New Oral Anticoagulants in Secondary Stroke Prevention: Apples and Oranges?

March 22, 2013

Collectively, three members of the new generation of oral anticoagulants are no more effective than warfarin for the secondary prevention of stroke in patients with nonvalvular atrial fibrillation, concluded an analysis that aggregated data from three large randomized trials [1]. The new agents did appear, however, to significantly reduce the risk of hemorrhagic stroke compared with the older drug.

Looking specifically at patients with a history of stroke or transient ischemic attack (TIA), the analysis also saw no relative difference in secondary stroke-prevention efficacy among the three agents, which were dabigatran etexilate (Pradaxa, Boehringer Ingelheim), apixaban (Eliquis, Bristol-Myers Squibb/Pfizer), and rivaroxaban (Xarelto, Bayer Healthcare/Janssen Pharmaceuticals).

Although this sort of analysis across trials for the different agents has been done before, it had yet to be performed in a high-risk, stroke secondary-prevention subgroup, Dr Partha Sardar (New York Medical College-Metropolitan Hospital Center, NY) told heartwire . Sardar had just presented the analysis at the American College of Cardiology 2013 Scientific Sessions.

It may be, Sardar said, that the newer oral anticoagulants would be best used in patients at increased risk of bleeding, leaving the less expensive warfarin for other patients.

In his presentation, Sardar acknowledged the conspicuous limitations of such an analysis, which combined subgroup data from RE-LY (testing dabigatran 110 mg or 150 mg twice daily), ARISTOTLE (apixaban 5 mg twice daily), and ROCKET-AF (rivaroxaban 20 mg once daily) on patients who had a history of cerebrovascular ischemia. Results for the 14 527 patients, he cautioned, should therefore be thought of only as generating hypotheses for exploration in prospective trials.

Outcomes of Oral Anticoagulation Therapy, New Agents vs Warfarin, in Patients* With Atrial Fibrillation and a History of Cerebrovascular Events

End point OR (95% CI)
All-cause mortality 0.90 (0.79–1.02)
Stroke 0.86 (0.73–1.01)
Hemorrhagic stroke 0.37 (0.19–0.72)
Disabling or fatal stroke 0.85 (0.71–1.04)
Systemic embolism 0.85 (0.74–0.99)
Gastrointestinal bleeding 1.17 (0.76–1.80)
Any major bleeding 0.84 (0.69–1.03

*Combined subgroup data from the RE-LY, ROCKET-AF, and ARISTOTLE trials.

But the analysis as presented at the meeting drew fire at its subsequent question-and-answer period, in part because of those expressed cautions against overinterpretation. And it outed conflicting views about the value of analyzing such differing trials together in the first place.

Dr Peter Kowey (Jefferson Medical College, Philadelphia, PA) rose and blasted the very idea of such analyses. "This is the second time that we've had to watch cross-trial comparisons presented," despite cautions about overinterpretation in both studies, he said, likely referring to just such an example in the Journal of the American College of Cardiology last year [2] and reported then by heartwire .

There were at least two other publications of similar analyses in 2012, with slightly varying conclusions [3,4].

"You really shouldn't be doing this, because even though you had all the disclaimers at the end of your paper about the limitations of [the analysis], it's human nature that people will look at it and think it means something," Kowey said.

Further, "I don't think we're going to see a head-to-head comparison among these drugs." It would take such a large trial that it wouldn't be feasible. "It's just not going to happen," he said.

"So saying, here's an idea, and we should explore it in a randomized trial, I think is disingenuous. I just wish we would stop doing this because it leads to false conclusions."

Sardar replied that there are few data on using these agents for secondary prevention of stroke, a randomized trial is needed, and so the analysis was performed to raise awareness of the issue.

To heartwire , Sardar acknowledged that a randomized comparison of the agents is unlikely to be conducted, and that's the very reason analyses like the one he reported can be useful. "In the absence of comparative data, we can compare hypothesis-generating results."

Neither Sardar nor his coauthors had disclosures. Kowey recently disclosed consulting for Sanofi, Boehringer Ingelheim, Merck, AstraZeneca, Johnson & Johnson, and Daiichi Sankyo and being on the speaker's bureau for Sanofi, Boehringer Ingelheim, and Johnson & Johnson.