Schug S. et al

◊ BACKGROUND:
Studies of parecoxib, the inactive prodrug of the cyclooxygenase-2 selective inhibitor valdecoxib, and valdecoxib for postoperative pain relief in patients undergoing coronary artery bypass graft surgery revealed an increased risk of cardiovascular (CV) adverse events compared with placebo. We conducted this study to address whether parecoxib and valdecoxib increased CV risk in noncardiac surgery patients.

◊ METHODS:
A pooled post hoc analysis was conducted using 2 large datasets: 17 controlled trials of parecoxib for noncardiac studies and 32 studies, including the 17 noncardiac parecoxib studies plus 15 studies of valdecoxib. The 32-study dataset provided 95% power to detect a twofold increase in the incidence of CV adverse events assuming a placebo group incidence of 1% (estimated from previous study data), and 69% power to detect a twofold increase from a 0.5% incidence.

◊ RESULTS:
The incidence of total CV events for the 17 parecoxib studies was 0.44% (13 of 2966) in patients who received parecoxib and 0.37% (7 of 1915) in those receiving placebo (P > 0.20). In the analysis of 32 studies, the incidence of total CV events was 0.40% (21 of 5285) in the parecoxib/valdecoxib group compared with 0.50% (16 of 3226) in the placebo group (P > 0.20). No significant differences in the incidence of total or any individual CV event category were observed between the parecoxib or parecoxib/valdecoxib and placebo groups in the two analyses. When patients were stratified by number of baseline CV risk factors, no significant difference in CV events was detected in parecoxib/valdecoxib patients compared with placebo.

◊ CONCLUSION:
In the largest analysis of the CV risk of cyclooxygenase selective inhibitors or nonsteroidal antiinflammatory drugs for perioperative pain management, parecoxib and valdecoxib were not found to increase the risk of CV adverse events after noncardiac surgery.

◊ Reviewed by A. Gupta, MD, PhD, FRCA
Associate Professor, Dept. of Anesthesiology and Intensive Care,
University Hospital, Örebro, Sweden

The study is well performed and the authors should be given credit for their thorough and painstaking extraction of data from these studies. The take-home message is that parecoxib/valdecoxib can be given to patients undergoing non-cardiac surgery, irrespective of cardiovascular risk factors based on a similar incidence of CV adverse events in those patients compared to placebo. Unfortunately, as in all retrospective studies and meta-analyses, the data is obtained from a mixed population of patients undergoing a variety of surgeries (minor or intermediate risk), receiving different doses of drugs (20 – 80 mg), comprising a wide age range (18-96 years) and predominantly women (75%). Furthermore, the present study is based on integrated data obtained from Pfizer Inc. until end 2004. We do not know if the dataset just comprised published studies or whether unpublished data from Pfizer has also been included. In addition, two of the authors are employed by Pfizer and all remaining authors have received research grants and honoraria from Pfizer.

There are some further concerns that I believe should be addressed. First, the negative results of this study (no difference between parecoxib/valdecoxib and placebo) arise from the calculation of statistical power of 80%, which was based on a demonstrated baseline CV event rate of 1% and its doubling when using parecoxib/valdecoxib. However, the event rate in this population of patients undergoing non-cardiac surgery was lower (approx. 0.5%), which, according to the authors own calculations, would give a power of only 69% for a two-fold increase in CV events. In other words, this study also was underpowered, and a substantially greater number of patients would have been needed to achieve adequate power. The authors admit this by stating that “the ability to make a definitive statement regarding the CV safety of parecoxib/valdecoxib is somewhat limited”.

Second, most patients included in this study underwent minor surgery associated with a low risk of CV events. In Table 2 the authors include, under major non-cardiac surgery, patients undergoing inguinal hernia repair as well as laparoscopic cholecystectomy, which I would certainly not consider a major surgery. Although it might be important to study patients at low risk for CV events, the number of patients needed for statistical reasons increases dramatically. Therefore, and in order to exclude the occurrence of a rare event, it is probably better to study a group of patients at greater risk, such as patients undergoing cardiac or vascular surgery. Indeed, previous studies on just this group of patients do suggest an increased risk of CV events. Therefore, caution should be urged in interpreting data from the present study. Although the presence and the number of CV risk factors did not seem to increase the risk of thromboembolic events, the number of patients studied with risk factors was too small to draw any conclusions.

Finally, although the authors argue against the “imbalance theory” and propose that cardiopulmonary bypass is largely responsible for the increase in thromboembolic events, their arguments are based on just a single study performed on rats. This theory would not explain the substantial body of evidence of an increased risk of CV events (myocardial infarction) in patients, who take COX-2 inhibitors but do not undergo surgery, as well as those who were at high risk and underwent cardiac surgery. It is, however, possible that COX-2 inhibitors vary in their impact on CV events with some (rofecoxib) having greater effects than others (celecoxib). Before we can discount the impact of parecoxib/valdecoxib on CV events in patients undergoing non-cardiac surgery, further larger and preferably prospective studies need to be performed. In the meantime, it is my belief that caution should be used in administering these drugs to patients at risk of thromboembolic events. The risk-benefit of parecoxib/valdecoxib should be assessed in each individual patient before administering these drugs. The data presented by the authors does not support that parecoxib/valdecoxib can be administered for pain relief, without further consideration, in patients undergoing non-cardiac surgery.