ABSTRACT
BACKGROUND: Salivary cortisol (sa-cortisol) and salivary cortisone (sa-cortisone) correlate well with serum cortisol (s-cortisol) but validated reference ranges for healthy individuals are lacking. OBJECTIVE: To establish cutoff levels for sa-cortisol and cortisone following cosyntropin testing, and assess their diagnostic utility in adrenal insufficiency (AI). METHODS: Steroids in saliva were assayed using liquid-chromatography tandem-mass-spectrometry (LCMS/MS) before and after administration of 250µg cosyntropin test in 128 healthy subjects (16 on oral estrogens) and 59 patients with suspected AI, of whom 26 were diagnosed with AI with conventional serum cortisol criteria. The cutoff level for AI was defined as the 2.5th centile in healthy subjects not receiving estrogens. Performance was evaluated by calculating diagnostic accuracy and analyzing receiver operating characteristic-curves. RESULTS: The sa-cortisol cutoff 60 minutes after cosyntropin stimulation was 12.6 nmol/L (accuracy 89%, sensitivity 85%, and specificity 90%). Sa-cortisone and the sum of sa-cortisol and cortisone exhibited poorer diagnostic performance than sa-cortisol. The correlation between sa-cortisol and s-cortisol was best described by a model incorporating two regression lines (R2 = 0.80). Segmented regression analysis identified a breakpoint at sa-cortisol 9.7 nmol/L and s-cortisol 482 nmol/L, likely corresponding to saturation of cortisol binding globulin (CBG). Healthy subjects on oral estrogens demonstrated a linear agreement between s- and sa-cortisol through all measurements. Seventeen healthy subjects repeated the test, with similar outcome, but reproducibility in terms of intraclass coefficient and correlation was poor. CONCLUSIONS: Sa-cortisol in cosyntropin-test has high diagnostic accuracy in detecting adrenal insufficiency, and is particularly useful in women on oral estrogens. A sa-cortisol > 12.6 nmol/L assayed with LCMS/MS 60 min after 250µg cosyntropin is normal.
