Application of CT attenuation value in prediction of screw loosening after short-segment lumbar pedicle screw fixation

English Keywords:Screw loosening  CT attenuation value  Osteoporosis

English Abstract:

  【Abstract】 Objectives: To explore the relationship between vertebral CT attenuation value and screw loosening after short-segment lumbar pedicle screw fixation, and to identify thresholds of bone density in HU for predicting screw loosening. Methods: The clinical data of patients who underwent short-segment lumbar pedicle screw fixation in our spinal center from July 2006 to June 2015 were retrospectively analyzed. The inclusion criteria included: (1)no more than 2 segments; (2)at least 2-year follow-up; (3)three dimensional reconstructive CT was done within 30 days before operation. 297 patients(104 males, 193 females) with a mean age of 54.3±12.5 years(21-80 years) and a mean follow-up of 36.1±16.5 months(24-110 months) were reviewed. Screw loosening and bone fusion with lumbar X-ray of final follow-up were assessed. According to the occurrence of screw loosening at UIV(upper instrumented vertebra) or LIV(lowest instrumented vertebra), patients were divided into screw loosening groups (screw loosening at UIV; screw loosening at LIV) and corresponding control groups. Moreover, the LIVs were divided on whether or not the LIV was at S1. The CT attenuation values of L1, UIV, LIV and S1 were measured. Patient data included age, gender, body mass index(BMI), history of diabetes, number of instrumented vertebrae, fusion methods at marginal segments, LIV at S1. According to the occurrence of screw loosening at UIV or LIV, patients were divided into screw loosening groups (screw loosening at UIV; screw loosening at LIV) and corresponding control groups. Inter-examination and inter-observer reliability were assessed by using the intraclass correlation coefficient(ICC). The role of CT attenuation values in screw loosening was explored with Logistic regression analysis. Receiver-operating characteristic curve analysis was used to evaluate the value of CT attenuation values in predicting screw loosening. Because of the relatively small sample size and non-normal distribution of CT attenuation value in screw loosening group, the median rather than the average of the CT attenuation value in screw loosening group was chosen to be the threshold forpredicting screw loosening. Results: Fifty-three patients suffered from postoperative screw loosening(17.8%, 53/297). 21 patients suffered from screw loosening at UIV, 276 patients were in the corresponding control group. 48 patients suffered from screw loosening at LIV, 249 patients were in the corresponding control group. 24 patients suffered from nonunion, the rate of union at upper interbody was 93.6%(278/297), and the rate of union at lowest interbody was 93.3%(273/297). Inter-examination and inter-observer reliability were excellent(ICC>0.8, P<0.001). There were statistically significant differences of CT attenuation values between screw loosening groups and control groups: the mean CT attenuation value of UIV was 87.3±41.9HU in the UIV-screw loosening group and 140.5±55.9HU in corresponding control group, P<0.05; the mean CT attenuation value of LIV at lumbar vertebrae was 121.9±39.9HU in the LIV screw loosening group and 152.2±54.5HU in corresponding control group, P<0.05; the mean CT attenuation value of LIV at S1 was 216.4±61.1HU in the LIV screw loosening group and 254.8±81.7HU in corresponding control group, P<0.05. The CT attenuation value was the independent risk factor of screw loosening. The CT attenuation value of marginal vertebrae was the independent risk factor of screw loosening at marginal vertebrae(UIV-CT attenuation value: OR, 0.979; 95%CI, 0.967-0.992; LIV-CT attenuation value: OR, 0.990; 95%CI, 0.983-0.998). The application of CT attenuation value in prediction of screw loosening was acceptable(AUC>0.6, P<0.05), the median CT attenuation value in UIV-screw loosening group, LIV-lumbar-screw loosening group and LIV-S1-screw loosening group was 75HU, 110HU, 220HU respectively. Conclusions: The CT attenuation value of marginal vertebrae is the independent influencing factor of screw loosening at marginal vertebrae, the lower the CT attenuation value, the higher the risk of screw loosening.