RESUMEN
This study estimates diagnostic performance of late-night salivary cortisol (LNSC) as measured by automated electrochemiluminescence immunoassay (ECLIA), evaluates the clinical implication of two consecutive LNSC measurements, and compares its accuracy with enzyme-linked immunosorbent assay (ELISA) and serum cortisol after low-dose dexamethasone suppression test (DST) in obese and overweight patients referred for suspected Cushing's syndrome (CS). One hundred twenty three consecutive obese and overweight referred patients and 98 healthy volunteers provided two saliva samples collected at 23:00 using a Salivette (Sarstedt, Germany), assayed by ECLIA (Cobas e601) and ELISA. The patients underwent DST and were further evaluated until CS was pathologically confirmed (n = 45) or excluded. Diagnostic performance of LNSC was evaluated by receiver operating characteristic (ROC) analysis. The total areas under the curve (AUC) were calculated to compare the different tests. We found that a cut-off value of 9.4 nmol/l can differentiate CS among obese and overweight patients with sensitivity of 84.4 % (95% CI 71.2-92.2), specificity of 92.3 % (95% CI 84.2-96.4), and diagnostic odds ratio of 65.1 (95% CI 20.4-207.6). No difference was found between AUCs from the first, second, and the mean from the two LNSC measurements (ECLIA), LNSC (ELISA), or DST. The single LNSC (ECLIA) and DST improved the sensitivity and specificity for concordant results up to 100 and 97.4 %, respectively. In conclusion, due to its automation and its comparable diagnostic performance, ECLIA is preferable as a first-line LNSC screening test for CS. The initial use of single LNSC followed by DST provides better diagnostic performance for concordant results.
