Cyclooxygenase-2 inhibitors: what went wrong?

This review describes the biology underpinning the development of selective cyclooxygenase-2 inhibitors, documenting the clinical experience from the pivotal gastrointestinal safety trials to their eventual withdrawal or labelling with cardiovascular safety warnings.

The elucidation of differences between the active sites of cyclooxygenase-1 and cyclooxygenase-2 allowed the targeted design of the selective cyclooxygenase-2 inhibitors known as coxibs. These were developed and marketed as non-steroidal anti-inflammatory drugs (NSAIDs) that had improved upper gastrointestinal safety compared with older non-selective NSAIDs. A large-scale study with arthritis patients to evaluate upper gastrointestinal safety, however, demonstrated that celecoxib was not superior in terms of upper gastrointestinal safety compared with the older non-selective NSAIDs that were used as comparators. In an equally large study with arthritis patients, a more selective cyclooxygenase-2 inhibitor, rofecoxib, did have improved upper gastrointestinal safety compared with the non-selective non-steroidal anti-inflammatory drug naproxen. Although concomitant clinical trial evidence emerged that rofecoxib increased cardiovascular risk, this was discounted by its pharmaceutical company owner. Despite the lack of improved upper gastrointestinal safety with celecoxib and the evidence of cardiovascular risk with rofecoxib, both agents had widespread clinical use for 4-5 years. This was not diminished by the publication of plausible eicosanoid-based biological mechanisms whereby selective cyclooxygenase-2 inhibition could increase cardiovascular risk. Finally, clinical trials involving patients with colorectal cancer and post-operative pain revealed increased cardiovascular risk with all members of this class of drug.

These events provide a case study of a failure of the medical journal literature to guide drug usage.