基于结构协变因果网络的阿尔茨海默病灰质萎缩机制研究OA

Objective:To jointly apply voxel-based morphometry(VBM)and causal structural covariance network(CaSCN)methods to identify characteristic brain atrophy regions in Alzheimer's disease(AD)and systematically reveal the underlying causal associations of gray matter structural changes and potential pathological propagation pathways.Materials and Methods:Data were obtained from the Alzheimer's Disease Neuroimaging Initiative(ADNI)database,including 32 patients with mild-to-moderate AD and 29 demographically matched healthy controls(HC).First,VBM was used to compare whole-brain gray matter volume differences between the AD and HC groups.To explore the association between structural changes and clinical symptoms,the correlation between gray matter volumes in these group-difference brain regions and the Mini-Mental State Examination(MMSE)score,which represents disease severity,was further analyzed.Subsequently,based on the VBM results,the brain region with the most significant atrophy was selected as the seed point.The CaSCN method was applied to conduct whole-brain bidirectional causal analysis,including assessing the seed-to-map(outward causal effects from the seed to the whole brain)and map-to-seed(inward causal effects from the whole brain to the seed)effects.This analysis aimed to reveal how the causal interaction patterns between brain regions evolve with increasing disease severity,characterized by the continuous change in MMSE scores.Finally,brain regions with significant causal connections from the above analyses were defined as regions of interest(ROIs)to construct an ROI-to-ROI causal network.The weighted out-degree and in-degree of nodes were calculated to quantify the causal driving and receiving capacity of each brain region within the network.Results:VBM analysis revealed significant gray matter atrophy in the AD group in the left hippocampus,right superior temporal pole,and right middle temporal gyrus[voxel-wise family-wise error(FWE)correction,corrected P＜0.001],and the volumes of these brain regions were positively correlated with MMSE scores(P＜0.05).Among these,the left hippocampus exhibited the most pronounced atrophy(t=11.72).Seed-based CaSCN analysis of the left hippocampus demonstrated that,in the map-to-seed direction,the bilateral parahippocampal gyri and left amygdala exerted positive causal effects on the left hippocampus[voxel-wise false discovery rate(FDR)correction,P＜0.05],corresponding to Granger causality(GC)values＞0.61 and Z-values＞3.85.In contrast,in the seed-to-map direction,no significant causal effects from the left hippocampus to the whole brain were observed(GC＜0.61,Z＜3.85).These findings suggest that the left hippocampus primarily serves as a hub for information convergence within the structural covariance network.Further ROI-based network analysis(based on brain regions showing significant map-to-seed connections under FDR correction)revealed that the left hippocampus emitted significant GC connections to the bilateral parahippocampal gyri,right anterior entorhinal cortex,and right hippocampus(FDR correction,P＜0.05,GC＞0.61),with the strongest output effect directed to the right parahippocampal gyrus(weighted out-degree value=1.03).However,no significant output from the left hippocampus to the ipsilateral amygdala was detected.Conclusions:In the mild-to-moderate stage of AD,the left hippocampus acts as a dynamic hub within the gray matter atrophy network.It not only receives pathological input from structures such as the bilateral parahippocampal gyri,right anterior olfactory cortex,and right hippocampus but also drives further pathological output to these regions,forming a local closed-loop network with positive feedback regulation.The unidirectional causal dissociation between the hippocampus and amygdala suggests a difference in their pathological timelines.The strong cross-hemispheric output from the left hippocampus to the right parahippocampal gyrus further supports the notion of AD as a whole-brain network disorder.This study deepens the understanding of AD atrophy mechanisms from a causal network perspective and provides a new basis for explaining its clinical heterogeneity and disease progression.