Waking salivary cortisone as screening test for adrenal insufficiency

Miguel Debono; Charlotte Elder; Jen Lewis; Jane Fearnside; Sharon Caunt; Simon Dixon; Richard Jacques; John Newell-Price; Brian Keevil; Richard Ross

doi:10.1530/endoabs.81.P7

Introduction: In many endocrine centres the 250μg Short Synacthen (Cosyntropin) Test (SST) is the reference standard for the diagnosis of adrenal insufficiency (AI)¹, but it is time consuming, expensive, and requires hospital attendance and venepuncture. The morning physiological peak of cortisol shortly after waking is a good predictor for a negative SST; however, a morning serum cortisol requires venepuncture. Serum cortisol and salivary cortisone correlate strongly², but salivary cortisone is far more convenient as can be collected at home, posted to the laboratory, and is stable at room temperature. We hypothesised that waking salivary cortisone (WSC) could predict a positive or negative SST in patients assessed for AI.

Methods: A prospective, diagnostic accuracy study of WSC was performed in 220 people at high risk of AI (majority on exogenous glucocorticoids). All patients collected a salivary cortisone sample on waking, and then attended the endocrine unit for a 250μg SST. AI was defined as a SST 30minute serum cortisol <430nmol/l. Baseline and 30minute serum cortisol was measured by LC-MS/MS and Roche Elecsys Cortisol II immunoassay, and WSC by LC-MS/MS. Questionnaires were used to assess patient views on home WSC vs SST in hospital. ROC curves were computed to assess diagnostic accuracy. Sensitivities, specificities, PPV and NPV (95%CI) were estimated.

Results: Insufficient salivary collection excluded 5.5% of samples. By SST the prevalence of AI in 208 subjects was 45% using LC-MS/MS and 44% using Immunoassay. Using the diagnostic thresholds in the table below with at least 95% sensitivity/specificity on valid WSC samples we achieved 96% NPV and 95% PPV, to exclude and confirm AI respectively. 82% of subjects preferred the home salivary cortisone.

Conclusion: Home WSC testing is a highly accurate screening tool for AI. We have shown that using WSC at 95% sensitivity/specificity avoids performing a SST in 76% (measured by LC-MS/MS) and 74% (measured by immunoassay) of patients at high risk for AI. WSC is also the preferred test by patients.

Table 1
WSC Threshold Predicts		Sensitivity	Specificity	PPV	NPV
SST 30 minute cortisol by LC-MS/MS
Exclude AI	≥17 nmol/l	96.81(90.96-99.34)	68.42 (59.05-76.81)	71.65 (62.98-79.29)	96.3 (89.56-99.23)
Confirm AI	<8nmol/l	78.72 (69.07-86.49)	96.49 (91.26-99.04)	94.87 (87.39-98.59)	84.62 (77.24-90.34)
SST 30minute cortisol by Immunoassay
Exclude AI	≥17nmol/l	96.7 (90.67-99.31)	66.67 (57.36-75.11)	69.29 (60.49, 77.17)	96.3 (89.56-99.23)
Confirm AI	<7nmol/l	75.82 (65.72-84.19)	96.58 (91.48-99.06)	94.52 (86.56-98.49)	83.7 (76.37-89.50)

References: ¹Elder CJ, Arch Dis Child 2012 97:870-873;²Debono M, JCEM 2016 101:1469-1477

Endocrine Abstracts

P7