Cortical neurodegeneration influences annual posterior white matter hyperintensity progression

Abstract Background For over three decades, the concomitance of cortical neurodegeneration and white matter hyperintensities (WMH) has sparked discussion about their coupled temporal dynamics (Garnier‐Crussard et al. 2023). Longitudinal evidence supporting this hypothesis remains nonetheless scarce (Ter Telgte et al. 2018). We integrated surface‐based morphometry and bivariate latent change score modelling (BLCSM) to examine interrelationships between individual WMH and cortical thickness changes over a one‐year period in cognitively unimpaired participants. Method We analysed baseline and 12‐month follow‐up data from cognitively unimpaired DELCODE participants (n=393; median age 70.31 [IQR 66.06‐74.87] years; 52% females). We used T2w FLAIR and T1w MPRAGE data to segment WMH (LST; Schmidt and Wink 2019) and estimate cortical thicknesses (CAT12; Gaser et al., n.d.). Using BLCSM in a vertex‐wise fashion, we tested whether baseline WMH volumes predicted cortical thinning rates and whether baseline cortical thickness predicted WMH volumes increases (Figure 1). Due to the posterior dominance of WMH in AD (Bernal et al. 2023), we focussed on parietal and occipital WMH. All models included age, sex, years of education, and total cardiovascular risk factors as covariates on baseline and change scores. We log‐10 transformed WMH volumes and corrected WMH volumes and thicknesses measurements for TICV via residualisation. We finally tested for moderation effects of CSF‐derived Aβ42/40 and pTau181 on cross‐domain coupling. Result BLCSM generally provided good data fits (RMSEA≤0.05, CFI≥0.095, SRMR≤0.05) across all analyses. The mean thickness across precuneal and superiorparietal cortices at baseline predicted the progression of WMH better than any other cortical region (Figure 2B; β Thick→ΔWMH =‐0.051, SE=0.011, Z=‐4.535, p‐value<0.001). This was especially evident with lower CSF‐derived Aβ42/40 (β Thick*Aβ42/40→ΔWMH =0.176, SE=0.053, Z=3.309, p‐value=0.001). WMH volume at baseline explained, in part, the level of thinning of parts of the fusiform cortex (Figure 2D; β WMH→ΔThick =‐0.139, SE=0.028, Z=‐5.030, p‐value<0.001). This interrelationship was moderated by CSF‐derived pTau181 (β WMH*pTau181→ΔThick =‐0.144, SE=0.053, Z=‐2.711, p‐value=0.007). Conclusion Cortical thinning and WMH progression may be mutually reinforcing processes that become reciprocally coupled prior to any detectable cognitive deficits. Hallmark AD proteins appear to moderate their interrelationships.