Sick guts hold different microbes, and proving the microbes did it is harder

Key takeaways

• When researchers reviewed 38 studies that moved human gut microbes into lab rodents, 36 reported that the disease moved too, a success rate the reviewers called implausible.
• C. diff infection is the clearest case of gut bacteria shaping disease: antibiotics weaken the gut’s resident defenders, and restoring a healthy community helps prevent repeat infections.
• Microbes from lean twins kept mice lean in a 2013 study, but a December 2025 analysis of 11 human trials found no clear weight or blood-sugar benefit from stool transplants.
• Gut microbes help make TMAO, a compound tied to heart attacks, yet a genetic study suggests diabetes and kidney disease may raise TMAO rather than the other way around.
• Pasteurized Akkermansia bacteria improved insulin sensitivity in a small pilot whose authors included inventors on patent applications for it, which makes it promising but early.

In May 2019, the Broad Institute announced results from a yearlong study of people with inflammatory bowel disease, or IBD. Researchers had followed 132 people and analyzed 2,965 samples of stool, blood and colon tissue¹.

A trade news site ran the story under the headline “Gut microbiome identified as cause of IBD, study finds”². The article beneath it described associations, not a cause². The study’s authors had found microbes and chemicals that shifted as the disease flared, which is a different thing¹.

That gap, between “different in sick people” and “makes people sick,” is the central problem in microbiome science. In 2020, four scientists put a number on it. They reviewed 38 studies that moved gut microbes from people into lab rodents. In 36 of them, 95%, the disease traits reportedly came along³.

Real biology is rarely that tidy. The reviewers concluded that these rodent experiments were generally overinterpreted³.

If you care about gut health, this matters. Some microbe-disease links are solid, and a few lead to real treatments. Others are early clues dressed up as findings. This is how to tell them apart.

Different microbes in sick people can mean three things

When scientists find that a disease comes with a different mix of gut microbes, the microbes could be playing one of three roles.

• Cause. The microbes help start or worsen the disease.
• Consequence. The disease, or its treatment, changes the gut, and the microbes shift in response.
• Bystander. Something else, such as diet or a medicine, changes both the microbes and the disease risk.

The 2019 IBD study shows why these are hard to separate. People with IBD had less diverse gut microbes. During active disease, they also had fewer of the chemicals microbes normally make, and a group related to Subdoligranulum bacteria dropped during inflammation¹. Crohn’s disease and ulcerative colitis changed the gut in different ways¹.

The authors offered explanations that run in both directions. Some changes could come from the disease itself: poorer nutrient absorption, more water in the bowel and more urgent bowel movements¹. Yet they also suggested the shifts made the gut ecosystem less stable, which could feed more immune flare-ups¹. Both can be true at once.

Scientists have three main ways to test cause

Each tool answers part of the question. None answers all of it.

Tool	How it works	What it does well	Where it falls short
Germ-free mice	Mice raised with no microbes get stool from people with and without a disease	Shows whether microbes alone can change a body	A human community changes once it lives in a mouse, and a mouse is not a small person4
Mendelian randomization	Uses gene variants that people inherit at random, which raise or lower a substance, as a natural experiment	Less tangled with lifestyle than ordinary studies	Works only when genes clearly shift the thing being studied
Randomized trials	People are assigned by chance to get microbes or a placebo	The strongest test of cause and benefit in people	Costly, often small, and short

The mouse tool deserves extra caution. In their 2020 review in the journal Cell, Jens Walter, Anissa Armet, Brett Finlay and Fergus Shanahan said many of those 38 studies stretched rodent results into causal claims about human disease³.

Their critique comes with a to-do list, as summarized by the site Gut Microbiota for Health. A sound study first confirms that the human microbes settled in the mice⁴. It draws on many donors, since people’s gut communities vary so much, and does its math on the number of donors rather than the number of mice⁴. It also avoids blending several donors’ stool into one dose, a shortcut that makes false alarms more likely⁴.

C. diff is the case where cause is clearest

Clostridioides difficile, or C. diff, is a germ that causes severe diarrhea. It shows what a proven microbe-disease link looks like.

The setup is well understood. Antibiotics disrupt the gut’s resident bacteria, and that disruption helps C. diff make its toxins⁵.

During antibiotic treatment, the gut also stops converting bile acids into “secondary” forms, which are toxic to C. diff spores and cells⁵. Meanwhile, the “primary” bile acids that remain encourage C. diff to grow and its spores to sprout⁵.

In mice, the gut’s resistance to C. diff returned about six weeks after antibiotics stopped, and the recovery tracked with rising short-chain fatty acids, compounds that resident bacteria make from fiber⁵.

The proof in people came from putting the community back. Transplanting gut microbes from healthy donors has had striking success against C. diff infections that keep coming back⁶. Our article on microbiome therapies covers the approved products and their trials.

That combination is what strong causal evidence looks like: a clear trigger, a known mechanism and a treatment that works when tested.

Obesity: microbes moved weight in mice, not in people

The obesity story began with striking mouse work. In a 2013 Science study led by Jeffrey Gordon at Washington University in St. Louis, researchers took gut microbes from female twin pairs in which one twin was obese and the other lean⁷.

Germ-free mice given the obese twin’s microbes gained more fat and developed metabolic problems. Mice given the lean twin’s microbes stayed lean⁷.

Then came a twist. When the two groups of mice shared cages, bacteria from the lean mice moved into the others and prevented weight gain⁷. But that happened only when the mice ate a diet low in saturated fat and high in fruits and vegetables. On a diet high in saturated fat and low in produce, the lean-type microbes didn’t take hold⁷.

The study was funded by the NIH and the Crohn’s and Colitis Foundation of America, and also by the food companies Kraft Foods and Mondelez International⁷.

Human results have been far less dramatic. A meta-analysis published in December 2025 pooled 11 randomized trials of stool transplants in people with overweight or obesity and metabolic problems, with 320 participants analyzed⁶.

In its main analysis, the change in body mass index was small and not statistically significant⁶. Insulin resistance, long-term blood sugar and cholesterol showed no clear benefit⁶. The authors rated the certainty of all this evidence as low and said the true effect remains uncertain⁶. One alternate analysis did find a small drop in BMI, so the negative result isn’t the final word either⁶.

The mouse twist may be the most useful lesson. In those mice, diet decided whether helpful microbes could even settle in.

Heart disease and TMAO: a link that may run backward

TMAO, short for trimethylamine N-oxide, starts in the gut. Gut microbes turn choline from foods such as eggs into a compound called TMA, and the liver converts it to TMAO⁸.

A 2013 Cleveland Clinic study in the New England Journal of Medicine showed the gut’s role clearly. After 40 healthy adults ate two hard-boiled eggs and a labeled tracer, TMAO climbed in their blood⁸. After six of them took antibiotics for a week, their TMAO became nearly undetectable; it reappeared when they were retested a month or more after stopping⁸.

The same team then followed 4,007 adults who had elective heart catheter tests for three years⁸. People with the highest TMAO had about 2.5 times the risk of a heart attack, stroke or death compared with those with the lowest⁸. After the researchers adjusted for standard risk factors, kidney function, medicines and the extent of artery disease, that shrank to about 1.4 times⁸.

A risk that shrinks when you adjust for other factors can be a sign of confounding, where a hidden factor explains part of a link.

A 2019 Mendelian randomization study in the journal Diabetes tested the question another way⁹. People whose genes predicted higher TMAO did not have more type 2 diabetes, heart disease, stroke or kidney disease⁹.

But type 2 diabetes and chronic kidney disease did appear to raise TMAO levels⁹. The authors said the heart link may reflect confounding or reverse causation⁹.

So microbes clearly help make TMAO. Whether TMAO itself harms your heart is still open. Our paleo diet article covers TMAO studies in people eating that way.

Colon cancer: a mouth microbe that keeps turning up in tumors

Fusobacterium nucleatum usually lives in the mouth. Scientists first noticed it in colorectal tumors more than a decade ago, and studies around the world have found the same thing since¹⁰.

In 2024, a Fred Hutchinson Cancer Center team including Susan Bullman and Christopher Johnston went further, in the journal Nature¹⁰¹¹. They compared 55 strains from colorectal cancers with 80 from people without cancer¹¹.

One subtype, called Fna C2, carried genes linked to faster growth, acid tolerance and settling in the gut¹¹. It was elevated in about half of tumors from 116 patients¹¹. Stool from 627 people with colorectal cancer held more of it than stool from 619 healthy people¹¹.

In mice, Fna C2 led to more intestinal tumors than a related subtype¹⁰¹¹. A post by Monica Bertagnolli on the NIH Director’s Blog said in its headline that the microbe “may drive” the progression of colorectal cancer¹¹.

That is careful wording, and it fits the evidence. The human data show a strong association, and the mouse data suggest a role. Researchers hope the subtype could someday help with screening or treatment¹⁰¹¹. None of that replaces standard colon cancer screening.

Akkermansia: from lab bench to supplement aisle

Akkermansia muciniphila is a gut bacterium that feeds on mucus⁵. After encouraging results in mice, a team at UCLouvain in Belgium tested it in people and published the results in Nature Medicine in 2019¹⁴¹⁵.

Forty overweight or obese adults with insulin resistance enrolled, and 32 finished, about a dozen or fewer per group¹⁴. They took live bacteria, pasteurized (heat-killed) bacteria or a placebo every day for three months¹⁴.

The pasteurized group showed the clearest changes. Compared with placebo, its insulin sensitivity improved about 29%, insulin levels fell about 34% and total cholesterol fell about 9%¹⁴. Weight dropped about 2.3 kilograms, roughly 5 pounds, but that change was not statistically significant¹⁴. The authors called the study an exploratory pilot that was not powered to reach firm conclusions about metabolism¹⁴.

The paper also disclosed that five authors were inventors on patent applications covering the use of Akkermansia in obesity and related disorders, and two had co-founded A-Mansia Biotech¹⁴. That company planned to sell the bacterium as a food supplement, with sales expected in 2021, according to a trade publication¹⁵. The publication’s headline suggested the supplement could cut heart disease risk for half of the Western population, a claim the three-month trial never tested¹⁵.

Context matters, too. In mice fed a diet short on fiber that microbes can use, Akkermansia was one of the bacteria that thinned the gut’s outer mucus layer⁵. A microbe isn’t simply good or bad. What it does depends on what it has to eat.

If you see Akkermansia on a label, check whether the product uses the form and amount that was tested. Ask your clinician before trying it if you take diabetes medicine.

How strong is the link? A ladder for microbiome claims

Use this ladder to place a headline. The higher the rung, the more you can rely on it.

Rung	Claim	Evidence level	Why
1. Cause and cure shown	Disrupted gut bacteria set up C. diff, and restoring them helps prevent repeats	Strong evidence	Clear trigger, known mechanism, effective treatment56
2. Mechanism shown, disease role open	Gut microbes are needed to turn choline from food into TMAO	Strong evidence	Antibiotics nearly erased TMAO in people8
3. Strong link, cause unclear	Fusobacterium is enriched in colorectal tumors	Strong evidence of a link	Found in tumors worldwide; cause in people unproven1011
3. Strong link, cause unclear	Gut microbes differ in IBD and shift with flares	Strong evidence of a link	Direction could run both ways1
4. Early clue	A Fusobacterium subtype drives tumor growth	Promising but early	Shown in mice11
4. Early clue	Pasteurized Akkermansia improves insulin sensitivity	Promising but early	One small pilot; several authors are inventors on patent applications14
5. Unproven in people	Stool transplants cause weight loss	Not enough evidence	No clear effect across 11 trials6
5. Unproven in people	TMAO causes heart attacks	Not enough evidence	Genetic test points to reverse causation9
6. Fails the test	If human stool makes mice sick, microbes cause the human disease	Evidence says no, or unsafe	95% transfer rate judged implausible34

Three questions place almost any claim on the ladder:

1. Was it people, mice or a lab dish? A mouse result belongs on rung 4 at best.
2. Did the study compare, or did it test? Comparing sick and healthy people shows a link. Only a trial that changes the microbes can show they matter.
3. Who paid, and who profits? Look for patents, company founders and product sales, as in the Akkermansia pilot¹⁴.

Questions readers ask

The bottom line

Gut bacteria really do differ in many diseases, and in one case, C. diff, both the cause and an effective treatment are clear. For obesity, heart disease, IBD and colon cancer, scientists have strong links and intriguing mouse results, but they haven’t yet shown that changing your microbes changes your risk.

Read microbiome claims with a ladder in mind. Ask whether the study tested people, whether it changed the microbes or only counted them, and who stands to profit. Meanwhile, the steps with the best support remain the ones you already value: good food, careful use of antibiotics and a clinician for symptoms that don’t add up.

Related reading

• What your gut bacteria do, given that they don’t outnumber you 10 to 1
• Stool became FDA-approved medicine for one infection, and other uses are still experiments
• Chocolate tracks with Nobel prizes, and other lessons in telling a link from a cause
• Only 1 in 20 treatments tested in animals wins approval for people
• The preventive care that costs you nothing, and the exceptions

Sources

1. Broad Institute. “Human Microbiome Project reveals how microbiome is disrupted during inflammatory bowel disease.” News release, May 29, 2019. broadinstitute.org
2. Drug Target Review. “Gut microbiome identified as cause of IBD, study finds.” Drug Target Review, undated web article on the 2019 study.
3. Walter J, Armet AM, Finlay BB, Shanahan F. “Establishing or Exaggerating Causality for the Gut Microbiome: Lessons from Human Microbiota-Associated Rodents.” Cell, 2020. research.ucc.ie
4. van den Wijngaard R. “Challenges and next steps in using human fecal microbiota transfer to germ-free mice models as a means of translating microbiome science to therapeutic medicine.” Gut Microbiota for Health (a site funded by Danone Research & Innovation and Biocodex), July 2, 2020. gutmicrobiotaforhealth.com
5. Ducarmon QR, Zwittink RD, Hornung BVH, van Schaik W, Young VB, Kuijper EJ. “Gut Microbiota and Colonization Resistance against Bacterial Enteric Infection.” Microbiology and Molecular Biology Reviews, 2019. pubmed.ncbi.nlm.nih.gov
6. Zhang Y, Cao J, Wang Y, Fan X, Deng R, Mi J. “Effects of fecal microbiota transplantation on glycemic and lipid profiles in overweight or obese patients with metabolic disorders: a systematic review and meta-analysis.” Frontiers in Endocrinology, December 2025. public-pages-files-2025.frontiersin.org
7. Washington University in St. Louis. “Altering mix of gut microbes prevents obesity, but diet remains key factor.” The Source, September 5, 2013. source.washu.edu
8. Tang WHW, Wang Z, Levison BS, et al. “Intestinal Microbial Metabolism of Phosphatidylcholine and Cardiovascular Risk.” New England Journal of Medicine, 2013. masspec.scripps.edu
9. Jia J, Dou P, Gao M, et al. “Assessment of Causal Direction Between Gut Microbiota-Dependent Metabolites and Cardiometabolic Health: A Bidirectional Mendelian Randomization Analysis.” Diabetes, 2019. pubmed.ncbi.nlm.nih.gov
10. Fred Hutchinson Cancer Center. Spotlight on Zepeda Rivera M, et al., “A distinct Fusobacterium nucleatum clade dominates the colorectal cancer niche,” Nature. Fred Hutch News, May 22, 2024. fredhutch.org
11. Bertagnolli MM. “Microbe Normally Found in the Mouth May Drive Progression of Colorectal Cancer.” NIH Director’s Blog, National Institutes of Health, post date not shown (on a 2024 study).
12. American College of Gastroenterology. “Irritable Bowel Syndrome.” Patient information, accessed October 2026. gi.org
13. U.S. Preventive Services Task Force. “Colorectal Cancer: Screening.” Final recommendation statement, May 18, 2021. uspreventiveservicestaskforce.org
14. Depommier C, Everard A, Druart C, et al. “Supplementation with Akkermansia muciniphila in overweight and obese human volunteers: a proof-of-concept exploratory study.” Nature Medicine, 2019 (full text with competing-interests statement, UCLouvain repository). research.dial.uclouvain.be
15. Haigh L. “Pasteurized bacteria supplement could ‘reduce the risk of heart disease in half of the West’s population.'” Nutrition Insight, July 4, 2019. nutritioninsight.com