Cardiovascular safety of COX-2 inhibition: time to inform systems of postmarketing surveillance?

Abstract

Since late summer 2004, regulatory and patient concerns over retrospective data on the overall safety of cyclooxygenase (COX)-2 inhibitors (coxibs), such as rofecoxib, valdecoxib and celecoxib, in patients with coronary disease have evolved in an ever-upward spiral of increasingly public rather than scientific debate. The episode was heralded to the wider world by the precautionary withdrawal of rofecoxib from the world market following discussions by the US Food and Drug Administration (FDA). A serious drop in the stock valuation of Merck Sharpe and Dohme Ltd (MSD) and a worldwide media frenzy followed. In particular, the US media reaction to this step was openly criticised by AJJ Wood (the Scottish clinical pharmacologist based at Vanderbilt who chaired the relevant FDA advisory panel on COX-2 inhibitors and is a potential new commissioner of the FDA). Regulatory concern over data on safety is not a new feature in the assessment of drugs following marketing. Paradoxically, adverse events involving a common disease such as coronary disease are harder to define than rare or idiosyncratic side effects. Unlike the latter, they tend to be below the limits of recognition of individual clinicians and most current systems of postmarketing surveillance. Specifically, recognition of an impact of concomitant drug treatment on coronary disease in patients with arthritides (where the background symptom burden is traditionally very high) is a good case study in the complexity of pharmacoepidemiology. These patients have a long-established excess incidence of coronary and vascular events recognized long before the era of COX-2 inhibitor use [1]. The safety and efficacy of any drug entity (new or old) is of course a product of the number and duration of patient exposures and the unpredictable impact of prescription to unselected patients (those not systematically studied during development). Clearly there were no data to suggest an adverse interaction with coronary disease or its treatment before marketing. In general, no drug therapy is anything other than relatively safe in this regard and, routinely, prescribers place a disproportionate amount of faith in the developmental and regulatory assessment process. What can be learnt from the regulatory process laid out in the postmarketing surveillance of COX-2 inhibition?