Background: The relationship of pre-treatment gene expression (GE) to long-term growth response in GHD is unknown. Prediction of long-term response to r-hGH therapy would allow better decision making about start and maintenance doses and hence cost:benefit.
Objective and hypotheses: To assess the relationship of baseline GE to response to r-hGH over 5 years of therapy in GHD children.
Method: Pre-pubertal children with GHD (n=50) were enrolled from the PREDICT studies (NCT00256126 and NCT00699855). Baseline whole blood GE was determined using Affymetrix U133 plus 2.0 microarrays and Gene Expression Barcode 3.01. Height velocity (HV) on r-hGH over the 5 years of therapy was used as the marker of growth response. Two groups of patients were defined on growth response over 5 years of treatment (G1) always above and (G2) always below the median. The effect of age, gender and distance to target height (DTH) were assessed. Network models and random forest analysis were used to relate GE to growth response using area under the curve (AUC) of the receiver operating characteristic. The robustness of GE markers was assessed using permutation testing (n=1000).
Results: There was no difference in age, gender and DTH (P>0.05) between the HV groups (G1 and G2) at the start of treatment. Uniquely expressed genes were identified (P<1×10−5) in G1 (n=69) and G2 (n=72). Network models prioritised 94 of these 141 genes. PIK3R1 expression (P=1.2×10−9), associated with cell proliferation, was related to G1. DDX58 expression (P=2.2×10−10), associated with RNA secondary structure, was related to G2. Baseline GE could predict growth response consistently above and below the median over 5 years with an AUC of 0.86 and 0.89 respectively.
Conclusion: We have identified genes expressed in pre-treatment GHD associated with growth response over 5 years of therapy. Further assessment to determine predictive value and functional significance of gene subsets is required.
10 - 12 Sep 2016
European Society for Paediatric Endocrinology