trailing stent were then successfully removed. The patient underwent placement of another 11.5F stent resulting in excellent palliative decompression. Distally migrated stents impacted in the duodenal wall are fortunately a rare event. The above technique allowed for stent retrieval when all other endoscopic techniques had failed. Although both the wire and the snare were passed through the same channel, the duodenoscope's elevator allowed for selective movement of the snare and ultimately the success of the procedure. It is possible that the same technique may have been used with a forward-viewing doublechannel therapeutic endoscope, but none was available. Admittedly a difficult technique, retrieval of the impacted stent with snare and guide wire prevented this rare complication from being much more devastating to this patient and should be considered when standard endoscopic techniques of stent retrieval fail. David S. Weinman, MD Sharp Rees-Stealy Medical Clinic San Diego, California
Balloon removal of an impacted esophageal meat bolus To the Editor: Meat impactions frequently occur in the elderly population. Numerous extracting devices have been used in this situation, including jumbo biopsy forceps, polypectomy snares, and basket retrievers, as well as specialized grasping instruments such as alligator, rat-tooth, and tenaculum forceps.I None of these work particularly well, especially when the meat bolus is partially digested or impacted in the subcricopharyngeal area. This letter describes just such a situation and a new technique that we have found useful for removing impacted meat boluses. A 79-year-old woman presented to the emergency department with acute onset of dysphagia after eating a steak dinner. She stated that this dysphagia had occurred on several occasions over the last 20 years, in one case requiring a difficult extraction with a rigid endoscope. She did not have weight loss, odynophagia, or abdominal pain. She was in no acute distress and had only mild hypersalivation. Her physical examination was unremarkable, as was her laboratory data. A barium swallow revealed a large, partially obstructing meat bolus just below the cricopharyngeal area. The patient was initially treated with intravenous glucagon with no improvement of her symptoms. Endoscopy was then performed, revealing a large piece of meat impacted below the cricopharyngeal area. The meat could not be advanced with forward pressure from the endoscope and because of its immediate subcricopharyngeallocation could neither be easily grasped nor maneuvered. A 15 mm diameter, 3 cm length through-the-scope balloon catheter (Microvasive, Watertown, Mass.) was cautiously advanced beyond the impaction under direct endoscopic guidance. The balloon was maximally inflated with saline and then withdrawn until resistance was encountered. At this point, the VOLUME 40, NO.3, 1994
patient was placed in steep Trendelenburg position and the scope and balloon withdrawn simultaneously. This motion successfully dislodged the meat bolus, displacing it into the oropharynx, where it was readily removed. Re-endoscopy of the patient revealed no obvious rings or obstructing lesions. The entire procedure was well tolerated. The above technique of foreign body removal has been performed by one of the authors on an additional three similar patients. It uses a well-established method for coin removal in infants 2 and has also been used in the past for gastric bubble removal. 3 The technique appears most useful for patients with impaction occurring in the subcricopharyngeal area. It is in this area where endoscopic visualization and manipulation is most difficult and the patient is at highest risk for aspiration. Many patients poorly tolerate repeated passage of the endoscope necessary for piecemeal removal. An average of 10 trips were required for complete removal in one study.4 Because of the high level of impaction, it is difficult or impossible to fully advance an overtube when working in this area. The obvious concern with this technique is aspiration of the foreign body, but in our experience this has not been a problem because of the size of the bolus, the patient being placed in steep Trendelenburg position, and rapid digital removal of the disimpacted meat bolus. We think that this technique represents a significant advance in the management of high esophageal meat impactions and should be considered for use in these situations. Jeffrey H. Goldman, MD Robert I. Goldberg, MD University of Miami, School of Medicine Division of Gastroenterology Mt. Sinai Medical Center Miami, Florida
REFERENCES 1. Goldberg RI, Manten RD. Foreign bodies and bezoars of the
upper gastrointestinal tract. In: Barkin JS, O'Phelan CA, eds. Advanced therapeutic endoscopy. New York: Raven Press, 1990:27 -37. 2. Dunlap LB. Removal of an esophageal foreign body using a Foley catheter. Ann Emerg Med 1981;10:101-3. 3. Blinder M, Goldberg RI, Barkin JS, Phillips RS. A new method for gastric bubble removal. Gastrointest Endosc 1987;33:243-4. 4. Rogers BRG, Kot C, Meiri S, Epstein M. An overtube for the flexible fiberoptic esophagogastroduodenoscope. Gastrointest Endosc 1982;28:256-7.
Improved gastric lavage via direct endoscopic visualization To the Editor: Current endoscopic options for the treatment of active UGI hemorrhage are effective for reducing rebleeding rates. 1 However, interventions for UGI hemorrhage rest on the precise identification of the source of bleeding. The endoscopist often encounters blood or clot that obscures the underlying mucosa from view. Lavage through large-bore tubes or suction through therapeutic gastroscopes may remove a portion ofthe adherent clot. We report a case in which both 391
modalities were combined simultaneously to facilitate visualization. A 70-year-old woman, with a history of gastric ulcer and currently using non-steroidal anti-inflammatory drugs, presented with dizziness, melena, and anemia. Nasogastric tube lavaged only "coffee grounds" from the stomach. Esophagogastroduodenoscopy revealed multiple erosions of the antrum with blood and clot obscuring the view of the cardia and fundus. Vigorous suctioning and repositioning the patient failed to improve visualization. The endoscope was removed, and a large orogastric (Edlich Gastric lavage tube 35F, Sherwood Medical, St. Louis, Mo.) tube was passed. Seven-liter tap water lavage was performed. On immediate endoscopy (Olympus IT-100, Olympus America Inc., Lake Success, N.Y.), the cardia remained covered with blood and clot. At this point, endoscopy and lavage were combined. The orogastric tube was inserted beside the patient's bite block and advanced into the stomach. The endoscope was advanced alongside the tube. Several advantages became apparent: (1) endoscopic visualization allowed appropriate targeting of the lavage tube; (2) endoscopic air insufflation created intragastric pressure that increased return oflavage fluid and clot; (3) the solid clot was too large for removal with endoscopic suction, yet was easily removed via the orogastric tube. This aggressive targeted lavage revealed the bleeding source. A 2 cm diameter gastric ulcer with a pigmented protuberance was seen in the cardia, allowing epinephrine injection around the lesion. After the procedure, the patient had stable hemodyamics without any evidence of further bleeding during her hospitalization. Literature review revealed no reports of simultaneous orogastric tube lavage and UGI endoscopy. It is possible that risks such as perforation, mucosal trauma, or pulmonary aspiration may be increased. However, the combination of modalities is warranted for major hemorrhage without a bleeding source identified after conventional lavage. Robert H. Goldklang, MD Steven Christensen, MD University of California, San Diego Medical Center San Diego, California
REFERENCE 1. Laine L. Multipolar electrocoagulation in the treatment of ac-
tive upper gastrointestinal hemorrhage: a prospective controlled trial. N Engl J Med 1987;316:1613-7.
Cytomegalovirus esophagitis in an immunocompetent host To the Editor: Cytomegalovirus (CMV) infection with gastrointestinal involvement is well documented in immunocompromised patients. Although several such cases involving the esophagus have been reported, acute erosive CMV esophagitis has rarely been described in immunocompetent hosts. 1 This case adds to the current literature. A 69-year-old woman was transferred to our hospital 3 weeks after developing acute pancreatitis following ERCP 392
Figure 1. CMV esophagitis. Vesicles not seen; arrow indicates site of probable denuded vesicle.
for suspected choledocholithiasis. Because of associated gastrointestinal hemorrhage, she had received 4 units of packed red blood cells, the most recent being 9 days earlier. Two days after transfer the patient developed odynophagia and hematemesis. Esophagogastroduodenoscopy (EGD) revealed esophagitis involving the lower four-fifths, characterized by marked ulcerations, increased friability, and scattered areas of mucopurulent membrane (Fig. 1). Biopsy specimen demonstrated severe hemorrhagic inflammation with viral inclusions characteristic ofCMV (Fig. 2). Serologic samplings revealed an anti-CMV IgM titer of ::::;1:10 and an anti-CMV IgG titer of> 1:2560; urine culture was also positive for CMV. Human immunodeficiency virus serologic sampling was negative. Total serum IgM and IgG levels were within normal limits. The patient was not anergic on delayed hypersensitivity skin testing with mumps antigen. Fourteen days after admission, a follow-up EGD showed no improvement. Histopathologic findings and culture of the second biopsy specimen were diagnostic of persistent CMV infection; ganciclovir therapy was then initiated at 5 mg/kg every 12 hours for 2 weeks. EGD performed on the 28th day after admission demonstrated complete resolution of infection. CMV serologic samplings repeated after 14 days showed an IgM titer of >1:10 and a persisting IgG titer of > 1:2560. The patient died 3 months after admission because of progressive complications of pancreatitis. Autopsy revealed no evidence of esophageal or disseminated CMV infection; furthermore, no underlying malignancy was identified. Esophageal infection with CMV has predominantly been described in immunocompromised patients such as human immunodeficiency virus-infected individuals 2 or organ transplant recipients. 3 A literature review revealed only one case of severe CMV esophagitis in a patient considered to be immunocompetent. 1 Our case is similar to that one in the following respects: no history of compromised immunity; no prior esophageal disease; concomitant major catabolic illness; recent receipt of blood transfusions; normal total immunoglobulin levels; and intact delayed hypersensitivity skin testing. Symptomatic CMV infections in immunocompetent individuals usually require supportive care only.4 Our patient GASTROINTESTINAL ENDOSCOPY