Ten lessons to be learned from the withdrawal of Vioxx� (rofecoxib)

Abstract

NSAIDs are one of the most ubiquitous class of drugs, with different compounds being sold as over the-counter drugs for a wide range of minor and major conditions. As a group, they rank first among causes of adverse drug reactions (ADRs) [1]. Severe gastrointestinal (GI) toxicity (i.e. upper GI bleeding, ulcers and perforation) due to aspirin and NSAIDs is probably the main cause of iatrogenic admission to hospitals [2], with important quantitative risk differ ences among NSAIDs [3-7] and with an overall attributable incidence of 150 cases/million/year [8]. In 1991, two COX-isozymes of prostaglandin G/H synthase (cyclooxigenase) were first characterized and named COX-1 (the constitutive form) and COX-2 (the inducible form). Inhibition of COX-2 was proved to be most directly implicated in reducing inflam mation, whereas inhibition of COX-1 has been re lated to significant adverse events in the GI tract [9]. This knowledge opened the door to a new class of compounds, the selective COX-2 inhibitors, which, if proved to be safer upon the GI tract that classical NSAIDs, could be potentially important 'blockbust ers' [9]. However, some words of caution are in order. First, selectivity inhibition should be regarded as a relative, rather than an absolute concept; the in vitro COX-1/COX-2 ratio differs from compound to compound (among classical as well as newer drugs); it is also critically dependent on the methodology used as well as on the concentration of the drug [9]. Secondly, NSAIDs as a group have potentially deleterious effects on the cardiovascular system; prostaglandins can counterbalance the vasoconstric tor effects of other mediators as angiotensin II, nor epinephrine or vasopressin on the kidney. They can also induce the appearance of oedemas and fluid retention, cause loss of blood pressure control and increase the risk of congestive heart failure. Thirdly, regarding its effect on platelet function, many physiopathological features of the COX-1/ COX-2 equation are still evolving. The well known antiplatelet effect of aspirin is achieved by irreversible acetylation of a serine residue at position 529 in COX-1, the only isoform of the enzyme expressed in platelets, inhibiting its main metabolite (thrombox ane A2), which has proaggregant properties. Prosta glandin I2, the predominant COX product in endothelium, has opposed effects: it inhibits platelet aggregation, causes vasodilatation and prevents the proliferation of vascular smooth-muscles. The inhi bition of these effects, previously thought to be mediated by COX-1, later has been proved to be COX-2 related [10]. If this is so, strong inhibition of COX-2 dependent prostaglandin I2 in endothelium could lead to cardiovascular damage. In 1999, the two first anti-inflammatory drugs based upon the concept of COX-2 selectivity, the so called 'coxibs' (rofecoxib and celecoxib) were intro duced into the American market. The success was impressive, and worldwide sales of rofecoxib reached US$ 2.5 billion in 2003 [11]. However, some aspects of the process were unusual. The pivotal randomised clinical trial (RCT) was only published over a year later than the commercial approval of the drug had been granted [12]. The VIGOR trial compared ro fecoxib (50 mg/od) with naproxen (1.000 mg/od) in 8.076 patients with rheumatoid arthritis. The trial showed an increased number of acute myocardial infarctions (AMI) (0.4% vs. 0.1%) in patients taking rofecoxib. Concerns about the cardiovascular safety profile of the drug were initially overcome by inter preting that the results of the trial revealed more an allegedly cardioprotective effect of naproxen rather than a cardiotoxic effect of rofecoxib [12]. This hypothesis has been repeatedly questioned be cause, in fact, epidemiological as well as 'in vitro' data on the effect on platelet aggregation and other mech anisms related to cardiovascular safety of NSAIDs (other than aspirin) are very scarce. Regulatory authorities in USA and EU reacted by including a precautionary sentence in the labelling of marketed coxibs, reflecting the findings from the VIGOR trial. Subsequently, the coxibs 'me-too' saga increased sub stantially: valdecoxib, parecoxib, etoricoxib, lumirac oxib... New results from a RCT comparing lumiracoxib vs. naproxen raised more concern about an excess risk of AMI of another coxib [13]. Meanwhile, a Merck-sponsored meta-analysis of randomised trials of rofecoxib [14] and some obser vational studies [15-17] appeared to support the an tiaggregant effect of naproxen, while other studies suggested that the increased cardiovascular risk of