Search results and study characteristics

The search strategy identified 978 citations, of which 318 were duplicates. After title and abstract screening, the full texts of 99 relevant citations were selected for a thorough analysis. Thereafter, 78 citations were excluded for the following reasons: (1) the full text of 6 citations was not available; (2) 4 citations were not RCTs; (3) the control group in 12 studies did not receive placebo or standard treatment; (4) AEs of 13 citations were not clearly described; (5) 42 were conference abstracts; and (6) the experimental group was autogenous IMT in 1 citation, resulting in a total of 21 studies for the qualitative synthesis. Then, after a sensitivity analysis, we removed a study that was considered the main source of heterogeneity. Finally, 20 articles [12,13,14, 23, 24, 33,34,35,36, 38,39,40,41,42,43,44,45,46,47,48] were included for meta-analysis (Fig. 1), involving 1132 patients: 603 in the IMT group and 529 in the control group. The details and characteristics of the included studies are presented in Table 1. All of these studies were RCTs from published from 2015 to 2020, five of which were open-label RCTs, one of which was a single-blind RCT and fifteen of which were double-blind RCTs.

Table 1 Basic characteristics of the included articlesBias risk assessment of the included RCTs

According to the guidelines of the Cochrane intervention system evaluation manual, 13 studies had a low risk of bias [12,13,14, 23, 24, 36, 39, 41, 44, 45, 47, 48]. Five of the studies were considered high risk because the blind method was not used [33, 37, 38, 42, 43]. One trial was considered high risk because the method was blinded only to the investigator [40], not to the subjects, and the reason for the loss of follow-up was not stated. Another study also did not state the reason for loss to follow-up [46]. In one study [34], the Data Monitoring and Safety Committee advised that the trial should be discontinued, as the treatments were futile. In addition, another study was terminated early due to an interim futility analysis and considered to have an unclear risk of bias because the allocation concealment was not explained [35]. Details of the risk-of-bias assessment by domain for each trial are shown in Fig. 2.

Fig. 2

a Risk of bias graph. b Risk of bias summary. “ + ” indicates study meets criteria. “?” indicates unclear if study meets criteria. “-” indicates study not meets criteria

Meta-analysis of the AEs of IMT

The characteristics of the participants and the number of patients with SAEs and CAEs are listed in Table 2. A total of 1132 patients were included in this study: 603 in the IMT group and 529 in the control group. The male-to-female ratio was 248:345 in the IMT group and 240:279 in the control group. The mean age of participants in all studies ranged from 33 to 75.7 years.

Table 2 The characteristics of the participants in RCTs

The total number of patients with SAEs in the IMT group was 28, but only 10 SAEs were definitely/probably/possibly related to IMT. IMT-related SAEs occurred mainly in 5 studies [34, 39, 41,42,43] (Table 3). There were no deaths in the IMT arm. Meta-analysis of IMT-related SAEs in these 21 studies found that heterogeneity existed in the results. We conducted a sensitivity analysis of the 21 studies and found that one study (Bajaj [37]) had a great impact on heterogeneity (Fig. 3a), so it was removed considering that this study was the main source of heterogeneity. Finally, it was concluded that no significant difference was found in the incidence of SAEs between the IMT group and the control group. The pooled RR for the IMT group compared with the control group was 1.36 (95% CI 0.56–3.31, P = 0.50) and evidence showed that there was no heterogeneity between these 20 studies (x2 = 1.87, df = 4, P = 0.76, I2 = 0%) (Fig. 3b). Thus, the fixed effect model was adopted.

Fig. 3

a Sensitivity analysis of the 21 studies related to SAEs. b SAEs of IMT group versus control group

Only 7 of the 20 studies reported the number of patients with CAEs [12, 13, 24, 35, 39, 42, 47], including 105 in the IMT group and 94 in the control group. There was no significant difference in the incidence of AEs between the IMT group and the control group (RR = 1.06, 95% CI  0.921–1.23, P = 0.43). There was no heterogeneity among the 7 articles (× 2 = 7.09, df = 6, P = 0.31, I2 = 15%) (Fig. 4a). Therefore, the fixed effect model was implemented. We also conducted a sensitivity analysis on the results and found no studies that had a large impact on heterogeneity (Fig. 1c).

Fig. 4

a Sensitivity analysis of the 7 studies related to CAEs. b CAEs of IMT group versus control group

Different diseases and AEs of IMT

Of the RCT studies on IMT included in this analysis, namely, 4 studies on rCDI, 5 studies on ulcerative colitis (UC), 5 studies on IBS, and 2 studies on obesity, there was only one study on other related diseases, such as hepatic encephalopathy, systemic sclerosis, multidrug-resistant Enterobacteriaceae, alcohol use disorder, and constipation (Table 2).

As shown in Table 2, a total of 177 rCDI patients were included in the RCTs analysed in this study, 82 of whom received IMT treatment, and only one patient had an SAE with septic clinical manifestations, including fever, convulsions, vomiting, and diarrhoea. No SAEs were observed in rCDI patients in the control group. In the four RCT studies on rCDI, IMT patients were most likely to experience AEs, including diarrhoea, abdominal pain, abdominal distention, nausea, vomiting, and constipation, but most of the AEs lasted for a short time and resolved spontaneously. Hota et al. found that abdominal pain and abdominal distention were equally common in the IMT and vancomycin groups, but abdominal pain, abdominal distention, mucoid stools, and foul-smelling stools were more common in the vancomycin group at a later stage [38]. In the RCT study by Kelly et al., it was found that autologous IMT was more prone to cause chills than donor IMT, while there was no significant difference in the incidence of other AEs [36]. After 8 weeks of follow-up after IMT treatment, Hvas et al. found that one patient had small bowel bacterial overgrowth after primary IMT and that there were no statistically significant changes in patients' body weight, plasma albumin, or haemoglobin [43]. Most RCT studies of IMT for rCDI did not find any recurrence or death during follow-up.

Of the enrolled UC patients, 171 received IMT, and 8 developed IMT-related SAEs, such as varying degrees of infection and disease progression. However, there was no significant difference in the incidence of SAEs between the IMT group and the control group (RR = 1.53, 95% CI 0.52–4.51, P = 0.45) (Fig. 5a). No individual donor or donor batch was significantly associated with the primary outcome or SAEs, although the study was not powered to evaluate this possibility [39]. Although other SAEs, such as intestinal perforation and cytomegalovirus infection, were found during the follow-up of these studies, they are not considered to be related to IMT [35]. Similarly, data collected in this study showed no significant difference in the number of CAEs associated with IMT treatment of UC. Paramsothy et al. showed that 32 of 41 IMT patients (78%) and 33 of 40 placebo patients (83%) had at least one AE during 8 weeks of IMT treatment. There was no significant difference in the number or type of AEs between the two groups. The most common AE was self-limiting gastrointestinal disorders [39]. In addition, Costello et al. found that three participants developed new anaemia (control, 2; IMT, 1), 2 had a mild elevation in alkaline phosphatase (control, 0; IMT, 2), and 4 had mild elevations of alanine aminotransferase (control, 3; IMT, 1). However, there were no significant differences between the IMT and control groups [41].

Fig. 5

a SAEs of IMT in patients with UC. b CAEs of IMT in patients with IBS

Huttner et al. conducted a study to evaluate whether oral antibiotics followed by IMT can eradicate intestinal carriage with the β-lactamase Enterobacteriaceae and carbapenemase-producing Enterobacteriaceae. Among the 21 patients in the IMT group, 19 (90%) experienced at least one CAE (overall, 104 CAEs). There were four SAEs, of which only one was possibly related to IMT (a patient with known liver cirrhosis and recurrent episodes of hepatic encephalopathy hospitalized two weeks after FMT for an episode of encephalopathy) [42].

In the included IBS studies, there were no SAEs in either the IMT group or the control group [12,13,14, 46, 47]. There was no significant difference in the incidence of CAEs in IBS studies (RR = 1.11, 95% CI 0.79–1.56, P = 0.54) (Fig. 5b). CAEs mainly include diarrhoea, abdominal pain, abdominal distention, constipation, and other gastrointestinal symptoms, among which the most common AE is diarrhoea [12], but most of these cases are self-limited. Some patients may experience transient fever. Dizziness and nausea in some patients may be related to the medication and instrumentation used during colonoscopy [13]. No IMT-related AEs were observed during follow-up after IMT treatment. In a study of IMT for slow-transit constipation included in this review, no SAEs were found. The CAEs reported were mainly in the IMT group (IMT, 50; control, 4), including exhaust, nausea, abdominal pain, and diarrhoea, which were transient. Since IMT was administered through the nasointestinal tube in this study, some AEs, such as nausea and dyspnoea, were considered to be related to the IMT delivery method [40].No SAEs were found in the IMT group in the RCT studies of IMT treatment for extraintestinal diseases such as hepatic encephalopathy, systemic sclerosis, alcohol use disorder, and obesity. In the IMT group of the hepatic encephalopathy study, one patient at day 84 post-IMT with acute kidney injury responded within 24 h to intravenous hydration and one at day 115 due to chest pain that was ruled out for an acute cardiac event. However, both cases were judged to be unrelated to IMT [37]. In the systemic sclerosis study, patients in the FMT group reported more postinterventional CAEs than the placebo controls, but all the CAEs were regarded as mild and transient, including abdominal bloating, diarrhoea, nausea, and constipation. There was one duodenal perforation in the control group that was related to endoscopy, not placebo [24]. In the obesity studies, most of the CAEs in the IMT group and the control group were mild abdominal pain and diarrhoea, with no significant difference between the two groups [23, 48].

Fresh or frozen faecal sample and AEs

Six [13, 33,34,35,36, 38] of the 20 studies used fresh faecal samples, and 13 [12, 14, 23, 24, 39,40,41,42,43,44,45,46,47,48] used frozen stool samples for IMT. The occurrence of SAEs (RR = 1.46, 95% CI  0.26–8.25, P = 0.67) and CAEs (RR = 0.83, 95% CI 0.24–2.89, P = 0.76) was not different between the IMT group and the control group when IMT group patients used fresh faecal samples. When frozen faecal samples were used for IMT, no difference was observed between the IMT group and the control group in the incidence of SAEs (RR = 1.32, 95% CI  0.47–3.73, P = 0.60) and CAEs (RR = 1.06, 95% CI  0.90–1.26, P = 0.46). The incidence of SAEs and CAEs was not different between faecal sample type subgroups (P(SAE) = 0.92, P(CAE) = 0.69). (Fig. 6, Table 4).

Fig. 6

a SAEs of different types of fecal sample. b CAEs of different types of fecal sample

Table 4 AEs of IMT by subgroupsIMT delivery methods and AEs

To observe the relations between delivery methods and AEs, the delivery methods in these studies were divided into upper gastrointestinal (UGI) delivery (via a nasoduodenal tube, nasojejunal tube, or gastroscope) [24, 35, 40, 42,