Dr Devi Nair:
Welcome everyone. I’m Dr Devi Nair, joining you at ACC 2026 in New Orleans from Arkansas. Today we will have the opportunity to take a deep dive into the CHAMPION‑AF trial that was just presented as a late‑breaking clinical trial at ACC 2026. With me today I have Dr Vivek Reddy from New York, who is going to discuss CHAMPION‑AF with me. Welcome, Vivek.

Dr Vivek Reddy:
Oh, thanks for having me, Devi.

Dr Devi Nair:
So I want to just dive in and ask you if you can give us a little background on what the trial design was for CHAMPION‑AF and what question CHAMPION‑AF was trying to address.

Dr Vivek Reddy:
Sure. So left atrial appendage closure has obviously been a therapy that people have used in patients who cannot take anticoagulation, and there are a couple of trials that have looked at this: PROTECT AF, PREVAIL, PRAGUE‑17, and more recently CLOSURE‑AF. But there was a question as to: many patients actually can take anticoagulants, but they do not want to for various reasons—either because of nuisance bleeding, because they just do not want to take a blood thinner because of the concern of what happens if they fall, and frankly cost; these are not cheap drugs, as we are well aware. So the question was: does it make sense to consider left atrial appendage closure as a real alternative to anticoagulation, not because the patients cannot take anticoagulation, but for these other reasons?

So that is the background, and CHAMPION‑AF was a very large randomized trial where patients who had atrial fibrillation and could take oral anticoagulants were randomized in a 1:1 allocation to either continued anticoagulation—basically non–vitamin K antagonist oral anticoagulants, mainly apixaban—or receiving left atrial appendage closure with a WATCHMAN FLX device. I should note that with the WATCHMAN FLX device the strategy was that after the device is implanted, the patients stayed on mainly NOACs—they could be on dual antiplatelet therapy, but mainly NOACs—for about three months, then they transitioned to low‑dose aspirin, and then at the four‑month time point they underwent imaging, mainly transoesophageal echocardiography, and assuming that was okay, they then continued on low‑dose aspirin. So those were the two randomized arms.

Dr Devi Nair:
Definitely a very different post‑implant antithrombotic regimen and imaging regimen than we have followed in earlier WATCHMAN trials.

Dr Vivek Reddy:
Yeah, it is different. But on the other hand, frankly, this is the sort of strategy that many of us—you, me, a lot of us—have been using for a long time. The idea of doing the TEE at six weeks while the patient is still on anticoagulation and then not doing another one until one year, I always found that a little troubling because the TEE at six weeks really is not that helpful: patients on anticoagulants are not going to have device‑related thrombus, and the peri‑device leak rates are going to be similar whether it is at six weeks or later. So the idea was: wait three months to allow time for endothelialization, and then stop the drug, wait another month. You could argue that that month is a vulnerable period, and technically yes, but prior data show the risk of stroke during that month is exceedingly low; in fact, there were no strokes in that time period. So you have to weigh the benefits of that versus saying, “Okay, I am going to do it at six weeks, stop the drug, and do nothing until the one‑year time point.”

I am actually very happy that CHAMPION‑AF was designed that way. It was a massive effort from lots of investigators, including you—the top enroller.
Dr Devi Nair:
Yes, we definitely tried.

Dr Vivek Reddy:
Yeah.

Dr Devi Nair:
Now, let us talk a little bit about the results and about the primary efficacy endpoint. The device arm did show that it was non‑inferior to oral anticoagulation, but there were some controversies around the absolute numbers. So can you just talk me through what that non‑inferiority design and the results that we saw mean to you?

Dr Vivek Reddy:
Yeah. So when you are thinking about left atrial appendage closure versus oral anticoagulants, you have to ask: what were we trying to do? None of us really believes that it is necessarily going to be superior for stroke prevention; NOACs are very good at preventing stroke. But we wanted to have something that works at least close to as well, but at the same time reduces what patients really care about—which is bleeding, in addition to stroke. So if you have a strategy where the stroke rate is not that much higher, but the bleeding is substantially reduced, then that is a reasonable alternative. It does not mean that you have to do it, but it is something to consider with patients.

So with that background, there were two important primary endpoints, both assessed at three years, in about 3,000 patients—roughly 1,500 in each arm. The first, which we call the primary efficacy endpoint, is a composite of stroke, systemic embolism, and cardiovascular death. The primary safety endpoint is non‑procedural major bleeding and clinically relevant non‑major bleeding.

For efficacy, this was designed as a non‑inferiority endpoint because we wanted the strategy to be at least as good or close to as good as NOACs. In a non‑inferiority trial, you have to set your non‑inferiority margin based on what is a clinically acceptable increase in events that physicians and patients would accept in exchange for other benefits. CHAMPION‑AF used a non‑inferiority margin of an absolute 4.8%, assuming a 12% three‑year event rate in the control arm.

At three years, the primary efficacy endpoint occurred in 5.7% of patients in the device arm versus 4.8% in the NOAC arm; the hazard ratio was 1.20 and the absolute difference was 0.9%, well within the 4.8% margin, so the trial met the non‑inferiority criterion. Cardiovascular death was virtually identical between the groups, with overlapping curves and a hazard ratio around 1.0. For stroke and systemic embolism specifically, the hazard ratio was about 1.4, so roughly a 40% relative increase in events, but in absolute terms the difference over three years was about 1%, which translates to an annualized difference of about 0.33% per year. That small absolute increase in ischaemic stroke and systemic embolism is one of the key points for debate.

On the bleeding side, the primary safety endpoint—non‑procedural major and clinically relevant non‑major bleeding—occurred in about 19% of patients in the NOAC group versus about 11% in the WATCHMAN FLX group over three years, corresponding to a hazard ratio of roughly 0.55, i.e. about a 45–50% relative reduction. There was a numerically lower rate of non‑procedural major bleeding alone with the device, though that component was not statistically significant on its own; the primary safety result was driven predominantly by a marked reduction in clinically relevant non‑major bleeding. As follow‑up is extended to five years, one would expect the bleeding curves to separate further.

One controversy is that the primary safety endpoint excluded procedure‑related bleeding; those events were captured in a separate procedural safety analysis. The composite major procedural complication rate was about 1%, with pericardial tamponade around 0.6% and procedure‑related stroke around 0.1%, which is consistent with contemporary registry data. Clinically, the rationale was that procedure‑related bleeding is managed very differently from spontaneous bleeding “in the wild,” which is why it was excluded by design, similar to OPTION and CATALYST. But in post‑hoc analyses, when you include both procedural and non‑procedural major and clinically relevant non‑major bleeding, you still see a significant reduction in bleeding with left atrial appendage closure, with a hazard ratio in roughly the 0.5–0.6 range.

The other controversy is the inclusion of clinically relevant non‑major bleeding in the primary safety endpoint. Clinically relevant non‑major bleeding is defined as bleeding that does not meet major criteria but requires medical evaluation, treatment, or temporary interruption of therapy; it is distinct from trivial or minor bruising. Importantly, many modern anticoagulation trials, including factor XI inhibitor trials, use the composite of major plus clinically relevant non‑major bleeding, because NOACs already perform well for life‑threatening bleeding and patient‑important non‑major bleeds drive quality of life, healthcare utilization, and adherence.

If you look at major bleeding alone and include procedural events, there is an early uptick of major bleeding in the device arm around the time of implantation, and then the curves cross over; by three years, there are numerically fewer major bleeds with left atrial appendage closure than with NOACs, although this did not reach statistical significance at that time point, and further divergence over longer follow‑up is expected. So the central question becomes whether the reduction in clinically relevant non‑major bleeding and the emerging reduction in major bleeding are “clinically relevant” enough for an individual patient to accept a very small absolute increase in ischaemic stroke or systemic embolism risk.

Dr Devi Nair:
CHAMPION‑AF was also presented in the context of CLOSURE‑AF, OCEAN, and OPTION. In the wake of all these trials, how does this clinical evidence come together for you? How is it going to change your practice next week?

Dr Vivek Reddy:
One important point for the audience is that there will be a late‑breaking presentation at HRS with a pre‑specified ablation subgroup analysis, because about half of the CHAMPION‑AF patients had undergone catheter ablation at some point prior to enrollment. That will be very informative for practice.

For patients who have not undergone ablation but are interested in getting off anticoagulation and are still eligible for NOACs—the prototypical CHAMPION‑AF population—this becomes a shared decision‑making scenario. As more data come out, including detailed subgroup and net clinical benefit analyses, we will better understand whether certain clinical profiles derive more benefit from left atrial appendage closure versus continued NOAC therapy.

Regarding ablation patients, the OCEAN and ALONE studies looked at discontinuing oral anticoagulation after successful atrial fibrillation ablation, with follow‑up of two to three years. They showed that in carefully selected patients with documented arrhythmia freedom, stopping anticoagulation resulted in very low event rates; stroke rates were numerically higher off oral anticoagulation but not significantly different, and overall absolute rates were extremely low. Those cohorts were predominantly CHA₂DS₂‑VASc scores of 1–2, with fewer than 30% at scores ≥3.

By contrast, OPTION enrolled a higher‑risk population—mostly CHA₂DS₂‑VASc ≥3—comparing left atrial appendage closure plus short‑term anticoagulation versus continued oral anticoagulation after atrial fibrillation ablation. So in lower‑risk (CHA₂DS₂‑VASc 1–2) ablation patients, it is reassuring that oral anticoagulation cessation can be considered after durable ablation, although long‑term risk beyond two to three years remains an open question. In higher‑risk patients (CHA₂DS₂‑VASc ≥3), left atrial appendage closure becomes an attractive option versus lifelong NOAC; the challenge is the intermediate group (around 2–3), where age and trajectory matter—a CHA₂DS₂‑VASc 2 patient who is 74 will soon be 3.

CLOSURE‑AF, on the other hand, enrolled a very high‑risk population and, despite not being very large, accrued a substantial number of events, increasing statistical power. It showed that ischaemic stroke rates were almost identical between left atrial appendage closure and medical therapy, which is reassuring in such a high‑risk group, but there were more bleeding events in the medical therapy arm, and bleeding appeared to drive a numerically higher mortality, although the mortality difference did not reach statistical significance. One key lesson is that bleeding is harmful, and prolonged dual antiplatelet therapy is probably worse than NOACs in this context, emphasizing the need for optimized post‑procedure antithrombotic regimens.

There are still open questions about CLOSURE‑AF: multiple devices were used, operator experience varied, and we need more granularity on disability from stroke, stroke subtypes, and timing. Those details will further refine how these data are integrated with CHAMPION‑AF and other trials in clinical decision‑making.

Dr Devi Nair:
Well, thank you, Vivek. That was a very enlightening discussion, and there is so much more to come from CHAMPION‑AF and so much more data‑dive that we will go into—and of course the five‑year results will come out in about two years. Definitely looking forward to that. And thank you all for joining us at ACC 2026 in New Orleans for this discussion on the CHAMPION‑AF trial. Thank you very much.