• Open Access

Effects of Meloxicam and Phenylbutazone on Equine Gastric Mucosal Permeability


  • Work was completed on the Wagga Wagga campus of Charles Sturt University, NSW, Australia. The results of this study have not been presented or published previously

Corresponding author: S.L. Raidal, School of Animal and Veterinary Sciences, Charles Sturt University, Wagga Wagga, NSW 2658, Australia; e-mail: sraidal@csu.edu.au.



Newer NSAIDs that more selectively target the induced isoform of the cyclooxygenase enzyme (COX2) activity might reduce adverse effects while preserving therapeutic benefits of these drugs.


To compare the effect of oral administration of multiple dose rates of meloxicam and phenylbutazone (PBZ) on gastric mucosal integrity in horses.


Twenty-five light breed horses.


In vivo toxicity study. Horses were randomly assigned to 5 treatment groups, receiving placebo, PBZ (4.4 mg/kg PO q12h day 1, 2.2 mg/kg PO q12h for 4 days, 2.2 mg/kg PO q24h for 9 days), or 3 dose rates of meloxicam (0.6 mg/kg q24h, 1.8 mg/kg q24h, 3.0 mg/kg q24h) for 14 days. Sucrose permeability testing was performed on Day 0 (before treatment) and on Day 13. All personnel involved with data collection or analysis were blinded to treatment.


Administration of PBZ at the above dose rate significantly increased gastric permeability to sucrose, evidenced by increased peak serum sucrose concentrations (280–1,580 pg/μL, P = .001) after treatment. Similar changes were not evident after administration of meloxicam at any dose rate tested, or in control horses (P > .05). Treatment was not associated with significant differences in ulceration of the squamous or glandular mucosa. Peak sucrose concentrations were not correlated with serum total protein or albumin concentrations (R2 = −0.07, P = .61, R2 = −0.08, P = .58, respectively).

Conclusion and Clinical Importance

These results suggest that PBZ was associated with greater compromise to gastric mucosal integrity than meloxicam.




nonsteroidal anti-inflammatory drugs




standard deviation


standard error of mean


analysis of variance


Nonsteroidal anti-inflammatory drugs (NSAIDs) are a heterogeneous family of pharmacological compounds that work to inhibit the activity of cyclooxygenase (COX), a key proinflammatory enzyme activity. However, COX derivatives also mediate a number of homeostatic mechanisms which contribute especially to the maintenance of gastrointestinal tract integrity and renal blood flow. As the COX enzyme activity exists as different isoforms, considerable effort has gone into the development of anti-inflammatory agents that preferentially target COX2. Gastrointestinal ulceration associated with NSAID administration in horses and other species likely occurs owing to interference with mucosal protective mechanisms. In vitro and ex vivo studies have demonstrated that COX1 is preferentially expressed in the healthy gastric mucosa of horses and COX2 is variably expressed in healing ulcers in this region,[1] or induced after bradykinin stimulation of tissue explants.[2] Hence it has been hypothesized that agents such as meloxicam, which preferentially targets COX2 in horses,[3, 4] might spare COX1 and hence have less detrimental effect on gastric mucosal integrity than nonselective NSAIDs such as phenylbutazone (PBZ).

Sucrose permeability testing has been explored as an alternative diagnostic tool to gastroscopy in horses,[5, 6] and has been utilized to investigate effects of NSAID administration in other species.[7, 8] A liquid chromatography/mass spectrometry technique was developed and validated for the quantitation of sucrose in equine serum[9] to compare the effect of meloxicam and PBZ treatments on gastric mucosal permeability. Findings were compared with endoscopic gastric ulceration scores and plasma protein concentrations determined in the broader context of a wider study on the safety of PO meloxicam in horses.[10]

Materials and Methods


This study was conducted as part of a larger investigation aimed at evaluating the safety of an oral formulation of meloxicam for horses.[10] The current study focused solely on possible changes to gastric permeability caused by the oral administration of meloxicam or PBZ in a subset of study horses (horses M1 to M25).

Horses and Experimental Design

Twenty-five light-breed (Thoroughbred/Standardbred) horses (13 geldings and 12 mares) were randomly assigned to 5 treatment groups, each consisting of 5 horses. Mean (±SD) age (7.9 ± 4.5 years) and body weight (514 ± 62 kg) were not significantly different between groups (P = .515 and P = .934, respectively). Group A (control) received a placebo (product vehicle only) for 2 weeks, Group B received the recommended daily dose of meloxicam1 (0.6 mg/kg PO q24h), Group C received 3 times the recommended dose (meloxicam 1.8 mg/kg PO q 24h), Group D received 5 times the recommended dose (meloxicam 3.0 mg/kg PO q 24h), and Group E received PBZ2 at the manufacturer's recommended dose rate (4.4 mg/kg PO q12h day 1, 2.2 mg/kg PO q12h for 4 days, 2.2 mg/kg PO q24h for 9 days). No significant abnormalities were evident on physical examination or laboratory evaluation (routine hematology and serum biochemistry) of horses at the commencement of the study, and horses received no medication for 2 weeks before entering the trial. All procedures were approved by the Animal Care and Ethics Committee at Charles Sturt University (ACEC Approval Number 07/127).

Horse care and clinical investigations have been described previously.[10] Briefly, horses were stabled with 1 hour of daily exercise (either free turnout or on a horse walker) for the duration of the study. All were maintained on 1.8% body weight of lucerne hay divided into 2 meals, and were examined twice daily by personnel blinded to treatment. Fecal output, fecal consistency, and food and water intake were noted at these times for each horse. Blood was collected twice weekly for routine hematology and serum biochemistry by a commercial laboratory.3 Gastroscopy (described below), abdominal ultrasound, and urinalysis were performed weekly (Days 0, 6, and 13) to assess gastrointestinal integrity and renal function, as described.[10] Fecal samples were collected for detection of fecal occult blood with guiaic slides,4 as previously described.[11] Horses in treatment Groups A and B continued in the trial for a further 4 weeks as part of related studies.[10]

Gastric Endoscopy

Horses were fasted overnight for gastroscopy on Days 0, 6, and 13 by a 3-m endoscope5 and were sedated immediately before the procedure (xylazine 0.4 mg/kg and acetylpromazine 0.02 mg/kg IV). The severity of gastric mucosal ulceration was subjectively assessed by 2 independent assessors blinded to treatment, as previously described.[10] In addition, the presence and severity of lesions in the pyloric antrum were noted and graded subjectively by a single examiner blinded to treatment.

Permeability Test Procedure

Gastric permeability was assessed for each horse on Day 0 and Day 13, before and after 14 days of NSAID or placebo administration, by determination of sucrose absorption as described previously.[9] Horses were fasted overnight (a minimum of 12 hours) before gastroscopy and then received 0.5 g/kg body weight of sucrose as a 10% solution in tap water via nasogastric intubation. Blood samples were collected from the jugular vein at 0 (before), 15, 30, 45, 60, 90, 120, and 240 minutes post sucrose administration from an aseptically placed intravenous catheter. Samples were stored at 5°C and serum was separated within 3 hours of collection and stored at −20°C for sucrose determination.

Sample Preparation

After thawing, serum samples for each horse were diluted, and protein precipitation performed by the addition of 0.01 mL of equine serum to 0.9 mL of diluent consisting of 90% acetonitrile and 10% milli-Q water containing 5,000 pg/μL of internal standard (trichlormethiazide, TCM). Diluted samples were vortexed and then centrifuged (10,000 × g for 10 minutes). The supernatant was transferred to 2 mL autosampler vials for HPLC analysis.

Liquid Chromatography/Mass Spectrometry

Analysis was performed by an Agilent 1200 ultra high-performance liquid chromatography coupled with an Agilent 6410QQQ triple quad mass spectrometer6 with detection by selected ion monitoring with mass spectrometry, as described previously.[9] Assay validation, specificity, and sensitivity have been described.[9] The limit of detection was 50 pg/μL, although concentrations as low as 11 pg/μL were detected. Study samples were run in duplicate, firstly by treatment group and then in randomized order to exclude potential confounding of results caused by sample handling. For each horse, Day 0 and Day 13 samples from each time point (T0, T15, T30, T45, T60, T90, T120, and T240 minutes), with a duplicate of T60 minute sample, were run concurrently. To minimize the risk of column carryover and remove residual products, column integrity was maintained by performing a routine column flush after completion of Day 0 and Day 13 samples for each horse and by washing the column with methanol then acetonitrile after analysis of samples from 5 horses.

Statistical Methods

Results were compared within and between treatment groups utilizing parametric and nonparametric statistics as appropriate after checking for normality and equal variance (SigmaPlot 11.07). Significant interactions between time of collection and treatment for clinical pathology parameters were evaluated by two-way repeated measures ANOVA, with posthoc pairwise comparisons performed by Tukey test. Peak serum sucrose concentrations were compared within treatment groups (Day 0 versus Day 13) and between groups on Day 0 and Day 13 by paired t-test and one-way ANOVA, respectively. Serum sucrose concentrations were examined for sampling time and day of collection (treatment) effects by two-way repeated measures ANOVA or repeated measures ANOVA on ranks. Correlations between peak serum sucrose concentration and endoscopic score for squamous and glandular mucosa, and between serum total protein concentrations, were compared across all data by the Pearson product moment correlation. Gastroscopy squamous and glandular mucosa scores for each horse were compared before and at the completion of treatment by ordinal logistic regression, as previously described,[10] and differences in gastroscopic observations between groups were compared by χ2 analysis. For all evaluations, P < .05 was considered significant.


Clinical Findings

Detailed clinical findings have been reported previously.[10] Group A (control) and Group B horses (treated with 0.6 mg/kg meloxicam PO q24h) demonstrated no change in any of the clinical parameters assessed (physical examination, body weight, hematology, serum biochemistry, or urinalysis) and no changes were detected on gastrointestinal or renal ultrasound. Fecal occult blood tests were positive for 5 horses on a single occasion, but did not correlate with gastroscopic or ultrasound findings consistent with gastrointestinal ulceration. Horses in treatment Group C (meloxicam 1.8 mg/kg PO q24h) and Group D (meloxicam 3.0 mg/kg PO q24h) developed clinical findings and laboratory changes consistent with NSAID toxicity, including right dorsal colitis (two Group D horses), ventral edema, loose feces, inappetance, leucopenia, neutropenia, hypoproteinemia, and hypoalbuminemia. Only 1 horse in each group (of 5 horses) remained unaffected by treatment. Fecal occult blood was positive on at least 1 occasion for 4 Group C horses and 1 Group D horse, including samples from 2 horses on Day 1 (before treatment), but did not correlate with endoscopic or gastroscopic findings. Three of 5 Group E (PBZ) horses remained well for the duration of the study. One horse demonstrated preputial edema associated with decreased serum total protein and albumin concentrations, which resolved after completion of the study, and 1 horse developed right dorsal colitis. Results of fecal occult blood were positive for 3 Group E horses, but did not correlate with endoscopic or ultrasound evidence of gastrointestinal ulceration.

Consistent with our findings for the larger group of horses,[10] Group C (meloxicam 1.8 mg/kg), Group D (meloxicam 3.0 mg/kg), and Group E (PBZ) horses in this study demonstrated significant reductions in serum total protein (P < .001) and albumin (P < .001). Other serum biochemistry parameters did not change in a clinically significant way between or within groups, and there were no abnormal urinalysis findings for any horse.

Effect of NSAID Treatment on Gastric Sucrose Permeability

Peak serum sucrose concentrations on Day 0 ranged from 160 to 625 pg/μL for all horses (median 361 pg/μL), and were significantly (P = .021) less than values on Day 13 (range 190–1,580 pg/μL, median 671.25 pg/μL) when compared by one-way analysis of variance on ranks. Time to peak sucrose concentration (mean ± SD) was 61.2 ± 23.0 minutes on Day 0, and 51.2 ± 8.8 minutes on Day 13. Across all sampling times, serum sucrose concentrations were not normally distributed and had unequal variance. Transformation of data was unsuccessful in addressing these limitations, precluding two-way repeated measures ANOVA of the entire data set. Within each group, separate repeated measures ANOVA on ranks showed a significant increase in serum sucrose concentrations for Group E on Day 13 (P = .001), with significant increases evident at 45 and 60 minutes when compared with 0 minute samples (Fig 1). This effect was not observed in Day 0 samples or in other groups at any time. Peak serum sucrose concentrations were normally distributed on Day 0 and on Day 13. A significant difference was not evident between groups for results obtained on Day 0 or on Day 13, when compared by separate one-way ANOVA (P = .067 and P = .398, respectively). Within each treatment group, peak serum sucrose concentrations were not significantly different between Day 0 and Day 13, when compared by paired t-test (P = .05 for Group E; P = .18 for Group E; P = .14 for Group C; P = .77 for Group B; P = .19 for Group A).

Figure 1.

Serum sucrose concentrations for each treatment group before (Day 0) and after 14 days of NSAID treatment (Day 13). Results on Day 0 were normally distributed and there were no differences between or within groups; hence pooled results are displayed as 25th and 75th percentiles, with 10 and 90% (error bars), median and outliers indicated. Mean values from each group are also displayed (±SEM). Results from Day 13 were not normally distributed and were compared within each group by separate one-way ANOVA on ranks, with results shown for each group as box and whisker plots with median, mean (thick bar), and 10th and 90th percentiles indicated. Significant effects, evident in Group E (PBZ) only, are indicated (*, P < .05).

Gastric Endoscopy

Gastroscopic findings from study horses have been reported in full.[10] Comparison of the proportion of horses within each group that developed increased mucosal ulceration during the course of this study by χ2 analysis demonstrated no significant difference between groups (P = .657 for lesions of the squamous mucosa, P = .406 for the glandular mucosa). Logistic regression demonstrated no changes in squamous or glandular ulceration scores for any group. As is evident in Figure 2, peak serum sucrose did not correlate with endoscopic squamous mucosal ulceration score (r= 0.21, P = .31) or with endoscopic glandular ulceration score (r= −0.06, P = .79). Similarly, peak serum sucrose concentration (Fig 3) did not correlate with total serum protein (r= −0.07, P = .61) or serum albumin concentrations (r= −0.08, P = .58).

Figure 2.

Relationship between serum sucrose concentration and endoscopic evaluation of gastric squamous (A) and glandular mucosa (B). There was no correlation between serum sucrose concentration and squamous or glandular lesions (P = .79 and P = .31, respectively).

Figure 3.

Relationship between serum sucrose concentration and serum total protein and albumin concentrations. There was no correlation between serum sucrose concentration and total serum protein or albumin concentrations (= .61 and P = .58, respectively).


This study identified a significant increase in serum sucrose concentration (P = .001) in horses that received PBZ treatment for 14 days. This effect was not observed before treatment or in other treatment groups, suggesting that PBZ, a nonselective COX inhibitor, had a greater effect on mucosal integrity than meloxicam. In horses, as in other species, meloxicam more selectively targets COX2.[3, 4] Studies comparing selective inhibition of COX1 and COX2 with nonselective COX inhibition have demonstrated in rats that inhibition of both isoenzymes is required for the induction of gastric ulceration,[12] and studies in other species have suggested potential benefits of COX1 sparing NSAIDs, including meloxicam.[7, 8, 13]

PBZ toxicosis is well recognized in horses, and our results are consistent with previous reports of PBZ toxicosis in horses and ponies. In this study, gastric permeability to sucrose could not be correlated with serum protein or albumin concentrations, despite significant changes in these parameters for horses receiving PBZ and higher doses of meloxicam. Because there was no evidence of renal protein loss from any horse, our findings suggest that protein loss occurred through other parts the gastrointestinal tract, such as the right dorsal colon. Ultrasound and, in 1 case, postmortem examination confirmed right dorsal colitis in 3 horses in the current study population. Taken together, these observations may indicate that the right dorsal colon or other parts of the intestine are more susceptible to NSAID-induced damage than the stomach.

Our finding of increased gastric mucosal permeability associated with PBZ administration is consistent with similar studies that have utilized permeability testing to demonstrate subtle changes in gastric mucosal integrity in dogs[7, 14] and human patients[8, 15] associated with NSAID administration. The lack of correlation between endoscopic assessment of gastric squamous or glandular mucosal damage and gastric mucosal permeability, as determined by sucrose absorption in the present study, may suggest limitations of endoscopic evaluation of mucosal integrity. Postmortem study has demonstrated that gastroscopy is likely to underestimate both the presence of glandular ulceration and the severity of nonglandular lesions.[16] Martineau and colleagues[17] observed that endoscopic scoring systems were useful for the assessment of the extent of disease and monitoring response to treatment, but that little work has been performed to correlate gross observations with microscopic changes. In a postmortem study of 21 horses, these authors also noted that histological changes correlated poorly with gross changes.[18] If NSAID-induced damage begins with ultrastructural changes before gastric ulceration,[19] or with generalized mucosal damage rather than discrete ulceration,[14] sucrose permeability may provide a more sensitive and objective method of assessing gastric mucosal integrity than gastroscopy.

Other investigators[5, 6] have successfully correlated peak serum sucrose concentration with ulceration grade. Factors such as small group sizes, the inability to run a cross-over study and particularly the relatively mild ulceration observed in our study, may have reduced our ability to relate endoscopic findings to changes in permeability. Serum sucrose concentrations in our study were consistent with values reported by Hewetson et al[5] for lesions of similar gastroscopic appearance. Correlation between gastroscopy findings and sucrose permeability is reportedly more likely with ulcer scores of grade 2 or greater.[6, 7]

Across all groups, a significant increase in peak serum sucrose concentration was observed (Day 0 results compared with Day 13), suggesting that factors such as housing and diet may have caused subtle increases in gastric permeability. Gastric ulceration in horses is widely regarded as a multifactorial condition. Stabling and periods of fasting required for sample collection and gastroscopy in the present study may have had an impact on the gastric mucosa.[20] Although we cannot positively exclude gastrointestinal hemorrhage on occasions where positive results for fecal occult blood were determined with guiaic slides, this technique did not appear to identify horses with gastric ulceration, right dorsal colitis, or protein loss as results were negative for horses with clinically apparent lesions (such as grade 2 glandular or squamous mucosal lesions, thickening and ulceration of the right dorsal colon, hypoproteinemia). Apparent positive results in otherwise healthy horses may have been caused by unrecognized peroxidases in fecal samples, as described previously.[21] The sucrose permeability test in this study suggested a difference in gastric toxicity between PBZ and meloxicam which was not appreciated on gastroscopic evaluation. Our findings suggest that selective COX inhibitors may have a lesser effect on gastric mucosal integrity than nonselective agents, and that changes in mucosal permeability to sucrose may be more objective and sensitive than gastroscopic findings for the identification of such damage.


The authors gratefully acknowledge the technical expertise of Paul Weston in the analysis of serum samples. Equine and veterinary science staff and students (Dr Fiona Schneiders, Sarah Hanlon, Danielle Arthur, Naomi Bakker, Whitney Chapple, Amanda-Lee Charman, Freya Colvern, Greg Dale, Fiona Edwards, Alistair Grant, Simone Healey, Shahid Khaflan, Trystan Keylock, Tara Mills, Amelia Pascoe, Emily Roberts, Anneliese Seagar, Alecia Sheridan, and Sarah Ward) who assisted with care of horses, sample collection, or both. Dr Fiona Schneiders evaluated gastroscopy videos.

Conflict of Interest Declaration: This project was unfunded, but was completed as an extension of studies funded by Troy Laboratories, which evaluated the safety of an oral formulation of meloxicam.


  1. 1

    Meloxicam Oral Suspension (30 mg/mL), Troy Ilium Pty Ltd, Smithfield, NSW, Australia

  2. 2

    Oralject P-Butazone Paste (200 mg/mL); Virbac Australia, Milperra, NSW, Australia

  3. 3

    Idexx Laboratories, Rydalmere, NSW, Australia

  4. 4

    Hemoccult Sensa, Beckman Coulter Australia Pty Ltd, Gladesville, NSW, Australia

  5. 5

    Olympus CV160, Austvet Endoscopy, Melbourne, Australia

  6. 6

    Agilent Technologies, Forest Hill, VIC, Australia

  7. 7

    Systat Software Inc, San Jose, CA