1. Citation
Mukherjee D, Nissen SE, Topol EJ. Risk of Cardiovascular Events Associated With Selective COX-2 Inhibitors. JAMA. 2001;286(8):954-959.
2. Background
The authors posit that COX-2 inhibitors may promote thrombosis by decreasing vasodilatory and antiaggregatory prostacyclin production.
3. Methods
This paper does not examine original data about selective COX-2 inhibitors and thromboembolic disease. Instead the authors evaluate two studies on gastrointestinal disease, one examining gastrointestinal outcomes of rheumatoid arthritis patients treated with COX-2 inhibitors (VIGOR; 8076 patients) and the other looking at safety of COX-2 inhibitors (CLASS;8059 patients).
VIGOR used a very high dose of rofecoxib (50 mg/day) in patients with rheumatoid arthritis, comparing them to patients taking naproxen (1000 mg/day). For this study, patients were not permitted to take aspirin. Although patients who needed aspirin for cardiac reasons were excluded, almost 4% of the studied patients did have indications (prior thromboembolic events) to take aspirin. The rofecoxib-treated patients who did have aspirin indications had a relative risk for a serious cardiovascular event almost five-fold that of the naproxen-treated patients. When there was no aspirin indication, the relative risk vs. naproxen was less than 2, apparently proving that removal of COX-1 inhibition, when it is indicated, is a bad thing. Although the authors present a rather dramatic graph illustrating the difference in cardiovascular event rates between the rofecoxib and naproxen groups, the actual incidence of cardiovascular events in those with and without aspirin indications rose from 0.5% to only 1.1%. The authors comment on the possibility that drug-induced hypertension contributed to the cardiovascular complications in the rofecoxib group, but they make no comment whatsoever about the very high dose of rofecoxib used.
In CLASS, celecoxib (800 mg/day) was compared to ibuprofen (2400 mg/day) and diclofenac (150 mg/day). Again, the supratherapeutic doses received no comment. Low dose aspirin was permitted. In this study, there is no difference between the COX-2 inhibitor and the other drugs.
Mukherjee, et al, also examined two smaller unpublished studies submitted to the Food and Drug Administration. Both had a small number of cardiovascular events and showed no differences in cardiovascular event rates between COX-2 inhibitors and non-selective NSAIDs.
4. Results
The authors conclude that VIGOR showed a relative risk of developing a thrombotic cardiovascular event (myocardial infarction, unstable angina, cardiac thrombus, resuscitated cardiac arrest, sudden or unexplained death, ischemic stroke, and transient ischemic attacks) with rofecoxib compared with naproxen was 2.38, although there was no significant difference in cardiovascular event (myocardial infarction, stroke, and death) rates between celecoxib and NSAIDs in CLASS. Their annualized MI rates for COX-2s in both CLASS and VIGOR were significantly higher than that in the placebo group of a meta-analysis of 23,407 patients in primary prevention trials (0.74% with rofecoxib and 0.80% with celecoxib).
5. Clinical Relevance/Commentary
With the VIGOR and CLASS data, Mukherjee and colleagues try (unsuccessfully in this reviewer's mind) to argue that the COX-2 inhibitors are thrombogenic. To support this argument further they compare the above studies with a meta-analysis of the United States Physicians' Health Study, the United Kingdom Doctors Study, the Thrombosis Prevention Trial, and the Hypertension Optimal Treatment trials. They performed this meta-analysis to establish an expected rate of cardiovascular events in other populations -- an argument so specious that it borders on the sinful. The populations of the meta-analyzed studies differ from those of the comparison studies in being (in some cases) much healthier, in being much more male and in being much more health-conscious. (Two of the studies concern aspirin use in healthy physicians.) One assumes, but the authors do not give us assurance, that the studies were adjusted for age before the comparisons were done. Nonetheless, it is absolutely predictable that the meta-analysis cases would have lower cardiovascular event rates than would the ill populations to whom the COX-2 inhibitors were given. The performance of the meta-analysis not only does not inform the question at hand, it is misleading.
COX-2 inhibitors do not prevent platelet aggregation and should not be used alone when cardioprotection via platelet aggregation inhibition is intended. It is this reviewer's practice to use a small dose of aspirin (81 mg) together with COX-2 inhibitors in patients at risk for hypercoagulability. Furthermore, this author's practice is to use rofecoxib and celecoxib in their authorized doses, 25 and 400 mg daily, respectively, and not in the double-recommended doses reported by Mukherjee, et al.
In other words, the contention that the COX-2 inhibitors are potentially thrombogenic is unsupported by the data offered. The additional faults of reporting on drugs used at twice the recommended dose and of failing to recognize the obvious point that cardioprotection provided by non-selective NSAIDs is not provided by selective COX-2 inhibitors effectively invalidate the conclusion of this study. The Mukherjee study is valuable in reminding physicians to use COX-2 inhibitors in appropriate doses and to add platelet active drugs, such as low dose aspirin, when offering COX-2 inhibitors to patients who might benefit from this protection.
