The Risk of Low Energy Vertebral Body Compression Fracture Can be Predicted by CT Hounsfield Units

Abstract

Introduction Vertebral compression fractures (VCF) are an increasingly common spinal pathology in the older adult population. Dual emission X-ray absorptiometry (DEXA) is frequently used to assess bone mineral density. Unfortunately, the sensitivity of DEXA in predicting fracture risk has been shown to be relatively poor, leading some to suggest the need for additional diagnostic studies that more accurately predict fracture risk. The objective of this study was to define the CT parameters that could be used to predict the risk of VCF. Materials and Methods A total of 100 consecutive patients undergoing CT scans for the workup of new onset axial back pain were enrolled in this study. CT scans utilized slice thickness of 0.5 mm. Tube voltage was 120 kV, current 300 mA, auto mAs range 180 to 400, 1.0 second/3.0 mm/0.5 × 32, and helical-pitch 21.0. Patients with a history of high-energy trauma or neoplasm were excluded. Bone density measurements were obtained in Hounsfield units (HU) at the T10–L5 levels from the cancellous portion of the vertebral body in the midsagittal, midcoronal, and axial planes and the average value was calculated. The presence of a single-level or multilevel VCF was identified by CT. Multilevel degenerative changes were characterized and recorded. Logistic regression was utilized to assess the relationship between the variables of bone density in HU, single- or multilevel VCF, and the presence of degenerative changes. Results HU were found to have a strong correlation to the risk of VCF. HU of less than 101 were associated with a significant increase in the rate of VCF, the parameters of logistic regression were В0 = 2.6254, p = 0.0013; B1 = − 0.0387, p < 0.0001. Goodness of fit: chi-square = 37.1180; p < 0.0001, percent of correctly predicted 81%. HU of less than 82 were associated with a significant increase in the rate of multilevel VCF, the parameters of logistic regression were В0 = 1.5934, p = − 0.0338; B1 = 0.0355, p = 0.00013. Goodness of fit: chi-square = 27.21870; p < 0.0001, percent of correctly predicted 86%. The analysis of graphs using third derivatives of estimated regression functions showed that bone density of 101 HU or less is associated with an increased risk of VCF, and bone density of less than 82 HU is associated with an increased risk of multilevel VCF. Hypertrophic degenerative changes were found to be associated with a decreased rate of VCF, the parameters of logistic regression were chi-square = 8.976428, p = 0.0027; В0 = 0.3514, р = 0.2434 (not significant); B1 = 1.3978, p = 0.0050, odds ratio = 4.04 [1.54–10.62]. Conclusion CT data can accurately define the risk of VCF and, therefore, presents a useful clinical tool to support the need for prophylactic medical therapies for osteoporosis or to provide information useful in counseling patients at risk for VCF. Although highly predictive, routine use of CT as a screening tool for osteoporosis is unsuitable because of the associated radiation exposure and cost of the study; however, when CT scans are obtained for other clinical indications, they offer the clinician a unique opportunity to easily obtain useful data that can be used to direct treatment and counsel patients with regard to fracture risk.