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. This podcast is intended for healthcare professionals only.
In This Week’s Podcast
For the week ending September 23, 2022, John Mandrola, MD comments on the following news and features stories.
Placement of a filtering device north of the valve that catches debris seems like a sure bet to reduce stroke. It does catch ugly-looking debris, but a previous trial showed no differences in MRI defects or cognition. The filtering device adds cost and requires separate arterial access. Its placement in the great vessels may dislodge debris that causes stroke.
The PROTECTED TAVR trial randomized 3000 patients to TAVI with or without embolic protection.
The endpoint was stroke—a symptomatic defect that correlated with imaging.
Stroke after TAVR occurred in 34 patients (2.3%) in the CEP arm vs 43 patients (2.9%) in the control arm. This difference of 9 strokes did not come close to reaching the accepted threshold of statistical significance.
The relative risk reduction was 21% with a confidence interval that went from 0.51 to 1.23 – a 49% decrease or a 23% risk increase. The p-value was .30.
Given the previous trial showing no benefit, and now this larger trial with hard outcomes, we can conclude that it doesn’t work.
No. That is not what happened.
First the New England Journal of Medicine (NEJM). The journal that just 4 weeks ago, made Axel Diederichsen, the PI of the DANCAVAS trial of cardiac screening, declare that a 5% reduction in death was non-significant because the upper bound of the CI reached 1.0- and a p-value was .06, allowed this statement in the conclusion of PROTECTED TAVR
Among patients with aortic stenosis undergoing transfemoral TAVR, the use of CEP did not have a significant effect on the incidence of periprocedural stroke, but on the basis of the 95% confidence interval around this outcome, the results may not rule out a benefit of CEP during TAVR.
The conclusion is correct but incomplete. It is true that the 95% CI allows for a 49% reduction in stroke, it also allows for a 23% increase. Why not state that as well?
But that is not all that happened to spin this trial:
At TCT, the authors –and panelists—drew attention to one of 15 secondary endpoints—disabling stroke.
Disabling stroke occurred in 8 patients in the CEP arm vs 20 patients in the control arm. The 0.8% lower risk had a 95% CI ranging from -1.5% to -0.1% and, thus, met statistical significance.
The delta here was 12 strokes—in a trial with 3000 patients.
The authors and panelists justify it on the grounds of plausibility-- because the device catches large debris and large debris is more likely to cause disabling strokes.
One problem is that stroke location can be as important. A tiny clot in a crucial area can be disabling.
Another problem is statistical – it’s one of 15 secondary endpoints in a trial with a decidedly non-significant primary endpoint. Never forget Richard Peto’s publication in the Lancet in ISIS-2 wherein he showed that astrological sign could determine aspirin’s benefit. His point was that if you look at enough outcomes after the primary, you will get some that turn positive—by chance alone.
The final issue with disabling stroke is that it requires more adjudication. And any time you add adjudication, you add a risk of bias. Especially using the mRS scale which is very subjective.
If we allow trials with 0.6% ARR and p-values of 0.30 to be spun positive by proponents, then why bother doing trials. Proponents of doing costly invasive procedures must bear the burden of proof.
If any editor of the NEJM accidently happens upon this podcast, I want to ask them: why now?
What about DANCAVAS, or PARAGON-HF or a host of other trials that came close but did not meet our accepted threshold of significance? Let us know. Please.
I am seeing a very worrisome trend in the RDN space. While there is no doubt that newer generation devices provide a statistically significant, albeit small, sham-resistant drop in BP over the short term—months. Andrew Foy’s group at Penn State has the best meta-analysis on this in Trends in CV Med. First author Mo Ahmed. In 10 RCTs with sham there is a 2.6 mmHg sham -resistant lowering of BP but in meta-regression, Studies that followed patients for longer than 4 months had numerically lower reductions in most BP outcomes.
I am deeply concerned with the push to convince us that this effect is durable over the long-term.
I’ve previously reported that interpretation of the follow up from SPYRAL-HTN-ON, which showed reductions of BP at 36 months are deeply, if not fatally flawed, by unmasking. Lest we forget the intense excitement regarding the non-sham-controlled pre SYMPLICITY III trials, which found massive reductions of BP.
Then boom. SMYPLICITY III, with its proper sham, showed no significant reduction.
At TCT, we heard results of a follow-up study from SYMPLICITY III.
Steve Stiles is one of the most experienced health reporters in the business. His lead exactly captured what the authors want us to know about this follow-up.
There's an intriguing plot twist in the story of SYMPLICITY HTN-3, the sham-controlled clinical trial that nearly put the kibosh on renal denervation (RDN) therapy as a promising approach to treatment-resistant hypertension (HTN).
The trial famously showed no benefit for systolic blood pressure (BP) from the invasive procedure at 6 months and 12 months, dampening enthusiasm for RDN in HTN for both physicians and industry. But it turns out that disappointment in the study may have been premature.
The procedure led to significant improvements in systolic BP, whether in-office or ambulatory, compared with a sham control procedure, in a new analysis that followed the trial's patients out to 3 years.
Steve then quotes PI of SYMPLICITY and this re-analysis Prof Deepak Bhatt
"These findings support that durable blood pressure reductions with radiofrequency renal artery denervation...”
Steve’s job is to report the news. The TCT news was...don’t worry folks, RDN provides durable control, and, better, the effects may actually improve over time.
While Steve reported it beautifully, my job is to critically analyze these conclusions.
And I will say with little doubt that these conclusions are surely wrong.
In the original SYMPLICITY III study RDN was tested in 2:1 randomization against a sham control. Researchers were careful to maintain blinding.
This trial found no significant difference in BP over 6 and 12 months. Now, I know, there are many reasons why this may have been – inadequate ablation for instance—but the fact remains that it did not reduce bp.
Now the SYMPLICITY III authors offer a follow-up study in which patients were unmasked at 6 months.
About 361 patients started in the active RDN arm.
About 170 patients started in the control arm-àAfter unmasking, 101 of the 170 had crossover to RDN and 70 did not.
The active arm was the RDN arm carried out over 3 years and the comparison arm included non-crossover patients and data from crossover patients at 6 months, before renal denervation, using the most recent pre-crossover blood pressure to impute subsequent follow-up data.
And the results were clear and massive:
RDN reduced office systolic blood pressure at 36 months of 26·4 mm Hg (SD 25·9) in the renal denervation group versus 5·7 mm Hg (24·4) in the control group (p≤0·0001)
A 21-mmHg difference. 24-hour ambulatory BP was also hugely and significantly reduced.
The authors most of whom are experienced trialists conclude:
This final report of the SYMPLICITY HTN-3 trial adds to the totality of evidence supporting the safety of renal artery denervation to 36 months after the procedure. From 12 months to 36 months after the procedure, patients who were originally randomly assigned to receive renal artery denervation had larger reductions in blood pressure and better blood pressure control compared with patients who received sham control.
I read the limitations paragraph and they hardly mention the elephant stomping about in the room: the unmasking.
So to sum this up another way: When under masking conditions there was no difference. But after unmasking, you observe better blood pressures in the RDN and higher BPs in the non-cross-overed group.
The reason there are placebos and blinding is to control bias.
The main problem here—as so clearly noted in the accompanying editorial from Drs Fanelli and Persu—is the unmasking. Once you do that, the game is over. Not only are patients unmasked and open to different behaviors and adherence, but the clinicians are also unmasked, opening up the possibility of performance bias.
You might ask –how do you know there was bias, Mandrola? My answer is that if there wasn’t bias in unmasked comparisons, we would never do placebo-controlled trials.
Here, bias from unmasking is 80% of the problem but another is imputation of blood pressures before RDN in the crossover arm. When you take the last BP before an unmasked patient in the control arm has RDN, you may be, as the editorialists write, crystallizing the highest BP. And thereby biasing in favor of RDN. When the trialists did not use imputation (see table s4), there were no longer differences in BP.
Yet more evidence for bias comes when looking at figure s5 in the appendix which graphs SBP in crossover and non-crossovered patients from baseline.
The patients who did not crossover had a reduction in blood pressure when masked in the first six months, but then had increases in blood pressure after unmasking. This strongly suggests a Hawthorne effect during blinding—meaning that during the trial when they did not know which therapy they had, and it could have been RDN, they had better adherence to lifestyle and meds.
I do not oppose RDN. Newer generation devices seem to have a sham-resistant effect over the short-term. There is clear potential because medical adherence is sub-optimal. But HTN treatment is for years. RDN is invasive and costly. The only way to prove it works is to maintain blinding over years. The argument that resistant HTN patients must be cross-overed to RDN is specious because it assumes RDN works—and I don’t think we know that yet.
I really worry that complicated analyses like this hide the fatal flaws inherent in unblinded observational comparisons.
LAAO – Two comments
A colleague called me this week to tell me they dealt with 3 Watchman complications on their service in one week.
The TCT conference featured a head-to-head randomized trial comparing Boston Scientific’s Watchman to Abbott’s Amulet. Two percutaneous devices used to occlude the left atrial appendage (LAA).
I highlight this study to illustrate what happens when lenient regulatory approval meets a rapturous audience of physicians who like doing procedures.
The trial results made news. It was a featured study at TCT. Yet there exist no compelling evidence that either device is better than DOACs or no treatment.
Recall that Watchman was approved despite not meeting noninferiority vs warfarin for the first co-primary endpoint of stroke, systemic embolism, and CVD in PREVAIL.
Patients who had contraindications to anticoagulation were excluded from the pivotal trials. And we have yet to generate any convincing evidence that this procedure helps these people.
The Amulet IDE trial randomized nearly 1900 patients to either device. Why not have a no treatment arm?
Rates for procedure-related complications were 4.5% with Amulet and 2.5% with Watchman. A no-treatment arm would start the follow-up period without the risk of these complications.
The second story on percutaneous LAAO is an update to my coverage last week on death rates after the procedure. Recall that Canadian and Spanish investigators found that in a contemporary multinational European series, 1 in 6 patients died in the first year. Based on what I have seen for referrals for Watchman in the US, I bet numbers of early death in this country are higher than that.
These patients have no chance to benefit from LAAO closure because the procedure is designed to reduce bleeding and stroke over years not months.
Patrice Wendling has excellent news coverage. One of the debates she covered is that the study authors called the procedure futile if the patient has early death. The editorialists push back on that.
I side with the study authors. If you do a preventive procedure for one disease, and the patient was sick enough to have died in less than a year from other diseases, than the people that have been helped are the device makers, doctors, and hospitals.
I realize many listeners may be tired of my railing against this procedure. I am sorry, but it is deeply troubling to me to see my colleagues embrace such an invasive procedure that is so devoid of convincing data.
Aspirin—especially in the presence of oral anticoagulation (OAC)—is, by far, the most common drug that I deprescribe.
There is just no evidence that use of both OAC and aspirin when there is no clear indication for benefit. There is, in fact, evidence of harm in the form of bleeding.
But doctors and patients tend to do it. The soft thinking goes: OAC for clot prevention from the low flow system of the LA or peripheral veins and aspirin for prevention of arterial thrombus in the coronaries or brain.
One of the hardest challenges in stopping aspirin is that people are attached to it. Univ of MI researchers led by Geoff Barnes, first author Jordan Schaefer, published a neat ecological study in JAMA Network Open on the effects of a quality initiative project in 6 AC clinics in different medical centers in MI.
The written paper is complicated. But the figures tell the story. Basically, they got AC clinics to identify patients on warfarin and aspirin who did not have an ndication – recent stent or CABG or stroke etc – for aspirin.
They then studied the effects of an aspirin deprescribing intervention---which was as simple as communication with the primary care clinician to say, hey, there isn’t a clear reason to give aspirin along with warfarin.
Again, the figures hold the key. In the 2-year pre-intervention period there was a steady decline in aspirin use. But not a steep decline. Then in the 24 months after the intervention, aspirin use plummeted.
Bleeding rates declined. Thrombotic rates did not.
I know, it’s observational. The way to answer this question is do a de-implementation RCT. It probably could be done pragmatically embedded in the EHR.
I would say, in the absence of clear aspirin indication, which are becoming less and less common, I say use OAC alone.
First, do no harm. Good work to the Michigan team.
© 2022 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Sep 23, 2022 This Week in Cardiology Podcast - Medscape - Sep 23, 2022.