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 12, 2023, John Mandrola, MD comments on the following news and features stories.
A Thank You
First, a quick thanks to Dr Apurva Vyas and the crew at Lehigh Valley Health for inviting me to give the Keynote last Saturday.
As always, I’m grateful for these invitations. It is an honor to be asked to speak and it is great to meet the young people training to do this great job of cardiology, but this was a particularly fun meeting as the sessions were notable in the robustness of their critical appraisal.
Subspecialists were critically appraising the literature in their fields. A vascular surgeon, for instance, questioned the value of operating on asymptomatic carotid lesions. An interventional cardiologist went through the studies of percutaneous coronary intervention (PCI) and stable coronary artery disease (CAD) and, wait for it, lauded the strength of modern medical therapy. Absent from this session were cheerleading sessions of the favorite drugs of the day.
Listener Feedback
I received an email from a full professor regarding my take on statins. The professor noted that the CTT (Cholesterol Treatment Trialists collaboration), a group that has done many meta-analyses of individual patient-level data, does not share source data with independent researchers. The professor also noted that CTT does not collect data on adverse effects of statins, and this is important because of the possibility of what Prof Hayward called unknowns unknowns of statin drugs. This week I will link to Hayward’s piece — one of the best I’ve read on the matter.
The professor also noted that CTT receives funding from industry. I looked it up, and it turns out that funding comes from professional societies in Australia and Britain such as the Australian and British Heart Foundations. CTT writes that they do not receive funding from industry. Of course, it is a bit of semantics because most professional societies exist and fund things with industry support.
My take on statins as a preference sensitive decision remains. Though I agree that a data source so important to cardiovascular (CV) health ought to be open to independent researchers.
By the way, I love listener feedback even when it disagrees with my take. Thank you. Now is a good time to remind everyone to consider rating and reviewing this podcast.
Left Atrial Appendage Occlusion
Europace has published a paper from American authors using an American database that has produced worrisome trends. This was an observational study using the National Inpatient Sample (NIS) from 2016-2020. They identified roughly 74,000 left atrial appendage occlusion (LAAO) device implants in the United States.
They stratified outcomes by CHA2DS2-VASc score. Approximately 63% of LAAO device implantations occurred in patients with CHA2DS2-VASc scores of 4 and ≥5.
Before I tell you the results I have two comments:
This is a good use of observational data. Science tells us what we can do, trials tell us what we should do, and registries tell us what we are doing.
The seminal trials that got this device approved were PROTECT and PREVAIL. Nearly two-thirds of PROTECT-enrolled patients with a CHA2DS2-VASc score of ≤ 2. PREVAIL enrolled higher risk patients, but the average CHA2DS2-VASc was 3.8. The point is that, even if you believe this device delivers a net benefit — I don’t, but even if you did — the trials enrolled far lower risk patients.
I’ve talked often on this podcast about the importance of co-morbidities in modifying treatment effects. And as you would expect, in this study, as the CHA2DS2-VASc increased, the authors observed increasingly prevalent co-morbidities, such as chronic kidney disease, peripheral vascular disease, and unintentional weight loss.
The first main finding was that the rate of complications increased with CHA2DS2-VASc score.
Overall complications:
7.4% in patients with a CHA2DS2-VASc of 3;
9.4% in patients with a CHA2DS2-VASc of 4;
14% in patients with a CHA2DS2-VASc of ≥ 5.Same with major complications:
5% in patients with a CHA2DS2-VASc of 3;
6.2% in patients with a CHA2DS2-VASc of 4;
8% in patients with a CHADSVASC of ≥ 5.In-patient death:
0.1% in patients with a CHA2DS2-VASc of 3 and 4;
0.3% (n =60) in patients with a CHA2DS2-VASc of ≥ 5.
The authors did some regression analyses adjusted for potential confounders and found that CHADSVASC 4 and ≥ 5 were independently associated with higher rates of overall complications.
For patients with a CHA2DS2-VASc ≥ 5, the adjusted odds ration of complications was 1.88 or near double the risk of an important complication.
Comments. Notwithstanding my concerns about the results of the seminal trials of this procedure, this is worrisome data. This shows something many of us see. The patients who are being referred for LAAO are often saddled with significant co-morbid conditions.
These observations tell us that having more co-morbid conditions — as evidenced by a higher CHA2DS2-VASc score — means the risk of serious complications are higher.
That is hardly controversial for any procedure. Why it is especially important here is that LAAO is a preventive procedure in which you hope the future probability of lower stroke and bleeding rates is greater than major complication rates.
If you choose patients with higher upfront risk, you create a higher bar for net benefit in the future.
But it’s even more than just treatment benefit vs harm. There is also the issue of the risk of the primary outcome. Here the outcomes of interest with LAAO are less stroke and less bleeding.
But, when you choose patients with higher CHA2DS2-VASc score, you invariably choose patients with more competing causes of stroke. An 80-year-old man with CHA2DS2-VASc of 5 and atrial fibrillation is likely to have other causes of stroke, such as carotid, intracranial, or aortic disease, or valvular heart disease.
We desperately need randomized controlled trial (RCT) level data on these older patients with co-morbid conditions. I am so worried about our patient selection for this procedure.
JACC Interventions published another notable study regarding the use of CT scans for guiding LAAO.
This was a subanalysis of the SWISS APERO RCT that compared Watchman and Amulet. I apologize for missing the RCT when it came out in 2021. Published in Circulation, the two devices for percutaneous closure of the appendage performed a bit differently from each other. The Amulet device had a lower peri-device leaks on transesophageal echocardiography (TEE) at 45 days but a higher procedural complication rate.
That sentence hardly makes news, right? Two competing devices behaving a bit differently is a common issue in cardiology. What struck me about this RCT, done in eight European centers, was the actual numbers.
At 45 days, the peri-device leak rate via TEE was higher with Watchman (27.5%) than with Amulet (13%).
27% vs 14% had peridevice leaks. In a trial, when docs know they are being studied.
This is similar to the LAAO registry data published by a Mayo group in JACC-Electrophysiology last summer. I’d expect the rate to be lower in a trial format than in real world data.
I want to do the same thing with the JACC-Intervention substudy, which broke out the effects of using CT as a guide to closure.
According to the study protocol ongoing at the time of the procedure, the first operators had (computed tomography angiography [CCTA] unblinded group) or did not have (CCTA blinded group) access to preprocedural CCTA images.
This paper compared results of the two groups. And as you would expect the group that used CT images had better results. CT is pretty darn good at looking at the appendage. (We’ve used it to adjudicate some cases where TEE struggles to sort out clot or no clot in the LAA)
But the actual numbers of successful closure without leaks is kind of sobering.
Long-term procedural success was seen in 84% of CT unblinded vs 72% blinded. So again, that is a lot of patients treated in the best-case scenario of a clinical trial who do not achieve long-term successful closure.
Again – there are so many things that have to go right for this procedure to deliver more benefit than harm. In other words, there is a lot of multiplication of probabilities.
If you are the patient, you have to avoid a procedural complication, then you have to get perfect closure at 45 days, have no bleeding on more intense short term anti-thrombotics, get good endothelialization of the device, and then not have a stroke from another source, such as intracranial or carotid artery.
Sorry about the seeming negativity here. I am open to you sending me non-negative papers. I haven’t come across too many.
The Promise of DNA
A few years ago, I was struck by the idea of using DNA to predict the future. That is an amalgam of DNA the so-called polygenic (many-gene) risk score (PRS). Don’t drift off.
We all know that there are diseases related to single genes. Think Long QT or Brugada Syndrome.
Atherosclerotic cardiovascular disease (ASCVD) may have a heritable component, but clearly there is no one gene that causes it.
But what if you could take oodles of single nucleotide polymorphisms or SNPs, each with some association with ASCVD and use these SNPs to form a PRS.
You’d need the ability to sequence DNA.
You’d need a lot of computing power. We have all that. The promise, or hope, is that you could use a PRS to risk-stratify people into low or high-risk groups. Think of it as a super high tech family history.
Many researchers have made PRS for different diseases, like obesity and diabetes. Here is a simple explanation:
First comes the genome-wide association study in which you find specific genetic markers – the SNPs — that are more common in people with heart disease.
Then you select and weight the SNPs according to the strength of the association with heart disease.
Then you sum up the weighted contributions of each SNP.
Boom. There is your PRS.
The promise is a quantification of a person’s genetic susceptibility to getting heart disease. Given the high prevalence of heart disease, a highly predictive and specific PRS would be so much better than a current pooled cohort equation which uses basic bedside data.
In 2018, a famous group of scientists in Boston published a study showing that PRS for common diseases identified individuals with risk equivalent to those with monogenic mutations. They concluded that it was time to think about including PRS in clinical care.
We now have a number of clinical studies of PRS, including one this month in a million veterans.
And I think Yogi Berra was closer to the truth: Yogi famously said, ”"It’s tough to make predictions, especially about the future.”
Does a PRS improve risk estimation over the traditional risk factors in veterans, for CAD and stroke events?
The study subjects were taken from a longitudinal cohort called the Million Veteran Program – a mega data bank that had PRS scores and clinical data.
The main outcomes were myocardial infarction (MI) and stroke.
The sample size was nearly 80,000; mean age 57 years, all without previous heart disease.
A strength of the study was that it included decent proportions of different races and ethnicities. Since these were veterans, it was 85% male.
There were statistically significant associations of the PRS with future MI and stroke. But the risk increase was very modest – only about 10% to 20% higher.
Net reclassification from adding the PRS to traditional risk scores, however, was modest.
And the C-statistic from adding PRS to traditional risk factor scores averaged about 0.01. In other words, almost nothing.
The numbers were slightly better for woman and in younger patients, but the absolute difference in prediction was small.
This is very similar to an older paper, published in JAMA in 2020, first author Jonathon Mosley. They also looked at longitudinal studies ARIC and MESA and assessed the additive predictive ability of PRS to the traditional risk score. And again, they found that the PRS is associated with CAD events but did not significantly improve discrimination, calibration, or risk reclassification compared with conventional parameters.
Comments. I realize this involves genetics and biology, but I mention it because of how wrong I was in the past to be so enamored by the early promises. I thought tons of genes sorted together. This should work.
But in actual practice, this extremely high-tech score isn’t substantially better than entering in your total cholesterol, HDL, blood pressure, age, and smoking history.
For those of you who want to know more about why — if I were smarter, I would have known the challenges and been more skeptical — I will link to an amazing paper from the late Cecile Janssens. She explained, in readable terms, why PRS will struggle to predict diseases like heart disease. The paper is called “Validity of polygenic risk scores: are we measuring what we think we are?” And it is open access in the journal Human Molecular Genetics.
Dr Janssens summarizes this issue by writing that “Models are a simplification of reality.” The construction of PRSs needs to acknowledge the biological reality, not create a new one.
In 2019, she called for “more comparative research to investigate the construct, content and criterion validity of PRS; to explore alternative ways of quantifying polygenic risk; and to rigorously compare new and current methods.”
Well, the comparative studies are being done and, as Gerd Gigenzer has noted, when there is uncertainty, fast and frugal prediction rules are often as good or better than complex models.
And, if he is correct, the fact that a traditional risk calculator is as good as a complex PRS, bodes somewhat positively for the lowly CHA2DS2-VASc score in AF.
Speaking of Influencing the Future – Statins In Older People
The American Geriatrics Society (AGS) is working on preliminary recommendations for deciding on statin therapy for primary prevention in adults older than 75 years. I cover this topic because statin use in older patients is a dilemma that comes up nearly every clinic day. The problem is that we don’t have much in the way of trial data.
Journalist Kathleen Doheny covered a session at the AGS meeting. She described the way leaders in geriatric medicine would help sort out the statin decision. Some of tools are wise; some extremely unwise.
One of the decision tools used clinical frailty scales. Ok, I like that. Consider co-morbid conditions as these will affect competing causes of mortality –besides heart disease.
Another of the decision tools they mentioned was coronary artery calcium (CAC) scans. My friends, consider this a public service advisory, DO NOT ORDER CAC SCANS in older people. Most have CAC, and all it will do is tempt doctors to do more testing.
At the risk of sounding like a know it all, I have two thoughts regarding primary prevention therapies in the elderly.
Remember what primary prevention is for: It is to reach old age. We treat blood pressure and cholesterol in 40 to 50-year-olds so they make it to 80.
Once you are 80, I am not sure primary prevention has as much meaning, because at age 80, in general, the notion of trying not to die of one disease might detract from living life.
But that is my view; some 80-year-olds won’t feel that way.
A second solution is to let patients decide. If they think a 25% reduction in future events is worth taking a pill every day, then so be it. Of course, you give them the information to make the decision, but ultimately their preferences make it easy.
I don’t think we need a guideline when there is no reliable data, and 80-year-olds differ so much in their conditions and preferences.
FDA Approves Dapagliflozin for Any Patient with HF
You all know that the SGLT2 inhiibitor dapagliflozin was approved for patients with heart failure with reduced ejection fraction (HFrEF). This new expanded indication is based on the DELIVER trial and a meta-analysis combining DELIVER (HF with preserved EF; HFpEF) and DAPA-HF (HFrEF).
I’ve covered DELIVER before:
6000 patients, mean EF 54%, dapagiflozin vs placebo.
Primary outcome of CV death or hospitalizations for HF (HHF) was 18% lower in relative terms and 3.1% lower in absolute terms.
But it was driven by HHF not CV death, same as EMPEROR-PRESERVED. The problem is always that HHF represents a small proportion of hospitalizations for patients with HFpEF.
In my opinion, it is not fair to meta-analyze DELIVER with DAPA-HF because these are two very separate types of patients and the benefits of using SGLT2 inhibitors in patients with HFrEF will make the drugs look better in patients with HFpEF.
But it’s hard to stew too much about this decision. It was going to happen. Indeed, there are probably some patients with HFpEF who may benefit, and many patients with HFpEF will have other indications for SGLT2 inhibitors, such as chronic kidney disease or diabetes.
What I would strongly oppose is any strong guideline recommendation or quality measure that nudges doctors to always use these drugs, because the data in patients with HFpEF is simply not strong enough for that.
© 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: May 12, 2023 This Week in Cardiology Podcast - Medscape - May 12, 2023.
Comments