Single Center and Multicenter Analysis of AEs in POEM

Haito-Chavez published in 2017 an international multicenter study on adverse events in association with POEM performed in a total of 1826 patients at 12 tertiary care academic centers between 2009 and 2015 [8]. All authors were expert endoscopists and pioneers in the field of POEM. They found 156 AEs occurring in 137 of 1826 patients (7.5% of patients). Mild, moderate, and severe AEs had a frequency of 116 (6.4%), 31 (1.7%), and 9 (0.5%), respectively. An AE was defined as any symptomatic event related to the POEM procedure itself or to anesthesia, requiring temporary stop of the procedure and/or further action to solve the event and/or to treat the symptoms [8]. Any event that prevented completion and/or resulted in prolongation of hospital stay required another procedure, or subsequent medical consultation was considered as AEs as well. The ASGE lexicon’s severity grading system was used to grade the AEs [37]. Incidental findings of capnoperitoneum, capnothorax, or capnomediastinum on post-procedure imaging and subcutaneous emphysema were not considered AEs. The authors included different multivariate analyses to find out predictors for AEs. They analyzed factors related to the patient including age, gender, Charlson comorbidity index, American Society of Anesthesiologists (ASA) class, history of antiplatelet or anticoagulation, immunosuppression drug or steroid use, and previous therapies including botulinum toxin injection, pneumatic dilation, and LHM. There was no significant association between these patient-related predictors and occurrence of AEs.

Multivariate analysis demonstrated that sigmoid-type esophagus (odds ratio (OR) 2.28, p = 0.05), endoscopist experience <20 cases (OR 1.98, p = 0.04), use of a triangular tip knife (OR 3.22, p = 0.05), and use of an electrosurgical current different than spray coagulation (OR 3.09, p = 0.02) were significantly associated with the occurrence of AEs [8].

The most common time of presentation of AEs was intraprocedural in 89 patients (57.1%). A total of 64 (41.0%) AEs presented during the first 48 h, and only 3 (1.9%) AEs presented after 48 h. The most common AEs that presented during the first 48 h were esophageal leak (n = 13), submucosal hematoma (n = 10) (Fig. 19.1a, b), and pneumonia (n = 8). A total of 51 (2.8%) inadvertent mucosotomies occurred, mostly closed by clips (Fig. 19.2a, b). Only three AEs occurred after 48 h. There was one case of empyema requiring thoracotomy and chest tube insertion. The two remaining cases were one patient with pneumonia and one patient with delayed bleeding, both of whom were treated conservatively [8].

As discussed most of the AEs were graded as mild in 116 (6.4%), followed by moderate and severe in 31 (1.7%) and 9 (0.5%), respectively.

Among the nine severe AEs, two were esophageal leaks, two bleeding episodes during tunneling (one resulted in conversion to LHM and one resulted in intensive care unit admission), one perforation, one aspiration pneumonia, one empyema, one capnomediastinum, and one severe cardiac arrhythmia. There were two patients with heavy bleeding during tunneling; one patient with secondary bleeding could not be managed endoscopically and required balloon tamponade with a Sengstaken–Blakemore tube. The second patient experienced intraprocedural bleeding with extensive submucosal hematoma that rendered completion of POEM impossible. LHM was performed successfully during the same session [8].

Among the 13 patients who presented with esophageal leak, there were two with severe esophageal leaks; one of them required surgery (washout surgery and drainage), while the second patient was treated with endoclipping. However, this latter patient progressed with a pleural effusion requiring insertion of a chest tube and then progressed with empyema requiring thoracotomy and drainage.

Overall inadvertent mucosotomy was the most common intraprocedural AE occurring in 51 patients, followed by insufflation related AEs in 28 patients (22 capnoperitoneum, 4 capnothorax, 1 pneumothorax, and 1 capnomediastinum), and bleeding during tunneling in 6 patients [8].

Other successful treatment of the 13 esophageal leaks included stent placement (n = 2) and endoscopically assisted vacuum therapy (n = 1). Three patients presented with contained leak into the submucosal tunnel and responded to conservative management.

Zhang and Zhou et al. presented their retrospective single-center analysis on only major perioperative adverse events (mAE) in 1680 patients who underwent POEM between August 2010 and July 2015 at Zhongshan Hospital, Shanghai, China [38]. They identified a total of 55 patients experiencing major adverse events (3.3%): they found delayed mucosal barrier failure (n = 13; 0.8%), delayed bleeding (n = 3; 0.2%), hydrothorax (n = 8; 0.5%), pneumothorax (n = 25; 1.5%), and miscellaneous (n = 6; 0.4%). Four patients (0.2%) required ICU admission. No surgical conversion occurred, and 30-day mortality was zero. In stepwise multivariate regression, institution experience of <1 year (odds ratio [OR] 3.85; 95%CI 1.49–9.95), air insufflation (OR 3.41; 95%CI 1.37–8.50), and mucosal edema (OR 2.01; 95%CI 1.14–3.53) were identified as related risk factors. After introducing CO₂ insufflation, the major Adverse Event rate declined to 1.9% (95%CI 1.2–2.7%) and seemed to plateau after 3.5 years at ~1%. The authors concluded that POEM appeared to be a safe procedure. Major adverse events were rare and could usually be managed effectively.

CO₂-Associated Problems and Anesthesiologic Considerations

Already in early series, the need for CO₂ insufflation instead of room air during POEM became evident [46, 47]. CO₂ may inadvertently track into surrounding tissues during POEM, causing systemic CO₂ uptake and tension capnoperitoneum. This in turn may affect cardiorespiratory function. Gas-associated AEs include also pneumomediastinum, subcutaneous emphysema, and pneumothorax. In a meta-analysis of Akintoye et al., subcutaneous emphysema was found in 7.5%, pneumothorax in 1.2%, pneumomediastinum in 1.1%, and pneumoperitoneum in 6.8% [48]. Important guiding parameters indicating the need for an intervention were significant abdominal distension, increased end-tidal CO₂ and increased peak airway pressure [40]. In cases of tension pneumoperitoneum, a Veress needle (or a 16–18 G intravenous cannula) is inserted through the abdominal wall para-umbilically respecting sterile conditions [15]. A 10–20 ml syringe is filled with saline and connected with the canula, and the plunger is removed. The appearance of bubbles shows a successful drainage of the capnoperitoneum. CO₂ is absorbed about 300 times faster than room air. Only gas-related events requiring an intervention should therefore be categorized as adverse events [15].

The endoscopist should try to reduce the CO₂ gas flow to the necessary minimum. The use of a low-flow CO₂ gas tube has been described helpful in this regard. In case of a pneumothorax with a volume of more than 30%, a thoracic drainage should be introduced for 2 or 3 days. In the rare case of capno-pericardium, a cardiac arrest may occur the way that anesthetists and endoscopists should be aware of this rare but possible complication [49].

Close anesthesiologic supervision of changes in airway pressures and hemodynamics are recommended, and an arterial line for monitoring of arterial blood gases can be considered [15, 50]. Important guiding parameters indicating the need for an intervention include significant abdominal distension, increased end-tidal CO₂, and peak airway pressure. Increasing minute ventilation is usually enough to manage an increase in end-tidal CO₂ levels associated with CO₂ insufflation [40]. Loeser et al. analyzed 173 consecutive POEM patients of a tertiary care single center in Germany over a 4-year period from an anesthesiologic standpoint [50]. During POEM, cardiorespiratory parameters increased from baseline: pmax 15.1 vs 19.8 cm H₂O, etCO₂ 4.5 vs 5.5 kPa [34.0 vs 41.6 mmHg], MAP 73.9 vs 99.3 mmHg, and HR 67.6 vs 85.3 min(−1) (p < 0.001 for each). Hyperventilation [MV 5.9 vs 9.0 L.min(−1), p < 0.001] was applied to counteract iatrogenic hypercapnia. Individuals with tension capnoperitoneum are treated with percutaneous needle decompression (PND; n = 55). They had higher peak pmax values [22.8 vs 18.4 cm H₂O, p < 0.001] than patients who did not require PND. After PND, pmax [22.8 vs 19.9 cm H₂O, p = 0.045] and MAP [98.2 vs 88.6 mmHg, p = 0.013] decreased. Adverse events included pneumothorax (n = 1), transient myocardial ischemia (n = 1), and subcutaneous emphysema (n = 49). The latter precluded immediate extubation in eight cases. Postanesthesia care unit (PACU) stay was significantly longer in individuals with subcutaneous emphysema than in those without (p < 0.001). The authors concluded that carbon dioxide insufflation during POEM produced systemic CO₂ uptake and increased intra-abdominal pressure. Changes in cardiorespiratory parameters included increased pmax, etCO₂, MAP, and HR. Hyperventilation and percutaneous abdominal needle decompression helped to mitigate some of these changes. Subcutaneous emphysema was common in 28.3% of cases and did delay extubation and prolong PACU stay.

Bleeding

Bleeding is a common side effect during any of the different steps of POEM, especially during submucosal tunneling (Figs. 19.1, 19.2, 19.3, and 19.4). Careful stepwise dissection will allow vessels to be visualized and to be prophylactically treated using cautious coagulation with the electrocautery knife itself or by means of a “Coag Grasper” (Olympus, Center Valley, PA, USA) using “Soft Coag” or low wattage “Forced Coag” current. Caution should be applied in case bleeding originates from a vessel running along the mucosal surface side of the tunnel in order to prevent secondary mucosal defects and perforation after coagulation. A gentle compression with the tip of the endoscope +/− cautious secondary coagulation is carried out in these cases. The placement of clips in the tunnel is usually avoided as secondary perforation of the covering mucosa should be feared.

Guidelines recommend to perform POEM without anticoagulant or antiplatelet therapy except for acetylsalicylic acid. It is recommended that all patients should have a blood type and antibody screening before starting the procedure [51, 52]. Postoperative bleeding apparently is infrequent. In a large series of Li et al. with 428 patients, delayed bleeding has been reported in 0.7% [53].

Secondary bleeding into the tunnel is infrequent (Fig. 19.5a, b). However, a massive hematoma in the tunnel can result in pressure necrosis of the mucosal flap with potentially disastrous consequences in case of wide perforation. A CT scan should be performed to discriminate a mere bleeding into the tunnel from additional mediastinal effusion. Li et al. reported on three patients (0.7%, 3/428) who experienced delayed bleeding in the submucosal tunnel after POEM. None of these patients had any predisposing factor to bleeding, such as hypertension, coagulation disorders, and antiplatelet/anticoagulant therapy before undergoing POEM. There were no special difficulties related to tunnel creation or myotomy performance in these cases. In one patient, a small hematoma was observed by CT before any clinical manifestation occurred; this patient then reported progressive serious retrosternal pain from the first day after surgery and vomited fresh blood on the third day. Two other patients suddenly vomited large amounts of fresh blood on the first and third days after the intervention, respectively; no submucosal hematoma was observed on CT scans before hematemesis occurred in these two patients. Emergency esophago-gastroscopy was performed immediately on all three patients, revealing a hematoma in the submucosal tunnel. After removing the metal clips from the mucosal entry, a large quantity of blood clots were discovered inside the submucosal tunnel and were removed. In the first patient, the bleeding source could not be identified, and a Sengstaken–Blakemore tube was directly placed into the stomach and lower esophagus to compress the bleeding sites. In the other two patients, active bleeding points were identified and coagulated with a hemostatic forceps in the forced coagulation mode. Almost all of the bleeding spots were from the cut muscular edges. A PPI, antibiotics, and hemocoagulase were administered to all three patients. Intermittent balloon deflation was performed every 24 h. The Sengstaken–Blakemore tube gastric balloon was permanently deflated on the first day after placement, and the esophageal balloon was deflated on the second day after insertion.

Benech et al. from Lyon, France, reported on successful conservative management [54]. The patient had experienced massive epigastric pain shortly after the procedure and showed a drop in hemoglobin from 14.2 to 11.2 g/dl. We had a similar case, managed conservatively (Fig. 19.5a, b).

Perforation

After dissection of the muscular layer, even a small mucosal defect can become potentially dangerous. In case such a mucosotomy is detected during submucosal tunneling, closure should be performed immediately as otherwise a significant increase of the defect may occur (Fig. 19.6a, b) [38]. Preoperative edema of the mucosa is suggested a risk factor for mucosal injury during intervention. Mucosal edema makes closure difficult and promotes perforation. Edema has been seen in 8% of patients in a retrospective study of over 1600 patients [38]. The endoscopic tunnel should be created very close to the muscular layers to avoid injury to the mucosal flap and because of a lower vascularity adjacent to the muscle [55]. Most perforations happen at the level of the lower esophageal sphincter due to a narrowing at the cardia. If a mucosotomy is identified, it should be closed immediately with endoscopic clips. Larger mucosotomies have been closed using a flexible endoscopic suturing device (OverStitch; Apollo Endosurgery, Austin, TX, United States) [56, 57]. Other salvage techniques used included fibrin glue and over-the-scope clips (OTSCs; Ovesco, Tuebingen, Germany) [58, 59]. In case of multiple ruptures which cannot be clipped, a covered retrievable stent may be used as rescue technique [60, 61].

Postprocedural Chest Pain

The most common periprocedural side effect is substernal chest pain. Data suggest an average mild to moderate chest pain after the procedure and during the following 3 days (4.6/10 immediately after POEM, 3.2–3.3/10 the following 2 days) [40]. As in tubular esophageal ESD, the application of a fentanyl patch, adapted to patients weight, age, and general condition, e.g., 25 mcg/g (12.5–50 mcg/h), applied at the beginning of the procedure, has been very valuable in our own experience over the last 5 years.

Infections and Pneumonia

In general index gastroscopy should be performed one to several days before the POEM procedure. In case signs of Candida esophagitis, a systemic antifungal treatment should be initiated immediately. Remaining material in the lower esophagus should be removed, and the patient is set on a strictly liquid diet 24–48 h before treatment. Single-shot antibiosis of, e.g., ceftriaxone plus metronidazole, is usually sufficient in a non-immunocompromised patient.

Sterility is still under debate as the endoscope is penetrating into a space in direct contact with the mediastinum and abdominal cavity. On the other hand, infectious complications have been reported less frequent as feared in the initial era of procedure [46, 47]. As a routine, we remove the endoscope with sterile gloves from the washing machine after reprocessing it shortly before the procedure. The same is done if a drying cabinet is used for storage. It is then placed into a tray with a sterile cloth inside and covered which a second sterile cloth until its use for the procedure. The use of a sterile coat and sterile gloves is recommended for the procedure [46, 47]. However, this practice varies from center to center and many units perform POEM with the endoscope processed and handled as for any other upper endoscopy. Single centers ask the patients to flush the mouth with chlorhexidine solution before the intervention [62].

Pleural Effusion

Pleural effusion is noticed in 5–40% of POEM patients. Depending on the size of effusion, laboratory findings plus clinical signs of infection (fever, etc.), antibiotics and early pleural drainage or just waiting for spontaneous absorption is indicated [42].

Reflux After POEM and LHM

The most common long-term adverse event with POEM seems to be gastroesophageal reflux (GER). As the premise behind the POEM procedure, similar to Heller myotomy, is to decrease lower esophageal sphincter pressure, it is not surprising that post-POEM GER is encountered [63]. Early studies were focused on technical feasibility and safety, with a short duration of follow-up. Furthermore, a large proportion of the early literature came from Asia, where GER is less prevalent. Finally, the consequences of asymptomatic or proton pump inhibitor (PPI)-responsive GER after POEM had not been clear at the time.

When objective data are reviewed, such as erosive esophagitis in EGD and/or an abnormal acid exposure on a pH study, the prevalence of GER after POEM appears to be in recent studies high and varies between 20% and 46% after POEM [51, 63–65]. Barrett’s metaplasia has been reported in first few cases as found earlier after Heller myotomy [66, 67].

In patients with a hiatal hernia, the risk for erosive esophagitis and GERD post-POEM seems increased [68]. If the rates can be compared to those seen with Heller myotomy plus partial fundoplication had been long time contradictory [69–71]. Kumbhari et al. note that when Heller myotomy was first introduced, it was not combined with an anti-reflux procedure and initially not deemed necessary [72]. Subsequently a high rate of GERD became evident, and a partial fundoplication became standard practice [70, 73, 74].

Kumbhari et al. analyzed results from seven tertiary academic centers (one Asian, two US, four European). POEM had been carried out in 467 patients during the 5-year study period. A total of 282 patients were included in the analysis. One hundred eighty-five patients were excluded because no pH study was performed at ≥3 months after POEM. A post-procedure DeMeester score of ≥14.72 was seen in 57.8% of patients. Multivariable analysis revealed female sex to be the only independent association (odds ratio 1.69, 95% confidence interval 1.04–2.74) with post-POEM GER. No intraprocedural variables were associated with GER. Upper GI endoscopy was available in 233 patients, 54 (23.2%) of whom were noted to have reflux esophagitis (majority Los Angeles grade A or B). GER was asymptomatic in 60.1%. The authors concluded that post-POEM GER was seen in the majority of patients. No intraprocedural variables could be identified to allow for potential alteration in procedural technique.

Repici et al. published a meta-analysis on gastroesophageal reflux disease after POEM as compared with laparoscopic Heller’s myotomy plus fundoplication published until February 2017 [65]. They identified 17 and 28 prospective studies, including 1542 and 2581 subjects who underwent POEM and LHM, respectively. Pooled rate of post-procedure reflux symptoms was 19.0% (95% CI, 15.7–22.8%) after POEM and 8.8% (95% CI, 5.3–14.1%) after LHM, respectively. Pooled rate estimate of abnormal acid exposure at pH monitoring was 39.0% (95% CI, 24.5–55.8%) after POEM and 16.8% (95% CI, 10.2–26.4%) after LHM, respectively. Rate of post-POEM esophagitis was 29.4% (95% CI, 18.5–43.3%) after POEM and 7.6% (95% CI, 4.1–13.7%) after LHM. At meta-regression, heterogeneity was partly explained by POEM approach and study population. They concluded that the incidence of reflux-disease appears to be significantly more frequent after POEM than after LHM with fundoplication. pH monitoring and appropriate treatment after POEM should be considered in order to prevent long-term reflux-related adverse events [65].

However, long-term results after LHM indicate that the antireflux effect of the fundoplication might only be of temporary nature. In their editorial, Rosch et al. asked the question “Will Reflux Kill POEM?” [66]. Rosch discusses that only one small randomized controlled trial (n = 43) has been published showing reflux rates of 9.1% versus 47.6% in the groups of Heller myotomy with and without Dor fundoplication, respectively [73]. Kummerow Broman et al. published the long-term symptomatic follow-up results on part of this group in 2018 [75]. They collected patient-reported measures of dysphagia and gastroesophageal reflux using the Dysphagia Score and the Gastroesophageal Reflux Disease-Health-Related Quality of Life (GERD-HRQL) instrument. Patient-reported re-interventions for dysphagia were verified by obtaining longitudinal medical records. Among living participants, 27/41 (66%) all completed the follow-up study at a mean of 11.8 years postoperatively. Median Dysphagia Scores and GERD-HRQL scores were slightly worse for Heller than Heller plus Dor but were not statistically different (6 vs 3, p = 0.08 for dysphagia; 15 vs 13, p = 0.25 for reflux). Five patients in the Heller group and six in Heller plus Dor underwent re-intervention for dysphagia with most occurring more than 5 years postoperatively. One patient in each group underwent redo Heller myotomy and subsequent esophagectomy. Nearly all patients (96%) stated that they would undergo operation again. The authors concluded that long-term patient-reported outcomes after Heller alone and Heller plus Dor for achalasia were comparable, providing support for either procedure [75].

There is no consensus on how to manage patients with symptomatic gastroesophageal reflux disease, but a primary attempt with low-dose PPIs seems to work well for most patients [8, 10, 32, 38–45]. In case of the necessity of a secondary fundoplication only a partial or “floppy” fundoplication is recommended in order to not impair esophageal emptying with secondary dysphagia again [8, 10, 32, 38–45]. Kumta et al. even reported one case of endoscopic fundoplication in an patient with gastroesophageal reflux symptoms refractory to proton pump inhibitors [76].

Requirements to Perform POEM

The first step for a “POEM learner” is an excellent knowledge of the specific thoracic and abdominal anatomy and the different steps of the procedure [77]. The second step is usually an “ex vivo” and “in vivo” training in the porcine model similar to ESD training [78]. The first clinical POEM cases in patients should be accompanied by an expert endoscopist from an external POEM referral center [79].