CardioExchange Archive

Once again, dabigatran (Pradaxa) has raised the wrath of the critics. Several articles (see here, here, and here) and an editorial published today in The BMJ raise more questions and concerns about the drug, which is the first of the new oral anticoagulants. Relying on new evidence along with previously disclosed data, Deborah Cohen, the investigations editor for The BMJ, casts doubt on the reliability of the data supporting the drug as well as the behavior and decisions of regulatory authorities, trial investigators, and employees of Boehringer Ingelheim (BI), the drug’s manufacturer. Although the articles focus on dabigatran, it is entirely possible that similar questions may also be raised about other new oral anticoagulants, including rivaroxaban and apixaban.

All the new oral anticoagulants are trying to displace warfarin, the long-standing and much-maligned standard of therapy. Warfarin has been vulnerable because it requires constant monitoring of its anticoagulant effect and dose adjustments. A chief advantage of dabigatran is that it does not require monitoring, but critics have maintained that this advantage may be somewhat illusory, and they have raised repeated concerns that dabigatran may cause excessive bleeding complications. In her articles, Cohen maintains that BI failed to fully disclose findings that support these concerns and that regulatory authorities, perhaps in their eagerness to approve an alternative to warfarin, relaxed their standards of scrutiny in approving the drug.

The most important new accusation in The BMJ articles is that BI “failed to share with regulators information about the potential benefits of monitoring anticoagulant activity and adjusting the dose to make sure the drug is working as safely and effectively as possible. The company also withheld analyses that calculated how many major bleeds dose adjustment could prevent.”

Analyses performed by the company, Cohen reports, found “that if the plasma levels of the drug were measured and the dose was adjusted accordingly major bleeds could be reduced by 30-40% compared with well controlled warfarin. The adjustment would have little or no effect on the risk of ischemic stroke.” The internal report concluded that “Optimally used (=titrated) dabigatran has the potential to provide patients an even better efficacy and safety profile than fixed dose dabigatran and also a better safety and efficacy profile than a matched warfarin group.”

BI responded to this allegation in a press release:

It is inappropriate to provide regulators simulations that are unreliable and have limitations. Post-hoc exploratory analyses are commonly performed to generate or test hypotheses and are not structured to direct patient management. Thus, they generally are not shared with regulators and first need to be tested in a clinical trial, as many hypotheses, such as the one discussed here, prove to be incorrect.

Regulators were aware of wide variations in the plasma levels of the drug. At the 2010 FDA advisory panel, one committee member, Darren McGuire, even commented:

“I’m struck by what my eyeball tells me about a five-fold variability [in plasma levels] within the 90% confidence [interval] of the 150-dose. That seems awfully big to me in a drug that we’re proposing to use without therapeutic monitoring.”  But, notes Cohen, “McGuire’s concerns were not pursued by the agency.” Cohen cites internal BI documents in which Stuart Connolly, a principal investigator of the pivotal RE-LY trial, expresses similar concerns about this variability.

“Our scientists determined, and the FDA concurred, that the research does not support making dosage decisions based on plasma concentrations—a conclusion based solely on science and patient welfare,” the BI spokesperson told Cohen.

The BMJ articles also provide additional details about multiple efforts to re-examine the RE-LY data that uncovered additional events not originally reported.

In an accompanying editorial, Blake Charlton and Rita Redberg write that the articles raise “serious questions about” dabigatran.The material “illuminates a lack of transparency about the safety of unmonitored dabigatran, compounded by the drug’s fickle pharmacokinetics, which can cause a fivefold variation of plasma concentration.” “Equally unsettling,” they write, “are data integrity issues.” The editorialists write that their “clinical concerns regarding dabigatran include potentially higher than reported bleeding risk; the possibility of undertreating or overtreating with fixed doses, especially in older patients and patients with changing renal function; the unknown value of monitoring dabigatran levels and adjusting the dose; and the lack of a specific reversal agent.”

It should be noted that similar charges against dabigatran have been brought previously. It is unclear whether the articles in The BMJ contain significant new information.

Stuart Connolly provided several comments in response to The BMJ articles:

1. To my knowledge there was never any withholding of information related to the concentration response relationship data. The concentration data was all filed with regulatory agencies who reviewed it independently. There was substantial discussion amongst the principal investigators and company representatives regarding many of our papers on the RELY data, with … disagreements from time to time and some vigorous discussions. The review of the concentration response data was no different. After much discussion we agreed that we were not in a position to recommend a fixed ‘therapeutic level’ for dabigatran. Indeed even if we had it would need to be prospectively verified with a new study. When some of us did actually try to target a therapeutic range for dabigatran in the subsequent valve study we found it difficult to do and the results were inconclusive.

2. In regards to the reviews of the RELY data, it is true that we did go back into the database to look for potentially unreported events and we did find some. However it is not surprising that we found events. In a trial of 18,000 patients with lots of co-morbid disease and multi-system failure from time to time, it is not surprising that if you look harder you will find more events even though the initial reporting of events was rigorous and met usual standards. The definition of bleeding used in these types of anticoagulation trials is quite comprehensive and all encompassing. In fact not only did we find very few ‘new’ events despite vigorous searching but the occurrence of these events was generally balanced across the three treatment groups suggesting an absence of any bias in reporting of events. The robustness of the main results of RELY is substantiated by the fact that adding these events had no impact whatsoever on the interpretation of the trial.

3. It is of note that the subsequent analyses by the FDA of their mini-sentinel and other databases showing the effects of dabigatran in clinical practice in the USA (mostly the higher dose) have strongly indicated that the effects in practice in tens of thousands of patients from a variety of types of practice are almost exactly what would be predicted by the RELY trial main results.

Therefore I remain entirely confident in the results of RELY and am confident that dabigatran performs in practice as predicted by RELY. There was no significant lack of transparency in the reporting of the RELY trial. I do not believe that we have sufficient evidence that adjusting the dose to achieve a specific therapeutic concentration would improve outcomes, although this would be an interesting hypothesis to test.