Boom: TAVR Low-Risk Trials Usher in a New Era in Cardiology

John Mandrola, MD.


March 18, 2019

The transcatheter aortic valve replacement (TAVR) trials presented today at the American College of Cardiology 68th Annual Scientific Session 2019 (ACC.19) reminded me of one of the reasons I was drawn to cardiology: not only did we do nifty things, but we also lead the way in generating evidence.

UCLA cardiologist Boback Ziaeian, MD, PhD, said it well in this tweet: "The investigators demonstrated how devices need to be approved by the FDA. Sickest patients first and [then] demonstrate safety and adverse events in lower and lower risk groups."

My friend and colleague Andrew Foy, MD, from Penn State, told me that even if the low-risk TAVR data were slightly worse than those for surgery, he thinks most patients would still choose the transcatheter route. People like the idea of a fix without a sternal saw.

But the TAVR data are not slightly worse. They are  impressive—albeit with caveats.

Sue Hughes has a recap of the PARTNER 3 and Evolut results.  Here is my short take:


The PARTNER 3 trial[1] enrolled about 900  truly low-surgical-risk patients; the average Society of Thoracic Surgeons (STS) score was 1.9%. The primary endpoint of death, stroke, or rehospitalization occurred in 42 of 496 patients (8.5%) in the TAVR arm and 68 of 754 patients (15.1%) in the surgical aortic valve replacement (SAVR) arm. That 6.6-percentage point absolute risk reduction met requirements for noninferiority and superiority.

Review of the individual endpoints, which required going to the paper's supplement, showed that TAVR benefit was largely driven by rehospitalization. Specifically, TAVR was associated with 5 fewer deaths, 8 fewer strokes, but 13 fewer rehospitalizations at 1 year.

In an email, Sanjay Kaul, MD, from Cedars-Sinai questioned the choice of the composite endpoint because rehospitalization biases the trial in favor of TAVR, given that rehospitalization (infection, arrhythmia, renal impairment, pleural and pericardial effusion, venous thromboembolism) is more likely to occur after SAVR. The data support his view; the rate of rehospitalizations in the SAVR arm was even greater at 30 days vs 1 year.

The choice of a softer endpoint deserves emphasis. Veteran journalist Patrice Wendling from | Medscape Cardiology was quick to point out to me that previous PARTNER trials,[2],[3] the Surtavi trial,[4] and even a future TAVR study used either death or death/stroke as primary endpoints.

This is a serious caveat because it tempts you to think the investigators, who are clearly proponents of transcatheter procedures, were going for a big win rather than a more objective comparison. Kaul added that PARTNER 3 relative to PARTNER-2 used less stringent noninferiority criteria and shorter—1-year vs 2-year—follow-up.  "One would have surmised that for a low-risk cohort, where durability of outcomes is a key issue, the follow-up would have been longer, the endpoint more or equally robust, and the noninferiority margin would have been much more stringent."


The Evolut trial enrolled nearly 1500 patients, also with an average STS score of 1.9%. The caveats with this trial begin even before I give you the results.

This paper is not the final 2-year report but a prespecified interim analysis at 1 year. One-year follow-up was available for 432 patients in the TAVR group and 352 in the surgery group; 24-month follow-up was available for 72 patients in the TAVR group and 65 patients in the surgery group.

The authors give the top-line results using an imputed estimate of event rates at 2-years.

The primary endpoint of death or disabling stroke occurred in 5.3% of the TAVR arm and 6.7% in the surgery group. This difference met noninferiority criteria but did not reach superiority.

In addition to reporting estimates of 2-year event rates, Evolut had two other major concerns.

The rate of pacemaker implantation was 17.4% with TAVR vs 6.1% with SAVR. That is a serious issue when the device is used in younger patients, who have longer periods of time to experience lead complications and the need for generator replacement(s). During the presentation, Michael Reardon, MD, from Houston Methodist, said that the newer self-expanding valve is designed to have less heart block.

Moderate or severe aortic regurgitation occurred in 3.5% of the patients in the TAVR group vs 0.5% in the SAVR group. This, too, is relevant for younger patients, who have longer time frames to develop left ventricular dilation.


Even with the caveats, TAVR performed well against SAVR in the short term. But low-risk patients are less sick and younger; their time horizon is longer. This makes durability of the valve a key issue.

In an email, Ajay Kirtane, MD, from Columbia University agreed on the need to answer the durability question, but "it's important to remember that many surgical valves do not have the same degree of durability data, especially because they also iterate."

He pointed me to a recently published report that dispels the simplistic notion that surgical bioprosthetic valves universally last a long time. This systematic review[5] found considerable variability in reporting structural valve deterioration with different definitions and inadequate long-term core-lab data. What's more, the smaller Danish NOTION trial of TAVR vs SAVR in patients at lower risk reported that through 6 years, the rate of structural valve deterioration was higher for SAVR than for TAVR (24% vs 4.8%; P<.001).[6] Reardon emphasized this point in the discussion after the trial presentation.

Another issue Kirtane highlighted was the fact that SAVR better addresses coronary issues with combined valve replacement and bypass surgery. "With younger patients who have TAVR, there is more time for coronary issues to manifest, and TAVR typically means less complete revascularization up front." He also noted that with certain TAVR valves, subsequent percutaneous transluminal angioplasties are harder to do.

Cognition effects play a bigger role in younger, healthier patients. I recently saw an MRI brain scan obtained shortly a TAVR. The patient had no obviously neurologic defects, but the degree of "white spots" due to debris could only be described as stunning. Alternatively, SAVR requires cardiopulmonary bypass. Cognition should be rigorously studied over the long term.


I am impressed. The progression of evidence generation for TAVR has been exemplary.

More notable, though, is the gain for patients with aortic stenosis. A patient who is treated with TAVR rarely goes to an intensive care unit, often goes home the next day, and is less likely to have a stroke or die.

That said, much work remains. We need long-term data; we must self-police and prevent indication creep; and we need to keep in mind that not all hospitals should be doing TAVR. The percutaneous vs surgical decision in low-risk aortic stenosis lends itself well to decision support tools, such as a decision aid.

The TAVR trials remind me of the cardiology of old—big gains.

We used to ask, When is TAVR appropriate? The question now becomes, When is TAVR not appropriate?


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.