Efficacy of different modalities of faecal microbiota transplantation in ulcerative colitis: systematic review and network meta-analysis

Abstract

Background:

While several small sample size randomized controlled trials suggested the superiority of faecal microbiota transplantation (FMT) over placebo in ulcerative colitis (UC), the most effective modality to perform FMT remains unknown.

Objectives:

To compare the efficacy of different modalities of FMT to induce clinical remission in patients with UC.

Data sources and methods:

We performed a systematic review and network analysis (sources: MEDLINE, Embase, Cochrane CENTRAL; random effects model) of randomized controlled trials including at least one arm of FMT in adult patients with active UC. The primary endpoint, that is, clinical remission (total Mayo score ⩽2 with Mayo endoscopic score ⩽1), was assessed between weeks 6 and 12. Results are expressed as relative risks with 95% confidence intervals, adjusted for bowel cleansing and pre-FMT antibiotics. Ranking of FMT modalities was calculated as their surface under the cumulative ranking (SUCRA).

Results:

Among the 12 selected studies, patients were exclusively bio-naïve in 4 studies (4/12), while between 9% and 32% had prior biologics exposure in the other trials. The risk of bias was low across all domains in seven studies. Contrary to upper gastrointestinal tract (GI) FMT (Relative risk (RR) = 1.1 (0.2–7.7)), oral capsule (RR = 7.1 (1.8–33.3)), lower GI FMT (RR = 4.5 (1.7–12.5) and combination of both (RR = 12.5 (2.1–100)) are more effective than placebo to induce clinical remission. The combination of lower GI FMT and oral capsule was significantly more effective than upper GI FMT to induce clinical remission (RR = 10.7 (1.1–104.2)). Combination of lower GI FMT and oral capsule ranked the highest for the induction of clinical remission (SUCRA = 0.93). Multidonor FMT did not perform better than single donor FMT. Autologous FMT ranked lower than placebo (SUCRA = 0.12 vs 0.22).

Conclusion:

The combination of lower GI and oral capsule FMT seems to be the best modality of FMT for patients with UC. In clinical trials, autologous FMT should be avoided due to a potential detrimental effect.

Trial registration:

PROSPERO registration number: CRD42023385511.

Plain language summary

Faecal microbiota transplantation in ulcerative colitis

Faecal microbiota transplantation (FMT) is effective to induce remission in patients with UC. Combination of lower GI FMT and oral capsule is the best modality. Upper GI FMT and autologous FMT are not suitable in patients with UC.

Introduction

Ulcerative colitis (UC) is a chronic inflammatory bowel disease that can involve the mucosa of the rectum and the colon. ¹ UC is a progressive disorder that can lead to long-term complications such as disease extension, motility disorders, and colorectal cancer. ² The current guidelines recommend to treat beyond symptoms owing to potential discrepancies between symptoms and inflammation. ³ Thus, the therapeutic target is to achieve a combination of symptom disappearance and endoscopic remission, so-called clinical remission in UC, to alter the natural history of the disease. ⁴ After 5-ASA and steroid failure or intolerance, advanced therapies are now the reference in patients with UC. ⁵ However, the therapeutic target is not reached in a substantial proportion of patients with such a strategy. Because the alteration of the gut microbiota, also called dysbiosis, is hypothesized as playing a key role in the pathophysiology of UC, ⁵ treatments targeting it could be a relevant alternative. After mixed results with antibiotics or probiotics,^6,7 Faecal microbiota transplantation (FMT) could be effective to induce remission in patients with UC. Several randomized controlled trials suggested the superiority of FMT over placebo in patients with active UC^8–11 or chronic pouchitis. ¹² Meta-analyses of these studies also confirmed that FMT could be a therapeutic option, even though the sample size of these trials remains small and the modalities of FMT are very heterogeneous.^13–19 Recently, the first international Rome consensus conference on gut microbiota and FMT in inflammatory bowel disease delivered specific statements and recommendations to provide guidance and general criteria required to promote FMT as a recognized strategy for the treatment of IBD. ²⁰ This group of FMT experts considered that future research is required to determine the optimal formulation and route of administration for FMT-based therapy in IBD. ²⁰

In this systematic review and network meta-analysis, we aimed to assess and compare the different modalities of FMT to induce clinical remission in patients with UC.

Methods

Search strategy and selection criteria

This systematic review and network meta-analysis were performed using a pre-established protocol and were registered online on the website of the International Prospective Register of systematic reviews (PROSPERO) with the following registration number: CRD42023385511. The search was performed independently by two authors (J.C. and A.B.) according to the PRISMA guidelines. Disagreement was resolved by discussion between them, leading to a consensual decision. The two authors searched MEDLINE, Embase, and the Cochrane CENTRAL without language restrictions for articles published between January 1, 1990, and July 14, 2022. The search algorithms are detailed in Supplemental Appendix 1. To be included in our network meta-analysis, the potential studies had to be randomized controlled trials with at least one arm investigating the efficacy of FMT regardless of administration modalities, enrolling patients >16 years old with a confirmed diagnosis of UC, with active UC according to total Mayo score at baseline (total Mayo score >2). ²¹ In addition, the total Mayo score had to be assessed between week 6 and week 12. Concomitant infection with Clostridioides difficile or pouchitis and severe acute colitis were not eligible. No restriction was applied based on disease severity or prior lines of treatment.

Outcome assessment

According to the STRIDE 2 consensus from the International Organization of Inflammatory Bowel Disease, ⁴ the primary endpoint of our network meta-analysis was clinical remission according to total Mayo score (total Mayo score <2), which is in fact a composite clinical and endoscopic endpoint.

Risk of bias assessment

Two authors (J.C. and A.B.) independently assessed the quality of the included trials using the Cochrane Risk of Bias tool, version 2.0. ²²

Data collection

Study data were collected and managed using REDCap (Research Electronic Data Capture) tools, which is a secure, web-based application designed to support data capture for research studies, providing (1) an intuitive interface for validated data entry; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages and (4) procedures for importing data from external sources. For each trial, we gathered the following data from each arm: number of patients, age, gender, total Mayo score, concomitant use of a stable dose of immunosuppressive or biological therapy, and prior exposure to biologics. For the arms focused on FMT, information about pre-treatment with antibiotics or bowel cleansing, as well as FMT modalities, was also collected.

Data are not publicly available but are available from the author upon reasonable request.

Statistical analyses

We performed a network meta-analysis using the frequentist model, with the statistical package ‘netmeta’ V.0.9–0 in R V.4.0.2, R core team, Vienna (Austria). We explored direct and indirect comparisons of efficacy between FMT modalities. Transitivity was checked to ensure similarity regarding patients, interventions, the comparator, and outcomes. Inconsistency was evaluated by comparison of direct evidence with the indirect evidence.

Heterogeneity among studies was evaluated by I² tests, with values greater than 50% suggesting substantial heterogeneity, as well as by τ² = 0% (global statistical heterogeneity across all comparisons). As previously described, we used estimates of τ² of approximately 0.04, 0.16, and 0.36 to consider the level of heterogeneity as low, moderate, and high. ²³

A random-effects model was used to give a more conservative estimate of the effect of individual modalities of FMT, allowing for any heterogeneity among studies. Results were expressed as relative risk with a 95% confidence interval, adjusted for bowel cleansing and pre-FMT antibiotics therapy. We calculated the relative ranking of FMT modalities for clinical remission as their surface under the cumulative ranking (SUCRA), which represents the percentage of efficacy achieved by an agent compared with an imaginary agent that is always the best without uncertainty. ²⁴ A higher SUCRA score means a higher ranking for efficacy outcomes. For our network meta-analysis, particular attention was paid to avoiding breaking randomization as recommended by Jansen et al. ²⁵ A sensitivity analysis, distinguishing single donor FMT and multidonor FMT (pooled samples from different donors), was scheduled for assessing the potential impact of pooled samples on FMT efficacy.

Role of the funding source

There was no funding source for this study.

Results

Bibliographical search and description of selected articles

Our search strategy generated 1877 citations, including 568, 1129, and 180 in Medline, Embase, and Cochrane, respectively (Figure 1). After removing duplicates (n = 305), 1572 studies were screened for eligibility, leading to the assessment of 24 studies in patients with UC. Among these studies, 12 were withdrawn owing to exclusion criteria (Figure 1). Eventually, 12 studies were retained for the analyses, encompassing 16 arms of patients treated with FMT. All of them were randomized controlled trials assessing the efficacy of inducing remission in patients with UC using FMT. Enrolled patients had active UC with a total Mayo score ⩾4 (Table S1). The control arms (three arms in one study) were placebo (5/12), autologous FMT (2/12), continuation of 5-ASA (3/12), or FMT with other route of administration (2/12; Table S1). Among the 16 arms of FMT, before receiving the faecal transplant, patients were treated with antibiotics in 4 studies (4/16), while bowel cleansing by PEG was performed in 6 studies (7/14; Table S2). FMT was used as an add-on therapy with stable doses of immunosuppressive and/or biological therapies in 0%–49% and 0%–33%, respectively (Table S1). Enrolled patients were exclusively bio-naïve in four studies (4/12), while the proportion of patients previously treated with at least one biologic ranged from 5% to 40% in the other trials (Table S1). Donor transplant resulted from pooled samples from different patients in three arms of treatment (3/16).

Figure 1.

Search strategy according to PRISMA guidelines.

Evaluation of risk of bias and heterogeneity

Risk of bias was low across all domains in seven studies (7/12) but was considered with some concerns in four studies (4/12; Table S3). Of note, one trial was classified as high risk of bias regarding the randomization process. Overall heterogeneity was negligible as illustrated by I² = 0% (0%–71%) and τ² = 0% (global statistical heterogeneity across all comparisons).

Induction of clinical remission

The network plot is provided in Figure 2. Endpoint was evaluated between 6 and 12 weeks after starting therapy. The results of our network meta-analysis are detailed in Table 1. Oral capsule (RR = 7.1 (1.8–33.3)), lower gastrointestinal tract (GI) FMT (RR = 4.5 (1.7–12.5)), and combination of both (RR = 12.5 (2.1–100.0)) were more effective than placebo to induce clinical remission. In contrast, upper GI FMT (RR = 1.1 (0.2–7.7)) as well as autologous FMT (RR = 0.8 (0.15–3.8)) failed to demonstrate any superiority over placebo. When comparing active treatment, the combination of lower GI FMT and oral capsule was significantly more effective than upper GI FMT to induce clinical remission (RR = 10.7 (1.1–104.2)). Based on the SUCRA score, the combination of lower GI FMT and oral capsule ranked highest for the induction of clinical remission (SUCRA = 0.93), followed by oral capsule (SUCRA = 0.82) and lower GI FMT (SUCRA = 0.72). Upper GI FMT (SUCRA = 0.25) and placebo (SUCRA = 0.22) had a similar rank (Figure 3). Interestingly, autologous FMT ranked lower than placebo (SUCRA = 0.12 vs 0.22).

Figure 2.

Network plot of the meta-analysis.

Table 1.

League table for inducing clinical remission in patients with UC.

Relative risks with 95% CIs in parentheses. NMA comparisons are below. Direct or indirect comparisons, column versus row, should be read from left to right. Boxes shaded green denote a statistically significant difference.

FMT, fecal microbiota transplantation; GI, gastrointestinal tract.

Figure 3.

SUCRA for FMT modalities to induce clinical remission in patients with UC. (A higher SUCRA score means a higher ranking for efficacy outcomes.)

FMT, faecal microbiota transplantation; SUCRA, surface under the cumulative ranking; UC, ulcerative colitis.

We performed a sensitivity analysis after splitting trials based on single or multidonor (pooled samples) FMT. We did not see any clear difference between single donor FMT and multidonor FMT (pooled samples) in the case of lower GI FMT or oral capsule (Table S4). Based on SUCRA score, the combination of lower GI FMT and oral capsule ranked highest for the induction of clinical remission (SUCRA = 0.88), followed by single donor lower GI FMT (SUCRA = 0.69), multidonor (pooled samples) lower GI FMT (SUCRA = 0.64), multidonor (pooled samples) oral capsule (SUCRA = 0.61), single donor oral capsule (SUCRA = 0.59; Figure 4).

Figure 4.

SUCRA for FMT modalities to induce clinical remission in patients with UC, distinguishing between FMT from pooled samples from multiple donors and single donors. (A higher SUCRA score means a higher ranking for efficacy outcomes.)

FMT, faecal microbiota transplantation; SUCRA, surface under the cumulative ranking; UC, ulcerative colitis.

Discussion

In this systematic review and network meta-analysis, we identified that the combination of lower GI and oral capsule is the most effective modality of FMT to induce clinical remission in patients with UC. We also observed that the upper GI route of administration is not suitable in this situation. Finally, we reported a detrimental effect of autologous FMT, leading to avoid its use as a control group in clinical trials.

In the absence of direct comparisons between treatments, network meta-analysis is an attractive tool to rank different medications to help physicians with therapeutic decision-making. ²⁵ Owing to multiple heterogeneous procedures of FMT across the different trials, the generalizability of data on FMT is currently difficult, and the best way of performing FMT remains questionable, as reminded by the recent publication from an international expert group. ²⁰ This group stated that future research is required to determine the optimal formulation and route of administration for FMT-based therapy in IBD. ²⁰ In this context, we decided to use network meta-analysis to attempt ranking FMT modalities to guide IBD physicians in clinical trials and in the near future in daily practice.

In our network meta-analysis, we reported that lower GI and oral capsule FMT are more effective than placebo. In contrast, we found that upper GI was not suitable for patients with UC. Zhao et al. ²⁶ suggested in a meta-analysis a trend for higher clinical improvement in the case of administration via the lower GI tract compared to administration via the upper GI (44.3% vs 31.7%), but without statistical comparisons between these two modalities. Another meta-analysis reported that frozen stools from multiple donors delivered via the lower GI had a better curative effect than placebo, while the difference in efficacy between mixed fecal samples from a single donor transplanted through the upper GI and placebo was not significant. ¹⁴ Unfortunately, no indirect comparison was performed to hierarchize these two routes of delivery. Conflicting results considering similar efficacy for all the different routes of delivery were found in other meta-analysis.^27–29 In our study, we confirmed the lack of efficacy after administering FMT via upper GI. It could be partly explained by an alteration of the faecal transplant by different mechanisms of defense against xenobiotics, such as gastric acidity, before reaching the area of interest, that is, the colon and rectum in UC. In contrast, oral capsules have been developed to surpass these potential dangers. Considering also a lower acceptability due to nasoduodenal tube, we recommend avoiding the upper GI tract as a delivery route for FMT in patients with UC. ²⁰

In our study, we described, for the first time, that combination of lower GI FMT followed by oral capsule seems to be the best modality of FMT to induce clinical remission in patients with UC. It remains unclear whether the superiority of this association is related to the route of delivery or to a more regular administration of FMT. However, the SUCRA calculation found that the combination ranked higher than the two procedures independently. Be that as it may, starting with lower GI FMT followed by oral capsule is a seductive procedure that is close to the common use of some biologics, combining intravenous infusion followed by subsequent subcutaneous injections. Pharmacokinetic studies could help to decipher more deeply the process of transplant engraftment according to the route of delivery.

Another interesting result is the fact that autologous FMT ranked lower than placebo, suggesting a detrimental effect. It could be intuitive to contemplate using autologous samples for FMT for a more physiologic control. However, infusing faecal samples with severe alteration of microbiota and pro-inflammatory compounds could contribute to maintain mucosal inflammation. Accordingly, autologous FMT is not suitable as a control arm and should be avoided in randomized controlled trials on FMT.

We scheduled a sensitivity analysis to investigate whether pooling samples from different patients may affect the efficacy of FMT in UC. From our network meta-analysis, including SUCRA ranking, we did not find any clear message regarding any superiority of multi versus single donor FMT. It is in line with data, considering that subgroup analyses of influencing factors showed no differences between pooled or single stool donors did not influence the efficacy of FMT in UC. ¹⁵ In contrast, Levast et al. reported a superiority of multi-donor FMT over single-donor FMT. The authors hypothesized that pooling samples from different donors enables to reduce donor effect and may lead to a gain in microbial diversity that could improve response to treatment. ¹⁹ Whether single or multidonor is the best option for FMT remains to be determined, but the possibility of individualizing the microbiota screening of donors to their patients prior to FMT administration may be important to obtain the best match and increase the rate of clinical remission.

Although we generated interesting results, our data must be interpreted with caution to avoid drawing overstated conclusions. For this, internal and external validity have to be evaluated. The heterogeneity across studies was low in our meta-analysis. Inclusion criteria were highly similar but mixed mild-to-moderate and moderate-to-severe UC. We did not assess the impact of UC severity on FMT efficacy. However, a previous meta-analysis reported a trend for a higher numerical rate of pooled clinical remission for mild-to-moderate UC was 0.46 (0.33–0.58), than for moderate-to-severe UC (0.31 (0.20–0.42)). ¹⁷ It did not mean that FMT is not effective in moderate-to-severe UC, but that FMT is less effective in patients with more severe disease, which is also the case for biologics. Dedicated trials investigating how to position FMT within the therapeutic armamentarium are warranted to clarify the eligible population for such a treatment. Another network meta-analysis reported that FMT was comparable with other agents, such as biologics, in achieving all efficacy outcomes. ¹⁶ In addition to efficacy, several other parameters have to be considered for therapeutic decision-making, such as safety profile, acceptability of the treatment by the patients ³⁰ and the cost for the healthcare system. A dedicated cost-effectiveness study should be conducted to properly assess the cost of this procedure, relying on direct and indirect costs such as donor screening, and to compare it with the cost of advanced therapies (biologics and small molecules).

In contrast, the endpoint (clinical remission defined as total Mayo score ⩽2) was consensual across studies and is in line with the recent STRIDE 2 guidelines. ⁴ In the same way, the time point for assessing remission was close between studies and ranged from week 6 to week 12. Some questions remain, such as the optimal dose and the rhythm of administration of FMT. A meta-analysis identified an impact of FMT dose with a threshold of 275 g. ²⁶

In addition, the interest in using pretreatment use of antibiotics or bowel cleansing is questionable, even though a meta-analysis found that pre-FMT antibiotics or bowel cleansing did not affect remission rates. ¹⁵

The main limitations of our meta-analysis are the small sample size of each individual study as well as the imprecise definition of UC severity and possible variation in definitions of clinical remission. In addition, each study may have different confounding variables within their population (prior exposure to biologics, UC duration) and their protocol that could not be taken into account in our meta-analysis. However, several strengths could be underlined, such as identification of a combination of lower GI FMT and oral capsule as the best modality for FMT, as well as the lack of efficacy using upper GI FMT and the potential detrimental effect of autologous FMT in UC, leading to helpful messages for future clinical trials.

Conclusion

FMT is effective to induce remission in patients with UC. A combination of lower GI FMT and oral capsule is the best modality. Upper GI FMT and autologous FMT are not suitable for patients with UC. Additional studies clarifying the eligible population regarding disease severity and the optimal dose and rhythm of administration are needed before the widespread use of FMT in daily practice.