Long-Term Oral Anticoagulant Therapy in Patients With Unstable Angina or Suspected Non–Q-Wave Myocardial Infarction
To the Editor:
A striking thing that emerges from the report by Anand et al of the OASIS pilot study1 relates to the differences in the control arms in phase 1 (randomization to coumadin or standard care a mean of 6 days after study entry) and phase 2 (randomization to coumadin or standard care a mean of 26 hours after study entry). It is difficult to reconcile why the placebo composite event rate of cardiovascular death/new myocardial infarction/refractory angina at 6 months in phase 1 (3.9%) is so much lower than the placebo composite event rate at 3 months in phase 2 (12.1%). How can getting placebo early or late make such a difference? At first glance, the protocol modifications instituted in phase 2 (cited by the authors by way of explanation) all relate to the dosing and administration of warfarin and ought not to really affect control event rates.
So what are the more subtle differences between the 2 control arms? In phase 1, heparin/hirudin therapy had been discontinued by the time randomization took place, whereas in phase 2 heparin/hirudin was still ongoing (and could be continued for up to 72 hours). Also, the “event clock” started much earlier in phase 2 than in phase 1. Could the higher control event rate in phase 2 be because the investigators were capturing far more early events that otherwise would have been counted as part of the heparin/hirudin randomization in phase 1?
Three important additional pieces of information are really necessary to help sort out this dilemma. First, what is the breakdown of individual components of the composite end points in phases 1 and 2? Is the driving end point (as I suspect) a much higher incidence of refractory angina in the phase 2 control patients? Second, what is the timing of the end-point events? Figure 2 (cumulative event rates in phase 2) indicates that all of the differences between groups occurred early (within 2 weeks or so), suggesting an early “window of vulnerability.” How do the phase 1 data (particularly in the control group) look in comparison? Might this not be construed as really an expression of a high number of recurrent events in the 26-hour-plus period after initiation of heparin/hirudin? Finally, were there any interactions between heparin/hirudin and coumadin/placebo in phase 2?
An even more intriguing question is raised by the phase 2 data in Figure 2. If the differences between coumadin and placebo all occur over the first 2 weeks or so, and the event curves then track parallel, couldn’t one argue that only 2 weeks of therapy with coumadin (rather than 3 to 6 months) may be necessary in patients with acute coronary syndromes? Again, this reinforces a “window of vulnerability,” a finite period within which there is a high risk for recurrent events.
It would be a mistake to broadly infer that coumadin therapy (particularly long-term therapy) is of benefit in acute coronary syndromes. I believe the more important message that emerges from the work of Anand et al is that there is an opportunity to improve the early-phase management of patients with acute coronary syndromes. The benefits of long-term coumadin therapy have yet to be established.
With additional data from the OASIS Pilot study, I hope we can begin to tease out the answers to some of the nagging questions
- Copyright © 1999 by American Heart Association
Anand SS, Yosuf S, Pogue J, Weitz JI, Flather M, for the OASIS Pilot Study Investigators: Long-term oral anticoagulant therapy in patients with unstable angina or suspected non–Q-wave myocardial infarction: Organization to Assess Strategies for Ischemic Syndromes (OASIS) pilot study results. Circulation. 1998;98:1064–1070.