Apr 03 2026 This Week in Cardiology

You're listening to This Week in Cardiology from org Medscape. Any views expressed are the presenter's own. When every decision matters, you need trusted medical intelligence. Medscape AI synthesizes peer-reviewed evidence and Medscape's expert validated content in seconds, delivering clinically verified answers you can trust. Join over 13 million Medscape members on the world's leading platform for clinicians.

Start using Medscape AI on the web or in the Medscape app. Medscape AI, trusted medical intelligence.

This is John Mandrola from the Heart.org Medscape Cardiology, and this is this week in cardiology for april third, twenty twenty six. This week. A big ACC recap number one, and I'll talk about five of the biggest trials, High Petho, Pro TAVI, Orbita CTO, Chips BCS3, and Champion AF. Let's start with the interventional treatment of PE, the high PTO trial, which was actually the first late breaking clinical trial. And at ACC, I sat up front during the presentation. High PIZO was the first trial.

I was there to hear Champion, which was to follow, and I wasn't listening that actively, but what I did hear was a pretty jubilant presenter, an even more enthusiastic trial discussants. Quote, A new era in P E care, I heard in my ears. I made note and I said I'm gonna look under the hood of this trial. Well What follows is my look under the hood. Even if you aren't a P interventionalist, you're probably not, I'm not, the there are really great trial lessons to be had here.

New England Journal of Medicine published the multinational open label trial of ultrasound facilitated catheter directed fibrinolysis. for m intermediate risk PE patients. The inclusion required an R V L V ratio greater than one, an elevated troponin, and two indicators of cardiorespiratory distress, such as systolic blood pressure, Less than one hundred ten or heart rate greater than one hundred, or respiratory rate greater than twenty.

One group got the ultrasound facilitated altoplase plus anticoagulation and the other group got anticoagulation alone. The primary outcome is a composite of PE-related death, cardiorespiratory decompensation, or symptomatic PE recurrence. Now, PE-related death and PE recurrence are straightforward, pretty hard endpoints. Cardiorespiratory decompensation was more complex and it included four items: cardiac arrest, signs of shock.

Placement of ECMO, intubation, or initiation of noninvasive mechanical ventilation. Or, and this one is the Humdinger, a national early warning score, a news score greater than 9 by two consecutive measures 15 minutes apart. Now there, I had to explain what cardiorespiratory decompensation entails.

Since most of the endpoints in this cardiac respiratory decompensation bucket were news scores greater than nine. Now I have to tell you about news, which of course I did not know about before looking at the trial. News is this many-point score with each item ranging from zero to three for bedside physiologic signs, including respiratory rate, O2 SAT, supplemental oxygen use, temperature, blood pressure, heart rate. Now note here that the new score was designed as a surveillance

an escalation trigger, meaning its value was in prompting doctors to act. It was never designed to be a blinded, adjudicated outcome measure in a trial. Also keep in mind when I tell you the results that high patho is an open label trial, so clinicians knew their patient's trial assignment. The new score also requires only two measures 15 minutes apart.

So in this trial, 4,300 patients were screened and enrolled only 544. That's an important thing to consider. Most patients were excluded for not meeting the inclusion criteria. The maker of the Ecosonic Endovascular System, Boston Scientifics, sponsored the trial and had representation on the Executive Committee. The co sponsors of the trial include the University of Mainz, Germany, and the PERT PERT Consortium, which is a nonprofit receiving money from industry.

The statistical analyses were performed by a contract research organization with a biostatistics plan by a Boston scientific employee. So, in sum, two of the three funders were from industry. The key result.

The two arms had each about two hundred seventy patients, and a primary outcome event occurred in eleven patients, four point zero percent in the intervention group. and twenty eight patients, ten point three percent in the control group, that's a relative risk reduction of zero point three nine,

Ninety five percent conference intervals clearly significant as they go from zero point two to zero point seven seven and the P value was very, very low, a a statistically significant result. And it's a win, right? That's what I heard from the stage when I was there. But but now we need to look at the drivers of the primary endpoint.

PE related death higher in the active arm, three versus one. Recurrent PE each one. All the difference in this endpoint was from cardiorespiratory decompensation. ten versus twenty eight. So let's look at the supplement and see what these ten versus twenty eight differences were in the cardiorespiratory decompensation bucket. Cardiac arrest six versus four. Signs of shock, five verses eight. ECMO one verses three. Intubation five verses six.

It's all in the news score. News score greater than nine. Two patients in the active arm versus 19. in the uh control arm. So the primary endpoint, my friends, was essentially driven by bedside scores of items to be judged by a human clinician, including things like the level of consciousness and respiratory rate, and these were only taken twice fifteen minutes apart. Hard outcomes such as PE related death was higher, and there was no difference in P recurrence.

What's more, 14 of the 19 patients in the control arm who had new scores greater than 9 had only that as their endpoints. On safety, major bleeding at 72 hours, 3.7% in the active arm, 1.5% in the control arm. That's a relative risk ratio or increase of 2.5x. The conference intervals uh spanned uh zero point eight to seven point nine, the P value was zero point one seven.

So the p value is non significant for harm, but the trial was almost certainly underpowered to detect bleeding differences. But still there were two point five times more bleeds in the intervention arm. Of course there were, it used thrombolytic therapy, so it should be no surprise. I don't think it should be cast off as noise either.

There was also slightly more all cause death in the intention to treat population at thirty days. It was six patients died in the active arm versus three. At thirty days, the six minute walk test was no different, hardly an improvement. All right, my comments as an outside observer The quote positive result is driven by soft physiologic score, a new score greater than nine, in an open label trial.

All the hard endpoints were clearly not different and they trended really against the intervention. Bleeding rates were higher, two point five times higher. In addition, this is a highly selected population where 87% screened were not enrolled. There was no overall mortality benefit benefit, and if anything, numerically went the wrong way for the Ecos catheter. Functional outcomes were also nearly identical. Twelve month follow-up is planned, but I don't expect any major differences.

I would also add that this trial was sponsored by industry and performed by PE interventionalists who likely believe in PE intervention. Now, I acknowledge that PE trials are very hard to do because the presentation is often variable And by nature, uh the selection has to be rigorous or you stand chance to have way too much noise. That is patients who are not sick enough to benefit or too sick to benefit. But I am struck by the lack of signal in important hard, unbiased outcomes.

The new score may be good for prediction, but does not seem like an unbiased trial endpoint, especially when the trial is open label. I would not call this a win for the ultrasound directed fibrounalytic catheter. Not at all. And my friends who are learning about trial appraisal, this is a great example of how important it is to look deeper than top line results. And you got to look into the supplement, look at the drivers of the primary endpoints.

I think the hype surrounding this trial actually approaches that of Champion.

All right, next topic is the ProTavi trial. The matter of stable coronary disease interests me greatly. That revascularization of these severe, gnarly looking lesions does not improve outcomes is one of the great mysteries of modern medicine, I think. I mean you see an LAD of eighty percent On an angiogram you're like, man, that looks bad. But if that patient is stable, there's never been a trial showing that putting in a stent reduces MI or death.

Trial after trial of PCI and cabbages are in cabbage are null in stable disease. Even during an MI, the preponderance of the evidence favors culprit only PCI and leaving the other stable lesions to be treated with medical therapy. Yet, yet, patients with aortic stenosis may be different. Patients with AS have LVH and likely reduced coronary flow reserves.

They have increased oxygen demand and they have relatively reduced subendocardial perfusion. And all of that is without any coronary stenosis. The myocardium uh distal tuos stenosis could even be worse. So the matter of revascularization of coronary flow limiting lesions before TAVER is a scientifically curious and clinically relevant question. And the only way to study this would be in an RCT because observational trials would be marred with tons of selection bias.

So at ACC, we heard the results of the pro-TAVI non-inferiority trial of deferring PCI or doing PCI before TAVI in patients who had severe aortic stenosis and coronary disease. The trial was done in twelve centers in the Netherlands and it was publicly funded. Here are the trial basics. Four hundred sixty six patients with significant CAD

who and aortic stenosis were randomized to either deferral of PCI or PCI. The primary endpoint was a composite of death, MI, stroke, and major bleeding at one year. This was tested with non inferiority. Now whenever someone says non inferiority, the two next questions that you should have are A, is it appropriate to test this with non inferiority? For instance, does the new treatment offer some less invasive or less costly approach? And B

How is the non inferiority margin decided upon? And in this case I think Non inferiority is a great way to test deferral of PCI as it's clearly less invasive and less costly. The expected rate of events on this trial was fifteen percent in both groups, and the non inferiority margin was set at eleven percent and risk difference. Absolute turn. No explanation was given in the paper as to where this came about. I looked multiple times, nothing. Why 11%?

This corresponds to a relative non-inferiority margin of 1.73, meaning deferral of PCI could be 73% worse on the primary endpoint and still be called non-inferior. Now, given non-inferiority testing, there is one major flaw of the composite endpoint, and I hope many of you see the flaw in that endpoint choice. If not, stay tuned. Now, before I tell you the results, let me tell you two background issues.

There's two things to say. One is the first rule of all trials, and that is think first if the trial is trying to answer an important clinical question or if it's simply to sell a product. This trial pro TAVI is publicly funded and the question of PCI before Tavi is a true modern day question. I added the modifier modern day because In days past, coronary stenoses were just bypassed at the time of surgical aortic valve replacement at little incremental risk to the patient.

The availability of TAVI makes PCI or deferral of PCI a critical question. The second thing to say is that the Notion three trial published in New England Journal of Medicine twenty twenty four. found that PCI of a severe coronary stenosis versus conservative management before Tavi

in patients who had AS and C A D found a twenty nine percent reduction in the MACE endpoint of death, MI, or urgent revascularization. Though the hazard ratio in Notion three of zero point seven one had ninety five percent conference intervals that went from zero point five one to zero point nine nine in the p value of just zero point zero four. Now if this was a true reduction, it came at the cost of a statistically significant fifty one percent increase in the first time.

in the rate of major bleeding. I tell you this because the Notion 3 trial results is an important prior for the Protavi trial. So here are the Protavi results. A primary endpoint of death and my stroke or major bleeding occurred in twenty four percent of the deferral group. Versus twenty six percent of the PCI group, the absolute risk difference using the Kaplanmeyer estimates was one point seven percent, and the all important ninety five percent conference interval.

went from minus nine point five to an upper bound of six point two percent, which is less than their declared non-inferiority margin of eleven percent. And so the deferral arm was declared non-inferior to the PCI arm. Other important findings was that major bleeding was substantially better, six percent versus fifteen percent. The hazard ratio there was zero point three nine. So the conclusion in the Lancet reads as follows Quote The Fural of PCI was not inferior to PCI before Tavi.

For the one year composite of mortality, MI stroke, and major bleeding, quote These findings suggest that an initial conservative strategy can be appropriate in selected patients. Although patient-tailored treatment decisions remain essential. Now, this is a different conclusion, the Notion 3, which found that PCI reduced mace efficacy versus conservative treatment.

So my comments, I hate to be negative again, but this is a faulty analysis of the data. I feel like a broken record, but in the non inferiority comparison. You should never put endpoints in a composite that are expected to go in opposite directions because they cancel each other out and then makes non-inferior easier to reach. One goes up, one goes down, and the end result is no change and no change is a positive result.

Here, the efficacy endpoints of PCI, death MI and stroke, and you have safety endpoints of bleeding. Let's look at how they trend and what would have happened to non-inferiority. The ischemic composite endpoint of death, MI, and stroke was numerically higher in the deferral group, 21% versus 16%. That's a Kaplanmeyer rate difference of four point three percent with ninety-five percent conference levels of wait for it.

Minus two point seven to eleven point four percent. The upper bound of eleven point four percent is greater than the non-inferiority margin of eleven percent. and thus would have missed non inferiority. Now, the number needed to harm here for this rate increase twenty two. What is more, the rate ratio, the relative risk of 1.28 or 28% more mace.

in the deferral group had a ninety five percent conversal upper bound of one point nine three. So the worst case scenario here, using rate ratio or relative risk, would be that deferral of PCI had a nearly two X greater chance of having MACE. This is similar to Notion, which also had a twenty nine percent percent reduction in MACE in the PCI arm versus the conservative arm.

Now let's just separate out safety in the pro TAVI trial. Bleeding was fifteen percent versus six percent in the PCI versus deferral group. That's two times X more bleeding in the PCI arm. So in sum, when you consider deferral as the new therapy or less invasive therapy. you have a substantially higher, up to two times more chance of having a mace event compared to revascarization. That surely doesn't look like non inferiority.

But on the other side, the it's at the cost of two two point five X more major bleeding, and this is surely due to dual antiplatelet therapy in older patients. Again, my look at the tabby data is as an outside observer, but in the case of a patient who has aortic stenosis and severe C A D This data suggests that doing PCI may decrease mace but increases bleeding. I would not go with the pro-tabi conclusion that deferral is non-inferior. I don't think it is.

But here's the thing perhaps these patients who have CAD. and aortic stenosis would be better served with surgery. BIPASS does not substantially increase operative risk, and you only require single antiplatelet after surgery. The recommendation may even be stronger in younger patients, as the recent Evolute seven year data found a three point six percent higher death rate in the TAVI arm versus the surgical aortic valve replacement arm.

I'd love to have a TAVI plus PCI versus Saver plus Cabbage trial of patients with both. All we have now are small underpowered subgroup analyses from some of the trials.

The next topic is the Orbita CTO trial. Now two RCTs of CTO PCI versus med. Decision CTO and Euro CTO have found no reduction in hard outcomes from doing PCI of chronic total occlusion. A CTO PCI is often offered to patients mostly for angina and quality of life improvement. The evidence for benefit in this subjective endpoint, however, have been inconclusive. Inconclusive because no trial has done the proper test of angina relief, and that includes a placebo procedure.

Orbita CTO led by doctor Sarosh Khan and doctor John Davies in Essex, UK, randomized fifty patients with single vessel CTO to PCI versus a placebo procedure. As it has been done in all the Orbita studies, extreme care was taken to maintain patient and doctor blinding. And I think their efforts were successful as the blinding index that they did revealed quite good blinding.

The inclusion criteria of this trial were strict, these were two point five millimeter or greater vessels that had a JCTO score less than three. There also had to be inducible ischemia and viability and suitability of the legion in question had to pass muster in a multidisciplinary team meeting. Orbita also included only experienced operators.

The primary outcome was an angina symptom score taken on a specially designed cell phone app called the Orbita app. Daily anginal symptom score included three components that is the episodes of angina and the anti anginal medicines.

and uh bad events like cardiac uh arrest or ACS. Follow up was twenty four weeks. The trial procedures were more similar to Orbita two, where antianginals were titrated up and then they were stopped at the time of the invasive procedure and the reasoning here is to isolate the true placebo resistant effect of the CTO PCI.

The primary result was that the CTO PCI improved angina symptom score. The odds ratio was very high at four point three eight, the probability of benefit ninety nine point six percent. CTOPCI also led to additional thirty point six days free from angina. And as for components of the primary endpoint, angina frequency was the driver. It was substantially lower whereas there was no significant difference in anti anginal medicine use.

CTOPCI improved other measures of angine as well, such as the CCS and SAQ domains of angine of frequency, physical limitation, and disease specific quality of life. In hospital complications included one patient in the active arm who had pericarditis and one patient in the placebo arm who had pseudoaneurism treated with thrombin injection. My comments conclude.

Come on, this is a superb trial. It's well conducted, it answered an important question, and the Orbita teams from the UK have consistently sh shown the cardiology community that placebo controlled trials can be done and answer important questions. As for the results, I would say the odds ratio of four for angina relief oversell the benefits a bit.

I did a 15 minute video that's going to come up on Metscape with the authors and they felt that the degree of anjun relief could be considered moderate. Now that's not nothing, but moderate is moderate. Now others have criticized the choice to stop anginal meds at the time of the procedure because this is not likely what one would do in practice. Recall that in Orbita one, PCI was tested as an add on to anginal medicines and it had no effect over placebo.

But in Orbita two, meds were stopped at the time of the PCI in an effort to scientifically answer the question of whether there is any placebo resistant effect, and there was. So Orbita CTO Like Orbita II shows that there is placebo resistant effect, but it remains unknown how much this would add incrementally to medical therapy. Now there is one interesting observation that may shed light on managing these patients in practice.

And that is that patients in the placebo arm of this study enjoyed an immediate improvement from the placebo procedure. Now, why would patients who are symptomatic on anginal medicines, who had their angel medicines stop? have a symptomatic improvement from a placebo procedure. Now, that could be placebo effect, because clearly uh the bigger the procedure, the more the placebo effect. But the other thing that doctor Davies mentioned is that it could be stopping the medicine.

It may be that higher perfusion pressure improves blood flow in collateral vessels. Well, the authors planned more analyses to help sort out this finding, but I found that interesting. Finally, it's important to translate this data wisely. These were highly selected patients, highly selected chronic total occlusions, and highly selected skilled operators.

This moderate angina relief should not be assumed to be there in more complex diseases and sicker patients. Kudos to the Orbita CTO team again.

All right, next trial is the Chips BCIS three trial. This is the microaxial flow pump. This is a device placed in the left ventricle and it provides liters of cardiac output. Engineering-wise, it is an incredible device, right? Taking blood from the LV, spinning it around and pumping it out to the aorta, and providing this cardiac output.

Now this device goes by the brand name Impella and today we break the rule against using brand names because Impella is so much easier to say than microaxial flow pump. After the Impellers approval in two thousand eight through the LAX five ten K program, It's had two main uses in cardiology. Uh in one is in acute MI related cardiogenic shock, and the second is in support of high-risk PCI.

I did a liter literature search and it was aided by large language models and it estimates that between 50 to 100,000 impellas have been used for high-risk PCI. And that is the topic of today. But as background, In twenty twenty four, the danger shock trial randomly assigned three hundred and sixty patients with MI related cardiogenic shock.

After screening twelve hundred patients over ten years to impelause or standard care, and DangerShock found a statistically fragile twenty six percent reduction in death with the device p value zero point zero four. The CHIPS BCIS3 trial randomized 300 patients with severe left ventricular dysfunction and extensive severe coronary disease across twenty-one centers in the UK to impell a supported PCI or standard PCI.

These were high-risk patients with a median LVEF of 27%. More than 80% of the patients required calcium modification. 75% had left main disease, and about a quarter had a chronic total occlusion. The primary endpoint was a hierarchical composite of death, disabling stroke, MI, cardiovascular hospitalization, or periprocedural myocardial injury, and it was analyzed with a win ratio.

And the results, over a median follow up of twenty two months, there were thirty six point six percent wins in the impella arm. versus forty-three percent in the standard care, the win ratio is zero point eight five, that that conference intervals go from zero point six three to one point one five, and this is clearly a non-significant end result.

But this no result hides, I think, the main finding. Death from any cause was 54% higher at Neampella arm, 32.6 versus 23.4. That's a 1.54 hazard ratio for death. The conference intervals were nearly significant, 0.99 to 2.41. However, cardiovascular death was ninety-one percent higher with the impellate-treated patients.

twenty six point seven percent versus fourteen point five, that hazard ratio for CV death one point nine, ninety-five percent conference intervals one point one to three point three. As for safety, major bleeding occurred in 10.8% of patients in the impelling arm versus 7.3% in the standard arm. That's a relative risk ratio of 1.48, conference interval 0.7 to 3.1.

So I wrote a column on this trial, note also my first rule about asking whether the purpose of a trial was to answer a question or sell a product. Here you have another example of a non-industry funded trial designed to answer a question. Now, while there was clearly no benefit in a comprehensive win ratio hierarchical endpoint, There was a twelve percent higher rate of C V death in the impella arm. That's twelve percent absolute. That's a number needed to kill

of eight. Recall that the number needed to kill in the CAS trial was twenty one, which means that impella was three times more deadly than antirhythmic drugs, post my The specific message here is obvious. Don't use Impella for high-risk PCI in patients like those in chips PCI-S3. And I think guidelines should clearly downgrade this to a class III harm.

But the larger message is ten times more important and goes back to my feelings about medical conservatism. This device has been on the market and used for this indication for 17 years. And now, after about a hundred thousand patients have been treated, we learn that it is clearly harmful. No, I'm not against innovation. I love innovation. But new invasive devices like Impella should be adopted after there's evidence.

The solution is bloody obvious to me. You pay for new devices only if the patient is part of a trial. Then, after one to two years, you have proof of benefit before 17 years of harming patients takes place. Now, a part of this blame, maybe a large part of the blame for the impelled debacle falls on us, doctors. Why are we so gullible to marketing? I mean, we went to school and we studied science for years. And yet we accept shoddy evidence or no evidence. I don't get it.

To those listening to this thinking, well, what about danger shock, John? Come on. There was benefit in patients with AMI and cardiogenic shock. My response to that is yes, yes, but it took ten years for P.I. Jacob Moeller to find 360 patients suitable for his trial. And even then, the twenty six percent reduction in death was extremely statistically fragile.

And one more thing about danger shock, David Brown pointed out another issue to me, and that is that in the as treated analysis of danger shock, which included nine patients in the impella arm who did not receive the device and three patients in the standard arm who did receive the device

Here, the hazard ratio was 0.77, but the conference intervals were 0.57 to 1.03, so that's non-significant. And this is notable because if a device is conferring benefit, You would expect more benefit, not less, in the analysis where you count only patients who receive the device. Therefore, the highly selective nature of danger shock, its fragile statistical benefit, and the non significant as treated analysis, I think calls strongly for another trial of impella and cardiogenic shock.

Again, I repeat, on what planet is it right to be doing these trials 17 years after the device has been on the market? My friends, we need to demand more, not less, from our industry partner.

All right, final topic today, and I know I've gone long, but it's a big ACC recap. And I'll have another ACC recap next week. This is Champion AF, Watchman versus Dox and Patience with AFib. And I left this for last because perhaps you've read my column on it or you've seen my post on X that uh came out when the trial was presented, and that post has a hundred and two thousand views. If you haven't heard any of this or you want to hear it again, here's the summary.

Champion AF randomly assigned three thousand patients with AF to either Watchman Flex. Percutaneous left atrial appendage closure device or doax. These were patients eligible to take long term anticoagulation.

Importantly, these were lower risk patients than in a similar German trial Clojure AF. In Champion, patients had a median Chadz vask of only three point five and a very low Hazblood score of only one point three, and contrast that with Clojure AF, where patients had a Chadzvask of five point two and Hazblood of three point one. Now both Clojure AF and Champion ask provocative questions. Provocative because millions of people worldwide

have atrial fibrillation and take oral anticoagulation to prevent stroke. And this practice of anticoagulation rests on one of the strongest evidence bases in all of medicine. Yet, of course, we all know the two major challenges of anticoagulation are bleeding and adherence. So everything that's wrong with the Champion AF trial, and there is a bunch, had to do with the endpoints and the analysis of the endpoints. It was a non inferiority comparison for efficacy.

Now that's debatable because Watchman is more invasive, but perhaps if it allowed a patient to avoid anticoagulation for life you could justify non inferiority, maybe. You could also justify showing percutaneous appendage closure should be superior to an accoagulation. So the efficacy endpoint in the trial was strokes of stemic embolism and C V death, and the first problem with that is that C V death

is not affected by either DOEC or Watchman, so it just adds noise. Yet the fatal problem with this endpoint was its analysis, namely the choice of non-inferiority margin of four point eight percentage points on the absolute scale. Now this was chosen based on an expected event rate of twelve percent, which means the authors would deem Watchmen Not inferior to Do X if the upper bound of the ninety-five percent confidence level or worst case scenario for stroke systemic embolism and C V death

was no more than four point eight percent higher than in the doac arm. But, but, Four point eight percent non inferiority margin has to be based on the twelve percent rate. That translated to a relative increase of one point four zero or forty percent higher. Well, Sanjay Call at ACC told me that most, if not all, d non inferiority drug trials use a non inferiority margin that's set in both absolute and relative terms, and it's gonna become obvious why that should be.

The champion AF authors did not do this. They just used the absolute risk difference. So here's what happened in the result. The incidence of the primary efficacy endpoint was far less than the predicted 12%. It came in at 4.8% in the Doak arm and 5.7% in the Watchman arm. That absolute risk difference between the two treatments was 0.9% higher than the Watchman arm. The ninety-five percent conference intervals go from minus zero point eight to two point six percent.

And because the upper bound of the ninety five percent conference interval was two point six percent less than four point eight percent, the authors declared non inferiority, and there it sits in the New England Journal of Medicine. But come on. The problem is that this non inferiority margin was based on a much higher predicted event rate. And when a rates come in much lower, as in this case, they came in at four point eight percent,

That 4.8% risk margin is far too generous. And that is why it's standard to set out both an absolute risk margin and a relative risk ratio. Like I just said, the four point eight percent absolute risk increase corresponds to a forty percent higher rate of the primary endpoint. So effectively the non inferiority margin should also be one point four zero or forty percent higher if we're using relative risk.

So you go to table two of the paper. Table two shows that the zero point nine percent higher rate of events on the watchman arm. corresponds to a relative risk increase of one point two zero. That has conference intervals. Those conference intervals go from zero point eight seven to one point six six. So had the standard non inferiority margin been used, the upper bound of one point six six is higher than one point four zero and non inferiority would not have been met.

So the components of the efficacy endpoint also looked worse for watchmen. Strokes went the wrong way. As it did in Protect, Prevail, and Closure AF in Champion, stroke was 46% higher in the Watchman Arm compared to Doak. There were 33 strokes in the DOAC group versus 50 in the Watchman arm. That's 2.6% versus 3.3%. That hazard ratio is 1.46. And the upper bound of that ninety five percent conference rule is two point two seven.

Now the difference in ischemic stroke was even more striking forty five twenty seven. That hazard ratio is 1.61, and the upper bound or worst case scenario could be that Watchman has a 2.59 higher rate of ischemic stroke than Doex. Now you ask about hemorrhagic strokes. You'd think surely hemorrhagic strokes would be less in the watchman arm. No, they occurred in five patients in both groups. My friends, antiplatelet therapy is not benign.

Some might say that the low stroke rates preclude making conclusions because of low power, and I would admit that technically that is true. But I would point you to the five percent rate of device related thrombus and the fact that twenty percent of patients had incompete closure of the appendage at four months.

Even Watchman proponent Christopher Ellis from Vanderbilt pointed this out in a post on X, and I'm gonna read it to you because Christopher Ellis is one of the true proponents of left atrial appendage closure. He writes, quote, honest, sobering fact here is that we need better devices before any push as frontline therapy.

Granted, CTA picks up more device related thrombus than TE, but no surprise DRT of four percent and a twenty percent leak rate has something to do with the ischemic stroke numbers. Surprising ICH was the same between groups, and it was lower than in the ocean trial. There you have it. Efficacy wasn't the only fatal flaw in this trial, safety was also hidden by the choice of primary safety endpoints.

Now, instead of simply counting all major bleeds in the two arms, the authors chose non procedural bleeding and quote clinically relevant non major bleeding as the primary safety endpoint. This strongly favored the watchman arm ten point nine percent versus nineteen percent and the authors claimed superiority

for safety. Yet, come on, no patient can exclude procedural bleeding, which is the most common complication from this procedure. And worse, In an OBEN-label trial, everyone knows, everyone knows, patients on DoX will complain of more non-major bleeds. things like gum bleeding and bruising and nosebleeds. The authors say these bleeds required a healthcare interaction, but unlike option authors, they did not list the actual events in the supplement, so we don't know what they were.

The proper safety endpoint of all major bleeds was called the secondary endpoint in this trial, and instead of testing to see if the new therapy, the watchman, was superior in safety, they tested this for non-inferiority. And the results are as they always are in previous trials of Watchman versus anticoagulation. When you count all bleeds, they're about the same. In champion, 83 patients bled in the watchman arm, 87 in the doac arm. That's a hazard ratio of 0.92, clearly non-significant.

But that's not the only way they obfuscated this. They obfuscated it further by creating an endpoint called Net Clinical Benefit Endpoint, where they used stroke, systemic embolism, and CV death, and then they combined it with not all major bleeding, but non procedural and non major bleeding. And of course this came out positive because non major non procedural bleeding favored the watchman arm. a better net benefit calculation, one that a common sense patient would do, would be

To count up that there were 17 more strokes in the Watchman arm and only four less bleeds. Now you don't even need statistics for that net benefit calculation. So my final comments. This is on the board as a win for Watchmen. This company was so confident that they even sent reps to our office to detail clinicians with free lunch before the trial was released with a little flyer that said, is your practice ready?

But while the trial sits in New England Journal of Medicine as a positive trial, it is clearly not, and I hope I've convinced you of that. Watchmen looked worse in this trial, worse in efficacy, and no better in safety. I hold out some hope that public critical appraisal will win the day. I think it would be a tragic mistake to use this trial data, especially in the face of strongly negative appendage data from Clojure F, to offer this therapy to patients with AF on anticoagulation.

Again, I reiterate the words of Doctor Christopher Ellis. Honest, sobering fact here is that we need better devices before any push as frontline therapy. And of course I remind everyone that oral anticoagulations for patients with AF is backed by some of the strongest evidence in all of medicine.

The old warfarin trials showed clear benefit over placebo or aspirin, and then the doac trials showed clear advantages over warfarin. Nearly a hundred thousand patients have been randomized in trials of inaccoagulation.

And I want to close by strongly pushing back against the notion that was pushed by proponents at the meeting that we should use this data for shared decision making with patients. If this trial is used for shared decision making to convince patients to have watchmen or appendage closure instead of anticoagulation, then shared decision making is dead.

The three overlapping circles of evidence-based medicine are clinical expertise, best evidence, and patient values. In this case, our clinical expertise and critical appraisal of the best evidence tell us. to dissuade, not encourage patients to choose inferior therapies over proven therapies. And we still don't know whether this device has any role at all in cardiology because patients who struggle with anticoagulation because of bleeding have not been studied in trial.

And based on the inferior results of Closure and Champion, I believe that we should have an appendage closure trial versus no anticoagulation arm in frail pa older patients who cannot take anticoagulation. And I believe that that trial, if it's ever done, would be stopped for harm in the watchman arm. And that's it for this week in cardiology, and I am sorry that I went over, but these were

really important trials. Uh next week I'll have a number of more trials. I'll try to keep it under thirty minutes. Perhaps it'll even take three podcasts to get all of ACC trials in. Until next week. This is John Mandrola from the Heart.org Medscape Cardiology. org, Medscape Cardiology. Any views expressed are the presenter's own and do not necessarily reflect the views of WebMD.