Endocrine Abstracts

Introduction: The diagnosis of Cushing’s syndrome (CS) is often challenging and requires the use of several diagnostic methods. Late-night salivary cortisol (LNSC) offers an easy, non-invasive screening method for CS. However, its availability is still limited and the cut-off values vary widely between laboratories. The aim of this study was to assess the diagnostic value of LNSC for CS in comparison with late-night serum cortisol (LNSerC) and 24-hour urinary free cortisol (UFC). We also aimed to verify the accuracy of manufacturer-provided cut-off value for LNSC.

Methods: Patients with suspected CS hospitalized in our department were retrospectively reviewed. Saliva was collected at 11 p.m. using a Salivette. At the same time blood samples for the determination of serum cortisol were obtained. Additionally, 24-hour urine collection for free cortisol assessment was performed. Salivary cortisol was determined using automated electrochemiluminescence assay - Elecsys Cortisol II (Roche Diagnostics). Upper reference limit for LNSC provided by the manufacturer was 11,3 nmol/l.

Results: A total of 69 patients were included in the study. 19 patients were diagnosed with endogenous CS (pituitary: 10, ectopic: 2, adrenal: 7), among them 2 were already treated and presented normal cortisol values. CS was excluded in the remaining 50 patients. 86% of the cohort were women. The mean age was 45,9, and mean BMI was 29.9 kg/m². Obesity was present in 40% of patients. Cortisol concentrations assessed with all analysed methods were significantly higher in patients with active CS compared to patients in whom CS was excluded: salivary cortisol (15 nmol/l vs 2.985 nmol/l, P<0.001), serum cortisol (523 nmol/l vs 170 nmol/l, P<0.001), and UFC (594.5 nmol/24 h vs 130 nmol/24 h, P<0.001). LNSC was not related to sex, BMI, smoking status or oral contraceptive (OC) use (however only 4 patients were using OC). Significant correlations between LNSC and LNSerC (P<0.001; r=0.832), and UFC (P<0.001; r=0.526) were identified. ROC curves comparison demonstrated no superiority of any of the analysed methods in the diagnosis of CS (LNSC vs LNSerC P=0.905; LNSC vs UFC P=0.619; LNSerC vs UFC P=0.623). AUC for LNSC was: 0.93043, P<0.001, for LNSerC: 0.933258, P<0.001, and for UFC: 0.901563, P<0.001. The cut-off point 11.3 nmol/l for LNSC provided sensitivity of 88.24% and specificity of 88.46%.

Conclusions: LNSC is a valuable method in the diagnosis of CS providing the accuracy comparable to LNSerC and UFC. Cut-off value provided by the manufacturer offers high accuracy in the diagnosis of CS.