Glutaraldehyde Colitis Following Endoscopy: Clinical and Pathological Features and Investigation of an Outbreak

Transcription

1 GASTROENTEROLOGY 1995;108: Glutaraldehyde Colitis Following Endoscopy: Clinical and Pathological Features and Investigation of an Outbreak A. BRIAN WEST,* SHIH-FAN KUAN,* MICHAEL BENNICK, t and SUZANNE LAGARDE* Departments of *Pathology and *Internal Medicine, Yale University, New Haven, Connecticut Although potentially noxious compounds are used routinely to disinfect endoscopes, reports of their inadvertent introduction to the gastrointestinal tract, usually attributed to the retention of disinfectant within endoscope channels, are rare. This case report describes the clinical features of glutaraldehyde-induced colitis and the pathology of the mucosal injury in four patients, in at least one of whom the disinfectant was not retained in the endoscope itself. Within 3 months, three patients experienced severe acute proctocolitis <6 hours after a sigmoidoscopy showing no abnormalities, performed in a small endoscopy unit. Investigation of the unit's protocols suggested that the most likely cause was retention of 2% glutaraldehyde disinfectant in the endoscope channels, and changes were made to prevent this. When a fourth case occurred 5 months later, the source of the glutaraldehyde was found to be the tubing connecting water bottles to the endoscopes, which was disinfected rigorously but flushed inconsistently between cases. Glutaraldehyde-induced colitis seems similar to ischemic colitis in biopsy specimens and cannot be diagnosed by histological analysis alone. Acute colitis occurring within 24 hours of a colonoscopy showing no abnormalities should be considered iatrogenic and should lead to an investigation of procedures in use for cleaning and disinfecting endoscopic equipment. oncern about the transmission of pathogens from C one patient to another via endoscopes has led to many reports on methods for disinfecting these instruments. 1 In the process of ensuring the death of microbes contaminating instruments at the end of each procedure, powerful chemicals are used to clean and disinfect them. Some of these chemicals contain compounds that are toxic to human mucosae. The protocols that have been developed for cleaning the equipment between each endoscopy usually remove these noxious compounds, and reports of toxic mucosal injury resulting from endoscopies are rare. 2-5 However, injury of this type may be underreported. Two compounds used in cleaning and disinfecting endoscopes have been implicated as causing colitis: hydrogen peroxide 2 and glutaraldehyde) '4 Glutaraldehyde is generally considered the disinfectant of choice, despite its known properties as irritant, toxin, and allergen, I and is commonly applied in concentrations similar to those used for fixation in electron microscopy (2%). In the cases reported, the toxic agent was retained in the channels of endoscopes as a result of inadequate flushing after disinfection and was therefore sprayed directly onto the mucosa during colonoscopy. This report describes four cases of severe colitis caused by glutaraldehyde, in at least one of which the irritant was not retained in the endoscope channels. Taken together, the clinical, endoscopic, and histological features of this form of colitis are characteristic and should enable the correct diagnosis to be made swiftly and with confidence. History of the Outbreak The problem arose in the sigmoidoscopy unit of a health maintenance organization that had been functioning satisfactorily for 12 years. It was used principally by three internists and one physician assistant and had a fully trained, experienced support staff. Approximately 1200 sigmoidoscopies were performed each year. Case Reports Patient 1 An 81-year-old woman underwent sigmoidoscopy 12 days after a transient episode of diarrhea and rectal bleeding that had completely resolved (Table 1). The examination showed no abnormalities. Within hours of the procedure, the patient had severe abdominal cramps and passed blood per rectum. The following morning, she had tachycardia, mild pyrexia, and a neutrophil leukocytosis. The abdomen was soft and unremarkable. Radiographs showed no free air. The patient was admitted for observation because of possible colonic perforation. Blood cultures, stool cultures, and Clostridium difficile toxin assay were negative. She was treated with broad-spectrum antibiotics. Computerized tomographic scan of the abdomen and a contrast enema showed thickening of the distal sigmoid, suggestive of segmental colitis. Her cramping and rectal bleeding gradually subsided, as did her leukocytosis and fever, and she was discharged on day 7. Her colorectal problems resolved and have not recurred during 29 months of follow-up by the American Gastroenterological Association /95/$3.00

2 April 1995 GLUTARALDEHYDE COLITIS FOLLOWING ENDOSCOPY 1251 Table 1. Clinical Features of Patients With Glutaraldehyde- Induced Colitis Patient no Age (yr) Sex F M F F Time of sigmoidoscopy a Day 0 Day 32 Day 84 Day 246 Endoscopist A B A A Nurse a b b c Place on endoscopy list b 5/6 3/5 5/6 5/5 Preoperative diagnosis BRBPR c Screen a Screen ~ Screen r Endoscopic findings (initial) Normal Diverticulosis Normal Normal Symptoms Onset within 6 hours Abdominal cramps Bleeding per rectum Tenesmus Findings Pulse (beats~rain) Temperature ( F) Abdomen Soft Soft Soft Soft White cell count (per#l) 27,700 20,000 14, % Segmented forms Hematocrit (%) atime (days) after sigmoidoscopy of patient 1. ~Sequence of patient on list/total no. of patients on list. COne episode of diarrhea with bright red blood per rectum. aroutine screening for colonic polyps. ~History of vague abdominal pain for several years. rhad previous hyperplastic polyp of rectum. Patient 2 One month after patient 1 underwent sigmoidoscopy, an asymptomatic 74-year-old white man underwent a routine screening sigmoidoscopy to 30 cm (Table 1). Other than diverticula, no abnormalities were seen. Within hours of the procedure, he had lower abdominal cramps, rectal bleeding, and tenesmus. The next day, he had tachycardia and mild pyrexia. His abdomen was unremarkable, but there was fresh blood in the rectum. Abdominal radiographs showed no free air. A contrast enema showed some spasm in the proximal sigmoid but no extravasation of contrast. Blood cultures were negative. He was treated empirically with oral ciprofloxacin. He continued to have tenesmus with occasional spots of bright red blood per rectum and suprapubic discomfort, but his white cell count decreased to 14,000/}.tL by day 2 to 6,800//.tL by day 15. During 28 months of follow-up, he has had no subsequent colitis. Patient 3 Seven weeks after patient 2 underwent endoscopy, a 61- year-old white woman underwent flexible sigmoidoscopy for evaluation of vague mild chronic abdominal pain (Table 1). The mucosa appeared normal up to 50 cm. Several hours later, the patient experienced severe lower abdominal cramps, bleeding per rectum, and tenesmus. That night, she had shaking chilis and a temperature of 102 F. On day 2, her temperature, pulse rate, and white cell count were elevated. Her abdomen was soft with active bowel sounds, but the rectum contained blood-tinged fluid. Abdominal radiography showed no abnormalities. The patient underwent a repeat limited sigmoidoscopy by a trained gastroenterologist using an out-of-center instrument on unprepared bowel. There was a sharply delineated patch of proctitis in the proximal rectum and extensive colitis with mucosal necrosis and contiguous areas of ulceration from 20 cm to at least 25 cm, beyond which the scope was not advanced. The mucosa between and distal to these areas appeared normal. Stool culture grew normal enteric flora, and blood cultures were negative. Biopsy specimens obtained from involved areas of rectum and sigmoid colon (20-25 cm) showed severe acute mucosal injury of the ischemic type (Figure 1; see Pathology). The histological features were similar to those described in colitis induced by hydrogen peroxide 2'6 and glutaraldehyde, 3'4 although the differential diagnosis included ischemia and colitis secondary to bacterial toxins. The clinical, endoscopic, and pathological findings were considered most suggestive of an acute chemical-induced colitis. The patient was given a liquid diet, and her symptoms resolved over the next 3 days. She has had no recurrence during 26 months of follow-up. Etiologic Investigations Not until the third case was brought to the attention of a staff gastroenterologist were the earlier cases recalled, and the clustering of similar cases was noted. Use of the endoscopy suite was suspended, procedures for cleaning, disinfecting and preparing the endoscopes were reviewed, and all personnel were interviewed. All three procedures resulting in colitis had been performed toward the end of the day's list, but endoscopes of two brands had been used, two endoscopists had performed the procedures, and no one member of the nursing staff had been involved in cleaning and preparation of instruments or in assisting with the sigmoidoscopies (Table 1). None of the other patients undergoing endoscopy on the same days had reported problems. On inquiry, it transpired that a new brand of disinfectant that contained glutaraldehyde at a concentration of 2.0% had recently been introduced (Cidex; Surgikos Inc., Arlington, TX), whereas the brand used previously contained only 0.2% glutaraldehyde (Coldspor; Metrex Research Corp., Parker, CO). Moreover, the cleaning procedure did not invariably dry the channels of the endoscopes completely. It appeared that if rinsing was inadequate, residual disinfectant might have been retained in the endoscope channels and sprayed directly onto the mucosa, causing localized toxic injury, as described previously for cleaning solutions containing hydrogen peroxide and glutaraldehyde. 2-4 Therefore, a rigorous protocol was instituted to ensure that chemicals were rinsed thoroughly from the surface and channels of the cleaned and disinfected endoscopes. In practice, after routine rinse cycles on the autodisinfector (Keymed II; Olympus, Boston, MA), water was flushed through the water channel manually, water and air were drawn through the suction channel, and the exterior of the endoscope was hand-rinsed and -dried. Patient 4 Five months after introducing this protocol, patient 4, a 46-year-old woman, underwent routine sigmoidoscopy with

3 1252 WEST ET AL. GASTROENTEROLOGY Vol. 108, No. 4 Figure 1. Biopsy specimens from the rectosigmoid of patient 3 obtained 30 hours after the first sigmoidoscopy, which showed no abnormalities. (A) Mild injury with lack of uniformity of cells in the surface epithelium, mucin depletion, and eosinophilia of the superficial lamina propria. (B) Oblique section through an area with more severe injury showing loss of epithelium from the mucosal surface and from the upper parts of the glands, exudation of neutrophils, and eosinophilia of the lamina propria. (C) Severe injury to the mucosal surface with neutrophil exudate but preservation of deep mucosa and gland bases. (D) Loss of entire glands and surface epithelium in some areas. Note the relatively sparse inflammatory infiltrate and the presence of microthrombi in some capillaries (arrow) (H&E; original magnifications: A, 460 ; B, 230 ; C, 145 ; D, 290 ). normal findings. Within hours, she experienced a sensation of rectal pressure accompanied by abdominal pain, temperature of 101 F, tenesmus, and frequent, small-volume watery bloodstained diarrhea. She was seen on the second day after the procedure. On sigmoidoscopy to 35 cm, a severe patchy procto- colitis was observed, involving mainly the rectum. Biopsy specimens from 15 cm contained normal colorectal mucosa, but specimens from 5 cm showed focal mucosal necrosis with an inflammatory exudate suggestive of resolving acute mucosal injury of the ischemic type (Figure 2; see Pathology). The

4 April GLUTARALDEHYDE COLITIS FOLLOWING ENDOSCOPY 1253 patient was treated with steroid foam enemas for 1 week. She made a good recovery, and 21 months later has had no recurrence of colitis. Further Investigations Patient 4 was the last patient to undergo sigmoidoscopy on the day of her initial procedure (Table 1), and the only one to experience symptoms after it. The endoscopic and histological features of the rectum at the second examination were strongly suggestive of a localized acute toxic proctocolitis. Review of the endoscopy suite records and interviews with the staff confirmed that the new procedures were being performed in all respects, and it was evident that the channels of the endoscopes were being properly cleaned, disinfected, rinsed, and dried. However, on inspecting the endoscopy suite procedure room, the unmistakable smell of glutaraldehyde was detected, although the disinfectant was stored and used only in the separate preparation room. Several physicians who used the suite had noted this smell sporadically. The source proved to be a small volume of glutaraldehyde lying in the tubing used to connect the water bottles to the endoscopes. The bottles and tubing were soaked in disinfectant after each procedure, but whereas the bottles were rinsed and dried, the tubing was sometimes incompletely flushed and retained a few milliliters... of disinfectant. Procedures for rinsing the tubing after disinfection by suctioning water through it and for drying it were instituted. No additional cases have occurred during the subsequent 20 months despite extensive use of the suite. Pathology Biopsy specimens from patient 3 and patient 4 showed similar features, with normal and severely injured mucosa in close proximity in the areas sampled. In the least affected areas, there was mucin depletion and loss of uniformity of the ceils in the surface epithelium, with some eosinophilia of the superficial lamina propria (Figure 1A). More advanced lesions showed breakdown of the epithelial monolayer, first on the mucosal surface and then descending progressively down the glands. At this stage, there was exudation of neutrophils through the disrupted epithelium (Figure 1B), with capillary hemorrhage and fibrin deposition in the lamina propria. The gland bases often were preserved (Figure 1C), although entire glands were lost in some areas (Figure 1D). Thrombi were seen in occasional capillaries in the lamina propria (Figures 1D and 2A). In the later stages, the epithelium of the remaining glands became regenerative (Figure 2A), and lymphocytes and plasma cells migrated into the lamina propria (Figure 2B). No evidence of vasculitis or of bacterial or fungal invasion was noted. The histology of these biopsy specimens is similar to that described by Qizilbash et al. 4 and by Durante et al. in an animal model. 3 The morphological features are indist-ingui~h± able from those seen in acute mucosal ischemia v-9 and from some stages of mucosal injury caused by certain toxin-producing bacteria, e.g., C. difficile 1 and enterohemorrhagic Escherichia coli,.1 although the clinical setting under which the Figure 2, Rectal biopsy specimens from patient 4 obtained 60 hours after the first sigmoidoscopy. (A) Loss of epithelium from the surface of the mucosa and the upper parts of the glands and a fibrin exudate covering the surface. Thrombi are present in some capillaries (arrows). Extravasation of eryth rocytes (lower left) and fibrin (upper half of field) are evident in the lamina propria. The remaining epithelium is regenerative (H&E; original magnification 460 ). (B) Low-power view of injured mucosa showing loss of glands and of surface epithelium, regeneration of the remaining glands, and a round cell infiltrate in the lamina propria (H&E; original magnification 150 ).

5 1254 WEST ET AL. GASTROENTEROLOGY Vol. 108, No. 4 specimens are obtained will enable a distinction to be easily made. Discussion With increasing emphasis on the necessity to disinfect endoscopic equipment thoroughly between procedures on different patients, the use of potentially toxic chemicals has become routine. There are few published reports of toxic reactions of the gastrointestinal mucosae to these agents. 2-4 However, after presentation of these cases at clinical conferences, several gastroenterologists have told us of instances in which patients experienced transient symptoms of acute colitis within 24 hours after colonoscopy of an uninflamed bowel, suggesting that less severe reactions than those experienced by our patients may not be uncommon. Early reports described the toxic effects of hydrogen peroxide on human colonic mucosa, 6'I2'13 and studies of rats 13 and dogs I4 confirmed the "ischemic" morphological pattern of the induced injury. Recently, Jonas et al. reported the development of acute colitis during colonoscopy in 21 patients as a result of exposure of the mucosa to H202 from residual cleaning solution in the endoscope channels. 2 White mucosal plaques with variable amounts of mucosal inflammation and frothy fluid in the lumen were actually seen to develop during the colonoscopic examination. This observation strongly supports the view that H202 was the noxious agent, although glutaraldehyde was also used in the endoscope cleaning protocol and a single case of colonoscopy-associated colitis was encountered after the use of H202 had been discontinued, implicating the aldehyde also as a primary mucosal toxin. Proctitis attributed to glutaraldehyde in cleaning solution retained in colonoscope channels was first reported by Qizilbash et al. 4 and Castelli et al. 5 in a group of 13 patients who had normal rectal mucosa on initial examination. Subsequently, Durante et al. 3 reported a similar occurrence in a group of 10 patients. In the latter series, inadequate flushing of the cleaned endoscopes was related to the work of one clinical assistant. In both series, the concentration of glutaraldehyde used in the cleaning solution was 2%. A novel feature of the present series is that in at least one of the cases, the source of the glutaraldehyde was not the endoscope channels but the tubing between the water bottle and endoscope. Toxin lodged there might have been expected to be diluted before reaching the mucosa; indeed, this may have been the case, although the narrow bore of the tubing and of the channels distal to it would impede mixing. It seems likely that when the weaker (0.2%) glutaraldehyde solution was in use in the endoscopy suite, it was diluted sufficiently not to cause significant damage to the mucosa. Although the smell of glutaraldehyde had been noted in the procedure room intermittently at that time, no case of postendoscopy proctocolitis was encountered. The more concentrated (2.0%) glutaraldehyde solution was apparently too strong to be diluted to nontoxic levels before application to the mucosal surface. The clinical features of glutaraldehyde-induced proctocolitis reported by Qizilbash et al. include tenesmus, bright red blood per rectum, mucus while straining, a feeling of constipation, and, in severe cases, chills. Symptoms developed within 24 hours of the procedure. 4 In the series of Durante et al., 3 the initial patient had bloody diarrhea and fever a few hours after the sigmoidoscopy. Subsequently, cases were identified in a retrospective survey if the patients had not had diarrhea before the procedure but had at least one episode within 6 hours after it; of these patients, 60% reported abdominal cramps or bloody diarrhea, and a minority had fever, vomiting, or nausea. 3 The clinical features of our four patients lie at the extreme of this spectrum (Table 1). All experienced severe abdominal cramps and rectal bleeding within a few hours of sigmoidoscopy. By the next day when they were examined, all four had fever, and the three who had white cell counts performed had significant neutrophil leukocytosis. It is probable that patients with milder symptoms did not bring them to the attention of their endoscopists. The pathological features of glutaraldehyde-induced proctitis were described in an unillustrated abstract by Qizilbash et al. as resembling ischemic injury, with "erosion and ulceration of superficial mucosa, loss of tubules, hemorrhage, and the presence of inflammatory exudate on the surface. ''4 A biopsy specimen from one of these patients was illustrated by Durante et al. 3 as a comparison with experimentally induced glutaraldehyde colitis in rats. The biopsy specimens of our third and fourth patients confirm and extend these observations. However, the histological features are not specific, and cannot be distinguished from features of ischemic proctocolitis or from some stages of certain toxic bacterial colitides, notably pseudomembranous colitis and hemorrhagic enterocolitis, on morphological grounds alone. Therefore, the diagnosis will be missed unless the pathologist is aware both of this entity and of the clinical circumstances leading to the biopsy. Glutaraldehyde remains the agent of choice for disinfecting endoscopic equipment despite its irritant, allergenic, and toxic qualities. 1'15 Exposure of patients and staff to this noxious compound should be minimized. In this report, we have shown the necessity to ensure its complete removal from all parts of endoscopic equipment, and we have described the clinical and pathological

6 April 1995 GLUTARALDEHYDE COLITIS FOLLOWING ENDOSCOPY 1255 features of severe glutaraldehyde-induced colitis. The severe and milder forms of this ill-reported entity should be more widely recognized so that preventive measures, and, should they fail, corrective measures, can be instituted. Addendum After submission of this paper, a detailed report of 6 patients with endoscope-induced colitis attributable to glutaraldehyde was published. 16 Retention of disinfectant in sigmoidoscopes because of inadequate rinsing and evacuation of the channels was identified as the cause. The pathological features of the colitis are described and illustrated and appear indistinguishable from those seen by us. These data, and an additional report of severe glutaraldehyde-induced proctitis following endorectal ultrasonography, iv lend support to our impression that glutaraldehyde-induced colitis following endoscopy occurs more commonly than is generally recognized. References 1. Babb JR, Bradley CR. The mechanics of endoscope disinfection. J Hosp Infect 1991; 18: Jonas G, Mahoney A, Murray J, Gertler S. Chemical colitis due to endoscope cleaning solutions: a mimic of pseudomembranous colitis. Gastroenterology 1988; 95: Durante L, Zulty JC, Israel E, Powers P J, Russell RG, Qizilbash AH, Morris JG Jr. Investigation of an outbreak of bloody diarrhea: association with endoscopic cleaning solution and demonstration of lesions in an animal model. Am J Med 1992;92: Qizilbash A, Castelli M, Seaton T. Post-colonoscopy chemical proctitis (abstr). Lab Invest 1986;54:51A. 5. Castelli M, Qizilbash A, Seaton T. Post-colonoscopy proctitis (abstr). Am J Gastroenterol 1986;81: Meyer CT, Brand M, DeLuca VA, Spiro HM. Hydrogen peroxide colitis: a report of three patients. J Clin Gastroenterol 1981; 3: Kilpatrick ZM, Farman J, Yesner R, Spiro HM. Ischemic proctitis. JAMA 1968; 205: Dawson MA, Schaefer JW. The clinical course of reversible ischemic colitis--observations on the progression of sigmoidoscopic and histologic changes. Gastroenterology 1971;60: McNeill C, Green G, Bannayan G, Weser R. Ischaemic colitis diagnosed by early colonoscopy. Gastrointest Endosc 1974; 20: Price AB, Davies DR. Pseudomembranous colitis. J Clin Pathol 1977;30: Griffin P, Olmstead L, Petras R. Escherichia coil 0157:H7-associated colitis. Gastroenterology 1990;99: Pumphrey RE. Hydrogen peroxide proctitis. Am J Surg 1951;81: Sheehan JF, Brynjolfsson G. Ulcerative colitis following hydrogen peroxide enema: case report and experimental production with transient emphysema of colonic wall and gas embolism. Lab Invest 1960; 9: Shaw A, Cooperman A, Fusco J. Gas embolism produced by hydrogen peroxide. N Engl J Med 1967;277: Beauchamp RQ. A critical review of toxicity of glutaraldehyde. Crit Rev Toxicol 1992;22: Rozen P, Somgen G J, Baratz M, Kimel R, Gilat T. Endoscopeinduced colitis--description, probable cause by glutaraldehyde, and prevention. Gastrointest Endosc 1994;40: Burtin P, Ruget O, Petit R, Boyer J. Glutaraldehyde-induced proctitis after endorectal ultrasound examination: a higher risk of incidence than expected? Gastrointest Endosc 1993;39: Received September 20, Accepted November 23, Address requests for reprints to: A. Brian West, M.B., M.R.C.Path., Department of Pathology, Yale University, 310 Cedar Street, P.O. Box , New Haven, Connecticut Fax: (203) Supported in part by grant P30 DK34989 to the Digestive Diseases Research Core Center, Yale University.