Bone Abstracts

Introduction and objective: Adolescent idiopathic scoliosis (AIS) is a three-dimensional spinal deformity with lateral curvature of ≧10 degree measured by Cobb angle. Previous studies reported that ~30% of female AIS patients had systemic osteopenia which could persist beyond skeletal maturity and that was one of the important prognostic factors. The aim of this study is to investigate the incremental prognostic value of osteopenia on curve progression in AIS as defined by Scoliosis Research Society (SRS) criteria using the decision curve analysis (DCA) and reclassification table.

Methods: Between 1997 and 2014, 450 newly diagnosed AIS girls aged 11–16 years old before skeletal maturity without prior treatment and with initial Cobb angle ≤40° were recruited. Bilateral hips were measured by dual-energy x-ray absorptiometry (DXA) at their initial clinic visit, followed by regular follow-up with detailed clinical and radiological assessments. Curve progression was defined as ≧6° increase at maturity (years since menarche≧2 and age≧16). Two regression models were compared in their prediction on curve progression. Model 1 (M1) included background variables of Cobb angle, age at clinic, menarche status (yes/no) and assigned brace treatment (yes/no) at first visit. Model 2 (M2) was same as M1 added with the osteopenia status (yes/no) using z-score of bone mineral density (BMD).

Results: Among all subjects, 92 (20.4%) were osteopenia with z-score of BMD≤−1 at femoral neck and 72 (18.7%) had curve progression (progressive cases), as defined by SRS criteria. Overall, M2 has better performance than M1 (AUC 0.641 vs 0.631, P=0.568, Nagelkerges R² 0.068 vs 0.052, P=0.035). DCA showed similar results (M2 had higher net-benefit values than M1). Besides, M2 improved the reclassification of high-risk progressive cases when compared to M1 (net reclassification improvement=0.123, P=0.003), e.g. ~12.3% better classification by including baseline osteopenia status in prediction model.

Conclusion: In conclusion, the prediction model with baseline osteopenia had better prediction performance than that of without. Despite reaching statistical significance, the relatively low value of R² indicates the presence of other bone health parameters that might have to be included to further improve the predictive power for curve progression.

Disclosure: The authors declared no competing interests.