May 19, 2023 This Week in Cardiology Podcast

John M. Mandrola, MD


May 19, 2023

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 May 19, 2023, John Mandrola, MD, comments on the following news and features stories.

ICD in Patients with NICM

Modern times have brought renewed interest in risk stratification for the implantable cardioverter defibrillator (ICD). But first, let me digress briefly on ICDs.

In days of old, cardiologists had simple rules for implanting an ICD. If the ejection fraction (EF) was less than 30% to 35% in a patient with heart failure (HF) on medical therapy, you recommended an ICD. Trials showed that this improved survival. Not cardiovascular (CV) death, but overall survival.

Of course, this rule also included making sure the patient had reasonable expectations of good survival, and making sure there was thorough discussion about benefits, harms and expectations.

Sadly, during the period of irrational exuberance of ICD implantation, after the SCD-HeFT and MADIT 2 trials, ICDs were overused. Fueled by brilliant marketing and keen use of key opinion leaders, EF < 30 morphed into, “you need this ICD.”

Well, progress has made ICD decision-making harder. A lot harder.

In 2017, HF trialists published an under-cited paper showing a robust decline in the rate of sudden death in patients with HF over the 20 years of HF trials. It showed that, “The cumulative incidence of sudden death at 90 days after randomization was 2.4% in the earliest trial and 1.0% in the most recent trial.”

  • That’s hugely important for ICD use, right? ICDs improve survival only if ventricular tachycardia/fibrillation (VT/VF) comprises a large proportion of the deaths.

  • Around the same time, Lars Kolber and colleagues reported the results of DANISH — a randomized controlled trial (RCT) of ICD implantation in patients with nonischemic cardiomyopathy (NICM), terrible EFs, but great background therapy.

  • The Kaplan Meier curves were near identical. No difference. None. ICDs, in a modern setting, did not reduce mortality. DANISH is inconvenient for the electrophysiology (EP) and device industry, so it has been largely ignored.

Guidelines still include ICD implantation in these patients. One trick proponents use to ignore DANISH is to meta-analyze it with older trials and say, look, in meta-analyses, ICDs are still beneficial. (Pro tip: Always be cautious of meta-analyses that include old and new trials.)

Here is the problem now though. It’s not so simple as ICDs work or don’t work in NICM. Here is why: In the old days, NICM simply meant patients without infarct-related CM.

NICM in the modern era includes a much more diverse group. You have sarcoid, amyloid, healed scar from myocarditis, hypertrophic, arrythmogenic right ventricular cardiomyopathy (ARVC), noncompaction, and genetic forms, such as familial dilated CM related to Lamin AC mutations.

We have advanced beyond the simple distinction of NICM. And within this diverse group of patients, there are surely gradients of risk. And sit down for this: Some of these patients with CM-not-due-to-old myocardial infarction (MI), will die of sudden death and have EFs > 35%. We have known for decades that EF is an imperfect risk stratifier.

Yet, in days of old, there was no interest in risk stratification beyond EF. You know why? Because it risked reducing the number of people who would get ICDs.

Modern times have brought renewed interest in risk stratification.

The group in Rochester, NY – home of the MADIT trials — has presented a paper on a simple risk score for predicting VT/VF in patients with NICM. First author Ido Goldenberg discussed the study at the 2023 Internal Medicine meeting of the American College of Physicians. Journalist Robert Fulton has coverage on | Medscape Cardiology site.

  • They had data on ≈ 1500 patients with NICM from the MADIT-RIT and MADIT CRT trials. Through sophisticated statistical methods using patient characteristics and outcomes, they came up with three risk groups – high, low, and intermediate risk.

  • The main four risk factors used was ICD vs cardiac resynchronization therapy (CRT) device; left ventricular (LV) EF; EF ≤ 25%; male sex; and black race.

  • How good was this score? The 5-year cumulative incidences of ventricular arrhythmia (VTA) were 15% in the low-risk group, 24% in the intermediate-risk group, and 42% in the high-risk group.

  • The burden of recurrent VTA in the high-risk and intermediate-risk groups was significantly higher when compared with the low-risk group.

  • The C-index of the model in the development cohort (internal validation) and in the external validation cohort were 0.712 and 0.691, respectively.

Comments. Before I say anything about efforts like this, I want to say we should absolutely have more work done in risk stratification. We should studied this 20 years ago. But no, we settled on EF, which is like treating all sore throats with antibiotics even though only a fraction will have strep throat.

Of course, the best way to study risk stratification is with a randomized controlled trial (RCT). Pick a group you think would benefit more from an ICD and randomly assign. I wonder what a trial of women-only would have shown. Women have a lower risk of sudden death, and higher rate of complications from ICD. This wasn’t done. Why? Because it might have decreased the number of people who got ICDs.

This risk score is not going to be good enough.

  • Low-risk patients still have a 15% risk of VT. That’s not good enough, especially when you are considering younger patients. The C-statistic suggests the risk score is slightly better than a coin toss.

  • The excitement of the day now is use of imaging to risk stratify. An Australian group has looked at ICD implants in Victoria and found a strong association between ICD implantation and reduced all-cause and cardiovascular mortality only in those patients with LV scar.

  • A trial in the United Kingdom called BRITISH will test ICD implant vs no implant in patients with CMR-based scar.

Scar on cardiac magnetic resonance imaging might identify a group that benefits, but the major headwind for ICDs will be good background therapy. And of course, in older people, competing causes of death.

Another interesting risk stratifier is genetics. It’s early yet, but already, there are gene markers, such as lamin A/C mutations that confer a high enough risk that an ICD would extend survival.

Few better areas in cardiology (ICD use) right now underscore the fact that evidence often has an expiration date.

Screening in HCM

My colleague, Brazilian EP Bruno Valdigem, has a nice recap of the TEMPO-HCM study presented at the European Heart Rhythm Association (EHRA) meeting. The study raises a bunch of good questions.

  • Spanish investigators — first author Juan Caro-Codon -- studied the issue of screening patients with hypertrophic cardiomyopathy (HCM) for both atrial fibrillation (AF) and non-sustained VT (NSVT).

  • NSVT is especially important to discover as it is one of the strongest risk predictors for sudden cardiac death (SCD) in these patients. But detecting AF is also important because of the increased risk of clots and stroke.

  • There are always going to be few RCTs in the HCM space because HCM is uncommon, although modern imaging and perhaps ECG screening has led to more diagnoses of HCM.

  • One of the main jobs of the doctor caring for these patients is risk stratification. Until recently we have used 24 to 48-hour ECG monitors to look for VT.

  • Yet, both AF and NSVT are infrequent arrhythmias. The idea of the Spanish team was to look at the impact of longer-term monitoring for these arrhythmias.

  • This was a prospective registry-type study at five centers. Patients with HCM had 30-day monitors. Guess what happened when they looked longer? Well, it was mixed.

  • For AF, in the first 24 hours only 6% of the 100 patients had AF detected. After thirty days, this increased to only 10%.

  • For NSVT, longer monitoring made a big jump; 8% had VT detected in the first 24 hours, but after 30 days, to the percentage rose to 62%. Wow.

  • When they looked at the time course of NSVT detection, there seemed to be a plateau. There was a steep rate of detection from day 0 to 15, but from 2 to 4 weeks, the rate of detection seemed to plateau.

  • The authors then made a very interesting leap: they took the increase in NSVT, applied current risk stratification parameters, and found that 22% of patients were potentially reclassified in a higher risk category.

    • 14% more would be in the “may consider ICD” category.

    • 8% would be in the “ICD should be considered” category.

Comments. I missed this study when it came out, so thank you Bruno. Please do read his column. He asks some great questions at the end.

You might think, come on Mandrola, of course you will detect more arrhythmia if you monitor longer, why is this a big deal?

  • Well, here is why: this study and its results represent a microcosm of the new challenges we face. A 30-day monitor is one thing, well, what about digital watches, implantable loop recorders that are always on — for years?

  • I am sure any clinician listening to this knows that more monitoring means picking up more arrhythmia. The problem comes in translating this to the patient.

  • AF is a good example. The benefits of oral anticoagulants (OAC) were established when all we had were symptoms and a 12-lead ECG. Now we pick up AF so much easier, and we don’t know that the benefits of OAC apply to AF episodes detected with a constant monitoring. In fact, it probably doesn’t. The NOAH AFNET-6 study of direct OACs for subclinical, short-duration AF was stopped for futility.

  • So, I would gently push back on the authors using this data with longer monitoring, to reclassify patients to higher risk. Why? Because the thresholds for NSVT and VT risk in patients with HCM were obtained with a lot less monitoring.

Now that we monitor longer, maybe we need new parameters for risk prediction. Is VT detected on a 1-day monitor the same risk as NSVT detected on a 30-day monitor? I doubt it is, but we need more data.

And it makes me think, and hopefully you too, about the challenges of translating all this digital data that we get.

The makers of digital devices aren’t motivated to correlate data with outcomes. Their selling points are. hey, look at all this data we can collect. I’ve scrolled through health apps in the clinic, and it’s shocking to me what people see on their phones. Data upon data. Yet many ignore their expanding bellies and inability to touch their toes.

Sometimes we find good things with this extra monitoring, but I strongly believe we need new normal values. Having or not having NSVT in the presence of HCM surely means different things when the monitor is 1-day vs 30 weeks vs 3 years of monitoring.

Blending EP and Structural after TAVI

Skeptical as I am, I believe transcatheter aortic valve replacement/implantation (TAVR, TAVI) has been (or is) one of the great achievements of modern cardiology.

Patients with severe aortic stenosis (AS) are some of the sickest, most symptomatic people we see. They are asking for and need our help. It is as pure as medicine gets. It’s about the exact opposite of screening healthy people.

Yet the established treatment, surgical aortic valve (AV) replacement, is no small thing. And many older high-risk patients, sick from AS, not with AS, could benefit a lot from TAVI. If we used TAVI in the oldest and sickest, the higher pacemaker rate post TAVI would be no issue.

But cardiologists and patients seek less invasive ways to replace stenotic AVs. The natural thing, therefore, is to extend TAVI to less sick patients. Iterative improvements in the valve and implant procedure make this a more feasible approach. But there are headwinds. There are always headwinds.

One of the stiffest headwinds to squishing the bad valve and inserting a new valve is that the conduction system lies adjacent to the AV. All AV replacements can lead to heart block and the need for a pacemaker, but rates of pacer implants after TAVI is higher than surgical AV replacement.

The journal JACC-Interventions published an observational study from a large group of Italian investigators, first author, Francesco Bruno, in which they used registry data to study the association between the percentage of RV pacing and adverse outcomes.

  • PACE-TAVI is an international, many-center registry of consecutive TAVI patients who had pacermakers inserted after TAVI.

  • For this non-random comparison the authors divided the 377 patients into those who had RV pacing (< 40%) vs those who had RV pacing (≥ 40%).

  • The endpoint of interest was a composite of CV death or hospitalization for HF (HHF).

  • Of the 377, 42% had ventricular pacing < 40% and 58% had ventricular pacing > 40%.

  • The authors did statistical adjustments and found that the risk of the composite endpoint of CV death/HHF was 2.76 -fold higher in the high percentage pacing group.

  • Taken individually, both CVD and HHF tracked significantly higher in the high percentage pacing group.

  • All-cause death was 2 times higher in the high-pacing group but did not reach significance because of wide confidence intervals.

  • Patients with ventricular pacing ≥ 40% showed a higher New York Heart Association functional class at 1 year and a nonsignificant reduction of left ventricular ejection fraction (P = 0.18) on 1-year echocardiography, while patients with ventricular pacing < 40% showed significant improvement (P = 0.009).

The authors concluded that in pts who have pacers and use them a lot to pace in the RV, outcomes are worse.

“These findings suggest the opportunity to minimize right VP through dedicated algorithms in post-TAVR patients without complete atrioventricular block and to evaluate a more physiological VP modality in patients with persistent complete atrioventricular block.”

Comments. I like this paper. Yes, of course, it is observational, and there may be confounders. Patients who need RV pacing may be inherently sicker and that is why they did worse.

But, let’s put on our Bayesian hats. Let’s think about priors. We know, from oodles of studies, both randomized trials (MADIT 2 had a signal of more HF in the ICD arm, likely because of RV pacing) and observational studies, that RV pacing increases the risk of pacing-induced cardiomyopathy. The incidence ranges from 10% to 25%.

So, our prior knowledge suggests that RV pacing is detrimental. And that too makes sense and is consistent with bio-plausible mechanisms. An RV-paced beat is clearly better than no beat, but pacing the RV muscle creates right and left contractile dyssynchrony. That’s because RV contraction occurs first, and then LV contraction.

An RV-paced beat is similar to a beat with a left bundle branch block, where the RV is activated way before the LV.

  • The solution to this problem is simple. EP teams should strongly consider conduction system pacing in these patients. I am near 100% left bundle area pacing.

  • Here is what happens. When you place a lead into the septum, you can capture the left bundle. This leads to a right bundle branch-morphology QRS and most important, LV activation is not delayed and dyssynchrony is avoided.

  • And there is another advantage: When you pace the ventricles without creating dyssynchrony, you don’t have to extend the PR interval. Cunduction system pacing (CSP) allows for a normal PR interval. Having normal AV synchrony is very much under-rated.

  • In our hospital, nearly 100% of post-TAVI heart block is treated with left bundle branch area pacing. I strongly suspect this will prove beneficial. It’s another reason why TAVI done at expert centers, with experienced EP docs, will likely have better outcomes.

I hope I’ve stirred you all up. Here I am promoting an unproven, untested pacing technique. Mandrola, where is your data. Perhaps you are entranced by the beautiful ECGs from CSP?

Well, there is some data, there are small trials, and non-random comparisons. But here at the Heart Rhythm Society meeting we are having our first investigator meeting for the Left vs Left trial of CSP vs CRT. It’s being studied, of course, with government — not industry — funding.

Stay tuned. But I will bet you all a big cappuccino that the left bundle branch proves superior to RV apical pacing.

Troponin Assays

Don’t laugh, but our hospital has just recently started using high-sensitivity troponin (hsTn) assays. It’s creating a bit of confusion. I am here to sort this out for you all.

JACC has published a study that sought answers as to how to define a normal Tn value. John McEvoy and colleagues led the analysis of frozen blood samples from individuals in the National Health and Nutrition Examination Study (NHANES) from 1999-2004.

They measured 99th upper limit normal (ULN), henceforth, just ULN, for three hsTnI assays and one hsTnT assay. There were about 12,000 individuals in the sample.

The first task was to find super healthy controls. They excluded people who said they had CV disease, abn BMI either high or low, those taking CV meds, those with diabetes or hypertension or high cholesterol; smokers, heavy drinkers, those with recent admissions, those with high BNP.

This left about 2700 people who formed the healthy reference subsample. The other 10,000 or so formed a second group — those without overt CV disease but some risks for CV disease — you know the type of people who present with chest pain syndromes.

Then they compared the ULN for each troponin assay in these adults. They also reported hsTn percentages by sex, age, and race/ethnicity.

  • Finding 1. In the healthy cohort, the ULN for all three hs-cTnI assays were lower than the manufacturer-reported ULN; in two cases, the 99th percentile ULN was less than one-half of the approved cutoffs. The ULN for the TnT were consistent with the manufacturer.

  • Finding 2. Healthy men had higher ULNs than healthy women across assays,

  • Finding 3. Older individuals showed higher values than younger. All were statistically significant.

  • Finding 4. No significant differences were observed when stratified by race or ethnicity.

  • Finding 5. In the general population without heart disease but with risk factors, eg, the other 10,700 individuals, the nationally representative ULN for hsTn were higher than the 99th percentile ULNs calculated in the healthy reference sample. And there were sex and age differences.

Comments. Be not afraid. This seems super confusing. But I am going to try and make it simple.

First, there is some flux in the exact choices of ULN. Sources of this uncertainty includes different assays, age, sex, but not race/ethnicity. These are good things to know.

But clinically, not much is changed, even though greater sensitivity has made troponin assays less specific. But one great thing is that in low-risk patients — those with very low likelihood presentations -- two normal hsTn levels have incredibly strong negative predictive value.

But don’t let these details on flux in the ULNs intimidate you.

  • The first and most important question you have to answer when seeing patients in the ED is, is this an acute coronary syndrome from acute occlusion, either ST-elevated MI (STEMI) or non-STEMI?

  • You can mostly make that on clinical grounds. Chest pain and ECG. Troponins are barely needed.

  • If there is a plaque rupture type MI, then the treatment pathway is clear. STEMIs to the lab, non-STEMIs get antithrombotic therapy and also go to the cath lab in most cases.

  • When plaque rupture isn’t suspected, clinicians ought not do troponins. This includes patients with SVT, or acute AF, or pneumonia, etc.

  • But even if there are troponin elevations, the treatment is to treat the underlying cause. Treat the underlying cause.

  • This includes type 2 MI from demand ischemia, seen in anemia or tachycardia in the presence of chronic coronary artery disease.

  • It also includes acute or chronic myocardial injury say from acute myocarditis, or CKD or even super strenuous exercise.

Clinicians need not worry too much about exact ULN, because we have to use a lot more than a biomarker.

Again, in the absence of plaque-rupture acute coronary syndrome, don’t do troponins, and if there are troponins, don’t be tempted to go the lab, treat the underlying cause.

In case you are interested, Andrew Foy and I wrote this explanation out in JAMA-IM in 2021.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.