Fecal microbiota transplantation in inflammatory bowel disease patients: A systematic review and meta-analysis

Objectives Current evidence on fecal microbiota transplantation for inflammatory bowel disease is inconclusive. We conducted a systematic review to gather evidence on the efficacy and safety of fecal microbiota transplantation for inflammatory bowel disease. Methods Systematic searches were conducted in PubMed, Scopus, and Web of Science. Clinical remission was considered as the primary endpoint. Pairwise meta-analyses were performed for the randomized controlled studies (Mantel Haenszel, random effects model). Proportion meta-analyses, accounting for weighted pooled rates reported in the interventional studies, were conducted using the mixed effects model. Subgroup analyses considering the type of stool, donor type, and disease subtype were also performed. Cumulative meta-analyses to assess further needs of evidence were conducted. Results Sixty studies were included, from which 36 could be synthesized in the quantitative analyses. Pairwise meta-analyses of six controlled trials showed significant differences in favor of fecal microbiota transplantation compared with placebo (clinical remission: RR 1.70 [95% CI 1.12, 2.56]; clinical response: RR 1.68 [95% CI 1.04, 2.72]). An overall clinical remission of 37%, overall clinical response of 54%, and a prevalence of 29% of adverse events were found for the interventional studies. Frozen fecal material and universal donors were related to better efficacy outcomes. In addition, Crohn’s disease patients seemed to benefit more from the procedure. Conclusions The comparative analyses demonstrated that frozen fecal material from universal donors may be related to a higher rate of clinical remission, especially for Crohn’s disease.

Introduction Inflammatory bowel disease (IBD) affects about 15% of the worldwide population [1]. Clinical signs and symptoms of IBD comprise abdominal pain, anemia, diarrhea, fecal urgency, nausea, chronic fatigue, weight loss, and gastrointestinal bleeding, producing vitamin deficiencies. Studies evaluating burden of diseases highlight the negative impact of active IBD on the patients' quality of life, especially concerning daily activities and well-being [2][3][4].
The goal of IBD treatment is to control the inflammatory process, achieving disease remission. Current therapies include corticosteroids, anti-tumor necrosis factor alpha (TNF-α) agents (e.g. vedolizumab and ustekinumab), aminosalicylates, immunomodulators (e.g. methotrexate), and surgery [5]. Despite the amount of available therapies, many patients are unresponsive to these treatments or present secondary failure during treatment. Hence, the development of new therapies and the investigation of alternative strategies are needed [6].
Fecal microbiota transplantation (FMT) arises as an alternative therapeutic strategy for the management of several gastrointestinal disorders, including IBD, since the gut inflammatory process is frequently associated with dysbiosis. FMT consists of the transfer of fecal material from a healthy donor, via the lower gastrointestinal tract or upper gastrointestinal tract, aiming to restore the individual's normal intestinal microbiota [1,7].
The use of FMT is well established for the treatment of recurrent Clostridium difficile infection (CDI), with a large body of evidence proving the efficacy of the transplant for this purpose [8]. Since FMT is currently being considered as an experimental treatment for IBD [9], it is highly relevant to critically analyze the available evidence regarding this strategy, to guide clinicians. Therefore, the aim of the present systematic review was to synthesize published data on the efficacy and safety of FMT in patients with IBD.

Ethics statement: Not applicable
A systematic review following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and the Joanna Briggs Institute recommendations was conducted. The search strategy and inclusion criteria were directed by the Population, Intervention, Comparator, Outcome, Study design (PICOS). Hence, we included interventional or observational studies evaluating the efficacy / effectiveness or safety of FMT in patients of any age or gender diagnosed with IBD (namely Ulcerative colitis, Crohn's disease, and Pouchitis [10][11][12]), with or without a treatment comparator and reporting response, remission or adverse events. Exclusion criteria comprised articles published in non-Roman characters, studies including IBD patients in which the indication of FMT was not IBD (e.g. CDI) and studies that did not address the outcomes of interest (i.e. clinical remission, clinical response, adverse events).
Searches were performed in PubMed, Scopus, and Web of Science without limits for timeframe or language (last update April 2020). The following descriptors, combined with the Booleans AND and OR, were used: "Fecal Microbiota Transplantation"; "Fecal Microbiota"; "Foecal Microbiota"; "Faecal Microbiota"; Transplant � ; encapsul � ; capsul � ; Fecal; Microbiota; Bacteriotherapy. Manual searches of the reference lists from the included studies were conducted. As we intended to perform a highly comprehensive search, descriptors for the study design were not included in the search strategy. The complete search strategy is available in the Supplemental Digital Content.
Two independent reviewers screened the titles and abstracts to identify irrelevant records. Then, full-text articles were appraised. In case of disagreements, a third reviewer acted as referee. A standardized form to collect data on the general characteristics (e.g. author, country) and clinical outcomes of the included studies was used. The methodological quality of included randomized controlled trials (RCTs) was assessed using the Cochrane risk-of-bias tool for randomized trials [13], and the quasi-experimental studies were appraised using the Newcastle-Ottawa Scale (NOS) [14].
The primary endpoint outcome was clinical remission, and the secondary outcomes were clinical response and any adverse event. For studies with comparative arms, a pairwise metaanalysis was performed using Review Manager, version 5.1 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Sweden). Results were expressed as risk ratios (RR) with a 95% confidence interval (CI). The Mantel Haenszel statistic and the random-effects model were applied. P-values lower than 0.05 (two-tailed) were considered indicative of a statistically significant difference between groups.
For both the comparative studies and single arm studies, a proportion meta-analysis of weighted pooled rates with a 95% CIs were calculated for each outcome, using a mixed effects model. In addition, subgroup analyses considering the type of stool (fresh or frozen), type of donor (relative/acquaintance or universal [i.e., a donor not related to the patient]), and type of IBD (Crohn or ulcerative colitis) were performed using Comprehensive Meta-Analysis (CMA) (Version 2.0, Biostat, Englewood, NJ). Both for the pairwise and prevalence meta-analyses, between-trial heterogeneity was assessed using the inconsistency relative index, I 2 (I 2 > 50% indicates high and significant heterogeneity) [15].
To evaluate whether further clinical trials are needed to provide more robust evidence on the effects of FMT for patients with IBD, a cumulative meta-analysis was performed for the primary outcome (clinical remission) using Comprehensive Meta-Analysis (CMA) (Version 2.0, Biostat, Englewood, NJ).
To assess the accuracy of the results and the effect of individual studies on data heterogeneity, sensitivity analyses were performed by the hypothetical and sequential removal of studies from the meta-analyses. No study was permanently removed. Funnel plot was applied to evaluate publication bias, along with Egger's test, for which p-value less than 0.05 indicates potential publication bias [16].

Results
The searches in the databases yielded 4,019 records after duplicates were removed. During the screening of titles and abstracts, 3,434 were considered irrelevant records, and, subsequently, 585 studies were assessed after full text appraisal. Two additional records were identified from manual searches, resulting in a total of 60 studies for the qualitative synthesis, with 36 that could be analyzed in the quantitative synthesis (Fig 1). These 60 studies included 27 quasiexperimental trials (interventional studies without randomization or without a control group), 9 RCTs, 5 cohort studies, 5 case series, and 14 case reports. The list of the 60 included references is provided in the Supplemental Digital Content (S1 Table). The 36 records included in the quantitative synthesis comprised all the quasi-experimental studies  and RCTs [45][46][47][48][49][50][51][52][53]. Data on these studies are provided in Table 1. The included RCTs presented an overall moderate risk of bias (S1 Fig). The mean NOS score for the quasi-experimental studies was 6 (ranging from 6 to 8).

Pairwise meta-analyses
Six of the nine included RCTs presented comparable data and were assessed in two meta-analyses, one for clinical remission and one for clinical response [45, 47-49, 52, 53]. For both outcomes, statistically significant differences were observed favoring FMT over placebo (355 patients-clinical remission: RR 1.70 [95% CI 1.12, 2.56], I 2 = 45%; clinical response: RR 1.68 [95% CI 1.04, 2.72], I 2 = 55%). The forest plots are presented in the Supplemental Digital Content (S2 and S3 Figs). In the sensitivity analyses, the study from Rossen et al. [47] was responsible for increasing heterogeneity in the clinical response meta-analysis. After removing this study, the I 2 value dropped to 0%, and more pronounced results favoring FMT over placebo were observed (clinical response: RR 2.13 [95% CI 1.45, 3.12]). The nasogastric route was used (upper gastrointestinal tract) in the study by Rossen et al., [47] whereas the other studies used colonoscopy and enema (lower gastrointestinal tract) [45,48,49]. In addition, donors of the   [47], while all fecal samples in the other three studies were collected from universal donors [45,48,49]. No other aspects that could explain the heterogeneity between the studies were observed. For the clinical remission analysis, no study was identified as responsible for the heterogeneity.

Proportions meta-analyses
Three outcomes were evaluated in the prevalence meta-analyses: clinical remission, clinical response, and any adverse event.
Clinical remission. The analysis of the 24 quasi-experimental studies assessing clinical remission [17-39, 42, 43] [36,48,53], and 66.5% for capsules [37,43], with a statistical difference observed among the groups (p = 0.036). Two articles did not report the type of stool used for the FMT [33, 34] ( Table 2, S4 Fig). A statistical difference was observed when only comparing studies that reported the use of fresh, frozen fecal material and capsules separately (p = 0.010).
The sensitivity analysis showed that the results did not change with the hypothetical removal of one study (S11 Fig). Any  [37,43]. In one study, both fresh and frozen material were used, resulting in a prevalence of 14.4% for any adverse event [36]. The studies by Suskind et al. did not report the stool type [33, 34]. A statistically significant difference between stool type groups was observed (p = 0.033) (  50]. A statistically significant difference was observed between IBD type subgroups (p = 0.047) ( Table 3, S14 Fig).
The sensitivity analysis showed that the results remained unchanged with the hypothetical removal of one study (S15 Fig). The PRISMA checklist for the present systematic review is presented in S2 Table.

Cumulative meta-analysis
The studies included in the cumulative meta-analysis were published between 2013 and 2020. The articles that reported the use of capsules [37,43] were excluded from this analysis since they exhibited pronouncedly higher frequencies of clinical remission. Cumulative meta-analysis shows that the proportion of clinical remission increased over the years and has not stabilized yet (S16 Fig). The funnel plot does not indicate potential publication bias (S17 Fig), which is confirmed by the Egger's test (p-value = 0.25).

Discussion
FMT for the management of patients with IBD demonstrated a response rate of 53.8% with a complete remission of 37%. The administration of frozen fecal material produced better results in terms of clinical remission when compared to fresh material. A randomized controlled trial conducted in Canada compared frozen and fresh FMT in patients with recurrent CDI concluding that the efficacy was similar for both types of stool (75.0% in the frozen FMT group and 70.3% in the fresh FMT group-modified intention to treat analysis) [54]. Our results produced positive evidence towards the use of frozen fecal material, which is coincident with some previous studies [55,56]. Nevertheless, this result should be interpreted with caution, since in all included studies the frozen fecal material was provided by universal donors, which may produce greater microbial diversity in the recipient enhancing the efficacy of the procedure [57]. In addition, for the studies that mentioned the stool processing, the freezing methods were slightly different, with protocols varying in the temperature of storage of the fecal material (-80˚C [27, 45,49] or -20˚C [21,39]). Frozen FMT offers several advantages, including immediate availability of the fecal material, cost-savings associated with the less donor screenings required, and the lower structural requirements of the practice setting where the transplantation is performed [54]. In addition, it is important to notice that the freezing and thawing process do not significantly alter the viable microbiota composition [58]. Although no significant difference was found when comparing relatives/acquaintances with universal donors, our results pointed to higher rates of clinical remission and clinical response when the fecal material was provided by universal donors. A potential reason of the lack of significance could be the large confidence intervals of the pooled proportions in relative/acquaintance and universal donors. Previous studies also showed no significant differences in outcomes between FMT from universal and from patient-identified donors [59]. These results reinforce the universal donor model, which promotes a cost-effective access to FMT [60].
Despite the significant difference observed between the type of stool groups in the adverse events assessment, pointing to a higher occurrence with frozen fecal material, this result should be interpreted with caution, since most of included studies do not standardize the adverse events report, which may generate bias. A similar situation was observed for the type of IBD, with a very small proportion of adverse events in patients with Crohn's disease. Hence, it is not possible to definitely infer that the stool characteristics and the type of IBD influence the occurrence of adverse events related to the procedure. No difference was obtained when assessing adverse events grouped by the type of donor. Most adverse events reported in the studies were mild and included diarrhea, abdominal pain, nausea, flatulence, and fever, which ceased within 24 hours after the transplant.
FMT has been widely investigated for other gastrointestinal disorders besides CDI, but few systematic reviews have been published indicating that this procedure is effective and safe for IBD management [61][62][63][64][65][66]. In 2012, Anderson et al. conducted a systematic review evaluating FMT in IBD patients wherein no meta-analytical calculations and no controlled trial were included. Despite the weak evidence available at the time of that study, their results pointed to the potential effectiveness and safety of FMT for the management of IBD [66]. Our pairwise meta-analyses support significant differences favoring FMT for clinical remission and clinical response. Nevertheless, there are still a reduced number of comparative studies assessing FMT for IBD. This may be the reason for the over-precautious position of some consensus that limited the use of FMT for IBD, and other non-CDI gastrointestinal disorders, to research settings only [8,67]. Narula et al. published a systematic review with meta-analysis, in 2017, evaluating FMT only for active ulcerative colitis. The authors gathered high-quality RCTs to assess clinical remission and endoscopic remission or response, including only four studies in their metaanalysis. Their findings demonstrated a higher clinical remission combined with endoscopic remission in the FMT group when compared to placebo, with no significant difference between FMT and placebo for serious adverse events [64].
A recent study compared the efficacy and safety of biological agents, tofacitinib, and FMT in ulcerative colitis through a systematic review and network meta-analysis of RCTs. The results of the study showed that all evaluated treatments were more effective than the placebos, with no statistical difference in the efficacy of biological agents, tofacitinib, and FMT. In addition, with the exception of infliximab, no active treatment increased the occurrence of adverse events when compared to the placebos. In conclusion, the authors point FMT and tofacitinib as promising alternatives for ulcerative colitis management [68].
Dang et al. performed a systematic review with meta-analysis comparing FMT with mixed probiotics therapy in ulcerative colitis patients. The authors included RCTs in their analysis showing that all treatments were superior to placebo with no increased risk of adverse events.
In addition, no statistical difference was observed between FMT and the probiotics in terms of clinical efficacy. Despite these promising results, the authors point to many unresolved issues regarding the clinical application of these alternatives, suggesting that more randomized controlled trials are needed [69].
Our results were also promising towards FMT and we additionally performed several subgroup analyses regarding the type of stool, type of donor and IBD subtype. The concerns raised by the aforementioned studies on the need of more clinical trials to confirm FMT effects are in line with the results of our cumulative meta-analysis, which shows that the proportion of clinical remission has not yet reached stability. This means that further studies would potentially increment the evidence on FMT for the treatment of IBD. It is observed that the proportion of clinical remission increased over the evaluated years, probably due to improvements in the protocols of FMT, especially concerning more stringent criteria for selection and screening of donors [70]. Despite de need for more studies assessing FMT efficacy, no publication bias was detected in our analysis, revealing the comprehensiveness of our search and the robustness of our results.
In 2017, Paramsothy et al. conducted a broad systematic review with meta-analysis to assess the effectiveness and safety of FMT in IBD patients [62]. Our results are in consonance with their findings for clinical remission and clinical response in Crohn's disease (52% clinical remission in the Paramsothy et al. study versus 47.6% in our study) and ulcerative colitis (33% versus 35% for clinical remission, respectively; 52% vs 54.6% for clinical response, respectively). Nevertheless, unlike Paramsothy et al., we preferred not to include conference abstracts to gather more reliable evidence. These authors conducted some subgroup analyses, however, their results for donor type were uncertain due to the small number of studies. In contrast to Paramsothy et al., the studies included in our systematic review allowed quantitative analyses of adverse events, demonstrating the benefits of FMT for patients with Crohn's disease and ulcerative colitis [62].
The Third European Evidence-based Consensus on Diagnosis and Management of Ulcerative Colitis, published by the European Crohn's and Colitis Organisation (ECCO), recognizes the encouraging results from RCTs evaluating the ability of FMT to achieve clinical remission in patients with active ulcerative colitis. Nevertheless, the ECCO guideline highlights the need for additional studies to identify the best strategy, taking into consideration the administration route and donor type [71]. Our study pointed to some crucial elements in FMT protocols to optimize the efficacy and safety in IBD patients. Conversely, the 3rd European Evidence-based Consensus on the Diagnosis and Management of Crohn's Disease does not mention FMT as a potential alternative for the treatment of this condition [72]. Notwithstanding, our results demonstrated that patients with Crohn's disease seem to achieve even better outcomes with the FMT than ulcerative colitis patients.
The present systematic review has some limitations. First, only interventional studies were analyzed, so further analysis on observational studies should be conducted to add information on the effectiveness of FMT in patients with IBD. Second, the analysis on adverse events should be interpreted with caution due to potential bias, since several included studies did not report specific events nor provided a detailed information on them. Finally, attention should be taken in the analysis on the type of stool, since this variable may be a confounding. The analysis and the literature show that stool from universal donors may provide better results with FMT procedure, and the included studies with frozen stool used universal donors.
In conclusion, our systematic review demonstrated the positive effects of FMT for IBD management with significant differences compared to placebo. Additionally, several highquality, non-randomized, quasi-experimental studies showed that FMT is a safe alternative with promising remission and response rates. We also found that the use of frozen fecal material from universal donors may be associated with better efficacy outcomes in IBD patients, especially those with Crohn's disease.