Minding the gap between cortisol levels measured with second-generation assays and current diagnostic thresholds for the diagnosis of adrenal insufficiency: a single-center experience

Purpose The current cut-offs for the diagnosis of adrenal insufficiency (AI) have been established using outdated immunoassays. We compared the cortisol concentrations measured with Roche Cortisol I (R1), the newly available Roche Cortisol II (R2), and liquid chromatography tandem mass spectrometry (LC-MS/MS), the gold standard procedure to measure steroids in patients undergoing the corticotropin (ACTH) test. Methods We enrolled 30 patients (age 47 +/- 21 years) referred to undergo the ACTH test (1 or 250 mu g). Cortisol was measured at 0, 30, and 60 min after stimulation with R1, R2, and LC-MS/MS. AI was diagnosed for R1-stimulated peak cortisol concentrations < 500 nmol/L. Results Mean cortisol concentrations measured with R2 and LC-MS/MS were comparable, while mean cortisol concentrations measured by R1 were higher than those of both R2 and LC-MS/MS (respectively, basal 411 +/- 177, 287 +/- 119, and 295 +/- 119 nmol/L; at 30 min, 704 +/- 204, 480 +/- 132, and 500 +/- 132 nmol/L; at 60 min, 737 +/- 301, 502 +/- 196, and 519 +/- 201 nmol/L, p <= 0.01 for R1 vs. both R2 and LC-MS/MS at each point). Considering the 500 nmol/L cortisol peak cut-off, AI was diagnosed in 5/30 patients using R1 and in 12/30 using R2 (+ 140%). Based on the correlation between R1 and R2, the threshold of 500 nmol/L became 351 nmol/L (12.7 mu g/dL) when cortisol was measured with R2, and 368 nmol/L (13.3 mu g/dL) with LC-MS/MS. Conclusions The use of more specific cortisol assays results in lower cortisol concentrations. This could lead to misdiagnosis and overtreatment when assessing AI with the ACTH test if a different cut-off for cortisol peak is not adopted.