Endocrine Abstracts

Background: Diagnosing adrenal insufficiency (AI) often requires complex multi-step testing which can be expensive and time consuming. In this study, we evaluated the diagnostic performance of baseline dehydroepiandrosterone sulfate (DHEAS) level in identifying patients with AI.

Methods: A single-center retrospective cohort study was performed of patients who underwent Cosyntropin stimulation testing (CST) between 2005 and 2023 and had a baseline DHEAS level obtained within 3 months of CST. Patients were excluded if they had congenital adrenal hyperplasia (CAH), oral estrogen use at the time of lab testing, were hospitalized within 1 month after CST, or died within 6 months after CST. AI was defined as post-CST peak cortisol <18 μg/dl and DHEAS level obtained closest to the time of CST was utilized. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were determined using a receiver operating characteristic (ROC) curve. Optimal cutpoint was determined to maximize both sensitivity and specificity. For further analysis, patients were stratified by gender and age; women ≥50 years old were assumed to be post-menopausal.

Results: Out of 1245 patients included in this study, 204 (16.4%) had an abnormal CST result. In these patients with AI, median DHEAS level was lower compared to patients without AI (21.2 vs 96.0 μg/dl, P<0.001). Similarly, patients with AI had a lower median baseline cortisol level than patients without AI (4.3 vs 8.0 μg/dl, P<0.001). Among all patients, baseline DHEAS level predicted an abnormal CST result with an accuracy rate of 73.7%. The optimal DHEAS cutoff point was determined to be 51 μg/dl with a sensitivity of 73.0%, specificity of 73.9%, PPV of 35.4%, and NPV of 93.3%. In our subgroup analysis, men had an optimal DHEAS cutpoint of 59.2 μg/dl with an accuracy rate of 72.2%, sensitivity 73.0%, specificity 72.0%, PPV 45.1%, and NPV 89.4%. While women <50 years old had an optimal DHEAS cutpoint of 63.9 μg/dl with an accuracy rate of 75.9%, sensitivity 75.9%, specificity 75.9%, PPV 29.2%, and NPV 96.0%, post-menopausal women had a lower optimal DHEAS cutpoint of 21.3 μg/dl with an accuracy rate of 71.1%, sensitivity 70.7%, specificity 71.2%, PPV 41.0%, and NPV 89.6%.

Conclusions: DHEAS is a valuable diagnostic test which can effectively rule out AI in many patients due to its high negative predictive value. For the majority of patients, a baseline DHEAS level >51 μg/dl makes a diagnosis of AI unlikely and additional dynamic testing may not be necessary.