Structural network disruption associated with molecular pathogenesis in Alzheimer's disease spectrum

325 

Abstract Submission 

Su Rim Ham¹, Hanna Cho², Han-Kyeol Kim², Sung-Woo Kim¹, Chul Hyoung Lyoo², Joon-Kyung Seong^1,3

¹School of Biomedical Engineering, Korea University, Seoul, Korea, Republic of, ²Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Korea, Republic of, ³Department of Artificial Intelligence, Korea University, Seoul, Korea, Republic of

Su Rim Ham  
School of Biomedical Engineering, Korea University
Seoul, Korea, Republic of

Hanna Cho  
Gangnam Severance Hospital, Yonsei University College of Medicine
Seoul, Korea, Republic of

Han-Kyeol Kim  
Gangnam Severance Hospital, Yonsei University College of Medicine
Seoul, Korea, Republic of

Sung-Woo Kim  
School of Biomedical Engineering, Korea University
Seoul, Korea, Republic of

Chul Hyoung Lyoo  
Gangnam Severance Hospital, Yonsei University College of Medicine
Seoul, Korea, Republic of

Joon-Kyung Seong  
School of Biomedical Engineering, Korea University|Department of Artificial Intelligence, Korea University
Seoul, Korea, Republic of|Seoul, Korea, Republic of

Alzheimer's disease involves structural and functional changes in the brain due to amyloid and tau deposition, ultimately leading to cognitive impairment [1]. While numerous medical imaging studies aim to explore the interactions among pathological markers and their relationship with connectivity disruption, a clear understanding is yet to be established [2,5]. In this study, we aim to investigate network disruption in Alzheimer's disease across early, late, and intermediate stages using multimodal imaging. Additionally, we explore the correlation between disrupted connections and pathology markers.

We constructed an FA (Fractional Anisotropy) weighted matrix based on tensors fitted using the free-water elimination (FWE) method applied to diffusion-weighted Imaging (DWI) data obtained from 147 participants at Gangnam Severance Hospital. The FWE method aims to enhance tractography by removing confounding free water from diffusion signals [4]. Amyloid and tau PET imaging were used to evaluate an individual's regional or global burden of proteins. The collected individuals were categorized into three groups based on amyloid positivity and disease state (43 amyloid-negative cognitively unimpaired (CU), 44 with mild cognitive impairment (MCI), and 60 with Alzheimer's disease (AD). Both patient groups were amyloid-positive). Comparison of connectivity between groups was conducted using network-based statistics (NBS) [6]. Additionally, cluster-based statistics (CBS) were performed to investigate significant correlations between connectivity disruption and amyloid, tau retention across the three groups [3]. The entire process of the study is shown in figure 1.

We observed significant disconnection through group-wise comparisons of edge strength. Subsequently, the representative region was defined as the region within the output subnetwork with the highest concentration of disrupted edges. In the comparison between CU and MCI groups, the representative region included the bilateral precuneus and right medial orbitofrontal, while in the comparison between MCI and AD groups, regions such as bilateral caudal anterior cingulate, bilateral precuneus, and cuneus. Comparing CU and AD groups to assess the overall disruption of connectivity throughout the disease, additional regions were identified, including hub nodes resulting from early or late changes (figure 2). Furthermore, we obtained subnetworks that explained the association between disruptions and accumulation of pathological markers. The precuneus, the most prominent region in CU and MCI comparison exhibited an association with global amyloid retention. In contrast, the temporal lobe, identified as the second most vulnerable in the CU and AD comparison, was confirmed to be associated with the accumulation of all protein types. On the other hand, the superior parietal region appeared to be associated with global tau. Hub regions such as bilateral caudal anterior cingulate, left paracentral, left postcentral, and bilateral cuneus either connected via edges associated with global SUVR or did not belong to the output subnetwork.

We aimed to identify primary regions of disconnection and areas vulnerable to the influence of biomarkers through multimodal imaging and cluster-based statistics. In the early stages of Alzheimer's disease (AD), the bilateral precuneus and orbitofrontal cortex appeared to be the most susceptible regions to disruption, particularly influenced by amyloid. As the disease progresses, association areas, especially in the parietal and temporal lobes, appear to undergo disconnection attributed to the effects of amyloid or tau burden. Finally, the anterior cingulate cortex and unimodal cortex (including the visual cortex, primary sensory areas, and motor areas) manifest in the late stage and appear to be minimally affected or least affected by protein influence.

Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) ¹

Multi-Modal Imaging ²

Other - Alzheimer's disease, Structural connectivity, Tau, Amyloid-beta, Network-based statistics