Constructing an Insulin-Like Growth Factor-based Prediction Model

The first phase of our study, previously approved by the UCLA IRB in August 2004, involved a retrospective chart review and collection of preexisting data. We analyzed charts of patients who underwent non-weight-based GH dose adjustments preceded and followed by a serum IGF-I level. Data collected included patient's age, gender, height, weight, linear height velocity, pubertal status, GH dose, IGF-I levels, disease condition (i.e., growth hormone deficiency, idiopathic short stature, small-for-gestational age) and recording of any adverse effects. Based on this data collected, we performed statistical analysis of the relationship between GH dose change and change in IGF-I level achieved, and have found that in prepubertal children, there is a significant relationship between the GH dose change and change in corresponding IGF-I level. We have subsequently constructed a mathematical prediction model that allows us to determine the GH dose change necessary to achieve a desired IGF-I level, and in this second phase of our study, we plan to apply this mathematical prediction model prospectively for guiding GH dose adjustments in prepubertal children being treated with GH therapy.

Children being treated with GH are typically followed in the outpatient Endocrinology clinic every 3-4 months, and our practitioners have routinely been obtaining serum IGF-I levels (via venipuncture in the outpatient laboratory) for monitoring purposes. Validation of our GH dose change/IGF-I change prediction model will require assessment of sequential serum IGF-I level measurements immediately preceding a GH dose change, and again within 1-4 months following the GH dose change. We plan to continue our retrospective and ongoing analysis of patients who have paired values of GH dose adjustments preceded and followed by a serum IGF-I level, in order that we may continue to fine-tune our prediction model for optimizing GH dose adjustments.