Vertebral computed tomography attenuation to case-find low bone density in people undergoing lung cancer screening

Lung cancer screening (LCS) participants are "high-risk" for osteoporotic fracture. We compared lumbar vertebral Hounsfield-unit CT-attenuation (VHU) from LCS images to standard dual-energy X-ray absorptiometry (DXA), quantitative CT (QCT) and clinical fracture risk tools to identify participants with low bone mineral density (BMD), prevalent vertebral fracture (VF) and incident osteoporotic fracture.

LCS participants were cross-sectionally analysed at a 2-year follow-up timepoint. BMD was assessed from paired LCS chest low-dose CT (VHU and spine QCT) and DXA scans (T-scores). FRAX and Garvan 10-year fracture risks were calculated from questionnaires. VF were assessed semi-quantitatively from CT scans; 2-year incident VF were identified by comparing baseline and 2-year follow-up LCS CT scans. Participants self-reported 2-year incident osteoporotic fractures. Method discrimination for low DXA-BMD (T-score<-1), prevalent VF and incident fractures were compared.

In 91 participants (M: F ratio 0.75), VHU moderately discriminated low DXA-BMD moderately (AUC 0.74), similar to QCT (AUC 0.73), FRAX (AUC 0.82) and Garvan (AUC 0.78)(p > 0.05). VHU performed similarly to DXA spine T-score, QCT, FRAX and Garvan in discrimination for both prevalent VF (AUC 0.71) and incident fractures (AUC 0.71). Correlation and inter-method reliability between VHU and DXA-T-scores were fair (R = 0.42, ĸ < 0.3).

In LCS, VHU is a convenient alternative to case-find screenees with occult osteoporosis, having comparable discrimination to standard methods for low BMD and osteoporotic fractures.

Within-population comparison of osteoporosis assessment methods, shows VHU can comparably 'case-find' LCS participants with occult osteoporotic disease and incident fractures, warranting larger prospective studies.