The performance and reproducibility of late-night salivary cortisol estimation by enzyme immunoassay for screening Cushing disease.

Objective: Our study aimed to establish a local reference range for late-night salivary cortisol (LNSC) using enzyme immunoassay (EIA) and to study the intra-individual reproducibility of LNSC.
Methods: Prospective study involving 30 healthy subjects (HS) with body mass index (BMI) <25 kg/m2, 37 obese/overweight subjects (OS) with BMI >25 kg/m2 and 28 patients with Cushing disease (CD). Salivary sampling was performed on 2 consecutive nights and assayed by EIA. The reference range was established using LNSC values of HS, and receiver operating characteristic (ROC) curves were used to determine diagnostic cutoffs.
Results: The mean LNSC level of CD was significantly higher than HS and OS (CD: 16.96 ± 9.11 nmol/L, HS: 1.30 ± 0.95 nmol/L, and OS 1.21 ± 0.78 nmol/L). A cutoff of 2.92 nmol/L differentiated CD from HS with 100% sensitivity and 96.7 % specificity, and a cutoff of 5.04 nmol/L yielded a specificity of 100% with a sensitivity of 96.4% to distinguish CD from OS. There was more intra-individual variability in HS (55%) than in CD (49%) and OS (22%). There was no difference in the sensitivity and specificity derived from the ROCs using day 1 values or the higher of the 2 LNSCs.
Conclusions: In our cohort, we found that LNSC assayed by EIA showed good sensitivity and specificity to screen patients suspected to have CD. Although intra-individual variability was significant, it did not hamper the diagnostic performance of the test.