Intra-procedural bleeding as a result of EMR or ESD can be managed by a variety of hemostatic techniques. Epinephrine injection in 1:10,000 dilution (or higher) can be injected in 1–2 ml aliquots to stop active bleeding, although it does not prevent delayed hemorrhage. If active bleeding occurs immediately after completion of lesion resection, placement of clips targeting the bleeding point is preferable to contact thermal coagulation (e.g., bipolar coagulation) since the former does not extend tissue injury. However, clip placement should be avoided if the EMR, for example, is not complete since the clips may interfere with subsequent resection of residual lesion and serve as a current conductor to deeper tissue layers during inadvertent snare wire contact with the clips. The selection of a particular TTS clip for hemostasis is primarily dependent upon device availability and operator preference since there are no prospective comparative trials demonstrating the superiority of one clip over another in the setting of intra-procedural iatrogenic bleeding. The use of a dedicated hemostatic forceps (Coagrasper, Olympus Corp., Tokyo, Japan) is ideal for hemorrhage that occurs in the midst of an EMR or ESD procedure. The technique involves grasping, tenting, and applying coagulation current for sealing of the vessel (Fig. 19.7). The suggested settings for the Coagrasper are a power of 50 W and 1–2 s pulse duration using a soft coagulation mode (Video 19.1).

Snare resection of a pedunculated polyp with a thick stalk can result in active bleeding, which can immediately be controlled by recapturing and constricting the stump with the snare for several minutes. If bleeding resumes upon loosening the snare, epinephrine can be injected within the stump to slow or stop bleeding, followed by definitive therapy. If feasible, mechanical hemostasis is preferable to avoid extending thermal injury. Clips can be used to close the end of the stump. A detachable snare is also an option if the length and location of the residual stump is suitable for loop placement. Occasionally, access to the stump is difficult for either clip or loop placement, and a contact thermal probe is used with light–moderate contact pressure of 3–5 s to achieve hemostasis, with suggested settings of 15 J for the heater probe and 12–15 W for the bipolar probe.

In patients who require endoscopic intervention for delayed post-polypectomy bleeding, assessment of the post-polypectomy ulcer site partly dictates the need for therapy, although the significance and rebleeding rates of stigmata of recent hemorrhage (SRH) within post-resection ulcers are not as well studied as SRH associated with bleeding peptic ulcers. Post-polypectomy ulcers with clean bases or flat pigmented spots are not generally treated, whereas endoscopic therapy is performed for post-polypectomy ulcers with non-bleeding visible vessels (Fig. 19.8), adherent clots, or active bleeding. Clip placement and contact (coaptive) coagulation, with or without epinephrine injection, are commonly used modalities for delayed post-polypectomy hemorrhage. However, TTS clips may be ineffective if the ulcer base is quite indurated due to insufficient clip closure force. Also, a bleeding vessel entrenched in a fibrotic ulcer base may not be amenable to Coagrasper coagulation. The indurated base, however, provides a safety cushion for the use of coaptive coagulation, such as a bipolar probe, which may be more suitable in this setting (Fig. 19.9). The over-the-scope clip (OTSC^® , Ovesco Endoscopy AG, Tübingen, Germany) provides greater compression force and tissue capture than TTS clips (Video 19.2), although the use of this device requires endoscope removal to fit the OTSC, which may not be practical in some actively bleeding cases. The use of a hemostatic spray (Hemospray, Cook Medical Inc., Bloomington, Indiana, US) requires active bleeding, and its role in the management of delayed post-polypectomy bleeding is currently being defined.

Esophageal perforation is associated with significant morbidity and mortality, especially when management is delayed for >24 h [34, 35]. In one systematic review, esophageal perforation occurred in 56/3071 (1.8 %) patients with achalasia who underwent pneumatic balloon dilation, with an incidence rate ranging from 0 % to 5.4 % [36]. Recent data suggest that dilation is effective and relatively safe for the treatment of strictures associated with eosinophilic esophagitis. A meta-analysis involving 525 patients and a total of 992 dilations showed that perforation occurred in 3 patients only (0.3 %; 95 % CI: 0–0.9 %) [37].

Endoscopic clips are generally successful at closing esophageal perforations [38], particularly when the perforation size is <1 cm [30]. Unlike chronic fistulas, acute esophageal perforations generally heal with clip closure alone within 1 week. For larger perforations, SEMS is a potential option. In general, larger diameter (23–28 mm) covered SEMS are employed particularly when there is not a stricture of shelf to anchor the stent [39]. As previously mentioned, achalasia patients with a dilated esophagus (diameter >3 cm) may not benefit from stent placement due to the lack of adequate sealing between the stent and the esophageal wall [40]. In one retrospective study, esophageal leaks and perforations were closed in 77.6 % of cases using SEMS [39]. A partially covered SEMS may be placed to seal an esophageal perforation, especially when the stent crosses the gastroesophageal junction, to minimize the risk of stent migration. However, utilization of a partially covered SEMS may be hampered by tissue ingrowth and embedment at the uncovered flanges of the stent, requiring the stent-in-stent technique for its eventual removal [39]. As discussed above, a more attractive alternative is to place a fully covered SEMS with endoscopic suture fixation of the stent to the esophageal wall to prevent its migration (Fig. 19.15). One prospective study of 33 patients with esophageal perforation, including 19 iatrogenic perforations, found that temporary placement of SEMS of different types was successful in as many as 97 % of cases [41]. Of note, stent extraction was uneventful in all cases when performed within 6 weeks of insertion, whereas stent extraction was complicated in 50 % of cases when it was performed after 6 weeks [41]. It is suggested, therefore, that the stent dwell time should be less than 6 weeks. Stent migration ranges from 11.1 % to 33 % among various studies and, therefore, ongoing patient monitoring is required for signs of stent migration following placement [39–42].

In addition to EMR or ESD, endoscopic submucosal tunnel dissection (ESTD) has been pioneered for en bloc excision of larger lesions (>5 cm) [4, 43]. These advanced resection techniques may result in esophageal perforation, although such a complication can be managed successfully in the hands of a skilled endoscopist without resorting to surgery. It has been suggested that perforation is to be expected (or even sometimes intended) with these enhanced resection techniques, and it should not simply be considered as an adverse event in a controlled setting [30]. In one study of 306 ESD and 171 EMR performed to remove esophageal neoplasms in 368 patients, esophageal perforations occurred in 7 (1.9 %) cases. All perforated patients were male and had undergone ESD, while no perforation occurred in the EMR group [44]. Perforations occurred intra-procedurally in 3 cases, after stricture dilatation in another 3, and due to food bolus impaction in the remaining patient [44]. In another study, perforations occurred in 4/58 (6.9 %) patients during ESD and was successfully managed conservatively following perforation closure with endoscopic clips [38, 45].

The incidence of perforation was higher following ESD for Barrett’s-associated adenocarcinoma (20 %; n = 25) compared to either esophagogastric junction (2.9 %; n = 103) or non-junction squamous cancers (2.7 %; n = 1335) [46]. In contrast, two studies found no perforations following EMR performed in 102 patients [47] and following 2513 EMR procedures in 681 patients with neoplastic appearing lesions in Barrett’s esophagus [48]. In another study, perforations occurred in only 3/185 (1.6 %) patients, which were successfully managed with clips [45].

Pneumomediastinum without overt perforation may occur after esophageal EMR and ESD, with an incidence of 6 % and 10 %, respectively [47, 49]. In a study where systematic radiographic imaging was performed in 58 patients who underwent esophageal ESD, mediastinal emphysema was detected in 18 (31 %) patients by chest CT as opposed to only 1 (1.7 %) patient by chest x-ray. The ESD-induced exposure of the muscularis propria (n = 32) was the only significant risk factor [50]. In all the reported cases, pneumomediastinum promptly regressed with conservative treatment alone [47, 49, 50].

Although post-procedural bleeding in the esophagus is mainly attributed to endoscopic resection of Barrett’s esophagus or squamous cell dysplasia/early carcinoma, the risk appears to be small without an overall significant impact on patient outcome or procedural safety [51–53]. In an early series of 216 EMR for dysplastic Barrett’s esophagus [53], the risk of post-EMR bleeding occurred in 1 % of patients, which was successfully managed endoscopically in all cases. In a larger series encompassing 1060 resections in Barrett’s patients, the overall risk of delayed bleeding was 2.1 %, which was also effectively managed endoscopically. The same risk has been observed during the learning curve of this procedure with less experienced endoscopists [54]. EMR of squamous dysplasia or early carcinoma appears to carry a lesser risk of bleeding than Barrett’s lesions [55, 56]. ESD is extensively used for the treatment of squamous cell carcinoma, especially in Asia. ESD does not appear to result in a greater risk of post-procedural bleeding relative to EMR [49, 57–59]. The same appears to be true with regard to ESD for Barrett’s esophagus based on a small series [60].

Endoscopic TTS or OTS clip approximation and closure of small gastric perforations (<2 cm) secondary to EMR or ESD are generally effective. For large perforations, the omental patch method should be considered [29]. Alternatively, endoscopic suturing may be an option if the device is available and the defect readily accessible.

In a large single-center Japanese series, perforation occurred in 121/2460 (4.9 %) cases who underwent gastric EMR [61]. The first 4 patients were treated with emergent surgery and the subsequent 117 patients were treated with endoscopic clips. The latter treatment was successful in 115 (98.3 %) patients, and salvage surgery was required in 2 patients due to failed endoscopic closure [61]. In another study [29], a gastric perforation was encountered in 7/789 (0.88 %) patients following EMR, with the defect diameter ranging from 4 to 25 mm. These cases were successfully managed with clips (range 3 to 11), with the addition of an omental patch in the patient with the largest perforation.

A systematic review encompassing12 studies and 3806 gastric lesions evaluated the safety of ESD compared with EMR for the resection of early gastric cancers [62]. A significantly higher perforation rate occurred in the ESD group (4.54 %) compared with the EMR group (1.03 %), with an estimated increased risk of 3.58 (95 % CI: 1.95–6.55). The perforation rate did not significantly differ according to the type of ESD knife used. In those who perforated, surgical intervention was required more frequently in the ESD group (11.7 %) than in the EMR group (6.2 %), although the difference was not statistically significant [62].

Risk factors for ESD-related perforation were assessed in a large single-center study involving 1123 lesions [63]. Perforation occurred during ESD of 27 (2.4 %) lesions, and resection of lesions in the proximal stomach was the only significant risk factor (OR 4.88, 95 % CI: 2.21–10.75) on multivariate analysis. Surgical intervention was needed in one patient 5 days after an ESD-associated perforation due to dehiscence of the defect despite initial successful closure with endoscopic clips [64].

Both EMR and ESD procedures have been found to be relatively safe even in cirrhotic and elderly patients [8, 9]. In one systematic study, perforation occurred in only 1 (1.6 %) of 68 cirrhotic patients with gastric neoplastic lesions removed by ESD, which was successfully treated with endoscopic clips [8]. A comparative study found that post-ESD perforation rate was not increased in the elderly (14/372; 3.8 %) relative to non-elderly patients (4/143; 2.8 %) patients [9]. However, in those who perforated, emergency surgery was required in 14.3 % of elderly patients as opposed to none in the non-elderly group.

Similar to the esophagus, EMR and ESD of gastric lesions are the most frequent causes of post-procedural bleeding. A systematic review of 12 studies involving 3806 early gastric cancers compared the efficacy and safety of ESD (n = 1734) to EMR (n = 2072) [62]. The overall bleeding rate was 7 % for both procedures, though the rate of immediate bleeding was over twofold higher in the ESD group compared to the EMR group. On the other hand, the delayed bleeding risk was slightly lower in the ESD group, but the difference was not statistically significant. None of the patients in the ESD group underwent surgery due to delayed bleeding. A similar risk profile for post-procedural bleeding was noted in another meta-analysis study [65]. In cirrhotic patients, post-ESD bleeding occurred in 8/61 (13.1 %) patients, with successful endoscopic treatment in all cases [8]. This data support similar previous findings with regard to ESD for early gastric cancer in the setting of liver disease [66–68].