Decoding neuroinflammation in Alzheimer’s disease: a multi-omics and AI-driven perspective for precision medicine

Shiyu Lin, Yijun Zhan, Ruiqi Wang, Jian PeiFrontiers in Immunology 2025 · DOI

This mini-review, published in Frontiers in Immunology, explores the shifting landscape of Alzheimer’s Disease (AD) research. It focuses on how high-resolution technologies—specifically single-cell RNA sequencing (scRNA-seq) and AI—are moving the field away from "one-size-fits-all" theories toward a precision medicine approach centered on neuroinflammation.

TL;DR

Researchers are using single-cell sequencing and AI to move beyond "bulk" brain analysis, uncovering specific, rare cell states that drive Alzheimer’s progression. By mapping the heterogeneous immune responses of microglia, astrocytes, and even peripheral immune cells, the field is laying the groundwork for individualized "precision medicine" treatments for AD.

What Problem Does This Paper Address?

Alzheimer’s Disease has traditionally been defined by the "big two" hallmarks: amyloid-beta (Aβ) plaques and tau tangles. However, clinical trials targeting these have seen limited success, partly because AD is incredibly heterogeneous.

The authors argue that chronic neuroinflammation is a central driver of the disease, but it isn't a uniform process. Traditional "bulk" tissue analysis (which grinds up brain samples) averages out the signals, effectively "hiding" rare but critical cell populations. This paper addresses the need to decode this cellular diversity to find better biomarkers and more specific therapeutic targets.

Methods

As a "Mini Review," this paper synthesizes recent breakthroughs rather than presenting a new lab experiment. The key methodologies discussed include:

  • scRNA-seq (Single-Cell RNA Sequencing): Profiling gene expression at the level of individual cells to distinguish between different functional states (e.g., homeostatic vs. reactive).
  • Spatial Transcriptomics: Mapping gene expression to specific physical locations in the brain tissue to see how cells interact near plaques.
  • Multi-omics Integration: Combining transcriptomics with proteomics (proteins), metabolomics (metabolites), and epigenomics to create a holistic "molecular signature" of the disease.
  • AI-Driven Analytics: Using machine learning to decode complex immune-inflammatory networks and identify "regulatory nodes" that could be targeted by drugs.

Datasets

The review draws on data from several landmark studies, including:

  • Human Brain Tissue: Post-mortem samples from AD patients compared to healthy controls.
  • AD Mouse Models: Specifically the 5xFAD and APP/PS1 models, which were instrumental in identifying "Disease-Associated Microglia" (DAM).
  • Single-cell Atlases: Reference datasets that categorize the transcriptional states of neurons, glia, and vascular cells in the aging brain.

Key Insights and Ideas

  • The DAM Transition: Microglia (the brain's immune cells) don't just "turn on." They undergo a specific transition from a homeostatic state to a Disease-Associated Microglia (DAM) state. This shift is often dependent on the TREM2 gene, which helps clear Aβ but can also drive inflammation if dysregulated.
  • Astrocyte Polarity: The paper highlights the "A1" (neurotoxic) vs. "A2" (neuroprotective) phenotypes in astrocytes. Targeting the signaling pathways that flip astrocytes into the toxic A1 state (like the NF-κB pathway) is a major area of interest.
  • The "Peripheral" Factor: AD isn't just a brain-isolated disease. scRNA-seq has revealed that peripheral immune cells—T cells, B cells, and Natural Killer (NK) cells—infiltrate the brain and interact with resident glia, significantly worsening neurodegeneration.
  • Oligodendrocyte Dysfunction: Beyond just providing insulation (myelin), oligodendrocytes in AD show significant metabolic stress. The paper notes that promoting cholesterol transport (via ApoE4 pathways) might help repair damaged myelin and improve cognition.

Key Results

  • Specific Biomarkers: The review identifies several high-potential markers: SPP1 in microglia (linked to synaptic loss), PLP1/ST18 in oligodendrocytes, and CACNG3/BDNF in excitatory neurons (positively correlated with cognitive function).
  • Pathway Targets: The JAK/STAT and NLRP3 inflammasome pathways are identified as critical "hubs" where multiple inflammatory signals converge, making them prime targets for AI-driven drug discovery.
  • Vascular Integrity: scRNA-seq has shown that endothelial cells in AD patients lose their "tight junction" stability, leading to blood-brain barrier (BBB) leakage and increased immune cell penetration.

Limitations and Future Work

The authors acknowledge that while scRNA-seq is powerful, it provides a "snapshot" in time. Future work needs to focus on:

  • Longitudinal Data: Understanding how these cell states evolve over decades of disease progression.
  • AI-Powered Lineage Tracing: Using AI to predict how one cell state transitions into another.
  • Clinical Translation: The biggest hurdle remains moving from "identifying a marker" to "developing a precision drug." The authors suggest that the next step is using AI to classify AD patients into subtypes based on their specific immune profiles, allowing for truly individualized therapy.

Summary updated

Abstract

Alzheimer's disease (AD) is a common neurodegenerative disease, which is characterized by β-amyloid (Aβ) deposition, Tau hyperphosphorylation, synaptic dysfunction and chronic neuroinflammation. Despite significant advances in research in recent years, effective therapeutic options remain limited. The development of single-cell RNA sequencing (scRNA-seq) has made it possible to analyze cellular heterogeneity in AD brain tissues at high resolution, breaking through the limitation of signal averaging in traditional large-scale tissue analysis. This technology has led to the discovery of novel disease-associated cell subsets, such as pro-inflammatory microglia and reactive astrocytes, and the identification of key molecular markers linked to disease progression. Integrating scRNA-seq with AI-driven analytics and multi-omics platforms further enhances our ability to decode the intricate immune-inflammatory networks underlying AD. This strategy is expected to achieve accurate classification and early diagnosis of AD subtypes, and promote the development of individualized treatment strategies based on individual molecular and immune characteristics.
BibTeX
@inproceedings{lin2025decoding,
  author = {Shiyu Lin and Yijun Zhan and Ruiqi Wang and Jian Pei},
  title = {{Decoding neuroinflammation in Alzheimer’s disease: a multi-omics and AI-driven perspective for precision medicine}},
  booktitle = {Frontiers in Immunology},
  year = {2025},
  doi = {10.3389/fimmu.2025.1616899},
}
Figure