The Value of Fecal Microbiota Transplantation in the Treatment of Ulcerative Colitis Patients: A Systematic Review and Meta-Analysis

Background and Aims Fecal microbiota transplantation (FMT) has challenged the traditional management of ulcerative colitis (UC) in recent years, while it remained controversial. We aimed to provide a systematic protocol of FMT treatment on UC. Methods Studies reporting on FMT treatment in UC patients were performed. A fixed-effect model was used to assess the efficacy of FMT. Results Eighteen studies were enrolled (n = 446). A pooled proportion of patients who received FMT had a significant efficacy compared to the placebo group (odds ratio (OR): 2.73, P = 0.002) with a low risk of heterogeneity (P = 0.59, I2 = 0%). The Mayo score decreased to 5 points in a state of mild–moderate activity after FMT treatment, and the optimal range of the Mayo score baseline was 6–9 for FMT administration. Then, the baseline of the Shannon diversity index (SDI) had a negative correlation with the clinical response rate (R = −0.992, P = 0.08) or remission rate (R = −0.998, P = 0.036), and the optimal diversity of bacteria was at 7 days to one month. Moreover, the colonoscopy delivery and unrelated fecal donor had slight superiorities of FMT treatment. Conclusion FMT treatment had a higher efficacy and shorter time-point of early assessment of effectiveness on UC patients compared to traditional therapies. And the optimal FMT delivery and donor were colonoscopy delivery and unrelated donor in clinical practice.


Introduction
Ulcerative colitis (UC), a subtype of inflammatory bowel disease (IBD), is a chronic, relapsing, and remitting disease characterized by the aggressive inflammation contributing to the destruction of the colonic mucosa [1]. Its main symptoms include bloody diarrhea, abdominal pain, urgent, and tenesmus [2,3], which produce a miserable influence on the quality of life. Meanwhile, about 3 million of people suffered from it [4], while the etiologies of UC still have remained unclear and were involved in immunologic, genetic, environmental, and gut microbial changes. Several studies demonstrated that the abnormal changes in gut microbiota (e.g., Firmicutes and Bacteroidetes) reduced the ability of the intestinal environment to fight pathogens and can be relevant with some disease conditions [5,6].
Fecal microbiota transplantation (FMT) was first reported formally in 1958 with treatment of pseudomembranous colitis [7]. Afterward, FMT has been applied for refractory Clostridium difficile infection (CDI) on the basis of the rebuilding of abnormal bacterial species of the human gut [8], which is more effective than antibiotics for recurrent CDI patients (87~90%) [9,10]. IBD patients were at a higher risk for microbiota disorder in the intestinal tract; therefore, FMT has been regarded as a potential treatment for IBD management and showed attractive results, especially for UC patients [11][12][13][14].
Numerous clinical trials have evidenced that FMT was characterized as a bacteria-driven therapy of maintaining remission (22%) of UC and preventing recurrence [15][16][17]. Additionally, Sun et al. [1] and Shi et al. [18] described that the donor selection, the administration type (e.g., enema, colonoscopy, gastroscopy, and nasogastric tube insertion), the time of FMT, and the microbiota relationship of recipients and donors lack systematic analysis. Given that there has been no systematic review focusing on UC subjects, we performed a systematic review with the most reliable evidence to assess the efficacy of FMT and establish a standard practical protocol of FMT administration for UC. Thus, we were supposed to take the Mayo score, SDI, Inflammatory Bowel Disease Questionnaire (IBDQ), and C-reactive protein (CRP) of pre-or post-FMT into account, and the FMT route of administration and donor selection were also considered for analysis.

Searching Strategies.
We searched for articles published in PubMed and Web of Science with the following MeSH terms for fecal microbiota transplant: "fecal microbiota transplantation," "faecal microbiota transplantation," "fecal transfusion," "fecal therapy," "microbiota implant," "flora implantation," "bacteriotherapy," and "FMT." Then, the results were combined using "AND" with studies identified by alternatives for UC: "ulcerative colitis," "UC," "colitis," "inflammatory bowel diseases," and "IBD." We enrolled all relevant articles to May 5, 2017, by reviewing the titles and abstracts about FMT in IBD or UC. Both parallel control and nonparallel control trials were included. Additionally, the reference lists of relevant articles were also scrutinized.

Data Collection and Quality Assessment of Studies.
All study selections and data extractions were performed by two reviewers independently, and disagreements were resolved by discussion. The data contained authors, publication dates, countries, number of patients, severity and duration of UC, frequency of FMT, routes of FMT administrated, participants' characteristics, and remission rates, and response rates of follow-up time. Any discrepancies were resolved through further discussion. The studies had to meet the following inclusion criteria: (1) clinical trials, (2) randomized controlled trials (RCTs) and nonrandomized trials, and (3) patients of any age with UC who underwent FMT; the exclusion criteria are (1) duplicate publication, (2) animal or in vitro trials, (3) articles that included Crohn's disease (CD) patients, (4) language other than English, and (5) case reports, reviews, letters to the editor, and conference abstracts.
The Newcastle-Ottawa Scale (NOS) was adopted to assess the quality of included studies [19]. And eight items evaluated the quality of articles from different sides. In total, scores of ≥4 were considered as high qualities while scores of <4 were considered as poor qualities [20]. The total of eight answers generated the final scores for each study.

Data Synthesis and Statistical
Analysis. The efficacy of FMT was assessed by clinical response and clinical remission. Clinical remission of UC was defined as Mayo score of <3; clinical response a decrease by 30% [13,21]. And the evaluation index of the Mayo score baseline, SDI, IBDQ, and CRP were recorded. The subgroup analysis of FMT protocol in UC patients included FMT route of administration, donor selection, and fecal fresh or frozen status. Then, we analyzed them.
We assessed statistical heterogeneity with the value of I 2 . Values of I 2 of 25, 50, and 75% were assumed to mean low, moderate, and high heterogeneity, respectively; values of I 2 < 50% indicated low heterogeneity with a high quality of results. In the absence of statistical heterogeneity (I 2 < 50%), we could use a fixed-effects model; otherwise, we used a random-effects model. Review Manager (version 5.2) and Stata (14.0) were applied for the analysis of the efficacy among RCTs and the subgroups, respectively. SPSS (version 24.0) and GraphPad Prism (version 6.0) were managed for statistical analysis and drawing. Pearson/Spearman's test was used for the correlation analysis. Results were expressed as mean ± SD or mean ± SEM. Statistical analysis was performed by variance (ANOVA) or Student's t-test [22,23]. All tests were two-tailed, and a value of P < 0 05 was deemed statistically significant difference.
Among the 18 studies we included, the median score was 6.61 with a range of 4 to 8 for each item based on the NOS scoring system. All studies fell in "high-quality study" (those of ≥4 scores). Overall, the quality of included studies was deemed eligible. The qualities of each study included in our review are shown in Table 2.

The Efficacy of FMT and Placebo Treatment in UC.
A total of 16 articles were reported on the efficacy of FMT in UC patients, and the overall clinical response rate and clinical remission rate were 46.18 ± 25.08% and 28.96 ± 22.39%, respectively. Among four RCTs, only three trials included the FMT group and placebo group; besides, the study of Rossen et al. [55] divided all cases into FMT-autologous and FMT-donor fecal groups as control and experimental groups. No heterogeneity was found (P = 0 590, I 2 = 0%), so a fixed-effect model was used; the FMT group had a possession of 94 among 187 cases, while 93 of 187 in the placebo group did not receive FMT treatment (e.g., water, amoxicillin, fosfomycin, and metronidazole). The patients that received FMT had a more apparent clinical response rate of 59.23% compared to the placebo group (39.13%) with a significantly statistical difference (OR: 2.73, 95% CI: 1.45-5.15, P = 0 002) ( Figure 2).

The Relation of Evaluation Index and Efficacy at Pre-/
Post-FMT in UC. The evaluation index of FMT efficacy in our review included the Mayo score, SDI, IBDQ, and CRP. We analyzed their changes at pre-FMT and post-FMT; meanwhile, we demonstrated that their baseline impacted the clinical response or remission rate at the intervention of fecal transplantation.
Six of eighteen studies reported the Mayo score of UC patients [21,42,[50][51][52]57] and included 108 cases. A total Mayo score of <3 was defined as clinical remission of UC, and a decrease of >30% was the clinical response of UC; Mayo scores equal to 3, 5, and 10 were assumed to represent mild, moderate, and severe active UC. Figure 3(a) shows that the Mayo score decreased to 5 at the endpoint of FMT among almost all of the studies, which were characterized as mildmoderate activity of UC. Interestingly, the Mayo score in the combination of the FMT and pectin groups (Wei et al. [57]) decreased to 2.25 ± 0.75 at the endpoint with clinical remission; at the same time, the FMT group showed a lower clinical response rate of 34.86% and a remission rate of 13.75% compared to the FMT + pectin group with a response rate of 70% and a remission rate of 30%. Figure 3(b) shows that the optimal range of the Mayo score baseline was 6-9 in terms of the clinical response or clinical remission of FMT.
There were three articles reported on the Shannon diversity index (SDI) with a measure of the colonic bacterial diversity on pre-FMT and post-FMT in UC patients [21,43,57]. Only 20 patients were enrolled in this analysis. The SDI of post-FMT reached the highest level from seven days to one month ( Figure 3(c)). In terms of clinical response and remission, both of them had significant negative correlations with an SDI baseline of pre-FMT in UC patients (Pearson: R response = −0 992, P response = 0 08; R remission = −0 998, P remission = 0 036) (Figure 3(d)).
Only three trials were related to IBDQ with a total of 94 cases [51,55,57]. It showed an increased tend at endpoint of IBDQ ( Figure 3(e)). And four studies reported on CRP value, of which three trials were eligible [49,51,58]. The decrease of CRP (D-value of CRP, mg/L) has a significant positive correlation with clinical response rate after FMT (Pearson: R response = 0 99, P = 0 027) (Figure 3(f)).

Author
Year Representativeness of the exposed of the colonoscopy route of administration was similar to that of enema. Our second subgroup analysis compared the efficacy of related fecal donor and unrelated donor applied in the FMT of UC patients. Five articles based on related donor [43,45,47,52,55] and two based on unrelated donor [54,57] were available and enrolled. A total of 131 subjects in the related donor group had a clinical remission of 27.79% with a low homogeneity (P = 0 23, I 2 = 31 1%), and the unrelated donor group (n = 51) had a higher clinical remission of 36.95% with a low homogeneity (P = 0 4, I 2 = 0 0%). All data are showed in Figure 4(b) and Table 3.
Finally, data on the bacterial fluid status were represented based on fresh fecal and frozen fecal transplantation. UC patients among six studies (n = 133) [45,48,51,52,57,58] used fresh feces with an overall clinical remission rate of 25.30%. As for the frozen fecal group, only Paramsothy et al. [54] claimed a remission rate of 43.9% (Table 3).

Discussion
To date, this article comprehensively summarized the efficacy and evaluation index in the treatment of FMT among UC patients, which identified 16 articles showing the clinical efficacy of FMT for our analysis. The use of FMT for the management of UC patients resulted in a higher efficacy of 59.23% compared to the meta-analysis of Costello et al. (49%) [59]. Importantly, the FMT route of colonoscopy administration (25.74%) is consistent with the study of Sun et al [1]. showing colonoscopy administration of 29.8%, and the unrelated donor (36.95%) had a significant effect for UC patients. We also mainly discussed that the baseline of the Mayo score and SDI played favorable roles on UC patients, influencing the clinical response and remission of FMT administration. Nonetheless, the results from pre-/post-FMT and subgroup analysis still remained controversial.

Evaluation Index.
The Mayo score was considered as the comprehensive system in almost all aspects, including defecation frequency, hematochezia, endoscopic evaluation, and physicians' score [60]. A score of <3 was considered as a clinical remission, and a reduction of >30% from baseline was a clinical response [57,61,62]. Our analysis showed that the Mayo score significantly decreased to 5 points at the endpoint after FMT, which was consistent with the mild-moderate activity (score of 3-10) of UC (Figure 3(a)); however, Wei et al. [57] investigated the effects of the combination therapy of FMT and pectin with a lower Mayo score and a higher efficacy, which explains why pectin delayed the loss of diversity of transplanted gut flora enhancing the effects of FMT in UC cases. Therefore, FMT needs to use adjunctive therapies to improve its effect. Moreover, the Mayo score baseline of 6-9 holds a significant clinical response. Both of the above results explained the optimal requirement of FMT in UC patients from the Mayo score point of view, which were linked to the complexity of pathogenesis in UC patients.
Moreover, our review has evidenced that, on the one hand, FMT had a short-time effect for the regulation of intestinal flora diversity at the following 7 days at the treatment of FMT, but one month later, bacterial diversity began to fall. Notably, the reason for the transient effect of FMT treatment is still bewildering. Furusawa et al. [63] explained that the following factors might influence the result in detail: the time of FMT administration [57], bowel preparation, and antibiotic usage [64,65]. Nevertheless, the higher the baseline of the Shannon index, the higher the clinical response or remission rate (Figure 3(d)). The verdict of the SDI baseline will provide the proof of assessment of the efficacy for the physician. On the other hand, the SDI of post-FMT reached up to the highest level between 7 days and one month, which was considered as the early efficacy of FMT assessment. The Toronto Consensus of 2015 [66] has evidenced that the time-points of early efficacy of 5-amino salicylic acid, corticosteroids, and anti-TNF mAbs were 1-2 months, 0.5 month, and 2-3 months, separately. Thus, it can be seen that FMT administration is superior to traditional therapies in the treatment of UC due to the shorter assessment time of efficacy.

Subgroup Analysis.
In our review, we found that colonoscopy administration appeared to have a superiority in treatment of UC via FMT administration. Firstly, considering the complex pathogenesis and clinical characteristics of UC, it was likely to start in the rectum and extend to the proximal colon; it therefore was best targeted by lower gastrointestinal delivery of FMT compared to upper gastrointestinal delivery (e.g., nasogastric tube insertion, nasojejunal tube insertion, gastroscopy, and esophagogastroduodenoscopy) [1]. Secondly, our results showed that it was as effective as enema Total (95% CI) Total events Heterogeneity. 휒 2 = 1.07, df = 2 (P = 0.59); I 2 = 0% Test for overall effect: Z = 3.11 (P = 0.002)  administration (25.74% versus 33.37%). And Ishikawa et al. [45] reported results that the administration of colonoscopy had a clinical remission rate of 52.94% in UC patients, and Paramsothy et al. [54] posited that the rate of clinical remission peaked at 43.9% via enema. At the same time, most patients in our enrolled articles often received unclear preintervention impacting the effect of different routes of FMT administration, and we had a difficulty in obtaining complete data and precise analysis after FMT via colonoscopy or enema. Moreover, some studies had suggested a slight superiority of FMT in CDI patients via colonoscopy but without sufficient evidence [9]. Most importantly, colonoscopy could not only deliver enough volume suspensions directly into the site of inflammation in the colon but also visualize the relevant gut lesion achieving two things at one stroke; in addition, there was intolerance to retention enema in some studies compared to colonoscopy due to multiple times of administration and the patients' tolerance itself. Therefore, the FMT treatment via colonoscopy will have a very good prospect in UC patients. However, additional high-quality and better-designed researches are needed for further investigating this procedural aspect.
In the review of fecal donor selection, it still remained debatable in FMT application. The phase II trial of Rossen et al. [55] had already reported that there was no statistically significant difference in clinical and endoscopic remission between UC patients who received FMT from unrelated and related donors. And Costello et al. [67] showed that the clinical remission of UC patients receiving an unrelated fecal donor of 50% was higher than for a related fecal donor of 17%. It was similar to our results of unrelated donor having a mildly higher clinical remission rate of 36.95%. As is generally known, to a great extent, related donors share common microbial species with the recipients (UC patients), which minimized the risk of adverse events associated with FMT administration. Nevertheless, the recipients who received related feces suffered from a recurrence of UC in patients; no data regarding recipients receiving unrelated feces exists in this issue. Then, unrelated donors had advantages of lower costs and simpler process compared to related donors. In light of these observations, unrelated feces may thus be preferable for UC patients.  (2017) Moayyedi (2015) Vinta (2017) Zhang (

Limitation.
FMT has concerned some ethical and medical technology standardization issues largely other than that seen as an upcoming treatment strategy for UC patients. Firstly, some patients were resistant to the use of FMT as a viable treatment option, unless they were only under the circumstance of no other way [68]. Moreover, a matter that should not be neglected was the essence of FMT treatment, organ transplantation, or biological therapy. Finally, a standard and detailed FMT procedure in clinical trials for researchers is still absent. Our analysis involved several limitations in this review: Firstly, our statistical analysis was based on individual articles because we did not have detailed data of each case for all the trials; secondly, the small number of cases among some trials was included by our review, and the evidence might be weak. Thirdly, among the articles about the evaluation index (e.g., Mayo score, SDI, IBDQ, and CRP), less than half of the total studies were enrolled. Finally, several included studies received the drug therapy before or after FMT and may influence the accuracy and specify our results.

Conclusion
We have demonstrated that FMT was an alternative therapy for UC with a certain efficacy. Then, the combination therapy of FMT and a lower baseline of microbial richness probably further contribute to the curative effect in preclinical and clinical practices.

Conflicts of Interest
All authors state that they have no conflict of interest.

Authors' Contributions
Yantian Cao, Fangfang Shen, and Bangjie Zhang conceived and designed the experiments. Yantian Cao and Yuanyuan Wu performed the experiments. Yantian Cao analyzed the data. Jie Wang and Yuanyuan Wu contributed reagents/ materials/analysis tools. Yantian Cao wrote the paper. Yantian Cao is the first author.