Short-term Use of Rofecoxib May Increase Cardiovascular RiskJun 28, 2006 | www.medscape.com
On June 26, the New England Journal of Medicine (NEJM) posted online an unsigned correction notice regarding the Adenomatous Polyp Prevention on Vioxx (APPROVe) study, as well as letters from experts who report on an additional analysis of original and follow-up data from APPROVe. The letters report an increased risk of confirmed thrombotic events associated even with short-term use of rofecoxib.
"The safety risk with rofecoxib occurs earlier and lasts longer than we originally thought," letter author Steven E. Nissen, MD, vice-chairman of cardiology at the Cleveland Clinic Foundation in Ohio," told Medscape. "Even those who used the drug for shorter periods have an increased cardiovascular risk, which is important medically and scientifically and has legal implications as well."
The original report of this rofecoxib chemoprevention trial in colorectal adenoma, which was published by NEJM in March of last year, showed an increase in cardiovascular risk for rofecoxib (Vioxx) compared with placebo only after 18 months of use. Findings from APPROVe led to Merck's withdrawal of Vioxx from the market in September 2004.
However, the analysis in the 2005 NEJM article did not use the statistical method described in the Methods section. Subsequent application of that method led to the June 26 correction, which deleted references to the 18-month threshold from the Abstract and Discussion section.
Although the Methods section of the original paper specified that the logarithm of time would be used for the test for proportionality of hazards, the reported results used linear time rather than the logarithm of time for this test. Analysis using the logarithm of time led to several changes. Visual inspection suggested that the Kaplan-Meier curves for rofecoxib vs placebo separated 18 months after randomization, but an overall test of the proportional-hazards assumption for the entire 36-month observation period did not reach statistical significance (P = .07).
"The changing relative risk over time (the topic of the correction) was a secondary analysis," John A. Baron, MD, MS, MSc, a professor of medicine at Dartmouth Medical School in Lebanon, NH, and coauthor of the original APPROVe report, told Medscape. "The conclusions that were drawn regarding that analysis aren't very different because of the correction. Potentially more interesting, I think, is the analysis of the extended follow-up, which is underway."
The correction notice specified that statements regarding an increase in risk after 18 months should be removed from the Abstract, specifically removing the claim that "during the first 18 months, the event rates were similar in the two groups." Similarly, the correction removes the following from the Discussion section: "the increased relative risk of adjudicated thrombotic events was first observed after approximately 18 months of treatment."
Despite these changes, the correction ends with the following revised statement: "We found an increased risk of confirmed thrombotic events associated with the use of rofecoxib. Visual inspection of the Kaplan-Meier curves suggested that there was an increased frequency of thrombotic events associated with rofecoxib therapy after 18 months."
A Wall Street Journal story on June 27 says that Merck stands by its assertion that the increased relative risk of cardiovascular events observed with rofecoxib was seen beginning after 18 months. The Journal quotes Theodore Mayer, co-lead outside counsel for Merck, as saying, "We believe our position is scientifically sound and we feel comfortable representing it."
According to Dr. Baron, the correction notice did not reflect a new analysis, but Merck has conducted a new analysis regarding the additional follow-up recently completed, and analysis of an extended follow-up is now underway.
In addition to the correction notice, the NEJM also posted 2 letters to the editor describing new analyses based on the recent public disclosure of data from a 12-month extension study of the APPROVe trial, as well as a statistical analysis and discussion of the correction, written by Stephen W. Lagakos, PhD, chair of biostatistics at the Harvard School of Public Health in Boston, Massachusetts. When contacted by Medscape, Dr. Lagakos declined to comment, saying that he did not have access to the original data.
Dr. Nissen's letter to the editor states that the new data reveal the full results of both the original study and the extension phase, including data tables and Kaplan-Meier curves. The original article reported cardiovascular event rates using "an unusual censoring rule" which excluded events if they occurred more than 14 days after the study drug was stopped. That article included a post-hoc hypothesis that curves for confirmed thrombotic events would not begin to diverge until after 18 months of exposure to rofecoxib.
However, using a conventional intention-to-treat analysis and several different end points, including the widely used end point of the Antiplatelet Trialists' Collaboration (APTC) study, the newly released report showed that the event curves begin to diverge much earlier, generally within 4 to 6 months or even after only 3 months of exposure to rofecoxib. Because patients who stopped the study drug early are likely to be people who had adverse reactions such as hypertension, heart failure, or renal dysfunction, Dr. Nissen points out that they represent a particularly vulnerable group.
"The APPROVe investigators censored all cardiovascular events occurring more than 2 weeks after stopping study drug," Dr. Nissen said. "This had a distorting effect, because those who stopped the drug had an increased risk of cardiovascular effects, and excluding these late effects made the drug look safer than it is. They should have shown both analyses."
The second letter to the editor, by Curt D. Furberg, MD, PhD, from Wake Forest University Health Sciences in Winston-Salem, North Carolina, notes that the update of the APPROVe trial data contains additional information about events in the subgroup of participants whose data were censored if they had an event more than 14 days after early discontinuation of the study medication.
Of 12 thrombotic events that occurred more than 14 days after the study drug was stopped but within 36 months after randomization, 8 were in the rofecoxib group. Analysis of the 3-year event data according to the intention-to-treat principle showed that including these events had a clear effect on the published survival curve for rofecoxib, making it more linear. The narrowing of the distance between the rofecoxib and placebo curves at 18 months almost disappeared, and statistical analysis showed no evidence of deviation from the proportional hazard over time.
"The release of the new data raises questions," Dr. Furberg writes. "At the time the APPROVe trial was submitted and published, was the complete data set available to the authors for an intention-to-treat analysis? Did they perform a proportionality test of the three-year event data before publication?"
A response letter from APPROVe investigators Robert S. Bresalier, MD, from the University of Texas MD Anderson Cancer Center in Houston, and Dr. Baron was submitted to the NEJM editors before the correction notice was posted. That letter assumes that Drs. Nissen and Furberg were referring to a preliminary analysis of new data that was released by Merck to the Food and Drug Administration on May 11, 2006. It states that the decision to perform a systematic event follow-up covering the period more than 14 days after the discontinuation of therapy was motivated by the cardiovascular findings reported in September 2004. These additional safety data were not available when the original article was published and were compiled for analysis only in the past few months. The letter further defends the censoring of event data after 14 days.
"Clearly, an in-depth analysis of the extended experience of the patients in the APPROVe Trial is indicated, and it is under way," Drs. Bresalier and Baron conclude. "It will include an independent statistical analysis of the cardiovascular data. Until that is completed and a formal report is peer-reviewed, speculations regarding what will be found are premature and may be misleading. However, it is clear that the main conclusion of the article — that 'among patients with a history of colorectal adenomas, the use of rofecoxib was associated with an increased cardiovascular risk' — is unaffected."
In the accompanying Perspective, Dr. Lagakos highlights methodological and statistical issues about the analysis and interpretation of a time-to-event end point in a randomized, placebo-controlled trial evaluating a long-term treatment. These issues include the appropriate period of follow-up for safety outcomes after treatment discontinuation, the purpose and implications of checking the assumption of proportional hazards, and what the results of a trial examining long-term use imply about the safety of a drug if it were given for shorter periods. He points out that some have misinterpreted the results of APPROVe to mean that treatment with rofecoxib for fewer than 18 months poses no excess cardiovascular risk.
"The risk [with rofecoxib] occurred earlier than Merck asserted in its published statements, and occurred even after the drug was stopped," Dr. Nissen said about the analysis he discusses in his letter. "This is a big surprise scientifically, because it suggests that the underlying mechanism of how the drug causes cardiovascular events may be more profound than we originally thought. Something is different about this drug from other drugs in its class."
Dr. Nissen reports no relevant financial relationships. Dr. Baron is a consultant for Merck.