Endocrine Abstracts

JOINT1287

Background: Diagnosis of adrenal insufficiency (AI) in children requires formal cortisol stimulation test which is time-consuming and costly. While morning cortisol levels has been proposed to predict hypothalamic–pituitary–adrenal (HPA) axis integrity in adults, data on children are limited. This study aimed to investigate the association between basal morning cortisol and peak cortisol response to the insulin tolerance test (ITT), the gold standard for diagnosing adrenal insufficiency, and to establish morning cortisol cutoff values for predicting AI in children.

Method: This retrospective study includes all children aged <18 years who underwent ITT to evaluate the cause of short stature. Subjects with prior steroid use and known pituitary or adrenal diseases were excluded. We collected data on age, sex, anthropometry, Tanner stages, glucose levels, baseline and peak cortisol levels, and cortisol increment (peak–baseline) during ITT. AI was defined as a peak cortisol level from ITT<18 mcg/dL measured by Chemiluminescence Immunoassay. The Mann–Whitney U test compared data between AI group and normal HPA group (nHPA). Spearman’s rho correlation assessed association between variables. ROC analysis determined optimal cortisol cutoff values for predicting AI.

Result: Of the 108 subjects, 70 (64.8%) were male. The median age was 9.4 years (4.3–17.2). AI was diagnosed in 50 (46.3%) of the subjects. Compared to the nHPA, the AI showed significantly lower baseline morning cortisol (9.7 [4.7] vs. 12.6 [6.4] mcg/dL, P=0.018), and a significant smaller cortisol increment during ITT (3 [7.9] vs. 9.7 [10.7] mcg/dL, P<0.001). Baseline morning cortisol levels were positively associated with peak cortisol levels from ITT (r=0.336, P<0.001), particularly in the AI group (r=0.535, P<0.001). Peak cortisol levels correlated with cortisol increment (r=0.45, P<0.001), but not with age, anthropometry, Tanner stage, or gender. Regarding the morning cortisol levels, the cutoff of < 6.3 mcg/dL showed highest sensitivity (34%) and specificity (86.2%), although the best cutoff to predict AI was <4.8 mcg/dL (sensitivity 22%, specificity 91.4%), and the best cutoff to predict nHPA was >16.7 mcg/dL (sensitivity 90% and specificity 31%)(AUC 0.632, P=0.013). The cortisol increment cutoff of <8.7 mcg/dL from ITT showed the best sensitivity (82%) and specificity (60%) to predict AI in children (AUC 0.737, P<0.001).

Conclusion: Basal morning cortisol levels and cortisol increment were well correlated with peak cortisol response to ITT and could be used to predict AI in children, potentially reducing the need for further unnecessary procedures. Further investigation in larger cohorts is warranted to validate these findings.