Please note that the text below is not a full transcript and has not been copyedited. For more insight and commentary on these stories, subscribe to the This Week in Cardiology podcast, download the Medscape app or subscribe on Apple Podcasts, Spotify, or your preferred podcast provider. This podcast is intended for healthcare professionals only.
In This Week’s Podcast
For the week ending April 4, 2025, John Mandrola, MD, comments on the following topics: ischemia with nonobstructive coronary arteries, cerebral embolic protection in TAVI, SGLT2 inhibitor in TAVI and an oral GLP-1 trial.
The WARRIOR Trial
I noted WARRIOR in my ACC preview column because the study of a mysterious condition of ischemia with non-obstructive coronary artery disease or INOCA is novel and a great place to spend NIH money.
The results aren’t formally published so I will keep the discussion brief, and focus on the concept of pragmatic trials. All trials have some components of pragmatism, but WARRIOR was really quite pragmatic because it randomized women with INOCA to either intensive medical therapy: high-dose statin and angiotensin-converting enzyme inhibitors (ACEI), or angiotensin receptor blocker (ARB) vs usual care.
This was a large trial of about 2500 women with symptomatic angina and no obstructive epicardial CAD. About half were diagnosed with Coronary Computed Tomography Angiography (CCTA) and the other half angiography. The primary outcome was MACE (death, stroke, hospitalization for angina or HF.)
I would normally tell you the baseline characteristics at this point. But instead I will tell you the results.
The rate of primary MACE events was actually higher in the intensive medicine strategy. The HR was 1.13 and the 95% CI went from 0.94 (a 6% reduction) to 1.37 (a 37% increase). The p-value was obviously non-significant, and the fact that the point estimate and most of the CI was above 1.0 pretty much excludes the chance that IMT was better.
There were no subgroups of note.
Now to the baseline characteristics: Nearly the same number of pts were taking statins. ≈71% in each arm. Nearly the same number of pts were taking high dose statins. Similar numbers of patients were taking ACE inhibitors or ARB 53%.
Another noteworthy finding was the tiny number of CV deaths, myocardial infaction (MI) and stroke or transient ischemic attack (TIA). More than 1200 patients in each arm and there were fewer than 15 of each CV death, MI and hospitalization for heart failure (HHF). The main driver of the endpoint was ≈300 events in each arm for hospitalization for chest pain (CP).
One thing that we learned that patients with INOCA who take these medicines may have lots of hospital visits for CP but the number of hard events is very low.
So, this was a non-significant result, but I don’t think it tested intensive medical therapy, because so many women in the control arm were on the same therapy. They call this “contamination”.
This is a common problem with pragmatic trials. When you study things in the real world, rather than with strict trial protocol, you can end up with the two groups on similar therapies. I don’t think we learned anything, despite the randomization and follow-up of 2,500 people.
Money was spent and little was learned. I would not conclude from this that intensive medical therapy is not helpful. Instead, I would say that since both arms got essentially the same therapy, the only conclusion is that the trial procedures preclude knowing anything about the intervention.
The biggest lesson from this trial is to be wary of super pragmatic trials. I used to think we could just randomize people within the realm of real-world practice, without strict protocols. Doing so would be perhaps more representative. The problem of course with this approach is noise. Patients in the control arm are free to take whatever they like.
I have come around to the idea that trials should be at least somewhat restrictive with their protocols. Yes, this might make a trial a special environment, one that makes translation to the clinic harder, but at least with such a restrictive trial, you can test two disparate strategies.
The authors’ discussion of the trial procedures issues will be interesting when the paper is published.
Cerebral Emoblic Protection: The BHF PROTECT -TAVI trial
The British Heart Foundation PROTECT TAVI was a great effort and trial. More than 3800 patients were assigned randomly to the cerebral embolic protection (CEP) device during transcatheter aortic valve implantation/replacement TAVI/TAVR. ≈ 3800 patients had TAVI without protection.
The CEP device is known to catch debris. Debris that would have lodged in the brain and caused stroke. The results of this randomized controlled trial (RCT) in the UK are clear: 2.1% of the CEP had stroke (the primary outcome) vs 2.2% in the no CEP arm. The actual numbers of patients were 81 and 82 in each group. The absolute risk difference was -0.02 and the 95% CI went from -0.68 to +0.63. The p-value was almost 1 at 0.94.
Disabling stroke was 1.2% vs 1.4%. Death was 0.8% and 0.7%. Adverse events were similar.
The authors also performed a special type analysis called complier average causal effect or CACE. I am not an expert and will probably not get this exactly right, but it is an instrumental variable analysis that attempts to mitigate bias in looking at a per-protocol type analysis.
For instance, in the trial, about 81% of the CEP group had the device deployed as designed but the trial was analyzed according to ITT. David Cohen helped explain this to me. He also attached a beautiful explainer paper from Drs Joshua Angrist, Robert Yeh, and colleagues in NEJM Evidence.
At the simplest level, you are dividing the average treatment effect by the difference in the rate of exposure to the treatment variable. So, in the case of PROTECT-TAVI, you divide the intention to treat (ITT) treatment effect on stroke (-0.1%) by the difference in exposure to the CEP device between the treatment and control groups (0.81), which yields an estimated compliance-adjusted treatment effect of 0.1/0.81 = 0.12%.
Which is not different from the ITT unlike in a trial like CABANA, the ITT and per-protocol treatment effects are similar.
The authors concluded that routine use of the CEP did not decrease the incidence of stroke during TAVI.
Comments:
These results were similar to PROTECTED TAVR which had 1500 hundred patients per arm (rather than 3800) and reported a stroke rate of 2.3% vs 2.9% in the control group. One positive from this “negative” trial was a signal of device benefit for disabling stroke (0.5% vs 1.3%).
It also appears that observational studies largely predicted the findings: Namely, David Cohen and colleagues used the STS/ACC/TVT registry to compare outcomes after TAVR with and without embolic protection in more than 414,000 patients. In a clever design using an instrumental variable analysis (site-level preference for device use as the instrument), they found a 13% statistically significant reduction in disabling stroke and a nonsignificant 10% reduction in all strokes with the device.
This result suggested rightly I think that PROTECT TAVI would a) have a very low rate of stroke and b) likely be underpowered. If you assume—from the observational study—that stroke rates would be 3% (they were slightly lower at 2.2%) and you then assume a relative risk reduction of 10%, a trial will have to have 100,000 patients to detect a difference.
Indeed, in the PROTECT TAVI trial, the 2.1 vs 2.2% stroke rates yield a Risk ratio of 0.99 but the 95% CI went as low as 0.73 (a 27% reduction) and 1.34 (a 34% risk increase).
It’s weird isn’t it. The trial has 3800 patients per arm, but because stroke rates are so low, the CI are wide, and given that wideness, you’d have to say the trial is not strongly informative. The device could have a huge 27% reduction or 34% risk increase.
The same comment holds for disabling stroke which was even lower in incidence at 1.2% vs 1.4%. Here the risk reduction (RR) was 0.89 but the 95% CI went to 0.60 (a 40% reduction) vs 1.31 (a 31% risk increase.)
While the 95% CI are wide, I think we can learn a few things relevant to stroke protection during TAVI. You can say the total stroke and disabling stroke rates in both trials are very low. When an outcome is that low, it’s always going to be hard to reduce. You will need huge numbers of patients. Even meta-analyzing the two trials will still lead to small numbers and minimally informative pooled effect.
That in and of itself is instructive. We can conclude therefore that since stroke rates are so low after typical TAVI, it is hard to justify placing a costly device that requires another arterial access and manipulation in the carotid.
I would think CEP devices would be rarely if ever used based on this data.
SGLT2 inhibitor in TAVI: DAPA TAVI
DAPA TAVI was a RCT comparing dapagliflozin after TAVI to TAVI alone. NEJM published the results of the Spanish trial.
Before I discuss the trial. The presenter at the late-breaker session at ACC, came out to Danza Kuduro a Spanish Portuguese song by Puerto Rican artist Don Omar and Portuguese-French singer Lucenzo (according to my shazam app). The song became a hit in many Latin American countries.
The audience of presumed scientists stood and began clapping to the beat. The presenter did a little dance as well. I will come back to this business after I tell you about the trial.
To be included, patients had to have a history of HF or one of the following: chronic kidney disease (CKD), diabetes (DM), left ventricular (LV) dysfunction.
The primary endpoint was CV death or worsening HF defined by an urgent visit at one year.
Patients were older—82, half female, mean EF 47%, and estimated glomerular filtration rate (eGFR) around 56 mL/min/1.73 m2. Slightly more than 70% of both arms were on a diuretic. About 600 patients were included in each group.
The primary outcome occurred in 15.0% in the dapagliflozin group and in 20.1% in the standard-care group (hazard ratio [HR], 0.72; 95% confidence interval [CI], 0.55 to 0.95; P=0.02)
As you would expect the driver of the primary endpoint was HF events. CV death and all-cause death were not significantly lower. Most of the HF events were “urgent heart failure” visits.
As usual, the authors don’t tell us total hospitalizations.
Comments
The trial was investigator initiated, and industry did not fund appear to fund it directly. That’s good. Also good is that it enrolled older patients. Few trials have patients mean age of older than 80. Also good was that half the trial were women.
The primary endpoint was positive, but I think there are serious limitations that preclude a new indication for routine use of SGLT2i. One is that the driver of the primary endpoint was a bias-prone heart failure event. It’s especially troublesome here because as the authors note, the trial was open label without a placebo. Thus, patients and caregivers knew the treatment assignment and that knowledge may have affected clinical decisions. Namely the decision to just add some oral diuretic over the phone (and not have a HF event) vs have the patient come in for a “HF visit or hospitalization.”
Second, like so many of the SGLT2i trials, the drug does not reduce CV death or all-cause death. I don’t think it can be called a disease modifying drug. Perhaps longer follow up than a year may have led to a lower death rate in the drug arm. And the short duration of the trial is also a limitation.
The authors do not present total hospitalizations. This is an important outcome because these are 82-year-old patients with multiple co-morbid conditions. HHF may represent only one cause of hospitalization and if total hospitalizations are not reduced then that would further lessen my enthusiasm. Note to HF trialists: please report all-cause hospitalizations.
Many of these patients had SGLT2i indications – low EF, DM or CKD. For those types of patients having TAVI, I would say, SGLT2i seems reasonable. But these results are not strong enough to warrant a new indication for drug.
Once again, I love trials. I laud clinicians who do trials. This one was investigator initiated, which is great, but the issue is that if we are going to use the results to start a new practice, trial procedures have to be robust and all important outcomes reported. The lack of blinding, the short follow-up and no report of total hospitalization, which precludes assessment of competing risks, weaken the ability to use this trial to change practice.
Finally, the business about the music. It was a nice song, but it is out of place in a presentation of science. I’m sorry but that’s the way I feel.
SOUL Study of Oral Semaglutide and CV outcomes in Diabetes
We know that Glucagon-like peptide-1 (GLP-1) drugs have shown benefit in patients with DM, CKD and atherosclerotic cardiovascular disease (ASCVD) with obesity. These were injectable GLP-1 drugs. The PIONEER 6 trial of oral semaglutide vs placebo in patients w DM established non-inferiority and CV safety.
The SOUL trial, presented at ACC and published in NEJM set out to establish efficacy of the oral GLP-1 semaglutide. This was a big Novo Nordisk led trial. Patients had to have DM with elevated HbA1c and at least one of a bunch of conditions, such as CAD, peripheral arterial disease (PAD) or CKD. Patients with end stage renal disease were excluded.
They were randomized 1-1 to oral semaglutide or placebo. Doses started low and were escalated to 14 mg.
The primary outcome was a three-point MACE of CV death, MI, stroke. There were many secondary outcomes.
There were 9600 pts randomized in 444 sites in 33 countries. Mean age was 66 years, and about a third were women. The mean follow up was nearly 4 years.
The Main result was positive.
A primary-outcome event occurred in 12.0% in the oral semaglutide group, vs 13.8% in the placebo group (HR, 0.86; 95% CI, 0.77 to 0.96; P=0.006),
Or 3.1 events per 100 person-years vs 3.7 events per 100 person-years.
Absolute risk reduction is about 1.8% or number needed to treat (NNT) of 55.
CV death was 6.2 and 6.6% - not statistically significant
MI was 4.1 and 5.6% a 26% reduction for GLP-1 (CI, 0.61-0.89)
Stroke was 3.0 and 3.3 – not significantly different.
Death from any cause 10.9 vs 12.0% - not significantly different
HF – 3.0 vs 3.5 – not significantly different
Major kidney events – not significantly different
Mean body weight was ≈ 3kg less in the semaglutide group.
Adverse events (mainly gastrointestinal) that led to permanent drug discontinuation occurred in 15.5% of the GLP-1 group and 11.6% of the placebo group.
The authors concluded that the drug was effective at lowering CV events.
Comments:
This was a positive trial. The primary was positive, mostly driven by lower rates of MI in the semaglutide group. But all components of the endpoint, and kidney outcomes were directly better on S.
The secondary composite kidney composite was lower with semaglutide but did not reach statistical significance, so the authors did not test components of that endpoint.
The authors spent some time explaining why the kidney outcomes were less impressive than in the FLOW trial.
One explanation could be that FLOW patients started with a much lower GFR. Another explanation could be the much lower bioavailability of oral vs injectable GLP-1. The safety profile looked fine—and similar to other GLP-1 trials.
Notable was that about 25% of this population of diabetic pts were on SGLT2i. For the purposes of translating this evidence, inclusion criteria required established heart or kidney disease. Many patients with T2D do not have these conditions.
The mean BMI of these patients was 31 and the weight loss was only 3 kg over 4 years.
I have to say though that the effect size was modest. An absolute risk reduction in nonfatal outcome of 1.8% with an NNT of 55 is decidedly modest. The p-value is persuasive and since the trial is directly consistent with other GLP-1 trials, I suspect it is real.
I don’t know enough about diabetes management to know where this will fall into the drug regimen.
At quick glance the degree of weight loss seems less than the injectable form. But patients may be more accepting of an oral form.
We shall see.
That’s it for now. I will recap more trials from ACC and EHRA next week including the big pulsed field ablation vs cryoballoon trial from Swiss authors, as well as the fluid restriction in heart failure trial.
© 2025 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Apr 04, 2025 This Week in Cardiology Podcast - Medscape - Apr 04, 2025.
Comments