Challenging diagnosis — Icterus associated with a single perforating duodenal ulcer after long-term nonsteroidal antiinflammatory drug administration in a dog

Abstract

A dog developed icterus, vomiting, and anorexia 2 wk after orthopedic surgery and treatment with meloxicam for approximately 1 y. Exploratory laparotomy revealed a single perforated duodenal ulcer. The most likely cause of the hyperbilirubinemia was intrahepatic cholestasis resulting from peritonitis associated with the perforation.

A 5-year-old, 55 kg, neutered male German shepherdcross was presented to the University of Saskatchewan, Western College of Veterinary Medicine (WCVM), for a ruptured cranial cruciate ligament of the left stifle. The dog had been administered meloxicam (Metacam; Boehringer Ingelheim, Burlington, Ontario), 0.1 mg/kg bodyweight (BW), PO, q24h, for chronic thoracic limb lameness for approximately 1 y. Prior to surgery, a routine data base was obtained and revealed the following abnormalities: leukocytosis with neutrophilia and left shift, slight elevation of alanine aminotransferase (ALT), and increased hematocrit (Table 1; 28/10). A tibial plateau levelling osteotomy was performed, and the dog was discharged from the hospital 2 d following surgery with instructions for the owner to continue the administration of meloxicam, 0.1 mg/kg BW, PO, q24h for another 8 d, to control postsurgical pain.

Table 1.

Hematological and serum biochemical values obtained on various dates (d/mo) from a dog with icterus associated with a single perforating duodenal ulcer after long-term nonsteroidal antiinflammatory drug administration

	Reference range	28/10a	15/11b	16/11	18/11	19/11	20/11	05/12
ALT	19 to 59 U/L	177	77	60	12	—	6	85
AP	9 to 90 U/L	—	114	118	155	—	220	113
GGT	0 to 8 U/L	—	—	—	1.5	—	—	9
Urea	3.5 to 11.4 MMOL/L	—	1.8	2.0	19	—	1.0	1.9
Albumin	28 to 38 G/L	—	19	16	—	—	21	—
Glucose	3.1 to 6.3 MMOL/L	—	—	—	—	—	2.9	—
Bile acid (pre/post)	0 to 10/20 UMOL/L	—	12	—	76	—	—	—
Total Bilirubin	1 to 4 UMOL/L	—	136	127	—	—	40	9
Total Solids	56 to 74 g/L	—	49	40	43	53	—	—
Total Protein	55 to 71 g/L	—	41	35	38	—	47	—
WBCs	4.8 to 13.9 × 109/L	17.2	32.3	36.4	30.02	45.9	—	19.8
Segs	3.0 to 10.0 × 109/L	13.588	26.809	28.028c	3.8	36.261	—	15.84
Bands	0.0 to 0.1 × 109/L	0.516	1.938	4.732	1.52	5.049	—	0.396
Eos	0.0 to 1.1 × 109/L	—	—	—	1.14	—	—	1.386
Lymphs	1.2 to 5.0 × 109/L	—	0.969	—	1.52	—	—	—
Monos	0.08 to 1.0 × 109/L	—	1.938	2.184	3.54	2.295	—	—
RBCs	5.2 to 8.2 × 1012/L	8.61	—	4.35	—	3.5	—	—
RDW	13.8% to 17.6%	—	—	—	79	13.5	—	17.9
Hgb	128 to 196 g/L	198	—	100	0.248	80	—	—
Hct	0.365 to 0.573 L/L	0.581	—	0.307	320	0.249	—	—
MCHC	335 to 357 g/L	—	326	327	22.4	323	—	331
MCH	22.5 to 25.5 pg	—	—	—	—	—	—	—
Chloride	103 to 118 MMOL/L	—	121	—	18/11	—	—	—

ALT — alanine aminotransferase; AP — alkaline phosphatase; GGT — gamma glutamyl transferase; WBCs — white blood cells; Segs — segmented neutrophils; Eos — eosinophils; Lymphs — lymphocytes; Monos — monocytes; RBCs — red blood cells; RDW — red cell distribution width; Hgb — hemoglobin; Hct — hematocrit; MCHC — mean corpuscular hemoglobin concentration; MCH — mean cell volume

^aprior to tibial plateau levelling osteotomy

^bday of initial presentation for ulcer

^c1 + toxic change

Two weeks later, the dog was referred again to the WCVM with a 3-day history of inappetance, vomiting, lethargy, tachycardia, and increased respiratory effort. The referring veterinarian had administered IV fluid therapy (lactated Ringer’s solution; Abbott Laboratories, Ville St. Laurent, Quebec); diazepam (Diazepam; Sabex, Boucherville, Quebec), 0.23 mg/kg BW, IV, q12h; meloxicam, 0.1 mg/kg BW, PO, q24h; and cefazolin (Cefazolin; Novopharm, Toronto, Ontario), 18 mg/kg BW, IV, q8h for 1 d, followed by enrofloxacin (Baytril; Bayer, Toronto, Ontario), 2.3 mg/kg BW, IV, q12h, and ampicillin (Ampicillin; Novopharm), 18 mg/kg BW, IV, q12h for 2 d. An in-house biochemical panel and complete blood cell (CBC) count performed by the referring veterinarian had revealed the following abnormalities: hyperbilirubinemia (25 μmol/L; reference range, 0 to 15 μmol/L), hypoglycemia (2.41 mmol/L; reference range, 4.28 to 6.94 mmol/L), elevated ALT (203 U/L; reference range, 10 to 100 U/L), and leukocytosis (17.3 × 10⁹/L, reference range, 3.6 to 11.5 × 10⁹/L).

At presentation, the physical examination revealed that the dog was weak but ambulatory, and had yellow-tinged sclera, mucous membranes, and skin. He had a weight-bearing lameness on the left pelvic limb. Abdominal bearing palpation elicited a painful response. Rectal temperature was 39.7°C, heart rate was 78 beats/min, and the dog was panting. Hematological abnormalities included a leukocytosis with neutrophilia and left shift, lymphopenia, and cytosis monocytosis. Serum chemical abnormalities included hypoproteinemia, hypochloremia, low urea and albumin, bilirubinemia, and elevation of alkaline phosphatase (AP) and ALT (Table 1; 15/11). Postprandial bile acids were within normal limits (12 μmol/L; reference range, 0 to 20 μmol/L). A coagulation panel revealed only clinically insignificant changes (activated partial thromboplastin time: 14.2 s; reference range, 9.6 to 13.8 s). Abdominal radiographs revealed a mild decrease in serosal detail in the mid-abdomen, consistent with peritoneal effusion. An abdominal ultrasonographic examination revealed peritoneal effusion and a hyperechoic liver (Figure 1). The gall bladder contained a moderate amount of inspis-sated, immobile bile sludge (Figure 2). Biliary obstruction was not evident. On abdominocentesis, 10 mL of a yellow, turbid, milky fluid was recovered (specific gravity: 1.016; total solids: 20 g/L; nucleated cells: 122 × 10⁹/L; red blood cells: 0.033 × 10¹²/L). Cytologic /examination revealed nonseptic, suppurative inflammation. As no etiologic agent was evident but possible bile pigments were seen, a tentative diagnosis of bile peritonitis was made; this warranted an exploratory laparotomy.

Figure 1.

Peritoneal effusion located in the region of the liver.

Figure 2.

“Sludge” in the gall bladder.

A transfusion of freshly thawed frozen plasma (360 mL) was initiated prior to surgery because of the hypoalbuminemia. The abdomen contained a moderate amount of brown tinged fluid, a sample of which was collected for aerobic and anaerobic culture. Other abnormal findings included discoloration of the liver, specifically in the right lateral and caudate lobes. A perforation (about 0.5 cm diameter) was present in the proximal part of the duodenum, just distal to the pylorus. This area of the duodenum also appeared thickened. About 3 to 4 cm of the duodenum, including the perforation and thickened area, was resected and submitted for histopathologic examination. A surgical stapling instrument (LDS; United States Surgical Corporation, Norwalk, Connecticut, USA) was used to ligate and divide the cranial pancreaticoduodenal artery, the common bile duct, and the major pancreatic duct. A gastroduodenostomy (Bilroth I) was performed. A cholecystoduodenostomy was performed on the antimesenteric surface of the descending duodenum, distal to the gastroduodenostomy site. Punch biopsies were taken from the liver and submitted for histopathologic examination. The abdomen was lavaged with sterile 0.9% sodium chloride (Abbott Laboratories) and the incision was closed routinely.

Following surgery, the transfusion of freshly thawed frozen plasma (120 mL) was continued and IV fluids (Normosol-R; Abbott Laboratories) were administered at 3 mL/kg BW/h. The abdominal fluid was negative on aerobic, anaerobic, and enrichment cultures. These results may have been due to the pre- and perioperative antimicrobial treatment. Antibiotic therapy with cephalexin, 22 mg/kg BW, IV, q8h, and enrofloxacin, 10 mg/kg BW, IV, q24h, was continued postoperatively. Additional treatment consisted of oxymorphone (Sabex), 0.05 mg/kg BW, IM, as needed); famotidine (Pepcid; Merck Frosst, Kirkland, Quebec), 0.5 mg/kg BW, IV, q12h; metoclopramide (Metoclopramide; Sabex), 0.25 mg/kg BW, IV, q6h; and heparin (Hepalean; Organon, Toronto, Ontario), 75 U/kg BW, SC, q8h. Oral feeding was initiated 48 h postoperatively. The dog was monitored in the intensive care unit for 3 d postoperatively and serial CBC counts and biochemical panels were obtained (Table 1; 16/11, 18/11, 19/11, 20/11).

Histologic evaluation of the liver biopsies confirmed cholestasis and hepatitis. Several canaliculi were plugged with bile and, in all zones, hepatocytes and Kupffer cells were swollen and contained light brown granular material. Hall’s staining confirmed that this material was bile. Staining for hemosiderin and lipofuscin (Perle’s Prussian Blue and Schmorl’s) gave negative results. Small, discrete inflammatory foci were disseminated throughout the parenchyma. These foci consisted of clusters of macrophages mixed with a few plasma cells and neutrophils. Staining with rhodanine revealed abundant copper granules within the macrophages, as well as mild copper accumulation in hepatocytes of all acinar zones. No organisms were associated with the inflammatory cells (results of staining with Brown and Brenn and periodic acid-Schiff were negative). Single necrotic hepatocytes were common, but there were no large foci of necrosis. The connective tissue of portal tracts and the capsule contained very mild neutrophil infiltrates. Moderate neutrophil infiltrates occasionally surrounded terminal venules. Mesothelial cells on the capsular surface were markedly swollen. Histopathologic examination did not reveal anything specific about the resected perforated duodenal tissue. The bile duct papilla was found to be intact; however, inflammation and fibrosis associated with the ulcer extended close to the papilla. No neoplastic cells or infectious organisms were present at this site. The submucosa contained a few large aggregates of lymphocytes and plasma cells. An exudate of fibrin and neutrophils was adhered to the serosal surface and mesentery. Histopathologic diagnosis was chronic duodenal ulceration (with acute perforation) and a subacute, multifocal, moderate, pyogranulomatous hepatitis, with cholestasis and copper accumulation.

The dog was discharged 6 d following the surgery with a 2-week course of metoclopramide, 0.25 mg/kg BW, PO, q6h; a 3-month course of famotidine, 0.5 mg/kg BW, PO, q12h; a 7-day course of enrofloxacin, 5 mg/kg BW, PO, q24h; and cephalexin, 22 mg/kg BW, PO, q8h. Follow-up calls 2, 4, and 6 mo after discharge revealed continued improvement. Occasional vomiting (2 to 3 times/wk) was noted for the first 3 mo after discharge.

The most common causes of gastroduodenal ulcer disease (GDUD) in dogs and cats include pharmaceuticals, neoplasia, liver disease, and shock (1). Systemically administered corticosteroids and nonsteroidal antiinflammatory drugs (NSAIDs) are the medications most commonly associated with GDUD. The pathophysiology of upper gastroduodenal erosion and ulceration caused by NSAIDs has been widely reported (2–6). The pathophysiology of ulcers caused by liver disease is not well described. Increased gastric acidity due to an inability of the liver to remove gastrin and histamine from the circulation has been suggested as a cause (1).

The combination of liver disease and long-term NSAID therapy, or either condition on its own, could lead to GDUD. The elevated liver enzyme in the preoperative workup suggests that the dog may have already had an underlying liver disease. However, even though the location of the ulceration would be consistent with underlying liver disease, the liver changes were not severe enough to cause ulcers. This is supported by normal bile acid test results. The histopathological finding of mild copper accumulation in hepatocytes of all acinar zones could indicate early stages of a copper storage disorder, or it could be secondary to cholestasis (7). Hepatocellular injury due to cholestasis (and possibly the toxic effects of copper) and inflammatory mediators or endotoxins entering the portal circulation from the duodenum could account for the hepatitis found on histopathologic examination.

Duodenal ulceration is another potential cause of the icterus in this patient. Extrahepatic cholestasis could result from inflammation and fibrosis near the bile duct papilla. In humans, gastroduodenal ulcer disease is rarely reported to cause obstructive jaundice (10). In our patient, no evidence of biliary obstruction was found on ultrasonography or exploratory surgery. The histopathologic changes noted in the mucosa and submucosa near the bile duct papilla would suggest that bile flow was possibly impeded. However, with an extrahepatic obstructive lesion, significant increases in AP and gamma glutamyl transferase (GGT) would have been expected (11).

Although many papers and textbooks describe GDUD after NSAID administration, to our knowledge, icterus has not been described as a clinical sign (5). Administration of NSAIDs is associated with both erosive lesions within the stomach (6) and duodenal ulcers (3,4). The environmental stress and periods of hypotension associated with surgery may have exacerbated possible preexisting duodenal erosion. In the patient described, existing ulceration and perforation led to peritonitis, which could be the cause of the observed intrahepatic cholestasis. Severe bacterial infection, in any extrahepatic location, is a well-recognized cause of intrahepatic cholestasis in humans. Possible pathophysiologic mechanisms include interference of endotoxins with bile flow, formation, or both (8). Marked bile retention within canaliculi and hepatocytes, Kupffer’s cells, or both, but with minimal hepatocellular inflammation and necrosis, is considered to be consistent with cholestasis due to nonhepatic bacterial infection in humans (9). This disease has not been well characterized in veterinary medicine, but it has been reported in 5 cases (9). Similar to our case, 1 type of infection was associated with peritonitis due to a rectal tear. All cases showed only slight increases in liver enzymes, accompanied by significant elevation of bilirubin and hypoalbuminemia. Intrahepatic cholestasis associated with peritonitis is supported by the histopathologic examination that did not reveal evidence of hepatic neoplasia or changes suggestive of chronic progressive hepatitis and by the normal liver function results.

Osteoarthritis and pre- and postoperative pain are commonly controlled with NSAIDs in veterinary medicine. Newer generations of NSAIDs that are considered to be COX-2 selective promise fewer side effects on the gastrointestinal tract. It is generally agreed that meloxicam is somewhat selective for COX-2 and, therefore, provides a lower frequency of adverse effects (12,13). Despite this, duodenal ulceration after meloxicam administration has been reported (5). It is advisable, even with newer, more COX-2 selective generations of NSAIDs, to monitor patients closely and to clearly inform owners about early signs of side effects and overdosing. It should also be noted that various factors, including dehydration, inadequate water intake, hypovolemia, renal disease, hypotension during surgery, and long-term administration, can exacerbate the toxicity of NSAIDs. CVJ