Mar 19, 2021 This Week in Cardiology Podcast

John M. Mandrola, MD


March 19, 2021

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.

In This Week’s Podcast

For the week ending March 19, 2021, John Mandrola, MD comments on the following news and features stories.


In the United States, the burden of COVID continues to decrease. Hospitalizations are down and the slope of the descent seems to be steepening, though one cautionary note is the state of MI, which has seen a rise in cases and hospitalizations.

The European Union is a different story. Cases are going up and my friend in France tells me that COVID-related ICU admissions are also increasing. Brazil also has increasing cases. Worse, some EU nations have paused vaccinations because of reports of clotting events after the Astra Zeneca shot. Twitter and the media were awash with stories emphasizing that correlation is not causation. In other words, you have to accept that there will always be a low-level baseline rate of clotting events in a population of millions of people who have received a shot.

Whether the difference in the EU vs US COVID situations is due to vaccine rollouts, variants, or just the stochastic nature of the virus, I do not know. I do know that, as this pandemic winds down, there will be a great tension in the dispersion of risk tolerance in society. My two bold predictions: 1) Our transition to an endemic SARS COV-2 is not going to be smooth. 2) In learning to live with a new respiratory pathogen, social media will be a net negative.

On the upside: My granddaughter went back to second grade this week. She was so happy to see her friends. It’s still only 2 days per week, but it is progress.

Fish Oil and AF

This week JAMA published the results of the VITAL-AF study which enrolled 25,000 participants and studied the question of whether marine omega-3-fatty acids and/or Vitamin D reduces atrial fibrillation (AF). The type of fish oil was an EPA-DHA combination. Recall that in the REDUCE-IT trial, it was pure EPA. For the fish-oil vs placebo comparison, incident AF events occurred in 3.7% vs 3.4% participants, respectively. Similarly, in the Vitamin D vs placebo arms there was no significant difference.

A notable observation on fish oil comes in the accompanying editorial from JAMA deputy editor Dr. Gregory Curfman. He lays out the strong case that fish oil may actually increase incident AF. In the STRENGTH trial, 4 grams per day of EPA/DHA fish oil was compared with corn oil for the reduction of CV events. CV events were not sig different but there was a 69% increase in risk of developing AF in the fish oil group. In the REDUCE-IT trial, 4 grams per day of purified EPA called icosapent ethyl (IE) was compared with mineral oil for the reduction of CV events. IE reduced CV events by 25% but there was a 35% increase in risk of AF with omega-3 fatty acids compared with mineral oil. In the OMENI trial of older patients with recent MI, 1.8grams per day of EPA/DHA combination was compared to corn oil for CV event reduction. There was no significant reduction of major adverse cardiac events (MACE) but again, AF incidence was 84% higher: 7.2% vs 4.0% developed AF.

With these large RCTs as priors, now consider the results of VITAL AF, which I just told you had a 9% increased rate of AF in the fish oil group compared with placebo. Recall also that VITAL AF used less than a gram per day of fish oil. On its own this 9% increase in AF wasn’t significant, but taking the evidence as a whole, Curfman argues that fish oil may exert a dose dependent increase in AF.

I agree, and I have been deprescribing fish oil like crazy in the clinic. Eat fish not fish pills seems a worthy slogan.

Ertugliflozin and Renal Outcomes

You all know I am quite positive on SGLT2 inhibitor drugs for the reduction of cardiovascular disease (CVD) complications in patients with diabetes (DM), for reduction in renal outcomes in patients with chronic kidney disease (CKD) and DM, and for the reduction of CVD and hospitalizations for heart failure (HHF) in patients with HF with reduced ejection fraction.

You also know there are numerous SGLT2 inhibitor drugs (empaglifozin, canagliflozin, dapagliflozin, etc) and when considered together, there sure seems to be a class effect. That said, we will soon see great marketing pressure to use one or the other drug. Our Pharmacy and Therapeutics committee has had robust debates on which SGLT2 inhibitor to have on formulary.

In December of 2020, the New England Journal of Medicine published a CV outcomes trial called Vertis CV for the SGLT2 inhibitor drug ertugliflozin. Vertis CV participants had DM and atherosclerotic CVD. The primary objective was to show noninferiority of the ertugliflozin to placebo in the primary endpoint of MACE—CVD, stroke, or myocardial infarction. There were also three glycemic substudies listed as primary objectives. Nearly the same rate of MACE occurred and the upper bound of the hazard ratio (HR; 1.11) and this was less than the noninferiority margin of 1.2, so it met noninferiority.

One of the secondary objectives was to look at the renal composite outcome of death from renal disease, renal replacement therapy (RRT), or doubling of the serum creatinine. This was the renal outcome that the trialists prespecified in the trial protocol and listed on Clinical—they also listed 118 total secondary endpoints!

In the main trial, the rate of this renal composite in patients who were in the ertugliflozin group was 3.2% vs 3.9% for those on placebo. This corresponds to a 19% reduction, HR = 0.81; but the 95% confidence interval (CI) ranged from 0.63 to 1.04; or said another way, the 95% CI ranged from a 37% reduction to a 4% increase in risk outcomes. Since the upper bound passed 1.0, the reduction is not felt to be statistically significant. Not only that, since the first key secondary outcome (CVD and HHF) was not statistically significant, “in accordance with the prespecified hierarchical testing procedure, further statistical testing of other outcomes was not performed.”

But now the journal Diabetologia has published a new analysis of the data from VERTIS CV investigators. For this analysis, the authors used a different renal composite endpoint. They swapped out the time to doubling of creatine and swapped in a sustained 40% reduction from baseline estimated glomerular filtration rate (eGFR) into the renal composite secondary endpoint. And voila: using this new composite—renal death, RRT, and 40% eGFR reduction—the event rates were 6% vs 9% and the HR was 0.66 with a CI that ranged from 0.50-0.88, which is a much larger effect size with a tighter CI.

While the authors write “all analyses reported here were pre-specified, either in the statistical analysis plan of the trial or in a separate analysis plan completed prior to database lock and unblinding,” keep in mind that the renal outcome was a secondary endpoint. And they also prespecified 118 other secondary endpoints. So, this is a reanalysis of a secondary endpoint with a swapped in different endpoint.

In the article, the authors cite many reasons why a 40% reduction in eGFR is a reasonable endpoint. For example, other similar trials have used this measure. But I find this is very dubious analysis. As it is, a secondary endpoint is considered exploratory; now you have an exploratory reanalysis of an exploratory endpoint.

I don’t think this is “p-hacking” (or selective reporting) because I don’t believe the researchers performed hundreds of different comparisons and picked ones that were statistically significant. With more than 100 secondary endpoints, it seems more like degrees of freedom. Statistician Andrew Gelman calls this a ‘garden of forking paths.’ “The mistake,” he writes, “is in thinking that, if the particular path that was chosen yields statistical significance, that this is strong evidence in favor of the hypothesis.”

The main point in discussing this analysis is not to dismiss the potentially beneficial effects of ertugiflozin on renal outcomes (and you can see from the original HR and 95% CI spanning 0.66-1.04 that this is likely a true effect) but to oppose in the strongest terms an after the fact (or post-hoc) reanalysis of data with a different endpoint swapped in. The authors cover themselves because they describe the problem with post-hoc analyses in the limitations paragraph, but this doesn’t stop them from concluding: “Among individuals with type 2 diabetes and atherosclerotic CVD, ertugliflozin reduced the risk for the pre-specified exploratory composite renal endpoint.”

Here is the problem with this: a major diabetes journal published it. It’s “peer-reviewed!” Reprints, Power Point slides, and info cards can be made with this more impressive risk reduction in renal outcomes. These marketing devices can then be shown in expos at future meetings, and along with burritos in unsuspecting doctors’ offices. It oversells the evidence and that is not a good thing.

Transcatheter Aortic Valve Replacement (TAVR) vs Surgical Aortic Valve Replacement (SAVR) for Patients With Aortic Stenosis (AS). | Medscape Cardiology features a 20-minute discussion between two experts in TAVR and SAVR, interventional cardiologist Wayne Batchelor and surgeon Tom Nguyen. I highly recommend this video. They discuss many of the issues surrounding this decision and include a brief synopsis of trial data, discussions on bicuspid valves, valve iterations, long-term durability, and the super-relevant discussion about heart team approach and tension between interventional cardiologists and surgeons.

Dr. Nguyen mentioned the catch-up phenomenon in the PARTNER 3 trial of low-surgical-risk aortic stenosis TAVR vs SAVR. That is: at one year the results strongly favored TAVR but then the curves for the primary endpoint of stroke, death, and hospitalization started coming together.

I’ve discussed this topic before, but it is worth a brief review: First, I consider myself an unbiased neutral observer. I don’t do either procedure. I do see a lot of these patients because of the electrical issues. Second, I believe TAVR has been an amazing technology, especially for older and intermediate to higher operative-risk patients. And it is still iterating.

Here are eight things that worry me about rapid acceptance of TAVR in younger lower-risk patients.

  • PARTNER 3 was the only trial using rehospitalization as a primary endpoint (PE).  All others used stroke and death. Adding hospital admissions in the PE was highly dubious because it biased in the short term against SAVR. No explanation has been given for this outlier. At the European Association for Cardio-Thoracic Surgery (EACTS) meeting, when I asked a trialist, he shrugged his shoulders and offered, because the Food and Drug Administration allowed it.

  • Relative to the number of PE events in the medium and high-risk trials, there have been smaller numbers of hard endpoints in the low-risk trials. Fewer events means more uncertainty. For example, 90% of the EVOLUTE-low-risk data was imputed. At the time that trial was published, only 10% of the patients had had 2 years of follow-up.

  • If you include both types of transcatheter valves (self-expanding and balloon-expanding) there is a higher rate of pacemakers. This is less of an issue in older patients but a huge issue in younger patients. Dr. Joseph Bavaria has shown data from real-world Society for Thoracic Surgery registry that the rates of pacing after TAVR has not decreased appreciably in the last 5 years.

  • Bavaria shows from the same registry that what he calls TAVR disasters have not decreased significantly over time.

  • In PARTNER 3 the rate of left bundle branch block (LBBB) was more than two-fold greater in TAVR (20% vs 9% in SAVR). LBBB is not a good thing to give to young patients.

  • In PARTNER 3, the rate of valve thrombosis was 2.6% in TAVR vs 0.7% in SAVR. This could influence anticoagulation decisions and is a bigger deal in younger patients. Whenever you look at longer-term data TAVR trends worse.

  • In the intermediate risk group, the 5-year results of PARTNER 2, just including the transfemoral cohort, the Kaplan-Meier curves for stroke and death after 2 years are seriously diverging against TAVR.

  • And a meta-analysis of 6 TAVR vs SAVR trials from Barilli and colleagues finds a time-varying effect on mortality. TAVI is related to better survival in the first months after implantation whereas, after 40 months, it is a risk factor for all-cause mortality.

Open heart surgery is clearly a bigger deal in the short-term, so many patients would still choose TAVR even with a slightly higher rate of death or stroke at 5 years. But not all would, and as Drs. Batchelor and Nguyen highlight, a thorough discussion with patients is crucial.


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.