Endocrine Abstracts (2015) 38 P22 | DOI: 10.1530/endoabs.38.P22

Effect of cortisol assay bias on the overnight dexamethasone suppression test: implications for the investigation of Cushing's syndrome

Rachel Brixey-McCann1, Sarah Tennant1, John Geen2,3, Annie Armston4, Julian H Barth5, Brian Keevil6, Aled Rees7 & Carol Evans1


1Department of Medical Biochemistry and Immunology, Cardiff and Vale University Health Board, University Hospital of Wales, Cardiff, UK; 2Department of Clinical Biochemistry, Prince Charles Hospital, Cwm Taf University Health Board, Abercynon, Rhondda Cynon Taff, UK; 3Faculty of Life Sciences and Education, University of South Wales, Cardiff, UK; 4Department of Clinical Biochemistry, Southampton General, Southampton, UK; 5Department of Blood Sciences, Leeds General Infirmary, Leeds, UK; 6Department of Clinical Biochemistry, University Hospital South Manchester, Manchester, UK; 7School of Medicine, Centre for Endocrine and Diabetes Science, Cardiff University, Cardiff, UK.


Background: NEQAS data demonstrate a divergence in bias of cortisol immunoassays over the last 10 years. Despite this, a serum cortisol of 50 nmol/l has been universally applied as the cut-off for the overnight dexamethasone suppression test (ONDST), the commonest screening test for Cushing’s syndrome.

Aims: To assess the effect of assay bias on interpretation of the ONDST and determine the necessity for a method-specific cut-off.

Methods: 120 serum samples sent for cortisol analysis as part of an ONDST were collected prior to disposal. Samples were anonymised and aliquots prepared for cortisol analysis by four different immunoassays (Abbott Architect, Roche E170, Beckman Access, and Siemens Centaur), and both cortisol and dexamethasone analysis by LC–MS/MS. Precision at three different cortisol concentrations (~30, 75, and 100 nmol/l), using sample pools prepared from patients not on interfering steroids, demonstrated inter-assay CV of <10% for all cortisol assays at concentrations above the lower reporting limit (LRL) of the assay. Case notes were reviewed to ascertain clinical indications (49% adrenal incidentalomas) and final diagnosis (13% confirmed Cushing’s).

Results: Cortisol concentrations for each patient sample above the LRL for each assay were compared to the corresponding LC–MS/MS cortisol. Mean biases were −19.5 nmol/l (Architect), 15.2 nmol/l (E170), 0.3 nmol/l (Access), and −3.9 nmol/l (Centaur). Dexamethasone was detected in 111 samples, with optimal concentrations (>5.6 nmol/l) in 88 samples; samples without dexamethasone (n=6) were removed from further analysis. Using the 50 nmol/l cortisol cut-off, 29/108 samples were screen positive by all methods, 60/108 screen negative, and 19/108 samples were discrepant, of which one subsequently had confirmed Cushing’s syndrome. Sensitivities and specificities at a 50 nmol/l cut-off varied from 87.5 to 92.5% for the Abbott, to 93.8 and 91.2%, respectively, for LC–MS/MS.

Conclusions: Assay bias affects performance of the ONDST and dexamethasone measurements are required for accurate interpretation.

Article tools

My recent searches

No recent searches.