Parecoxib – getting to the heart of the matter

Abstract

Merck Sharp & Dohme voluntarily withdrew rofecoxib (Vioxx®; Merck Sharp & Dohme Ltd, West Drayton, UK), its first cyclo-oxygenase-2 (COX-2) inhibitor, in September 2004 following the unexpected finding of a doubling of the risk of cardiovascular events in patients after 18 months of treatment [1]. Inevitably, this has provoked much comment in both the scientific journals and in the lay press regarding the safety of other drugs in this class. Much of the data generated by the subsequent clinical trials has concentrated on the consequences of the chronic use of rofecoxib and celecoxib (Celebrex®; Pfizer Ltd, Tadworth, UK). However, a recent study by Lévesque and colleagues has found that the cardiovascular risk associated with rofecoxib occurs over a much shorter time period than was previously thought [2]. Parecoxib (Dynastat®; Pfizer Ltd) continues to be licensed in the United Kingdom for use in the short-term management of postoperative pain. Although no trials have been published which focus on the safety of parecoxib during short-term peri-operative use, absence of evidence is not evidence of absence. In light of this recent finding, is it wise to continue treating patients with a drug whose safety in this setting is unclear? Parecoxib is the intravenous prodrug of valdecoxib (Bextra®; Pfizer Ltd) and is the only injectable COX-2 inhibitor currently available. Parecoxib exerts its analgesic action by conversion to valdecoxib, which is known to have comparable COX-2 selectivity to rofecoxib [3]. The marketing and sale of valdecoxib itself, however, were suspended by Pfizer in April 2005, both in the European Union and in the United States, in order for risk-benefit discussions to be carried out with the US Food and Drug Administration. The theory behind the action of selective and non-selective non-steroidal anti-inflammatory drugs (NSAIDs) is well known [4, 5]. Selective inhibition of COX-2 suppresses prostacyclin, a vasodilator and platelet inhibitor, but leaves the actions of thromboxane-A2, responsible for platelet aggregation and vasoconstriction, unopposed. This has the potential to produce a prothrombotic state, with platelet aggregation, vasoconstriction and ultimately thrombosis. The prothrombotic tendency associated with use of the COX-2 inhibitors first became apparent in 2000 when the Vioxx Gastrointestinal Outcomes Research Study (VIGOR; 8076 patients) [6] demonstrated that the relative risk of developing a thrombotic cardiovascular event with rofecoxib compared with naproxen was 2.38 (95% confidence interval (CI) 1.39–4.00). The Adenomatous Polyp Prevention on Vioxx Study (APPROVe; 2586 patients) [7] and the Adenoma Prevention with Celecoxib Study (APC; 2035 patients) [8] were both stopped early when it was found that the patients in these studies had substantially increased relative risks of cardiovascular events. In the former study the relative risk of death from cardiovascular causes or myocardial infarction was 3.4 (95% CI 1.2–9.7), whilst in the latter the risk of a thrombotic cardiovascular event was 2.80 (95% CI 1.44–5.45). Both studies demonstrated increased cardiac risk after at least 12 months of treatment, with higher risk at higher doses. The authors of the VIGOR study commented that 38% of the patients who suffered myocardial infarction met the criteria for low-dose aspirin therapy, but were not taking it. In each of these three studies the patients who developed thrombotic events were those with more cardiovascular risk factors at the outset. Until Lévesque and colleagues' recent publication it has been assumed that much of the cardiovascular risk associated with COX-2 inhibitors was related to chronic use. Their study found that, in elderly patients (≥ 66 years) without previous myocardial infarction, the risk of having a myocardial infarction was greatest within 6–13 days (median 9 days) of first-time rofecoxib users starting treatment (relative risk (RR) 1.67, 95% CI 1.21–2.30) [2]. This risk continued for the first 7 days after drug discontinuation (RR 1.23, 95% CI 1.05–1.44). The risk was not shown to increase with increased length of treatment, perhaps as a result of those at greatest susceptibility already having died. The increase in thrombotic risk in first-time celecoxib users was not significant (RR 1.29, 95% CI 0.90–1.83). Comparisons can be drawn with the VIGOR, APPROVe and APC studies as the patients suffering myocardial infarctions were more likely to be at increased risk for thrombotic events due to the presence of underlying atherosclerotic disease. A Danish population-based case-control study published in 2005 identified increased early risk of myocardial infarction in new users (= 30 days) of rofecoxib and celecoxib compared with non-users [9]. The relative risks were adjusted for various, but not all, cardiovascular risk factors and were 2.52 (95% CI 1.74–3.64) and 2.13 (95% CI 1.45–3.13), for rofecoxib and celecoxib, respectively. The evidence that these adverse cardiovascular events represent a class effect of COX-2 selective drugs, rather than a unique effect of rofecoxib itself, is becoming increasing persuasive. If this is indeed the case, then surely patients are at a similar risk of cardiovascular events following a few days of peri-operative parexcoxib administration. It could even be argued that the risk in these patients may be greater as they are more likely to be in a stress-induced hypercoaguable state. The current evidence available to answer this question provides conflicting conclusions. To date, three randomised placebo-controlled trials have been conducted to assess the safety of parecoxib and valdecoxib; two following coronary artery bypass grafting (CABG) surgery [10, 11] and one following non-cardiac surgery [12]. The first CABG study (462 patients) allocated patients to receive either IV parecoxib (40 mg twice daily (BD)) for a minimum of 3 days followed by oral valdecoxib (40 mg BD), or placebo/placebo in a 14-day trial. The incidence of cerebrovascular events and myocardial infarctions was greater in the parecoxib/valdecoxib group, but the difference did not achieve statistical significance [10]. The second, larger CABG study (1671 patients) allocated patients to one of three groups; IV parecoxib (initial dose of 40 mg, then 20 mg BD) for a minimum of 3 days followed by oral valdecoxib (20 mg BD) for the remainder of the 10-day treatment period, IV placebo followed by oral valdecoxib, or IV placebo followed by oral placebo. The relative risk of a cardiovascular event, including myocardial infarction, cardiac arrest and pulmonary embolism, in the patients given parecoxib and valdecoxib compared with placebo, was 3.7 (95% CI 1.0–13.5) [11]. The third trial (1062 patients) was conducted on patients undergoing major orthopaedic, abdominal, gynaecological or non-cardiac thoracic surgery. Patients were randomly allocated into two groups; IV parecoxib (initial dose of 40 mg, then 20 mg BD) for a minimum of 3 days followed by oral valdecoxib (20 mg BD) for the remainder of the 10-day period, or IV placebo followed by oral placebo. No significant differences in the overall safety profile were found between the two groups, although it is worth noting that the majority of these patients were not considered to be at increased risk for cardiovascular disease [12]. A meta-analysis presented at the recent European Society of Anaesthesiology Annual Meeting examined the incidence of thrombotic cardiovascular events associated with the short-term (= 10 days) use of parecoxib compared with placebo after major surgery [13]. Patients were stratified for pre-existing cardiovascular risk factors. Nineteen studies (including the two aforementioned CABG studies) were reviewed and the authors calculated that the analysis had a 92% power to detect a two-fold increase in thrombotic events from an assumed baseline rate of 1%. No statistically significant differences in total or individual cardiovascular thrombotic events between parecoxib and placebo were found. A further review was carried out, after excluding the two CABG studies, with 77% power to detect the same outcome. The incidence of cardiovascular events did not differ significantly between the two groups; 0.44% (13/2966) for parecoxib and 0.37% (7/1915) for placebo. In addition, following stratification for the number of pre-existing cardiovascular risk factors, no significant differences in the 17 non-cardiac studies were found. These analyses, however, were only powered to detect a two-fold increase in the incidence of cardiovascular thrombo-embolic events; this may result in smaller, but clinically important, increases being missed. The total number of cardiovascular events available for analysis was also relatively small. Other short-term studies have been conducted with COX-2 selective drugs in non-cardiac (gynaecological, orthopaedic and laparoscopic abdominal) patients with the principal aim of assessing postoperative analgesia scores [14–16]. Although no serious adverse cardiovascular events were uncovered, these studies were not primarily designed or powered to determine cardiovascular risk and clinically meaningful differences in adverse events may not have been detected. What do the results of these studies actually tell us? The increased relative risk of thrombotic cardiovascular events has now been demonstrated to occur with the long-term use of rofecoxib and celecoxib, and the relatively short-term use of parecoxib/valdecoxib in patients with known atherosclerotic heart disease undergoing CABG. So, in the first instance the recommendation that patients with cardiovascular disease should not be treated with COX-2 selective inhibitors seems clear. Furthermore, it may be that the concomitant use of aspirin will confer some protection for these higher risk patients; however, using an additional non-selective NSAID will negate the much sought-after gastrointestinal advantages. Regrettably, it remains unclear whether parecoxib, our only injectable COX-2 selective inhibitor, is safe for short-term peri-operative use in patients at low risk of cardiovascular events. A study designed to gather this evidence would not be an easy undertaking. To ensure sufficient power to detect relatively small numbers of adverse thrombotic events in a population at a low level of risk, a very large number of patients would need to be randomised. Alternatively, waiting for the results of voluntary reporting, such as the ‘yellow card system’, will be lengthy and unlikely to be representative of the true scale of the problem. It is important, of course, to bear in mind that many therapies have risks and side-effects, and for certain patient populations the available alternatives may be worse. In fact, recent research shows that even the traditional NSAIDs are associated with moderate increases in the risk of thrombotic vascular events [17]. It would seem prudent, however, that until data are produced which support the safety of parecoxib in the peri-operative setting, careful consideration of the risk/benefit ratio for each individual patient is required.