New dabigatran safety data from RE-LY unveiled ...

Arrhythmia/EP
New dabigatran safety data from RE-LY unveiled as watchdog group seeks answers
December 8, 2010 | Steve Stiles

Vancouver, BC - Last month, without fanfare, investigators for the RE-LY trial published a reanalysis of adverse-event data, conducted to answer lingering questions the FDA had about dabigatran (Pradaxa, Boehringer Ingelheim) before it would approve the drug. The letter to the editor published in the November 4, 2010 issue of the New England Journal of Medicine [1] provides new details on neurologic events, bleeding, and infarctions that were not mentioned in the trial's primary publication in 2009 [2].

Taken at face value, the reanalysis addresses a nagging concern about RE-LY, a significantly increased risk of MI despite a significant drop in stroke rates, among patients who took dabigatran. But its investigators say nothing has changed about the trial's messages.

The new information comes in the wake of questions raised by researchers at a respected, independent drug-safety monitoring group in Canada that recently aired its concerns in the media [3] and, earlier this year, in the medical literature [4,5] about the completeness of published safety data from the >18 000-patient trial that led to the new oral antithrombin's recent approval in the US and Canada to prevent stroke in patients with atrial fibrillation (AF) [6,7], as covered by heartwire.

RE-LY revisited

As part of its review of dabigatran, the US FDA had asked RE-LY investigators to reevaluate their database "for possible underreporting of events," according to the published letter. The agency received the new numbers in early 2010 and considered them in its review of the drug prior to granting approval.

The reanalysis found scores of previously unidentified, sometimes serious adverse events that had not been part of the trial's published outcomes data but that—when added to the database—had no substantial effect on the originally reported primary end point of stroke or peripheral embolic events.

The newly identified events included strokes and transient ischemic attacks, other embolic events, instances of major bleeding, and silent MIs evident by electrocardiography.
These events did not lead to any changes in the interpretation of the results. In fact they strengthen the findings as they show the robustness of the overall results.

"This in-depth analysis found a few events related to exhaustive analysis of the RE-LY database, which is not unexpected in a trial of this size," RE-LY lead investigator Dr Stuart J Connolly (McMaster University, Hamilton, ON) wrote in an email to heartwire. "These events did not lead to any changes in the interpretation of the results. In fact, they strengthen the findings as they show the robustness of the overall results."

RE-LY compared two doses of the new anticoagulant with conventionally administered and monitored warfarin. For the stroke/embolic-events primary end point, as heartwire reported last year, there was an annualized relative risk (RR) of 0.66 (95% CI 0.53-0.82; p<0.001) for dabigatran at 150 mg twice daily vs warfarin. It changed negligibly after inclusion of the new adverse-event data: RR 0.65 (95% CI 0.52-0.81; p<0.001). The solidly nonsignificant RR for major bleeding didn't budge at all, staying at 0.93 (95% CI 0.81-1.07).

On the other hand, the primary report's increased risk of MI among patients who received that dabigatran dosage—RR 1.38 (95% CI 1.00-1.91; p=0.048)—went from significant to nonsignificant after adding the new data: RR 1.27 (95% CI 0.94-1.71; p=0.12).

At face value, the change eliminates a primary residual concern about RE-LY, that dabigatran apparently upped the risk of MI even as it was pulling the stroke risk down.

But RE-LY investigator Dr Lars Wallentin (Uppsala Clinical Research Center, Sweden), a coauthor on the letter to the journal, downplays the impact of the revised outcomes based on the new adverse-event numbers.
Our feeling is that we have just added some noise to the analysis.

The follow-up analysis, he observed for heartwire, "added a small number of MIs to both [treatment] groups, and the significance of the MI end point vanished. But our feeling is that we have just added some noise to the analysis. [Although] there is still a very small increase in MI, no longer significant, we still maintain there might be somewhat less protection against MI with dabigatran than with warfarin. But at this stage we have no statistical evidence for that."

Even without the silent-MI data, and going on the originally reported numbers, he said, the elevation in MI risk with dabigatran "is still very low. Maybe two more per 1000 patients will have an MI on dabigatran than on warfarin. . . . We know warfarin protects against MI, so some of this protection may be lost with dabigatran." Although, Wallentin noted, the greater MI rate didn't translate into increased CV mortality: that was significantly reduced for patients who received dabigatran.

"There was also a reduction in stroke and stroke mortality," he said, "so the small increase in MI seems to be of limited clinical relevance."

Complications you wouldn't expect

Questions about adverse events in RE-LY have focused on more than just MIs and bleeding. Publically reported adverse-event data for RE-LY are missing numbers for "total serious adverse events," according to clinical pharmacist Dr Aaron Tejani, a researcher at the Therapeutics Initiative at the University of British Columbia, Vancouver.

Therapeutics Initiative conducts evidence-based safety checks on drugs available in Canada and has been lauded for studies that led to restrictions on drugs that have been shown to cause more harm than good [8,9], including rofecoxib (Vioxx, Merck).

Not knowing total serious adverse events, Tejani said to heartwire, makes it impossible to gauge the full clinical impact of dabigatran relative to warfarin in the trial and therefore to adequately appreciate the drug's risks.

Yet dabigatran at 150 mg twice daily was reported to show a "net clinical benefit" in the trial compared with warfarin. Tejani pointed out that was based predominantly on adverse events such as death, MI, stroke, peripheral embolism, bleeding, and other conditions with recognized associations with anticoagulation therapy.
We know that, especially with new drugs, there can be complications you wouldn't expect that are serious.

"But we know that, especially with new drugs, there can be complications you wouldn't expect that are serious, that you need to look for," he said. "Some might say, What if it's something obscure that you wouldn't normally pin on dabigatran? Well, if there's a higher risk of it in the dabigatran group, because you've done a randomized controlled trial, you have to at least, on the surface, attribute it to dabigatran—and then do some more investigation."

Tejani and his colleagues published their concerns about the trial's adverse-event data and other issues in a letter in the March-April 2010 issue of the Canadian Journal of Hospital Pharmacy [4]. He says the reply from RE-LY investigators in the subsequent July-August issue [5] didn't adequately answer his questions.

"Although they claimed a net clinical benefit for dabigatran, we didn't feel that was justified."

Having unsuccessfully asked RE-LY investigators to see the total serious-adverse-event data he was seeking, Tejani said he turned to Boehringer Ingelheim, the trial's sponsor.

The only way he could see the data, the company told him, was in the setting of a formal review by a hospital considering dabigatran for its formulary. In that case, he said, he would have to sign a confidentiality agreement forbidding that he share the data with anyone but the hospital decision makers. Tejani declined.

[Calls to the company by heartwire to clarify its policies on data sharing were not returned by the time this story published.]

Ascertainment containment

Tejani also contends that the multilayered adjudication process for interpreting the new clinical events, which is spelled out in the FDA's own medical review of dabigatran, left the reanalysis vulnerable to ascertainment bias that could have favored outcomes with the new drug.

The process called for three successive tiers of data review, the first two unblinded; identified events in the unblinded first review were assessed at the unblinded second tier, and then the blinded third tier made the final calls.
We're saying that there's enough concern, enough differences in interpretation of the same story, that we need to look at the drug's safety more closely.

The reviewers behind the FDA document acknowledge that the process was subject to possible ascertainment bias but concluded that "at this time, we think the data are of sufficient quality to allow substantive review." The agency went on to grant dabigatran market approval.

Tejani points out that Boehringer Ingelheim had significant input into the design of RE-LY, randomization in the trial was unblinded, "a lot of the reassessment of outcomes was done by unblinded investigators," and most if not all of the main publication coauthors had disclosure statements showing different kinds of financial relationships with the companies. "We're saying that there's enough concern, enough differences in interpretation of the same story, that we need to look at the drug's safety more closely."

Infarcts and gastric bleeds

Tejani maintains that the MI signal in RE-LY is a concern, especially when considered along with other suggestions of dabigatran-related side effects and adverse events.

For example, although the risk of major bleeding was similar for dabigatran at 150 mg twice daily and warfarin in RE-LY, he points out, gastrointestinal bleeding was significantly more common with dabigatran.

The primary RE-LY paper [2] puts the RR for any kind of gastrointestinal bleeding for dabigatran at 150 mg twice daily vs warfarin at 1.50 (95% CI 1.19-1.89). In the FDA review, the corresponding RR for adjudicated major GI bleeding was 1.47 (95% CI 1.17-1.85) and for life-threatening GI bleeding was 1.62 (95% CI 1.17-2.26).
The public and physicians can be pretty confident that this drug has been subjected to the proper reviews.

Connolly, asked to respond to a story in the media [3] about Tejani's concerns, emailed heartwire to say: "The FDA approved the drug for AF after their extensive reviews of safety. The labeling in the United States and Canada is very transparent about the increased rate of gastrointestinal bleeding with dabigatran and also lists the rates of myocardial infarctions on dabigatran and warfarin. . . . Dabigatran has undergone extensive review of its safety and efficacy and received approval in Canada and USA based on those reviews done by the agencies themselves—not based on our analyses. So the public and physicians can be pretty confident that this drug has been subjected to the proper reviews."

But Tejani insists that the gastrointestinal bleeding and MI signal in the trial have not been emphasized enough to clinicians or the public, and all RE-LY data relating to safety should be openly available.

"We're looking for full transparency. We're not saying that dabigatran is necessarily an unsafe drug. All we're saying is that it's hard to tell whether it's good, bad, or neutral, based on only one large trial. So wouldn't it be smart to repeat the experiment and see which is correct?"

Shelley Wood and Sue Hughes contributed to this report.
RE-LY was sponsored by Boehringer Ingelheim; disclosures for Connolly, Wallentin, and other individual coauthors of the RE-LY primary paper are included in the report [2]. Tejani reports no conflicts of interest.