ESPE Abstracts (2018) 89 P-P2-237

Predictors of Poor Response to Growth Hormone Therapy in Children with Short Stature - Evidence from Neural Prediction Model for Final Height

Joanna Smyczynskaa, Urszula Smyczynskab, Maciej Hilczera,c, Renata Stawerskaa & Andrzej Lewinskia,d


aDepartment of Endocrinology and Metabolic Diseases, Polish Mother’s Memorial Hospital, Research Institute, Lodz, Poland; bDepartment of Biostatistics and Translational Medicine Medical University of Lodz, Lodz, Poland; cDepartment of Pediatric Endocrinology, Medical University of Lodz, Lodz, Poland; dDepartment of Endocrinology and Metabolic Diseases, Medical University of Lodz, Lodz, Poland


Prediction of poor response to growth hormone (GH) therapy in children with short stature is an important issue for personalized approach to treatment. Recently, our research group has published prediction models derived with neural networks. The main predictors of final height (FH) in our model were: patient’s height SDS at therapy onset (H0SDS) and pre-treatment IGF-I and IGFBP-3 concentrations but not the results of GH stimulation tests; pre-treatment growth rate was also a significant variable. The aim of present study is to analyze the main predictors of poor and good growth response to GH therapy in children with wide range of GH secretion. Analysis comprised 133 children (89 boys, 44 girls) with short stature, 101 with GH deficiency (GHD) and 32 with idiopathic short stature (ISS), treated with GH up to FH. In all children 20 auxological and hormonal parameters were assessed before treatment, in 1st year of therapy and at FH (for details see: Smyczynska U et al. doi.org/10.1530/EC-17-0277). According to the increase of FH SDS with respect to H0SDS below or over 1.0 S.D., the patients were classified as poor and good responders, respectively. As all but one poor responders were GH-deficient, further comparison between GHD and ISS was performed only for good responders. Both groups had similar H0SDS but in poor responders it was significantly higher than in good responders (−1.29±0.79 vs. −1.75±0.78, P=0.03) while corrected by target height (TH) SDS; pre-treatment growth rate (HSDS V0) was significantly better in poor than in good responders (−0.09±0.20 vs. −0.25±0.21, P=0.002). Poor responders had insignificantly higher IGF-I SDS than good responders before treatment (−1.17±0.96 vs. −2.07±1.07, P=0.06) and in 1st year of therapy (1.04±0.93 vs. 0.52±1.12), however IGF-I increase was lower in poor than in good responders (2.21±0.95 vs. 2.59±0.97). There were no similar differences for IGFBP-3. The only significant difference between GHD and ISS was that in GH secretion. More severe deficit of height with respect to TH, decrease of height SDS before treatment and more severe IGF-I deficiency were the main predictors of good response to GH therapy, with no difference between GHD and ISS. In poor responders decreased IGF-I sensitivity should be taken into account. Neural models are useful for identification of variables that should be subjected to further analysis.

Article tools

My recent searches

No recent searches.