Akkermansia Muciniphila

The verdict

Is Akkermansia Muciniphila worth taking?

It is not a probiotic, and that is the most important thing to understand about it. The form that works in humans is pasteurized — heat-killed. When researchers compared live against pasteurized cells in mice, they reported the result as unexpected: pasteurization enhanced the effect. The active component turned out to be a protein on the bacterium's outer membrane that survives the heat. You are not buying a live culture. You are buying a dead one, deliberately, and the regulatory specification requires it.

So the question people ask about it — an oxygen-sensitive gut anaerobe, how much could possibly survive manufacture, a shelf and a stomach? — has a clean answer. Nothing is meant to survive. The UK regulator's specification for the authorized product sets the viable count below 10 colony-forming units per gram. That is a requirement that it be sterile for practical purposes.

The evidence is better than most things in this guide and weaker than the marketing. The famous result — a 28.6 percent improvement in insulin sensitivity — comes from a 32-person study its own authors labeled a proof-of-concept exploratory pilot. The weight and fat-mass changes from that same study, the numbers that sell the product, came in at p=0.091 and p=0.092: not significant. Then the largest trial, 142 people over four months, missed its primary endpoint outright, and everything positive reported from it is explicitly exploratory subgroup analysis.

What does hold up: a 90-person trial in 2026 met its pre-specified primary outcome for weight-loss maintenance, with regain of 1.2 kg against 3.2 kg on placebo (p=0.012). That is a real result on a real endpoint, and it is the strongest single piece of evidence here.

And a pattern across three separate trials that nobody puts on a label: it appears to work mainly in people who have little Akkermansia to begin with. In participants who already had plenty, colonization was poor and nothing improved. There is no consumer test that tells you which group you are in.

The bottom line. Moderate is the right grade, which is not something I say often on these pages. There is one clean positive trial, one failed primary endpoint, and a credible mechanism with a named molecule behind it. If you are going to try it, the honest framing is weight-regain prevention after losing weight rather than weight loss, the dose that was tested is 10 to 30 billion cells a day, and the polyphenol-rich foods that raise your own Akkermansia are cheaper and better evidenced as a first step.

What it does

What Akkermansia Muciniphila does in the body

Akkermansia muciniphila lives in the mucus layer lining your gut and eats mucin, the glycoprotein that mucus is made of. That sounds alarming and is not: by grazing on the mucus layer it stimulates the host to produce more of it, and a thicker mucus layer is a better barrier. It makes up somewhere around one to four percent of the gut bacteria in a healthy adult and it is the only representative of its phylum there.

The mechanism has a named molecule, which is rare in this field. A 2016 paper in Nature Medicine isolated Amuc_1100, a protein from the bacterium's outer membrane. It signals through Toll-like receptor 2, it is stable at the temperatures used for pasteurization, and on its own it partly reproduced the metabolic benefits of the whole organism. That is a specific, testable mechanism rather than a general appeal to gut health.

It also explains the inversion. If the active component is a heat-stable surface protein rather than a living metabolism, then killing the cells costs you nothing and gains you a great deal: a product that is stable, that does not need to be kept alive through manufacture and shipping, and that cannot colonize anything. The same 2016 paper found pasteurized cells outperformed live ones in mice at reducing fat mass, insulin resistance and dyslipidemia. The authors called it unexpected. It was also the finding that made a commercial product possible, because the live organism's oxygen sensitivity and its animal-derived growth medium had been the obstacles.

Which means “next-generation probiotic” is the wrong label and “postbiotic” is the right one. A probiotic is a live organism intended to establish itself. This is a dose of bacterial structure intended to be recognized by your immune system on the way past. Those are different products with different questions attached, and almost everything written about this compound asks the probiotic questions.

One honest complication: the live-versus-pasteurized question is not fully closed. A 2025 trial in 130 overweight adults compared live cells, a postbiotic preparation and placebo over eight weeks, and reported that the live form outperformed the postbiotic on several metabolic markers — noting in its own abstract that this contradicts the mainstream view. One trial in a minor journal does not overturn the picture, but it is on the record.

The evidence

What the research actually shows

Moderate, and here that grade is earned rather than generous. Four randomized trials in humans, one clean positive, one clean negative on its own primary endpoint, and a consistent pattern underneath all of them. Taken in order:

Depommier 2019, Nature Medicine. n=40 enrolled, 32 completed, 10¹⁰ cells daily for three months, overweight and obese insulin-resistant adults. This is the study everything else rests on, and its authors called it a proof-of-concept exploratory study. Pasteurized cells improved insulin sensitivity by 28.62 ± 7.02 percent (p=0.002), reduced insulinemia by 34.08 percent (p=0.006) and total cholesterol by 8.68 percent (p=0.02). The body-composition numbers did not reach significance: body weight −2.27 kg at p=0.091, fat mass −1.37 kg at p=0.092, hip circumference −2.63 cm at p=0.091. Live cells moved in the right direction but did not reach significance on the metabolic endpoints either. And the paper states that the overall gut microbiome structure was unaffected — it did not change the microbiome it was given to change.

Suenaert 2026, Gut Microbes. n=142, multicenter, four months, 30 billion pasteurized cells daily, adults with metabolic syndrome. This is the largest and best-powered trial, and its primary endpoint failed: whole-body insulin sensitivity by the Matsuda index did not differ from placebo in the intention-to-treat population. The paper is clear that what follows is “subsequent exploratory analyses,” which found improvements in hepatic insulin sensitivity in prediabetic participants (12 percent, p=0.05) and in those aged 63 and over (p=0.05), and a greater GLP-1 excursion after a glucose load (p<0.01).

Mount 2026, Nature Medicine. n=90, weight-loss maintenance. Participants did an eight-week low-energy diet to lose at least 8 percent of body weight, then 24 weeks of normal eating with pasteurized cells or placebo. The pre-specified primary outcome was met: weight regain of 1.2 ± 0.7 kg against 3.2 ± 0.4 kg on placebo (p=0.012), and greater net weight loss from baseline (p=0.009). The authors list their own limitations plainly — a relatively short intervention, and no comparison group given modified strains lacking the active components. This is the best evidence for the compound and it is for keeping weight off, not taking it off.

Zhang 2025, Cell Metabolism. n=58, overweight and obese type 2 diabetes, 12 weeks, live cells. The trial as a whole was null: both groups lost weight and lowered HbA1c with no significant difference between them. The interesting part is why. Participants who started with low Akkermansia showed high colonization and significant reductions in weight, fat mass and HbA1c. Participants who started with plenty showed poor colonization and no clinical improvement. The authors then reproduced the split in germ-free mice given human stool from each group.

That baseline pattern is the one consistent thread. Three separate trials report it: the Cell Metabolism trial above, the 142-person trial where the low-baseline subgroup was the one that improved, and the weight-maintenance trial, which found initial Akkermansia abundance was associated with the cardiometabolic response. It is the most useful finding in the whole literature and the least usable, because there is no test a consumer can order to find out where they stand.

And the association the whole field was built on does not look the way it is usually told. Lower Akkermansia is widely reported in obesity and type 2 diabetes. But a 2017 study in Diabetes Care of 112 adults, matched for sex, age and BMI category, found Akkermansia was 3.4 times more abundant in people with type 2 diabetes who were taking metformin than in those who were not (p=0.003) — and found no significant difference between people with diabetes and people without (p>0.1). In that cohort the bacterium tracked the drug, not the disease. The authors treat metformin as a modifier of the relationship rather than a confounder of it, and that is the careful reading. It is still a result worth knowing before accepting that low Akkermansia causes metabolic disease rather than accompanying its treatment.

Reading the label

Why the number on the bottle is not a CFU count

Every other probiotic on the shelf is sold in CFU — colony-forming units, a count of bacteria capable of growing into a colony, which is to say a count of living cells. Akkermansia products are labeled in TFU, total fluorescent units, which counts cells by staining them whether they are alive or dead.

That is not a marketing trick. It is the only measurement that makes sense for a product specified to contain no viable organisms. But it does mean a shopper comparing “20 billion” on one bottle with “20 billion” on another is comparing two different measurements of two different things, and there is no conversion between them.

The practical consequence is dose. The trials used 10 to 30 billion cells a day. Retail products range from one billion a day up to powders quoting 100 billion per gram, and a product supplying a billion cells daily is a tenth of the lowest dose ever tested. If you are going to buy this, the number to match is the trial number, and it needs to be a daily figure rather than a per-gram one.

One more thing from the regulatory file, stated neutrally because I cannot resolve it. The UK assessment derived a safe level of 4.8 × 10⁸ cells per kilogram of body weight per day from a 90-day rat study with an uncertainty factor of 200, which it converts to about 34 billion cells a day for a 70 kg adult. It then authorizes up to 40 billion cells a day. The authorized maximum is above the derived figure and the document does not explain the gap. It may be an intentional rounding against the heavier end of the adult range; it may be an oversight. Either way the human trials used 10 to 30 billion with no adverse effects reported, which sits below both numbers.

Food first

Getting Akkermansia Muciniphila from food

This is the one entry in the guide where the food answer is not a consolation prize. You cannot eat Akkermansia — it is not in any food, it has never been on an approved list of strains that can be added to food, and the live organism is difficult enough to grow that its oxygen sensitivity was one of the obstacles that pushed the whole field toward the pasteurized product. What you can do is feed the Akkermansia you already have, and the human data on that is more specific than you would expect.

What raised it in human trials. A systematic review of 29 dietary intervention trials in 1,444 people found Akkermansia abundance increased with caloric restriction, and with pomegranate extract, resveratrol, inulin, polydextrose, yeast fermentate and sodium butyrate. A separate 2024 meta-analysis of six microbiome studies, 821 stool samples from 451 participants, found galacto-oligosaccharides increased it in healthy people.

What lowered it — and this one matters. The same review found a low-FODMAP diet decreased Akkermansia abundance. That is worth sitting with, because low-FODMAP is the standard dietary approach for irritable bowel syndrome, and the foods it removes — onion, garlic, leek, asparagus, chicory, legumes — are precisely the inulin and galacto-oligosaccharide sources that feed this bacterium. There is a real tension there and nobody mentions it. Separately, lignans appear to push abundance the opposite way from other polyphenols, which puts flaxseed and sesame in an unexpected column.

Translating the polyphenol classes into a shopping list. Anthocyanins, flavan-3-ols, flavonols, flavanones, stilbenes and phenolic acids all raised Akkermansia in vivo. In a kitchen that means dark berries, cranberry, Concord and other dark grapes, red cabbage and black rice for the anthocyanins; cocoa, green tea and apples with the skin on for the flavan-3-ols; onions, kale and capers for the flavonols; citrus peel and pith for the flavanones; grape skin for the stilbenes; and coffee for the phenolic acids. The strongest single animal signal was Concord grape polyphenols, which dramatically increased abundance in mice, and cranberry polyphenols, which raised it roughly fivefold.

The counterintuitive part, which is the useful part. A 2024 study comparing dietary antioxidants found that only grape polyphenols produced an Akkermansia bloom, and concluded that it was their poor bioavailability that did it — the fraction that fails to be absorbed in the small intestine is the fraction that reaches the colon where this bacterium lives. So the polyphenols that work here are the ones that never make it into your bloodstream. That inverts the usual advice about choosing the most bioavailable form of anything, and it is why whole fruit, skins and peels beat a purified extract designed for absorption.

And it is why peels and skins are not a garnish. Keep the skin on the apple. Press whole grapes rather than straining. Use the citrus pith, not just the zest and juice. Cook red cabbage with a little acid to hold its color, because the color is the anthocyanin. None of this is exotic and all of it is better evidenced, cheaper and more pleasant than a capsule.

What the guide adds

The part that is not on this page

It appears to work mainly in people who are short of it, and you cannot find out whether you are. Three trials now report that the response depends on baseline Akkermansia abundance, with poor colonization and no clinical benefit in people who already have plenty. There is no consumer stool test validated to answer this question, and the trials measured it with metagenomic sequencing rather than anything you can order.

The largest trial missed its primary endpoint. Everything positive reported from the 142-person four-month trial is explicitly labeled exploratory subgroup analysis by its own authors. That does not make those findings wrong, but it is a different category of evidence from a met primary endpoint, and it is usually quoted as though it were the same thing.

Nothing persists after you stop. This follows from the mechanism rather than from a study: a product specified to contain essentially no viable cells cannot colonize anything. Whatever it does, it does while it is passing through. There is no trial of what happens after discontinuation.

No trial has been run in a healthy, normal-weight person. Every participant in every trial was overweight or obese, insulin-resistant, prediabetic, diabetic, or had metabolic syndrome. If you are metabolically healthy and taking this for prevention or longevity, there is no study of you.

Longest exposure is about seven months. The weight-maintenance trial ran eight weeks of dieting plus 24 weeks of maintenance. Everything else ran three to four months. There is no long-term human data.

Safety looks good within those limits. No serious treatment-related adverse events were reported in any trial. The UK regulator assessed pasteurized Akkermansia as safe as a novel food at up to 40 billion cells a day for people aged 12 and over, based on a 90-day rat study and the human trial data. It was authorized in the EU in 2022. I could not confirm from the assessment body text whether pregnancy and breastfeeding were specifically addressed, so anyone pregnant or breastfeeding should treat it as untested rather than cleared.

The cost question is the real one. This is among the more expensive supplements available, it is sold on a 32-person exploratory pilot plus one positive maintenance trial, and the dietary route that raises your own abundance has a systematic review behind it. If money is finite, the order of operations is not in doubt.

About the author

Who graded this

Daniel Edlund is a Le Cordon Bleu-trained private chef in Los Angeles, a certified personal trainer and strength coach, and an internationally licensed massage therapist. Thirty years of strength training and fifteen years cooking in private estate kitchens. He is the author of The Complete Supplement Guide, which grades 147 entries against the clinical literature, and of The EASY Way to Organic Cooking, first published in 2013.

More about Daniel →

This page is information, not medical advice. Supplements interact with prescription medication and with each other, and the right answer depends on your own bloodwork and history. Talk to your doctor before starting anything, particularly if you are pregnant, nursing, managing a health condition, or already taking medication.

Also in Gut Health & Digestion

Related entries

• Apple Cider Vinegar — Emerging evidence
• Colostrum — Moderate evidence
• Prebiotics — Strong evidence
• DGL — Moderate evidence
• Berberine — Strong evidence
• Zinc Carnosine — Moderate evidence

All 147 entries, graded →