Cortisol Feedback During the HPA Quiescent Period in Patients With Major Depression

Subjects were recruited by advertisement and from outpatient services at Stanford University Medical Center, McLean Hospital, and Washington University School of Medicine. Screening included a psychiatric interview, administration of the Structured Clinical Interview for DSM-III-R (SCID) (7), clinical ratings, and a medical evaluation with a physical examination. Ratings were conducted by staff who had demonstrated adequate interrater reliability; kappa exceeded 0.85 for depression items on the SCID and the Hamilton Depression Rating Scale (8). All subjects were required to have no active medical problems, no weight loss of 5 lb or greater in the previous month, and no use of any medications for 2 weeks. Major depressive disorder patients were also required to have: 1) a diagnosis of unipolar major depressive disorder in a current depressive episode without psychotic features, 2) a minimum score of 21 on the 21-item Hamilton depression scale, 3) a minimum score of 7 on the Thase Core Endogenomorphic Scale (9), 4) no use of fluoxetine within the past 6 weeks, 5) no use of electroconvulsive therapy within the past 6 months, and 6) no drug or alcohol abuse within the past 6 months. Comparison subjects were also required to have no history of psychiatric disorders and no first-degree relatives with axis I disorders, according to the Research Diagnostic Criteria family history assessment (10). Standard dietary restrictions, described previously (11), were observed. After complete description of the study to the subjects, written informed consent was obtained.

The protocol was conducted in General Clinical Research Centers at Stanford University Medical Center, Brigham and Women’s Hospital, and Washington University School of Medicine. Subjects had an intravenous line placed in each arm. Baseline HPA axis hormone levels were monitored for 24 hours; the results of this baseline study have been reported elsewhere (11). Subsequently, at 7:00 p.m., blood was drawn through an intravenous line to measure cortisol and ACTH levels. The subjects were then randomly assigned to receive 15 mg of cortisol (hydrocortisone sodium succinate) or placebo, infused over 2 hours beginning immediately after the 7:00 p.m. baseline blood draw. Cortisol and placebo were administered by using a double-blind procedure. Every 30 minutes from 7:30 p.m. to 11:00 p.m., blood was drawn through the intravenous line not used for the infusion to measure cortisol and ACTH levels. Some subjects were randomly assigned to receive CRF instead of cortisol or placebo, and these results are not reported here. Cortisol levels were determined by radioimmunoassay (12) and ACTH levels by immunoradiometric assay (12) in the Endocrinology Laboratory at Brigham and Women’s Hospital.

For each subject individually, ACTH levels from 7:00 p.m. to 11:00 p.m. were modeled, by using a least squares procedure, as having a linear decline, an assumption based on the known diurnal variation of HPA axis activity (13). The ACTH response to administration of cortisol or placebo was represented by the slope of the modeled linear decrease. Values of the ACTH slope were subjected to a two-way analysis of covariance (ANCOVA), in which diagnosis (major depressive disorder versus no psychiatric disorder) and treatment (cortisol versus placebo) were factors and the 7:00 p.m. baseline ACTH level, sex, and age were covariates. The hypothesis that patients with major depressive disorder have an abnormal ACTH response to cortisol was tested by examining the effect of the interaction of diagnosis and treatment.

The subjects assigned to receive placebo included 18 depressed patients (11 women and seven men) and 14 comparison subjects (four women and 10 men). Those assigned to receive cortisol included 11 patients (six women and five men) and 11 comparison subjects (six women and five men). The mean ages for the placebo subjects were 40.0 years (SD=10.2) for the depressed patients and 39.7 years (SD=12.7) for the comparison subjects. The mean ages for the subjects who received cortisol were 52.6 years (SD=15.4) for the depressed patients and 40.9 years (SD=13.4) for the comparison subjects. The mean Hamilton depression scale scores were 25.6 (SD=4.3) for the depressed patients who received placebo and 24.6 (SD=3.9) for the depressed patients who received cortisol. Subjects’ weight and the area under the curve for cortisol over time after administration of cortisol did not differ significantly between groups. Infused cortisol increased serum cortisol levels to maximum values having a mean of 24.5 μg/dl (SD=5.7) in the patients and 22.5 μg/dl (SD=10.3) in the comparison subjects.

As Figure 1 shows, cortisol produced a substantial and comparable decrease in ACTH levels in both the patients and the comparison subjects, while there was relatively less change in ACTH over time in the placebo groups. Baseline ACTH levels were higher for both groups receiving placebo than for the groups receiving cortisol, reflecting considerable interindividual variability in ACTH secretion. The fit of the linear model was represented by mean R² values of 63.2% (SD=26.5%) for the comparison subjects who received cortisol, 64.7% (SD=21.6%) for the depressed patients who received cortisol, 32.2% (SD=29.1%) for the comparison subjects who received placebo, and 31.7% (SD=30.9%) for the depressed patients who received placebo. Mean values for ACTH slope were –0.52 pg/ml per hour (SD=1.82) for the comparison subjects who received placebo, –1.28 pg/ml per hour (SD=1.69) for the depressed patients who received placebo, –1.91 pg/ml per hour (SD=1.54) for the comparison subjects who received cortisol, and –2.14 pg/ml per hour (SD=1.27) for the depressed patients who received cortisol. Analysis of the ACTH slope revealed a highly significant effect of treatment (F=33.3, df=1, 47, p<0.001), but no significant effects of diagnosis (F=0.4, df=1, 47, p=0.50) or of the interaction of diagnosis and treatment (F=0.7, df=1, 47, p=0.40).

After administration of placebo, the ACTH decline appeared steeper in the patients than in the comparison subjects. However, this difference reflected the greater proportion of women in the group of depressed patients who received placebo, compared with the comparison subjects who received placebo, since the ACTH slope was steeper for women than for men (F=8.5, df=1, 47, p<0.01). When the ACTH slopes of patients and comparison subjects receiving placebo were compared by using ANCOVA with the same covariates as in the principal analysis, the group differences were not significant (F=0.1, df=1, 27, p=0.80).

To our knowledge, this is the first study of cortisol feedback mechanisms in depressed patients studied during the quiescent period of HPA axis activity. No differences in ACTH response to cortisol between patients with major depressive disorder and healthy comparison subjects were found. Power was computed for the statistical model employed in this study by using Cohen’s method (14) and by assuming a relatively large effect size, indicated by f=0.40. With this approach, the test of the effect of the diagnosis-by-treatment interaction on ACTH slope had a power of 0.79, close to the usual standard of 0.80. Thus, our study provides some evidence that patients with unipolar, nonpsychotic major depressive disorder do not have a defect in cortisol feedback evaluated during evening hours.

Some limitations of this study should be noted. A single dose of cortisol may not be sensitive enough for detecting group differences in the feedback response, and a dose-response study design would have advantages. Although studying cortisol feedback during the HPA axis quiescent period may be valuable, the relationship of our findings to those of other investigators is not clear because of the absence of an additional condition in which feedback is assessed at the time of peak activity. Finally, research with a very limited number of subjects with endocrine disorders has suggested that there may be a distinct phase of feedback occurring over less than 15 minutes (15). Thus, 30-minute sampling of ACTH levels may not be frequent enough to uncover abnormalities in feedback.

Received Oct. 18, 2000; revision received April 17, 2001; accepted May 10, 2001. From the Department of Psychiatry, Washington University School of Medicine and Metropolitan St. Louis Psychiatric Center, St. Louis, Mo.; the Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, Calif.; and the Division of Endocrinology/Hypertension, Brigham and Women’s Hospital and Harvard Medical School, Boston. Address reprint requests to Dr. Posener, Department of Psychiatry, Washington University School of Medicine, 660 S. Euclid Ave., Box 8134, St. Louis, MO 63110; [email protected] (e-mail).Supported by NIMH grant MH-50604 (Dr. Schatzberg), a National Alliance for Research on Schizophrenia and Depression Young Investigator Award (Dr. Posener), and NIH General Clinical Research Center grants RR-00070, RR-00036, and RR-02635.The authors thank Christine M. Blasey, Ph.D., for statistical consultation.