Anti-reflux mucosectomy for gastroesophageal reflux disease in the absence of hiatus hernia: a pilot study

Introduction

As a treatment for gastroesophageal reflux disease (GERD), proton pump inhibitors (PPIs) are the mainstay of medical therapy [1,2]. Laparoscopic fundoplication is generally advised when symptoms are poorly controlled with PPIs and is regarded as a gold standard of treatment, with excellent control in the short- and midterm [3,4]. Long-term results, however, remain equivocal [5,6]. Following on from the principles of surgical fundoplication, a variety of endoscopic procedures for GERD have been proposed to achieve non-surgical control. Literature data at the date support that the delivery of radiofrequency energy to the gastroesophageal junction (Stretta) is safe, effective, durable, and repeatable if necessary and serves an unmet need for many GERD sufferers. The SAGES guidelines for GERD published in early 2013 gave for this procedure high grade of evidence and strong recommendation for treating GERD in well-selected patients [7].

Commonly employing tissue-approximating devices (staple- or suture-based), a range of response rates have been demonstrated with these techniques [7-13]. To date, however, no endoscopic procedure has been widely accepted as a standard treatment of GERD, either (primarily) due to insufficient symptom control or the requirement of costly proprietary devices.

In 2003, we reported a case of circumferential mucosal resection of the distal esophagus and cardia for high-grade dysplasia (HGD) in a short segment of Barrett’s esophagus [14]. In this case, the primary objective was complete excision of Barrett’s mucosa with HGD. The patient happened also to present with repeat chest pain and regurgitation (DeMeester score 5) with significant hiatal hernia (flap valve Grade 3). A circumferential mucosal resection extending to include a 2 cm wide portion of the gastric cardia was completed to ensure an adequate distal margin to the Barrett’s mucosa. We postulated that this would also reduce reflux symptoms by creating a relative stricture at gastric cardia. As expected, excellent control of reflux was indeed obtained by scar formation at the level of gastric cardia, associated with normalization of 24-h pH monitoring [15]. The unobstructed passage of food through the gastroesophageal junction was preserved, although multiple balloon dilatations were required to control initial stricture formation. Eventually, the Barrett’s mucosa was totally replaced by new squamous epithelium in the healing process [15]. After more than 10 years of follow up, the patient still remains asymptomatic, without requiring PPI prescription and no recurrence of Barrett’s epithelium.

This case series suggests that endoscopic anti-reflux mucosectomy (ARMS) may represent an effective anti-reflux procedure, with the added advantage of requiring no additional devices and leaving no artificial prostheses in situ.

Patients and methods

Patient selection

PPI-refractory GERD without sliding hernia seems the most appropriate indication for this procedure, regardless of the presence or absence of Barrett’s esophagus. We selected 10 consecutive such patients presenting to the outpatient clinics at our institution for this procedure (Table 1).

Table 1

Patient background

Results

In all cases, mucosal reduction of the scheduled area was completed without any immediate complications (bleeding or perforation).

GERD symptoms were significantly improved after ARMS (DeMeester score: heartburn 2.7-0.3, P=0.0022; regurgitation 2.5-0.3, P=0.0022; total, 5.1-0.8, P=0.0022) (Fig. 1). The flap valve score was also significantly improved after ARMS from 3.2 to 1.2 (P=0.0152) (Fig. 2).

Figure 1

DeMeester score before and after anti-reflux mucosectomy (ARMS). The upper and lower bars show the standard deviation. The square indicates mean value. The DeMeester scores significantly improved after cardiac mucosal reduction (heartburn, P=0.0011; regurgitation, P=0.0011; total, P=0.0011)

Figure 2

Flap valve grade before and after anti-reflux mucosectomy (ARMS). The flap valve grade score significantly improved after cardiac mucosal reduction (P=0.0065)

In this series, the initial 2 patients underwent circumferential ARMS, removing short-segment Barrett’s esophagus also including HGD. In one patient (case no. 1) circumferential ARMS was carried out in multi-fragment EMR manner. In the other patient (case no. 2) circumferential ARMS was completed by ESD technique. In both cases, 2 cm-wide circumferential mucosal resection at the gastric cardia was completed and visually confirmed in the retroflexed view (Fig. 3A). Endoscopic balloon dilatation was repeated several times in the subsequent 3 months due to stricture formation. Further dilatation was not required once re-epithelialization was complete. In these cases, almost all reflux symptoms disappeared after mucosal healing (DeMeester score 0). Retroflex view from stomach at this stage demonstrated a tight and robust junction with the procedure having created mucosal fold convergence at the lesser curve of the stomach (Fig. 3B). After 10 and 3 years of follow up respectively, neither patient has experienced recurrence of symptoms or recurrence of endoscopic evidence of esophagitis or Barrett’s esophagus (Fig. 3C, D). Although circumferential ARMS resulted in very good long-term control of GERD symptoms , it was related to stricture formation necessitating repeat balloon dilation. Hence, we considered these initial two cases as “anecdotal” in attendance of further studies.

Figure 3

Endoscopic follow up of circumferential anti-reflux mucosectomy (ARMS) (retroflexed views). (A) Immediately after circumferential ARMS. Approximately 2 cm-wide gastric cardia mucosa was circumferentially resected by cap-endoscopic mucosal resection method [18]. (B) Appearance at 3 years. A tight gastro-esophageal junction. Convergence of three gastric folds was observed along the lesser curve of the stomach. (C) More than 10 years after circumferential ARMS. Appearance is similar to Fig. 3B. (D) More than 10 years after circumferential ARMS (forward view). Chromoendoscopy with Lugol’s solution demonstrated well-stained squamous epithelium with neither recurrence of esophagitis nor Barrett’s esophagus

The remaining 8 consecutive patients underwent crescentic ARMS; 6 received ESD, and 2 multi-fragment EMR. In all cases, hemi-circumferential ARMS was completed with approximately 2 cm cardia mucosal resection.

Before treatment, endoscopic retroflexed view demonstrated substantial opening of EGJ (flap valve Grade 2 or 3) but no sliding hernia was identified (Fig. 4A). Half- to two-thirds circumferential ARMS was carried out (Fig. 4B). The standard dose PPI was prescribed for the subsequent 40 days and then discontinued. These patients were started on a soft diet the morning after the procedure and were switched to a regular diet after the 2nd day. Follow-up endoscopy was scheduled at 2 months after ARMS. The EGJ in retroflex view demonstrated a significantly improved appearance (flap valve Grade 1) (Fig. 4C, D). Reflux symptoms were absent, with no medication. The DeMeester score fell to 0 in 7 cases and to 1 in 1 case.

Figure 4

Endoscopic follow up of crescentic anti-reflux mucosectomy (ARMS). (A) Before ARMS. Endoscopy in retroflexion demonstrated significant hiatal hernia (Flap valve score 3) but no sliding component. Chest pain and regurgitation were prominent in this case (DeMeester score 5) and symptoms were not controlled by double dose proton pump inhibitor. (B) Immediately after procedure. Endoscopy in retroflexion showed two-thirds circumferential artificial ulcer. ARMS was centered at lesser curve and the mucosal flap valve at greater curve was preserved. (C) Appearance at 2 months. Mucosal valve was re-shaped and well-defined (Mucosal flap valve Grade 1). (D) Alternative endoscopic view at 2 months. Mucosal valve as appeared as though “stitched” at the lesser curve of gastric cardia

Fraction time absorbance (>0.14) was controlled from 52% to 4% in Bilitec (P=0.05). In 24-h esophageal pH monitoring the fraction of time at pH <4 improved from 29.1% to 3.1% (P=0.01), (Fig. 5). Fraction time absorbance (>0.14) of bile reflux was controlled from 52% to 4% (P=0.05) (Fig. 6).

Figure 5

Changes of acid reflux after anti-reflux mucosectomy (ARMS). The fraction of time at pH <4 was improved from 29.1% to 3.1% (P=0.05)

Figure 6

Changes of bile reflux after anti-reflux mucosectomy (ARMS). Fraction time absorbance (>0.14) was controlled from 52% to 4% in Bilitec (P=0.05)

Operating time on average was 76 min (42-124, N=3) in the group of piecemeal EMR and 127 min on average (98-176, N=7) in the group of ESD, EMR is faster than ESD (P=0.05). In all cases, PPI prescription could be discontinued with no ill effects.

Discussion

Our results suggest a potential anti-reflux effect of ARMS. The mechanism is presumed to be due to scar formation after healing of the mucosal defect [14,15]. On the gastric side, this induces narrowing of the gastric cardia opening, while preserving and/or re-creating a robust his angle. We also postulate some remodeling of mucosal flap valve as an effective antireflux mechanism at this anatomical level.

As demonstrated in Fig. 4, after ARMS, the lesser curve of the gastric cardia takes on an almost “mechanically-stitched” appearance. The mucosal flap is rebuilt and looks well-defined. Furthermore, the lesser curve side of the EGJ is shortened with scar formation, and greater curve of EGJ (his angle side) is kept non-scarred and therefore retains its flexibility as a mucosal flap valve. Another major advantage of this procedure is that it requires no proprietary equipment and leaves no artificial prostheses in situ (which could generate a foreign body reaction in the future).

Although possible complications may include perforation or bleeding, the techniques of EMR/ESD used in this procedure have become standardized and popularized. These risks are likely to be minimal in “expert” hands and, even if experienced, can be resolved using standardized endoscopic recovery techniques such as coagulation forceps or clips for bleeding and perforation [18-20]. Furthermore, piecemeal EMR could be a less time-consuming and acceptable (sole) technique in this procedure, particularly if the specimen involves no dysplasia.

The limitation of this pilot study is that, in almost half of cases, pH monitoring was not accepted by the patients after procedure. In order to demonstrate efficacy, the results of this pilot study need to be confirmed by larger ones with long-term follow up. For this perspective, manometric studies before and after the procedure are highly recommended, and pH-impedenzometry might be used as the only measurement for reflux so ameliorating the compliance to instrumental studies after the procedure.

The quantity of mucosa to be resected to induce appropriate (“not too tight and not too loose”) scar formation is a key issue in this procedure. Our two cases seem to indicate that total circumferential resection causes too tight a junction, which requires repeat endoscopic dilation [21,22], while subtotal dissection, which we have termed crescentic, produces better results in this regard, while still resulting in symptom control. In the initial 2 cases circumferential ARMS was eventually done because of multiple HGD in short-segment Barrett’s, but circumferential ARMS should be avoided from the point of post-procedural stricture. In this latter technique, the remaining mucosal area should be estimated as twice the width of the endoscope circumference when viewed in retroflexion from stomach. The ideal range of mucosal reduction may be altered by several factors such as the extent of laxity at the EGJ as well as esophageal contractile function. In the case of hypomotility of the esophageal body, for instance, half circumferential mucosal reduction may be recommended.

In this pilot study, the crescentic ARMS was centered on the lesser curve. As this kept the mucosa on the angle side intact, but greater curve ARMS is also technically possible. A prospective study is necessary to clarify which ARMS technique produces the best results.

Furthermore, the length of ARMS may also influence outcomes. In this series, mucosal reduction was carried out in 1 cm esophageal site and 2 cm gastric side as this followed from the first successful clinical case. As the major antireflux contribution may be from the mucosal resection on the gastric side, the overall length could also vary.

Finally, the long-term results of ARMS warrant close attention. Excellent control of GERD in our first clinical case of ARMS even after more than 10 years suggests that this technique may achieve long-lasting results.

Case selection is important, and we have not extended the study at this stage as most GERD in Japanese population is generally mild and can be easily controlled by PPI medication, whereas most refractory cases have a substantial hiatal hernia. This is well-treated by laparoscopic Nissen fundoplication [4,5]. We feel that the most appropriate indication for ARMS is intractable GERD with no sliding hernia. Patients with PPI-refractory GERD without a sliding hiatal hernia and possibly short-segment Barrett’s esophagus seem to be the best candidates for the ARMS treatment.

In conclusion, this first clinical series of ARMS for GERD with no sliding hernia showed excellent short- and midterm control of GERD. Future, larger studies with objective assessment and long follow up are warranted.