Author + information
- William B White, MDa ()
In the recent perspective published by Bing and Lomnicka in the Journal(1), several hypotheses were given for why cyclooxygenase-2 (COX-2) inhibitors may cause cardiovascular events. The investigators even stated in their abstract that their report “confirms evidence that selective nonsteroidal anti-inflammatory drugs [NSAID] such as celecoxib can lead to thrombotic cardiovascular events.” In fact, there are no data provided by Bing and Lomnicka (1) nor from the clinical literature that clinical cardiovascular events, defined as acute myocardial infarction, stroke and cardiovascular death, are increased owing to the COX-2 inhibitor, celecoxib. Thus, I believe that their study is potentially misleading to the readership of the Journal.
Using well-known basic pharmacology literature as a resource, Bing and Lomnicka (1) stated that selective COX-2 inhibitors attenuate the production of prostacylin, but do not alter thromboxane A2 levels and therefore may theoretically tip the balance in favor of thrombosis. Thus, certain types of high-risk patients treated with COX-2 inhibitors could be predisposed to increases in cardiovascular events. To further support their hypothesis, however, they use the highly controversial post hoc analysis of data from Mukherjee et al. (2), which suggested that the COX-2 inhibitors celecoxib and rofecoxib had a higher myocardial infarction event rate compared to an entirely unrelated, separate cohort of generally healthy individuals in the placebo arm of four primary prevention trials using aspirin (2). The sources of the pooled analyses for the COX-2 inhibitors from the analysis of Mukherjee et al. (2) derived from the Celecoxib Long-term Arthritis Safety Study (CLASS) (3) and Vioxx Gastrointestinal Outcomes Research (VIGOR) (4) trials using celecoxib and rofecoxib, respectively, and two clinical trials that compared rofecoxib with a nonselective NSAID, nabumetone. The CLASS and VIGOR trials were conducted in approximately 8,000 arthritis patients each and compared the gastrointestinal safety of the COX-2 inhibitors versus the widely used NSAIDs, ibuprofen and diclofenac (in CLASS) and naproxen (in VIGOR) for a median period of about nine months in each trial.
A number of errors made by Mukherjee et al. (2) have now been documented by numerous letters to the editor of JAMA in December 2001. For example, patients in the CLASS trial who were treated with low-dose aspirin (owing to prior cardiac or cerebrovascular disorders) were compared to placebo patients who had no known prior history of myocardial infarction (MI) from four primary prevention trials evaluating the beneficial effects of aspirin. The annual MI rates reported for celecoxib in CLASS and rofecoxib in VIGOR by Mukherjee et al. for the entire group were 0.7% to 0.8% compared to a rate of 0.52% of MI observed in the placebo group from the primary prevention trials. When the patients who were nonusers of aspirin in the CLASS trial (about 3,100 patients) on celecoxib were assessed, the incidence of MI was just 0.3%.
We recently reported on an extensive analysis of these thrombovascular events in the CLASS trial (5); that study showed no evidence that high doses of celecoxib (400 mg twice daily) increased the risk of acute MI, stroke, or venous thromboembolic events compared to the conventional NSAIDs, ibuprofen or diclofenac. This was true for the entire study population, both in patients not taking aspirin and in patients taking aspirin. Similarly, there have been no data from the premarketing clinical trials that show an increase in cardiovascular events on celecoxib versus placebo nor celecoxib versus other nonselective NSAIDs (6) . One would have to conclude, then, that there are no clinical outcome data that support the hypothesis and statement of Bing and Lomnicka (1) that the COX-2 inhibitor celecoxib causes cardiovascular events.
- American College of Cardiology Foundation