Disconnect between COX-2 selective inhibition and cardiovascular risk in preclinical models.

INTRODUCTION Secondary pharmacology profiling is routinely applied in pharmaceutical drug discovery to investigate the pharmaceutical effects of a drug at molecular targets distinct from (off-target) the intended therapeutic molecular target (on-target). Data from a randomized, placebo-controlled clinical trial, the APPROVe (Adenomatous Polyp Prevention on VIOXX, rofecoxib) trial, raised significant concerns about COX-2 inhibition as a primary or secondary target, shaping the screening and decision-making processes of some pharmaceutical companies. COX-2 is often included in off-target screens due to cardiovascular (CV) safety concerns about secondary interactions with this target. Several potential mechanisms of COX-2-mediated myocardial infarctions have been considered including, effects on platelet stickiness/aggregation, vasal tone and blood pressure, and endothelial cell activation. In the present study, we focused on each of these mechanisms as potential effects of COX-2 inhibitors, to find evidence of mechanism using various in vitro and in vivo preclinical models. METHODS Compounds tested in the study, with a range of COX-2 selectivity, included rofecoxib, celecoxib, etodolac, and meloxicam. Compounds were screened for inhibition of COX-2 vs COX-1 enzymatic activity, ex vivo platelet aggregation (using whole blood from multiple species), ex vivo canine femoral vascular ring model, in vitro human endothelial cell activation (with and without COX-2 induction), and in vivo cardiovascular assessment (anesthetized dog). RESULTS The COX-2 binding assessment generally confirmed the COX-2 selectivity previously reported. COX-2 inhibitors did not have effects on platelet function (spontaneous aggregation or inhibition of aggregation), cardiovascular parameters (mean arterial pressure, heart rate, and left ventricular contractility), or endothelial cell activation. However, rofecoxib uniquely produced an endothelial mediated constriction response in canine femoral arteries. CONCLUSION Our data suggest that rofecoxib-related cardiovascular events in humans are not predicted by COX-2 potency or selectivity. In addition, the vascular ring model suggested possible adverse cardiovascular effects by COX-2 inhibitors, although these effects were not seen in vivo studies. These results may also suggest that COX-2 inhibition alone is not responsible for rofecoxib-mediated adverse cardiovascular outcomes.