Background:
Osteoporosis is under-diagnosed in Australia, with 75% of subjects at risk of fragility fractures are not investigated or treated (1). Opportunistic osteoporosis screening in patients undergoing computed tomography (CT) scans can potentially avoid additional imaging and optimise patient convenience (2). We aim to assess the diagnostic accuracy of L1 vertebral HU measurement on abdominal CT scans with L1 DEXA T-score as the reference standard.
Methods:
Consecutive patients who underwent abdominal CT were included if they underwent DEXA scans within a 12-month period. Density values (HU) from CT at L1 were measured in the axial plane over a central 2 cm area of L1 trabecular bone. Pearson’s correlation coefficient was used to quantify the correlation between L1 density and T-score. T scores were analysed categorically as per WHO definitions using different HU thresholds.
Results:
The study comprised 460 patients (42.8% males, 57.2% females). The prevalence of osteoporosis was 11.5%. CT density correlated with DEXA T-score (r=0.372, p<0.001).The AUC across CT density thresholds at L1 to distinguish osteoporosis from osteopenia and normal BMD was 0.65 for the entire cohort. In females, using a threshold of 190 HU detected T-scores ≤ -2.5 with a NPV of 94.8%, O.R. = 4.5, p < 0.01) and T-scores ≤ -1.5 with a NPV of 80.5%, O.R. = 4.1, p < 0.01). In males, a threshold of 180 HU detected T-scores ≤ -2.5 with a NPV of 100%, O.R. = 10.1, p < 0.01) and T-scores ≤ -1.5 with a NPV of 86.8%, O.R. = 2.6, p = 0.03).
Conclusion:
L1 HU values less than 190 and 180 increased the odds of osteoporosis diagnosis in an Australian female and male cohort. The use of abdominal CT to screen for osteoporosis is feasible and can improve the diagnosis rate of osteoporosis leading to reduced fracture risk.