Context: The comparison of variability, reproducibility, and diagnostic performance of late-night salivary cortisol (LNSF) and urinary free cortisol (UFC) using concurrent and consecutive samples in Cushing's syndrome (CS) is lacking. Objectives, Patients, and Methods: In a prospective study, we evaluated 3 simultaneous and consecutive samples of LNSF by RIA and UFC by liquid chromatography associated with tandem mass spectrometry in Cushing's disease (CD) patients (n = 43), adrenal CS patients (n = 9), and obese subjects (n = 18) to compare their diagnostic performances. In CS patients, we also performed a modified CS severity index.
Results: There was no difference in the coefficient of variation (percentage) between LNSF and UFC among the 3 samples obtained for each patient with Cushing's disease (35 ± 26 vs 31 ± 24), adrenal CS (28 ± 14 vs 22 ± 14), and obesity (39 ± 37 vs 48 ± 20). LNSF confirmed the diagnosis of hypercortisolism even in the presence of normal UFC in 17.3% of CS, whereas the inverse situation was not observed for UFC. The area under the receiver-operating characteristic curves for LNSF was 0.999 (95% credible interval [CI] 0.990-1.00) and for UFC was 0.928 (95% CI 0.809-0.987). The ratio between areas under the curve was 0.928 (95% CI 0.810-0.988), indicating better performance of LNSF than UFC in diagnosing CS. There was no association between the CS severity index and the degree of biochemical hypercortisolism.
Conclusion: Our data show that despite similar variability between both methods, LNSF has a superior diagnostic performance than UFC and should be used as the primary biochemical diagnostic test for CS diagnosis.