Introduction

Gastroparesis (GP) is defined as an objective delay in gastric emptying which can manifest with fullness, nausea, vomiting, bloating, and upper abdominal pain. It is often chronic and debilitating which is frustrating for both physician and patient alike as there are few therapeutic options that are considered effective. The incidence may also be rising as indicated by a sharp increase in gastroparesis-related hospitalizations since the year 2000 [1].

Diabetes mellitus (DM) is commonly associated with gastroparesis. In some studies looking at patients with diabetes and generalized upper abdominal symptoms, the incidence of gastroparesis was over 50% [2]. Other etiologies include gastrointestinal surgery, collagen vascular diseases, and neurological disorders. Many cases have no detectable underlying abnormalities and are considered idiopathic. Some reports suggest that idiopathic cases may be even more frequent than diabetic gastroparesis as a whole.

Traditionally, patients are managed with a step-up approach beginning with dietary modifications, control of psychological stressors, and medical treatment. Medical treatments, including promotility agents, are only modestly effective at controlling symptoms. Prokinetic agents, such as metoclopramide, do not consistently improve gastric emptying despite being shown to improve some of the symptoms [3]. In addition, they neither decrease the length of stay or readmission rate related to GP [4]. Metoclopramide, the only FDA-approved prokinetic agent in the USA, also carries a black box warning due to the risk of tardive dyskinesia. Other prokinetic agents such as erythromycin, a macrolide antibiotic, have well documented promotility effects but limited long-term benefit. They primarily act as motilin receptor agonists [5].

Interventional therapeutic modalities such as gastric electrical stimulation, laparoscopic pyloroplasty, or gastrostomy may be considered for patients with refractory symptoms. However, those interventions are either too invasive or not very effective. Recently, there has been a tremendous interest in minimally invasive gastric drainage procedures aimed at the pylorus, the most promising of which is the gastric peroral endoscopic myotomy (G-POEM) [6••].

Mechanisms of gastroparesis

Gastric motility is a complex sequence, comprised of many highly coordinated motor components. The phenomenon of delayed gastric emptying, therefore, is multifactorial and not entirely understood. Some factors identified that can contribute to gastric dysmotility include vagal and sympathetic nerve dysfunction, damage to the enteric nervous system, and loss of interstitial cells of Cajal (ICCs), which are considered the pacemakers of the gastric body [3]. These individual components may result in impaired fundic accommodation, antral hypomotility, gastric dysrhythmias, and pyloric compliance. It is unknown which of these elements, either alone or in combination, are the predominant problem in this disease.

In normal food processing, particles seem to cycle from the fundus to the body, with each successive pass reducing them in size. As these smaller particles enter the antrum, intrinsic motor mechanisms of the antrum push the food particles into the duodenum as the pylorus relaxes. ICCs play a distinct role in regulating gastric contractions of the fundus, body, and antrum through slow wave propagation. In GP, full thickness biopsies and histologic analysis have revealed a generalized loss of ICCs regardless of the etiology [7]. On the neural side, autonomic dysfunction and injury to the vagus nerve has been a well-associated mechanism of gastroparesis, especially in DM. This emphasizes the importance of acetylcholine and the response of the stomach to cholinergic stimuli. Vagus nerve dysfunction reduces pyloric relaxation and thereby prohibits passage of foods, which are effects similar to the consequences of subdiaphragmatic vagotomy. In addition, it also seems that nitronergic pathways exist which are dependent on nitric oxide (NO) as an inhibitory neurotransmitter. Similar to cholinergic stimulation, NO plays an important role allowing for pyloric relaxation.

The pyloric apparatus

The pylorus represents the last final obstacle for food particles to pass, and as such has a distinct role in gastric emptying and gastroparesis. There has been a specific interest in the pyloric apparatus in the delay of gastric emptying. Early on, Mearin et al. had shown that a large proportion of diabetic patients (> 50%) experienced episodes of unusually prolonged and intense contractions he termed “pylorospasm” [8]. Typically, antral-duodenal phasic waves need to be coordinated with relaxation of the pylorus. Normal relaxation of the pylorus is dependent, as are most smooth-muscle sphincters, on neuronal nitric oxidase synthase (nNOS)—the depletion of which is a well-described phenomenon in achalasia, resulting in hypertonicity and an impaired relaxation of the lower esophageal sphincter [9].

These types of functional manometric studies have been difficult to reproduce because the technique Mearin et al. described, required fluoroscopically placing a multilumen pneumohydraulic perfusion assembly with side ports across the antroduodenal junction and monitoring the patients for 5 h [8]. Recently, there have been advances in technology that have simplified this analysis. Devices, such as the endolumenal functional lumen imaging probe (EndoFLIP; Crospon Inc., Galway, Ireland), allow us to assess sphincter pressure, cross-sectional area, and distensibility separately. Making the distinction between sphincter pressure and distensibility is important, as distensibility represents compliance which seems to correlate better with symptoms in similar conditions such as achalasia and may better predict response to treatment as well [10].

Gourcerol et al. used the EndoFlip device to assess functional characteristics of the pylorus and whether they correlated with GP symptoms and emptying studies. Their results suggest that fasting pyloric compliance, and not simply the resting pressure, is decreased in GP. In addition, fasting pyloric compliance was correlated not only with gastric emptying but also with symptoms and quality of life impairment [11]. Similarly, Malik et al. used EndoFlip to assess the pylorus with similar findings and also showing that pyloric diameter and CSA were negatively correlated with symptoms of early satiety and postprandial fullness. There is greater variability amongst patients’ distensibility, but it was clear that low distensibility and a lower CSA were highly correlated with GP symptoms [12••]. While Malik et al. studied distensibility at various balloon volumes (range 20–50 mL), it seems that 40 mL allowed them to consistently reproduce the differences noted in their study. A specific cutoff of distensibility has not been established, but it appears that symptoms and GP findings seem to increase when distensibility is less than 10 mm2/mmHg.

Pyloric interventions

Considering the key role of the pylorus in gastric emptying, there has been a significant interest in pyloric interventions for relieving symptoms of gastroparesis. Some of the treatments aimed at the pylorus include dilation, botulinum toxin A injection, surgical pyloroplasty/pyloromyotomy, pyloric stenting, and more recently G-POEM. These bring the promise of a simple pyloric intervention to help patients with gastroparesis and perhaps avoid a more complex gastric drainage procedure.

Pyloric interventions are aimed at reducing obstruction at the level of the pylorus without affecting any remaining contractility the antrum may have in this disease. Therefore, the propulsive force towards the pylorus is not affected. A simple proof of concept is the use of botulinum toxin injected directly into the pyloric sphincter. Miller et al. first published on ten patients with idiopathic gastroparesis that received 80–100 U of botulinum toxin in four quadrants of the pylorus [13]. There was a significant decrease of solid gastric retention at 4 h from approximately 27 to 14% after intervention. Three patients had no response, but of the patients that did well, symptom improvement was maintained even at 6 months. Similarly, 20-mm balloon dilation of the pylorus seems to have a short-term effect on improving compliance and symptoms, but is not durable [11].

A more direct approach to relieving pyloric obstruction involves cutting the pyloric sphincter. The laparoscopic Heineke-Mikulicz pyloroplasty (LP) has been used as a treatment for gastroparesis and is well established as far as safety and efficacy. A typical LP creates a 5-cm incision across the pylorus including some of the antrum and duodenum. This incision is made longitudinally, and the hand-sewn closure is performed transversely starting from the center widening the pyloric channel significantly. Shada et al. reviewed 5 years of LP data from a single database and found that 86% of patients experienced an improvement in gastric emptying study (GES) with normalization in 77% [14]. On the other hand, pyloromyotomy involves just splitting the muscle of the pylorus surgically and has not been well established in gastroparesis. This sort of approach has been used more extensively studied for hypertrophic pyloric stenosis.

Finally, another important proof of concept has been the use of transpyloric stenting for GP. Khashab et al. reported a series of 30 patients that received a pyloric stent for refractory GP [15••]. Again, gastric emptying and symptoms were improved, but this did not seem to be a viable long-term solution. Stents can neither be left in place long term nor can one expect that they will not migrate at some future time. Endoscopic suturing seems to limit the possibility of migration, but this is not guaranteed.

Given the potential of pyloric interventions for improving symptoms and gastric emptying, a simple minimally invasive approach for cutting the pylorus would be highly desirable and likely effective long term.

G-POEM

Technique

In 2013, Khashab et al., described the first G-POEM case in humans for a patient that had failed other medical therapies but had responded well to a transpyloric stent [16]. Since then, there have been several series in the literature suggesting that the G-POEM technique is both safe and effective and has an impact on both symptoms and gastric emptying at times [16,17,18,19,20].

The G-POEM technique is based on the principles of submucosal tunneling and dissection and is similar in many ways to the technique described for peroral endoscopic myotomy for achalasia [21]. During the procedure, the patient is positioned supine or slightly left lateral and using a gastroscope, the stomach is evacuated of any contents and lavages. It is often beneficial to have patients on a liquid diet for 48 h, and there is no consensus on whether an antibiotic solution is required for lavage of the stomach. Generally, however, broad-spectrum intravenous antibiotics are administered at the start of the procedure. A 9-mm gastroscope fitted with an integrated water jet (e.g., Olympus GIF-HQ190; Tokyo, Japan), a transparent cap (11.3 mm, Olympus, Tokyo, Japan), and carbon dioxide insufflation are preferred. A small submucosal injection is made 5 cm from the pylorus along the axis of the greater curvature in the region of the antrum (Fig. 1b). Our preference is to use a solution of normal saline with either 0.25% indigo carmine solution or 0.1% methylene blue. Hetastarch can be added to the solution at a final concentration of 3% to improve retention of fluid within the space. This site of mucosal entry falls in line with scope as one pushes towards the pylorus. This 4–6 o’clock approach is considered posterior. Other attempts at a more anterior approach, similar to what is done laparoscopically, have been described but can result in suboptimal angulation. Gonzalez et al. described a lesser curvature approach, but again their experience may not be generalizable [19••].

Fig. 1
figure 1

Stepwise pictorial of G-POEM. A posterior site along the axis of the stomach is selected (a), and an initial bleb (b) and mucosotomy are created (c). The tunnel is entered and fibers are dissected using a blended current (d) and soft coagulation is performed of larger caliber vessels (e). Alternately, a spray coagulation setting (f) can be used to reduce bleeding, but care must be taken to avoid injury of the mucosa which in this case is along the superior aspect of the tunnel (g). The pylorus is identified as a short-thick band of white muscle (h) and a myotomy is performed either anterograde or retrograde fashion of the circular muscle (i). The mucosotomy and tunnel are inspected for bleeding and mucosal injury (j). The pylorus is visibly more relaxed (k) after the myotomy. Finally, closure of the mucostomy is performed using either clips (l) or endoscopic suturing.

Full size image

As with POEM, a mucosotomy is created at the site of the initial submucosal bleb using a preferred endoscopic submucosal dissection knife (e.g., Olympus Hook Knife, Olympus triangular tip knife, or Erbe T-Type Hybrid Knife) (Fig. 1c). This is most often performed longitudinally; however, a more transverse approach may favor closure with endoscopic suturing. A blended current is typically preferred (e.g., Erbe EndoCut Q 3:1:1, or Dry Cut 80 watts effect 3/4) (Fig. 1d). The initial entry is not rushed so as to prevent early bleeding or muscular injury at this point. A submucosal tunnel is created towards the pylorus. This is performed using the same solution as with the mucosotomy. Large caliber blood vessels can be preemptively coagulated using a coagulative grasper (Olympus, Tokyo, Japan) and a soft coagulation setting (Erbe soft coag 30–50 watts) (Fig. 1e). The dissection of fibers can be performed with the same blended current as used to start the mucosotomy. Some endoscopists prefer a higher powered coagulation current to reduce bleeding within the tunnel from small caliber vessels (e.g., Erbe Spray Coag 40 watts, effect 2) (Fig. 1f). It is imperative to prevent mucosal injury while in the tunnel.

Identifying the pylorus may require exiting and reentering the tunnel several times. There have been several techniques described for recognizing the end of the pyloric muscle ring including the use of a fluoroscopically visible clip [22]. In most cases, however, simply looking for evidence of the submucosal dye at the level of the pylorus is sufficient. The pylorus is usually a well-defined band of smooth muscle and is generally not long as compared to the lower esophageal sphincter (Fig. 1h). The length of the myotomy, therefore, is generally short and in most case series ranges between 1.5 and 2 cm. In fact, pyloric topography with EndoFlip confirms that the pyloric channel in most cases is about 1.5 cm [12••]. Care is taken not to injure the duodenal mucosa at this point as it can present with a delayed leak and/or infections.

Finally, a myotomy is performed either in an anterograde or retrograde fashion. The inner circular and oblique muscle bundles are dissected preserving the outer longitudinal fibers to ensure separation from surrounding vital structures, especially the larger vasculature surrounding the duodenum (Fig. 1i). It is important to proceed cautiously therefore, as deep penetrating branches of these vessels can be injured resulting in bleeding. Our preference is to perform the myotomy in retrograde fashion utilizing EndoCut Q current and an insulated-tip knife (IT-2 or IT Nano, Olympus) as the non-conducting tip protects the duodenum which runs close and perpendicular to the pyloric ring. As with POEM, the initial mucosotomy is closed with either clips or endoscopic sutures (Fig. 1l).

Postoperative care

Postprocedure care has not been standardized. Our practice is to monitor patients in the hospital at least overnight. Intravenous antibiotics are continued while in the hospital, and patients are usually transitioned to oral antibiotics for 7 days upon discharge. An upper gastrointestinal series is obtained on the next day and, if no leaks, the patient is maintained a liquid and a soft mechanical diet for 2 weeks. In addition, acid suppression with proton pump inhibitors is required for at least 4 weeks.

Procedural outcomes

Several large series have been reported for G-POEM. In total, about 130 patients have been reported on and many more cases have been performed worldwide [16,17,18,19,20]. However, there is only one prospective study available to date and certainly no randomized studies (Table 1).

Table 1 Up to date summary of international registries for patients undergoing G-POEM
Full size table

Following the first report by Khashab et al., Shlomovitz et al. reported on endoscopic pyloromyotomy in seven patients in which the etiology of gastroparesis was either postsurgical or idiopathic [16]. The procedure was technically successful in all patients without complications. Six of these seven patients had clinical improvement, and 80% of those with follow-up had improved emptying on a standard 4-h gastric emptying study. One patient failed to respond clinically but also subsequently had no improvement after a laparoscopic pyloroplasty. The described technique was termed as a peroral pyloromyotomy (POP) but was identical to a G-POEM.

Following this study, a larger series of 30 patients was reported by Khashab et al., with a mixture of diabetic, postsurgical, and idiopathic GP [17••]. This was a multicenter study in centers skilled with submucosal tunneling and POEM. The overall technical success was 100% with minimal adverse events and an 86% clinical response. In addition, there was an improvement in the majority of the emptying studies and complete normalization in 47% of the patients.

The largest study to date included 47 patients that had exhausted prior medical therapy and again had a mixture of etiologies [18]. In this particular study, the approach of the G-POEM was along the lesser curvature which was favored by this group. As with other studies, there was a significant improvement in the postprocedure emptying study and in the Gastroparesis Cardinal Symptom Index Score (GCSI-Score). Retention of solids was reduced on average from 37 to 20% at 4 h, but few if any patients truly had completely normalized emptying in this study. This is in contrast to previous studies that have shown up to 70% normalization of gastric emptying following the procedure. It is not clear if this is related to variation in technique (i.e., using a lesser curvature approach) or patient selection. This raises an important and unsettled point with regard to G-POEM, and that is that we do not yet know which patients will respond.

The only prospective study to date was recently reported by Gonzalez et al. [19••]. A total of 29 patients underwent G-POEM with 100% technical success and had a median follow-up of 10 months. The clinical success rate was noted to be 79% at 3 months and 69% at 6 months. The GES normalized in 70% of the cases. Univariate analysis suggested that there was a higher chance of clinical failure (OR 1.8) in females, and that etiologies other than diabetes were associated with a higher chance of long-term success.

Overall, published studies suggest that the G-POEM technique holds tremendous promise, but for a subset of patients, it may have a little to no effect. It is not yet clear how to best select patients for success, although it does seem that patients with idiopathic or postsurgical GP have a greater chance of improving.

Patient selection

G-POEM holds significant promise as a minimally invasive option for patients that fail to respond to dietary modifications and medical therapy. While there are no head-to-head comparisons between LP and G-POEM, they both achieve approximately 70% normalization of gastric emptying. Some questions remain regarding technique—i.e., whether there is an advantage to a greater curvature versus lesser curvature approach. The more important question, however, is who are the patients that will respond best?

It makes most sense that those whose pylorus has a low cross-sectional area, smaller diameter and lower compliance are likely to be the best candidates. This can only be defined by EndoFlip currently, and we do not have a clear understanding of what entirely constitutes “abnormal” regarding these parameters. One surrogate marker for good response could be an initial response to botox or transpyloric stenting. This has not been tested prospectively, but this is currently an evolving practice. There is concern that repeated Botox injection can cause fibrosis making any future interventions more difficult. Therefore, if a patient with GP responds to a trial of botox to any degree, it is more advisable to proceed with a more definitive endoscopic intervention such as G-POEM rather than repeated Botox treatments. If local expertise for G-POEM is not available, transpyloric stenting can be considered.

Conclusion

G-POEM represents an evolution of submucosal endoscopy for a condition that has been difficult to treat and has had few therapeutic options to date. Similar to LP, endoscopic pyloric interventions for gastroparesis seem to hold particular promise and have the ability to normalize GES. G-POEM is not only minimally invasive; it is also durable. Gastroparesis is a ubiquitous disorder, and it is expected that only a subset of patient will respond to G-POEM. Future studies should focus on optimal patent selection and factors that are associated with positive response. Currently, the best surrogates are response to Botox and/or transpyloric stenting. The most promising test, however, may be pyloric topography using EndoFlip as it can offer concrete functional measurements and is easy to perform. The combination of improved patient selection, standardized techniques, and improved tools will only build on the significant early success of G-POEM.