Muscle fat-free area on lumbar MRI outperforms VBQ score for osteoporosis detection

The work, published in BMC Musculoskeletal Disorders, compared the two imaging biomarkers head to head on routinely acquired lumbar scans and tested whether combining them improved discrimination beyond standard clinical variables.

Osteoporosis is conventionally diagnosed with DXA, which reports bone mineral density as T-scores at the femoral neck, total hip, and lumbar spine. In this study, the lowest T-score was defined as the minimum of the femoral neck, total hip, and individual L1–L4 T-scores, and osteoporosis was defined as a lowest T-score of −2.5 or below.

The imaging protocol centered on two families of measurements. The second comprised quantitative paraspinal muscle parameters measured on axial T2-weighted MRI: cross-sectional area, fat-free cross-sectional area, and fat fraction of the erector spinae plus multifidus complex and of the psoas major.

Patients with osteoporosis had higher vertebral bone quality scores, a difference that reached statistical significance at P = 0.019, and significantly lower cross-sectional and fat-free cross-sectional areas of both muscle groups, with all comparisons at P < 0.001. Correlation analysis reinforced the pattern. The lowest DXA T-score showed only a weak inverse correlation with the vertebral bone quality score, at ρ = −0.273, but moderate positive correlations with muscle cross-sectional and fat-free areas, the strongest being psoas major fat-free area at ρ = 0.518. The fat-free areas of the erector spinae plus multifidus and the psoas major were strongly correlated with each other, at ρ = 0.768, suggesting the two muscle groups carry overlapping information about bone status.

Multivariable logistic regression sharpened the contrast between the biomarkers. After adjustment for age, sex, and body mass index, only the fat-free cross-sectional area parameters remained independently associated with osteoporosis, whereas the vertebral bone quality score was no longer independently associated. This indicates that much of the apparent relationship between the bone score and osteoporosis in unadjusted analyses could be accounted for by basic demographic and anthropometric variables, while the muscle measure retained its association on its own.

Receiver operating characteristic analysis quantified the diagnostic gap. Psoas major fat-free area achieved an area under the curve of 0.761 and erector spinae plus multifidus fat-free area 0.759, both clearly better than the vertebral bone quality score at 0.611. Pairwise comparisons were performed with the DeLong test, and internal validation used bootstrap resampling. The combined model pairing the vertebral bone quality score with psoas major fat-free area produced the highest area under the curve among imaging-based models, 0.785, but this did not differ significantly from psoas major fat-free area alone, at P = 0.328, and bootstrap validation yielded an optimism-corrected area of 0.776.

The clinical variables themselves were informative. A basic model containing age, sex, and body mass index achieved an area under the curve of 0.768. Adding psoas major fat-free cross-sectional area to that model raised the area to 0.791, a statistically significant improvement by likelihood-ratio testing at P = 0.002. Adding the vertebral bone quality score to the same clinical model did not produce a significant improvement, at P = 0.098. In a separate likelihood-ratio test, adding the vertebral bone quality score to psoas major fat-free area alone did significantly improve model fit, at P = 0.002, and the combined model significantly outperformed the vertebral bone quality score alone, with a DeLong P < 0.001.

The authors conclude that paraspinal muscle fat-free cross-sectional area, particularly of the psoas major, was independently associated with osteoporosis and demonstrated higher diagnostic performance than the vertebral bone quality score. Although the combined models showed numerically higher areas under the curve, the incremental value of adding the vertebral bone quality score to fat-free area was not statistically significant, and whether these biomarkers provide complementary information requires further validation. They position fat-free cross-sectional area as a potentially useful MRI-derived quantitative imaging biomarker for osteoporosis identification using routinely acquired axial T2-weighted lumbar MRI.

Several caveats frame the findings. The study was retrospective and cross-sectional, so it establishes associations and diagnostic discrimination within a single cohort rather than causal effects or prospective screening performance. The sample of 150 patients is modest, and the reported optimism-corrected area under the curve of 0.776 for the combined model reflects internal validation by bootstrap resampling rather than external testing in an independent population. Measurement reliability was assessed with intra- and inter-observer intraclass correlation coefficients, but the analysis relied on quantitative muscle measurements that require dedicated post-processing of axial images. The authors also note that the version shared ahead of publication is citable and carries a permanent DOI but remains subject to further edits before the final Version of Record appears.

For clinicians, the practical message is that muscle composition on lumbar MRI may carry more diagnostic signal for osteoporosis than the vertebral bone quality score in this cohort, and that a single psoas measurement added meaningful discrimination to age, sex, and body mass index. Whether fat-free cross-sectional area can shorten the path from an incidental MRI finding to a DXA referral in routine practice will depend on validation in larger, independent, and prospectively recruited cohorts.

Subject of Research: Comparing paraspinal muscle fat-free cross-sectional area with the vertebral bone quality score for identifying osteoporosis on lumbar MRI

Article Title: Diagnostic performance of paraspinal muscle fat-free cross-sectional area versus vertebral bone quality score for osteoporosis identification on lumbar MRI

Article References: Zhang, Y., Ding, G., Li, Z., Yan, F., Wen, J., Wang, Z., & Zhang, H. (2026). Diagnostic performance of paraspinal muscle fat-free cross-sectional area versus vertebral bone quality score for osteoporosis identification on lumbar MRI. BMC Musculoskeletal Disorders. https://doi.org/10.1186/s12891-026-10536-2

Image Credits: AI Generated

DOI: 10.1186/s12891-026-10536-2

Keywords: osteoporosis, lumbar MRI, vertebral bone quality score, paraspinal muscles, fat-free cross-sectional area, psoas major, DXA, T-score, ROC analysis, imaging biomarker, BMC Musculoskeletal Disorders, Diagnostic