ESPE Abstracts

GHD is the most common endocrine disorder in children with short stature (~1 in 4000). Appropriate diagnosis should synthesize auxological, biochemical and neuroradiological data. However, until now there are significant controversies, especially from the laboratory interpretation. A false positive diagnosis can lead to a significant wasted expenditure and unnecessary exposure to rare potential adverse effects. Given these open issues about biochemical diagnosis of GHD in childhood once reaching adult, patients should be re-assessed to define persistent adulthood GHD. We aim to identify predictive factors for permanent growth hormone deficiency.

Methods: GHD patients followed and re-tested when reaching adult height with an insulin provocative test, defining permanent with a peak <15 mIU/L. Collected information included perinatal, at diagnosis and at follow-up anthropometry, biochemical data, pubertal data, mid-parental height, MRI. Statistical analysis: Descriptive anthropometric and hormonal characteristics analysis, prognostic factors for persistent GHD evaluated using a univariate logistic regression.A multivariate stepwise model was developed.

Results: 101 GHD patients using a cut off of 20 mUI/l in 2 provocative tests, who started GH treatment at a mean age of 8.1± 0.4 years, with a height of -2.25±0.8 and BMI -0.27± 0.1 SDS. At final height 29 (28.7%) were defined as persistent GHD. The persistent GHD group had a shorter height SDS (-2.56±0.1 vs -2.13±0.1, P<0.001), a higher BMI SDS (0.13±0.27 vs.-0.42±0.13, P<0.05) and a lower GH peak mIU/l (8.3±1.0 vs 12.8±0.5, P<0.001, means of 2 stimulation tests) at diagnosis, and at final height a lower IGF1 ng/ml (232.± 19.9 vs. 331±9.1, P< 0.001) and a higher BMI SDS (-0.15± 0.27 vs. -0.73±0.13, P< 0.01). The OR for persistent GHD of patients with height < -3 SDS and between -3/-2.5 SD was 7.3 and 3.2 times vs. those taller than >- 2.5 SDS. By logistic regression analysis the best predictive model included height and BMI SDS and the two GH peak and MRI at diagnosis. Evaluated by ROC curve this model has a discrimination capacity of AUC of 0.9, a 89% sensitivity, 81% specificity and a 94.2% positive predictive value, showing that a shorter height SDS at baseline, adjusted by the other covariates showed 4.4 times risk of persistent GHD at adulthood. Nevertheless, an abnormal MRI was the best predictor (OR 10.7 of persistent GHD) after adjustment for the other covariates.

Conclusion: A shorter height SDS at diagnosis is a good predictor of persistent GHD, especially when accompanied by higher BMI, lower GH peaks and abnormal pituitary MRI.