Hypothesis

Age‑related inflammation (inflammaging) is driven not by stochastic damage to host tissues but by a measurable loss of topological complexity in the coupled gut‑microbiome‑brain metabolomic network. Persistent homology of longitudinal multi‑omics data will show that specific 1‑dimensional homology classes (loops) linking microbial‑derived short‑chain fatty acids, vagal neurotransmitters, and neuroimmune genes decay with age, and that this decay predicts systemic inflammatory biomarkers better than host gene‑expression changes alone.

Mechanistic Basis

The gut‑brain axis operates as a high‑dimensional signalling manifold where microbial metabolites (e.g., butyrate, tryptamine) activate enteroendocrine cells, which signal via the vagus nerve to modulate microglial phenotype and cytokine production in the brain. In youth, this manifold exhibits robust topological features: persistent loops that reflect feedback cycles between microbial synthesis, host receptor expression, and neural firing patterns. Age‑related dysbiosis perturbs metabolite concentrations, causing certain loops to collapse or become noisy, thereby breaking coordinated anti‑inflammatory signalling. The resulting loss of topological invariants reduces the system’s capacity to buffer inflammatory stimuli, leading to elevated circulating IL‑6, TNF‑α, and CRP.

We extend prior work showing that persistent homology captures spatial structure in gene expression for kidney disease [3] and immune cell states [2] by applying it to paired metagenomic, metabolomic, and transcriptomic time series from the same individuals. The hypothesis predicts that the birth–death intervals of H1 features (loops) linking, for example, fecal butyrate levels → colonic FFAR2 expression → vagal afferent firing (measured via heart‑rate variability) → hippocampal BDNF transcription will show a characteristic shortening with age.

Predictions and Tests

1. Cross‑sectional test: In a cohort aged 20–90, compute persistence diagrams from paired fecal 16S rRNA, plasma metabolome, and peripheral blood mononuclear cell (PBMC) transcriptomes collected at the same visit. The average persistence (lifespan) of H1 loops connecting microbial SCFA pathways to NF‑κB target genes will negatively correlate with age (Spearman ρ < –0.4, p < 0.001).
2. Longitudinal test: Over a 2‑year follow‑up, individuals exhibiting a >20% reduction in the persistence of the butyrate‑FFAR2‑vagal‑BDNF loop will develop a ≥30% increase in plasma IL‑6 compared with those whose loop persistence remains stable (hazard ratio 2.1, 95% CI 1.5–3.0).
3. Intervention test: Administration of a defined butyrate‑producing probiotic for 12 weeks will increase the persistence of the targeted loop by ≥15% in participants >65 years, accompanied by a significant drop in CRP (paired t‑test, p < 0.01).
4. Falsification: If topological metrics show no age‑related decline or fail to predict inflammatory outcomes after adjusting for confounders (diet, BMI, antibiotics), the hypothesis is refuted.

Potential Confounds and Controls

Control for host genetics by including polygenic risk scores for inflammatory diseases as covariates. Use diet records to adjust metabolite levels. Exclude participants with recent antibiotic use or gastrointestinal infections. Apply batch‑correction methods (ComBat) before constructing persistence diagrams to mitigate technical variation.

By framing inflammaging as a topological breakdown of a microbiome‑brain signalling circuit, this hypothesis offers a concrete, quantifiable biomarker (persistence of specific homology loops) that links microbial ageing to host inflammation, providing a clear avenue for mechanistic intervention and diagnostic development.