Nov 12, 2021 This Week in Cardiology Podcast

John M. Mandrola, MD


November 12, 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 on Apple Podcasts, Spotify, or your preferred podcast provider.

In This Week’s Podcast

For the week ending November 12, 2021, John Mandrola, MD comments on the following news and features stories.

Left Atrial Appendage Closure

At the TCT (Transcatheter Cardiovascular Therapeutics) 2021 meeting, Pavel Osmancik from Prague presented the 4-year results of PRAGUE-17, a noninferiority (NI) trial of left atrial appendage closure (LAAC) vs direct oral anticoagulants (DOAC) in high-risk patients with atrial fibrillation (AF).

Recall that PRAGUE-17 was conducted in ten centers in the Czech Republic; 415 patients were randomly assigned and last year the investigators published results of 21-month follow-up. These were 72-year-old patients; CHADSVASC scores, 4.7; and HAS-BLED scores of 3.0. The original trial design was of NI, and the NI margin was 1.47.

The most controversial aspect was the broad composite endpoint that included both efficacy (stroke prevention) and safety (bleeding) endpoints. This creates a serious bias issue because in NI trials, no difference is a positive for the new therapy, and when you have end points you expect to track in opposite directions, they cancel each other out and make NI easier to reach.

That is exactly what happened in the 2-year results: there were 47 primary outcomes in the DOAC arm vs 38 in the LAAC arm and the hazard ratio (HR) of 0.84 had a confidence interval (CI) of 0.53 to 1.31 and since the worst-case scenario of 1.31 was less than 1.47, PRAGUE-17 was declared NI. For the efficacy endpoints — stroke, transient ischemic attack (TIA), cardiovascular death (CVD), and systemic embolism (SE), there were four more events in the drug arm, but in the safety column of device complications and bleeding, there were two more events in the device arm.

In the 4-year results, the rate of primary endpoints was 11.9% in the DOAC arm vs 8.60% in the LAAC for an HR of 0.81 and CI of 0.56 to 1.18, which met NI.

  • There were 13 stroke and SE events in the DOAC arm vs 14 in the LAAC arm. So, no difference.

  • There were 40 bleeds in the DOAC arm vs 29 in the LAAC arm.

  • There were 0 procedural complications in the DOAC arm vs 9 in the LAAC including 2 procedural deaths.

The authors concluded that, “In long-term follow-up of PRAGUE-17, LAAC remains non-inferior to NOACs for preventing major cardiovascular, neurological or bleeding events. Furthermore, non-procedural bleeding was significantly reduced with LAAC.”

Comments and Translation. The authors’ first conclusion is technically true as they defined the NI criteria, but the same existential criticism applies: you should not test NI this way. NI is only used if a new therapy offers some advantage (safety, convenience, cost) over a new therapy. LAAC may do that, with its purported (but yet unproven) ability to get patients off AC and have less bleeding.

The way would you test this is the same way DOACs were tested against warfarin. You test NI in the efficacy endpoint of stroke — only stroke, not CV death. Then you test superiority in safety, which in the DOAC vs warfarin trials was bleeding, but in the DOAC vs LAAC trial, safety would have to be bleeding plus procedural complications.

If LAAC was superior to DOAC in safety and not worse in stroke prevention, you’d have a winner. You clearly don’t have that here.

  • In the LAAC arm, the total bleeds of 29 plus the nine procedural complications is 38 events, nearly the same as the 40 bleeds in the DOAC arm.

  • For efficacy you have basically the same number of stroke events in both arms.

In the accompanying editorial, Faisal Merchant from Emory outlined other important critical appraisal points:

  • First, he pointed out that device-related thrombus and peri-device leaks are an Achilles heel for LAAC. These can present late after the implant and, when present, increase the risk of stroke and require AC, which defeats the purpose of the device.

    • In PRAGUE-17, the planned transesophageal echocardiographic assessment at 6 and 18 months were not done because of COVID, which is no knock on the authors. Merchant’s point was that this was a lost opportunity to learn more about these important headwinds for LAAC.

  • Second, Merchant noted that 13% of those in the DOAC arm stopped AC, some for bleeding, others for unknown reasons. But also 9% of patients in the LAAC arm were treated with oral AC at some point, some for device related thrombus (DRT) or peri-device leak (PDL), others for unknown reasons.

    • This is also an existential problem for LAAC because the whole point of the device is to provide stroke prevention without AC. If you start an AC in a patient with a device, this defeats the purpose of taking the upfront procedural risk.

    • What’s more, consider that many patients have LAAC precisely because they bleed on oral ACs. If they then develop a DRT and have to take an AC, you have put them in an even more precarious position then they were in before.

  • Third, Merchant also noted another problem with the composite endpoint: asymmetry. In PRAGUE-17, a 2 gm drop in hemoglobin is given the same weight as stroke.

In the end, you must read beyond the headlines and abstracts. PRAGUE-17 shows that the device does not reduce stroke events, and if you combine the upfront risks plus bleeding, there is no safety benefit.

It takes a mighty effective therapy to overcome an upfront major complication risk of 5% (in real world data, it approaches 8% to 9%), and thus far, LAAC does not come close. Unless there are incredible iterative gains, I remain unshaken in my belief that this procedure, percutaneous LAAC, as it is practiced now, will remain one of our field’s greatest mistakes.

Transcatheter Aortic Valve Replacement – SURTAVI

Also at TCT, we learned the 5-year results of the SURTAVI trial of self-expanding transcatheter aortic valve replacement (TAVR) vs surgical AVR (SAVR) for intermediate risk patients with symptomatic aortic stenosis (AS).

Recall that the original SURTAVI results, published in NEJM in 2017, found that the rate of the primary outcome of stroke or death at 2 years was 12.6% with TAVR vs 14% with SAVR. These met NI. Surgery had more acute kidney injury (AKI), AF, and transfusion requirements but TAVR had more pacemakers and residual aortic insufficiency (AI). A word on patients—the mean age was 80 years.

  • At 4 years, the Kaplan-Meier (KM) curves for stroke or death were nearly super-imposable: 31.3% for TAVR vs 30.8% for SAVR.

  • There were slightly lower rates of stroke in the TAVR arm but it did not reach significance.

  • New pacemaker rates: 35% for TAVR, 15% for SAVR—which was highly significant.

  • VA slightly better with TAVR.

  • More AI with TAVR.

  • Kansas City Cardiomyopathy Questionnaire: No difference in long-term follow-up.

  • The conclusions were that over 5-years, hard outcomes were similar.

Comments. I agree with the authors’ conclusions. The KM curves with this self-expandable valve are quite similar to the transfemoral cohort of PARTNER 2A. However, there are some internal validity concerns. There were many more withdrawals and loss to follow-up from the trial in the SAVR arm — about 27% withdrawn or lost in the surgical arm vs 11% in the TAVR arm. I am not sure I heard an explanation for these disparities. There is likely nothing nefarious here, but when the paper is published, I wonder if there will be some discussion of how differential withdrawal and lost to follow-up effects confidence in the results.

Nonetheless, taking the results as they are, we can say that:

For 80-year-old patients similar to those enrolled, the primary outcomes look quite similar, and since TAVR is less invasive, this seems a choice many patients would make.


  • For me, the entire issue of TAVR vs SAVR is the translation to younger patients. One in three patients in the TAVR arm ended up with a pacemaker. That’s not a big issue for an 80-year-old but it is for younger patients.

  • The presence of AI is also not likely an issue for an octogenarian, but it is for younger patients.

  • Then there is matter of longer-term outcomes and valve durability. A 70-year-old with aortic stenosis has to care much more about longer-term data including valve durability.

Just so you know it’s not some random podcaster electrophysiologist with these concerns about translation, in our news story, Steve Stiles cites the esteemed Harvard cardiologist and former American College of Cardiology (ACC) president, Patrick O’Gara who said:

“We need to be really cautious in translating these outcomes to 'younger' patients at lower risk. Of the problems that we face, indication creep is close to the top of the list...I think it's not really fair for us to say 60-year-old patients will fare equally well, or that the trade-off between a pacemaker implant and the need for a repeat procedure is perceived the same in folks who are 15 years younger.”

Renal Denervation

There were two notable studies on renal denervation to discuss this week. First, a tiny bit of background. The idea behind renal denervation (RDN) is that, by blocking neural activity from the kidney, you reduce blood pressure (BP) via a number of mechanisms.

Initial enthusiasm for RDN stemmed from non-controlled studies. But then a sham-controlled study called SYMPLICITY HTN-3 found no significant effect. Now, though, through iterations of the denervation device and procedure, and more careful control of medications, RDN is making a comeback. I’ve previously discussed the RADIANCE-HTN series of trials. In May of this year, Lancet published the 2-month results of RADIANCE-HTN-Trio study in which patients with resistant hypertension (HTN) were given a three-drug polypill.

  • More than half of “resistant” patients were no longer resistant when given the polypill. Those that remained HTN were randomized to RDN or sham procedure.

  • In blinded follow-up at 2 months, RDN reduced daytime ambulatory BP by about 4.5 mmHg.

At TCT, we heard results of a second phase of the study, from months 2 to 6. This was a pre-specified extension of the main study, and it was clever. Patients and doctors were still blinded.

So those who remained hypertensive were given, via a protocol, stepwise intensification with doses of first spironolactone and then bisoprolol. The three main findings were that

  • The aldosterone antagonists were used in 40% of those in the RDN vs 61% of those in the sham-control arm.

  • The sham-control arm patients were on disproportionately higher dosages of spironolactone.

  • Home measured systolic BP fell by 4 mmHg more in the RDN group despite using less add-on meds, though there were no differences in office BP.

This study has not been published and we will need to see the paper, but it looks like there is a modest additive effect of RDN out to 6 months. To me, the main take-home can be summed up in one modifier: “modest.” Still, nearly half of those who had RDN still required spironolactone. And the difference in BP is modest. Both groups get a huge effect from spironolactone, which is a heck of lot less pricey and invasive than a procedure requiring femoral artery access.

A Meta-analysis. JACC-Cardiovascular Interventions published a second notable study. This was a meta-analysis of the seven sham-controlled published RDN studies, with the goal of informing the effect of RDN in patients taking medications and in those not taking medications.

  • The seven trials included about 1300 patients and different catheters, different types of patients.

  • The overall mean effect on ambulatory systolic BP was -3.61 mmHg and there was no significant heterogeneity.

  • The overall mean effect on office BP was a bit higher at 5.9 mmHg—also no heterogeneity.

  • When separating out trials that included patients taking meds vs those off meds, the reduction in ambulatory systolic BP was similar at slightly more than 3 mmHg.

  • There was no evidence that the use of concomitant antihypertensive medications had a significant impact on the effect of denervation on ambulatory systolic BP.

  • There was also no effect of first vs second generation catheters.

  • Complications were low.

The authors concluded that RDN “...could be helpful at various points within the overall strategy of BP management, including as an alternative to additional medications for patients with resistant hypertension already taking several medications or as a first-line therapy for drug-naive patients with hypertension.”

Comments. The obvious and main point was that the extremely modest reduction of 3 to 6 mmHg of systolic BP occurred at 4.5 months. BP management is for years, not months. While proponents will say RDN is always on; for example, you don’t need to take pills. But I would counter that we don’t know whether the modest reduction in BP will last.

And sadly, the authors point out that we aren’t likely to find out because trials require a strict protocol of medication and limited clinician freedom to make dose changes. Such strictness is necessary to make precise estimates of the effect size. “The way to truly assess long-term effects of denervation on BP would be to have patients undergo a double-blinded placebo-controlled procedure and be willing to remain blinded in the long term. Such an approach is impractical.”

Pause there and think about the implications of the cost efficacy of this procedure. It might be very hard to get a handle on that without knowing long-term data. David Kandarzi wrote the accompanying editorial, and he emphasized the difficulty in pooling such different studies, because of things like variance in BP criteria, different procedural techniques, follow-up duration, and trial conduct. He then points out that the benefits of BP turn on lots of things: the initial severity of BP, the underlying CV risk and co-morbidity, and time within treatment range.

His next point is key: we need the ability to identify predictors of response. This is going to be important because there are millions of patients with HTN who, if offered a procedure rather than lifelong pills, would jump at the chance. I think we need a lot more data on RDN, which is being gathered. Kudos to the researchers.

These early studies show a modest effect that seems present on or off meds. There is a biologic effect that lasts at least 4 months. But In addition to the knowledge gained from further studies I also wonder if this would be a classic case of looking for heterogeneity of treatment effects. Do age, baseline BP, renin levels, etc matter? To contain the costs of this procedure we must find a sweet spot of who benefits most.

AHA Preview

The American Heart Association (AHA) meeting will happen this weekend. Both Patrice Wendling and I have written previews. The meeting is virtual, which is understandable given its large size and international nature. A number of notable trials will be presented.

Leading off is the AVATAR trial, the second RCT looking at watchful waiting vs SAVR in patients with severe but asymptomatic AS. Last year, the South Korean RECOVERY trial found surgical therapy far superior. If AVATAR is positive, the dogma of waiting for severe AS to cause syncope, heart failure or chest pain will be over. One external validity or generalizability question is, if the trial is positive, will it also apply to TAVR since both trials studied SAVR vs watchful waiting?

Another surgical trial will assess the benefits of tricuspid repair at the time of mitral valve surgery. It’s controversial whether or not added work on the tricuspid helps patients. The primary endpoint of this trial is tricky — it is a composite of death, reoperation for tricuspid valve regurgitation (TR), or the progression of TR from baseline by two grades, or the presence of severe TR at 24 months. I hope you notice the problem: that latter component, the degree of TR, is a soft surrogate endpoint, one that is tough to estimate. There will be numerous papers on AF. Greg Marcus from UCSF is on the docket with two elegant trials, one on caffeine and AF. That group is doing superb work. There is another trial on curbing post-cardiac surgery AF.

And finally we get to hear the results of the AMAZE trial of epicardial LAAC with the LARIAT device as an add-on for AF ablation. The company announced that the trial missed its primary end point but I am looking forward to the data.

My colleague at Medscape, F Perry Wilson from Yale will present an intriguing trial called REVEAL AF, which looks at the value of alerting clinicians about the prognosis of heart failure (HF) patients. I am drawn to this trial because I think the prognosis of HF patients is both under- and overestimated. So much emphasis is placed on guideline-directed medical therapy that sometimes I think we underuse arrhythmia and palliative care consultants There can be no cardiology meeting these days without SGLT2 inhibitor trials. AHA will feature two more. And Marc Sabatine of Brigham will present the TIMI group’s independent analysis of the left main percutaneous coronary intervention vs coronary artery bypass graft trials. That will be an attention getter. | Medscape Cardiology will be there to cover the meeting. These are great journalists doing important work.


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.