Neuroendocrine Responses to Inhibitors of Steroid Biosynthesis in Patients With Major Depression Resistant to Antidepressant Therapy

Beverley E Pearson Murphy, MD, PhD¹, A Missagh Ghadirian, MD², Veena Dhar, MD³

Objective: Patients with major depression frequently have high cortisol levels and resistance to dexamethasone. We sought to determine to what extent major depression might be influenced by inhibitors of steroid biosynthesis and to study the endocrine changes produced.

Method: After drug washout, 20 treatment-resistant patients with major depression were given aminoglutethimide, metyrapone, and/or ketoconazole, along with a small dose of cortisol for 8 weeks. Hamilton Depression Rating Scale (HDRS) ratings, 8:00 AM cortisol, dehydroepiandrosterone sulfate (DHAS), adrenocorticotropin (ACTH), and testosterone levels were followed weekly or oftener. A dexamethasone suppression test (DST) was conducted before and after treatment.

Results: Seventeen patients (85%) completed the course of treatment, and a significant mean drop (P £ 0.0001) of 50% in the HDRS score occurred by 7 weeks of treatment. Cortisol levels fluctuated widely and were often still high after the patient had improved clinically. Dehydroepiandrosterone sulfate levels fell more uniformly and were found to be a useful indicator of compliance and, to some extent, efficacy with aminoglutethimide and ketoconazole therapy. The correlation between DHAS and HDRS (r = 0.94) was significant (P = 0.02). Testosterone levels in men fell with ketoconazole but returned promptly to normal at the end of treatment. Adrenocorticotropin levels were normal or elevated, depending on the assay used, and rose (P = 0.07; n = 13) in most subjects during therapy. Of the 6 responders who had nonsuppressor DSTs before starting therapy, 5 had reverted to normal 1 to 2 weeks following cessation of therapy (P = 0.0006).

Conclusions: Abnormal metabolism of adrenocortical steroids may perpetuate depression, and alterations of synthesis or metabolism of these steroids may lead to a remission.

(Can J Psychiatry 1998;43:279–286)

Key Words: antiglucocorticoid, depression, cortisol, aminoglutethimide, metyrapone, ketoconazole, adrenocorticotropin, dehydroepiandrosterone

Patients with major depression frequently have high cortisol levels and resistance to their suppression by dexamethasone; however, the diurnal rhythm, while blunted, is maintained, and cortisol levels are suppressed with high doses of dexamethasone (1). Such patients lack the physical and other biochemical manifestations of Cushing’s syndrome. Conversely, patients with Cushing’s syndrome are often depressed (about 70%), and about 10% are suicidal (2). These psychiatric changes occur regardless of the etiology. Depression in such patients is often resistant to antidepressant medication but responds to treatment of the underlying cause. The psychiatric symptoms of Cushing’s syndrome may also be produced by the exogenous administration of corticosteroids, although these differ from those of the endogenous form in that they predominantly tend to euphoria, while those associated with endogenous Cushing’s syndrome are mainly depressive (1–3). Indeed, dexamethasone has been shown to improve depressive symptoms in some cases (4).

Because of the response of depression in Cushing’s syndrome to decreasing adrenal steroid production either medically (with the medications used here) or surgically, and because of the similarities of the depression in the 2 conditions, we speculated that major depression might respond to antiglucocorticoid medications (1,5).

To explore this possibility, a clinical trial of steroid-suppressive therapy was undertaken in a series of 20 patients with major depression resistant to conventional therapy. Preliminary results and the analysis of the Hamilton Depression Rating Scale (HDRS) data have been published previously, and the neuroendocrine data and details of treatment are presented here (5–7). This was the first clinical trial of such therapy in major depression, and confirmatory studies of the clinical efficacy of the medications used here have appeared recently (8–11).

In this study, 11 (65%) of the 17 subjects who completed the treatment responded markedly; 2 (12%) responded partially; and 8 (47%) maintained their improvement for 5 months to more than 5 years without further antidepressant therapy.

Method

Sample

Criteria for inclusion were described previously (6). Briefly, the patients were 18 to 65 years old, had a diagnosis of severe major depression, as defined by the Diagnostic and Statistical Manual (DSM- III-R) and confirmed by 2 psychiatrists, had a score ³ 20 on the 21-item HDRS, were drug-resistant (no response to 2 or more antidepressant drugs [cyclic or monoamine oxidase inhibitors] given in adequate dosage for at least 4 weeks), had no major medical problems, and women were infertile or using adequate nonhormonal means of contraception (12,13). The protocol was approved by the local ethics committee, and informed written consent was obtained from each patient.

The patient population comprised 13 women and 7 men with mean ages of 42.6 ± 11.3 SD years (range 26 to 60) and 51.4 ± 12.5 years (range 34 to 64) respectively; this age difference was not significant (P = 0.13) (Table 1). Eleven patients were nonpsychotic and depressed, 3 of whom had experienced manic episodes. Nine had psychotic features, and all but 2 of the patients (#4,5) had at least 5 melancholic features.

Two patients (#14,16) had been diagnosed as having primary hypothyroidism and were on satisfactory replacement therapy, which was continued throughout the study. Three were receiving medications (captopril, hydrochlorothiazide) for mild hypertension, and these were withdrawn with no adverse effects. One patient (#12) was receiving azathioprine (125 mg daily) for immunosuppression of a renal transplant done 16 years previously, and this medication was continued. Apart from mild hypertension, no patient had any signs of Cushing’s syndrome. Routine pretreatment biochemical and hematological data were all normal.

Control subjects were 15 apparently healthy hospital personnel of similar age who were not taking medications and who had no history of mental problems.

Laboratory Investigations

Cortisol, dehydroepiandrosterone sulfate (DHAS), adrenocorticotropin (ACTH), routine serum biochemical tests (glucose, creatinine, electrolytes, liver function tests), and hematology (hemoglobin, white blood cell count, platelets, smear) values were determined between 7:30 and 8:30 AM on several occasions before treatment, once or twice  weekly during treatment, and weekly for 2 months following treatment. A dexamethasone suppression test (DST) (1 mg dexamethasone at bedtime; samples drawn at 8:00 AM and 4:00 PM the following day) was carried out before and approximately 2 weeks following treatment. A DST was considered nonsuppressor if either value exceeded 136 nmol/L.

Thyroid function tests (serum thyroxine and triiodothyronine, triiodothyronine uptake, and thyrotropin) were done before treatment, weekly during treatment, and following treatment. Serum testosterone levels were determined in the men.

Routine determinations were made by the hospital laboratories. Adrenocorticotropin was determined using a commercial kit (RSL), and some samples were split and also assayed using another commercial kit (DSL). All of the assays had intraassay and interassay coefficients of variation less than 5% and 10% respectively. Unexpectedly high cortisol levels were confirmed using a different antibody (r = 0.85 for 46 samples, P £ 0.0001).

Clinical Psychiatric Evaluation

Patients’ assessments included clinical evaluation to confirm the diagnosis and follow-up evaluation weekly by 2 physicians who independently used the 21-item HDRS; their total scores correlated well (r = 0.90, P £ 0.00l). A marked response was defined as a reduction in the total score of at least 50% to a HDRS score £ 15, with persistence of improvement to the end of the treatment period. A partial remission was considered to be a reduction of 20% to 50%, with a clear improvement in behaviour and the ability to cope well at home.

Blood pressure, pulse, and temperature were monitored 4 times daily for the first 2 weeks, twice daily thereafter while the patient was in hospital, on each weekly outpatient clinic visit, and every 1 or 2 weeks for 2 months after stopping treatment. Treatment was initiated in hospital, where patients remained as long as required.

Psychopharmacological Schedules

Before starting the therapy, all medications (with the exceptions previously noted) were withdrawn during a washout period of 3 days to 3 weeks, depending on the medications used previously. Initially, it was legally possible to use only aminoglutethimide (AG) and metyrapone (MET). After the sixth patient had been treated, permission was obtained to use ketoconazole (KC). Aminoglutethimide was given in a dose of 500 to 1750 mg daily, MET 250 to 2000 mg daily, and KC 400 to 1200 mg daily. All were started at the minimum dose, which was increased (by 250 mg increments for AG and MET and by 200 mg for KC) every 4 days until the DHAS fell below 3 nmol/L, the HDRS fell below 15, or the patient developed troublesome side effects. If improvement was slow (no clear reduction in HDRS after 3 weeks and poor response in DHAS), one of the other agents was added. For the first 6 patients treated, AG was used as the first medication and MET was added if required; subsequently, KC was the first medication used.

Twenty mg of cortisol (hydrocortisone) was given at bedtime to ensure that there was no sudden drop in glucocorticoids and also to decrease the ACTH levels induced by hypothalamo-pituitary-adrenal feedback (22,24). By 8:00 AM the next morning, the effect on the cortisol level of the cortisol given at bedtime would be expected to be negligible because it is rapidly absorbed and has a short half-life. Since aldosterone levels are known to be decreased by these medications, serum electrolytes were carefully monitored.

Supplementary Medication

Occasional medication was permitted for insomnia (chloral hydrate), anxiety (diazepam), and headache (acetominophen).

Statistical Analysis

Groups were compared using repeated measures ANOVA and also by Student t-test. To facilitate comparisons, the mean data for the hormonal levels in each subject were determined for 2-week periods.

Results

Early Terminations and Compliance With the Protocol

Three patients did not complete the treatment: one because she took an overdose of an antidepressant (#11), another because of side effects (#8), and a third because of noncompliance (#20). Thus the completion rate was 17 out of 20 or 85%.

Side Effects

Side effects were mild to moderate (Table 2). If they became troublesome, the first medication was decreased and a second medication was started. Termination was required in only 1 patient (#8) because of catatonia, which was probably unrelated. One patient (#14) developed a rash with AG, which is a common occurrence with this medication, which rarely requires discontinuation (14,15), and refused to take the medication. Metyrapone was substituted.

Patients tended to lose weight. Of the 10 patients who were adequately documented, 8 lost weight and 2 gained weight, the mean change being –2.5 ± 0.79 SE kg (range +1 to –7)(P = 0.0125).

Routine Laboratory Tests

There were few changes. In 2 cases (#10,18), when the sodium dropped slightly below normal, 9-fluorohydrocortisone 0.05 mg daily was added until the end of treatment. Liver function tests rose slightly in most patients but exceeded the acceptable limit (twice the upper limit of normal) in only 1 patient on KC (#19). When the patient was on AG, the liver function tests returned to normal.

Pretreatment Hormonal Levels

The average 8:00 AM cortisol level before therapy was 714 ± 252 SD nmol/L, significantly (P = 0.03) above that for 15 controls (551 ± 101 SD nmol/L [range 422 to 736]). Eleven of the 20 patients (55%) and 1 normal subject had levels above the upper limit of the normal 8:00 AM range (248 to 690 nmol/L). Cortisol levels in psychotic patients (861 ± 346 SD) tended to be higher than those in nonpsychotic patients (640 ± 219 SD; P = 0.15). Also, the difference between responders (685 ± 248) and nonresponders (746 ± 330) was not significant (P = 0.67).

Dehydroepiandrosterone sulfate mean levels were 4.4 ± 2.3 SD µmol/L in women (normal range 2.2 to 8.1) and 4.4 ± 2.1 µmol/L in men (normal range 2.7 to 10.9). Three women and 2 men (25%) had low levels. Mean levels of psychotic and nonpsychotic patients did not differ (P = 0.77); nor did those of responders and nonresponders (P = 0.52). The correlation between age and DHAS levels was not significant (r = –0.25, P = 0.28).

Adrenocorticotropin (RSL) levels (n = 13, seven of whom had cortisol levels ³ 690 nmol/L, were within the normal range (12.6 ± 3.6 SD pmol/L compared with 16.4 ± 9.4 SD pmol/L in 15 controls (P = 0.15). The difference between psychotic (16.2 ± 5.7 SD pmol/L) and nonpsychotic patients (11.6 ±3.5) was borderline (P = 0.10). Responders and nonresponders did not differ (P = 0.31).

There was a highly significant correlation between ACTH and cortisol levels (r = 0.80, P = 0.002) and a significant correlation between cortisol and DHAS levels (r = 0.49, P = 0.01) but not between ACTH and DHAS (r = 0.04, P = 0.9).

When some samples were split and assayed for ACTH using the DSL method, the DSL values were always higher than the RSL values for the patients (by an average 6.2 pmol/L, n = 14), although mean values for the control subjects were lower (9.5 ± 4.7 SD, range 5 to 17 pmol/L compared with 16.4 ± 9.4 SD, range 8 to 33 pmol/L)(P £ 0.025). The cortisol levels on the same samples were clearly higher in the patients (P  £ 0.001). Because the patients’ samples were run in many different assays, it is difficult to attribute this discrepancy to technical error. Since most of the samples were determined only by the RSL method, only the RSL values are given.

Testosterone was measured in the 7 men; the mean pretreatment level was 15.6 ± 6.4 SD nmol/L (normal 10 to 25 nmol/L). One man (#19) had a low level of 6.8 nmol/L.

Pretreatment Relationship Between Hypercortisolemia and Nonsuppression to Dexamethasone

Although nonsuppressors tended to have higher 8:00 AM cortisol levels (846 ± 333 nmol/L, n = 9) compared with suppressors (669 ± 213, n = 11; P = 0.17), 2 psychotic patients (#17,19) with pretreatment cortisol levels exceeding the normal range had normal responses to dexamethasone.

Pretreatment DST cortisol levels in responders (315 ± 317 nmol/L, n = 13) tended to be higher than those in nonresponders (143 ± 126, n = 7; P = 0.10).

Changes in Depression With Treatment

The changes in the whole group are shown in Figure 1. A highly significant drop in the mean HDRS occurred (P £   0.0001 by ANOVA, either one-way, including all 20 subjects [F = 11.4], or by repeated measures for the 17 completers [F = 14.3]). A fall from 26.8 to 20.0 (P £ 0.05 by t-test) occurred over the first 2 weeks; a further fall to 17.5 by 2 to 4 weeks (P £ 0.01); and a maximal fall to 13.4 (50% of initial mean score) by 6 to 8 weeks (P £ 0.0001), after which the level remained steady.

In Figure 2, responders (n = 11) are compared with nonresponders (n = 4; P £ 0.0001 by ANOVA). For the responders, there was a larger fall in the first 2 weeks, from 26.6 to 18.1 (P £ 0.01 by t-test), with a further fall to 13.6 by 2 to 4 weeks (P £ 0.0001) and little change thereafter. Nonresponders showed no significant changes by ANOVA, although there was a significant fall at 1 week by t-test (P = 0.016), which was presumably a placebo effect.

The nonpsychotic unipolar responders (6/7) did better than the psychotic unipolar group (5 marked, 2 partial/8)(Figure 3) because the response in the nonpsychotic group occurred earlier (maximal 2 to 4 weeks compared with 6 to 8 weeks for the psychotic group), with a lower maximal response (10 compared with 15), which was better sustained. The 2 bipolar responders did not reach a maximum response until 5 and 8 weeks, and they quickly relapsed after treatment was stopped. One patient (#13) failed to respond to AG + MET but did respond to a second course of KC + MET.

Figure 1. Changes in all subjects with treatment. Means ± SE. ANOVA with repeated measures: P £ 0.0001 for HDRS; P £ 0.002 for DHAS; P = NS for cortisol.

Figure 2. Changes in responders (left) and nonresponders (right). Means ± SE. ANOVA with repeated measures: P £ 0.0001 for HDRS in responders; NS in nonresponders; P = 0.0005 for DHAS in responders; NS in nonresponders; P = NS for cortisol in responders; P = 0.03 in nonresponders.

Figure 3.

Changes in unipolar nonpsychotic responders (left) (n = 6) and unipolar psychotic responders (n = 5): Means ± SE. ANOVA with repeated measures: P £ 0.0001 for HDRS in nonpsychotic; P = 0.01 for psychotic; P = 0.004 for DHAS in both groups; P = NS for cortisol in both groups. For HDRS, # 13 is omitted; for DHAS, # 14 is omitted.

Changes in Hormonal Levels With Treatment

Mean cortisol levels did not vary significantly in the group as a whole, nor in the responders. Levels tended to rise in the nonresponders (P = 0.03). In all groups, there were large fluctuations, which often did not fall consistently until late in treatment or even in the few weeks following treatment. Patients who had prolonged remissions of more than 1 year off treatment, however, showed a consistent and significant fall to normal levels (from 758 ± 156 nmol/L to 507 ± 102, P = 0.009) (Figure 4). The level in the psychotic responders also gradually fell from 783 ± 302 to 600 ± 220, but this fall was not significant. Of the more than 300 8:00 AM cortisol levels determined during the treatment period, only 1 patient (#3), who remained asymptomatic, had values [79,80,107] below normal.

Figure 4. Change in 8:00 AM cortisol levels in 5 responders with prolonged remissions. ANOVA by repeated measures: P = 0.0016.

In the whole group, DHAS levels fell (P £ 0.002), and there was a significant correlation (r = 0.94, P = 0.02) between DHAS and HDRS. This drop was slightly more marked in the responders (P = 0.0005), and the correlation with HDRS was similar (r = 0.95, P = 0.01), but it failed to reach significance in the nonresponders (P = 0.06), although 1 patient (#7) had a very marked response. In the unipolar groups (psychotic and nonpsychotic), the drop was also significant (P = 0.004); however the lowest point was reached earlier in the nonpsychotic group (1 to 2 weeks) than in the psychotic group (5 weeks); in these small groups, the correlation with HDRS was of borderline significance (r = 0.86, P = 0.06 psychotic;  r = 0.80, P = 0.10 nonpsychotic). One of the 2 bipolar responders had a low value of DHAS, which fell even lower with the treatment, while the other responder had an erratic response. Of particular interest was patient #13, who failed to respond clinically to AG (when the DHAS level was only transiently decreased) but did respond to ketoconazole when her DHAS fell markedly. Her cortisol levels and her DST reverted to normal.

Adrenocorticotropin levels varied widely, rising to levels over 50 pmol/L in 2 patients (#4,20) but only slightly in most. Using individual patient means, levels during treatment had a group mean of 28.7 ± 49.7 SD pmol/L, the very large SD was due to the much higher levels in patients #4 and #20. After conversion to logarithms, the paired t-test was of borderline significance (P = 0.045). Excluding these 2 patients, the level rose an average of 3.29 pmol/L (P = 0.07).

Mean testosterone levels in the 4 men receiving ketoconazole fell from 15.6 nmol/L to 11.2 nmol/L during the first 4 weeks and to 5.1 nmol/L during the latter 4 weeks of treatment (P = 0.05). There was a prompt rise after stopping treatment, however, to a mean of 23.5 ± 6.7 nmol/L. Levels in the 3 men receiving AG did not change significantly.

Changes in DSTs With Treatment

Three patients refused the final DST: 2 nonresponders and 1 drop-out; all 3 were suppressors. Five of 6 responders who were nonsuppressor before treatment had reverted to normal after treatment. Including the 1 patient who responded clinically but remained nonsuppressor (#4), the mean level fell significantly (P = 0.006) from 600 ± 100 SE nmol/L to 165 ± 77 SE nmol/L. All pretreatment suppressors remained suppressor, while nonresponding nonsuppressors remained nonsuppressor.

Discussion

It is estimated that 10% to 15% of patients given antidepressants drop out of medication trials in the first 3 weeks (16). Our completion rate of 17 out of 20 patients (85%) after 8 weeks, therefore, compares favourably.

The incidences of pretreatment hormonal abnormalities, 55% hypercortisolemia and 50% nonsuppressors, were similar to those of other studies (1). There was a 25% incidence of low DHAS levels, but we were unable to find any comparable data in depressed patients. Osran and others reported a loss of diurnal variation in DHA (the unsulfated form of DHAS) in depression, and Reus and others found no differences in 4:00 PM DHAS levels in depressed patients compared with controls (17,18).

Mean pretreatment ACTH levels were lower but not significantly different from control levels, although mean cortisol levels were clearly higher (P = 0.03). The discrepancy in the pretreatment results of the 2 ACTH methods supports the possible presence of a factor similar to, but not identical with, ACTH, which crossreacts in the DSL but not the RSL assay. Both assays give higher values than those obtained using the Nichols Institute assay (normal range 2 to 11 pmol/L). Such findings are in accord with those of other studies (19–21).

Our response rate of 65% (76% including the 2 partial responders) compares well with the 50% response rate obtained in refractory patients by electroconvulsive therapy (ECT) and by adjunctive lithium (16). We have previously noted that there was no significant difference in efficacy among the medications used; however, this study was not designed to detect such a difference (7).

In this study, it appeared that improvement in mood occurred well before the cortisol levels normalized in the hypercortisolemic patients. Because the 8:00 AM cortisols were taken approximately 10 hours after the evening dose of medication, and because these drugs are short-acting, this is not surprising; the feedback mechanism stimulates ACTH to maintain the cortisol levels (22–24). Wolkowitz and others, who measured cortisol levels at 4:00 PM in hypercortisolemic depressed patients treated with KC, noted a fall of 26% in the cortisol levels but no significant correlation with the HDRS ratings (P = 0.12) (8). Santen and others found a linear relationship (r = 0.79, P £ 0.001) between the percentage of suppression of urinary free cortisol and the percentage of suppression of DHAS in patients with breast cancer receiving AG (14). They found it necessary to give 40 mg daily of cortisol along with AG to reduce ACTH levels to low normal, which was more than the usual 20 to 30 mg daily replacement dose.

The administration of cortisol prevents any acute drop of cortisol, which is most likely to occur during the first few days. Since aldosterone synthesis is also affected, it may be necessary to give 9 a-fluorocortisol if electrolyte changes or hypotension occur. In our study, subnormal cortisol levels were seen in only 1 patient who was asymptomatic, and electrolytes remained normal throughout, except in 2 patients whose serum sodium dropped slightly but who did not become hypotensive. In recent reports of favorable effects of steroid biosynthetic inhibitors in depressed patients, most investigators have given only KC without cortisol. O’Dwyer and others, however, gave 30 mg daily of cortisol (7.5 mg 4 times daily) along with MET with good results (9). Therefore, particularly in an outpatient situation, it seems prudent to give cortisol along with these medications at least in the initial few weeks of treatment.

Dehydroepiandrosterone sulfate levels, in contrast to cortisol levels, fell consistently in most subjects receiving KC and AG, at least in the early weeks of treatment, and served as an indicator of compliance and, to some extent, of efficacy because there was a significant correlation with HDRS. Collection of urine for free cortisol was attempted but abandoned because most of these patients were too severely affected to cooperate. Changes in both HDRS and DHAS were most rapid in the unipolar responders, with the DHAS reaching its lowest point after 1 week and the HDRS at 3 weeks. They fell more slowly in the psychotic group, with the DHAS being lowest at 5 weeks and the HDRS at 7 weeks. No significant change in DHAS levels occurred in the nonresponders. All but one of the responders had a marked drop in DHAS level. Possibly the blockade was insufficient in those patients with a small drop or no drop. It should be noted that the fall in DHAS occurs with AG and KC but not with MET. The change in DHAS levels with AG and KC served to confirm that the medications were affecting the steroid biosynthetic pathway. Although the DHAS levels in some patients began to rise before the end of treatment, this rise did not seem to affect the clinical remissions that had already occurred. This loss of blockade after some weeks or months is commonly seen with these medications and poses a problem in the medical treatment of Cushing’s syndrome (24).

Although there was a significant drop in testosterone levels in men receiving KC, this was rapidly reversed in the first 2 weeks after stopping treatment, as has been observed by others (22).

The change in the DST response was highly significant (P = 0.006), with reversion of 5 out of 6 nonsuppressors to suppressors. The one patient who had a clinical response and remained nonsuppressor relapsed after 8 months.

The actions on steroidogenesis of the drugs used here are not confined to the adrenal gland but also affect the testes, ovaries, and brain. In particular, KC lowers testosterone levels, as seen here, while AG lowers estradiol levels (15,22,25). Hirsutism can be a troublesome side effect of MET because of an increase in adrenal androgens (24). Aminoglutethimide and KC also have poorly understood effects on steroid catabolism, so that urinary steroid metabolite excretion is disproportionately low compared with plasma cortisol levels (15).

Another possible approach to steroid inhibition in depression is the use of glucocorticoid receptor antagonists such as RU 486 (26). In addition to its effects on steroid metabolism, KC acts as a glucocorticoid receptor antagonist. In patient #13, KC appeared to be more effective than AG; a low dose of MET was added only late in both courses of treatment.

The available studies of antiglucocorticoid therapies in patients with major depression suggest that they may be effective in inducing remissions in major depression as well as in treating Cushing’s syndrome (27). Since Cushing’s syndrome has many features in common with major depression and may be intermittent, the possibility has been entertained that major depression may possibly represent a variant of Cushing’s syndrome. At times, it may be difficult to distinguish between the two (19). The case report of Ravaris and others also suggests this situation (28).

It is desirable to confirm these findings in a double-blind, placebo-controlled trial. For this we suggest starting with a small dose of KC and gradually increasing the dosage according to the patient’s HDRS response and tolerance. If the study is done on inpatients under close supervision, cortisol can be omitted. If it is done on outpatients, we prefer to give the small dose of cortisol at bedtime to ensure that serious adrenal insufficiency does not occur. Dehydroepiandrosterone sulfate levels should be done weekly as an index of blockade retrospectively, but they need not be available to the treating physician. Cortisol levels are not clinically useful. Electrolytes and liver function tests should be done weekly for the first few weeks. The treatment period of 8 weeks seems to be appropriate but might be extended.

References

1. Murphy BEP. Steroids and depression. Journal of Steroid Biochemistry and  Molecular  Biology 1991;38:537–58.

2. Starkman MN. The HPA axis and psychopathology: Cushing’s syndrome. Psychiatric Annals 1993;23:691–701.

3. Reus VI, Wolkowitz WM. Behavioral side effects of corticosteroid therapy. Psychiatric Annals 1993;23:703–8.

4. Arana GW, Santos AB, Laraia MT, McLeod-Bryant S, Beale MD, Rames LJ, and others. Dexamethasone for the treatment of depression: a randomized, placebo-controlled, double-blind trial. Am J Psychiatry 1995;152:265–7.

5. Murphy BEP.Treatment of major depression with steroid suppressive therapy. J Steroid Biochem Mol Biol 1991;39:239–44.

6. Murphy BEP, Dhar V, Ghadirian AM, Chouinard G, Keller R. Response to steroid suppression in major depression resistant to antidepressant therapy. J Clin Psychopharmacol 1991;11:121–6.

7. Ghadirian AM, Engelsmann F, Dhar V, Filipini D, Keller R, Chouinard G, and others. The psychotropic effects of inhibitors of steroid biosynthesis in depressed patients refractory to treatment. Journal of Biological Psychiatry 1995;37:369–75.

8. Wolkowitz OM, Reus VI, Manfredi F, Ingbar J, Brizendine L, Weingartner H. Ketoconazole administration in hypercortisolemic depression. Am J Psychiatry 1993;150:810–2.

9. O’Dwyer A-M, Lightman SL, Marks MKN, Checkley SA. Treatment of major depression with metyrapone and hydrocortisone. J Affect Disord 1995;33:123–8.

10. Anand A, Malison R, McDougle CJ, Price LH. Antiglucocorticoid treatment of refractory depression with ketoconazole: a case report. Biol Psychiatry 1995;37:338–40.

11. Thakore JH, Dinan T. Cortisol synthesis inhibition: a new treatment strategy for the clinical and endocrine manifestations of depression. Biol Psychiatry 1995;37:364–8.

12. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 3rd ed. Revised. Washington (DC): American Psychiatric Association; 1987.

13. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960;23:55–62.

14. Santen RJ, Wells SA, Runic S, Gupta C, Kendall J, Rudy ED, and others. Adrenal suppression with aminoglutethimide: part I. Differential metabolism as a rationale for use of hydrocortisone. J Clin Endocrinol Metab 1977;45:469–79.

15. Santen RJ, Wells SA. The use of aminoglutethimide in the treatment of patients with metastatic carcinoma of the breast. Cancer 1980;46:1066–74.

16. American Psychiatric Association. Practice guideline for major depressive disorder in adults. Am J Psychiatry 1993;150(Suppl):1S–26S.

17. Osran H, Reist C, Chen C-C, Lifrak ET, Chicz-Demet A, Parker LN. Adrenal androgens and cortisol in major depression. Am J Psychiatry 1993;150:806–9.

18. Reus VI, Wolkowitz OM, Roberts E, Chan T, Turetsky N, Manfredi F, and others. Dehydroepiandrosterone (DHEA) and memory in depressed patients. Neuropsychopharmacology 1993;9(Suppl):66S.

19. Starkman MN, Schteingart DE, Schork MA. Discordant changes in plasma ACTH and  -lipotropin/-endorphin levels in Cushing’s disease patients with depression. Psychoneuroendocrinology 1992;17:619–26.

20. Goverde JM, Martens GH, Pesman GH, Smals GH. Multiple forms of bioactive and immunoreactive adrenocorticotropin in human pituitary and blood of patients with Nelson’s syndrome. J Clin Endocrinol Metab 1993;77:443–7.

21. Gibson S, Crosby SR, Stewart MF, Jennings AM, McCall E, White A. Differential release of proopiomelanocortin-derived peptides from the human pituitary: evidence from a panel of two-site immunoradiometric assays. J Clin Endocrinol Metab 1994;78:835–41.

22. Trachtenberg J, Zadra J. Steroid synthesis inhibition by ketoconazole: sites of action. Clin Invest Med 1988;11:1–5.

23. Martikainen H, Heikkinen J, Ruokonen A, Kauppila A. Hormonal and clinical effects of ketoconazole in hirsute women. J Clin Endocrinol Metab 1988;66:987–91.

24. Verhelst JA, Trainer PJ, Howlett TA, Perry L, Rees LH, Grossman AB, and others. Short and long-term responses to metyrapone in the medical management of 91 patients with Cushing’s syndrome. J Clin Endocrinol Metab 1991;35:169–78.

25. Pont A, Graybill JR, Craven PC, Galgiani JN, Dismukes WE, Reitz RE, and others. High-dose ketoconazole therapy and adrenal and testicular function in humans. Arch Intern Med 1984;144:2150–3.

26. Murphy BEP, Filipini D, Ghadirian AM. Possible use of glucocorticoid antagonists in the treatment of major depression: preliminary results using RU 486. J Psychiatry Neurosci 1993;18:209–13.

27. Murphy BEP. Antiglucocorticoid therapies in major depression: a review. Psychoneuroendocrinology 1997;22(1 Suppl):125S–132S.

28. Ravaris CL, Sateia MJ, Beroza W, Noorday DL, Brinck-Johnsen T. Effect of ketoconazole on a hypophysectomized, hypercortisolemic, psychotically depressed woman. Arch Gen Psychiatry 1988;45:966–7.

Résumé

Objectif : Les patients qui souffrent de dépression majeure présentent souvent des taux élevés de cortisol ainsi qu’une résistance à la dexaméthasone. Nous avons cherché à connaître l’effet des inhibiteurs de la biosynthèse des stéroïdes sur la dépression majeure, et étudié les changements endocriniens produits.

Méthode : Après une période de sevrage, 20 patients résistant au traitement et souffrant de dépression grave ont reçu de l’aminoglutéthimide, de la méryrapone et/ou de la kétoconazole, de même qu’une faible dose de cortisol pendant 8 semaines. Chaque semaine ou plus souvent, on a suivi, à l’aide de l’échelle de dépression de Hamilton (HAMD), les niveaux de cortisone, de sulfate de déhydroépiandrostérone (DHAS), de corticotrophine (ACTH) et de testostérone, prélevés à huit heures du matin. Un test de freinage de la dexaméthasone (DST) a été administré avant et après le traitement.

Résultats : Dix-sept patients (85 %) ont achevé la série de traitement, et on a enregistré une baisse moyenne significative (P £ 0,0001 de 50 %) des résultats à l’échelle HAMD, à la 7 ième semaine de traitement. Les niveaux de cortisol ont considérablement fluctué et demeuraient souvent élevés, même après une amélioration de l’état clinique du patient. Les niveaux de sulfate de déhydroépiandrostérone ont baissé plus uniformément et se sont révélés un bon indicateur de conformité et, jusqu’à un certain point, d’efficacité avec le traitement à l’aminoglutéthimide et à la kétoconazole. La corrélation entre le DHAS et l’HAMD ( r = 0,94) était significative (P = 0,02). Chez les hommes, les niveaux de testostérone ont baissé à cause de la kétoconazole, mais sont rapidement revenus à la normale au terme du traitement. Les niveaux de corticotrophine étaient normaux ou élevés, selon le dosage, et ont augmenté (P = 0,07 ; n = 13) chez la plupart des sujets durant le traitement. Des 6 répondeurs à qui l’on avait administré des DST non- suppresseurs avant le début du traitement, 5 sont revenus à des niveaux normaux 1 ou 2 semaines après la fin du traitement (P = 0,0006).

Conclusion : Il se peut que le métabolisme anormal des stéroïdes corticoïdes perpétue la dépression. La modification de leur synthèse ou de leur métabolisme peut occasionner la rémission.

Manuscript received May 1997, revised and accepted August 1997.

¹Professor of Medicine, Obstetrics and Gynecology, and Psychiatry, McGill University; Senior Physician, Montreal General Hospital; Director, Psychoendocrinology Unit, McGill Research and Training Building; Montreal, Quebec.

²Professor, Department of Psychiatry, McGill University, Montreal, Quebec.

³Research Assistant, Psychoendocrinology Unit, McGill University, Montreal, Quebec; presently Private Practitioner, Monroeville, PA.

Address for correspondence: Dr BE Pearson Murphy, Montreal General Hospital, 1650 Cedar Avenue, Montreal, QC H3G 1A4
email: md98@musica.mcgill.ca