Scientists Find Gut-Lung Immune Pathways That Drive Lung Scarring in IPF

Idiopathic pulmonary fibrosis has always lived up to its name: idiopathic, meaning unknown cause, a label that has frustrated patients and clinicians alike. A new multi-omics study published in June 2026 by Zhao J, Liu Z, Kuang L, and colleagues now offers some of the most specific causal leads yet, linking gut bacteria, immune cell subtypes, inflammatory proteins, and plasma metabolites to IPF risk through a rigorous genetic analysis framework.

Nine Immune Cell Phenotypes Causally Linked to IPF Risk

The central finding is simultaneously striking and complicated: nine immune cell phenotypes showed significant causal associations with IPF, and they do not all point in the same direction. Regulatory T cells (Tregs) illustrate the tension well. Certain protective Treg subsets appear to reduce IPF risk, while activated CD4+ Tregs actually increase susceptibility to the disease. That distinction matters clinically, because therapies aimed broadly at boosting or dampening Tregs could help some patients while harming others.

CD39+ CD4+ T cells emerged as a protective force, associated with suppression of a specific bacterial signal, while CD39+ CD8br Tregs were linked to promotion of an anti-fibrotic factor. The picture that emerges is not a single immune switch being flipped the wrong way, but a network of cell types with conflicting effects that must be understood individually before any intervention can be designed safely.

How Gut Bacteria May Fuel Lung Scarring

Nine gut microbiota taxa also showed causal associations with IPF in this analysis, a finding that adds weight to growing evidence on gut microbiota dysbiosis in IPF onset and progression. The gut-lung axis, the idea that microbial activity in the intestine can shape immune responses in the lung, has been gaining traction in respiratory research, and recent reviews highlight it as a core regulatory link in pulmonary fibrosis pathogenesis.

The mediation analysis gets specific about how one of those bacterial signals operates. Roughly 16.8% of the effect of activated CD4+ Tregs on IPF susceptibility was mediated by Phylum SAR324, a bacterial group whose name alone signals how far outside clinical familiarity this research sits. The inflammatory proteins TNFSF12 and CCL25 mediated an additional 10.8% and 4.94% of that same immune cell effect, respectively, suggesting multiple parallel routes by which gut-derived or immune-derived signals reach the fibrotic lung.

How Mendelian Randomization Supports Causal Claims

Most microbiome research struggles with the chicken-and-egg problem: does a disrupted gut microbiome cause disease, or does disease disrupt the gut microbiome? Mendelian randomization addresses this by using genetic variants as natural proxies for exposures, sidestepping the confounding and reverse causality that plague conventional observational studies. As researchers have argued, MR offers a way to enhance causal inference in gut microbiome research that standard epidemiology simply cannot match.

This study applied bidirectional MR and mediation analysis to GWAS summary data from the FinnGen dataset, one of the most comprehensive population-level genomic resources available. The scope was substantial: 731 immune cell phenotypes, 473 gut microbiota taxa, 91 inflammatory proteins, 179 plasma lipid species, and 1,400 metabolites were all analyzed alongside IPF cases. Sensitivity analyses were conducted to check the robustness of the findings.

HLA-DR+ CD33- myeloid cells, another immune cell type implicated in the study, modulated IPF via two plasma metabolites: LIFR, which mediated 22.1% of the effect, and decadienedioic acid (C10:2-DC), which mediated 8.38%. Three inflammatory proteins, one lipid species, and 29 metabolites in total showed causal links to IPF, pointing to metabolic pathways as potential intervention targets alongside the immune ones.

Limitations of a Genetics-First Approach

The study's authors are clear that the causal relationships identified here remain associations inferred from genetic data, not confirmed mechanisms. Additional experimental studies, meaning cell culture work, animal models, and eventually clinical trials, will be needed to validate whether the pathways identified are genuinely driving fibrosis or simply correlated with it through genetic architecture.

Mendelian randomization, while methodologically powerful, still cannot fully escape all sources of bias, particularly when genetic instruments are weak or when pleiotropy, a single gene affecting multiple traits, is present. The complexity of the multi-omics network identified here also means that any single mediator's contribution, like SAR324's 16.8% share of the activated CD4+ Treg effect, should be read as a proportion of a partially explained pathway, not a complete mechanistic account.

What These Findings Could Mean for IPF Treatment

IPF currently has only two approved therapies, pirfenidone and nintedanib, both of which slow progression without reversing or curing the disease. Median survival after diagnosis remains roughly three to five years. That stark clinical reality is why identifying new mechanistic leads, even preliminary ones, draws serious attention.

The pathways surfaced here suggest that precision immunomodulation, targeting specific Treg subsets rather than the immune system broadly, and modulating particular gut bacteria rather than the microbiome as a whole, could be viable therapeutic directions. Proteins like TNFSF12 and CCL25 are pharmacologically tractable targets that already have research programs attached to them in other disease contexts. Whether those programs can be redirected toward IPF, informed by the mediation percentages reported here, is the question that experimental follow-up will need to answer.

Evidence rating

EMERGING — 82/100

Caveats:

• Mendelian randomization infers causality from genetic proxies but cannot fully replace experimental validation — these findings remain unconfirmed in laboratory or clinical models.
• All analyses derive from GWAS summary statistics (FinnGen), not direct patient samples, so individual-level confounding cannot be fully excluded.
• Funding was not disclosed by the authors.
• Independent replication in separate cohorts has not yet been reported.

Bidirectional Mendelian randomization study using large-scale GWAS summary data (FinnGen), analyzing over 2,800 omics variables to infer causal relationships between gut microbiota, immune cell phenotypes, inflammatory proteins, metabolites, and IPF risk. The MR design supports directional inference over standard observational association, but all findings are derived from genetic proxies rather than direct biological measurement, and the proposed causal chains require experimental validation. This is a single-study analysis with no independent replication reported to date.