Glial Transcriptomic Convergence in Alzheimer’s and Parkinson’s Disease with Independent Retinal and Aqueous-Humor Evidence
Glial Transcriptomic Convergence in Alzheimer’s and Parkinson’s Disease with Independent Retinal and Aqueous-Humor Evidence
Where did the research take place?
The study site has not been established. Author addresses may differ from where the research occurred.
Publication status: preprint
A plain-language reading has not been prepared for this paper yet.
Original abstract
Alzheimer’s disease (AD) and Parkinson’s disease (PD) are usually studied as distinct neurodegenerative disorders, yet both involve molecular changes that extend beyond vulnerable neuronal populations. We asked whether the two diseases share reproducible transcriptional organization across homologous human glial cell classes and whether components of that shared signal are independently represented in retinal and ocular proteomic datasets. We reanalyzed five human single-nucleus RNA-sequencing datasets using donor-level pseudobulk models for astrocytes, microglia, oligodendrocyte precursor cells (OPCs), and oligodendrocytes. Disease effects were estimated and replicated within AD and PD before cross-disease comparison. Across matched genes, AD and PD showed modest but permutation-supported positive effect-vector concordance in all four cell classes. Concordance was strongest in oligodendrocytes (Spearman ρ = 0.253; 60.7% directional agreement), whereas no individual gene survived a conservative AD–PD conjunction FDR criterion. Reactome analysis identified 29 pathways with formal same-direction AD–PD FDR convergence in one glial class and a broader set of 665 pathways with recurrent same-direction enrichment across at least three classes. The recurrent pathway structure exceeded a profile-preserving permutation null (95% interval, 411–476; empirical P = 1.0 × 10⁻⁴), supporting a global excess of recurrent pathway organization without implying significance for every individual recurrent pathway. We then integrated healthy retinal expression, AD retinal proteomics, and PD aqueous-humor proteomics with the brain-concordance results. Of 17,745 genes in the integrated candidate universe, 14,469 were detected in at least two healthy retinal donors. The AD retinal reanalysis tested 4,995 genes and identified 227 nominal and two FDR-significant effects; the PD aqueous-humor reanalysis tested 5,691 proteins and identified 2,713 nominal and 1,803 FDR-significant differences. No candidate reached Class A, which required AD–PD brain concordance, healthy-retinal expression, and FDR-level support in both ocular disease datasets. The highest-ranked multimodal candidates included SCAMP5, XYLB, RABEPK, ITGAM, CD38, CTCF, UBTD2, and NAPEPLD. Together, these results support a distributed, cell-class-dependent molecular architecture shared between AD and PD. Independent retinal and ocular measurements provide an additional layer for prioritizing components of this shared biology, but they do not establish clinically validated retinal or ocular biomarkers.