Bariatric–metabolic surgery remains the most effective method of obesity treatment providing long term weight loss and improvement of obesity-related diseases. According to the International Federation for the Surgery of Obesity and Metabolic Diseases, the number of bariatric procedures performed worldwide in 2013 exceeded 460,000. The most commonly performed bariatric procedures are gastric bypass (GB), sleeve gastrectomy (SG), and laparoscopic adjustable gastric banding (LAGB) [1, 2].

Although bariatric procedures are effective, they have various degrees of success and complication profiles that are unique to the procedure type. Overall, bariatric surgery has a low incidence of serious complications of approximately 4% and mortality rate of 0.1% [3, 4]. The Agency for Healthcare Research and Quality (AHRQ) and recent clinical studies report significant improvements in metabolic and bariatric surgery safety, which is mainly associated with the increased use of laparoscopy and advances in surgical techniques [3, 4] Among all complications, fistulas and leaks are major adverse events which increase post-operative morbidity and mortality, especially in the acute phase [5]. The incidence of leaks after SG has been reported to be approximately 1.06% [6]. Post-SG leak can lead to the development of gastric fistula over time. Fistulas after SG occur in 0.2% to 2.5% of cases and are most commonly located at the proximal third of the gastroplasty [6, 7]. Leaks are also the major complications of GB, occurring in 0.7% to 5% of patients. They are usually located at the gastrojejunal anastomosis, but have also been noted at the distal esophagus, gastric pouch, remnant stomach, blind jejunal limb, and jejunojejunal anastomosis [8, 9].

The treatment of fistula/leak may involve surgical, endoscopic, and/or radiological procedures [3]. Over the last years, the management evolved with the development and improvement of several endoscopic techniques including self-expanding metal (SEMS) and plastic stents, clipping techniques [including the use of through the scope clips (TTSC) and over-the-scope clips (OTSC)], tissue sealants, suturing systems (OverStitch System®), and internal drainage techniques [3, 10,11,12]. The use of endoscopic therapies has gained popularity over time and tends to be more standardized among expert teams. The available literature contains many case reports, case series and only a few retrospective observational cohort studies assessing the use of different endoscopic techniques to treat post-bariatric leaks and fistulas. The results of these studies are inconclusive as they report the use of individual techniques as monotherapy or in combination with surgery or other endoscopic technique, mainly based on a given center’s experience. There are no prospective, randomized studies on this topic. Meta-analyses and systemic reviews on this topic are also limited [13,14,15].

With this in mind, we performed a meta-analysis and systematic review of self-expanding stents, clipping techniques, or tissue sealants used as primary strategies in the treatment of leak/fistula after SG, GB, and LAGB with the aim to assess technical characteristics, successful closure rate, and technique-related complications.

Materials and methods

This systematic review and meta-analysis was performed in accordance with the guidelines formulated in the Cochrane Handbook for Systematic Reviews of Interventions [16]. The authors followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines for systematic reviews and meta-analyses.

Search strategy

A systematic literature search was conducted by two researchers (R. P. and S-S. A) to identify and appraise studies of endoscopic management of anastomotic leaks and fistulas after bariatric surgery. MEDLINE (PubMED) and SCOPUS databases were searched from inception to February 2019. The formulas used to search the MEDLINE (PubMED) and SCOPUS databases are showed in Supplementary Materials. The reference lists of review articles were hand-searched for additional relevant studies. Inclusion and exclusion criteria were determined by two researchers (R. P. and S-S. A). All authors independently determined studies eligible for meta-analysis and systematic review. Institutional review board approval and written consent for this paper was not required.

Definitions

Based on the initial review of the literature, we found that both the definition of leak and the time intervals (acute, early, late, chronic) varied. There was also no universal definition of fistula. While some authors defined it as late or chronic leak, others used the term leak and fistula interchangeably. Therefore, in the inclusion criteria, we used the general definition of leak and fistula as an endoscopic or radiologically confirmed dehiscence of anastomosis or leakage of gastrointestinal content from a surgical join between two hollow viscera or through a suture line around an organ or the presence of a luminal content collection next to the anastomosis [17].

Eligibility criteria

Included studies employed trials involving endoscopy in the management of anastomotic leaks and/or fistulas after bariatric surgery. Only full-text articles, focused on self-expanding metal stents or tissue sealants or clipping techniques, published in English were considered. Randomized controlled trials, non-controlled clinical trials, observational cohort studies, and case series (≥ 3 cases) were considered eligible for the meta-analysis. The inclusion criteria were as follows: (1) research on patients with fistulas and leaks after bariatric procedures, including gastric sleeve, gastric bypass, or laparoscopic adjustable gastric banding, (2) studies in which the stent or clipping technique or tissue sealant application was the preferred method of endoscopic leak/fistula closure, (3) studies in which stents were used after previous unsuccessful attempts at endoscopic or surgical treatment. The following studies were excluded: (1) studies evaluating only combined endoscopic methods; (2) studies without clear data and/or description of therapy are used; and (3) studies focusing on endoscopic methods other than the stent or clipping technique or tissue sealant application. In addition, a meta-analysis of studies on self-expanding stents included only the studies specifying the frequency of stent migration.

Extracted data and subgroup analysis

The following data were extracted: (1) study characteristics (author name, publication year, type of study); (2) data on participants (sample size, age, gender); and (3) data on interventions, success, and adverse events of endoscopic therapy.

Data on stents, clipping techniques, and tissue sealants were analyzed and presented separately. A meta-analysis of studies reporting the use of stent in the treatment of leak/fistula was performed to assess the frequency of successful leak/fistula closure (defined as the percentage of patients with successful leak/fistula closure, confirmed by endoscopy or contrast X-ray after), and stent migration. In addition, systemic reviews of studies reporting the use of clipping techniques and tissue sealants were performed to determine the efficacy and complications of such therapies.

Statistical methods

A random-effects model described by DerSimonian and Laird was used to aggregate the study data [18]. For zero-score events, the continuity correction was performed by adding a correction factor of 0.5. Proportions of overall successful leak/fistula closure, stent migration as well as successful leak/fistula closure in gastric sleeve and gastric bypass group were given with 95% confidence intervals that are based on exact binomial Clopper–Pearson method [19]. Statistical heterogeneity between the studies was evaluated with the Q Cochrane’s statistics and the I2 coefficients, which showed contribution of heterogeneity relative to the whole for each study. The publication bias was examined by visual inspection of funnel plots and formally with Begg’s test with continuity correction [20]. Furthermore, sensitivity analysis was performed for parameters showing significant heterogeneity. The analysis was performed using the STATA software, version 14.2 (forest plot and Begg’s test) and PQStat software, version 1.8.0 (visual interpretation of funnel plot and sensitivity analysis). The significance level of 0.05 was assumed.

The efficacy of clipping techniques or tissue sealants was presented as a percentage of leak/fistula closure and the frequency of technique-related complications.

Results

The initial database search identified 3757 reference articles, in which 65 relevant articles were selected and reviewed (Fig. 1 and Supplementary material: Figure S9 show the search results).

Fig. 1
figure 1

Flowchart for search strategy and selection of eligible studies for systemic review and meta-analysis

Full size image

Self-expanding metal and plastic stents

Data focused on the efficacy of self-expanding stents in the treatment of leaks and fistulas after bariatric surgery were extracted from 40 studies (493 patients) that met the inclusion criteria. We did not find randomized controlled trials and non-controlled clinical trials. Therefore, only cohort observational studies and case series were included in the meta-analysis. All selected studies were published between 2006 and 2019 (Table 1) [21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61].

Table 1 Studies assessing self-expanding stents in the treatment of post-bariatric leak/fistula eligible for meta-analysis
Full size table

The median body mass index (BMI) of patients varied between 32 (30–42) and 56.6 (44–65) kg/m2. Sixteen studies provided information on the interval between surgery and leak diagnosis; this period ranged from 4.6 (3–7) to 142.3 (7–252) days. The time between leak diagnosis and stent placement was reported by 6 studies and ranged from 5 (3–10) to 82 (5–367) days. The time between surgery and stent placement was reported by 6 studies and ranged from 14 (7–21) to 95 (13–395). The median interval between implantation and removal of the stent was reported in 22 studies and varied between 15 (14–16) and 121.7 (18–341) days. Leaks were most often located within the gastroesophageal junction (GEJ) near the proximal end of the staple line or in the distal portion of the esophagus, or in the upper third of the gastric stump. The mean estimated defect size was reported only in one study and was 1.18 cm [38].

Overall proportion of successful leak closure

The overall proportion of successful leak/fistula closure was 89% (95% CI, 85–92%) (Fig. 2). However, the funnel plot, sensitivity analysis and the Begg’s test suggested an existing bias and asymmetry between the studies (Supplementary material: Figure S1). Therefore, several studies were excluded [48, 52, 62]. After excluding above studies, the remaining studies were homogenous (I2 = 0.00%, p = 0.77) and the overall proportion of successful leak or fistula closure did not change significantly—92% (95% CI, (90–95%) (Supplementary material: Figure S2).

Fig. 2
figure 2

Forest plot for successful leak closure. After exclusion of outliers: overall proportion of successful leak closure = 92% (95% CI, 90–95%), test of heterogeneity I2 = 0.00% (p = 0.77)

Full size image

Successful leak closure in gastric sleeve group

Thirty-seven studies (Table 1) reported the effectiveness of SEMS after gastric sleeve (344 patients). The proportion of successful leak closure in gastric sleeve group was 92% (95% CI, 88–95%), I2 = 0.00% (p = 0.81)—Fig. 3, Supplementary material: Figures S3 and S4.

Fig. 3
figure 3

Forest plot for successful leak closure in gastric sleeve group. After exclusion of outliers: overall proportion of successful leak closure = 92%, (95% CI, 88–95%), homogeneity coefficient was I2 = 0.00% (p = 0.81)

Full size image

Successful leak closure in gastric bypass group

Fifteen studies (Table 1) reported the effectiveness of SEMS after gastric bypass (149 patients). The proportion of successful leak closure in gastric bypass group was 96%, (95% CI, 91–100%), I2 = 0.00% (p = 0.58)—Fig. 4, Supplementary material: Figures S5 and S6.

Fig. 4
figure 4

Forest plot for successful leak closure in gastric bypass group. After exclusion of outliers: overall proportion of successful leak closure = 96%, (95% CI, 91–100%), homogeneity coefficient was I2 = 0.00% (p = 0.58)

Full size image

Stent migration

The overall proportion of stent migration was 23% (95% CI, 17–30%) (Fig. 5). However, a significant heterogeneity between the studies was observed (I2 = 72.41%, p < 0.01). Also, the funnel plot, the sensitivity analysis, and the Begg’s test suggested an existing bias and asymmetry (Supplementary material: Figures S7 and S8). Therefore, several studies, which contributed to heterogeneity the most and were located outside the funnel, were excluded [21, 23, 24, 26, 34, 38, 40, 63]. After excluding the above studies, the remaining studies appeared to be homogeneous (I2 = 3.36%, p = 0.41) and the overall proportion wasn’t changed, but the precision increased and the confidence interval become narrower—proportion of stent migration = 23% (95% CI, 19–28%). The contribution of individual studies to the total proportion of stent migration was comparable.

Fig. 5
figure 5

Forest plot for stent migration. After exclusion of outliers: overall proportion of stent migration = 23% (95% CI, 19–28%), test of heterogeneity I2 = 3.36% (p = 0.41)

Full size image

Clipping techniques

We identified a total of 46 papers by our initial search protocol and performed a manual review for relevant articles describing the use of clips in the treatment of leak/fistula after bariatric therapies (Supplementary material, Figure S9) [40, 53, 64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91]. Among them, we found 17 articles reporting clipping as the preferred endoscopic therapy for leak/fistula closure. There were seven case reports including a total of 9 patients and ten retrospective studies including a total of 89 patients. Additional therapies used are noted in Table 2. Patients age varied between 21 and 67 years. The leak and/or fistula were the complications of SG in seven studies, GB in nine studies, and LAGB in two studies.

Table 2 Clipping in the management of post-bariatric leak/fistula—characteristics of studies, included patients, and leak/fistula occurrence
Full size table

The time interval between surgery and diagnosis of leak/fistula was reported in 13 studies and ranged from 1 day to 38 years. Among them, 10 studies reported leak/fistula occurrence within a month after bariatric procedure. The clinical manifestations of leak/fistula were shown in 13 studies (53 patients) and included abdominal pain, fever/peritonitis, nausea/vomiting, and abdominal collections. Weight regain as the main clinical symptom was reported by 8 patients (4.2%) (Table 2). In 1 study, 4 of 19 patients did not report any symptoms and fistula was diagnosed in control imaging examinations [75].

The location of leak/fistula was described in all except one study. The most common locations of leak/fistula were proximal staple line, gastrogastric, and gastroesophageal junction (Table 3). In addition, there were three gastrocutaneous fistulas and one esophagobronchial fistula reported by one study [77]. The size of the leak/fistula was assessed by six studies and varied between 3 and 20 mm. Thirteen studies (85 patients) used the OTSC system for leak/fistula closure and only one system was required in most of the studies. The size of OTSC was specified in five studies (16 OTSC in 13 patients) and included: 12/6t (8 OTSC), 12/6gc (6 OTSC), 14/6t (1 OTSC), and 11/6t (1 OTSC). Four studies (13 patients) reported the use of other endoclips including TriClip, Quick Clip, and Resolution Clips (1–6 clips per patient) [53, 68, 70, 79]. Prior therapies used before clipping included external drainage (seven studies), revision surgery (three studies), argon plasma coagulation (one study) (Tables 2 and 3). In addition, we presented the data regarding OTSC therapy from the study describing the use of endoscopic closing techniques alone versus endoscopic drainage (with or without other endoscopic techniques) [72]. We excluded from our analysis studies or study results reporting the concomitant use of stent or stent placement before clipping (Supplementary material, Figure S9) [40, 66, 70, 72, 75, 77, 78, 81,82,83,84,85,86,87,88,89,90,91].

Table 3 Clipping in the management of post-bariatric leak/fistula—characteristics of leak/fistula and endoscopic procedures
Full size table

Different definitions of therapy success were applied in the reviewed studies. Overall, successful closure of a leak/fistula with the OTSC system was achieved in 57 of 85 patients (67.1%). However, five studies reported the need for additional treatments after primary OTSC therapy to achieve secondary success such as: multiple attempts with OTSC [74, 76], other endoscopic therapies (SEMS, standard clips, glue, suturing) [73, 76, 77, 80], and surgery [73, 76, 80]. Only two studies reported complications related to the OTSC system including anchor migration (1 patient), mediogastric stenosis (1 patient), and one complication related to delivery system (anchor blocked within the clip) [77, 80]. Among four studies describing the use of other clips for leak/fistula closure, successful treatment was reported in 9 of 13 patients (69.2%) (Table 3). In one study, 4 of 8 patients (50%) with therapy failure were referred for surgical repair [79].

The follow-up imaging included radiology and/or endoscopy in all studies and was performed in different intervals varying from 2 days to 12 months. Post-closure follow-up ranged from 14 days to 46 months (Table 3). Three case reports provided information about follow-up body weight and reported weight loss in two patients and stable body mass in one patient [64,65,66].

Tissue sealants

We found 10 case series comprising 63 patients treated by the application of tissue sealant—fibrin glue—as a single endoscopic method for leak/fistula closure after bariatric surgery (Tables 4 and 5). The time between bariatric operation and sealant application varied from 1 to 144 days (median for 6 reports: 12.5 days) [92,93,94,95,96,97].

Table 4 Tissue sealants in the management of post-bariatric leak/fistula—characteristics of studies, included patients, and fistulas
Full size table
Table 5 Tissue sealants in the management of post-bariatric leak/fistula—characteristics of endoscopic procedures
Full size table

In 8 reports (25 patients), the sealant was delivered via endoscopic access with a 100% success rate in fistula closure. Some authors passed a cytology brush into the fistula orifice to clean away debris and loose granulation tissue before application of fibrin glue [98]. In 2 studies (38 patients) the sealant was delivered via combined percutaneous and endoscopic access with 95.8–92.8% rate of fistula closure [96, 98] (Table 4).

The number of sessions per patient needed for successful fistula closure was reported in 9 studies comprising of 49 patients [92,93,94,95, 97,98,99,100,101]. Success was achieved in 48 patients: after 1 session in 22 (45.82%) patients, after 2 sessions in 10 (20.83%) patients, after 3 sessions in 9 (18.75%) patients, after 5 sessions in 2 (4.17%) patients, after 6 sessions in 4 (8.33%) patients, and after 9 sessions in 1 (2.08%) patient (Table 5). One study reported only a median number of 4 sessions needed for closure [96]. In the majority of reports the sessions of sealing were repeated every 2 to 3 days. Only in a report from Assalia et al. [98] were sessions scheduled at 2 week intervals.

The exact volume of fibrin glue used per session was reported in 6 studies and ranged from 2 to 10 ml (median 4 ml). In 6 reports, endoscopic treatment was combined with a total parenteral nutrition and in addition with somatostatin or octreotide in 3 reports [93,94,95,96,97, 100].

The complications of fibrin glue applications were reported in only one study and included pain and fever in 3 of 24 patients (12.5%) [98].

Discussion

In our systematic review and meta-analysis, we focused on the most commonly used endoscopic therapies such as self-expanding stents, clipping, and tissue sealants used as the preferred endoscopic method for leak and/or fistula closure. In general, we found a high efficiency of self-expanding stents, clipping techniques, and fibrin glue in closing post-bariatric leak/fistula.

The effectiveness of self-expanding stents was the subject of two previously published meta-analyses [13, 14]. Eight years have passed since the publication of the first of them. Therefore, it summarizes the preliminary results of the use of stents non-strictly designed for the treatment of leaks and fistulas after bariatric surgery. In addition, the significance of this meta-analysis is limited by the relatively small sample size. A second meta-analysis included publications from 2006 to 2016 presenting more recent results. However, out of the 28 studies included in this meta-analysis, only 4 studies used stents designed specifically for treating leaks after bariatric surgery. Such stents were used in 11 studies included to our meta-analysis. Seven of these studies were published in 2017–2019. The use of stents dedicated to bariatric patients is becoming a standard therapy approach, therefore, our meta-analysis better reflects the current results of treatment of post-bariatric leaks with self-expanding stents. In fact, the results of endoscopic stent treatment obtained in our meta-analysis were significantly better compared to the previous meta-analysis 92% versus 72.8% and 96% versus 76.1% in GS and GB group, respectively. We can speculate that this was due to more frequent use of stents designed to treat post-bariatric leaks. Unfortunately, most publications did not provide results of the effectiveness of closing leaks and fistulas in relation to the stent used. Therefore, we were unable to perform a separate analysis.

Clipping techniques including the OTSC system were also effective for leak/fistula closure. The overall successful closure was achieved in 67% of patients. In the previously published systematic review, 86.3% of patients treated with the OTSC system had an overall successful leak/fistula closure [15]. However, it included studies with concomitant or previous additional endoscopic procedures such as stent placement. Therefore, the results of this systematic review demonstrate the effectiveness of endoscopic combined techniques rather than the clips themselves. In our systematic review, an overall successful leak/fistula closure was significantly lower compared to previous systematic review. However, we did not analyze studies reporting concomitant stenting or other endoscopic methods. Moreover, Shoar et al. analyzed the effectiveness and safety of the OTSC system only, while in our study we looked at other clips systems.

The most commonly used sealant for fistula closure is fibrin glue—a tissue-compatible adhesive working in a double manner. It mechanically occludes the stomach wall defect and plays a predominant role in wound healing, inducing cellular response to tissue damage, forming matrix-building strands, which promote neovascularization and fibroblast proliferation [102]. Some bariatric surgeons routinely use fibrin sealant to facilitate healing of stapled closures and anastomoses as a prevention of leaks [103]. In our study, fibrin glue was highly effective, but in most cases repeated sessions were necessary to achieve final closure. The main reasons of tissue sealing failure was due to the huge orifice of the fistula [95] or non-compliance of the patient who did not appear regularly on scheduled sessions of fibrin glue application [98]. Although the cost or the fibrin glue is considerable, the cost of one session of tissue sealing is more than five times lower than that of stent insertion and more than 6 times lower than OTSC [95]. Among other types of tissue sealants there are cyanoacrylate glue and SurgiSIS [99, 104]. Cyanoacrylate, a synthetic glue working as a mechanical sealant, has the advantage of having high adhesive and high antibacterial properties, and thus is suitable for application in infectious sites. It is eliminated by hydrolysis after a significant time period (1–6 months), and only a small quantity of the glue is needed [95]. The cost of one portion of cyanoacrylate is approximately six times lower than one portion of fibrin glue [95]. Despite these advantages, the poor mechanical properties of the film, brittle nature, possible proinflammatory effect as well as the risk of damage of the endoscope because of its rapid polymerization make cyanoacrylate a second-choice method. There are fewer studies on those sealants than on fibrin glue, usually combining multiple techniques. Thus, we were unable to extract the specific effectiveness of tissue sealant instead of the effectiveness of other methods or combined endoscopic therapy from these studies [82]. Kotzampassi et al. demonstrated high efficacy of 96.8% of cyanoacrylate in a heterogeneous series of 8 patients with fistulas after various types of surgery [95]. The volume of cyanoacrylate delivered was 0.5 or 1 ml for one session. Total volume applied was 0.5–4 ml (median 1.5) in a median of 2 sessions (range 1–4 sessions). However, the study does not provide detailed information about the type of surgery or fistula location. On the other hand, Vilallonga et al. reported high efficiency of transcutaneous application of cyanoacrylate in combination with stent implantation in the treatment of gastrointestinal fistulas [104]. SurgiSIS (Wilson-Cook, Winston-Salem, NC) is an acellular matrix biomaterial formulated from the porcine small intestine submucosa that stimulates proliferation and formation of fibroblasts in the region of wounds and incorporates into the scar without initiating a foreign body inflammatory reaction. Strips of soaked SurgiSIS material are captured within a specially designed polypectomy snare and loaded into the endoscope outside of the patient. This quite inconvenient delivery method is balanced by high efficacy. The rate of 5 to 20 mm wide fistulas closure after 3 sessions was achieved in 20 of 25 patients [99].

The use of endoscopic therapies demands precise visualization of the internal fistula orifice, which can be a great challenge. In many patients, its presence is confirmed late, after numerous radiologic and endoscopic examinations. Proper selection of patients seems to be critical for favorable outcomes. Patients qualified to endoscopic therapy were hemodynamically stable and in many cases the leak was controlled by percutaneous drainage. Septic patients with uncontrolled gastrointestinal leaks or peritonitis should be treated surgically. The success of endoscopic therapies in the management of leak/fistula also depends on the defect’s size. In general, self-expanding stents allow closing the largest leaks and fistulas. The studies reported use of clipping including the OTSC system for fistulas not larger than 20 mm. In addition, thin and clean fistula canals facilitated quick closure with fibrin glue application. However, most studies lacked information about fistula orifice size. Due to the heterogeneity of the endpoints of the included studies, we were also not able to perform additional analyses on: the optimal time to start endoscopic therapy, and the length of endoscopic treatment. Some of the studies provided information on the duration of stent dwelling time. Although the time range was very wide, the period of maintaining stents in the gastrointestinal tract recommended by most authors was 6–8 weeks.

To ensure the clarity of results, we did not include studies assessing complex endoscopic techniques. However, reviewing the publications from recent years, we found some excellent research on the endoscopic treatment of fistulas and leaks after bariatric surgery [84, 105,106,107]. These studies show an important trend in the treatment of post-bariatric leaks. Currently, complex endoscopic surgical treatment or combined treatment with simultaneous or sequential use of several endoscopic methods seems optimal in management of the complications of bariatric surgery. Therefore, future research should focus on assessing the effectiveness of complex therapies rather than individual endoscopic methods. Rebibo et al. compared the results of endoscopic management of large gastric leaks or gastric stenoses associated with gastric leaks using covered stents with endoscopic combined treatment using covered stents and double-pigtail stents [51]. The median time to gastric leak closure was shorter, the number of endoscopic procedures, the stent migration rate and the failure rate was significantly lower in the double-pigtail plus covered stent group compared to the covered stent group only. Shebab et al. performed a retrospective analysis of 81 patients with leaks after SG or GB treated with Mega stent (an ultra-large covered stent) alone or in combination with OTSC, which was applied in 46% of patients [84]. The OTSC was placed simultaneously with the stent or after stent removal. The final leak closure was achieved in 82% of patients with a mean of three endoscopic procedures per patient. The authors concluded that the approach combining stents with OTCS is highly effective, but Mega stents should be used with caution as most of observed complications were associated with stent placement. The authors mentioned that clip placement in the treatment of post-bariatric leaks is less effective as the leaks are surrounded by fibrotic and inflamed tissue and clips can act as a foreign body limiting healing. Therefore, they preferred the use of stent as the first method, and clips when the leak persist after stent therapy [84].

Cost-effectiveness is another important issue related to the treatment of post-bariatric complications. Unfortunately, no cost analysis was carried out in any of the studies included in our systematic review and meta-analysis. Theoretically, the use of endoscopic methods can contribute to reducing the costs associated with reoperation and the patient's stay in the ICU. However, the cost-effectiveness of individual endoscopic methods also varies. Cosse et al. have recently shown that double-pigtail stents for the treatment of gastric leak is more cost-effective than covered stents and should be proposed as the standard regimen whenever possible [105].

We did not find endoscopic procedure-related mortality. In our meta-analysis, the frequency of stent migration, which is the most common self-expanding stent-related complication was 23%. In contrast, two recently published meta-analyses reported stent migration rates to be 16.94% and 30.5% [13, 14]. The use of endoscopic suturing or clip application to fix esophageal stents and prevent migration have been reported only in a few studies. Therefore, we were not able to perform a comparative analysis between studies that used or did not use these techniques. The analysis of the relationship between stent type and frequency of stent migration was not possible due to the large variety of stent types used. Nevertheless, based on the individual results of some studies, it seems that the use of partially covered stents reduces the risk of migration. These stents increase tissue hyperplasia which can complicate their removal. Other severe complications of stent implantation were rare. One reported gastrointestinal perforation [38]. Two studies showed minor bleeding as a complication of stenting [32, 34]. Some patients reported dysphagia and vomiting as a result of esophageal stenosis due to tissue hyperplasia, obstruction of the stent with food, or stent collapse [21, 27, 34, 37, 40]. Mild retrosternal discomfort, nausea, excessive salivation which resolved spontaneously after a few days were the most commonly reported symptoms associated with stents [21, 37, 41]. The first published studies reported difficulties in stents removal [21, 24]. Nevertheless, stent removal is currently easily performed due to the improvement of stent design and stent removal techniques.

Clipping or fibrin glue – related complications were also rare, although reported only by a few studies without adequate follow-up information. Only one study reported pain and fever in 12.5% of patients after fibrin glue application [98]. Two studies described a few complications associated with clipping such as anchor migration, tear, mediogastric stenosis, and one complication related to the delivery system (anchor blocked within the clip) [77, 80].

Our study has several limitations. First, due to the high heterogeneity of the included studies and missing data, we could not analyze some of the assumed endpoints. The funnel plot, sensitivity analysis and the Begg’s test suggested an existing bias and asymmetry between the studies assessing stents. Therefore, several studies had to be excluded from the meta-analysis. Moreover, our analysis was limited by the inclusion of retrospective studies, case series, and case reports due to lack of prospective, randomized controlled trials. There is also a risk of bias associated with the publication of only those studies in which the effectiveness of stents, clips and tissue sealants has been confirmed. Several retrospective studies reported the efficacy of leak/fistula closure using different techniques, probably dependent on the experience and capabilities of medical centers. In addition, reports describing the clipping techniques or fibrin glue in the treatment of leak/fistula used different definitions of therapy success and follow-up time. And finally, we included studies on post-operative leaks as well as chronic fistulas, so the timespan from surgery to diagnosis of the defect was wide, as well as the timespan from the diagnosis of leak to stent implantation. Interestingly, all reviewed methods were shown to be effective in both acute leaks as well as chronic fistulas.

In summary, to the best of our knowledge, our meta-analysis and systematic review is currently the largest analysis of the efficacy and safety of endoscopic treatment of post-bariatric leaks and fistulas.

In conclusion, there is the most evidence of the effectiveness of self-expanding stents in the treatment of post-bariatric leaks and fistulas. However, despite the use of new stent designs, the frequency of stent migration remains high. Post-bariatric fistulas and leaks with an orifice size of up to 20 mm can also be successfully treated with clips, preferably OTSC. In turn, application of fibrin glue allows closing narrow fistulas. However, it may require multiple sessions to achieve leak closure. There is an urgent need for RCTs to assess the efficacy and safety of both individual as well as combined endoscopic methods in the treatment of post-bariatric leaks and fistulas.