Efficacy and safety of fecal microbiota transplantation for the treatment of diseases other than Clostridium difficile infection: a systematic review and meta-analysis

ABSTRACT The intestinal microbiome has been identified as a key modifier for a variety of health conditions. Fecal Microbiota Transplantation (FMT) has emerged as a fast, safe, and effective means by which to modify the intestinal microbiome and potentially treat a variety of health conditions. Despite extensive research of FMT for CDI, there is a lack of clarity informed by systematic synthesis of data regarding the safety and efficacy of FMT for other health conditions. This systematic review used PRISMA guidelines and was prospectively registered with PROSPERO (CRD42018104243). In March 2020, a search of MEDLINE, EMBASE, and PsycINFO was conducted. We identified 26 eligible studies. A meta-analysis of FMT for active Ulcerative Colitis (UC) showed that FMT significantly improved rates of clinical remission (OR = 3.634, 95% CI = 1.940 to 6.808, I2 = 0%, p < .001), clinical response (OR = 2.634, 95% CI = 1.441 to 4.815, I2 = 33%, p = .002) and endoscopic remission (OR = 4.431, 95% CI = 1.901 to 10.324, I2 = 0%, p = .001). With respect to Irritable Bowel Syndrome, a meta-analysis showed no significant change in symptoms following FMT (p = .739). Hepatic disorders, metabolic syndrome, and antibiotic-resistant organisms were conditions with emerging data on FMT. Serious adverse events (AE) were more often reported in control group participants (n = 43) compared with FMT group participants (n = 26). There were similar rates of mild to moderate AE in both groups. Preliminary data suggest that FMT is a potentially safe, well-tolerated and efficacious treatment for certain conditions other than CDI, with evidence for active UC being the most compelling.


Introduction
The intestinal microbiome has emerged as a modifiable target for treating a variety of health conditions thought to be associated with dysregulated microbiome profiles. 1 The intestinal microbiome is believed to have a key role in modifying immunity, inflammation, and -by extensiona plethora of health conditions. [2][3][4] There is now substantial research interest 5 into interventions that might target the gut microbiome to improve chronic diseases, including diet, supplementary prebiotics, probiotics, antibiotics, short-chain fatty acids, and Fecal Microbiota Transplantation (FMT). 6,7 FMT is a technique in which gut bacteria are transferred from a healthy donor to a patient, with the goal of introducing or restoring a stable microbial community in the gut. FMT has been established as an effective means of rapidly modifying the intestinal microbiota and may therefore have potential as a treatment for the many health conditions linked with the intestinal microbiome. 8 FMT is already widely practiced as a highly effective treatment for recurrent Clostridium difficile infection (CDI). 9,10 A wealth of new research is investigating whether FMT may be used to treat other health conditions linked to the intestinal microbiome, 11,12 including gastrointestinal, [13][14][15][16][17] autoimmune, 18,19 metabolic, 20,21 and neuropsychiatric [22][23][24] conditions. There is also promising preclinical evidence supporting the use of FMT in conditions other than CDI, including Major Depressive Disorder, 25,26 schizophrenia, 27 and cardiometabolic syndrome. 28 While reviews of FMT for specific indications such as IBS [29][30][31] and IBD exist, [32][33][34] to date there have been no comprehensive reviews evaluating and synthesizing the entire body of data for both the efficacy and safety of FMT for all conditions other than CDI. This systematic review and metaanalysis addresses the question of whether FMT is safe and effective at treating health conditions other than CDI in humans.

Protocol and registration
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were adhered to as a methodological template for this review. The protocol for this Systematic Review was prospectively registered with PROSPERO (CRD42018104243).

Search strategy and eligibility criteria
The PICO approach (population, intervention, comparator, outcomes) was used to guide the search strategy for this review. The PICO criteria used are outlined below: • Population: Humans participants of any age with any acute or chronic health condition other than CDI. Studies were included only if participants were followed up for at least two weeks post-FMT. • Intervention: All possible variations of human FMT were included. For the purposes of this review, FMT was defined as any process by which a fecal microbiota suspension was transferred from the gastrointestinal tract of a healthy individual into another person with the aim of treating a health condition. 35 • Comparator: Studies were included if they utilized a control group. • Outcomes: When reporting on efficacy, this review used primary outcome measures as described by each study. When the primary outcome did not relate to efficacy, the secondary outcomes relating to clinical efficacy were noted, but results were only considered significant when the primary outcome measure related to clinical efficacy and was statistically significant vs the control intervention. Adverse events (AE) were reported as presented by the included study.
In March 2020, searches were carried out using MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Health Technology Assessment Database, Allied and Complementary Medicine (AMED) and PsycINFO. Reference sections of previously published randomized trials, systematic reviews, and meta-analyses on this and related topics were also searched. Forty-two iterations of the term "FMT" were identified and used as search terms: FMT or fecal microbiota transplant* or fecal microbiota transplant* or microbiota transfer or microbiome transfer or microbiota transplant* OR microbiome transplant* or microbial transplant* or microbial transfer or fecal transplant* OR fecal transplant* or feces transplant* OR feces transplant* or stool transplant* or stool transfer or fecal flora transplant* OR fecal flora transplant* or microflora transplant* OR fecal flora transfer or fecal flora transfer OR fecal bacteriotherapy OR fecal bacteriotherapy OR feces bacteriotherapy OR feces bacteriotherapy OR rectal bacteriotherapy OR fecal flora bacteriotherapy OR donor fecal OR donor stool OR donor feces OR donor fecal or donor feces fecal transfer OR fecal transfer OR fecal reconstitution OR fecal reconstitution OR flora reconstitution OR microbiome reconstitution OR feces reconstitution or feces reconstitution. The following modifiers were applied: studies relating to humans and published in English.

Study selection
The following study types were included: randomized controlled trials (RCTs), non-randomizedcontrolled studies, and observational studies with a comparator arm. In the case of observational studies with a comparator arm, only prospective cohort studies were included in order to assess temporality. Reviews, abstracts, conference papers, and posters were excluded.
Two investigators (JG and JD) independently performed the searches using Rayyan software. JG performed initial screening to identify potentially eligible studies. Articles were first screened by title and abstract. Remaining articles were further scrutinized by full-text review. JD acted as a secondary reviewer and was blind to JG's screening outcomes. Where there was a lack of consensus between the two reviewers, the senior author (WM) acted as a third reviewer to make a final decision on whether the study met inclusion criteria.

Risk of bias assessment
Methodological heterogeneity was evaluated by comparing included data using the 'risk of bias' tables. The Cochrane Risk of Bias tool was used to assess the risk of bias in randomized trials. JG and JD independently evaluated risk of bias. 36

Statistical analysis
Data from individual trials were to be combined, and a meta-analysis performed only if the data were amenable. Patient groups, disease entity, and outcome measures needed to be sufficiently similar in order for synthesis to occur. Our meta-analyses were conducted in Comprehensive Meta-Analysis (version 3.3.070) 37 using a Mantel-Haenszel random-effects model to account for heterogeneity between studies. The I 2 -statistic was used as an indicator of heterogeneity. A value of 0% indicates no observed heterogeneity, and larger values indicate increasing heterogeneity. Due to the limited number of studies included in each meta-analysis, no sensitivity or subgroup analyses were performed. Furthermore, no test of publication bias was performed due to the limited number of trials. 38

Assessment of microbial "engraftment"
This review also assessed whether successful "engraftment" occurred of the donor microbiome in the recipient. For the purposes of this review, the term "engraftment" was ascertained according to the following key concepts: • Was a change in recipient microbiota observed following FMT?
If a change in recipient microbiota was observed, then: • Was this change toward the donor microbiota?
• To what extent/how significant was that change? • For how long did the microbiota changes persist following cessation of FMT?

Study selection
The systematic search identified 5,495 de-duplicated studies, of which 26 met eligibility criteria for inclusion (see Figure 1).

Study characteristics
As per Table 1, of the 26 articles that were included in the final review, 20 were Randomized Controlled Trials (RCTs), two of which were open-label, three single-blind and 15 double-blind. The remaining six studies encompassed non-randomizedcontrolled studies (n = 3), case-control studies (n = 2), and cohort studies (n = 1). Eight studies investigated the use of FMT for Inflammatory Bowel Disease (IBD), six for functional gut disorders, four for hepatic disorders, three for metabolic syndrome, two for antibiotic-resistant organisms, and one each for "pouchitis," obesity without metabolic syndrome, and Human Immunodeficiency Virus (HIV). A total of 1149 participants were enrolled in the included studies, with a mean of 44 participants per study (sample sizes ranged from 6 to 165). When broken down by disorder, there were a total of 463 participants in studies relating to IBD, 424 for functional gut disorders, 104 for metabolic syndrome/ obesity, 109 for hepatic disorders, 60 for antibioticresistant organisms, and 14 for other disorders. Study follow-up periods varied from 2 weeks to 12 months. The largest sub-groups by disorder were UC (n = 6) and irritable bowel syndrome (IBS) (n = 5). These groups were large enough to allow for meta-analyses.

Dose
Dosing was inconsistently described. Twelve studies did not provide clear information regarding amount of stool used. Fourteen studies reported on the initial sample of fresh stool, whilst eight described the amount of "suspension" used, which consisted of filtered stool diluted with normal saline and sometimes mixed with a cryoprotectant such as glycerol. Doses of 12 g-250 g of fresh stool were reported in the 14 studies that did provide these data.

Adjunctive treatments
A wide range of adjunctive treatments were employed. Fourteen studies used bowel preparation, 14 56 and four did not specify. 46,57,63,64 Screening protocol Donor screening protocols overall were incompletely and poorly described. Where screening was stated as occurring, the methods for screening were frequently not provided. However, the more recent studies tended to have better reporting of screening protocols and more comprehensive screening. Fourteen studies specifically screened for metabolic risk factors, but only seven specifically described screening for mental illnesses.

Efficacy
Results were categorized by disorder and are summarized in Table 2. Of the 26 included studies, 10 reported significant results for their primary outcome measures, where these related to clinical efficacy. These 10 studies related to functional gut disorders, 14,40,61,64 Hepatitis B, 63 IBD, 53,54,57 antibiotic-resistant organisms, 46 and metabolic syndrome. 52 The evaluated conditions were highly heterogeneous, even within groups. Nonetheless, it was possible to perform meta-analyses for two groups of disorders: IBS, and active UC.

Inflammatory bowel disease
There were eight studies of IBD, six of which were of active UC, the remaining two being of Crohn's Disease (CD), 45 and maintenance of remission in UC. 60 Sokol et al 45 conducted a randomized, singleblind, controlled trial comparing colonoscopic FMT with placebo in 17 adults with CD. There was a significant decrease in CD symptoms in the FMT group compared with placebo (p = .03). 45 Sood et al 60 conducted a double-blind, randomized-controlled trial (RCT) of colonoscopic FMT compared with placebo as maintenance treatment for inactive UC. The study did not find a significant difference in the primary outcome measure (steroid-free clinical remission) between groups (p = .111); however, significant betweengroup differences were reported in endoscopic remission (p = .026), histological remission  Control: Antibiotics only. POM: Mayo Score at the 90-day follow up point between the two groups, wherein a reduction in total Mayo score by three or more was considered a clinical response, and a score of two or less was considered remission P-values/significance not described.
10/17 (59%) of the FMT group achieved a clinical response and four participants (24%) achieved clinical remission compared to 1/10 (10%) in the control group achieving partial response. P-values not provided, hence significance unclear.
(Continued) There was no significant change in fecal calprotectin levels as data were only available for 5 participants.

Meta-analysis for ulcerative colitis subgroup
Six studies reported on active UC, which was sufficient to perform a meta-analysis for clinical remission, clinical response, endoscopic remission, and endoscopic response. Outcome measures were heterogeneous, were collected at different time points (between 7 weeks and 90 days), and used differing definitions of clinical response/remission and endoscopic remission/response. Five of six used Mayo score, whilst one used Clinical Activity Index (CAI) score. Definitions and data for clinical remission and response are summarized in Supplementary Table 1, and endoscopic remission and response are summarized in Supplementary  Table 2.

Functional gut disorders
In a trial of nasojejunally delivered FMT given daily for six days in adjunct to treatment as usual (TAU) for slow-transit constipation, Tian et al 61 reported a clinical remission rate of 36.7% for the FMT group compared with 13.3% for the TAU control group (p = .04).
Five studies reported on IBS, which was sufficient to perform a meta-analysis for clinical response and average change in IBS-SSS. Different studies used different definitions of clinical response: four used IBS-SSS, and one used GSRS-IBS, mostly at 3 months. Definitions and data for clinical response and change in IBS-SSS are summarized in Supplementary Table 3.

Hepatic disorders
Of the four studies in hepatic disorders, three had significant results for clinical efficacy favoring FMT over control, whilst the fourth did not report significant outcomes. In a trial of nasoduodenally delivered FMT every 4 weeks (for 1-7 treatments) plus TAU for Hepatitis B, Ren et al 63 reported that four of the five participants achieved clearance of HbeAg, whereas all 13 of the TAU controls continued to have a positive HbeAg titer (p = .0002). Bajaj et al 48 conducted an open-label RCT investigating FMT for recurrent hepatic encephalopathy using a retention enema compared with TAU and reported a significant improvement in two measures of cognitive outcomes in favor of FMT (p < .01 for both). Similarly, in a single-blind RCT of encapsulated FMT compared with placebo capsules, Bajaj et al 49 found a significant improvement in cognitive outcomes for the FMT group but not the placebo group (p = .02). Philips et al 62 conducted a retrospective cohort study comparing FMT with three control groups (steroids, nutritional support or pentoxifylline) for the treatment of severe alcoholic hepatitis, but found no significant improvement in the primary outcome of 90-day survival (p = .179).

Metabolic syndrome or obesity without metabolic syndrome
Four studies evaluated FMT for the treatment of metabolic syndrome or obesity without metabolic syndrome, and only one of these had significant results for clinical efficacy regarding the primary outcome. The remaining three showed significant results for secondary outcome measures, all in favor of FMT. Vrieze et al 52 conducted a double-blind pilot RCT of nasoduodenally delivered FMT compared with autologous FMT for metabolic syndrome and reported a significant improvement in week 6 peripheral insulin sensitivity (p < .05) in favor of FMT, but not in hepatic insulin sensitivity (p = .08), diet composition, resting energy expenditure, or counter-regulatory hormones. In a double-blinded RCT examining nasoduodenally delivered FMT for metabolic syndrome, Kootte etal 39 did not find significant differences in their primary outcome measure (change in intestinal microbiota in relation to insulin sensitivity at 18 weeks), nor did they observe a significant change in BMI or SCFA levels at any study time point. In terms of secondary outcomes, change in fecal microbiota composition at 6 weeks associated with improved peripheral insulin sensitivity (from 25.8 [19.3-34.7] to 28.8 [21.4-36.9] mmol kg/ 1 min/1, p < .05) in the allogenic FMT group, whereas autologous FMT had no effect (from 22.5 [16.9-30.2] to 20.8 [17.6-29.5] mmol kg/1 min/ 1, p > .5).
Smits et al 43 conducted a double-blind pilot RCT of nasoduodenally delivered FMT compared with autologous FMT for TMAO production in participants with metabolic syndrome and did not find a significant difference between groups. In a double-blind pilot RCT of encapsulated FMT compared with placebo capsules for obesityrelated biomarkers in participants with obesity but without metabolic syndrome, Allegretti et al found a significant between-group difference in area under the curve at week 12 for leptin compared with baseline (p = .001), but no significant change for other biomarkers of obesity. 50

Antibiotic-resistant organisms
Two studies evaluated FMT for the treatment of colonization of antibiotic-resistant organisms, one demonstrating significant clinical efficacy of FMT over control and the other without significant outcomes. In a retrospective matched case-control study of nasogastric FMT compared with TAU for Carbapenemase-Producing Enterobacteriaceae (CPE), Saidani et al 46 reported a significant delay in negativation of rectal swab cultures 2-weeks post-FMT compared with TAU (p < .001). Huttner et al 59 conducted a multicentre, randomized, open-label, superiority trial of nasogastric or encapsulated FMT (treatment was site dependant), compared with TAU for CPE and Extended spectrum beta-lactamase (ESBL), but did not identify a significant between-group difference in the primary outcome measure for clinical efficacy (p-value not provided).

Other conditions
Two studies were not able to be grouped with the others. They evaluated FMT for the treatment of individuals with HIV and antibiotic-dependant pouchitis, respectively. Neither showed clinical efficacy.

Safety data
There were variable quality and completeness of reporting of safety data for both serious adverse events (SAEs) and mild to moderate AEs, across studies (see Table 3). Studies had a follow up period ranging from four weeks 55 to 1 year . 14 SAEs Of the 26 included studies, 23 provided clear descriptions of SAEs. A total of 69 SAEs were reported from 12 studies; 26 occurred in participants allocated to receive FMT, and 43 in participants in the control groups (see Supplementary  Table 4). Of the 26 SAEs that occurred in participants allocated to receive FMT, all but one was deemed unlikely to be related to the intervention. Twenty of these SAEs occurred in participants who received FMT via colonoscopy or enema, and six in those receiving FMT endoscopically or via capsules. When broken down by specific disorder, 17 of these SAEs occurred in participants with inflammatory bowel disease, three in participants with hepatic encephalopathy, five in participants with antibiotic resistant organisms and one in a participant with IBS.

Mild to moderate AEs
Due to the inconsistent quality and completeness of reporting of mild to moderate AE, it was only possible to pool/summarize data for a small number of included studies (see Supplementary  Table 5). These studies were related to IBS, (n = 4), UC (n = 2), slow transit constipation (n = 1), hepatic encephalopathy (n = 1), and metabolic syndrome (n = 1). As such a cross-indication assessment of adverse events was not possible as the data were insufficiently reported across disorders. Similar rates of mild to moderate AEs were observed in participants allocated to FMT compared to the control groups (see Supplementary  Table 5). However, the following AEs were more common in participants receiving allogenic FMT compared with those allocated to control groups: nausea (reported in 80% of FMT recipients compared with 72% in control groups), constipation (reported in 17.4% of FMT recipients compared with 2.4% in control groups), diarrhea (reported in 16.8% of FMT recipients compared with 6.7% in control groups), transient, or low-grade fever (reported in 8.4% of FMT recipients compared with 3.0% in control groups) and vomiting (reported in 5.9% of FMT recipients compared with 2.9% in control groups).
Incomplete reporting precluded comparison of AE rates between different routes of FMT; however, encapsulated FMT appears to have been the besttolerated route.

Successful microbial "engraftment"
Microbiome analysis pre-and post-FMT was performed in 23 of the 26 included studies. All microbiome analyses used 16s RNA sequencing. The data relating to "engraftment" are summarized in Table 4. All 23 of 23 studies which measured microbiome analysis reported change in microbiome following FMT. Fourteen of the 23 studies reported whether the change in microbiota was toward the donor and, of these, 11 confirmed that the recipient microbiome did move toward the donor microbiome. The remaining three studies did not report significant results.
Reporting on the extent or significance of microbiota changes was inconsistent across studies and the complexity of microbiome data analysis has meant it was not possible to answer the question of the extent to which the recipient microbiome changed toward the donor, as no clear quantification was provided by the included studies. As such, these data are not reported in Table 4.
Regarding longevity of the observed changes in recipient microbiota, it was not possible to answer this question in this review, as included studies either did not follow-up recipients for long enough, or did not measure microbiota changes frequently enough to be able to state the duration for which any changes were observed. However, with these limitations in mind, it appears that the demonstrated microbiome changes were transient and appeared to last between 2 weeks to 1 year following the intervention.

Correlation of "engraftment" with clinical findings
Fourteen of 23 studies reported on associations between "engraftment" and clinical outcomes, and of these, 12 studies had statistically significant results with 10 reporting a significant association between successful engraftment and clinical efficacy and two reporting no association between efficacy and engraftment. These data are summarized in Table 4.

Risk of bias assessment
According to the Cochrane Risk of Bias tool 36 (see  Supplementary Table 6), nine studies were evaluated as "low risk," 14,[40][41][42]49,53,54,58,60,64 six as "some concerns," 39,42,43,45,50,52,57 and five as "high risk." 48,[57][58][59]61 Studies rated "high risk" were: Tian et al, 61 due to incomplete reporting across most domains and inadequate randomization processes; Moayyedi et al 57 due to likely inadequacy of blinding of participants as water enemas were used as placebo, which would likely be easily differentiated from true FMT by recipients; Bajaj et al 48   the studies were open label, with an absence of blinding; and Herfarth et al, 51 due to an absence of a statistical pre-analysis plan, and the fact that the trial was ceased after only six participants were randomized.

Statement of principal findings
This review identified FMT trials for conditions other than CDI, with promising, albeit mixed, outcomes regarding efficacy and safety. Meta-analysis of UC studies found FMT to be superior to control conditions for active disease in terms of endoscopic remission, clinical remission, and clinical response. In contrast, meta-analysis of the five IBS studies did not yield significant results regarding symptoms or clinical response. Regarding clinical efficacy in other applications of FMT, studies were too heterogeneous to perform meta-analyses, but four yielded evidence of clinical efficacy in slow-transit constipation, Hepatitis B, colonization of CPE, and insulin sensitivity in metabolic syndrome. The impact of FMT on psychiatric outcomes was assessed in three studies of IBS patients, with one of these finding significant improvements.
This review also found that FMT was safe and well tolerated. Similar rates of mild to moderate AEs were observed in participants who received FMT compared to those allocated to control groups, while SAEs were more commonly reported in participants allocated to control/placebo groups. Not all studies assessed or reported whether FMT results in successful engraftment of the donor microbiome into the recipient, but a majority of those that did report it confirmed a move toward the donor microbiome following FMT and that these changes persisted for up to 1 year. Furthermore, four of the five studies that reported on association between microbiome changes and clinical efficacy, four of five confirmed such an association. This suggests that FMT alters the recipient microbiome, and that it is possible that this change is a contributing factor to clinical efficacy.

Strengths and weaknesses of the review
This review is the first systematic review to evaluate both safety and efficacy of FMT for all disorders other than CDI. This review aimed to recruit higher quality studies by excluding uncontrolled studies, which represent a majority of studies in this field. Whilst other reviews have been conducted with respect to safety of FMT for indications other than CDI, these reviews are either not recent, 65 or were restricted to a single indication such as IBS [29][30][31] and IBD. [32][33][34] With respect to efficacy, whilst other reviews have been published for single indications, such as IBS [29][30][31] and IBD, [32][33][34] there have been no holistic reviews looking at all indications other than CDI. As far as the authors of this review are aware, this review also represents the most up to date systematic review and metaanalysis of the safety and efficacy of FMT for IBS.
However, the 26 studies included were heterogeneous and of mixed quality, with several using open-label designs and small samples. Encouragingly, more recently published studies appear to be of higher quality, using more robust study designs (such as double-blinded RCTs), and with clearer and more complete descriptions of study methodology.
It was possible to conduct meta-analyses for both IBS and active UC. However, due to the lack of consensus regarding outcome measures and small number of included studies, the results of these meta-analyses should be considered preliminary at this stage. Further, due to the low numbers, tests for publication bias (e.g. eggers regression and funnel plots) were unable to be carried out.
We were unable to undertake a quantitative analysis on the level of engraftment, given the gaps in data in the included studies. Future studies should evaluate microbial engraftment as a result of FMT, allowing for a systematic assessment. This paper evaluated safety data across a range of indications, finding broadly that FMT is well tolerated and safe. However, due to the poor quality and incompleteness of reporting in several papers, a cross-indication analysis of safety data was not possible. We recommend future FMT studies report more clearly on mild to moderate and SAE.
With respect to SAE, these were observed more frequently in control group participants than those allocated to receive FMT, and of the SAE observed in FMT recipients, most considered to be unrelated to FMT. In understanding this finding, it should be noted that most SAE were likely due to the underlying disease process rather than the FMT procedure, a majority of reported SAE were flares of the disease in inflammatory bowel disease participants. Thus, FMT may have prevented disease flares.

Implications for clinicians and policymakers
With respect to active UC, our meta-analysis revealed that FMT appears to be clinically efficacious compared to control conditions. Four of the six included studies used a gold-standard doubleblind placebo-controlled RCT design, and all six included studies favored FMT over control conditions regarding clinical efficacy, notwithstanding limitations described above. Thus, the quality and consistency of outcomes appear to favor FMT in the treatment of active UC, making this is a promising area for research attention.
Evidence supporting the application of FMT in the treatment of IBS is more equivocal. Five studies included in this review showed mixed outcomes, with three reporting that FMT was favorable, and two finding that control conditions were more effective than FMT. Possible reasons for these mixed findings are discussed in depth in other review papers, 29-31 which note, inter alia, that route of delivery, choice of placebo (i.e. inert vs autologous FMT), and patient group may have contributed. [29][30][31] Suffice to say, there is no strong evidence at this stage that FMT is efficacious for the treatment of functional gut disorders.
This review also identified two additional studies relating to inflammatory bowel disease yielding significant outcomes regarding clinical efficacy, and four studies that evaluated FMT for metabolic syndrome or obesity that focused on biological outcomes rather than clinical efficacy. As such, these are identified as conditions of interest for further research only.
Regarding safety, FMT appears generally to be a safe and well-tolerated treatment, with orally administered FMT appearing to be the best-tolerated route. However, it is also important to note that on 13 th June, 2019 the American-based Food and Drug Administration (FDA) released a statement warning of the risks of FMT. They reported two cases (both immunocompromised patients) in whom antibioticresistant organisms, (specifically, Extended Spectrum Beta Lactamase-producing Escherichia coli (E.coli)) were transferred via FMT, resulting in one death. In these cases, donor feces were not screened for antibiotic-resistant organisms. 66 A further warning was issued on 12 th March 2020 advising of six cases of additional transmission of antibiotic-resistant organisms via FMT provided by a US-based stool bank (enteropathogenic Escherichia coli in two cases and Shigatoxin-producing Escherichia coli in four cases) and two deaths that occurred in recipients of FMT, but in which FMT may not have been the cause of death. 67 It is now standard across widely accepted protocols in the United Kingdom, United States, and Australasia to screen thoroughly for antibioticresistant organisms. These recent serious incidents highlight the importance of adhering to rigorous screening protocols, such as the Openbiome Protocol in the US, 68 the British Guidelines for donor screening, 69 or the Australasian guidelines. 70

Unanswered questions, challenges for the field of FMT research, and future directions
Of all the uses of FMT for conditions other than CDI, the most promising at this stage is for active flares of UC. Further large scale, high-quality studies, utilizing consistent data points to measure primary outcomes, are urgently called for. Other indications with some promise include metabolic syndrome/obesity, antibiotic-resistant organisms, and certain hepatic disorders. Whilst published outside of the time range of the search performed for this review, a recent Phase I study investigating FMT for Alcohol Use Disorder showed safety and efficacy. 71 Thus, we watch with interest the growing field of FMT for hepatic disorders.
However, on the whole, studies in these emerging areas are heterogeneous and generally of poor quality, with most using open-label designs and only one study using clinical efficacy as a primary outcome measure. Again, further high-quality research, using larger sample sizes and doubleblinded, placebo-controlled designs, and that use clinical efficacy endpoints as a primary outcome, are needed for these emerging indications. Furthermore, no studies evaluated the use of FMT for psychiatric conditions, an area of great importance given growing interest and data supporting the relationship between mental health and the gut microbiome ("the microbiota-gut-brain axis"). 7,72 Much also remains unknown about the ideal methodological design for studies of FMT for conditions other than Clostridium difficile infection. For one, an important question remains around choice of placebo. Nine studies selected an inert placebo, such as water mixed with glycerol and food dye, whilst eight opted for autologous FMT as a placebo. There is some evidence to suggest that even autologous FMT may have an impact on gut microbiota, 58 which may confound results. This needs to be further explored but presents an argument against using autologous FMT as a placebo in future studies aimed at determining efficacy of FMT compared with an inactive control. Followup periods to assess long-term safety, engraftment, and metagenomics are also an important consideration for study design. The authors of this review suggest a follow up period of at least 6 months to adequately monitor for safety and long-term AE.

Conclusions
This systematic review and meta-analysis provide preliminary data that FMT may be safe and effective for several conditions other than CDI. Preliminary meta-analyses suggest efficacy for outcomes related to active ulcerative colitis but not IBS. Hepatic disorders, metabolic syndrome/obesity, and antibiotic-resistant organisms were also identified as emerging areas of interest for FMT research. Regarding safety, there was little difference in SAEs between participants allocated to receive FMT and those allocated to control groups. With respect to mild to moderate adverse events, similar rates were also observed in treatment and control groups. These encouraging pilot outcomes provide preliminary support for further highquality research in these areas.