Journal of Advances in Medicine and Medical Research

Background: Alzheimer’s disease is a progressive neurodegenerative disorder characterized by complex spatial and molecular alterations, necessitating advanced spatial transcriptomic approaches to better understand disease progression across Braak stages.

Aims: The study aims to examine how spatial organization of gene expression and cell-type relationships changes across Alzheimer’s disease (AD) progression using the SEA-AD MERFISH spatial transcriptomics dataset.

Study Design: Secondary computational discovery study using publicly available human brain spatial transcriptomics data.

Place and Duration of Study: Biomedical Sciences Division, STEM Science Center, Englewood Cliffs, NJ, USA; analysis of the publicly released SEA-AD dataset was conducted using archived data.

Methodology: Sixty-nine middle temporal gyrus sections from 27 donors spanning Braak stages 0–VI were grouped as low Braak (0–III) or high Braak (IV–VI). Moran’s I was used as the primary quantitative spatial autocorrelation metric, and a spatial shift summary based on section-level median difference, Cliff’s delta, and Braak-stage correlation was used to rank disease-associated spatial changes. Neighborhood enrichment and co-occurrence metrics were calculated with Squidpy. Group differences were evaluated using Mann-Whitney U tests with Benjamini-Hochberg false discovery rate correction, and Pearson and Spearman correlations with Braak stage were calculated.

Results: Twenty-two genes showed significantly altered spatial autocorrelation between low and high Braak groups, and 13 genes showed progressive correlation with Braak stage. NRG1, PEX5L, FGF13, SEMA3E, and SCUBE1 consistently showed higher Moran’s I values in advanced stages. L6b neurons displayed significant self-enrichment, and oligodendrocyte-L4 intratelencephalic neuron co-occurrence increased in high Braak tissue.

Conclusion: AD progression was associated with increased spatial clustering of selected genes and selective reorganization of cell-type neighborhoods. These findings support Moran’s I and related spatial-shift measures as hypothesis-generating spatial biomarkers that require validation in additional brain regions and by protein-level or orthogonal in situ methods.