On a Source Error in Biagioli et al., “Beyond Ketosis”

The review by Biagioli et al. reframes dietary therapy in epilepsy as modulation of the microbiota–gut–brain axis rather than ketosis alone. That correction is useful. One line of interpretation inside it is not, and it carries a clinical recommendation that readers may act on before the evidence supports it.

The review associates ketogenic therapy with a reduction in microbial diversity and treats that reduction as a trade-off against gut ecosystem stability, one that may become maladaptive over time. That characterization rests most directly on a cohort the review itself cites, a study that reported no significant change in alpha diversity after three months of ketogenic diet and left open whether the microbial shift is necessary for the treatment to work. The review elsewhere invokes preclinical evidence that ketogenic-associated microbiota mediate seizure protection. Its later reading of the same class of remodeling as ecosystem harm leaves that tension unresolved.

The review also carries a concrete source error. It attributes to treatment responders a set of taxa that its cited pediatric study reported as enriched in non-responders, which reverses the clinical meaning of the finding.

Correcting the record matters because clinicians managing drug-resistant epilepsy rely on the published literature, and advice to add fiber for the purpose of preserving microbial diversity during treatment could work against a mechanism implicated in seizure control. The concern reaches past epilepsy. Clinical fields now applying ketogenic therapy to other conditions draw on the same evidence base, and an unsupported claim about the microbiome travels with it.

I submitted the following as a formal Comment to Nutrients. The journal declined to send it out for review. I am publishing it here.

---

Comment on Biagioli et al. Beyond Ketosis: Dietary Therapies and the Microbiota–Gut–Brain Axis in Epilepsy. Nutrients, 2026, 18, 2151

Nicole Laurent, Independent clinician-researcher

Abstract

Biagioli et al. usefully frame dietary therapy in epilepsy through the microbiota–gut–brain axis rather than ketosis alone. This comment examines their interpretation of ketogenic diet-associated microbial remodeling, focusing on claims that reduced microbial diversity may represent an unfavorable or maladaptive trade-off. It argues that the cited evidence does not establish preservation of microbial diversity as a therapeutic target, that the review misattributes to treatment responders a set of taxa that its cited pediatric study reported as enriched in non-responders, and that fiber recommendations to preserve diversity rest on an unestablished target, with the burden on those who advise the change to show it does not compromise seizure control.

Keywords: ketogenic diet; epilepsy; gut microbiota; microbiota–gut–brain axis; microbial diversity; dietary fiber; treatment response; seizure control

The reframing of dietary therapy as modulation of the microbiota–gut–brain axis is a useful correction to a ketosis-only account of how these diets work. One line of interpretation in the clinical and translational section works against the review’s own evidence and against the sources it cites, and it carries a clinical recommendation that warrants scrutiny before readers act on it.

The reduction in microbial diversity that the review associates with ketogenic diet (KD) therapy is framed in terms of “potential trade-offs” against gut ecosystem stability and as a “potentially unfavorable” alteration of the gut ecosystem. Prolonged therapy, the review adds, “could become maladaptive over time.” Each characterization treats preservation of microbial diversity as the direction of benefit, without establishing that a higher-diversity state is the appropriate endpoint for a patient whose seizures were uncontrolled before treatment. The review hedges the claim, but the hedging does not cure the premise. Terms such as “potentially unfavorable” and “could become maladaptive” still treat departure from baseline diversity as presumptively suspect, even though the review has not shown that the pretreatment microbial state is the appropriate endpoint for a patient whose seizures were uncontrolled in that state.

The review supplies the evidence that unsettles that reference. A ketogenic-associated microbial community is presented, in mouse models, as necessary and sufficient for seizure protection. Germ-free and antibiotic-treated animals did not achieve KD-mediated protection, co-colonization with KD-associated Akkermansia muciniphila and Parabacteroides restored protection in antibiotic-treated mice, and both fecal transfer of the KD microbiota and treatment with these two taxa conferred seizure protection to control-diet animals. Microbial remodeling appears in the same discussion among the changes that may contribute to seizure control, alongside enrichment of taxa such as Bacteroides and Akkermansia (Biagioli et al., 2026). The same treatment-associated microbial shift cannot be invoked as part of the therapeutic mechanism and then treated as evidence of ecosystem harm without explaining how those two readings are to be reconciled.

“The gut microbiota is necessary and sufficient for seizure protection in two mouse models of intractable epilepsy.”

Olson et al., 2018. https://doi.org/10.1016/j.cell.2018.04.027

The trade-off claim rests on a narrower Lindefeldt et al. cohort than the review’s generalization allows. In twelve children treated for three months, alpha diversity was not significantly changed, although taxonomic and functional composition changed. The authors reported decreased relative abundance of Bifidobacterium, Eubacterium rectale, and Dialister, increased relative abundance of Escherichia coli, and functional shifts including reductions in carbohydrate-metabolism pathways. They raised concern about loss of health-promoting, fiber-consuming bacteria while leaving open whether those microbial shifts are necessary for the diet’s therapeutic effect. The review keeps the concern and drops the open question, generalizing from compositional remodeling to reduced overall diversity and potential ecosystem harm.

“Alpha diversity is not changed significantly during the [ketogenic] diet.”

Lindefeldt et al., 2019. https://doi.org/10.1038/s41522-018-0073-2

A second cited pediatric study also needs more careful handling. Zhang et al. (2018) reported numerically lower alpha-diversity indexes after six months of ketogenic therapy, with Chao1, Shannon, and Simpson indexes decreasing from 139.2, 4.2, and 0.9 before KD to 126.2, 3.9, and 0.8 after KD, although the alpha-diversity comparisons were not statistically significant. More importantly, the taxa described in the review as enriched in responders were reported by Zhang et al. as enriched in the non-responsive group. These taxa were Clostridiales, Clostridia, Ruminococcaceae, Lachnospiraceae, Alistipes, and Rikenellaceae.

“Zhang et al. observed that six months of KD induced microbial shifts in responders, including enrichment of Alistipes, Lachnospiraceae, Ruminococcaceae, and Rikenellaceae, associated with ≥50% seizure reduction.”

Biagioli et al., 2026. https://doi.org/10.3390/nu18132151

This is not a minor taxonomic discrepancy. It attributes to treatment response a microbial profile that Zhang et al. associated with non-response.

“Compared to effective patients, the non-responders after KD had significantly increased relative abundances of the Clostridiales, Clostridia, Ruminococcaceae, Lachnospiraceae, Alistipes, and Rikenellaceae.”

Zhang et al., 2018. https://doi.org/10.1016/j.eplepsyres.2018.06.015

The recommendation therefore depends on an inference the cited evidence has not established. Readers are advised that ketogenic protocols should, where feasible, incorporate fiber-rich, microbiota-supportive foods to preserve microbial diversity, a correction aimed at the same class of microbial shifts that the review presents as potentially involved in seizure control, qualified by the phrase “without compromising therapeutic efficacy” (Biagioli et al., 2026). The qualification concedes that the correction could compromise efficacy. Recommending fiber to preserve microbial diversity presupposes that diversity preservation is the target for a person in treatment, which is the proposition under dispute. A recommendation to add fiber for the purpose of preserving or redirecting the treatment-associated microbiome is itself a microbiota-targeted intervention, not a neutral dietary housekeeping measure.

The same caution applies in the other direction. Current evidence does not establish the long-term safety of the ketogenic-associated microbial state, and neither the pretreatment microbiome nor the treatment-associated microbiome can be designated the healthy endpoint for a treated patient on current evidence. Without evidence identifying the appropriate microbial endpoint, directional language and directional intervention go beyond what is known. Olson et al. (2018) conclude that further studies are needed before microbe-based treatments can be safely and effectively applied clinically. Lindefeldt et al. (2019) likewise state that fiber supplementation might seem advisable, but first ask whether the KD-induced microbial changes are part of the therapeutic effect.

“Supplementing patients on KD with such fibers might seem advisable… However, we first need to understand the role that changes in the gut microbiota during KD play in its therapeutic effect.”

Lindefeldt et al., 2019. https://doi.org/10.1038/s41522-018-0073-2

That is the caution the review’s clinical language should have preserved.

For a child who reached seizure control on a ketogenic protocol, advice to add fiber for the stated purpose of preserving microbial diversity treats diversity itself as the outcome to protect, even though the diet-associated microbial community may be part of the seizure-control signal. The burden of showing that such intervention does not compromise seizure control belongs to whoever recommends the change. The review should have preserved that uncertainty rather than converting it into directional clinical advice.

---

Neither the pretreatment microbiome nor the treatment-associated microbiome has been established as the right endpoint for a patient who reached seizure control on ketogenic therapy. Until that is known, advice to move the microbiome toward diversity preservation runs ahead of the evidence, and the burden of showing the change does not cost seizure control belongs to whoever recommends it.

A source error in a founding review does not stay put. The taxa the review assigns to responders were the non-responder taxa in the study it cites, and that reversal will be read and built upon by people who never open the original. Left uncorrected in the literature that newer fields now draw from, a claim like this shapes treatment decisions well past the condition where it started. This could influence metabolic psychiatry practice in a way that the literature does not support and may harm patients.

MDPI’s published ethics policy commits the publisher to a standard it set for itself.

“[MDPI] recognizes our responsibility to correct scientifically relevant errors or ethical issues that have been brought to our attention.”

MDPI Research and Publication Ethics, Updating Published Papers. https://www.mdpi.com/ethics

The error is now on record with the publisher. Their published policy commits them to correcting scientifically relevant errors brought to their attention, and this is one. What remains is for that standard to be applied to this case, in the ordinary course. My hope is that the publisher follows up and issues a correction.

References

Biagioli, V., Matera, M., Imola, I., Mela, F., Lemmi, D., Verrotti, A., & Striano, P. (2026). Beyond ketosis: Dietary therapies and the microbiota–gut–brain axis in epilepsy. Nutrients, 18(13), 2151. https://doi.org/10.3390/nu18132151

Lindefeldt, M., Eng, A., Darban, H., Bjerkner, A., Zetterström, C. K., Allander, T., Andersson, B., Borenstein, E., Dahlin, M., & Prast-Nielsen, S. (2019). The ketogenic diet influences taxonomic and functional composition of the gut microbiota in children with severe epilepsy. npj Biofilms and Microbiomes, 5, 5. https://doi.org/10.1038/s41522-018-0073-2

Olson, C. A., Vuong, H. E., Yano, J. M., Liang, Q. Y., Nusbaum, D. J., & Hsiao, E. Y. (2018). The gut microbiota mediates the anti-seizure effects of the ketogenic diet. Cell, 173(7), 1728–1741.e13. https://doi.org/10.1016/j.cell.2018.04.027

Zhang, Y., Zhou, S., Zhou, Y., Yu, L., Zhang, L., & Wang, Y. (2018). Altered gut microbiome composition in children with refractory epilepsy after ketogenic diet. Epilepsy Research, 145, 163–168. https://doi.org/10.1016/j.eplepsyres.2018.06.015

Epilepsy Dietitians Already Deliver What Psychiatry Needs