New post-hoc analyses of ROCKET-AF confirm the main results of the controversial trial, according to a paper published in the New England Journal of Medicine. The papers are a response to disturbing questions about the trial raised last fall. But the new analyses will not fully satisfy some critics of the trial who are calling for independent analysis of the trial.
ROCKET AF was the pivotal trial that compared rivaroxaban (Xarelto, Johnson & Johnson) to warfarin in patients with atrial fibrillation. Last year it became known that the portable device used to monitor and calibrate warfarin usage in the trial had been seriously defective. The device– the Alere INRatio and INRatio2 PT/INR Monitor System– was the subject of a broad class 1 FDA recall. The FDA said that the devices “may provide an INR result that is lower than the expected result obtained using a laboratory INR method” in patients with certain conditions, including abnormal hematocrit levels, conditions associated with raised fibrinogen levels, and bleeding or unusual bruising.
In an interview last fall, Duke’s Manesh Patel, a member of the ROCKET AF executive committee, said that the ROCKET AF investigators first became aware of the problem in October and immediately began a series of analyses to assess the impact of the problem. They said that the results were “consistent with the results from the original trial and do not alter the conclusions of ROCKET AF” but they did not then release the details of their study.
Now the details of their analyses have been published in the New England Journal of Medicine. In their studies the investigators identified patients who had conditions linked to the recall of the INR device and asked whether the results of the trial were any different in this group, and specifically whether rivaroxaban appeared to do better because warfarin was not used properly.
“These results are consistent with the overall trial findings and indicate that possible malfunction of the point-of-care device used for INR measurement in the ROCKET AF trial that potentially led to lower iNR values than would be obtained by laboratory testing did not have any significant clinical effect on the primary efficacy and safety outcomes in the trial,” the investigators concluded.
In an interview today Patel said the new results are broadly consistent with a separate analysis conducted by the drug’s sponsors and presented to the FDA. In a statement, the drug’s sponsors said that the “findings are in line with the sensitivity analyses conducted by Bayer and Janssen, which also affirm the results of the ROCKET AF study and the positive benefit-risk profile of” rivaroxaban.
One expert, Sanjay Kaul (Cedars-Sinai), who served on the FDA advisory committee that reviewed ROCKET AF, said that the “investigators have done the appropriate due diligence.” He noted that “the observation of a higher relative risk of major bleeding with rivaroxaban in those with recalled conditions (HR of 1.18) contradicts the claim that underestimation of INR by the recalled device could have amplified warfarin bleeding thereby enhancing the safety of rivaroxaban.”
But in a separate paper published in BMJ, associate editor Deborah Cohen quotes several leading experts who say that the trial data should be made available for independent analysis by outside investigators not involved with the trial.
One of these experts, Harlan Krumholz (Yale Open Data Access project) praised the analysis of the trial but said this is “the perfect case” to support the recent ICMJE proposal that would mandate sharing of trial data. Krumholz said that there are still remaining questions about the trial. “I think the question of whether the study is valid is still unanswered. There should be a note associated with this study that explains to readers that one arm was managed with measures from a device that was subsequently recalled.” Krumholz told the BMJ that although Johnson & Johnson has agreed to make trial data available, its partner on the drug, Bayer, has refused.
Cohen also called for an independent analysis of the trial data. Both Cohen and Krumholz also want to know why the investigators did not perform a comparison of the point of care tests with blood samples stored in the core lab. In an email Cohen told me:
Their reanalysis has a fundamental flaw. It assumes the the defect in INRatio only affects certain subgroups of patients. The investigators could have and should have evaluated this assumption. They have a huge dataset to evaluate the performance of the device. They took split samples at 12 and 24 weeks and tested those with both INRatio and lab tests. The investigators should not only compare the point of care measurements with lab results in those patients in the recall notice, but also include evaluate those without a recall related condition.
I asked Patel about this question. He said the trial investigators did not perform this analysis because their studies were designed to address the problems identified in the specific patient population identified in the FDA recall. But, he said, the INR data from the core lab is available for analysis and may be the subject of future studies.
Patel expressed frustration at some of the criticism directed at the ROCKET AF investigators. He noted that it only took four months from the first notification of the problem with the device to the publication of the paper in the New England Journal of Medicine.