ROCKET AF Investigators Respond to New Questions About Point-of-Care Warfarin Monitoring Device

Continuing to address concerns about the device used to measure international normalized ratio (INR) values in ROCKET AF, trial leaders have released additional analyses of stored blood samples showing that even though there are discrepancies between point-of-care and laboratory measurements in some patients, they do not seem to have had an impact of the overall conclusions of the study.

Indeed, when patients with discrepant values are excluded, rivaroxaban (Xarelto; Janssen Pharmaceuticals) remains noninferior to warfarin in terms of stroke or systemic embolism, with no difference seen in the rate of major or nonmajor clinically relevant bleeding, Manesh Patel, MD (Duke Clinical Research Institute, Durham NC), a member of the ROCKET AF executive committee, and colleagues report in correspondence published online July 6, 2016, ahead of print in the New England Journal of Medicine.

“We conclude—although there are limitations—that we again don’t see anything that takes us away from the primary conclusions of the trial that rivaroxaban is noninferior to warfarin,” Patel told TCTMD. “I am reassured.”

Patel added that trial leaders are open to suggestions about other ways to think about and study the issue, but said that there are no further analyses planned.

The History

In December 2014, Alere issued an alert through an FDA recall notice about its INRatio and INRatio2 PT/INR Monitor System, which was used to measure INR levels in patients in both the warfarin and rivaroxaban arms of ROCKET AF. The company informed users of the device that patients with certain conditions—including anemia with a hematocrit less than 30%, unusual bleeding or bruising, and conditions associated with elevated fibrinogen levels—should not be tested because of the possibility of getting INR results lower than would be obtained with laboratory testing.

The concern is that an inaccurate reading on the low side might induce physicians to increase warfarin doses to raise INR levels, potentially resulting in excess bleeding in the warfarin arm and a more favorable assessment of rivaroxaban’s safety.

Trial investigators have said that they learned about the issue in October 2015. In response, they performed a post hoc analysis, which showed that the overall results of the trial were similar both in patients with and in those without any of the conditions listed in the FDA recall notice.

Shortly after, the European Medicines Agency (EMA), having reviewed the issue, determined “that there is sufficient evidence to conclude that the benefit/risk balance remains unchanged and favorable for treatment with rivaroxaban in the prevention of thromboembolism in nonvalvular atrial fibrillation.”

The FDA is also exploring the issue, in regard both to the Alere systems used in ROCKET AF and to other similar point-of-care tests.

New Questions

In a letter to the editor published alongside the latest from Patel et al in NEJM, J. Robert Powell, PharmD (University of North Carolina at Chapel Hill), indicates that the issue regarding ROCKET AF remains unsettled.

“Although the EMA found insufficient evidence to alter the benefit-risk conclusion of the original rivaroxaban study, the agency did not address several important questions,” he writes. “How many warfarin-treated patients had major bleeding as a result of this error in the point-of-care device? Did this error result in a change in the percentage of time that patients receiving warfarin were in the therapeutic range? Answers to these questions could alter the relative benefit-risk relationship between rivaroxaban and warfarin. They could also diminish the validity of the ROCKET AF study.”

Patel and colleagues responded with the additional analyses, which were based on stored blood samples drawn on the same day that point-of-care tests were performed at 12 and 24 weeks of follow-up. The samples, which represent just 6% of the point-of-care INR tests administered during the trial, were frozen and sent to a central laboratory as part of a pharmacodynamic assessment.

A discrepancy between the point-of-care and lab values was found in 13% of samples obtained at either 12 or 24 weeks and in 4% of samples at both time points. Discrepancies were not related to the presence of conditions listed in the FDA recall notice.

The researchers found that both bleeding and stroke were more frequent in warfarin-treated patients with discrepant values, whereas the bleeding rate was higher but the stroke rate was lower in rivaroxaban-treated patients with discrepant values.

And finally, they showed that excluding patients with discrepant INR values did not change the conclusions of the trial.

“However, we acknowledge the limitations of these analyses,” Patel et al write. “To be fully informative, we would need to provide paired central-laboratory and point-of-care INR values throughout the trial, and these values are not available.”

Powell told TCTMD that the new analyses do not fully address the lingering questions about the point-of-care device, adding that he doesn’t think it can be resolved one way or another with further studies.

“There’s uncertainty around what the results are because you can’t accurately understand the impact of this error,” he said, suggesting that there should be an asterisk next to the trial akin to the one placed next to sports records broken by steroid users.





Related Stories:

  • Patel MR, Hellkamp AS, Fox KAA, et al. Point-of-care warfarin monitoring in the ROCKET AF trial [correspondence]. N Engl J Med. 2016;Epub ahead of print.

  • Powell JR. Point-of-care warfarin monitoring in the ROCKET AF trial [correspondence]. N Engl J Med. 2016;Epub ahead of print.

  • Patel reports receiving grant support from Janssen and Johnson & Johnson and personal fees from Bayer and Janssen during the conduct of the study. He also reports receiving grant support from AstraZeneca, CSI, Heartflow, Maquet, Medtronic, and the National Heart, Lung, and Blood Institute and personal fees from AstraZeneca, CSI, Genzyme, Medtronic, and Merck unrelated to this study.
  • Powell reports no relevant conflicts of interest.

We Recommend