Constructing a precise prediction model for screening growth hormone deficiency in children in a single laboratory...

Growth hormone deficiency (GHD) requires accurate diagnosis, but the current gold standard, the growth hormone (GH) stimulation test, is highly invasive and resource-intensive, thus necessitating effective screening methods. Insulin-like growth factor 1 (IGF-1) is the most commonly utilized predictive marker for GHD. However, the consistency and accuracy of serum IGF-1 tests can be affected by patient characteristics and technical issues in blood analysis, which complicates its use as a reliable standalone marker. This study investigated the performance of the IGF-1/age ratio as a predictive marker for GHD and constructed a precise prediction model to mitigate these sources of variability in a single laboratory setting.

This cross-sectional study included 352 children aged 3 to 9 years who presented to Xi'an Children's Hospital for short stature and underwent the GH stimulation testing, which served as the gold standard for GHD diagnosis (peak GH cutoff of 10 µg/L). Key exclusion criteria were body mass index (BMI) ≥25 kg/m², history of GH therapy, secondary sexual characteristics, or chronic systemic diseases. Serum IGF-1 levels were measured using the IMMULITE^® 2000 Immunoassay System at our single center. The predictive performance of the IGF-1/age ratio was evaluated both alone and in combination with other markers. A precise prediction model was developed using a logistic regression algorithm, with feature selection guided by the Lasso method, and its performance was assessed using area under the curve (AUC) and calibration analysis. We designed a reference heatmap to facilitate its clinical use.

A total of 117 participants (33.2%) exhibited GHD based on the GH stimulation test. The GHD group exhibited significantly lower levels of IGF-1 (82.19 vs. 155.69 ng/mL, P<0.001) and IGFBP-3 (3.63 vs. 4.09 µg/mL, P<0.001) and a significantly greater bone age delay (calculated as bone age - chronological age; -1.64 vs. -1.06 years, P<0.001) compared to the non-GHD group. Among all evaluated markers, the IGF-1/age ratio demonstrated the highest predictive performance, with an AUC of 0.921 [95% confidence interval (CI): 0.894-0.947]. Combined use with other markers further improved the performance. The prediction model using insulin-like growth factor binding protein 3 (IGFBP-3), IGF-1/IGFBP-3 ratio, IGF-1/age ratio, and BMI showed good discrimination (AUC of 0.936, 95% CI: 0.913-0.960) and good calibration (Hosmer-Lemeshow test P value of 0.74).

Based on data from this single-center study, our results suggest that the IGF-1/age ratio is a promising predictive marker that may be used to effectively stratify GHD risk and facilitate fast screening prior to definitive GH stimulation testing.