The main technical challenge of ESD is control of the specimen during dissection from the underlying muscle layer. As an alternative to the standard technique, von Delius et al. reported “endoscopy of the submucosal space” in 2007, in which they reversed the order of the mucosal incision and division of submucosal fibers [24]. In a pig model, they entered the submucosal space through a mucosotomy and created a tunnel under the lesion. Once the mucosa had been completely separated from the underlying muscle, they completed the resection by post-cutting the mucosa and were able to successfully resect lesions of various sizes, including a complete circumferential donut.

Overall, procedure-related bleeding and perforation were each less than 3% over 38 studies and 2223 lesions. There were no ESD-related mortalities [1].

In three separate meta-analyses of 1500 or more lesions each, procedure-related bleeding ranged from 4 to 7%, while perforation ranged from 4 to 5%. There were no mortalities [17, 25, 26].

A systematic review by Repici et al. of 2841 lesions found an overall procedure-related bleeding rate of 2% and perforation rate of 4%. There were no mortalities [27].

In a pooled analysis of 38 studies and more than 2200 lesions (including 970 squamous cell carcinoma, 346 adenocarcinoma arising in Barrett’s esophagus, 678 squamous cell + adenocarcinoma, and 185 submucosal tumors), the en bloc resection rate ranged from 81 to 100%, with an overall average of 96%. The R0 resection rate was 85% with a local recurrence of 0.4% [1].

In three meta-analyses of 1495, 1734, and 1916 lesions, the en bloc resection rate was 92%, with an R0 resection rate of 82–92%, and local recurrent of <1% [17, 25, 26].

A systematic review by Repici et al. of 2841 lesions found an overall en bloc resection rate of 96%, R0 resection rate of 88%, and local recurrence of <0.1% [27].

POEM is generally performed using a standard-sized gastroscope with addition of an auxiliary water jet (e.g. Olympus GIF-Q260J, Tokyo, Japan) and a distal cap (FujiFilm DH-28GR or Olympus MH-588) to help maintain a clear operative field and facilitate dissection. Low- or medium-flow carbon dioxide insufflation is used due to a higher risk of complications with room air or high-flow CO₂ [41, 42].

Patients are generally placed on a liquid diet in the days prior to the procedure to minimize the residual contents and treated empirically with an antifungal due to the high rate of esophageal stasis in achalasia patients. Most centers administer perioperative antibiotics and proton pump inhibitors, and upper endoscopy is often performed immediately prior to induction of anesthesia to suction any esophageal contents and reduce the risk of aspiration.

POEM is performed under general anesthesia with a cuffed endotracheal tube, which may help protect against aspiration and may reduce the incidence of capnothorax by providing positive intrathoracic pressure. Some cases have been reported using conscious sedation, but this resulted in longer procedure times and a higher rate of complications such as bleeding, perforation, and pneumothorax [41].

The procedure can be performed in the left lateral decubitus position, though anecdotal evidence suggests this may exacerbate anatomic distortions in patients with advanced sigmoid achalasia. The supine position minimizes distortions. It also allows for monitoring of tense capnoperitoneum, which may occur in 16% or more of cases, and facilitates needle decompression if necessary [43].

Inspection begins in the proximal esophagus, where extrinsic compression from the trachea (anterior, 12 o’clock), left main bronchus, aortic arch (anterolateral), and spine (posterior, 6 o’clock) can often be identified and used to maintain orientation. There may be a tight area with resistance to passage of the endoscope just proximal to the EGJ, and the proximal esophagus may demonstrate tertiary contractions.

The location of the EGJ is noted by measuring the distance from the incisors. Many patients will demonstrate a tight area just proximal to the EGJ, and there may be slight resistance to passage of the endoscope. Normal physiologic tightness will often be noted at the upper esophageal sphincter.

The point of entry depends on the planned myotomy length. The mean length of the total myotomy (esophageal + gastric) in published series ranges from 5.4 to 14.4 cm [41, 44]. The entry point is generally chosen 10–15 cm proximal to the EGJ, as first described by Inoue et al. [38], which allows the submucosal tunnel to extend 2–3 cm proximal to the myotomy and may protect against full-thickness perforation.

The procedure can be performed at any “clock” position, with no studies yet demonstrating a clear advantage of one location over the others. The anterior approach, as utilized in the first 500 cases by Inoue et al., avoids the gastric sling fibers and may reduce the risk of post-POEM reflux, but at the expense of an increased risk of major procedural bleeding from branches of the left gastric and left phrenic arteries [38, 39, 45]. The posterior approach avoids some of the larger blood vessels and preserves the anterior anatomy for a straightforward surgical myotomy if indicated in the future, but may disrupt the gastric fling fibers, and theoretically increase the risk of post-POEM reflux. Greater curvature myotomy has also been described, but is technically more challenging [46]. The angle of His serves as a consistent landmark (Fig. 3.4), which may be particularly useful in cases of distorted anatomy from advanced sigmoid achalasia, severe fibrosis from the previous myotomy, ESD, or repeated pneumatic balloon dilations, or inflammation from fungal esophagitis.

After the location for the mucosotomy is chosen, a “saline lift” is performed with an injection needle to expand the potential space between the mucosa and the circular muscle fibers (Fig. 3.5a). This protects against unintended full-thickness myotomy, facilitates entry of the endoscope into the submucosal space, and, if indigo carmine or methylene blue dye is used, helps to delineate the anatomy. Some centers also include epinephrine in the solution, though there is a risk of cholinergic side effects [45].

The mucosa is incised in a longitudinal direction using a cutting electrical current (Fig. 3.5b), and the submucosal fibers are dissected using spray coagulation. Creation of the submucosal tunnel can be performed using the Triangle Tip (TT) Knife (Olympus KD-640L) (see Fig. 3.5c). Multiple submucosal injections can aid in dissection. Alternate knives such as the Hybrid Knife (ERBE 20150-060, Tübingen, Germany) may reduce the need for multiple instrument exchanges and reduce the average procedure time [47]. The HookKnife™ can facilitate dissection when scarring or inflammation is encountered. During dissection, small perforating vessels may be controlled using the dissecting knife, while larger vessels can be controlled using a coagulating forceps (Coagrasper, Olympus America).

The esophageal myotomy can be performed in anterograde or retrograde direction, with no difference in outcomes (Fig. 3.5d) [48]. The myotomy extends from 2 to 3 cm distal to the mucosal incision. This leaves a short segment of intact mucosa overlying intact muscle, protecting against full-thickness perforation in the event of mucosal closure dehiscence.

There is no clear superiority of either selective circular or full-thickness myotomy. The full-thickness approach is most similar to the surgical myotomy, in which both longitudinal and circular muscle fibers are divided; however, in an international POEM survey, only one-sixth of centers preferred the full-thickness myotomy [40]. Clinical outcomes, complication rates, and rates of post-POEM reflux appear to be similar regardless of whether both muscle layers are transected [49, 50].

The main risk factor for clinical failure of POEM is incomplete myotomy on the gastric side. A gastric myotomy length of 2–3 cm is recommended. Endoscopic landmarks of the gastric side include narrowing followed by widening of the submucosal tunnel, identification of palisade vessels (longitudinally arranged vessels that are characteristic of the gastroesophageal junction), and blue discoloration of the gastric mucosa on retroflexed view in the true lumen [40]. A number of adjuncts have been developed to ensure a complete gastric myotomy, particularly when endoscopic landmarks are likely to be inaccurate, as in cases of distorted anatomy, scarring from prior procedures, or prior esophagitis.

One technique, first described by Baldaque-Silva et al, utilizes two endoscopes to ensure adequate length of the submucosal tunnel, placing one endoscope in the tunnel and observing the transillumination with the other endoscope in retroflexed view of the gastric cardia [51]. A prospective, randomized study involving 100 patients found that the use of this technique resulted in a significant increase in average gastric myotomy length [52]. Further advantages included minimal increase in procedure time, no increases in the rate of complications, and no need for specialized equipment or additional training.

Alternate techniques include placement of a radiopaque clip to mark the EGJ followed by fluoroscopy to measure the distance from the clip to the tip of the endoscope, or the use of the EndoFLIP device to measure EGJ distensibility and guide intra-operative decision making [53–56].

After confirming adequate myotomy length and ensuring hemostasis, the mucosa is closed with hemostatic clips in the majority of cases (Fig. 3.5e). Some authors advocate the use of an antibiotic solution, which is flushed through the endoscope, prior to exiting the submucosal tunnel. Addition of endoloops, endoscopic sutures, over-the-scope clips, or fully covered metal stents may be required if the mucosa is inflamed, macerated, or otherwise difficult to close.

Water-soluble contrast esophagram is obtained on POD #1 to exclude a leak. As many as one-third of patients may demonstrate delayed emptying in the early postoperative period, but this does not appear to correlate with treatment failure in the long term, which calls into question the true value of a contrast esophagram aside from ensuring there is not a full-thickness leak [57]. Some centers also perform upper endoscopy to assess for the development of mucosal necrosis, submucosal hematoma, or dislodgement of the hemostatic clips with dehiscence of the mucosal closure. If partial thickness mucosal necrosis or submucosal hematoma is present, oral intake is delayed until resolution can be confirmed.

Postoperative CT scans are likely to demonstrate nonspecific inflammation or collections of gas, which are considered normal postoperative findings. CT is not recommended in asymptomatic patients [58].

Clear-liquid diet is resumed on POD #1, with advancement to pureed diet on POD #2–3, and regular diet as early as POD #4. An oral PPI is prescribed for at least 1 month, though some centers continue indefinitely [39]. Patients may be discharged home as early as POD #1 [59, 60].

We conduct the initial follow-up visit at 2 months postoperatively and may include upper endoscopy, high-resolution manometry, and timed barium study in the evaluation. Some centers also include an esophageal pH study. If the PPI has been discontinued, it is resumed in the presence of subjective reflux symptoms or endoscopic findings of esophagitis. Subsequent follow-up is conducted at 1 year postoperatively and then annually thereafter.

Events related to insufflation are relatively common, with rates as high as 30% capnoperitoneum, 11% capnothorax, 5% mediastinal emphysema, 36% subcutaneous emphysema, and one case report of tension capnopericardium (personal communication). Only 8% of patients with capnoperitoneum and 3% of patients with capnothorax require decompression, however [43, 61]. Rates also appear to be technique-dependent, with the use of air insufflation or high-flow CO₂ insufflation resulting in higher rates of adverse events than low- or medium-flow CO₂ insufflation [41, 42]. Capnothorax, capnomediastinum, and capnoperitoneum are generally self-limited. Tense capnoperitoneum may result in increased end-tidal CO₂ or increased ventilator peak pressures, with a theoretical risk of abdominal compartment syndrome. When abdominal decompression is necessary, a large-gauge angiocatheter or Veress needle can be used to decompress the peritoneal cavity. The case of tension capnopericardium required a brief period of chest compressions; however, the patient recovered without complications.

Minor procedural bleeding is common and can generally be controlled using a hemostatic forceps or the knife with a coagulating current. Compared to standard dissection, the Hybrid Knife may reduce the number of minor bleeding episodes [63]. Only one case of severe bleeding has been reported early in the POEM experience; this was controlled with hemostatic forceps, and there have been no reports of procedural bleeding that could not be controlled endoscopically [39].