Multi‐modal Imaging‐based peudotime analysis for AD progression

• Published in: Alzheimer's & dementia Vol. 19; no. S17
• Main Authors: He, Bing, Risacher, Shannon L., Saykin, Andrew J., Yan, Jingwen
• Format: Journal Article
• Language: English
• Published: 01.12.2023
• ISSN: 1552-5260; 1552-5279; 1552-5279
• DOI: 10.1002/alz.077057

Background Alzheimer’s disease (AD) progresses along a continuum and begins many years before symptom onset. AD takes a long progression time, making it hard to study the entire spectrum of AD development. Pseudotime analysis widely used in cell lineage analysis have been applied to AD gene expression data to understand the molecular changes along progression. In this work, we performed a neuroimaging‐based pseudotime analysis to explore the progression patterns of different imaging modalities. Method Data were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (https://adni.loni.usc.edu/). Data from three imaging modalities were collected including MRI (N = 1684), Amyloid (N = 1041), and Tau (N = 540) (Table.1). Out of those, 223 subjects have all three modalities. We tested four different pseudotime analysis tools: PHATE, MONOCLE3, Slingshot, and Destiny. MRI, Amyloid, Tau, and the combined multi‐modality data were used to estimate the pseudotime respectively. Only PHATE returned non‐reversed pseudotime and clear disease trajectories Result Amyloid derived pseudotime best captures the continuum of disease progression and differentiate diagnosis groups the best (Fig.1(b,f)). It also showed better correlation with clinical cognition scores than the composite amyloid measure (Fig.2(b,d)). Multi‐modality derived pseudotime reveals turning point of Amyloid and Tau. We found that the accumulation of Amyloid starts to accelerate when the composite SUVR reaches ∼1.2, which agrees well with existing findings (Fig.2(g,h)). Such turning point does not exist for amyloid derived pseudotime. It also reveals temporal order of brain changes within and across modalities (Fig.2(e,f)), where we observed overall amyloid changes starts early (i.e., small pseudotime) and accelerates earlier than overall brain atrophy and Tau deposition. With a closer look into Tau regions, we found Tau deposition in BRAAK 1 regions accumulates in a faster rate than those inBRAAK34 and BRAAK56 regions. Conclusion Pseudotime analysis has a great potential for us to better understand the progression of Alzheimer’s disease. New approaches to fully leverage the longitudinal imaging data could provide a refined view of imaging changes and their co‐occurance pattern along AD progression. Longitudinal data in pseudotime analysis could also help capture the subtle changes in very early stage and therefore would likely provide some valuable insights to improve early diagnosis.