Subjects in the MD-AF group were 8 consecutive female outpatients presenting to the Depression Clinic of the Clarke Division of the Centre for Addiction and Mental Health who met the following DSM-IV criteria for MD-AF: mood reactivity and at least 2 of increased appetite or weight gain, hypersomnia, leaden paralysis or interpersonal rejection sensitivity.1
Eleven normal controls were recruited via posters and newspaper advertisements at the University of Toronto. They had no history of psychiatric illness and were matched as closely as possible to the depressed patients on age and body mass index (BMI).
None of the subjects were pregnant, and all had regular menstrual cycles in the 3 months before the study. Menstrual phase was documented by self-report and was defined as: day 0–5, menstrual; day 5–14, follicular; and day 14 to menses, luteal. To control for possible effects of menstrual cycle on cortisol measures, all subjects were tested during the follicular phase. Subjects were excluded if they were medically ill, taking corticosteroids or actively abusing substances. None of the study subjects were taking antidepressants at the time of the study.
Each subject was given an oral and written summary of the purposes, procedures and potential risks of the project and each gave informed consent. Ethics approval was obtained from the University of Toronto.
As it was not known which dose of dexamethasone would be optimal to demonstrate differences in cortisol suppression, each subject was challenged twice — once with 0.25 mg and once with 0.5 mg of dexamethasone, in random order.
Before undergoing the first dexamethasone challenge, subjects in the atypical depression group were administered the 29-item Structured Interview Guide for the Hamilton Depression Rating Scale (SIGH-SAD).13
This version of the HDRS includes an 8-item subscale to assess atypical symptoms of depression.
On the day of each challenge, a baseline plasma cortisol sample was drawn by venipuncture at 8 am. Dexamethasone was self-administered by the study subjects at 11 pm, and postchallenge plasma cortisol samples were drawn at both 8 am and 3 pm the next day. The morning sample was taken before breakfast to avoid the confounding effect of food intake on plasma cortisol levels. The second dexamethasone challenge was completed 1 week after the first.
All blood samples were drawn at the hospital's clinical laboratory. To standardize blood drawing, subjects arrived 15 minutes before each procedure. Plasma cortisol levels were measured via radioimmunoassay by a technician blind to the nature of the study.
Demographic variables of the 2 groups were compared with unpaired t-tests. Across all subjects, the 2 prechallenge plasma cortisol levels taken 1 week apart were highly correlated (r = 0.72, p < 0.01). To assess the relation of baseline cortisol levels to key demographic and clinical variables, we calculated a mean overall baseline cortisol level (i.e., mean CORT-0) for each subject by averaging the 2 prechallenge values. The relation of mean CORT-0 to key demographic and clinical variables was then assessed using Pearson correlations.
Because of the small sample size and high degree of variability in plasma cortisol levels, nonparametric statistics (i.e., Mann-Whitney U tests) were used to compare cortisol levels and cortisol percent change scores across the 2 study groups. Individual prechallenge baseline measures (not the mean CORT-0 described above) were used for these analyses. Percent change scores were calculated for each postchallenge time point as: ([postchallenge value – prechallenge value]/[prechallenge value]) х 100. Where applicable, correlations between percent change scores and other study variables were assessed.