Leaky Barriers and the Mind: What Blood Biomarkers Really Reveal About Psychiatric Disorders

A sweeping new analysis has put one of psychiatry’s most provocative hypotheses to a rigorous statistical test, and the results are more nuanced than the headlines that have long surrounded the so-called gut–brain connection. In a systematic review and meta-analysis published in BMC Psychiatry, researchers led by Oguzhan Kılınçel of İstanbul Gelişim University examined whether two molecular gatekeepers of the body’s barrier systems—zonulin, a putative marker of intestinal permeability, and claudin-5, a core structural protein of the blood–brain barrier—show consistent alterations in people living with psychiatric disorders. Their conclusion is a careful blend of signal and skepticism: circulating zonulin levels do appear elevated on average across psychiatric populations, but the evidence is far too fragile, and the measurement tools far too imprecise, to declare a definitive link between leaky barriers and mental illness.

The idea of a barrier dysfunction axis rests on an elegant biological premise. The human body is lined and walled by epithelial and endothelial barriers that decide what passes from the outside world into the interior. In the gut, a single layer of epithelial cells, sealed by protein complexes called tight junctions, separates the trillions of microbes and dietary antigens in the intestinal lumen from the immune-rich tissue beneath. In the brain, the blood–brain barrier performs an analogous gatekeeping role, with specialized endothelial cells joined by tight junctions that exclude most blood-borne molecules from the neural parenchyma. If both barriers become permeable, the theory goes, microbial products such as lipopolysaccharide could translocate from the gut into circulation, trigger systemic inflammation, and potentially compromise the brain’s defenses—creating a plausible mechanistic bridge between digestive physiology and psychiatric symptoms.

Zonulin has become the most contested character in this narrative. Discovered in the context of studies on cholera toxin and coeliac disease, it is often described in the literature as a modulator of intestinal tight junctions, and commercial enzyme-linked immunosorbent assays promise a simple blood measurement of gut leakiness. Claudin-5, by contrast, is an established molecular workhorse: it is the tight junction protein whose expression is indispensable for the selective permeability of the blood–brain barrier, and circulating fragments or soluble forms have been proposed as indirect windows onto barrier integrity. The Turkish research team reasoned that if a genuine barrier dysfunction axis operates in psychiatry, then these two biomarkers, measured in blood, should show coordinated and reproducible deviations in patients compared with healthy controls.

To test this, the investigators followed the PRISMA 2020 reporting framework and searched PubMed/MEDLINE, Web of Science, and Scopus for observational studies published between January 2020 and February 2026 that measured serum zonulin or serum claudin-5 in psychiatric patients and healthy comparison groups. Fifteen studies met the eligibility criteria, spanning diagnoses that included schizophrenia, major depressive disorder, bipolar disorder, obsessive-compulsive disorder, pediatric obsessive-compulsive disorder, and attention-deficit/hyperactivity disorder. The team pooled standardized mean differences using Hedges’ g, a statistic that corrects for small sample sizes, and applied a random-effects model with the DerSimonian–Laird estimator, which assumes that the true effect may differ from study to study rather than being a single fixed quantity.

The zonulin result was striking at first glance. Ten studies contributed to the pooled estimate, and psychiatric populations showed substantially higher assay-reported circulating zonulin levels than healthy controls, with a Hedges’ g of 0.97 and a 95 percent confidence interval of 0.46 to 1.48, corresponding to a p value below 0.001. In practical terms, a standardized difference approaching one suggests that the average patient’s zonulin value sat near the eighty-third percentile of the control distribution—a large effect by conventional benchmarks in biomedicine. Yet the same analysis revealed a critical caveat: the heterogeneity statistic, I², reached 91.2 percent, meaning that more than nine-tenths of the variability between studies reflected genuine differences in populations, methods, and assays rather than random sampling noise alone.

The claudin-5 picture was even more equivocal. Eight studies entered that meta-analysis, and the pooled difference between patients and controls was small and statistically non-significant: Hedges’ g of 0.25, with a confidence interval spanning from −0.21 to 0.71 and a p value of 0.280. Heterogeneity was again substantial at 89.2 percent, and the study-level estimates pointed in both directions, with some studies reporting higher claudin-5 in patients and others reporting lower values. For a protein so central to blood–brain barrier architecture, this scatter suggests either that circulating claudin-5 is a poor proxy for what happens at the barrier itself, or that barrier changes in psychiatric illness are too heterogeneous, episodic, or illness-stage-dependent to register in a pooled cross-sectional comparison.

The authors applied the GRADE framework to rate the certainty of the evidence, and the verdict was sobering: very low certainty for both biomarkers. Publication bias was probed with funnel plots and Egger’s test, and robustness was stress-tested with leave-one-out sensitivity analyses, in which each study was removed in turn to see whether the pooled estimate collapsed. Subgroup analyses attempted to parse the heterogeneity by diagnosis and other study characteristics. None of these procedures could rescue the findings from their fundamental fragility, and the team was explicit that the elevated zonulin signal cannot be interpreted as direct proof of increased intestinal permeability in psychiatric disorders.

A central reason for that caution lies in the analytical chemistry of zonulin measurement itself. Commercial zonulin ELISA kits have been shown in independent laboratory work to cross-react with other proteins, and there is ongoing debate about whether the assays detect a single defined molecule at all. The meta-analysis therefore deliberately framed its outcome as assay-reported circulating zonulin, a phrase that acknowledges the measurement may capture a family of related or unrelated proteins rather than the specific permeability regulator implied by the name. When the underlying assay is uncertain, a large pooled effect size inherits that uncertainty wholesale. This is a recurring lesson in biomarker research: statistical significance in a meta-analysis is only as meaningful as the validity of the instrument that generated the raw numbers.

What, then, remains of the barrier dysfunction axis? The authors position it explicitly as a hypothesis-generating framework rather than a tested mechanism. The axis was not directly evaluated in any of the included studies, because none simultaneously measured gut permeability, blood–brain barrier integrity, and clinical outcomes within a single longitudinal design. Cross-sectional blood biomarkers, however convenient, cannot establish the direction of causality, cannot distinguish cause from consequence of illness or its treatment, and cannot capture the dynamic fluctuations that a truly leaky barrier might exhibit during acute relapse versus remission. The field’s next steps, the analysis implies, should include standardized assay validation, longitudinal cohorts, and studies that pair circulating biomarkers with direct functional measures of permeability.

For a science-attentive public, the study is a case study in disciplined skepticism. It neither dismisses the gut–brain axis as fashionable speculation nor endorses the popular leap from elevated blood proteins to a leaky gut driving depression or psychosis. Instead, it quantifies exactly how much the evidence supports—and how much it does not—using transparent effect sizes, confidence intervals, and heterogeneity statistics that anyone can scrutinize. The elevated zonulin signal across heterogeneous psychiatric populations is a genuine and reproducible observation worth pursuing; the leap to barrier failure as a mechanism of mental illness remains, for now, an unproven and testable hypothesis. In a research landscape crowded with overconfident claims about the microbiome and the mind, that measured tone may be the most newsworthy finding of all.

Subject of Research: Circulating zonulin and claudin-5 as biomarkers of gut and blood–brain barrier dysfunction in psychiatric disorders

Article Title: The proposed barrier dysfunction axis in psychiatric disorders: a systematic review and meta-analysis of circulating zonulin and claudin-5

Article References: Kılınçel, O., Bulut, F., Aslan, E. A., & Kılınçel, Ş. (2026). The proposed barrier dysfunction axis in psychiatric disorders: a systematic review and meta-analysis of circulating zonulin and claudin-5. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08694-7

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08694-7

Keywords: zonulin, claudin-5, gut–brain axis, blood–brain barrier, intestinal permeability, psychiatric disorders, meta-analysis, systematic review, tight junctions, schizophrenia, major depressive disorder, biomarkers