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 October 27, 2023, John Mandrola, MD comments on the following news and features stories.
Feedback
CVD Risk Reduction. First, a brief addendum to my report last week of the Million Hearts Model randomized controlled trial (RCT).
The idea of the Million Hearts Model is good. It is to pay health care organizations to assess and reduce cardiovascular (CV) risk. At least initially it was focused on the highest risk patients.
This was an RCT in which organizations were randomly assigned to standard care vs the Million Hearts Model of incentives to measure and lower CV risk. The primary endpoint was a major adverse cardiac event (MACE)(. These were primary prevention patients.
I need to clarify some of my coverage. After This Week in Cardiology aired, I spoke at length about this study with my friend and colleague, Andrew Foy, from Penn State. And he taught me some important nuances about this very complicated study.
Last week I said that the program did not pay for achieving anything. Only to measure risk. I liked that. But that wasn’t quite right. Million hearts did pay to measure risk — using risk calculators, which I feel are underused, at least in my neighborhood.
But they also paid extra for reducing risk. At least for part of the study:
“CMS (Center for Medicare and Medicaid Services) made performance-based risk-reduction payments: CMS paid each organization $0, $5, or $10 PBPM (per beneficiary per month) for each high-risk beneficiary with an annual risk reassessment, with monthly payment amounts dependent on mean risk score change across all of the organization’s high-risk beneficiaries reassessed.”
That’s an important methodological point, right? Because this creates an incentive, a nudge, if you will, for clinicians to convince patients to take a statin or more meds for hypertension (HTN).
I say it nudged for more medication because, in the risk equations, the only modifiable risks are blood pressure (BP) and cholesterol, at least for non-smokers. Hence more pills.
The thing about primary prevention is it ought to be largely concordant with a patient’s goals. If there are rewards for getting patients to have lower LDL and BP, then the doctor is motivated to treat more aggressively, which is fine if the patient wants it, but some patients don’t care about the extra risk lowering. Maybe it’s not worth it for them.
Andrew Foy also made two important points about the results.
Yes, there was a relative risk reduction in organizations that had were randomly assigned to the intervention group. Specifically, it was a 3.3% lower rate of CV events in the intervention arm vs the control arm (adjusted hazard ratio [HR], 0.97 [90% confidence interval (CI), 0.93 - 1.00]; P = .09)
But Foy also taught me to study Table 3 of the results more closely. Two teaching points arise:
One is that the benefits stemmed from medium-risk patients. Not high-risk patients. Recall that Million Hearts was initially going to measure effects on high-risk patients, but there weren’t going to be enough events, so they expanded to medium-risk patients.
I used to think the idea that medium-risk patients were more likely to benefit was counter-intuitive, but Foy’s research has shown that, oftentimes, the higher-risk patients benefit less from CV therapies — probably because of competing risks, and/or increased risk of harms.
The other important point from Million Hearts was also in Table 3: Health Care Utilization. Specifically, all-cause hospitalization, which was higher in the Million Hearts group vs standard care. The P-value was in fact the lowest of any in the study: 0.005. And again, the higher rate of all-cause hospitalizations was driven mostly by high-risk patients, though both were higher.
Why is that important? Because all-cause hospitalization is a pretty unbiased endpoint — you are admitted to the hospital or not. So, this tells that the extra treatment in the Million Hearts model may have led to a very small reduction in CV events, but at the cost of more total hospitalizations.
Foy believes this study is a wash for the Million Hearts intervention. One final caveat, the intervention group did have a 4% lower all-cause death rate. But I think this is likely statistical noise, because a) there are small numbers of events, and b) when you go to the supplement, there are large components of non-CV death that are lower in the intervention group.
Bottom line, this was a complicated study, that doesn’t convince me that nudges at the organization level have any great impact. Especially on higher risk patients.
ENRICH AF Feedback. Wow, I am getting a lot of notes from big investigators. At 3:00 am Friday, just hours before I taped this podcast, I heard from the ENRICH AF investigators.
Recall that last week I covered a meta-analysis of four trials looking at restarting oral anticoagulation (OAC) after an intracerebral hemorrhage (ICH).
I was struck by the quite positive results in that the primary thrombosis outcome was 32% lower in the group of patients restarted on OAC and overall major bleeding was not significantly different. Also not different was an endpoint of death or dependence.
It made me think that this blanket idea that if you had an ICH you can never take an OAC again is foolish.
Also relevant to the discussion was a trial called ENRICH AF, which was not included in the meta-analysis because it is still ongoing. The authors of this study of edoxaban vs no edoxaban in patients with atrial fibrillation AF and an ICH history published a letter in the Lancet announcing that an early review of the data found that patients with lobar intracerebral bleeding and convexity subarachnoid bleeding will stop receiving edoxaban because of excess bleeding. I had said the trial was terminating.
The authors clarified that ENRICH AF will continue to enrollpatients with qualifying ICH (i.e., other than lobar intracerebral hemorrhage and convexity subarachnoid hemorrhage). Based on their current predictions, the study results will be available in the first half of 2025.
ENRICH AF is important because it is the largest trial of OAC in patients with AF who survived an ICH. The trial has already enrolled 750 patients — that is more than the meta-analysis I mentioned.
TAVI vs SAVR
The annual Transcatheter Cardiovascular Therapeutics (TCT) meeting provided a lot of new data on the transcatheter aortic valve implantation (TAVI) vs surgical aortic valve replacement (SAVR) issue, this time in low-surgical risk patients with severe aortic stenosis (AS). I will cover two big trials — the PARTNER 3 5-year results and EVOLUT-Low-Risk (LR) trial 4 year results.
PARTNER 3. This was TAVI with the balloon-expandable valve vs SAVR in low risk patients. The primary endpoint, which was different than all other primary endpoints in TAVI trials, was stroke, death, and re-hospitalization at 1 year. They did both noninferiority (NI) testing and superiority testing at the time. The trial enrolled about 1000 patients, mean age, 73 years.
The original 1 year results strongly favored TAVI, 8.5% vs 15.1% for primary outcome events. This was largely driven by the rehospitalization rate, which of course would favor TAVI, as surgical patients are expected to have a more difficult short-term course.
At 5 years, these results are a bit different. The primary endpoint remained. There was also second primary endpoint — a hierarchical composite that included death, disabling stroke, nondisabling stroke, and number of rehospitalization days, analyzed by the win ratio.
I am not sure why, but the authors analyzed the primary endpoints this time with superiority testing. Was TAVI superior was now the question, whereas in the original trial it was NI.
Now, for the primary endpoint, rather than 8.5% vs 15.1% for TAVI vs SAVR, as it was in year 1, it was 22.8% vs 27.2%, TAVI vs SAVR. The 6.6% absolute risk reduction was decreased to 4.4% after 5 years. This difference no longer met statistical significance for superiority.
The nonsignificance also held true for the second primary endpoint; the hierarchical win ratio was 1.17 with CI going from 0.9-1.5 and a high P-value of 0.25.
Components of the primary endpoint require comment.
Death was now higher in the TAVI arm, 10% vs 8.2%.
Stroke was a bit lower in TAVI, 5.8% vs 6.4%.
Hemodynamic performance and the rate of bioprosthetic valve performance were similar.
Although this was concerning:
Clinically significant valve thrombosis, according to VARC-3 criteria, occurred in 12 patients (2.5%) in the TAVI group and in 1 patient (0.2%) in the surgery group.
A note on the death analysis: There was a clear crossing of Kaplan Meier (KM) curves. While TAVI clearly had an early advantage, at 1 year the curves show gradually separating benefit with SAVR.
When you look at what’s called a landmark analysis, starting at year 1, there is a 61% higher death rate with TAVI vs SAVR. It’s not quite significant statistically but the curves are remarkable.
Authors Conclusion:
Among low-risk patients with severe, symptomatic aortic stenosis who underwent TAVR or surgery, there was no significant between-group difference in the two primary composite outcomes.
Comments. First the good, then the uncertain, then the critical.
Good: TAVI trialists have done a nice job studying this new technique. I wish all aspects of cardiology had this degree of evidence to analyze. The TAVI trialists have also stimulated more research on surgical valves, which was sorely lacking pre-TAVI.
Good: TAVI looks quite decent at 5 years. Mean age of patients was mid-70s. Not having your chest opened and getting these results is somewhat compelling.
Uncertain: These are bioprosthetic valves; 5 years is good, but you really start seeing things with bioprosthetic valves at 8 to 10 years. The higher rate of valve thrombosis in TAVI is worrisome, as are the separating death curves. We still need more data.
Uncertain: One piece of data that is there but not published is the type of surgical valve. Some are better than others. I wonder if some of the bad outcomes in the SAVR arm are clustered among surgical valves that are no longer used. It would be nice to see that data. If the data were open, we’d be able to look at that. See discussion of EVOLUT-LR.
Critical: The conclusions as written in the New England Journal of Medicine are wrong. You don’t say there were no significant differences in a superiority trial. You say TAVI failed to meet superiority over SAVR at 5 years. Saying it the first way may be true, but it is not true to the science. Saying it the first way spins the results to the side of TAVI.
Critical: The choice to add re-hospitalization in the primary endpoint of PARTNER 3 is sort of unbecoming. Unbecoming because it obviously biases against SAVR and everyone knows it. All the other trials measured stroke and death. It tempts neutral people like me to think cynically.
Critical: The death curves are separating. Yes, while many patients may not care about a ≈ 2% absolute higher rate of death at 5 years, many would care. There are plausible reasons why this could be true: valve thrombosis, pacemakers, etc. In a perfect world, patients would be shown these landmark analyses as part of their informed consent. I am not sure that will happen.
EVOLUT LR. This trial was similar in some ways to PARTNER 3. TAVI vs SAVR. TAVI was with the self-expanding valve. It enrolled 1400 patients, mean age 74 years. Low surgical risk. Primary endpoint far better, disabling stroke and death.
The original trial was published in 2019, with 2-year results, though many patients had not reached 2 years follow-up so there was a lot of imputing of data.
Bottom line: For the primary endpoint, it was 5.3% in the TAVI group and 6.7% in the SAVR group (posterior probability of noninferiority > 0.999).
The EVOLUT LR authors have published yearly results. At TCT, we heard the 4-year results in the form of a research letter published in the Journal of the American College of Cardiology (JACC).
The main results for the primary endpoint at 4 years was 10.7% in the TAVI arm vs 14% in the SAVR arm. That’s an HR of 0.74, a 26% reduction; CI, 0.54 to 1.00 and the P-value is exactly 0.05.
The KM curves are totally different than in PARTNER 3. In EVOLUT LR, the differences seem to be getting better over time with TAVI. The ARR with TAVI at 4 years is 3.4% vs only 1.8% at 1 year.
Rates of the primary endpoint components:
Death: 9.0% TAVI vs 12.1% SAVR (P = 0.07).
Stroke: 2.9% TAVI vs 3.8% SAVR (P = 0.32).
New permanent pacemaker implantation was significantly higher in the TAVI group: 24.6% vs 9.9%; P < 0.001.
Clinical or subclinical valve thrombosis: 0.7% TAVI vs 0.6% SAVR; P = 0.84.
Aortic valve performance looked a bit better in the TAVI arm (less mean gradients) but surgical valves had less aortic insufficiency.
Authors conclusions:
TAVI patients in the Evolut Low Risk trial continue to show excellent outcomes for the primary endpoint and significantly better hemodynamics than SAVR patients through 4 years.
The difference in all-cause mortality or disabling stroke seen early in the Evolut Low Risk trial continues to diverge in favor of TAVI, with a 26% reduction in hazard for death or disabling stroke through 4 year.s
My Comments. The authors note a very large and differential amount of loss to follow-up in the surgical arm. This is similar to other TAVI trials, but it’s nearly double for the surgical group. Totaling 11% of all surgical patients. This adds quite a lot of uncertainty because the primary outcome had a P-value of 0.05 and it would not take many events to push that to not significant.
I also wonder what is different about these patients? I say that because surgeon Victor Dyan noted that the 4-year death rate in the surgical arm of EVOLUT LR is 12% vs only 5.6% in PARTNER 3. That’s curious and suggests something about patient selection in these two trials.
Yet, the neutral view of EVOLUT LR is that it looks quite promising. I agree with the authors that we need more data. I don’t know why the TAVI death and stroke curves are looking better over time for EVOLUT LR and worse for PARTNER 3. If you do, let me know.
The higher pacer rate in EVOLUT really worries me. Pacemaker complications increase over time. So let’s see at 5, 8, and 10 years.
Another area of uncertainty is the surgical valves. I went back and looked at the 3-year results of EVOLUT LR, published early in 2023. Three of the re-interventions in the SAVR arm were for a TRIFECTA valve which some surgeons tell me is considered inferior and off the market. So, it makes me wonder about an analysis of events by surgical valve in the SAVR arm. I realize that both groups have valves that iterate, so I am not saying we should make too much of this, but it is curious to me that the trial compares one type of valve (self-expanding TAVI) to a hodge-podge of surgical valves.
The other thing I wonder about is whether we should have a head to head comparison of self-expanders vs balloon expanders in patients with suitable anatomy for both?
Myocardial Viability
If myocardial viability as a medical test is not dead, it is surely actively dying.
One of my favorite trials, the REVIVED-BCIS trial has reported a prespecified secondary analysis.
REVIVED BCIS is an awesome trial—it shredded the common dogma that patients with heart failure should have coronary angiography to find coronary artery disease (CAD) so that CAD can be fixed and outcomes improved.
REVIVED BCIS took patients with a low ejection fraction (EF;27%), ischemic cardiomyopathy, and extensive multi-vessel CAD that was amenable to PCI and myocardial viability and randomly assigned them to either percutaneous coronary intervention (PCI) or tablets.
Imagine the angiograms; all those perfectly PCI-able lesions. The optimal medical therapy (OMT) group had them left alone.
Boom. At the end of 3.5 years, there was not a shred of difference in the primary outcome of death or hospitalizations for heart failure (HHF).
You should put this study in your phone, Save it. Study it. One of the most important trials of our time. It’s unassailable — or at least I haven’t seen a good critique of this trial.
Well, JAMA Cardiology has published a substudy looking to determine the effect of the extent of viable and nonviable myocardium on the effectiveness of PCI, on prognosis and improvement in left ventricular function after PCI.
Recall that you had a to have evidence of viability in at least four myocardial segments that were dysfunctional at rest to be randomized.
Viability was determined in the main study by cardiac magnetic resonance (CMR) and dobutamine stress echocardiogram (DSE0 or nuclear imaging but for this study they used only CMR and DSE because the numbers in the nuclear arm were so small. These were read in blinded fashion.
They then separated the degree of viability into tertiles. Low viability, medium viability, and lots of viability.
The primary outcome of this subgroup analysis was death and HHF. Secondary outcomes were death, CV death, HHF, and improved LV function at 6 months.
Mean age of patients was 69 years. Of the 700 pts in the main trial, this analysis included 610, all but 90 pts.
In the main trial analysis, primary outcome events occurred in 36% of the two groups. Absolutely no difference. That’s the main trial results.
The main finding of this substudy was: There was no interaction between the extent of viable myocardium and the effect of assignment to PCI vs OMT on occurrence of the primary outcome or any of the secondary outcomes.
Boom. Read that again.
Figure 2 plotted the primary outcome events based on how much viability there was. The curves were super-imposable no matter the extent of viability.
Also surprising was that the amount of viable myocardium or the extent of nonviable myoocardium did not affect the rate of the primary outcome or CV death.
The one predictor they did see is that greater amounts of scar predicted the occurrence of a primary outcome and all secondary outcomes. But that’s not a surprise – scar predicts bad things. But again, there was no interaction with treatment effect. This was irrespective of treatment assignment.
None of the viability characteristics interacted with the effect of assignment to PCI vs OMT on the likelihood of improvement in left ventricular function
CMR did not do better: Sensitivity analyses based on a late gadolinium enhancement transmural threshold less than or equal to 50% also showed no association between the extent of viability and primary outcome, as well as no interaction with assignment to PCI vs OMT.
For the 479 participants assessed with CMR imaging, scar burden did not interact with the effect of assignment to PCI vs OMT on the risk of the primary outcome or any secondary outcome.
The authors concluded that, This study found that viability testing does not identify patients with ischemic cardiomyopathy who benefit from PCI.
My Comments. The idea of viability makes so much sense. Muscle can be healthy, hibernating and not working well because of ischemia, or scarred and dead.
For years, we thought, let’s find that second category; let’s then revascularize the areas with muscle that is still alive but poorly functioning because of ischemia. It is why so many patients with HF are referred for angiography.
And we think we see this clinically, because EF often gets better. But we also treat these patients medically or with cardiac resynchronization therapy.
But the first clue that myocardial viability may be a figment of wishful thinking came from an analysis of STICH, an RCT of coronary artery bypass grafting (CABG) vs medical therapy in patients with ischemic cardiomyopathy (CM), and multivessel CAD.
The main results of STICH, like REVIVED, found that CABG was no better than tablets, for reducing a primary outcome of death.
But a substudy looking at viability in STICH was quite revealing. About half the patients in STICH had viability scans. Of these, half were randomly assigned to CABG and half to meds.
After adjustment of other variables, the extent of viability had no association with mortality.
The criticism of this was that it only included about half of STICH patients.
Now we have an analysis of REVIVED BCIS, a more modern trial in which nearly 90% had viability testing. And it had no modification of the treatment.
Scar predicted worse outcomes, but we already knew that.
I think this substudy, taken together with STICH, suggests strongly that we save our money on viability tests.
The preponderance of the data say that revascularization does not improve outcomes in patients with ischemic CM. Treat these patients medically. Reserve revascularization for symptomatic angina. And don’t be tempted to try and find that perfect patient who will improve with revascularization.
Final Note: There is much more to report from TCT. I will do more next week. I will have a lot to say about TRILUMINATE; Watchman TAVR; and a super sobering comparison study of Watchman vs Amulet.
© 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: Oct 27, 2023 This Week in Cardiology Podcast - Medscape - Oct 27, 2023.
Comments