L., Williams, J.B.W., Gibbon, M. and First, M.B. (1988) Structured clinical interview for DSM-III-R. Biometrics Res. Dept., New York, NY. Targum, S.D. (1983) The application of serial neuroendocrine challenge studies in management of depressive disorders. Biol. Psychiatry 18, 3- 19. Targum, S.D., Bymes, S.M. and Sullivan, SC. (1982) The TRH stimulation test in subtypes of unipolar depression. J. Affect. Disord. 4, 29-34. Widerlov, E., Wide, L. and Sjostrom, R. (1978) Effects of tricyclic antidepressants on human plasma levels of TSH and prolactin. Acta Psychiatr. Stand. 58, 449-456. Winokur, A., Amsterdam, J.D., Berwish, N. and Phillips, J. (1989) The relationship between the change in thyrotropin response to TRH and therapeutic response to DMI in depressed patients. 44th Meeting of the Society of Biological Psychiatry, San Francisco, CA. Winokur, A., Amsterdam, J.D., Oler, J., Mendels, J., Snyder, P.J., Caroff, S. and Brunswick, D.J. (1983) Multiple hormonal responses to TRH administration in depressed patients and healthy volunteers. Arch. Gen. Psychiatry 40, 525-53 I.