MRI-based Endplate Bone Quality score independently predicts cage subsidence following transforaminal lumbar interbody fusion

Cage subsidence following transforaminal lumbar interbody fusion (TLIF) has closely correlated with poor vertebral bone quality. Studies have shown better predictive value for cage subsidence by measuring bone density at specific site. However, few studies have been performed to examine the relationship between site-specific MRI bone assessment and cage subsidence in patients who have undergone lumbar interbody fusion. The association between MRI-based assessment of endplate bone quality and cage subsidence after TLIF remains unclear.To study the predictive value of MRI-based endplate bone quality (EBQ) score for cage subsidence following TLIF, using QCT bone densitometry as a reference standard.A retrospective study.A total of 280 adult patients undergoing single-segment TLIF for degenerative lumbar spine disease from 2010 to 2020 at our institution who had preoperative T1-weighted MRIs.Cage subsidence, disc height, endplate bone quality (EBQ) score, bone mineral density, fusion rate.The retrospective study reviewed patients who underwent TLIF at one institution between March 2010 and October 2020. Cage subsidence was measured with postoperative lumbar X-rays based on the cage protrusion through into the superior or inferior end plate or both by more than 2 mm. The EBQ score was measured from preoperative T1-weighted MRI in accordance with the previously reported method.Cage subsidence was observed in 42 of the 280 patients. Bone densitometry with quantitative computed tomography was visibly reduced in the subsidence group. The mean EBQ scores of the lumbar endplate bone was 4.3±0.9 in nonsubsidence and 5.0±0.6 in subsidence. On multivariate logistic regression, the difference between the two groups was remarkable. Risk of cage subsidence increases significantly with higher EBQ scores (odds ratio [OR]=2.063, 95% confidence interval [CI] 1.365-3.120, p=.001) and was an independent factor in predicting subsidence after TLIF. On receiver operating characteristic curve, the AUC for the EBQ score was 0.820 (95% confidence interval [CI]: 0.755-0.844) and the most suitable threshold for the EBQ score was 4.730 (sensitivity: 76.2%, specificity: 83.2%).Higher EBQ scores measured on preoperative MRI correlated significantly with cage subsidence following TLIF. Performing EBQ assessment prior to TLIF may be a valid method of predicting the risk of postoperative subsidence.