Clinical utility of insulin-like growth factor binding protein-3 in the evaluation and treatment of short children with suspected growth hormone deficiency

We have shown previously that serum insulin-like growth factor binding protein-3 (IGPBP-3) levels have good predictive value for complete, but not partial, growth hormone deficiency (GHD). In this study, we compare TGFBP-3 levels in short children previously divided into groups on the basis of their post-stimulation GH levels. Complete GHD (N = 59) included those children with peak post-stimulation GH < 5 μg/l. The partial GHD group (N = 49) had post-stimulation GH peaks of > 5 μg/l but < 10 μg/l. The normal children with short stature (N = 103) had post-stimulation GH peaks > 10 μg/l. Partial GHD and normal children with short stature also were divided into either low IGF-I or normal IGF-I subgroups. The clinical sensitivity of IGFBP-3 for complete GHD was 92%, whereas its sensitivity for partial GHD was 39%. For partial GHD, among those with low IGF-I (N = 19) 68% were also low for IGFBP-3, while 80% of those with normal IGF-I (N = 30) were also normal for IGFBP-3, The clinical specificity of IGFBP-3 for normal children with short stature was 69%. For these groups, among those with low TGF-I (N = 22) 73% also were low for IGFBP-3, while 80% of those with normal IGF-I (N = 81) also were normal for IGFBP-3. In addition, we tested whether IGFBP-3 can predict the response to GH treatment in prepubertal children by comparing pretreatment IGFBP-3 with the height gain achieved by 1 year of GH treatment. The incremental growth velocity during treatment correlated significantly with the pretreatment IGFBP-3 SD score (N = 46 r = -0.80, p < 0.005). The baseline IGFBP-3 so score for all subjects correlated (N = 171, r = 0.51 p < 0.0001) with height. These data suggest that IGFBP-3 may reflect GH secretion status in most children being evaluated for GHD and that a low pretreatment IGFBP-3 so score predicts improved growth during the first year of GH treatment.