Lumbar spine trabecular bone score (TBS), a gray-level texture measure derived from spine dual-energy X-ray absorptiometry (DXA) images, is a bone mineral density (BMD)-independent risk factor for fracture. An unresolved question is whether TBS is sufficiently responsive to change over time or in response to widely used osteoporosis therapy at the individual level to serve as a useful biomarker. Using the Manitoba DXA Registry, we identified 11,643 individuals age 40 years and older with two fan-beam DXA scans performed on the same instrument within 5 years (mean interval 3.2 years), of whom 6985 (60.0%) received antiresorptive osteoporosis medication (majority oral bisphosphonate) between the scans. We examined factors that were associated with a change in lumbar spine TBS, lumbar spine BMD, and total hip BMD exceeding the 95% least significant change (LSC). Change exceeding the LSC was identified in 23.0% (9.3% increase, 13.8% decrease) of lumbar spine TBS, 38.2% (22.1% increase, 16.1% decrease) lumbar spine BMD, and 42.5% (17.6% increase, 24.9% decrease) total hip BMD measurement pairs. From regression models, the variables most strongly associated with significant change in TBS (decreasing order) were tissue thickness change, acquisition mode change, weight change, and spine percent fat change. Consistent with the insensitivity of TBS to oral antiresorptive therapies, use of these agents showed very little effect on TBS change. In contrast, for both spine BMD change and total hip BMD change, osteoporosis medication use was the most significant variable, whereas tissue thickness change, acquisition mode change, and weight change had relatively weak effects. In summary, change in spine TBS using the present algorithm appears to be strongly affected by technical factors. This suggests a limited role, if any, for using TBS change in untreated individuals or for monitoring response to antiresorptive treatment in routine clinical practice with the current version of the TBS algorithm.
