Neutral Trial, Positive Spin: PROTECTED TAVR

John M. Mandrola, MD


September 23, 2022

The debris-catching cerebral embolic protection (CEP) device should reduce stroke. It actually stops large debris from entering the brain's circulation after transcatheter aortic valve replacement (TAVR). We know this because we can see the debris that would have infarcted brain tissue.

A neutral observer might argue that there is no need to study this device against no protection. TAVR cannot be done without creating north-flowing debris; most post-TAVR MRI brain scans document defects consistent with emboli. Why would you leave patients unprotected?

Proponents of equipoise would point to the low stroke rates in recent TAVR trials in low-risk patients (3.4% in EVOLUT and <1% in PARTNER 3). They would also cite the fact that a previous trial of the same cerebral protection device (Sentinel, Boston Scientific) vs standard care found no difference in the primary endpoint of reduction of new lesion volume on MRI scans or improvement in neurocognitive function. Although the device did safely capture embolic debris in nearly all patients.


PROTECTED TAVR randomly assigned 3000 patients to TAVR with or without embolic protection. The primary endpoint was stroke within 72 hours of the procedure or before discharge.

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.

In the New England Journal of Medicine (NEJM) paper, the authors report the difference in stroke rates as –0.6%, with a 95% confidence interval of –1.7 to 0.5 and a P value of .30.

At the Transcatheter Cardiovascular Therapeutics (TCT) 2022 conference presentation, and in the NEJM paper, the authors focused on the secondary endpoint of disabling stroke—1 of 15 prespecified secondary endpoints. Disabling stroke occurred in 8 patients (0.5%) in the CEP arm vs 20 patients (1.3%) in the control arm. The 0.8–percentage point lower risk had a 95% CI ranging from –1.5  to –0.1 percentage points and thus met statistical significance (because the CI did not cross 1.0).

The authors and trial discussants made the point that the device is more likely to catch larger debris, and larger debris tends to cause larger areas of brain infarction. The signal of lower rates of disabling stroke, therefore, has biologic plausibility.

The conclusion paragraph in NEJM also made news (emphasis mine).

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.

Stroke Size vs Stroke Location

I have three areas of concern in this trial's interpretation.

The first is the primary endpoint. When planning PROTECTED TAVR, the investigators believed that a 2–percentage point reduction in stroke would be a clinically important difference.

The observed 0.6–percentage point absolute risk reduction in the trial, along with the 95% CI that only went to 1.7 percentage points essentially excludes that 2–percentage point  reduction. You might argue that had the trial recruited more patients, they may have passed the 2–percentage point  threshold, but we don't know that. The CI also includes a risk of harm, so more recruited patients may have pushed the endpoint in that direction.

The P value of .30 for the primary endpoint is remarkable. This means that if you assume CEP has no benefit (the null hypothesis), and you repeated this trial numerous times, you would see these results nearly 1 in 3 times. In other words, these results are not at all surprising given the assumption of no benefit.

A second area of concern is the emphasis on the secondary endpoint of disabling stroke.

I spoke to a stroke neurologist, who told me that although large debris tends to cause larger strokes, which would increase the probability of disabling stroke, another important variable in determining function is stroke location. He said you could have a huge stroke in the right frontal lobe, and it could cause no symptoms; however, a tiny stroke in a key area could be disabling. He believes that location is also important when it comes to brain infarcts.

Another problem with pushing a disabling stroke benefit is statistical. Disabling stroke was a secondary endpoint in a trial with a decidedly nonsignificant primary endpoint. Disabling strokes accounted for about half the events in the composite primary endpoint. This leads to even more uncertainty in detecting signal from noise.

Yet another issue with disabling strokes as an endpoint is that it requires more adjudication than the primary endpoint of any stroke. The authors defined disabling stroke as a modified Rankin Scale (mRS) score of  2 or greater or an increase of at least 1 point from the prestroke baseline. The mRS score is subjective, and there is substantial observer variability in its assessment.

Similar Findings, Inconsistent Conclusions

My third area of concern is that the conclusion in the NEJM paper allowed the phrase "the 95% confidence interval includes the possibility of benefit."

In the PARAGON-HF trial, sacubitril-valsartan led to a 13–percentage point reduction of the primary endpoint of cardiovascular death or heart failure hospitalization vs valsartan alone in patients with heart failure and a preserved ejection fraction. The P value was .06 and the conclusion of that NEJM paper read (emphasis mine):

Sacubitril–valsartan did not result in a significantly lower rate of total hospitalizations for heart failure and death from cardiovascular causes….

And a month ago, the DANCAVAS trial of screening for heart disease found a 5–percentage point reduction in death. The P value was .06 and the CI ranged from 0.90 (a 10–percentage point reduction) to an upper bound of 1.00 (no reduction). The conclusion of that NEJM paper read:

After more than 5 years, the invitation to undergo comprehensive cardiovascular screening did not significantly reduce the incidence of death….

Why the sudden change for PROTECTED TAVR? Will journal editors go back and change the conclusions for other trials with nonsignificant primary endpoints?

Although it is true that the CI in PROTECTED TAVR did not rule out benefit from the CEP device, it is also true that it did not rule out harm. Why does the conclusion focus only on the benefit side of the interval?

I've communicated with statistical experts, and many don't like the way journal editors force investigators to dichotomize results into something works or it does not. They prefer communicating the possibilities within the bounds of the 95% CI. I do too. The problem is that nuance is harder than yes/no conclusions.

But there has to be consistency.

Two Conclusions

I am excited about the possibilities of adding nuance to trial results. Declaring success or failure on the basis of arbitrary statistical thresholds never made sense. We shall see what leading journals do with future trials.

In the end, though, clinicians must decide. For me, in the case of embolic protection after TAVR, this trial and others did not show that the probability of benefit is substantial enough to warrant its routine use. If we lower the threshold of evidence needed to accept new therapies, I am afraid we will see even more low-value care.

John Mandrola practices cardiac electrophysiology in Louisville, Kentucky, and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence. 

Follow John Mandrola on Twitter

Follow | Medscape Cardiology on Twitter

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


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.