Synopsis

Keywords: Neurodegeneration, Alzheimer's Disease, Cerebrovascular, radiological

Evidence suggests that cardiovascular risk factors and cerebrovascular disease contribute to the Alzheimer’s disease (AD) pathophysiology. It is still unclear to what extent vascular and amyloid pathology have a synergistic or independent influence in preclinical AD stages. We used structural equation models in a cohort of cognitively unimpaired individuals to investigate if cerebrovascular pathology mediates the relationship between cardiovascular risk factors and CSF Aβ1-42, and show that cerebrovascular pathology accelerates downstream AD markers (p-Tau181 and hippocampal volume) by increasing amyloid pathology.

Introduction

Epidemiological and clinical studies have shown considerable overlap between cerebrovascular disease and Alzheimer’s disease. However, it is unclear whether vascular pathology represents an independent biological event or an integrated part of the amyloid cascade, facilitating hyperphosphorylated tau (P-tau) accumulation and subsequent neurodegeneration in AD.
We investigated the impact of cardiovascular risk factors and cerebral small vessel disease (CSVD) on amyloid pathology and their combined effect on the downstream AD biomarkers P-tau and hippocampal neurodegeneration.

Methods

Data were drawn from the European Prevention of Alzheimer's Dementia (EPAD) cohort baseline data vIMI release (n=1592, Table 1). Cardiovascular risk was computed with the Framingham risk score (FRS). The MRI visual assessment included a 0-4 scale for PVS in the basal ganglia (BG) and centrum semiovale (CS)¹, the Fazekas scale (0-3) for periventricular and deep WMHs², presence (0-1) of lobar or deep cerebral microbleeds (CMBs³), and lacunes (0,1,2,>2), according to the STRIVE criteria⁴. Global and lobar periventricular and deep WMH volumes and hippocampal volumes were derived from FLAIR and T1w MRI sequences. Aβ_1-42 and p-Tau₁₈₁ concentrations were obtained from CSF. Linear models were used to study the association between i) FRS and CSVD; ii) FRS and Aβ_1-42; iii) CSVD and Aβ_1-42 and p-Tau₁₈₁. We added an interaction term between CSVD indices and Aβ_1-42 for predicting p-Tau₁₈₁, to study possible synergic effects.
Structural equation modeling (SEM) was used to model the association between 4 observed variables: FRS, Aβ_1-42, P-tau₁₈₁, and hippocampal volume (HCV); and one latent variable measuring the cerebrovascular component. The latent “CSVD-burden” variable was created using confirmatory factor analysis on all radiological indices. Initially, we tested the mediating effect of the CSVD-burden component in the association between FRS and Aβ_1-42. Subsequently, we built a SEM model including a direct relationship of FRS with the latent CSVD-burden factor; the reciprocal relationship between the latter and Aβ_1-42; the effect of both amyloid and the latent factor to P-tau₁₈₁ and hippocampal atrophy; the effect of P-tau₁₈₁ on hippocampal atrophy. In addition to direct effects, the model also estimated the indirect (mediating) effects of amyloid and the CSVD-burden on downstream AD markers (Figure 1).
All models were corrected for age and sex.

Results

Cohort characteristics are reported in Table 1. We observed an association between FRS and CSVD (Table 2). Higher FRS scores were also associated with lower levels of Aβ1-42 (β=-0.18; p<0.001) and higher levels of P-tau181 (β=0.21; p<0.001) and T-tau (β=0.22; p<0.001). Lower levels of Aβ1-42 were further observed in relationship with both BG (p<0.001) and CS (p<0.01) PVS; higher periventricular (p<0.001) and deep (p<0.001) Fazekas scores; the presence of lobar (β=-0.51; p<0.001) and deep (β=-0.64; p<0.05) microbleeds; the presence of 2 or more lacunes (p<0.001); and higher regional WMH volumes (all p<0.001; Figure 2)
Stronger association between Aβ1-42 and P-tau181 in participants with higher Fazekas DWMH scores (coefficients per group: 0:β=-0.03, p=0.57; 1:β=-0.08, p<0.05; 2:β=-0.36, p<0.001; 3:β=-0.21, p=0.12); higher Fazekas PVH score (coefficients per group: 0:β=0.01, p=0.85; 1:β=-0.21, p<0.001; 2:β=-0.32, p<0.001; 3:β =-0.46, p<0.05); and higher regional WMH (all p<0.005) volumes were observed (Figure 3).
The CSVD-burden latent factor fully mediated the association between FRS and Aβ1-42 (direct effect: β=-0.01; p=0.34; indirect effect: β=-0.03; p<0.001). This relationship was thus not included in the subsequent model.
In the SEM model, we found a significant effect of FRS on the CSVD-burden latent component (β=-0.01; p<0.001), and a significant effect of the latter on Aβ1-42 (β=-2.64; p < 0.05). In turn, levels of Aβ1-42 were not significantly predictive of the CSVD-burden latent factor (p=0.22). We observed a significant direct effect of Aβ1-42 on P-tau181 levels (β=-0.14; p<0.001) and HCV (β=0.06; p<0.05), and of P-tau181 levels onto hippocampal volumes (β=-0.05; p<0.05). The CSVD-burden latent factor was not directly related to P-tau181 and HCV, and did not show mediating effects. However, a significant indirect effect of the latent factor on both P-tau181 (indirect effect: β =0.36; p < 0.05) and HCV (indirect effect: β=-0.24; p < 0.05) through Aβ1-42 was observed.

Conclusion

Our work shows that CSVD mediates the effect of cardiovascular risk factors on amyloid deposition, highlighting the importance of monitoring cerebrovascular pathology in patients at risk of AD. Moreover, our findings show that cerebrovascular pathology might accelerate the amyloid cascade in the preclinical phases of the disease, accelerating subsequent tau accumulation and hippocampal degeneration, stressing the importance of considering individualized vascular protective treatment for the prevention or deceleration of AD.

Acknowledgements

No acknowledgement found.