Reduced incidence of gastroduodenal ulcers with celecoxib, a novel cyclooxygenase-2 inhibitor, compared to naproxen in patients with arthritis

THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2001 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.

Vol. 96, No. 4, 2001 ISSN 0002-9270/01/$20.00 PII S0002-9270(01)02303-6

Reduced Incidence of Gastroduodenal Ulcers With Celecoxib, a Novel Cyclooxygenase-2 Inhibitor, Compared to Naproxen in Patients With Arthritis J. L. Goldstein, M.D., P. Correa, M.D., W. W. Zhao, Ph.D., A. M. Burr, M.S., R. C. Hubbard, M.D., K. M. Verburg, Ph.D., and G. S. Geis, Ph.D., M.D. Section of Digestive and Liver Diseases, University of Illinois at Chicago, Chicago, Illinois; Department of Pathology, LSU Medical Center, New Orleans, Louisiana; and Clinical Research and Development, G. D. Searle & Co., Skokie, Illinois

OBJECTIVE: Nonsteroidal anti-inflammatory drugs (NSAIDs) block prostaglandin production by inhibiting cyclooxygenase (COX); they are believed to cause gastroduodenal damage by inhibiting the COX-1 isoform and to have analgesic and anti-inflammatory effects by inhibiting the COX-2 isoform. As compared to conventional NSAIDs, celecoxib, a COX-2 specific inhibitor, has been shown in previous single posttreatment endoscopy studies to be associated with lower gastroduodenal ulcer rates. In response to concerns that such studies may under-represent ulceration rates, the present serial endoscopy study was designed to compare cumulative gastroduodenal ulcer rates associated with the use of celecoxib to those of naproxen, a conventional NSAID. METHODS: In this double-blind, parallel-group, multicenter study, 537 patients with osteoarthritis (OA) or rheumatoid arthritis (RA) were randomized to treatment with celecoxib 200 mg b.i.d. (n ⫽ 270) or naproxen 500 mg b.i.d. (n ⫽ 267) for 12 wk. Gastroduodenal damage was determined from esophagogastroduodenoscopy after 4, 8, and 12 wk of therapy. Arthritis efficacy was evaluated with Patient’s and Physician’s Global Assessments. RESULTS: Gastroduodenal ulcer rates after celecoxib and naproxen treatment were 4% versus 19% in the 0 – 4 wk interval (p ⬍ 0.001), 2% versus 14% in the 4 – 8 wk interval (p ⬍ 0.001), and 2% versus 10% in the 8 –12 wk interval (p ⬍ 0.001), respectively. After 12 wk of treatment, the cumulative incidence of gastroduodenal ulcers was 9% with celecoxib and 41% with naproxen. In the celecoxib group, gastroduodenal ulcers were significantly associated with Helicobacter pylori status (p ⬍ 0.05), concurrent aspirin usage (p ⫽ 0.001), and a history of ulcer (p ⫽ 0.010), but not with disease type (OA/RA), age, gender, other relevant medical histories, or concurrent corticosteroid or diseasemodifying antirheumatic drugs usage (p ⬎ 0.05). Celecoxib produced a significantly lower incidence rate of both gastric (p ⬍ 0.001) and duodenal (p ⬍ 0.030) ulcers. The two agents produced similar improvements in Patient’s and Phy-

sician’s Global Assessments of arthritis efficacy. The incidence of adverse events and withdrawal rates did not differ significantly between treatments. CONCLUSIONS: As compared to naproxen (500 mg b.i.d.), use of celecoxib (200 mg b.i.d.), a COX-2 specific agent, at the recommended RA dose and twice the most frequently prescribed OA dose, was associated with lower rates of gastric, duodenal, and gastroduodenal ulcers but had comparable efficacy, in patients with OA and RA. (Am J Gastroenterol 2001;96:1019 –1027. © 2001 by Am. Coll. of Gastroenterology)

INTRODUCTION Nonsteroidal anti-inflammatory drugs (NSAIDs) are commonly used to manage pain and inflammation associated with osteoarthritis (OA) or rheumatoid arthritis (RA) (1). Although these agents often provide relief of symptoms, their use is associated with well-known risks. Potentially serious adverse effects of NSAID therapy, most notably including ulceration of GI mucosa and associated hemorrhage, and interference with platelet function, compromise the utility of these agents for many patients (2–11). The likely mechanism of analgesic and anti-inflammatory actions of NSAIDs is the blockade of prostaglandin synthesis through inhibition of cyclooxygenase (COX). Recent research shows that there are at least two forms of the COX enzyme. COX-1 is produced constitutively in virtually all body tissues, including the gut, kidney, and platelets (12, 13). In the upper GI tract, COX-1-mediated prostaglandin synthesis plays an important role in maintaining mucosal integrity (14). COX-2, in contrast, is normally found at low levels in most body tissues, but is expressed at high levels in inflamed tissues (15–18). Mediators of inflammation (e.g., tumor necrosis factor, interleukin-1) have been shown to up-regulate the expression of COX-2 in activated macrophages and synoviocytes (19 –25). The finding that glucocorticoids inhibit up-regulation of COX-2 further supports the role of COX-2 as an inducible enzyme in

1020

Goldstein et al.

inflammation pathways (25). Taken together, these observations provide a basis for understanding how nonselective inhibition of both COX-1 and COX-2 by NSAIDs produces the well-known profile of anti-inflammatory effects coupled with the development of gastric, duodenal ulcers, and ulcerrelated and bleeding-related adverse events. Celecoxib, the first in a new class of COX-2 inhibitors, has a high specificity for COX-2 enzyme at therapeutic doses (26). Experience in clinical trials conducted to date has consistently shown that use of celecoxib has been associated with improvement in the signs and symptoms of OA and RA with no appreciable effects on platelet function nor significantly increased gastroduodenal ulcer rates as compared to placebo (27–29). Using monthly serial endoscopy, the primary objective of the current study was to compare the cumulative incidence of upper GI ulceration among OA and RA patients receiving 12 wk of treatment with celecoxib, at the recommended RA dose and twice the most frequently prescribed OA dose versus naproxen. It is noteworthy that this trial differs from previously reported studies that evaluated ulcer rates in OA and RA patients at a single time point. For this current trial, ulcer incidence rates determined on a monthly basis are reported in response to concerns that performing a single posttreatment endoscopy at 12 wk may under-represent the true ulceration rates associated with celecoxib or NSAID treatment. The rationale for this concern is that single posttreatment endoscopies fail to detect endoscopic ulcers that develop and resolve spontaneously over a 12-wk period of exposure.

MATERIALS AND METHODS Patients Men and women of legal age of consent were eligible to participate in the study if they had a diagnosis of active OA or RA for at least 3 months and required chronic therapy with NSAIDs. All patients were to have a Functional Capacity Classification of I to III at baseline (30), i.e., patients were excluded if they were largely or wholly incapacitated, permitting little or no self-care. Patients were excluded from the study if they had: any other inflammatory arthritis, acute gout, active GI disease (e.g., inflammatory bowel disease), a diagnosis of (or required treatment for) any acute upper GI ulceration within 30 days before the first dose of study medication, or if they had taken naproxen (ⱖ1000 mg/day) within 30 days before the first dose of study medication. Patients with endoscopic ulcers (ⱖ3 mm in diameter with depth) documented at the time of pre-enrollment screening esophagogastroduodenoscopy (EGD) were excluded. Women of childbearing potential were required to have confirmed use of adequate contraception and a negative serum pregnancy test within 7 days before the first dose of study medication.

AJG – Vol. 96, No. 4, 2001

Study Protocol This protocol was a multicenter, double-blind, parallelgroup study conducted at 75 sites in the United States. The study protocol was approved by the Institutional Review Boards at each participating center. Each patient provided written informed consent. PRETREATMENT PERIOD. At the baseline visit, patients provided a full medical history, and underwent a physical exam and clinical laboratory testing, including complete blood count, BUN, creatinine, total bilirubin, alkaline phosphatase, AST, ALT, and creatinine kinase (Covance Central Laboratory Services, Indianapolis, IN). A serological test for Helicobacter pylori (H. pylori) antibodies was also performed at the baseline visit (FlexSure HP, Smith Kline Diagnostics Inc, Palo Alto, CA). EGD was performed within 7 days before the first dose of study medication to determine eligibility for the study and to provide baseline endoscopic data. Patients discontinued NSAIDs at or before endoscopy, and prescription and over-the-counter anti-ulcer medications 1 day before screening EGD. Patient’s and Physician’s Global Assessments of Arthritic Condition and Functional Capacity Classification were also performed within 7 days before the first dose of study medication (31, 32). Patients were randomly assigned to treatment in the order in which they were enrolled at each site, according to a computer-generated randomization schedule. Separate schedules were prepared for each type of arthritis (i.e., OA or RA). TREATMENT PERIOD. Patients returned for evaluation at the end of wk 4 (day 28 ⫾ 5 days), wk 8 (day 56 ⫾ 5 days), and wk 12 (day 84 ⫾ 5 days). An EGD was performed at each visit; in addition, an unscheduled endoscopy could be performed at any time at the discretion of the physician if a patient had a sign or symptom indicative of GI ulceration. For consistency, the same endoscopist who had performed a patient’s baseline examination performed that patient’s follow-up endoscopies. In each case, the endoscopist used the same equipment and measuring devices and was blinded to treatment. At the wk 12 EGD, antral biopsies were taken for a CLOtest (Ballard Medical Products, Draper, UT) and for histology (hematoxylin-eosin, modified Steiner’s) (33). Histology was considered positive if one or both stains demonstrated the presence of organisms. Only patients who tested positive on both CLOtest and histology were considered H. pylori positive. Similarly, only patients who tested negative on both tests were considered H. pylori negative. All other patients were excluded from the subsequent analysis evaluating the influence of H. pylori status on the development of ulcers. Patients were considered to have completed therapy if they received the full 12 wk of therapy without developing a gastric, pyloric channel, or duodenal ulcer at the time of a scheduled or unscheduled endoscopic examination.

AJG – April, 2001

Treatment Patients were randomized to one of two treatment groups: celecoxib 200 mg b.i.d. or naproxen 500 mg b.i.d. Patients assigned to celecoxib treatment received celecoxib 200-mg capsules and placebo capsules matching naproxen 500-mg capsules. Patients assigned to receive naproxen treatment received naproxen 500-mg capsules and placebo capsules matching celecoxib 200-mg capsules. All patients were instructed to take study medication twice daily with breakfast and dinner. Concomitant Medications Use of drugs other than study medication was discouraged during the treatment period. The following drugs were specifically prohibited: NSAIDs (other than aspirin 325 mg/day for nonarthritis-related reasons); prescription or over-thecounter anti-ulcer drugs (including misoprostol); antibiotics to treat H. pylori infection; anticoagulants; antacids; and antineoplastics, other than methotrexate (ⱕ20 mg/wk), or azathioprine for the treatment of RA. Acetaminophen could be used throughout the study; oral corticosteroids were allowed only if there was no change in the dosing regimen during the study. Measures of Efficacy Arthritis was assessed at each visit by the Patient’s and Physician’s Global Assessments, with a range of 1 (very good/asymptomatic) to 5 (very poor/severe, intolerable symptoms) (31, 32). Treatment efficacy was also measured by the incidence of withdrawal resulting from lack of arthritis efficacy. Statistical Analysis All randomized patients who took one dose of study medication were included in the analyses (intent-to-treat cohort). Homogeneity between treatment groups with respect to age, height, weight, and vital signs was analyzed by a two-way analysis of variance with treatment and center as factors. Homogeneity between treatment groups in terms of gender and race was analyzed by the ␹2 test. Differences between treatment groups with respect to medical history were analyzed using Fisher’s exact test. Every randomized patient who received at least one dose of study medication was included in the safety analysis. Based on endoscopy findings, investigators assigned gastroduodenal mucosal scores on a modified Lanza scale of 0 (no visible lesions/normal mucosa) to 7 (ulcer/break in the mucosa ⱖ3 mm in diameter with unequivocal depth) with intermediate scores according to the number of petechiae or erosions. Erosions were defined as any break in the mucosa without depth (34). The ulcer incidence and ulcer/erosion incidence (above prespecified endoscopy scores of 7 and 2, respectively) and their score distributions were analyzed with Cochran-Mantel-Haenszel (CMH) tests, stratified by baseline status. For gastroduodenal rates, gastric rates, and duodenal rates, the differential effect of H. pylori status (as determined from the

Upper GI Safety of Celecoxib

1021

FlexSure HP and CLOtest/biopsy results) were examined by both analysis of covariance (with treatment, center, H. pylori status as factors and baseline as covariate) and CMH (stratified by baseline and treatment). When “for cause” endoscopies were performed (i.e., in response to signs or symptoms of ulcer), any ulcers found were reported and adverse events were generated. If ulcers were found, they were carried over and included in the ulcer rates and the next regularly scheduled visit. Similarly, discontinuation resulting from an ulcer revealed by an unscheduled endoscopic examination was considered discontinuation caused by an adverse event. Changes from baseline and mean values of Patient’s and Physician’s Global Assessments were analyzed by the CMH test (stratified by center) and analysis of covariance (with treatment and center as factors and baseline value as covariate), respectively. The incidence of withdrawal resulting from lack of arthritis efficacy was analyzed by Fisher’s exact test. Sample Size Calculation Based on the clinical assumption that the gastroduodenal ulceration rate was 3% for celecoxib and 11% for naproxen 500 mg b.i.d., it was determined that a sample size of 200 patients per treatment group was sufficient to detect a difference of this magnitude between treatment groups using a two-sided test with a 0.05 significance level and 80% power.

RESULTS Patient Demographics and Disposition In total, 536 of the 537 randomized patients received at least one dose of study medication, 269 patients in the celecoxib group and 267 patients in the naproxen group. At baseline, the treatment groups were well matched with respect to patient demographics and clinical characteristics (Table 1). Patients ranged in age from 22 to 86 yr, with a mean of 57 yr. The majority of patients was white (83– 84%), women (67%), and had negative FlexSure HP serological test results (67%). Furthermore, the treatment groups did not differ with respect to height (p ⫽ 0.272), weight (p ⫽ 0.728), or vital signs (p ⱖ 0.150). The percentage of OA and RA patients in each treatment group, as well as the duration of disease, did not differ significantly. Disease duration ranged from 4 months to 50 yr. Furthermore, the treatment groups did not differ with respect to their self-reported medical history; 4 –5% had a history of GI bleeding (p ⫽ 0.411), 20 –21% had a history of gastroduodenal ulcer (p ⫽ 0.749), 13–14% had a history of GI intolerance to NSAIDs (p ⫽ 0.800), and 50 –56% had a history of cardiovascular disease (including edema, hypertension, etc.; p ⫽ 0.167). There were no significant differences in the distribution of baseline endoscopy scores between treatment groups in either the stomach (p ⫽ 0.176) or duodenum (p ⫽ 0.579) (Table 2). A single patient was inadvertently randomized

1022

Goldstein et al.

AJG – Vol. 96, No. 4, 2001

Table 1. Demographics and Baseline Characteristics* Celecoxib 200 mg b.i.d. (n ⫽ 270) Mean age (range) Sex (%) women Race (%) Caucasian Black Hispanic Other Type and duration of arthritis, no. (mean ⫾ SD yr) OA RA Past medical history (%) GI bleeding Gastroduodenal ulcer† GI NSAID intolerance Cardiovascular disease Pretreatment aspirin use Glucocorticoid use, baseline Glucocorticoid use, history Positive for H. pylori antibody, (%)‡

56.7 (23–86) 67

Naproxen 500 mg b.i.d. (n ⫽ 267) 57.7 (22–84) 67

83 12 4 1

84 13 4 0

194 (9.5 ⫾ 8.3) 76 (11.6 ⫾ 10.0)

195 (11.0 ⫾ 9.6) 72 (8.9 ⫾ 7.1)

4 21 13 56 15 14 48 33

5 20 14 50 12 9 41 33

OA ⫽ osteoarthritis; RA ⫽ rheumatoid arthritis; NSAID ⫽ nonsteroidal anti-inflammatory drug. * Numbers reflect all randomized patients. One patient in the celecoxib group did not take study medication; therefore, tables and figures relating to posttreatment results show 269 patients in the celecoxib group. † History of past gastroduodenal ulcer ⬎30 days before randomization. ‡ By FlexSure H. pylori serological test.

and entered in the naproxen group with a gastric ulcer at baseline endoscopy. Table 2. Baseline Endoscopy Results* Celecoxib Naproxen 200 mg b.i.d. 500 mg b.i.d. (n ⫽ 270) (n ⫽ 267) p Values Stomach Endoscopy score (mean ⫾ SD) Distribution (%) 0 (no visible lesions) 1 (1–10 petechiae) 2 (⬎10 petechiae) 3 (1–5 erosions) 4 (6–10 erosions) 5 (11–25 erosions) 6 (⬎25 erosions) 7 (ulcer) Duodenum Endoscopy score (mean ⫾ SD) Distribution (%) 0 (no visible lesions) 1 (1–10 petechiae) 2 (⬎10 petechiae) 3 (1–5 erosions) 4 (6–10 erosions) 5 (11–25 erosions) 6 (⬎25 erosions) 7 (ulcer†)

1.1 ⫾ 1.48

1.1 ⫾ 1.55

60 11 3 19 6 1 ⬍1 0

60 11 2 21 4 1 ⬍1 1

0.3 ⫾ 0.86

0.2 ⫾ 0.74

85 7 1 6 ⬍1 ⬍1 0 0

88 6 1 4 ⬍1 0 0 0

0.176

0.579

* Numbers reflect all randomized patients. One patient in the celecoxib group did not take study medication; therefore, tables and figures relating to posttreatment results show 269 patients in the celecoxib group. † Defined as ulcer/break in the mucosa ⱖ3 mm in diameter with unequivocal depth.

Of the 536 patients who entered the study, 359 (67%) completed 12 wk of treatment, including 210 (78%) patients in the celecoxib group and 149 (56%) patients in the naproxen group. In the celecoxib group, 20 (7%) patients discontinued because of adverse events, 17 (6%) because of treatment failure, and 12 (4%) because of identification of an ulcer at the scheduled 4- or 8-wk endoscopies (i.e., not “for cause” endoscopies). In comparison, 67 (25%) patients in the naproxen group discontinued because of an identified ulcer at the 4- or 8-wk endoscopies, 24 (9%) patients because of adverse events, 11 (4%) because of treatment failure, nine (3%) because of protocol noncompliance, and one patient because of a pre-existing protocol violation (ulcer at baseline). In addition, two (⬍1%) patients in the celecoxib group and six (2%) patients in the naproxen group were lost to follow-up. The withdrawal rates caused by treatment failure (p ⫽ 0.332) and adverse events (p ⫽ 0.431) did not differ significantly between treatments nor did the time to withdrawal because of these reasons. Endoscopy OVERALL ULCER RATES. Significantly more patients in the naproxen group developed gastroduodenal ulcers than in the celecoxib group. Gastroduodenal ulcer rates observed after celecoxib and naproxen treatment were 4% versus 19% in the 0 – 4 wk interval (p ⬍ 0.001), 2% versus 14% in the 4 – 8 wk interval (p ⬍ 0.001), and 2% versus 10% at the 8 –12 wk interval (p ⬍ 0.001), respectively (Fig. 1). During the 12 wk of the study, gastroduodenal ulcers developed in 9% of patients in the celecoxib group and 41% of patients in the naproxen group (p ⬍ 0.001) (Fig. 2).

AJG – April, 2001

Upper GI Safety of Celecoxib

1023

Table 3. Week 12 Endoscopy Results* Celecoxib Naproxen 200 mg b.i.d. 500 mg b.i.d. (n ⫽ 270) (n ⫽ 267) p Values

Figure 1. Incidence of gastroduodenal ulcers determined by endoscopy after 4, 8, and 12 wk of treatment with celecoxib 200 mg b.i.d. or naproxen 500 mg b.i.d. Ulcers determined at the time of “for cause” (unscheduled) endoscopies were carried forward and reported with the scheduled endoscopy results.

During the 12 wk of the study, gastric ulcers developed in 6% of patients in the celecoxib group and 37% of patients in the naproxen group (p ⬍ 0.001) (Fig. 2). Furthermore, cumulative duodenal ulcer rates after 12 wk were significantly lower in the celecoxib group (4%) compared to the naproxen group (12%, p ⬍ 0.002) (Fig. 2). Among patients with a gastric endoscopy score of 0 – 6, the distribution shifted significantly (p ⬍ 0.001) to higher scores in the naproxen group compared to the celecoxib group (Table 3). A similar shift in distribution was also observed in duodenal scores (p ⫽ 0.023).

Figure 2. Cumulative incidence of gastroduodenal, gastric, and duodenal ulcers determined by endoscopy after 12 wk of treatment with celecoxib 200 mg b.i.d. or naproxen 500 mg b.i.d. Ulcers determined at the time of “for cause” (unscheduled) endoscopies were carried forward and reported with the scheduled endoscopy results.

Stomach Endoscopy score (mean ⫾ SD) Distribution (%) 0 (no visible lesions) 1 (1–10 petechiae) 2 (⬎10 petechiae) 3 (1–5 erosions) 4 (6–10 erosions) 5 (11–25 erosions) 6 (⬎25 erosions) 7 (ulcer) Duodenum Endoscopy score (mean ⫾ SD) Distribution (%) 0 (no visible lesions) 1 (1–10 petechiae) 2 (⬎10 petechiae) 3 (1–5 erosions) 4 (6–10 erosions) 5 (11–25 erosions) 6 (⬎25 erosions) 7 (ulcer†)

1.1 ⫾ 1.98

3.8 ⫾ 2.85

65 9 5 9 4 1 ⬍1 6

25 6 1 20 6 3 ⬍1 37

0.7 ⫾ 1.65

1.2 ⫾ 2.35

79 6 2 7 ⬍1 ⬍1 ⬍1 4

71 7 1 8 0 1 0 12

⬍0.001

⬍0.001

* Numbers reflect all randomized patients. One patient in the celecoxib group did not take study medication; therefore, tables and figures relating to post-treatment results show 269 patients in the celecoxib group. † Defined as ulcer/break in the mucosa ⱖ3 mm in diameter with unequivocal depth.

EFFECT OF BASELINE COVARIATES AND RISK FACTORS. The incidence of gastroduodenal ulcers in the celecoxib group appeared to be higher among patients with positive H. pylori status (Fig. 3). By 12 wk of treatment, gastroduodenal ulcers developed in four of 31 (12.9%) celecoxib-treated patients who were positive for H. pylori as

Figure 3. Effect of H. pylori status on the incidence of gastroduodenal ulcers after 12 wk of treatment with celecoxib 200 mg b.i.d. or naproxen 500 mg b.i.d.

1024

Goldstein et al.

AJG – Vol. 96, No. 4, 2001

Table 4. Gastroduodenal Ulcer Incidences by Patient Subgroup* Celecoxib 200 mg b.i.d. (n ⫽ 269) Age ⬍65 yr ⱖ65 yr Sex Women Men History of: GI NSAID intolerance Yes No Gastroduodenal ulcer Yes No GI bleeding Yes No Cardiovascular disease Yes No H. pylori serology Positive Negative Low dose aspirin use Yes No Glucocorticoid use Yes No Disease OA RA

Naproxen 500 mg b.i.d. (n ⫽ 267)

7.1 (11/153) 12.1 (7/58)

37.7 (55/146) 47.1 (32/68)

7.7 (11/143) 10.3 (7/68)

44.3 (66/149) 32.3 (21/65)†

13.3 (4/30) 7.7 (14/181)

55.2 (16/29) 38.6 (71/184)

18.2 (8/44) 6.0 (10/167)†‡

63.6 (28/44) 34.9 (59/169)†‡

25.0 (2/8) 7.9 (16/203)

54.5 (6/11) 40.1 (81/202)

10.7 (13/121) 5.6 (5/90)

41.3 (45/109) 40.0 (42/105)

12.9 (4/31) 2.9 (4/136)†

29.6 (8/27) 30.2 (32/106)

24.1 (7/29) 6.0 (11/182)†

42.9 (12/28) 40.3 (75/186)

7.7 (2/26) 8.6 (16/185)

40.0 (8/20) 40.7 (79/194)

8.4 (13/154) 8.8 (5/57)

45.2 (70/155) 28.8 (17/59)

NSAID ⫽ nonsteroidal anti-inflammatory drug; OA ⫽ osteoarthritis; RA ⫽ rheumatoid arthritis. * Entries are percentages (no. of subjects with ulcer/total). † p ⬍ 0.05 for difference within treatment group. ‡ p ⬍ 0.05 between celecoxib and naproxen.

determined by the combination of both CLOtest and histology, but in only four of 136 (2.9%) patients with negative H. pylori results (p ⫽ 0.023). Similarly, when H. pylori status was based only on the baseline FlexSure HP test, gastroduodenal ulcers developed in 10 of 73 (13.7%) celecoxibtreated patients with positive results and in eight of 137 (5.8%) patients with negative results (p ⫽ 0.047). In comparison, H. pylori status did not significantly affect the percentage of patients who developed gastroduodenal ulcers in the naproxen group. The incidence of gastric ulcers in the celecoxib group was significantly higher among patients with positive H. pylori findings by a combination of both CLOtest and histology (p ⫽ 0.026), but not by the FlexSure HP test (p ⫽ 0.494). Furthermore, there was no association between H. pylori status and the incidence of duodenal ulcer in the celecoxib group, or either gastric or duodenal ulcer in the naproxen group. The incidence of gastroduodenal ulcers was further analyzed by stratification according to potential risk factors (Table 4). The percentage of celecoxib patients who devel-

oped ulcers was significantly higher among those who used low dose aspirin (24.1%) than among those who did not (6%, p ⬍ 0.001). Similarly, the incidence of gastroduodenal ulcers was higher among those with a history of gastroduodenal ulcer (18.2%) than among those without such history (6%, p ⫽ 0.010). In contrast, ulcer incidence in the celecoxib group was not significantly related to patients’ age (⬍65 yr vs ⱖ65 yr, p ⫽ 0.260), gender (p ⫽ 0.501), history of GI intolerance to NSAIDs (p ⫽ 0.282), history of GI bleeding (p ⫽ 0.077), cardiovascular disease (not requiring aspirin prophylaxis) (p ⫽ 0.170), current use of steroids (p ⫽ 0.969), or disease-modifying antirheumatic drugs (p ⫽ 0.833). In the naproxen group, the incidence of ulcers was significantly related only to a history of gastroduodenal ulcers (63.6% vs 34.9%, p ⬍ 0.001) and to female gender (44.3% vs 32.3%, p ⫽ 0.048). The percentage of OA and RA patients in the celecoxib group who developed ulcers was similar. By 12 wk of treatment, 13 (8.4%) of 154 OA patients and five (8.8%) of 57 RA patients developed ulcers (p ⫽ 0.977). In contrast, the percentage of patients who developed ulcers in the naproxen group, although not significant, tended to be higher among OA (45.2%) than among RA patients (28.8%, p ⫽ 0.079) (Table 4). Efficacy Because this study was primarily focused on safety, it is important to recognize that the efficacy of celecoxib 200 mg b.i.d. and naproxen 500 mg b.i.d. were equivalent at all time points within the trial. The equivalency in efficacy was assessed using the Patient’s and Physician’s Global Assessments. In both the celecoxib and naproxen groups, the mean Patient’s Global Assessment score equally improved from 2.9 at baseline to 2.5 after 12 wk of treatment, whereas the mean Physician’s Global Assessment score improved from 2.9 at baseline to 2.4 after 12 wk. Of 269 patients in the celecoxib group, 128 (48%) improved by at least 1 grade in the Patient’s Global Assessment score, 97 (36%) remained unchanged, and 44 (16%) had an increase of at least 1 grade. Similarly, of 267 patients in the naproxen group, 120 (45%) improved by at least 1 grade, 109 (41%) remained unchanged, and 38 (14%) worsened. Similar results occurred in the Physician’s Global Assessment. The difference between treatments was not statistically significant for either the Patient’s (p ⫽ 0.809) or Physician’s (p ⫽ 0.997) Global Assessment scores at the 4-, 8-, or 12-wk interval. Safety The most commonly reported adverse events included dyspepsia, headache, abdominal pain, diarrhea, sinusitis, upper respiratory infection, flatulence, and nausea (Table 5). A total of 19 (7%) patients in the celecoxib group and 24 (9%) patients in the naproxen group experienced an adverse event that led to withdrawal. A total of 92 (34%) patients in the celecoxib group and 107 (40%) in the naproxen group

AJG – April, 2001

Upper GI Safety of Celecoxib

Table 5. Incidence of Adverse Events Celecoxib Naproxen 200 mg b.i.d. 500 mg b.i.d. (%) (%) Total adverse events Total causing withdrawal Most common adverse events Dyspepsia Headache Abdominal pain Diarrhea Sinusitis Upper respiratory tract infection Flatulence Nausea GI adverse events Total Causing withdrawal

70 7

70 9

13 12 9 8 6 6 4 3

10 17 12 6 3 5 5 9

34 4

40 6

reported GI adverse events, most of which were mild to moderate in severity. Of these, 11 (6%) patients in the celecoxib group and 16 (4%) patients in the naproxen group experienced a GI adverse event that led to withdrawal. Overall, celecoxib was associated with a trend for a lower incidence of dyspepsia, abdominal pain, diarrhea, flatulence, and nausea combined than naproxen (37% vs 42%). However, there were no statistical differences for the total number or percent of GI or non-GI-related adverse events, withdrawal rates or any of the individual adverse events reported. Two clinically significant upper GI ulcer complications occurred in the naproxen group — a pyloric channel obstruction and an ulcer with evidence of GI bleeding. No clinically significant adverse GI events occurred in the celecoxib group. With the exception of decreased Hct (ⱖ5% from baseline: 13% celecoxib, 12% naproxen) and Hb (ⱖ2 g/dl from baseline: 2% celecoxib, 3% naproxen), no clinically relevant changes in laboratory values were observed in more than 2% of patients in either treatment group.

DISCUSSION Our study compared the incidence of gastroduodenal ulcers associated with up to 12 wk of treatment with celecoxib 200 mg b.i.d. with that of naproxen 500 mg b.i.d. in patients with OA and RA, by endoscopy performed at monthly intervals. A number of endoscopy trials lasting from 1 month to 1 yr have demonstrated a gastroduodenal ulcer incidence ranging from 15% to 40% with such commonly used NSAIDs as diclofenac, naproxen, ibuprofen, and piroxicam (35– 40). The endoscopic data parallel the results of case control studies, cohort studies and meta-analyses of ulcer complications occurring with conventional NSAID exposure. These studies demonstrate that the risk for hospitalization or death caused by gastroduodenal ulceration and upper GI bleeding is increased in patients taking NSAIDs and that the risk for an ulcer complication is further in-

1025

creased by such factors as age, sex, type of arthritis, duration of NSAID use, NSAID dose, and history of upper GI ulceration and bleeding (3– 6, 40 – 43). Serial endoscopy was performed in the current study to maximize the detection of gastroduodenal ulcers, particularly those that develop but subsequently resolve before the end of treatment. This strategy reduces the risk of underestimating the true ulceration rate associated with celecoxib or naproxen treatment. Two recent randomized, comparative, 12- and 24-wk endoscopy studies in RA patients have shown similar results to those in the current study (27, 29). The incidence of gastroduodenal ulcers in the 12-wk study was 4% in the placebo group, 4% in the celecoxib 200-mg b.i.d. group, and 26% in the naproxen 500-mg b.i.d. group (p ⬍ 0.05 vs all other groups) (27). Furthermore, the incidence of ulcers in patients treated with celecoxib doses as high as 400 mg b.i.d. (6%) was similar to placebo and significantly lower than naproxen (p ⬍ 0.05). In the 24-wk study, gastroduodenal ulcers were detected endoscopically in 15% of patients treated with diclofenac slow release 75 mg b.i.d. and in 4% treated with celecoxib 200 mg b.i.d. (p ⬍ 0.001) (29). The results of the present study confirm and extend these previous observations by showing that celecoxib produces less gastroduodenal damage than a conventional NSAID during long-term treatment of arthritis patients, some of whom had traditional risk factors such as history of ulcers, cardiovascular disease, and GI bleeding. Serial endoscopies performed at monthly intervals demonstrated that celecoxib treatment resulted in a significantly lower incidence of ulcers than naproxen over each interval, and that the cumulative rate of ulcer development was also significantly lower. In addition to causing significantly fewer ulcers, celecoxib produced significantly less gastric and duodenal erosions than naproxen. The pathogenesis of ulcers that developed in patients given celecoxib could not be established in the present study. Celecoxib-treated patients with a positive H. pylori finding determined by a combination of both CLOtest and histology were associated with a significantly higher incidence of gastric ulcers than those with a negative finding. On the other hand, duodenal ulceration appeared to be independent of H. pylori status. This finding should be interpreted with caution, however, as it is based on data from only eight celecoxib-treated patients with duodenal ulcers. In naproxen-treated patients, ulcer incidence was independent of H. pylori status, the impact of this factor being far outweighed by the ulcerogenic potential of naproxen. A substantial body of controlled, cohort, and case-control studies confirm that there is an increased prevalence of gastric and duodenal ulcer and of associated complications in subjects treated with low dose aspirin. The overall percentage of ulcers/erosions in patients treated with aspirin at therapeutic doses ranges from 10% to 50% with a relative risk of bleeding ranging from 1.8% to 15% (44). In the

1026

Goldstein et al.

present study, celecoxib-treated patients who received concomitant aspirin (ⱕ325 mg/day) were at significantly higher risk of developing gastroduodenal ulcer than the subgroup who did not receive aspirin. This is in marked contrast with the naproxen group, in which the gastroduodenal ulcer rate appeared to be overshadowed and independent of aspirin use. Female sex, age, alcohol consumption, smoking, cardiovascular disease, presence of RA, history of GI ulceration or bleeding, and corticosteroid use have all been implicated as risk factors for the development of NSAID-induced symptomatic ulcers or ulcer complications, such as GI bleeding, perforation, or gastric outlet obstruction (45, 46). It has not been established, however, whether these are risk factors for gastroduodenal ulceration per se or whether they create a predisposition for ulcers to deteriorate to ulcer complications in susceptible individuals. In the present study, a history of gastroduodenal ulcers was associated with a significantly higher incidence of ulcers in both the celecoxib and naproxen groups. However, the influence of this factor was less in the celecoxib-treated patients; celecoxib treatment was associated with a significantly lower rate of ulceration when compared to naproxen. In the celecoxib group, patients with a history of GI intolerance to NSAIDs or GI bleeding appeared to be at increased risk of developing duodenal ulcers, though even in these subgroups, the incidence of duodenal ulcers was lower than that observed in the naproxen group. There was no evidence of a relationship between either of these risk factors and the incidence of gastroduodenal or gastric ulcers in either treatment group. The results of both Patient’s and Physician’s Global Assessments in the current study suggest that celecoxib 200 mg b.i.d., the highest recommended RA dose and twice the most frequently prescribed OA dose, provides similar efficacy to naproxen 500 mg b.i.d. in treating the signs and symptoms of arthritis. These findings are consistent with the results of two recent pivotal studies in patients with OA of the knee and in RA patients, in which celecoxib 200 mg b.i.d. significantly reduced arthritic signs and symptoms according to standard arthritis efficacy measures, including Patient’s and Physician’s Global Assessments (27, 28). The current study also suggests that celecoxib may be better tolerated than naproxen. The overall incidence of adverse events between the two groups was similar, though slightly fewer patients in the celecoxib group discontinued therapy because of adverse events. Furthermore, fewer patients in the celecoxib group reported GI adverse events and withdrawals because of GI adverse events compared with the naproxen group. Notably, two serious upper GI ulcer complications (a pyloric channel obstruction and an ulcer with evidence of GI bleeding) were reported in the study, both in patients receiving naproxen. We conclude that celecoxib, a COX-2 specific inhibitor that does not inhibit COX-1 at therapeutic dosages, is as efficacious but produces significantly less gastroduodenal

AJG – Vol. 96, No. 4, 2001

damage in arthritis patients than naproxen, a conventional NSAID that inhibits both COX-1 and COX-2. Though the correlation between endoscopic findings and the development of upper GI ulcer complications remain to be fully established in prospective, long-term comparative studies, the present study suggests that celecoxib offers a rational alternative to conventional NSAIDs for the treatment of arthritis.

ACKNOWLEDGMENT This work was supported by G. D. Searle & Co. and Pfizer Inc. Reprint requests and correspondence: J. L. Goldstein, M.D., Professor of Medicine, Section of Digestive and Liver Diseases, University of Illinois at Chicago, 840 South Wood (M/C 787), Room 734, CSB, Chicago, IL 60612-7323. Received June 21, 2000; accepted Nov. 1, 2000.

REFERENCES 1. Roth SH, Bennett RE. Nonsteroidal anti-inflammatory drug gastropathy. Recognition and response. Arch Intern Med 1987;147:2093–100. 2. Fries JF. NSAID gastropathy: The second most deadly rheumatic disease? Epidemiology and risk appraisal. J Rheumatol 1991;18:6 –10. 3. Borda IT. The spectrum of adverse gastrointestinal effects associated with non-steroidal anti-inflammatory drugs. In: Borda IT, Koff RS, eds. NSAIDs — a profile of adverse effects. Philadelphia: Hanley and Belfus, 1994:25– 80. 4. Garcia-Rodriguez LA, Jick H. Risk of upper gastrointestinal bleeding and perforation associated with non-steroidal antiinflammatory drugs. Lancet 1994;343:769 –72. 5. Gabriel SE, Jaakimainen K, Bombardier C. Risks for serious gastrointestinal complications related to use of nonsteroidal anti-inflammatory drugs: A meta-analysis. Ann Intern Med 1991;115:787–96. 6. Langman MJS, Weil K, Wainwright P, et al. Risks of bleeding peptic ulcer associated with individual non-steroidal anti-inflammatory drugs. Lancet 1994;343:1075– 8. 7. Carmichael J, Shankel SW. Effects of nonsteroidal anti-inflammatory drugs on prostaglandins and renal function. Am J Med 1985;78:992–1000. 8. Whelton A, Stout RL, Spillman PS, et al. Renal effects of ibuprofen, piroxicam, and sulindac in patients with asymptomatic renal failure. Ann Intern Med 1990;112:568 –76. 9. Schlondorff D. Renal complications of nonsteroidal anti-inflammatory drugs. Kidney Int 1993;44:643–53. 10. Houston MC. Nonsteroidal anti-inflammatory drugs and antihypertensives. Am J Med 1991;90(suppl 5A):42S– 47S. 11. Burch JW, Stanford N, Majerus PW. Inhibition of platelet prostaglandin synthetase by oral aspirin. J Clin Invest 1978; 61:314 –9. 12. Merlie JP, Fagan D, Mudd J, et al. Isolation and characterization of the complementary DNA for sheep seminal vesicle prostaglandin endoperoxide synthase (cyclooxygenase). J Biol Chem 1988;263:3550 –3. 13. De Witt DL, Smith WL. Primary structure of prostaglandin G/H synthase from sheep vesicular gland determined from the

AJG – April, 2001

14.

15.

16.

17.

18.

19. 20.

21.

22. 23. 24. 25. 26.

27.

28.

29.

complementary DNA sequence. Proc Natl Acad Sci USA 1988;85:1412– 6. Sasaki E, Pai R, Halter F, et al. Induction of cyclooxygenase-2 in a rat gastric epithelial cell line by epiregulin and basic fibroblast growth factor. J Clin Gastroenterol 1998;27(suppl 1):S21–S27. Xie W, Chipman JG, Robertson DL, et al. Expression of a mitogen-responsive gene encoding prostaglandin synthase is regulated by mRNA splicing. Proc Natl Acad Sci USA 1991; 88:2692– 6. Kujubu DA, Fletcher BS, Varnum BC, et al. TIS10, a phorbol ester tumor promoter-inducible mRNA from Swiss 3T3 cells encodes a novel prostaglandin synthase/cyclooxygenase homologue. J Biol Chem 1991;266:12866 –72. Masferrer JL, Zweifel BS, Manning PT, et al. Selective inhibition of inducible cyclooxygenase 2 in vivo is anti-inflammatory and nonulcerogenic. Proc Natl Acad Sci USA 1994; 91:3228 –32. Seibert K, Zhang Y, Leahy K, et al. Pharmacological and biochemical demonstration of the role of cyclooxygenase 2 in inflammation and pain. Proc Natl Acad Sci USA 1994;91: 12013–7. Fu JY, Masferrer JL, Seibert K, et al. The induction and suppression of prostaglandin H2 synthase (cyclooxygenase) in human monocytes. J Biol Chem 1990;265:16737– 40. Lee SH, Soyoola E, Chanmugam P, et al. Selective expression of mitogen-inducible cyclooxygenase in macrophages stimulated with lipopolysaccharide. J Biol Chem 1992;267: 25934 – 8. Lin AH, Bienkowski MJ, Gorman RR. Regulation of prostaglandin H synthase mRNA levels and prostaglandin biosynthesis by platelet-derived growth factor. J Biol Chem 1989; 264:17379 – 83. Masferrer JL, Zweifel BS, Seibert K, et al. Selective regulation of cellular cyclooxygenase by dexamethasone and endotoxin in mice. J Clin Invest 1990;86:1375–9. Masferrer JL, Seibert K, Zweifel B, et al. Endogenous glucocorticoids regulate an inducible cyclooxygenase enzyme. Proc Natl Acad Sci USA 1992;89:3917–21. Kujubu DA, Herschman HR. Dexamethasone inhibits mitogen induction of the TIS10 prostaglandin synthase/cyclooxygenase gene. J Biol Chem 1992;267:7991– 4. De Witt DL, Meade EA, Smith WL. PGH synthase in enzyme selectivity; the potential for safer nonsteroidal anti-inflammatory drugs. Am J Med 1993;95(suppl 2A):40S– 44S. Penning TD, Talley JJ, Bertenshaw SR, et al. Synthesis and biological evaluation of the 1,5-diarylpyrazole class of cyclooxygenase-2 inhibitors: Identification of SC-56835 (celecoxib). J Med Chem 1997;40:1347– 65. Simon LS, Weaver AL, Graham DY, et al. Anti-inflammatory, and upper gastrointestinal effects of celecoxib in rheumatoid arthritis: A randomized controlled trial. J Am Med Assoc 1999;282:1921– 8. Bensen WG, Fiechtner JJ, McMillen JI, et al. Treatment of osteoarthritis with celecoxib, a cyclooxygenase-2 inhibitor: A randomized controlled trial. Mayo Clin Proc 1999;74:1095– 105. 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: 2106 –11.

Upper GI Safety of Celecoxib

1027

30. Steinbroker O, Traeger CH, Batterman RC. Therapeutic criteria in rheumatoid arthritis. JAMA 1949;140:659 – 62. 31. Ward J, Williams HJ, Egger MJ, et al. Comparison of auranofin, gold sodium thiomalate and placebo in the treatment of rheumatoid arthritis. Arthritis Rheum 1983;26:1303–15. 32. Cooperating Clinics Committee of American Rheumatism Association. A seven-day variability study of 499 patients with peripheral rheumatoid arthritis. Arthritis Rheum 1965;8:302– 34. 33. Garvey W, Fathi A, Bigelow F. Modified Steiner for the demonstration of Spirochetes. J Histotech 1985;8:15–7. 34. Lanza FL, Royer GL Jr, Nelson RS. Endoscopic evaluation of the effects of aspirin, buffered aspirin, and enteric-coated aspirin on gastric and duodenal mucosa. N Engl J Med 1980; 303:136 – 8. 35. Agrawal NM, Van Kerckhove HEJM, Erhardt LJ, et al. Misoprostol coadministered with diclofenac for prevention of gastroduodenal ulcers. A one-year study. Dig Dis Sci 1995; 40:1125–31. 36. Roth SH, Bennett RE, Caldron PH, et al. Endoscopic evaluation of the long term effects of diclofenac sodium and naproxen in elderly patients with arthritis. Clin Drug Invest 1995;9:171–9. 37. Roth SH, Bennett R, Caldron P, et al. Five-year comparative endoscopic evaluation of arthritis patients treated with nabumetone versus naproxen. Arthritis Rheum 1991;34(suppl): S157 (abstract). 38. Roth SH, Tindall EA, Jain AK, et al. A controlled study comparing the effects of nabumetone, ibuprofen and ibuprofen plus misoprostol on the upper gastrointestinal tract mucosa. Arch Intern Med 1993;153:2565–71. 39. Taha AS, Hudson N, Hawkey CJ, et al. Famotidine for the prevention of gastric and duodenal ulcers caused by nonsteroidal anti-inflammatory drugs. N Engl J Med 1996;334: 1435–9. 40. Cheatum DE, Arvanitakis C, Gumpel M, et al. An endoscopic study of gastroduodenal lesions induced by nonsteroidal antiinflammatory drugs. Clin Ther 1999;21:992–1003. 41. Langman MJS. Epidemiologic evidence on the association between peptic ulceration and anti-inflammatory drug use. Gastroenterology 1989;96:640 – 6. 42. MacDonald TM, Morant SV, Robinson GC, et al. Association of upper gastrointestinal toxicity of non-steroidal anti-inflammatory drugs with continued exposure: Cohort study. Br Med J 1997;315:1333–7. 43. Garcia´-Rodrı´guez LA, Cattaruzzi C, Troncon MG, et al. Risk of hospitalization for upper gastrointestinal tract bleeding associated with ketorolac, other nonsteroidal anti-inflammatory drugs, calcium antagonists and other hypertensive drugs. Arch Intern Med 1998;158:33–9. 44. Dobrilla G, Benvenuti S, de Guelmi A. The epidemiology of the gastroduodenal damage induced by aspirin and other nonsteroidal anti-inflammatory drugs. Recent Prog Med 1997;88: 202–11. 45. Griffin MR. Epidemiology of nonsteroidal anti-inflammatory drug-associated gastrointestinal injury. Am J Med 1998;104: 23S–29S. 46. Hansen JM, Hallas J, Lauritsen JM, et al. Non-steroidal antiinflammatory drugs, and ulcer complications: A risk factor analysis for clinical decision-making. Scand J Gastroenterol 1996;31:126 –30.