Fecal Transplants Beyond C. diff: What Is Approved, What Is Promising, and What Got Harder to Get

There is one thing in this field that nobody seriously disputes. In an adult who has had Clostridioides difficile infection come back after standard antibiotics, restoring the gut's microbial community works. Trials have repeatedly reported prevention of further recurrence at rates antibiotics alone do not reach, and the finding has been durable enough to move through the regulatory system: the FDA approved Rebyota, a rectally administered microbiota suspension, in late 2022, and Vowst, an oral capsule of purified bacterial spores, in 2023. In February 2024 the American Gastroenterological Association published a clinical practice guideline suggesting select use of fecal microbiota-based therapies in immunocompetent adults after completing standard-of-care antibiotics for recurrent C. diff.

That is the whole of the settled territory. Everything else described as “fecal transplant medicine” — for inflammatory bowel disease, irritable bowel syndrome, autism, metabolic disease, cancer — has sat somewhere between a promising signal and an untested inference.

What makes 2026 worth writing about is that the inference finally started getting tested. In the space of nine months the field produced its first randomized, placebo-controlled trial of FMT in a solid tumour, a European regulatory rejection of its most advanced product, and a head-to-head comparison suggesting the stool itself may not be the necessary ingredient. Read together, those three results do not tell a single story. They tell three, and the field's honest position is somewhere in the middle of them.

What holds up

The C. diff case holds up because mechanism and outcome are unusually well matched. Repeated broad-spectrum antibiotics strip the colon of the organisms that ordinarily keep C. diff from establishing itself. Reintroducing a complete microbial community restores what clinicians call colonization resistance. The therapy does one specific thing, the disease is caused by the absence of that thing, and the endpoint — did the infection come back within eight weeks — is objective and easy to count. Few interventions in medicine are this cleanly aligned.

The approved products matter in a way that is easy to miss. Rebyota and Vowst are licensed biologics with standardized manufacturing, donor screening, and lot-level testing. They are not the same thing as conventional FMT — stool from a screened donor, processed and delivered by colonoscopy or enema at a hospital. The AGA guideline covers both; the approvals cover only the manufactured products, and only for preventing recurrence rather than treating an active infection.

The second-strongest indication is one most coverage skips: steroid-refractory acute graft-versus-host disease, the condition in which a stem-cell transplant recipient's donated immune cells attack their gut. The biological logic is close to the C. diff case — these patients have had their microbial ecology destroyed by conditioning chemotherapy and antibiotics — and a September 2026 review in Onco by Al-Omari and colleagues cites a meta-analysis putting pooled clinical remission at 64 percent (95% CI, 51-77%) across prospective studies.

Read that confidence interval before the number. Every study in it was single-arm.

Where the randomized data landed

Until this year, the oncology case for FMT rested almost entirely on small uncontrolled cohorts. The hypothesis was never that microbes treat cancer; it was that gut composition shapes how well checkpoint inhibitors work. Two small 2021 trials in Science reported that stool from patients who had responded to anti-PD-1 therapy could produce responses in some melanoma patients whose disease had resisted it. A 2023 phase I trial in Nature Medicine reported a 65 percent objective response rate in 20 previously untreated advanced melanoma patients. In January 2026 the phase 2 FMT-LUMINate trial reported a 75 percent objective response rate in its 20-patient first-line melanoma cohort, and the phase 1 PERFORM trial reported acceptable safety in 20 patients with metastatic renal cell carcinoma. Encouraging, uncontrolled, and — in first-line melanoma, where combination immunotherapy already produces high response rates — impossible to attribute.

TACITO changed the shape of that argument. Published in Nature Medicine in April 2026 by Serena Porcari and colleagues in Rome, it was the first randomized trial of targeted FMT in metastatic renal cell carcinoma: 45 patients on first-line pembrolizumab and axitinib, randomly assigned double-blind to donor FMT or placebo FMT, three administrations over six months, with donor material drawn from two screened long-term complete responders to immunotherapy.

The primary endpoint missed. Twelve-month progression-free survival was 70 percent with donor FMT versus 41 percent with placebo — close to, but not across, the prespecified threshold (P = 0.053). The secondary endpoint did not miss: median progression-free survival was 24.0 months versus 9.0 (HR 0.50; P = 0.035). The objective response rate was 52 percent versus 32 percent. Median overall survival was 41.0 months versus 28.3, which did not reach significance. No adverse events were attributed to donor FMT.

A trial that misses its primary endpoint and hits a secondary one is the exact shape of a result that becomes a press release in one direction and a footnote in the other.

Both readings are available here, and the honest summary is that TACITO is a positive signal from a trial too small to settle anything — which is what its own authors say. It is also, for the first time, a signal with a placebo arm underneath it.

What does not hold up

The clearest negative finding is the one the field discusses least: what works for C. diff mostly does not transfer. The 2024 AGA guideline recommends against conventional FMT for inflammatory bowel disease and irritable bowel syndrome outside clinical trials. That is not a neutral “more research needed.” It is a guideline panel reading the randomized evidence and concluding the certainty is too low and the heterogeneity too high for routine use.

Ulcerative colitis is the most studied case and the most instructive. Multiple randomized trials have induced remission in some patients, and reviews continue to find a signal, particularly with lower-GI delivery and multi-donor preparations. But the trials are small, the protocols differ on nearly every variable that could matter — donor selection, number of infusions, route, antibiotic pretreatment, diet — and meta-analyses rate the certainty of evidence as low. A 2026 randomized trial testing whether added dietary fiber improved FMT outcomes in mild-to-moderate UC found no benefit from the fiber, and was terminated early when the manufacturer discontinued the FMT product.

Europe supplied the year's sharpest reminder that a promising signal is not an approval. MaaT013, a pooled-donor microbiota product for steroid-refractory acute GvHD, met the primary endpoint of its single-arm phase III ARES trial with a day-28 gastrointestinal overall response rate of 62 percent. On June 25, 2026, the EMA's Committee for Medicinal Products for Human Use adopted a negative opinion on its conditional marketing authorization. The committee's objection was not the response rate. It was that in patients receiving multiple concomitant therapies for GvHD, a single-arm design cannot attribute the observed effect to the drug. MaaT Pharma requested re-examination and made its oral explanation to the committee on September 14, 2026. The committee did not move: it issued a negative trend vote, maintaining that in the absence of a randomized controlled trial the benefit-risk balance of the product cannot be sufficiently characterized. A formal opinion is expected on September 18.

That is the same objection, in a different jurisdiction, that the oncology cohorts invite: not “is the number good” but “how would you know the number came from the treatment.”

The supply problem, and the thing that may replace it

The access picture has changed more than the science since this site last touched the microbiome. OpenBiome, the nonprofit stool bank that supplied conventional FMT material to US hospitals, voluntarily suspended distribution in late September 2024 and stopped shipping frozen preparations at the end of that year. The cause was regulatory rather than any finding about OpenBiome's own product. FDA guidance finalized in November 2022 states plainly that the enforcement discretion allowing a physician to perform FMT for treatment-refractory C. diff without an investigational new drug application “does not apply to FMT that is obtained from a stool bank.” The stated reason is that centralized manufacturing exposes many patients to a single donor and a single set of practices — and the guidance cites its own March 2020 safety alert, issued after six patients developed E. coli infections traced to a stool-bank product, as the illustration. Centralized supply is precisely what the policy declines to wave through. The AGA publicly urged that access be preserved.

The practical result is a field where the approved products cover one indication, and the cheap, flexible, investigator-driven version that made the research possible is much harder to obtain — at the moment the research questions got interesting. The UC fiber trial terminated for lack of product is not an anomaly; it is the shape of the problem.

Which makes the year's quietest result arguably its most consequential. In June 2026, a Mount Sinai team published in Nature Medicine a randomized phase 1b trial comparing a manufactured 15-strain live biotherapeutic, MTC01, against FMT derived from the same donor, in 18 patients with recurrent C. diff. The trial's primary outcome was safety, which it met — ten adverse events across eight patients, evenly split between arms, none treatment-related. Efficacy was a secondary outcome: prevention at eight weeks in seven of nine on the manufactured consortium and eight of nine on donor stool. Eighteen patients, in a trial not designed to measure efficacy, settles nothing about efficacy. What it gestures at is a supply chain: a defined product grown in a fermenter, with known strains and reproducible lots, rather than a screened human being.

The counter-evidence is worth holding alongside it. Defined single-strain approaches have repeatedly produced interesting clinical signals while failing their own microbiological hypotheses — the CBM588 trials in renal cell carcinoma showed efficacy signals in two small randomized phase I studies but missed their prespecified endpoint of increased Bifidobacterium abundance in both. The field does not yet reliably know which components of a stool sample do the work, which is an awkward foundation on which to manufacture a replacement for it.

What a reader should take from this

If you have had recurrent C. difficile infection and completed standard antibiotics, microbiota restoration is well-evidenced and guideline-supported, and two FDA-approved products exist. That recommendation is strong enough to act on.

For inflammatory bowel disease, IBS, fatigue, autoimmune, metabolic or neurological illness, the honest answer has not changed: FMT is a research question, not a treatment. The gastroenterology societies say so explicitly. A clinic selling fecal transplant for these conditions outside a registered trial is ahead of the evidence, and the cost is not only money — it is exposure to an uncontrolled biological product with real, if uncommon, infectious risk. The FDA's own 2020 safety alert, prompted by six E. coli infections traced to screened stool-bank material, is what that risk looks like when the supply chain is professional.

What did change in 2026 is worth stating precisely, because it will be overstated elsewhere. In cancer, the field graduated from uncontrolled cohorts to one randomized trial that missed its primary endpoint and hit a secondary one — genuine progress, and not yet an answer. In GvHD, a regulator looked at a 62 percent response rate and declined to approve it, on the grounds that nobody could show where the response came from. And in C. diff, a manufactured bacterial consortium performed about as well as the donor stool it was derived from, in eighteen people.

The pattern across all three is the same, and it is not the pattern the wellness market reports. The results are getting more rigorous, and as they do, the effects are getting harder to attribute rather than easier. That is what a field looks like when it is being taken seriously. It is also why the correct summary remains narrow: one disease with strong evidence and approved products, a second with a plausible case and no controlled trials, and everything else still open.

This article is general information, not medical advice. See our Editorial Standards.