Comparison of two methods for measuring salivary cortisol.

To the Editor: The concentration of cortisol in the saliva reflects the activity of the hy-pothalamic-pituitary-adrenal axis (1), and an increased late-night salivary cortisol has Ͼ95% sensitivity and spec-ificity for Cushing syndrome (1, 2). We compared a commonly used modification of a RIA for serum cor-tisol with a new enzyme immunoas-say (EIA) specifically designed to measure salivary cortisol in 352 saliva samples. The first samples (n ϭ 195) were sent to our laboratory to screen for Cushing syndrome; all were collected at 2300 from 90 patients (age range, 17–77 years; 74 women and 16 men). The second set of samples (n ϭ 104) was obtained at 2300 and 0700 from healthy, elderly patients (n ϭ 52; age range, 67– 82 years; 21 women and 31 men) who were part of a large, longitudinal study (3). The third set of samples (n ϭ 53) was obtained at 2300 and 0700 from a group of apparently healthy individuals (n ϭ 27; age range, 12– 63 years; 16 women and 11 men). The study was approved by the appropriate Institutional Review Boards, and consent was obtained. Saliva was sampled as described previously (2, 4) with a collecting device (Salivettes with no preservative; Sarstedt). Salivary cortisol was measured by two methods. The serum cortisol RIA (Coat-a-Count TKCO; Diagnostic Products) was used as commonly modified for the measurement of sal-ivary cortisol (2, 4). The salivary cor-tisol EIA (product no. 10-67100; Diagnostic Systems Laboratories) was used as instructed without modification. The sample volumes were 200 and 25 ␮L for the modified RIA and EIA, respectively, and the incubation times were 3 h for the RIA and 45 min for the EIA. The EIA calibrators ranged from 2.8 to 276 nmol/L and were provided as cortisol in a solution of bovine serum albumin. The lower detection limit of the EIA was 0.4 nmol/L. The intraassay imprecision (CV) was 9. The regression lines for the data from the three groups of individuals were not different from each other, so the combined data are shown in Fig. 1. The slope of the regression for RIA salivary cortisol values Ͻ11 nmol/L was not significantly different from the slope for the complete data set (Fig. 1, inset). Because the EIA showed consistently higher sali-vary cortisol concentrations, we assayed the RIA calibrators in the EIA, and vice versa. The EIA gave values for the RIA calibrators that were the same as the …