Mar 20 2026 This Week in Cardiology

You're listening to This Week in Cardiology from org Medscape Cardiology. Any views expressed are the presented. Hi everyone, this is John Mandrola from the Heart.org Medscape Cardiology, and this is this week in cardiology for March 20th, 2026. This week, some listener feedback, the huge publication of the closure AF trial, of left atrial appendage closure versus best medical therapy. A preview of the Champion AF study and the controversial ACCAJA lipid treatment guidelines.

Today I have kind of a longer discussion on fewer topics. Left atrial appendage closure and lipid treatment are so important and and so common, I felt like we needed more detailed discussion on these topics. First though some listener feedback on the matter of the Cobra.

trial of Apixaban versus riveroxaban for VTE and PE, in which Apixaban had a fifty four percent statistically significantly lower rate of major bleeding Clinical pharmacist Lorenz Vanderlinden writes from Leuven, Belgium to remind me that two drugs initial regimens were not actually loading doses. Loading dose implies filling an apparent distribution volume to reach a target concentration faster.

Instead, the higher doses of these two doacks, the Pixaban and Riveroxaban, were actually lead-in doses to provide more intensive anticoagulation, which is sort of analogous to the therapeutic doses of parental anticoagulation such as inoxoperin or heprin that was used before warfarin. These higher lead-in doses allows the body to resorb or remove the large initial thrombus faster.

El Doac manufacturers address the dosing through distinct strategies. While most doac are comparable in terms of half life. The lead in dose riveroxiban. was a three week lead in, fifteen milligrams BID, and the thirty milligram total dose is one point five times the twenty milligram maintenance, was largely based on phase two data And that showed that BID dosing optimized early clot regression with low recurrences at three weeks. A Pixaban has a seven day lead in, it's ten milligrams BID.

And this twenty milligram total dose, which is two times the ten milligram maintenance dose, was based on phase two data, partially modeled to mirror the ratio between therapeutic and prophylactic low molecular weight heparin doses. Vander Linden writes that the distinction between loading and lead in clarifies that the initial higher dose is intended to match the intensity of parental therapy during the highest risk period, rather than simply accelerating steady state kinetics.

And as always, I'm a student of pharmacology and I appreciate this clarification. Still, though, my idea that the differing lengths of higher dose doac therapy could have made the difference in the bleeding rates. I am not sure, therefore, that one drug or the other drug is safer. Certainly though the Apixaban regiment as it is now is superior than the Revoroxaban Regiment.

Okay, the major topic today is closure AF published in the New England Journal of Medicine. Closure AF was a non-industry funded German RCT of left atrial appendage closure versus best medical therapy, mostly DOAC therapy in AF patients at high bleeding and stroke rates. Professor and PI Olf Landmesser, whom I debated once at ESC and was super nice to me, presented the data at AHA meeting in November, and I covered the preliminary results in November.

The Winning Journal of Medicine published the manuscript Wednesday along with an excellent and extremely candid editorial. Do read the editorial, it's excellent. Oh, it should be stated that left atrial appendage closure has been done since 2010 in Europe, with German cardiologists being some of the earliest adopters. So these are a group of doctors very experienced in left atrial appendage closure. W globally, more than six hundred devices have been placed.

And Clojure AF is the first properly powered trial comparing appendage closure to DoX. And I know that's crazy. You would think before we put in 600,000 devices. We would have a strong evidentiary basis, but we don't. In the United States, the wording from the payers is sufficiently vague that many patients with atrial fibrillation are offered left atrial appendage closure instead of anticoagulation for very soft indication.

You don't believe me, but I'll tell you, at Western AFib a couple weeks ago, an electrophysiologist showed her indications for appendage closure. Which included patients who could injure themselves from athletics, say cycling, uh those patients who had bruising, and those who could not pay for DOAX. These were all indications to do appendage closure. Needless to say, I push back on those indications. But her indications are not unusual. I see patients frequently from other hospitals.

and the softness of the indication for this appendage closure procedure shocks me. Even seeing promotions for appendage closure in hospital lobbies, on social media, TV news, and the gist of these advertisements. is that left atrial appendage closure is just another way to prevent stroke. So don't tell me that Clojure AF does not generalize to the standard population, because Closure AF clearly does generalize and I'll go into more of the patient characteristics in a second.

In closure AF, 912 patients underwent randomization. The primary endpoint was a composite of stroke, could be ischemic or hemorrhagic. systemic embolism, major bleeding, or cardiovascular unexplained death. It was tested for non inferiority with a hazard ratio margin of one point three. The population was elderly, the mean age seventy eight, mean Chadsvask five point two, has bled of three point zero. A few features are notable.

Twelve percent of patients had had previous intracranial hemorrhage, thirty percent had had major bleeding, and twenty-five percent had stage four CKD.

So these were elderly ill patients who are exactly who are getting these procedures, at least in the U.S. Again, external validity of closure AF is excellent. A few words on trial procedures. Watchman devices were used in fifty four percent, amulet forty two percent, the newest iteration of Watchman Flex was used in forty four percent.

I mentioned this because proponents think that Watchman Flex is clearly better. So I can show you a slide from the IDE trial which does not support that contention, at least in terms of outcomes. Flex may feel better, but its empiric results are not impressive compared with, say, the anticoagulation results from the DOAC trial.

Post-implantation anti-thrombotic management was individualized based on bleeding risk and treating physician judgment, but followed a structured protocol and it used DAP. D dual antiplaylet therapy was the primary post-implant protocol. This is kind of an important point because some newer studies suggest that DOAGs or low dose DOEX may be better, though that data is somewhat preliminary and quite new. Certainly wasn't available at the time of the trial began.

The quote physician-directed arm was pragmatic and left to doctor choice. In the end, 85% of patients received a doac and 3% received a vitamin K antagonist. Notably, get this, 7.5% received no antithromotic therapy at all. The trialists don't provide a DOAC type breakdown. They specifically state that all four of the DOAC drugs have comparable profiles.

A word on the statistical plan and power. Clojure AF was designed as an event-driven trial with a target of four hundred and sixty-seven primary endpoint events to achieve eighty percent power for non-inferiority. The original plan called for 1,586 patients based on an assumed annual control group event rate of 11.2 per 100 patient years, and that was derived from appendage closure registry.

Recruitment, however, was stopped early at 912 patients after published data from Prague 17 suggests that the original sample size may already be overpowered. The analytic plane plan changed from AHA presentation to the New England Journal of Medicine publication. The original plan was to analyze the non inferiority using Cox proportional hazards model and use the non inferior margin of one point three.

New England Journal reviewers must have felt that the plots deviated from the proportional hazard assumptions, particularly for major bleeding, and the main analysis uh was changed to a restricted mean survival time. RMST is more robust when the proportionality assumption is violated as it captures cumulative time to event experience without assuming constant hazard ratios over time. RMST is simply the average amount of time a patient lives free of the primary endpoint.

Now the authors also made some complicated adjustments to account for competing events and to avoid over precision from site level clustering at the forty two different hospitals. And don't ask me to explain these because they're too complicated for me. The main results, here it is. Over a median follow-up of three years, a primary endpoint event occurred in 155 patients. That's 16.8.

per 100 patient years in the device group versus 127 patients, that's 13.3 per 100 patient years in the medical therapy group. The difference in the restricted mean survival time was a minus zero point three six years, and the ninety-five percent conference intervals went from minus zero point seven to minus zero point zero one. And non-inferiority was not demonstrated. The p-value for that was 0.44, so very high p-value.

And this too was an intention to treat analysis, but you remember, and you know, that in a non-inferiority trial the per protocol analysis is the more conservative estimate. This is opposite to a superiority trial where the intention to treat is the more conservative. The per protocol analysis is the more conservative because in a non-inferiority trial, you want to show that it's non-inferior in the patients who got the device or the new treatment.

So in the more conservative per protocol analysis of closure AF It more strongly favored the medical arm. Here it was sixteen point six versus twelve point five. uh per 100 patient years and the restricted mean survival time was minus 0.40 and the conference intervals were Minus zero point seven six to minus zero point zero four, which is even more strongly in f in favor of um medical therapy over left atrial appendage closure.

The reason that the P value for non inferiority was so high and not significant is because the device event rates were twenty six percent higher than the medical arm. Sixteen point eight divided by thirteen point three is equal to one point two six. So the point estimate of one point two six is nearly at the non inferiority margin of one point three.

I did some calculations with exponents and standard error and an online calculator, and you can get an upper bound here, 95% upper bound of 1.59. I'm not saying it's exactly that, but it's well above. the non-inferiority margin of one point three. Now, multiple trial experts online also noted that since the 95% conference interval for the restricted mean survival time was less than zero, then the device arm could be declared inferior to medical therapy.

This verbiage was not used in a New England Journal of Medicine paper, but technically the device was both not non inferior and inferior to best medical therapy. It's an important distinction and it gets a little complicated because Uh therapy can be not non-inferior. But that doesn't necessarily make it inferior. In this case, however, you can declare that left atrial appendage closure was both not non-inferior and inferior to best medical therapy.

Let me tell you about the components of the primary endpoint in the 440 patient arm. So each arm had 440 patients. So ischemic stroke was 18 appendage arm 15 best medical therapy. All stroke, including hemorrhagic stroke, were 27 and 27, so identical. Stroke systemic embolism, basically the same, twenty nine and twenty eight patients. Bleeding, surprisingly, was higher. 7.4% in the appendage closure arm, 6.2% best medical therapy, cardiovascular death.

Higher in the appendage closure arm, 99 patients versus 81. That's 9.5 versus 7.7%. Device implant data in the 446 patients in the uh device arm. Procedural success was high at 96%. However, 20 patients or 4.4% of patients had peridivice leaks, which are bad because they increased stroke risk. Five patients had pericardial tamponade. There were eighteen patients who had a major bleeding during the procedure, one device embolization requiring surgery, one procedural TIA and the

One patient had a peripheral embolism and there were two periprocedural deaths within seven days. Two deaths. Remember, this is a preventive procedure. So my comments. Look, every trial has its imperfections and I've never seen this much critical appraisal on social media. Perhaps the critical appraisal out on social media is so robust because closure results upend the standard practice.

Yet I would argue that these signals are not surprising to me, given the very weak evidence from the seminal trials. Again, I'll say it, but I'll be brief. Protect failed to pass FDA muster. Even though Protect sits in journals as a positive trial FD8 tossed it out. prevail, missed its non-inferiority endpoint in the main co-primary endpoint of stroke, systemic embolism, and CV unexplained death. That is, Watchman was not non-inferior to warfare.

Proc seventeen made non inferiority but was woefully underpowered and was largely uninformative. And the option trial was one of the most bold trials I've ever seen in my life. Options main safety endpoint excluded procedural bleeding and included nuisance non-major bleeding in an unblinded trial. My point, therefore, in approaching the closure data is that the priors here should be very pessimistic, and the null data from Clojure AF only should strengthen your prior. It shouldn't be a surprise.

Okay, first point, the net adverse clinical events, primary endpoint in closure, includes efficacy, stroke and systemic embolism, and safety. And these are endpoints that should cancel each other out and make non-inferiority easier to reach. Yet even in this endpoint, which was biased towards non-inferiority, left atrial appendage closure was still not non-inferior. Another comment on the primary endpoint.

It included CV death, which is not affected by either treatment, and therefore that should just add events and again bias towards non inferiority. But it did not in this case. It favored best medical therapy, and much has been made about the ninety nine versus eighty one CV unexpected death rates favoring the BMT arm, best medical therapy.

I think this is most likely noise, a chance event. Remember that higher CV death occurred in the warfront arm of Protect, and that was largely driving that Charles that trials non-inferiority signal. It was proven to be noise and protect, and that favored left atrial appendage closure. So it's likely noise in the closure trial favoring best medical therapy. CV death does not belong in the primary endpoint of any left atrial appendage closure trials.

It was not in the DOAC versus Warfare trials because stroke prevention is not likely to affect mortality or C V mortality. Of course, trialists add CV death to increase power and decrease sample size. And you can't blame the closure authors because they merely followed what Protect and Prevail and Proc 17 did. And a nose a noisy signal can go for you or against you.

Now once again, left atrial appendage closure is found not to protect against stroke versus best medical therapy. The rates were the same. And the rates were at three percent. That's far lower than what would be expected by Chad's vast score of five, and that's further proof. that TASVASC is no longer predictive in the modern era. There's simply been a temporal decline in stroke rates, which is important.

Okay, fourth point. Bleeding events were higher, and they were the largest driver of left atrial appendage closure missing non-inferiority. Now, this may seem mysterious, but it is not if you think about it. First of all, you cannot exclude periprocedural bleeding. These were substantial enclosure. Eighteen of the seventy bleeds occurred during the procedure, many requiring transfusion.

Now I've seen a number of proponents of appendage closure on social media say that closure centers were not skilled because we don't have these numbers of complications at our center. I find this criticism foolish. Most people don't track their complications, and if you and if you don't track it, it's easy to forget bad cases.

Second, closure was done in more than forty centers in Germany, not all of which were academic top centers. This approximates what is done in the United States where many, if not most of the Left atrial appendage closure procedures are done in low volume centers. I've been in this business thirty years and I don't buy the we don't have complications rebuttal.

Plus, I've been to many German centers firsthand. These are excellent operators. And furthermore, left atrial appendage closure began far earlier in Europe and Germany were early adopters. So I think these folks know how to put in Left atrial appendage closure devices. Further, the signal of procedural harm lies at the crux of the problem for patients who hope to have a net benefit from this procedure. Namely, this is a preventive procedure, right? Patients start in a negative probability hole.

Because any probability of procedural risk has to be outweighed by future lower stroke or bleeding rates. And this was absolutely not seen in closure. Bleeding rates may be slightly lower in the last part of the trial compared against best medical therapy, but the differential was not enough to overcome the initial procedural bleeding. This is exactly what was seen in the Holmes meta-analysis of protect and prevail.

When they meta-analyzed, protect, and prevail, and they included procedural bleeding, there is not a signal of bleeding benefit. Now, endpoints called non procedural bleeding are a scientific embarrassment because no patient can exclude the risk of the procedure. In the case of closure, two patients died from the procedure and one had to have surgery to remove an embolized device. This underscores the negative probability gamble that patients take when they have this procedure.

Point number five, I've also seen criticism saying that left atrial appendage closure was not intended to replace DOAC because if an AF patient who is at risk for stroke truly cannot safely tolerate long term enticoagulation, It remains reasonable as a two way recommendation to place one of these devices.

My response to this is A, these patients have been excluded from the trials. That is, patients who absolutely cannot take long term anticoagulation have not been studied. And despite 600,000 devices placed, we don't know. If left atrial appendage closure is better than no anticoagulation in truly ineligible patients, I don't think left atrial appendage closure beats no therapy in frail patients who bleed easily, but I don't know either, because it's not been studied.

My second rebuttal to this is that oodles, perhaps the majority of patients who have appendage closure in the U.S., are nudged into having it as a replacement for DOAC. Like I said earlier about the slide at Western AFib, where a doctor counted bruising, athletic endeavors, potential falls, and the cost of doax as reasons to do left atrial appendix closure. You even see this in the advertisements where language suggested appendage closure as a l as a long-term alternative to DOEX.

Point number six, finally, next weekend we will hear the results of Champion AF trial, which will randomize three thousand patients to a Watchman Flex or DOAC at one hundred and forty two sites. I will tell you now that Champion will deliver positive results. I guarantee it, based on a choice of endpoints. It's very similar to the option trial. The industry sponsored trial is designed to sell Watchmen units. Why? Three reasons.

One, the efficacy endpoint of champion AF is stroke, systemic embolism, or CV or unexplained death. Again. We know that CV and unexplained death are not affected by either treatment, and so it adds noise, which makes non inferiority easier to read. And you can expect to find low stroke rates because stroke rates are just lower now and it'll be low stroke rates in both arms and just stroke and systemic embolism will be underpowered to declare non inferiority.

Now the second primary endpoint of champion AF is really, really problematic. This is a safety endpoint, and it'll include non-procedural bleeding and non-major uh clinically important bleeding, and it'll be tested for superiority. And it'll be the same exact thing we see with option. It will be positive. Because patients randomized to the oral anticoagulation arm in an unblinded trial will complain of more non major bleeding, such as bruising and oral bleeding, just as it was an option.

What's more, it is scientific malfeasance to exclude procedural bleeding because no patient can exclude it. Exclude it. I can't believe that this is done. Now the final problem with champion, and you have to know this going into it. is that the secondary endpoint of safety is all bleeding, all major bleeding, and such an endpoint ought to be tested for superiority. Why? Because a new therapy should offer an advantage over the older established therapy.

But not in champion. It will be tested for non inferiority, which is wrong, just as it was an option, where left atrial appendage closure was non inferior to oral inoculation in the all bleeding safety endpoint. In champion, watchmen will be non inferior to oral anticoagulation, but we need it to be superior because a new treatment that's not worse in efficacy ought to offer a superior safety advantage. This is how it was t tested in the

Doac vs. Warfin trials and it's how it should be tested here. So when that secondary endpoint of all major bleeding comes out non-inferior, you should ask whether it's superior. And if it's not, then it doesn't offer an advantage. So when champion comes out, I recommend you ignore the hype and positive top line results and look at stroke and systemic embolism rates for efficacy.

and pretend the secondary end point of all bleeding is actually the primary safety endpoint. Then assess whether it is superior. I suspect it'll be similar to option in that stroke and systemic embolism will be lower than expected and underpowered and bleeding will not be superior. And if that is the case and you take closure and champion together, the use of appendage closure should be greatly reduced.

The next trial, of course, that I'd want to do is left atrial appendage closure compared to no anticoagulation in older, frail patients who cannot take long-term anticoagulation. And now my final final comment comes from Professor David Capadano, the editor in chief of Euro Intervention, who wrote What seems to be missed in this accumulation of arguments is straightforward.

The burden of proof lies with left atrial appendage closure, not with the control arm. The issue is the strength of the evidence supporting appendage closure, not medical therapy, which remains the reference standard.

Okay. Final topic today is the lipid guidelines. Oh my. This is another buzz that occurred last week, and it was the release of the ACC AHA Lipid Treatment Guidelines. Nearly a thousand people have posted about it on X. 56 news sites have covered it. The document is 134 pages. It includes 45 brand new recommendations. Fourteen complicated treatment flow charts and eight revised recommendations from the last guideline in 2018.

I'm gonna have more to say on this in the coming weeks, and I can't possibly go over every aspect, there's just too many, so I have a few summary comments. First. Of the 53 Class I recommendations, only 10 or 22% were backed by randomized controlled trials. Most of the class one recommendations are level of evidence B non-randomized. I think this is awful because class one recommendations.

signifies something that you should do. That benefit is way greater than risk. Class one recommendations can become standard of care and not following standard of care can bring medical legal trouble. Class one recommendations therefore should be carefully chosen. That's not what happened with these guidelines. Here are some examples.

Incidental coronary artery calcium seen on a CT scan done for something else gets a class one recommendation to discuss and consider initiating or intensifying lipid lowering therapy. The two citations from that recommendation go one to a single center observational study and another to an observational study of carotid plaque. Not a single trial supports this contention. Not one, yet it is class one.

Now doctor James Stein from University of Wisconsin, a preventive cardiologist, academic preventive cardiologist, wrote about a few other uh problematic i issues on his substack. Coronary articals scoring gets a class one recommendation. In adults at intermediate ten year risk of five to ten percent using the prevent equation.

and selected borderline risk adults, a CAC score is given a class one recommendation When the treatment decision remains uncertain and if CAC is greater than zero, initiation of lipid lowering therapy clock carries a class one B non-randomized recommendation, particularly if the score is one hundred units or above. But doctor Stein is correct in that no RCT has proven that CAC driven strategy reduces downstream cardiac events.

He doesn't mention it, but I will, and Madan Kavas won and two trials. which included coronary calcium measures in a comprehensive screening program in Denmark, one of the best places to do screening studies, did not significantly reduce cardiac outcomes. Now if you can't show benefit in Denmark, Where excess angiography and stents really don't occur, it's surely not going to benefit US patients where CAC is often a hotline to angiography and PCI.

doctor David Brown pointed out on X that we have multiple RCTs for screening mammography, PSA, chest CTs, Flex SIGs, but none for coronary artery calcium. But yet it's still class one. Stein Stein also notes that the guideline issues a class one recommendation for panel based genetic testing for pathogenic or likely pathogenic variants in any adult with possible, probable or definite familial hyperclosaremia.

Again, there is not an RCT supporting such a practice. He writes, and I agree that if LDL is greater than 190, you can just treat the patient. But that's not the only genetic overreach. The authors also put genetic testing in the form of a PRS or polygenic risk score in a box as a risk enhancing feature in decision making alongside lipoprotein little A. And high sensitivity CRP. Again, there are no trials testing these as risk enhancers. And making them class one puts added pressure to use them.

Okay, another poor evidence recommendation, in my opinion, is to use the prevent score instead of the PCE pulled cohort equation. But get this, prevent predicts far lower event rates than the PCE. It may be more accurate, but If you can't keep the same treatment thresholds, millions of U.S. adults fall out of the statin eligibility threshold. So what did the authors do? They simply adjusted down the thresholds and now give the nod for statins at a prevent risk greater than 3%.

Next problem. There are eight new recommendations regarding specific levels of prevent risk and LDL cholesterol numbers. Not one, but eight. Stein also notes the conflict of interest in the writers of the guidelines, specifically the He notes that of thirty-one writing committee members, about half carry relevant relationships with pharmaceutical manufacturers, imaging companies, digital health companies, or genetic testing companies, often multiple of these simultaneously.

Several members hold relevant relationships spanning essentially the entire landscape of novel lipid lowering agents that this guideline governs. Over two thirds of the twenty nine member peer review committee, which is meant to provide independent assessments of the writing committee conclusions, disclode relevant conflicts of interest. He writes, and I'm want to quote him specifically, quote.

From the patient's standpoint, the issue is trust, what a guideline gives class one recommendations to cornear article calcium screening and genetic testing for people with high LDL cholesterol values. and you combine that with all of the conflicts of interest, what you ultimately do is decrease trust in the guidelines, trust in effective therapies, and even trust that the lipid basis of atherosclerosis is real.

The conflict of interest issue is always there, but I think it's even more relevant given the overcalling of so many class recommendations which border on, I think, expert opinion. Finally, I want to reiterate that I am a preventive cardiologist. I tell my patients who have had atherosclerotic vascular disease or at high risk for disease that heart disease remains one of the most preventable fatal conditions.

I say to my patients that they can improve their odds greatly by incorporating a healthy lifestyle and taking preventive meds such as blood pressure and cholesterol lowering tablets. That said, though, I also ascribe to Dr. Victor Montori's minimally disruptive care, and that's why I have such a problem with the utter complexity of this guideline.

Patients are human beings and should be living life outside of being a patient. When you make something this complex, it's less likely to be followed. There's no way that doctors can follow 14 flowcharts for treatment. Simple, simple, especially in the absence of evidence of complexity, is the ideal way. In the 14 complex treatment flow charts are a morass of arbitrary cutoffs.

For instance, there's a cutoff for LDL cholesterol greater than 190. But is an LDL cholesterol of 189 any different from 191? Likewise. There are recommendations for CAC 0 to 99 and 100 to 299. But is a CAC of 99 any different than one 101? No. This minutiae is crazy and it's distracting and it's non-evidence-based. And don't get me started on CAC of Zero, as this is a serious false negative problem wherein it could lead to less treatment in susceptible people.

I say give us four recommendations, one for primary prevention, one for secondary prevention, one for special circumstances such as a super high LDL or triglyceride, and maybe one for people with diabetes. I also say jettison this 30-year time frame. There is no data. Uh to and to think that you can modify health with pills over that time frame is the height of hubris. I'll have more to say about the guideline coming soon.

Okay, next weekend is the ACC meeting in New Orleans, and this week in cardiology, all tons of studies to discuss. I'll have a preview uh ACC column coming next week. And if I can get through TSA screening, I plan on being in New Orleans. In fact, I think Metscape is planning on Ask Me Anything at the booth. I'll let you know more about that on next Friday this week in cardiology.

Tomorrow is the first day of spring. It's been a bad winter here in Louisville. I say happy spring to all, and I hope you ride fast this year. Until next week, this is John Mandrola from the Heart.org Medscape Cardiology. You're listening to This Week in Cardiology. org Medscape Cardiology Any views expressed are the same.