SUN-258 Prediction of Height Two Years after Start Treatment in Children with Growth Hormone Deficiency

Background: Various prediction models for the growth response to recombinant human growth hormone (rhGH) have been published, but none of these have been able to incorporate data on adherence to rhGH treatment. Previous studies have shown that suboptimal adherence negatively affects the growth response in the first two years of rhGH treatment. The availability of the easypod electronic drug-delivery device which continuously assesses adherence, enables to develop a prediction model taking adherence into account. Aim: To develop a model for prediction of height standard deviation score (HSDS) two years after start treatment, using information available at the start of GH treatment and adherence in the first two years. Patients and Methods: We included prepubertal children with idiopathic isolated growth hormone deficiency (GHD) participating in the easypod connect observational study (ECOS), a 5-year, Phase IV open-label study to continuously assess real-world adherence via the easypod device. Inclusion criteria were GHD, naïve to rhGH treatment, age <10y in girls and <12y in boys, height and adherence available in the first and second year, target height, birth weight, gestational age, and information on GH dose. Linear regression analyses were performed to construct a prediction model for HSDS 2y after start treatment, with age at start, HSDS at start, target height (TH) SDS, birth weight SDS, GH dose, gestational age (<37 weeks vs ≥37 weeks), sex, and adherence in the first and second year as potential predictors. Stepwise backward regression analyses were performed to simplify the model without a statistically significant loss of fit with a model fit criterion of p<0.1 for each predictor. Results: In total, 132 patients complied with the inclusion criteria. Mean (SD) HSDS was -2.4 (0.9) and age at start 6.5 (2.0) yrs. The prediction model was: HSDS at 2y after start treatment = 0.4699 - 0.0627 x Age at start + 0.7904 x HSDS at start + 0.1216 x TH SDS + 0.0418 x birth weight SDS + 0.1659 x sex (male) + 0.0050 x adherence in the second year. P-values were <.001 for HSDS at start and adherence in the second year, .002 for Age at start, and .07 for TH SDS, birth weight SDS and sex. This model explained 76% of the total variance. For example, if a girl starts with GH treatment at 6.5y with a HSDS of -2.4, TH SDS of 0, birth weight SDS of 0 and 100% adherence in the second year, her predicted HSDS 2y after start treatment is -1.3 SD. If she misses 2 injections/week in the second year, the predicted HSDS 2y after start treatment is -1.5 SD. Conclusion: We have developed a prediction model for height two years after start treatment for children with GHD. This model can be used to study the impact of information available at the start of GH treatment and adherence in the second year. More research is needed to explore the effect of individual patterns of adherence on growth in real word settings.