Novel MRI-based vertebral bone quality score as a predictor of cage subsidence following transforaminal lumbar interbody fusion

Abstract

OBJECTIVE Decreased bone mineral density as measured by dual-energy x-ray absorptiometry (DEXA) has been reported to be associated with cage subsidence following transforaminal lumbar interbody fusion (TLIF). However, DEXA is not often available or routinely performed before surgery. A novel MRI-based vertebral bone quality (VBQ) score has been developed and reported to be correlated with DEXA T-scores. The authors investigated the ability of the VBQ score to predict cage subsidence and other risk factors associated with this complication. METHODS In this retrospective study, the authors reviewed the records of patients who had undergone single-level TLIF from March 2014 to October 2015 and had a follow-up of more than 2 years. Cage subsidence was measured as postoperative disc height loss and was graded according to the system proposed by Marchi et al. The MRI-based VBQ score was measured on T1-weighted images. Univariable analysis and multivariable binary logistic regression analysis were performed. Ad hoc analysis with receiver operating characteristic curve analysis was performed to assess the predictive ability of the significant continuous variables. Additional analyses were used to determine the correlations between the VBQ score and T-scores and between the significant continuous variables and the amount of cage subsidence. RESULTS Among 242 patients eligible for study inclusion, 111 (45.87%) had cage subsidence after the index operation. Multivariable logistic regression analyses demonstrated that an increased VBQ score (OR 14.615 ± 0.377, p < 0.001), decreased depth ratio (OR 0.011 ± 1.796, p = 0.013), and the use of kidney-shaped cages instead of bullet-shaped cages (OR 2.766 ± 0.358, p = 0.008) were associated with increased cage subsidence. The VBQ score was shown to significantly predict cage subsidence with an accuracy of 85.6%. The VBQ score was found to be moderately correlated with DEXA T-scores of the total hip (r = −0.540, p < 0.001) and the lumbar spine (r = −0.546, p < 0.001). The amount of cage subsidence was moderately correlated with the VBQ score (r = 0.512, p < 0.001). CONCLUSIONS Increased VBQ scores, posteriorly placed cages, and kidney-shaped cages were risk factors for cage subsidence. The VBQ score was shown to be a good predictor of cage subsidence, was moderately correlated with DEXA T-scores for the total hip and lumbar spine, and also had a moderate correlation with the amount of cage subsidence.