Gut Bacteria May Be Quietly Driving Three of the Leading Causes of Blindness

Three of the world's most common causes of irreversible blindness, age-related macular degeneration (AMD), diabetic retinopathy (DR), and retinal vein occlusion (RVO), have long been treated as separate problems requiring separate local fixes injected directly into the eye. A new study published in Ophthalmology Science by Wei and colleagues proposes a different frame: all three diseases may share a common molecular vulnerability routed through the gut, and the metabolites bacteria produce inside us could be both the source of damage and a lever for treatment. This is early, computational research, and the authors are explicit that experimental validation in animals and humans still lies ahead.

Shared Gene Targets Across AMD, Diabetic Retinopathy, and RVO

The researchers began by cataloguing disease-associated genes from three major databases, pulling 2,747 genes linked to AMD, 1,102 to DR, and 1,097 to RVO. They then identified 226 gut microbial metabolites from the gutMGene database and used two computational target-prediction tools to estimate which human proteins those metabolites interact with, generating a list of 1,518 predicted host targets. Crossing those lists revealed something striking: 38 genes appeared in the metabolite-related gene sets for all three diseases simultaneously. That convergence, across conditions with different triggers and patient populations, suggests the gut is not influencing each disease in isolation but may be pulling on a shared set of biological levers.

The 38 shared genes clustered around four proteins that emerged repeatedly as high-centrality nodes in the protein-protein interaction network: IL6, TNF, AKT1, and TP53. IL6 and TNF are well-known inflammatory signalling proteins; the authors describe them here as potential "gut-to-systemic inflammatory mediators" capable of triggering ocular inflammatory cascades. AKT1, by contrast, appears to play a protective role, sitting at a junction where incoming signals regulate the survival of retinal pigment epithelium cells, photoreceptors, and the delicate blood vessels that nourish them.

Why the Gut-Retina Connection Could Change Treatment Thinking

Current clinical management for all three diseases centres on intraocular injections, a physically direct but logistically demanding approach. The gut-eye axis concept, which proposes that gut microbiota influence systemic inflammation, oxidative stress, and immune homeostasis through circulating metabolites, opens a different therapeutic lane. Rather than suppressing inflammation inside the eye after it has already started, systemic modulation of gut-derived signals might reduce the underlying pressure that drives chronic retinal disease in the first place.

Reviews of microbial metabolites and ocular health have flagged this possibility before, but translating it into drug targets requires knowing which specific molecules and pathways matter. This study attempts to do that mapping computationally. The enrichment analyses pointed to a recurring cluster of pathways: NF-κB, MAPK, PI3K-Akt, TNF, VEGF, Toll-like receptor, and NOD-like receptor signalling. The authors frame these as a shared "microbial signal sensing, inflammatory regulation, and stress response" functional axis that runs across all three blinding diseases.

The disease-specific subnetworks also showed distinct features. The AMD subnetwork was associated with Escherichia coli and Bifidobacterium; the DR subnetwork with Bacteroides and Enterococcus faecalis; and the RVO subnetwork with Lactobacillus and Clostridium, along with a metabolite called phenylacetic acid. Whether these microbial associations reflect causal relationships or statistical co-occurrence in database entries is a question the study cannot answer on its own.

How Network Pharmacology Maps a Complex Biological System

The methodological backbone here is integrative network pharmacology, an approach that constructs large-scale maps of relationships among diseases, genes, and potential drug-like molecules rather than focusing on a single target at a time. The team built what they call an M-S-M-T network, linking microbiota to substrates to metabolites to targets, using STRING for protein interaction data, Cytoscape for network prioritisation, and a minimum interaction confidence score of 0.7 to filter out weaker predicted connections.

For each candidate metabolite, the authors then ran ADMET evaluations (assessing absorption, distribution, metabolism, excretion, and toxicity). Four metabolites came through with favorable profiles: indole-3-propionic acid, tryptamine, phenylalanine, and indole-3-lactic acid. Indole-3-propionic acid was specifically noted for its safety and anti-inflammatory potential. None of the key metabolites showed predicted hERG channel blockade, hepatotoxicity, or carcinogenicity in the computational models, which is a necessary but far from sufficient marker of clinical safety.

Open Questions About Retinal Accumulation and Causality

The study's authors are candid about where the analysis stops. The most important gap is pharmacokinetic: there is no direct evidence in this work that any of the identified metabolites actually reach the retina at biologically effective concentrations. The mechanisms by which gut signals precisely activate immune responses in the retina remain incompletely understood, and the blood-retinal barrier adds a further filtering step that tissue-distribution modelling has not yet accounted for here.

Beyond pharmacokinetics, the study cannot distinguish how much of any metabolite signal comes from gut bacteria versus dietary sources, since many of these compounds have dual origins. The recurring appearance of broadly connected proteins like AKT1 and IL6 as hub targets is also worth interpreting cautiously: highly connected proteins often surface in network analyses simply because they are nodes in many known pathways, not necessarily because they are the most disease-relevant targets in a specific biological context.

The enriched pathways, the authors note, may partly reflect shared chronic cellular stress across all three diseases rather than mechanisms unique to the gut-retina connection. Future work will need to combine stable isotope tracing (to track metabolite origin), animal models with direct retinal measurements, and prospective clinical cohorts before any of these computational predictions can be tested as therapeutic hypotheses.

Next Steps Toward Probiotic and Metabolite-Based Retinal Therapies

Despite its preliminary nature, the study lays out a concrete research agenda. The authors suggest that targeted probiotic formulations or prebiotic supplementation could, in principle, offer long-term, gentle, and reversible modulation of retinal disease risk by shifting the balance of gut-derived metabolites toward the protective end of the spectrum. Metabolites and their associated microbial taxa are also proposed as candidate biomarkers for early screening and for tracking treatment response, a potentially valuable application given how few systemic markers currently exist for AMD and RVO progression.

For the three conditions at stake, which collectively affect hundreds of millions of people and consume enormous healthcare resources through repeated intraocular injection regimens, even a modest systemic intervention that slows disease onset or reduces injection frequency would represent a meaningful clinical advance. The pathway from this computational map to that outcome is long, but the map itself is more detailed than anything previously available for the gut-retina connection across all three major blinding diseases simultaneously.

Evidence rating

INVESTIGATIONAL — 61/100

Caveats:

• This is a computational/in silico study only — no animal experiments, no human participants, and no lab validation were performed.
• All target predictions are derived from database mining and similarity-based algorithms, not direct experimental binding or functional assays.
• The study cannot establish whether identified metabolites reach biologically relevant concentrations in retinal tissue.
• No causal relationships are demonstrated; all proposed mechanisms remain hypothetical and require experimental validation.
• Funding was not disclosed by the authors.

This is a computational network pharmacology study with no human participants or experimental validation. All target–metabolite associations are predicted using database mining and similarity-ensemble algorithms, meaning the study identifies plausible mechanistic hypotheses rather than demonstrating causation. The framework has not been independently replicated, and key predictions regarding retinal metabolite concentrations and in vivo efficacy remain untested.

References

1. Bin Wei, Jingfen Zhang, Yichong Duan, Na Cheng, Xiaorong Wu (2026). Integrative Network Pharmacology Analysis of Gut Microbial Metabolite–Mediated Gut–Eye Axis Regulation in Major Blinding Retinal Diseases. Ophthalmology Science. https://doi.org/10.1016/j.xops.2026.101284
2. Kinga Szymańska, Karolina Sałasińska, Agnieszka Młynarczyk, Justyna Miszczak, Weronika Dmoch, Piotr Maciejewicz (2026). The Gut-Eye Axis and Microbiome in Ophthalmic Diseases: A Narrative Review. Journal of Clinical Medicine. https://doi.org/10.3390/jcm15103563
3. Yvonne Nguyen, Josephine Rudd Zhong Manis, Nicole Marie Ronczkowski, et al. (2024). Unveiling the gut-eye axis: how microbial metabolites influence ocular health and disease. Frontiers in Medicine. https://doi.org/10.3389/fmed.2024.1377186
4. Fatima Noor, Muhammad Tahir ul Qamar, Usman Ali Ashfaq, Aqel Albutti, Ameen S. S. Alwashmi, Mohammad Abdullah Aljasir (2022). Network Pharmacology Approach for Medicinal Plants: Review and Assessment. Pharmaceuticals. https://doi.org/10.3390/ph15050572
5. The gut-eye axis in retinal degeneration – Lighthouse Scholar. frontiersin.org. https://www.frontiersin.org/journals/cellular-and-infection-microbiology/articles/10.3389/fcimb.2024.1539255