Live vs Pasteurized Akkermansia: What the Research Show

In 2004 a Dutch team named a bacterium that eats the mucus lining of your gut. In 2017 somebody heated it until nothing was alive, gave it to mice anyway, and it still worked. Twenty-two years later, two companies are selling opposite answers to the same question.

Lit Review Friday · So What's the Deal with Akkermansia Anyways? · Published 2026 · Reading time ~20 minutes

---

The Organism That Refused to Behave Like a Probiotic

Most probiotic stories are simple. You find an organism, you show it does something useful, you keep it alive long enough to reach somebody's gut, and you sell it. The living cell is the product. Everything about manufacturing, packaging and shelf life exists to protect that.

Akkermansia muciniphila broke that story, and it broke it from the inside, in the labs of the people who had spent their careers on it. The organism works when it is dead. Not as well in every setting, not in every trial, and not for every mechanism it possesses. But well enough, in humans, that a patent issued in 2026 covers both the living and the pasteurized form.

This is the story of how that happened over twenty-two years, what the evidence actually supports as opposed to what gets repeated about it, and what it says about a field now trying to bottle organisms that die on contact with air.

---

2004: Somebody Names a Mucus Eater

The founding paper is Derrien and colleagues, 2004, in the International Journal of Systematic and Evolutionary Microbiology. A team at Wageningen in the Netherlands isolates an anaerobic organism from human stool that grows on mucin, the heavily glycosylated protein that makes up intestinal mucus. They name it Akkermansia muciniphila, after the Dutch microbiologist Antoon Akkermans. Muciniphila means mucus loving.

In 2004 that reads as bad news. The mucus layer is the physical barrier between the contents of your colon and the cells of your colon. An organism whose entire metabolic strategy is eating that layer sounds like something you would want less of, not more.

Worth noting for what comes later: Willem de Vos is on this paper. His name appears on nearly every chapter of this story, right through to the company selling the product and the patent that may decide who else can.

---

Why Is Eating Your Mucus Layer a Good Thing?

The turn comes in 2013. Everard and colleagues, working with Patrice Cani at UCLouvain, publish in PNAS. They report that Akkermansia is depleted in obesity in both mice and people, and that restoring it in mice improves the metabolic picture: better barrier function, less endotoxemia, improved glucose handling.

There is a second thing about this organism that is easy to miss. Nearly every other bacterium in your gut is eating what you ate. This one is eating you. It runs on mucin your own goblet cells secrete, which makes it the rare community member whose food supply does not depend on your diet at all. And what it releases when it grazes, acetate and propionate, is exactly what the butyrate producers downstream consume. It converts a substance your body makes into raw material for somebody else's metabolism. That is what "keystone" actually means: remove it, and processes you were not watching stop working.

And the mucus eating gets read a second way. Consider what happens to a mucus layer that nothing grazes. It is not a wall, it is a secretion, continuously produced by goblet cells and continuously turned over. Grazing is part of how that turnover happens. A layer that accumulates without renewal is not healthier. It is stagnant.

This is the pattern we keep returning to on this show, and it is worth stating flatly: it is not the name on the bug. Ask what an organism makes, and ask what conditions it is making it in. The same organism, performing the same behavior, reads as helpful or harmful depending entirely on context and dose. Akkermansia is the cleanest available example. It has been reported as elevated in multiple sclerosis, as necessary for immunotherapy response, and as damaging when it over-colonizes a gut whose community has been flattened by antibiotics. All of those can be true, because none of them are properties of the organism. They are properties of a relationship.

---

2017: The Control That Misbehaved

Plovier and colleagues publish in Nature Medicine in 2017. The stated goal is mechanism: what, specifically, about this organism produces the metabolic effect? They identify a protein on its outer membrane, designated Amuc_1100, and show that the purified protein alone improves metabolism in obese and diabetic mice. It acts through TLR2, one of the pattern-recognition receptors your epithelial and immune cells use to read microbial surfaces.

Then the result that reorganized the field. They also tested the pasteurized bacterium, heated until nothing in it was alive. It worked too. Amuc_1100 is thermostable at pasteurization temperature. The structure that carries the signal survives the heat that kills the cell.

Nobody set out to demonstrate that the dead organism works. It was a control condition behaving in a way the field's operating assumption did not predict.

---

Does Dead Akkermansia Really Beat Live Akkermansia?

This is the most repeated claim in the entire Akkermansia literature, and it deserves a much closer reading than it usually gets.

The source is Depommier and colleagues, 2019, in Nature Medicine: a proof-of-concept exploratory study in overweight and obese volunteers, three arms, placebo against live Akkermansia against pasteurized Akkermansia. The reported result is that the pasteurized arm separated from placebo on insulin sensitivity and the live arm did not.

Read the arms themselves and the picture changes. The live arm's point estimate for improvement in insulin sensitivity was larger than the pasteurized arm's. It failed to reach statistical significance because only nine participants completed that arm, which left the confidence interval around the estimate roughly three times wider.

There is a second detail almost nobody mentions. On insulin resistance score, a different glucose measure reported in the same figure, the live arm did separate from placebo, at p = 0.046, with a larger absolute difference from placebo than the pasteurized arm showed. So it is not that the live preparation failed while the pasteurized one succeeded. Each separated from placebo, on a different measure.

This matters beyond one paper, because it is the difference between "heat-killing preserves the active component in this organism" and "heat-killing is better." The first is supported. The second is not, and the evidence that it is not a general rule is more direct than most people realize.

The same protocol, the opposite result

In 2022, Le Roy and colleagues, working in the same UCLouvain research lineage, characterized a different human gut commensal called Dysosmobacter welbionis, published in Gut. Live, it prevented diet-induced obesity in mice. Then they pasteurized it at 70 degrees Celsius for 30 minutes, the identical protocol that improved Akkermansia, and the benefit was completely abolished.

Same temperature. Same duration. Same research program. Opposite outcome. Heat-killing is not a principle you can apply to a category. It is a result you get one organism at a time, because what survives depends on which molecule was carrying the activity in the first place.

And heat-killed preparations do fail in humans. In 2026, a tyndallized Clostridium butyricum postbiotic that had performed well in animal models went into a randomized, double-blind, placebo-controlled trial in 96 adults with knee osteoarthritis, dosed for twelve weeks. Both groups improved. Neither beat the other, with no significant between-group difference on any clinical endpoint or biomarker. The trial was industry-authored, which makes a null result more credible rather than less.

---

Meanwhile, in San Francisco

While the Belgian and Dutch groups were working out mechanism, a different bet was being made in California.

Pendulum Therapeutics, founded in 2013 as Whole Biome, was started by three scientists out of Pacific Biosciences: Colleen Cutcliffe, who holds a PhD in biochemistry and molecular biology from Johns Hopkins, John Eid and Jim Bullard. The Pacific Biosciences background is the part worth pausing on. Long-read DNA sequencing was the tool that finally let researchers resolve which organisms were actually present in a gut sample at sufficient resolution to build a product around. They came out of the room where that capability was being built.

Their clinical work, published by Perraudeau and colleagues in 2020 in BMJ Open Diabetes Research and Care, is a twelve-week randomized, double-blind, placebo-controlled study in 76 adults with type 2 diabetes. The tested formulation, WBF-011, is a consortium rather than a single organism: inulin plus Akkermansia muciniphila, Clostridium beijerinckii, Clostridium butyricum, Bifidobacterium infantis and Anaerobutyricum hallii.

A p-value of exactly 0.0500 is worth showing rather than rounding, and so is the fact that it is one-sided. The trial's stated analysis plan tested only for improvement rather than for a difference in either direction, which means a one-sided 0.0500 corresponds to roughly a two-sided 0.10. It sits precisely on a conventional threshold, but not the one most readers picture. That is an honest picture of where the evidence stands at the frontier of this field, and the authors say so themselves. It is also worth noting the manufacturing context the paper describes: producing consistent live-cell concentrations of strict anaerobes in a facility operating under current Good Manufacturing Practice was itself one of the study's stated objectives. Keeping these organisms alive is not a solved problem. It is the problem.

Pendulum has raised more than $100 million across its rounds from investors including Sequoia Capital, Khosla Ventures, Meritech Capital, True Ventures and the Mayo Clinic.

---

Is There Any Science on the Side of Keeping It Alive?

Yes, and this is where the story stops being a straightforward argument for postbiotics.

In 2021, Yoon and colleagues published in Nature Microbiology a description of a different protein from Akkermansia, called P9, which induces GLP-1 secretion and improves glucose homeostasis. P9 is secreted. It is not a structural component of the cell surface. It is something the organism manufactures and releases, which means a dead cell is not producing any.

In June 2026, a group at MD Anderson added a different kind of evidence to the live side, and it arrived from an unexpected direction. Barrodia and colleagues, writing in PNAS, were studying why fasting protects the small intestine from radiation damage. Fasting enriched Akkermansia. Depleting the organism abolished the protection. Putting it back restored survival and intestinal integrity. That is a necessity test and a sufficiency test in the same experiment.

Fasting raised propionic acid, one of the short-chain fatty acids Akkermansia liberates from mucin. Propionate then drove histone acetylation in intestinal stem cells, remodeling which promoters and enhancers those cells could reach and expanding a population of primed, persister-like stem cells. A bacterial metabolite reached into the nucleus and changed what a cell was capable of becoming.

This is a mouse study, and radiation injury is a narrow setting, so hold it loosely. But notice what the axis requires. Propionate has to be made, not carried, and making it requires an organism that is still metabolizing. Nothing in this pathway works with a dead cell.

Once you have all of this on the table, the live-versus-dead question stops being a matter of opinion and becomes a question you can answer mechanism by mechanism.

The honest answer to "does it work dead" is therefore partly, and you can specify which parts. That is a considerably more useful answer than either camp's headline, and it is the answer the evidence actually supports.

---

2026: Two Trials, One Hit, One Miss

In 2026 the same senior research group published two human randomized trials of pasteurized Akkermansia that reached different conclusions.

The one that worked

Mount and colleagues, Nature Medicine, 2026. Ninety adults with overweight or obesity completed an eight-week low-energy diet achieving at least 8% weight loss, then entered a 24-week maintenance phase on either pasteurized Akkermansia muciniphila or placebo. The primary outcome was body-weight change during maintenance.

Regain was 1.2 ± 0.7 kg on the pasteurized organism against 3.2 ± 0.4 kg on placebo, p = 0.012. Net weight loss from baseline to end of maintenance was greater in the treated group by 3.1 ± 0.7 kg, p = 0.009. No treatment-related serious adverse events. Initial Akkermansia abundance was associated with cardiometabolic response.

The one that missed, which is the more interesting of the two

Suenaert and colleagues, Gut Microbes, 2026. A double-blind, placebo-controlled, multicenter trial across Ireland and Germany in 142 adults with metabolic syndrome, four months of daily pasteurized Akkermansia. The primary endpoint, whole-body insulin sensitivity measured by the Matsuda index, did not differ from placebo in the intention-to-treat population.

Then the exploratory analyses, which need care. Hepatic insulin sensitivity improved in the prediabetic subgroup by 12% at three months, and in participants aged 63 and older, but at p = 0.054 that did not reach significance, and the paper itself calls it almost significant. Post-challenge GLP-1 excursion did increase over placebo across the whole group at p < 0.01, though that was measured at day 90 and the effect had faded by the four-month endpoint. And most usefully, in participants whose baseline Akkermansia gene counts were low, there were significant improvements in insulin sensitivity and GLP-1 excursion, reduced trunk fat, and a weight trend at p = 0.06.

The authors offer their own explanation for the miss, and it may be the most useful sentence in the paper. This cohort arrived with unexpectedly high baseline Akkermansia, in some cases exceeding geographically matched healthy controls, which they write likely limited the room for improvement in the overall analysis. The trial did not so much fail to find an effect as enrol a population with little room to show one.

The authors' own conclusion is that the most significant improvements occurred in subjects with low baseline intestinal Akkermansia, who were, in their words, apparently truly in need of this intervention.

This is the finding to carry forward, because it generalizes far beyond one organism. A supplement that supplies something is most likely to register in someone who lacks it. Trials that enroll unselected populations will therefore dilute real effects into null results, and almost nobody knows their own baseline for any of this. It is also a commercially inconvenient thing to be true, which is one reason it is worth stating clearly.

One more thing belongs on the record here. The scientist who co-named this organism in 2004 is a co-founder of the company selling the pasteurized version today, and is a senior author across nearly every paper described above. That is a real conflict of interest and it should factor into how you weigh the evidence. It should also factor in that the trial which missed its primary endpoint came from that same group, and they published it.

---

What Is Coming Next: The Keystones We Still Cannot Manufacture

In April 2026 a practitioner-channel brand launched a product that is worth understanding regardless of what you think of it, because it shows exactly where the manufacturing frontier now sits.

It contains five keystone commensal organisms drawn from a single screened healthy donor: Akkermansia muciniphila, Bifidobacterium adolescentis, Bifidobacterium longum, and then two that the company says have never appeared in a commercial product before, Faecalibacterium prausnitzii and Roseburia intestinalis. Those last two are among the most significant butyrate producers in the human colon, and they are extremely oxygen sensitive. Getting them into a shelf-stable capsule required what the company describes as a patented oxygen-free manufacturing process.

Two details on the label repay attention.

Behind these are the organisms with strong data and no product at all. The next-generation probiotic literature keeps naming the same candidates: Christensenella minuta, Anaerobutyricum hallii, Prevotella copri, Ruminococcus bromii, several Bacteroides species, and Parabacteroides goldsteinii, whose autoclaved form retains anti-inflammatory activity and which we covered in an earlier episode. Most of them are strict anaerobes. Most of them are difficult or impossible to manufacture at scale with current methods.

There is a real irony in the timing. One part of the field is deploying patented engineering to keep fragile anaerobes alive from factory to gut, at precisely the moment another part is producing human trial data suggesting that for at least one organism, the dead preparation carries much of the effect. Both bets are being placed simultaneously, by serious people, with real evidence behind each. Neither is being dishonest. They are working on different parts of the same problem.

---

What This Means for How We Think About Postbiotics

The Akkermansia story is the clearest worked example we have of a principle that keeps recurring across this series: the active ingredient is frequently the chemistry, not the cell.

But the version of that principle this literature actually supports is narrower and more interesting than the slogan. It is not that dead is better. It is that different mechanisms have different requirements, and heat-killing preserves some and destroys others, organism by organism and molecule by molecule. Dysosmobacter lost everything at the same temperature that Akkermansia survived. The tyndallized C. butyricum trial found nothing in 96 people.

And underneath that sits the finding from the trial that missed. Whether an intervention registers depends on the state of the ecosystem receiving it. This is the same shape as the argument we made about butyrate in the previous episode: the precursor is necessary but never sufficient, and the environment decides what actually gets made. The food is the easy part. The hard part is the ecosystem.

Where this leaves the postbiotic argument

If a substantial part of what a beneficial organism does is carried by molecules rather than by colonization, then supplying those molecules directly is a reasonable thing to investigate. That is the logic behind full-spectrum postbiotics, and this literature is consistent with it.

We work on that thesis at Thaena. A full-spectrum postbiotic derived from screened healthy human donors is an attempt to capture something a single organism cannot: the emergent chemistry that only appears in a functional community, delivering more than 10,000 molecular signals rather than one protein from one species. We believe that is a plausible mechanism. It is a thesis, not a proven clinical outcome, and the human trial data needed to validate it is part of what we are working toward. We have not run that trial. We would like to.

It also needs saying plainly that nothing in this article is evidence about ThaenaBiotic. The trials described here studied a different preparation, a single organism, in populations selected for metabolic conditions, and their results belong to that preparation. ThaenaBiotic is one tool in your toolbox, and the interesting thing about this literature is what it teaches about the category, not what it implies about any product in it.

---

The Honest Limitations

The foundational human study was exploratory and small. Depommier 2019 describes itself as proof-of-concept. Nine completers in the live arm is not enough to conclude anything about the live preparation, in either direction.

The most-cited comparison was never made. Pasteurized and live were each compared to placebo, not to each other.

The 2026 subgroup findings are exploratory. The Gut Microbes trial missed its pre-specified primary endpoint. Everything reported after that is hypothesis-generating and requires prospective confirmation in a trial designed to enroll on baseline Akkermansia status. Several of the reported p-values sit at 0.05 or 0.06.

There is a substantial commercial conflict of interest running through the primary literature, disclosed above and worth weighing.

Much of the mechanistic work is in mice. Plovier 2017 is a mouse study. The Dysosmobacter comparison is a mouse study. Mouse metabolic models do not translate reliably to human outcomes.

Akkermansia is not uniformly beneficial. It has been reported as elevated in multiple sclerosis cohorts, and over-colonization following antibiotic disruption has been associated with excessive mucin consumption and barrier damage. The context-dependency is real and is not resolved.

---

Frequently Asked Questions

Does Akkermansia need to be alive to work?

It depends which mechanism you mean. A structural surface protein survives heating and can still be read by the receptors your gut cells use, because those sensors detect molecular shapes rather than whether a cell is alive (Plovier 2017). Other functions, including a secreted protein and the organism's mucin metabolism, require a living cell.

Should I buy the live Akkermansia or the pasteurized one?

No published trial has compared the two head to head, so the evidence cannot rank them. What it can do is tell you which functions each preparation retains. The pasteurized form keeps Amuc_1100, the heat-stable surface protein. The live form keeps P9, the mucin grazing and the cross-feeding, all of which require an organism that is still metabolizing. They are not the same product with different processing; they carry different subsets of the organism's activity. Which one suits you is not a question the current literature answers.

Why do some probiotics work when they are heat-killed and others do not?

Because heat preserves some kinds of molecules and destroys others, and different organisms carry their activity in different molecules. The same 70-degree protocol that preserved the benefit of Akkermansia completely abolished the benefit of Dysosmobacter welbionis (Le Roy 2022). Heat-killing is not a category property.

Should everyone take an Akkermansia supplement?

The 2026 Gut Microbes trial did not find a benefit across its overall study population; the measurable improvements concentrated in participants who started with low levels of the organism. Most people do not know their own baseline, which is a genuine limitation of the current evidence rather than a reason to assume either way. The trials were built to find the average person, and averages are not what you are. This is a question for a clinician who knows your situation.

What is the difference between a probiotic and a postbiotic?

A probiotic delivers live organisms with the intent that they act while alive in your gut. A postbiotic delivers non-living microbial material, or the molecules microbes produce, with the intent that the chemistry does the work. The Akkermansia literature is unusual because the same organism has been studied in both forms.

Why are Faecalibacterium and Roseburia so rarely found in probiotics?

Both are strict anaerobes that do not survive exposure to oxygen, so conventional manufacturing kills them. They only reached the market in 2026, in very small doses relative to hardier organisms, and only with specialized oxygen-free production.

What does AFU mean on a probiotic label?

Active Fluorescent Units, a flow-cytometry count of metabolically active cells. It differs from Colony Forming Units, which counts cells that grow into colonies on a plate. For organisms that grow poorly on plates, AFU may describe the product more accurately, but the two units are not interchangeable when comparing labels.

---

The Bottom Line

So which one should you buy, the live one or the heat-killed one? Try both. The science genuinely does not pick for you, and that is not a dodge. Amuc_1100 is structural and survives the heat, so the pasteurized preparation carries it. P9 is secreted, and the mucin grazing and the cross-feeding both require a metabolizing organism, so the live preparation carries those. Two groups of serious scientists, each right about the part they can see, and as with most microbiome science, probably each holding only part of the story.

In the interest of disclosure: Thaena makes a postbiotic, so weigh this however you want to. It does not change what the papers say.

Twenty-two years after somebody first named this organism, with two companies, more than a hundred million dollars and a granted patent behind it, nobody can tell you which preparation to buy. That is genuinely where the field still is. So take one for a few months and pay attention. Then take the other. If one of them does something you can actually notice, that is real information, and it is the kind no trial was going to hand you, because a trial is built to find the average person. The low-baseline finding is the tell: the people who responded were the ones who were missing it.

The bacterium that started all of this grazes the lining of your gut and refuses to behave the way a probiotic is supposed to behave. That is not a conclusion. It is a field in the middle of finding out, and the only gut it gets settled in is yours.

---

References

1. Derrien M, Vaughan EE, Plugge CM, de Vos WM. Akkermansia muciniphila gen. nov., sp. nov., a human intestinal mucin-degrading bacterium. Int J Syst Evol Microbiol. 2004;54(5):1469–1476. https://doi.org/10.1099/ijs.0.02873-0
2. Everard A, Belzer C, Geurts L, et al. Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity. Proc Natl Acad Sci USA. 2013;110(22):9066–9071. https://doi.org/10.1073/pnas.1219451110 · FREE FULL TEXT
3. Plovier H, Everard A, Druart C, et al. A purified membrane protein from Akkermansia muciniphila or the pasteurized bacterium improves metabolism in obese and diabetic mice. Nat Med. 2017;23(1):107–113. https://doi.org/10.1038/nm.4236
4. Depommier C, Everard A, Druart C, et al. Supplementation with Akkermansia muciniphila in overweight and obese human volunteers: a proof-of-concept exploratory study. Nat Med. 2019;25(7):1096–1103. NCT02637115. https://doi.org/10.1038/s41591-019-0495-2 · FREE FULL TEXT
5. Perraudeau F, McMurdie P, Bullard J, et al. Improvements to postprandial glucose control in subjects with type 2 diabetes: a multicenter, double blind, randomized placebo-controlled trial of a novel probiotic formulation. BMJ Open Diabetes Res Care. 2020;8(1):e001319. NCT03893422. https://doi.org/10.1136/bmjdrc-2020-001319 · FREE FULL TEXT
6. Yoon HS, Cho CH, Yun MS, et al. Akkermansia muciniphila secretes a glucagon-like peptide-1-inducing protein that improves glucose homeostasis. Nat Microbiol. 2021;6(5):563–573. https://doi.org/10.1038/s41564-021-00880-5
7. Le Roy T, Moens de Hase E, Van Hul M, et al. Dysosmobacter welbionis is a newly isolated human commensal bacterium preventing diet-induced obesity and metabolic disorders in mice. Gut. 2022;71(3):534–543. https://doi.org/10.1136/gutjnl-2020-323778 · FREE FULL TEXT
8. Barrodia P, Saw AK, Jeter-Jones SL, et al. Fasting primes small intestinal regeneration after damage via a microbiome-metabolite-chromatin axis. Proc Natl Acad Sci USA. 2026;123:e2529215123. https://doi.org/10.1073/pnas.2529215123 · FREE FULL TEXT
9. Mount S, Canfora EE, Jocken JW, et al. Pasteurized Akkermansia muciniphila Muc^T for weight loss maintenance in people with overweight and obesity: a controlled randomized trial. Nat Med. 2026. NCT05417360. https://doi.org/10.1038/s41591-026-04394-7
10. Suenaert P, Segers A, Rymenans L, Devroye H, Moll JM, Cani PD, de Vos WM. Effect of pasteurized Akkermansia muciniphila Muc^T on insulin sensitivity, body composition, and GLP-1 production in subjects with metabolic syndrome: impact of low baseline gut Akkermansia levels. Gut Microbes. 2026. NCT05114018. https://doi.org/10.1080/19490976.2026.2690689 · FREE FULL TEXT
11. Lee HB, Lee J, Jeong K, Yang J, Jung YH, Moon JS. Randomized, double-blind, placebo-controlled trial of ID-CBT5101, a tyndallized Clostridium butyricum postbiotic, in adults with mild-to-moderate knee osteoarthritis. J Microbiol Biotechnol. 2026;36:e2512024. https://doi.org/10.4014/jmb.2512.12024 · FREE FULL TEXT
12. Salminen S, Collado MC, Endo A, et al. The International Scientific Association of Probiotics and Prebiotics (ISAPP) consensus statement on the definition and scope of postbiotics. Nat Rev Gastroenterol Hepatol. 2021;18(9):649–667. https://doi.org/10.1038/s41575-021-00440-6 · FREE FULL TEXT

This post accompanies the Lit Review Friday episode of Learn Something with Thaena.