Artifactual effects of bone area from dual-energy X-ray absorptiometry (DXA) confound bone mineral density (BMD) measurements and in turn complicate fracture risk assessment. Trabecular bone score (TBS) from the lumbar spine DXA image was found to be minimally affected by bone area, which helps to improve fracture risk assessment.
PURPOSE: We hypothesized that TBS is less sensitive to the effects of skeletal size than lumbar spine BMD. Our aim was to compare the effects of DXA-derived lumbar spine bone area on lumbar spine TBS and BMD, and how this affects subsequent fracture risk.
METHODS: We included individuals aged  ≥  40 years with baseline lumbar spine TBS and BMD. Lumbar spine bone area was stratified into sex-specific quintiles. Least squares mean (LSM) TBS and BMD were estimated according to lumbar spine bone size quintile. Models were adjusted for baseline covariates. Incident major osteoporotic fractures (MOF) were assessed from population-based linked healthcare data. The effect of TBS on lumbar spine bone area was assessed by comparing the change in model chi-square before and after including TBS in Cox regression models.
RESULTS: The study population included 56,586 individuals (mean age 63 years, 90% female). Increasing lumbar spine bone area quintile was associated with greater BMD (smallest-largest quintile difference 13.2%), while TBS showed minimal sensitivity to bone area (difference 0.6%). These findings were largely unchanged when fully adjusted for multiple covariates (smallest-largest quintile difference 11.3% for BMD, 0.3% for TBS). TBS was significantly associated with incident MOF in all regression models independent from other covariates, including bone area and BMD. Adjusting for TBS reduced but did not eliminate the effect of bone area.
CONCLUSION: TBS is largely unaffected by variation in DXA-derived lumbar spine bone area. Inclusion of TBS may attenuate the confounding effect of lumbar spine bone area on BMD and fracture risk assessment.