Late-Night Salivary Cortisol: Cut-Off Definition and Diagnostic Accuracy for Cushing’s Syndrome in a Portuguese Population

Introduction: Diagnosis of Cushing’s syndrome remains a challenge in clinical endocrinology. Even though late-night salivary cortisol is used as screening tool, individualized cut-off levels for each population must be defined. Material and Methods: Three groups of subjects were studied: normal subjects, suspected and proven Cushing’s syndrome. Salivary cortisol was measured using an automated electrochemiluminescence assay. The functional sensitivity of the assay is 0.018 μg/dL. The diagnostic cut-off level was defined by Receiver Operating Characteristic curve and Youden’s J index. Results: We studied 127 subjects: 57 healthy volunteers, 39 patients with suspected and 31 with proven Cushing’s syndrome. 2.5th - 97.5th percentile of the late-night salivary cortisol concentrations in normal subjects was 0.054 to 0.1827 μg/dL. Receiver Operating Characteristic curve analysis showed an area under the curve of 0.9881 (p < 0.0001). A cut-off point of 0.1 μg/dL provided a sensitivity of 96.77% (95% CI 83.3 - 99.92%) and specificity of 91.23% (95% CI 80.7 - 97.09%). There was a significant correlation between latenight salivary cortisol and late-night serum cortisol (R = 0.6977; p < 0.0001) and urinary free cortisol (R = 0.5404; p = 0.0025) in proven Cushing’s syndrome group. Discussion: The mean ± SD late-night salivary cortisol concentration in patients with proven Cushing’s syndrome (0.6798 ± 0.52 μg/ dL) was significantly higher (p < 0.0001). In our population, the late-night salivary cortisol cut-off was 0.1 μg/dL with high sensitivity and specificity. Conclusion: Late-night salivary cortisol has excellent diagnostic accuracy, making it a highly reliable, noninvasive, screening tool for outpatient assessment. Given its convenience and diagnostic accuracy, late-night salivary cortisol may be added to other traditional screening tests on hypercortisolism.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.