Purpose: To evaluate whether retrospective analysis of standard abdominal computed tomography
(CT) scans in middle-aged and elderly patients can provide reliable information regarding
decreases in vertebral trabecular bone mineral density (BMD), possible vertebral body
fracture, and stage of its severity.
Methods and Materials: A total of 119 patients (7 men, 112 women) aged 50 to 83 years (mean: 64.6; standard
deviation [SD]: 9.1), who underwent abdominal CT examination as well as a lumbar spine
dual-energy X-ray absorptiometry (DXA) test within 90 days at the Riga East Clinical
University Hospital between 2012 and 2019 were included in this retrospective study.
The exclusion criteria were lumbar spine implants, spinal inflammation, and malignancies.
Lumbar spine trabecular bone radiodensity values were measured in CT scans from levels
L1 to L4. Circular region of interest, 20 mm diameter, was drawn in the axial plane
in the middle of the vertebral body excluding cortical margins. The DXA T-score of
the lumbar spine served as the reference standard. Linear regression analysis and
independent sample Kruskal-Wallis tests were applied to the obtained data.
Results: Using abdominal CT, we found that 21 patients had 1 mild, 8 moderate, and 12 severe
lumbar vertebral body fractures at L1–L4. Of these patients, DXA showed vertebral
height loss in just three patients: one moderate and two severe fracture cases. DXA
sensitivity for vertebral fracture assessment compared with CT was < 0.15. Linear
regression with moderate correlation was found between mean lumbar vertebral body
CT Hounsfield (HU) measurements and mean DXA BMD estimates (correlation coefficient
[r] = 0.63). However, it was different in L1, L2, L3, and L4 vertebrae: 0.69, 0.64,
0.58, and 0.50, respectively. Pairwise comparison analysis for L1–L4 vertebrae showed
no significant difference (p < 0.05) in CT HU measurements (except for the L1 and L3 pair). However, DXA BMD measurements
were significantly different (except for the L3 and L4 pair). CT values for normal
bone density (DXA T-score ≥ 1.0) was ≥ 140 HU [95% confidence interval]). A statistically
significant difference (p < 0.05) was not found between osteopenia (DXA T-score between − 1.0 and − 2.5) and
osteoporosis (DXA T-score ≤ − 2.5) for CT values < 140 HU. The sensitivity, specificity,
positive predictive values, negative predictive values, and accuracy of CT HU for
diagnosing low bone density were 0.89, 0.69, 0.85, 0.76, and 0.82, respectively.
Conclusion: Retrospective analysis of standard abdominal CT scans provides useful and reliable
information regarding vertebral bone density and allows differentiating between normal
and low bone density. This method, avoiding additional irradiation and costs, can
serve as the first screening method for patients with clinically silent spinal osteopenia.
Patients with lumbar spine trabecular bone radiodensity < 140 HU have to undergo an
additional radiologic examination (quantitative CT) to evaluate the severity of decreased
bone density and potential vertebroplasty planning.
