PURPOSE: To evaluate the relationship between preoperative imaging features of endplate degeneration and incomplete fusion after LLIF and to establish a multivariable predictive model for preoperative risk stratification.
METHODS: Patients who underwent LLIF between January 2021 and January 2025 were retrospectively reviewed. Preoperative MRI and CT were used to evaluate imaging features of endplate degeneration, including Modic classification, Schmorl's node, endplate cartilage erosion, sub-endplate trabecular sclerosis, and bony endplate Hounsfield unit (HU). Univariate and multivariate logistic regression analyses were performed to identify independent risk factors for incomplete fusion. A nomogram was constructed based on the final model, and its performance was evaluated using receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis (DCA).
RESULTS: The mean age was 59.7 ± 9.5 years. At the 12-month follow-up, incomplete fusion, defined as BSF grade 1 pseudarthrosis or BSF grade 2 partial fusion, was observed in 33.3% (84/252) of patients. Multivariate analysis identified smoking history (OR 2.623, 95% CI 1.187-5.799, p = 0.017), osteoporosis (OR 2.231, 95% CI 1.002-4.967, p = 0.049), sub-endplate trabecular sclerosis (OR 7.598, 95% CI 1.498-17.545, p = 0.014), Modic type III changes (OR 7.290, 95% CI 1.853-16.679, p = 0.025), and elevated bony endplate HU (OR 5.454, 95% CI 1.032-15.822, p = 0.046) as independent predictors of incomplete fusion. The predictive model demonstrated good discrimination (AUC = 0.883).
CONCLUSIONS: Imaging features of endplate degeneration, particularly sub-endplate trabecular sclerosis, Modic type III changes, and elevated bony endplate HU, are independent predictors of incomplete fusion after LLIF. Incorporating these local structural factors with systemic risk factors may improve preoperative risk assessment.
Read the article: PubMed · Publisher (DOI)