TAVR vs Surgery in Low-Risk Aortic Stenosis: What the Long-Term PARTNER 3 and Evolut Data Show

Both landmark trials found transcatheter aortic valve replacement (TAVR) non-inferior to open surgery early in low-risk patients, and that verdict has largely held through six to seven years of follow-up. What has shifted is the durability picture. Reintervention rates now diverge by valve platform, and a primary endpoint that looked settled in 2019 reads differently once the curves have more time to separate. The long view does not overturn the early result so much as add the chapter that early results cannot contain.

Where the low-risk question came from

TAVR was first approved for patients too sick or too high-risk for surgery, then worked its way down the risk ladder. The low-risk step was the decisive one, because most people with severe aortic stenosis are not high-risk, and because a younger, healthier patient has more years for a valve to wear out. Two trials anchored that step. PARTNER 3 randomized 1,000 low-risk patients to a balloon-expandable transcatheter valve or surgery, and Evolut Low Risk tested a self-expanding transcatheter valve against surgery in a comparable population. Both reported early non-inferiority for their primary endpoints, and both were built to keep following patients for a decade because durability was the open question from the start.

What the seven-year PARTNER 3 data show

At seven years, presented at TCT 2025 and published in the New England Journal of Medicine, PARTNER 3 found similar rates of its primary composite of death, stroke, or rehospitalization: a Kaplan-Meier estimate of 34.6 percent with TAVR and 37.2 percent with surgery. The confidence intervals overlap, so the honest reading is that the two strategies looked broadly comparable rather than one clearly beating the other.

The components are worth separating, because a composite can hide as much as it reveals. Death was 19.5 percent with TAVR versus 16.8 percent with surgery, a numerical gap that favors surgery but sits well within the play of chance for a trial this size. Stroke was nearly identical at 8.5 versus 8.1 percent. Rehospitalization ran the other way, 20.6 percent with TAVR versus 23.5 percent with surgery. On durability, aortic-valve reintervention was 6.7 percent with TAVR and 6.0 percent with surgery, and bioprosthetic valve failure was low and similar in both arms, close to 7 percent either way. For the balloon-expandable device, the durability signals stayed close.

What the six-year Evolut data show

Evolut Low Risk, reported through six years, also met non-inferiority for its primary endpoint of death or disabling stroke: 23.3 percent with TAVR and 20.4 percent with surgery, a difference of 2.9 percentage points that was not statistically significant. So far the two trials agree.

The reintervention data are where they part company. In Evolut, reintervention was already numerically higher after TAVR at six years, 5.5 versus 3.3 percent, and by seven-year follow-up the gap reached statistical significance, roughly 9.8 versus 6.0 percent, with a subdistribution hazard ratio near 1.68. The mechanism matters: the investigators attributed the excess mostly to bioprosthetic aortic regurgitation, meaning the valve leaking, rather than to the valve re-narrowing. That is a different failure mode from the one many people picture when they think of a worn-out valve.

Why the two trials point in slightly different directions

It is tempting to average the two into a single verdict, but the more useful lesson is that they used different devices and the durability data followed the device. PARTNER 3 tested a balloon-expandable valve; Evolut tested a self-expanding one. The Evolut analysis carries an explicit caveat that it reflects earlier-generation self-expanding valves and earlier implantation techniques, along with heterogeneous choices of surgical prosthesis in the comparison arm. Newer valves and refined technique may not behave the same way, which is exactly why reintervention numbers from trials whose patients were enrolled years earlier cannot be read as the final word on today's hardware.

This is the deeper point for anyone appraising trial evidence. A primary endpoint answers the question the trial was powered to answer, usually a short-to-medium-term composite. Durability endpoints such as reintervention and structural valve deterioration answer a different question that only long follow-up can address, and they can move after the headline has been written. Both things can be true at once: TAVR was non-inferior at the primary readout, and the reintervention curves deserve continued watching, especially for younger patients whose valves must last decades.

How to read this if aortic stenosis is on your radar

For a person weighing options, the trials do not crown a universal winner. They frame a genuine trade-off that depends on anatomy, age, expected longevity, valve type, and the possibility of a future valve-in-valve procedure inside whichever valve goes in first. A numerically higher reintervention rate on one platform is meaningful for a 65-year-old in a way it may not be for someone considerably older. These are decisions made with a heart team that can see the imaging and the individual, not from a single trial statistic.

This article is educational and is not medical advice. Planned ten-year data from both trials should sharpen the durability comparison further, and until then the responsible summary is that early non-inferiority has held while the long-term reintervention story is still being written, and it is being written differently for different valves.