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 on Apple Podcasts, Spotify, or your preferred podcast provider.
In This Week’s Podcast
For the week ending January 7, 2021, John Mandrola, MD comments on the following news and features stories.
First, I want to say Happy New Year and again express my thanks to you for listening to This Week in Cardiology (TWIC). What a start we have to 2022 with the Omicron surge.
The old John would have opined briefly on this. The new John, however, has decided that no one is persuaded by any pandemic-related commentary; everything pandemic-related is politically charged and too dangerous to speak on; and it’s best to move straight to cardiology topics.
Sudden Death
One thing we electrophysiologists tend to forget — me included — is that nearly all patients in whom we implant pacers and implantable cardioverter-defibrillators (ICDs) end up dying with them. Sometimes the cause of death in these patients is obvious — cancer, heart failure, pneumonia, and so on. But sudden and unexplained death is also common, even though nearly all modern devices have the capability of detecting arrhythmias. So there is a bunch of learning to be had here.
The Melbourne Australia EP group has published a very nice paper in JACC Clinical Elecrtophysiology chronicling a 15-year experience looking at post-mortem interrogation of cardiac devices. The devices included pacers, ICDs, and loop recorders (ILRs). The sample size was 260 patients, median age 73.
After the autopsy, about seven in ten of these deaths were deemed cardiac.
Interrogation directly informed the cause of death in half the cases.
Ventricular tachycardia (VT) or fibrillation (VF) was the terminal rhythm in nearly half the group and was more common in patients with ICDs than pacers. None of the ILR patients had VT/VF as the terminal rhythm (thank goodness since we often implant those for syncope).
Only 10% of patients who died from VT/VF had non-sustained VT (NSVT) in the 30 days before death.
Potential cardiovascular implantable electronic device (CIED) malfunction was seen in 8% of cases. Issues included failure to appropriately treat ventricular arrhythmias (n = 13), lead issues (n = 3), or battery depletion (n = 6).
Of the 13 cases identified to have potential undertreatment of ventricular arrhythmias, two devices had major technical issues identified due to faulty leads for which a recall had been issued; six had not appropriately recognized a ventricular arrhythmia and therefore withheld defibrillation; and five had appropriately withheld therapy as the ventricular arrhythmia was at a rate below the designated therapy thresholds set by the treating cardiologist.
In several cases, the absence of an arrhythmia carried medicolegal implications: for example, in eight fatal car crash cases, only one patient had a ventricular arrhythmia documented on their CIED.
Comments. These are important findings. Device malfunction was directly implicated in three patients, and in six patients, there was concern regarding misrecognition of rhythm. This finding of device malfunction as causal in death has been described by German and UCSF teams.
Here is why these observations are important: there is this notion that ICDs are near perfect and require little maintenance. Neither is true.
The best way to mitigate these problems is laser-like focus on programming and follow-up. Much attention is focused on the implant decision and implant procedure. But these are arguably the easy parts. It’s almost as if we need a timeout for programming as much as for pre-op.
Patients with these devices should understand going into the procedure the importance of remote follow-up. I also strongly believe in a medically conservative approach to hardware choice. As a younger doc, I was totally burned by the Sprint Fidelis disaster. We were wooed by new technology. I now use tried and true leads. There are performance reports and there are leads that have proven performance. I stand by these leads. Math professor Nassim Taleb calls it the Lindy effect — named after a deli that has withstood the test of time.
The second important observation is the lack of significant pre-mortem abnormalities — which is highly notable because this is a population enriched for having heart disease and high risk. Only 10% who died from VT/VF had NSVT in the 30 days before death. Thus, 90% of patients who died from VT/VF had no NSVT in the 30 days before death. Only one patient had a shock for VT/VF in the 30 days before.
This lack of signal has implications for the move towards more monitoring. Remote monitoring may be great for detecting lead and device malfunction, but it’s unlikely to improve outcomes from finding warning arrhythmias, of which there seem to be few. The corollary here is the deluge of messages cardiologists receive regarding NSVT in patients with pacers and ICDs. Right now, these self-terminating arrhythmias instill quite a bit of fear, and fear means stress testing, echocardiograms, and then catheterizations. Studies like these make me wonder how much NSVT — and for that matter nonsustained short lived atrial fibrillation (AF) — is “normal” in these patients. I sense that a lot of it is expected and non-risk stratifying.
Final comment on the S-ICD vs transvenous ICD. Almost all the devices in this study were transvenous devices. S-ICDs have higher defibrillation thresholds and more difficulty sensing. I think it will be very important to do similar post-mortem analyses in anyone with an S-ICD who dies suddenly.
Inclisiran
Just before Christmas, the US Food and Drug Administration (FDA) approved inclisiran as an adjunct to statins for further reduction of LDL cholesterol (LDL-C). Inclisiran is a first-in-class, small-interfering RNA agent. Not only is it novel in drug type it is novel in delivery as it is given by SC injection initially, then 3 months later, and then every 6 months. The drug works by “silencing” RNA involved in making PCSK9, which has a role in controlling the number of LDL-C receptors.
The drug is indicated for patients on maximally tolerated statins who have known heart disease or heterozygous familial hypercholesterolemia (HeFH) who require additional LDL-C reduction. The approval is based on the three ORION clinical trials. These trials, the ORION 9 trial of patients with HeFH, and Orion 10 and 11 trials of patients with atherosclerotic cardiovascular disease (ASCVD) or risk-equivalent patients, all showed LDL-C reductions of about 50%.
Some interesting tidbits:
“Peak plasma levels of inclisiran occur approximately 4 hours after dosing, and most is excreted through the kidney.”
After uptake, inclisiran is bound to the RNA-induced silencing complex in liver-cell cytoplasm.
Inclisiran is no longer detectable in plasma within 24 to 48 hours after dosing.
A theoretical concern is that any adverse effects (AE) are staying a while because the LDL lowering effects decay slowly, over months. But the authors write that the trials included thousands of injections and person-years of exposure and AE were similar.
As for cardiac events, the three trials had LDL reduction as the primary endpoint, which regular listeners know is merely a surrogate marker for cardiac (CV) events. CV events were actually lower in the inclisiran arm, but the authors say, rightly so, that there were too few events to make meaningful conclusions.
Comments. My first thought is that it was peculiar that FDA approved a novel drug based solely on LDL-C reduction. But this is probably because the FDA considers LDL-C a reliable enough surrogate marker to give market approval. Also, a meta-analysis of the three ORION trials with 3,660 patients found a 24% reduction in CV events. That same meta-analysis also found no signal of AE. The ORION 4 trial is a CV outcomes trial of inclisiran vs placebo in patients on statins. It is scheduled to enroll 15,000 patients.
While all signals look good, and LDL-C is a strong surrogate, patients do not complain about a cholesterol level. Reduced myocardial infarctions, strokes, and death are the patient-centered outcomes. In practice, FDA approval of this drug may not matter for now because it will be costly and until there are outcomes benefits, insurers are unlikely to cover it. But if ORION 4 delivers even mildly improved CV outcomes, the once every 6-month delivery will be hard to beat. Plus, it’s not a high bar to beat evolocumab and alirocumab, as the hazard ratios (HR) in FOURIER and ODYSSEY were only 0.85, a 15% reduction in nonfatal CV outcomes. There is a bit of race going on here. MERCK has an oral PCSK9 inhibitor in development. Phase 1 data was presented at the American Heart Association meeting and it was favorable, but inclisiran outcomes will be way ahead of it.
Finally, it has been a long time since I have been in molecular biology class, but this seems like a truly remarkable scientific advance. It’s pretty darn cool.
DOAC Choices and the Perils of Observational Comparisons
The choice of direct acting oral anticoagulant (DOAC) comes up often. I don’t know about your practice but for me, it’s apixaban or rivaroxaban. I haven’t started dabigatran as a new prescription in years. Patients often ask which of the two DOACs are best. Both drugs were tested in non-inferiority trials against warfarin, and both drugs proved non-inferior to warfarin for stroke and systemic embolism. They’ve never been compared head-to-head in an RCT.
Many of us think apixaban is the superior agent because the point estimates of the HR for apixaban vs warfarin looked better than rivaroxaban vs warfarin. And since the pharmacokinetics are similar, apixaban’s twice daily dosing seems to make more sense.
Here’s the thing though: the trials enrolled different types of patients, were carried out in different centers, and we should never ever discount the real-world value of once-daily dosing over twice daily dosing.
Vanderbilt authors have published a comparison of the two drugs. Did they randomize tens of thousands of patients? No. They harnessed a US Medicare claims data base and performed a retrospective cohort study of 581,000 patients with AF who a clinician chose to give either rivaroxaban or apixaban.
They found the adjusted primary outcome of major ischemic stroke/systemic embolism and hemorrhagic intracranial bleeding rate for rivaroxaban was 16.1 per 1000 patient years vs 13.4 per 1000 patient years with apixaban.
The risk difference was 2.7 per 1000 patient years and the HR was 1.18 with the confidence interval (CI) ranging from 1.12 to 1.24. In other words, rivaroxaban was 18% worse.
But in fact, for every outcome rivaroxaban looked slightly better.
The authors did tons of adjustments and sensitivity analyses. This included adjustment for 208 covariates — such as demographics, clinical features (diabetes or no diabetes), measures of frailty, etc. But then they wrote this in the first line of the limitations:
Residual confounding by unmeasured factors, including geographic variation in the preferences of patients and physicians, is possible.
There is also the matter of misclassification of outcomes inherent in claims data, and the matter of those who discontinued therapy.
I want to focus on the non-randomization issue. My friends, we have discussed this many times: no matter how many variables you control for, you can only control for those on a spreadsheet. But tons more goes into a clinician’s decision to use one drug or another. If you want to draw causal conclusions from a comparison of drugs, you need randomization to balance out the measured and unmeasured co-variates.
You simply cannot make conclusions from this sort of data. Yes, you can use the priors from the DOAC vs warfarin trials, and the pharmacokinetic data, so it’s plausible that apixiban is better than rivaroxaban, but you can’t discern it without randomization. And don’t be persuaded by the 580,000; large numbers do not mitigate bias from confounding. In fact, it may worsen it. Though I don’t think you need it, perhaps you want more than residual confounding to ignore these findings. Here is another clue: As pointed out by Dr Manesh Patel in the Medscape news coverage, the unadjusted rates of events favored rivaroxaban and apixaban came out ahead only after adjustments.
The good news is that a little birdie told me at a recent conference that there may be a large-scale randomized trial comparing these two drugs. That is the only way to tell. Think about this: you could simply embed randomization into an electronic health record. For now, both drugs have utility. I truly don’t know if there is a difference. Every time I prescribe apixiban, thinking it may be better, I remember how many times I forget to take the second dose of my asthma puffer. Do not discount the value of once-daily dosing.
Alcohol and VT/VF
The Adelaide electrophysiology group published the results of an observational study of just over 400,000 individuals from the UK biobank, looking at the association of total and beverage-specific intake with ventricular arrhythmia (VA) and sudden cardiac death (SCD) events.
The UK Biobank is a rich source of observational data. It is a prospective cohort of about a half a million individuals ages 40-69 years across the United Kingdom. Individuals were recruited from assessment centers between 2006 and 2010. They completed a questionnaire and had basic anthropomorphic measurements made. Interviews were done to assess comorbidities. Then the researchers can use death and hospital records to assess outcomes.
The knock on the UK Biobank is that it may suffer from healthy volunteer bias, and therefore may not be generalizable to the population as a whole.
The main findings of this study:
No clear association was found with total alcohol consumption and incident VA; however, consumption of a greater amount of sprits was associated with increased VA risk.
A U-shaped association was seen for total alcohol consumption and sudden death risk. U-shaped means intake of low amounts of alcohol looked protective (lower SCD risk).
Beverage-specific analyses of SCD risks delivered contrasting results: high intake of beer, cider, and spirits was associated with greater risks while red and white wine was associated with lower risks.
No substantial association was found of an interaction between alcohol and sex for VA and SCD.
We learn in the discussion section that while there is a ton of data on alcohol and AF, there hasn’t been a lot of data published on alcohol and VA. The main finding to discuss here is that you’d think alcohol would have the same relationship with VA and SCD. But here there was no real association with VA but a U-shaped protective effect at low doses in SCD.
I think I agree with the authors who write that there is limited specificity in SCD diagnoses for a fatal event. They cite the USCF autopsy study I mentioned earlier, which found that 40% of all SCD were confirmed as non-arrhythmic in origin. The beverage-specific contrasting associations with SCD are novel. But it’s hard to sort out causality because clearly it’s possible that people who drink lots of beer and spirits may materially differ from those who drink white wine.
This is a good first step in understanding the relationship of alcohol and VA. And it seems to further reinforce the crucially important concept that toxicity is in large part a matter of dosage.
© 2022 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 7, 2021 This Week in Cardiology Podcast - Medscape - Jan 07, 2022.
Comments