Extended Thromboprophylaxis in Medically Ill Patients - Episode 7

MAGELLAN Trial Review and the Risk of Bleeding

Published on: 

Deepak Bhatt, MD, MPH: Those are really good points. Alex, did you want to add?

Alex C. Spyropoulos, MD, FACP, FCCP, FRCPC: What I’d like to do is really build on these are critical concepts that both Gary and Mike have introduced. And now we’re much more mature in our understanding of extended thromboprophylaxis. What are the patient populations at risk and what are the benefits of extended prophylaxis? The only thing that I’d like to do is introduce both the original MAGELLAN trial as well as the MAGELLAN subpopulation, which gained FDA approval for rivaroxaban for the indication of extended prophylaxis.

Much like the APEX trial, MAGELLAN randomized patients in the first 72 hours of their hospitalization to either extended duration prophylaxis with rivaroxaban given for 31 to 39 days, or the standard of care approach using enoxaparin, a low molecular weight heparin, given 40 mg once daily. There were, unlike APEX, 2 co-primary efficacy end points, 1 at day 10 trying to establish noninferiority with enoxaparin, and then a secondary end point at day 31 to 39, trying to establish superiority over placebo. And the MAGELLAN trial met both of those co-primary end points. At the end of the day, it was noninferior to enoxaparin, and then in the outpatient phase or post-hospital discharge phase, there was about a 23% or so risk reduction favoring rivaroxaban from 5.7% to about 4.4% and, again, mostly driven by asymptomatic proximal deep vein thrombosis found by screening ultrasonography.

Now, the catch of that was there was a price to pay, so you did rob Peter to pay Paul. With that strategy, major bleed rate went from about 0.4% to 1.1%. So there was an increased risk of major bleeding, producing unfavorable net clinical benefit. So what was done is, as a group of trialists, we went looking at the MARINER trial database. We established some key risk factors for bleeding, and these were 5 key risk factors for bleeding: history of dual antiplatelet therapy, history of a recent bleed within 3 months, history of active gastrointestinal ulcer, history of active cancer, and pulmonary cavitation of bronchiectasis. These were some of the key exclusionary criteria that established such a low bleed risk population within MARINER, and I believe, Mike, that there was very similar exclusionary criteria within the APEX trial as well.

C. Michael Gibson, MS, MD: We wanted to send you a fruit basket because you identified all those high-risk patients, and we took those off the table. We also though took a lesson from MAGELLAN on the D-dimer. That’s where a lot of the benefit was derived, and that was really why we went with that higher risk group.

Alex C. Spyropoulos, MD, FACP, FCCP, FRCPC: I think based on those key lessons learned from MARINER, what we did then is we applied those 5 key exclusionary criteria retrospectively in the MAGELLAN database, and I think we found some important conclusions. The efficacy of rivaroxaban was maintained. If anything, maybe there was a more even pronounced risk reduction in the subpopulation. By the way, we only excluded about 20% of the MAGELLAN population. We were left with 80% of patients to do these analyses, a robust population.

So the efficacy was maintained. If anything, there was a more pronounced effect. So about 5.7% versus 3.9%, 32% risk reduction. Then the safety was markedly reduced, so all of a sudden that almost 3-fold increased risk of major bleeding was cut in half in both phases, the hospital-based phase and the post-discharge phase, and no longer statistically significant. The number of fatal bleeds was reduced dramatically, numerically, in both treatment phases.

Then all of a sudden the data looked very favorable. The rates of major bleeding were 0.5% and 0.7%, almost identical between both populations. As Mike had described, when one conducted formal benefit-risk analysis, based on pairs of efficacy and safety outcomes that were similar in clinical severity, we were trying to compare apples to apples. We compared VTE [venous thromboembolism]-related death, for example, with fatal bleeding. We compared symptomatic VTE and VTE-related death with critical sites and fatal bleeding, etcetera. Or the primary efficacy end point, which included asymptomatic disease with major bleeding. We found in general that we were able to prevent anywhere from 2 to 10 major or fatal thromboembolic events for every major and fatal bleed incurred.

An overall, very favorable net clinical benefit. And I think now as clinicians we have 2 drugs in our armamentarium that we can, number 1, confidently give, and then number 2, safely give the patient, both in the inpatient phase, but I think much more importantly where I sit, in the post-discharge phase.

Gary Raskob, PhD: We’ve talked about net benefit and NNT [number needed to treat], but just to put a risk on it, we just presented at the American Heart Association this weekend the pooled analysis of MAGELLAN and MARINER.

Deepak Bhatt, MD, MPH: Oh, so literally hot off the presses.

Gary Raskob, PhD: Yes.

Deepak Bhatt, MD, MPH: That’s terrific.

Gary Raskob, PhD: For the extended out-of-hospital phase in the analysis of 16,000 patients, the risk of a fatal or a critical site bleed, like an intracranial bleed—a serious bleed that clinicians really are worried about—was 0.05%, or 5 per 10,000 people. I think the real message that we want to emphasize is that we have now an easy to use, simple approach, and the safety of these oral anticoagulants has been much improved now. We know how to select the patients and the drugs are there to use for this, so we can really reduce the burden of VTE if we apply this.

Deepak Bhatt, MD, MPH: That’s a good message.

Gregory Piazza, MD, MS: Deepak, I would add that some of this work is critical if we’re going to incorporate the assessment of risk of VTE but also risk of bleeding into medical informatics. These 5 factors that increase the risk for a major bleeding can be easily incorporated either through machine learning or into some sort of a computerized decision support tool. That will make the process of getting providers to recognize risk and to prescribe the right therapy more precise, so we’re not necessarily alerting them to give prophylaxis to a patient that might be high risk for bleed. I think that’ll be well accepted.

Transcript edited for clarity.