Q&A: An Indirect Comparison of Andexanet Alfa, Prothrombin Complex Concentrate in Patients With NOAC-Related Bleeding

March 30, 2020

The American Journal of Managed Care® discussed an indirect comparison of andexanet alfa and prothrombin complex concentrate therapy with Alexander T. Cohen, MBBS, MSc, MD, Guy’s and St Thomas’ Hospitals NHS Foundation Trust.

The American Journal of Managed Care® (AJMC®) discussed an indirect comparison of andexanet alfa and prothrombin complex concentrate (PCC) therapy with Alexander T. Cohen, MBBS, MSc, MD, Guy’s and St Thomas’ Hospitals NHS Foundation Trust. Cohen’s study, titled “30 Day Mortality Following Andexanet Alfa in ANNEXA-4 Compared with Prothrombin Complex Concentrate (PCC) Therapy in The ORANGE Study for Life-Threatening Non-Vitamin K Oral Anticoagulant (NOAC) Related Bleeding,” found that adjusted mortality rates were lower in patients receiving andexanet alfa than in matched patients receiving PCC. The abstract was released through the American College of Cardiology (ACC) and World Congress of Cardiology (WCC) Annual Meeting—ACC.20/WCC Virtual.

AJMC®: Can you discuss what causes NOAC-related bleeding and how it can prove fatal?

Cohen: Well, NOAC-related bleeding is no different from any other anticoagulant related bleeding as it's not actually new bleeds, they're just bleeds that we might have in our normal daily life or in other circumstances that are somewhat amplified by the presence of an anticoagulant. The NOAC-related bleeds occur in different locations to other anticoagulant related bleeds, but they can occur anywhere like all bleeds.

For example, NOAC-related bleeds are less likely to be bleeding into the brain, but they might be as likely or more likely to occur from an epithelial surface like the gastrointestinal tract, or the urothelial tract. Although they occur less frequently, and they're less severe, they do result in fatalities, and we know that if you do bleed into the brain, you probably have about a 30-40% chance of dying. If you bleed into the gut, that's somewhere between 10-30%, depending on other things like your comorbidities or how bad the bleed is.

So, NOAC-related bleeding is related to all the things that other bleeds are related to. It’s related to, for instance, being either traumatic or spontaneous. If it’s spontaneous, it may be related to underlying pathology, such as ulcers in the gut or damage to the lungs with infections that lead to bleeding or chronic lung disease or cancers that are revealed by the fact that they're bleeding anyway, but they bleed more when you’re anticoagulated—or maybe related to kidney failure or liver failure. It’s related to age, we know that older people bleed more, and people with a history of bleeding are more likely to bleed.

NOAC-related bleeding is still a common complication of these anticoagulants because although their safety profile is better, we are more confident in using NOACs in perhaps an older and perhaps a sicker population than what we might have used in previous days with vitamin K antagonist anticoagulants. That’s because the risk benefit is different. So, we’ve changed the population a little bit now as to who we're treating with anticoagulants.

AJMC®: Recommendations for treatment of NOAC-related bleeding include off-label PCC therapy, as well as an approved antidote, andexanet alfa. Can you explain what warranted an indirect comparison of the efficacy of both options and subsequent 30-day mortality findings?

Cohen: There’s no good evidence that PCCs are effective in treating anticoagulant related bleeding, and that’s true, not just for NOACs, but it’s also true for vitamin K antagonists. In the only randomized trial of PCCs in vitamin K antagonist related bleeding, there was no difference in clinical outcomes, no difference in transfusions, and no difference in mortality. So, although PCCs are indicated for vitamin K antagonist bleeds, there's very little evidence for giving them from a clinical point of view.

This is also true for NOAC-related bleeds. If we look at the data, there have been single arm studies where PCCs have been given for NOAC-related bleeds, and there are indications that it might either make no difference or be minimally effective, but they are not strong indications and this probably just depends on the population you're treating. In other words, there's no clear evidence that PCCs have a role. If we look at the experimental studies in patients that have been given non-vitamin K or NOAC anticoagulants, we can see that the PCC doesn’t seem to reverse the coagulopathy in patients that are getting therapeutic or high therapeutic levels. It will only reverse the coagulopathy in patients that have very relatively low levels of the NOACs and that’s important because most of the patients that bleed on NOACs have the therapeutic or high therapeutic levels. So, experimentally, it’s shown the PCCs don’t affect the coagulation and reverse the anticoagulation.

Whereas we know that in patients that have received NOACs and given andexanet alfa, for both apixaban and rivaroxaban you get a fast reduction in anti-Xa levels and you get a fast reduction in anticoagulation as evidenced by a correction of thrombin potential, which is a global assay for coagulation. So, there’s just more evidence for andexanet alfa, and furthermore, within the ANNEXA-4 study, we could see very clearly that there was a moderate correlation between the drop in anti-Xa levels and the effect of hemostasis or stopping bleeding in the brain of patients that had intracerebral hemorrhages or brain hemorrhages. So, there’s evidence there just looking within the study.

We also saw that those patients that didn’t get a response in the ANNEXA-4 study had median levels of anticoagulants that were much higher than those that did get a response, which once again, tells you that there’s probably a dose effect and it’s indicative that perhaps in some patients, you might need more of the andexanet alfa than what we gave in the study. So, there are a number of things that point to the efficacy of andexanet alfa in treating anticoagulant- or NOAC-related bleeding. There’s nothing like that sort of evidence with PCC.

As there wasn’t any evidence for comparing andexanet alfa with PCC, we performed a propensity-score matched analysis from 2 different databases—1 being the ANNEXA-4 database for andexanet alfa and the other being the ORANGE study database for PCC. We showed that in those patients, that were as well matched as we could do within the limits of the studies, andexanet alfa was associated with a roughly 50% reduction in total mortality compared with PCCs. This is true for all the patients where there was about a 57% reduction, or the intracranial hemorrhage patients where there was about a 2-thirds reduction in mortality, and there was also a non-significant trend for reductions in mortality in the patients with GI hemorrhages.

So, it seems based on the data we have, albeit with limitations because these are observational data—what we call comparative effectiveness, it seems that andexanet alfa is associated with lower mortality than PCCs for life threatening and critical site bleeds.

AJMC®: What is the significance of including patients with a varied history of conditions such as atrial fibrillation and venous thromboembolism?

Cohen: The importance of doing this is because anticoagulated patients are a mixed bunch. They’re mainly made up of patients who have atrial fibrillation, but there’s also a significant proportion that have venous thromboembolism. There are other reasons why people receive NOACs apart from those 2 conditions, but they're far less common. By having a good case mix of patients with both atrial fibrillation and venous thromboembolism, and having a case mix of patients who have intracranial bleeds, GI bleeds, and bleeds in other sites like bleeds into the eyes, joints, or into the pericardium or the retroperitoneal space—this gives us a much broader picture of bleeding and makes the findings from our propensity-score matched analysis more generalizable.

So, the significance of including the patients with varied histories, varied bleeds, and varied drugs, because they could be on apixaban or rivaroxaban, is that this really covers a much broader range of patients and therefore the findings will be more generalizable. In the future, it would of course, be interesting to look at the subgroups and see how they behave separately. So, compare the different anticoagulants, compare the different indications such as atrial fibrillation or venous thromboembolism, and compare the different types of bleeds, which we did in our subgroup analyses where we looked at intracranial bleeds separately, GI bleeds separately, but we couldn't look at other bleeds separately because there are only 8 cases of other types of bleeds in the PCC arm of the propensity-score matched analysis.

AJMC®: Do some patients react differently to treatment and are any specific conditions more at risk for adverse events?

Cohen: We found in the ANNEXA-4 study that perhaps the patients with the higher levels of anticoagulants didn't respond as well to those that had therapeutic levels or high levels. It was the ones with the very high levels, but that didn’t respond as well. We also know that patients that have intracranial bleeds have a much higher mortality rate than patients that have gastrointestinal bleeds. Also, of course, the mortality rate with other bleeds might be much lower or nonexistent. For instance, if you bleed into a knee joint or hip joint, that’s not normally associated with mortality unless there is a complication of that bleed.

We also have to consider that patients will behave differently depending on their comorbidities. So, for instance, patients that have renal failure or renal impairment, are going to metabolize the drug much more slowly so the drugs are going to hang around more. Those sorts of patients may require different dosing, but we don’t know about that yet, we’re hypothesizing that they might need extra doses or higher doses. People are more at risk if they have comorbidities, and if they've had recent thrombotic events, that’s a risk too, because when you reverse the anticoagulation, they’re back at risk of having another thrombotic event.

We expect that these patients will have a fairly high baseline of thrombotic events, because they’re prothrombotic in the first place—that’s why you’re anticoagulating them. Then when you reverse the anticoagulation, you’re putting them acutely back in that prothrombotic state. When we’re bleeding, our bodies muster our ability to coagulate to stop the bleeding. So it’s sometimes a perfect storm when you have someone who’s bleeding, who’s highly thrombotic, and then you reverse the anticoagulation, you may see a return to thrombosis.

The other questions arise as to whether reversal agents may be prothrombotic. We haven’t seen any clear evidence of that at this stage. We think that the thrombosis rates we’re seeing in the observational data, and in the ANNEXA-4 study are consistent with what we’d see in patients whom we reverse anticoagulation.

AJMC®: As the study was an indirect comparison of the 2 therapies, what further research may be considered to confirm the study findings?

Cohen: So, the study was an indirect comparison of the 2 therapies, and therefore open to confounding in particular, but also some biases. To minimize the confounding, we match the patients as closely as possible, but we couldn’t match on a number of things—like for instance, the volume of the bleed or the severity of the bleed with respect to the hemodynamic response, that was all measured in ANNEXA-4, but it wasn’t measured in the ORANGE study.

To deal with that, there are other studies going on at the moment, 2 of which I think will be important. One will be presented in the next couple of months, which is another indirect comparison where the patients in ANNEXA-4 were able to be compared to the patients on PCC, but in this case, they knew about the volumes of the bleeds so they could match for the volumes of the bleeds. So, studies like this of comparative effectiveness with observational data, need as much information as possible about the patients, and this new study I think will provide that.

Of course, the ultimate test of comparing 2 therapies is a randomized controlled trial. At the moment, we’re conducting a randomized controlled trial of andexanet alfa versus PCCs. This is the ANNEXA-I study, we’re actively recruiting into that study now, and we will need hundreds of patients, but it’s heading in the right direction and we have many, many centers all around the world recruiting into that study.