ABSTRACT NUMBER: ESOC2026A1195 CORTICAL SURFACE AREA PARTIALLY MEDIATES THE EFFECT OF GENETICALLY PREDICTED C-REACTIVE PROTEIN LEVELS ON VASCULAR DEMENTIA

Abstract Background and aims Emerging evidence suggests inverse associations between circulating C-reactive protein (CRP) levels and dementia risk in certain contexts. Whether similar associations extend to vascular dementia (VaD) and the neural substrates remains unclear. Given that cortical morphometry is a key structural correlate of cognitive impairment, we investigated whether genetically predicted CRP levels causally influence cortical structure and mediate VaD risk. Methods Two-sample Mendelian randomization (MR) was performed using genome-wide association study summary statistics for CRP, global and regional brain imaging-derived phenotypes (IDPs), and VaD. Two-step MR mediation analysis quantified the indirect effects of CRP via cortical morphometry on VaD. Results Genetically predicted CRP levels were inversely associated with VaD risk (odds ratio [OR]=0.749, p=0.001) and its subcortical subtype (OR=0.711, p=0.010). Global cortical surface area was positively associated with the risk of VaD (OR=1.32, p=0.008) and subcortical VaD (OR=2.08, p=2.01×10-4), mainly driven by inferior parietal (OR=4.55, p=3.82×10-5) and posterior opercular regions (OR=2.92, p=1.18×10-5). Higher CRP levels were causally associated with reduced cortical surface area in these regions (FDR-corrected p<0.05). Mediation analyses indicated that regional cortical surface area partially mediated the effects of CRP on VaD risk, accounting for 17.8–29.3% of the total effect (p<0.05). Conclusions Genetically predicted CRP levels are associated with cortical surface area and lower VaD risk, with regional cortical structure partially mediating this association. These findings suggest that CRP-related biological processes may reflect context-dependent immune or systemic states relevant to vascular cognitive impairment, highlighting cortical surface area as a potential intermediate neuroanatomical biomarker. Conflict of interest Jie Zhang, Kaisi Ren, Haonan Xie, Mengsha Hu, and Shuchang Zhong: nothing to disclose.