Several meta-analyses, varying in size and study design, have been performed. Three of these analyses revealed no increase in CV events with this agent, compared with placebo or other NSAIDs, although one analysis in 2006 did find an increase in the rate of MIs with celecoxib.

In one analysis, OA and RA trials of four weeks or more in 0.70-1.61; P = .79) or placebo (RR, 0.85; 95% CI, 0.23-3.15; P = .81). No differences were seen with the use of ASA or in terms of the underlying arthritis type.

In another meta-analysis, the authors compared tolerability of various doses of celecoxib and associated adverse events in patients with arthritis. They evaluated 31 clinical trials, ranging from 2 to 52 weeks, that had been performed in patients with OA and RA. No difference in the number of MI events was seen with celecoxib (0.18%) versus the comparators (0.12%) (RR, 1.4; 95% CI, 0.88-2.2).

White et al. recently reported their findings of an analysis of celecoxib in more than 41,000 patients at the 2006 American College of Cardiology meet-ing. For adjudicated events, celecoxib versus placebo (RR, 1.11; 95% CI, 0.47-2.67; P = .81) or other NSAIDs (RR, 0.90; 95% CI, 0.60-1.33; P = .59) did not show an increase in CV events. These findings were similar to the two prior analyses.

Caldwell et al. evaluated four placebo-controlled trials involving more than 4,000 patients. Several outcomes, including composite CV events and MI rates with celecoxib, were assessed. For the combined CV events, celecoxib did not show an duration were considered, allowing data from 13 New Drug Application (NDA) studies and the Celecoxib Long-term Arthritis Safety Study (CLASS) and Successive Celecoxib Efficacy and Safety Study (SUCCESS) postmarketing trials to be evaluated. A total of 31,879 patients were included in the study analyses. When the same endpoints from the Antiplatelet Tri-alists’ Collaboration were used, no significant differences were found between celecoxib and NSAIDs (RR, 1.06; 95% CI, increased risk (OR, 1.38; 95% CI, 0.91-2.1); however, for MI rates only, it did show an increased risk (OR, 2.26; 95% 1.0-5.1).

In a secondary analysis performed by the same authors, similar results were found in a comparison of celecoxib and other NSAIDs. Unlike prior analyses, this was the first analysis to find an increased risk with celecoxib.

Combined Rofecoxib and Celecoxib

Another analysis, which included the VIGOR and CLASS trials, evaluated the risk of CV events associated with both rofecoxib and canadian celecoxib. Rofecoxib demonstrated an increased risk when compared. In addition, when results of the VIGOR trial were compared to those of a placebo group in a meta-analysis of more than 23,000 patients, the annual MI rate for rofecoxib was significantly higher (0.74% vs. 0.52% for naproxen, respectively) (P = .04).

Unlike earlier reports showing no increased risk of CV events in the CLASS trial, when the authors used the same placebo comparator, the annual MI rate with celecoxib (0.80%) was significantly higher than with placebo (0.52%) (P = .02).

Other Coxibs

Few analyses have been performed to assess the risk of CV events with other coxibs. Two pooled analyses of randomized trials of valdecoxib were performed; one analyis included 8,000 patients with RA or OA, and the other included 3,000 patients with RA. No differences in the incidence of thrombotic events were found between the valdecoxib and comparator groups.

Pooled data were evaluated from more than 6,700 patients receiving in all phase 2, 3, and 4 studies lasting four weeks or more. Etoricoxib drug showed no increased risk of CV events over that for placebo (RR, 1.1; 95% CI, 0.32-3.81), canadian naproxen (RR, 1.70; 95% CI, 0.91-3.18) or other non-naproxen NSAIDs (RR, 0.83; 95% CI, 0.26-2.64).

In a meta-analysis of 22 randomized clinical trials evaluating lumiracoxib, no increased risk of CV events, as defined by the Antiplatelet Trialists’ Collaboration, occurred compared with placebo (RR, 0.88; 95% CI, 0.34-2.25), naproxen (RR, 1.49; 95% CI, 0.94-2.36) or non-naproxen NSAIDs (RR, 0.83; 95% CI, 0.46-1.51).

Considering these analyses, the risk of CV events appears to be highest in patients taking rofecoxib compared with other coxibs. Many of the trials differed in study design, methods of recording safety outcomes, and comparator agents, thereby making it difficult to draw firm conclusions. It is noteworthy that the primary outcomes and study power were based on the assessment of GI risk, a slowing progression of Alzheimer’s disease, treatment of arthritic pain, or the secondary prevention of colorectal cancer—but not CV risk. Therefore, differences noted in these clinical studies and other analyses should be evaluated cautiously. Long-term, randomized clinical trials that are powered to show CV risks are urgently needed.

Possible Explanations for Cardiovascular Outcomes Physiology of COX-2 Inhibition on the Cardiovascular System

As we have begun to understand the physiological mechanisms of COX isoenzymes, an appreciation of the potential negative effects of inhibiting the COX-2 enzyme has been recognized. Theoretical adverse effects include problems with salt and water balance, wound healing, and thrombosis. The COX-1 enzyme is responsible for thromboxane production, the promotion of platelet adhesion, and vasoconstriction; the COX-2 enzyme results in prostacyclin formation and the consequential inhibition of thrombosis and vasodilation (Figure 2).

Figure 2 Nonselective nonsteroidal anti-inflammatory drugs

Figure 2 Nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit both the COX-1 and COX-2 enzymes to varying degrees,whereas coxibs predominantly block the COX-2 pathway. Nonselective NSAIDs can maintain a vasomotor and thrombotic balance,but coxibs may disrupt this balance,allowing for vasoconstriction and thromboxane-induced thrombosis to occur.Coxib = selective cyclooxygenase-2 inhibitor; COX = cyclooxygenase.(Adapted with permission from Bates ER,et al. Circulation 2005;111:267-271. Lippincott Williams & Wilkins.)

Ideally, these enzyme systems maintain a homeostatic balance between bleeding and thrombosis. When nonselective NSAIDs are given, they inhibit both COX-1 and COX-2 activity, and a balance of thrombotic activity is maintained. By contrast, when coxibs are given, they are selective to the COX-2 enzyme, affecting the ability of prostacyclin to limit thrombosis and allowing thromboxane-unopposed thrombosis to occur. These physiological effects may account for a portion of the increase in CV and renovascular events associated with coxibs.