Faecal microbiota transplantation : a review of FMT as an alternative treatment for Clostridium dif fi cile infection

Clostridium difficile infection (CDI) is the most common aetiology of hospital acquired infections, the leading cause of nosocomial diarrhoea and a significant clinical and economic burden. Recommended treatment for CDI is prescription of broadspectrum antibiotics. Antibiotic treatment disrupts natural gut microbiota allowing C. difficile to colonize. Clostridium difficile is also resistant to antibiotics and may persist in the gastrointestinal tract for months causing recurrent disease. There is a pressing need for alternate therapies. Faecal microbiota transplantation (FMT) is a promising treatment option whereby healthy donor faecal samples are infused into a patient thus, theoretically, restoring normal bowel conditions. Until recently there has been a paucity of randomized controlled trials and reports of FMT efficacy have been limited to case studies and series. Here, a meta-analysis of 684 patients treated with FMT across 23 independent studies was conducted, reporting a mean resolution rate of 90.4%. Three RCTs are included that, although methodologically limited, appear to verify reports of high efficacy in case reports with minimal adverse effects. FMT appears to be a safe, highly efficacious and affordable treatment option. Future research should focus on potential long-term safety concerns and optimizing protocols for donor screening, patient selection and FMT administration.


Introduction
Clostridium difficile infection (CDI) is the most common aetiology of hospital acquired infections superseding methicillinresistant Staphylococcus aureus (MRSA) (Miller et al., 2011).CDI, the leading cause of nosocomial diarrhoea, is responsible for significant morbidity and mortality.A United States Centers for Disease Control and Prevention (CDC) study estimated that during 2011, in the US alone, there were ~453 000 cases of CDI resulting in ~29 000 deaths (Lessa et al., 2015).The economic burden of CDI in the US is ~$4.8 billion per annum (Vonberg et al., 2008;Dubberke and Olsen, 2012).works against anaerobic bacteria (Cohen et al., 2010).The favoured drug for first case moderate-severe episodes or where previous metronidazole treatments have failed is a glycopeptide, vancomycin (Cohen et al., 2010).Between 2004 and2007, treatment failures of metronidazole increased by 16-38%, vancomycin failures remained between 1% and 6% (Zar et al., 2007).
Intestinal microflora defends itself from harmful microorganisms by a mechanism known as colonization resistance (Wilson, 2005).For example, Bacteroides thuringiensis DPC 6431-a bacterial strain isolated from human faecesproduces a bacteriocin (Thuricin CD) that directly targets C. difficile and consequently prevents C. difficile from infecting the host (Rea et al., 2010).Antimicrobial treatments disrupt colonization resistance thus creating favourable conditions for the successful colonization of C. difficile (Olson et al., 1994;Owens et al., 2008).Transmission rates are likely to increase as carrier patients contaminate their environment, in turn increasing the risk of successful C. difficile colonization in susceptible patients such as the immunocompromised (Owens et al., 2008).CDI is now a recognized risk associated with any antimicrobial treatments (e.g.Clindamycin (Pear et al., 1994;Johnson et al., 1999;Gerding, 2004); Cephalosporins (Johnson et al., 1999;Gerding, 2004); Fluoroquinolones (McCusker et al., 2003;Gaynes et al., 2004;Loo et al., 2005)).Global antibiotic consumption increased by 36% from 2000 to 2010 with significant increases in Brazil, China, India, South Africa and Russia (Van Boeckel et al., 2014).In the US, prescription rates of broad spectrum antibiotics increased two-fold during the same time frame (Lee et al., 2014).With widespread reports of antimicrobial resistance, increasing burdens on healthcare systems and rapidly evolving strains of C. difficile associated with increased virulence and epidemic infection rates (Loo et al., 2005), there is a pressing need for alternative therapies.
The aim of faecal microbiota transplantation (FMT) is simple; to counteract the susceptibility to infection of microbial flora in the gut caused by antibiotics, which allows the pathogenesis of CDI, by reinstating natural microbial colonies in the patient by faecal transplantation from a healthy donor.Faeces from a healthy donor is introduced thus, theoretically, restoring normal bowel conditions (i.e.resolution, indicated by (i) C. difficile negative stool assays and/ or (ii) relief of diarrhoea).Using orally administered stool for diarrhoea treatment can be traced back to fourth century China (Zhang et al., 2012).The first example of FMT via enema for treatment of pseudomembranous colitis was described in 1958 (Eiseman et al., 1958).FMT was first used for CDI treatment in 1983 where the patient presented immediate resolution and was asymptomatic at the 9-month follow-up (Schwan et al., 1984).With reports of resolution rates of ~90%, FMT has recently surged in popularity both within the literature and the public eye (Gough et al., 2011;Kassam et al., 2013;Sha et al., 2014).Screening processes for donors are rigorous in order to prevent the transmission of communicable diseases but concerns remain about the transmission of diseases such as norovirus (Schwartz et al., 2013).Despite this, there is still great variation within screening protocols (Moayyedi et al., 2014), even after appeals for standardization (Bakken et al., 2011;Allen-Vercoe et al., 2012).The main gap within the literature, until recently, was randomized controlled trials (RCTs), needed to confirm the efficacy, safety and optimal protocols of FMT.The first RCT was conducted by van Nood et al. (2013) and more recently RCTs have been conducted by Youngster et al. (2014) and Cammarota et al. (2015).Despite growing excitement surrounding FMT, there is little literature addressing concerns of potential long-term safety such as the onset of obesity (Alang and Kelly, 2015), or our relative lack of knowledge about human microbiome interactions despite recent advances (Kamada et al., 2013).

Synthesis FMT delivery procedure
Due to the relative lack of RCTs, there is presently no agreement on the most effective transplantation route.Resolution of CDI has been achieved using colonoscopy (Borody et al., 2003;Keller et al., 2009;Arkkila et al., 2010;Garborg et al., 2010;Rohlke et al., 2010;Yoon and Brandt, 2010;Wilcox, 2011;Brandt et al., 2012;Hamilton et al., 2012;Kelly et al., 2012;Matilla et al., 2012;Kelly et al., 2014;Cammarota et al., 2015), rectal enemas (Paterson et al., 1994;Borody et al., 2003;Silverman et al., 2010;Kassam et al., 2012), nasogastric tubes (Aas et al., 2003;Macconnachie et al., 2009;Rubin et al., 2013), nasojejunal tubes (Borody et al., 2003;Keller et al., 2009), nasoduodenal tubes ( van Nood et al., 2013), orally administered frozen capsules (Youngster et al., 2014) and gastroscopy (Garborg et al., 2010) (Table 1).Postigo and Kim (2012) found no significant difference in efficacy between colonoscopy and nasogastric deliveries.However, there may have been sample selection bias in the data collected for this meta-analysis.Since FMT is not the standard treatment for recurrent CDI, the small homogenous population of patients receiving FMT may have introduced an unavoidable sample selection bias into the original papers used in this analysis.Furthermore, by limiting their searches to English-language publications the results cannot be generalized to the wider population of reported CDI cases treated with FMT.The meta-analysis was also fairly small and only included 12 studies with a total of 182 patients, 148 of whom received colonoscopy treatment.There are also pathophysiological differences between stool infusion via nasogastric tubes and colonoscopy or enema (Kassam et al., 2013).Nasogastric delivery has potential for bacterial overgrowth in achlorhydric individuals (Yoon and Brandt, 2010) but the impacts of this on the GI tract are not well documented (Guo et al., 2012).Alternatively, rectal enemas are both safe and cheap but it is not known if proximal colonization can be achieved by rectal repopulation of the microflora (Kassam et al., 2013) rectal sphincters in elderly patients may also reduce the chances of faecal infusion retention and hence reduce treatment efficacy (Kassam et al., 2012).Colonoscopy is an attractive option, not least because it is able to deliver large volume FMT along the entire colon.Nevertheless, the procedure is less safe than rectal enemas with an increased risk of perforation (Kassam et al., 2013) or aspiration (Kelly et al., 2014).Colonoscopy is also more expensive and time consuming.Procedure costs and safety of any FMT treatment should be carefully considered.

Patient selection
There are no official guidelines for criteria that patients must meet to be deemed eligible for FMT, but guidelines for clinical testing have been put forward by the Faecal Microbiota Transplantation Workgroup (Bakken et al., 2011).These guidelines propose selection criteria for both patients and donors.Patients must display recurrent or relapsing CDI with (i) at least three episodes of mild-to-moderate CDI and treatment failure of a minimum 6-week course of vancomycin, or (ii) a minimum of two severe (requiring hospitalization) episodes of CDI; be unresponsive to standard therapy for ≥1 week during moderate CDI, or a severe infection with no response to standard treatment after 48 h (Bakken et al., 2011).

Donor screening
A major risk associated with FMT is the potential transmission of infectious diseases.Accordingly, screening processes of potential donors are rigorous (Bakken et al., 2011).
Serology tests of blood samples from potential donors screen for HIV types 1 and 2, syphilis and hepatitis A, B and C viruses.Stool is screened for C. difficile toxins, parasites and other enteric pathogens (e.g.Giardia, Cryptosporidium, Isospora and Rotavirus), and Helicobacter pylori (Bakken et al., 2011).Donor selection is then a process of elimination based around specific exclusion criteria (Table 2).A recent study by Hamilton et al. (2012) described a standardised method for the preparation and use of frozen stool samples.
Of particular interest was the fact that samples appeared to retain their efficacy against CDI despite being cryopreserved.The implications of this include use of universal donors who will reduce costs associated with screening process for multiple patients (Borody and Khoruts, 2012).Frozen samples have since been used in an RCT comparing the efficacy of colonoscopy and nasogastric tubes for FMT (Youngster et al., 2014).

Safety
Adverse effects of FMT are low (Table 1), indicating that FMT is a safe procedure.However, no studies have sought adverse effects (Moayyedi et al., 2014).Within the studies reviewed, 18 (78%) reported no adverse effects at all.Of the five studies that reported adverse effects, three were mild transient effects (e.g.flatulence, irregular bowel movements and abdominal pain) all of which resolved after 2-3 weeks (Hamilton et al. 2012;Kelly et al., 2014).Brandt et al. (2012) reported that one patient who was unresponsive to the treatment was not further treated and died in a care home of an unclear cause.Kelly et al. (2014) used FMT to treat CDI in immunocompromised patients.Of the 80 patients treated, 12 presented with serious adverse effects (death or hospitalization) ≤12 weeks post FMT.Ten were hospitalized.Two deaths occurred, one as a result of aspiration during the colonoscopy and one unrelated death.No patients suffered infections directly related to FMT.Three patients reported mild bowel discomfort and five irritable bowel disease (IBD) patients experienced disease flare post-FMT.Although this study did not report any adverse effects in relation to the infusion of faecal material there was a complication with the colonoscopy in one patient.Aside from the surgical complication, this study showed promise for the treatment of immunocompromised patients with FMT.The sample size of 80 was the largest of the studies reviewed here.To confirm their results there is a need for a well-designed RCT that includes immunocompromised patients.A review by Kassam et al. (2013) postulated that there may be a relationship between FMT via nasogastric or nasojejunal tube and increased risk of complications such as upper GI bleeding, enteritis and peritonitis.There are also case reports concerning the transmission of norovirus following FMT (Schwartz, Gluck, Koon, 2013).Although transmission was found not to be related to FMT, the study highlighted concerns about screening protocols.

Limitations of case reports
Literature detailing clinical use and results of FMT is, at present, almost exclusively limited to case studies and reports.In this review, 20/23 (87%) articles reviewed are case series or reports.Case reports cannot be used to generate information on rates or ratios of treatment success due to the lack of cause-and-effect relationships and representative population samples.However, case series can be used to further strengthen arguments for cause-and-effect relationships and indeed generate hypotheses.A major publication bias in FMT treatment trials may exist, with only the case reports and series reporting positive results reaching publication (Albrecht et al., 2005).Another limitation of case reports/series is that they are written retrospectively.This can lead to an absence of relevant medical data which was not recorded at the time.Furthermore, the experiments have no repeatability, it is impossible to recreate duplicate situations.Experiments or observations under similar settings and circumstances can be amalgamated to form a case series.The inherent selection bias introduced in case studies/reports can be minimized by conducting RCTs.

Randomized controlled trials
The first RCT conducted by van Nood et al. (2013) , or vancomycin and bowel lavage (13).They demonstrated a resolution rate of 81% in the FMT group compared to a resolution rate of 31% for those receiving just vancomycin.There was no significant difference between the vancomycin with bowel lavage and the vancomycin-only treatment groups suggesting that bowel lavage alone does not play a significant role in the resolution of CDI.The study was cut short by interim analysis based on the overwhelming difference between treatments and the success of FMT.There are several limitations associated with the study.Firstly, the study was non-blinded.Although RCTs remove the selection bias of case studies if they are non-blinded they do not prevent differing treatment of study groups such as biased assessment and differential co-interventions (Karanicolas et al., 2010).Furthermore, treatment of patients by clinicians may differ between blinded and non-blinded trials (Schulz and Grimes, 2002).Bias may also factor into data collection, outcome adjudicators and statistical analyses (Karanicolas et al., 2010).
One way to potentially overcome the issue of blinding in this study may be to administer the vancomycin via the nasoduodenal tube.Administering the treatments in the same way could blind the study physicians as to which patients were receiving FMT and which were receiving vancomycin.In order to prevent vancomycin treatment being identified in the third treatment arm by the administration of bowel lavage, it may be prudent to introduce a fourth treatment arm in which patients receiving faecal infusions also receive bowel lavage.However, this alternative methodology brings new limitations as the administration of medication via enteral feeding tubes is complicated by a number of issues (White and Bradnam, 2015).Firstly, if the drug in question is not prescribed for administration via a feeding tube then administration of the drug via this route falls outside of the product licence.This means that the manufacturers are no longer accountable for any adverse effects of the drug and the administering physician assumes liability.Other concerns arise when considering the risk of drug errors linked with feeding tube administration of drugs via intravenous syringes; unsuitable drug formulations causing blockages of feeding tubes; shared use of tablet crushing tools increasing risk of cross-contamination; and inappropriate handling of drug powders leading to occupational exposure (White and Bradnam, 2015).In order to proceed with this suggested study format, careful consideration would have to be given to the potential consequences, both legal and professional, of administering vancomycin via the feeding tube.This would include consultation with pharmacists, drug manufacturers, patients, relatives and other healthcare professionals.Protocols of how to administer vancomycin via this route are provided in the Handbook of Drug Administration via Enteral Feeding Tubes (White and Bradnam, 2015).The small sample size of 17 patients receiving the study treatment is another major limitation.Youngster et al. (2014) inoculated patients with faecal material that had been previously frozen.This was the first RCT to compare treatment routes (nasogastric tube versus colonoscopy) and to use frozen inoculate.They found that both treatment routes were equally effective.Like van Nood et al. (2013), this was a non-blinded trial with a small sample size (20) thus suffering the same limitations.However, this was another important step in the literature which showed that cryopreserved faecal samples retain their efficacy.This may have far-reaching implications for collection, storage and use of donor stool samples.Cammarota et al. (2015) followed a similar study design to van Nood et al. (2013).They randomized treatment arms of FMT or vancomycin between 39 patients, 20 and 19, respectively.The FMT treatment arm had a resolution rate of 90% compared with 26% in the vancomycin treatment group.This study was also non-blinded and shares the associated limitations of this with the two other RCTs.
Small sample sizes within the current RCTs limit generalization of results to wider populations.Equally, non-blinded study designs pose significant limitations; however, recurrent CDI presents a clinical challenge for which blinded studies may not be practicable even if they are desirable (Cammarota et al., 2015).Despite this, the results of these studies can provide cautious optimism for the future of FMT.All three studies showed FMT to have high efficacy across varying treatment routes and, importantly, superior efficacy to the currently ubiquitous antibiotic treatments.Furthermore, there is promising signs for the use of cryopreserved faecal samples (Youngster et al., 2014) and solid foundations for further studies to build on these results and test their reliability.

Long-term concerns
One danger with the shortage of well-designed RCTs is the tendency for people to draw conclusions too rapidly.While the average resolution rate appears to be around 90% (Kassam et al., 2013;Sha et al., 2014) there are currently no RCTs that have monitored long-term effects of FMT.Obesity and metabolic syndrome are two examples of conditions in which the microbiome plays a part (Tsai and Coyle, 2009;Nicholson et al., 2012).A recent case study documented a 32-year-old woman who developed new-onset obesity following successful FMT treatment for CDI (Alang and Kelly, 2015).This was noteworthy as she had no history of obesity and the donor also experienced notable weight gain, leading the authors to hypothesize that obesity onset was related to FMT, something which is supported by animal models (Ridaura et al., 2013).No comparable data from microbiome sequencing of the patient and donor exists.This is an important limitation to this case study as cause and effect cannot be proven.Other theoretical long-term problems associated with FMT need to be investigated.Some specialists already err on the side of caution rejecting donors with high body mass indexes or a history of colon cancer (Leading, 2014).

Meta-analysis
Twenty-five articles were identified for review.Of these, 23 articles were included in the analysis composed of journal articles (78%) and abstracts (22%).Two articles were inaccessible online and the authors were unavailable when contacted.Only studies with a minimum of 5 patients receiving FMT were included.In total, 684 patients were treated with FMT for CDI across 23 studies with a mean resolution rate of 90.42% (Fig. 1; Table 1).These results are in line with other reviews (Kassam et al., 2013 (89.7%);Sha et al., 2014 (90.7%)).The mean sample size for included studies was 30 (range 6-80).Of the 23 articles reviewed, three were RCTs (van Nood et al., 2013;Youngster et al., 2014;Cammarota et al., 2015).These studies are significant as they are the first representation of the potential efficacy of FMT as a treatment option for CDI.Many previous reviews have previously highlighted the absence or paucity of RCTs in the literature (Brandt et al., 2012;Kassam et al., 2013;Moayyedi et al., 2014;Sha et al., 2014).The most common procedure used, colonoscopy ( 14), was used exclusively in 10 studies.Enemas were used in five studies, exclusively in three, of which selfadministered enemas were used in one (Silverman et al., 2010).Nasogastric tubes were used exclusively in three studies.Nasojejunal tubes (2), nasoduodenal tubes (1), orally administered tablets (1) and gastroscopy (1) were also used.Faust et al. (2002) did not report an FMT route.Follow-up periods ranged from 60 days to 5 years and were not reported in four studies.
In the 23 studies reviewed, there were numerous combinations of stool donors (Table A1, Appendix).Family and relatives were used as donors in 16 studies.Spouses/intimate partners were the donors in six studies.Unrelated healthy donors provided stool in eight studies.One donor was a member of the clinic staff (Aas et al., 2003).Close friends (4), and household members (2) also provided stool.Donor relation to the patient was not reported in two studies (Keller et al., 2009;Kelly et al., 2014 (abstract only)).

Conclusions
FMT for the treatment of CDI appears to be highly efficacious (90.42%) regardless of the infusion route.This is consistent with the findings of other reviews (Kassam et al., 2013;Sha et al., 2014).In most of the studies reviewed (14/23) clinicians appear to favour colonoscopy, although a recent RCT showed that colonoscopy and nasogastric infusion were equally effective (Youngster et al., 2014).Clear guidelines for practitioners to follow in terms of donor selection are required.At present too much variation within the selection procedures may lead to the omission of certain screening tests and the potentially life-threatening transmission of infectious diseases, such as norovirus (Schwartz et al., 2013).This is of particular concern in the treatment of immunocompromised patients.Hamilton et al. (2012) described a standardized method for the preparation and use of frozen samples for FMT while retaining the efficacy of the treatment.These findings were supported in an RCT (Youngster et al., 2014).Benefits of using frozen samples include: reduced screening costs through use of universal donors, and improved storage, ease of access and transportation of samples between hospitals.Although there are numerous case studies highlighting the short-term success of FMT, long-term safety concerns still need to be addressed.Our knowledge of human microbiome interactions is still in its infancy (Kamada et al., 2013); there may be unexpected long-term complications following FMT, such as obesity (Alang and Kelly, 2015), which need to be further studied.These potential complications should be carefully considered before use of FMT to treat children and young adults.Despite the multitude of reports demonstrating the high efficacy of FMT, it is yet to become a treatment option available to many patients.Clinicians are rightly cautious to offer FMT as an option due to the shortage of RCTs.However, this has led to reports of well-meaning relatives offering stool samples for uncontrolled, self-administered enemas after seeking advice online (Brandt, 2013;Youngster et al., 2014).This is dangerous in many regards, not least that the stool samples are unscreened.Patients treating themselves in this manner are likely to lack understanding of the potential adverse effects and risks associated with FMT (Brandt, 2013).There is a paucity of literature directly comparing multiple routes of FMT and their efficacy; more RCTs are required in order to find the optimal procedure.There is some evidence that treatment routes do not differ significantly (Postigo and Kim, 2012;Youngster et al., 2014), but FMT route should be based on individual patient situations (Owens et al., 2013).Future research should continue to focus on long-term effects of FMT and designing robust RCTs with larger sample sizes to test the efficacy of FMT.Efforts should be made to make these trials blinded, although this may be impractical.FMT remains a highly promising treatment option that appears to be efficacious, safe in the short-term and affordable.
CDI by infusion of faeces via a duodenal tube (17 patients), vancomycin