Faecal transplant: what it can do, where it stands and why it is not something to try at home

Susana Fuentes and her team characterised the stool flora of nine people from the Amsterdam trial before the transfer, afterwards and across ten weeks of follow up using a phylogenetic microarray, and compared it with that of their donors.

Before treatment, Proteobacteria and Bacilli dominated at low diversity. Afterwards the picture shifted towards Bacteroidetes and Clostridium groups including butyrate producing bacteria, and resembled that of the donors. This shift persisted across the entire follow up.

For you this means: no single germ is being fought. An ecological niche that has fallen empty is being occupied again. That is a different idea from the one most people associate with gut treatment.

A team around Menno Bénard evaluated every step of their own donor screening at the Amsterdam UMC from January 2018 to August 2021 and calculated the costs.

393 interested people entered pre screening. 202 dropped out at that stage already. Of the 191 who completed the detailed questionnaire, 43 were excluded. In parasite screening, 91 of 148 dropped out, mostly because of Dientamoeba fragilis and larger amounts of Blastocystis. In the further stool tests, 18 of 57 were excluded, mainly because of Helicobacter pylori and ESBL producers. In the end 38 of 393 were approved, that is 10 per cent, at a cost of 64,112 euros for those 38.

For you this means: in a university hospital, nine out of ten volunteers fail. The idea that your own partner or a sibling is somehow healthy enough does not hold up statistically.

Nicole Dubois and colleagues evaluated the reasons for rejection among 7,968 prospective donors at a stool bank, stage by stage.

In the end 134 people qualified, that is 1.7 per cent. Published rejection rates of donor programmes lie between 90 and 96 per cent. The most common reasons at the initial check were a BMI of 30 or above (37.0 per cent), logistics (20.5 per cent) and travel history (15.5 per cent). Despite passing the initial check, 569 of 781 were still turned down at the in person appointment, primarily because of conditions that are linked with the microbiome. Only a small proportion failed at the laboratory stage.

For you this means: most rejections do not come from an abnormal laboratory value, but from conversation and examination. That is exactly the part a stool test bought online cannot replace.

Zhi-Dong Jiang and colleagues gave 72 people with at least three episodes of a Clostridioides difficile infection, double blind and via colonoscopy, either fresh, frozen or freeze dried material from a small donor pool.

Response over two months was 100 per cent for fresh (25 of 25), 83 per cent for frozen (20 of 24) and 78 per cent for lyophilised (16 of 23). Fresh versus freeze dried differed statistically, P equals 0.022. Stool diversity was back at day 7 with fresh and frozen material, with freeze dried material only at day 30.

For you this means: if even a standardised freeze drying step in a laboratory measurably costs response, then the idea that this could be reproduced in a kitchen is not plausible.

Dina Kao and her team randomised 116 adults with recurrent Clostridioides difficile infection to capsules or colonoscopy.

In the per protocol analysis both groups reached 96.2 per cent, the difference was zero per cent, non inferiority was met. Mild adverse events were less frequent in the capsule group (5.4 versus 12.5 per cent). 66 per cent of the capsule group rated the experience as not at all unpleasant, compared with 44 per cent after colonoscopy.

For you this means: the capsule is the more comfortable route, medically equivalent. It is not the more harmless substance, though. It is the same material with the same requirements for donor testing, only packaged differently.

Els van Nood and her team in Amsterdam randomised people with recurrent Clostridioides difficile infection to three arms: a short vancomycin course plus bowel irrigation plus infusion of donor material through a nasoduodenal tube, or 14 days of vancomycin alone, or vancomycin plus bowel irrigation.

After the interim analysis the trial was stopped. 13 of 16 people in the infusion arm (81 per cent) had no further diarrhoea after a single dose. Of the remaining three, all received a second dose with material from a different donor, two of them successfully, making 15 of 16 (94 per cent). Vancomycin alone: 4 of 13 (31 per cent). Vancomycin plus irrigation: 3 of 13 (23 per cent). P less than 0.001 in each case. Adverse events did not differ, apart from mild diarrhoea and cramps on the day of the infusion.

For you this means: stopping early is not a flaw here, it is the opposite. An independent committee halts a trial when it becomes unethical to keep withholding the superior treatment from the control group. That does not happen often.

Colleen Kelly and colleagues gave 46 people with three or more recurrences, after a complete vancomycin course, either donor material or their own stool as a control, via colonoscopy.

In the intention to treat analysis, 20 of 22 with donor material stayed free of symptoms (90.9 per cent) compared with 15 of 24 with their own material (62.5 per cent), P equals 0.042. All nine people who relapsed after the autologous dose were free of symptoms after a subsequent donor transfer. Notably, the autologous dose performed very differently at the two centres, 9 of 10 versus 6 of 14.

For you this means two things. The effect holds up even against a credible sham treatment. And the difference between the centres is an early hint at how strongly the details of the procedure shape the result.

Christian Lodberg Hvas and his team in Aarhus compared the transfer after a short vancomycin course with fidaxomicin and with vancomycin, that is with the best and with the usual antibiotic option.

For the combined endpoint of clinical settling and a negative toxin PCR after eight weeks, 17 of 24 (71 per cent) were in the transfer arm, 8 of 24 (33 per cent) on fidaxomicin and 3 of 16 (19 per cent) on vancomycin. For purely clinical response the figures were 92 versus 42 versus 19 per cent. One serious event was classified as possibly related to the procedure.

For you this means: the comparison is fair, because fidaxomicin is not the weakest but the strongest drug comparator. The German guideline highlights precisely this trial.

Simon Mark Dahl Baunwall and colleagues pooled 45 studies on recurrent Clostridioides difficile infection, with GRADE assessment and a pre registered protocol.

After eight weeks the clinical effect was 91 per cent for repeated transfer (95 per cent confidence interval 89 to 94, 24 studies, 1,855 people) and 84 per cent for a single transfer (80 to 88, 43 studies, 2,937 people). Against vancomycin, the number needed to treat for repeated use was 1.5 (1.3 to 1.9). The evidence for repeated use was rated as high quality.

For you this means: a number needed to treat of 1.5 means that, arithmetically, one and a half people have to be treated for one additional person to gain a benefit. For many preventive medicines this number runs into the tens or hundreds. That is the order of magnitude we are talking about here.

Yvette van Beurden and colleagues followed up all 39 people who were treated according to this protocol in Amsterdam between 2010 and 2016, with structured telephone interviews over 3 to 68 months.

The primary response rate was 82 per cent. In 5 people, procedure related complications occurred, all of them regurgitation or vomiting. One person died of pneumonia one week after the transfer, and a causal link with the procedure could not be excluded. No long term side effects were reported.

For you this means: the route through the upper digestive tract has dangers of its own, above all aspiration. This one sentence from the publication is missing from every advice page I have found.

Paul Feuerstadt and a large team tested, in the ECOSPOR III trial, a product made of purified Firmicutes spores from donor material in 182 people with at least three episodes, after completed standard antibiotic therapy, four capsules a day for three days against placebo.

A recurrence within eight weeks occurred in 12 per cent on the product compared with 40 per cent on placebo. Relative risk 0.32 (0.18 to 0.58), P less than 0.001. The delivered species were detectable from week 1 onwards, together with bile acid profiles that inhibit the germination of spores. Side effects were mostly mild to moderate and similarly frequent in both groups.

For you this means: here a biological observation has become a defined medicinal product. This trial is the basis of the 2023 US licence for an oral microbiome based product.

Sahil Khanna and colleagues tested, in PUNCH CD3, a broadly composed microbial consortium from human stool as a single enema against placebo, in adults with at least one recurrence.

The modelled success rate was 70.6 versus 57.5 per cent, the estimated treatment effect 13.1 percentage points, the posterior probability of superiority 0.991. More than 90 per cent of those with success at eight weeks held that result to six months, and that in both groups. Treatment related adverse events were more frequent in the active arm, mostly mild gastrointestinal complaints.

For you this means: the sentence from above applies here too. With a less desperate starting point the gap shrinks. This trial is the basis of the 2022 US licence for a microbiome based enema product.

Paul Moayyedi and colleagues gave 75 people with active ulcerative colitis a weekly enema with donor material or with water for six weeks.

Remission was reached by 9 of 38 (24 per cent) versus 2 of 37 (5 per cent), risk difference 17 per cent. Adverse events did not differ. Two details are at least as important as the main result: the trial had been stopped early by the data monitoring committee for futility, and 7 of the 9 remissions came from the material of one single donor.

For you this means: anyone quoting only the 24 versus 5 per cent leaves out two thirds of the story. This trial is at once a positive result, an early stop and the origin of the super donor question.

Sudarshan Paramsothy and colleagues in Australia combined a colonoscopic dose with enemas on five days a week over eight weeks. Each enema came from pooled material of three to seven unrelated donors.

After eight weeks, 11 of 41 (27 per cent) reached the endpoint versus 3 of 40 (8 per cent), risk ratio 3.6, P equals 0.021. Adverse events occurred in about four out of five people in both groups, mostly self limiting gastrointestinal complaints. Microbial diversity rose and stayed elevated.

For you this means: that was 40 enemas in eight weeks. Anyone wondering whether a single procedure could influence a chronic bowel inflammation will find the answer of the research here. And even at this intensity, 27 per cent reached remission.

Aamer Imdad and colleagues pooled all randomised trials on this topic in ulcerative colitis and Crohn's disease for the Cochrane Collaboration, with searches up to December 2022.

For remission induction in ulcerative colitis the risk ratio was 1.79 (1.13 to 2.84) at low certainty of the evidence. Endoscopic remission was 1.45 (0.64 to 3.29), so the confidence interval includes the null effect. Serious adverse events: 1.77 (0.88 to 3.55) at very low certainty. For maintenance of remission: 2.97 (0.26 to 34.42), also very low certainty.

For you this means: possibly an effect in active ulcerative colitis. The long term benefit is unclear. And the safety question is rated at very low certainty, which amounts to saying: we cannot derive anything robust from it.

Magdy El-Salhy and colleagues in Norway randomised 165 people with irritable bowel syndrome to placebo (their own stool), 30 grams or 60 grams of donor material. All of the material came from one single, very carefully characterised donor.

After three months, 23.6 per cent in the placebo group responded, 76.9 per cent in the 30 gram group and 89.1 per cent in the 60 gram group, P less than 0.0001 in each case. Fatigue and quality of life improved as well. The authors explicitly conclude that a well defined donor with a favourable microbial signature was the precondition for this result.

For you this means: 89 per cent from a trial with one single donor is not proof of a procedure. It is the strongest conceivable hint that it may have come down to that one donor.

Gianluca Ianiro and colleagues pooled five randomised trials with 267 people and combined the relative risk of still being symptomatic after treatment.

Across all trials it was 0.98 (0.58 to 1.66), that is no effect. The pattern underneath is revealing: placebo capsules performed better than capsules with donor material in two pooled trials (RR 1.96, 1.19 to 3.20). Donor material via colonoscopy, by contrast, performed better than autologous material (RR 0.63, 0.43 to 0.93). Via a nasojejunal tube there was a trend (RR 0.69, 0.46 to 1.02).

For you this means: there is not one procedure with one result. Route of delivery and processing change the sign. And of all things, the form you would be most likely to obtain yourself, namely capsules, performed worse than placebo.

Anne Vrieze and colleagues gave men with metabolic syndrome, in randomised fashion, a small bowel infusion with microbiota from lean donors or with their own, and measured insulin sensitivity with the clamp technique.

Six weeks after the allogenic infusion, the median rate of glucose disappearance rose from 26.2 to 45.3 micromol per kilogram per minute, together with an increase in butyrate producing bacteria.

For you this means: what was measured was a laboratory parameter over six weeks in a small group of men. Weight loss was not an endpoint and was not shown.

Jasmohan Bajaj and colleagues randomised men with liver cirrhosis and recurrent hepatic encephalopathy to standard therapy alone or standard therapy plus a single enema with material from a deliberately selected donor whose microbiota contained precisely the groups missing in this condition.

In the standard group, 8 of 10 people had a total of 11 serious events, in the transfer group 2 of 10, P equals 0.02. Five people in the standard group and none in the transfer group developed a further episode, P equals 0.03. Cognition improved in the transfer group, and not in the standard group.

For you this means: what is interesting here is less the result than the method. The donor was not selected as healthy in some general way, but by microbial profile, matching the gap in the recipients. And even so: 10 against 10 people, open label design.

Brooke Wilson and colleagues in Auckland gathered the evidence that the success of a transfer might depend on the diversity and composition of the donor material, and examined the concept of keystone species along the way.

Their finding: in recurrent Clostridioides difficile infection, efficacy in the available trials is high and largely independent of who donated. In chronic conditions linked with a dysbiosis it is modest and the response far more variable, and several trials point to a dependence on the donor. As a model for the future they propose a transfer matched deliberately to the particular gap.