A knee bone age-based predictive nomogram for the poor response to growth hormone therapy in children with idiopathic short stature during mid-to-late puberty.

purposeTo determine whether MRI-based knee bone age (KBA) could be a predictor for growth hormone therapy (GHT) response in idiopathic short stature (ISS) children during mid-to-late puberty, and to develop a KBA-based model for predicting the risk of poor GHT response (pGHR).

methods80 adolescents aged over 10 years, at Tanner stage 3 or above, with height below − 2 SDS and a GH peak ≥ 7 ng/dL were included in the study, dividing into pGHR (∆HTSDS<0.5SDS) and good GHT response (gGHR, ∆HTSDS≥0.5SDS) groups. Univariable logistic regressions investigated associations between KBA, clinical features and pGHR risk. Multivariable logistic regression was established for nomogram development. Internal validation was performed using 1000-time self-bootstrapping. Nomogram performance was evaluated through discrimination, calibration, and clinical utility.

resultsKBA, chronological age, and Tanner stage were positively associated with the risk of pGHR, while the distance between predicted adult height and pretreatment height in SDS (PAHSDS–HTSDS) was negatively associated. Male with KBA > 14.5 years and female with KBA > 12.5 years had higher risk of pGHR. Multivariable logistic regression identified KBA, sex, and PAHSDS–HTSDS as independent predictors. A nomogram developed by these predictors showed good calibration and discrimination (AUC = 0.953). The score of internal validation also revealed good discrimination (AUCmean=0.960).

conclusionsKBA is a reliable predictor for GHT response in ISS children during mid-to-late puberty. A simple and predictive nomogram including KBA, sex and PAHSDS–HTSDS was developed to predict the pGHR risk, exhibiting good discriminative ability and clinical utility, which may assist clinicians make clinical decisions and minimize unnecessary interventions.