Systematic Review: Adverse Events of Fecal Microbiota Transplantation

Background Fecal microbiota transplantation (FMT) is a microbiota-based therapy that shows therapeutic potential in recurrent or refractory Clostridium difficile infections and other intestinal or extra-intestinal disorders. Nonetheless, adverse events (AEs) remain a major challenge in the application of FMT. Aim To review the AEs of FMT and to address the concerns of safety during the procedure. Methods Publications were retrieved in the databases of Medline, Embase and Cochrane Library. AEs were classified according to their causality with FMT or their severity. Results A total of 7562 original articles about FMT were identified in this study, 50 of them fulfilled the inclusion criteria. Totally 78 kinds of AEs were revealed enrolled in these 50 selected publications. The total incidence rate of AEs was 28.5%. Among the 42 publications, 5 kinds were definitely and 38 kinds were probably related to FMT. The commonest FMT-attributable AE was abdominal discomfort, which was reported in 19 publications. For upper gastrointestinal routes of FMT, 43.6% (89/204) patients were compromised by FMT-attributable AE, while the incidence dropped to 17.7% (76/430) for lower gastrointestinal routes. In contrast, the incidences of serious adverse events (SAEs) were 2.0% (4/196) and 6.1% (40/659) for upper and lower gastrointestinal routes, respectively. A total of 44 kinds of SAEs occurred in 9.2% patients, including death (3.5%, 38/1089), infection (2.5%, 27/1089), relapse of inflammatory bowel diseases (0.6%, 7/1089) and Clostridium difficile infection (0.9%, 10/1089). Conclusion Consequently, both AEs and SAEs are not rare and should be carefully monitored throughout FMT. However, high quality randomized controlled trials are still needed for the more definite incidence of AEs of FMT.

Fecal microbiota transplantation (FMT), also known as fecal bacteriotherapy or intestinal microbiota transplantation, is defined as the perfusion of treated feces from a healthy donor via the upper or lower gastrointestinal route [13]. About 1700 years ago, Ge Hong, a well-known traditional Chinese medicine doctor, firstly described the use of human fecal suspension by mouth for patients with food poisoning or severe diarrhea [14]. In 1958, Eiseman et al applied FMT to treat antibiotic-associated diarrhea [15]. Since Schwan et al reported the first FMT therapy for CDI in 1983 [16], the application of FMT in CDI has been practiced extensively [17][18][19]. The effective rate of FMT for recurrent or refractory CDI was over 90% [20,21]. Although FMT is still regarded as an investigational agent and requires an investigational new drug (IND) application, the US FDA has already recommended FMT as an alternative therapy for recurrence of CDI after a pulsed vancomycin regimen [22]. Moreover, FMT shows remarkable therapeutic potential in diverse conditions [13,23] including inflammatory bowel diseases (IBD) [24,25], irritable bowel syndrome (IBS) [26][27][28], metabolic diseases [4,29,30], neuropsychiatric conditions [31], autoimmune diseases [32,33], allergic disorders [34,35], and chronic fatigue syndrome [36].
Although patients benefit from FMT, concerns about this emerging strategy remain to be addressed, including long-term outcomes of FMT and the AEs. So far, the exact roles of the gut microbiota in FMT are not yet fully understood. And the AEs that happen during or after FMT still perplex clinicians and fundamental researchers. Hence, we systematically reviewed the AEs of FMT in all related publications aiming to elucidate the causality between FMT and the AEs. Furthermore, the AEs of FMT were divided into different degrees according to the severity and SAEs were emphatically introduced to arouse attention in FMT.

Information Sources and Search Strategy
Electronic databases for literature search included the Medline, Embase, and Cochrane Library. The last search was run on July 2015. The complete string used for the electronic search is shown in Table 1. All the deriving terms were combined by the Boolean operator "OR" to assure the identification of studies regarding FMT. articles in accordance with the inclusion criteria and exclusion criteria. Disagreement was resolved by a third investigator (CHL). Original full-text articles, letters to the editor, abstracts of scientific conferences, case reports and case series which were published between 1913 and 2015 were reviewed. Studies involving the AEs of FMT for human sujects of any age were included into this study. Studies evaluating treatments with cultured bacteria other than human feces, animal studies and non-original reports (reviews, systematic reviews, meta-analyses, editorials, etc) were excluded.

Data Collection and List of Items
Data extraction was conducted according to the above mentioned inclusion and exclusion criteria and cross-checked by the two independent investigators (WSN and XMQ). When publications included patients from a previous study and newly enrolled ones, only the latter were brought into the study. Items of this study were listed as follows: (i) the study characteristics (the first author, the year of publication, the length of follow-up); (ii) the patients (the number, the reason for FMT, the prior therapy); (iii) the relationship between donors and recipients; (iv) the FMT procedure (the patient preparation for FMT, the weight of infused stools, the route of infusion, the number of infusion); (v) the detailed descriptions of AEs (the onset time, the causality between AEs and FMT, how the AEs relieve and the outcomes).

Definition of causality between the AEs and FMT
The relationship between the AEs and FMT were categorized into four types as previously dsecribed with minor modifications: definitely related, probably related, possibly related, and unrelated to FMT [37].
Definitely related. AEs caused by endoscopic procedure during FMT; an event that follows a reasonable temporal sequence from FMT exposure; that follows a known or expected response pattern to the FMT; that is confirmed by stopping the FMT procedure; and that is not explained by any other reasonable hypothesis.
Probably related. An event that follows a reasonable temporal sequence from FMT procedure; that follows a known or expected response pattern to the FMT; that is confirmed by stopping the FMT procedure; and that is unlikely to be explained by the known characteristics of the subject's clinical state or by other interventions.
Possibly related. An event that follows a reasonable temporal sequence from FMT procedure; that follows a known or expected response pattern to FMT; but that could readily have been caused by a number of other factors.

Definition of severity of AEs
Adverse events (AEs). AE is defined as any untoward medical occurrence in a patient after administration of FMT that does not necessarily have a causal relationship with this treatment. Therefore, an AE can be any unfavorable and unintended sign (including an abnormal laboratory finding), symptom, or disease temporally associated with FMT, whether or not related to the FMT [37].
Serious adverse events (SAEs). A SAE is any adverse experience occurring during or after FMT that results in any of the following outcomes: death, life-threatening experience, inpatient hospitalization or prolongation of existing hospitalization, persistent or significant disability or incapacity, congenital anomaly or birth defect, or an important medical event [37].

Included studies
A total of 7562 original articles about FMT were identified. Among them, 534 were selected for further assessment according to the titles. After reviewing the full text articles, 50 of them fulfilled the inclusion criteria among which 16 were case series, 9 were case reports and 4 were randomized controlled trials (Fig 1). The excluded articles were presented in S1 Appendix. Besides, conference abstracts and letters to the editor were included. The included studies were published during the period from 1998 to 2015, with a span of 18 years. The follow-up time after FMT ranged between 2 weeks to 68 months.

Patient characteristics
In the selected 50 publications, a total of 1089 patients were treated with FMT (age range: 1-95 years). Among them, 831 patients were affected by refractory or recurrent CDI, of whom 106 had concomitant IBD; 235 were affected by independent IBD; 1 affected by early onset colitis; 9 affected by antibiotic-associated diarrhea (AAD), of whom 6 had concomitant CDI; and 13 affected by IBS. Overall, 78 kinds of AEs were reported to happen on 310 patients during or after FMT. The majority of AEs were presented as mild symptoms such as abdominal disomfort, diarrhea, transient fever, nausea, vomiting and constipation. Each AE was cited once for one patient since the AE always recorded only once during a multiple FMT in one course of treatment.
Since AEs usually overlapped, we could not obtain the exact total number of patients with AEs in a publication. Hence, we took the number of patients with the most frequent kind of AEs from each publication for further calculating the overall incidence of AEs. Based on the above mentioned statistical principles, the overall incidence of AEs was 28.5% (310/1089). The incidences of AEs in CDI and non-CDI (IBD, AAD, IBS and early onset colitis) were 28.0% (233/831) and 29.8% (77/258) respectively.

Causality between AEs and FMT
Many factors could be involved in the development of AEs, including the individual difference of recipients, donors, methods of administration and regimen of FMT. The causality between AEs (including SAEs) and FMT was analyzed according to the description as above [37]. As a result, AEs were described to be attributable to FMT in 42 publications (Tables 2 and 3). Five kinds of AEs were reported to be definitely related to FMT in 5 publications. Thirty-eight kinds of AEs, probably related to FMT, were reported in 35 articles and were considered as results of temporary systemic immune response to the applied bacteria. In addition, 25 kinds of AEs were reported to be possibly related to FMT in 13 articles. Finally, 38 kinds of AEs were reported to be unrelated to FMT in 22 articles (Table 4).
Donors and AEs. So far, there is no unified standard to screen the stool samples from donors. The following donor screening tests were applied to the donors in the selected 50 publications: viral screenings (hepatitis A virus, hepatitis B virus, hepatitis C virus, Epstein-Barr virus, human immunodeficiency virus, treponema pallidum, and cytomegalovirus), stool tests for Clostridium difficile toxin, and routine bacterial culture for enteric pathogens (Escherichia coli, Salmonella, Shigella, Yersinia, Campylobacter), parasites and ova. However, the donors who were in the latent period of infection could not be excluded by the above screening tests and thus these donors might contribute to the development of infectious AEs.
In addition, individual differences of donors may also lead to AEs. For example, one patient (a 1-year-old girl) developed fever, vomiting and tachycardia after receiving the fecal transplant from her brother, while she well tolerated the transplant from her niece [38]. For the infrequent infection of cytomegalovirus in FMT receipt, it may be caused by the donors with young age [39]. Finally unrecognized pathogens that were carried by donors might induce AEs [40].
Related donors (family members) for FMT were reported in 11 publications. Unrelated donors were reported in 9 publications.
Preparation and route of administration and AEs. For FMT via upper gastrointestinal routes, the recipients were prepared in fasting condition. For FMT via lower gastrointestinal routes, bowel lavage and/or antibiotics were given to the recipients before FMT. However, no association of AEs with the preparation of FMT was found in the 50 publications. Of note, in Bota Cui's report, the recipients who took metoclopramide before FMT manifested fewer AEs, suggesting metoclopramide might potentially help avoid the AEs [41].
The routes of administration are listed as follows according to the frequency that they were used: colonoscopy (26 publications), retention enema (8 publications), nasogastric tube (6 publications), nasojejunal tube (5 publications), gastroscopy (2 publications), sigmoidoscopy (1 publication), nasoduodenal tube (4 publications), enteroscopy (1 publication),  (Tables 2 and 3). Among the 78 kinds of AEs, 5 kinds were definitely related to endoscopic manipulation. Of these, nasal stuffiness, sore throat, rhinorrhea and upper gastrointestinal hemorrhage happened on a total of 8 patients in 4 publications, which were attributable to upper gastrointestinal routes administration. It seems that patients are likely to be injured by invasive endoscope procedures for upper gastrointestinal routes of FMT administration.
The commonest attributable AE was abdominal discomfort for both upper and lower gastrointestinal routes, including abdominal pain, increased stool frequency, flatulence, bloating, cramps and other nonspecific symptoms. For upper gastrointestinal routes of administration in 12 publications, 29.9% (61/204) patients (in 9 publications) were reported to suffer abdominal discomfort after FMT. For lower gastrointestinal routes in 22 publications, 13.0% (56/430) patients (in 10 publications) developed abdominal discomfort after FMT. The upper gastrointestinal routes were therefore more likely to develop abdominal discomfort compared with lower gastrointestinal. The second commonest attributable AE was transient fever which was happened on 3.4% (7/204) and 2.8% (12/430) patients for upper and lower gastrointestinal routes of FMT administration, respectively ( Table 2).

Classification of AEs based on severity
Mild to moderate AEs such as abdominal pain, abdominal cramping, flatulence, increased stool frequency, constipation, vomiting, belching, fever and transient increase of C-reactive protein (CRP) were reported in most of the selected 50 publications and ususally did not cause critical clinical outcome. Hence, we paid emphatic attention to SAEs and listed 44 kinds of SAEs that were reported in 27 publications (Table 5), of which 18 kinds were associated with FMT procedure. Totally 9.2% (100/1089) patients developed SAEs. The incidences of SAEs were 2.0% (4/196) and 6.1% (40/659) for upper and lower gastrointestinal routes respectively, which suggest that lower gastrointestinal routes of FMT administration induce more SAEs compared with upper routes. The FMT-attributable (definitely, probably and possibly related) SAEs included death, pathogen infections, IBD flare, auto-immune diseases, and FMT procedure related injury, etc, while the FMT unrelated SAEs covered death or hospitalization caused by underlying conditions. The commonest SAEs were death, severe infections and relapse of CDI and IBD.
As the most devastating SAEs, death happened on 38 patients in 10 publications (Table 6) and the mortality rate was 3.5% (38/1089). Of these deaths, 1 was definitely related, 2 were possibly related, and 35 were unrelated to FMT. The death that was definitely related to FMT was caused by aspiration during sedation of colonoscopy [42]. The other two deaths were associated with infections which might be outcomes of either FMT procedures or underlying immunocompromised status. Except for the above 3 patients, no evidence supported the notion that   Twenty-seven patients were reported to be hospitalized or die for infection in 12 publications (CDI was not included) ( Table 7). The incidence of severe infection was 2.5% (27/1089). Among the 27 cases of severe infection, 8 cases were probably or possibly related to FMT and the remaining 19 cases were unrelated to FMT.Out of the 8 cases of severe infection, 2 were viral infection, 2 were bacteriemia infection, and the remaining 4 were infection of unknown pathogens. The pathogens that caused the 2 cases of viral infection were cytomegalovirus [39] and norovirus [43] respectively and the pathogens that caused the 2 cases of bacteriemia infection were Escherichia coli, Proteus mirabilis, Citrobacter koseri, and Enterococcus faecium [44,45]. The cytomegalovirus infection happened after home FMT and was suspected to be probably related to a child donor without strict donor screening. The noroviurs infection was speculated to be probably related to environmental contamination by an endoscopy suite employee. IBD flare happened on 7 patients (4 UC and 3 CD) post-FMT in the 50 selected publications (Table 8) and its incidence reached 0.6% (7/1089). Most patients who suffered from IBD flare were those with low immunity, such as kids, aged people and immunocompromised ones. Of note, patients with IBD flare were administered FMT via lower gastrointestinal routes, such as colonoscopy, sigmoidoscopy or enema. Therefore, IBD flare should arouse attention when FMT was administered via lower gastrointestinal routes. So far, the association between donors and IBD flare has not been defined, for some unrecognized pathogens from donors' stool might also lead to IBD flare.
Although unrelated to FMT, CDI relapse was still another unignorable SAE, the incidence of which reached 0.9% (10/1089) in the selected 50 publications (Table 7).

Discussion
The human gastrointestinal tract harbors the largest number of microbes in the human body, which is referred to as the gut microbiota. Perturbations in the gut microbiota have been associated with conditions as diverse as gastrointestinal diseases and even systemic disorders [46]. As a microbiota-targeted therapy, FMT shows promise in controlling bacteria-associated disorders, especially recurrent or refractory CDI. Nevertheless, with the growing application of FMT, safety evaluation for FMT is increasingly urgent and potential risks of FMT must be paid attention to. Previous studies focus on the effectiveness of FMT treatment on CDI [47], IBD [25] and other digestive and nondigestive disorders [23,48], lacking of emphasizing AEs of FMT. Landy et al [48] reviewed publications about FMT therapy for gastrointestinal diseases that were published before 2011 and did not find any reports regarding FMT related AEs. A more recent systematic review about FMT for IBD treatment (2012) summarized AEs that were reported in just three citations [25]. Lately, two randomized controlled trials of FMT for CDI [49] and UC [50] reported a high incidence of AEs, suggesting that under-reporting AEs     Adverse Events of FMT of FMT may exist in many other cases. Therefore, there is an urgent need to systematically review and analyze the characteristics of AEs of FMT to evaluate the safety of the procedure.
Here, we selected 50 publications in which AEs of FMT were reported. Totally 78 kinds of AEs happened on 310 patients following FMT treatment. The AEs of FMT were divided into two major categories, namely related and unrelated to FMT. Our analytical results showed that the AEs related to FMT (including definitely, probably and possibly related) accounted for a larger proportion than the AEs unrelated to FMT. Moreover, we found that SAEs related to FMT such as death, viral and bacterial infections, transient relapse of IBD, were not rare and therefore deserved attention and consideration in the procedure of FMT.
Human Microbiome Project (HMP) has sampled the microbiome of many people to get a better idea of variability, and how microbes work together in complex communities. HMP implied that because the microbiome is more varied than the genome, and easier to modify, it gives a more logical starting point for individual treatments [51]. As observed by Angelberger et al, most AEs may be caused by the applied bacteria into the gut [52], which, in our opinion, could be further supported by the notion that most patients receiving FMT were under the conditions of impaired intestinal mucosal barrier and severe inflammation. In a recent observational cohort study of FMT for treating recurrent CDI, mild transient fever happened on two patients receiving FMT [53]. The authors speculated that FMT itself rather than glycerol used in the frozen preparations caused the AE, which was in agreement with Angelberger's opinion. Though all donors underwent blood and stool tests before FMT as recommended by FDA, some unrecognized infective agents might cause AEs of FMT. Since variability in donor microbiotas exists, it is necessary to establish a better donor screening methodologies. Moreover, the inclusion criteria of FMT donors for recurrent or refractory CDI have been established, but an agreement of the inclusion criteria of donors for IBD, IBS, metabolic diseases, and other extra-CDI have not been reached, which might engender potential risks for AEs of FMT [54,55]. A recent placebo-controlled randomized trial demonstrated that FMT with the donation of two donors were more effective but with milder AEs than the other donors. Sequencing analysis of the microbiota was conducted for the two donors and they turned out to have similar taxonomic profiles [56]. Previous study also demonstrated that genetic variation in immune genes could result in variability in susceptibility to enteric infection in germfree mice [57]. Thus, genetic variation may paly a key role in variability in microbiota composition, susceptibility to enteric infection, response to FMT treatments, and even AEs. The emerging metagenomics, genetic and microbiota screening methodologies could be useful for identifying better donor sources for FMT therapies in the future [58].
Route of fecal infusion is another concern in FMT that may lead to AEs. Lower gastrointestinal routes, including colonoscopy, sigmoidoscopy, and retention enema, were more widely used than upper gastrointestinal routes. We found that the patients who received FMT treatment via upper gastrointestinal routes were more likely to develop AEs than those who received FMT treatment via lower gastrointestinal routes (43.9% vs. 20.6%). To avoid injury associated AEs during endoscopic process, noninvasive and patient-acceptable routes can be chosen for FMT treatment. Actually, a recent pilot study in which frozen capsules FMT was administered orally for patients with recurrent CDI demonstrated a high incidence of diarrhea resolution (overall 90%) but few mild AEs such as abdominal cramping and bloating [59]. Therefore, capsules would potentially make FMT procedure safer by avoiding procedurerelated complications as well as availability for long-term usage. Furthermore, the encapsulated FMT can be accessible to a wider range of patients, especially to those who cannot withstand the endoscopic procedure. Hence, large randomized controlled studies for the safety and therapeutic efficacy of encapsulated FMT are warranted.
Though this systematic review provides a handful of valuable messages for clinical application of FMT, some limitations need to be addressed. First, the incidences of AEs might be underestimated. On one hand, since AEs usually overlapped, we took the number of patients with the most frequent kind of AE from a publication for further calculating the overall incidence of AEs. On the other hand, transient or mild AEs were sometimes ignored by researchers, resulting in the missing data of AE occurrence. Secondly, some potential confounding factors such as the health conditions of the donors, the time span from FMT exposure to the onset of the AEs and the outcomes of AEs could have substantial impacts on the classification of AEs. Thirdly, there was subjective nature in the classification of AEs. In most instances, it is impossible for us to obtain the original data of the publications we selected. Therefore we categorized AEs according to the authors' subjective description or FDA general definitions of AEs for drugs or therapy [37].

Conclusion
Though FMT was validated to be a beneficial therapeutic strategy, we should pay enough attention to AEs of FMT. In order to prevent or treat AEs during or after FMT, more clinical trials and fundamental research are urged to elucidate the exact mechanism of how FMT causes AEs and set up a guideline on how to handle FMT-related AEs in different situations.