Evidence for altered intestinal permeability in human research

Evidence for altered intestinal permeability in human disease has expanded considerably.

Changes in functional permeability tests, circulating or faecal biomarkers, epithelial-injury markers and microbial-translocation measures have been reported across gastrointestinal, metabolic, inflammatory, cardiovascular and neuropsychiatric conditions.^¹–⁵

Taken together, this literature supports an important conclusion: intestinal-barrier biology is altered in association with a range of human conditions.

But that conclusion is narrower than it is sometimes presented.

Human evidence does not yet support the idea that a single state called “increased intestinal permeability” is a uniform mechanism across diseases. Nor does the observation of altered permeability or a barrier-associated biomarker establish that barrier dysfunction caused the condition in which it was measured.

The central problem is therefore not whether associations exist.

They do.

The harder question is:

What has human evidence actually earned the right to say about those associations?

Human relevance comes with reduced experimental control

Human research provides something no cell system or animal model can provide: direct evidence from human physiology and disease.

But that relevance comes with a methodological cost.

Age, diet, medication, adiposity, infection, physical activity, alcohol exposure, smoking, stress, sleep, disease activity and many other factors may influence intestinal physiology. Microbial composition, immune state and metabolic health differ substantially between individuals.

These factors are not experimental noise around an otherwise uniform system.

They are part of human biology.

The challenge is that many of them can also confound the relationship being studied.

If people with a disease have higher concentrations of a barrier-associated biomarker than healthy controls, several explanations remain possible. The disease may alter the intestinal barrier. A treatment may affect both disease and barrier biology. Diet or medication may differ between groups. Inflammation may influence the barrier and the biomarker simultaneously. The measured marker may reflect a process related to permeability rather than permeability itself.

A human association is therefore biologically meaningful without necessarily being mechanistically specific.

“Intestinal permeability” is not one outcome

A second problem is measurement heterogeneity.

Human studies described as investigations of intestinal permeability may use lactulose–mannitol or other sugar-probe tests, serum zonulin-related assays, I-FABP, LBP, D-lactate, diamine oxidase, endotoxin-related measures, citrulline or combinations of these approaches.^⁶–¹³

These measurements do not interrogate the same biological property.

Functional probe tests estimate the passage of defined molecules under specified conditions.

I-FABP is associated primarily with enterocyte injury.

Citrulline can provide information related to functional enterocyte mass.

LBP and related measures concern exposure and host response to bacterial products.

Zonulin-related assays raise distinct questions concerning tight-junction regulation and analytical specificity.

Recent reviews of human permeability assessment increasingly emphasise this multidimensionality rather than treating these measures as interchangeable.^⁶,⁷

This matters when studies are compared or combined.

If one study defines increased permeability by a sugar-probe ratio and another defines it by serum zonulin, both may use the same label while measuring substantially different biological phenomena.

The apparent consistency of the terminology can therefore exceed the consistency of the evidence.

Cross-sectional studies establish association, not direction

Much of the human literature is observational and cross-sectional.

Such studies are valuable.

They can identify populations in which barrier-associated measures differ, quantify associations with symptoms or disease severity and generate mechanistic hypotheses.

What they generally cannot establish is causal direction.

Consider a study in which people with an inflammatory disorder show both altered intestinal permeability and increased systemic inflammation.

At least three broad possibilities remain:

intestinal-barrier alteration contributes to inflammation;

inflammation alters intestinal-barrier function;

or another process influences both.

All three may even occur simultaneously.

A cross-sectional study cannot determine which sequence came first.

This becomes especially important when the intestinal barrier is incorporated into larger mechanistic narratives involving microbiota, immune activation, metabolism or brain function.

The more steps in the proposed pathway, the more evidence is required to establish the direction of each step.

Temporality strengthens inference

Longitudinal studies improve this situation because they allow biological changes to be placed in time.^¹⁴

If altered permeability precedes disease onset, the interpretation differs from a finding that appears only after disease is established.

If permeability changes alongside disease activity and normalises during remission, another type of relationship becomes plausible.

If a barrier-associated measure predicts subsequent clinical change, the temporal sequence becomes more informative still.

But temporality is not causality.

A factor can precede an outcome without causing it.

Longitudinal research reduces some forms of ambiguity; it does not remove confounding or establish mechanism on its own.

This is one reason why human evidence becomes strongest when longitudinal observation, functional measurement and experimental biology converge.

Intervention studies ask a stronger causal question

Intervention studies can move causal inference further.

If an intervention changes a well-validated measure of intestinal permeability and that change is accompanied by a predicted downstream biological effect, the evidence becomes more informative than a cross-sectional association.

Even here, interpretation requires discipline.

An intervention may have multiple effects.

Dietary interventions can alter nutrient exposure, microbiota, microbial metabolites, immune activity, body weight and intestinal physiology simultaneously. Probiotics or prebiotics may affect several biological pathways. Pharmacological interventions may have systemic effects unrelated to the barrier.

A change in permeability after intervention therefore demonstrates responsiveness.

It does not necessarily prove that permeability mediated every accompanying clinical effect.

Recent systematic reviews illustrate the problem. A 2024 review of dietary interventions in healthy people found relatively few eligible studies and substantial diversity in interventions and permeability methods; the authors concluded that strong evidence for broad dietary effects on intestinal-barrier permeability was lacking.^¹⁵ Another 2024 systematic review examining prebiotics found similarly limited direct human evidence and highlighted the need for larger studies using standardised permeability protocols.^¹⁶

Negative or uncertain evidence is informative.

It tells us where biological plausibility has moved ahead of demonstrated human effect.

Disease association is not a single category of evidence

Reports of altered permeability now span many clinical fields.

That breadth can create the impression of a common causal mechanism.

But evidence differs substantially between conditions.

In some diseases, particularly those involving the gastrointestinal mucosa, barrier alterations may be closely connected to local pathology.

In others, permeability changes may occur downstream of systemic inflammation, metabolic dysfunction or treatment.

In still others, evidence may consist mainly of differences in indirect biomarkers.

These situations should not be treated as equivalent simply because the phrase increased intestinal permeability appears in each literature.

A 2024 systematic review and meta-analysis in cardiovascular disease illustrates both the strength and the interpretive challenge. The review included 13 studies and 1,321 participants and found higher levels of several reported permeability-associated markers in patients than controls.^¹⁷ Yet the pooled category included biologically different measures — including LPS, D-lactate, zonulin, DAO, LBP, I-FABP and a sugar-probe measure.

The finding supports altered gut-barrier-associated biology in cardiovascular disease.

But combining different markers does not make them measurements of one identical mechanism.

That distinction matters.

Systematic reviews expose the structure of the evidence

Individual studies show observations.

Systematic reviews reveal the architecture of a field.

They show whether different investigators have measured the same biological property, whether populations are comparable, whether methods are standardised, whether findings replicate and whether the evidence base is large enough to support quantitative synthesis.

This was central to my own systematic review and meta-analysis of zonulin-dependent intestinal permeability in children diagnosed with mental disorders.^¹⁸

We screened 833 records and identified five observational studies meeting the inclusion criteria, representing 402 participants. The studies included children diagnosed with ADHD, autism spectrum disorder and obsessive–compulsive disorder. Four studies contributed serum zonulin data to the meta-analysis.^¹⁸

The pooled analysis found higher mean zonulin concentrations in children with mental disorders than in controls.

That is the statistical result.

The interpretation requires another step.

What our meta-analysis did — and did not — establish

The findings supported an association between the clinical populations studied and zonulin-related measurements.

They did not establish that all children with these diagnoses had increased functional intestinal permeability.

They did not establish that the measured zonulin values represented direct quantitative measurements of paracellular flux.

They did not establish that altered intestinal permeability preceded the mental disorder.

And they did not establish that intestinal permeability caused the psychiatric or neurodevelopmental phenotype.

The review itself identified a limited evidence base and methodological inconsistencies, including concerns related to the assessment of zonulin.^¹⁸

This distinction is particularly important in microbiota–gut–brain research.

It is tempting to convert a cross-sectional difference into a complete mechanism:

altered microbiota → increased zonulin → increased intestinal permeability → systemic inflammation → altered brain function → clinical phenotype.

Each arrow is biologically plausible.

The complete sequence is not demonstrated merely because variables at two or three points in that sequence differ between cases and controls.

Every arrow requires evidence.

Meta-analysis cannot repair conceptual heterogeneity

Pooling studies can increase statistical power.

It cannot make biologically different measurements equivalent.

If studies use different assays, different specimens, different age groups, different disease definitions or different analytical platforms, a pooled estimate may still conceal substantial conceptual heterogeneity.

The same applies when different biomarkers are grouped under a common heading such as “intestinal permeability markers”.

Statistical heterogeneity can be quantified.

Conceptual heterogeneity requires scientific judgement.

This is one reason the quality of a meta-analysis depends on more than the number of included participants or the precision of the pooled estimate.

The biological meaning of the measurements must also be sufficiently coherent to justify combining them.

Biomarker validity limits epidemiological inference

Human evidence can never be stronger than the validity of the measurement on which it depends.

This becomes particularly important for biomarkers.

If an assay does not specifically measure the molecule it is assumed to measure, uncertainty enters the evidence chain before any epidemiological analysis begins.

If a marker reflects epithelial injury rather than permeability, an association with disease cannot automatically be interpreted as altered paracellular transport.

If a marker reflects microbial exposure or immune response, it may be highly biologically relevant while still answering a different question.

The distinction between analytical validity, biological validity and clinical association should therefore remain explicit.

A statistically robust association does not correct an ambiguous biomarker.

Population matters

Human evidence is also population-specific.

A finding in children cannot automatically be generalised to adults.

A result in people with active inflammatory disease may not apply during remission.

A finding in severe obesity may not describe metabolically healthy individuals.

Medication use, dietary pattern, geography and microbial ecology may all influence the observed relationship.

This is especially relevant to microbiome-associated barrier research, where background exposures differ substantially across populations.

Generalisation should therefore be treated as an empirical question, not an automatic consequence of statistical significance.

Healthy populations are informative

Studies in healthy people provide an important counterweight to disease-focused research.

If intestinal permeability is a dynamic physiological property, variation should exist even in the absence of disease.

Diet, exercise and other exposures may produce transient changes without implying pathology.

A 2024 systematic review examining dietary effects on intestinal-barrier permeability in healthy individuals found only 12 eligible intervention studies from 3,725 records. Evidence differed by intervention and was generally limited; the authors concluded that there was no strong overall evidence that diet or dietary interventions consistently increase or decrease intestinal-barrier permeability in healthy people.^¹⁵

This is scientifically important.

It reminds us that the absence of a simple intervention effect is part of the evidence base too.

Human evidence becomes stronger through triangulation

The strongest human inference rarely comes from one marker or one study design.

Confidence increases when several independent forms of evidence converge.

A functional permeability alteration is demonstrated with a validated method.

The finding is reproduced.

A plausible biological mechanism is supported experimentally.

Temporal evidence places the alteration before or alongside the relevant outcome.

Intervention modifies the proposed pathway in the predicted direction.

And the clinical association persists after important confounders are addressed.

Not every research question will satisfy all of these conditions.

The point is not to create an impossible standard.

It is to recognise what each additional layer contributes.

Evidence becomes stronger not simply because there is more of it, but because different forms of evidence constrain alternative explanations.

Absence of proof is not proof of absence

Scientific caution should not be confused with dismissal.

Methodological limitations do not mean the intestinal barrier is irrelevant to human disease.

They mean the magnitude, direction and causal importance of its role must be established rather than assumed.

A heterogeneous literature may contain a genuine biological signal.

An imperfect biomarker may still correlate with meaningful physiology.

An observational association may correctly identify a pathway that later proves causal.

But these possibilities are reasons to continue investigating, not reasons to describe the mechanism as already established.

Uncertainty is not the opposite of knowledge.

It is part of its structure.

What has the evidence earned the right to say?

Human research supports several conclusions.

Intestinal-barrier-associated measures differ between clinical and healthy populations in multiple research contexts.^{¹–⁴,¹⁷}

Functional permeability can be altered in humans.

Barrier-related biomarkers can provide useful biological information.

The intestinal barrier interacts with inflammatory, metabolic, microbial and environmental processes.^¹⁹

But the literature also supports equally important limits.

Different permeability methods are not interchangeable.^⁶,⁷

A biomarker is not automatically a functional permeability measurement.

Cross-sectional association does not establish temporal order.

Temporal order does not by itself establish causality.

A pooled association does not remove assay limitations or biological heterogeneity.

And evidence that intestinal permeability is altered in a disease does not establish that permeability caused the disease.

The question is therefore not whether the intestinal barrier matters.

It clearly does.

The question is what role it plays, in whom, at what stage, measured by which method, and with what level of causal evidence.

That is a more difficult question.

It is also the one human research now needs to answer.