Development of diagnostic algorithm for Cushing’s syndrome: a tertiary centre experience

Abstract

Purpose No consensus exists as the gold standard for Cushing’s Syndrome (CS) screening. This study aimed to evaluate the diagnostic accuracy and utility of late-night salivary cortisol (LNSC) and cortisone (LNSE), overnight dexamethasone suppression test (ODST), and urinary free cortisol (UFC) in developing a screening algorithm for CS. Methods A retrospective, single-centre analysis on 93 adult patients referred to the Oxford Centre for Diabetes, Endocrinology, and Metabolism for CS evaluation (2017–2022). Data were analysed using binomial logistic regression and area under the receiver-operating curve (AUROC). Results Fifty-three patients were diagnosed with CS. LNSC (sensitivity 87.5%, specificity 64.9%, AUC 0.76), LNSE (sensitivity 72.4%, specificity 85.7%, AUC 0.79), and ODST (sensitivity 94.7%, specificity 52.1%; AUC 0.74) demonstrated comparable effectiveness for CS diagnosis. Their combined application increased diagnostic accuracy (AUC 0.91). UFC was not statistically significant. Pre-test clinical symptom inclusion improved screening test performance (AUC LNSC: 0.83; LNSE: 0.84; ODST: 0.82). For CD diagnosis, LNSE + LNSC (AUC 0.95) outperformed ODST. Combining these with ACTH levels < 12.6 pmol/L perfectly distinguished MACS (AUC 1.00). ODST (AUC 0.76) exhibited superior performance (sensitivity 100.0%, specificity 52.2%) in MACS detection. Conclusions LNSC, LNSE, and ODST are robust tools for CS screening, with their combined use offering the highest diagnostic precision. LNSE, especially when used with LNSC, is highly effective for CD diagnosis, exceeding ODST accuracy. ODST is preferable for MACS identification. Integrating ACTH levels markedly improves differentiation between CD and MACS. Conversely, UFC shows limited diagnostic utility.