1. 2021;24;E493-E500Evaluation of the Use of CT Attenuation for the Prediction of Subsequent Vertebral Fracture in Patients with Osteoporosis
   Retrospective Review
   De-sheng Wu, PhD, Shan-Jin Wang, PhD, Tao Hu, PhD, Yu-yang Yi, MD, Shu-bao Zhang, MD, and Hao-Wei Xu, MD.

BACKGROUND: Subsequent vertebral fracture (SVF) is one of the most common complications of percutaneous vertebral augmentation (PVA), which leads to lower back pain in patients. Low bone mineral density (BMD) is an independent risk factor for SVF. BMD measured using computed tomography (CT) trabecular attenuation correlates closely with BMD.

OBJECTIVES: This study aims to analyze the risk factors of SVF after PVA and to estimate the predictive role of CT trabecular attenuation.

STUDY DESIGN: A retrospective review.

SETTING: Department of spinal surgery in an affiliated hospital of a medical university.

METHODS: A total of 515 patients were retrospectively enrolled between January 2015 and December 2019 into a 5-year follow-up investigation. Trabecular attenuation (Hounsfield units [HU]) was retrospectively measured at L1 on preoperative lumbar or thoracic CT scans, and the receiver operating characteristic (ROC) curve was used to evaluate its value for the prediction of SVF. Kaplan–Meier analysis and Cox proportional hazards regression were performed to identify the risk factors for SVF.

RESULTS: A total of 166 patients (32.2%) experienced SVF. ROC curve analysis demonstrated that an L1 trabecular attenuation of <= 95 HU has a sensitivity of 70.5% and a specificity of 79.9% for the prediction of SVF. Kaplan–Meier analysis showed that L1 trabecular attenuation <= 95 HU was significantly associated with lower SVF-free survival (P = 0.001; log-rank test). Multivariate analysis demonstrated that advanced age (hazard ratio [HR] = 1.03, P = 0.022), low body mass index (HR = 0.83, P = 0.001), diabetes status (HR = 1.50, P = 0.024), antiosteoporosis drugs use (HR = 0.65, P = 0.031), and decreased L1 trabecular attenuation (HR = 0.95, P = 0.001) were risk factors for SVF.

LIMITATIONS: A single-center retrospective study of a consecutive cohort of patients may include the inevitable bias. We periodically reviewed the full-length x-ray of the spine at every 3 months of follow-up visit, which we may miss some patients with SVF without low back pain.

CONCLUSIONS: SVF is highly prevalent in patients with osteoporotic vertebral fracture who undergo single-level PVA. Low L1 trabecular attenuation is associated with a significant reduction in SVF-free survival, and when their L1 trabecular attenuation is <= 95 HU, patients may be at higher risk of SVF.

KEY WORDS: Computed tomography, Hounsfield units, vertebral fracture, osteoporosis, percutaneous vertebral augmentation