Assay-Specific Diagnostic Thresholds for Adrenal Insufficiency: A Retrospective Comparison of Monoclonal Cortisol Measurements Using Roche Elecsys Cortisol II and Tosoh AIA-CL2400.

Background The diagnosis of adrenal insufficiency traditionally relies on peak cortisol values during stimulation testing, with a cut-off of 18 µg/dL established using polyclonal antibody assays. However, newer monoclonal antibody-based assays show reduced cross-reactivity with other glucocorticoids. In April 2022, our hospital changed the cortisol measurement method from Roche Elecsys Cortisol II to Tosoh AIA-CL2400, both using monoclonal antibodies but with potentially different measurement characteristics. In addition to stimulation test results, physicians initiate hydrocortisone replacement based on the patient's clinical condition. Methods We retrospectively reviewed 226 patients who underwent adrenocorticotropic hormone (ACTH) or corticotropin-releasing hormone (CRH) stimulation testing for suspected adrenal insufficiency between March 2016 and September 2023. Patients with baseline cortisol ≥18 µg/dL or receiving steroid medications other than hydrocortisone were excluded. Assessment variables included baseline cortisol, baseline ACTH, peak cortisol, and the change in cortisol during stimulation tests. Clinical diagnosis of adrenal insufficiency was defined by initiation of hydrocortisone replacement therapy. Receiver operating characteristic (ROC) curve analysis was performed to determine optimal cut-off values. Background factors included gender, age, presence of adrenal insufficiency symptoms, and structural or functional abnormalities of the hypothalamus-pituitary-adrenal (HPA) axis. Simple and multiple logistic regression analyses were conducted with the diagnosis of adrenal insufficiency as the dependent variable. Results The mean age was 54 years, and 75 (33.2%) of the participants were male. Adrenal insufficiency symptoms were present in 118 (52.2%), while 135 (59.7%) had either structural or functional abnormalities of the HPA axis. When comparing values between the two assay periods, peak cortisol was significantly different (mean: 14.6 vs. 16.8 µg/dL for Elecsys Cortisol II vs. AIA-CL2400). The proportion of patients with cortisol peaks <18 µg/dL differed (128 (68.1%) vs. 18 (47.4%)), but hydrocortisone initiation remained similar (82 (43.6%) vs. 14 (36.8%)). Multiple logistic regression analysis identified peak cortisol, presence of symptoms, and structural or functional abnormalities of the HPA axis as significant factors associated with the clinical diagnosis. ROC curve analysis revealed that the optimal cut-off value increased from 14.5 µg/dL before the assay change to 17.0 µg/dL after the change. When stratified by symptom presence, cut-off values were 14.5 µg/dL (with symptoms) and 13.1 µg/dL (without symptoms) before the change, and 17.0 µg/dL (with symptoms) and 15.1 µg/dL (without symptoms) after the change. When stratified by structural or functional abnormalities of the HPA axis, cut-off values were 13.7 µg/dL (with abnormalities) and 14.5 µg/dL (without abnormalities) before the change, and 17.5 µg/dL (with abnormalities) and 16.4 µg/dL (without abnormalities) after the change. Conclusion Our study provides two key messages: First, differences in assay methodology can significantly alter cortisol values and appropriate diagnostic cut-offs, even when transitioning between monoclonal antibody-based assays. Second, both peak cortisol and symptom presence were crucial for clinical diagnosis of adrenal insufficiency, with potentially lower cut-off values being appropriate for asymptomatic cases.