Vioxx Trouble Started Early and Stayed Late

OXFORD, England, July 26 -- The cardiovascular risk of rofecoxib (Vioxx), the now-withdrawn Cox-2 inhibitor, increased from the start of treatment rather than after an 18-month delay, found researchers here.

The findings support a long-standing claim of rofecoxib critics, a claim that was disputed by Merck on the basis of the drug-maker's analysis of the Adenomatous Polyp Prevention on Vioxx (APPROVe) trial.

When the trial was published in the New England Journal of Medicine in 2005 after the drug's withdrawal in September 2004, Kaplan-Meier curves suggested that the risk was entirely accounted for by patients who took the drug for 18 or more months, a threshold that the company and its lawyers embrace.

Even after the statistical analysis was admitted to be in error and revised in May 2006 to demonstrate a chance that risk began earlier in treatment, the company held the line that there was no evidence to support harm before 18 months.

Now in an analysis of a large colorectal cancer prevention trial, a 2.66-fold excess risk was seen within eight months of rofecoxib treatment (P=0.04) or sooner, found David J. Kerr, M.D., of the University of Oxford here, and colleagues.

Half of all events occurred among patients treated for less than 12 months, they reported in the July 26 issue of the New England Journal of Medicine.

And, the results may implicate other Cox-2 inhibitors as well, commented Elliott M. Antman, M.D., of Brigham and Women's Hospital and Harvard, acting as a spokesperson for the American Heart Association.

"Based on our understanding of the biology there's no reason to believe that other Cox-2 inhibitors would have a longer period to harm," he said, though he noted that other drugs in the class may have less cardiovascular risk than rofecoxib.

"We should not feel safe saying we don't need to worry until 18 months have elapsed," he added.

It was true at the time that there was no evidence to support earlier harm of rofecoxib and not much information to support or refute that claim, commented Steven E. Nissen, M.D., of the Cleveland Clinic, who was one of the early rofecoxib critics.

So to clarify the duration of drug exposure responsible for increased cardiovascular risk, Dr. Kerr and colleagues analyzed outcomes in the Vioxx in Colorectal Cancer Therapy: Definition of Optimal Regime (VICTOR) trial.

This trial began like APPROVe, attempting to prevent colorectal cancer with rofecoxib treatment. However, when the drug was withdrawn from the market, the researchers halted treatment and continued to follow patients for two years after the trial closed.

Before September 2004, 2,434 patients at 151 hospitals in Britain who had undergone potentially curative surgery for colorectal cancer were randomized to 25-mg rofecoxib daily or placebo.

Patients in the rofecoxib group had slightly greater baseline cardiovascular risk than the placebo group based on predefined risk factors, but the only significant difference was a higher prevalence of diabetes (8.7% versus 5.6%, P=0.003).

Treatment continued for a median 7.4 months in the rofecoxib group and 8.2 months in the placebo group before the trial was closed and treatment discontinued. A third of all patients received assigned medication for at least a year.

The researchers found 16 confirmed cardiovascular thrombotic events in 15 patients receiving rofecoxib during or within 14 days after the treatment period versus seven events in six patients in the placebo group.

The relative risk of a cardiovascular thrombotic event was 2.66-fold higher with rofecoxib than placebo (95% confidence interval 1.03 to 6.86, P=0.04). Adjustment for cardiovascular risk factors attenuated the association, but the point estimate remained similar (RR 2.41, 95% CI 0.93 to 6.26, P=0.07).

Eight of the events were cardiac, three were peripheral vascular, and five were cerebrovascular in the rofecoxib group. In the placebo group, there were four, one, and one of these events, respectively, and two hemorrhagic events as well. Three of events overall occurred in conjunction with other non-steroidal anti-inflammatory drugs.

Moreover, Kaplan-Meier plots showed that the curves started to separate almost from day one, Dr. Nissen noted.

"The risk begins almost immediately and just gets progressively greater over time," he said.

An additional 14 adjudicated events were added from study closure to 24 months of follow-up for a total of 21 events in the rofecoxib group and 14 in the placebo group.

The relative risk of cardiovascular thrombotic events over the entire period from the beginning of treatment out to 24 months after trial closure was 1.50-fold higher with rofecoxib than placebo (95% CI 0.76 to 2.94, P=0.24).

Death from cardiovascular causes did not differ significantly between groups (five in the rofecoxib group versus seven in the placebo group).

But, the lack of significance for longer-term risk was not reassuring, Dr. Antman said.

"The trends here are all consistent with a signal of harm," he said. "Whether or not it's statistically significant has more to do with sample size."

Furthermore, other studies have suggested residual risk remains after rofecoxib is discontinued, Dr. Nissen said.

"It's important to know about the time to onset of risk because there are other drugs being developed in the class," Dr. Nissen noted. "If new drugs come along that have this problem, we need to recognize that they can cause trouble fairly quickly."

Merck commented that the "limited data from prematurely terminated studies such as ...VICTOR need to be interpreted with caution, and that assessments of cardiovascular risk with Vioxx must take into account the large amount of randomized, placebo-controlled clinical trial data from the Vioxx research and development program."

Dr. Kerr and another researcher reported receiving grant support from Cancer Research UK. One researcher also reported receiving consulting fees from Merck and Johnson & Johnson, while another reported serving as an advisor to Merck and to Novartis and as a consultant to lawyers representing Merck. Drs. Antman and Nissen reported no relevant conflicts of interest.