Nonsteroidal anti-inflammatory drugs (NSAIDs) are among

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5: Celecoxib Plus Aspirin Versus Naproxen and Lansoprazole Plus Aspirin: A Randomized, Double-Blind, Endoscopic Trial JAY L. GOLDSTEIN,* BYRON CRYER, FOUAD AMER, and BARBARA HUNT *University of Illinois at Chicago, Chicago, Illinois; University of Texas Southwestern Medical School, Dallas, Texas; and TAP Pharmaceutical Products Inc, Lake Forest, Illinois Background & Aims: Patients requiring low-dose aspirin along with nonsteroidal anti-inflammatory drugs are at increased risk for gastrointestinal injury. This study compared the incidence of gastroduodenal ulcers in patients treated with low-dose aspirin and a cyclooxygenase-2 selective nonsteroidal anti-inflammatory drug or a nonselective nonsteroidal anti-inflammatory drug plus the proton pump inhibitor lansoprazole. Methods: Subjects 18 years or older with osteoarthritis, without gastroduodenal ulcer or erosive esophagitis at baseline endoscopy, and a cardiovascular indication for prophylaxis low-dose (81 or 325 mg) aspirin were prescribed open-label aspirin and blindly randomized to celecoxib 200 mg/day or naproxen 500 mg twice daily plus lansoprazole 30 mg once daily. Endoscopy was performed at 12 weeks or early termination. Results: One thousand forty-five subjects were randomized and received at least 1 dose of study medication, and 854 (n 426 celecoxib, n 428 naproxen plus lansoprazole) subjects with both baseline and final visit endoscopies were evaluable for the primary efficacy analysis. Among these subjects, the rate of endoscopically confirmed gastroduodenal ulcers was not different in the celecoxib (9.9%) and naproxen plus lansoprazole (8.9%; treatment difference [95% confidence interval], 1.0% [ 2.9% to 4.9%]) groups. Conclusions: In patients with osteoarthritis taking low-dose aspirin, the use of celecoxib or naproxen plus lansoprazole resulted in similar rates of gastroduodenal ulceration. Nonsteroidal anti-inflammatory drugs (NSAIDs) are among the most frequently prescribed medications, with more than 100 million prescriptions written in 2003 in the United States. 1 However, the use of NSAIDs increases the risk of gastroduodenal ulcers (GDUs) and ulcer complications by 2-fold to 5-fold. 2 5 An estimated 62 million Americans have cardiovascular disease, and recent studies indicate that nearly 75% or more of those who have a history of myocardial infarction and 25% of the general population older than 50 years take aspirin at doses less than or equal to 325 mg per day. 6 9 Like NSAIDs, aspirin use increases the risk for upper gastrointestinal ulcer complications by between 1.5-fold and 4-fold The cyclooxygenase-2 (COX-2) selective NSAIDs were designed to provide pain relief comparable to that of nonselective NSAIDs, with a reduced risk for gastrointestinal events and complications. Numerous studies have shown significantly lower rates of endoscopic GDUs and ulcer complications with COX-2 selective NSAIDs compared with traditional NSAIDs Not surprisingly, a large proportion of patients take both aspirin and either a COX-2 selective or nonselective agent. Evidence indicates that gastrointestinal injury and complications with aspirin plus a nonselective NSAID exceed those of either agent alone. 15,21 A 2001 prescription claims that a database survey indicated that nearly half of long-term COX-2 users take concomitant aspirin for cardioprotection. 22 However, the gastrointestinal benefit of COX-2 selective NSAIDs in this population is questionable. In a 12-week endoscopic study, Laine et al 21 reported that the ulcer incidence was significantly higher in those with 81 mg of aspirin plus rofecoxib 25 mg once daily (16%) than with aspirin alone (7%) but was not significantly different than the incidence in those treated with aspirin plus ibuprofen 800 mg 3 times daily (17%). In contrast, in a 1-week endoscopic placebo-controlled study comparing celecoxib 200 mg once daily and naproxen 500 mg twice daily in subjects taking 325 mg of aspirin daily, Goldstein et al 15 noted that aspirin use alone was associated with a significantly lower endoscopic ulcer rate compared with aspirin plus celecoxib use (8% vs 19% ulcer incidence, respectively). However, the celecoxib plus aspirin ulcer incidence was still significantly lower than the 27% ulcer incidence in those treated with naproxen plus aspirin. Proton pump inhibitors (PPIs) have been evaluated for the prevention of NSAID-associated gastrointestinal injury. For example, in prospective endoscopic trials, PPIs have been shown to be significantly better than placebo 23 or H 2 -receptor antagonists 24,25 and not statistically different from high-dose misoprostol 5 in reducing the incidence of ulcers even in high-risk patients (eg, history of ulcer) taking NSAIDs. Furthermore, Chan et al 26 compared the rate of rebleeding in subjects taking omeprazole plus diclofenac or celecoxib alone. No significant difference in the 6-month probability of recurrent bleeding was observed between the 2 treatment groups (6.4% vs 4.9%, respectively). PPI co-therapy has also been shown to be effective in reducing the ulcer rebleeding rate in aspirin use alone 27 and in endoscopic trials in NSAID users. 12 Many guidelines recommend 1 of 2 strategies for those patients receiving aspirin and who require NSAID therapy; these strategies are use of a COX-2 selective NSAID or a nonselective NSAID plus a PPI The current study is the first prospective randomized head-to-head comparison designed to compare these 2 strategies in reducing endoscopic ulcer rates in subjects with osteoarthritis receiving low-dose (81 or 325 mg/ day) aspirin therapy. We hypothesized that at the end of a Abbreviations used in this paper: CI, confidence interval; COX-2, cyclooxygenase 2; GDU, gastroduodenal ulcer; PPI, proton pump inhibitor by the AGA Institute /07/$32.00 doi: /j.cgh

2 1168 GOLDSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No week study in an aspirin-using population, endoscopic ulcer rates would be lower in subjects treated with a PPI and a nonselective NSAID when compared with those treated with a COX-2 selective NSAID. Methods Study Population This study took place between July 22, 2003 and July 28, Subjects 18 years or older with an osteoarthritis diagnosis, an expectation of requiring ongoing NSAID use for at least 3 months, and a clinical indication for daily low-dose aspirin use for cardiovascular prophylaxis were eligible to participate in this multicenter, double-blind, parallel, active-drug control, randomized study. The institutional review board at the 81 study sites approved the protocol, and all subjects voluntarily signed an informed consent/health Insurance Portability and Accountability Act authorization form at the screening visit before any study-related procedures. Study Interventions Study eligibility was determined in a pre-study screening visit that occurred 2 days to 2 weeks before the baseline visit. During the screening visit, subjects underwent a complete medical and social history and physical examination. At the screening visit, subjects were queried as to their use of either a nonselective or a COX-2 selective NSAID during the prior 90 days. During the 90 days prior, subjects could take up to one NSAID daily at a therapeutic dosage. Additional NSAID(s) could be taken but could not exceed 12 doses per month; of those 12 doses per month, a subject could not exceed 7 doses per week. Subjects with the following conditions were also excluded from study participation: those who required anticoagulant agent or corticosteroid use; had a history of gastric or duodenal surgery other than a simple oversew of an ulcer during the past 1 year; had a known history of GDU, gastrointestinal bleed within the prior year, or coexisting esophageal disease; had a serum creatinine exceeding 2.0 mg/dl; had a history of cancer within the prior 5 years (except basal cell carcinoma); had an anticipated need for surgery or other invasive procedures during the study period; had clinically abnormal laboratory, biochemical, or hematologic parameters; and (if female) were pregnant or lactating or not currently using an acceptable form of birth control. Subjects meeting screening eligibility requirements underwent a baseline endoscopy. Those found to have esophageal erosions or ulcers with a mucosal break, an endoscopy-confirmed gastric or duodenal ulcer, with an ulcer defined as a discrete lesion with appreciable depth and 3 or more mm in diameter, or 10 or more gastroduodenal erosions (defined as superficial break in the mucosa 3 mm with or without exudates) were excluded. To be randomized in this trial, eligible study subjects were required to test negative for Helicobacter pylori. In the initial protocol, subjects were required to test negative on the basis of the CLO test (Delta West, Western Australia) performed at the baseline endoscopy. If negative, subjects were randomized, and no further H pylori testing was performed. After approximately 30% (318/1045) of subjects were randomized in the trial, the protocol was amended so that H pylori serology was obtained at the screening visit before the baseline endoscopy. This amendment was initiated to reduce the number of unnecessary baseline endoscopies in potentially H pylori positive subjects. If seronegative at the screening visit, subjects proceeded to the baseline endoscopy, and no further H pylori testing was performed. If seropositive, subjects could be treated with eradication therapy, as determined by the investigator. After a 2-week minimum after eradication regimen completion, subjects could reconsent and undergo a baseline endoscopy. For this subgroup, confirmatory H pylori status was established at study end by using a CLO test at the final visit endoscopy. Although H pylori positivity was an exclusion criterion, subjects who ultimately tested positive by CLO at final endoscopy were not excluded from the efficacy analyses on the basis of this criterion. Treatment Randomization Subjects meeting study eligibility requirements were randomly assigned in a 1:1 ratio with a block size of 4 to receive either celecoxib 200 mg once daily (celecoxib) or naproxen 500 mg twice daily plus lansoprazole 30 mg once daily (naproxen plus lansoprazole) according to the concealed randomization schedule computer-generated by TAP before the study start. All subjects received open-label aspirin 81 or 325 mg once daily. The first dose of study drug was administered within hours of completing the screening endoscopy. Identical-appearing blister packs with the active medication or matching dummy overencapsulated in a gray capsule to maintain blinding were dispensed along with open-label, oncedaily aspirin 81 or 325 mg (dose based on investigator discretion). Subjects were instructed to self-administer blinded study medication with full glasses of water in the morning before breakfast (celecoxib, naproxen-dummy, and lansoprazoledummy in the celecoxib arm or celecoxib-dummy, naproxen, and lansoprazole in the naproxen plus lansoprazole arm) and in the evening before or after dinner (naproxen-dummy in celecoxib arm or naproxen in the naproxen plus lansoprazole arm). Low-dose aspirin was to be taken every morning with the blinded study drug. Open-label antacid tablets (Gelusil; Warner-Lambert, Morris Plains, NJ) were also dispensed, and subjects self-administered these as needed for minor upper gastrointestinal symptom relief, not to exceed 12 tablets during a 24-hour period. Efficacy Variables The predefined primary efficacy variable was the percentage of subjects in each treatment group with GDUs by endoscopy at the final visit (week 12 or early discontinuation). The predefined secondary efficacy variables were joint pain severity and the proportions of subjects with gastrointestinal complications. Treatment Assessments Subjects returned to the study site at weeks 4, 8, and 12 (or final visit) for treatment efficacy and safety assessments. At each visit, vital signs and study drug and aspirin compliance (by pill counts) were assessed. With a nonvalidated measure of joint pain, subjects were instructed to pick a whole number corresponding to their pain intensity by using individual boxes labeled Subjects were given the following subjective guide: 1 3 corresponds to mild, 4 7 to moderate, and 8 10 to severe pain intensity. Subjects underwent a brief physical examination (weeks 4 and 8 only) and laboratory evaluations (hematology,

3 October 2007 NSAID/PPI PLUS ASPIRIN VS COXIB PLUS ASPIRIN 1169 serum chemistry, urinalysis). Treatment-emergent adverse events were monitored, including all adverse events that occurred during the study, irrespective of whether the events were thought to be related to the study drug. Any adverse events including dyspepsia that were reported or observed by physical examination or laboratory assessments were rated for severity (mild, moderate, or severe) and relationship (definite, probable, possible, unlikely, not related) to study drug. Endoscopy was performed at week 12 or earlier at investigator discretion if the subject was thought to be having a gastrointestinal complication or was terminating the study prematurely. Assessments of Efficacy The primary efficacy parameter (GDU incidence at the final visit) was assessed by endoscopy performed at the final visit. The secondary efficacy variable of gastrointestinal complications was assessed at each visit by monitoring for signs and symptoms indicative of a gastrointestinal complication. Subjects were queried for the presence of severe abdominal pain, melena, hematemesis, and signs and symptoms of gastrointestinal obstruction (defined as a significant change in bowel habits or severe abdominal distention). Suspected gastrointestinal complications were categorized as bleed, perforation, obstruction, or non-event according to pre-established definitions. Diagnoses of suspected gastrointestinal obstructions were to be confirmed by endoscopy or radiographic results. Subjects experiencing gastrointestinal complications were discontinued from the study. Statistical Analysis The sample size was determined by assuming that 11% of subjects receiving celecoxib and aspirin and 5% of subjects receiving naproxen plus lansoprazole and aspirin would experience GDUs. 5,12,24,25 Assuming that the dropout rate would not exceed 20%, complete data would be available for at least 384 subjects in each treatment group, providing at least 80% power at the 0.05 significance level to detect a difference in the GDU rate at the final visit. Treatment group homogeneity at baseline was analyzed with one-way analysis of variance for continuous variables including age, height, and weight and Fisher exact test for categorical variables including gender and race. Efficacy was assessed among evaluable subjects, defined as those who received at least 1 dose of study drug, were free of GDUs, had fewer than 10 erosions at the screening endoscopy performed within 3 days of study drug initiation, and completed a post-baseline endoscopy not more than 5 days after the last dose of study drug. Fisher exact test was used to compare the proportions of subjects with GDUs and with each of the 3 predefined gastrointestinal complications (bleed, perforation, gastric outlet obstruction). Cochran-Mantel-Haenszel methodology was used to compare the proportions of subjects with GDUs between treatment groups stratified by various baseline characteristic levels. Post hoc analyses with Fisher exact test were performed, evaluating response within aspirin dose subgroups, within subgroups previously treated and untreated with NSAIDs for 90 days, and among all randomized subjects (intent-to-treat). In the intent-to-treat analysis, subjects without a final endoscopy were assumed to have a GDU. In post hoc analyses, one-way analysis of variance was used to compare the mean percent change in joint pain severity from baseline to weeks 4, 8, and 12 between treatment groups. Adverse events were coded by using the Medical Dictionary for Regulatory Activities. Comparisons between treatments of the proportion of subjects reporting any adverse event and of proportions with specific reported adverse events were made with Fisher exact test. Results Subject Disposition and Characteristics A total of 1045 subjects were randomized into the 12-week study and received at least 1 study drug dose (516 celecoxib and 529 naproxen plus lansoprazole). A range of 1 93 subjects per site were enrolled at the 81 U.S. investigative sites (Appendix). Of those enrolled, 163 (79 celecoxib and 84 naproxen plus lansoprazole) subjects discontinued treatment prematurely, with adverse events as the most common reason (n 68). Figure 1 illustrates the subject disposition and reasons for discontinuation. The 2 treatment groups were similar with respect to baseline characteristics, including demographics, dose of aspirin use, number of erosions at baseline endoscopy, negative H pylori status (baseline or confirmed at final visit), osteoarthritis duration, baseline joint pain severity score, and NSAID use (previous 90 days) (Table 1). Of 516 subjects randomized to celecoxib, 33 were H pylori positive at screening. Of the 33 subjects, 6 were CLO-positive and as such were inadvertently randomized to the trial (2 were included in the primary analysis, and 4 were excluded because no final endoscopy was performed). After the amendment, 27 subjects were seropositive at the time of randomization, and 6 were subsequently excluded because they did not have either baseline or final endoscopy consistent with the protocol. The remaining 21 seropositive subjects were included in the primary analysis, of which 2 were CLO-positive at the final endoscopy. Of 529 subjects randomized to naproxen plus lansoprazole, 23 were H pylori positive at screening. Of the 23 subjects, 7 were CLO-positive and as such were inadvertently randomized to the trial (2 were included in the primary analysis, and 5 were excluded because no final endoscopy was performed). After the amendment, 16 subjects were seropositive at the time of randomization, and 4 were subsequently excluded because they did not have either baseline or final endoscopy consistent with the protocol. The remaining 12 seropositive subjects were included in the primary analysis, of which 1 was CLO-positive at the final endoscopy. Eighty-nine percent of celecoxib (457/516) and 89% of naproxen plus lansoprazole (468/529) subjects were prescribed 81 mg aspirin per day. More than 93% of subjects in each treatment group (94.2% [486/516] celecoxib and 93.4% [494/ 529] naproxen plus lansoprazole) took at least 90% of blinded study drug, as determined by pill count, and only 3.5% and 3.2% of celecoxib and naproxen plus lansoprazole subjects, respectively, were less than 80% compliant with blinded study drug. Similar compliance rates were reported with daily aspirin therapy; 90.9% (469/516) and 87.9% (465/529) of celecoxib and naproxen plus lansoprazole subjects, respectively, reported taking at least 90% of daily aspirin doses.

4 1170 GOLDSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 10 Figure 1. Subject disposition. Efficacy Population Figure 1 also illustrates the primary analysis subset. Among the 1045 randomized subjects, 854 (426 celecoxib and 428 naproxen plus lansoprazole) were evaluable for the primary efficacy analysis. Subjects were excluded from the primary efficacy analysis (14 in the celecoxib group and 17 in the naproxen plus lansoprazole group) if they did not have their first endoscopy consistent with protocol ( 3 days from study drug initiation); these subjects received drug, continued in the study, and were included in the safety analysis. In addition, subjects who did not complete a final visit endoscopy within 5 days per protocol (76 in the celecoxib group and 84 in the naproxen plus lansoprazole group) were also excluded from the primary efficacy analysis. Gastroduodenal Ulceration Rate at Week 12 or Final Visit On the basis of the primary efficacy analysis, the GDU incidence at the final visit was not different between those treated with celecoxib (9.9%, 42/426) and those treated with naproxen plus lansoprazole (8.9%, 38/428; treatment difference [95% confidence interval (CI)], 1.0% [ 2.9% to 4.9%]; P.64) (Figure 2). No significant differences between the treatment groups were observed in the GDU incidence when controlled for age, gender, race, height, weight, tobacco use, alcohol or caffeine use, or number of baseline erosions. Among older subjects ( 65 years), the GDU incidences were 10.1% (11/109) in the celecoxib and 14.3% (16/112) in the naproxen plus lansoprazole groups ( 4.2% [ 12.8% to 4.4%]). In the intentto-treat population (Figure 1), in which subjects who did not have a final endoscopy were assumed to have a GDU, the GDU incidence remained not significantly different between treatment groups (20.3% [105/516] for the celecoxib group, 18.0% [95/529] for the naproxen plus lansoprazole group; 2.4% [ 2.4% to 7.2%]). Post hoc comparisons in the primary efficacy population of the separate rates of duodenal, gastric, and both a gastric and duodenal ulcer(s) also showed no statistically significant treatment differences at the final visit. Two subjects in the celecoxib group and one in the naproxen plus lansoprazole group had both gastric and duodenal ulcers. Excluding those with duodenal ulcers, 31 subjects in each treatment group had gastric ulcers (7.3% vs 7.2%), which was not statistically significant (0.0% [95% CI, 3.4% to 3.5%]; P.99). Excluding those with gastric ulcers, duodenal ulcers were present in 9 (2.1%) and 6 (1.4%) subjects, respectively (0.7% [ 1.1% to 2.5%]; P.45). In a post hoc analysis, we stratified our data on the basis of the subject s self-reported history of NSAID use (either nonselective or COX-2 selective within 90 days before baseline) versus non-use. Among those who reported using either a nonselective or COX-2 selective NSAID, no difference in the GDU rate was seen between celecoxib-treated and naproxen plus lansoprazole treated subjects (7.5% [23/305] vs 10.6% [34/320], respectively; treatment difference [95% CI], 3.1% [ 7.6% to 1.4%]). In the NSAID-naïve population (n 229), the GDU rate in the celecoxib group was 15.7% (19/121) as compared with 3.7% (4/108) in the naproxen plus lansoprazole group (12.0% [4.6% 19.4%]; P.003). In a predefined analysis with the efficacy population stratified by aspirin dose, no statistically significant difference was shown in GDU rates between the celecoxib and naproxen plus lansoprazole arms (P.65). In addition, no statistically significant difference was observed in post hoc analyses when comparing the celecoxib with the naproxen plus lansoprazole arms in the aspirin subgroups (81 mg aspirin: 9.6% [36/376] and 7.8% [30/383], 325 mg aspirin: 12% [6/50] and 17.8% [8/45], respec-

5 October 2007 NSAID/PPI PLUS ASPIRIN VS COXIB PLUS ASPIRIN 1171 Table 1. Baseline Subject Characteristics Characteristic Celecoxib (N 516) tively). In an additional post hoc analysis, subjects in the naproxen plus lansoprazole arm had a statistically significantly lower rate of GDUs in the 81-mg (7.8% [30/383]) compared with the 325-mg (17.8% [8/45]) aspirin group (treatment difference [95% CI], 9.9% [ 21.4% to 1.5%]; P.046); no statistically significant difference was seen between aspirin doses in the celecoxib group (9.6% [36/376] for 81 mg and 12.0% [6/50] for 325 mg; 2.4% [ 11.9% to 7.1%]; P.61). Gastrointestinal Complications Lansoprazole/ naproxen (N 529) Male (%) Age, mean (y) (SD) 56.7 (10.84) 56.7 (11.49) Body mass index, mean (SD) 30.9 (8.07) 31.1 (7.14) Race (%) White Black Hispanic Asian Other Low-dose aspirin use (%) 81 mg mg No. of erosions at baseline (%) Missing Negative H pylori status confirmed at baseline or final visit a (%) Duration of osteoarthritis, (108.21) (97.92) mean (mo) (SD) Baseline joint pain severity 5.8 (2.17) 5.8 (2.00) score, mean (SD) No prior NSAID use for days (%) Substance use (%) Alcohol Caffeine Tobacco SD, standard deviation. a Otherwise, status positive or unknown. One subject experienced a bleeding gastrointestinal complication. This 57-year-old woman, randomized to naproxen plus lansoprazole, presented with bloody diarrhea after day 34 of treatment. On the basis of an endoscopy and colonoscopy performed on day 36, she was noted to have moderate hemorrhagic colitis. After management with clear liquids and antibiotics, the symptoms resolved. Joint Pain Assessment As seen in Table 1, baseline joint pain assessment was not different between the treatment groups. In post hoc analyses, no statistically significant differences in mean percent change from baseline in joint pain severity were found between the 2 study arms at any treatment visit. By week 12, the degree of response was not different between the treatment groups (week 12 visit mean [standard deviation] scores, % change from baseline, respectively: 4.0 [2.3], 18.2% vs 3.7 [2.3], 25%; treatment difference [95% CI]: 6.8% [ 3.1% to 16.7%]). Adverse Events The percentage of subjects reporting any adverse event was similar between the celecoxib (53%, 272/516) and naproxen plus lansoprazole (57%, 302/529) treatment groups. The most frequently reported ( 5% incidence) treatment-emergent adverse events for celecoxib-treated and naproxen plus lansoprazole treated subjects, respectively, were upper respiratory tract infections in 9% (46/516) and 11% (58/529); dyspeptic symptoms (ie, dyspepsia, epigastric discomfort, eructation) in 10% (54/516) and 7% (35/529; P.027); diarrhea in 4% (20/516) and 7% (38/529; P.022); abdominal pain in 6% (31/516) and 6% (32/529); and nausea and vomiting in 6% (33/516) and 6% (33/529). One cardiac event (mild palpitations considered by the investigator as probably related to study drug) was seen in 1 subject randomized to naproxen plus lansoprazole. Six subjects in the celecoxib group reported serious adverse events; 4 of these subjects were prematurely discontinued. Four subjects in the naproxen plus lansoprazole group, all of whom were prematurely discontinued, reported serious adverse events. No subject died during the study. Discussion The medical need to reduce the risk of NSAID-associated gastric ulcers and ulcer complications is confounded by the fact that many patients have also been advised to take low-dose aspirin as cardiovascular prophylaxis. 33 Clinical trials indicate that the use of COX-2 or nonselective NSAIDs with co-prescribed aspirin increases the risk of developing endoscopic GDUs 14,21,34,35 or GDU complications as compared with the non aspirin-using population. 14,20 Notably, in the TARGET 20 and CLASS 14 trials, aspirin use resulted in ulcer complication rates that were not different as compared with their nonselective NSAID comparators. Although these results suggest that add- Figure 2. Incidence of gastroduodenal ulcers at final visit.

6 1172 GOLDSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 10 ing aspirin to a coxib negates the potential gastrointestinal benefit, endoscopic trials do not uniformly support this conclusion. For example, in 2 separate prospective randomized 1-week endoscopic trials, the ulcer rates in aspirin users taking 325-mg or 81-mg daily doses and celecoxib were significantly lower than rates seen in naproxen plus aspirin users. 15,36 To address the important clinical question of aspirin use in patients requiring anti-inflammatory therapies, this large, multicenter, double-blind, prospective study was conducted to compare 2 common strategies used for minimizing the risk in subjects who received daily low-dose aspirin therapy and also required chronic anti-inflammatory therapy. Although we hypothesized, powered for, and anticipated that the GDU rate in the naproxen plus lansoprazole arm would be significantly lower than in the celecoxib arm, the results of this trial did not confirm this hypothesis, because there was no statistical difference between the 2 arms (Figure 2). As such, the primary end point was not met. As detailed in the statistical analysis section, we anticipated the 11% GDU rate in the coxib arm on the basis of previous studies of celecoxib and valdecoxib. 37,38 In fact, the 9.9% GDU rate observed in this trial was not dissimilar from what we expected. On the other hand, the 8.9% GDU rate observed in those randomized to naproxen plus lansoprazole was higher than the 5% anticipated rate on the basis of prior studies with a PPI combination. 24,25,39,40 Interestingly, the results of this current, larger prospective study conflict with the 0% ulcer rate reported at 12 weeks in a post hoc analysis 5 of a larger trial 12 of subjects receiving less than or equal to aspirin 325 mg plus lansoprazole 30 mg plus nonselective NSAIDs. The higher than anticipated rate of ulcers in the naproxen plus lansoprazole arm in the current trial cannot be explained by differences in baseline demographics, because our population was of similar age, H pylori status, and ulcer risk as compared with populations in other studies. 12,14,15,18,29,32,35,37,38 Therefore, on the basis of the lack of significant differences in our study design and population, it remains unclear why this study did not meet its anticipated end point; this finding should be reproduced in additional endoscopic trials because of the importance of this clinical scenario. An interesting but unexpected finding in a post hoc analysis of subjects with prior NSAID or COX-2 selective NSAID use within the past 90 days was that there was no significant difference between the 2 treatment groups. In contrast, in NSAID-naïve patients, the difference between the 2 arms was statistically significant, with lower ulcer rates found in the NSAID plus PPI arm. The reason for this finding is unclear, and any hypothesis would be speculative. However, independent of the mechanism, these results are intriguing and provide some evidence suggesting a potential protective effect of prior NSAID exposure and could potentially guide the choice of a safer strategy on the basis of whether the patient was recently using NSAIDs. Prospective studies addressing NSAID-naïve versus non-naïve patients are needed to better define this finding and potential clinical utility. A major limitation in this trial is that the end point is based on endoscopic findings. The use of endoscopic findings as surrogate markers for ulcer complications in non aspirin-using populations has been questioned, and the validity of using this marker is further questioned in the aspirin-using population. In part, this might be related to aspirin s additional antiplatelet effects and impact on bleeding ulcer complications. As such, the generalizability of our endoscopic findings to upper gastrointestinal ulcer complications needs to be confirmed. An additional limitation is that we did not directly measure the incremental and possibly differential impact of aspirin use in the 2 arms. An important clinical question that was not addressed is whether aspirin negates the benefit of a COX-2 selective NSAID because no patients were exposed to celecoxib alone. Although our trial results were based on an endoscopic end point, this question has been most recently addressed by the MEDAL trial, a randomized comparison of the COX-2 inhibitor, etoricoxib, with the traditional NSAID, diclofenac, in which it was observed that while taking etoricoxib, low-dose aspirin users had a lower incidence of noncomplicated, but not complicated, ulcers when compared with diclofenac users taking low-dose aspirin. 41 The results of the MEDAL trial, in part, reproduce the results of the CLASS and TARGET trials in which aspirin use resulted in ulcer complication rates that did not differ between the COX-2 selective NSAID arm to its nonselective NSAID comparator. 14,20 Another limitation is that we did not compare endoscopic damage related to NSAID plus aspirin use with or without a PPI. Data from the MEDAL trial suggest that PPIs benefit nonselective NSAID or COX-2 selective NSAID plus aspirin users in reducing the rate of complicated, but not uncomplicated, events. Conclusions In subjects with osteoarthritis taking low-dose aspirin, the use of celecoxib or naproxen plus lansoprazole resulted in rates of gastroduodenal ulceration that were not significantly different. Because of the importance of the clinical scenario of co-administration of anti-inflammatory agents plus aspirin, the endoscopic findings need to be reproduced and correlated to clinical outcomes before our results are used in daily clinical practice. Appendix Principal Investigators The following is a list of the principal investigators: R. E. Benkert, Wheat Ridge, CO; M. Beshay, Mission Hills, CA; S. A. Bookbinder, Ocala, FL; V. L. Bralow, Philadelphia, PA; R. B. Branum, Ft Smith, AR; A. L. Brodsky, Dallas, TX; D. Campbell, Kansas City, MO; C. M. Chappel, Kissimmee, FL; J. J. Cohen, Hallandale, FL; J. Crout, Austin, TX; J. E. Dodd, Jr, Jackson, MS; J. D. Durden, Tallasee, AL; M. Ellerbusch, Northport, AL; M. Ellman, Chicago, IL; R. D. Emkey, Wyomissing, PA; P. A. Eweje, Jacksonville, NC; M. J. Fairfax, Phoenix, AZ; I. F. Fenton, Vernon Hills, IL; J. J. Fiechtner, Lansing MI; C. L. Fisher, Jr, Newport News, VA; D. Fitz-Patrick, Honolulu, HI; F. C. Fowler, Charlotte, NC; R. E. Gaona, Sr, San Antonio, TX; W. P. Gilbert, Johnson City, TN; J. S. Gimbel, Phoenix, AZ; J. L. Goldstein, Chicago, IL; S. F. Gordon, Atlanta, GA; D. Graham, Houston, TX; W. L. Gray, Spokane, WA; B. Hill, Chaska, MN; J. A. Hoekstra, Richmond, VA; P. A. Holt, Baltimore, MD; E. W. Hood, Atlanta, GA; J. P. Huff, San Antonio, TX; V. S. Jayanty, Houston, TX; S. P. Kafka, Duncansville, PA; R. Kalb, Perrysburg, OH; N. M. Kassman, Statesville, NC; Y. Khronusova, Las Vegas, NV; M. Kohen, Port Orange, FL; R. A. Krause, Chattanooga, TN; R. J. Lapidus, Wheat Ridge, CO; W. E. Larson, Lakewood, WA;

7 October 2007 NSAID/PPI PLUS ASPIRIN VS COXIB PLUS ASPIRIN 1173 T. W. Littlejohn III, Winston-Salem, NC; B. D. Long, Cleveland, OH; A. Marcadis, Boynton Beach, FL; H. W. Marker, Memphis, TN; P. D. Matz, Medford, OR; D. C. McCluskey, Mogadore, OH; J. I. McMillen, Mechanicsburg, PA; J. R. Medoff, Greensboro, NC; R. V. Movva, Moline, IL; M. J. Noss, Cincinnati, OH; D. J. Pambianco, Charlottesville, VA; E. B. Portnoy, Westlake Village, CA; R. W. Powel, Newark DE; R. E. Pruitt, Nashville, TN; A. J. Ramsay, Pembroke Pines, FL; A. N. Reddy, Jacksonville, FL; D. L. Rider, San Francisco, CA; D. S. Riff, Anaheim, CA; S. H. Roth, Phoenix, AZ; C. K. Saadeh, Amarillo, TX; D. R. Schumacher, Columbus, OH; F. E. Scott, Santa Maria, CA; N. R. Shah, Hollywood, MD; P. Sharma, Kansas City, MO; V. K. Sharma, Scottsdale, AZ; E. A. Sheldon, Miami, FL; W. J. Shergy, Huntsville, AL; M. E. Shirley, Omaha, NE; M. Sisay, Evansville, IN; D. E. Stillwagon, The Woodlands, TX; N. Stollman, Oakland, CA; J. Tesser, Phoenix, AZ; M. L. Throne, Atlanta, GA; M. S. Touger, Birmingham, AL; M. W. Warren, Lancaster, PA; H. T. Williams, Birmingham, AL; L. Willis, Oklahoma City, OK; D. B. Wright, Cordova, TN; S. Zakko, Bristol, CT. References 1. Ansani NT, Fedutes BA, Vogt M, et al. Rheumatologic medication utilization in 2003: prescribing trends and retail sales costs. Poster presented at Annual Scientific Meeting of the American College of Rheumatology/Association of Rheumatology Health Professionals; October 18, 2004; San Antonio, TX. A correction was published at fedutes_nsaid.asp. 2. Laine L. Approaches to nonsteroidal anti-inflammatory drug use in the high-risk patient. Gastroenterology 2001;120: Bjorkman DJ. Current status of nonsteroidal anti-inflammatory drug (NSAID) use in the United States: risk factors and frequency of complications. Am J Med 1999;107(Suppl 6A):S3 S8. 4. Ofman JJ, MacLean CH, Straus WL, et al. A metaanalysis of severe upper gastrointestinal complications of nonsteroidal antiinflammatory drugs. J Rheumatol 2002;29: Graham DY, Agrawal NM, Campbell DR, et al. Ulcer prevention in long-term users of nonsteroidal anti-inflammatory drugs: results of a double-blind, randomized, multicenter, active- and placebocontrolled study of misoprostol vs lansoprazole. Arch Intern Med 2002;162: Antithrombotic Trialists Collaboration. Collaborative meta-analysis of randomized trials of antiplatelet therapy for prevention of death, myocardial infarction, and stroke in high risk patients [published correction appears in BMJ 2002;324:141]. BMJ 2002; 324: Rodondi N, Vittinghoff E, Cornuz J, et al. Aspirin use for the primary prevention of coronary heart disease in older adults. Am J Med 2005;118: Paulose-Ram R, Hirsch R, Dillon C, et al. Frequent monthly use of selected non-prescription and prescription non-narcotic analgesics among US adults. Pharmacoepidemiol Drug Saf 2005;14: Brown JB, Delea TE, Nichols GA, et al. Use of oral antithrombotic agents among health maintenance organization members with atherosclerotic cardiovascular disease. Arch Intern Med 2002; 162: Weil J, Colin-Jones D, Langman M, et al. Prophylactic aspirin and risk of peptic ulcer bleeding. BMJ 1995;310: Garcia Rodriguez LA, Hernandez-Diaz S, de Abajo FJ. Association between aspirin and upper gastrointestinal complications: systematic review of epidemiologic studies. Br J Clin Pharmacol 2001;52: Goldstein JL, Huang B, Amer F, et al. Ulcer recurrence in high-risk patients receiving nonsteroidal anti-inflammatory drugs plus lowdose aspirin: results of a post hoc subanalysis. Clin Ther 2004; 26: Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis: VIGOR Study Group. N Engl J Med 2000;343: Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS study: a randomized controlled trial Celecoxib Long-term Arthritis Safety Study. JAMA 2000;284: Goldstein JL, Lowry SC, Lanza FL, et al. The impact of low-dose aspirin on endoscopic gastric and duodenal ulcer rates in users of a non-selective non-steroidal anti-inflammatory drug or a cyclooxygenase-2-selective inhibitor. Aliment Pharmacol Ther 2006; 23: Hawkey C, Laine L, Simon T, et al. Comparison of the effect of rofecoxib (a cyclooxygenase 2 inhibitor), ibuprofen, and placebo on the gastroduodenal mucosa of patients with osteoarthritis: a randomized, double-blind, placebo-controlled trial the Rofecoxib Osteoarthritis Endoscopy Multinational Study Group. Arthritis Rheum 2000;43: Laine L, Harper S, Simon T, et al. A randomized trial comparing the effect of rofecoxib, a cyclooxygenase 2-specific inhibitor, with that of ibuprofen on the gastroduodenal mucosa of patients with osteoarthritis: Rofecoxib Osteoarthritis Endoscopy Study Group. Gastroenterology 1999;117; Goldstein JL, Eisen GM, Agrawal N, et al. Reduced incidence of upper gastrointestinal ulcer complications with the COX-2 selective inhibitor, valdecoxib. Aliment Pharmacol Ther 2004;20: Rordorf C, Kellett N, Mair S, et al. Gastroduodenal tolerability of lumiracoxib vs placebo and naproxen: a pilot endoscopic study in healthy male subjects. Aliment Pharmacol Ther 2003;18: Schnitzer TJ, Burmester GR, Mysler E, et al. Comparison of lumiracoxib with naproxen and ibuprofen in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), reduction in ulcer complications: randomised controlled trial. Lancet 2004;364: Laine L, Maller ES, Yu C, et al. Ulcer formation with low-dose enteric-coated aspirin and the effect of COX-2 selective inhibition: a double-blind trial. Gastroenterology 2004;127: Cox ER, Frisse M, Behm A, et al. Over-the-counter pain reliever and aspirin use within a sample of long-term cyclooxygenase 2 users. Arch Intern Med 2004;164: Scheiman JM, Yeomans ND, Talley NJ, et al. Prevention of ulcers by esomeprazole in at-risk patients using non-selective NSAIDs and COX-2 inhibitors. Am J Gastroenterol 2006;101: Hawkey CJ, Karrasch JA, Szczepanski L, et al. Omeprazole compared with misoprostol for ulcers associated with nonsteroidal antiinflammatory drugs: Omeprazole versus Misoprostol for NSAID-induced Ulcer Management (OMNIUM) Study Group. N Engl J Med 1998;338: Yeomans ND, Tulassay Z, Juhasz L, et al. A comparison of omeprazole with ranitidine for ulcers associated with nonsteroidal anti-inflammatory drugs: Acid Suppression Trial: Ranitidine versus Omeprazole for NSAID-associated Ulcer Treatment (ASTRONAUT) Study Group. N Engl J Med 1998;338: Chan FK, Chung SC, Suen BY, et al. Preventing recurrent upper gastrointestinal bleeding in patients with Helicobacter pylori infection who are taking low-dose aspirin or naproxen. N Engl J Med 2001;344: Lai KC, Lam SK, Chu KM, et al. Lansoprazole for the prevention of recurrences of ulcer complications from long-term low-dose aspirin use. N Engl J Med 2002;346:

8 1174 GOLDSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No Laine L. Proton pump inhibitor co-therapy with nonsteroidal antiinflammatory drugs: nice or necessary? Rev Gastroenterol Disord 2004;4(Suppl 4):S33 S Chan FK, Graham DY. Review article: prevention of non-steroidal anti-inflammatory drug gastrointestinal complications review and recommendations based on risk assessment. Aliment Pharmacol Ther 2004;19: Fendrick AM, Garabedian-Ruffalo SM. A clinician s guide to the selection of NSAID therapy. Pharmacy and Therapeutics 2002; 27: Recommendations for the medical management of osteoarthritis of the hip and knee: 2000 update American College of Rheumatology Subcommittee on Osteoarthritis Guidelines. Arthritis Rheum 2000;43: Goldstein JL. Challenges in managing NSAID-associated gastrointestinal tract injury. Digestion 2004;69(Suppl 1): Patrono C, Garcia Rodriguez LA, Landolfi R, et al. Low-dose aspirin for the prevention of atherothrombosis. N Engl J Med 2005;353: Emery P, Zeidler H, Kvien TK, et al. Celecoxib versus diclofenac in long-term management of rheumatoid arthritis: randomised double-blind comparison. Lancet 1999;354: Goldstein JL, Correa P, Zhao WW, et al. Reduced incidence of gastroduodenal ulcers with celecoxib, a novel cyclooxygenase-2 inhibitor, compared to naproxen in patients with arthritis. Am J Gastroenterol 2001;96: Goldstein JL, Aisenberg J, Berger M, et al. Effects of concomitant aspirin (81 mg qd) on incidence of gastric and/or duodenal ulcers in healthy subjects taking celecoxib or naproxen: a randomized, placebo-controlled trial (abstract). Gastroenterology 2006;130(Suppl 2):A Goldstein JL, Agrawal NM, Silverstein F, et al. Influence of H pylori (Hp) infection and/or low dose aspirin (ASA) on gastroduodenal ulceration in patients treated with placebo, celecoxib or NSAIDs (abstract). Gastroenterology 1999;116:A Goldstein JL, Kent J, Shu V, et al. Valdecoxib is less ulcerogenic than conventional NSAIDs in osteoarthritis (OA) and rheumatoid arthritis (RA) patients with high risk of peptic ulcer disease (abstract). Gastroenterology 2002;122(Suppl 1):A Cullen D, Bardhan KD, Eisner M, et al. Primary gastroduodenal prophylaxis with omeprazole for non-steroidal anti-inflammatory drug users. Aliment Pharmacol Ther 1998;12: Ekstrom P, Carling L, Wetterhus S, et al. Prevention of peptic ulcer and dyspeptic symptoms with omeprazole in patients receiving continuous non-steroidal anti-inflammatory drug therapy: a Nordic multicentre study. Scand J Gastroenterol 1996;31: Laine L, Curtis SP, Cryer B, et al. Upper gastrointestinal clinical events and symptoms in the double-blind multinational etoricoxib and diclofenac arthritis long-term (MEDAL) program. Lancet 2007; 369: Address requests for reprints to: Jay L. Goldstein, MD, Professor of Medicine, Vice Head for Clinical Affairs, Department of Medicine, University of Illinois at Chicago, 840 S Wood St (m/c 787), Room th Floor, Chicago, Illinois jlgoldst@uic.edu; fax: (312) This study was sponsored by TAP Pharmaceutical Products Inc, Lake Forest, IL (LAN ; ClinicalTrials.gov Identifier: NCT ). Dr Goldstein has served as a consultant to and has received honoraria, travel expenses, educational grants, and research grants from TAP Pharmaceutical Products Inc, Astra Zeneca, Merck, Novartis, Pozen, Takeda/Sucampo, GlaxoSmithKline, Given, and Pfizer. Dr Cryer has served as a consultant to and has received honoraria from TAP Pharmaceutical Products Inc, Pfizer, Merck, and Astra Zeneca. Fouad Amer and Barbara Hunt are employees of TAP Pharmaceutical Products Inc. Elizabeth Crane, MA, and Ingrid Hallman, MPH, TAP Pharmaceutical Products Inc, as well as Susan Ruffalo, PharmD, MedWrite, Inc, Newport Coast, CA provided professional medical writing assistance. Marsha Raanan, MS, TAP Pharmaceutical Products Inc, contributed to the statistical analysis.