initial diameter of the stricture: dilation should be started with the smallest balloon, with subsequent gradual increases in balloon diameter and evaluation of the mucosa for signs of injury between each dilation. 5,6 Proposed mechanisms of gastric emphysema include forceful entry of air into the stomach wall through a new or a pre-existing break in the gastric mucosa. 4 The latter may arise from neoplastic or peptic ulceration, or as a result of instrumentation. Insufflation of the stomach with air during endoscopy may then exacerbate any mucosal defect and force air into the wall. High intra- abdominal pressure generated during protracted retching or severe vomiting may also cause air to dissect along the tissue planes of the stomach. 4 Gastroparesis has been associated with neuromuscular dysfunction. 7 It is sug- gested by us that intramural air may affect the myenteric plexus and/or smooth muscle layer of the stomach and thereby impair gastric motor function, with resulting acute gastroparesis. Gastric emphysema may be evident on a plain abdom- inal radiograph as a thin radiolucency outlining a distended stomach, or it may be detected by CT. Management is conservative, mainly nasogastric suction. In case series, the clinical course usually has been benign, with resolution of the emphysema. 4 Prophylactic systemic administration of broad-spectrum antibiotics may be indicated, because of the risk of bacterial contamination of the gastric wall if the emphysema has occurred secondary to instrumentation as opposed to spontaneous development in association with pulmonary disease. Gastric emphysema must be distinguished from em- physematous gastritis, a condition in which the intramural gas is produced by organisms invading the stomach wall. Patients with emphysematous gastritis typically are severely ill; manifestations include fever, rigors, abdominal tenderness, leukocytosis, and shock. 4 However, the radio- graphic appearance often is similar to that of gastric emphysema, and thus the differential diagnosis is based on clinical manifestations, including the absence of a cause, such as instrumentation, in cases of emphysematous gastritis. In summary, gastric emphysema can occur as a conse- quence of endoscopic balloon dilation of the distal esophagus and/or the cardia. Although this complication appears to be rare, it should be considered in patients with signs of upper-GI obstruction after endoscopy. Conserva- tive management appears to be satisfactory. However, more cases are required to establish the outcome for patients in whom this complication occurs as a result of balloon dilation. Michael Sproat, MBBS, Bsc Jim Huddy, MRCP James Wafula, FRCR Queen Elizabeth Hospital Woolwich, London T. Paulose George, MD, MRCP North East Wales NHS Trust Wrexham United Kingdom REFERENCES 1. Taub S, Rodan BA, Bean WJ, Koerner RS, Mullin DM, Feng TS. Balloon dilatation of esophageal strictures. Am J Gastroenterol 1986;81:14-8. 2. ASGE. Complications of upper GI endoscopy. Gastrointest Endosc 2002;55:784-93. 3. Clouse RE. Complications of endoscopic gastrointestinal dila- tion techniques. Gastrointest Endosc Clin N Am 1996;6: 323-41. 4. Lee S, Rutledge JN. Gastric emphysema. Am J Gastro- enterol 1984;79:899-904. 5. ASGE. Esophageal dilation. Gastrointest Endosc 1998;48: 702-4. 6. Langdon DF. The rule of three in esophageal dilation. Gastrointest Endosc 1997;45:111. 7. Hornbuckle K, Barnett JL. The diagnosis and work-up of the patient with gastroparesis. J Clin Gastroenterol 2000;30: 117-24. Is metallic clip application reliable for perforations of the stomach caused by EMR? To the Editor: We read with interest the article of Tsunada et al. 1 on endoscopic closure with metallic clips of perforations caused by EMR. We compliment them for performing such a difficult intervention without morbidity, even when the perforation was as much as 25 mm in size. Tsunada et al. 1 address this problem appropriately and provide a meaning- ful contribution that substantiates the use of metallic clips for closure of EMR-related perforations. Unfortunately, however, there are certain points that require clarification. A satisfactory and secure closure can be created if the submucosal layers at a site of perforation are approxi- mated. It has been shown in an experimental study that intraluminal clip application brings the mucosal layers together but not the submucosa, muscularis propria, and serosa; whereas, by comparison, conventional repair of a perforation with sutures approximates all layers of the gastric wall. 2 In terms of the support of sutures, the submucosa is the most important layer. 3 The muscularis propria, formed by packed smooth muscle with collagen, serves as an intramuscular tendon and a source of strength. 4 The holding capacity of clips applied to the mucosa is extremely poor, because the submucosa and muscularis propria are not captured with an endoscopically applied clip. 2 Tsunada et al. 1 did not state whether simple closure by application of a metallic clip includes the sub- mucosal layer. As demonstrated by Mustoe et al., 5 the intact gastric mucosa contributes only 14% of the total breaking strength of the fresh gastric wound. Although an omental patch technique that includes the application of clips has been described in a porcine model by Hashiba et al., 6 there are no data that demonstrate the LETTERS TO THE EDITOR VOLUME 60, NO. 4, 2004 GASTROINTESTINAL ENDOSCOPY 669 Letters to the Editor