Estrogen priming effect on growth hormone (GH) provocative test: A useful tool for the diagnosis of GH deficiency

We have studied the effect of estradiol (E-2)on the GH-insulin-like growth factor (GH-IGF) axis in 15 prepubertal GH deficiency (GHD) children and 44 prepubertal or early pubertal children with idiopathic short stature (SS). All of them received a daily dose of micronized E-2 (1 or 2 mg) or placebo, for 3 days, before a sequential arginine-clonidine test. In SS children, G H maximal responses were 17.8 +/- 10.9 on placebo and 27.9 +/- 14.5 mug/L o n estrogen (P < 0.0001). The lower 95% confidence limits for GH maximal res ponse changed from 3.7 <mu>g/L (without E-2) to 8.3 mug/L (on E-2). In GHD children, no significant stimulatory effect of estrogen on GH levels was ob served. After placebo, a cut-off limit of 3.7 mug/L (the lower 95% confiden ce interval limit) resulted in 73% sensitivity, 95% specificity, and an ove rall 90% diagnostic efficiency. After E-2, a cut-off limit of 8.3 mug/L res ulted in a sensitivity of 87%, a specificity of 98%, and a diagnostic effic iency of 95%. After placebo, 68% of SS showed normal IGF-I levels, and the mean did not c hange on E-2 (13.7 +/- 6.3 vs. 14.3 +/- 6.8 nmol/L, not significant). In 93 % of SS, IGF binding protein (IGFBP)-3 levels were normal during placebo. O n E-2, mean IGFBP-3 did not change (2.63 +/- 0.70 vs. 2.70 +/- 0.70 mg/L, n ot significant). In 14 of 15 GHD patients, IGF-I values were below normal o n placebo, and the mean of the group did not change after E-2. During place bo, 13 of 15 GHD children presented low IGFBP-3 values. During E-2, there w as a small significant increase in IGFBP-3 values (1.06 +/- 0.58 vs. 1.20 /- 0.69 mg/L, P < 0.02). The highest diagnostic efficiencies for IGF-I and IGFBP-3 were observed during placebo (75% and 91%, respectively). We conclude that GH stimulation tests after E-2 priming had the highest dia gnostic efficiency. Our findings suggest that the effect of estrogen primin g on GH stimulated levels, by reducing the number of false nonresponders, m ight be useful to better discriminate between normal and abnormal GH status in SS children.