Dec 19 2025 This Week in Cardiology

You're listening to This Week in Cardiology from TheHeart.org, Medscape Cardiology. This podcast is intended for healthcare professionals only. Any views expressed are the presenter's own and do not necessarily reflect the views of WebMD or Medscape. Hi everyone, this is John Mandrola from the heart.org Medscape Cardiology, and this is This Week in Cardiology for December 19th, 2025. This week...

The MI paradox seen in risk scores to celebrate trial of a new subcutaneous glycoprotein 2B3A inhibitor with its interesting endpoint. the SURPASS-CVOT trial, and the bad story of adonexit-alpha, the reversal agent for factor Xa inhibitors. First, an announcement. This week in cardiology...

is off for its usual two-week break for the holidays. I will return on January 9th. First topic, what do risk scores mean for individuals? Or, perhaps a better question, Is a first MI preventable? Cardiologist Wayne Kessmeyer sent me a very interesting small study that I would not have seen from Jack Advances, and I want to discuss it today.

The lead author, Dr. Anna Mueller, and her team at the Mount Sinai Morningside Hospital in New York looked at risk scores in 465 patients who were less than age 65 who had had their first MI. Their goal was to characterize how well the ASCVD risk scores and symptoms reliably identify patients at risk for a first ACS event. The teaser, I would say, is that...

Not very well, actually. Now, this study is elegant in its simplicity. They had clinical information on 465 patients who had had their first MI. They then tell us the distribution of the risk scores. And I tell you, it's an eye-opener. And I think it really has some humility-inducing messages. I say that because we put a lot of weight on these risk scores. Okay. For the standard ASCVD risk prediction and by pool cohort equation, the authors found that low risk patients, less than 5%,

153 out of 465 patients who had their first MI had low risk, less than 5%. That's 33%. Those with a 5 to 7.5% risk. This was 12% of first MIs, intermediate 28%, greater than 20% comprised only 10% of first MIs. And then, of course, the category of LDL greater than 190 or diabetes, where you wouldn't use the pooled cohort equation, these were 17%.

Okay, using the PREVENT score, which I've talked about, this is this new score that presumably has better calibration. So 45% of first MIs had a low risk, less than 5%. 16% of first MIs had risk calculated at 5 to 7.5%. And 23% were intermediate risk. Only 3%. had a prevent score greater than 20%, and only 13% had the category of LDL greater than 190 or diabetes, where you wouldn't use the risk score. Now, what about timing of symptoms?

Timing of first ever symptoms of chest pain or dyspnea before the event. 53% had symptoms on the same day, 7% less than two days, and about a quarter, two to seven days, a smattering longer than that. So the authors made some summary comments. Using the pooled cohort equation, they wrote, statin or imaging were not recommended in 45% of the first ACS patients. Using the PREVENT score, statin or imaging were not recommended in 61% of the patients who had their first MI.

Nearly two-thirds of first MI patients in the last five years, so a contemporary time period, had symptoms only in the first 48 hours before the event. Now, the author's discussion, which I want to note, and is quite good, except one thing that I disagree with that I'll talk about the end. They write that the risk scores have enhanced CVD detection at the population level. But they have very limited role in individuals. And I think that's right. And I think we forget it.

Because remember, if a score has a C statistic of like 0.72, that is awfully close to 0.5, which is a coin toss. The authors also mention this MI paradox, wherein risk scores successfully identify high-risk groups at the population level, but they fall short in low-risk people. due mostly to the issue of mathematics. And that is that even though low-risk individuals have a small percentage chance of having an MI, there are so many more people in this low-risk category.

that they contribute a large absolute number of heart attacks. As the authors note, quote, the large denominator of low risk asymptomatic individuals mean that even a small percentage of events in this group

results in a considerable absolute number of MIs. In their study, 45% of patients, nearly half, who had their first MI would have been classified as way too low a risk to warrant statin therapy or further testing using the pulled cord equations and would prevent that number jump to almost two-thirds of patients.

Now, I want to make a correlation here. This is very similar to the ICD and sudden death problem. To wit, we can identify high-risk people who have low ejection fraction and heart failure, and those seminal trials done 20 years ago showed that an ICD... improves survival though a footnote here is this needs to be retested in new trials and that's being done in the profit trial over in germany but the vast majority of cardiac arrest cases occur in lower risk individuals

who don't have heart failure, or a terrible ejection fraction, or even known heart disease. Dr. Robert Meyerberg described this in the New England Journal of Medicine, and I'm going to link to that paper that he wrote. Now also... A UMC Amsterdam-led group also reported on this issue early this year in the European Heart Journal.

And they report this sort of same MI paradox in a much bigger cohort of patients. And they found that nearly one in five of the patients who had an MI, first MI, had no standard modifiable risk factors. or no doctor visits beforehand. And it's exactly as the Mount Sinai authors found. About two-thirds of these patients were not taking preventive medicines.

The Mount Sinai authors, and this is where I disagree, mentioned that a key to resolving this MI paradox might come with imaging. Quote, detecting early atherosclerosis, the disease itself may offer a more effective and personalized approach to treatment, which enables earlier intervention to halt plaque progression and reduce the likelihood of future events. I am not so sure about this. Okay, when you look at CAC scores,

Such as when you put the CAC score into the MESA risk calculator that you can look up online, it really doesn't add that much predictive ability. Maybe a little. I bet you'd see the majority of MIs in low risk. cac individuals but of course this is a testable hypothesis with large databases i mean maybe someone has data on this

But I think you'd see the same MI paradox would exist because the majority of MIs in young people are not going to occur in heavily calcified lesions. They're going to occur in people with low CAC scores or CAC zero. Now, one reaction to these findings might be the urge to treat younger people with lower risk. Like maybe we should lower the 10-year risk. Don't use 7.5%, use 5% or 2.5%.

Basically, that would make almost anyone alive and middle-aged statin eligible. And I don't think that is a good answer either. Others have suggested better risk markers. Okay, like maybe if we use... genetic polygene risk scores or adding CRP or inflammatory markers. In fact, there is a Reynolds risk score with a CRP and family history added.

And these risk scores with adding more might help on the margin, I think. But I think there's still going to be a huge numbers of low-risk people that'll have the majority of MIs. Now, you might tell me that AI models will magically predict those will have an MI, but maybe so. Maybe I'm just not imagined enough, but I doubt it. Perhaps the best we can do is something much more humble, and that is...

recommend healthy living to our patients. We say, don't ignore your symptoms. And then as a policy, we fund capable cath labs to care for STEMIs. Of course, having enough cath labs is sort of like having enough pilots to fly airplanes. This requires rigorous training of doctors and staff, and it requires the funding to keep them available. You have to peep

Pay people well to be on call. Interventional cardiologists are well paid, no doubt, but given their call schedules, I could argue that they are underpaid. Right now, it's never been a better time to have an MI. The technology is amazing. But I hope that does not change in the future. I hope that available high-tech cath labs are increasingly available, not decreasingly available.

All right, second topic, speaking of access to emergent MIPCI care, the CELEBRATE trial of a new glycoprotein 2B3A inhibitor called zalunfaban. was published in the New England Journal of Medicine Evidence Journal. Now, you don't know what this drug is, and neither did I because it's not yet FDA approved, and I'm sorry if I'm pronouncing it wrong. Zalunfaban is a glycoprotein 2B3A inhibitor designed for subcutaneous administration on first medical contact in patients with a suspected STEMI.

Glycoprotein 2B3A, remember, is the final common pathway of platelet aggravation. It includes thrombin-induced activation. And these inhibitors are more potent than P2Y12 inhibitors.

and they have been associated with improved infarct-related artery patency, better LV function, especially with earlier treatment. However, these first generation... GP2A3Bs have important limitations, and that is the need for continuous infusion, and more notably, the increased risk of severe bleeding and potential GP2A-induced thrombocytopenia. So you enter this new drug, Xelonfoban. It's a new generation small molecule designed for single dose sub-Q administration at first medical contact.

It achieves near complete platelet inhibition within 15 minutes and it's antiplatelet. effect diminishes rapidly the plasma half-life is about one hour and thus a reduction of antiplatelet effect by approximately 50 within two hours Therefore, the drug may have a more favorable pharmacokinetic pharmacodynamic and safety profile for the early treatment of STEMI patients compared with these earlier first-generation glycoprotein 2B3As.

Now, it all sounds great, doesn't it? Sounds perfect. Well, of course, you need an RCT. And that's what I'll talk about now, the CELEBRATE trial. Now, even though this trial is studying a drug that's not yet approved,

It has a really interesting endpoint to discuss. So Celebrate had 2,500 patients with STEMI, and there were three arms, one placebo arm and two with... the different doses of the drug one slightly lower one slightly higher the primary endpoint was the super interesting thing this was a hierarchical proportional odds model that ranked seven endpoints from worst to best

So it went death, stroke, MI, stent thrombosis, new hospitalization for heart failure, larger infarct size with troponins, or no endpoint through 30 days. The primary safety endpoint was bleeding. The authors combined the two drug arms with the different doses versus placebo, and they got a positive result. for the the new drug the adjusted odds ratio is 0.79 that's a 21 reduction

And the conference intervals go from 0.65 to 0.98. The p-value is at 0.03. So the 21% reduction was statistically positive, but pretty fragile. Should we incorporate this through drugs? Should FDA approve it? It's a positive RCT. Now it's funded by the company that makes it, of course. And here are the components. Death, 2.3 versus 2.2%. So absolutely no difference. Stroke, 0.7 versus 0.8%. No difference. MI. 1.9% versus 1.2%, so slightly higher for the drug.

Stent thrombosis, better. 0.2% versus 1%. That's an odds ratio of 0.18, but it's only 3 versus 8 events. Heart failure hospitalizations. Better with the new drug, 6.5 versus 8.1%. And MI, large MI, that's greater than 30 times the upper limit of normal of troponin. 85% versus 88%, so a little better. Odds ratio 0.77. And then, of course, this funny endpoint at the end, which is no major MACE endpoint, which is basically event-free survival, 13.3 versus 9.8. That's an odds ratio of 41% better.

Secondary endpoints were mostly comparable, though there was slightly better flow at index angiography in the active drug arm. Also, surprisingly, there was no safety signal of bleeding. Rates of major bleeding were low, 1.2 versus 0.8%, although mild to moderate bleeds were higher with the active drug, 2.6%.

So this is going to be a tough call for regulators, I think. This is an easy drug to administer. Boom, one subcutaneous injection. The positive results have occurred on top of really good MI care. And as the authors point out, the median time from drug to angiography was relatively short in this trial, 36 minutes. That's shorter than typical real-world delays.

Longer delays could potentially amplify the benefits of the Zell-Unflamin. But my question is the end point. Yes, it's positive. But the major things like death, stroke, MI, were similar. MI was even a bit higher for the active arm, though stent thrombosis was better, but the numbers were tiny, only 3 versus 8 events. The driver of the endpoint was heart failure hospitalizations and troponin.

And of course, event-free survival in general, no major endpoint. Now you want a drug though to move death, stroke, or MI. Less important to me seems a hospitalization for heart failure and even very little importance I would put on troponin levels. The other headwind was not only the matter of clinically important differences like death, MI, and stroke, but the statistical signal was fragile.

The 21% reduction in the composite had an upper bound of the 95% conference interval was 0.98, so close to 1.0, and the p-value was 0.03. So it would not take many... event switches or non-events to swing to non-significant. I put this trial up on Frank Harrell's discussion forum, which is a statistics forum.

And to my surprise, the statisticians felt like it was pretty positive. The thinking here is that such a hierarchical endpoint allows capture of the most data, so you don't waste data. And of course, that's true. But the clinical translation is what we do as clinicians. Now, the other translational matter, of course, is going to be cost. Surely this new drug will be pricey. And do these results justify a higher cost?

I'd say probably not, but we shall see. I also wonder whether FDA would require two trials, because certainly another large trial would help. Again, as I mentioned in the first study, we have come so far in cardiology, so it's hard to do much better. Let me know what you think about this.

All right. Third topic today is that the SURPASS, CVOT trial, is published in the New England Journal of Medicine. I first covered this August 1st, 2025 podcast when Lily put out a... press release about the drug. Surpassed CVOT was truzepatide, which of course is the GIP-GLP1 dual agonist, versus dual glutide, which is trulicity. in patients who had diabetes and established cardiovascular disease. Now, as I said, I discussed it previously on the October 1st podcast.

The trial began in 2020 and enrolled about 13,000 patients and went on for about five years. Now, first, a little background. Dulaglutide was shown to reduce cardiovascular outcomes in patients with type 2 diabetes. and either establish cardiovascular disease or high risk for cardiovascular disease. This was the Rewind trial in Lancet 2019, and I'll link to that.

Now, these results were close on the primary endpoint of MI stroke and CV death. It was 12% duoglutide versus 13.4% placebo. That hazard ratio was only 0.88. So only a 12% reduction with a p-value of 0.026. Now, since duoglutide had shown this benefit, the next trial, of course, with trisepatide had to be against duoglutide, not placebo.

The SURPASS CV outcomes trial was designed as non-inferiority, which is questionable, and I'll go over that. These patients in SURPASS were 64 years old, 28% female. They're from all over the world. This was a multi... multi-continent trial, and all had some established cardiovascular disease, two-thirds with coronary disease, the others with peripheral artery disease.

The BMI of these patients was high at 33. Also, the hemoglobin A1c was high at 8.4. The primary endpoint of MI stroke CV death was 12.2% terzepatide. Versus 13.1% dual glutide. That relative risk reduction hazard ratio is 0.92. That's an 8% reduction. And the conference intervals go from 0.83 to 1.01. So let's pause there and think for a second. An 8% reduction with an upper bound of just over 1.

So if this was a typical superiority trial, it would have not met statistical significance. But it was a non-inferiority design, so terzepicide could be declared non-inferior. Let's look at the components of the endpoint. CV death, 5.6% trisempotide, 6.2% dual glutide, an 11% reduction. MI. 4.7 versus 5.4 percent. Hazard ratio there, 0.56. Not significant, but, you know, in the right direction. Stroke about the same. All cause deaths, which...

people are going to talk about was 8.6% terzepatide, 10.2% to a glutide. So that has a ratio of 0.84. And the conference intervals go 0.75 to 0.94. Now, this is a secondary endpoint that's not tested formally, but clearly that would be in the range of statistically significant. Two other secondary endpoints that were important is that weight loss was greater in the terzepatite arm, 11.6 kilograms versus minus 4.8 kilograms.

And hemoglobin A1c dropped a lot more too, 1.6 versus 0.9. So the conclusion that sits in the New England Journal of Medicine, quote, Among patients with type 2 diabetes and cardiovascular disease, terzepatide was non-inferior to dual glutide with respect to a component endpoint of CV death, MI, or stroke. So my comments. At the time in August when the press release came out, I wrote that the key opinion leader, Dr. Muthu Vadaganathanen,

Rodon asks, quote, the game has changed. The surpassed CVOT trial met its primary and secondary endpoints in the first head-to-head CV outcomes trial. Dr. V noted that there was a 16% lower risk of all-cause mortality. Yet, at the time, Dr. Sanjay Kahl noted that SURPASS was powered for a 15% relative risk reduction in MACE. and the primary efficacy endpoint comes out only at 8% lower, and the 95% conference intervals went from 0.83. So it barely contained that powered endpoint of 0.85.

Call also noted that the superiority of the comparator duoglutide had not actually been established in this patient population. Now, I told you that the Rewind trial of duoglutide versus placebo was positive, and it was. But Call noted that the subgroup of patients who had established vascular disease, which was about a third of patients, The hazard ratio of dual glutide versus placebo was 0.87 with conference intervals that were 0.74 to 1.02.

Sanjay Kahl also asked what to make of the 16% reduction in all-cause mortality. It's a good question because you only have an 8% reduction in mace and you want... are going to say that there's a 16% reduction in all-cause mortality. Now, despite the fact that terzepatide reduced A1c, it reduced weight, it slowed CKD, but...

There were no significant difference in CV events, so it's hard to explain all-cause mortality. And that made me say in August that all-cause mortality was likely a statistical noise issue. That p-value was not adjusted for multiple testing, but more important, I thought, was that if a drug is a cardiac disease modifier, then CV death and CV outcomes should drive the reduction in death. But...

But my friend, Dr. Vank Murthy, an academic cardiologist at the University of Michigan, made some nice comments to me on Signal this morning. We were talking on Signal. as i typed out some of my thoughts for the podcast and vanck points out that a the point estimates are all in the same direction right cv death was 0.89 all cause death was 0.86 and

He also points out hyperinsulinemia and obesity are both linked to cancer risks. Consider that a more potent GLP agent may reduce alcohol intake or smoking. and this may lead to less cancer and pneumonia, a driver of mortality. Less alcohol could reduce car crashes, for instance. Veig notes that unlike many cardiovascular drugs, this class of drugs...

The GLPs have manifold impact, and it could seriously improve non-cardiovascular disease events as well. It's a mistake, he thinks, to consider only CV events. Now, Murphy agrees that secondary endpoints are less reliable than primary endpoints, but he would not dismiss them entirely, especially when directionally they're similar.

Now, I would also note that there were some other trial procedure issues to think about in this trial. One was that terzepatide was titrated to maximum dose until glutide had a fixed one dose. The other thing that really bothers me is that the non-inferior design is strange. Non-inferiority designs are used for interventions that offer something else, something less invasive or less costly.

or less risky. And I don't see any of that true with terzepatide. In my mind, this should have been a superiority trial, as non-inferiority is a lower bar to cross. That said, that terzepatide is clearly a more potent drug, there's greater weight loss, a modestly lower rate of mace, and the all-cause death signal may not be noise, it may be real.

So to me, I guess trisepatide is the drug of choice here, but not by that much. All right, final topic today is a humdinger, and it's a great lesson in evidence-based medicine, and that is that... The FDA has pulled the factor Xa reversal agent at an exit alpha from the market. This is a factor Xa reversal agent was studied way back in 2019.

in a trial called ANEXA-4. I said trial, but it's not really a trial because it only had one arm. It's had 352 patients who all had acute major bleeding within 18 hours after administration of a factor Xa inhibitor.

The patients all received Ebola's adnexit followed by a two-hour infusion. It had a co-primary outcomes where the percent change in anti-factor Xa activity after... adnexa-alpha treatment, and the percentage of patients with excellent or good hemostatic efficacy at 12 hours after the end of the infusion, with hemostatic efficacy adjudicated on the basis of some pre-specified criteria.

These were older adults, age 77, all had established cardiovascular disease. Bleeding was predominantly intracranial in about 64%, GI in 26%. The reversal agent... Basically, successfully reduced anti-factor Xa activity. Excellent or good hemostasis occurred in 82% of patients who could be evaluated. Within 30 days, death occurred in 49 patients, or 14%, and a thrombotic event occurred in 10% of patients. However, reduction in anti-factor Xa activity was not.

not predictive of hemostatic efficacy overall, but was modestly predictive in patients with intracranial hemorrhage. Notice, my friends, no comparison group. 10% of the group had thrombotic events, including 14 ischemic strokes. There were also 35 cardiovascular deaths, yet the drug was approved for use, and it cost many thousands per dose.

I had never had a chance to use it, mainly because many pharmacies didn't carry it. Okay, then we get the Anexa I trial, or maybe Anexa I trial, of Adnexit Alpha. versus standard care. So finally we get a comparison. This was published last year in New England Journal of Medicine 2024. The late Stuart Connolly is the first author.

This had about 260 patients with intracranial hemorrhage in each arm. The primary endpoint was hemostatic efficacy that was defined by expansion of the hematoma volume by 35% or less. at 12 hours after baseline an increase in the score on the nihs scale of less than seven at 12 hours and no receipt receipt of rescue therapy between three and 12 hours Safety endpoints were thrombotic events and death. Safety endpoints were thrombotic events and death, and that's going to come up.

The trial was stopped early for benefit. The primary endpoint, which was measures of hemostatic efficacy, all favored adnexit alpha. However, thrombotic events were higher. 10.3% versus 5.6%, that absolute risk increase of 4.6% absolute risk increase actually met statistical significance. And there were 17 versus 4 ischemic strokes, or 6.5% versus 1.5%. And there were 11 versus 4 MIs.

In the end, the difference in stroke severity by modified ranking scale was not different. The editorialist at the time wrote, quote, Longer term outcomes at three or more months, a standard measure in most stroke clinical trials, including those for intracranial hemorrhage, were not reported. Okay, so... Hematologist Richard Buka has a great thread on X that I will link to it.

He wrote that the 2019 study had no control arm and that 10% had thrombotic events and most patients stopped bleeding, but was it due to the adnexid alpha or just patients stopped bleeding? Without a control arm, the trial was meaningless, but Adnexa Alpha plowed on anyways, and it quickly gained conditional approvals from regulators across the world.

Despite deficiencies in evidence, there was a massive buy-in from key opinion leaders. Luca notes that the 2018 CHESS guidelines had the comment that in a patient with serious bleeding, A specific reversal agent should be used instead when available. The guidelines went on and said that general hemostatic agents are less effective and have not been shown to improve outcomes and are potentially prothrombotic. I find that a very dubious statement because no comparison had been made yet.

the comparative trial wasn't published until 2024. Bucca also notes that the 2020 ACC guidelines consensus statement also recommended Adnexit Alpha. So it turns out that the 2018 FDA approval was contingent on an RCT being performed. But before that, several highly confounded observational studies showing a massive benefit for adnexin-alpha were published. But Buca notes that the Annexa eye trial was not powered for clinical outcomes, even though these were worse in the reversal agent arm.

He attached an FDA slide from the FDA meeting showing that while the trialist adjudicated a 10.9% thrombotic event rate in a trial, the FDA reviewers found nine more events, or 15%. in their trial analysis. So he writes that this trial is a huge lesson for evidence-based medicine, but one that will probably be swept aside and forgotten.

He said there should have been a meaningful RCT in the late 2010s, but instead we allowed this drug to come to market and remain untested. Bucca writes that I have seen the evidence skewed, slanted, tarted up, spun in so many ways. A year ago, marketing was pulled, but the drug kept on being sold. First sold in 2018, he writes, Annexa had been earning revenue for nearly eight years. $400 million between 2021 and 2023.

Only now are real-world studies confirming the thrombosis rate, something the aforementioned propensity score match studies failed to find. In other words, you couldn't see those thrombotic events in the observational studies. Bucco says that he's done numerous talks about his skepticism about reversal, which also extends to PCC, by the way. After one, a key opinion leader challenged him's argument and said that if you're bitten by a snake, you need the antidote.

And then he writes, there you go. That shows you how misguided that we can be. Now, I also want to point out that the current director of the CBER at FDA, my friend Vinay Prasad, did a podcast on this study, a plenary session, in 2020, and I'll link to it. And Vinay expressed extreme doubts about the single arm study of the reversal agent. He ended up being correct then.

based on this critical look at the data. And I must say that I was wrong not to be skeptical as well. This got by me, and it's a great lesson in evidence-based medicine. We definitely should have had an RCT done first. At the time, it was felt to be unethical, but gosh, was that wrong? And I think it's often wrong when people say there's not agripoise enough to do a proper trial. So...

A few quick closing notes. My top 10 of 2025 cardiology stories is out. It's a long post. But it was a great year for studies in 2025. You can read the post and sections. Take a look at that. Finally, I hope everyone has a nice holiday season. It's cold and windy and rainy here in Kentucky. i'm resigned to doing my trainer rides though we expect warm one warm day over the holidays and maybe i can get my canyon arrow road out for a ride on the road

I want to seriously thank everyone for listening this year. As always, I learn a lot. And it's a fitting end of the year to finish with a story that I flubbed on for not being skeptical enough. adnex at alpha factor 10a reversal story. Now, while I am the second oldest cardiologist at my hospital, I have never enjoyed this job more.

We have such great tools in EP, and it's fun learning all aspects of cardiology. Last month, one of my partners was off, and I actually hit the 95th percentile for the number of procedures done. And pulse field ablation has truly shocked me in a good way. It's so much easier to ablate atrial fibrillation. We'll see you all next year.

You're listening to This Week in Cardiology from TheHeart.org, Medscape Cardiology. This podcast is intended for healthcare professionals only. Any views expressed are the presenter's own and do not necessarily reflect the views of WebMD or Medscape.