Late-night salivary cortisol in normal subjects and in patients with Cushing's syndrome

Background Late-night salivary cortisol is used as a screening test for Cushing's syndrome (CS) in many European and American countries. However, its utility has not been studied in an Asian-Indian population. Objective To establish the reference range in Asian-Indians and to evaluate its usefulness in the diagnosis of CS. Methods Three groups of subjects were studied: normal subjects, patients with suspected CS, and patients with proven CS. All participants collected saliva at 23:00 h using a Salivette. Salivary cortisol was measured using an automated electrochemiluminescence assay. Results There were 56 normal subjects, 40 patients with suspected CS, and 30 with proven CS. Of the 40 with suspected CS, three were confirmed to have CS. The remaining 37 served as control patients. The 975th centile of the late-night salivary cortisol concentrations in normal subjects was 10.87 nmol/l. The mean +/- SD 23:00 h salivary cortisol concentration in control patients and those with confirmed CS was 3.21 +/- 2.36 nmol/l and 32.33 +/- 44.14 nmol/l, respectively. All the control patients and 30.3% (10/33) of patients with CS had a salivary cortisol concentration of <10.87 nmol/l. With the use of a receiver operating characteristic curve, a cut-off of 4.55 nmol/l gave a sensitivity of 93.9% and specificity of 81.1%. However, as this cut-off is less than the functional sensitivity of the assay, it may not be clinically applicable. Conclusions The reference range for late-night salivary cortisol in our population was <10.87 nmol/l. With this cut-off, the sensitivity was 69.2% and specificity 100%. Even though this automated electrochemiluminescence assay is easy and quick to use, its clinical utility in measuring the low salivary cortisol concentrations needs further investigation.