Rofecoxib and early relapse of inflammatory bowel disease: an open-label trial

Authors

  • L. Biancone,

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author
  • C. Tosti,

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author
  • A. Geremia,

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author
  • D. Fina,

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author
  • C. Petruzziello,

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author
  • S. Emerenziani,

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author
  • F. Pallone

    1. Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Center of Excellence for the Study of the Genomic Risk of Complex Multifactorial Diseases, Università degli Studi di Roma ‘Tor Vergata’, Rome, Italy
    Search for more papers by this author

Dr L. Biancone, Cattedra di Gastroenterologia, Dipartimento di Medicina Interna, Università degli Studi di Roma ‘Tor Vergata’, Via Montpellier, 1, 00133 Rome, Italy.
E-mail: biancone@med.uniroma2.it

Summary

Background : The safety and efficacy of selective cyclo-oxygenase-2 inhibitors in inflammatory bowel disease are under investigation.

Aim : To assess, in a prospective, open-label trial, the efficacy and safety of rofecoxib (12.5 mg/day) in inflammatory bowel disease patients and controls.

Methods : The inflammatory bowel disease group included 45 inactive patients (25 Crohn's disease; 20 ulcerative colitis) with associated arthralgia. The control group included 30 dyspeptic patients. The efficacy and safety of rofecoxib were assessed in inflammatory bowel disease patients and controls before and after treatment (range, 3 days to 3 months).

Results : In inflammatory bowel disease, nine of the 45 patients (20%) required rofecoxib withdrawal due to gastrointestinal symptoms inducing clinical relapse, which subsided on drug discontinuation. The percentage of patients requiring rofecoxib discontinuation was comparable in patients with Crohn's disease and ulcerative colitis (20% vs. 20%), but was higher in inflammatory bowel disease patients than in controls (20% vs. 3%; P < 0.001). In inflammatory bowel disease, arthralgia relief was reported by 32 patients (71%): complete relief by eight patients (18%) and partial relief by 24 (53%). Thirteen patients (29%) reported no benefit. A comparable percentage of inflammatory bowel disease patients and controls reported arthralgia relief (71% vs. 70%).

Conclusions : Rofecoxib appears to control arthralgia in almost two-thirds of inflammatory bowel disease patients. Side-effects requiring drug discontinuation are observed, however, in almost one-quarter of patients.

Introduction

Non-steroidal anti-inflammatory drugs (NSAIDs) inhibit prostaglandin synthesis by inhibiting the enzyme cyclo-oxygenase (COX).1–3 The use of these drugs has been associated with symptom onset and relapse of quiescent inflammatory bowel disease.4, 5 Acute flares of Crohn's disease and ulcerative colitis are frequently associated with NSAID use.6 These drugs are therefore usually not indicated in patients with inflammatory bowel disease, thus leading to difficult management of common flu-like symptoms and minor arthralgia in these patients. The induction of upper gastrointestinal lesions by NSAIDs is also well established.7

New NSAIDs have therefore been developed which specifically inhibit COX-2, whilst sparing COX-1.8 As a result of these properties, the COX-2 inhibitors, celecoxib and rofecoxib, show anti-inflammatory activities, whilst being less harmful to the gastrointestinal tract.9–11 Because of the specific COX-2 inhibitor activity, celecoxib and rofecoxib may be associated with a lower risk of exacerbation in patients with inflammatory bowel disease.

In a prospective, open-label study, we assessed the efficacy and safety of the COX-2 inhibitor, rofecoxib, in inactive inflammatory bowel disease patients with associated arthralgia in comparison with dyspeptic patients with associated arthralgia used as controls.

Materials and methods

Study population

The clinical characteristics of and efficacy of rofecoxib in Crohn's disease and ulcerative colitis patients are reported in Tables 1 and 2, respectively. Forty-five inflammatory bowel disease patients undergoing regular follow-up at the Gastroenterology Unit of the University ‘Tor Vergata’, Rome (Italy) were enrolled and followed up prospectively. The inflammatory bowel disease group included 25 patients with Crohn's disease (10 males, 15 females; median age, 53 years; range, 28–70 years) and 20 patients with ulcerative colitis (14 males, six females; median age, 50 years; range, 28–69 years). The control group included 30 patients attending the same unit for dyspeptic symptoms not related to inflammatory bowel disease (12 males, 18 females; median age, 58 years; range, 18–78 years). None of the controls had a familial history of inflammatory bowel disease.

Table 1.  Clinical characteristics, efficacy and safety of rofecoxib in Crohn's disease patients
PatientIBDIBD siteRofecoxib durationRofecoxib discontinuation (Y/N)Arthralgia relief
Complete (Y/N)Partial (Y/N)None (Y/N)
  1. C, Colon; CD, Crohn's disease; I, ileum; IBD, inflammatory bowel disease; I-PA, pre-anastomotic ileum; N, no; Y, yes.

 1CDI-PA10 daysNNYN
 2CDI-C1 monthNYNN
 3CDI-C1 monthNNYN
 4CDI-PA2 weeksNNNY
 5CDI-PA5 daysNNNY
 6CDC4 weeksNYNN
 7CDI-PA4 weeksNNNY
 8CDI-PA8 weeksNNYN
 9CDI-C10 daysNYNN
10CDI-PA4 weeksNNNY
11CDI10 daysY (abdominal pain)NYN
12CDI-PA1 weekY (diarrhoea)NYN
13CDI1 monthNNYN
14CDI-PA3 daysNYNN
15CDI-PA4 weeksY (abdominal pain)NNY
16CDC2 weeksNNNY
17CDI4 weeksNYNN
18CDI-PA5 daysY (abdominal pain)NYN
19CDI1 monthNNYN
20CDI-PA5 daysY (abdominal pain)NYN
21CDI-C1 monthNNYN
22CDI-PA2 weeksNNYN
23CDI-C1 monthNNYN
24CDI3 monthsNYNN
25CDI1 weekNNYN
Table 2.  Clinical characteristics, efficacy and safety of rofecoxib in ulcerative colitis patients
PatientIBDIBD siteRofecoxib durationRofecoxib discontinuation (Y/N)Arthralgia relief
Complete (Y/N)Partial (Y/N)None (Y/N)
  • IBD, inflammatory bowel disease; N, no; UC, ulcerative colitis; Y, yes.

  • Epigastric pain subsided with 48 h of treatment with omeprazole (40 mg/kg).

 1UCTotal2 weeksY (heartburn,  bloody stools)NNY
 2UCDistal3 weeksNYNN
 3UCLeft1 monthNNNY
 4UCLeft10 daysNNYN
 5UCPouch3 weeksY (diarrhoea)NYN
 6UCDistal2 weeksNNNY
 7UCDistal8 weeksNNYN
 8UCDistal2 weeksN (epigastric pain)*NYN
 9UCTotal2 weeksNNNY
10UCDistal2 daysY (diarrhoea)NNY
11UCDistal2 weeksNNYN
12UCTotal20 daysNNNY
13UCLeft20 daysNNYN
14UCDistal2 weeksNNYN
15UCLeft2 daysY (abdominal pain)NNY
16UCDistal2 weeksNNYN
17UCDistal2 weeksNNYN
18UCDistal3 weeksNYNN
19UCLeft3 monthsNNYN
20UCLeft1 weekNNYN

The inclusion criteria for the inflammatory bowel disease group were as follows: (i) patients undergoing regular follow-up with a diagnosis made according to standard criteria; (ii) any inflammatory bowel disease site, extent and duration; (iii) no therapeutic changes in the last 3 months; (iv) clinically inactive Crohn's disease (Crohn's disease activity index of < 150)12 or ulcerative colitis (Mayo score of < 4)13 lasting for more than 6 months; (v) arthralgia lasting for at least 3 months.

The inclusion criteria for the control group were as follows: (i) dyspeptic symptoms requiring clinical attendance as out-patients at the same gastrointestinal unit; (ii) concomitant arthralgia requiring medical advice, lasting for at least 3 months.

The exclusion criteria for both groups were as follows: (i) age of less than 18 years and or greater than 80 years; (ii) concomitant use of NSAIDs; (iii) active peptic ulcer; (iv) previous use of COX-2 inhibitors.

As shown in Table 1, in the Crohn's disease group, the disease involved the distal ileum alone in 18 patients (as recurrent disease in 12), the distal ileum and colon in five patients and the colon only in two patients. Concomitant drug treatments at enrollment were mesalamine by mouth (2.4 g/day) in 23 patients and sulphasalazine (3 g/day) in two patients. Of the 23 patients treated with mesalamine, concomitant treatments included: methotrexate (15 mg/week intramuscularly) in one patient, budesonide (9 mg/day) in one patient, metronidazole (1 g/day by mouth) and ciprofloxacin (1 g/day by mouth) in one patient, metronidazole (1 g/day by mouth) in one patient and proton pump inhibitors (omeprazole 20 mg/day) in one patient. The indications for rofecoxib in Crohn's disease patients included peripheral arthralgia with or without tenosynovitis (n = 18), arthralgia in the shoulder (n = 4) and arthralgia in the cervical spine (n = 3). The previous use of NSAIDs (> 3 months before enrolment) was associated with epigastric pain in two patients.

As shown in Table 1, in the ulcerative colitis group, the disease extent was distal in 10 patients, left-sided in 6 patients and total in 3 patients, and 1 patient had pouchitis. Concomitant drug treatments at enrollment were mesalamine by mouth (2.4 g/day) in 12 patients (five patients were also receiving 2 g mesalamine enema), sulphasalazine (3 g/day) in seven patients (one patient was also receiving 2 g mesalamine enema) and VSL#3 (1 packet containing 900 billion bacteria/day) in the patient with pouchitis. The indications for rofecoxib treatment in the ulcerative colitis group were peripheral arthralgia (n = 18), arthritic pain in the shoulder (n = 1) and arthritic pain in the cervical spine (n = 1).

In the control group, concomitant drug medications included domperidone (10 mg/day) in five subjects. All 30 controls were referred to the unit with dyspeptic symptoms and the indications for rofecoxib treatment were peripheral arthralgia (n = 19), arthralgia in the cervical spine (n = 7), arthralgia in the lumbar spine (n = 3) and headache (n = 1). One control reported epigastric pain when using NSAIDs (> 3 months before enrolment).

Study protocol

Tables 1 and 2 show the clinical characteristics of and the efficacy and safety of rofecoxib treatment in Crohn's disease and ulcerative colitis patients, respectively. The characteristics of arthralgia were recorded at visit 1 and at the end of treatment in each inflammatory bowel disease patient and control. In inflammatory bowel disease patients, the clinical activity (Crohn's disease activity index for Crohn's disease; Mayo score for ulcerative colitis)12, 13 was assessed at the following points: (i) at entry into the study; (ii) at the end of the study; (iii) at the time of dropout in patients showing side-effects. Drug-related side-effects during and after treatment were reported. In inflammatory bowel disease patients, no change in drug treatment for intestinal disease was allowed during the study period. The study protocol was approved by the local ethics committee. All enrolled inflammatory bowel disease patients and controls gave written informed consent to rofecoxib. No financial support was obtained for this study.

Medication

All inflammatory bowel disease patients and controls were treated with rofecoxib (12.5 mg/day) for at least 3 days (range, 3 days to 3 months). Rofecoxib could be discontinued by the patient at any time during the study period due to side-effects or to a lack of efficacy of the drug.

Statistical analysis

Statistical differences between the frequency of side-effects or arthralgia relief in inflammatory bowel disease patients and controls were assessed using the chi-squared test.

Results

Adverse events

Inflammatory bowel disease.  During rofecoxib treatment, gastrointestinal symptoms were observed in 10 of the 45 inflammatory bowel disease patients (22%), requiring drug discontinuation in nine (20%) (Table 3). The nine patients discontinued rofecoxib due to abdominal pain (n = 5), diarrhoea with no blood (n = 3) or bloody stools and heartburn (n = 1), which subsided promptly on drug discontinuation. All nine patients showed inflammatory bowel disease relapse (Crohn's disease activity index of > 150 or Mayo score of > 4) at the time of dropout. One ulcerative colitis patient showed mild epigastric pain after 2 days, which subsided after 48 h of omeprazole (20 mg/day) treatment and did not require rofecoxib withdrawal. In order to assess the possible relation between rofecoxib-induced side-effects and inflammatory bowel disease localization, patients were grouped according to colonic (22 patients: 20 ulcerative colitis and two Crohn's disease) or small bowel (23 Crohn's disease patients, including the caecum in five) involvement (Figure 1). Side-effects were observed in five of the 23 patients with small bowel involvement (22%), all requiring drug discontinuation, and in five of the 22 patients with colonic inflammatory bowel disease (23%), requiring drug discontinuation in four (18%). The frequency of dropout was therefore comparable in patients with colonic vs. small bowel inflammatory bowel disease (Figure 1).

Table 3.  Induction of gastrointestinal symptoms requiring drug discontinuation and efficacy of rofecoxib in inflammatory bowel disease patients and controls
Disease groupGI symptoms requiring drug discontinuationArthralgia relief (%)
CompletePartialNone
  • CD, Crohn's disease; GI, gastrointestinal; IBD, inflammatory bowel disease; UC, ulcerative colitis.

  • No statistical differences between Crohn's disease and ulcerative colitis.

  • *

     P < 0.001, inflammatory bowel disease vs. controls.

IBD (n = 45)9/45 (20%)*8/45 (18%)24/45 (53%)13/45 (29%)
CD (n = 25)5/25 (20%)*6/25 (24%)13/25 (52%)6/25 (24%)
UC (n = 20)4/20 (20%)*2/20 (10%)11/20 (55%)7/20 (35%)
Controls (n = 30)1/30 (3%)6/30 (20%)15/30 (50%)9/30 (30%)
Figure 1.

Histograms showing the percentage of inflammatory bowel disease patients, grouped according to pure colonic (n = 22; 20 ulcerative colitis, two Crohn's disease) or small bowel (including the caecum in five) (n = 23; all Crohn's disease) involvement, requiring rofecoxib discontinuation due to side-effects. The percentage of dropouts did not differ significantly between the two groups of inflammatory bowel disease patients grouped according to disease localization (22% vs. 18%). IBD, inflammatory bowel disease.

Crohn's disease.  As shown in Table 1, of the 25 Crohn's disease patients treated with rofecoxib, five (20%) discontinued the drug due to diarrhoea with no blood (n = 1) or abdominal pain (n = 4), inducing clinical relapse (Crohn's disease activity index of > 150).12 These symptoms appeared within 10 days and completely subsided on drug discontinuation. In these five patients, arthralgia relief was reported to be partial by four, and one patient reported no benefit.

Ulcerative colitis.  Of the 20 ulcerative colitis patients treated with rofecoxib, four (20%) discontinued the drug due to the onset, within 3 weeks, of diarrhoea, abdominal pain or bloody stools and heartburn (Table 2). In all four patients, the symptoms subsided on drug discontinuation. Partial arthralgia relief was observed in one of these ulcerative colitis patients. Overall, five of the 20 ulcerative colitis patients (25%) showed side-effects, as one patient reported mild epigastric pain within 2 days that did not require drug discontinuation; it subsided with 48 h of treatment with omeprazole (20 mg/day).

Controls.  In the control group, one patient (3%) discontinued the drug after 2 days due to the abrupt onset of epigastric pain not relieved by omeprazole (40 mg/day), with no arthralgia relief. In the remaining 29 controls, no side-effects requiring drug withdrawal were observed. However, five showed gastrointestinal symptoms not requiring drug withdrawal: four (13%) reported epigastric pain subsiding with 48 h of omeprazole treatment (40 mg/day), and one (3%) reported the sporadic appearance of loose stools with no blood during treatment. In these five controls showing gastrointestinal side-effects not requiring drug discontinuation, rofecoxib induced partial arthralgia relief.

Inflammatory bowel disease patients vs. controls.  The percentage of inflammatory bowel disease patients and controls showing gastrointestinal symptoms was comparable (22% vs. 20%) (Figure 2). However, although nine of the 10 inflammatory bowel disease patients showing side-effects required drug discontinuation (9/45 = 20%), only one of the six controls required rofecoxib withdrawal (1/30 = 3%). The percentage of patients requiring drug withdrawal was therefore higher in the inflammatory bowel disease group than in the control group (P < 0.001) (Figure 2).

Figure 2.

Histograms showing the percentage of inflammatory bowel disease patients and controls with gastrointestinal symptoms during rofecoxib (12.5 mg/day) treatment. The overall frequency did not differ significantly between inflammatory bowel disease patients and controls (22% vs. 20%). However, although in nine of the 10 inflammatory bowel disease patients the appearance of these gastrointestinal symptoms required drug discontinuation (9/45 = 20%), only one of the six controls showing side-effects required rofecoxib withdrawal (1/30 = 3%). The frequency of dropouts was therefore higher in inflammatory bowel disease patients than in controls (20% vs. 3%; P < 0.001)*. IBD, inflammatory bowel disease.

Efficacy

Inflammatory bowel disease.  Complete or partial efficacy was reported by 32 of the 45 inflammatory bowel disease patients (71%). In particular, arthralgia relief was reported to be complete by eight patients (18%) (6 Crohn's disease, 2 ulcerative colitis) and partial by 24 patients (53%) (13 Crohn's disease, 11 ulcerative colitis); no benefit was reported by 13 patients (29%) (6 Crohn's disease, 7 ulcerative colitis) (Figure 3). The efficacy of rofecoxib did not differ between patients grouped according to pure colonic (n = 22) or small bowel (n = 23) inflammatory bowel disease involvement (efficacy: global, 64% vs. 78%; complete, 14% vs. 22%; partial, 50% vs. 56%; none, 36% vs. 22%; P = N.S. for all).

Figure 3.

Histograms showing the percentage of inflammatory bowel disease patients and controls with complete, partial or no relief of arthralgia when using rofecoxib (12.5 mg/day). The proportion of patients showing complete, partial or no response did not differ significantly between the two groups (arthralgia relief in inflammatory bowel disease patients vs. controls: complete, 18% vs. 20%; partial, 53% vs. 50%; no response, 29% vs. 30%; P = N.S. for all groups). IBD, inflammatory bowel disease.

Crohn's disease.  When considering the whole group of 25 Crohn's disease patients, arthralgia relief induced by rofecoxib was reported to be complete by 6 patients (24%) and partial by 13 patients (52%) (4 dropped out due to side-effects); no benefit at all was reported by the remaining 6 patients (24%) (1 dropout) (Table 3).

Ulcerative colitis.  Of the 20 ulcerative colitis patients, rofecoxib induced complete arthralgia relief in 2 (10%) and partial relief in 11 (55%) (one dropout due to bloody stools); no benefit was reported by the remaining 7 patients (35%) (3 dropouts: 1 due to diarrhoea, 1 due to abdominal pain and 1 due to bloody stools and heartburn) (Table 3).

Controls.  Rofecoxib induced either complete or partial arthralgia relief in 21 of the 30 controls (70%), and was reported to be complete by six (20%) and partial by 15 (50%); nine controls (30%) reported no benefit at all.

Inflammatory bowel disease vs. controls.  As shown in Figure 3, the percentage of inflammatory bowel disease patients and controls showing complete, partial or no arthralgia relief did not differ significantly between the two groups.

Discussion

The use of NSAIDs is a risk factor for upper and lower gastrointestinal side-effects.14–17 Two COX isoforms have been recognized (COX-1 and COX-2), and recent findings suggest the existence of a third COX enzyme.18, 19 COX-1 is constitutively expressed by many tissues, whereas COX-2 is expressed only at very low levels in most tissues, but is up-regulated at sites of inflammation. NSAIDs inhibit both COX-1 and COX-2 activity.20 NSAID-associated gastropathy appears to be mainly related to the suppression of COX-1 activity,19 although no gastric injury has been reported when using selective COX-1 inhibitors only.20 NSAID-induced gastrointestinal lesions appear to be mainly related to the inhibition of prostaglandin synthesis as a result of COX-1 suppression.21 COX-1 converts unsaturated fatty acids to prostaglandins, exerting protective effects on the gastrointestinal mucosa.22 NSAIDs also exert antiplatelet effects related to COX-1 suppression, including the inhibition of platelet aggregation and thromboxane synthesis,14 which are involved in NSAID-induced gastrointestinal bleeding. NSAID-induced COX-1 inhibition may therefore induce gastrointestinal lesions by interfering with prostaglandin synthesis and platelet function.14

NSAID use has been associated with the onset or exacerbation of inflammatory bowel disease,23, 24 proctitis induced by the use of suppositories,25 an increased risk of inflammatory bowel disease colitis,26 emergency admission to hospital for inflammatory bowel disease colitis,6 and relapse of experimental models of colitis. As a result of these findings, NSAIDs are currently not indicated in inflammatory bowel disease, although several conditions, including associated arthralgia, may suggest their efficacy.

As the inhibition of COX-1 activity appears to be responsible for NSAID-induced gastrointestinal lesions, the use of selective COX-2 inhibitors should not induce gastrointestinal complaints, as suggested by several studies.27–29 Nevertheless, increasing evidence has indicated that selective COX-2 inhibitors, including rofecoxib, may also induce gastrointestinal events, although at a lower frequency than NSAIDs.14–19 It has recently been reported that 77 patients would need to be treated with rofecoxib, instead of naproxen, to avert one serious gastrointestinal event.14

As arthralgia is a frequent complaint in inflammatory bowel disease and very few data are available on the efficacy and safety of selective COX-2 inhibitors in these patients, we addressed this issue in the first prospective study to investigate the effects of rofecoxib in inactive inflammatory bowel disease patients in comparison with controls. Dyspeptic patients with arthralgia were used as controls, in order to assess the role of pre-existing small bowel or large bowel lesions in rofecoxib-induced gastrointestinal complaints. Our open-label trial suggests that rofecoxib-induced lower gastrointestinal events are more frequent in patients with pre-existing small/large bowel lesions. In addition, our results suggest that rofecoxib may induce inflammatory bowel disease relapse, although arthralgia relief is reported by a high proportion of patients (71%) (comparable with controls: 70%), in agreement with previous findings in controls.27–29

In our series, of the 45 inflammatory bowel disease patients enrolled, nine (20%) discontinued rofecoxib due to gastrointestinal symptoms. The observation that the symptoms appeared a few days after starting treatment and subsided promptly on drug withdrawal supports their assignment as drug-related side-effects. The inflammatory bowel disease-related side-effects of rofecoxib are suggested both by the characteristics of the symptoms in previously inactive patients and by their appearance only in inflammatory bowel disease patients. Indeed, only one of the 30 controls (3%) discontinued rofecoxib (due to epigastric pain), although five (17%) showed epigastric pain relieved by proton pump inhibitors. It is noteworthy that the frequency of gastrointestinal symptoms was comparable in inflammatory bowel disease patients and controls (22% vs. 20%). However, although nine of the 10 inflammatory bowel disease patients showing side-effects required drug discontinuation, only one of the six controls discontinued rofecoxib. The higher frequency of dropouts due to side-effects in inflammatory bowel disease patients vs. controls may also be related to the higher state of alert by both patients and physicians to the appearance of gastrointestinal complaints in inflammatory bowel disease patients when compared with controls.

The type of gastrointestinal symptoms induced by rofecoxib differed between inflammatory bowel disease patients and controls: all nine inflammatory bowel disease patients showed inflammatory bowel disease-related symptoms, with clinical relapse (Crohn's disease activity index of > 150 or Mayo score of > 4), whereas five of the six controls showing side-effects reported epigastric pain only. The percentage of inflammatory bowel disease patients showing rofecoxib-induced gastrointestinal complaints did not differ between patients with pure colonic and small bowel involvement (23% vs. 22%). This observation is in agreement with the results from other studies on NSAID-induced side-effects in humans.14–16

In animal models of inflammatory bowel disease, higher COX-2 levels are observed in the colonic mucosa. Long-term celecoxib-induced COX-2 inhibition or COX-2 deficiency in mice has been associated with small intestinal pathology despite normal prostaglandin E2 levels,30 suggesting that long-term celecoxib-induced COX-2 inhibition may induce intestinal pathology and that COX-2 may play a role in the maintenance of small intestinal integrity.30

The results of our uncontrolled trial are in substantial agreement with the few others available. In a retrospective study, rofecoxib or celecoxib was used in 37 inflammatory bowel disease patients; the efficacy was found to be 59% and 14% discontinued the drug due to side-effects; 24% of patients discontinued COX-2 inhibitors due to either side-effects or to a lack of efficacy.31 In a letter, inflammatory bowel disease relapse was reported in two patients (one Crohn's disease, one ulcerative colitis) treated with celecoxib.32 Acute flare-up of ulcerative colitis associated with rofecoxib and cytomegalovirus infection33 and exacerbation of Crohn's colitis using rofecoxib34 have also been described in case reports. The first prospective study investigating the efficacy and safety of rofecoxib in inflammatory bowel disease-associated peripheral arthritis and arthralgia reported adverse events in 12.5% of patients, requiring drug discontinuation in 9%.35 These percentages are lower than those observed in our inflammatory bowel disease population.35 However, in agreement with our findings, gastrointestinal symptoms appeared after a few days of treatment and subsided promptly on drug discontinuation; in addition, improvement of joint functionality was observed in a comparable proportion of patients (57%).35 The different characteristics of the inflammatory bowel disease populations may account for minor discrepancies between the two studies, including the number of tested patients (45 vs. 32 inflammatory bowel disease: 20 vs. six ulcerative colitis; 25 vs. 20 Crohn's disease; zero vs. six indeterminate colitis), type and extent of inflammatory bowel disease and length of remission.35 Taken together, the results from our study are in substantial agreement with previous reports.31–36 Three main observations can be made: (i) possible inflammatory bowel disease flare-up and gastrointestinal complaints when using COX-2 inhibitors; (ii) early onset of side-effects; and (iii) prompt disappearance of gastrointestinal events on drug discontinuation. These findings suggest that COX-2 inhibitors should not be used in active inflammatory bowel disease until their safety and efficacy in patients with this condition have been assessed by controlled trials.

The mechanisms underlying COX-2-induced relapse in inflammatory bowel disease are still uncertain. Prostaglandins are protective in the gastrointestinal tract and NSAID-induced gastrointestinal lesions and inflammatory bowel disease relapse appear to be related to the inhibition of prostaglandin synthesis mediated by COX-1 suppression.19, 21 However, in vitro studies have indicated that, not only NSAIDs, but also selective COX-2 inhibitors reduce the production of prostanoids by colonic biopsies from inflammatory bowel disease patients and controls.37 This in vitro finding suggests that selective COX-2 inhibitors may also induce gastrointestinal lesions mediated by lower prostaglandin levels. Moreover, although the expression of COX-1 is comparable in colonic epithelial cells from inflammatory bowel disease patients and controls, COX-2 expression is induced only in inflammatory bowel disease, suggesting its role in tissue healing.38 In accordance with this, lower gastrointestinal events (42.7% of all gastrointestinal events) have been reported in rheumatoid arthritis patients using rofecoxib, although at a frequency 54% lower than in patients taking NSAIDs (naproxen).14 In contrast with NSAIDs, rofecoxib (50 mg) shows no effect on the small intestinal permeability.19, 21 This mechanism therefore does not appear to be related to COX-2 inhibitor-induced inflammatory bowel disease relapse. Certain risk factors (NSAIDs, steroids, age > 65 years, Helicobacter pylori) increase COX-2 inhibitor-induced gastrointestinal lesions.19, 39, 40 Moreover, as COX-2 is constitutively expressed in the kidneys, being involved in renin release and sodium handling,41 COX-2 inhibitors may affect renal function, inducing salt and water retention.42 This effect may be worsened by the concomitant use of steroids. These observations suggest the need for careful follow-up of older inflammatory bowel disease patients taking both COX-2 inhibitors and steroids, who are at higher risk of developing gastrointestinal and cardiovascular events.

Unfortunately, no controlled trials have compared the frequency of inflammatory bowel disease relapse induced by COX-2 inhibitors and NSAIDs. Therefore, no appropriate comparisons can be made between the two types of drug in this regard. Controlled trials in rheumatoid arthritis patients and controls have indicated a lower toxicity of COX-2 inhibitors.9–11, 27, 29 The results from the present study cannot be compared with previous investigations using NSAIDs in inflammatory bowel disease, as they have included mainly retrospective analyses and have used different study designs, patient populations and end-points. Nevertheless, one prospective study similar to ours reported that the use of inhibitors of prostaglandin synthesis (including paracetamol) was the only event that was more frequent in inactive ulcerative colitis patients undergoing relapse (76%) vs. patients maintaining remission (39%; P < 0.01).43 The percentage of patients relapsing on rofecoxib in our study was lower than that found in the study using prostaglandin synthesis inhibitors (76%), when considering all inflammatory bowel disease patients (20%) and ulcerative colitis patients only (20%).43 Preliminary data from a prospective study, however, also indicate a lower (25%) frequency of inflammatory bowel disease relapse when using NSAIDs.44

In conclusion, the results from the first prospective open-label study investigating the safety and efficacy of rofecoxib in inactive inflammatory bowel disease patients in comparison with controls suggest that almost two-thirds of patients show improvement in arthralgia with this COX-2 inhibitor. However, clinical relapse of inflammatory bowel disease requiring drug discontinuation was observed during the first few days of treatment in almost one-quarter of patients. The present findings also suggest that the use of the selective COX-2 inhibitor, rofecoxib, may be safer than NSAIDs in patients with clinically inactive inflammatory bowel disease, although comparative controlled trials are required in order to address this issue. Therefore, there is a need for careful follow-up of inactive inflammatory bowel disease patients during the first few days of treatment with rofecoxib, as a result of possible clinical relapse requiring drug discontinuation.

Ancillary