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 March 8, 2024, John Mandrola, MD, comments on the following news and features stories: plastics and heart disease, the MINT trial letters-to-the-editor and Bayes theorem, and Brugada syndrome.
Plastics and Heart Disease
The New England Journal of Medicine (NEJM) published an observational study that explored potential linkage of microplastics and nanoplastics (from now on, I’ll refer to these just as plastics) and cardiac events.
This publication was a big event. Nearly 100 news sites have covered it. It’s all over social media. And I need not say how ubiquitous plastics are in our world.
The reason to explore the cardiac question is that a) plastics are ubiquitous, and b) numerous observational preclinical and animal studies have suggested links of plastics to potential cardiac issues.
The mechanisms are speculative but include oxidative stress, inflammation, and apoptosis in endothelial cells.
The many study authors from three centers in Italy, first author Dr. Raffaele Marfella, used an interesting design. They took resected plaque from patients who had undergone carotid endarterectomy (CEA) and assessed them for plastics using biophysics, such as electron microscopy.
These were all asymptomatic carotid lesions. (An aside could be why asymptomatic carotids should have CEA. It’s a good question. Some countries (the United States, for instance) do a lot of asymptomatic carotids, other countries do very few. We won’t get into that debate today.)
The authors chose asymptomatic lesions to maximize the chances that patients survived the post-procedure period and to minimize interpatient variation in plaque phenotypes.
They then followed 307 patients for future cardiac events. In all, 257 completed follow-up.
They had two groups of patients: One group of 150 patients had plastics detected in carotid plaque and 107 patients had no plastics. The primary cardiac endpoint was a composite of myocardial infarction (MI), stroke, or death.
The most common plastic found was polyethylene (58%); 12% of patients had measurable amounts of polyvinyl chloride. The electron microscopy pictures, which the authors shared in the paper, show jagged-edged foreign particles.
Over 33 months, the main finding was that patients who had plastics detected were at higher risk for a primary end-point event than those in whom these substances were not detected (hazard ratio [HR], 4.53; 95% confidence interval [CI], 2.00 to 10.27; P < 0.001).
I know; it’s a huge effect size for a cardiac event. To support potential causality, the authors told us there were increased levels of four inflammatory markers in plastics-positive patients.
In sum, this observational, non-random comparison study combined biophysics with epidemiology.
As with many observational studies, there was an attempt to compare groups and assess causality. Of course, the authors say this was just an association; it cannot prove causality. But these words are just veneer. What they said in the conclusions was that patients with plastic detected had higher rates of events.
The NEJM also included an 80-reference review article and a scary editorial. That review included all the things that can be done to limit exposure to plastics.
The lead in the conclusion of the review article:
There is an urgent need for the clinical community to address the growing burden of exposure to EDCs, largely derived from petrochemicals, in order to prevent a broad range of associated health harms.
And then this:
In addition to counseling their patients, clinicians can be critical advocates for policy changes to both decarbonize and detoxify the economy in order to address the combined health threats of petrochemical-derived EDCs and climate change.
The editorialist, Dr. Philip Landrigan, wrote that:
The report by Marfella et al. provides evidence that microplastics and nanoplastics are associated with cardiovascular disease outcomes in humans.
And …
…the finding of microplastics and nanoplastics in plaque tissue is itself a breakthrough discovery that raises a series of urgent questions.
Comments. I am not a biophysics expert. I approach this problem as a skeptical, not cynical, appraiser of evidence.
I want to start by saying that it is surely plausible that plastics could have a causal role in atherosclerotic disease. I worry a lot about these exposures. For instance, don’t forget that at the American College of Cardiology meeting in April we will learn the results of the TACT 2 trial of chelation to remove heavy metals in diabetic patients.
The positive signals from the TACT 1 trial were met with great skepticism because heavy metals are not thought of as a typical risk factor for vascular disease. If TACT 2 is positive, not only will we have a new therapy in diabetes, more importantly we may have discovered a heretofore unrecognized cardiac risk.
My point is that the ubiquity of endocrine disrupting chemicals in our lives makes it a potential black swan-like risk. If you told me 10 years from now, that plastic exposure was a health risk, I would not be surprised.
But. This paper should not change our feelings about the risks of plastics. Scientifically, it is a terribly limited analysis. The authors mention some of these limits.
Why the editorial did not mention these surprises me greatly. Where were the editors here? Aren’t editorials supposed to be like the sober judges in the room?
This was a highly selected, small group of patients from Italy. There is no mention about how patients were screened and recruited.
There were an extremely small number of events. I don’t care what the P-value was; there were 2 vs 10 nonfatal MIs. There is a huge chance of noise vs signal here.
Speaking of noise, Table s3, which depicts the regression of variables for the primary cardiac outcome, reveals the fatal flaw. Okay, the authors tell us the presence of plastics yields an HR of 4.5. Almost 5 times the risk that is statistically significant by CIs.
Yet, the presence of known cardiovascular disease does not increase the risk of cardiac events (HR 1.65 with CI that go from 0.8-3.4)? Also, LDL cholesterol is protective in this analysis; HR 0.96 (0.93-1.00).
I am sorry, a 5 times greater risk of cardiac events is way too large an effect size in a study where having heart disease does not increase risk.
The other issue: If plastics are so ubiquitous why don’t all the plaques have detectable plastics? Why only 58%?
The authors describe the possibility of lab contamination, lack of socioeconomic data, and the possibility of residual confounding. But this was the last paragraph of the discussion.
The purpose of critically appraising this paper is multi-fold:
First, it’s a good educational exercise. I learn from doing it. And maybe you do too.
Second, the proposed policy changes scare the heck out of me. I may have missed it, but in the editorial and review article, I did not see much mention about the negative externalities of changing society. Rich people have no major issue changing to glass dishes, but what about the costs of shipping much heavier containers?
And, one of the main plastics exposures is IV tubing. Are people proposing giving less medical care?
Dr. Tracey Woodruff, who authored the review article, concluded by saying clinicians can be critical advocates for policy change to decarbonize and detoxify the economy to address the combined threats of petrochemical-derived endocrine disrupters and climate change.
I find this a shocking leap, from a flawed observational study in 250 Italian patients with carotid disease to a major policy change.
What exactly are doctors supposed to tell policy makers? I have no idea how to decarbonize and detoxify the economy without creating billions of dollars in potential waste and possibly massive societal harm.
Have smart people and journal editors learned nothing from previous policy mistakes? For example, the hospital readmissions penalties, which may be increasing mortality. Or more obvious, the prolonged school closures during the pandemic, which nearly everyone now agrees caused immense damage to young people, especially the most vulnerable in society.
The Italian authors performed a modest study. It’s a start in understanding the potential role of plastics in vascular disease. I’d equate it to the first half mile of a marathon.
To say more about it, to even suggest major policy changes is foolish and, to be honest, a trust-shredder.
The MINT Trial Letters-to-the-Editor and Bayes Theorem
NEJM published two remarkable letters-to-the-editor and perhaps an even more remarkable authors’ response regarding the liberal vs restrictive transfusion strategy in patients with acute MI and anemia trial (MINT).
I present this topic because it is an amazing learning chance, and because I am not sure myself how to think about it.
Recall first that the MINT trial enrolled 3500 patients with acute MI who had anemia.
The liberal transfusion strategy used transfusion for a Hgb less than 10g/dL vs a restrictive strategy that required Hgb < 7-8 g/dL. Obviously, more blood was transfused in the liberal arm. And the mean Hgb was 1.3-1.6 g/dL lower in the restrictive arm.
A primary-outcome event (MI or death in 30 days) occurred in 295 patients (16.9%) in the restrictive-strategy group and in 255 (14.5%) in the liberal-strategy group; that’s 40 more patients or an absolute risk increase of 2.4%.
Here are the statistics: The risk ratio (RR) modeled with multiple imputations for incomplete follow-up, 1.15, or 15% higher. The 95% CI, 0.99 to 1.34; P=0.07).
The two components, death and MI had similar RRS. For death it was 1.19 (0.96-1.47) and for MI it was 1.19 with (0.94-1.49)
Pause there. Higher rates of both composite endpoints. But both with CI ever so slightly below 1 (and 1 means no difference, whereas below 1 means better). The P-value — otherwise known as the surprise nature of the results given the hypothesis that there were no differences between the strategies — was low at 0.07 but slightly higher than the accepted threshold of 0.05.
NEJM had the authors write this conclusion:
In patients with acute myocardial infarction and anemia, a liberal transfusion strategy did not significantly reduce the risk of recurrent myocardial infarction or death at 30 days. However, potential harms of a restrictive transfusion strategy cannot be excluded.
The problem with all this hedging is that when you — the clinician — has a patient with an MI and a Hgb of 8.2, you have to make a decision to transfuse or not. A P-value of 0.07 is one consideration, but so is that there were 40 more primary outcome events in the restrictive strategy. And the upper bound of the 95% CI included a 34% higher rate of death or MI in a month if you don’t transfuse.
That intro brings me to the two letters-to-the-editor that addressed the cloudiness of the results.
One was from two French authors: Drs. Chapalain and Aubron. They wrote about the tension between lack of significance and 15% higher rate of death or MI with the restrictive strategy.
They noted the reverse fragility index represents the number of events that would have changed a nonsignificant result into a significant one.
In the MINT trial, if 2 additional patients among the 1749 in the restrictive-strategy group had had a primary-outcome event — or 2 additional patients among the 1755 in the liberal-strategy group had not had a primary-outcome event — the results would have been significant in favor of the liberal transfusion strategy.
This calculation shows how fragile was the lack of significance.
I like the fragility index, but I will also tell you that purists find it unnecessary, for two reasons: a) because proper understanding of the P-value renders the fragility index obvious, and b) it assumes that reaching 0.05 means anything different from 0.07.
The second letter, from three Missouri authors lead by Dr. Mirza Khan, presented a Bayesian re-analysis of the MINT trial results. Bayesian analysis of trials is just as we do with medical tests, such as stress tests. In a stress test, we multiply the prior belief (low or high risk) by the test results (likelihood ratio) to achieve a posterior belief. A positive test in a patient with super low prior risk may not cause your posterior belief to change regarding the probability of coronary artery disease.
It’s the same in the famous breast cancer mammogram example. What is the probability that a 50-year-old woman with a positive mammogram has breast cancer?
You are told the incidence of breast cancer is 1% (which means 99% do not have breast cancer). The prior then is 1%. The sensitivity of the test is 90% and the specificity is 91%. So that seems like a great test.
Gerd Gigerenzer asked a bunch of doctors this question. Most gave the wrong result; most docs said her probability was 90%, which was way off.
It’s not 90%. It’s actually 10%. Here is why: Tests (and trials) don’t tell you the answer, they update your knowledge or belief. In the case of the mammogram (90% sensitivity and 91% specificity) it gives a likelihood ratio (LR) of about 10. LR is roughly the true positive (or sensitivity) divided by false positive, which is 1 minus specificity. Or using numbers 0.9 for true positives and 0.091 for false positives. Which is about 10.
So, if you multiple the prior 1% incidence or 0.01 x LR of 10, you get about 0.1 or 10%. That was the right answer to the Gigerenzer question.
I present this example to explain why the prior matters so much in how we interpret test and trial results. (I hope you also noted how hard it is to screen for low incidence conditions because even with highly accurate tests, most positive tests are false positives.
Anyways, back to MINT and the letters from the Missouri docs. They offered a Bayesian re-analysis. This allows for a probability of benefit (LR) given the data. They then multiply that LR x three different prior beliefs. A non-informative belief, a skeptical belief, and an enthusiastic belief.
In the non-informative belief, you are basically plotting the results as they are. And when I say plotting, it means drawing a bell curve, like probability density function. This allows you to predict the probabilities of any harm from the restrictive strategy or whatever harm you choose. In this case the probability of any harm (a RR >1) was 94%.
When they chose a skeptical prior, that is a curve centered on a RR toggling around 1, then the probability of harm is less — now it 91%.
However, if they use an enthusiastic prior, that is a belief that restrictive is harmful relative to liberal, the probability of harm goes to nearly 100%.
They write: “In sum, the trial provides evidence that a restrictive transfusion strategy has a high probability of causing harm in patients with myocardial infarction and anemia.”
But the best part of the story is the authors’ response. Basically, they agree. They don’t say this, but one can imply that the NEJM required them to stay in the box regarding strict verbiage stemming from the rules of Null Hypothesis Significance Testing
They say that both the fragility index and Bayesian analysis basically argue for adopting a liberal transfusion strategy for these patients.
Comments. I am torn about this, because to me, MINT showed that the restrictive strategy looked harmful relative to liberal transfusions. Of course, there is a tradeoff because blood products can be in short supply. But from a single patient perspective, it looks beneficial to take a liberal transfusion strategy.
But I am also sensitive to some who argue that Bayesism can be dangerous because you can make non-significant results seem positive. A drug company that does a trial and the results do not make significance, despite proper power, could run a Bayesian analysis and find a 60% to 70% probability that the drug is beneficial.
I am not saying I have the right answer. Only that overly restrictive binary conclusions from trials are fraught. Please do weigh in.
Brugada Syndrome
The European Heart Journal has published a nice observational study of 370 patients with Brugada syndrome (BrS) who had implantable loop recorder (ILR) implantations. This was a multicenter study throughout Europe. They followed the patients for about 3 years. I like the paper; there are no causal conclusions or non-random comparisons.
Instead, there are nice descriptors of what happens to patients with BrS, which is a common question, especially for those of us in US centers where BrS is uncommon and kind of scary.
The authors were careful about making the diagnosis of BrS. They had to meet ECG criteria. Here is a summary of the findings:
The mean age was 43 years, about one-third were female; 60% had proband status and 30% had a family history of BrS. About one-third had a typical SCN5A variant. Half had syncope as a symptom; 22% had palpitations; and 25% had no symptoms.
Thus, these were two groups — those implanted for symptoms and those who had no symptoms.
ILR implantation in symptomatic patients with BrS identified an arrhythmic event in nearly 30% of the patients, with significant clinical implications in 70% of them.
The majority of detected arrhythmias are brady-arrhythmias (BAs) or atrial arrhythmias (AAs), while the rate of sustained ventricular arrhythmias was low.
True arrhythmic syncope is infrequent in patients with unexplained syncope and mostly caused by BAs. Specifically, 77% of BrS patients who had syncope had no arrythmia on the ILR. That is important.
Only 8 of 32 patients with arrythmia and syncope had ventricular tachycardia (VT). Most were BA and AA. The authors add that the yearly incidence of fatal VT in patients with BrS who do not have an implantable cardioverter-defibrillator (ICD) indication is low at 1.1%.
This is important, as the authors discuss, because syncope in patients with BrS often triggers an ICD. But previous studies in patients with ICDs have shown very low appropriate shock rates and substantial rates of inappropriate therapies. This data certainly refutes guideline recommendations to use an ICD because most syncope in BrS in this study had no arrhythmia and the arrhythmia that did occur was mostly not ventricular.
Symptom status can be used to guide ILR implantation. That is because the detection of significant arrhythmias in asymptomatic patients was very low: 89% of ILRs in asymptomatic patients found no arrythmia. Only 4 % found Bas and 7% found AAs, with no ventricular arrhythmias.
The paper is open access and worth looking at. The first author is Marco Bergonti.
The more I look at BrS, the more I worry that we overtreat many of these patients.
© 2024 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Mar 08, 2024 This Week in Cardiology Podcast - Medscape - Mar 08, 2024.
Comments