ESPE Abstracts

Introduction: The diagnosis of growth hormone deficiency (GHD) in childhood is challenging, because of the lack of a true gold standard and the relatively poor performance of available diagnostic testing. Many young children undergo unnecessary growth hormone stimulation tests (GHST).

Objective: The aimof this study was to design and validate a predictive Score to diagnose children with GHD.

Patients and Methods: This was multicentric study conducted from 2019 to 2023. Prepubertal short children aged 3-11 years from 5 hospital in Algiers, Algeria, were included They all had a clinical and hormonal evaluation for short stature (including 2 growth hormone stimulation tests and an MRI. They were divided in 2 groups: GHD patients and Idiopathic short stature patients (ISS). Using the results of the study, a multivariable prediction model was designed and validated using the TRIPOD guidelines, and a score was developed for the diagnosis of GHD. A different cohort of short children was used to validate the score

Results: The study compared 329 GHD children with 392 ISS children. The optimal thresholds, when used alone were:

• Child’s height ≤-3.16 SD (area under the curve (AUC):0.594, Se:35.56%, Sp:85.71%)
• Discrepancy between the child’s height SD and the mid-parental height (MPH) SD ≤2.38 SD (AUC:0.630, Se:36.36%, Sp:86.61%),
• Decrease of the growth velocity ≤-1.93 SD (AUC:0.608, Se:37.50%, Sp:81.12%) and ≤2.25 (AUC:0.717, Se:51.30%, Sp:83.82%). The study identified some key risk factors for the diagnosis of GHD through multivariate analysis, including: the discrepancy between the child’s and MPH, the decrease of growth velocity, the bone age delay, the IGF1 levels, and the anatomic pituitary abnormalities. A scoring system was developed using logistic regression, with different thresholds indicating the need for the GH stimulation testing, the diagnosis of GHD and predicting its persistence. The scoring system demonstrated a high performance, with an AUC ranging from 0.833 to 0.925, a sensitivity from 70.45% to 72.5%, and a specificity from 89.1% to 94.8%. In a different cohort of 275 short children, the prediction score showed a specificity of 92.98%, a sensitivity of 93.12%, and an error rate of 6.91%, with high positive predictive (0.98) and negative predictive (0.78) values.

Conclusion: Growth hormone deficiency (GHD) should be diagnosed through a comprehensive evaluation of clinical, auxological, imaging and hormonal data. Our score is highly sensitive and specific for the diagnosis of GHD in prepubertal children. Its routinely use will reduce the number of GHST in children. Further studies are needed to validate it.