Do gastrotomies require repair after endoscopic transgastric peritoneoscopy? A controlled study

ORIGINAL ARTICLE: Experimental Endoscopy

Do gastrotomies require repair after endoscopic transgastric peritoneoscopy? A controlled study Simon Bergman, MD, MSc, Daniel J. Fix, BS, Kevin Volt, MD, Jason C. Roland, MD, Lynn Happel, MD, Kevin M. Reavis, MD, Theodore J. Cios, BS, Vincent Ho, BS, Alan Evans BS, Vimal K. Narula, MD, Jeffrey W. Hazey, MD, W. Scott Melvin, MD Columbus, Ohio, USA

Background: The optimal method for closing gastrotomies after transgastric instrumentation has yet to be determined. Objective: To compare gastrotomy closure with endoscopically delivered bioabsorbable plugs with no closure. Design: Prospective, controlled study. Setting: Animal laboratory. Subjects: Twenty-three dogs undergoing endoscopic transgastric peritoneoscopy between July and August 2007. Interventions: Endoscopic anterior wall gastrotomies were performed with balloon dilation to allow passage of the endoscope into the peritoneal cavity. The plug group (n ⫽ 12) underwent endoscopic placement of a 4 ⫻ 6-cm bioabsorbable mesh plug in the perforation, whereas the no-treatment group (n ⫽ 11) did not. Animals underwent necropsy 2 weeks after the procedure. Main Outcome Measurements: Complications related to gastrotomy closure, gastric burst pressures, relationship of burst perforation to gastrotomy, and the degree of adhesions and inflammation at the gastrotomy site. Results: After the gastrotomy, all dogs survived without any complications. At necropsy, burst pressures were 77 ⫾ 11 mm Hg and 76 ⫾ 15 mm Hg (P ⫽ .9) in the plug group and no-treatment group, respectively. Perforations occurred at the site of the gastrotomy in 2 of 12 animals in the plug group and in none of the 11 dogs in the no-treatment group (P ⫽ .5). Finally, there were minimal adhesions in all dogs (11/11) in the no-treatment group and minimal adhesions in 3 and moderate adhesions or inflammatory masses in 9 of the 12 animals in the plug group (P ⫽ .004). Limitations: Small number of subjects, animal model, no randomization. Gastrotomy trauma during short peritoneoscopy may not be applicable to longer procedures. Conclusions: After endoscopic gastrotomy, animals that were left untreated did not show any clinical ill effects and demonstrated adequate healing, with fewer adhesions and less inflammation compared with those treated with a bioabsorbable plug. (Gastrointest Endosc 2010;71:1013-7.)

The Natural Orifice Surgery Consortium for Assessment and Research, a Society of American Gastrointestinal and

Endoscopic Surgeons/American Society for Gastrointestinal Endoscopy working group, identified gastric closure as

Abbreviations: NOTES, natural orifice transluminal endoscopic surgery; NTG, no-treatment group; PG, plug group.

doi:10.1016/j.gie.2010.01.025

DISCLOSURE: The following authors disclosed financial relationships relevant to this publication: J.W. Hazey: Clinical advisory board member for Covidien. W.S. Melvin: Scientific advisory board member for Stryker, scientific advisory board member for Endogastric Solutions, scientific advisory board member for W.L. Gore. All other authors disclosed no financial relationships relevant to this publication Copyright © 2010 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00

www.giejournal.org

Received April 4, 2008. Accepted January 7, 2010. Current affiliations: Center for Minimally Invasive Surgery, Department of Surgery, The Ohio State University, Columbus, Ohio, USA. This work was presented at the SAGES annual meeting, April 2008, Philadelphia, Pa. Reprint requests: W. Scott Melvin, MD, Division of General Surgery, The Ohio State University, Room N729, Doan Hall, 410 West Tenth Avenue, Columbus, OH 43210.

Volume 71, No. 6 : 2010 GASTROINTESTINAL ENDOSCOPY

1013

NOTES gastrotomy closure

Bergman et al

a fundamental challenge to the safe introduction of natural orifice transluminal endoscopic surgery (NOTES). It was agreed that, given the safety of current laparoscopic approaches, before embarking on human trials studying transgastric peritoneal access, gastric closure techniques should reach 100% reliability in animal models.1 In a recent study, we demonstrated that the application of a bioabsorbable plug device resulted in durable closure of gastric perforation with physiologic healing of the injury site in all animals.2 Other simple closure techniques involve the use of endoscopic clips or loops, although these only offer mucosal reapproximation.3,4 Full-thickness endoscopic suturing and stapling devices are currently under study and have generally been successful in closing gastrotomies after endoscopic transgastric peritoneoscopy.5-8 However, these instruments may be more difficult to handle and position and are certainly more costly. The objective of this controlled survival study was to evaluate, in a canine model, gastrotomy closure using endoscopically delivered bioabsorbable plugs compared with gastrotomies left to heal without any closure. Primary outcomes are complications related to gastrotomy closure, gastric burst pressures, relationship of burst perforation to gastrotomy, and degree of adhesions and inflammation at the gastrotomy site.

METHODS Twenty-three male dogs, weighing 40 to 60 lb, were assigned to the no-treatment group (NTG) (n ⫽ 11) or the plug group (PG) (n ⫽ 12). The procedure to be performed was selected by the surgical team, and the animals were chosen by the veterinary team without knowledge of group assignment. After approval by the Institutional Laboratory Animal Care and Use Committee, all animals were fasted for 12 hours before the procedure. They underwent general endotracheal anesthesia and were placed in the supine position on the operating room table, and a single dose of parenteral antibiotics was administered (500 mg cefazolin). Before use, the endoscope (11-mm singlechannel therapeutic endoscope, model GIF1T100; Olympus Medical Systems, Center Valley, Pa) was treated with Klenzyme solution (Steris, Mentor, Ohio) and regular liquid hand-washing soap. It was inserted perorally and passed into the stomach, which was insufflated with air. The anterior gastric wall was identified by external palpation of the abdominal wall. A needle-knife (Cook Medical, Bloomington, Ind) was used to create a 5-mm anterior wall gastrotomy through which a 450-cm guidewire (Boston Scientific, Natick, Mass) was inserted. The needle-knife was then exchanged over the guidewire for a 15-mm CRE wire-guided balloon dilator (Boston Scientific). The balloon was positioned so that it was halfway into the peritoneal cavity and was then inflated to 15 mm (8 atm) for 60 seconds. The balloon and the endoscope were brought together and passed through the gastric wall as a unit. If 1014 GASTROINTESTINAL ENDOSCOPY

Volume 71, No. 6 : 2010

Take-home Message ●

Endoscopic repair and closure of gastric perforations are not a well-documented process. Transagastric access to the peritoneal cavity requires a secure and reliable closure method. An endoscopically placed, bioabsorbable plug may be an option for gastrotomy closure. However, it must be remembered that control animals with gastrotomies left to heal without any type of closure did well in all cases. The control data are important when interpreting the data of any closure techniques evaluated in a similar context.

there was difficulty in introducing the endoscope into the peritoneal cavity and air leaked through the gastrotomy, the tense pneumoperitoneum that occurred was decompressed by percutaneous insertion of a 14-gauge catheter in the supraumbilical midline. Peritoneoscopy was considered complete after all 4 quadrants had been visualized. Iatrogenic injuries from access were noted. Finally, the stomach was desufflated, and the endoscope was removed. The NTG dogs recovered from anesthesia without further intervention. The PG dogs underwent a second upper endoscopy during which an endoscopic grasper was used to wedge a 4 ⫻ 6-cm piece of bioabsorbable polyglycolic acid mesh (Dexon; Covidien, Mansfield, Mass) into the gastrotomy. It was positioned so that it was protruding halfway into the stomach and halfway into the peritoneal cavity, as illustrated in Figure 1. In all cases, the friction between the makeshift plug and the gastrotomy edges seemed sufficient to hold it in place and obviated fixation to the gastric wall. The stomach was desufflated while visualizing adequate plug placement. Procedure time was defined as the time between initial endoscope insertion and final endoscope removal. Both groups were fed a regular diet on the first postoperative day and were killed 14 days after the procedure. Weights were recorded 1 to 4 days before the necropsy. The weight difference was calculated by subtracting the preoperative weight from the prenecropsy weight. The study protocol allowed early killing at the discretion of the veterinary team for any animal exhibiting early clinical deterioration after the procedure. At necropsy, the canine stomachs were explanted. The gastrotomy site and the extent of exterior adhesions were documented. The area of adhesions around the gastrotomy was measured, and the presence of an inflammatory mass was noted. The degree of adhesions and inflammation was described as none (no adhesions or inflammatory masses), minimal (adhesions ⬍1 cm2 and no inflammatory mass), moderate (adhesions ⬎1 cm2 and no inflammatory mass), or severe (presence of an inflammatory mass). The duodenum was then tied to a water tap with umbilical tape. A catheter, connected to a pressure transducer (Hoswww.giejournal.org

Bergman et al

Figure 1. A bioabsorbable mesh is positioned by an endoscopic grasper so that it is protruding halfway into the stomach and halfway into the peritoneal cavity.

pira, Lake Forest, Ill), and an anesthesia monitor (Smiths Medical PM, Waukesha, Wisc) were tied to the esophagus with its tip in the gastric lumen. The stomachs were then filled with water at a low, steady flow of approximately 700 mL/min, and intraluminal gastric pressures were continuously recorded until a perforation occurred. Any lowpressure leaks were noted. The burst pressure was defined as the final recorded pressure before perforation. After this, the location of the burst perforation was noted.

Statistical analysis Continuous data are expressed as mean ⫾ standard deviation and analyzed with Student t test. Categorical data are expressed as proportions (percentage) and analyzed with Yates’ corrected ␹2 test. A P value ⬍5% was set as the criterion for statistical significance. Analysis was performed by using Microsoft Excel 2004 for Macintosh.

RESULTS Between July and August 2007, 23 dogs underwent endoscopic transgastric peritoneoscopy. The NTG dogs (n ⫽ 11) had a mean preoperative weight of 47 ⫾ 4 lb. Procedure time was 37 ⫾ 18 minutes, and all dogs survived without any signs of surgical complications until necropsy. Prenecropsy weight for the NTG dogs was 47 ⫾ 6 lb, which was not statistically different from the preoperative weight (P ⫽ .3). The PG dogs (n ⫽ 12) weighed 44 ⫾ 7 lb preoperatively, and procedure time was 47 ⫾ 15 minutes. All dogs survived without any signs of surgical complications until the necropsy. Prenecropsy weight for the PG dogs was 47 ⫾ 8 lb, which also lacked statistical significance compared with preoperative weights (P ⫽ .2). www.giejournal.org

NOTES gastrotomy closure

During ex vivo examination of the stomach, all NTG gastrotomies were located on the anterior surface of the stomach. Mesh plugs could not be found in 2 PG dogs, and in 1 of them, the gastrotomy site had healed completely and could not be identified. There was 1 gastrotomy on the posterior greater curvature, and the remaining 10 of 12 PG gastrotomies were located on the anterior surface of the stomach. There were minimal gastrotomy site adhesions (⬍1 cm2 area) in all NTG animals. On the other hand, 3 of the 12 PG dogs had minimal adhesions, whereas 9 of 12 had adhesions greater than 1 cm2 area or an adherent inflammatory mass (P ⫽ .004). All gastrotomies in both groups had sealed and did not leak at low filling pressures. Burst pressures were not performed in 1 NTG animal because of technical issues. Burst pressures were 76 ⫾ 15 mm Hg and 77 ⫾ 11 mm Hg (P ⫽ .9) in the NTG and PG animals, respectively. Burst perforations occurred at the site of the endoscopic gastrotomy in none of 10 (0%) NTG dogs and in 2 of 12 (17%) PG dogs (P ⫽ .5). A statistical comparison of the 2 groups is summarized in Table 1.

DISCUSSION The Natural Orifice Surgery Consortium for Assessment and Research has identified intestinal closure as one of the fundamental hurdles to the successful clinical introduction of endoscopic transluminal surgery: “If NOTES is to reach human trials, a 100% reliable means of gastric closure must be developed.”1 Peritoneal access enterotomies must clearly be shown to cause no additional morbidity over the established safety profiles of conventional transabdominal access routes, whether they are open or laparoscopic approaches.1,8,9 Without it, ethical and patient safety concerns will confine NOTES to the animal laboratory and prevent significant foray into mainstream clinical practice. Current gastrotomy closure techniques involve either plug/patch or reapproximation techniques. Our group recently reported results demonstrating the success of the former approach with a commercial bioabsorbable hernia plug. Twelve dogs in which an open gastrotomy was plugged all survived without leak and showed physiologic healing at the injury site on ex vivo examination. On histologic examination, there was robust initial ingrowth through 8 weeks and eventual attenuation, possibly reflecting collagen turnover and remodeling at 12 weeks, as would be expected in the accepted time course of wound healing with healthy subjects.2 Similarly, in a rat gastric perforation model, a porcine-derived small intestine submucosal patch was used successfully for closure in all animals. Histologic examination revealed the patch to be an effective scaffolding agent, as evidenced by early regeneration of normal gastric mucosa at its edges.10 Finally, a nitinol self-expandable, double-umbrella-shaped device with an inner nonpermeable polyethylene terephthalate patch was deployed in gastrotomies after endoscopic Volume 71, No. 6 : 2010 GASTROINTESTINAL ENDOSCOPY

1015

NOTES gastrotomy closure

Bergman et al

TABLE 1. Comparison of the no-treatment group and the plug group No treatment group (n ⴝ 11)

Plug group (n ⴝ 12)

P value

Procedure time (min)

37 ⫾ 18

47 ⫾ 15

.2

Weight difference (lb)

0⫾2

2⫾6

.3

0/11 (0%)

0/12 (0%)

1.0

Burst pressure (mm Hg)

76 ⫾ 15

77 ⫾ 11

.9

Burst at gastrotomy site

0/10 (0%)

2/12 (17%)

.5

Mortality/early euthanasia

Adhesions/inflammation Minimal

.004 11/11 (100%)

3/12 (25%)

Moderate

0/11 (0%)

5/12 (42%)

Severe

0/11 (0%)

4/12 (33%)

transgastric peritoneoscopy. Five of 6 animals thrived for several weeks, and 1 died of sepsis.11 Finally, others have described pulling omentum into the gastrotomy and clipping it in place to provide a repair conceptually similar to the Graham patch.12 In this controlled study, we compared bioabsorbable plug closure with leaving gastrotomies open. The animals left to heal without treatment had equivalent gastric burst pressures and healed with a lesser inflammatory reaction than the dogs closed with the plug. In 2 of the PG animals, the repair was the weakest site of the stomach. This suggests that a foreign body, bioabsorbable or not, may be detrimental to gastrotomy healing, either because of a greater associated inflammatory reaction or a stenting phenomenon that may delay tissue ingrowth or create a path for bacteria to travel from the gastric lumen to the peritoneal cavity. The 2 missing plugs likely dislodged after placement. Our technique may be improved by securing the mesh to the gastric wall with clips or a suturing device. The greater degree of adhesions observed in the PG is interesting but not unexpected, given the presence of a foreign body and its associated inflammatory reaction, well described in the hernia literature. Postoperative adhesions can have serious clinical consequences because they contribute to the development of small-bowel obstruction and add considerable risk to reoperative surgery, whether transluminal or transperitoneal. In recent years, modern surgical techniques, such as laparoscopic surgery, have been successful in reducing this risk. Similarly, NOTES investigators should seek to refine gastrotomy closure techniques that do not promote intraperitoneal adhesion formation. Simple reapproximation techniques have had significant success. As described in their seminal paper, Kalloo et al3 closed their gastrotomies with endoscopic clips, and the animals recovered uneventfully with no signs of leak. Similar results have also been reported after reapproxima1016 GASTROINTESTINAL ENDOSCOPY

Volume 71, No. 6 : 2010

tion of the gastric edges by using only several endoscopic loop applications.4 The argument against such repairs is that providing only mucosal approximation of the gastric wall violates the surgical dogma that seromuscular apposition is necessary for proper healing. The facts that excellent results have been reported with simple mucosal gastrotomy repair and that, in this study, animals with unrepaired gastrotomies all survived are not entirely surprising. Similar results were found in a swine colon perforation model with pigs randomized to no closure or endoscopic clip closure. All control animals recovered fully, and, at the time of necropsy, perforations were found to have healed spontaneously.13 We used a canine rather than a swine model because, in our opinion, the anatomy of the canine stomach is more analogous to the human stomach. Furthermore, we believe that swine have a superior resistance to clinical peritonitis. In humans, conservative nonoperative management of perforated gastroduodenal ulcers is considered by some to be part of the treatment algorithm.14 Several comparative studies, including 1 randomized, controlled study, found success rates of 50% to 90%, with morbidity and mortality rates equivalent to those of surgical management.15-18 Factors that predict failure of this approach are generally absent in the healthy, nondiseased stomach subjected to endoscopic enterotomies: age older than 70 years old, history longer than 9 to 24 hours before treatment, use of immunosuppression or nonsteroidal antiinflammatory drugs, concomitant illness, presence of shock, and large pneumo-/hydroperitoneum.16-21 This represents a significant issue for investigators in this field who must address the need for appropriate animal models and control groups, as well as more sensitive outcome measures, to rigorously establish the superiority of one closure technique over another. In addition to the anatomic, physiologic, and immunologic differences between canine models and humans, other limitations include small group sizes and the nature of the NOTES procedure that we performed. A short www.giejournal.org

Bergman et al

peritoneoscopy may cause significantly less gastrotomy site trauma than would a longer procedure requiring extreme endoscope maneuvering, organ manipulation, and specimen extraction. Endoscopic repair of gastric perforations with a bioabsorbable plug is associated with greater adhesive and inflammatory reactions than controls, with no leaks and equivalent burst pressures. In a canine model, gastric perforations are effectively sealed either by an endoluminal plug technique or without therapy. Nevertheless, in humans, some type of repair will always be indicated. To allow NOTES to progress via a transgastric or transcolonic approach, adequate closure techniques remain to be developed. Testing and validation of new devices and procedures will have to be compared with the low failure rates seen in controls and must always be effective. REFERENCES 1. Rattner D, Kalloo A. ASGE/SAGES working group on natural orifice translumenal endoscopic surgery. Surg Endosc 2006;20:329-33. 2. Cios TJ, Reavis KM, Renton DR, et al. Gastrotomy closure using bioabsorbable plugs in a canine model. Surg Endosc 2008;22:961-6. 3. Kalloo AN, Singh VK, Jagannath SB, et al. Flexible transgastric peritoneoscopy: a novel approach to diagnostic and therapeutic interventions in the peritoneal cavity. Gastrointest Endosc 2004;60:114-7. 4. Katsarelias D. Endoloop application as an alternative method for gastrotomy closure in experimental transgastric surgery. Surg Endosc 2007; 21:1862-5. 5. Bergstrom M, Swain P, Park PO. Early clinical experience with a new flexible endoscopic suturing method for natural orifice transluminal endoscopic surgery and intraluminal endosurgery. Gastrointest Endosc 2008;67:528-33. 6. Magno P, Giday SA, Dray X, et al. A new stapler-based full-thickness transgastric access closure: results from an animal pilot trial. Endoscopy 2007;39:876-80.

www.giejournal.org

NOTES gastrotomy closure 7. McGee MF, Marks JM, Onders RP, et al. Complete endoscopic closure of gastrotomy after natural orifice translumenal endoscopic surgery using the NDO Plicator. Surg Endosc 2008;22:214-20. 8. Sclabas GM, Swain P, Swanstrom LL. Endoluminal methods for gastrotomy closure in natural orifice transenteric surgery (NOTES). Surg Innov 2006;13:23-30. 9. McGee MF, Rosen MJ, Marks J, et al. A primer on natural orifice transluminal endoscopic surgery: building a new paradigm. Surg Innov 2006; 13:86-93. 10. de la Fuente SG, Gottfried MR, Lawson DC, et al. Evaluation of porcinederived small intestine submucosa as a biodegradable graft for gastrointestinal healing. J Gastrointest Surg 2003;7:96-101. 11. Perretta S, Sereno S, Forgione A, et al. A new method to close the gastrotomy by using a cardiac septal occluder: long-term survival study in a porcine model. Gastrointest Endosc 2007;66:809-13. 12. Feretis C, Kalantzopoulos D, Koulouris P, et al. Endoscopic transgastric procedures in anesthetized pigs: technical challenges, complications, and survival. Endoscopy 2007;39:394-400. 13. Raju GS, Ijaz A, Xiao SY, et al. Controlled trial of immediate endoluminal closure of colon perforations in a porcine model by use of a novel clip device. Gastrointest Endosc 2006;64:989-97. 14. Putcha RV, Burdick S. Management of iatrogenic perforation. Gastroenterol Clin N Am 2003;32:1289-309. 15. Kristensen ES. Conservative management of 155 cases of perforated peptic ulcer. Acta Chir Scand 1980;146:189-93. 16. Songne B, Jean F, Foulatier O, et al. Non operative treatment for perforated peptic ulcer: results of a prospective study [in French]. Ann Chir 2004;129:578-82. 17. Dascalescu C, Andriescu L, Bulat C, et al. Taylor’s method: a therapeutic alternative for perforated gastroduodenal ulcer. Hepatogastroenterology 2006;53:543-6. 18. Crofts TJ, Park KG, Steele RJ, et al. A randomized trial of nonoperative treatment for perforated peptic ulcer. N Engl J Med 1989;320:970-3. 19. Anonymous. Conservative management of perforated peptic ulcer. Lancet 1989;2:1429-30. 20. Buchler P, Oulhaci W, Morel P, et al. Results of conservative treatment for perforated gastroduodenal ulcers in patients not eligible for surgical repair. Swiss Med Wkly 2007;137:337-40. 21. Irvin T. Mortality and perforated peptic ulcer: case for risk stratification in elderly patients. Br J Surg 1989;76:215-8.

Volume 71, No. 6 : 2010 GASTROINTESTINAL ENDOSCOPY

1017