Following the publication of two large studies VIGOR* and CLASS* of the selective COX-2 inhibitors rofecoxib (Vioxx) and celecoxib (Celebrex), doubts remained regarding safety advantages of these drugs (see also a-t 2000; 31: 107 and 2001; 32: 35-6). Now it has become clear that in both studies essential risk data where withheld from publication.
CARDIOTOXICITY: The published data of the VIGOR study reflect the cardiovascular risk potential of rofecoxib as compared to the non-steroidal antirheumatic drug (NSARD) naproxen (e.g. Proxen) in an insufficient and distorted manner. Following an interim analysis at the end of 1999, concerns were expressed regarding excess deaths and adverse cardiovascular events, as well as elevated blood pressure associated with rofecoxib. (1) In the publication issued in November 2000, the authors reported merely about an increased myocardial infarction rate instead of all cardiovascular complications that arose. A significantly increased risk of myocardial infarction with rofecoxib treatment, as compared to naproxen, was found only in those patients who would have benefitted from acetylsalicylic acid (ASA, e.g. Aspirin) for their cardiovascular disease but were not allowed to take ASA based on the study's protocol and who, therefore, should not have participated at all. (2) The provided data could lead to the conclusion, that rofecoxib does not render an increased cardiovascular risk in patients who do not require ASA.
However, a safety update submitted to the American Food and Drug Administration (FDA) by the sponsor on 13 October 2000 - six weeks before the publication of the study - clearly states that severe thrombotic cardiovascular events increased significantly also in the aspirin not indicated patients of the rofecoxib group. These events included myocardial infarction, unstable angina pectoris, cardiac arrest, sudden death or ischemic stroke. (1) Relative risk (RR) in these patients was 1.89 (95% confidence interval [CI] 1.03-3.45). (3) In patients who might have benefitted from aspirin use, RR increased even fivefold (RR 4.89; 95% CI 1.41-16.88), with a total RR for the entire rofecoxib group of 2.38 (95% CI 1.39-4.00). There was also a trend towards an increased rate of congestive heart failure adverse experiences (19 vs. 9; RR 2.11; p=0.065). (1) Rofecoxib caused an increase in blood pressure: The mean increase in systolic blood pressure was 4.6 mmHg, the mean increase in diastolic blood pressure was 1.7 mmHg as compared to 1 and 0.1, respectively, under naproxen. 1 37 patients died, 22 (0,54%) in the rofecoxib group, 15 (0,37%) in the naproxen group. The presentation of the overall mortality rate in the original publication favoured rofecoxib by only reporting rounded numbers (0.5% and 0.4%). (2)
The potential mechanism of injury, blocking the synthesis of prostacyclin (PGI2), a platelet aggregation inhibitor and vasodilator, by selective COX-2 inhibition without effect on the COX-1 dependent platelet activating factor thromboxane A2 (TXA2), suggests a class effect. At present, however, clinical data do not support this. In the CLASS study with celecoxib, where low-dose ASA use was permitted, no significant differences in adverse cardiovascular event rates, as compared to ibuprofen (e.g. Brufen) and diclofenac (e.g. Voltaren) were reported. Independent direct comparison between the two COX-2 inhibitors is not available. Therefore, American authors compared the annualized myocardial infarction rates in the rofecoxib (0.74%) and celecoxib (0.80%) groups in VIGOR and CLASS with those under placebo (0.52%) in four primary prevention of myocardial infarction trials with aspirin. (3) However, in our view, comparing patient groups from different studies is flawed by so many uncertainties, that the results become inconclusive.
GASTROINTESTINAL TOLERANCE: By now, the authors of the CLASS study are accused of an attempt to mislead the public. Only results from the first six months of the study were submitted for publication to the Journal of American Medical Association (JAMA). The author of the accompanying editorial also received only the half year data, although the study, which ran over one year, had already been completed at the time of publication. (4) The event rates of gastrointestinal ulcer complications during the first six months were expressed as annualized incidence rates in the original publication. Overall this resulted in a trend in favour of celecoxib as compared to the standard NSARDs diclofenac (e.g. Voltaren) and ibuprofen (e.g. Brufen). In patients who did not take aspirin, the COX-2 inhibitor fared significantly better. (5) The actual results after one year, which were reported to the FDA only, do not show any benefit of celecoxib. Almost all ulcer related complications in the second half of the study occured in the celecoxib group. Even in patients who did not take aspirin, there was no difference between the COX-2 inhibitor and the two standard NSARDs in this respect. (6)
SUMMARY: Parallels are striking: Both studies publicly reported only the more advantageous results, while the complete data pool showing less favourable results was reported to the FDA only. In our view, this strategy was used to provide the manufacturers with a timely advantage in order to establish their products on the market. (7) The obvious commercial interests reflected in VIGOR and CLASS by tampering with scientific results and disregarding safety concerns for patients, undermine confidence in the seriousness and scientific quality of all study data presented to the public.