This work was conducted at the University of Georgia College of Veterinary Medicine and was partially funded by Schering-Plough. The research was presented as a poster at the 2008 ACVIM Forum.
Effects of Firocoxib and Tepoxalin on Healing in a Canine Gastric Mucosal Injury Model
Version of Record online: 27 NOV 2008
Copyright © 2008 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 23, Issue 1, pages 56–62, January/February 2009
How to Cite
Goodman, L., Torres, B., Punke, J., Reynolds, L., Speas, A., Ellis, A. and Budsberg, S. (2009), Effects of Firocoxib and Tepoxalin on Healing in a Canine Gastric Mucosal Injury Model. Journal of Veterinary Internal Medicine, 23: 56–62. doi: 10.1111/j.1939-1676.2008.0226.x
- Issue online: 6 JAN 2009
- Version of Record online: 27 NOV 2008
- Submitted May 28, 2008; Revised July 12, 2008; Accepted September 29, 2008.
- Gastric healing;
Background: Little is known about the effect of dual cyclooxygenase (COX) and lipoxygenase inhibition on canine gastric mucosal healing.
Objective: This study compares the effects of putative dual COX and 5-lipoxygenase inhibition with that of COX-2 selective inhibition on gastric mucosal lesion healing in dogs.
Animals: Six normal adult mixed-breed research dogs.
Methods: Gastric body and pyloric lesions were induced by endoscopic biopsy. Dogs were treated with tepoxalin, firocoxib, or placebo for 7 days in a randomized 3-way crossover study design. Healing was evaluated on days 2, 4, and 7 of treatment by endoscopic lesion scoring. Eicosanoid concentrations in plasma and at the lesion margins were determined on days 2, 4, and 7. Repeated measures analyses were performed. All hypothesis tests were 2-sided with P < .05. Multiple comparisons were adjusted using Tukey's test.
Results: Significant treatment differences were noted in the pyloric lesion area measurements. Overall, the firocoxib group had larger lesions than the placebo (P= .0469) or tepoxalin (P= .0089) groups. Despite larger pyloric lesions in the firocoxib group, mucosal prostaglandin production did not differ significantly from placebo. In contrast, the tepoxalin group had significantly lower pyloric mucosal prostaglandin production compared with the firocoxib (P < .0001) or the placebo (P < .0001) groups but pyloric lesions were not significantly larger than those of the placebo group (P= .7829).
Conclusion: COX-2 inhibition by firocoxib slowed wound healing by a mechanism independent of prostaglandin synthesis. Suppression of mucosal prostaglandin production by tepoxalin did not alter mucosal lesion healing compared with placebo.
The process of gastric mucosal lesion healing is complex and the mechanisms by which it occurs are not fully understood.1,2 The role of the cyclooxygenase (COX) isoenzymes in the healing process has been studied. Although cyclooxygenase-1 (COX-1) plays an important role in mucosal homeostasis, it may not play a major role in mucosal wound healing under normal conditions. Ulcer healing has been shown to be unaffected in COX-1-deficient mice or in mice treated with a COX-1-specific inhibitor.3 In contrast, inhibition of cyclooxygenase-2 (COX-2) has been linked to slower gastric mucosal lesion healing.4 Although present at negligible concentrations in healthy tissue, COX-2 expression is increased on the edges of gastric ulcers5 and is thought to influence several steps in mucosal wound healing. Unlike the COX isoenzymes, the roles of 5-lipoxygenase (5-LO) and its product leukotriene B4 (LTB4) in mucosal lesion healing are largely unknown. Most research has focused on their roles in gastrointestinal lesion induction.
5-LO metabolizes arachidonic acid to proinflammatory paracrine hormones such as LTB4. Produced primarily by neutrophils and macrophages, LTB4 functions as a potent chemoattractant, directing leukocyte migration toward sites of inflammation.6–8 In addition, LTB4 activates neutrophils, increases their adhesion to the vascular endothelium, and enhances the release of proinflammatory cytokines by macrophages and lymphocytes.9,10
LTB4 is thought to play a role in nonsteroidal anti-inflammatory drug (NSAID)-induced gastrointestinal mucosal damage.11 Studies have shown that administration of NSAIDs is associated with enhanced gastric mucosal synthesis of LTB4.12 Increased LTB4 concentration may stimulate chemotaxis and adhesion of leukocytes to the mucosal vascular endothelium resulting in gastrointestinal mucosal damage.11 Consistent with this hypothesis, leukocytes appear to be activated and attracted to the gastrointestinal tract during treatment with NSAIDs.13–15 This has been shown to be an early and pivotal event in the process of NSAID-induced mucosal injury.16–18 Furthermore, leukotriene inhibition mitigates NSAID-induced gastropathy.19 When dual COX and 5-LO inhibitors such as tepoxalin are administered to rats, indomethacin-induced neutrophil adhesion and migration in the mesenteric venules does not occur.20,21 Pretreatment with lipoxygenase inhibitors and leukotriene antagonists protects against mucosal damage in rats treated with either aspirin or indomethacin.22
The purpose of the study reported here was to evaluate the effect on mucosal wound healing of dual COX and 5-LO inhibition compared with that of COX-2 selective inhibition. In the model utilized in the study, gastric lesions were induced by biopsy, and subsequently tepoxalin,a a dual inhibitor, and firocoxib,b a COX-2-selective inhibitor, were administered. We hypothesized that COX-2 selective inhibition would delay gastric mucosal wound healing compared with both placebo and dual COX and 5-LO inhibition.
A 2nd goal of the study was to evaluate the in vivo activity of firocoxib and tepoxalin in the context of wound healing by measuring eicosanoid concentrations in the blood and gastric mucosa. The production of thromboxane B2 (TxB2) by platelets reflects COX-1 function, whereas plasma prostaglandin E2 (PGE2) production by lipopolysaccharide (LPS)-stimulated WBCs is an indicator of COX-2 activity.23 Gastric mucosal prostaglandin synthesis reflects both COX-1 and COX-2 activity because both COX isoforms and their associated prostanoid synthases are capable of producing mucosal prostaglandins.24,25 In the rat small intestine, inhibition of COX-1 alone results in a transient decrease in mucosal PGE2 concentration, which rebounds within 12 hours, presumably due to subsequent induction of COX-2 and production of prostaglandins.24 Therefore, inhibition of both isoforms is necessary to decrease mucosal prostaglandin E1 (PGE1) and PGE2 synthesis for longer than 12 hours.
Materials and Methods
Six adult neutered male and female dogs aged 5–10 years were the subjects in this study. The dogs are part of a research colony at the University of Georgia and have chronic unilateral osteoarthritis of the stifle joint because of previously induced unilateral cranial cruciate ligament injury. They have no history of gastrointestinal disease. Before inclusion in the study, all dogs were determined to be clinically normal based on physical examination findings and results of a plasma biochemical analysis, CBC, and urinalysis. In addition, endoscopic evaluation of the stomach was performed on day 0 of each treatment period and all dogs included in the study had normal findings. Dogs were fasted for 12 hours before each endoscopy. All dogs were cared for according to the principles outlined in the NIH Guide for the Care and Use of Laboratory Animals.
In a randomized, blinded, crossover study design, all dogs were anesthetized on days 0, 2, 4, and 7 of each of the 3 treatment periods. Venipuncture and endoscopic evaluation and biopsy of the gastric mucosa were performed and samples were used for eicosanoid measurements. TxB2 concentrations as well as PGE2 and LTB4 stimulation were measured in the blood samples. PGE1 and PGE2 synthesis as well as LTB4 concentrations were evaluated in pyloric and gastric body mucosal biopsy specimens.
Dogs were premedicated with acepromazine (0.03 mg/kg IM) and atropine (0.02 mg/kg IM). Anesthesia was induced with propofol (3–4 mg/kg IV to effect) and was maintained with isoflurane. On day 0, gastric body and pylorus lesions were induced by endoscopic biopsy. In each dog, 6 biopsy specimens were taken from the pyloric antrum and 6 biopsy specimens were taken from the fundus. Each biopsy specimen weighed a minimum of 2.0 mg. If an excessively small amount of tissue was obtained when a biopsy was performed, an additional biopsy was performed at the same location until at least 2.0 mg of tissue had been removed. The dual purpose of these biopsies was to induce lesions of consistent size and depth as well as to obtain tissue samples necessary for baseline mucosal eicosanoid determinations. Once during the study period, representative biopsy specimens were taken from sites other than these lesions and were submitted to a single pathologist for histologic analysis of biopsy depth. After lesion induction, the dogs were randomized and given therapeutic doses of tepoxalin (20 mg/kg PO once on day 1 and then 10 mg/kg q24 h), firocoxib (5 mg/kg PO q24 h), or placebo for 7 days.
On days 2, 4, and 7 of the treatment period, endoscopy was performed and subjective assessments were made with regard to lesion depth and mucosal erythema. Endoscopic scoring guidelines are listed in Figure 1. Objective measurements of the mucosal lesion area were determined by software analysisc of the photographs of each lesion. The photographs were taken perpendicular to the plane of the lesion at a standardized distance of 3 mm from the surface of the lesion as determined by a marked probe. In addition, the lesion margins of 2 antral lesions and 2 fundic lesions were biopsied on each of days 2, 4, and 7. A minimum of 2.0 mg of tissue was sampled from each lesion and was used for eicosanoid measurements. Thus, by day 7, the margin of all 6 lesions had been sampled no more than once during the study period. Once the margin of a lesion was sampled, the lesion was no longer scored for wound healing. A 21-day washout period was observed between treatments. The duration of the washout period was influenced by a study in which 4 mm2 pyloric mucosal lesions induced by surgical excision in nonmedicated rats healed completely within 2 weeks.26
Blood TxB2synthesis– 6 mL of blood was collected in a siliconized glass tube, immediately placed into a 37 °C water bath, and incubated for 1 hour. Indomethacind at a final concentration of 30 μM was added to stop further thromboxane synthesis and TxB2 was measured as previously described.27
Blood PGE2synthesis– 500 μL of sodium-heparinized blood was placed in a microcentrifuge tube. Bacterial LPS(Escherichia coli serotype 127:B8)e was added to each tube to stimulate PGE2 production, and PGE2 was measured as previously described.27
Gastric mucosal PGE1synthesis– all the endoscopic biopsy specimens were processed within 8 minutes after collection from the stomach. Specimens weighing <2.0 mg were discarded. Prostaglandin synthesis was stimulated by mincing as previously described.27
Gastric mucosal PGE2synthesis– samples were prepared identically to the aforementioned PGE1 protocol.
A repeated measures model that recognized multiple observations as belonging to the same dog was used to test for differences in gastric healing measurements among drugs and days. Multiple comparisons were adjusted using Tukey's test. An unstructured covariance structure was used in all repeated measures models. All hypothesis tests were 2-sided and the significance level was α= .05. The repeated measures analysis was performed by PROC MIXED in SAS.f
In this randomized study design, dogs received firocoxib, tepoxalin, or placebo in each treatment period. After randomization, 5 different sequences of drugs were administered. Two dogs received tepoxalin first followed by placebo and then firocoxib. One dog each received tepoxalin followed by firocoxib and then placebo; firocoxib followed by placebo and then tepoxalin; firocoxib followed by tepoxalin and then placebo; and finally placebo followed by tepoxalin and then firocoxib.
During the study period, 2 dogs in the tepoxalin treatment group vomited once, 1 dog in the firocoxib treatment group vomited once, and 1 dog in the firocoxib treatment group vomited twice. No change in appetite was noted in any dog at any time during the study. The representative pyloric and gastric body biopsies evaluated by a single pathologist were found to consist primarily of mucosa with some specimens penetrating into the submucosa.
Objective and Subjective Endoscopic Results
There were significant overall treatment differences for the pyloric lesion area measurements (P= .0081) with the firocoxib treatment group having larger lesions than either the placebo (P= .0469) or the tepoxalin (P= .0089) treatment groups. This treatment difference occurred regardless of the sequence of drug administration. Overall pyloric lesion sizes in dogs treated with tepoxalin were not significantly different from those in dogs that received placebo (Fig 2).
No significant overall treatment differences were noted in the following: gastric body lesion size (P= .0915), pyloric lesion erythema (P= .0880), body lesion erythema (P= .4508), pyloric lesion depth (P= .1204), or body lesion depth (P= .5673).
Plasma PGE2 Concentration
Overall plasma PGE2 concentrations were significantly lower in animals treated with firocoxib (P <.0001) or tepoxalin (P < .0001) compared with those that received placebo. Overall tepoxalin and firocoxib plasma PGE2 concentrations were not significantly different (Fig 3).
Serum TxB2 Concentration
Overall serum TxB2 concentrations were significantly lower in dogs treated with tepoxalin compared with those that received firocoxib (P < .0001) or placebo (P <.0001). There were no significant overall differences in TxB2 concentrations in dogs that received placebo compared with those treated with firocoxib (Fig 4).
Plasma and Mucosal LTB4 Concentrations
No significant decrease was noted in plasma, pyloric mucosa, or body mucosa LTB4 concentrations in dogs that received placebo, firocoxib, or tepoxalin at any time in this study (Fig 5). In addition, in the tepoxalin group, there was no significant decrease between baseline plasma and mucosal LTB4 concentrations compared with those on days 2, 4, and 7.
Pyloric and Body Mucosal PGE1 and PGE2 Synthesis
Overall pyloric and body mucosal PGE1 and PGE2 synthesis was significantly lower in dogs treated with tepoxalin compared with those that received placebo or firocoxib (P < .0001 for all comparisons). In dogs that received placebo or firocoxib, there was no significant difference in either PGE1 or PGE2 synthesis in the pylorus or gastric body on days 2, 4, or 7 (Fig 6).
Firocoxib and tepoxalin have different effects on gastric mucosal wound healing as determined by changes in mucosal lesion area measurements. The results of this study indicate that, in vivo, firocoxib is highly COX-2 specific, does not alter mucosal prostaglandin concentrations compared with placebo, but slows mucosal wound healing compared with both tepoxalin and placebo. Therefore, in this model, the selective suppression of COX-2 by firocoxib slowed mucosal wound healing by a mechanism independent of prostaglandin synthesis. In contrast, tepoxalin nonspecifically inhibits both COX isoforms, significantly decreasing the mucosal synthesis of prostaglandins, but does not alter mucosal wound healing compared with placebo. Furthermore, tepoxalin did not measurably decrease mucosal LTB4 production. Therefore, despite its suppression of prostaglandin mucosal synthesis and its inability to inhibit mucosal LTB4 production, tepoxalin favored lesion healing compared with firocoxib through an unknown mechanism.
COX-2 inhibition hinders several steps in the mucosal wound-healing process by prostaglandin-dependent and prostaglandin-independent mechanisms. For example, COX-2 but not COX-1 selective NSAIDs inhibit gastric epithelial cell proliferation and migration by a prostaglandin-independent mechanism.1,30 In addition, both COX-2 selective NSAIDs and nonselective NSAIDs inhibit angiogenesis by prostaglandin-dependent and prostaglandin-independent mechanisms.31 Although PGE2 produced by COX-2 has been shown to stimulate vascular endothelial growth factor expression,32 the prostaglandin-independent anti-angiogenic mechanisms are not fully understood. In the study reported here, COX-2 inhibition in firocoxib-treated dogs led to slower wound healing by a mechanism independent of prostaglandin synthesis.
Tepoxalin administration, despite its inhibition of both COX isoforms, favored wound healing compared with firocoxib. This is surprising given that prostaglandin-dependent mechanisms of mucosal lesion healing may have been impaired by the significantly decreased capacity for mucosal prostaglandin synthesis. Furthermore, the mechanism by which tepoxalin favored wound healing compared with firocoxib did not involve the inhibition of mucosal LTB4 synthesis. Tepoxalin administration did not lower mucosal or blood LTB4 concentrations as compared with placebo and there was no significant decrease in the tepoxalin group between LTB4 concentrations at baseline compared with those on days 2, 4, and 7 in either mucosal biopsy specimens or blood. In this canine model, tepoxalin administration favored wound healing compared with firocoxib despite potential inhibition of prostaglandin-dependent mucosal healing mechanisms and no suppression of mucosal LTB4 production.
In this study, administration of therapeutic doses of tepoxalin to dogs for 7 days did not measurably inhibit 5-LO activity. Although tepoxalin is considered a dual COX and 5-LO inhibitor in both humans and dogs, in humans, it is rapidly converted to a metabolite that loses it 5-LO inhibitory activity.33 A similar scenario may occur in dogs. In contrast to the findings in our study, an in vivo canine study found that tepoxalin significantly decreased LTB4 concentrations in the blood and the gastric mucosa at day 10 of treatment but not on day 3.29 It is unknown whether a similar treatment effect would have been noted in our study if the tepoxalin treatment duration had been extended to 10 days.
The in vivo COX-inhibitory activity of firocoxib and tepoxalin was evaluated by measuring eicosanoid concentrations in the blood and the gastric mucosa. The results indicate that firocoxib is highly COX-2 selective, having little effect on COX-1 activity. Firocoxib caused the greatest decrease in plasma PGE2 production but had no significant effect on serum TxB2 concentrations or mucosal PGE1 and PGE2 synthesis compared with placebo. In contrast, tepoxalin was inhibitory to both COX isoenzymes, significantly decreasing serum TxB2 concentrations, mucosal PGE1 and PGE2 production, and plasma PGE2 synthesis compared with placebo.
The gastric mucosal injury model developed by this laboratory and used for the 1st time in this study appears to be effective in evaluating treatment effect on mucosal healing. Based on histologic evaluation of representative biopsy specimens, the wounds generated in this model involved primarily mucosa and occasionally penetrated into the submucosa. Their depth was sufficient to monitor wound healing over a 7-day period without causing clinical changes in the study dogs. A limitation to this model is that the subjective lesion depth was difficult to score endoscopically, a problem noted in other studies utilizing endoscopic scoring of gastric ulcerations. A study of horses comparing endoscopic and histopathologic characterization of gastric ulcers suggested that the endoscopist could not accurately predict ulcer depth, especially in ulcers that appeared superficial.34 It is therefore possible that a treatment difference in pyloric lesion depth was present but not detected.
Because prostaglandin production can occur during tissue biopsy,35 a concern with this model is the potential for falsely increased prostaglandin concentrations in response to tissue handling. To minimize this effect, all biopsy specimens were processed within 8 minutes of collection. Any biopsy specimen processed outside this time limit was eliminated from the study. In addition, the original supernatant from each biopsy specimen was discarded. New buffer was added, and the samples were vortexed for 3 minutes to induce prostaglandin synthesis. Therefore, prostaglandin concentrations in these samples are a function of the ability of the sample to produce prostaglandin, and not of baseline prostaglandin concentrations in the sample.
Limitations of this study include small sample size resulting in an increased possibility of type 2 error and a lack of data regarding Helicobacter spp. infection status. Gastric helicobacter-like organisms have been associated with increased gastric mucosal PGE2 production36,37 and are commonly found in clinically healthy dogs. Prevalence rates of 100% have been reported in both random-source dogs38 and in laboratory beagles.39 Therefore, it is likely that all the dogs in this study were infected with Helicobacter spp. organisms. Given the crossover study design, the effects of Helicobacter spp. infection on eicosanoid mucosal concentrations would have been consistent in all 3 treatments for each dog.
In summary, in this study, firocoxib-treated dogs had larger pyloric mucosal lesions than dogs that received either tepoxalin or placebo despite having mucosal prostaglandin concentrations that were no different from placebo. This finding suggests that COX-2 inhibition led to slower mucosal wound healing by a mechanism independent of prostaglandin synthesis. In contrast, dogs treated with tepoxalin had very low mucosal prostaglandin synthesis capacity and mucosal LTB4 concentrations that were not decreased compared with placebo, but did not experience delayed mucosal wound healing. Therefore, in this canine model, tepoxalin administration favored wound healing compared with firocoxib despite the potential inhibition of prostaglandin-dependent mucosal healing mechanisms and no suppression of mucosal LTB4 production. Additional studies are needed to understand the mechanisms by which mucosal lesion healing is altered in the presence of compounds that inhibit prostaglandin synthesis.
aZubrin, Schering-Plough Corporation, Kenilworth, NJ
bPrevicox, Merial, Duluth, GA
cSigmaScan Pro 5, Aspire Software International, Ashburn, VA
dCayman Chemical Co, Ann Arbor, MI
eSigma Chemical Co, St Louis, MO
fSAS V 9.1, Cary, NC