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 Dec 20, 2024, John Mandrola, MD, comments on the following news and features stories: Strong listener feedback, multiple studies in atrial fibrillation (AF) ablation, a DANISH trial substudy of implanted cardioverter defibrillators (ICDs) in patients with nonischemic cardiomyopathy (NICM), and more data on eye complications of GLP-1 agonists.
First, as always, we take the next 2 weeks off for the holidays. I will see you back in the first week of January.
CCTA Feedback
I received perhaps my strongest rebuttal this week regarding my negative view of coronary computed tomography angiography (CCTA). I have published on this with Andrew Foy. In JAMA-Internal Medicine, our meta-analysis looked at trials comparing CCTA vs functional testing in patients presenting with suspected disease.
The main finding was that CCTA relative to functional testing increased diagnosis of coronary artery disease (CAD), increased angiography and percutaneous coronary intervention (PCI), and had minimal reduction of outcomes. In fact, we would have found no reduction in any outcomes if we excluded SCOT-HEART but were forced by reviewers to include it. SCOT-HEART did not belong in our meta-analysis because 85% of patients in that trial had both functional testing and CCTA.
My concern about CCTA use in the United States is that it leads to massive over diagnosis and overuse of revascularization, which does not reduce hard clinical outcomes. Hospitals are rushing to acquire CCTA machines because it is like a cash machine.
But. This appears to be an American phenomenon.
I received a strong rebuttal from Professor Bjarne Linde Nørgaard, a professor at the Aarhus University hospital in Denmark.
He admonished me, as did my mentor, William Miles, that a test or procedure is almost never the problem. The problem, instead, is its misuse by humans. I will tell you parts of what Norgard said. Keep in mind that Denmark has a public health system and doctors are paid salaries that do not depend on the number of procedures they do.
In Denmark, CTA has been the preferred test in patients with de novo symptoms suggestive of CAD over the past more than 10 years.
In Denmark, referral to CTA is based on a patient face-to-face office consultation. The CTA investigation is performed by experienced cardiologists and dedicated techs. Decision-making on post-CTA downstream management (e.g. no treatment/testing,OMT [optimal medical therapy], cath+OMT) is based on the CTA result/anatomy risk (plaque burden and morphology, stenosis severity/location) taking into account also the patient risk profile, symptoms, treatment (many not yet on OMT), and patient preference at the discretion of the CTA operating cardiologist.
Following implementation of CTA testing as a first-line test in western Denmark (which has now plateaued on approximately 10.000 scans a year) we have since 2015 seen a 33% decline in the use of invasive cath’s, and a 14% reduction in PCI´s. These findings may of course also be driven by the results from landmark trials like COURAGE, BARI-2D, ISCHEMIA, REVIVED BCIS 2 unambiguously showing the value of OMT relative to revascularization in stable CAD.
In 2023 at Aarhus University Hospital (having the largest CTA production in Western Denmark) we performed 1400 CTA´s. Despite the fact that 75% were diagnosed with CAD, including 24% with significant 1,2 or 3-vessel CAD, only 13% were referred to cath (of whom 65% were revascularized). Often, I see patients with "severe" 3-vessel or diffuse disease (high CAC scores) without LM [left main] involvement and only mild symptoms, not yet on OMT. In the majority of such cases we recommend an OMT strategy, we talk to the patient again after another 2 months, if the patient is well without side effects to OMT we stick with the OMT strategy, while in the event of continuing or worse symptoms we typically — taking into account patient preference — will refer to cath. This is a safe strategy in a cohort of stable CAD patients with an overall good prognosis.
These experiences from Denmark clearly show that the benefit (or harm) by CTA should be viewed in context of the health care organization in which it is applied. CTA is the best first line test in symptomatic patients providing important information on whether there is disease or not, on the CAD phenotype and as guidance for the revascularization procedure in the few patients needing this, that is, in the right hands and in the right organization.
In fact, I knew this, but I am so overwhelmed with the overuse of imaging and its downstream testing here in the United States, that I had forgotten what the principal investigator of DANCAVAS, Dr. Axel Diederichsen, had told me about their experience with CAC imaging as part of that study’s screening program. He said that there it was not a worry to find CAC because in Denmark this would lead to medical therapy, not a cascade of ischemia testing and treatment.
Thanks to Dr. Norgard for reminding me that there are parts of the world that practice based on evidence. Perhaps because Danes are smarter but perhaps Danes have better aligned incentives.
Listener Feedback on My Negative Take of the OPTION Trial
Briefly, the OPTION trial compared stroke and bleeding endpoints in patients who had AF ablation plus direct oral anticoagulants (DOAC) vs AF ablation plus left atrial appendage closure ( LAAC). The noninferiority (NI) trial delivered positive results (NI was met for efficacy) and superiority for reduction in bleeding in the LAAC arm, but it was one of the most flawed trials I have reported on in the last 10 years. I wrote a post before the results were publishing chronicling the flaws and how these would lead to positive results.
This week I received two messages about improving critical appraisal in trainees. One from a senior program director and the other from a young doctor who just joined an academic program.
I appreciated your take on the OPTION trial. Surely enough, we had a journal club this week from one of our trainees who touted that we have an option that is non inferior to leaving the LAA alone when ablation is offered. Sure enough, everyone on the virtual meeting call applauded the science/the presentation without inquiry, and then signed off.
My question for you is how do you suggest comfortably pushing back on the unabashed enthusiasm for these discussions? I agree with your take about non-inferiority and endpoint selection for this trial. However, it is hard to convince a large audience of impressionable trainees to 'think beyond the abstract conclusion.' Do you have suggestions here?
I have stewed about this for a few days, hoping to come up with an easy actionable answer. I reached out to others in academia.
And I don’t have a simple solution. Here is my best effort:
First, question the presenter on basic methodology. For instance, what is the effect of adding death to a composite endpoint of a NI trial, when death will not be affected by either treatment?
Second, as it is with publications, every journal club should have a section on limitations. These should be robust.
Third, the only way to affect impressionable trainees is to foster a culture of skepticism without cynicism. I would have journal clubs on classic trials, such as CAST and WHI, so trainees can learn the concept of medical reversals. During the pandemic I sat in on journal clubs at Penn State and I believe Andrew Foy successfully created a skeptical group. Then, these trainees go to other programs and continue the same culture.
This is a goal of this podcast: to give permission for skepticism.
AF Ablation
Another listener called me out for missing two ablation trials from the American Heart Association meeting. Grin. I did not really miss them; I just wasn’t very impressed.
JAMA published the PROMPT AF trial, which was a randomized comparison of pulmonary vein isolation (PVI) alone vs PVI plus optimized linear ablation aided by infusion of alcohol (ETOH) into the vein of Marshall (VOM) in patients with persistent AF.
The clinical relevance here is that patients with persistent AF have worse outcomes after ablation compared with patients with paroxysmal AF. The largest trial, called STAR-AF-2, found that, in persistent AF, PVI alone yielded similar results to PVI plus ablation of common femoral artery endarterectomy and PVI plus linear ablation.
For persistent AF, only two trials have shown slightly improved efficacy of additional ablation added to PVI alone. One trial was the COVEREGE trial of combined radiofrequency ablation and surgical ablation, and the other trial was called VENUS, with ETOH infusion into the VOM. In VENUS, both PVI and PVI plus VOM had tons of ablation outside the PVs. The group with VOM had significantly better freedom from AF (the primary endpoint) 49% vs 38%. But it was close; the P-value was 0.04.
A word on ETOH infusion into the VOM. It is doable. I have done it. But it’s labor intensive unless you do it all the time. It requires contrast, guiding catheter, PCI wires and balloons, and high quality x-ray. The VOM is variable and not always there. If you can do it, the degree of destruction along the mitral isthmus (MI) and anterior ridge of the left sided PVs is substantial. The destruction can even extend to the roof and parts of the posterior wall.
Okay, the PROMPT AF trial was a cleaner trial than VENUS. The control group got PVI alone. That is it. The active arm got ETOH infusion of the VOM to assist linear lesions that mostly included the MI, but also the left atrial roof, and cavotricuspid isthmus in the right atrium.
The primary endpoint was freedom from AF without antiarrhythmic drugs at 1 year.
They enrolled about 500 patients so, more than VENUS.
At 1 year, 71% had freedom from AF in the linear ablation group with ETOH vs 61.5% in the PVI alone group. Hazard ratio (HR) 0.73; confidence interval (CI) 0.54 to 0.99, and P-value 0.045.
Notable was that 7 patients assigned to the intervention group experienced pericarditis or pericardial effusion, compared with none in the PVI alone group.
Comments. This is a nice effort. Good on the investigators from multiple centers in China.
My interpretation is that linear ablation with ETOH into the VOM is a lot of work for very little gain. Recurrence rates of 30% vs 40% are hardly clinically meaningful. There were also 7 pericardial events with the extra ablation.
I think VOM is a nice trick that is useful for ablating the MI to eliminate peri-mitral flutter, and if you don’t practice it, it’s hard to be good at it.
So my recommendation is to learn the technique. But PROMPT AF does not convince me to adapt it for a normal workflow. Interesting also was that no subgroup, even those with more advanced LA disease, seemed to benefit more from the extra ablation. Which is interesting because PVI alone in more advanced disease performed quite well — as it did in STAR AF 2.
The other thing to note is that pulsed field ablation (PFA) is set to transform LA ablation. PFA will potentially allow us to do more durable PVI, with posterior wall isolation and or roof lines.
A final summarizing point on AF ablation, which is a variant of something I have said many times. PVI can be amazing for patients with minimal atrial disease who have focal drivers from the PVs. It’s almost curative.
But for the vast majority of patients, who come by AF by having cardiometabolic risk factors, who have atrial structural or electrical disease, or said another way, whose AF stems from scar, it is unlikely that creating more scar will be disease modifying.
The next major advance in AF will not be a better way to kill heart cells, but a better understanding of what causes AF.
AHA Also Featured a Trial From Multiple Centers in Japan
The CRRF-PeAF trial was a comparison of cryoballoon (CB) vs radiofrequency (RF) ablation in patients with persistent AF. Recall that Fire and Ice was RF vs CB in patients with paroxysmal AF and there was absolutely no difference.
CRRF-PeAF also found no difference in freedom from AF with either of the two technologies. Both arms had oodles of ablation outside of PVI so it’s hard to sort out true PVI effects.
The results are not published yet, and I can say more when published, but again, I emphasize that the answer to persistent AF is unlikely to come from one or the other ablation strategy.
SVC Isolation
Circulation Arrhythmia and Electrophysiology published a small randomized controlled trial (RCT) from Spanish centers called the CAVAC trial.
This was PVI alone vs CB PVI with superior vena cava (SVC) isolation, the idea being the SVC can be like a PV and have focal drivers.
The freedom from AF was no different but one in five patients in the SVC isolation arm ended up with phrenic nerve (PN) paralysis and sinus node disease. Fortunately, all but one PN injury was transient and no sinus node injury required pacing.
Obviously doing SVC isolation with CB is hazardous due to the phrenic nerve. But SVC isolation as an adjunct should probably be tested formally with PFA since electrical energy does not seem to harm the phrenic as much.
DANISH Trial Substudy
In talks with patients about the ICD decision, often CRT-P (with pacemaker) vs CRT-D (with defibrillator), I mention the 2016 DANISH trial, ICD vs medical therapy in 556 patients with NICM. Median LV ejection fraction (EF) 25%. After nearly 6 years of follow-up there was no difference in mortality. None. Background therapy was excellent.
One of the most controversial aspects of cardiology, in my opinion, is how little this trial has influenced care, at least here in the United States, even though there was a provocative subgroup analysis, which showed a strong heterogenous treatment effect (HTE) based on age, where younger patients benefited and older patients trended toward harm. While the age-related HTE, made sense, it is still a subgroup analysis within a nonsignificant trial.
This month, the authors of DANISH have published, in Circulation, as a research letter, a post-hoc study looking at the LVEF and the ICD in NICM.
The authors write that “there are no detailed reports on the efficacy of ICD-implantation according to LVEF in patients with HFrEF.” So they examined the effects of the primary prevention ICD according to LVEF in DANISH.
Before I tell you the results, I want to push back a bit on the notion that there are no detailed reports on the efficacy of the ICD according to LVEF. That is because both the SCD-HeFT and MADIT 2 authors have published subgroup analyses showing that most of the ICD benefit in those trials were in patients with lower EFs.
In the DANISH substudy, the authors explored three associations. Note I said associations because these were just that.
The effect of the ICD vs usual care according to continuous EF.
ICD effect according to EF in each age-category (ie < 70 or > 70).
ICD effect according to CRT use or nonuse.
First finding:
There was no significant interaction between EF examined as a continuous variable. Think of the figure with LVEF on the x-axis and HR on the y-axis and then splines. Basically the ICD effect remains flat at 1.0 no matter the EF, from 10% to 35%,
There was also no significant interaction between EF and cardiovascular (CV) death or sudden death (SCD), though there was a trend towards better ICD effect for SCD prevention (at better EFs). Which makes sense because the patients with lower EFs are more likely to die of pump failure.
Now let’s go to the age categories and LVEF. But before I say anything, remember that in the nonsignificant overall DANISH study, there was an ICD benefit in young people.
The question now is, was that modified by EF?
The first answer was no.
In patients less than 70 years, the LVEF did not significantly modify the ICD benefit for mortality, CV death, or SCD.
However, in older patients, there was a positive P-for-interaction based on LVEF. For older patients there was HR point estimate around 3.0 for the lowest EF. That is, ICD is harmful relative to usual care. But as the EF got better, the ICD turned beneficial. The weird thing was that the absolute differences 54% vs 56% were less than all other categories.
This finding makes sense, right? If you believe the main trial HTE, that is younger patients with NICM do better with an ICD, while older patients trend toward harm, you’d also expect older patients with the lowest EF to be most likely to incur harm from the device.
Final finding in patients with and without a CRT, the LVEF did not significantly modify the ICD effect.
The authors add the appropriate caveat that these findings should be considered as hypothesis-generating.
Other limitations include lack of generalizability to all patients with HFrEF and lack of central adjudication of LVEF measurements, which were investigator-reported.
In conclusion, in patients <70 years, LVEF did not significantly modify the beneficial effect of ICD-implantation. However, in patients aged >70 years, those with a higher, but not a lower, LVEF benefited from ICD-implantation.
The latter finding supports the notion that patients with a lower LVEF, especially those over 70 years, are more likely to die from “pump failure” or terminal HF and therefore less likely to derive benefit from an ICD.
Comments. DANISH is one of the most important trials in modern cardiology. It proves Rod Tung, Peter Zimetbaum, and Mark Josephson correct in their famous critical appraisal prediction paper published in 2008, where they argued that ICD benefit in NICM should be questioned.
Indeed they were correct. The problem with DANISH is not the trial but our interpretation. The trial showed not a hint of mortality benefit from the ICD. In my opinion, it should have changed guidelines. Instead it was just added to meta-analyses of 20 year old trials and the benefit remained. But this is misguided. The ICD did not provide an average benefit to patients in DANISH because background therapy was so good. It is even better now, 10 years on.
Proponents of the ICD use the data both ways. If you have a young person with an NICM, you cite the subgroup analysis, which is compelling, but it’s still a subgroup analysis from a nonsignificant trial. I believe that subgroup analyses should be used not for clinical decisions, but rather to plan future trials. If you use subgroups to make clinical decisions, then you have to accept astrological signs as a modifier of aspirin benefit in acute myocardial infarction.
Proponents of the ICD can also dismiss DANISH when they want to implant an ICD in an older patient with NICM because they can just say the meta-analyses are positive. And my patient, they’ll say, isn’t able to take such great background therapy.
So, in sum, while the findings of this association study make sense, I don’t like the authors conclusion with words like “in the < 70 years old group, LVEF did not modify the beneficial effect of the ICD.“
This study over calls its own precision. For if we argue that subgroups deserve our caution, subgroups of subgroups should deserve even more caution.
Eye Disease and GLP-1 Agonists
In July I covered a methodologically weak study published in JAMA-Ophthalmology in which the authors associated semaglutide use with a rare but serious eye condition called non-arteritic anterior ischemic optic neuropathy (NAION).
The article blew up with media coverage, but it was really a weak study marred by surveillance bias and likely collider bias.
I had just been to Denmark and visited with researchers who showed me the basics of how to use their national registry.
I sent my friend Soren Diederichsen an email and thought the Danish registry might be a good way to explore this association.
Soren emailed back with two studies now out using the Danish registry. One is weak and simply identified all persons with type 2 diabetes (T2D) and was stratified by use of nonuse of semaglutide.
Semaglutide exposure was associated with a higher incidence rate (0·228 vs. 0·093 per 1000 person-years, P < 0·001) and independently predicted a higher risk of upcoming NAION (HR 2·19, 95% confidence interval 1·54 − 3·12)
This paper made adjustments, but was a pretty weak paper and could clearly have included sicker patients getting the drug.
Anton Pottegard led a more rigorous analysis, and their paper is published as a preprint. I will link to it.
This was a bi-national (Denmark and Norway) active comparator, new-user cohort study with propensity score weighing to adjust for confounding, fixed-effect meta-analysis, and a supplementary self-controlled analysis (symmetry analysis). Let me try to translate, though it is not easy.
Their main outcome was Incidence rates of NAION among semaglutide and SGLT-2 inhibitors users, along with HRs and incidence rate differences (IRDs) with 95% confidence intervals.
Estimates from Denmark and Norway were pooled using a fixed-effects model (inverse variance weighting).
The supplementary self-controlled study is a cool thing where they look at the rate of NAION one year before and one year after semaglutide exposure. It outputs a symmetry ratio (SR).
First, they found 45,000 eligible users of semaglutide for T2D in Denmark and 17,000 in Norway.
They observed 32 cases (24 in Denmark, 8 in Norway) of NAION events among semaglutide users.
This yielded an unadjusted incidence of 2.2 per 10,000 person years among Danish semaglutide initiators. They then compared that to SGLT-2 inhibitor initiators and found an unadjusted incidence of 1.2 per 10,000. The corresponding unadjusted incidence rates in Norway were 2.90 and 0.92.
After adjustment, they found a pooled HR of 2.81 (95% CI 1.67 to 4.75) and IRD of +1.41 per 10,000 person years (95% CI +0.53 to +2.29)
Results were consistent across sensitivity and supplementary analysis, with a notably stronger association observed in a post hoc per-protocol analysis (HR 6.35; 95% CI 2.88 to 14.0).
In the supplementary self-controlled study, SRs for NAION were 1.14 (95% CI 0.55 to 2.36) in Denmark and 2.67 (95% CI 0.91 to 8.99) in Norway. Both of these were not significant because the CI include 1.
The authors concluded that these findings “support an association between use of semaglutide for type 2 diabetes and risk of NAION, with a more than two-fold increased hazard ratio. However, the absolute risk of NAION remains low among semaglutide users. Analyses of an association between semaglutide for obesity and NAION were inconclusive.”
Comments. The latter study is stronger. The comparison of semaglutide users and SGLT-2 inhibitor-users does not make it random but it does a decent job of making baseline characteristics similar because the choice of the two drugs is similar in T2D patients.
It is also possible that since GLP-1 agonist drugs hve fewer CV events, they may live longer and be exposed to other events like NAION.
Nonetheless, if the risk is real, it is very low in absolute terms and must be balanced against the reduction in major CV outcomes in patients with T2D, chronic kidney disease, or obesity.
However, notable is that millions of people will soon take these drugs, so if the risk is not limited to T2D, it could be important from a public health point of view.
What’s more, keep in mind, that such a risk could never be discovered in an RCT. This is a great use of observational data.
© 2024 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Dec 20, 2024 This Week in Cardiology Podcast - Medscape - Dec 20, 2024.
Comments