Glucocorticoid induced adrenal insufficiency in children: Morning cortisol values to avoid LDSST

Abstract

Objectives: Glucocorticoid-induced adrenal insufficiency (GI-AI) is a common side effect of glucocorticoid therapy. However, its diagnosis currently relies on the realization of a Low Dose Short Synacthen Test (LD-SST) that requires an outpatient hospital and several blood samples. Our goal was to evaluate whether morning cortisol values could predict the response to LD-SST, in children, to avoid useless dynamic tests and facilitate diagnosis of glucocorticoid induced adrenal insufficiency.

Study design: We recorded data of 91 pediatric patients who underwent a LD-SST in our center between 2016 and 2020 in a retrospective observational study. We selected LD-SST realized following administration of supra-physiologic doses of glucocorticoids during more than 3 weeks and performed at least four weeks after treatment was stopped. Adrenal deficiency was defined as a plasma cortisol concentration inferior to 500 nmol/l at LD-SST.

Results: Glucocorticoid-induced adrenal insufficiency was diagnosed in 60% of our cohort. Morning cortisol values were predictive of the response to the LD-SST (AUC ROC 0.78). A plasma cortisol concentration of less than 144 nmol/l predicted glucocorticoid induced adrenal insufficiency with a specificity of 94% and a value over 317 nmol/l predicted recovery of the HPA axis with a sensitivity of 95%. We did not find any other predictive factor for glucocorticoid-induced adrenal insufficiency.

Conclusions: Morning cortisol values can safely assess recovery of the HPA axis in children treated chronically with glucocorticoids. Using these thresholds, more than 50% of LD-SST could be avoided in children.

Keywords: adrenal insufficency; cortisol; glucocorticoid withdrawal; low dose short synacthen test; pediatric.

Figure 1

Flow chart of the study.

Figure 2

Morning plasma ACTH levels according to the result of the LD-SST. Distribution of ACTH dosages is represented with median and interquartile range depending on the results of the LD-SST. LD-SST was considered positive when cortisol at T20 raised above 500 nmol/L otherwise it was considered negative. Association between ACTH and GI-AI did not reach statistical significance using Analysis Of Variance (ANOVA, p = 0.18, ns).

Figure 3

Morning plasma cortisol levels according to the result of the LD-SST (A) morning plasma cortisol value was associated to the LD-SST (ANOVA, ***p < 0.0001). LD-SST was considered positive when cortisol at T20 raised above 500 nmol/l otherwise it was considered negative. The full line represents the value associated with GI-AI (Se = 0.47, Sp = 0.94). The dotted line represents the value associated with a Normal response to LD-SST (Se = 0.95, Sp = 0.42). B) ROC curve for morning plasma cortisol level with LD-SST as the reference test. Area under the curve is 0.7795 (IC95 [0.69; 0.88], p value < 0.001).