Background: Previous studies suggest that patients with rheumatoid arthritis (RA) may have a higher risk of cognitive impairment and show central nervous system imaging abnormalities. However, systematic evidence on RA-related brain structure on T1-weighted MRI and white matter microstructure on diffusion MRI (dMRI) remains limited in large population-based cohorts. In addition, the mechanisms underlying the interaction between peripheral inflammation and the central nervous system in these neuroimaging abnormalities are still unclear.

Objectives: This study aimed to systematically characterize cortical morphometric measures on T1-weighted MRI and white matter microstructural alterations on dMRI in RA using the UK Biobank (UKB) cohort. We further aimed to explore peripheral inflammatory signals and central molecular mechanisms potentially linked to these imaging findings using a collagen-induced arthritis (CIA) rat model, providing a basis for mechanistic interpretation and biomarker discovery for RA-related white matter abnormalities.

Methods: We included 976 RA cases and 61,430 non-RA controls from UKB, defined using First Occurrences (mean age 65.4 years). For T1-weighted MRI, regional cortical area, volume, and thickness were extracted using FreeSurfer across multiple parcellation schemes, including aparc.a2009s, Desikan–Killiany, DKT, and Brodmann areas. Group differences were assessed using linear regression with HC3 robust standard errors. Models were adjusted for demographic characteristics, lifestyle factors, body mass index, assessment center effects, and socioeconomic covariates. For dMRI, tract-based spatial statistics (TBSS) was applied to analyze ICVF, ISOVF, OD, FA, MD, and MO, with multiple testing correction using the Benjamini–Hochberg procedure (FDR < 0.05). All analyses were conducted using UKB baseline data, and RA participants were not further stratified by disease duration, disease activity, or medication use. In the animal study, 6 female Wistar rats were included as controls and 9 female Wistar rats were used to establish a CIA model. After T1 and dMRI acquisition, the prefrontal cortex, hippocampus, cerebellum, and plasma were collected. Exploratory proteomic profiling was performed using the Astral 24-minute workflow.

Results: T1-weighted MRI analyses showed no significant differences in area, volume, or thickness between the RA and control groups across parcellation schemes. In contrast, dMRI TBSS revealed widespread white matter microstructural abnormalities in RA. A total of 34 ICVF, 25 FA, and 27 MD white matter tracts remained significant after Benjamini–Hochberg correction (FDR < 0.05). Among them, 17 tracts showed consistent group differences across ICVF, FA, and MD. These alterations were mainly distributed in fronto-subcortical pathways involved in motor and cognitive regulation, including the external capsule, corona radiata, superior longitudinal fasciculus, cingulum, and internal capsule-related regions. They were also observed in transcallosal pathways, including the genu, body, and splenium of the corpus callosum, as well as sensory–cerebellar integration pathways, including the sagittal stratum and the right inferior cerebellar peduncle. No significant tracts were identified for ISOVF, OD, or MO after correction. Proteomic analyses in the CIA rat study are ongoing, with results expected in early February.

Conclusions: In this large UKB analysis, RA was not associated with overt cortical morphometric differences on T1-weighted MRI but was associated with stable and widespread white matter microstructural alterations on dMRI. The affected tracts primarily involved fronto-subcortical regulatory, transcallosal connectivity, and cerebellar integration pathways, suggesting inflammation-related axonal microstructural injury. Ongoing plasma and brain tissue proteomics in the CIA rat model may help identify key molecular signals linking peripheral inflammation to white matter abnormalities, providing mechanistic insights and potential biomarkers for RA-related central nervous system changes.

REFERENCES: NIL.

Acknowledgments: NIL.

Disclosure of Interests: None declared.