Gastrointestinal Safety of Coxibs and Outcomes Studies: What s the Verdict?

Vol. 23 No. 4S April 2002 Journal of Pain and Symptom Management S5 Proceedings from the Symposium The Evolution of Anti-Inflammatory Treatments in Arthritis: Current and Future Perspectives Gastrointestinal Safety of Coxibs and Outcomes Studies: What s the Verdict? Loren Laine, MD GI Division, Department of Medicine, University of Southern California School of Medicine, Los Angeles, California, USA Abstract Although nonsteroidal anti-inflammatory drugs (NSAIDs) are widely used, their main limitation is gastrointestinal (GI) injury. Two double-blind, randomized, outcomes trials were conducted to determine the incidence of clinical GI events with the coxibs, rofecoxib and celecoxib, compared with nonselective NSAIDs. The VIGOR study (VIOXX Gastrointestinal Outcomes Research) compared rofecoxib with naproxen, and the CLASS study (Celecoxib Long-term Arthritis Safety Study) compared celecoxib with ibuprofen and diclofenac. The VIGOR trial revealed a relative risk reduction for clinical upper GI events of 50% with rofecoxib, and a 60% reduction in complicated events. In the CLASS study, the 23% reduction in complicated ulcers with celecoxib did not reach statistical significance (P 0.45), although when all clinical events were examined, the 34% reduction was significant (P 0.04). However, low-dose aspirin use, which was allowed in the CLASS study, may have influenced the results. A subgroup analysis in the patients who did not take aspirin revealed a nonsignificant 45% reduction in complicated events with celecoxib (P 0.19), and a 47% reduction in complicated and symptomatic ulcers (P 0.02). J Pain Symptom Manage 2002;23:S5 S10. U.S. Cancer Pain Relief Committe, 2002. Key Words Celecoxib, coxibs, gastrointestinal, rofecoxib, safety, COX-2 inhibitors Introduction Nonsteroidal anti-inflammatory drugs (NSAIDs) are among the most widely used drugs in the world. They are effective in the treatment of pain and inflammation. The main factor limiting their use is concern about the development of gastrointestinal (GI) tract injury. The Address correspondence to: Loren Laine, MD, GI Division, Department of Medicine, University of Southern California School of Medicine, 2025 Zonal Avenue, Los Angeles, CA 90033, USA. Accepted for publication: November 19, 2001. primary reason for the development of cyclooxygenase (COX)-2-specific inhibitors, or coxibs, was to decrease the risk of GI toxicity. Earlier studies have documented that coxibs, in contrast to traditional nonselective NSAIDs, do not inhibit prostaglandin production in the gastric mucosa, do not increase fecal red blood cell loss, and do not increase intestinal permeability. 1 3 Furthermore, large, double-blind, randomized, endoscopic trials in patients with arthritis document that the coxibs, rofecoxib and celecoxib, significantly decrease rates of gastric and duodenal ulcers at 3 to 6 months as compared with nonselective NSAIDs, with rates similar to placebo at 3 months. 4 7 U.S. Cancer Pain Relief Committee, 2002 0885-3924/02/$ see front matter Published by Elsevier, New York, New York PII S0885-3924(02)00368-8

S6 Laine Vol. 23 No. 4S April 2002 What health care providers and their patients are most concerned with is the development of clinically relevant GI tract lesions (i.e., ulcers with bleeding, perforation, or obstruction, or symptomatic ulcers that lead a physician in standard clinical practice to perform a diagnostic evaluation such as endoscopy). Although endoscopic ulcers are likely to be a surrogate for the development of clinically relevant ulcers, only large-scale outcomes studies can directly evaluate whether coxibs really do decrease the incidence of clinical GI events as compared with nonselective NSAIDs. Two landmark studies that assessed whether coxibs improve GI safety in terms of clinical GI events have recently been published. 8,9 This article will review the design and results of those two studies. Coxib GI Outcomes Studies: VIGOR and CLASS Both studies were double-blind, randomized trials in approximately 8000 patients. The study assessing celecoxib was the Celecoxib Longterm Arthritis Safety Study (CLASS) trial, 8 and the study evaluating rofecoxib was the VIOXX Gastrointestinal Outcomes Research (VIGOR) trial. 9 In both studies, twice the maximum approved chronic dose of the coxib was compared with standard doses of nonselective NSAID comparators. The higher doses were chosen to provide a more rigorous evaluation of the hypothesis that coxibs will improve GI safety. As shown in Table 1, the VIGOR study included only patients with rheumatoid arthritis, whereas 73% of the patients in the CLASS study had osteoarthritis and 27% had rheumatoid arthritis. 8,9 The NSAID comparator in the VIGOR study was naproxen, and the NSAID comparators in the CLASS study were ibuprofen and diclofenac. The most important difference in the design of the trials is related to the use of low-dose aspirin. Because low-dose aspirin significantly increases the risk of GI complications, 10 use of this agent could increase the risk of GI events in both study groups, potentially obscuring a difference between the coxib and nonselective NSAID groups. For this reason, use of low-dose aspirin was not allowed in the VIGOR trial. On the other hand, low-dose aspirin was allowed in the CLASS study to better simulate a real-world experience, and 21% of patients in the CLASS study took low-dose aspirin. The median duration of the VIGOR study was 9 months, with a maximum of 13 months. The investigators for the CLASS study analyzed and published on GI events for a 6-month period, although the CLASS trial extended for a longer period, with median and maximum durations of 9 and 13 months, respectively. Full CLASS trial results later became available through the US Food and Drug Administration (FDA). 11 The primary end points and secondary end points in the two studies were similar. In the VIGOR study, the primary end point was all clinical upper GI events (bleeding, perforation, obstruction, and symptomatic ulcers), and the subset of the most serious complicated events (hemodynamically significant bleeding, perforation, and obstruction) constituted the secondary end point. 9 In contrast, the primary end point of the CLASS study was complicated ulcers, and the secondary end point included complicated and symptomatic ulcers. 8 Methodologic Assessment of Trials A number of trial characteristics should be evaluated when assessing methodologic quality Table 1 Study Designs of VIGOR (rofecoxib) and CLASS (celecoxib) Trials 8,9 Parameter VIGOR (N 8076) CLASS (N 7968) Drug/Dose Rofecoxib 50 mg QD Celecoxib 400 mg BID Patient type RA OA (73%), RA (27%) Comparator/Dose Naproxen 500 mg BID Ibuprofen 800 mg TID Diclofenac 75 mg BID Low-dose aspirin No Yes (21%) Duration Median: 9 months 6 months reported Maximum: 13 months (median: 9 months, maximum: 13 months) BID twice daily; CLASS Celecoxib Long-term Arthritis Safety Study; OA osteoarthritis; QD once daily; RA rheumatoid arthritis; TID 3 times daily; VIGOR VIOXX Gastrointestinal Outcomes Research.

Vol. 23 No. 4S April 2002 GI Safety of Coxibs and Outcomes Studies S7 Table 2 Baseline Characteristics of Patients in the VIGOR Trial (Adapted from Ref. 9, with permission) Characteristic Rofecoxib (N 4047) Naproxen (N 4029) Mean age (years) 58 58 Female (%) 80 80 Prior UGI event (%) 8 8 Steroid use (%) 56 56 UGI upper gastrointestinal; VIGOR VIOXX Gastrointestinal Outcomes Research. of therapeutic trials. First, were patients randomized and was the allocation concealed? It is important to be certain that investigators enrolling patients in the trial do not know what the next treatment to be assigned is, because this can introduce bias into the study. Both VIGOR and CLASS did have concealed allocation of randomization. 8,9 In addition, the patients, those caring for the patients, and those gathering data on the patients should be blinded to the treatment. In both studies, appropriate blinding was carried out. Full accounting of patients during the course of the study is also characteristic of a study of highest methodological quality. The number of withdrawals and the reasons for each withdrawal with end point information should be reported. In the VIGOR study, the rates of discontinuation were similar between the two groups with 29% of patients from each group withdrawing from the study 16% for adverse events, 9 6% for a lack of efficacy, and 7% for other reasons. In the CLASS study, 57% of patients withdrew from the study 24% for adverse events, 18% for lack of therapeutic efficacy, and 14% for noncompliance. 11 Related to the issue of full accounting of patients is the type of analysis performed. The analyses in the VIGOR trial included all patients randomized. 9 The CLASS study excluded patients not receiving study medication and patients who developed GI events during the first two days of the study. 11 When evaluating the results of a study, one should initially check the number of patients screened for participation in the study. If a large proportion of those screened were not enrolled, then one may question the generalizability of the study. However, in both the VIGOR and CLASS studies, more than 80% of those screened were included in the study. 8,9 In the CLASS study, 91 of the 8059 patients randomized did not receive study medications and were excluded from subsequent analyses. The next step in the methodologic assessment is to determine whether the study groups were comparable to confirm appropriate randomization. It is important to know that cointerventions, which could affect the primary outcome (e.g., aspirin use, corticosteroid use), also were comparable between the two groups. As shown in Tables 2 and 3, the coxib and nonselective NSAID groups in both studies were extremely well matched. 8,9 Methodological assessment of the mechanism for determining the primary end point is also important. In both cases, objective criteria were predefined for the end points, and committees blinded to the treatment adjudicated them. Studies should also provide information on the number of potential events reported by investigators and examined by the end point committees as well as the number actually included as confirmed events. For example, if investigators reported that 1000 patients experienced events, but 900 of these patients were excluded, we might have concern about the meaning and generalizability of the results. In the VIGOR study, investigators reported that 190 patients had upper GI events and 177 Characteristic Table 3 Baseline Characteristics of Patients in the CLASS Trial (Adapted from Ref. 8, with permission) Celecoxib (N 3987) Ibuprofen/Diclofenac (N 3981) Mean age (years) 61 60 Female (%) 69 69 Past ulcer/gi bleed (%) 8/2 8/2 Steroid use (%) 31 30 Hp seropositive (%) 39 38 CLASS Celecoxib Long-term Arthritis Safety Study; GI gastrointestinal; Hp Helicobacter pylori.

S8 Laine Vol. 23 No. 4S April 2002 Table 4 Results of VIGOR Trial: GI End Points (Rates per 100 patient-years) (Adapted from Ref. 9, with permission) GI end point Rofecoxib (N 4047) Naproxen (N 4029) (95% CI) Reduction (%) P Value Clinical UGI events 2.1 4.5 0.5 (0.3 0.6) 50 0.001 Complicated UGI events 0.6 1.4 0.4 (0.2 0.8) 60 0.005 Any GI bleeding 1.1 3.0 0.4 (0.3 0.6) 60 0.001 CI confidence interval; GI gastrointestinal; UGI upper gastrointestinal; VIGOR VIOXX Gastrointestinal Outcomes Research. were confirmed by the independent adjudication committee; 53 of the 177 events were complicated. 9 The end point committee in the CLASS study had 1,527 patients with reported upper GI events, and 111 met criteria for a clinical ulcer. 11 Forty-four of the 111 were complicated and six of these were not included in the analysis because of predefined exclusion criteria. Results of the VIGOR Trial Results from the VIGOR trial are presented in Table 4. 9 The rates are expressed as events per 100-patient years, which is comparable to an annualized incidence, although based on results from a median of 9 months. Significant reductions are seen for all clinical events, complicated events, and GI bleeding episodes. NSAIDs can cause both upper and lower GI tract bleeding, and the VIGOR trial revealed significant decreases in upper GI bleeding and in bleeding beyond the upper GI tract with rofecoxib compared with naproxen. For rofecoxib, the relative risk reduction for clinical upper GI events was 50% and for complicated events, 60%. Expression of results in clinical trials using the number needed to treat (NNT) provides a useful method of putting the results into clinical perspective. The NNT tells us how many patients would need to be treated with a new therapy instead of the control therapy to avert one bad outcome or to achieve one good outcome. The NNT for rofecoxib was 41 (i.e., 41 patients would need to be treated with rofecoxib instead of naproxen to prevent one clinical upper GI event over 1 year). 9 The NNTs for complicated events and any GI bleeding episode were 125 and 52, respectively. To put the use of NNTs into context, we can examine results from other medical situations. A review of the four large-scale, double-blind, placebo-controlled trials of statin drugs in patients with or without coronary heart disease or hypercholesterolemia revealed that the NNT for prevention of a coronary event ranged from 63 to 256 in 1 year. 12 Determining if a new therapy works only in specific subgroups is also important. For example, past history of GI events, older age, and steroid use are risk factors for NSAID-associated complications. 13 We want to know whether coxibs will have similar efficacy in patients with and without high-risk features. The treatment benefit of rofecoxib compared with naproxen was maintained across all subgroups analyzed. Subgroup analysis in the VIGOR study showed no significant differences, except between Helicobacter pylori (Hp)-positive and Hp-negative patients. 9 An even greater decrease in GI events with rofecoxib was demonstrated in Hp-negative patients compared with Hp-positive patients. GI End Point Table 5 Results of CLASS Trial: GI End Points (Rates per 100 Patient-Years) 11 Celecoxib (N 3987) NSAIDs (N 3981) (95% CI) Relative Risk Reduction P Value Complicated ulcers 0.7% 1.0% 0.77 (0.41 1.46) 23% 0.45 Complicated symptomatic ulcers 1.9% 2.8% 0.66 (0.45 0.98) 34% 0.04 CI confidence interval; CLASS Celecoxib Long-term Arthritis Safety Study; GI gastrointestinal; NSAIDs nonsteroidal anti-inflammatory drugs.

Vol. 23 No. 4S April 2002 GI Safety of Coxibs and Outcomes Studies S9 Table 6 Results from CLASS Trial: GI End Points of Patients Not Using Aspirin (Rates per 100 Patient-Years) 11 GI End Point Celecoxib (N 3105) NSAIDs (N 3124) (95% CI) Relative Risk Reduction P Value Complicated ulcers 0.4% 0.8% 0.55 (0.23 1.30) 45% 0.19 Complicated symptomatic ulcers 1.2% 2.2% 0.53 (0.31 0.90) 47% 0.02 CI confidence interval; CLASS Celecoxib Long-term Arthritis Safety Study; GI gastrointestinal; NSAIDs nonsteroidal anti-inflammatory drugs. Results of the CLASS Trial Results from the full analysis of the CLASS trial are shown in Table 5, presented as rates per 100 patient-years. 11 The reduction in complicated events, the primary end point, did not quite reach the traditional level of statistical significance, although when all clinical events were examined, the 34% risk reduction was significant (P 0.04), with an NNT of 111 patients. However, as mentioned above, low-dose aspirin use was allowed in the CLASS study, and may have influenced the results. Therefore, a subgroup analysis was conducted, which included the 79% of patients who did not take aspirin (Table 6). 11 The 45% relative risk reduction in complicated events with celecoxib in those not using aspirin was not statistically significant (P 0.19), but the significant benefit in terms of all clinical upper GI events remained (RR 0.53 [0.31 0.90], P 0.02). The NNT for prevention of a clinical upper GI event (symptomatic and complicated ulcers) with celecoxib was 100. There were no significant differences between the celecoxib and nonselective NSAID groups among the 21% of study patients taking low-dose aspirin. However, this study was markedly underpowered to study the effects of celecoxib versus nonselective NSAIDs in patients taking low-dose aspirin. Such a study would require more patients than either the CLASS or VIGOR study. Conclusions The results of the VIGOR and CLASS GI outcomes studies demonstrate that coxibs reduce clinical upper GI events by approximately 50% to 60%, as compared with traditional nonselective NSAIDs, in arthritis patients who do not take aspirin. The number needed to treat with a coxib instead of a nonselective NSAID to prevent a clinical upper GI event is approximately 40 to 100. References 1. Cryer B, Gottesdiener K, Gertz B, et al. In vivo effects of rofecoxib, a new cyclooxygenase (COX)-2 inhibitor, on gastric mucosal prostaglandin (PG) and serum thromboxane B 2 (TXB 2 ) synthesis in healthy humans [abstract]. Gastroenterology 1999; 116:A141. 2. Hunt RH, Bowen B, Mortensen ER, et al. A randomized trial measuring fecal blood loss after treatment with rofecoxib, ibuprofen, or placebo in healthy subjects. Am J Med 2000;109:201 206. 3. Bjarnason I, Sigthorsson G, Crane R, et al. COX-2 specific inhibition with MK-0966 25 or 50 mg Q.D. does not increase intestinal permeability: a controlled study with placebo (PBO) and indomethacin 50 mg T.I.D (INDO) [abstract]. Am J Gastroenterol 1998;93:1670. 4. Laine L, Harper S, Simon T, et al. for the Rofecoxib Osteoarthritis Endoscopy Study. A randomized trial comparing the effect of rofecoxib, a cyclooxygenase 2-specific inhibitor, with that of ibuprofen on the gastroduodenal mucosa of patients with osteoarthritis. Gastroenterology 1999;117:776 783. 5. Hawkey C, Laine L, Simon T, et al. for the Rofecoxib Osteoarthritis Endoscopy Multinational Study. Comparison of the effect of rofecoxib (a cyclooxygenase 2 inhibitor), ibuprofen, and placebo on the gastroduodenal mucosa of patients with osteoarthritis: a randomized, double-blind, placebocontrolled trial. Arthritis Rheum 2000;43:370 377. 6. Simon LS, Weaver AL, Graham DY, et al. Antiinflammatory and upper gastrointestinal effects of celecoxib in rheumatoid arthritis: a randomized controlled trial. JAMA 1999;282:1921 1928. 7. Emery P, Zeidler H, Kvien TK, et al. Celecoxib versus diclofenac in long-term management of rheumatoid arthritis: randomised double-blind comparison. Lancet 1999;354:2106 2111. 8. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis. The CLASS study: a randomized controlled trial. JAMA 2000;284:1247 1255. 9. Bombardier C, Laine L, Reicin A, et al. for the VIGOR Study. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in

S10 Laine Vol. 23 No. 4S April 2002 patients with rheumatoid arthritis. N Engl J Med 2000;343:1520 1528. 10. Weil J, Colin-Jones D, Langman M, et al. Prophylactic aspirin and risk of peptic ulcer bleeding. BMJ 1995;310:827 830. 11. CLASS Advisory Committee. Briefing Document. Available at: www.fda.gov/ohrms/dockets/ac/01/ briefing/367761_01_searle.pdf. Accessed Feb. 8, 2001. 12. Kumana CR, Cheung BMY, Lauder IJ. Gauging the impact of statins using number needed to treat. JAMA 1999;282:1899 1901. 13. Gabriel SE, Jaakkimainen L, Bombardier C. Risk factors for serious gastrointestinal complications related to use of nonsteroidal antiinflammatory drugs: a meta-analysis. Ann Intern Med 1991; 115:787 796.