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 9, 2026, John Mandrola, MD, comments on the following topics: the limits of knowing coronary artery disease anatomy, fish oil and AF risk, a new drug for PSVT, and maybe I was wrong about a drug for AF conversion (the RAFF4 trial).
Welcome back all. I hope you had a nice holiday. It’s warm and rainy here in Kentucky and as it goes in the winter here, we go from warm days to bitter cold in a matter of hours. This week was nice, but next week we have proper winter.
Prediction of CAD Is Hard — Even if You Know the Anatomy
On the last podcast of 2025, I covered a single-center study looking back at 465 patients who had had a first myocardial infarction (MI) to determine how well risk scores predicted the event. The main take-home message was that it was pretty hard. Most patients who had had an MI were from low-risk categories.
The authors sort of called it the “MI paradox,” wherein risk scores fall short. They wrote:
Even though low-risk individuals have a small percentage chance of having an MI, there are so many people in the low-risk category that they contribute a large absolute number of heart attacks. As the authors note, "the large denominator of low-risk, asymptomatic individuals means that even a small percentage of events in this group results in a considerable absolute number of MIs.
You would think knowledge of coronary anatomy, such as the numbers of diseased vessels or the presence of noncalcified plaque, would help predict first MI.
Yet a study of more than 30,000 individuals from 6 hospitals in Sweden who had coronary computed tomography angiography (CCTA) images also reveal the difficulty in predicting first coronary events. JAMA published the study.
The cohort of patients was from the SCAPIS study — Swedish Cardiopulmonary Bioimage Study. This was a general population-based prospective study designed to extensively characterize more than 30,000 individuals to obtain information that might be used to improve prevention strategies for cardiovascular disease.
For this study the authors basically compared two groups. Those with no coronary event (N = 24,791) vs those with a first coronary event (N = 304) patients. All had had CCTA.
The first metric was discrimination between those with and without an event.
Using the C-statistic, that is, a test that distinguishes between people who will have an event versus those who won't — it's like a report card grade where 0.5 is random guessing (flipping a coin) and 1.0 is perfect prediction.
The C-statistic for the model using only the pooled cohort equation (PCE) was 0.73 (95% CI, 0.700-0.756) and increased to 0.76 (95% CI, 0.736-0.792) for the PCE in combination with the coronary artery calcium score (CACS) and to 0.78 (95% CI, 0.750-0.807) for the PCE in combination with CCTA data.
These may be statistically significant, but that is not much different in clinical terms. What’s more, when the authors broke down the c-stat for subgroups according to PCE-defined low, moderate and high risk, improvement in discrimination with CCTA was statistically positive in the low-risk group but not in the moderate or high-risk groups.
What about reclassification? The paper reports the NRI or net reclassification index, which is a fraught measure, in my humble opinion. Here is what Claude says about NRI:
The NRI is a statistical measure used to evaluate whether a new predictive model improves risk classification compared to an existing model by quantifying the proportion of individuals correctly reclassified into higher or lower risk categories.
Basically: the NRI here of 13.3% sounds good, but NRI is a simple sum of relative improvement in reclassification among those with events who had net correct upward reclassification over those without events who had net incorrect upward reclassification
Yet the NRI belies the absolute number of people with events vs without events. Recall that my first description of the two groups revealed one group of nonevent people had 25,000 and the other group had 300.
So, for people with events (~300) the CCTA data correctly reclassified up 44 people (16%) and incorrectly reclassified down 4 people (1.4%) and they get a net reclassification correct of about14%.
For people without events (more than 10,000), CCTA incorrectly reclassified up 236 patients (or 2.3%) and correctly reclassified down 66 people (or 0.6%), and the net correct therefore is -1.6%
Now, even though 14% plus a -1.6% ends up being a positive NRI of around 12%-13%, the absolute numbers reveal that when you look at the total people reclassified to higher risk, it’s not good:
-
Correctly reclassified (had events): 44
-
Incorrectly reclassified (no events): 236
-
TOTAL = 280 people moved to higher risk
But of these 280 people, only 44 (or 16%) were correctly reclassified vs 236 (or 84%) were incorrectly reclassified.
So translating that roughly to clinical decisions, is that if you get a CCTA and are moved to a higher risk category (perhaps because you have disease), you have about a 1 in 6 chance (16%) this is correct and a 5 in 6 chance (84%) this is incorrect.
And so the authors conclude, and let me paraphrase and partially quote you their first paragraph in the discussion:
In a group without established CAD, we found that addition of data from CCTA to a model based on the PCE and the coronary artery calcium score (CACS) improved discrimination modestly for coronary events during 8 years of follow-up.
These results show that CCTA data have a value beyond risk factors and the CACS in identifying individuals at high risk of coronary events.
But, however, the observed absolute risk in the population was low, and very few individuals had a risk level above the currently recommended thresholds for preventive intervention. Therefore, the clinical value of adding CCTA to improve risk prediction needs to be further evaluated in populations at higher risk, and in parallel with a health economic evaluation.
Everything they write is technically true, but I think could be restated in a more sobering tone.
What I take from this data — as a clinician or as a middle-aged person who could have an MI — is that CCTA improves the c-statistic by a meaningless amount. I mean, if a test on you has a c-stat of 0.73 or 0.78, that is meaningless.
And, as for reclassification with CCTA, if you find disease, you would think this helps reclassify. But it’s wrong way more than it is correct — 5 out of 6 vs 1 out of 6.
The authors do consider the downsides of CCTA — the radiation, the potential risks of contrast, and downstream overtreatment from revascularization of stable disease.
I would also add that perhaps the Swedish environment makes it harder for CCTA. That is, in this study of 8 years, only 1.2% of peple had an event. But still…incorrect reclassification occurring in 1.6% of 10,000 people is a much bigger number (170) vs correct reclassification in 14.2% of 300 (or 40 people).
I doubt this would change much if the event rate was, say, 3% or 5%.
Maybe I am a total luddite when it comes to prediction and anatomy. I think we know how to prevent an MI: don’t smoke, eat well, exercise, and if a simple pooled cohort equation predicts high risk and you want to avoid an event, take a statin. Adding scans doesn’t help but it sure helps those who make the machines and bill for the data — as well as the doctors and hospitals who do the follow-up tests.
Fish Oil and Atrial Fibrillation (and as a bonus we learn again about analytic flexibility)
Medicine is full of mysteries. One is that recent RCTs of fish oil used as reducers of CV risk have found (surprisingly) an increase in risk of atrial fibrillation (AF). A meta-analysis of RCTs (first author Jia) found a clear increase in AF risk, most notably a graded association based on dose. This is powerful causal data in my opinion because randomization balances known and unknown potential confounders.
Adding to the causation story is yet another trial — not covered in the Jia meta-analysis — called RESPECT EPA, which was a Japanese RCT of EPA (1800 mg of icosapent ethyl administered daily) vs control (no placebo) in about 4000 patients at 95 sites in Japan. The trial found a 21% reduction in a major adverse cardiovascular events (MACE) endpoint, but it barely missed statistical significance. But new-onset atrial fibrillation was significantly higher in the EPA group (3.1% vs 1.6%; P =.017).
Adding further to the strength of this observation is an observational analysis from the UK Biobank published in the European Journal of Preventive Cardiology in 2022 (first author Zhang) which did find that reported fish oil use was associated with a 10% higher AF risk.
But...But...
The Journal of the American Heart Association has published a provocative observational study from the UK Biobank correlating omega-3 levels and reported fish oil supplement with AF risk in a very large sample of adults in the UK.
Among participants without prevalent AF, a random sample of 261,108 had data on plasma omega‐3 levels and 466,169 reported about fish oil supplement use.
To make the story consistent, you’d want observational data to support the increased risk seen in the RCTs.
But that is not what the Kansas City research team found. In their UK biobank sample, omega-3 levels were inversely associated with AF risk (HR per IQR levels, 0.89). And, after adjustment, they found no increase in AF risk with fish oil supplementation.
You might be asking, wait, you just told me that a UK biobank observational study found a positive association with fish oil supplement use and AF, but a different group found no association. What gives?
What gives is the crucially important issue of analytic flexibility — namely that in the Kansas City study of UK biobank, they adjusted for age as a continuous variable while the earlier study by Zhang et al adjusted for age as dichotomous variable.
The Kansas City authors argue in JAHA that their adjustment technique is better. But I (and likely you) have no idea about adjustment techniques in observational research. The point though is that if association data is this flexible, it tells me that it’s weak evidence.
I therefore continue to weigh the RCT observations — that fish oil supplementation increases AF risk — more strongly.
Your comeback might be…John, how do you explain this mechanistically? How does a fish pill increase AF risk? My answer is that I don’t know, but it doesn’t matter. If enough randomized comparisons show the same thing, we have causation. For example, we accept that SGLT2 inhibitors improve outcomes in some patients, but we aren’t sure how they work.
The final argument against my RCT-bolstered belief that fish oil increases AF risk comes from one of the strongest thinkers in all of cardiology — professor Stanley Nattel, who wrote, along with Dr Michelle Samuel, that the RCT findings may be biased because most of the trials did not adjust for the competing risks of death. I will link to their argument, but basically, they observe that in most of the fish oil RCTs, the risk of death is many times greater than risk of AF, and if fish oil had a modest effect on lowering the rate of death, the finding of increased AF may be biased.
My problem with that argument is that none of the fish oil trials were powered to assess the risk of death — they all had multicomponent MACE endpoints. And I find it unlikely that fish oil substantially affects all-cause mortality. So adjustment is unlikely to change the AF signal.
But their point is well-taken and highlights the complexity of evidence translation.
In sum, I don’t doubt that observational studies of reported fish oil supplement use or omega-3 levels could identify a group of people who do other things to avoid AF. The RCTs have a clear signal of increased AF risk. There is a dose-dependent response. And I don’t think there is any substantial CV benefit to fish pills based on the negative trials and problems with the positive trials (bad control group in the REDUCE-IT trial).
So I tell people who come to me for help with their AF to eat fish, not fish pills. It may not help, but there seems little cost in stopping fish oil supplements to see what happens.
A New Drug for Paroxysmal Supraventricular Tachycardia
While we were on holiday, the FDA approved etripamil, which rhymes with verapamil and is — get this — a nasal spray to stop paroxysmal supraventricular tachycardia (PSVT). It’s a rapid acting calcium channel blocker, like verapamil but up the nose.
The supporting RCT was called RAPID and was published in Lancet back in 2023.
SVT conversion rates by 30 min were 64% (63/99) with etripamil and 31% (26/85) with placebo (hazard ratio 2.62; 95% CI, 1.66–4.15; P < .0001). Median time to conversion was 17.2 min (95% CI, 13.4–26.5) with the etripamil regimen versus 53.5 min (95% CI, 38.7–87.3) with placebo.
There are local side effects in the nose but no major adverse effects.
The drug will be called Cardamyst. I am not sure how popular it will be. Though I think it may have a role depending on how expensive it is.
Just a minute to opine on PSVT treatment. There is a temptation in some of my colleagues to just say PSVT is curable; patients should just have an ablation and be done with it. I think this is a bad take.
The older I get the more nervous I get ablating PSVT. Why? Because these are often young people. And because most of the time it is atrioventricular nodal reentrant tachycardia (AVNRT), which is usually easy to ablate, but occasionally harrowing due to the close proximity (and variable anatomy) to the AV node. A mistake in this area can cause AV block and lifelong dependence on a pacemaker.
Yes, I know, people will say: if you are careful and experienced and perfect, you should be able to avoid iatrogenic AV block. But I will tell you this is like a Black Swan event. And I will tell you that everybody is human. It’s super rare but catastrophic when it occurs.
So, my approach to typical patients with PSVT is to first reassure them that while it is highly bothersome, it is benign and almost never causes cardiac arrest. Remove fear.
Then we go over ways to break the tachycardia — for instance, lowering sympathetic stimulation then doing vagal maneuvers. The initial lowering of sympathetic levels by sitting or lying down is crucial because vagal maneuvers are unlikely to work when people have a sympathetic tone of 11 out of 10.
Some PSVT patients — those who have hypertension and another reason to take meds — can have their PSVT treated with meds. Though daily meds are a very inelegant solution for an intermittent problem.
Then we discuss ablation, and while it is true that most often it is easy and highly curative, there are finite risks.
Ultimately, the treatment of patients with PSVT is, I think, one of the best examples in all of medicine of a preference-sensitive decision. I emphasize this strongly. Ablation is great, but what a patient chooses is ultimately up to them.
And that is where a drug like etripamil comes in. For some patients with infrequent PSVT, a sniff on nasal inhaler once in a while may be preferrable to an invasive procedure.
AF Conversion With Vernakalant
Speaking of drugs for conversion of rapid new-onset tachycardias, The BMJ has published the RAFF4 trial of vernakalant vs procainamide for the acute conversion of AF.
For my US colleagues, you likely have never heard of the drug vernakalant, which is atrial-selective antiarrhythmic that works primarily by blocking multiple ion channels, with particular effects on atrial tissue. Namely, it blocks atrial-specific potassium channels (especially Kv1.5, which carries the ultra-rapid delayed rectifier current IKur) and also inhibits sodium channels in a frequency-dependent manner, prolonging atrial refractoriness and slowing conduction to terminate atrial fibrillation while having minimal effects on ventricular tissue.
Now, US colleagues have never heard of it because the FDA rejected the drug. I was on the Ad-Committee that recommended against it on its second review in 2019. The main problem was that a) there are plenty of ways to convert AF in the ED, and b) there was a major safety event in the trial.
While I am a strong advocate for the give-peace-a-chance approach to acute AF — and I believe that my emergency medicine and cardiology colleagues are way too quick to shock people with AF — I recognize the advantage of quickly getting someone out of AF. Because AF is a buzzkill. But I hate shocking people. Even now, after thousands of times, I find shocking people displeasing. What’s more, if you wait many of these patients convert on their own.
But if there was a drug that was safe and easy, I can see the benefit. In fact, I use procainamide to convert AF fairly often — especially in the EP lab when AF makes SVT ablation impossible.
RAFF4 was a multicenter Canadian trial. Patients (N = 350) had new-onset AF. It was simple: procainamide vs vernakalant and the primary endpoint was simple and conversion to sinus rhythm.
Vernakalant was more effective (62.4% v 48.3%; adjusted absolute difference 15.0%, 95% CI, 4.6% - 25.0%, P = .005; adjusted odds ratio 1.87, 95% CI, 1.2 - 2.9, P =.006). Also, time to conversion was faster (21.8 vs 44.7 minutes).
Fewer patients in the vernakalant group underwent attempted electrical cardioversion (33.7% vs 44.2%; odds ratio 0.62, 95% CI, 0.39 - 0.96, P =.033).
Adverse events were similar in both groups, were generally mild and brief, and most patients were discharged home.
Subgroup analysis strongly favored vernakalant for conversion in patients younger than 70 years (73.3% vs 47.2%; adjusted odds ratio 3.1, 95% CI, 1.7 - 5.5, P =.001, interaction P = .005).
My Comments
I don’t know. I remember that ad-com meeting. There were strong and influential people on it. I may have been swayed by these leaders.
RAFF4 was clearly positive for vernakalant, which is far easier to give than procainamide. Vernakalant is given in a 10 minute infusion vs a 60 minute slow infusion of procainamide — which then hangs around for hours.
The drug is approved for use in Europe and Canada, and I am not sure how often it is used. But the trial seems highly supportive of the drug. I’d be open to relooking at it, because acute conversion with a well-tolerated drug in the emergency department would be kind of nice.
My colleague and friend Rita Redberg co-wrote an accompanying editorial, and they stand with our ad-com against vernakalant. They cite alternatives for cardioversion and the fact that a 350-patient RCT cannot allay safety concerns.
These are good points, but I would counter the alternative arguments that the alternatives are not without safety concerns. Ibutilide, for instance, has a 2%-3% rate of torsades de pointes, which is way worse than AF. Cardioversion requires heavy sedation, and I have seen serious post-shock rhythm disasters. And of course there is amiodarone and its potential safety issues, such as hypotension.
Maybe some Euro listeners could weigh in. I feel like maybe FDA was too strict on this one.
A Quick Note on HFpEF
JACC Case Reports has published a nice commentary on the matter of heart failure with preserved ejection fraction (HFpEF), but I’ve gone on long enough so I will save that for next week.
© 2026 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 09, 2026 This Week in Cardiology Podcast - Medscape - Jan 09, 2026.
Comments