Opportunistic Detection of Lumbar Scoliosis on DXA Images in Postmenopausal Women

Background: This study aimed to determine the prevalence of DXA-detected lumbar scoliosis in postmenopausal women based on dual-energy X-ray absorptiometry (DXA) scans and identify associated risk factors. Methods: A total of 261 postmenopausal women aged ≥50 years who underwent lumbar spine DXA before June 2021 were included. Lumbar scoliosis was defined as a Cobb angle ≥ 10° measured from DXA images. Logistic regression analysis was performed to evaluate associated risk factors. Diagnostic performance of DXA-based Cobb angle measurements was assessed in the radiographic validation subgroup using radiography as the reference standard. Results: The prevalence of DXA-detected lumbar scoliosis was 14.9% (39/261; 95% CI, 10.8-19.9%). Increasing age was significantly associated with scoliosis, while body mass index, bone mineral density, and T-scores at the lumbar spine, hip, and femoral neck were not. DXA and radiographic Cobb angle measurements demonstrated strong agreement (ICC = 0.91, 95% CI 0.73-0.96), with a mean difference of -2.63°. Diagnostic accuracy was 82.1%, with sensitivity 62.1%, specificity 97.4%, PPV 94.7%, and NPV 77.0%. ROC analysis demonstrated good discriminative performance (AUC = 0.88, 95% CI, 0.79-0.98); an exploratory cutoff of 6.5° yielded the highest Youden index. Conclusions: DXA-detected lumbar scoliosis was identified in 14.9% of postmenopausal women undergoing DXA. DXA-based Cobb angle measurements demonstrated strong agreement with radiographic assessment and may facilitate opportunistic case detection of likely lumbar scoliosis during routine BMD assessment.