Adding liver R2 quantification to proton density fat fraction MRI of vertebral bone marrow improves the prediction of osteoporosis

• Published in: European radiology 2022-10, Vol.32 (10), p.7108-7116
• Main Authors: Lu, Feng, Zhao, Yan-Jun, Ni, Jian-Ming, Jiang, Yu, Chen, Fang-Ming, Wang, Zhong-Juan, Zhang, Zhui-Yang
• Format: Article
• Language: English
• Subjects: Biomarkers; Bone density; Bone marrow; Bone mass; Chemical equilibrium; Diagnostic Radiology; Dual energy X-ray absorptiometry; Imaging; Internal Medicine; Interventional Radiology; Liver; Magnetic Resonance; Magnetic resonance imaging; Medicine; Medicine & Public Health; Menopause; Neuroradiology; Osteopenia; Osteoporosis; Predictions; Proton density (concentration); Radiology; Reclassification; Risk analysis; Risk factors; Ultrasound; Vertebrae
• ISSN: 1432-1084; 0938-7994; 1432-1084
• DOI: 10.1007/s00330-022-08861-w

Objectives To assess the predictive value of the combination of bone marrow (BM) proton density fat fraction (PDFF) and liver R2* for osteopenia and osteoporosis and the additional role of liver R2*. Methods A total of 107 healthy women were included between June 2019 and January 2021. Each participant underwent dual-energy X-ray absorptiometry (DXA) and chemical shift-encoded 3.0-T MRI. PDFF measurements were performed for each lumbar vertebral body, and R2* measurements were performed in liver segments. Agreement among measurements was assessed by Bland-Altman analysis. Receiver operating characteristic (ROC) curves were generated to select optimised cut-offs for BM PDFF and liver R2*. Univariable and multivariable logistic regressions were performed. The C statistic and continuous net reclassification improvement (NRI) were adopted to explore the incremental predictive ability of liver R2*. Results Bone mass decreased in 42 cases (39.3%) and nonbone mass decreased in 65 cases (60.7%). There were significant differences among the age groups, menopausal status groups, PDFF > 45.0% groups, and R2* > 67.7 groups. Each measurement had good reproducibility. The odds ratios (95% CIs) were 4.05 (1.22–13.43) for PDFF and 4.34 (1.41–13.35) for R2*. The C statistic (95% CI) without R2* was 0.888 (0.827–0.950), and with R2* was 0.900 (0.841–0.960). The NRI resulting from the combination of PDFF and R2* was 75.6% ( p < 0.01). Conclusion The predictive improvement over the use of BM PDFF and other traditional risk factors demonstrates the potential of liver R2* as a biomarker for osteopenia and osteoporosis in healthy women. Key Points • Liver R2* is a biomarker for the assessment of osteopenia and osteoporosis. • Liver R2* improved the ability to predict osteopenia and osteoporosis. • The intra- and interobserver measurements showed high agreement.