Computed Tomography-Based Prediction of Lumbar Pedicle Screw Loosening.

Abstract

Pedicle screw loosening is one of the main complications after pedicle screw fixation. However, there are few reliable measures for prediction of screw loosening. The current study was carried out to find an effective method to use preoperative CT scanning as a predictor of screw loosening in the elderly patients and provide guidance for preoperative surgical planning. Patients who were treated with lumbar pedicle screw fixation procedure in our department for degenerative lumbar disorders between January 2015 and January 2021 were retrospectively included in the current study. CT scan attenuation of each vertebra was measured with Hounsfield units (HU). Screw loosening was determined in postoperatively X-ray tests. One-way analysis of variance (ANOVA) and receiver operating characteristic (ROC) curve analysis were carried out with IBMSPSS 24.00 software. Screw loosening was observed in 44 of 215 patients (124 male, 91 female, average age 58.4 ± 7.6 years) during a mean follow-up time of 19.0 ± 11.2 months (range 12-32 months). No significant differences were found among the patients concerning patient gender, BMI, habit of smoking, and whether or not the patient had diabetes or suffered from spondylolisthesis (P > 0.05). The average HU value of lumbar vertebra was 122.4 ± 32.8 HU in the screw loosening group and 142.4 ± 38.2 HU in the control group, and the difference was significant (P < 0.01). ROC curve analysis revealed that the average HU value of L1-L5 has a relatively larger area under the curve (AUC) of 0.689 (95% CI: 0.605-0.773). With the sensitivity of 68% and specificity of 57%, a HU cut-off value of ≤124 HU is a plausible cut-off point to predict screw loosening. A prospective CT scan HU value-based prediction can be used to decide whether or not to use screw augmentation methods. A cut-off L1-L5 average HU value of 124 HU can be used as an independent risk factor for screw loosening in instrumented lumbar vertebra. More predictive indexes should be involved to achieve higher sensitivity and specificity in future clinical practice.