Diagnosing hyper- and hypocortisolism using saliva samples : pitfalls and how to avoid them

Abstract: Background: Cushing's syndrome (CS) is caused by high cortisol secretion whereas insufficient cortisol secretion is called adrenal insufficiency (AI). Both are rare diseases with substantial diagnostic delay, and high morbidity and mortality even though effective treatment is available. This thesis aims to improve diagnostic tests for CS and AI using analyses of cortisol and its inactive metabolite cortisone in saliva samples.Methods: Papers 1 and 2 are based on a reference cohort including 155 individuals and 22 patients with CS. Salivary samples were collected at late-night (23:00 hours ± 15 minutes) and after a 1-mg overnight dexamethasone suppression test (DST). In Paper 1, reference intervals for salivary cortisol and cortisone analyzed with liquid chromatography-tandem mass spectrometry (LC-MS/MS) were established for late-night and post-DST samples. Diagnostic accuracy for CS was calculated using the established reference intervals. Potential effects of age, comorbidities, season, and sampling time point were also studied. In Paper 2, different analytical methods for measurement of salivary cortisol (3 LC-MS/MS and 3 immunoassays) and salivary cortisone (3 LC-MS/MS assays) were compared regarding reference intervals and diagnostic accuracies for CS. Paper 3 elucidated the potential effect of liquorice consumption, blood contamination, and topical hydrocortisone handling prior to sampling on salivary cortisol and cortisone. Paper 4 investigated whether salivary cortisol and cortisone are less affected than plasma cortisol by estrogen-containing oral contraceptive (OCs) in women undergoing a short Synacthen test (SST) by comparing the response in women with (n=41) and without OCs (n=46).Results: Paper 1 established reference intervals for salivary cortisol and cortisone at 23:00 hours and after DST. Using the upper reference limits as cut-offs, the diagnostic tests rendered high diagnostic accuracy for CS using salivary cortisol (sensitivity 90–95 %, specificity 96 %). There was no seasonal variation and no significant difference between samples collected at 22:00 vs 23:00 hours. Salivary cortisone showed a higher diagnostic accuracy for CS (sensitivity 100 % and specificity 94–95 %) and was less affected by other comorbidities compared to salivary cortisol. Paper 2 showed very high agreement between the three LC-MS/MS methods and that measuring salivary cortisol with immunoassays resulted in higher cortisol concentrations than with LC-MS/MS. However, using the newly established reference limits for each method, all had high diagnostic accuracy for CS. Late-night salivary cortisone analyzed with the LC-MS/MS methods and salivary cortisol analyzed with the Roche immunoassay showed the highest diagnostic accuracies. Paper 3 showed that liquorice consumption increased late-night salivary cortisol, which was sustained for up to 6 days, whereas no effect was seen on salivary cortisone. Salivary cortisol, but not cortisone, was increased by contamination of saliva with ≥0.5 % blood, which could be revealed by a clearly visible red discoloration of the saliva. Handling of topical hydrocortisone before saliva sampling affected salivary cortisol to a much higher degree than salivary cortisone. Paper 4 showed that women using OCs have considerably higher plasma cortisol levels during an SST, whereas salivary cortisol and salivary cortisone were lower compared to controls. However, the lower reference limits were not significantly different for salivary measurands, with salivary cortisone slightly more robust, opting for a common cut-off to exclude AI regardless of OCs.Conclusion: Using the reference intervals calculated for several clinically used analytical methods showed high diagnostic accuracy for CS, with cortisone showing the highest accuracy. Analyzing salivary cortisone was not affected by liquorice consumption or blood contamination. Salivary cortisone was least affected by OCs during an SST. In summary, salivary cortisone is very useful in the diagnostic work-up for CS and AI.