Serum IGF-I and IGFBP-3 concentrations do not accurately predict growth hormone deficiency in children with brain tumours

OBJECTIVE The growth hormone (GH)-dependent growth factors insulin-like gro wth factor-I (IGF-I) and IGF-binding protein-3 (IGFBP-3) may be superior to provocative GH testing in diagnosing GH deficiency (GHD) in children. in a dults with brain tumours (BT) and GHD, however, provocative GH testing more accurately reflects GHD than either IGF-I or IGFBP-3. We assessed growth f actor levels in children with GHD due to BT with respect to brain tumour ty pe, pubertal stage, growth velocity, bone age delay, and body mass index (B MI). DESIGN Retrospective case review of all patients followed at our centre wit h GHD following treatment of BT. PATIENTS 72 children (51 M, 21 F) with BT diagnosed with GHD by clinical an d auxological criteria, including provocative GH testing, in whom pre-GH tr eatment IGF-I and IGFBP-3 revels were obtained. MEASUREMENTS Auxological data, including height, weight, growth velocity, a nd pubertal stage; and biochemical data, including GH response to provocati ve GH testing and pre-GH treatment serum IGF-I and IGFBP-3 concentrations. RESULTS IGF-I levels were normal (above -2 SD) in 19 of 70 children (27%), and IGFBP-3 levels were normal in 21 of 42 (50%). In children with GHD, pub ertal stage correlated significantly with both IGF-I (r = 0.328, p < 0.006) and IGFBP-3 (r = 0.364, P < 0.02) Normal IGF-1 levels were found in 1/15 c hildren with craniopharyngioma (Cranio) (7%), 10/30 with primitive neuroect odermal tumours (PNET) (33%), and 5/12 children with hypothalamic/chiasmati c glioma (HCG) (42%) (P < 0.05). IGFBP-3 levels were normal in 4/13 Cranio patients (31%), 8/15 PNET patients (53%), and 6/8 HCG patients (75%) (P = n s). Tanner staging varied significantly among tumour types: mode = 1 for Cr anio and PNET vs. mode = 3 for HCG (P < 0.03). BMI did not differ between p atients with low vs. normal growth factor levels. CONCLUSIONS Low IGF-I levels were more predictive of growth hormone deficie ncy than low IGFBP-3 levels in our brain tumour patients, but both were poo r predictors of growth hormone deficiency in children with hypothalamic-chi asmatic glioma and in pubertal children. Serum IGF-I and IGFBP-3 levels, th erefore, do not always reflect growth hormone deficiency in children with b rain tumours, particularly in those with hypothalamic-chiasmatic glioma or those already in puberty.