Risk factors of cortical bone trajectory screw loosening after single segment posterior lumbar interbody fusion and construction and validation of a predictive model

English Keywords:Posterior lumbar interbody fusion  Screw loosening  Cortical bone trajectory  Nomogram  Coronal angle  Cortical bone contact layer

English Abstract:

  【Abstract】 Objectives: To investigate and analyze the risk factors of screw loosening after single segment posterior lumbar interbody fusion(PLIF) with cortical bone trajectory(CBT) screw, and to establish and validate a visualized nomogram predictive model. Methods: A retrospective study was conducted on 102 patients undergone single-segment PLIF in our hospital from March 2020 to June 2023, and a total of 357 CBT screws were placed, averaged 3.5±0.3 per patient. The postoperative follow-up time was 2.0-35.0 months, with a median of 22.5 months. On the basis of lumbar CT scans, screw loosening was defined as a continuous transparent area exceeding 1mm, surrounded with thin sclerotic areas. The patients were divided into loosening group of 45 cases and non-loosening group of 57 cases, with 117 screws loosening and 240 screws non-loosening. The demographic data[gender, age, body mass index(BMI), bone mineral density(BMD)], operative indicators(operative time, blood loss, lumbar fusion grade, Oswestry disability index), and radiological parameters[screw fixation to S1(FS1), screw trabecular Hounsfield unit(HU), screw sagittal angle(SA), coronary angle(CA), and cortical bone contact layer(CBCL)] were compared between groups. Lasso regression and multivariate logistic regression models were used to screen the optimal independent risk factors to screw loosening, and then a nomogram predictive model was constructed with R software. The internal predictive stability[consistency index(C-index)], prediction differentiation(sensitivity and specificity), prediction probability accuracy(degree of agreement), and net clinical benefit of the model was evaluated with Bootstrap method, receiver operating curve(ROC), calibration curve, and decision curve analysis(DCA). Results: The BMD in loosening group was significantly less than that in non-loosening group(P<0.05), and there were no statistical differences in the other demographic data and surgical indexes between two groups(P>0.05). Compared with non-loosening group, the FS1 percentage and CA in the loosening group significantly increased, while CBCL significantly decreased(P<0.05). BMD(OR=0.44, 95%CI=0.21-0.89, P<0.001), FS1(OR=3.12, 95%CI=2.03-4.54, P<0.001), CA(OR=1.45, 95%CI=1.12-2.03, P<0.001), and CBCL(OR=0.67, 95%CI=0.32-0.88, P<0.001) were the independent risk factors of screw loosening after single-segment PLIF. The total score of the model was 140 points. The bootstrap method calculated the C-index of 0.856(P=0.125), ROC showed area under curve(AUC) of 0.847(95%CI=0.801-0.912, P<0.001). The calibration curve and DCA both showed good consistency in the model′s predictive result and clinical net benefit ratio. Conclusions: The BMD, FS1, CA, and CBCL are the independent risk factors for screw loosening after single-segment PLIF with CBT screws, and the nomogram model constructed has certain application values in early clinical screening of high-risk patients for screw loosening.