Bone density measured on sagittal reconstructed CT is highly correlated with axial CT but both measurements are only moderately correlated with DEXA T-scores

Abstract

Background

During the preoperative evaluation of a patient being considered for spinal surgery, Dual-energy x-ray absorptiometry (DEXA) has been traditionally used to diagnose poor bone mineral density (BMD) as a risk factor. As ordering a DEXA can add cost and delay diagnosis, spine surgeons have more recently began to use Hounsfield Units (HU) measured on computed tomography scans (CT) as a measure of BMD. The aim of our study was to evaluate associations between DEXA and HU on lumbar spine CT scans.

Methods

Forty-two patients (32 female, 10 male, mean age = 67.7 years) with DEXA and lumbar spine CT scans performed within one year of each other were identified. DEXA T-scores were collected from the hip, forearm and L1-L4. HU was determined using the maximum region of interest within the cancellous area in the mid-vertebral body from L1-L4 in the sagittal and axial planes.

Results

Using the lowest T-score, 8 (19 %) cases were osteoporotic and 25 (60 %) were osteopenic. Statistically significant differences in HU were seen in osteoporotic cases (Axial HU = 59.2, Sagittal HU = 61.1, p = 0.006) compared to osteopenic (Axial HU = 119.8, Sagittal HU = 122.9) and normal cases (Axial HU = 141.2, Sagittal HU = 142.3). There were moderate associations between the spine T-scores and CT HUs (Axial HU:r² = 0.50, Sagittal HU:r² = 0.49, p < 0.001), weak associations between the Axial HU (r² = 0.48, p < 0.000) and Sagittal HU (r² = 0.48, p < 0.000) with hip T-scores, and no correlations with forearm T-scores. There were strong associations between Axial HU and Sagittal HU (r² = 0.98, p < 0.001).

Clinical relevance

The results of the current study show a strong association between the sagittal and axial vertebral HU measurements, which supports the clinical use of either measurement technique. The weak correlation between T-scores and HU is consistent with prior studies and warrants future studies to determine which modality will better predict postoperative mechanical failures in patients undergoing spinal surgery.

Level of Evidence

III.