Sie sind auf Seite 1von 9

ORIGINAL CONTRIBUTION JAMA-EXPRESS

Gastrointestinal Toxicity With Celecoxib vs


Nonsteroidal Anti-inflammatory Drugs
for Osteoarthritis and Rheumatoid Arthritis
The CLASS Study: A Randomized Controlled Trial
Fred E. Silverstein, MD Context Conventional nonsteroidal anti-inflammatory drugs (NSAIDs) are associ-
Gerald Faich, MD ated with a spectrum of toxic effects, notably gastrointestinal (GI) effects, because of
inhibition of cyclooxygenase (COX)-1. Whether COX-2–specific inhibitors are asso-
Jay L. Goldstein, MD ciated with fewer clinical GI toxic effects is unknown.
Lee S. Simon, MD Objective To determine whether celecoxib, a COX-2–specific inhibitor, is associ-
Theodore Pincus, MD ated with a lower incidence of significant upper GI toxic effects and other adverse ef-
fects compared with conventional NSAIDs.
Andrew Whelton, MD
Design The Celecoxib Long-term Arthritis Safety Study (CLASS), a double-blind, ran-
Robert Makuch, PhD domized controlled trial conducted from September 1998 to March 2000.
Glenn Eisen, MD Setting Three hundred eighty-six clinical sites in the United States and Canada.
Naurang M. Agrawal, MD Participants A total of 8059 patients (Š18 years old) with osteoarthritis (OA) or
William F. Stenson, MD rheumatoid arthritis (RA) were enrolled in the study, and 7968 received at least 1 dose
of study drug. A total of 4573 patients (57%) received treatment for 6 months.
Aimee M. Burr, MS
Interventions Patients were randomly assigned to receive celecoxib, 400 mg twice
William W. Zhao, PhD per day (2 and 4 times the maximum RA and OA dosages, respectively; n= 3987);
Jeffrey D. Kent, MD ibuprofen, 800 mg 3 times per day (n= 1985); or diclofenac, 75 mg twice per day
(n= 1996). Aspirin use for cardiovascular prophylaxis (Š325 mg/d) was permitted.
James B. Lefkowith, MD
Main Outcome Measures Incidence of prospectively defined symptomatic upper
Kenneth M. Verburg, PhD GI ulcers and ulcer complications (bleeding, perforation, and obstruction) and other
G. Steven Geis, PhD, MD adverse effects during the 6-month treatment period.

F
Results For all patients, the annualized incidence rates of upper GI ulcer complications
OR PATIENTS WITH MUSCULO - alone and combined with symptomatic ulcers for celecoxib vs NSAIDs were 0.76% vs
skeletal disorders, conven- 1.45% (P=.09) and 2.08% vs 3.54% (P=.02), respectively. For patients not taking as-
tional nonste r oi dal anti - pirin, the annualized incidence rates of upper GI ulcer complications alone and combined
inflammatory drugs (NSAIDs) with symptomatic ulcers for celecoxib vs NSAIDs were 0.44% vs 1.27% (P=.04) and
are a mainstay of clinical care.1-3 Well- 1.40% vs 2.91% (P=.02). For patients taking aspirin, the annualized incidence rates of
established limitations of NSAID upper GI ulcer complications alone and combined with symptomatic ulcers for celecoxib
therapy, however, include the risk of vs NSAIDs were 2.01% vs 2.12% (P=.92) and 4.70% vs 6.00% (P=.49). Fewer celecoxib-
treated patients than NSAID-treated patients experienced chronic GI blood loss, GI in-
developing significant injury to the up-
tolerance, hepatotoxicity, or renal toxicity. No difference was noted in the incidence of
per gastrointestinal (GI) tract.4-10 The cardiovascular events between celecoxib and NSAIDs, irrespective of aspirin use.
annualized incidence rate of symptom-
Conclusions In this study, celecoxib, at dosages greater than those indicated clini-
atic GI ulcers and ulcer complications cally, was associated with a lower incidence of symptomatic ulcers and ulcer compli-
in NSAID users ranges from 2% to 4% cations combined, as well as other clinically important toxic effects, compared with
(1%-2% for ulcer complications NSAIDs at standard dosages. The decrease in upper GI toxicity was strongest among
alone).11-15 NSAID-related ulcer com- patients not taking aspirin concomitantly.
plications are estimated to lead to JAMA. 2000;284:1247-1255 www.jama.com

Author Affiliations and Financial Disclosures are listed Lefkowith, MD, Pharmacia Clinical Research and De-
For editorial comment see p 1297. at the end of this article. velopment, 4901 Searle Pkwy, Bldg A3E, Skokie, IL
Corresponding Author and Reprints: James B. 60077 (e-mail: James.B.Lefkowith@monsanto.com).

©2000 American Medical Association. All rights reserved. (Reprinted) JAMA, September 13, 2000—Vol 284, No. 10 1247

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

107000 hospitalizations and 16500 ing RA or OA evident for at least 3 point evaluation for 2 months or until
deaths yearly in the United States.10 months and were expected to require study termination.
NSAIDs inhibit cyclooxygenase continuous treatment with an NSAID
(COX), the enzyme responsible for con- for the duration of the trial. Patients Treatment
version of arachidonic acid to prosta- were excluded from study participa- Patients were randomly assigned to
glandins.16 COX exists in 2 isoforms.17 tion if at screening they had active GI, receive treatments (celecoxib, 400 mg
COX-1 is a ubiquitous constitutive iso- renal, hepatic, or coagulation disor- twice per day; ibuprofen, 800 mg 3 times
zyme producing prostaglandins respon- ders; malignancy (unless removed sur- per day; or diclofenac, 75 mg twice per
sible for homeostatic functions such as gically with no recurrence within 5 day) on a 2:1:1 basis by an interactive
maintenance of GI mucosal integrity.17 years); esophageal or gastroduodenal voice response system (ClinPhone, Not-
COX-2 is largely a cytokine-induced iso- ulceration within the previous 30 days; tingham, England) according to a com-
zyme producing prostaglandins that me- history of gastric or duodenal surgery puter-generated randomization sched-
diate pain and inflammation.17 NSAIDs other than an oversew; or known im- ule. All treatment regimens were
inhibit both COX-1 and COX-2 to vary- mediate-type hypersensitivity to COX-2 blinded and double dummy. Treat-
ing degrees.18,19 Thus, the therapeutic ef- inhibitors, sulfonamides, ibuprofen, or ment assignment for 3 patients was
fects of conventional NSAIDs are de- diclofenac. Women were excluded if unblinded by study site personnel dur-
rived from inhibition of COX-2, while the they were pregnant, might have be- ing trial conduct (1 at the investiga-
adverse effects of these agents, particu- come pregnant, or were lactating. tion site, 2 via the interactive voice
larly in the upper GI tract, arise from in- response system). None of these patients
hibition of COX-1 activity. Study Protocol experienced a study outcome event. One
Celecoxib, a COX-2–specific inhibi- This prospective, randomized double- celecoxib patient experienced diver-
tor, recently was approved by the US blind trial was conducted at 386 cen- ticular bleeding; 2 patients (1 cele-
Food and Drug Administration (FDA) ters in the United States and Canada coxib and 1 diclofenac) experienced
for symptomatic treatment of rheuma- from September 1998 to March 2000 non–GI-related adverse events; and in
toid arthritis (RA) and osteoarthritis in accordance with the principles of no instance was the treatment assign-
(OA). To determine whether the COX-2 good clinical practice and the Declara- ment made known to personnel of the
specificity of celecoxib is associated with tion of Helsinki. The protocol was ap- drug company (Pharmacia, Skokie, Ill)
lower COX-1–related adverse effects, we proved by the institutional review board or to members of the oversight com-
compared celecoxib administered at at each study site, and all patients pro- mittees prior to final review of all end
2 and 4 times the maximum FDA- vided written informed consent. Prior points by a GI events committee.
approved effective dosages for RA and to enrollment, patients completed a
OA, respectively, with commonly used physical examination and clinical labo- Concomitant Medications
therapeutic dosages of ibuprofen and di- ratory testing. After a baseline visit, fol- NSAIDs (except for stable dosages of as-
clofenac. The dosage of celecoxib ex- low-up clinic visits took place at weeks pirin up to 325 mg/d); antiulcer drugs
ceeded the maximum dosage approved 4, 13, and 26 after the initial dose of (except for occasional antacid use); an-
by the FDA for OA and RA to permit a medication, and every 13 weeks there- tibiotics used alone or in combination
safety assessment of the higher dos- after. All patients were provided an op- with omeprazole, lansoprazole, and ra-
ages. However, based on previous stud- portunity to complete a minimum of 6 nitidine for treatment of Helicobacter py-
ies, 20,21 exceeding the dosages ap- months of treatment. lori infection; and antineoplastics (ex-
proved by the FDA would not improve Patients withdrawing from study par- cept methotrexate or azathioprine for
patients’ symptom relief. The dosages of ticipation prior to 6 months were clas- RA) were prohibited during the study.
ibuprofen and diclofenac were based on sified as follows: preexisting violation Use of oral, intramuscular, and intra-
prescription data; 48% and 60% of OA of entry criteria, protocol noncompli- articular glucocorticoids and disease-
and RA patients, respectively, who re- ance (investigator-defined failure to modifying antirheumatic drugs was per-
ceived ibuprofen were prescribed a dos- comply with the requirements of the mitted.
age of at least 2400 mg/d, and 36% and protocol, eg, failure to take at least 70%
57% of OA and RA patients, respec- of the study medication in any 13- Clinical Assessments
tively, who received diclofenac were pre- week interval), treatment failure (in- Investigators were instructed to iden-
scribed a dosage of at least 150 mg/d.22 vestigator-defined failure of study medi- tify and report all potential upper GI ul-
cation to control arthritis signs and cer complications. Evaluation of such
METHODS symptoms), or adverse effect (investi- events was outlined in a prespecified al-
Study Population gator-defined signs or symptoms un- gorithm structured to reproduce clini-
Outpatients aged 18 years or older were related to arthritis; see “Clinical As- cal practice norms. Evaluation was
eligible to participate in the study if, on sessments” herein). These patients required for any of the following pre-
screening, they were diagnosed as hav- nonetheless were followed up for end- sentations: hematemesis; melena; acute
1248 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved.

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

hypovolemia/hypotension; develop-
Table 1. Protocol-Specified Definitions and Adjudication Criteria for Ulcer Complications
ment of postural dizziness, lighthead-
Event Criteria for Confirmed Event
edness, or syncope; history of dark
Gastric or duodenal Perforated lesion requiring surgery. Could involve a laparoscopic
stool, hematochezia, or anal or rectal perforation repair, but only if evidence of the perforation was unequivocal,
bleeding; development of new anemia such as free air in the abdomen visible on radiograph or
peritoneal signs on physical examination.
(defined as a hematocrit level outside
Gastric outlet obstruction Gastric outlet obstruction requiring diagnosis by investigator;
of the reference range) or a decrease in diagnosis was required to be supported by endoscopy (eg,
hematocrit of at least 5 percentage ulcer with a tight edematous pyloric channel) or by
radiographic results (eg, dilated stomach, delayed barium
points; development of dyspepsia, ab- emptying with clinical evidence of outlet obstruction and with
dominal pain, or nausea or vomiting; an ulcer in the channel, severe outlet narrowing and edema)
or development of occult blood- Upper gastrointestinal Hematemesis with a lesion (ulcer or large erosion) on endoscopy
positive stools. Endoscopy was encour- bleeding or radiograph
Lesion (ulcer or large erosion) on endoscopy with evidence of
aged to document bleeding lesions but active bleeding or stigmata of a recent hemorrhage (visible
could also be performed if indicated by vessel or clot attached to the base of an ulcer)
Melena with a lesion (ulcer or large erosion) on endoscopy or
the investigator’s clinical judgment. radiograph
All documentation relating to poten- Occult blood-positive stool with a lesion (ulcer or large erosion)
on endoscopy or radiograph and with evidence of serious
tial ulcer complications was forwarded bleeding, including at least 1 of the following:
to a GI events committee (J.L.G., G.E., Decrease from baseline in hematocrit of Š5 percentage
points or in hemoglobin of >15 g/L
N.M.A., and W.F.S). The committee col- Postural vital sign changes (increase in heart rate of
lectively reviewed each case in a treat- Š20/min and/or decrease in systolic blood pressure
ment-blinded fashion and assigned it by of Š20 mm Hg and/or in diastolic blood pressure
of Š10 mm Hg)
unanimous consensus as either meet- Transfusion of Š2 units of blood
ing or not meeting the definition of an Blood in stomach on endoscopy or nasogastric aspiration
upper GI ulcer complication (TABLE 1).
Symptomatic ulcers consisted of cases respectively. Clinically significant in the protocol as consisting of all pa-
that did not meet the definition of an ul- changes in alanine aminotransferase tients who received at least 1 dose of as-
cer complication but did have endo- (ALT) and aspartate aminotransferase signed study medication. An addi-
scopic or x-ray evidence of a gastric or (AST) were predefined as increases to tional prespecified analysis was
duodenal ulcer as judged by the com- at least 3 times the upper limit of nor- performed on the population of pa-
mittee. All patients with symptomatic ul- mal. Trial safety (eg, serious adverse ef- tients not taking aspirin (since aspirin
cers or ulcer complications were with- fects) was monitored in a treatment- use was a predefined risk factor for GI
drawn from the study and included in blinded fashion during the study by the events). Time-to-event analyses of up-
the analysis as having had a study end data safety monitoring board (G.F., per GI ulcer complications alone or com-
point. T.P., A.W., and R.M.). bined with symptomatic ulcers were per-
Adverse effect data were collected at formed based on cumulative event rates
each visit (and as reported spontane- Statistical Analysis (symptomatic ulcers and/or ulcer com-
ously) using the following question: Sample size calculations were based on plications) for the 6-month study pe-
“Since your last visit, have you expe- the assumption that the annualized in- riod and are expressed as annualized in-
rienced or do you currently have any cidence of upper GI ulcer complica- cidence rates (number of events per 100
symptoms that are not associated with tions would be 0.3% for celecoxib and patient-years of exposure or percentage).
your arthritis?” All affirmative re- 1.2% for NSAIDs. To detect this differ- The log-rank test was used to compare
sponses were recorded regardless of se- ence with a 2-sided .05 significance level time-to-event curves among treatment
verity or relationship to study drug. with statistical power of 85% and as- groups. Based on the recommendation
Laboratory data were also collected at suming a 35% withdrawal rate, a sample of the GI events committee and as speci-
each visit and as indicated according to size of approximately 4000 patients was fied by the protocol a priori, upper GI
the investigators’ discretion. Clini- required for the celecoxib group and ulcer complications were defined as a
cally significant changes in hemato- 2000 patients were needed for each of study end point (ie, an uncensored
crit and hemoglobin were predefined the 2 NSAID groups. event) if they occurred within the
as decreases of at least 10 percentage Homogeneity of the treatment groups 6-month treatment period and oc-
points and 20 g/L, respectively. Clini- at baseline was analyzed using the y2test curred 48 hours after the first dose day
cally significant changes in serum urea for categorical data and 2-way analysis or before 14 days after the last known
nitrogen and creatinine were pre- of variance with treatment and center ef- dose of study drug (to avoid confound-
defined as values at 6-month fol- fects for continuous-valued data. Statis- ing due to prestudy or poststudy NSAID
low-up of at least 40 mg/dL (14.3 tical analyses were conducted on the in- use). Patients who had upper GI ulcer
mmol/L) and 1.8 mg/dL (159 µmol/L), tent-to-treat population, defined a priori complications outside of the specified
©2000 American Medical Association. All rights reserved. (Reprinted) JAMA, September 13, 2000—Vol 284, No. 10 1249

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

time frame were censored for purposes except for liver enzyme elevations, for
Figure 1. Flowchart of Patient Disposition at
6 Months of time-to-event analysis. This recom- which results are presented separately.
mendation was based on the pharma-
cologic washout period for most com- RESULTS
9764 Patients Screened
mon NSAIDs and evidence in the A total of 8059 patients were random-
literature of carryover effects of NSAIDs ized (FIGURE 1). Ninety-one patients
8059 Randomized in terms of GI toxic effects.8,23 Analyses did not receive study drug (32 were ran-
were conducted with and without these domized and found to be ineligible prior
censored patients. The effects of poten- to administration of study drug; 59
3987 Received Celecoxib 3981 Received NSAID
Treatment Treatment
tial risk factors for the development of withdrew consent prior to taking study
44 Did Not Receive 1985 Received an ulcer complication (including but not drug). Of these 91 patients, 44 were ran-
Celecoxib as Ibuprofen
Assigned 1996 Received
limited to concurrent aspirin use) were domized to celecoxib and 47 were ran-
Diclofenac analyzed by Cox proportional hazards domized to NSAIDs.
47 Did Not Receive
NSAID as Assigned
models. The incidences of treatment- A total of 7968 patients received at
emergent adverse effects or clinical labo- least 1 dose of medication. Of these,
ratory changes in the different treat- 3987 patients were treated with cele-
1611 Withdrawn 1784 Withdrawn coxib, 400 mg twice per day, and 3981
ment groups during the 6 months were
732 Adverse Events 822 Adverse Events
503 Treatment 589 Treatment compared using the Fisher exact test. All patients were treated with NSAIDs (1985
Failures Failures P values and 95% confidence intervals received ibuprofen, 800 mg 3 times per
376 Study 373 Study
Noncompliance Noncompliance (CIs) are 2-sided. No significant differ- day, and 1996 received diclofenac, 75 mg
ences in adverse events were noted by twice per day). The celecoxib and NSAID
sex, so results are presented with women groups had 1441 and 1384 total patient-
2376 Completed Study 2197 Completed Study
and men combined. Adverse events for years of exposure, respectively. Base-
diclofenac and ibuprofen were similar line characteristics did not differ signifi-
cantly between groups (TABLE 2). More
than 20% of the patients were taking
Table 2. Baseline Patient Characteristics* low-dosage aspirin (Š325 mg/d). Ap-
Celecoxib Group NSAID Group proximately 57% of the patients
Characteristics (n = 3987) (n = 3981) (n=4573) completed 6 months of treat-
Age, mean (range), y 60.6 (20-89) 59.8 (18-90) ment (Figure 1). More patients in the
>65 y, % 39.1 37.3 NSAID treatment group withdrew from
>75 y, % 12.2 11.4 the study for either adverse effects
Women, % 68.5 69.1 (n= 822 [20.6%]) or lack of therapeu-
Race/ethnicity, % tic efficacy (n= 589 [14.8%]) than did
White 88.5 87.9
Black 7.5 8.2
celecoxib-treated patients (n = 732
Hispanic 2.7 2.8
[18.4%] and n= 503 [12.6%], respec-
Asian 0.7 0.8 tively; P=.01 and P=.005; Figure 1). No
Other 0.6 0.6 patients were lost to follow-up (ie, a
Primary rheumatoid arthritis, % 27.3 27.5 cause of withdrawal was determined for
Duration of disease, mean (SD), y all patients who withdrew).
Osteoarthritis 10.3 (9.7) 10.1 (9.9)
Rheumatoid arthritis 11.3 (9.9) 10.7 (9.6) GI Toxicity
NSAID therapy at study entry, % 81.4 81.6 A total of 260 cases were selected by the
Ibuprofen 21.7 20.9 GI events committee for adjudication.
Diclofenac 13.6 14.0 The committee identified 35 upper GI ul-
Potential risk factor, % cer complications and another 48 cases
History of gastrointestinal bleeding 1.7 1.5
that represented symptomatic but un-
History of gastrointestinal ulcer 8.4 8.1
complicated gastroduodenal ulcers
Helicobacter pylori infection, % 38.5 38.2
(TABLE 3). Four upper GI ulcer compli-
Tobacco use, % 15.8 14.9
cations (2 in celecoxib-treated patients
Alcohol use, % 30.9 30.1
Concurrent medications, %
and 2 in NSAID-treated patients) were
Aspirin (Š325 mg/d) 20.9 20.4 censored according to predetermined cri-
Corticosteroids 30.6 29.5 teria (see “Methods” section). The re-
Anticoagulants 1.1 1.1 maining 177 cases not meeting the defi-
*NSAID indicates nonsteroidal anti-inflammatory drug. nition of gastroduodenal ulcer or ulcer
1250 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved.

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

complication were assigned a diagnosis 1441 patient-years) and 3.68% (51 group). The annualized incidence of up-
from the categories listed in Table 3. events/1384 patient-years) for cele- per GI ulcer complications in non–
The annualized incidence of upper GI coxib and NSAIDs, respectively aspirin users was significantly lower
ulcer complications in celecoxib- (P=.03).Corticosteroid use was not sig- with celecoxib vs NSAIDs (0.44% [5
treated patients was 0.76% (11 events/ nificantly associated with the incidence events/1143 patient-years] vs 1.27% [14
1441 patient-years) vs an incidence of of upper GI ulcer complications in ei- events/1101 patient-years]; P =.04; Fig-
1.45% (20 events/1384 patient-years) ther treatment group (RR, 0.2 and 0.6 for ure 2B). The RR for celecoxib com-
for patients taking NSAIDs (P =.09; patients treated with celecoxib and pared with NSAIDs was 0.35 (95% CI,
FIGURE 2A). The relative risk (RR) for NSAIDs, respectively; P=.13 and P=.27). 0.14-0.98). The annualized incidence
celecoxib compared with NSAIDs was
0.53 (95% CI, 0.26-1.11). The annu- GI Toxicity With Aspirin Use
Figure 2. Annualized Incidence of Upper
alized incidence of upper GI ulcer com- Based on time-to-event analyses using Gastrointestinal Tract Ulcer Complications
plications plus symptomatic ulcers with a Cox proportional hazard model, low- Alone and With Symptomatic
celecoxib was 2.08% (30 events/1441 dosage aspirin use was found to have a Gastroduodenal Ulcers
patient-years) vs 3.54% (49 events/ significant effect on the incidence of up-
1384 patient-years) for patients tak- per GI ulcer complications in celecoxib- A All Patients
Celecoxib
NSAIDs
ing NSAIDs (P=.02; Figure 2A). The RR treated patients. Within the celecoxib
6
for celecoxib compared with NSAIDs treatment group, the RR of an upper GI
was 0.59 (95% CI, 0.38-0.94). ulcer complication was 4.5 with low- 5

Annualized Incidence, %
P = .02
Inclusion of the 2 censored events in dosage aspirin use: 6 events in 833 pa- 4 49/1384
each group did not alter the interpreta- tients taking low-dosage aspirin vs 5
3
tion of results. For upper GI ulcer com- events in 3154 non–aspirin users P = .09
30/1441
plications, the rates without censoring (P=.01). Low-dosage aspirin use did not 2
20/1384
were 0.90% (13 events/1441 patient- have a significant effect on the rate of up- 11/1441
1
years) and 1.59% (22 events/1384 pa- per GI ulcer complications in patients
tient-years) for celecoxib and NSAIDs, receiving NSAIDs (RR, 1.7; P = .29). 0

respectively (P=.11). For upper GI ul- When the non–aspirin-using co-


cer complications plus symptomatic ul- hort was examined, 2 upper GI ulcer B Patients Not Taking Aspirin
cers, the rates were 2.22% (32 events/ complications were censored (1 in each 6

5
Annualized Incidence, %

Table 3. Adjudicated Cases Meeting and Not Meeting Prespecialized Definitions of P = .02
4
Gastroduodenal Ulcers and Ulcer Complications*
32/1101
Celecoxib Group NSAID Group 3
(n = 3987) (n = 3981) P = .04
2
Total No. of cases adjudicated 111 149† 14/1101 16/1143
No. of adjudicated cases not meeting the definition 1
5/1143
of a gastroduodenal ulcer or ulcer complication
Esophageal disease 23 21 0
Gastroduodenitis 12 21
Colonic or small bowel disease 10 7 C Patients Taking Aspirin
P = .49
Nonulcer bleeding 10 17 17/283
6
Miscellaneous GI symptoms 18 20
Anemia 5 12 5 14/298
Annualized Incidence, %

Cholelithiasis 1 0
4
Total 79 98
P = .92
No. of adjudicated cases meeting the definition 3
of a gastroduodenal ulcer or ulcer complication 6/298 6/283
Gastroduodenal ulcers 19 29 2

Ulcer complications‡ 13 22 1
Upper GI bleeding 10 20
Perforation 0 0 0
Ulcer Complications Symptomatic
Gastric outlet obstruction 1 0 Ulcers and
Total 32 51 Ulcer Complications

*NSAID indicates nonsteroidal anti-inflammatory drug; GI, gastrointestinal.


†P<.001 vs celecoxib group. Numbers above bars indicate events per patient-
‡Four ulcer complications (2 in the celecoxib group and 2 in the NSAID group) were censored from the analysis be- years of exposure. NSAIDs indicates nonsteroidal anti-
cause of the timing of the event based on a priori–specified definitions. inflammatory drugs.

©2000 American Medical Association. All rights reserved. (Reprinted) JAMA, September 13, 2000—Vol 284, No. 10 1251

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

of upper GI ulcer complications plus tion of results. For upper GI ulcer com- For patients taking aspirin (Figure
symptomatic ulcers in patients not tak- plications, the rates without censoring 2C), the annualized incidences of symp-
ing aspirin was also significantly lower were 0.52% (6 events/1143 patient- tomatic ulcers and/or upper GI compli-
with celecoxib than with NSAIDs years) and 1.36% (15 events/1101 pa- cations were not significantly different
(1.40% [16 events/1143 patient- tient-years) for celecoxib and NSAIDs, in patients taking celecoxib vs NSAIDs.
years] vs 2.91% [32 events/1101 pa- respectively (P=.05). For upper GI ul- For upper GI complications, the ob-
tient-years]; P =.02; Figure 2B). The RR cer complications plus symptomatic ul- served rates were 2.01% for patients tak-
for celecoxib compared with NSAIDs cers, the rates were 1.49% (17 events/ ing celecoxib vs 2.12% for patients tak-
was 0.48 (95% CI, 0.28-0.89). 1143 patient-years) and 3.00% (33 ing NSAIDs (6 events/298 patient-
Inclusion of the 1 censored event in events/1101 patient-years) for cele- years vs 6 events/283 patient-years,
each group did not alter the interpreta- coxib and NSAIDs, respectively (P=.02). respectively; P = .92). For upper GI ul-
cer complications plus symptomatic ul-
cers, the observed rates were 4.70% for
Table 4. Adverse Effects During the 6-Month Treatment Period* patients taking celecoxib vs 6.00% for
All Patients Patients Not Taking Aspirin patients taking NSAIDs (14 events/298
patient-years vs 17 events/283 patient-
Adverse Effects Celecoxib Group NSAID Group Celecoxib Group NSAID Group years, respectively; P=.49). Including the
(n = 3987) (n = 3981) (n = 3154) (n = 3169)
Gastrointestinal 2 censored events (1 in each group), the
Dyspepsia 575 (14.4) 640 (16.1)† 427 (13.5) 496 (15.7)† rates were 2.35% and 2.47%, respec-
Abdominal pain 387 (9.7) 522 (13.1)† 286 (9.1) 395 (12.5)† tively, for upper GI ulcer complica-
Diarrhea 373 (9.4) 392 (9.8) 288 (9.1) 293 (9.2) tions and 5.03% and 6.36%, respec-
Nausea 277 (6.9) 370 (9.3)† 213 (6.8) 277 (8.7)† tively, for upper GI ulcer complications
Constipation 68 (1.7) 234 (5.9)† 48 (1.5) 172 (5.4)† plus symptomatic ulcers.
Total 1250 (31.4) 1465 (36.8)† 942 (29.9) 1127 (35.6)†
Withdrawals 345 (8.7) 427 (10.7)† 252 (8.0) 321 (10.1)† Other Adverse Effects
Hepatic Adverse effects with an incidence of at
Elevated serum ALT 23 (0.6) 88 (2.2)† 18 (0.6) 68 (2.1)†
least 5% in either treatment group dur-
Elevated serum AST 18 (0.5) 73 (1.8)† 13 (0.4) 60 (1.9)†
Total 24 (0.6) 93 (2.3)† 18 (0.6) 72 (2.3)†
ing the 6-month treatment period were
Withdrawals 2 (<0.1) 46 (1.2)† 2 (<0.1) 36 (1.1)†
GI symptoms, upper respiratory tract
Bleeding-related
infection or related symptoms, head-
Anemia 81 (2.0) 175 (4.4)† 59 (1.9) 123 (3.9)† ache, and rash. Adverse effects caus-
Ecchymosis 28 (0.7) 32 (0.8) 22 (0.7) 26 (0.8) ing withdrawal with an incidence of at
Hematochezia 17 (0.4) 40 (1.0)† 11 (0.3) 29 (0.9)† least 1% in either treatment group were
Total 123 (3.1) 238 (6.0)† 90 (2.9) 171 (5.4)† GI symptoms, rash, and elevated trans-
Withdrawals 16 (0.4) 26 (0.7) 13 (0.4) 19 (0.6) aminase levels. For these categories, ce-
Renal lecoxib was associated with equiva-
Peripheral edema 113 (2.8) 138 (3.5) 90 (2.9) 108 (3.4)
lent or lower incidences of adverse
Hypertension 66 (1.7) 90 (2.3)† 50 (1.6) 65 (2.1) effects and withdrawals compared with
Increased creatinine level 28 (0.7) 48 (1.2)† 20 (0.6) 33 (1.0) NSAID therapy, with the exceptions of
Total 200 (5.0) 263 (6.6)† 155 (4.9) 198 (6.2)† rash and pruritus (TABLE 4).
Withdrawals 44 (1.1) 41 (1.0) 37 (1.2) 32 (1.0) Serious adverse effects (representing
Cardiovascular
Cerebrovascular accident 5 (0.1) 10 (0.3) 3 (<0.1) 5 (0.2) hospitalizations or malignancies de-
Myocardial infarction 10 (0.3) 11 (0.3) 3 (<0.1) 4 (0.1) tected during the 6-month treatment pe-
Angina 24 (0.6) 22 (0.6) 10 (0.3) 7 (0.2) riod) were reported for 4.3% of cele-
Total 37 (0.9) 39 (1.0) 16 (0.5) 14 (0.4) coxib patients (172 events/3987
Withdrawals 12 (0.3) 13 (0.3) 9 (0.3) 5 (0.2) patients) and 4.2% of NSAID patients
Cutaneous (168 events/3981 patients). The most
Rash 218 (5.5) 103 (2.6)† 180 (5.7) 91 (2.9)† common serious adverse effects in pa-
Pruritus 91 (2.3) 59 (1.5)† 72 (2.3) 44 (1.4)† tients taking celecoxib and NSAIDs were
Urticaria 22 (0.6) 14 (0.4) 18 (0.6) 13 (0.4) accidental fractures (7 and 8 events, re-
Total 298 (7.5) 163 (4.1)† 241 (7.6) 136 (4.3)† spectively), back pain (8 and 8 events,
Withdrawals 109 (2.7) 49 (1.2)† 92 (2.9) 43 (1.4)† respectively), pneumonia (9 and 9
*Data are given as No. (%) of patients. Categories are nonadditive; patients may have experienced more than 1 ad- events, respectively), cardiac failure (9
verse event in each category. NSAID indicates nonsteroidal anti-inflammatory drug; ALT, alanine aminotransferase;
and AST, aspartate aminotransferase. and 10 events, respectively), myocar-
†PŠ.05 vs celecoxib group.
dial infarction (10 and 10 events, re-
1252 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved.

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

spectively), and coronary artery dis- of normal was several-fold and statis-
Figure 3. Patients With Decreases in
ease (9 and 7 events, respectively). No tically significantly higher in patients Hematocrit and/or Hemoglobin at 6 Months
serious rashes or unexpected serious ad- receiving NSAIDs than those receiv-
verse events were observed in patients ing celecoxib. The incidence of ALT el- Celecoxib
P <.001 NSAIDs
taking celecoxib. evation for diclofenac was 3.2% vs 0.3%
The overall incidence of GI symp- for ibuprofen; for AST, it was 1.8% vs 4.0 135/3651 P <.001
toms experienced by patients taking ce- 0.1%, respectively. Similarly, investi- 3.5
107/3487
lecoxib was significantly lower than by gators reported a significantly higher in- 3.0
those taking NSAIDs, as was the rate of cidence of adverse effects related to el-

Incidence, %
2.5
withdrawal due to GI intolerability evated ALT and AST with NSAID
(Table 4). Of the most commonly re- 2.0
treatment (Table 4). Study withdraw- 58/3701
ported GI adverse effects, dyspepsia, ab- 50/3582
als due to such elevations were also 1.5

dominal pain, nausea, and constipa- higher in patients receiving NSAIDs 1.0
tion were significantly less common (Table 4). Overall, 97% of ALT and AST 0.5
with celecoxib than with NSAIDs, al- abnormalities occurred in patients re-
0
though there was no difference in the ceiving diclofenac. All Patients Patients Without
incidence of diarrhea (Table 4). The overall incidence of renal ad- GI Disease

The overall incidence of bleeding- verse effects, and the incidence of in- Data are shown for patients with decreases from pre-
related adverse events, and specifically, creased creatinine and hypertension in treatment levels in hematocrit of 10 percentage points
or more, in hemoglobin of 20 g/L or more, or both.
anemia and hematochezia, experienced particular, were significantly lower in Results for the entire study population are shown on
by patients taking celecoxib was signifi- patients receiving celecoxib than in the left. On the right, results for all patients exclud-
cantly lower than that among patients those receiving NSAIDs (Table 4). Also, ing those with an upper gastrointestinal (GI) ulcer com-
plication, symptomatic ulcer, or other diagnosed GI
taking NSAIDs for all patients and for significantly more patients receiving disease are shown. NSAIDs indicates nonsteroidal anti-
those not taking aspirin (Table 4). Simi- NSAIDs exhibited clinically signifi- inflammatory drugs. Numbers above bars indicate the
number of patients with the event per total number
lar results were noted for patients tak- cant elevations in serum creatinine of patients in the treatment group.
ing aspirin; the incidences of all bleeding- and/or serum urea nitrogen levels than
related adverse events were 4.0% and with celecoxib (Figure 4).
8.3% for patients taking celecoxib and The overall incidence of cardiovas- Figure 4. Patients With Increases in Serum
NSAIDs, respectively, and for anemia cular events, and the incidences of ce- Creatinine and/or Serum Urea Nitrogen and
were 2.6% and 6.4%, respectively rebrovascular events and myocardial in- With Elevations in ALT and AST at 6 Months
(P<.001 for both comparisons). Cele- farction in particular, were similar in
coxib was also associated with a lower the 2 treatment groups (Table 4). No Celecoxib
NSAIDs
incidence (P<.001) of clinically mean- treatment-related differences in such P <.001
ingful reductions in hematocrit and/or events were apparent in the cohort of 2.0 P = .03
64/3659
hemoglobin for the entire patient co- patients not taking aspirin for cardio- 1.8
57/3659
1.6
hort than NSAIDs (FIGURE 3). A lower vascular prophylaxis (Table 4). Inci- 1.4 P <.001
Incidence, %

incidence was noted both in patients not dence of myocardial infarction in pa- 1.2
36/3702
36/3657
taking aspirin (1.3% vs 3.4% in patients tients taking either celecoxib or NSAIDs 1.0
0.8
taking celecoxib and NSAIDs, respec- was 0.3%, with 95% CIs of 0.12% to 0.6
tively; P<.001) and patients taking as- 0.46% and 0.14% to 0.49%, respec- 0.4
7/3702
pirin (2.6% vs 4.9% in the 2 groups, re- tively. For patients not taking aspirin, 0.2 4/3702
0
spectively; P = .02). This difference incidence of myocardial infarction in ALT 3 AST 3 Increase in
Times Times Urea Nitrogen
persisted when all cases selected by the patients taking celecoxib was less than ULN ULN 40 mg/dL
GI events committee for adjudication 0.10% (95% CI, 0.02%-0.28%) and was and/or
were excluded from the analysis, thus re- also 0.10% (95% CI, 0.03%-0.32%) in Creatinine
1.8 mg/dL
moving all patients with ulcer compli- patients taking NSAIDs.
cations, symptomatic ulcers, or other di- Data are shown for patients with increases from pre-
agnosed GI disease (Figure 3). Mean COMMENT treatment levels in serum creatinine to 1.8 mg/dL (159
µmol/L) or more, in serum urea nitrogen to 40 mg/dL
serum iron–iron binding capacity ra- This study determined that celecoxib, (14.3 mmol/L) or more, or both; and for patients with
elevations in alanine aminotransferase (ALT) and as-
tios increased in patients taking cele- a COX-2–specific inhibitor, when used partate aminotransferase (AST) to a level of at least 3
coxib and decreased in patients taking for 6 months in a dosage 2 to 4 times times the upper limit of normal (ULN). In the nonste-
NSAIDs (1.4% vs −2.3%; P=.007). the maximum therapeutic dosage, is as- roidal anti-inflammatory drug (NSAID) group, 97%
of ALT and AST abnormalities occurred in patients
As shown in FIGURE 4, the inci- sociated with a lower incidence of com- taking diclofenac. Numbers above bars indicate the
dence of serum ALT or AST elevations bined clinical upper GI events than number of patients with the event per total number
of patients in the treatment group.
that exceeded 3 times the upper limit comparator NSAIDs (ibuprofen and di-
©2000 American Medical Association. All rights reserved. (Reprinted) JAMA, September 13, 2000—Vol 284, No. 10 1253

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

clofenac) used at standard therapeutic In addition to the assessment of GI Despite these caveats, however, our
dosages. effects, the present study determined results demonstrate that celecoxib, at a
In this study, patients taking NSAIDs that the increased dosage of celecoxib dosage 2- to 4-fold greater than the maxi-
had significantly higher rates of symp- used in this study did not change the mum therapeutic dosages and those ap-
tomatic ulcers or ulcer complications adverse effect profile observed at lower proved for labeling for RA and OA, is as-
than did patients taking celecoxib, but dosages.20,21,35 sociated with a lower rate of upper GI
the rate for ulcer complications did not Of note, celecoxib-treated patients toxic effects compared with standard
differ. The statistically indistinguish- had a significantly lower incidence of therapeutic dosages of NSAIDs. This
able rate of ulcer complications asso- clinically significant decreases in finding supports the COX-2 hypoth-
ciated with celecoxib and NSAIDs ap- hemoglobin and/or hematocrit com- esis that COX-2–specific agents ex-
pears to be a function of the higher- pared with NSAID-treated patients, hibit decreased GI toxic effects.17,39 De-
than-expected event rate observed in even when patients with upper GI spite the high dosage used, other adverse
the celecoxib group. The previously re- ulcer complications, symptomatic effects did not emerge. Our findings thus
ported annualized incidence rate for ul- ulcers, and other GI diseases were have significant implications with re-
cer complications in patients taking ce- excluded. Celecoxib was also better spect to drug therapy for the symptom-
lecoxib (used for the sample size tolerated than NSAIDs, as evidenced atic treatment of RA and OA.
determination) was 0.2%, obtained by the decreased incidence of GI Author Affiliations: Department of Medicine, Uni-
from pooled analyses of 14 random- symptoms and withdrawals for such versity of Washington, and Partner Frazier & Co, Se-
ized controlled trials.15 symptoms. attle (Dr Silverstein); Pharmaceutical Safety Assess-
ments Inc, Narberth, Pa (Dr Faich); Department of
This increased ulcer complication The clinical consequences of NSAIDs Medicine, Section of Digestive and Liver Diseases, Col-
rate was likely attributable to higher- on kidneys are heterogeneous, and, at lege of Medicine, University of Illinois at Chicago, Chi-
cago, Ill (Dr Goldstein); Division of Rheumatology and
than-anticipated concurrent low- present, the relative importance of Metabolic Bone Disease, Department of Medicine, Beth
dosage aspirin use. The percentage of COX-1 and COX-2 in the human kid- Israel Deaconess Medical Center, Harvard Medical
School, Boston, Mass (Dr Simon); Divisions of Rheu-
patients using low-dosage aspirin for ney is not well defined.36 Regardless, ce- matology and Immunology (Dr Pincus) and Gastro-
cardiovascular prophylaxis was nearly lecoxib appeared to be associated with enterology (Dr Eisen), Vanderbilt University School of
Medicine, Nashville, Tenn; Department of Medicine,
double that seen in other clinical trials significantly less renal toxicity com- Johns Hopkins University, Baltimore, Md (Dr Whelton);
that we have conducted recently, al- pared with NSAID therapy in this study. Department of Biostatistics, Yale University, New Ha-
beit within the range reported for the Although it has been hypothesized ven, Conn (Dr Makuch); Division of Gastroenterol-
ogy, Duke University, Durham, NC (Dr Agrawal); Di-
general population.24 Low-dosage as- that COX-2–specific inhibitors might vision of Gastroenterology, Washington University, St
pirin therapy has clearly been associ- increase the risk of cardiovascular Louis, Mo (Dr Stenson); and Clinical Research and De-
velopment, Pharmacia, Skokie, Ill (Ms Burr and Drs
ated with serious GI ulcer complica- thromboembolic events via inhibition Zhao, Kent, Lefkowith, Verburg, and Geis).
tions.25-29 of vascular prostacyclin synthesis with- Financial Disclosures: Ms Burr and Drs Zhao, Kent,
Lefkowith, Verburg, and Geis are employees of Phar-
In contrast, analysis of non–aspirin out a corresponding inhibition of plate- macia, manufacturer of celecoxib. In addition, all
users alone demonstrated that cele- let thromboxane, no such increase was authors who are not Pharmacia employees are paid
consultants for Pharmacia.
coxib was associated with a signifi- evident in the current study. 37 In both Funding/Support: This study was funded by Phar-
cantly lower incidence of symptom- the entire study population and the co- macia.
atic ulcers and/or ulcer complications hort not taking aspirin (who would con- Acknowledgment: We acknowledge the contribu-
tions of Dave Callison, Kirsten Katz, Beverly Paperi-
compared with NSAIDs. The rate of ul- jecturally be at greatest risk of such an ello, and June Kaiser, who monitored the study, and
cer complications in non–aspirin us- effect), the incidence of cardiovascu- Clem Maurath, who provided statistical support.
ers taking celecoxib (0.44%) is similar lar events, particularly myocardial in-
REFERENCES
to the background rate of ulcer com- farction, was comparable between the
1. The American College of Rheumatology Clinical
plications observed in patients not tak- celecoxib and NSAID groups. Guidelines. Atlanta, Ga: American College of Rheu-
ing NSAIDs or aspirin in the general Despite the size and duration of this matology; 1997.
population (0.1%-0.4%).8,9,11,12,30-33 trial, the populations of patients with 2. Wolfe F, Zhao S, Lane N. Preference for nonste-
roidal anti-inflammatory drugs over acetaminophen
The observed incidences of symp- OA and RA are much larger and therapy by rheumatic disease patients: a survey of 1799 pa-
tomatic ulcers and/or ulcer complica- continues for substantially longer than tients with osteoarthritis, rheumatoid arthritis, and fi-
bromyalgia. Arthritis Rheum. 2000;43:378-385.
tions were not significantly different in 6 months.38 Moreover, many patients 3. Pincus T, Swearingen C, Cummins P, Callahan LF.
patients taking celecoxib vs NSAIDs with OA and RA have comorbid ill- Preference for nonsteroidal anti-inflammatory drugs
versus acetaminophen and concomitant use of both
who were also taking concomitant low- nesses (eg, active GI disease) that would types of drugs in patients with osteoarthritis. J Rheu-
dosage aspirin. Data from endoscopic have excluded them from the current matol. 2000;27:1020-1027.
trials suggest that there may be a sig- study. Consequently, the results of this 4. Gabriel SE, Jaakkimainen L, Bombardier C. Risk for
serious gastrointestinal complications related to use of
nificant but smaller risk reduction in study do not address the occurrence of nonsteroidal anti-inflammatory drugs: a meta-
patients taking low-dosage aspirin, but rare adverse events, nor can they be ex- analysis. Ann Intern Med. 1991;115:787-796.
5. Henry D, Lim LL, Garcia RL, et al. Variability in risk
this remains to be proven in terms of trapolated to all patients seen in gen- of gastrointestinal complications with individual non-
clinical outcomes.34 eral clinical practice. steroidal anti-inflammatory drugs: results of a col-

1254 JAMA, September 13, 2000—Vol 284, No. 10 (Reprinted) ©2000 American Medical Association. All rights reserved.

Downloaded From: on 05/14/2018


GI TOXICITY WITH CELECOXIB VS NSAIDS FOR ARTHRITIS

laborative meta-analysis. BMJ. 1996;312:1563- 17. Crofford LJ, Lipsky PE, Brooks P, Abramson SB, buffered product. Lancet. 1996;348:1413-1416.
1566. Simon LS, Van de Putte LBA. Basic biology and clini- 30. Garcia RL, Jick H. Risk of upper gastrointestinal
6. Griffin MR, Piper JM, Daugherty JR, Snowden M, cal application of specific cyclooxygenase-2 inhibi- bleeding and perforation associated with individual
Ray WA. Nonsteroidal anti-inflammatory drug use and tors. Arthritis Rheum. 2000;43:4-13. nonsteroidal anti-inflammatory drugs. Lancet. 1994;
increased risk for peptic ulcer disease in elderly per- 18. Gierse JK, Koboldt CM, Walker MC, Seibert K, 343:769-772.
sons. Ann Intern Med. 1991;114:257-263. Isakson PC. Kinetic basis for selective inhibition of cy- 31. Lanza LL, Walker AM, Bortnichak EA, Dreyer NA.
7. Langman MJ, Weil J, Wainwright P, et al. Risks of clooxygenases. Biochem J. 1999;339:607-614. Peptic ulcer and gastrointestinal hemorrhage associ-
bleeding peptic ulcer associated with individual non- 19. Gierse JK, Hauser SD, Creely DP, et al. Expres- ated with nonsteroidal anti-inflammatory drug use in
steroidal anti-inflammatory drugs. Lancet. 1994;343: sion and selective inhibition of the constitutive and in- patients younger than 65 years: a large health main-
1075-1078. ducible forms of human cyclooxygenase. Biochem J. tenance organization cohort study. Arch Intern Med.
8. Macdonald TM, Morant SV, Robinson GC, et al. 1995;305:479-484. 1995;155:1371-1377.
Association of upper gastrointestinal toxicity of non- 20. Simon LS, Weaver AL, Graham DY, et al. Anti- 32. Hallas J, Lauritsen J, Villadsen HD, Gram LF. Non-
steroidal anti-inflammatory drugs with continued ex- inflammatory and upper gastrointestinal effects of ce- steroidal anti-inflammatory drugs and upper gastro-
posure: cohort study. BMJ. 1997;315:1333-1337. lecoxib in rheumatoid arthritis: a randomized con- intestinal bleeding: identifying high-risk groups by ex-
9. Smalley WE, Ray WA, Daugherty JR, Griffin MR. trolled trial. JAMA. 1999;282:1921-1928. cess risk estimates. Scand J Gastroenterol. 1995;30:
Nonsteroidal anti-inflammatory drugs and the inci- 21. Bensen WG, Fiechtner JJ, McMillen JI, et al. Treat- 438-444.
dence of hospitalizations for peptic ulcer disease in el- ment of osteoarthritis with celecoxib, a cyclooxygen- 33. Rodriguez LAG, Cattaruzzi C, Troncon MG, Agos-
derly persons. Am J Epidemiol. 1995;141:539-545. ase-2 inhibitor: a randomized controlled trial. Mayo tinis L. Risk of hospitalization for upper gastrointesti-
10. Wolfe MM, Lichtenstein DR, Singh G. Gastroin- Clin Proc. 1999;74:1095-1105. nal tract bleeding associated with ketorolac, other non-
testinal toxicity of nonsteroidal anti-inflammatory 22. Scott-Levin Physician Drug and Diagnosis Audit steroidal anti-inflammatory drugs, calcium antagonists,
drugs. N Engl J Med. 1999;340:1888-1899. [database]. Newtown, Pa: Scott-Levin; 1997. and other antihypertensive drugs. Arch Intern Med.
11. Gutthann SP, Garcia RL, Raiford DS. Individual 23. Paulus HE, Bulpitt KJ. Nonsteroidal anti- 1998;158:33-39.
nonsteroidal anti-inflammatory drugs and other risk inflammatory drugs. In: Weyand CM, Wortmann RL, 34. Goldstein JL, Agrawal NM, Silverstein FE, et al.
factors for upper gastrointestinal bleeding and per- eds. Primer on the Rheumatic Diseases. Atlanta, Ga: Influence of H pylori (Hp) infection and/or low dose
foration. Epidemiology. 1997;8:18-24. Arthritis Foundation; 1997:422-426. aspirin (ASA) on gastroduodenal ulceration in pa-
12. Singh G, Rosen Ramey D. NSAID-induced gastro- 24. Physician advice and individual behaviors about tients treated with placebo, celecoxib or NSAIDs. Gas-
intestinal complications: the ARAMIS perspective— cardiovascular disease risk reduction—seven states and troenterology. 1999;116(suppl 4):G0759
1997: Arthritis, Rheumatism, and Aging Medical Infor- Puerto Rico, 1997. MMWR Morb Mortal Wkly Rep. 35. Emery P, Zeidler H, Kvien TK, et al. Celecoxib ver-
mation System. J Rheumatol Suppl. 1998;51:8-16. 1999;48:74-77. sus diclofenac in long-term management of rheuma-
13. Silverstein FE, Graham DY, Senior JR, et al. Mi- 25. Stalnikowicz DR. Gastrointestinal bleeding dur- toid arthritis: randomised double-blind comparison.
soprostol reduces serious gastrointestinal complica- ing low-dose aspirin administration for prevention of Lancet. 1999;354:2106-2111.
tions in patients with rheumatoid arthritis receiving non- arterial occlusive events: a critical analysis. J Clin Gas- 36. Whelton A. Nephrotoxicity of nonsteroidal anti-
steroidal anti-inflammatory drugs: a randomized, troenterol. 1995;21:13-16. inflammatory drugs: physiologic foundations and clini-
double-blind, placebo-controlled trial. Ann Intern Med. 26. Singh G. Gastrointestinal complications of pre- cal implications. Am J Med. 1999;106:13S-24S.
1995;123:241-249. scription and over-the-counter nonsteroidal anti- 37. McAdam BF, Catella LF, Mardini IA, Kapoor S,
14. Paulus HE. FDA arthritis advisory committee meet- inflammatory drugs: a view from the ARAMIS data- Lawson JA, FitzGerald GA. Systemic biosynthesis of
ing: serious gastrointestinal toxicity of nonsteroidal an- base. Am J Ther. 2000;7:121. prostacyclin by cyclooxygenase (COX)-2: the human
tiinflammatory drugs; drug-containing renal and bili- 27. Roderick PJ, Wilkes HC, Meade TW. The gastro- pharmacology of a selective inhibitor of COX-2. Proc
ary stones; diclofenac and carprofen approved. Arthritis intestinal toxicity of aspirin: an overview of ran- Natl Acad S c i U S A. 1999;96:272-277.
Rheum. 1988;31:1450-1451. domised controlled trials. Br J Clin Pharmacol. 1993; 38. Lawrence RC, Helmick CG, Arnett FC, et al. Es-
15. Goldstein JL, Silverstein FE, Agrawal NM, et al. 35:219-226. timates of the prevalence of arthritis and selected mus-
Reduced risk of upper gastrointestinal ulcer compli- 28. Weil J, Colin JD, Langman M, et al. Prophylactic culoskeletal disorders in the United States. Arthritis
cations with celecoxib, a novel COX-2 inhibitor. Am aspirin and risk of peptic ulcer bleeding. BMJ. 1995; Rheum. 1998;41:778-799.
J Gastroenterol. 2000;95:1681-1690. 310:827-830. 39. Verburg KM, Maziasz TJ, Weiner E, Loose L,
16. Vane JR. Inhibition of prostaglandin synthesis as 29. Kelly JP, Kaufman DW, Jurgelon JM, Sheehan J, Geis GS, Isakson PC. COX-2 specific inhibitors:
a mechanism of action for the aspirin-like drugs. Na- Koff RS, Shapiro S. Risk of aspirin-associated major up- definition of a new therapeutic concept. Am J Ther.
ture. 1971;231:235-239. per-gastrointestinal bleeding with enteric-coated or In press.

©2000 American Medical Association. All rights reserved. (Reprinted) JAMA, September 13, 2000—Vol 284, No. 10 1255

Downloaded From: on 05/14/2018

Das könnte Ihnen auch gefallen