International Application Form ___________________________________________________________________________ ___________________________________________________________________________ ___________________________________________________________________________ Street ___________________________________________________________________________ Town ___________________________________________
Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne COMPARISON OF UPPER GASTROINTESTINAL TOXICITY OF ROFECOXIB
AND NAPROXEN IN PATIENTS WITH RHEUMATOID ARTHRITIS
CLAIRE BOMBARDIER, M.D., LOREN LAINE, M.D., ALISE REICIN, M.D., DEBORAH SHAPIRO, DR.P.H., RUBEN BURGOS-VARGAS, M.D., BARRY DAVIS, M.D., PH.D., RICHARD DAY, M.D., MARCOS BOSI FERRAZ, M.D., PH.D., CHRISTOPHER J. HAWKEY, M.D., MARC C. HOCHBERG, M.D., TORE K. KVIEN, M.D., AND THOMAS J. SCHNITZER, M.D., PH.D., FOR THE VIGOR STUDY GROUP ABSTRACT
tinal events occur in 2 to 4 percent of patients who ly used medications in the world.1 A major are taking nonselective nonsteroidal antiinflammatory factor limiting their use is gastrointesti- drugs (NSAIDs). We assessed whether rofecoxib, a nal toxicity. Although endoscopic studies reveal that selective inhibitor of cyclooxygenase-2, would be as- gastric or duodenal ulcers develop in 15 to 30 percent sociated with a lower incidence of clinically important of patients who regularly take NSAIDs,2 the chief con- upper gastrointestinal events than is the nonselective cern is clinically important gastrointestinal problems, NSAID naproxen among patients with rheumatoid such as bleeding. It has been estimated that more than 100,000 patients are hospitalized and 16,500 die each Methods
year in the United States as a result of NSAID-asso- were at least 50 years of age (or at least 40 years of age and receiving long-term glucocorticoid therapy) and who had rheumatoid arthritis to receive either Most NSAIDs inhibit both cyclooxygenase-1 and 50 mg of rofecoxib daily or 500 mg of naproxen twice cyclooxygenase-2, isoenzymes involved in the synthe- daily. The primary end point was confirmed clinical sis of prostaglandins.5 Cyclooxygenase-1 is constitu- upper gastrointestinal events (gastroduodenal perfo- tively expressed and generates prostanoids involved in ration or obstruction, upper gastrointestinal bleeding, the maintenance of the integrity of gastrointestinal and symptomatic gastroduodenal ulcers).
mucosa and platelet aggregation,6 whereas at sites of Results
Rofecoxib and naproxen had similar effica- inflammation, cyclooxygenase-2 is induced to generate cy against rheumatoid arthritis. During a median fol- prostaglandins that mediate inflammation and pain.7 low-up of 9.0 months, 2.1 confirmed gastrointestinal The antiinflammatory effects of nonselective NSAIDs events per 100 patient-years occurred with rofecoxib, (those that inhibit both cyclooxygenase-1 and cyclo- as compared with 4.5 per 100 patient-years with na-proxen (relative risk, 0.5; 95 percent confidence inter- oxygenase-2) therefore appear to be mediated through val, 0.3 to 0.6; P<0.001). The respective rates of com- the inhibition of cyclooxygenase-2,8 whereas their plicated confirmed events (perforation, obstruction, harmful effects in the gastrointestinal tract as well as and severe upper gastrointestinal bleeding) were 0.6 their antiplatelet effects are believed to occur primar- per 100 patient-years and 1.4 per 100 patient-years ily through the inhibition of cyclooxygenase-1.5 (relative risk, 0.4; 95 percent confidence interval, 0.2 Agents that selectively inhibit cyclooxygenase-2 have to 0.8; P=0.005). The incidence of myocardial infarc- antiinflammatory and analgesic effects that are simi- tion was lower among patients in the naproxen groupthan among those in the rofecoxib group (0.1 percentvs. 0.4 percent; relative risk, 0.2; 95 percent confidenceinterval, 0.1 to 0.7); the overall mortality rate and the From the Institute for Work and Health, Mount Sinai Hospital, and the rate of death from cardiovascular causes were simi- University Health Network, Toronto (C.B.); the Gastrointestinal Division, Department of Medicine, University of Southern California School of Conclusions
Medicine, Los Angeles (L.L.); Merck, Rahway, N.J. (A.R., D.S.); the Fac- ulty of Medicine and Research Division, Universidad Nacional Autonoma treatment with rofecoxib, a selective inhibitor of cy- de Mexico, and Hospital General de Mexico, Mexico City, Mexico (R.B.-V.); clooxygenase-2, is associated with significantly fewer University of Texas–Houston School of Public Health, Houston (B.D.); the clinically important upper gastrointestinal events than Department of Clinical Pharmacology, University of New South Walesand St. Vincent’s Hospital, Sydney, Australia (R.D.); the Division of Rheu- treatment with naproxen, a nonselective inhibitor.
matology, Department of Medicine, Escola Paulista de Medicina, Univer- sidade Federal de São Paulo, São Paulo, Brazil (M.B.F.); the Division of 2000, Massachusetts Medical Society.
Gastroenterology, School of Medical and Surgical Sciences, University Hos-pital, Nottingham, United Kingdom (C.J.H.); the Division of Rheumatol-ogy and Clinical Immunology, University of Maryland, Baltimore (M.C.H.);Oslo City Department of Rheumatology, and Diakonhjemmet Hospital, Oslo,Norway (T.K.K.); and the Office of Clinical Research and Training, North-western University School of Medicine, Chicago (T.J.S.). Address reprint re-quests to Dr. Bombardier at the Institute for Work and Health, 250 BloorSt. E., Suite 702, Toronto, ON M4W 1E6, Canada, or at email@example.com.
Arthur Weaver, M.D., Arthritis Center of Nebraska, Lincoln, was another Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. U P P E R GAST R O I N T E ST I N A L TOX I C I T Y O F R O F EC OX I B A N D N A P R OX E N I N PAT I E N T S W I T H R H E U M ATO I D A R T H R I T I S
lar to those of nonselective NSAIDs,9-12 but they in- platelet aggregation, which is mediated by cyclooxygenase-1, there duced significantly fewer ulcers in endoscopic tri- was a possibility that the incidence of thrombotic cardiovascular events would be lower among patients treated with nonselective 12-15 Whether such a decrease in the number of cyclooxygenase inhibitors than among those treated with cyclooxy- ulcers translates into a similar decrease in the number genase-2–selective inhibitors. Therefore, cardiovascular events were of clinical gastrointestinal events is a matter of con- also assessed for a future meta-analysis by independent committees troversy. We performed a prospective, randomized, whose members were unaware of the patients’ treatment assign- double-blind comparison of rofecoxib and naproxen ments. A separate analysis of these events, however, was not spec-ified in the study design.
in more than 8000 patients with rheumatoid arthritis.
Study End Points
Patients who had potential clinical upper gastrointestinal events Study Population
were evaluated and treated according to the standard practice of Patients with rheumatoid arthritis who were at least 50 years the physicians who were caring for them. Patients who stopped old (or at least 40 years old and receiving long-term glucocorticoid taking the study medication before the study ended were followed therapy) and who were expected to require NSAIDs for at least until the end of the study to determine whether an upper gastro- one year were eligible. Patients were excluded if they had a history intestinal event had occurred. Only events that were confirmed by of another type of inflammatory arthritis, upper gastrointestinal the end-point committee according to prespecified criteria (Table 1) surgery, or inflammatory bowel disease; an estimated creatinine and that occurred during treatment or within 14 days after the dis- clearance of 30 ml or less per minute; a positive test for fecal occult continuation of treatment were included in the primary analysis.
blood (this test was performed at base line in all patients); an un- In addition, the protocol called for the analysis of all episodes stable medical condition; a history of cancer or alcohol or drug of gastrointestinal bleeding, including confirmed and unconfirmed abuse in the five years before the study; a history of cerebrovas- episodes of upper gastrointestinal bleeding, and bleeding from a cular events in the two years before the study; or a history of my- site beyond the duodenum that resulted in hospitalization, dis- ocardial infarction or coronary bypass in the year before the study.
continuation of treatment, or a decrease in the hemoglobin level Patients with morbid obesity and those who required or who had been receiving treatment with aspirin, ticlopidine, anticoagulants,cyclosporine, misoprostol, sucralfate, or proton-pump inhibitors or Assessment of Efficacy
treatment with histamine H –receptor antagonists in prescription- strength doses were also excluded from the study. Patients enrolled For each patient both the investigator and the patient answered in the study were not thought to require the use of these agents a Global Assessment of Disease Activity question at base line (af- ter the discontinuation of prestudy NSAIDs), 6 weeks, 4 months,and 12 months and at the end of the study or when treatment was Study Design
discontinued. The score can range from 0 (“very well”) to 4 (“verypoor”), and higher scores indicate more disease activity. The Mod- The study was conducted at 301 centers in 22 countries. Three ified Health Assessment questionnaire was administered only to to 14 days after discontinuing NSAIDs, eligible patients were ran- patients enrolled at centers in the United States at base line, at six domly assigned to receive either 50 mg of rofecoxib (Vioxx, Merck, weeks, and at the end of the study or when treatment was discon- Whitehouse Station, N.J.) once daily or 500 mg of naproxen (No- tinued. This questionnaire evaluates the extent of functional dis- vopharm Biotech, Toronto) twice daily. The groups were stratified ability in eight types of tasks performed on a daily basis. The level according to the presence or absence of a history of gastroduode- of effort required to perform each task is assessed on a 4-point nal ulcer, upper gastrointestinal bleeding, and gastroduodenal per- scale on which a score of 0 indicates no difficulty in performing the foration. Blinding was achieved through the use of a matching pla- task and a score of 3 indicates an inability to perform the task.16 Higher scores indicate more severe disability.
Patients were permitted to take acetaminophen, non-NSAID analgesic medications, glucocorticoids, and disease-modifying drugs Statistical Analysis
(e.g., methotrexate) to control their rheumatoid arthritis. Patientswere also allowed to take antacids and H -receptor antagonists in The primary hypothesis was that the risk of confirmed upper the following maximal doses: ranitidine, 150 mg daily; famotidine, gastrointestinal events (gastroduodenal perforation or obstruction, 20 mg daily; cimetidine, 400 mg daily; and nizatidine, 150 mg upper gastrointestinal bleeding, and symptomatic gastroduodenal daily. Nonstudy NSAIDs were not allowed. After randomization, ulcers) would be lower among patients who were taking rofecoxib the patients returned to the clinic at six weeks and at four months than among those who were taking naproxen. Secondary hypoth- and every four months thereafter until the end of the study. Patients eses were that the risk of confirmed complicated events (perforation, were contacted by telephone at week 10 and every four months obstruction, and severe upper gastrointestinal bleeding) and the thereafter. Compliance was assessed by pill counts at clinic visits risk of both confirmed and unconfirmed upper gastrointestinal and by questioning of patients during the scheduled telephone calls.
events would be lower among patients who were taking rofecoxib.
Serum was obtained from all patients for Helicobacter pylori test- Cox proportional-hazards analysis was used to compare the ef- ing (HM-CAP, Enteric Products, Stonybrook, N.Y.). Investigators fect of treatment; the presence or absence of a history of gastro- were not informed of the results of these tests during the study.
intestinal events was a stratification factor in the analysis.17,18 The The institutional review board or ethics review committee at scores for the Global Assessment of Disease Activity question and each center approved the protocol, and all patients gave written in- Modified Health Assessment questionnaire were analyzed in terms formed consent. A steering committee oversaw the study design, of the mean change from base line during the treatment period.
conduct of the trial, analyses of data, and drafting of this report. This The primary population for analysis comprised all randomized pa- committee was composed of 14 members, 2 of whom were employ- tients. Subgroup analyses were conducted with use of Cox regres- ees of the sponsoring pharmaceutical company. An independent data sion analysis.17,18 Interactions between treatments and subgroups and safety monitoring board monitored the patients’ safety. An in- were assessed to determine whether the effect of rofecoxib as com- dependent, external (end-point) committee whose members were pared with that of naproxen was consistent in the subgroups. We unaware of the patients’ treatment assignments reviewed the data assessed data on general safety by evaluating 95 percent confidence to determine which patients had reached the study end points. Be- intervals of the differences in the proportions of the treatment cause highly selective cyclooxygenase-2 inhibitors do not inhibit groups with each adverse event.19 All statistical tests were two-sided.
Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne TABLE 1. CRITERIA FOR GASTROINTESTINAL EVENTS.
CRITERIA REQUIRED FOR CONFIRMATION OF EVENT
Evidence of perforation on endoscopy, at surgery, on radiography (evidence of free intraabdominal air or extravasation of contrast medium), or at autopsy Occurrence of nausea and vomiting »24 hours postprandially and evidence of narrowing of distal portion of stomach or duodenum as a result of a nonmalig-nant ulcer on endoscopy, at surgery, on radiography, or at autopsy Episode of hematemesis or aspiration of bloody gastric fluid witnessed by health care provider; episode of melena witnessed by health care provider; evidence of active bleeding on endoscopy, at surgery, or on angiography; positive test for fecal occult blood with documented upper gastrointestinal lesion judged to be the source and associated with either clinically significant bleeding or decrease in volume* or evidence of visible vessel, clot, or pigmented spot on ulcer at endoscopy; or episode of hematemesis or melena reported by patient with up-per gastrointestinal lesion judged to be the source and associated with either clinically significant bleeding or decrease in volume* or evidence of visible ves-sel, clot, or pigmented spot on ulcer at endoscopy Evidence of gastric or duodenal ulcer on endoscopy, at surgery, on contrast- enhanced radiography of the upper gastrointestinal tract, or at autopsy *A decrease in volume was defined by the finding of a decrease in hemoglobin of at least 2 g per deciliter; by the finding of an orthostatically induced change in pulse of more than 20, change in systolic blood pressure of more than20 mm Hg, or change in diastolic blood pressure of more than 10 mm Hg; by the finding of other evidence of a clinicallysignificant reduction in circulatory volume (e.g., clinically significant hypotension that is corrected by volume replace-ment); or by the need for blood transfusion.
Adverse Gastrointestinal Events
Characteristics of the Patients
Confirmed upper gastrointestinal events occurred Between January 1999 and July 1999, we screened in 177 patients. In 53 of these patients the event was 9539 patients and enrolled 8076; 4047 were ran- complicated. An additional 13 patients had events that domly assigned to receive rofecoxib, and 4029 to re- were reported by investigators but that were judged ceive naproxen. The major reasons for exclusion were by the end-point committee to be unconfirmed.
a contraindication to prolonged NSAID therapy (in The time to the development of a confirmed up- the case of 246 patients), a positive test for fecal oc- per gastrointestinal event is shown in Figure 1. The cult blood (203 patients), and a failure to meet in- rates per 100 patient-years and incidences of the pre- clusion criteria (355 patients). The median follow- specified clinical events are shown in Tables 4 and 5, up was 9.0 months in both treatment groups (range, respectively. The relative risk of confirmed upper 0.5 to 13). A total of 5742 patients (71.1 percent) gastrointestinal events for patients in the rofecoxib continued to take their assigned medication until group as compared with those in the naproxen group the end of the study. Rates of discontinuation were was 0.5 (95 percent confidence interval, 0.3 to 0.6; similar in the two groups: 29.3 percent in the rofe- P<0.001), whereas the relative risk of complicated coxib group (16.4 percent because of adverse events, confirmed upper gastrointestinal events was 0.4 (95 6.3 percent because of a lack of efficacy, and 6.6 per- percent confidence interval, 0.2 to 0.8; P=0.005).
cent for other reasons) and 28.5 percent in the naprox- The relative risk of complicated upper gastrointesti- en group (16.1 percent because of adverse events, nal bleeding for patients in the rofecoxib group as 6.5 percent because of a lack of efficacy, and 5.9 per- compared with those in the naproxen group was 0.4 cent for other reasons). Ninety-nine percent of the (95 percent confidence interval, 0.2 to 0.7; P= patients in both groups took their medication on at 0.004), whereas the relative risk of bleeding beyond least 75 percent of the study days. The base-line char- the duodenum was 0.5 (95 percent confidence in- acteristics were similar in the two groups (Table 2).
A per-protocol analysis of the 7925 patients with- Efficacy
out substantial protocol violations demonstrated rel- Rofecoxib and naproxen had similar efficacy against ative risks of confirmed upper gastrointestinal events rheumatoid arthritis (Table 3). In addition, the rates and complicated confirmed events of 0.4 (95 per- of discontinuation of treatment owing to a lack of cent confidence interval, 0.3 to 0.6; P<0.001) and efficacy were low in both groups (6.3 percent in the 0.4 (95 percent confidence interval, 0.2 to 0.7; P= rofecoxib group and 6.5 percent in the naproxen 0.003), respectively. The results of an intention-to- treat analysis of all confirmed upper gastrointestinal Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. U P P E R GAST R O I N T E ST I N A L TOX I C I T Y O F R O F EC OX I B A N D N A P R OX E N I N PAT I E N T S W I T H R H E U M ATO I D A R T H R I T I S
tients with glucocorticoid therapy at base line (rela- TABLE 2. BASE-LINE CHARACTERISTICS OF THE PATIENTS.*
tive risk, 0.4; 95 percent confidence interval, 0.2 to0.6). The relative risks in these subgroups and the oth- ROFECOXIB
er prespecified subgroups (defined according to sex, race or ethnic group, and location of study center) CHARACTERISTIC
were not significantly different, indicating that there was no significant interaction between the treatments Treatment with rofecoxib was associated with a significantly lower incidence of clinical gastrointestinal events regardless of the results of serologic tests for H. pylori. However, the relative risks of clinical events among H. pylori–negative patients and H. pylori–pos- itive patients were significantly different (P=0.04, data not shown). Finally, the relative risk of gastrointesti- nal events remained significantly lower (0.1; 95 per- cent confidence interval, 0.02 to 1.0) in the rofecoxib group than in the naproxen group even in a subgroup at very low risk (i.e., patients who were younger than Treatment for rheumatoid arthritis — no. (%) 65 years, who were negative for H. pylori, who had no history of a clinical gastrointestinal event, and who were not taking glucocorticoids at base line).
History of clinical gastrointestinal events The safety of both rofecoxib and naproxen was similar to that reported in previous studies.20,21 The mortality rate was 0.5 percent in the rofecoxib group and 0.4 percent in the naproxen group. The rate of death from cardiovascular causes was 0.2 percent in both groups. Ischemic cerebrovascular events occurred in 0.2 percent of the patients in each group. Myo- cardial infarctions were less common in the naproxen *Plus–minus values are means ±SD. NSAIDs denotes nonselective non- group than in the rofecoxib group (0.1 percent vs.
0.4 percent; 95 percent confidence interval for the †A low dose was defined as a maximal daily dose of 150 mg of ranitidine, difference, 0.1 to 0.6 percent; relative risk, 0.2; 95 20 mg of famotidine, 400 mg of cimetidine, and 150 mg of nizatidine. percent confidence interval, 0.1 to 0.7). Four percent ‡Scores can range from 0 (“very well”) to 4 (“very poor”). Higher scores indicate more disease activity.
of the study subjects met the criteria of the Food and §According to the American College of Rheumatology’s system of clas- Drug Administration (FDA) for the use of aspirin for sification, functional class I indicates complete ability to perform usual ac- secondary cardiovascular prophylaxis (presence of a tivities of daily living, and class IV indicates limited ability to perform usualactivities of daily living.
history of myocardial infarction, angina, cerebrovas-cular accident, transient ischemic attack, angioplasty,or coronary bypass)22 but were not taking low-doseaspirin therapy. These patients accounted for 38 per-cent of the patients in the study who had myocardial events throughout the study, including those that oc- infarctions. In the other patients the difference in the curred at any time after the discontinuation of treat- rate of myocardial infarction between groups was not ment, were similar and remained statistically signifi- significant (0.2 percent in the rofecoxib group and 0.1 percent in the naproxen group). When the data Subgroup analyses showed the following relative showing a reduction in the rate of myocardial infarc- risks of clinical gastrointestinal events among the tion in the naproxen group became available after the patients in the rofecoxib group as compared with completion of this trial, Merck, the manufacturer of those in the naproxen group: patients with no prior rofecoxib, notified all investigators in ongoing studies gastrointestinal events (relative risk, 0.5; 95 percent of a change in the exclusion criteria to allow patients confidence interval, 0.3 to 0.7), patients with prior to use low-dose aspirin. There was no association be- gastrointestinal events (relative risk, 0.4; 95 percent tween hypertension and myocardial infarction; only confidence interval, 0.2 to 0.8), patients with no a single patient (in the rofecoxib group) had both glucocorticoid therapy at base line (relative risk, 0.7; hypertension and a myocardial infarction as adverse 95 percent confidence interval, 0.4 to 1.2), and pa- Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne TABLE 3. EFFECTIVENESS OF ROFECOXIB AND NAPROXEN FOR RHEUMATOID ARTHRITIS.*
CHANGE IN SCORE DURING TREATMENT
*Plus–minus values are means ±SD.
†The values were calculated by analysis of variance in a model that included treatment assignment and presence or absence of a history of gastrointestinal events and the base-line value as covariates. CI denotes confidence interval.
‡Scores can range from 0 (“very well”) to 4 (“very poor”). Higher scores indicate more disease activity.
§Scores can range from 0 (no difficulty in performing a task) to 3 (unable to perform the task). Higher scores indicate more severe disability. The questionnaire was administered only to patients enrolled at centers in the United States (1735in the rofecoxib group and 1732 in the naproxen group).
Figure 1. Cumulative Incidence of the Primary End Point of a Confirmed Upper Gastrointestinal Event
among All Randomized Patients.
The most common adverse events leading to dis- en group (3.5 percent vs. 4.9 percent). The rates of continuation of treatment, excluding the gastroin- discontinuation for any gastrointestinal events, includ- testinal end points, were dyspepsia, abdominal pain, ing gastrointestinal end points, were also significant- epigastric discomfort, nausea, and heartburn. In the ly lower in the rofecoxib group than in the naproxen rofecoxib group, significantly fewer patients discon- group (7.8 percent vs. 10.6 percent). The incidence of tinued treatment as a result of any one of these five adverse effects related to renal function was low and upper gastrointestinal symptoms than in the naprox- was similar in the two groups (1.2 percent in the ro- Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. U P P E R GAST R O I N T E ST I N A L TOX I C I T Y O F R O F EC OX I B A N D N A P R OX E N I N PAT I E N T S W I T H R H E U M ATO I D A R T H R I T I S
TABLE 4. INCIDENCE OF GASTROINTESTINAL EVENTS IN THE TREATMENT GROUPS.
TYPE OF EVENT
All episodes of gastrointestinal bleeding †The analysis includes 13 events that were reported by investigators but were considered to be unconfirmed by the ‡The analysis includes six events that were reported by investigators but that were considered to be unconfirmed by TABLE 5. INCIDENCE OF CONFIRMED UPPER GASTROINTESTINAL EVENTS.*
ALL COMPLICATED CONFIRMED
TYPE OF UPPER
ALL CONFIRMED UPPER
*Patients may have been included in more than one column, but each is counted only once in the †Perforations and obstructions are complicated events by definition.
‡Two confirmed upper gastrointestinal events occurred after only one dose of rofecoxib and most likely resulted from prior use of nonselective nonsteroidal antiinflammatory drugs. §The cause or source of bleeding was gastric ulcers in five patients, duodenal ulcers in five, and other upper gastrointestinal source in three. One patient in the rofecoxib group had both a gastricand a duodenal ulcer.
¶The cause or source of bleeding was gastric ulcers in 16 patients, duodenal ulcers in 9, and other upper gastrointestinal sources in 7.
fecoxib group and 0.9 percent in the naproxen group); nificantly lower rates of clinically important upper gas- only 0.2 percent of patients in each group discon- trointestinal events and complicated upper gastrointes- tinued treatment because of these adverse effects.
tinal events than did treatment with a standard dose(1000 mg per day) of naproxen. We also found that DISCUSSION
the incidence of complicated upper gastrointestinal We found that, in patients with rheumatoid arthri- bleeding and bleeding from beyond the duodenum tis, treatment with rofecoxib at twice the maximal dose was significantly lower among patients who received approved by the FDA for long-term use resulted in sig- rofecoxib. Only 41 patients would need to be treated Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne with rofecoxib rather than naproxen to avert one clin- Whether ulcers identified by endoscopic examina- ical upper gastrointestinal event in a one-year period.
tion are markers of a clinical gastrointestinal event Although the optimal dose of rofecoxib for the treat- has been a matter of controversy. The relative reduc- ment of rheumatoid arthritis has yet to be determined, tion in the risk of such ulcers in two identical studies data from a prior study indicate that maximal efficacy that compared 50 mg of rofecoxib daily with 800 mg of ibuprofen three times a day was 71 percent at six A prospective study that compared NSAIDs alone months.13,14 Thus, our findings support the concept with NSAIDs plus misoprostol reported that 0.95 that the results of endoscopic studies of ulcers can percent of patients with rheumatoid arthritis who be extrapolated to clinical gastrointestinal events.
were taking an NSAID alone had upper gastrointes- In prior endoscopic studies, the frequency of ulcers tinal complications over a period of six months,24 with was similar in patients taking rofecoxib and those a relative reduction in the risk of such complications taking placebo.13,14 We could not include a placebo with combination treatment of 40 percent during group, and no studies have yet assessed whether a this period. These results are similar to our finding cyclooxygenase-2–selective inhibitor or the combi- of a cumulative incidence of serious upper gastroin- nation of nonselective NSAIDs plus gastroprotective testinal events over a six-month period of 0.75 per- drugs (such as misoprostol and proton-pump inhib- cent in the naproxen group and a relative reduction itors) will achieve similar results.
in risk of 67 percent in the rofecoxib group (data not Gastrointestinal symptoms are extremely common shown). A 50 percent reduction in the incidence of with NSAID therapy, as demonstrated by the fact clinically important upper gastrointestinal events with that, in our study, the five most common adverse rofecoxib as compared with a nonselective NSAID events leading to the discontinuation of treatment was also found in a prespecified combined analysis were upper gastrointestinal symptoms. Although gas- of eight double-blind studies that included 4921 pa- trointestinal symptoms are poorly correlated with tients with osteoarthritis, none of whom received glu- endoscopic findings of ulcers or clinical gastrointes- cocorticoids.25 Patients with rheumatoid arthritis have tinal events,27 significantly fewer patients in the ro- a higher risk of upper gastrointestinal events than do fecoxib group than in the naproxen group discontin- patients with osteoarthritis.4 Thus, the results of our ued treatment because of gastrointestinal symptoms.
study extend the results of the combined analysis to The overall mortality rate was similar in the two a group of patients at higher risk of bleeding.
groups, as were the rates of death from gastrointes- The results of a randomized, double-blind com- tinal events and from cardiovascular causes. The rate parison of the cyclooxygenase-2–selective inhibitor of myocardial infarction was significantly lower in the celecoxib and the nonselective NSAIDs ibuprofen and naproxen group than in the rofecoxib group (0.1 diclofenac were recently published.26 In this trial of percent vs. 0.4 percent). This difference was primarily 7968 patients, 73 percent of whom had osteoarthri- accounted for by the high rate of myocardial infarction tis and 27 percent of whom had rheumatoid arthri- among the 4 percent of the study population with tis, data were reported from the first 6 months of a the highest risk of a myocardial infarction, for whom study period that extended for up to 13 months.
low-dose aspirin is indicated.22 The difference in the Treatment with celecoxib was associated with a non- rates of myocardial infarction between the rofecoxib significant (P=0.09) trend toward a decrease in the and naproxen groups was not significant among the incidence of the primary end point (complicated ul- patients without indications for aspirin therapy as cers and erosions) and a significant decrease (P=0.02) in the incidence of the secondary end point (compli- Naproxen inhibits the production of thrombox- ane by 95 percent and inhibits platelet aggregation The incidence of clinically important gastrointes- by 88 percent, and this effect is maintained through- tinal events was lower in the rofecoxib group than in out the dosing interval28; therefore, the effects of reg- the naproxen group in all subgroups we examined.
ular use of naproxen may be similar to those of as- Concomitant glucocorticoid and NSAID therapy has pirin. Flurbiprofen, another NSAID that is a potent been reported to be associated with a higher risk of a inhibitor of platelet-derived thromboxane, led to a 70 clinical gastrointestinal event than is NSAID therapy percent reduction in the rate of reinfarction as com- alone.4 Therefore, a larger reduction in the incidence pared with placebo among patients in whom acute of events might have been expected in the subgroup myocardial infarction was successfully treated with that received both an NSAID and glucocorticoids.
There was a greater reduction in the relative risk of Analyses of 7535 patients in double-blind trials events in the subgroup of patients who were taking comparing rofecoxib with placebo and other NSAIDs glucocorticoids at base line than in the subgroup of (diclofenac, ibuprofen, and nabumetone) that do not patients who were not taking glucocorticoids at base produce sustained, maximal inhibition of platelet ag- line, but the difference between the groups was not gregation revealed similar rates of myocardial infarc- tion in all groups30 (and unpublished data). Thus, our Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. U P P E R GAST R O I N T E ST I N A L TOX I C I T Y O F R O F EC OX I B A N D N A P R OX E N I N PAT I E N T S W I T H R H E U M ATO I D A R T H R I T I S
results are consistent with the theory that naproxen We are indebted to Christopher T. Brett, Dayna Carroll, Russell has a coronary protective effect and highlight the Cender, Edward Chafart, Laurine Connors, Gary Gartenberg, Nicole fact that rofecoxib does not provide this type of pro- Gillies, Richard Holmes, Paige Reagan, Douglas J. Watson, DeboraP.L. Weiss, and Qinfen Yu for their substantial contributions. tection owing to its selective inhibition of cyclooxy-genase-2 at its therapeutic dose and at higher doses.
The finding that naproxen therapy was associated The following persons participated in the study: International Steering
with a lower rate of myocardial infarction needs fur- Committee: C. Bombardier (chair), L. Laine (cochair), A. Reicin,
ther confirmation in larger studies.
M. Hochberg, R. Day, T. Capizzi, P. Brooks, R. Burgos-Vargas, B. Davis,
M. Ferraz, C. Hawkey, T.K. Kvien, T. Schnitzer, A. Weaver; Data Safety
In summary, the use of the cyclooxygenase-2– Board: M. Weinblatt (chair); D. Bjorkman, J. Neaton, A. Silman, R. Stur-
selective inhibitor rofecoxib resulted in significantly rock; End-Point Adjudication Committee: M. Griffin (chair), D. Jensen,
fewer clinically important upper gastrointestinal events M. Langman; Investigators: Argentina: E. DiGiorgio, H. Laborde, O.
Messina, A. Strusberg; Australia: J. Bertouch, R. Day, G. McColl, P. Ryan,
than did treatment with naproxen, a nonselective S. Sharma; Brazil: R. Bonfiglioli, C. Borges, J. Brenol, N. da Silva, M. Fer-
NSAID. The two drugs had similar rates of clinical raz, F. Lima, C. Moreira, G. Novaes, A. Pessoa, F. Petean, S. Radominski,
A. Samara, B. Souza; Canada: T. Anastassiades, M. Atkinson, A. Beaulieu,
M. Bell, M. Camerlain, J. Canvin, M. Cohen, J. Hanly, J. Karsh, T. Mc-
Carthy, W. Olszynski, P. Patel, Y. Pesant, W. Pruzanski, K. Siminovitch, J.
Thome, V. Verdejo-Aguilar; Chile: L. Barria, C. Fuentealba, L. Massardo,
P. Riedemann, L. Roca, H. Rossi; China (Hong Kong): C. Lau; Colom-
Dr. Bombardier has received clinical research support from Aventis Phar- bia: M. Abello, P. Chalem, M. Jannault, J. Molina; Costa Rica: R. Alpizar,
ma, Merck Research Laboratories, and Merck Frosst Canada. She has H. Garcia-Sancho; Czech Republic: L. Konecna, K. Pavelka, M. Sealac-
served as a consultant to Knoll Pharmaceutical, Aventis Canada, Schering kova, J. Vitova, R. Ahora; Ecuador: R. Villacis; Finland: M. Hakala, P.
Canada, Merck Research Laboratories, and Wyeth–Ayerst.
Hannonen, T. Helve, M. Nissila; Germany: R. Alten, M. Gaubitz, E.
Dr. Laine has received clinical research support from Merck, Wyeth– Gromnica-Ihle, H. Hantzschel, G. Hein, M. Keysser, R. Kurthen, B. Lang, Ayerst, and AstraZeneca. He has served as a consultant to Merck, Searle, F. Mielke, U. Muller-Ladner, C. Richter, M. Schattenkirchner, M.
Johnson & Johnson, AstraZeneca, Wyeth–Ayerst, and Tap Pharmaceuti- Schneider, H. Sorenson, E. Waldorf-Bolten, H. Watnatz, S. Wassenberg; Guatemala: E. Julian, R. Palomo; Israel: D. Buskila, A. Nahir, I. Rosner;
Dr. Burgos-Vargas has received clinical research support from Merck Malaysia: S. Yeap; Mexico: R. Burgos, I. Garcia de la Torre, F. Irazoque,
Sharp & Dohme, Pfizer, and Boehringer Ingelheim. He has served as a J. Arozco Alcala; New Zealand: P. Jones, L. McLean, C. Rajapakse; Nor-
consultant to Merck and has been a member of a speakers’ bureau spon- way: H. Gulseth, K. Helgetveit, A. Johannessen, C. Kaufmann, O. Knud-
srod, T. Kvien, K. Mikkelsen, B. Rossebo, G. Sidenwall; Peru: A. Calvo
Dr. Davis has served as a consultant to Mirvant, Merck Research Labo- Quioz, M. Gustavo, A. Hilgado, L. Portocarrero, E. Vera Bejar; Poland:
ratories, Parke-Davis, and SmithKline Beecham.
J. Badurski, J. Brzezicki, M. Bykowska, I. Fiedorowicz-Fabrycy, J. Gaweda, Dr. Day has received clinical research support from Merck, Searle, Pfizer, P. Gluszki, B. Konieczna, S. Mackiewicz, J. Pazdur, L. Szczepanski, J.
Roche, and Amgen. He has served as a consultant to Merck, Searle, and Szechinski; South Africa: I. Anderson, D. Gotlieb, A. Jacovides, A.
Pfizer. He has been a member of speakers’ bureaus sponsored by Merck Lubbe, G. Mody, M. Tikly; United States: Alabama: G. Divititorio, S.
Sharp & Dohme, Searle, Pfizer, and SmithKline Beecham.
Fallahi, D. McLaine, W. Shergy, M.S. Touger, G. W illiams; Arizona: B.
Dr. Ferraz has received clinical research support from Bristol-Myers Harris, P. Howard, D. Michel, L. Taber, J. Tesser, M. Maricic; California:
Squibb, Merck Sharp & Dohme, and Aventis Pharma. He has served as a E. Boiling, M. Brandon, G. Dolan, R. Dore, M. Greenwald, D. Hasel- wood, M. Keller, K. Kolba, C. Multz, R. Reid, D. Stanton, C. Weidmann, Dr. Hawkey has received clinical research support from AstraZeneca, S. Weiner, J. Higashida; Colorado: S. Kassan, R. Lapidus, D. Scott, J.
Asta Medica, Boehringer Ingelheim, Boots Healthcare International, Cell Thompson, C. Van Kerchov; Connecticut: E. Feinglass, J. Green, R.
Tech, Eisai, Elan, Merck Research Laboratories, NicOx, and Novartis. He Schoen; Florida: M. Beilian, C. Chappel, J. Forstot, P. Freeman, N. Gaylis,
has served as a consultant to AstraZeneca, Abbott, Cell Tech, Eisai, Merck B. Germain, M. Guttierrez, M. Heier, T. Johnson, M. Kaufman, R. Levin, Research Laboratories, NicOx, Novartis, Parke-Davis, and Synthelabo M. Lowenstein, A. Marcadis, M. Mass, S. Matthews, H. McIlwain, J. Mil- Pharmacie. He has been a member of speakers’ bureaus sponsored by As- ler, M. Nunez, H. Offenberg, J. Popp, A. Reddy, W. Riskin, A. Rosen, Y.
traZeneca, Boehringer Ingelheim, Boots Healthcare International, Takeda, Sherrer, K. Stark, A. Torres; Georgia: J. Lieberman; Iowa: M. Brooks;
Wyeth Lederle, and Merck Research Laboratories.
Idaho: F. Dega, C. Scoville, C. Wiesenhutter; Illinois: F. Dietz, I. Fenton,
Dr. Hochberg has received clinical research support from Merck and M. Pick, R. Trapp, J. Zuzga, M. Ellman, G. Liang; Indiana: C. Arsever,
Quintiles (Aventis Pharma). He has served as a consultant to Aventis Phar- R. Khairi, M. Stack, R. Fife; Kansas: N. Becker, P. Ginder, R. Lies; Lou-
ma, Biomatrix, Merck, Negma Laboratories, Procter & Gamble, Roche, isiana: W. Eversmeyer, P. Sedrish, L. Serebro; Maine: L. Anderson, S.
and Wyeth–Ayerst. He owns stock in Johnson & Johnson, Eli Lilly, Merck, Block; Maryland: R. Marcus, D. McGinnis, N. Wei, T. Zizic, M. Hoch-
Procter & Gamble, and Schering-Plough.
berg; Massachusetts: C. Birbara, A. Dahdul, S. Helfgott, J. Hosey, S. Mil-
Dr. Kvien has received clinical research support from Merck, Searle, ler, R. Rapoport; Michigan: P. Coleman, R. Roschmann, H. Uhl, J.
Aventis Pharma, and Schering-Plough. He has served as a consultant to Taborn; Missouri: T. Weiss; Montana: S. English; Nebraska: W. Palmer,
Merck, Searle, Aventis Pharma, and Schering-Plough. He has been a mem- A. Weaver; Nevada: S. Miller; New Hampshire: B. Samuels, J. Trice; New
ber of a speakers’ bureau sponsored by Merck and Aventis Pharma.
Jersey: S. Golombek, H. Hassman, R. Hymowitz, C. Knee, D. Macpeek,
Dr. Schnitzer has received clinical research support from Abbott, Boeh- R. Marcus, G. Rihacek, J. Rohlf, D. Worth; New Mexico: K. Bordenave;
ringer Ingelheim, Johnson & Johnson, McNeil Consumer Products, New York: P. Chatpar, M. Farber, A. Kaell, A. Porges, P. Riccardi, C.
Merck, Novartis, Ortho-McNeil, Parke-Davis, Searle, and Wyeth–Ayerst.
Ritchlin; North Carolina: G. Arnold, W. Chmelewski, G. Collins, W.
He has served as a consultant to Boehringer Ingelheim, Merck, Novartis, Harperm, R. Harrell, T. Littlejohn, G. Schimizzi, R. Senter, A. Kashif; Ortho-McNeil, Searle, and SmithKline Beecham. He has been a member Ohio: T. Isakov, B. Long, D. Mandel, B. Rothschild, D. Schumacher, R.
of speakers’ bureaus sponsored by Boehringer Ingelheim, Merck, Ortho- Sievers, S. Wolfe, K. Hackshaw, M. Hooper, R. Rothenberg; Oklahoma:
C. Codding, R. Hynd, J. McKay, L. Jacobs; Pennsylvania: R. Kimelheim,
Dr. Weaver has received clinical research support from Merck, Searle, Im- A. Kivitz, W. Makarowski, G. McLaughlin, J. McMillen, R. Reese, J. Weis- munex, Wyeth–Ayerst, Aventis Pharma, Pharmacia–Upjohn, Eli Lilly, berg; Rhode Island: J. Scott; South Carolina: C. Barfiels, M. Brabham,
Connetics, Parke-Davis, Procter & Gamble, Pfizer, Hoffmann–LaRoche, K. Flint; Tennessee: C. Arkin, R. Gupta, R. Krause, H. Marker, T. Namey,
Centocor, Amgen, Cyprus Bioscience, Helsinn, Novartis, and Boehringer L. Robinson, P. Wheeler, B. Tanner; Texas: A. Brodsky, F. Burch, W.
Ingelheim. He has served as a consultant to Merck, Searle, Immunex, Wy- Chase, A. Chubick, D. Halter, J. Rutstein; Utah: J. Booth; Virginia: C.
eth–Ayerst, Aventis Pharma, Pharmacia–Upjohn, Eli Lilly, Connetics, Fisher, A. Goldstein; Washington: W. Eider, R. Ettlinger, R. Levy; Wis-
Parke-Davis, Procter & Gamble, Pfizer, Hoffmann–LaRoche, Centocor, consin: M. Pearson, G. Fiocco.
Amgen, Cyprus Bioscience, Helsinn, Novartis, and Boehringer Ingelheim.
He has been a member of speakers’ bureaus sponsored by Merck, Searle,Immunex, Wyeth–Ayerst, Aventis Pharma, Pharmacia–Upjohn, Eli Lilly, REFERENCES
Connetics, Parke-Davis, Procter & Gamble, Pfizer, Hoffmann–LaRoche,Centocor, Amgen, Cyprus Bioscience, Helsinn, Novartis, and Boehringer 1. Misoprostol for co-prescription with NSAIDs. Drug Ther Bull 1990;
Ingelheim. He is on the board of directors of MGI Pharma.
Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne 2. Laine L. Nonsteroidal anti-inflammatory drug gastropathy. Gastrointest
15. Emery P, Zeidler H, Kvien KT, et al. Celecoxib versus diclofenac in
Endosc Clin North Am 1996;6:489-504.
long-term management of rheumatoid arthritis: randomised double-blind 3. Wolfe MM, Lichtenstein DR , Singh G. Gastrointestinal toxicity of non-
comparison. Lancet 1999;354:2106-11.
steroidal antiinflammatory drugs. N Engl J Med 1999;340:1888-99. [Er- 16. Pincus T, Summey JA, Soraci SA Jr, Wallston KA, Hummon NP. Assess-
ment of patient satisfaction in activities of daily living using a modified Stan- 4. Singh G, Rosen Ramey D. NSAID induced gastrointestinal complica-
ford Health Assessment Questionnaire. Arthritis Rheum 1983;26:1346-53.
tions: the ARAMIS perspective — 1997. J Rheumatol Suppl 1998;51:8-16.
17. Cox DR. Regression models and life-tables. J Stat Soc [B] 1972;34:
5. Cryer B. Nonsteroidal anti-inflammatory drugs and gastrointestinal dis-
ease. In: Feldman M, Scharschmidt BF, Sleisenger MH, eds. Sleisenger and 18. Idem. Partial likelihood. Biometrika 1975;62:269-76.
Fordtran’s gastrointestinal and liver disease: pathophysiology/diagnosis/ 19. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New
management. 6th ed. Vol. 1. Philadelphia: W.B. Saunders, 1998:343-57.
6. Meade EA, Smith WL, DeWitt DL. Differential inhibition of prosta-
20. Vioxx (rofecoxib tablets and oral suspension). Whitehouse Station,
glandin endoperoxide synthase (cyclooxygenase) isoenzymes by aspirin and other non-steroidal anti-inflammatory drugs. J Biol Chem 1993;268:6610- 21. Naproxen. Nutley, N.J.: Roche Laboratories, 1999 (package insert).
22. Genuine Bayer aspirin. Morristown, N.J.: Bayer Consumer Care Divi-
7. Fosslein E. Adverse effects of nonsteroidal anti-inflammatory drugs on
the gastrointestinal system. Ann Clin Lab Sci 1998;28:67-81.
23. Schnitzer TJ, Truitt K, Fleischmann R , et al. The safety profile, toler-
8. Mitchell JA, Akarasereenont P, Thiemermann C, Flower RJ, Vane JR.
ability, and effective dose range of rofecoxib in the treatment of rheuma- Selectivity of nonsteroidal antiinflammatory drugs as inhibitors of consti- toid arthritis. Clin Ther 1999;21:1688-702.
tutive and inducible cyclooxygenase. Proc Natl Acad Sci U S A 1993;90: 24. Silverstein FE, Graham DY, Senior JR , et al. Misoprostol reduces se-
rious gastrointestinal complications in patients with rheumatoid arthritis 9. Day R , Morrison B, Luza A, et al. A randomized trial of the efficacy
receiving nonsteroidal anti-inflammatory drugs: a randomized, double- and tolerability of the COX-2 inhibitor rofecoxib vs ibuprofen in patients blind, placebo-controlled trial. Ann Intern Med 1995;123:241-9.
with osteoarthritis. Arch Intern Med 2000;160:1781-7.
25. Langman MJ, Jensen DM, Watson DJ, et al. Adverse upper gastrointes-
10. Cannon GW, Caldwell JR , Holt P, et al. Rofecoxib, a specific inhibitor
tinal effects of rofecoxib compared with NSAIDs. JAMA 1999;282:1929-33.
of cyclooxygenase 2, with clinical efficacy comparable with that of diclofen- 26. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity
ac sodium: results of a one-year, randomized, clinical trial in patients with with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis osteoarthritis of the knee and hip. Arthritis Rheum 2000;43:978-87.
and rheumatoid arthritis: the CLASS Study: a randomized controlled trial. 11. Ehrich EW, Schnitzer TJ, McIlwain H, et al. Effect of specific COX-2
inhibition in osteoarthritis of the knee: a 6 week double blind, placebo 27. Singh G, Ramey DR , Morfeld D, Shi H, Hatoum HT, Fries JF. Gas-
controlled pilot study of rofecoxib. J Rheumatol 1999;26:2438-47.
trointestinal tract complications of nonsteroidal anti-inflammatory drug 12. Simon LS, Weaver AL, Graham DY, et al. Anti-inflammatory and up-
treatment in rheumatoid arthritis: a prospective observational cohort study. per gastrointestinal effects of celecoxib in rheumatoid arthritis: a random- ized controlled trial. JAMA 1999;282:1921-8.
28. VanHecken A, Schwartz JI, Depré M, et al. Comparative inhibitory
13. Laine L, Harper S, Simon T, et al. A randomized trial comparing the
activity of rofecoxib, meloxicam, diclofenac, ibuprofen and naproxen on effect of rofecoxib, a cyclooxygenase 2-specific inhibitor, with that of ibu- COX-2 versus COX-1 in healthy volunteers. J Clin Pharmacol 2000;40: profen on the gastroduodenal mucosa of patients with osteoarthritis. Gas- 29. Brochier ML. Evaluation of flurbiprofen for prevention of reinfarction
14. Hawkey C, Laine L, Simon T, et al. Comparison of the effect of rofe-
and reocclusion after successful thrombolysis or angioplasty in acute myo- coxib (a cyclooxygenase 2 inhibitor), ibuprofen, and placebo on the gas- cardial infarction: the Flurbiprofen French Trial. Eur Heart J 1993;14:951-7.
troduodenal mucosa of patients with osteoarthritis: a randomized, double- 30. Daniels B, Seidenberg B. Cardiovascular safety profile of rofecoxib in
blind, placebo-controlled trial. Arthritis Rheum 2000;43:370-7.
controlled clinical trials. Arthritis Rheum 1999;42:Suppl:S143. abstract.
Downloaded from www.nejm.org on August 28, 2010. For personal use only. No other uses without permission. Copyright 2000 Massachusetts Medical Society. All rights reserved.
Characteristics of Recipients of Free Prescription Drug Samples: A Nationally Representative Analysis| Sarah L. Cutrona, MD, MPH, Steffie Woolhandler, MD, MPH, Karen E. Lasser, MD, MPH, David H. Bor, MD, Danny McCormick, MD, MPH, and DavidFree prescription drug samples are used widely Objectives. Free prescription drug samples are used widely in the United States . in the United States. The