Assessment of paraspinal muscle fat infiltration using Hounsfield units in patients with lumbar degenerative disease: validation against MRI and handgrip strength

Paraspinal muscle fat infiltration (FI) worsens clinical outcomes in patients with lumbar degenerative diseases. This study aimed to evaluate the ability of Hounsfield Units (HU) to assess FI, using MRI and grip strength as reference standards, and to examine intra- and inter-observer reliability and sex-related differences. We retrospectively reviewed patient records who underwent elective lumbar surgery for lumbar degenerative disease. FI was measured on preoperative MRI using the Image J threshold function, while HU was obtained from preoperative CT scans using the Picture Archiving and Communication System to outline regions of interest in the paraspinal muscles. Handgrip strength was measured using a spring dynamometer. Restricted cubic spline was used to detect potential nonlinear relationships. Multivariable linear regression was performed to identify factors influencing HU. Propensity score matching (PSM) was applied to explore sex differences in HU at comparable levels of FI. A total of 252 independent patients (121 males and 131 females) were included in this study, with handgrip strength data available for a subset of 102 (40.48%) patients from the same cohort. Female patients exhibited significantly higher FI, lower paraspinal muscle HU, and a smaller functional cross-sectional area of the lumbar muscles compared to male patients. The inter-observer reliability of HU was higher than that of FI (HU: 0.980 vs. FI: 0.891). Paraspinal muscle HU was strongly negatively correlated with FI (r = − 0.82) and significantly positively correlated with handgrip strength (r = 0.58). Multivariable linear regression showed that sex and FI were independent determinants of HU. After accounting for the independent effect of sex on HU, PSM was performed for 76 male–female pairs with comparable levels of FI. HU was significantly higher in males than in females (33.64 vs. 21.76 HU; mean difference − 11.88 HU (95% CI: − 15.38 to − 8.38 HU, P < 0.001). HU was significantly correlated with MRI-derived FI and handgrip strength, while demonstrating higher inter-observer reliability and greater sensitivity to sex-related differences than conventional FI measurements. Future studies should further establish and validate sex-specific HU thresholds using clinically relevant outcomes.