Gastric or duodenal perforation and secondary septic peritonitis following therapeutic nonsteroidal anti-inflammatory drug administration

1 INTRODUCTION

The use of nonsteroidal anti-inflammatory drugs (NSAIDs) is common in both veterinary and human medicines. These drugs are used for both acute and chronic pain and they are proven to have anti-inflammatory, antipyretic, and analgesic properties.^1-4 These agents alter the body's inflammatory response through competitive inhibition of cyclooxygenase (COX), an enzyme within the arachidonic acid pathway. Inhibition of COX results in blockade of proinflammatory prostaglandins from arachidonic acid, accounting for the majority of the NSAID's therapeutic effects.^5-9 This blockade of COX also results in blockade of protective prostaglandins, and account for the majority of adverse effects.^{3, 10} Protective prostaglandins are important promoters of gastric bicarbonate and mucous production and also inhibit gastric acid secretion. In addition, they also promote epithelial cell migration and maintain mucosal blood flow.^{1, 6, 7, 11, 12} The 2 main isoforms of the COX enzyme are defined as COX-1 and COX-2, although a third has more recently been identified. The COX-1 isoform is expressed constitutively and accounts for the production of the majority of protective prostaglandins.^{8, 9, 13} The COX-2 isoform is an induced form and produces proinflammatory prostaglandins as well as other inflammatory mediators.^{8, 9, 14} Therefore, blockade or targeting of COX-2 alone may be advantageous therapeutically.¹⁵

Selective COX-2 NSAIDs were developed in part to reduce the adverse gastrointestinal (GI) effects recognized with nonselective COX inhibition.¹⁵ However, both selective COX-2 and nonselective NSAIDs continue to be associated with adverse effects.^16-24 Despite the increased use of COX-2-selective NSAIDs, there continue to be reports of gastric or duodenal perforation (GDP) associated with NSAID use in veterinary medicine.^{1, 17, 23, 25, 26}

In addition to the aforementioned mechanisms of GI injury, there are also prostaglandin-independent mechanisms that cause the stomach and duodenum to be at higher risk of ulceration and perforation after NSAID administration.^{7, 26} The major prostaglandin-independent mechanism occurs due to trapping of NSAIDs in the gastric mucosal epithelial cells due to alterations in pH. This results in mitochondrial and cellular injury, which leads to mucosal damage and increased permeability.^{7, 27, 28} A similar mechanism occurs in the small intestine where NSAIDs result in increased permeability; however, the enterocytes are also exposed to luminal contents such as bile acids, proteolytic enzymes, pancreatic secretions, and intestinal bacteria. The exposure to intestinal bacteria can also result in neutrophil chemotaxis and subsequent inflammation and ulceration.^{7, 29}

There have been veterinary case reports and studies describing GDP following specific NSAID administration, and veterinary studies comparing the adverse effects of different types of NSAIDs; however, there have not been any veterinary studies reporting on the frequency of perforation with use of different NSAIDs in dogs.^{1, 3, 17, 25, 26} The goal of this study was to report on which NSAIDs were associated with GDP in dogs presented to a university veterinary teaching hospital and to report on the frequency of the most commonly prescribed NSAIDs by the corresponding referring veterinary community during the same time period.

2 MATERIALS AND METHODS

Medical records were searched from January 2007 to March 2020 for dogs diagnosed with GDP. Dogs were included if they had a diagnosis of GDP documented during surgery and received NSAIDs within 7 days of the diagnosis of perforation. Dogs were excluded if they were determined to have other underlying conditions that could predispose to GDP. Dogs with liver dysfunction as assessed by increased bile acids, ammonia concentration, or other markers of liver dysfunction were excluded. Dogs with primary renal azotemia diagnosed based on increases in blood urea nitrogen or creatinine concentration not associated with prerenal or postrenal pathology were excluded. Dogs with GI neoplasia based on a biopsy or cytology or a previous diagnosis of neoplasia known to predispose to GI ulceration were excluded. Dogs with GI foreign body or gastric dilatation volvulus were also excluded as were dogs that did not have a complete medical record for review. Data collected included signalment, physical examination findings, NSAID type, NSAID dose (mg/kg), duration of prescription, locations of perforation (gastric vs. duodenal), and whether the dose was consistent with labeling instructions. NSAID overdose was defined as any dose 10% or greater than the high end of the labeled dose. In addition, concurrent medications including corticosteroids and other NSAIDs were also recorded.

Nine hundred and fourteen surveys were distributed to the referring veterinarians of Michigan State University Veterinary Teaching Hospital via fax and email. One hundred and sixty surveys were returned. The objective of the survey was to determine which NSAIDs were prescribed most frequently during the study period (Supporting Information S1). Two time periods were assessed to attempt to enable evaluation of recently developed NSAIDs such as grapiprant, which became available in March 2016. Referring veterinarians were not asked to provide specific numbers of NSAID prescription during the 2 time periods.

3 RESULTS

A total of 73 dogs were identified as having GDP during the study period. Of those cases, 39 were excluded because they did not receive NSAIDs, 1 dog had a predisposing factor (congenital intrahepatic portosystemic shunt), 2 dogs ingested large accidental overdoses of NSAIDs, and 1 dog had an incomplete medical record. Thirty dogs met full inclusion criteria and were reviewed. The median age was 6 years (range, 1–14 years). Median weight was 37.1 kg (17–92.7 kg). GDP occurred in the duodenum in 17 of 30 (56.7%) dogs, and in the pyloric region of the stomach in the remaining 13 of 30 (43.3%) dogs. Duration of prescription ranged from a single dose to 3 years with a median of 14 days.

Four of 30 dogs were administered more than 1 NSAID within 7 days of GDP. All 4 dogs received carprofen and an additional NSAID including deracoxib (2), ketoprofen (1), and aspirin (1). Only 1 dog had a washout period of 5 days between administration of carprofen and deracoxib. The remaining dogs received both NSAIDs concurrently. Three of 30 dogs received a combination of an NSAID and a corticosteroid within 7 days of GDP. All 3 dogs received oral prednisone. No dog receiving NSAIDs and steroids concurrently had a washout period. One dog received 2 NSAIDs (carprofen, firocoxib) and a steroid (dexamethasone sodium phosphate) within 7 days of GDP with no washout period. Four of 30 dogs received an overdose of an NSAID within 7 days of GDP. Three dogs received 2 times the labeled dose and 1 dog received 3 times the labeled dose with a median of 2 times the labeled dose. Overdoses involved different NSAIDs including meloxicam (n = 2) and firocoxib (n = 2). One dog received both an overdose of an NSAID and concurrent prednisone administration. Eighteen of 30 dogs received only 1 NSAID at the labeled dose. None of these 18 dogs had received carprofen, grapiprant, or aspirin within 7 days of GDP. Meloxicam was administered in 44.4% (8/18), firocoxib in 27.8% (5/18), deracoxib in 16.7% (3/18), and piroxicam in 11.1% (2/18) of these dogs with GDP.

Concurrent medications or supplements included tramadol in 11 dogs, gabapentin in 5 dogs, and 1 dog received each of the following: glucosamine, polysulfated glycosaminoglycan, enrofloxacin, cephalexin, cefpodoxime, chlorpheniramine, enalapril maleate, azathioprine, famotidine, and amantadine. Eighteen of 30 dogs received NSAIDs for orthopedic disease. Nine of those 18 dogs had recent orthopedic surgery, none of which were secondary to fracture repair. Four of 30 received NSAIDs for malignant neoplasia (leiomyosarcoma of the urinary bladder [n = 1], transitional cell carcinoma of the urinary bladder [n = 1], eyelid tumor [n = 1], and soft tissue sarcoma [n = 1]). Four of 30 dogs received NSAIDs for pain control after soft tissue surgery (lipoma removal [n = 1], neuter and gastropexy [n = 1], cryptorchid neuter [n = 1], and ovariohysterectomy [n = 1]), and 4 of 30 received NSAIDs for various other reasons.

The survival rate was 66.7% (20/30). Nine dogs were euthanized and 1 dog died naturally. Of the 10 dogs that were either euthanized or died, 70% (7/10) had duodenal perforations and 30% (3/10) had gastric perforations. All dogs that were euthanized or died were first taken to surgery and were euthanized or died intraoperatively or postoperatively. Dogs were euthanized or died due to complications including the need for biliary rerouting (n = 5), gastric or intestinal incisional dehiscence (n = 2), continued hemodynamic deterioration (n = 2), and acute respiratory distress syndrome (n = 1).

One hundred and sixty surveys (17.5%) were returned from referring veterinarians. COX-2 NSAIDs were the most commonly prescribed NSAID. Nonselective COX NSAIDs (enteric coated aspirin and piroxicam) were prescribed by fewer than 2% of referring veterinarians returning the survey over both time periods. Carprofen was the most frequently prescribed NSAID during both time periods evaluated (2007–2015 and 2016–2020) and was prescribed by 69.4% (111/160) of referring veterinarians during 2016–2020 and 71.8% (115/160) of referring veterinarians in the earlier study period (2007–2015). When the 2 study periods are combined, carprofen was prescribed most frequently in 70.6% (226/320) from 2007 to 2020. This was followed by meloxicam at 10.6% (34/320), deracoxib at 8.4% (27/320), firocoxib at 7.8% (25/320), aspirin at 1.5% (5/320), and others (grapiprant, piroxicam, etodolac) at 0.9%.

4 DISCUSSION

NSAIDs have previously been associated with GDP in dogs in previous veterinary reports.^{1, 17, 30} This study is the first reporting on GDP in clinical cases receiving different types of NSAIDs in dogs and the frequency of prescription of NSAIDs in the same referring veterinary community and the same time period.

COX-2-specific NSAIDs were the most commonly prescribed NSAID in the current study. Meloxicam has been shown to have selectivity for COX-2 inhibition in vitro.¹⁵ However, despite preferential selectivity for COX-2, meloxicam has been previously associated with GDP.^{17, 23, 30} Five dogs have been described with GI perforation associated with meloxicam administration. These patients received varying doses of meloxicam, but 3 of 5 dogs did receive doses according to the labeled dose.¹⁷ Multiple additional case reports of GDP in dogs receiving meloxicam have been published, including dogs that received labeled doses of meloxicam and a dog receiving an additional NSAID (aspirin) in addition to meloxicam.^{23, 30} In the current study, meloxicam was associated with the highest number of GDP in dogs receiving only 1 NSAID at the labeled dose (8/18) and in all dogs included in the study (11/30). In light of this, the survey results in the current study revealed that meloxicam was the second most commonly prescribed NSAID in both time periods (12% and 9%, respectively).

Firocoxib, like meloxicam, is considered a COX-2-selective NSAID and has a 384-fold selectivity for COX-2 in canine blood.³¹ This is the highest COX-2 selectivity of any NSAID in this study. Despite this significant selectivity for COX-2 firocoxib was associated with 28% of GDPs in dogs receiving only 1 NSAID at the labeled dose in this study. GDP secondary to firocoxib administration has not been specifically evaluated in dogs. In a study evaluating the long-term efficacy and safety of firocoxib in the treatment of dogs with osteoarthritis, 2.6% of dogs (1/39) died from a perforating duodenal ulcer a few days after receiving a double dose of firocoxib.³² Previous studies have failed to find significant adverse effects on the GI tract, based on endoscopic evaluation, following firocoxib administration.^{33, 34}

In the current study deracoxib was prescribed in 17% (3/18) of dogs with GDPs in the dogs receiving only 1 NSAID at the prescribed dose. Deracoxib is another COX-2-selective NSAID and has also been previously associated with GI ulceration and GDP. Deracoxib has an inhibitory concentration (IC₅₀) COX-1 to COX-2 ratio of 12, which is similar to carprofen, and represents more COX-2 selectivity. ^{31, 35} A retrospective study described 29 dogs diagnosed with GI tract perforation following the use of deracoxib. Many (90%) of these dogs received a higher than labeled dosage or at least 1 other NSAID or corticosteroid in close temporal association with deracoxib.¹ However, 10% of those dogs that received the appropriate dose of deracoxib with no other ulcerogenic medications were diagnosed with GI perforation. An additional case series also describes 3 dogs with duodenal perforation following deracoxib administration.³⁶ Labeled dosage was followed in 2 of 3 dogs in this case series and the remaining dog only received a slightly higher than the labeled dose (2.5 mg/kg) for a more chronic time period (21 days postoperatively). This suggests that overdoses may not be the only contributing factor to GDP following deracoxib administration in dogs. Firocoxib and deracoxib had relatively low frequency of prescription compared to carprofen in this study, particularly in the most recent time period, 2016–2020. However, these drugs accounted for the second and third highest incidence of perforation.

Piroxicam was the only nonselective COX NSAID to be associated with GDP in the 2 of 18 dogs that received only 1 NSAID at the labeled dose in this study. Both dogs were receiving piroxicam as an adjunctive treatment for urothelial tumors. No referring veterinarians listed piroxicam as the most commonly prescribed NSAID in either time period. There have been no studies specifically evaluating GDP following the use of piroxicam in dogs. The small sample of these cases makes it difficult to draw associations between piroxicam and GDP.

One of the most interesting results of the current study was that carprofen, another selective COX-2 NSAID, was by far the most commonly prescribed NSAID, yet it was not associated with any GDP in the 18 dogs that received only 1 NSAID at the labeled dose. Carprofen has been previously evaluated in multiple studies in dogs and has consistently been uncommonly associated with serious GI adverse effects.^{3, 18, 37, 38} In a study evaluating the adverse effects following long-term oral administration of several different NSAIDs on the GI tract, carprofen was associated with the lowest frequency of adverse effects.³⁹ An overdose of 10 times the recommended daily dose for 14 days only yielded hypoalbuminemia in 2 of 8 (25%) dogs and melena in 1 dog in a clinical field trial by the manufacturer. In the current study, carprofen was associated with 5 perforations, 1 in combination with an overdose of aspirin, 1 in combination with firocoxib and dexamethasone sodium phosphate, 1 in combination with ketoprofen, and 2 in combination with deracoxib. Although carprofen was infrequently associated with adverse GI side effects, it is only 1.74 times more selective than COX-2 and COX-1, which is much lower than both meloxicam and firocoxib.^{15, 31} These findings again support the notion that COX-2 selectivity is unlikely to be the only factor contributing to adverse GI side effects. When used alone and at labeled doses, carprofen is uncommonly associated with gastroduodenal perforation.

In the current study, only 60% (18/30) of dogs received only 1 NSAID at the labeled dose. Two dogs were prescribed firocoxib at a dose higher than the labeled dose, 1 dog received meloxicam twice daily due to owner error, and 1 dog received 10 times the prescribed dose of meloxicam due to a human pharmacy error. The reason for overdose of firocoxib in 1 dog remains unknown. The role of overdose in this study remains unknown as the study was not designed to make this comparison.

Four of thirty dogs in this study received 2 NSAIDs within 7 days of GDP. Two of these dogs received an additional NSAID by their owners to provide more pain coverage. The remaining 2 dogs received 2 NSAIDs at the recommendation of a veterinarian, 1 of which received an injection postoperatively and then was sent home on a different oral NSAID, and the other dog was prescribed 2 oral NSAIDs. All 3 dogs receiving an NSAID and a steroid were prescribed these medications by a veterinarian. In 2 of 3 dogs, a steroid was prescribed due to an immune-mediated process unrelated to the GI tract. The 1 dog receiving 2 NSAIDs and a steroid was given an injection of a steroid and an injection of an NSAID following a negative explorative surgery for a possible foreign body and then was sent home on a different oral NSAID. The reason for prescription remains unknown. Based on these findings, the prescription of multiple NSAIDs at 1 time or concurrent NSAID and steroid administration should be avoided.

This study had several limitations. The retrospective nature of this study precluded any assessment of clinical decisions made surrounding prescription of an NSAID and any other concurrent medications. Although we were able to evaluate for the frequency of prescription of individual drugs, a direct comparison between prescribing frequency and incidence of GDP could not be determined. Additionally, survey results were based on referring veterinarian response and may not accurately reflect frequency of prescription across the referring veterinary community.

5 CONCLUSION

Importantly, labeled dosing for NSAIDs should be followed and NSAIDs should not be administered concurrently with other NSAIDs or with steroids, and care should be taken to avoid overdoses of NSAIDs. Carprofen is unlikely to be associated with gastroduodenal perforation when given as a single NSAID at the labeled dose. Prospective large-scale monitoring is needed to further evaluate adverse effects associated with specific NSAID administration.

CONFLICT OF INTEREST

The authors declare no conflict of interest.