Screening for Cushing's syndrome in obese women with and without polycystic ovary syndrome.

Obesity and insulin resistance, menstrual abnormalities and clinical and biochemical signs of hyperandrogenism are common features in women with polycystic ovary syndrome (PCOS) and Cushing's syndrome (CS). Further, an overdrive of the pituitary-adrenal axis has been documented in PCOS and this condition is often present in women with CS. For this reason, screening for hypercortisolism is often needed in obese women with polycystic ovaries. The aim of this study was to compare the diagnostic value of different screening tests for CS in a population of obese premenopausal women with PCOS and without PCOS (OB) and in a group of patients with CS. We reviewed retrospectively the case records of 117 obese women of reproductive age (60 PCOS and 57 OB, BMI 25.1-70.1, 13-45 yr) who were screened for CS at our Institution in the years 1995-2001 and turned out to be free of the disease. Data were compared with those of 58 premenopausal obese women with active CS (BMI 25.1-50.2 kg/m2, 18-45 yr). Screening for CS was performed by urinary free cortisol (UFC) (three consecutive 24-h urine collections), cortisol circadian rhythm (blood samples taken at 08:00-17:00-24:00 h), and 1 mg overnight dexamethasone suppression test (DST). A 24:00 h plasma cortisol (MNC) of 207 nmol/l, a UFC of 221 nmol/day and plasma cortisol after DST of 50 nmol/l and 138 nmol/l were taken as cut-off values for the diagnosis of CS. As expected, patients with CS showed elevated basal and post-dexamethasone plasma cortisol and UFC levels (p < 0.001 vs OB and PCOS). PCOS had higher UFC (p < 0.005) but not MNC and post-DST plasma cortisol levels compared to OB. DST showed the greatest specificity and diagnostic accuracy in differentiating CS from PCOS and OB (both p < 0.05 vs MNC and UFC, according to the 138 nmol/l criterion) while MNC and UFC displayed a similar discriminatory value. However, by using a lower threshold (50 nmol/l) as response criterion, there were no diagnostic differences between DST and the other tests. Specificity and diagnostic accuracy of UFC measurement was lower in PCOS than in OB (both p < 0.05) whilst there were no differences between groups for DST and MNC. Similarly, the area under the ROC curve relative to DST, giving an estimate of the inherent diagnostic accuracy of the test, was slightly greater than those of MNC and UFC (z = 0.694 and z = 0.833 for DST vs MNC and UFC, respectively, both p = NS). These results indicate that the 1-mg DST and MNC are unaffected by the presence of PCOS and can be safely used to screen for CS premenopausal obese women with PCOS, while caution should be exercised in interpreting mildly elevated UFC levels in these patients.