Gastrointestinal Perforation Associated With Endoscopy in Cats and Dogs

Introduction

Gastrointestinal endoscopy is commonly performed for the diagnosis and treatment of gastrointestinal disease in cats and dogs. Endoscopy is a minimally invasive diagnostic procedure that is used to evaluate the mucosal surfaces of the esophagus, stomach, small intestine, and colon. Endoscopy can also be used to obtain specimens for histopathology, cytology, and microbial analysis.^1,2 Endoscopes are also used to assist in removal of esophageal and gastric foreign bodies and to assist in dilation of esophageal and colonic strictures.^1,2 Complications secondary to gastrointestinal endoscopy are rare, but can include gastrointestinal perforation, hemorrhage from laceration of major blood vessels, and anesthetic complications associated with gastric over distension and decreased venous return.² Perforation during endoscopy is a recognized complication of endoscopic foreign body retrieval and esophageal stricture dilation.^3–5 However, gastrointestinal perforation occurs rarely in cats and dogs undergoing routine diagnostic endoscopy and biopsy. Either poor biopsy technique or forceful insertion of the endoscope without visualization of the lumen have been cited as risk factors.⁶ The ob jective of this retrospective study was to evaluate the prevalence, associated underlying disease, anatomic location, and outcome of gastrointestinal perforations associated with diagnostic endoscopy in cats and dogs.

Materials and Methods

Medical records of cats and dogs with gastrointestinal perforation that occurred during endoscopy at the authors’ hospital between April 1993 and March 2010 were reviewed. Those dates were chosen to include all searchable gastrointestinal and colonoscopic examinations in the study authors’ endoscopy database. Perforations that occurred as a result of either foreign body manipulation or dilation of esophageal strictures were excluded because published retrospective studies have addressed complication rates associated with those procedures.^3–5 Information from medical records included signalment, clinical signs at the time of hospital admission, duration of clinical signs, physical examination findings, results of clinical laboratory tests, diagnostic imaging findings, endoscopic findings, location of perforations, histopathological findings, and clinical diagnosis. Perforation was defined as iatrogenic if it occurred either during endoscopic maneuvering or biopsy. Perforation was defined as secondary to chronic ulceration if it occurred during insufflation, an ulcer was detected at laparotomy, and the ulcer was determined to be preexisting based on histologic evidence of fibrosis around the ulcer margins.⁷ The presence of fibrosis indicates chronic ulceration of >3–5 days in duration.⁷

In preparation for gastrointestinal endoscopy, food was withheld for a minimum of 12 hr. In preparation for colonoscopy, food was withheld for a minimum of 24 hr and at least 2 warm-water enemas were administered, and some dogs had two doses of polyethylene glycol colonic lavage solution^a (20 mL/kg per os) at least 2 hr apart on the afternoon of the day before the procedure. Gastrointestinal endoscopy and colonoscopy were performed in left lateral recumbency under inhalant general anesthesia. Premedications varied, and were chosen by the preference of the attending anesthesiologist. A flexible videoendoscope^b (8.5 mm in diameter, 2.2 mm channel, 103 cm in length) was used for gastroduodenoscopy in cats. A flexible videoendoscope^c (9.4 mm in diameter, 2.8 mm channel, 103 cm in length) was used for gastroduodenoscopy in dogs, and a flexible videoendoscope^d (8.6 mm in diameter, 2.8 mm channel, 140 cm in length) was used for colonoscopy in cats and dogs. Isotonic electrolyte solution was administered by indwelling IV catheter throughout each procedure, and routine cardiovascular and respiratory parameters were monitored.

For gastrointestinal endoscopy, the esophagus, stomach body, angularis, antrum, pylorus, and duodenum were systematically visualized as the scope was advanced to several centimeters past the duodenal bile duct papilla. As the endoscope was withdrawn back into the stomach, it was retroflexed to view the cardia, the fundus, and lesser curvature. For colonoscopy, the endoscope was advanced from the rectum to the cecocolic region. The distal ileum was examined whenever possible. Insufflation was used as needed to optimize viewing of mucosal surfaces during the exam. Distilled water was infused through the irrigating channel of the endoscope as needed to clear mucus, bile, hair, food particles, or feces that obscured the mucosa. Endoscopic mucosal biopsies were obtained using flexible forceps passed through the endoscope’s operating channel from each region, including the duodenum, pyloric antrum, gastric body, and cardia/fundic region for gastroduodenoscopies, and the colon and ileum for colonoscopies. Supervision of inexperienced endoscopists consisted of close observation and instruction during the procedure by an experienced endoscopist. When necessary, control and advancement of the endoscope was performed by an experienced endoscopist if the inexperienced endoscopist was unable to do so safely.

Treatments, outcome, and survival times were recorded for all cats and dogs discharged from the hospital, and the cause of death and results of postmortem examinations were recorded when available. Histological slides of sections of biopsy specimens from each case were reviewed by a single pathologist for confirmation of the histopathological diagnosis using published standards.⁸

Results

A total of 377 cats and 1240 dogs were identified by a retrospective search of medical records as undergoing gastrointestinal endoscopy, colonoscopy, or both procedures during the 17 yr study period. This included 267 cats and 812 dogs each undergoing a single gastrointestinal endoscopy, 37 cats and 249 dogs undergoing a single colonoscopy, and 73 cats and 179 dogs undergoing both procedures. No cases had repeated endoscopic procedures.

Six cats and one dog were identified as having gastric or intestinal perforation associated with upper gastrointestinal endoscopy (Table 1). Perforations occurred in the stomach (fundus and body) and duodenum only. No patients were identified as having perforation associated with colonoscopy. Gastrointestinal perforation occurred in 1.6% of cats and 0.1% of dogs that underwent endoscopy. In three of seven cases (cases 1, 2, and 3), the perforations were associated with preexisting ulcers. In four of seven cases (cases 4–7), the perforations were considered iatrogenic.

TABLE 1 Summary of Clinical Signs, Diagnoses, and Outcomes of Six Cats and One Dog With Gastrointestinal Perforation Associated With Endoscopy

CM, castrated male; DSH, domestic shorthair; IBD inflammatory bowel disease; N/A, not applicable; SF, spayed female.

View Fullscreen

Of the six cats with perforation associated with endoscopy, three were castrated males and three were spayed females. Breeds included domestic shorthair (n = 3), Siamese (n = 2), and Himalayan (n = 1). The mean age of the cats was 12.3 yr (range, 8–17 yr). The mean body weight of the cats was 3.3 kg (range, 2.2–4.9 kg). The dog was a 10 yr old spayed female cocker spaniel weighing 10.3 kg.

Clinical signs reported by owners at presentation included vomiting (n = 5), inappetence (n = 4), weight loss (n = 4), lethargy (n = 4), diarrhea (n = 2), and melena (n = 1) as summarized in Table 1. The mean duration of clinical illness was 10 mo (range, 3 days to 2 yr). The physical examination abnormalities were also listed in Table 1 and included cachexia (n = 4), dehydration (n = 4), gas-filled intestinal loops (n = 2), heart murmur (n = 2), palpably thickened intestinal loops (n = 1), pale mucous membranes (n = 1), and renal asymmetry (n = 1). One cat defecated during abdominal palpation.

Complete blood cell counts and serum biochemical evaluation were performed at the time of hospital admission in four cats and one dog. The remaining two cats (cases 1 and 6) had a complete blood cell count performed within 1 wk prior to admission. Clinically relevant laboratory abnormalities for each case have been listed in Table 2. Those include anemia (n = 3), leukocytosis (n = 3), thrombocytosis (n = 1), thrombocytopenia (n = 1), as well as hypoalbuminemia (n = 3), hypoglobulinemia (n = 3), and hypocalcemia (n = 2). Serum cobalamin and folate concentrations were measured in two cats (cases 2 and 4) and were abnormal in case 4. Increased serum gastrin concentrations were found in cases 2 and 3. Reference ranges were based on published studies.^9–11

TABLE 2 Selected Results of Complete Blood Cell Count and Serum Biochemical Analysis in Six Cats and One Dog With Gastrointestinal Perforation Associated With Endoscopy

*Reference ranges for gastrin levels were based on previously published reference ranges.^9–11

N/A, not applicable; NP, not performed; PCV, packed cell volume.

View Fullscreen

Abdominal radiographs were taken prior to endoscopy in one cat (case 7) and the dog, and were suggestive of gastric foreign body in both. Pertinent findings for abdominal ultrasound, endoscopy, and exploratory laparotomy in each case have been listed in Table 3. Abdominal ultrasound was performed prior to endoscopy in three cats and the dog. Abnormalities included altered layering of the small intestinal wall (n = 1), diffusely thickened gastric and small intestinal walls (n = 1), thickened and corrugated duodenum (n = 1), gas in the intestinal wall (n = 1), enlarged mesenteric lymph nodes (n = 1), and trace peritoneal effusion (n = 1). Endoscopic examination findings were reported in the medical record for the stomach (n = 6) and for the duodenum (n = 5), and visible lesions were found in all animals examined. Abnormal findings in the stomach included thickened gastric folds (n = 2), gastric mass (n = 1), irregular gastric mucosa (n = 1), multiple gastric nodules (n = 1), and thickened gastric folds with nodular appearance (n = 1). Abnormal findings in the duodenum included thickened duodenal folds (n = 2), hyperemic duodenal mucosa (n = 1), multifocal duodenal ulcers (n = 1), irregular mucosa with nodules (n = 1), and multifocal hemorrhages (n = 1). Full endoscopic examination was not completed in case 2 because of severe abdominal distension (pneumoperitoneum) on initial insufflation of the stomach. Duodenoscopy was not completed in case 5 because the gastric mass prevented entry through the pylorus.

TABLE 3 Results of Abdominal Ultrasound, Endoscopy, and Surgery in Six Cats and One Dog With Gastrointestinal Perforation Associated With Endoscopy

NP, not performed; NR, not recorded; SI, small intestine.

View Fullscreen

In all cases, perforation was diagnosed either during or immediately after the endoscopic procedure. Abdominal distension, which did not decrease with suctioning air from the stomach, was observed during the endoscopic procedure in all cases, and tense pneumoperitoneum was noted in three cats (cases 1, 2, and 4). The perforation site was visualized during endoscopy in two cases. In case 4, a full-thickness tear was identified at a duodenal biopsy site, and in case 6, the peritoneal cavity was visualized through a tear in the wall of the gastric fundus as the endoscope was withdrawn from the duodenum into the stomach. In five cases, perforation was diagnosed by abdominocentesis, which yielded air, indicative of pneumoperitoneum. Abdominal radiographs confirmed pneumoperitoneum in the three cases in which they were performed (Figure 1). The endoscopic manipulations that were associated with perforation in each case have been listed in Table 3 and included initial gastric insufflation at entry into the stomach (n = 1), insufflation of the duodenum during duodenoscopic examination (n = 2), endoscopic maneuver of withdrawing the endoscope from the duodenum back into the pylorus (n = 1), biopsy of a gastric mass (n = 1), and biopsy of the duodenum (n = 1). The specific endoscopic manipulation associated with the perforation was not recorded in one case.

View Figure

• Download as Powerpoint
• Download Figure

Following a diagnosis of pneumoperitoneum, immediate exploratory laparotomy and resection (cases 1–5) or repair (cases 6 and 7) of the perforation was performed in all cases. Findings at exploratory laparotomy were included in Table 3. The perforations were located in the duodenum (n = 4), gastric fundus (n = 1), gastric body (n = 1), and both gastric fundus and duodenum (n = 1). Additional abnormal findings at exploratory laparotomy included multifocal duodenal ulcers (n = 2), a gastric mass (n = 1), evidence of chronic ulceration with gray fibrous tissue surrounding the duodenal perforation (n = 3), focal peritonitis (n = 3), a nodular pancreas (n = 1), and a pancreatic mass (n = 1).

All cases were treated as hospital inpatients. Case 5 was diagnosed with large cell lymphoma, based on impression smear cytology of the gastric mass, and was euthanized intraoperatively due to the perceived poor prognosis. Postoperative treatments in the five other cats and the dog included IV fluids (n = 6), opioid pain medications (n = 4), antibiotics (enrofloxacin^e in four cases, ampicillin Na/sulbactam Na^f in three cases, ampicillin^g in two cases, and cefoxitin^h in one case), and an H₂-receptor blockerⁱ in five cases. Case 6 was also treated with cyclophosphamide, vincristine, cytosine arabinoside, and prednisolone based on the thickened appearance of the gastric walls and the high clinical suspicion for lymphoma. Two cats (cases 1 and 2) were also managed with esophageal tube feedings. The mean duration of hospitalization, including the day of the endoscopic procedure, was 8.3 days (range, 5–17 days).

Histopathological diagnoses were based on surgical biopsies (all cases), necropsy (cases 2 and 5), and immunohistochemistry (case 5) and included severe inflammatory bowel disease (case 1); small cell (low-grade) lymphoma (case 4); gastric B cell (high-grade immunoblastic) lymphoma with transmural infiltration (case 5); chronic lymphoplasmacytic, eosinophilic and neutrophilic gastritis (case no.6); severe inflammatory bowel disease with lymphocytic portal hepatitis (case 7); multifocal duodenal ulcers with no underlying etiology apparent (case 2); and chronic pancreatitis (case 3). In both cats diagnosed with severe inflammatory bowel disease, the reviewing pathologist's opinion was that small cell lymphoma could not be definitively excluded but that histopathology was most consistent with inflammatory bowel disease.

The final clinical diagnosis differed from the histopathological diagnosis in three cases. Case 6 re-presented 5 mo after the original diagnosis for anorexia, weight loss, and lethargy, and was diagnosed with lymphoma based on cytology obtained via ultrasound-guided aspirate of a gastric mass. Case 2 was originally diagnosed with multifocal duodenal ulcers with no underlying etiology apparent from surgical biopsies, but was later diagnosed with gastrinoma based on an increased serum gastrin concentration and pancreatic carcinoma found at necropsy. The dog (case 3) was histologically diagnosed with chronic pancreatitis based on surgical biopsies of a pancreatic mass visualized during exploratory laparotomy, but had a clinical diagnosis of gastrinoma based on markedly increased serum gastrin concentration. Final clinical diagnoses included lymphoma (cases 4–6), severe inflammatory bowel disease (cases 1 and 7), and gastrinoma (cases 2 and 3). Five cats and one dog (six of seven patients included in the study) survived the initial hospitalization and were discharged. Long-term follow-up data were available for five cats (Table 1).

Discussion

Gastrointestinal perforation is a rare complication of routine diagnostic endoscopy. In the current study, the prevalence of perforation associated with endoscopy was 1.6% in cats and 0.1% in dogs. Gastrointestinal perforation has been reported infrequently in the veterinary literature. An article that reviews the history of veterinary endoscopy mentions a small cat that developed an 8 cm duodenal tear following forceful insertion of the endoscope. That cat recovered uneventfully with surgical repair, but no further details were provided.¹² Pneumoperitoneum associated with endoscopic biopsy was reported previously in a cat with intestinal lymphoma.¹³ Previous retrospective studies and review articles on endoscopy in cats and dogs allude to the large number of patients endoscoped without complication. In the first published review of veterinary endoscopy, 350 consecutive endoscopies (285 upper gastrointestinal exams and 66 colonoscopies) were performed without perforations.¹⁴ Two other studies in dogs reported >90 and >50 gastrointestinal endoscopies, respectively, without perforation as a complication.^15,16

The most common site of perforation in the current study was the duodenum, but perforation also occurred in the gastric body and gastric fundus. Perforations during endoscopic procedures in the current study were associated with insertion and manipulation of the endoscope, insufflation, and biopsy. Colonic perforation did not occur in any animal in this study population.

Risk analysis was not performed in the current study due to small case numbers, but the majority of endoscopy-associated perforations at the authors’ institution occurred in cats, even though more endoscopic exams were performed in dogs. Perforation occurred only in cats with severe infiltrative small intestinal disease (lymphoma, inflammatory bowel disease) and in cats and dogs with preexisting gastrointestinal ulceration. Although the cause was unknown, the authors speculated that unique characteristics of feline anatomy, tissue tensile strength, and elasticity may have contributed to predisposition to perforation. It was also possible that increased tissue friability contributed to iatrogenic perforation in four cats with underlying diffuse small intestinal infiltrative disease (inflammatory bowel disease, lymphoma).^17–19

Perforation of preexisting duodenal ulcers during endoscopy was identified in two cats and one dog, suggesting that duodenal ulceration increases the risk for endoscopy-associated perforation. In each of those cases, the ulcers were determined to be preexisting based on histopathological evidence of marked fibrosis around the ulcer margins, indicating chronicity of the lesions (>3–5 days).⁷ All cases with chronic duodenal ulceration developed pneumoperitoneum secondary to routine insufflation during endoscopy. Due to evidence of chronic fibrosis around the ulcer margins, as well as surrounding focal peritonitis, the study authors speculated that in each of those cases, the ulcer perforated prior to endoscopy but was likely covered by an omental seal that became disrupted when the gastrointestinal tract was distended with air during endoscopic insufflation.²⁰ Animals with chronic ulcers in the current study either had underlying infiltrative intestinal disease (severe inflammatory bowel disease) or gastrinoma, known risk factors for gastroduodenal ulceration.^17–20

Compared with the published literature, this study revealed a higher perforation rate. Previously cited risk factors for perforation during endoscopy include use of excessive force in manipulating the scope around corners or use of poor biopsy technique. Increased risk of perforation can also occur with deep gastric ulcers, especially if biopsies are taken from the center of an ulcer crater.² It is feasible that the variable experience level of trainee endoscopists at a teaching hospital such as the authors’ may increase the perforation rates associated with endoscopic manipulation and biopsy. However, all endoscopic procedures reported herein were supervised by experienced endoscopists, and perforation also occurred in one case during manipulation by an endoscopist with >20 yr experience. Furthermore, those original retrospective studies on diagnostic endoscopy were published in the 1970s and early 1980s, when endoscopes were less maneuverable and endoscopy was a new diagnostic technique in veterinary gastroenterology.^14–16

The current study suggests cats may be at increased risk for perforation associated with endoscopy compared with dogs and that duodenal perforation is the most common location for perforation. Earlier studies may have a lower incidence of perforation because of a lower number of duodenoscopies performed and a lower number of cats undergoing endoscopy. A higher number of duodenal exams were performed in the current study compared with the published literature, and the previous retrospectives focused mainly on dogs.^14–16 For example, in one series, esophagoscopy and gastroscopy were performed in all cases, but the duodenum was entered in only 6 cats and 30 dogs (13% of gastrointestinal exams).¹⁴ In a study of 90 dogs, esophagoscopy and gastroscopy were performed, but no duodenal exams were performed.¹⁵ In another study of 50 dogs, the rate of successful entry into the duodenum was higher, at an estimated 85%.¹⁶ Thus, the authors of this report suspect that patient factors were more likely than operator factors to contribute to the perforation rate in the current study. As previously discussed, the authors suspect that diseased gastrointestinal tissue, especially in cats, may predispose to perforation despite appropriate endoscopic technique.

The lack of colonic perforations in the current study may be related to anatomic characteristics of the colon, such as elasticity and luminal diameter, and to the distribution of infiltrative diseases in cats. Inflammatory bowel disease and small cell lymphoma are common diagnoses in cats undergoing gastrointestinal endoscopy, and both diseases more commonly affect the small intestine than the colon.^21,22 Iatrogenic perforations during colonoscopy have been rarely reported in cats and dogs. One author reported three colonic perforations over a 17 yr period.²³ Another study reported one perforation out of 355 flexible colonoscopic procedures.²⁴ That dog had a strictured region, and perforation occurred as the endoscope was being passed in an orad direction without direct visualization of the lumen. The authors state that the perforation may have been avoided if less force and a smaller diameter colonoscope had been used, underlying granulomatous colitis may have also been a factor in the perforation.²⁴

Endoscopy-associated gastrointestinal perforation caused pneumoperitoneum to develop during the endoscopic procedure in all seven cases reported herein. Gastrointestinal perforation was diagnosed based on endoscopic visualization, abdominocentesis, abdominal radiography, or a combination of these diagnostic findings. These results suggest that endoscopic visualization of the perforation is less likely than detection of unexplained abdominal distention during the procedure that is unassociated with gastric distention and does not decrease with suctioning of air from the stomach. In this situation, abdominocentesis should be performed immediately to determine if free abdominal air is present, which indicates gastrointestinal perforation. Any cat or dog with pneumoperitoneum associated with endoscopy should be suspected of perforation. Overdistension of the stomach with air can occur during endoscopic insufflation, and can cause abdominal distension, which can mimic pneumoperitoneum. For that reason it is important to evacuate air from the stomach prior to abdominocentesis. In the current study, either abdominocentesis or abdominal radiographs were diagnostic for pneumoperitoneum in all cases evaluated.

All animals in the current study had immediate surgery following gastrointestinal tract perforation, and six of the seven animals survived postoperatively. One cat was euthanized intraoperatively due to anticipated poor prognosis following a cytologic diagnosis of gastric large cell lymphoma. The animals with evidence of chronic gastrointestinal perforation had no fatal endoscopic complications. Nevertheless, the cat with gastrinoma and chronic duodenal ulceration was euthanized 3 wk after hospital release due to recurrence of clinical signs, as reported by the owner. The dog with gastrinoma was lost to follow-up. The remaining four of six animals that survived to hospital release had either long-term survival >8 mo or later died of unrelated causes.

Perforation associated with endoscopy has a good prognosis in the current study with a survival rate of 86% to hospital discharge and 57% to >8 mo. Reasons for the high survival rate to hospital discharge may include the prompt surgical correction and the supportive care that all seven cases received. Thus, risk factors for perforation at endoscopy may not necessarily be a contraindication to endoscopy, but prompt diagnosis and emergency surgery are essential if perforation occurs. Gastrointestinal perforation at endoscopy may have an excellent long-term prognosis when associated with benign or treatable underlying disease.

Limitations of this study include the retrospective nature, the small number of cases identified, and the lack of standardization of diagnostics and treatment.

Conclusion

Diagnostic endoscopy is associated with a low incidence of gastrointestinal perforation. With prompt recognition and surgical repair, gastrointestinal perforation associated with endoscopy has a good prognosis. Infiltrative small intestinal disease in cats and preexisting duodenal ulceration in both species may predispose to gastrointestinal perforation. Gastrointestinal endoscopy is not contraindicated in patients with a higher risk of endoscopic perforation as long as the patient can be treated with immediate surgical repair.