Penn Researchers Add More Evidence To Demonstrate Role Of COX Inhibitors In Heart-disease Risk

Six years ago Garret FitzGerald, MD, Director of the Institute for Translational Medicine and Therapeutics at Penn, raised the possibility that selective COX-2 inhibitors might predispose patients otherwise at risk for an increased incidence of heart attack and stroke. This proposal was based initially on his studies of how celecoxib (Celebrex) and rofeocoxib (Vioxx) worked in human volunteers.

The first unequivocal evidence of this risk emerged with the Merck-sponsored APPROVe study of Vioxx, leading to withdrawal of the drug in September 2004. Evidence implicating a second member of the class, valdecoxib (Bextra) was presented by FitzGerald in a lecture at the American Heart Association (AHA) in November 2004. This work – a collaboration led by Curt Furberg of Wake Forrest University, along with Bruce Psaty of the University of Washington and FitzGerald – appears online January 17 in Circulation and in the January 25th print edition of the journal.

In this first study, the researchers used a conventional statistical approach called meta-analysis to combine the findings of two trials to obtain a stronger estimate of the risk of heart attack plus stroke than is possible from looking at either trial alone. Their analysis suggests that the COX-2 inhibitor Bextra elevated the combined incidence of heart attack and stroke three-fold in the study population of CABG patients.

Two placebo-controlled trials of Bextra and parecoxib (Dynastat) were performed in patients undergoing coronary artery bypass graft surgery. These studies were sponsored by Pfizer, Inc. In both cases, intravenous Dynastat, which is converted to Bextra within minutes in the body, was given before oral dosing with Bextra. The first involved roughly 400 patients and dosing lasted 14 days. Concern was prompted by an apparent cardiovascular signal and the FDA did not grant approval to Dynastat, despite clear evidence of pain relief from the Dynastat/Bextra combination compared to placebo. A second, larger study was performed in which the dose of Bextra was reduced, as was the dosing period to 10 days. The figures from this study are now incorporated in a revised drug label for Bextra available at the FDA website (www.fda.gov).

This result is consistent with the original mechanism proposed by FitzGerald in 1999 that COX-2 inhibitors may be problematic for those at risk for heart disease. COX-2 is the main source of a fat – prostacyclin or PGI2 – which protects the heart from factors that activate the clotting system, harden the arteries, and raise blood pressure. "Although a clinical trial is a crude detection system for uncommon side effects, such as the cardiovascular risk from coxibs, we predicted that a signal would be detected faster and in smaller studies in patients with activated clotting systems," says FitzGerald. "It is well known that this is true of coronary artery bypass graft surgery patients." Studies of sufficient size and duration to detect the expected rates of cardiovascular events in arthritis patients have not been performed with Bextra.

While FitzGerald stresses that the emergence of a cardiovascular signal with a second COX-2 inhibitor rendered the argument for a class effect of the risk "compelling," the early cessation of a placebo-controlled trial of celecoxib (Celebrex) in December 2004 by the National Cancer Institute (because of an excess incidence of heart attack and stroke) appeared to put the matter beyond dispute.

A related paper from the FitzGerald group also appears online in the same issue of Circulation. Karine Egan, PhD, a postdoctoral researcher in the FitzGerald lab, and colleagues at Penn and the Wistar Institute studied mice genetically prone to hardening of the arteries (atherosclerosis) and showed that another fat – thromboxane or TxA2, this time produced by COX-1 – accelerates atherosclerosis. Indeed, they showed a drug that blocks TxA2 slowed this process at its early stages, although it seemed ineffective once atherosclerosis was well established. "This is of particular interest, as low-dose aspirin prevents heart attack and stroke by blocking COX-1 formation of TxA2 in blood cells called platelets," says FitzGerald. "Its effects or those of a TxA2 blocker on hardening of the arteries has never been studied well in humans."

Egan and colleagues failed to detect a benefit from COX-2 inhibition in the atherosclerosis-prone mice, but they reasoned that a potential benefit might be masked by the effect of the drug. Although platelet COX-1 is the major source of TxA2, this fat can also be formed by COX-2 in cells called macrophages that invade the hardening vessel wall. Given what they had seen with the TxA2 blocker, FitzGerald's group figured that they might need to add the TxA2 blocker to see the anti-inflammatory benefit of the COX-2 inhibitor.

Adding the COX-2 inhibitor not only failed to add to the beneficial effects of the TxA2 blocker, it caused disturbing changes in the makeup of the atherosclerotic plaques. "We were amazed," states Egan. "Addition of the COX-2 inhibitor caused changes that, if they occurred in humans, would result in a loss of stability of the plaque, making it more likely to rupture and activate clotting, causing heart attack or stroke."

"We must always be cautious projecting the results of studies in mice to clinical outcomes in humans," says FitzGerald. "While low-dose aspirin works by switching off TxA2, a TxA2 blocker might act in subtle, but importantly different ways. However, in so far as it mimics the effects of low-dose aspirin, these results would have disturbing implications for patients at high cardiovascular risk treated with aspirin and a coxib."

FitzGerald adds: "The clear emergence of a cardiovascular hazard from COX-2 inhibitors in patients, the weak rationale for a study of their protective properties in the first instance, and now this evidence from mice would indicate to me that a trial in high-risk patients, such as that proposed for Celebrex is, at best, ill advised."

###

This paper was funded in part by grants from the National Institutes of Health, Servier Laboratories, and Merck Research Laboratories. Servier and Merck had no influence on the design or interpretation of the studies, and the authors have no competing financial interests.