Portal de Periódicos da CAPES

The NSAID roller coaster: more about rofecoxib

2006; Wiley; Volume: 62; Issue: 3 Linguagem: Inglês

10.1111/j.1365-2125.2006.02761.x

1365-2125

Jeffrey K Aronson,

Pharmacogenetics and Drug Metabolism

‘The essence of experimentation is never to ignore the unusual.’ In 1982 John Robert Vane shared the Nobel Prize with Sune Bergström and Bengt Ingemar Samuelsson for ‘discoveries concerning prostaglandins and related biologically active substances’. Bergström and Samuelsson isolated and characterized the structures of the misnamed prostaglandins (in the male reproductive tract they mainly come from the seminal vesicles) and showed that they are synthesized from arachidonic acid; Samuelsson also discovered thromboxane and the leukotrienes; Vane showed that aspirin and the compounds that have become known as NSAIDs (non-steroidal anti-inflammatory drugs), such as indometacin, inhibit the production of prostaglandins from arachidonic acid by inhibiting the enzyme cyclo-oxygenase [1]. Although it seemed possible at that time that the actions of the NSAIDs could all be attributed to inhibition of cyclo-oxygenase, some observations could not be completely reconciled with this hypothesis. For example, there was variation from NSAID to NSAID in the risk of gastrointestinal ulceration and its associated complications, first described in 1938 when Douthwaite and Lintott gave 16 patients three tablets of aspirin crushed in one ounce of water and saw ‘dusky red’ changes in the stomach at gastroscopy [2]. The risk is lowest with ibuprofen and highest with azapropazone, and there is a 15-fold difference between the two [3]. The most parsimonious explanation for this and other inconsistencies was that cyclo-oxygenase existed in more than one isoform, as Flower & Vane had hypothesized early on, in an attempt to explain the mode of action of paracetamol [4]. Eventually in the early 1990s it was shown that there were indeed at least two different types of cyclo-oxygenase, COX-1 and COX-2 [5]. COX-3 remains a gleam in the eye of some pharmacologists [6]. In 1996 Vane received the Hanbury Memorial Medal, a prestigious award that was instituted in 1881 and is currently made every five years for ‘high excellence in the prosecution or promotion of original research in the natural history and chemistry of drugs’; the previous winner had been Sir James Black. In the lecture that he gave to celebrate this award, Vane talked about the potential uses of highly selective inhibitors of COX-2. When the isoforms of cyclo-oxygenase were discovered, the beneficial anti-inflammatory effects of the NSAIDs had been attributed to inhibition of COX-2, while the antiplatelet effects and the gastrointestinal and renal adverse effects were attributed to COX-1. Indeed, the rank selectivity of the COX-2 inhibitory effects of NSAIDs correlated reasonably well with the rank risk of serious upper gastrointestinal complications [7]. Although it was soon realized that this concept was flawed [8], Vane predicted that highly selective COX-2 inhibitors would be safer than the classical NSAIDs and that in addition to their use in painful arthropathies they would also prove to be effective in colorectal cancer, in which COX-2 is overexpressed [9]. But the bad news was yet to come; it broke just before Vane died in November 2004 [10]. By the late 1990s new highly selective inhibitors of COX-2 started to appear – the list of compounds that have been given International Non-proprietary Names by the WHO now includes celecoxib, cimicoxib, deracoxib, etoricoxib, firocoxib, lumiracoxib, mavacoxib, parecoxib, robenacoxib, rofecoxib, tilmacoxib, and valdecoxib, known collectively as the coxibs. The first hint that the coxibs might be associated with adverse effects that were not necessarily associated with non-selective NSAIDs came with reports of colitis [11–13]. In this issue of the Journal Backman et al. report another adverse effect of rofecoxib: it inhibits CYP1A2 and should therefore not be used with other drugs that are metabolized by this enzyme, such as tizanidine (the subject of the study), clozapine and olanzapine, lidocaine and ropivacaine, melatonin, riluzole, tacrine, and zolmitriptan [14]. The effect is large and for some drugs could be dangerous. However, the really bad news came when rofecoxib was withdrawn in September 2004 [15] following the emergence of evidence in the APPROVe study that it was associated with an increased risk of cardiovascular morbidity, from myocardial infarction and stroke [16]. This happened amid controversy over the interpretation of the results of an earlier study, VIGOR, which had shown an increased risk of cardiovascular events compared with naproxen [17, 18], discounted as showing a supposed beneficial effect of naproxen. Since then valdecoxib has also been withdrawn because of potentially life-threatening skin reactions, although it has also been associated with an increased risk of cardiovascular events after coronary artery bypass grafting [19]. In the wake of this, EMEA has issued guidelines on use of the COX-2 inhibitors that are still available [20]: all COX-2 inhibitors should be contraindicated in patients with established ischaemic heart disease or stroke and in patients with peripheral arterial disease; physicians should be cautious when prescribing COX-2 inhibitors for patients with risk factors for heart disease, such as hypertension, hyperlipidaemia, diabetes, and smoking; the lowest effective dose should be used for the shortest possible duration; the balance of cardiovascular and gastrointestinal risks should be carefully considered for patients who do not have heart disease but are taking low-dose aspirin; patients with a history of drug allergies may be at greater risk of hypersensitivity skin reactions, which are rare but serious and sometimes fatal. One would expect changes in the prescribing of NSAIDs in the light of these events, and in this issue of the Journal Williams et al. report the Scottish data [21]. Not surprisingly, following the withdrawal of rofecoxib there was an immediate increase in the number of prescriptions of celecoxib; the increase, however, was short-lived – after 3 months it fell dramatically. At the same time the numbers of prescriptions of ibuprofen and diclofenac rose. Prescriptions for etodolac and meloxicam also rose – most prescribers probably do not realize that they too are coxibs in all but name. The total number of prescriptions for all the NSAIDs included in this study seems to have fallen by about 10%, reading the data from the authors’ graph directly, although they do not give the actual numbers. This may reflect increased disillusionment with the use of NSAIDs, since the cardiovascular safety of the non-selective NSAIDs has also recently been questioned [22]. Adverse effects of drugs can (I would say should) be classified by three attributes: the relation between the dose at which they occur and the dose at which the therapeutic effect occurs, the time-course over which they occur, and the susceptibility of the individual [23]. An important question about the COX-2 inhibitors is therefore the dose at which their adverse cardiovascular effects occur. McGettigan et al. provide some information in this issue of the Journal[24]. In a case-control study of 328 patients with acute coronary syndrome and 478 controls they found a significant dose-effect relation (odds ratio for high doses versus low doses = 2.8); a comparison with standard NSAIDs suggested that the coxibs may be protective at low doses and harmful at higher doses. This study will continue, with the intention of recruiting 1200 patients and 1800 controls, and we look forward to seeing the complete results. The interim results suggest that the adverse cardiovascular effects of rofecoxib are either mediated by a toxic effect on COX-2 or by a mechanism other than inhibition of COX-2; that would be consistent with the hypothesis that increased plasma and tissue aldosterone is involved [22]. The time-course of the adverse cardiovascular effects of rofecoxib is not clear. In VIGOR the risk was noted after 9 months and in APPROVe after 18 months. However, newer data suggest that the risk of a first myocardial infarction is greater in the first 9 days and does not increase thereafter [25]; these data are hard to interpret and there were several potentially confounding factors. The important susceptibility factors have yet to emerge, but presumably the factors that increase the risk of cardiovascular disease in the general population are involved. The NSAIDs are on a roller coaster ride. The non-selective drugs provide many patients with much needed pain relief in many conditions, but their adverse effects can be intolerable. The COX-2 inhibitors were initially greeted with great enthusiasm, in the expectation that they would be safer. However, the seeds of doubt about rofecoxib were then sown by VIGOR. At that stage the manufacturers, Merck, could have elected to limit the recommended use of rofecoxib to short-term therapy; well-informed patients could have decided whether they should take the risk of longer-term use. Merck chose not to do this, preferring instead to believe that rofecoxib was safe. In John Vane’s words, taken from J H Burn, they ignored the unusual. They have paid the price. The fate of all the COX-2 inhibitors, as well as the non-selective NSAIDs, remains to be discovered. Newer drugs may emerge with different benefit to harm profiles. But I suspect that the roller coaster has a long way to go.