Clostridium difficile infection (CDI) represents a major clinical and economic burden on healthcare systems . The management of treatment failure and recurrent CDI poses a significant challenge, which drives the cost of healthcare because of their associated morbidity and mortality . Recurrent CDI is reported in 10-30% of patients after initial treatment, with recurrence approaching 60% after the third episode of CDI . Restoration of gut diversity through fecal microbiota transplantation (FMT) has emerged as an effective treatment for CDI-associated diarrhea in patients with recurrent disease after initial antibiotic therapy . A variety of approaches to FMT have been studied, including outcomes associated with donor selection, fecal samples and various delivery techniques, including upper or lower gastrointestinal routes . We herein review current knowledge on FMT and in particular its relationship to donor selection, use of frozen or fresh fecal samples, and delivery systems. We also review the cost-effectiveness of FMT by various delivery systems.
Materials and methods
In this narrative review, studies were searched in electronic databases according to article titles, abstract contents, and relevance in the field of FMT through the end of April 2018. The databases used in this review included PubMed, MEDLINE, Google Scholar, and Cochrane Databases. The main terms applied were Clostridium difficile, Clostridium difficile infection, fecal microbiota transplantation, fecal transplantation, fecal bacteriotherapy, intestinal microbiota transplantation, floral reconstitution, infectious diarrhea, donor selection, related donor, fresh stool, frozen stool, capsules, colonoscopy, enema, and management. We also manually searched references to identify additional relevant studies. Articles published in languages other than English were excluded.
Related versus unrelated donors
Early FMT practices offered patients the opportunity to select relatives as a source of stool for transplantation [6,7]. In the event the patient could not identify a donor, a healthy volunteer was chosen. The selected volunteers were often other hospital patients, medical students, or residents [6-9]. Early reviews noted the preference for related donors [10-12] due to their genetic similarity and shared environment, both of which are known to influence gut diversity . A shared environment also reduced the risk of transferring infectious agents to the patient [10,14,15]. However, it was noted that there was no rationale for excluding healthy volunteers , and subsequent systematic reviews [4,17] and meta-analysis  did not report a statistically significant difference in outcomes based on donor selection. Recently, with the use of stool banks, the use of unrelated donors has increased significantly [19,20].
There are limited data on direct comparisons of related and unrelated donors. Inferences from randomized controlled trials (RCTs) are also limited, as those trials were not structured to investigate the donor-patient relationship and their outcomes (Table 1) [5,23-29]. Retrospective studies from the United States  and Finland  were limited by small sample sizes and found no significant difference in outcomes after eight weeks and one year, respectively (100% vs. 78.6% resolution; 70% vs. 92% resolution, related and unrelated donors respectively). Six RCTs used exclusively unrelated or related donors, or a mixture [5,23,24,26-28], while two RCTs used almost exclusively related donors [25,29]. The rate of resolution in 7 of 8 RCTs ranged from 87-96.2%; one study that utilized a single enema infusion reported a resolution of 43.8%. When compared to the resolution rate of other studies that utilized a single enema infusion, it did not appear to be an outlier (47.9%  and 51.6% ). Overall, current evidence indicates that the success of FMT does not depend on the donor-patient relationship; however, larger studies should be designed to explore this area in greater detail.
Table 1 Summary of randomized controlled trials of fecal microbiota transplantation for recurrent Clostridium difficile infection
The advantages of using unrelated donors stem from their easy access and availability from stool banks. It can take weeks to screen a donor , thus time and capital are saved by treating patients with donations from an established source. Furthermore, accessibility to FMT is increased by removing the burden of donor screening and processing from healthcare providers [21,32,33]. However, ethical concerns regarding the privacy and ownership of stool have been raised [34,35]. It is unclear how these issues affect patients or stool banks.
Additional benefits of using unrelated donors include the avoidance of unpleasing or difficult conversations with related donors [36,38,39]. A survey of 183 FMT-naïve patients found that 28% of patients found selecting their own donor too unappealing to consider FMT as a treatment . On the other hand, experience from another institution found that, when given a choice after switching to using frozen stool from unrelated donors, all patients preferred the unrelated donor option . These data suggests that some patients struggle with discussing FMT with related donors and may prefer using unrelated donors.
A third option, the autologous donor, involves stool taken from an individual when their disease is in remission [40,41]. Stool can be used as treatment for that individual when the disease relapses. Although implementing this form of personalized medicine has been made easier by the creation of stool banks, its practicality and effectiveness have yet to be evaluated. Regardless of the donor utilized, the recommendation from consensus and working groups around the world emphasizes that every donor should be carefully and thoroughly screened before FMT is performed [42-46].
Frozen versus fresh fecal preparation
Another significant evolution in the practice of fecal transplantation has been the use of frozen fecal matter. The earliest records can be traced to a report in 1998 by a group in Norway [47,48]. Frozen cultures had been used 9 years previously by Tvede and Rask-Maden to treat 5 patients, but their transplant material was referred to as “synthetic stool” as it was a limited combination of cultured bacterial strains . The overall efficacy of the studies performed in Norway ranged from 83-100% [17,47,48].
Since then, most studies involving frozen fecal matter have reported an overall CDI cure rate between 81-100% [5,21,22,24,28,31,50-54] (Table 2). Two retrospective analyses [21,22] and three RCTs [5,24,28] directly compared fresh and frozen stool. All five studies reported primary outcomes that did not differ between fresh and frozen stool. No significant differences were noted, despite the range in the storage time of frozen fecal matter (1 week to 6 months). One small case series involving 3 patients  and a pilot clinical trial involving 7 patients who received frozen stool  reported lower resolution rates (67% and 71.4% respectively). Besides the significant limitation of small sample sizes, the case series involved patients suffering from severe CDI, one who refused a second transplant and ultimately passed away from fulminant disease . The pilot trial treated patients suffering an initial episode of CDI, some of whom were concomitantly or recently on antibiotics for other indications, with a protocol that did not include a bowel lavage . These differences may have contributed to a lower resolution rate. Regardless, current evidence appears to indicate similar efficacy among frozen and fresh fecal preparations.
Table 2 Summary of studies of fecal microbiota transplant utilizing frozen stool
A central question related to the use of frozen stool is the viability of the microbiome over time. Costello et al tested bacterial viability after six months of storage . The study showed that the microbiome remained largely unchanged after six months. CDI resolution has been achieved following 10 months of storage, while other studies have reported successful outcomes of FMT after six months [28,50] and up to 1 year of storage . Similar results have been reported in mice models, though a decline in quantity and diversity was observed after seven months of storage . Given the benefits of frozen fecal transplant material and evidence to support its durability, clinicians trained in the use of FMT may be more inclined to use frozen preparations.
In 2015, an innovation in fecal preparation involved the use of freeze-dried stool . Also known as lyophilized stool, a fresh suspension is transferred to a vacuum drier that produces a powdered product, which is then encapsulated. The motivation behind the creation of lyophilized stool was to develop a more concentrated and conveniently stored product with the hope that patients could utilize it at home. Various studies, including a case report , a retrospective study , pragmatic cohort study  and an RCT , have reported on the use of lyophilized stool with resolution rates ranging from 78-100%. The RCT by Jiang et al found lyophilized stool to be statistically less effective than fresh stool (78% vs. 100%, P=0.022) but equally effective as frozen stool (78% vs. 83%, P=0.255) in resolving recurrent CDI . Further studies are needed to determine the efficacy and safety of lyophilized stool.
Until 1990, enema was the method of choice for fecal transplant. Since then, various routes have been tested; however, the most optimal route of administration remains unclear. Delivery can be broadly classified into upper and lower gastrointestinal routes. Common upper gastrointestinal routes include esophagogastroduodenoscopy via enteric tubes (nasogastric, nasoduodenal, and nasojejunal tubes), and oral capsules [32,52,61-64]. Common lower gastrointestinal methods include colonoscopy and enemas [21,65-67]. There are benefits and pitfalls for each modality and physicians often choose one that best fits a combination of their patients’ needs, their expertise and availability (Table 3).
Table 3 Summary of studies of fecal microbiota transplantation delivered via oral capsules
The major advantage that colonoscopy offers over other modalities is the ability to visualize the entire colon [40,68-72]. It also enables reliable delivery of stool to affected segments of the bowel [71,73,74] and possibly better retention of stool . This can be particularly important, since it has been observed that the proximal colon has more severe pseudomembranous changes . Furthermore, colonoscopy can deliver larger amounts of stool per transplant procedure [40,71,74,76], associated with higher success rates . Bowel preparation before the procedure is suggested to increase the likelihood of resolution of CDI by decreasing the number of spores and residual organisms . However, there is a risk of bowel perforation, and those who are severely ill may not tolerate the procedure or anesthesia [69,77,78,79].
Enemas are less invasive, easier to perform, and relatively less expensive . The ease and simplicity of an enema allows it to be used in the hospital, ambulatory offices, and even at home [81,82]. However, there are greater concerns regarding retention of stool, as well as the risk that stool would not reach beyond the splenic flexure , which may require multiple infusions to achieve efficacy . Greater facility time and personnel are sometimes required to maintain and rotate the patient between various positions to compensate for poor retention . Additionally, patients with poor sphincter tone or issues with incontinence may augment concerns about retention .
Upper gastrointestinal routes are typically faster, less expensive and better tolerated compared to colonoscopy, though not as esthetically pleasing to some patients [12,15,16,61,68,70]. Typically, less stool sample is used when upper gastrointestinal routes are used, possibly to avoid regurgitation as well as better retention rates [61,84]. On the other hand, this has also been suggested as a reason why some transplants fail . Common concerns include degradation by the gastric milieu, aspiration, hemorrhage and perforation of the upper gastrointestinal system [68,85-87]. This was supported by Wang et al who demonstrated in a systematic review that the rate of adverse events was more common when upper compared to lower gastrointestinal routes were used (43.6% vs. 17.7%) .
The most recently developed mode of stool delivery is in the form of oral capsules [88,89]. It involves the delivery of stool mixed with a cryoprotectant, most commonly glycerol, and double- or triple-encapsulated to protect the stool from stomach acidity. Capsules are minimally invasive, convenient, and eliminate the risk of perforation by endoscopic procedures. Additionally, capsules are more esthetically pleasing, as patients have shown a preference for this mode of delivery over others [27,37].
Studies comparing the different routes of FMT have had mixed results, most likely due to sample size and research design . Postigo et al compared the efficacy of upper and lower routes of delivery. Using a pooled analysis of 182 patients, the study did not show a significant difference between lower gastrointestinal and upper gastrointestinal delivery of FMT 95% (95% confidence interval [CI] 92-97%) vs. 88% (95%CI 82-94%), respectively (P=0.162) . An observational follow-up study by Gundacker et al found that FMT by nasogastric tube (NGT) was less effective than colonoscopy . Subgroup analysis revealed that NGT was equally effective for patients who were severely ill (Charleston comorbidity index [CCI]>5) or moderately ill (CCI<5), although there was a trend towards increasing effectiveness with colonoscopy for severely ill patients. However, a small randomized study of 20 patients by Youngster et al found colonoscopy and NGT to be equally successful . A larger study by Kao et al involving 116 patients (57 patients randomized to the capsule group and 59 to the colonoscopy group) found that the use of capsules was non-inferior to colonoscopy . The study showed that 96.2% of patients achieved prevention of recurrent CDI after a single treatment in both groups (capsule=51/53, colonoscopy=50/52; difference 0%; 1-sided 95%CI -6.1% to infinity; P<0.001). Evidence from these comparisons largely indicates that upper gastrointestinal methods of delivery can be equally effective as lower gastrointestinal methods. Furthermore, the high efficacy of FMT capsules, combined with their convenience, safety and esthetics, make them an attractive option for many patients.
Cost-effectiveness of FMT
CDI places a large burden on the healthcare system and makes it imperative to develop guidelines that emphasize cost-effective therapies [93-97]. A meta-analysis estimated the total financial burden of inpatient management of CDI in the United States to be $6.3 billion in 2015 . In Europe, the economic burden of CDI was estimated to be roughly €3 billion , with costs per episode ranging between €5798 and €11,202 . Given the effectiveness of FMT in the management of recurrent CDI, several studies have evaluated its cost-effectiveness [99-104].
The first cost-effective analysis compared three types of FMT (colonoscopy, duodenal infusion, and enema) and standard antibiotic therapy in the US. The analysis began with the first episode of recurrence and utilized contemporary guidelines to model two additional episodes of recurrence. At a willingness-to-pay threshold of $50,000/quality-adjusted life year (QALY), they found that FMT via colonoscopy was cost-effective compared to vancomycin and dominant (both cheaper and more effective) compared to the other therapies . Of the several assumptions that went into the model, the choice to model FMT with only a single infusion of stool per treatment should be noted. If multiple infusions were permitted, the resolution rates for FMT via enema or duodenal infusion might have been higher (0.813 vs. 0.926 and 0.815 vs. 0.94, respectively), which may have significantly altered the conclusions of the study.
Two other cost-effectiveness analyses have led to mixed conclusions. A decision-analytic model involving Markov processes compared colonoscopy, enema, NGT and standard antibiotic therapy, beginning with the first recurrence of CDI. The analysis, from the perspective of the ministry of health in Ontario, Canada, concluded that FMT via colonoscopy dominated all other strategies. Sensitivity analysis revealed it to be dominant in 87% of simulations. FMT via enema was preferred if colonoscopy was not available . Another decision-analytic model, from France, compared various methods of FMT (colonoscopy, duodenal infusion, enema), vancomycin and fidaxomicin, for the second recurrence of CDI. Using the WHO’s commission on macroeconomics and health to set the willingness-to-pay threshold at $32,000, they found FMT via enema to be highly cost-effective (incremental cost-effectiveness ratio [ICER] $18,092/QALY) compared to vancomycin. Although FMT via colonoscopy was still cost-effective according to WHO guidelines, it was not as cost-effective as FMT via enema (ICER $73,653/QALY) .
Other cost-effectiveness studies for recurrent CDI have limited their scope to comparing specific methods. One study compared FMT via colonoscopy to vancomycin for the third recurrence of CDI. They found that FMT dominated vancomycin unless the cure rate with vancomycin was greater than 90% or the costs of FMT exceeded $3,205 (2011 USD) . Another study used Markov’s model to compare two methods of FMT (nasoduodenal infusion and colonoscopy) and vancomycin for the first recurrence of CDI. From the perspective of the healthcare system in Australia, the study found both methods of FMT to be dominant over vancomycin, with an estimated savings of over AU$1.37 million (2015 Australian dollars) per year. While no significant differences were reported between the various methods of FMT, the study did not incorporate the risks associated with nasoduodenal infusion into their model .
Additionally, a decision-analytic model by Varier et al compared FMT via colonoscopy to metronidazole and vancomycin, from the perspective of a third-party payer. Given the lack of published data on the effectiveness of FMT for initial CDI, the authors assumed that the procedure was as effective as it was for recurrent CDI . At a willingness-to-pay threshold of $100,000 (2011 US dollars), the base case of FMT was not cost-effective when compared to metronidazole (ICER $124,964/QALY) but was dominant over vancomycin. FMT would have dominated metronidazole if the cost of the procedure was less than $584, the cost of metronidazole was over $559, or the cure rate with metronidazole was less than 71% . Recently, a small RCT in Norway showed that, when FMT was used as primary therapy for CDI, overall response to treatment was achieved in 7 (n=9) patients in the FMT group (78%; 95%CI 40-97), compared with 5 (n=11) in the metronidazole group (45%; 95%CI 17-77) (P=0.20) . However, the study did not show statistical significance. Longitudinal studies are needed to define the role of FMT in the management of primary CDI. Currently, it appears that FMT may not be a cost-effective strategy for initial cases of CDI.
FMT is an effective and robust strategy for treating recurrent CDI. Our narrative review has shown that the success of FMT does not depend on the donor-patient relationship. To this end, the use of unrelated donor stool is becoming widely accepted with the creation of stool banks. However, further studies are needed to clarify storage requirements and the optimal freezing time before declining microbial function and diversity. Furthermore, several FMT techniques are used in the process of fecal transplantation. Based upon current data, we believe colonoscopy to be the most effective strategy. However, capsule FMT offers patients a more convenient and esthetically pleasing option. Cost-effectiveness models comparing these various approaches support the use of fecal transplant using colonoscopy over antibiotic therapy for treating recurrent CDI. However, there remains a knowledge gap regarding the cost-effectiveness of capsule FMT.