Intraoperative perforation rates of EMR and ESD were reported to be 0% to 1.6% and 2.6% to 6.9%, respectively.9,18-21 The risk of intraoperative perforation is much higher in cases with large tumor sizes or for procedures in less experienced facilities.22,23 Esophageal ESD is technically challenging because of the following reasons: (1) the narrow lumen of the esophagus decreases the efficacy of gravity counter-traction;8 (2) a resected specimen retracts distally, thereby making it difficult to maintain good traction and orientation;8 (3) the thin wall of the esophagus increases the risk of perforation;8 and (4) localized muscle layer defects lead to unexpected extensive perforations, although these are extremely rare.24 Traction-assisted ESD, which helps maintain a good endoscopic view, is useful in overcoming the challenges mentioned above.25 According to CONNECT-E, a multicenter randomized controlled trial (RCT), no perforation occurred in traction-assisted ESD using dental floss, whereas 4.3% of the cases had perforations in conventional ESD.26 The ESD/EMR guidelines for esophageal cancer published by the JGES weakly recommend traction-assisted ESD.16

Clip closure technique and conservative therapy (e.g., fasting and antibiotic therapy) are the first lines for intraoperative perforation.27 Clip placement should begin at the defect’s end distal to the endoscope to provide an optimal endoscopic view.28 Recently, the usefulness of the application of a polyglycolic acid (PGA) sheet has also been reported.29,30

Delayed perforations after esophageal ESD can occur, owing to tissue necrosis and degeneration due to excessive energy during the procedure.31 A literature that describes the management of delayed perforations is limited to case reports,31,32 in which the temporary use of self-expanding metallic stents (SEMSs) is reported as a nonsurgical treatment.31 There is no consensus on whether surgical intervention or an endoscopic approach should be performed to manage delayed perforations after esophageal ESD; therefore, a therapeutic decision should be made based on the patient’s condition.

Delayed bleeding after esophageal ESD is a relatively rare complication with an incidence rate of 0% to 1.3%.21,22,33 The risk factors for delayed bleeding are not well described in previous reports. Delayed bleeding after esophageal ESD is usually managed with endoscopic hemostasis and does not need surgical intervention.33

According to Shi et al.,34 the rate of post-ER stricture in patients who underwent esophageal ER for superficial esophageal carcinoma was 11.6% (42/362), and the median time from ER to stricture was 28 days (21–90 days). Moreover, the incidence of stricture increases as the circumferential range of the mucosal defect increases (≤1/2 circumferential mucosal defect, 0.7%; 1/2–3/4 circumferential mucosal defect, 27.6%; and ≥3/4 circumferential mucosal defect, 94.1%). A widespread mucosal defect involving three-quarters of the circumference of the esophagus is a strong risk factor for the development of post-ER stricture.34-36 Several preventive methods and management strategies have been developed for post-ER stricture (Table 1).37-44

The J-WEB/EGC cohort revealed that intraoperative perforations during gastric ESD occurred in 2.3% of lesions (218/10,821); however, most of them could be treated with conservative management, and only seven cases required emergency surgery.15 Endoscopic clip closure is the first-line therapeutic choice for intraoperative perforations.56,57 Other methods (such as endoloop and endoclip closure method,58 over-the-scope clip [OTSC] method,59 and PGA sheet and fibrin glue60) are used for perforations that are difficult to close completely with the simple clip closure method. The American Gastroenterological Association’s expert review recommends using standard through-the-scope clips (TTSCs) or OTSCs for perforations ≤2 cm and endoscopic suturing or a combination of TTSCs and endoloops for perforations >2 cm.28 The most important factor regarding the indication for emergency surgery is the condition of the patient, including vital signs and symptoms.27,28

Risk factors for intraoperative perforations during gastric ESD include the involvement of the upper third of the stomach, severe fibrosis, invasion depth, and tumor size ≥20 mm.61,62 ESD for gastric tumors in the upper third of the stomach is challenging, owing to frequent intraoperative bleeding and thinner wall thickness than that of other parts of the stomach.62 CONNECT-G, a multicenter RCT, demonstrated a lower incidence of perforations in traction-assisted ESD than in conventional ESD (0.3% vs. 2.2%, p=0.01). The traction-assisted technique can reduce the risk of perforations by improving the visibility of the submucosal dissection plane, which helps in the identification of muscle direction during submucosal dissection.63

Delayed perforations after gastric ESD are rare, but can cause peritonitis, which requires emergency surgery. A single-center retrospective study of 4,943 cases showed that delayed perforations after gastric ESD occurred only in seven (0.1%) cases, and the median time until delayed perforations was 11 hours (range, 6–172 hours).64 Gastric tube cases, lesions located in the upper stomach, and excessive electrocautery are associated with delayed perforations.64,65 Symptoms of peritoneal irritation and rebound tenderness are critical physical signs, indicating a necessity for emergency surgery.65 Although emergency surgery can be required with a high probability,65,66 several cases of delayed perforations were successfully managed conservatively.67,68 Suzuki et al.64 reported that early detection of the onset of delayed perforations within 24 hours after the procedure might be helpful in avoiding emergency surgery. Although no report has described effective prevention methods for delayed perforations, owing to their rarity, complete closure of the mucosal defect theoretically contributes to reducing the incidence of delayed perforations.

Delayed bleeding is a major complication related to gastric ESD. Although administering proton pump inhibitors (PPIs) and prophylactic coagulation after ESD are reportedly effective for preventing delayed bleeding,69-71 the incidence of delayed bleeding remains as high as 2.6% to 5.8%.15,72,73 A meta-analysis indicated that significant risk factors for delayed bleeding after gastric ESD were the use of antithrombotic agents, chronic kidney disease, resected specimen size >30 mm, and use of histamine-2 receptor antagonists instead of PPIs.74 Hatta et al.75 reported the BEST-J score, which is a novel scoring system for predicting delayed bleeding using risk factors, such as comorbidities, tumor factors, and use of antithrombotic agents. This prediction model stratifies the risk of delayed bleeding into four levels, according to the total score of risk factors: low-risk (0–1 point; 2.8%), intermediate-risk (2 points; 6.1%), high-risk (3–4 points; 11.4%), and very high-risk (≥5 points; 29.7%) (Table 2).75 Notably, the management of antithrombotic agents in the BEST-J score is based on the 2014 JGES guideline,76 which does not include the management of direct oral anticoagulants.

Second-look endoscopy (SLE) has been empirically performed to prevent delayed bleeding; however, three RCTs revealed that routine SLE did not reduce delayed bleeding in patients with an average risk of bleeding.77-79 Furthermore, in a single-center retrospective study, there was no statistical difference in the incidence of delayed bleeding between SLE and non-SLE groups in patients taking antithrombotic agents (21.7% vs. 21.9%, respectively).80 There is no prospective study investigating the efficacy of SLE in a high-risk group, and the indication of SLE in high-risk patients remains unclear.

Several researchers have developed innovative approaches, such as tissue shielding methods (PGA sheets and fibrin glue) (Fig. 1),81,82 endoscopic closure of artificial ulcers (the endoloop and endoclip closure method, OTSCs, and hand-suturing method),83-86 and the use of hemostatic powder,87,88 for the prevention of delayed bleeding. These methods are particularly used in high-risk patients.

Kataoka et al.81 conducted an RCT that investigated the efficacy of PGA sheets and fibrin glue in high-risk patients. There was no significant difference in delayed bleeding rates between PGA and control groups (4.5% vs. 5.7%, p>0.99). Although this study enrolled high-risk patients, delayed bleeding occurred in only 5.7% of the control group; this incidence rate is equivalent to that of low-risk patients.81 This result implies that the inclusion criteria were not optimal. Therefore, further well-designed and advanced large-scale studies are required to provide evidence for PGA sheets in the prevention of delayed bleeding.

Endoscopic closure of mucosal defects (including the endoloop and endoclip closure method,83 OTSCs,84 and hand-suturing method85,86) has a clinical potential to reduce the risk of delayed bleeding (Table 3). However, it is difficult to achieve complete closure of a gastric mucosal defect and avoid dehiscence because of the thickness of the gastric wall. Endoloop and endoclip closure method is an easy-to-perform and low-cost procedure, but it is not satisfactory in keeping the mucosal defect closed.83 Ego et al.83 reported that the endoloop and endoclip closure method did not prevent delayed bleeding in patients with antithrombotic therapy compared with control individuals (11.5% vs. 11.9%, p=0.89), and sustained closure rate in patients who underwent SLE was 47.8% (33/69). Endoscopic hand-suturing (EHS) demonstrated an incredibly high rate of complete closure (97%–100%) and sustained closure of mucosal defects (84%–100%).85,86 Akimoto et al.85 reported that there was no delayed bleeding in patients taking antithrombotic agents without perioperative cessation (0%, 0/20). EHS is a promising method for achieving complete and sustained closure, thereby allowing for the prevention of delayed bleeding. However, EHS is a technically challenging and time-consuming procedure. Further improvement is required to utilize EHS in clinical practice. Polysaccharide hemostatic powder (PHP) is a novel topical hemostatic method used for non-variceal upper gastrointestinal bleeding,89 and its efficacy for preventing post-ER bleeding has been investigated.87,88 Jung et al.87 conducted an RCT to investigate the efficacy of PHP in preventing post-ESD bleeding among patients with a high risk of bleeding (resected specimen size >40 mm and regular use of antithrombotic agents). This study showed that there was no significant difference in post-ESD bleeding rates between PHP and control groups (5.5% vs. 7.1%, p=0.74). However, a subgroup analysis of patients who discontinued antithrombotic agents during ESD showed that post-ESD bleeding rate tended to be lower in the PHP group than in the control group (0% vs. 6.3%, p=0.06).

Considering the increasing population of patients taking antithrombotic agents, the development of an effective preventive method for delayed bleeding is desirable. However, there is no RCT demonstrating the efficacy of preventive approaches. This topic is still under investigation, and further innovative methods are required.

It is reported that stricture occurs in 0.8% to 2.5% of cases following gastric ESD.90-93 An extensive mucosal defect that comprises three-quarters of the circumference in the antrum or cardia is a significant risk factor for stenosis.91 An EBD is empirically performed to alleviate stricture and stasis symptoms; however, it carries a certain risk of perforation (7.8%–14.3%).91,94 Moreover, cases of post-gastric ESD stricture require repeated EBD until the stricture is eliminated, and there are cases of strictures refractory to repeated EBD. Contrary to esophageal stricture, there is limited evidence for preventive methods, such as local steroid injection and prophylactic EBD for stricture following gastric ESD.94,95 Further investigations should be conducted to determine the optimal approach for preventing strictures.

A multicenter retrospective observational study, including 3,107 cases (187, 579, 1,324, and 1,017 cases of cold polypectomy, UEMR, conventional EMR, and ESD, respectively) revealed that intraoperative perforation and delayed perforation rates were significantly higher in an ESD group than in other groups.103 Fukuhara et al.106 analyzed the clinical course of 32 patients with intraoperative perforations during duodenal ESD. Either simple closure using clips or the string clip suturing technique was attempted to close a perforation, and complete closure was achieved in 13 of 32 patients (40.6%). They found that if the whole mucosal defect, including the perforation, was closed completely, the clinical course was significantly better than that of incomplete closure and was equivalent to that of cases without perforations. Moreover, incomplete closure cases with an endoscopic nasobiliary and pancreatic duct drainage tube (ENBPD tube) could be managed without additional intervention. However, one of eight cases with neither complete closure nor an ENBPD tube required percutaneous drainage for an abscess. An ENBPD tube might prevent mucosal defect exposure to pancreatic and bile juices and help to avoid a worse clinical course in a case without complete closure of the mucosal defect.

Delayed perforation is a more severe complication than intraoperative perforation and requires emergency surgery. A meta-analysis revealed that the incidence of delayed perforations in a closure group tended to be lower than that in an unclosed group, although it was not statistically different (1.6% vs. 3.8%, p=0.13; risk ratio, 0.39; 95% CI, 0.12–1.32).107 Complete closure of the mucosal defect might help to reduce the risk of delayed perforations; however, it is sometimes difficult to achieve complete closure for a lesion in the medial or anterior wall and in lesion size >40 mm.108 In order to overcome these difficulties, various endoscopic closure techniques (the clip with string method, endoloop and endoclip closure method, OTSCs, and PGA sheets) were developed.109-113 Moreover, external drainage of bile and pancreatic juices using an ENBPD tube might be a therapeutic option for preventing delayed perforation in cases of difficult complete mucosal defect closure. Fukuhara et al.114 reported that no patients developed delayed perforation (0/21) with immediate insertion of ENPBD tubes after duodenal ESD, whereas 4.1% (2/49) of the patients who did not undergo ENBPD tube insertion developed delayed perforations. It should be noted that ENBPD has a risk of post-endoscopic retrograde cholangiopancreatography pancreatitis and mucosal defect perforation when inserting a side-viewing endoscope. Therefore, the indications for ENBPD should be limited to cases in which endoscopic closure is impossible or difficult.

The incidence of delayed bleeding in cold polypectomy, UEMR, conventional EMR, and ESD are 0% to 0.5%, 0% to 2.1%, 0% to 2.6%, and 4.7% to 5.2%, respectively.98-105 A meta-analysis revealed that the incidence of delayed bleeding was significantly lower in a closure group than in an unclosed group (2.0% vs. 17.3%, p<0.01; risk ratio, 0.14; 95% CI, 0.06–0.33).107 Simple endoscopic clip closure is the standard method for closing mucosal defect after ER of duodenal neoplasms. However, it is technically challenging to achieve complete closure of extensive mucosal defects, particularly after ESD. Thus, various endoscopic closure techniques (the clip with string method, endoloop and endoclip closure method, OTSCs, and PGA sheets) are reported to prevent delayed bleeding after ER of duodenal neoplasms (Fig. 2).109-113 These studies are single-arm descriptive studies or case reports with small samples. A clinician can determine a preferred method based on procedural difficulty and device availability and cost. A database analysis using propensity-score matching demonstrated that vonoprazan significantly reduced the incidence of delayed bleeding compared with PPIs following duodenal ESD.115 However, it should be noted that this study did not consider the effect of endoscopic preventive procedures after duodenal ESD.

In summary, endoscopic closure of post-ER mucosal defect is highly recommended for preventing delayed bleeding. The indications for ER of duodenal neoplasms should be determined based on the technical difficulty of defect closure, as well as resection.