Use of Collagen X Biomarker as a Predictor for Response to Growth Hormone Therapy

Collagen X biomarker (CXM) is a by-product of type X collagen that is deposited in the hypertrophic zones of growth plates and released during endochondral ossification. Prior studies have shown a correlation between annualized height velocity (AHV) and CXM values as well as between change in CXM in relation to change in AHV over time. We aim to test these correlations in children undergoing growth hormone (GH) therapy, and to evaluate whether short-term changes in CXM at 4 weeks post treatment initiation predict first year growth response to treatment.

We conducted a prospective noninterventional study of GH-naïve patients, >3 years old who were about to start GH therapy (any formulation under any indication). We collected historical heights from 6 to 18 months prior to enrollment, along with anthropometrics, CXM (measured by ELISA in serum) and IGF-1 values at enrollment, 4, 12, and 52 weeks post GH start. Normality of dependent variable (AHV or change in AHV) was tested using the Shapiro-Wilk normality test and a visual inspection of histograms. Linear regression models were used to assess each relationship. Due to a limited sample size, interactions were not assessed.

A total of 28 patients (9 females, 12 with GH deficiency, 5 with idiopathic short stature [ISS], 3 SGA, 3 SHOX, 2 Turner syndrome, 3 other) were enrolled and followed for the duration of the study. Mean age was 9.12 years with height Z score of -2.28 at baseline. AHV increased from 4.85 cm/y at baseline to 8.69 cm/y after treatment, and CXM increased from 19.11 ng/mL to 25.04 ng/mL after 52 weeks. Baseline CXM at enrollment was associated with baseline AHV (slope = 0.117; 95% CI 0.044-0.019; p = 0.002). However, the same was not true when looking at CXM values at 52 weeks in relation to the treatment AHV (slope = 0.024; 95% CI (-0.046 - 0.094); p = 0.52). Change in CXM from baseline to 4 (slope = 0.017; 95% CI -0.168 - 0.201); p = 0.86) or 52 weeks (slope = 0.03; (-0.081 - 0.148); p = 0.57) was also not predictive of change in AHV over 52 weeks despite adjusting for sex and age.

Our study confirms prior findings of CXM correlating with baseline AHV. However, that correlation was not significant during the 1 year of GH therapy. This is possibly due to multiple cofounders including differences in GH dosing, underlying diagnoses, and inability to confirm universal adherence to therapy, with some subjects having prolonged pauses in therapy for different reasons. Short-term change in CXM did not accurately predict growth response. Future prospective studies of a larger scale would be needed to investigate the role of CXM as a surrogate for GH therapy response.