You are on page 1of 9


Video Capsule Endoscopy to Prospectively Assess Small Bowel

Injury With Celecoxib, Naproxen Plus Omeprazole, and Placebo

*University of Illinois at Chicago, Chicago, Illinois; ‡Oregon Health and Sciences University, Portland, Oregon; §The Mount Sinai Medical
Center, New York, New York; 储VA Greater Los Angeles Healthcare System and the David Geffen School of Medicine at UCLA, Los Angeles,
California; and ¶Pfizer, Inc, Peapack, New Jersey

Background & Aims: Data indicate that cyclooxygenase- examine suspected small bowel pathologies, its utility
2–specific inhibitors cause less gastroduodenal mucosal has been limited by technical difficulties, relative dis-
damage than nonspecific NSAIDS, but their effects on comfort for the patient, and incomplete evaluation of the
the small bowel mucosa are less well recognized. In a small bowel. A noninvasive video capsule endoscopy
multicenter, double-blind, placebo-controlled trial with (VCE) system providing detailed digital imaging of the
video capsule endoscopy (VCE) we prospectively evalu-
small bowel has been recently approved by the Food and
ated the incidence of small bowel injury in healthy sub-
Drug Administration and is recommended as a first-line
jects treated with celecoxib compared to naproxen plus
omeprazole. Methods: We randomly assigned subjects
diagnostic tool for detecting small bowel pathologies
with normal baseline VCEs to celecoxib 200 mg twice (Given Diagnostic System; Given Imaging Ltd, Yo-
daily (n ⴝ 120), naproxen 500 mg twice daily plus qneam, Israel).8 –10
omeprazole 20 mg once daily (n ⴝ 118), or placebo (n The gastroduodenal ulcer complications associated
ⴝ 118) for 2 weeks. The primary end point was the with nonspecific NSAID use have been attributed to
mean number of small bowel mucosal breaks per cyclooxygenase-1 (COX-1) inhibition.11 In contrast,
subject. Results: Baseline VCE found small bowel lesions COX-2–specific inhibitors spare COX-1 at therapeu-
in 13.8% (57/413) of screened subjects, who became tic doses, and their use has been associated with lower
ineligible for randomization. The mean number of small rates of endoscopic gastroduodenal ulcers and ulcer
bowel mucosal breaks per subject and the percentage complications.12–18 Because COX-1 is constitutively
of subjects with these mucosal breaks were 2.99 ⴞ expressed in the human small bowel,19 we hypothe-
0.51, 55% for naproxen/omeprazole compared to 0.32
sized that COX-2–specific inhibitors would also be
ⴞ 0.10, 16% for celecoxib and 0.11 ⴞ 0.04, 7% for
less injurious than nonspecific NSAIDs in the small
placebo (P < .001, both comparisons). The magnitude
of the difference between celecoxib and placebo was bowel. Although there have been no prospectively
small but statistically significant (P ⴝ .04). Conclusions: designed enteroscopic studies directly comparing
Among healthy subjects with lesion-free baseline VCEs, these 2 drug classes with regard to small bowel mu-
celecoxib was associated with significantly fewer small cosal injury, this hypothesis is supported by the
bowel mucosal breaks than naproxen plus omeprazole. greater rate of fecal 51Cr RBC loss reported in healthy
This study also showed that the background incidence of subjects treated with the nonspecific NSAID ibupro-
small bowel lesions in healthy adults is not insignificant fen compared to the COX-2–specific inhibitor rofe-
and should be considered in future trials with VCE. coxib.20 Moreover, and of clinical relevance, in the
Celecoxib Long-term Arthritis Safety Study, signifi-
ata from uncontrolled case series and epidemiologic cantly more patients receiving therapeutic doses of
D studies have shown that nonsteroidal anti-inflam-
matory drugs (NSAIDs) may cause small bowel lesions,
ibuprofen or diclofenac had decreases in hematocrit
ⱖ10% and/or decreases in hemoglobin ⱖ2 g/dL com-
including ulcerations.1– 4 Clinically relevant complica- pared to those receiving a supratherapeutic dose of
tions of so-called small bowel NSAID-associated enter- celecoxib (400 mg twice daily).12 This difference per-
opathy may include the development of webs, strictures/
obstructions, acute bleeding, perforations, and occult Abbreviations used in this paper: COX, cyclooxygenase; NSAID, non-
gastrointestinal bleeding, which may lead to the devel- steroidal anti-inflammatory drug; VCE, video capsule endoscopy.
© 2005 by the American Gastroenterological Association
opment of iron-deficiency anemia.1–7 Although small 1542-3565/05/$30.00
bowel enteroscopy has been used in clinical practice to PII: 10.1053/S1542-3565(04)00619-6

Table 1. Small Bowel Lesion Assessment Table known or suspected ulcerogenic medication (including aspirin or
Category Description other NSAIDs) within 2 weeks before the baseline VCE (run-in
period); and use of antiulcer medications (ie, sucralfate, antacids,
1 Petechiae histamine-2-receptor antagonists, proton pump inhibitors, and
2 Mucosal break without hemorrhage
misoprostol) within 4 weeks before the baseline VCE, which
3 Mucosal break with hemorrhage
4 Web/stricture without mucosal break and without corresponds to at least 2 weeks before the initial screening visit
hemorrhage and entry into the run-in period.
5 Web/stricture without mucosal break and with
hemorrhage Study Design
6 Web/stricture with mucosal break and without
hemorrhage This was a prospective, randomized, double-blind, triple-
7 Web/stricture with mucosal break and dummy, placebo-controlled, multicenter study. After the initial
hemorrhage screening visit, subjects underwent a 2-week run-in period, dur-
8 Presence of blood (hemorrhage) without
visualized lesions
ing which time alcohol consumption and use of any known or
9 Other lesions suspected ulcerogenic medications or antiulcer agents were pro-
hibited. At the baseline visit, eligible subjects underwent a base-
NOTE. There were 9 categories of small bowel lesions. Each lesion line VCE. If any mucosal break or blood was observed in the small
was evaluated and assigned a category. Mucosal breaks included any
lesion that appeared as an erosion or ulcer, regardless of perceived
bowel, or if small bowel visualization time was less than 2 hours,
size. Hemorrhage was defined as visible blood. the subject was not eligible for randomization. Eligible subjects
were subsequently randomly assigned to receive celecoxib (Cele-
brex; Pfizer, Inc, New York, NY) 200 mg twice daily, naproxen
sisted even when potential upper gastroduodenal ulcer (Naprosyn; Roche Laboratories, Inc, Nutley, NJ) 500 mg twice
complications were removed from the analysis, sug- daily plus omeprazole (Prilosec; AstraZeneca, Wilmington, DE)
gesting that this anemia might have resulted from 20 mg once daily, or placebo for 2 weeks. After this 2-week
occult blood loss from beyond the upper gastrointes- treatment period, each subject underwent VCE for determination
tinal tract. of the primary and secondary end points of the study. A com-
To better understand and quantify the differential puter-generated randomization technique was used to assign sub-
small bowel injury pattern of nonspecific NSAIDs and jects in a 1:1:1 ratio. During the study, NSAIDs (including
COX-2–specific inhibitors, we designed a prospective, topical agents and aspirin), known or suspected ulcerogenic
double-blind, randomized study to evaluate the inci- agents, analgesics (including opioids), and antiulcer medications
were prohibited. Acetaminophen up to 4 g daily up to twice per
dence of small bowel lesions in healthy subjects
week was permitted for general pain relief. The Institutional
treated with celecoxib compared to naproxen plus
Review Boards at each of the 9 centers approved the protocol, and
omeprazole or placebo by using VCE. We purposely all subjects provided written informed consent at screening.
elected to study the NSAID in combination with a
proton pump inhibitor in an attempt to achieve bal- Video Capsule Endoscopy Methodology
ance among the treatment groups with regard to the
VCE was performed by using the Given M2A video
upper gastrointestinal tract risk of gastroduodenal capsule system (Given Imaging Ltd).8 This system has been
mucosal injury,21–23 and because the protective effects described in detail elsewhere.24 Smoking was prohibited for 24
of proton pump inhibitors would not be expected to hours before the procedure was initiated, and only liquids were to
occur within the remainder of the small bowel beyond be ingested beginning at lunchtime of the day before the proce-
the duodenum. dure. No purges were given, and no prokinetic medications were
administered. Subjects fasted 12 hours before swallowing the
Patients and Methods capsule. A light snack was permitted 4 hours after capsule inges-
tion. Data were collected for up to 8 hours after capsule ingestion.
Healthy adults (18 –70 years of age) who had no evidence
Repeat VCEs as a result of system failures (image loss or low
of either a mucosal break or blood in the small bowel at the
signal) were permitted only if capsule excretion was confirmed.
baseline VCE and who had no clinically significant laboratory or
physical examination abnormalities were eligible for randomiza-
Methodology for Review of Video Capsule
tion. Important criteria for exclusion were active gastrointestinal
disease or history of gastrointestinal ulcers or bleeding; known or
suspected complete or partial stenosis of the small intestine; prior Each local investigator as well as each member of the
gastric or intestinal surgery (resection); established delayed gastric Small Bowel Events Committee were required to attend a
emptying or diabetic gastroparesis; hemoglobin ⬍10 g/dL standardized training session on the use of the Given Diag-
(women) or 12 g/dL (men); positive fecal occult blood test results nostic System and to pass a certification examination to ensure
at screening; frequent (more than 3 times per week) use of aspirin proficiency with the use of the Given Diagnostic System and
or other NSAIDs within 2 weeks before screening; use of any the evaluation of the M2A video images.

Figure 1. Trial profile.

The investigator at each center reviewed and interpreted whether they believed the image met the criteria of a
the baseline (pretreatment) video capsule images for deter- mucosal break. The members of the independent Small
mination of eligibility into the trial. If no mucosal break or Bowel Events Committee (J.L.G., G.M.E., I.M.G., and
blood was observed, the subject was randomized to the B.L.), who were blinded to subject, site, and treatment,
2-week, blinded treatment phase and, after that, underwent then reviewed and adjudicated to consensus these blinded,
the end-of-study VCE. The investigator at each center (who marked videos, according to the prespecified Small
was blinded to patient treatment) then reviewed and eval- Bowel Lesion Assessment criteria (Table 1) for determina-
uated the end-of-study VCEs and marked all suspected tion of the primary and secondary end points. Figure 1
small bowel abnormalities. These investigators were in- describes the flow of the patients during the duration of the
structed to mark any significant lesion, regardless of trial.

Study End Points wise comparisons. All hypothesis tests were conducted by
using a Type I error rate of 5%, and all tests were 2-tailed,
The primary end point was the mean number of
with two-sided 95% confidence intervals provided. SAS
small bowel mucosal breaks (defined as any break in the
Version 8.2 (SAS Institute, Inc, Cary, NC) was used for all
mucosa, as depicted by categories 2, 3, 6, and 7 in Table 1)
per subject. There was no attempt to differentiate between
ulcers and erosions, because video capsule technology can-
not accurately assess either lesion size or depth.24 There
were 3 secondary end points: the percentage of subjects with Patients
1 or more small bowel mucosal breaks; the total number of
Between September 2002 and March 2003, we
small bowel lesions with or without hemorrhage (categories
screened 462 healthy adult subjects from 9 centers (8
2, 3, 4, 5, 6, 7, and 9; Table 1); and the percentage of
subjects with visible blood in the small bowel but without centers in the United States and 1 center in Israel). A
visualized lesions in small bowel (category 8, Table 1). We total of 40 subjects did not meet general inclusion
also performed 2 post hoc analyses. One was an analysis of and/or exclusion criteria, and in 9 other subjects the
the mean number of mucosal breaks among those subjects baseline videos were technical failures. Of the remain-
with at least 1 mucosal break. The second was an analysis of ing 413 subjects, 57 (13.8%) were found to have
the distribution of small bowel mucosal breaks across ter- lesions at the baseline VCE and were not eligible for
tiles, which divided the area between the duodenum and randomization. The remaining 356 subjects under-
cecum into 3 equal parts, on the basis of each subject’s went randomization: 120 to celecoxib, 118 to
specific small bowel transit time. Any video in which the naproxen plus omeprazole, and 118 to placebo (Figure
cecum could not be identified was excluded from this 1). Adverse events resulted in the withdrawal of 4
specific analysis. Overall safety was assessed by physical celecoxib subjects and 1 naproxen/omeprazole subject.
examination, laboratory tests, and observed or reported
A total of 351 subjects completed the study and
adverse events.
underwent a final video capsule examination. Techni-
Statistical Analysis cal failures of the video occurred in 12 subjects. The
remaining 339 subjects constituted the modified in-
Descriptive statistics were used for subject demo- tention-to-treat cohort. These subjects (as well as all
graphics and clinical characteristics and for the outcome
randomized subjects) had similar baseline characteris-
variables. The primary end point (number of small bowel
tics (Table 2).
mucosal breaks) and the secondary end point (number of
small bowel lesions with or without hemorrhage) were not Primary End Point
normally distributed, and therefore their rank scores were
analyzed with Cochran-Mantel-Haenszel test, stratified by
As shown in Table 3, the mean number of small
site. The nonparametric test was a Kruskal-Wallis test for bowel mucosal breaks per subject was significantly
the overall comparison and a Wilcoxon rank sum test for the higher in the naproxen/omeprazole group (2.99) com-
pairwise comparison. The remaining secondary end points pared to the celecoxib (0.32) and placebo groups
(percentage of subjects with mucosal breaks and percentage (0.11) (P ⬍ .001 for both comparisons). The difference
of subjects with visible blood without visualized lesions) between celecoxib and placebo was also significant (P
also used Cochran-Mantel-Haenszel test, stratified by site, ⫽ .04). A representative lesion meeting the criteria for
for overall and pairwise comparisons. A post hoc analysis category 2 (mucosal break without hemorrhage) is
with a nonparametric test (Friedman chi-square) was per- shown in Figure 2.
formed to assess whether within each treatment group a
difference existed in the distribution of small bowel muco- Secondary and Other End Points
sal breaks across tertiles, and pairwise comparisons across A significantly higher percentage of subjects in
treatments were analyzed by a Wilcoxon rank sum test. All the naproxen/omeprazole group (55%) had mucosal
analyses were performed on the modified intention-to-treat breaks compared to those in the celecoxib (16%) and
cohort (defined as all randomized subjects who completed 2
placebo (7%) groups (P ⬍ .001 for both comparisons);
weeks of double-blind therapy and who had valid end-of-
study VCEs). Because of current interest in the gastrointes-
the difference between celecoxib and placebo was also
tinal mucosal effects of COX-2–specific inhibitors com- significant (P ⫽ .04). The number of small bowel
pared to nonspecific NSAIDs, we also performed a post hoc lesions with or without hemorrhage was significantly
analysis of the mean number of mucosal breaks among the higher in the naproxen/omeprazole group (4.18) than
subgroup of subjects who had at least 1 mucosal break. The in the celecoxib (0.50) and placebo groups (0.14) (P ⬍
nonparametric test was a Kruskal-Wallis test for the overall .001 for both comparisons); there was no difference
comparison and a Wilcoxon two-sample test for the pair- between celecoxib and placebo (P ⫽ .09) (Table 3).

Table 2. Baseline Demographic Characteristics of the Subjects

All randomized subjects Modified intention-to-treat cohort

Celecoxib Naproxen/omeprazole Placebo Celecoxib Naproxen/omeprazole Placebo

Characteristic (n ⫽ 120) (n ⫽ 118) (n ⫽ 118) (n ⫽ 115) (n ⫽ 111) (n ⫽ 113)

Age (y) 33.9 ⫾ 10.6 32.8 ⫾ 10.1 33.6 ⫾ 11.0 33.9 ⫾ 10.8 33.1 ⫾ 10.3 33.6 ⫾ 11.0
Gender, no. of subjects (%)
Male 51 (42) 47 (40) 42 (36) 50 (43) 46 (41) 39 (35)
Female 69 (58) 71 (60) 76 (64) 65 (57) 65 (59) 74 (65)
Race, no. of subjects (%)
White 93 (78) 92 (78) 85 (72) 88 (77) 87 (78) 84 (74)
Black 9 (8) 12 (10) 12 (10) 9 (8) 12 (11) 11 (10)
Asian 10 (8) 8 (7) 16 (14) 10 (9) 7 (6) 15 (13)
Othera 8 (7) 6 (5) 5 (4) 8 (7) 5 (5) 3 (3)

NOTE. Plus-minus values are means ⫾ standard deviations.

aOtherincludes other racial groups and where the race was not allowed to be asked.

The percentage of subjects with blood (without omeprazole. The mean number (⫾ standard error) of
visualized lesions) in the small bowel (category 8, mucosal breaks for the celecoxib group was 2.06 ⫾
Table 1) was similar for celecoxib (7%) and naproxen/ 0.48, which was significantly lower (P ⫽ .004) than
omeprazole (8%), and each of these treatments was the mean observed in the naproxen/omeprazole group
significantly different from placebo (1%) (P ⱕ .01 for (5.44 ⫾ 0.79), but not significantly different from
both comparisons) (Table 3). When including all sub- that observed in the placebo group (1.50 ⫾ 0.33) (P ⫽
jects with hemorrhage (categories 3,5,7, and 8; Table .58 for celecoxib vs placebo). The difference between
1), the percentages remained the same for celecoxib naproxen/omeprazole and placebo was also significant
and placebo and increased to 14% for naproxen/omep- (P ⫽ .01). The distribution of subjects with 1 or more
razole. Only the difference between the active treat- mucosal breaks is shown in Figure 3.
ment groups and placebo, however, was significant (P An additional post hoc analysis found that there was
ⱕ .01 for both comparisons). no statistically significant difference in the distribu-
Post Hoc Analyses tion of mucosal breaks throughout tertiles of the small
In a post hoc analysis, we calculated the mean bowel within each treatment group (Table 4). Further-
number of mucosal breaks among those subjects who more, the pattern of the treatment difference for the
developed at least 1 mucosal break; there was a sig- comparison of naproxen/omeprazole to celecoxib and
nificant difference between celecoxib and naproxen/ placebo was consistent within each tertile and similar

Table 3. Small Bowel Lesions After Two Weeks of Treatment

End point Celecoxib (n ⫽ 115) (n ⫽ 111) Placebo (n ⫽ 113) P valuea

Primary end point

Number of small bowel mucosal breaks ⬍.001b,c
0.32 ⫾ 0.10 2.99 ⫾ 0.51 0.11 ⫾ 0.04
per subject .04d
Secondary end points
Subjects with small bowel mucosal ⬍.001b,c
18 (16) 61 (55) 8 (7)
breaks, no. of subjects (%) .04d
Number of small bowel lesions with or ⬍.001b,c
0.50 ⫾ 0.19 4.18 ⫾ 0.72 0.14 ⫾ 0.05
without hemorrhage per subject .09d
Subjects with blood in the small bowel, .75b
without visualized lesions, no. of 8 (7) 9 (8) 1 (1) .007c
subjects (%) .01d

NOTE. Modified intention-to-treat cohort. Plus-minus values are unadjusted means ⫾ standard errors.
aP value based on Wilcoxon rank sum method for the end points relating to number of lesions and on Cochran-Mantel-Haenszel for end points

relating to percentage of subjects with small bowel lesions or blood.

bCelecoxib vs naproxen/omeprazole.
cNaproxen/omeprazole vs placebo.
dCelecoxib vs placebo.

.004). These findings are consistent with published re-

ports of the beneficial gastroduodenal effects of COX-2–
specific inhibitors compared to nonspecific NSAIDs
alone.12–18 We also performed a post hoc analysis of
mucosal break distribution within the small bowel to
investigate whether any treatment had a local irritant/
topical effect on the proximal small bowel versus a more
systemic effect throughout the small bowel. There was
no statistically significant difference in the distribution
of mucosal breaks throughout tertiles of the small bowel
within each treatment group. In addition, across treat-
ments, the pattern in the individual tertiles was similar
to that observed for the primary end point.
Small bowel injury and clinically relevant compli-
cations associated with the use of nonspecific NSAIDs
are well recognized.1–5 In the landmark study assess-
ing small bowel injury at the time of autopsy, the
prevalence of nonspecific small bowel ulcerations was
8.4% in patients taking nonspecific NSAIDs within 6
Figure 2. Mucosal break without hemorrhage (category 2 lesion), as months before death (n ⫽ 249) as compared to 0.6%
defined in Table 1. Note the loss of mucosal integrity with surrounding in control subjects (no reported NSAID exposure) (n
⫽ 464) (P ⬍ .001).2 Because the nonspecific NSAID
group constituted a broad range of documented
NSAID use (intermittent or daily use within 6 months
to the overall results (which did not control for tertile before death), the results likely do not represent a true
location). However, in each tertile, the difference be- incidence. The findings of our study (55% of healthy
tween celecoxib and placebo was no longer statistically subjects in the nonspecific NSAID group developed
significant, possibly because of fewer small bowel mu- mucosal breaks) are in better agreement with those of
cosal breaks in each tertile. a recently reported small bowel VCE study in arthritis
General Safety patients (n ⫽ 40).25 In that study, Graham et al25
found that 58% of nonspecific NSAID users (duration,
All treatments were well tolerated, and there
⬎3 months) had small bowel lesions compared to 17%
were no significant differences among the groups in
of nonusers (taking acetaminophen alone or nothing).
the incidence of adverse events. Adverse events caused
This preliminary study, however, did not compare
study withdrawal in 4 celecoxib-treated subjects (gas-
COX-2–specific inhibitors with nonspecific NSAIDs.
trointestinal symptoms, renal calculi, cholelithiasis,
pneumonia) and 1 naproxen/omeprazole-treated sub-
ject (allergic reaction). 18

16 celecoxib

Discussion 14
naproxen+om eprazole

We found a 9-fold lower mean number of small 12

No. of Subjects

bowel mucosal breaks per subject in healthy individuals 10

who received 2 weeks of treatment with celecoxib 200 8

mg twice daily compared to the combination of naproxen 6

500 mg twice daily plus omeprazole 20 mg once daily 4
(0.32 vs 2.99; P ⬍ .001). The celecoxib group also had 2
a significantly smaller percentage of subjects with mu- 0
cosal breaks (16% vs 55%, P ⬍ .001). In addition, in our 1 2 3 4 5 6 - 10 11 - 15 >15
No. of Mucosal Breaks
post hoc analysis among subjects who developed at least
1 mucosal break, the mean number of breaks was signif- Figure 3. Distribution of mucosal breaks among subjects with at
least 1 mucosal break. The mean number of mucosal breaks was
icantly greater in the naproxen/omeprazole group than in greater in subjects treated with naproxen ⫹ omeprazole than in
the celecoxib group (5.44 vs 2.06, respectively; P ⫽ subjects treated with either celecoxib or placebo.

Table 4. Number of Small Bowel Mucosal Breaks per Tertile of the Small Bowel Length
Tertile Celecoxib (n ⫽ 115) (n ⫽ 111) Placebo (n ⫽ 113) P valuea

First 0.13 ⫾ 0.04 1.28 ⫾ 0.28 0.07 ⫾ 0.04
Second 0.06 ⫾ 0.03 1.02 ⫾ 0.23 0.01 ⫾ 0.01
Third 0.13 ⫾ 0.08 0.68 ⫾ 0.17 0.02 ⫾ 0.01
Pe .08 .23 .37

NOTE. Modified intention-to-treat cohort. Plus-minus values are unadjusted means ⫾ standard errors.
aP value based on Wilcoxon rank sum test.
bCelecoxib vs naproxen/omeprazole.
cNaproxen/omeprazole vs placebo.
dCelecoxib vs placebo.
eP value based on Friedman ␹2 test.

A surprising and unexpected finding in our study bition of COX-1 by celecoxib at therapeutic doses29
was that 13.8% (57/413) of healthy subjects were not (as used in this study) or by COX-2–specific inhibitors
eligible for randomization because of the detection of in general, when taken in therapeutic or suprathera-
small bowel lesions at the baseline VCE. The primary peutic doses.29 –33 Importantly, although there was a
lesion in most of these subjects (44/57) was a mucosal statistically significant difference between celecoxib
break. Thus, 10.65% (44/413) of healthy subjects had and placebo with respect to the mean number of
mucosal breaks during the run-in period of the study mucosal breaks per subject and the percentage of
when no medication was given, which is not dissimilar subjects with mucosal break, the magnitude of the
from the 7% of placebo subjects who developed mu- difference is small compared to the differences ob-
cosal breaks during the 2-week study period. Al- served when celecoxib and placebo are compared to
though the clinical consequences and applicability of naproxen/omeprazole (Table 3).
mucosal breaks in healthy subjects and in subjects Our study had several limitations. First, it was a
taking anti-inflammatory agents are unknown, our short-term study and was conducted in a relatively
findings support our methodology of assessing small young and healthy population. As such, the general-
bowel lesions and, furthermore, emphasize the need izablity of our results to a disease-relevant population
for a control group in future prospective studies ex- using these drugs long-term is not clear. However, on
amining potential drug-induced effects on the small the basis of similar findings in the gastroduodenal
bowel. The use of a blinded events committee also mucosa, it is likely that our results might be extrap-
strengthens the reliability of our methodology. olated to an older, higher risk population who take
In several of the end points measured in our study, NSAIDs on a chronic basis.12,34,35 Second, although
celecoxib was found to be statistically significantly we used state-of-art VCE to assess small bowel abnor-
different from placebo. Although the cause is un- malities, this technique lacks the capability to mea-
known, it might possibly be explained by the up- sure lesion size and depth. Nevertheless, VCE is su-
regulation or importance of COX-2 in healing injured perior to other small bowel examinations in detecting
mucosa.26,27 Celecoxib might have inhibited this ef- mucosal disease and is the only currently available
fect, whereas placebo did not. Higher plasma concen- technique to reasonably address the issue of NSAID-
trations of celecoxib as a result of a genetic polymor- associated enteropathy in a systematic way.36 –39
phism of CYP2C9 might possibly increase the Moreover, it is important to recognize that the clinical
incidence and/or severity of adverse events.28 How- relevance of the reported mucosal breaks in this trial
ever, genotype screening was not performed, and the has not been established. Further studies will be re-
possible influence of this polymorphism in our popu- quired to correlate the presence of mucosal breaks
lation is unknown. A third possible explanation might with the development of outcomes such as the devel-
be related to COX-1 inhibition, which is constitu- opment of anemia, overt bleeding, and other small
tively expressed in the small bowel.19 Celecoxib might bowel events such as perforations. Finally, although
have caused some degree of COX-1 inhibition. How- our primary and 2 of our secondary end points dem-
ever, platelet studies do not support significant inhi- onstrate a difference between the celecoxib and

naproxen/omeprazole arms, the third secondary end small bowel ulceration in patients receiving non-steroidal
anti-inflammatory drugs. Lancet 1991;337:520.
point (the percent of subjects with blood in the small 5. Smale S, Tibble J, Sigthorsson G, et al. Epidemiology and differential
bowel without visualized lesions) showed no difference diagnosis of NSAID-induced injury to the mucosa of the small intes-
between these 2 arms. We believe that this latter tine. Best Pract Res Clin Gastroenterol 2001;15:723–738.
finding might be due to the low number of events. 6. Ihse I, Lunderquist A, Akerman M. Chronic bleeding from a pri-
mary non-specific small intestinal ulceration localized by angiog-
Moreover, we believe that this end point is the least raphy. Acta Chir Scand 1978;144:189 –192.
specific because of the lack of visualized mucosal 7. Walker CO, Sorsdahl OA. Anemia due to nonspecific ileal ulcer-
breaks. ation. South Med J 1968;61:127–128.
8. Lewis B, Goldfarb N. The advent of capsule endoscopy: a not-so-
In conclusion, our study extends the understanding futuristic approach to obscure gastrointestinal bleeding. Aliment
of the effects of nonspecific NSAIDs in the small Pharmacol Ther 2003;17:1085–1096.
bowel and provides insight into the small bowel mu- 9. Bowel disorders detection system approved. FDAnews Daily Bul-
letin July 11, 2003. Available at:
cosal effects of COX-2–specific inhibitors. We found
bulletin/1_134/news/15318-1.html. Accessed April 26, 2004.
that subjects treated with celecoxib had significantly 10. Swain P. Wireless capsule endoscopy. Gut 2003;52(suppl
fewer small bowel mucosal breaks than those treated 4):iv48 –50.
with the combination of naproxen and a proton pump 11. Crofford LJ, Lipsky PE, Brooks P, et al. Basic biology and clinical
application of specific cyclooxygenase-2 inhibitors. Arthritis
inhibitor. Further studies are needed to determine the Rheum 2000;43:4 –13.
clinical implications of small bowel mucosal breaks in 12. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal tox-
the healthy untreated population as well as in the icity with celecoxib vs nonsteroidal anti-inflammatory drugs for
osteoarthritis and rheumatoid arthritis: the CLASS study: a ran-
population of chronic users of NSAIDs and COX-2–
domized controlled trial: Celecoxib Long-term Arthritis Safety
specific inhibitors. Study. JAMA 2000;284:1247–1255.
13. Bombardier C, Laine L, Reicin A, et al. Comparison of upper
gastrointestinal toxicity of rofecoxib and naproxen in patients
Appendix with rheumatoid arthritis: VIGOR Study Group. N Engl J Med
Study Design Committee: J. Goldstein, B. Lewis, G. 2000;343:1520 –1528.
Eisen, I. Gralnek, J. Fort, S. Zlotnick. Small Bowel Events 14. Deeks JJ, Smith LA, Bradley MD. Efficacy, tolerability, and upper
gastrointestinal safety of celecoxib for treatment of osteoarthritis
Committee: J. Goldstein, B. Lewis, G. Eisen, I. Gralnek.
and rheumatoid arthritis: systematic review of randomised con-
Data analysis and interpretation: J. Goldstein, J. Fort, S. trolled trials. BMJ 2002;325:619 – 626.
Zlotnick, L. Simms, C. Chang. Data collection: J. Fort, S. 15. Goldstein JL, Silverstein FE, Agrawal NM, et al. Reduced risk of
Zlotnick, L. Simms, C. Chang. Writing Committee: J. upper gastrointestinal ulcer complications with celecoxib, a novel
Goldstein, J. Fort, L. Baer. COX-2 inhibitor. Am J Gastroenterol 2000;95:1681–1690.
16. Sikes DH, Agrawal NM, Zhao WW, et al. Incidence of gastroduo-
The following investigators participated as site-specific
denal ulcers associated with valdecoxib compared with that of
Principal Investigators in the study and the authors would ibuprofen and diclofenac in patients with osteoarthritis. Eur J
like to recognize their extremely collegial and collaborative Gastroenterol Hepatol 2002;14:1101–1111.
work: Israel: Nadir Arber, Tel-Aviv Sourasky Medical Cen- 17. Goldstein JL, Stenson W, Agrawal N, et al. Valdecoxib is associ-
ter, Tel-Aviv, Israel; United States: James Aisenberg, Re- ated with a significantly lower incidence of ulcer complications
search Associates of New York, New York, NY; Russell and symptomatic ulcers in arthritis patients compared to NSAID
(abstr #T1403). Gastroenterology 2002;122(suppl 1):A469.
Brown, University of Illinois Chicago, Chicago, IL; Wilmot 18. Langman MJ, Jensen DM, Watson DJ, et al. Adverse upper gas-
C. Burch, Vanderbilt University Medical Center, Nashville, trointestinal effects of rofecoxib compared with NSAIDs. JAMA
TN; David Cave, St Elizabeth’s Medical Center of Boston, 1999;282:1929 –1933.
Brighton, MA; Gareth Dulai, VA of Greater Los Angeles 19. Kargman S, Charleson S, Cartwright M, et al. Characterization of
Healthcare System, Los Angeles, CA; Douglas Faigel, Or- prostaglandin G/H synthase 1 and 2 in rat, dog, monkey, and
human gastrointestinal tracts. Gastroenterology 1996;111:445–
egon Health & Sciences University, Portland, OR; John 454.
Johanson, Rockford Gastroenterology Associates, Rockford, 20. Hunt RH, Bowen B, Mortensen ER, et al. A randomized trial mea-
IL; Jonathan Leighton, Mayo Clinic Arizona, Scottsdale, suring fecal blood loss after treatment with rofecoxib, ibuprofen, or
AZ. placebo in healthy subjects. Am J Med 2000;109:201–206.
21. Chan FK, Hung LC, Suen BY, et al. Celecoxib versus diclofenac
and omeprazole in reducing the risk of recurrent ulcer bleeding in
References patients with arthritis. N Engl J Med 2002;347:2104 –2110.
1. Lengeling RW, Mitros FA, Brennan JA, et al. Ulcerative ileitis 22. Hawkey CJ, Karrasch JA, Szczepanski L, et al. Omeprazole com-
encountered at ileo-colonoscopy: likely role of nonsteroidal pared with misoprostol for ulcers associated with nonsteroidal
agents. Clin Gastroenterol Hepatol 2003;1:160 –169. anti-inflammatory drugs: OMNIUM study Group. N Engl J Med
2. Allison MC, Howatson AG, Torrance CJ, et al. Gastrointestinal 1998;338:727–734.
damage associated with the use of nonsteroidal anitiinflamma- 23. Yeomans ND, Tulassay Z, Juhasz L, et al. A comparison of
tory drugs. N Engl J Med 1992;327:751–754. omeprazole with ranitidine for ulcers associated with nonsteroi-
3. Morris AJ, Wasson LA, MacKenzie JF. Small bowel enteroscopy in dal anti-inflammatory drugs: ASTRONAUT Study Group. N Engl
undiagnosed gastrointestinal blood loss. Gut 1992;33:887– 889. J Med 1998;338:719 –726.
4. Morris AJ, Madhok R, Sturrock RD, et al. Enteroscopic evidence of 24. Lewis B, Gostout C. The performance of capsule endoscopy. In:

Halpern E, Jacob H, eds. Atlas of capsule endoscopy. Haifa, compared to naproxen in patients with arthritis. Am J
Israel: Rahash Printing Ltd, 2002:15–20. Gastroenterol 2001;96:1019 –1027.
25. Graham DY, Qureshi WA, Willingham F, et al. A controlled study of 36. Herrerias JM, Caunedo A, Rodrigues-Tellez M, et al. Capsule
NSAID-induced small bowel injury using video capsule endoscopy endoscopy in patients with suspected Crohn’s disease and neg-
(abstr #146). Gastroenterology 2003;124(suppl1):A37. ative endoscopy. Endoscopy 2003;35:564 –568.
26. Baatar D, Jones MK, Pai R, et al. Selective cyclooxygenase-2 37. Liangpunsakul S, Chadalawada V, Rex DK, et al. Wireless cap-
blocker delays healing of esophageal ulcers in rats and inhibits sule endoscopy detects small bowel ulcers in patients with nor-
ulceration-triggered c-Met/hepatocyte growth factor receptor in- mal results from state of the art enteroclysis. Am J Gastroenterol
duction and extracellular signal-regulated kinase 2 activation. 2003;98:1295–1298.
Am J Pathol 2002;160:963–972. 38. Eliakim R, Fischer D, Suissa A, et al. Wireless capsule video endos-
27. Sigthorsson G, Simpson RJ, Walley M, et al. COX-1 and 2, intestinal copy is a superior diagnostic tool in comparison to barium follow-
integrity, and pathogenesis of nonsteroidal anti-inflammatory drug through and computerized tomography in patients with suspected
enteropathy in mice. Gastroenterology 2002;122:1913–1923. Crohn’s disease. Eur J Gastroenterol Hepatol 2003;15:363–367.
28. Kirchheiner J, Stormer E, Meisel C, et al. Influence of CYP2C9 39. Fireman Z, Mahajna E, Broide E, et al. Diagnosing small bowel
genetic polymorphisms on pharmacokinetics of celecoxib and its Crohn’s disease with wireless capsule endoscopy. Gut
metabolites. Pharmacogenetics 2003;13:473– 480. 2003;52:390 –392.
29. McAdam BF, Catella-Lawson F, Mardini IA, et al. Systemic
biosynthesis of prostacyclin by cyclooxygenase (COX)-2: the
human pharmacology of a selective inhibitor of COX-2. Proc
Natl Acad Sci U S A 1999;96:272–277 [published erratum Address requests for reprints to: Jay L. Goldstein, MD, University
appears in Proc Natl Acad Sci U S A 1999;96:5890]. of Illinois at Chicago, Department of Medicine, Room 1020, 840
30. Leese PT, Hubbard RC, Karim A, et al. Effects of celecoxib, a new South Wood Street, M/C 787, Chicago, Illinois 60612. e-mail:
cyclooxygenase-2 inhibitor, on platelet function in healthy adults: a; fax: (312) 413-0342.
randomized controlled trial. J Clin Pharmacol 2000;40:124 –132. Individual investigators are listed in the Appendix.
31. Leese PT, Talwalker S, Kent JD, et al. Valdecoxib does not impair Supported by a grant from Pharmacia Corporation and Pfizer, Inc.
platelet function. Am J Emerg Med 2002;20:275–281. Drs Goldstein and Eisen are consultants to Pfizer and have received
32. Van Hecken A, Schwartz JI, Depré M, et al. Comparative inhibitory travel expenses and honoraria; Dr. Lewis is a consultant to Pfizer and
activity of rofecoxib, meloxicam, diclofenac, ibuprofen, and has received travel expenses. He is also a member of the Medical
naproxen on COX-2 versus COX-1 in healthy volunteers. J Clin Advisory Board for Given Imaging, Ltd, and has received honoraria; Dr
Pharmacol 2000;40:1109 –1120. Gralnek is a consultant to Pfizer and has received honoraria and travel
33. Wight NJ, Gottesdiener K, Garlick NM, et al. Rofecoxib, a COX-2 expenses and is supported by a VA HSR&D advanced research career
inhibitor, does not inhibit human gastric mucosal prostaglandin development award; Drs Zlotnick and Fort are former employees of
production. Gastroenterology 2001;120:867– 873. Pfizer, Inc.
34. Simon LS, Weaver AL, Graham DY, et al. Anti-inflammatory and We are deeply indebted to Lorinda Simms MSc, BSc (Hon), and
upper gastrointestinal effects of celecoxib in rheumatoid arthri- Chi-Hsing Chang, PhD, for their expertise and invaluable assistance
tis: a randomized controlled trial. JAMA 1999;282:1921–1928. with the statistical analysis of the data and to Lorraine R. Baer,
35. Goldstein JL, Correra P, Zhoa WW, et al. Reduced incidence of PharmD, for her invaluable assistance in the preparation of the manu-
gastroduodenal ulcers with celecoxib, a novel COX-2 inhibitor, script.