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 January 10, 2025, John Mandrola, MD, comments on the following news and features stories. Alcohol (ETOH) and atrial fibrillation (AF), GLP-1 agonist trial representativeness, diagnosing coronary artery disease (CAD), changing stroke rates in patients with AF, and blanking periods after AF ablation.
Surgeon General Causes Hoopla Regarding ETOH Warning
While we were all on holiday, the US Surgeon General made news by warning about the risks of ETOH consumption. The specifics of the note was a “direct” link between ETOH and increased cancer risk.
Alcohol consumption is the third leading preventable cause of cancer in the United States, after tobacco and obesity, increasing risk for at least seven types of cancer. While scientific evidence for this connection has been growing over the past four decades, less than half of Americans recognize it as a risk factor for cancer. For breast cancer specifically, 16.4% of total breast cancer cases are attributable to alcohol consumption.
A robust online discussion then ensued, specifically on X. Economist and professional explainer Emily Oster tried to sort it out in 1200 words. Her post and others went to a big Lancet meta-analysis published in 2018, “Alcohol use and burden for 195 countries and territories, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016.”
I cite this paper because it is a classic example wherein combining hundreds of biased studies do not remove bias.
The authors attempt to describe the population attributable risk of ETOH. In other words, what would be the effect on mortality if ETOH was removed from the population.
Their methods were complex, including nearly 700 data sources of individual and population-level ETOH consumption, plus nearly 600 observational studies.
The authors then made correlations between ETOH intake or non-intake and 23 specific health outcomes.
The main thing to say about this paper involves the graphs. Gosh did they make impressive-looking graphs.
Figure 4, for instance, plots standard drinks per day on the X-axis and relative risk on the Y-axis. Each observational study is a point. They draw a correlation line, but it looks like noise.
The authors even list their limitations:
Time course is only a year. People change their intake patterns.
People estimate their own use; any doctor knows this is ridiculous.
People differ in other characteristics that are associated with drinking. No amount of adjustment fixes this.
Combining all these biased studies does not remove the bias. You could combine 1000 observational studies. It doesn’t matter. In fact, that is perhaps the most important lesson I have learned about observational studies: the number of individuals does not fix bias. Unlike randomized controlled trials (RCTs), where large numbers of events help reduce uncertainty, in biased observational studies, big numbers do nothing to remove bias.
As far as ETOH and health go, observational studies can’t really sort out general health correlations.
We can probably say two things: don’t start drinking ETOH for health; the French paradox is also an exercise in confounding. Second: avoid doses of ETOH that lead to drunkenness.
That said I don’t agree with the Surgeon General’s warning. The observational studies are not definitive enough for general health warnings.
One thing we can say, however, is that we have good data on ETOH and AF.
Bradford Hill would be happy because we have three lines of data supporting a causal role of ETOH in AF. There are population-level data showing associations between AF and ETOH intake, mechanistic data from the UCSF group showing ETOH has pro-arrhythmic effects in the atrium, and Alex Voskobonick’s RCT showing that patients randomly assigned to ETOH abstinence had less AF than those who did not abstain.
If patients have AF, I tell them about the data on ETOH. I state that I am not a preacher, just a teacher and evidence-based advisor. The data on ETOH and AF is exemplary and patients ought to know about it.
GLP-1 Agonists for Obesity : Trial vs Real World
JAMA-Internal Medicine published a neat research letter from Drs. Bessette and Anderson from the University of Pittsburgh.
Their question was, how generalizable are the clinical trials of GLP-1 drugs when used for weight loss in non-diabetic patients?
I like the concept because applying clinical trials is one of the toughest jobs for the modern clinician. In fact, I find the misapplication of trial data to patients not in trials to be one of the most common mistakes in evidence-based medicine.
It turns out that the weight loss trials have numerous exclusion criteria such as chronic medical and psychiatric conditions.
They took a representative sample from a National Health and Nutrition Examination Survey (NHANES) sample of about 9000 patients with obesity.
They then applied inclusion/exclusion criteria from the trials. For each medication, they calculated the proportion of individuals meeting the US Food and Drug Administration (FDA) label indication who also met trial exclusion criteria.
Of the approximately 9000 non-diabetic patients with obesity in their sample about 90% met FDA label criteria.
But of those “eligible,” approximately one in three patients also met exclusion criteria. The proportion meeting exclusion criteria was higher in adults older than 60 years.
Major depressive disorder, malignant neoplasms, liver disease (tirzepatide only), and uncontrolled hypertension were the most common exclusion criteria.
Comments. I cover this paper for two reasons. One is the specific nature of GLP-1 prescribing, which many of us will soon be doing because obesity surely is (or will become) a cardiac disease.
But the main reason I cover it is that trial generalizability is a core concept in evidence translation.
It’s important to think not only about who was excluded from trials, but also who was included.
For example, we are frequently asked to consider implanted cardioverter-defibrillators (ICDs) in patients with low ejection fractions (EFs) and heart failure (HF), who also have advanced kidney disease. Advanced severe chronic kidney disease (CKD) was an exclusion criterion for the clinical trials. What’s more, subgroup analyses find that any degree of CKD associates with no ICD benefit, likely because of competing risks and increased treatment-related harm. Therefore, I think it is a mistake to apply trial-level evidence for ICD benefit to patients with advanced CKD.
But it is also true that not all trial exclusion criteria are created equally.
For instance, in the ICD trials, patients were excluded based on timing — 90 days from revascularization or 40 days post myocardial infarction (MI). However, we now have at least some evidence that some patients with HF and low EF who received an ICD within this time frame can still benefit from the ICD. The Heart Rhythm Society, American College of Cardiology, and American Heart Association published a document outlining ICD implantation in patients not well represented in clinical trials. For instance, consider a patient who previously qualified for an ICD based on low EF who gets a percutaneous coronary intervention of a distal circumflex artery, which will not change the EF. This patient meets exclusion criteria but likely gets the same benefit.
Left atrial appendage occlusion (LAAO) is another example. The most common indication for percutaneous LAAO are for patients with contra-indications to oral anticoagulants (OACs). But these were patients excluded from the seminal clinical trials. To me, this is mind-boggling, because the seminal trials were hardly favorable, and now hundreds of thousands of patients receive this device in the absence of RCT evidence. The medical community has embraced a procedure for patients who were specifically excluded from seminal trials.
The GLP-1 paper describes large numbers of patients with exclusion criteria from clinical trials. The authors write:
Until there is evidence from high-quality postmarketing studies, the FDA should consider updating labeling to advise caution on generalizing the safety and effectiveness of GLP-1 and GLP-1/GIP to populations excluded from pivotal trials.
I hope my examples show how tricky this statement is. On the one hand, it is true that patients with exclusion criteria from clinical trials may not benefit from the therapy. But on the other hand, having an exclusion does not necessarily equate to no benefit. In the case of GLP-1 agonists, payers will likely not allow payment for these patients, but clinicians have a tougher call.
This is where clinical judgement comes in. Thank goodness there is still a role for judgement. I can’t tell you the right answer, but the message is to look at both exclusion criteria and Table 1 patient characteristics to translate evidence.
Stress Testing and CAD Severity
ISCHEMIA trial authors have published an interesting study relating severity of ischemia and CAD anatomy.
This is a bit of a throwback study from the 1990s when I was at Indiana University and we learned the following rule of thumb: When a patient had a positive stress test for ischemia — back then it was usually stress ECG — the cath would likely show CAD. But there was a small chance of severe left main or three vessel disease and similarly small but not tiny chance of normal coronaries. In other words, the correlation was real but not super strong. I seem to remember this rule: 10% left main, 80% regular CAD, 10% normal.
Indeed, some things in cardiology don’t change.
Circulation Interventions has published an observational analysis of ISCHEMIA trial data. Recall that ISCHEMIA was a huge RCT that tested an early invasive vs conservative medical therapy approach to patients with positive stress tests — moderate to severe ischemia. The kind of patients who, in our shop, go directly from stress lab to cath lab.
After 3+ years, there was no significant difference in major adverse cardiac events between the two strategies, despite the fact that slightly more than two-thirds of patients in the conservative arm never got a coronary angiogram. Imagine. No cath after finding moderate to severe ischemia. It’s heresy. Kidding. It’s not heresy, it’s evidence.
The one (monster) caveat in ISCHEMIA was that patients had a blinded cardiac computed tomography angiography (CCTA) and those with left main or no CAD were not randomized. I call it a monster caveat because, in real life, the thing you worry about with ischemic stress tests is, exactly, missing a left main. If that happens, you may find yourself in litigation.
The substudy of note involved about 3600 patients with interpretable stress tests and CCTA.
The results were as they were in the 1990s. Ischemia severity was correlated with CAD extent, but the correlation was weak.
The correlation coefficient was 0.27 overall, slightly better with nuclear (r2 = 0.4), terrible with stress echo (r2 =0.15) and plain stress ECG r2 = 0.18). Even extent of scar on nuclear imaging had a weak correlation with CAD extent.
I would note also that this is probably the best case scenario because it involves an important trial and motivated centers.
Comments. You might wonder why I mention a study that confirms findings from 40 years ago. Well, I mention it because a) it’s good to confirm lessons because they don’t always stay the same — beta-blockers after MI; ICDs in nonischemic cardiomyopathy.
The second reason I comment on this paper is that proponents of CCTA use it to favor anatomic imaging over functional testing for the diagnosis of CAD. They have a point. In both ISCHEMIA and COURAGE, anatomic severity, not degree of ischemia, was associated with the risk of death or MI (though both trials enrolled only patients with ischemia on stress testing so there may be a bit of collider bias).
So, once again, I struggle with the issue of diagnosing coronary disease with CCTA or functional stress testing — the CCTA test itself vs its dodgy interpretation.
In my zip code, I love the functional stress test because it is less likely to lead to downstream testing. Something our JAMA-Internal Medicine meta-analysis of RCTs of functional testing vs CCTA has confirmed.
What is more, there is a decent chance that if you reproduce symptoms on a treadmill and it is ischemia, the CAD is likely causal not incidental.
But it is also true that if you are after a diagnosis — -does this patient have CAD or not, and what is the risk — CCTA is probably better.
In Northern Europe, a doctor holds to the evidence and treats atherosclerosis medically and with lifestyle. In many parts of the US, CAD on CCTA leads to coronary angio so as to fix the CAD. Maybe it’s my experience as an AF doc, but so often, I see incidental CAD treated.
I have come to accept that the reaction to CCTA is a human problem not an imaging problem. I think it was Neils Bohr who famously said science progresses one funeral at a time. I think it’s the same with treating stable CAD in many areas of the United States — it will only progress to Northern European standards with funerals.
One last thing regarding stable CAD. When I visit our cath labs (mostly as a curious journalist) and I see a nasty right coronary artery or left anterior descending lesion, and the doctor says it’s asymptomatic, my brain plays two movies. One, it’s just an angiogram. Treat it medically, PCI if medical therapy fails. The other, crap that looks terrible. I will take the stent.
I mention this because, despite reading COURAGE, BARI 2D, REVISED BCIS, and ISCHEMIA, it’s hard to dissociate empiricism with the emotion of seeing a gnarly angiogram.
Changing Stroke Rates With AF
JACC-Electrophysiology has published an important paper from Finnish authors who describe changing ischemic stroke rates in patients with AF.
One of the revelations from recent subclinical AF (SCAF) studies, ARTESIA and NOAH, was the extremely low rates of ischemic stroke — approximately 1% per year. This low rate — despite high CHADVASC scores — was the likely reason for direct OAC drugs failing to provide a net benefit over placebo or aspirin.
Low stroke rates in SCAF were explained by saying that SCAF is likely different from old-fashioned clinical AF, where a patient must have AF symptoms that then require an ECG.
But what if the decline in ischemic stroke rates has other causes? Perhaps, AF-related ischemic stroke is just like sudden arrhythmic death, decreasing over time in patients with HF. We call these questions temporal trends. Observational registry-based studies are an excellent way to study these questions.
Led by Dr Konsta Teppo, a group of Finnish researchers used the national registry to study temporal trends in ischemic stroke rates with incident AF before oral anticoagulation.
Pause there and think about how important this question is. Every day in AF clinic we have to decide whether the base rate of ischemic stroke is high enough to warrant OACs, relative to bleeding risk.
Of course, we have the overly simplistic formulas from CHADSVASC and now CHADSVA scores, such that a score of 2 =.2.2% yearly risk and 3 is 3.3%. But I would encourage everyone to read Gene Quinn and colleagues’ study in Circulation 2017 where they showed massive variation in ischemic stroke across cohorts of patients with AF. To me, Quinn’s is a game-changing paper because it infuses us with huge uncertainty about untreated ischemic stroke risk in patients with AF
I gather that Finnish registry is similar to the DANISH registry and allows study of patients with new AF who have yet to receive a prescription for an OAC.
The authors had about 40,000 patients in three groups; AF without OAC in 2007 to 2010, AF without OAC in 2011 to 2014, and AF without OAC in 2015 to 2018.
They used a 28-day quarantine period after the initial diagnosis, to avoid finding AF with other serious illnesses that could impact stroke risk.
They then truncated follow-up at 2 years to avoid finding progression of risk factors.
Follow-up stopped after incident diagnosis of AF for ischemic stroke, death, or initiation of OAC (as they were interested in untreated risk).
The main findings looking at the three groups over time:
Age and risk factors increased.
But overall ischemic stroke rates decreased by 25%.
This was driven mostly by a 32% decrease in ischemic stroke in women vs a 7% decrease in ischemic stroke in men.
When looking at ischemic stroke rates at various CHADSVA levels, the rates of a score of 1 remained low and did not change.
Most of the decrease in ischemic stroke rates occurred in patients with higher CHADSVA levels. For instance, In patients with a CHADSVA score of 3, the overall ischemic stroke rate decreased by 26%, driven by a 39% decrease in women, whereas in men no change was observed.
When they looked at age, most of the ischemic stroke events occurred in older patients and in those older than 80 years, they observed a 29% decrease in ischemic stroke, more pronounced in women (33% decrease) vs men (15%) decrease.
Comments. This is a super nice study. First, temporal trends is a great use of observational studies. National registries are indeed national treasures for they allow studies like this.
Second, knowledge of baseline risk is at the core of medical decision making. We don’t use anticoagulants in patients without risk factors not because their stroke risk is zero, but because the stroke rate is so low that any tiny absolute risk reduction could be outweighed by a) the disutility of taking a pill every day and b) increase in bleeding.
The same concept holds in ICD decisions. We require a low EF for ICD implantation because patients with higher EFs have lower base rates of ventricular fibrillation. The ICD still works in these patients but the absolute risk reduction is small and outweighed by ICD harms.
Right now, in AF care, we delude ourselves into thinking we know the base rate of ischemic stroke in patients with AF. This Finnish study shows that the base rate of ischemic stroke is falling, despite an aging cohort who has more risk factors. What’s more, in this study, patients with the highest stroke rates had the most decrease in ischemic stroke incidence.
This data doesn’t tell us when to treat and when not to treat, for that is an individual decision based on a mixture of evidence, judgement, and patient values. But this data does help explain why ischemic stroke rates were so low in ARTESIA and NOAH. Perhaps it is simply that AF is a less risky condition now than it was 20 years ago. Whether this is due to better ways to detect AF when we diagnose it in its early phases or some other factor, it doesn’t really matter.
What matters is that growing evidence suggests that ischemic stroke rates in AF are falling. That’s important to know. It should make us humble about base rates of untreated ischemic stroke, it might push us to consider changing treatment thresholds, and could allay some of our fear when patients decline OAC.
Finally, and this is directed at LAAO enthusiasts, in this registry, the most recent incidence of ischemic stroke in a CHADSVA 5 patient was 86 per 1000 patient-years. If you change that to 100-patient years, that is about 8% per year.
While that seems high, there is a 92% chance that nothing happens.
And this ia why I have long believed that the proper control arm in a contemporary trial of patients with AVERROES-like patients (AF patients who cannot or don’t want to take OAC) should be LAAO vs no OAC (no therapy). Those who get no OAC and no device may have an elevated stroke risk, but they incur no bleeding harm from drugs or procedures. Given the falling rates of ischemic stroke, the do-nothing-risky arm would be a tough comparator.
Patients’ Lives Don’t Pause for Blanking Periods
One thing that has always bothered me about AF ablation studies is this blanking period after the procedure where inflammation from the ablation can cause AF. I usually tell patients it’s about 4 weeks. But many of the trials use 3 months.
Three months is quite a long time if you are the patient having AF.
This is the topic of a short paper I wrote with Mohammed Ruzieh and Andrew Foy in American Heart Journal Plus-Cardiology Research and Practice.
We offered four reasons why AF ablation trials should count all events and not exclude the 90 days after a procedure.
Reason 1: Patients with arrhythmias in the blanking period are often symptomatic. Some require cardioversion. If a trial measures mortality, non-lethal things like cardioversion don’t matter that much. But AF ablation trials are measuring symptomatic events. So symptomatic events are symptomatic events, whenever they occur.
Reason 2: Many AF ablation trials now measure AF burden. In CABANA and EARLY AF, the ablation arm had substantially lower AF burden vs the antiarrhythmic drug arm. However, since roughly half of ablation patients have AF in the blanking period, adding those would reduce the difference in AF burden. In other words, if you are measuring AF burden, excluding a huge chunk of time in one group biases in favor of ablation.
Reason 3: Few other trials in cardiology have blanking periods after procedures. Consider the ISCHEMIA trial, which I mentioned earlier. In this trial, all causes of MI were counted, regardless of whether they were directly related to the revascularization procedure or not. Had revascularization-related infarctions not been counted, the early invasive arm would have demonstrated a significant reduction in myocardial infarction (Hazard ratio 0.67, 95% confidence interval 0.53–0.83). Since all MIs were included in the analysis, there was no significant difference between the treatment arms.
Reason 4: The inclusion of a blanking period in an RCT of AF ablation introduces complexities in interpreting the results and poses challenges for shared decision-making because patients are concerned about all symptomatic events.
Let me know what you think.
© 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: Jan 10, 2025 This Week in Cardiology Podcast - Medscape - Jan 10, 2025.
Comments