Comparing the predictive value of CT-derived hounsfield units and MRI-derived bone quality scores for cage subsidence following spinal fusion: a systematic review and meta-analysis

Cage subsidence is the most common hardware-related complication of spinal fusion, often associated with suboptimal outcomes. Poor vertebral bone quality has repeatedly been suggested as a key risk factor. Therefore, we aim to evaluate and compare the predictive value of indirect bone quality metrics, Hounsfield units and vertebral bone quality (VBQ) scores, for cage subsidence following spinal fusion. Standard medical databases were searched to identify studies comparing HU or VBQ scores between patients with and without subsidence. Standard mean differences (SMD), diagnostic odds ratios (DOR), sensitivity, specificity, and area under the curve (AUC) were pooled. Meta-regression was performed to assess the influence of covariates. A total of 28 studies with 3,729 patients were included. Patients with subsidence had significantly lower HU (SMD: -1.22; 95% CI: [-1.5, -0.95], Z = 8.82, P < 0.001) and higher VBQ scores (SMD: 1.14; 95% CI: [0.91, 1.37], P < 0.001). Pooled diagnostic performance was high for both HU (AUC: 0.88, DOR: 19.3) and VBQ (AUC: 0.84, DOR: 14.8). Sensitivity and specificity were comparable and exceeded 79% for both metrics. Meta-regression identified no significant effect modifiers in multivariate models. Both HU and VBQ scores demonstrate strong and comparable diagnostic accuracy for predicting cage subsidence. While HU seemed to exhibit slight superiority, VBQ provides a radiation-free, cost-effective alternative with comparable performance. Incorporating these metrics into preoperative planning may improve risk stratification and guide surgical planning in spinal fusion.