Diagnostic Accuracy of Morning Salivary Cortisol in the Assessment of the Hypothalamic-Pituitary-Adrenal Axis Recovery after Prolonged Corticosteroid Therapy in Children

Abstract

<b><i>Introduction:</i></b> Assessment of the hypothalamic-pituitary-adrenal (HPA) axis is necessary after prolonged glucocorticoid therapy withdrawal. Salivary cortisol reflects 65% of the free circulating cortisol fraction. Saliva collection is non-invasive and child friendly. <b><i>Objective:</i></b> We aimed to evaluate the diagnostic accuracy of morning salivary cortisol (mSAF) to determine HPA recovery after prolonged corticosteroid therapy in children. <b><i>Methods:</i></b> We conducted a prospective, validation study in 171 paediatric patients (mean ± SD age: 13.0 ± 4.4 years) who received glucocorticoids for &gt;4 weeks (median and interquartile range: 11 [7–14] months) and were referred for therapy withdrawal. Serum and saliva samples were collected between 8 and 9 a.m. on the same day. Cortisol was measured by an electrochemiluminescence immunoassay (ECLIA) 48 h after cessation of glucocorticoid therapy. Serum cortisol ≥193 nmol/L was used as the reference cut-off value for HPA recovery after glucocorticoid withdrawal and mSAF as the index test. <b><i>Results:</i></b> The cut-off concentration obtained by ROC for mSAF was ≥5.0 nmol/L. True positive and true negative results were observed in 85/171 and 40/171 children, respectively. The false-positive rate was low (3/171, 1.7%); however, false-negative results were observed in 43/171 (25%) children. The main ROC results (95% CI) were area under curve: 0.98 (0.96–0.99), sensitivity: 0.66 (0.57–0.75), specificity: 0.93 (0.81–0.99), positive predictive value: 0.97 (0.90–0.99), negative predictive value: 0.48 (0.37–0.59), LR+: 9.5, and diagnostic accuracy: 73.1%. <b><i>Conclusion:</i></b> The present study supports that mSAF ≥5.0 nmol/L by ECLIA is a non-invasive biomarker for the assessment of HPA recovery after prolonged glucocorticoid therapy in paediatric patients, with a positive predictive value of 97%. This proposed cut-off should be further validated using gold standard techniques for steroid quantification such as liquid chromatography-tandem mass spectrometry.