Association between high-density bone composition and vertebral fracture risk in patients with type 2 diabetes: a...

Early identification of high-risk fracture populations is clinically vital for comprehensive diabetes management. This study aimed to investigate whether the vertebral bone composition is associated with vertebral fracture risk in patients with type 2 diabetes mellitus (T2DM). This retrospective longitudinal study included 405 T2DM patients (135 with new-onset vertebral fractures and 270 matched controls) from a cohort of 937 individuals with at least two spinal computed tomography (CT) scans between July 2019 and July 2022. Vertebral bodies (T12 level) were segmented on baseline CT images into three regions based on Hounsfield Unit (HU) thresholds: Area A (< 110 HU, low density), Area B (110-160 HU, median density), and Area C (> 160 HU, high density). The areas of these regions and the mean vertebral CT value were measured. Univariate and multivariate Cox regression analyses were performed to identify independent risk factors for vertebral fractures. Predictive models were developed and evaluated using time-dependent receiver operating characteristic (ROC) curves, C-index, calibration, and decision curve analysis. After adjusting for potential confounders, both a smaller Area of C (adjusted Hazard Ratio [aHR] = 0.99 per mm² increase; 95% confidence interval (CI): 0.99-0.99) and a lower vertebral CT value (aHR = 0.99 per HU increase; 95% CI: 0.99-0.99) were independent predictors of vertebral fractures. Patients with a C-area < 308.7 mm² had a 4.32-fold higher fracture risk (95% CI: 2.61-7.32), and those with a vertebral CT value ≤ 100.7 HU had a 1.58-fold higher risk (95% CI: 1.06-2.63). The combined model (C-area + CT value) demonstrated superior predictive performance, with area under the curve (AUC) values of 0.775, 0.796, and 0.822 at 2, 3, and 4 years, respectively, outperforming models based on either variable alone (C-index: 0.751 for combined vs. 0.713 for C-area and 0.659 for CT value). Smaller high-density bone area (Area C) is a significant and independent risk factor for vertebral fractures in T2DM patients. A model combining high-density area and vertebral CT value provides a more robust prediction of fracture risk than traditional bone mass metrics alone, suggesting the potential for improved risk stratification in clinical practice.