RESUMO
BACKGROUND: Adrenal insufficiency (AI) is a life-threatening condition which requires long term glucocorticoid replacement. The insulin tolerance test (ITT) is the current gold standard test for diagnosis of secondary AI, but the widely accepted cut-off value of a peak cortisol of less than 500 nmol/L assumes that anyone who does not reach this value has AI and thus requires full replacement. The cut-off used to diagnose AI is also founded on outdated assays. Use of this cut-off in an era of more specific immunoassays therefore risks misdiagnosis, subsequent unnecessary glucocorticoid exposure and associated adverse effects with increased mortality risk. DESIGN, PATIENTS AND MEASUREMENTS: This retrospective analysis assessed 300 ITT cortisol responses using the Abbott Architect and Alinity analyser platforms in patients with suspected AI over a period of 12 years (August 2010 to January 2022), at a tertiary centre. RESULTS: Patients were classified as having AI or not, based on a comprehensive clinical review of electronic patient records from the point of test to the present day by a panel of pituitary and adrenal specialists. Using the current institutional cut-off value of 500 nmol/L, receiver operating characteristic analysis identified a 100.0% sensitivity and 43.6% specificity (area under the curve 0.979). Using a lower cortisol threshold value of 416 nmol/L on the Abbott analyser platform maintained a sensitivity of 100.0% and improved the specificity to 86.7%. CONCLUSION: This data supports lowering the Abbott analyser ITT peak cortisol threshold to 416 nmol/L. Use of this improved cut-off avoids unnecessary glucocorticoid replacement therapy in 104 (34.7%) of individuals in this study. All patients remained well with at least 1 year longitudinal follow up of glucocorticoid replacement.