Ultra-processed Foods and Muscle Fat Infiltration at Thigh MRI: Data from the Osteoarthritis Initiative

Background The relationship between high ultra-processed food (UPF) consumption and musculoskeletal health remains underexplored. Purpose To assess the relationship between UPF consumption and thigh muscle fat infiltration (MFI) in a population at risk for knee osteoarthritis (KOA), using thigh MRI. Materials and Methods This cross-sectional secondary analysis uses baseline data from the Osteoarthritis Initiative (February 2004 to October 2015). Participants were at risk for KOA but free of radiographic osteoarthritis (Kellgren-Lawrence grade ≤ 1) and pain in either knee or hip. MFI was assessed using Goutallier grade (grade 0 no fatty streaks to grade 4 >50% fatty signal intensity) on axial T1-weighted spin-echo MRI scans. Standardized sum Goutallier grades were calculated for bilateral flexors, extensors, and adductors and all thigh muscles. Standardized proportion of UPF in daily diet was calculated using the NOVA diet classification and a self-reported food frequency questionnaire for the preceding 12 months. The relationship between standardized UPF consumption and Goutallier grades was tested using linear regression models adjusted for sociodemographic characteristics, total daily calories, smoking, physical activity, depression, and either body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) or abdominal circumference (AC). Interaction between UPF consumption and sex was also assessed. Results A total of 615 participants were included (mean age, 59.5 years ± 9 SD; 340 women; mean BMI, 27 ± 4.3; mean AC, 98.9 cm ± 12.1); UPFs constituted a mean 41.4% ± 13.4 of their diet. In BMI-adjusted models, greater UPF consumption was associated with higher MFI in all thigh muscles (β = 0.108 95% CI: 0.029, 0.186; P = .007), bilateral flexors (β = 0.111 95% CI: 0.030, 0.192; P = .008), and bilateral adductors (β = 0.122 95% CI: 0.040, 0.204; P = .004). Stronger relationships were observed for all muscle groups in AC-adjusted models (all muscles: β = 0.134 95% CI: 0.052, 0.215, P = .001; flexors: β = 0.133 95% CI: 0.050, 0.217, P = .002; extensors: β = 0.097 95% CI: 0.015, 0.178, P = .02; adductors: β = 0.147 95% CI: 0.061, 0.232, P = .001). Sex interactions were not statistically significant (all P > .05). Conclusion Regardless of sex, high UPF consumption in individuals at risk for KOA was associated with higher muscle fat content on thigh MRI scans. ClinicalTrials.gov identifier: NCT00080171 © RSNA, 2026 Supplemental material is available for this article. See also the editorial by Bredella in this issue.