Validação da atenuação tomográfica de L1 no rastreamento oportuno de osteopenia e osteoporose em estudos convencionais de tomografia computadorizada

Abstract

Objective: To evaluate the diagnostic accuracy of L1 vertebral tomographic attenuation in the opportunistic screening of osteopenia and osteoporosis in Brazilian patients who underwent conventional tomographic scans for other clinical indications. Methods: Observational, analytical, and retrospective study conducted in Rio Grande do Sul, Brazil, at two imaging diagnosis centers. Adult patients who underwent unenhanced computed tomography (CT) studies that included the first lumbar vertebra (L1) and bone mineral densitometry (BMD) studies between the period of 2016 and 2023 were included, with a maximum interval of six months between them. Patients with extensive degenerative changes in L1, vertebral fractures in L1, osteolytic or osteoblastic lesions in L1, patients with surgical metallic materials in the lumbar spine, or CT studies with kilovoltage (kV) other than 120 were excluded. T-scores from BMDs and the average trabecular attenuation of L1 in axial, sagittal, and coronal planes were collected. The average L1 attenuation in each BMD group was compared using the Kruskal- Wallis test. The correlation between BMD T-scores and L1 tomographic attenuation was assessed using Spearman's correlation coefficient (ρ). ROC curves were obtained to determine the areas under the curve (AUC), diagnostic cutoff points and their diagnostic accuracies. The interobserver agreement was assessed using the intraclass correlation coefficient (ICC). Results: Out of a total of 115,509 CTs and 16,090 BMDs evaluated, 500 patients met the inclusion criteria (464 women and 36 men; mean age: 63.6 ± 10.6 years), with a mean interval between exams of 63.2 days. The prevalence of osteoporosis in the sample was 25.8%. The mean L1 attenuation on CT was significantly lower in osteopenic and osteoporotic patients compared to those with normal bone density on BMD in all tomographic planes (p < 0.001). There was a positive linear correlation between densitometric T-scores and the average L1 attenuation in all evaluation planes (p < 0.001). The sagittal plane showed the highest correlation coefficient with densitometric T-scores (ρ = 0.573; p < 0.001), the largest AUC in the identification of osteoporosis (AUC = 0.785; p < 0.001) and abnormal bone density (AUC: 0.791; p < 0.001), and the highest interobserver agreement (ICC = 0.992; p < 0.001) compared with the axial and coronal planes. An average L1 attenuation equal or below 100 HU in the sagittal plane was able to identify abnormal bone density (osteopenia or osteoporosis) with a specificity of 96,4% and a positive predictive value of 96%, whereas an average attenuation above 180 HU showed a sensitivity of 95% and a negative predictive value of 95,1% to detect osteoporosis; there was significant overlap between the bone density categories among patients with L1 attenuations between 100 and 179 HU on CT. In addition, patients with L1 sagittal attenuation equal or below 100 UH showed significantly higher prevalence of vertebral fractures on CT (prevalence ratio = 4.6; p < 0.001). Conclusion: Opportunistic tomographic evaluation of L1 can identify patients with altered bone density (osteopenia or osteoporosis). Therefore, it allows selecting, without additional cost or radiation, patients who will benefit from formal densitometric assessment.