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 July 28, 2023, John Mandrola, MD comments on the following news and features stories.
Antiarrhythmic Drugs and CAD
Before I tell you about this study, let me recount a meeting I had years ago with a group of brand new advanced practice clinicians. I was tasked with teaching about the use of evidence.
Just for grins, I asked this question: If you had access to the chart of every patient in this hospital who took flecainide and every patient who took sotalol, and the electronic health record was perfect, could you tell anything about outcomes?
Nearly every person nodded yes. But of course, this isn’t true because there are baseline differences. That’s why you need randomization. I explained that, and I think they got it.
A study published in JACC: Clinical Electrophysiology from the electrophysiology (EP) group at Emory has stirred many emotions. This was a retrospective chart review. Their question centers on the use of class 1C antiarrhythmic drugs (AAD; flecainide and propafenone) in patients with coronary artery disease (CAD).
I know; just saying that sentence sounds scary, because the therapeutic fashion for the past 2 decades has been never to use these drugs in, not only patients with CAD, but also patients with any structural heart disease, however that is defined.
This topic overflows with concepts from evidence-based medicine. Even if you’ve never used these drugs or are not in cardiology, this is an incredible topic for discussing critical appraisal and evidence translation.
First you have the critical appraisal of the CAST trial – a study I cite as one of the most important studies of a generation, in all medical fields.
Then you have the critical appraisal of the Emory study.
And finally, there is the issue of applying trial data to patients.
Before I start, let me say what trials are for. Trials give us average effect sizes in the population studied, in the environment of the trial, at the time the study was done.
The patient in front of us is often different from those in trials, and that is where David Sackett’s three circles of evidence-based medicine (EBM) come in: Remember, it’s the mixture of a) clinical judgement, b) best evidence, and c) patient preferences.
Let’s start with CAST.
You probably know that CAST randomly assigned patients who were post myocardial infarction (MI) and who had premature ventricular contractions (PVCs) to a 1C AAD or placebo.
You probably also know the trial found a large increase in the risk of death or cardiac arrest.
The number needed to kill was about 29.
CAST upended the ensconced practice of suppressing PVCs after MI.
More importantly, it glaringly showed the value of randomized controlled trials (RCTs) and the danger of making causal inferences from the many observational studies that showed how PVCs are associated with higher mortality risk.
However, CAST was so scary that it led to a decades-long prohibition on use of 1C drugs in patients with CAD, or even any structural heart disease. Guideline documents reiterated this proscription. Here is a typical case that stemmed from this practice.
A 65-old-man has had his atrial fibrillation (AF) 100% suppressed with flecainide for years. He tells his family doctor he feels fatigued. He gets a nuclear stress test which comes back positive. At cath he gets a stent to the mid-circumflex. The artery is open; the left ventricular (LV) ejection fraction (EF) is normal; the QRS is normal.
But because he technically has CAD and that is considered structural heart disease, his cardiologist stops the flecainide, even though he is nothing like a post-MI patient with an EF of 25% in the 1990s.
His AF comes roaring back, and sotalol (a class III drug) is ineffective. The patient now gets AF ablation. Left atrial flutter recurs. He gets another ablation.
You see the problem. He’s now had to undergo two expensive and risky procedures for a problem that was beautifully controlled with a well-tolerated medicine. CAST is an important trial, but experts have massively over-extended its findings.Consider the following facts:
CAST enrolled patients in the late 1980s, well before the current stent technology.
Background medical therapy now is many times better.
Only one in three patients in CAST were on beta blockers (BB).
Subsequent post-hoc studies found that BB use reduced the increased risks of AAD.
The average EF in CAST was about 30%.
Those facts make it extremely difficult to apply CAST to patients treated with BB and statins, who have normal EF, and no ongoing or inducible ischemia.
In 2010, Dan Kramer and the late Mark Josephson wrote an editorial titled “Three Questions for Evidence-based Cardiac Electrophysiologists.” It’s a great piece. Their first question was just what I mentioned about applying CAST to current patients with different kinds of CAD.
Now let’s move to the Emory study. Emory is a big place, so they have lots of patients. They did a retrospective chart review from 2005 to 2021.
About 3400 patients were taking a 1C drug. The investigators then created a control arm of about 2200 patients who took sotalol or dofetilide. Obviously, this is not even close to a randomized comparison. It’s not prospective either. It’s a look back.
They excluded patients with ventricular tachycardia (VT), implantable cardioverter-defibrillators (ICDs), or non-revascularized MI.
They created three groups of CAD — none, non-obstructive, obstructive.
They then ascertained clinical outcomes, including the primary endpoint of death, VT/VF, or admission for heart failure (HF).
They did a regression (correlation) analysis looking at the effect of 1C use on event-free survival across varying degrees of CAD.
Obviously, there were huge differences in baseline characteristics, but they adjusted for baseline variables.
4.6% of patients on 1C drugs had a primary endpoint vs 9.4% on a class III drug. The ≈ 5% had a P-value of 0.001.
The rate of all-cause mortality was similar between the class 1C and III groups (1.2% vs 1.7%; P = 0.205).
After adjustment for the covariates in the model, class 1C use was independently associated with a better event-free survival compared with class III use (hazard ratio [HR]: 0.40; 95% confidence interval [CI]: 0.30-0.52; P < 0.001). Seriously. That is not a misstatement. 1C drugs look massively protective!
In patients with obstructive-CAD, patients on 1Cs did worse: (HR for the primary endpoint was 3.80; 95% CI: 1.67-8.67; P = 0.001)
But for patients with no CAD, there was no significant increase in hazard.
My friend and former Indiana University doc, Gerald Naccarelli wrote the editorial. He cited a number of similar contemporary observational studies that also found no increase in hazard with type 1C use and CAD.
Comments. Here we have a conundrum. We clearly have over-extension of the 1980 to 1990 vintage CAST trial, done in patients with sick LVs, most with scar. A person who applies evidence should not, could not, apply that data to contemporary patients with mild CAD or revascularized CAD and normal ventricles.
But that is our situation. A shining example of the weakness of expert opinion and sclerosis of therapeutic fashion.
But.
Studies like this are not the way to upend wrong-headed expert opinion. I like this group at Emory, but this study is potentially dangerous.
I mean they found a 60% reduction in serious outcomes in patients taking 1C drugs.
The chance that this represents residual confounding – ie, healthier patients got 1C drugs compared with sotalol and dofetilide — is about 99.99%.
If that is the case, why do such a study? I read the methods and it looked to me like very basic attempts at matching. I did not see attempts at propensity matching. Nothing about falsification endpoints, nothing on e-value, or use of instrument variables.
The answer to upending dumb expert opinions on 1C AAD prohibition in patients with minimal heart disease or mild CAD is not to do a flawed retrospective chart reviews. An RCT would answer the question but that is unlikely. Short of that, I wonder if we simply need new experts. Experts who think better. Experts who take care of patients more regularly. Experts who don’t do medical malpractice work. If you do med-mal work, you should not be allowed on a guideline document or expert consensus statement.
GLP-1 Agonists
Last summer, The New England Journal of Medicine (NEJM) published the results of the SURMOUNT-1 RCT of once-weekly tirzepatide for the treatment of obesity. At 1.5 years, the glucagon-like peptide-1 (GLP-1) receptor agonist induced substantial weight loss vs placebo. Go look at the main figure. Patients in the 10 to 15 mg dosing arm had more than 20% loss in weight. Think about it; 20% in less than 2 years, a 250 lb person is now 200 lbs.
Last month, the Lancet published the SURMOUNT-2 results. Again, at 1.5 years, patients with type 2 diabetes and obesity sustained substantial weight loss. The delta here (about 10%) was less than in SURMOUNT -1. The authors suggest weight loss in patients with type 2 diabetes is more challenging.
Well, the maker of the drug, Lilly, has announced topline results of the SURMOUNT-3 and 4 studies.
To get the flavor of the results let me read the lede of journalist Mitchel Zoler’s news story:
“ The tirzepatide weight-loss juggernaut barreled forward with a report on July 27 of positive top-line results from a pair of pivotal trials in adults with obesity or overweight but without diabetes.”
This was from the press release:
“Participants in SURMOUNT-3, after 12 weeks of intensive lifestyle intervention, achieved an additional 21.1% mean weight loss with tirzepatide for a total mean weight loss of 26.6% from study entry over 84 weeks.
“Participants in SURMOUNT-4 achieved 21.1% weight loss during a 36-week tirzepatide lead-in period and an additional 6.7% weight loss during a 52-week continued treatment period, for a total mean weight loss of 26.0% over 88 weeks.”
Both studies will be presented at future meetings.
Tirzepatide is available on the market as it received US Food and Drug Administration (FDA) approval in May 2022 for the indication of improving glycemic control in people with type 2 diabetes.
The twist for SURMOUNT-3 was that adults with obesity had to lose 5% of body weight in the run-in period with lifestyle interventions. Then they could enter the trial.
SURMOUNT-4 was also pretty nifty:
The first part was tirzepatide vs placebo. This led to a 21% body weight loss. Then patients were randomly assigned to continued tirzepatide vs placebo.
At the end of the 1-year randomized phase, those who continued tirzepatide had an average additional weight loss of 6.7%, while those who switched to placebo had an average 14.8% weight gain during the 52-week phase, producing a placebo-adjusted weight loss with tirzepatide of 21.4% for this phase.
In a related story, Mitchel Zoler also reports that semaglutide use has surged in US adults who have type 2 diabetes.
An abstract presented at the European Association for the Study of Diabetes reported that semaglutide was the now the top incretin-based drug class.
The main finding was that in January to March 2022, 57% of US adults with type 2 diabetes who were prescribed an incretin-based treatment were taking a GLP-1 agonist and 39% were taking a DPP-4 inhibitor.
These usage rates sharply diverged from 4 years earlier in January to March 2018 when two-thirds were taking a DPP-4 inhibitor and one-third were taking a GLP-1 agonist.
GLP-1 agonists now represent nearly one-third of all meds prescribed to patients with type 2 diabetes.
We also learned that semaglutide has the early lead over tirzepatide.
Zoler tells us about the tirzepatide SURPASS-CVOT (cardiovascular outcomes trial) which has enrolled more than 13,000 patients with type 2 diabetes and measures CV outcomes. Results are due next year.
My friends, SURPASS-CVOT is going to be super-important, because if tirzepatide had both huge weight loss and clinically significant major adverse cardiac events (MACE) reductions, that would be something. Plus, if you can show 20% weight loss leads to fewer MACE, then that lends credence to obesity as a serious disease.
Comments. First, I don’t like pre-presentation announcements. This may be couched in regulatory rules – for investors. But it’s a convenient marketing technique because it presets the conversation. Kahneman/Tversky call this anchoring. I understand it to be that you start the interpretation of the studies in a positive light.
Second, look what FDA can get companies to do. I don’t understand why this isn’t the norm with device approval procedures. When tirzepatide gains approval we will have many thousands of people who have been randomized.
Third, these drugs present us with a real thinking exercise, don’t they? If you see obesity as a medical disease, and weight loss as the ultimate goal, these drugs get the job done. In the short run, they are surely lower risk than obesity surgery. In the absence of more outcomes trials, weight loss remains a surrogate outcome.
But.
If you see obesity as more of a societal/norm issue — to wit: the differences in physical appearance in European vs American airports is striking — then the notion of long-term use of injections makes you wonder whether our norms should be adjusted. And perhaps these norms should change in schools. Childhood obesity saddens me.
What’s more, while we will have RCTs with these drugs for months to a couple of years, there will surely be the issue of unknown unknowns with long term use, because it looks like stopping the drug brings weight gain.
The Subcutaneous ICD
Age has a way of changing your view on things. Implantable devices is one of those things.
When I was young, during the era of irrational exuberance for ICDs, after MADIT 2 and SCD-HeFT, when key opinion leaders were saying we didn’t have enough EP docs to implant all those “needed” ICDs, I was at least somewhat swept up in the excitement.
I knew there could be long term downsides. But the present took precedence over the future.
Now that I am older and have practiced long enough to see patients who have dual coil ICD leads that have gone bad or been abandoned and then got infected at generator change, I have grown more concerned about long-term complications from transvenous leads.
The subcutaneous ICD (S-ICD) offers a tantalizingly plausible solution. You can still get the backup ICD, but everything is subcutaneous. The lead and the ICD.
But there are serious downsides. Not having a lead inside the heart results in poor sensing, which leads to higher rates of double counting and inappropriate shocks. No ability to pace. And of course, the ability to effectively defibrillate the heart can become an issue.
This week in the Journal of the American College of Cardiology, a large group of authors published a post approval study of the S-ICD system.
86 centers in the United States enrolled patients in this FDA-mandated single arm study that was intended to be a real-world registry.
The primary endpoint was overall shock efficacy in converting VT or VF through 5 years. The primary safety endpoint was complications from the system.
More than 1600 patients were enrolled and median follow-up was 4.2 years.
The abstract says they met the primary efficacy and safety goals. Shock efficacy was 98.4%. And it remained stable over the years.
S-ICD related complications were 6.6%, electrode related complications were less than 1%. Only 1.6% had their device replaced by a transvenous system.
The authors concluded:
“ These results demonstrate the 5-year S-ICD safety and efficacy for a large, diverse cohort of S-ICD recipients.”
Let’s dig a little deeper since this study earned a supportive editorial that will likely be used in marketing material from the company.
The inappropriate shock (IAS) rate was 6.7% at 1 year and 15.8% at 5 years. That is quite high, and it was due to cardiac oversensing. DANISH, an ICD trial that recruited patients about a decade ago, had an IAS of 5.9%. Most of these were due to AF rather than sensing problems. S-ICDs are often implanted in younger patients. In DANISH the IAS due to oversensing was less than 1%.
The other problem with this study is that it was voluntary. I could find no discussion about mandatory enrollment. You could be an implanting center and decide who gets in the study. This is a massive difference from the mandatory Society of Thoracic Surgeons registry of transvenous aortic valve replacement. If you are an S-ICD proponent, you might not enroll patients you had the most difficulty with.
Voluntary registry studies are super problematic due to this systemic bias.
Another problem with this study is the huge number of dropouts; 1643 began the study and 665 finished.
I subtracted from the original number the 288 who died and calculated that 50% of the cohort exited the study. Yes, of this voluntary study, 50% were not followed.
The most common reasons for exiting the study before the 5-year follow-up were an inability to be followed at the study center and loss to follow up.
Comments. I’ve implanted a handful of these devices. They have plausible reasons for use. In young people who you believe benefit from VF backup. But the evidentiary support for this device is extremely weak. We can make no reliable conclusions from this study regarding overall efficacy. How can you if more than 50% drop out of a voluntary study.
We do know that people in this study, from expert centers, and likely of S-ICD proponents, had a 1 in 6 chance of receiving a high voltage shock while wide awake.
Another problem with the evidentiary support for the S-ICD is that there were adjudication issues in the PRAETORIAN trial of transvenous (TV) -ICD vs S-ICD. This study, published in NEJM in 2020, was a noninferiority (NI) trial comparing the S-ICD as the new treatment and TV-ICD as the standard.
It reported and is widely cited as meeting NI in its primary composite endpoint of Device-related complications and inappropriate shocks.
But. Rita Redberg, Bogdan Enache and I have documented in a published Letter to the Editor, that the PRAETORIAN would have missed NI if the investigators had used the their original definition of appropriate shock as a shock for VT > 180 beats per minute, a definition which they used in a previous study.
Of the 83 patients in the S-ICD group who had shocks labeled as “appropriate,” 11 had shocks for VT below the cutoff of 180 beats per minute solely because of oversensing of cardiac signals. For example, ventricular tachycardia at a rate of 150 beats per minute could be sensed as 300 beats per minute.
We estimate that if approximately half of these oversensed ventricular tachycardias had been labeled as inappropriate, noninferiority would not have been met.
See what I mean? When the GLP-1 drugs come to market, we will have many-fold more data than we do for the S-ICD device.
I don’t understand why the FDA has such lenient standards for implantable devices relative to drugs. The makers of these devices have the money to fund appropriate trials.
Post-market studies are better than no post-market studies, but treatment of something so serious as VF should come with a mandatory registry, wherein every patient is enrolled.
Even better would have been a reimbursement with evidence policy. The post market study had 1600 patients. Imagine if these were randomly assigned to TV-ICD vs S-ICD. We would have doubled the number in PRAETORIAN.
© 2023 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Jul 28, 2023 This Week in Cardiology Podcast - Medscape - Jul 28, 2023.
Comments