Performance of afternoon (16:00 h) serum cortisol for the diagnosis of Cushing's syndrome.

Abstract

Introduction: Late-night salivary cortisol (LNSC), which assesses the loss of the circadian rhythm of cortisol, is one of the first-line tests performed to diagnose Cushing's syndrome (CS). Unfortunately, access to LNSC is still limited in some institutions. Alternatively, midnight serum cortisol can be measured, often combined with additional cortisol sampling over 24 h. This study investigates the performance of afternoon serum cortisol (F16h) for the positive diagnosis of hypercortisolism.

Methods: Retrospective study including consecutive patients evaluated for suspicion of CS by at least two tests among urine-free cortisol (UFC), overnight 1-mg dexamethasone suppression test (DST), and midnight serum cortisol (F00h). Patients assessed for adrenal incidentalomas or Cushing's disease recurrence were excluded.

Results: Of the 589 included patients (41.3 % women, mean age 50.7 ± 16.3 years), 49 (8.3 %) were diagnosed with CS. F16h demonstrated significant correlations with DST, UFC, and F00h (r=0.24, p<0.001, r=0.41, p<0.001 and r=0.42, p<0.001, respectively). The optimal cut-off of the F16h was 218 nmol/L, achieving 83.7 % sensitivity, 67.4 % specificity, and a 97.8 % negative predictive value. The area under the ROC curve (AUC) for the F16h did not differ from UFC (0.83 vs. 0.79, p=0.3), yet its sensitivity was higher using the optimal diagnostic threshold. The AUC for the F16h was significantly lower than that for F00h and DST.

Conclusions: Given the limited diagnostic accuracy of both F16h and UFC, particularly in mild to moderate CS, efforts should prioritize expanding access to reliable LNSC assays for circadian rhythm assessment. Meanwhile, F16h may help rule out CS when LNSC is unavailable.