Objective: It is presently unclear whether the accuracy of midnight serum cortisol (F24) in the diagnosis of Cushing's syndrome (CS) may be replicated under usual conditions of clinical care. The aim of the present study was to assess retrospectively the effectiveness of F24 for confirming the diagnosis in a consecutive series of 106 patients, in 78 of whom a definitive diagnosis of CS was made.
Design and methods: We have compared the results of F24, urinary free cortisol (UFC) and the overnight 1 mg dexamethasone suppression test (DST) with the definitive clinical diagnosis. Receiver operating characteristic (ROC) analysis has been performed to define the best cutoff values, the sensitivity (Se) and the specificity (Sp) of the tests.
Results: The best cutoff value for F24 was 8.3 microg/dl (Se 91.8%; Sp 96.4%). The best cutoff value for the DST was 4.0 microg/dl (Se 89.2%; Sp 90.9%). The best cutoff value for UFC was 238 microg/24 h (Se 73.2%; Sp 96.3%). The area under the curve of F24 was significantly greater than that of UFC, both in the overall series (P = 0.004) and in the subgroup of patients with mild CS (P = 0.02). The differences were analyzed by means of the two-tailed students's t-test. With the thresholds generated by the ROC analysis, UFC would have failed to achieve the correct diagnosis in a significantly higher percentage of cases than F24 (20.4% vs 7.9%; P = 0.01). The difference was analyzed by means of the chi-squared test with Yates correction.
Conclusions: The present results show that F24 has excellent effectiveness in the diagnostic procedures for CS in stressed conditions (patients studied in a hospital ward in a nonsleeping state). The test appears to be accurate also for patients with mild hypercortisolism.