Recently, the three-year follow-up results of the Evolut Low-Risk study, an American endeavor and one of the studies supporting the indication of TAVI for patients with low surgical risk, has been published. This study utilized the Medtronic® platform prosthesis and paved the way for FDA approval to cover the full indication spectrum for patients with symptomatic aortic stenosis. The study’s results have been presented in three follow-up series: perioperative and at one year, two years, and, most recently, three years. Previous versions showed no superiority of TAVI over surgical aortic valve replacement (SAVR). However, the study was initially designed as a non-inferiority trial, and for the first two years, both therapeutic options were considered equivalent across all analyzed clinical events, whether individual or composite.

The authors’ primary conclusion from the three-year outcomes indicated that, for the first time, TAVI demonstrated an almost statistically significant benefit (p = 0.05) in the composite event of mortality and disabling stroke, with a trend towards diverging curves that suggest an increasing benefit for TAVI compared to SAVR.

The conclusions and results obtained warrant a detailed analysis, which we provide below. This will help to understand the reasons behind this finding within the published material and foresee the successive results that will likely be published as the study continues to be of interest.

COMMENTARY:

There is a well-known saying in Spain, “The seeds of today’s troubles were sown in the past” and the Evolut Low-Risk study is a clear example. To begin our analysis, we must go back to 2019 when the study’s first version was published. It was a randomized, multicenter design that included 1,468 patients with severe aortic stenosis, randomized 1:1 (734 per group) to TAVI (98% transfemoral) or SAVR.

The first notable observation is that only 86 U.S. centers participated, contributing between one and 130 patients each. Such variability likely includes centers with very low volumes of activity, which may have impacted the results we will discuss. Selection criteria were balanced for both techniques from a clinical standpoint. However, from a technical standpoint, TAVI results were safeguarded by specific anatomical requirements for the ascending aorta (absence of aneurysm), the aortic root and sinuses of Valsalva (appropriate for TAVI deployment), the left ventricular outflow tract (free of calcification or other distortions), septal hypertrophy, maximum and minimum sizes (30 mm and 18 mm, respectively), and absence of bicuspid aortic valve.

Regarding patient numbers, 11 patients in the TAVI group and 56 in the surgical group were excluded from the initial recruitment, with analysis performed by treatment rather than by intention-to-treat. This resulted in the inclusion of 725 patients in the TAVI group and 678 in the SAVR group. No crossover introduced additional selection bias, but the decision to reject patients was influenced by study coordinators, who removed 20 patients from the surgical group; 33 patients also refused to sign informed consent. The study does not elaborate on the content of the information provided or the reasoning behind these arbitrary decisions. Interestingly, candidates for SAVR who were not treated showed no significant differences from treated patients, although they had a lower incidence of variables such as COPD. It is essential to note that these were low-risk surgical patients who could have largely undergone SAVR with favorable outcomes.

In the surgical group, results were “within mortal reach,” with cardiopulmonary bypass and aortic clamp times of 93.3 and 68.7 minutes, respectively. Most prosthetic implants were between 23 and 29 mm, indicating a “comfortable” aortic root size. Although concomitant valvular disease was an exclusion criterion, 178 patients required additional procedures, including 13.6% of all SAVR patients who underwent surgical revascularization. Over 10% also required procedures on the aortic root, including augmentations or replacements (BioBentall-De Bono). The study does not specify the types of prostheses used, approaches, or surgical protocols followed. Only bioprostheses, stented or non-stented, available in various U.S. centers, were permitted.

These data contrast and limit comparability with the TAVI group, where only 6.9% of patients required coronary intervention (not necessarily concurrent with TAVI). Embolic protection devices were used in 1.2% of cases (off protocol), and two TAVI prostheses were necessary in 1.2% of cases, with an extremely low surgical rescue rate of 0.6%. The implanted prostheses included various evolutionary generations, such as Evolut R® (74.1%) and Evolut PRO® (22.3%), which offer reduced paravalvular leakage, smaller profiles, suitability for larger aortic rings, and improved navigability compared to the first-generation CoreValve® (3.6%). These differences reflect the varying care in selecting centers and procedures between the two study arms.

The initial follow-up version showed an early penalty in 30-day all-cause mortality, which, though not statistically significant, was a difference carried throughout the follow-up. It was a typical initial handicap for the surgical alternative over the interventional one, as often seen in this type of study design. Mortality was 0.5% for TAVI and 0.8% for SAVR, totaling eight cases. Misfortunes seemed to cluster in this group initially, with causes of death including a rupture at the aortic cannulation site, two coronary ostia obstructions (one requiring mechanical circulatory support while the other died in surgery), one severe stroke, two cardiac arrests (etiology unspecified), one cardiac tamponade after pacemaker lead removal, one mesenteric ischemia, and one multiorgan failure of unspecified origin.

Postoperative events were as follows for TAVI vs. SAVR: cardiovascular mortality 0.5% vs. 0.6%; stroke 2.1% vs. 1.9%, with disabling stroke 0.4% vs. 0.9%; rehospitalization 0.9% vs. 1.1%; prosthetic thrombosis 0.1% in both groups; new aortic valve procedure 0.2% vs. 0.4%; mild paravalvular leakage 36% vs. 3%; moderate-severe 3.4% vs. 0.4%.

The one-year follow-up data were not fully available until the two-year publication, covering only 432 TAVI and 352 SAVR patients. One-year results officially reported data for 706 TAVI (out of 725 initially enrolled, -19 patients) and only 628 SAVR (out of 678 initially enrolled, -50 patients). Comparable results for TAVI vs. SAVR included cardiovascular mortality of 1.7% vs. 2.6%, stroke of 4% vs. 4.2% (disabling: 0.8% vs. 2.1%), rehospitalization of 3.6% vs. 6.7%, prosthetic thrombosis of 0.3% for both groups, and new valve procedure of 0.7% vs. 0.6%. Paravalvular leakage artifacts began to appear, with an apparent reduction in mild degree 33.9% vs. 2.5% but not in moderate-severe degree 3.6% vs. 0.6%.

Although full follow-up was unavailable, the study performed a Bayesian statistical analysis to predict two-year results, a largely theoretical exercise showing a rush for partial results. The actual two-year follow-up data included 685 TAVI (-40 patients) and 594 SAVR (-84 patients). The data, expressed as percentages without absolute values, included all-cause mortality of 4.5% in both groups, cardiovascular mortality of 2.8% vs. 3.5%, stroke of 4.9% vs. 5.3% (disabling: 1.1% vs. 3.5%), rehospitalization of 5.4% vs. 7.9%, and prosthetic thrombosis of 0.6% vs. 0.5%.

Finally, in the current three-year study, the authors struggled to cover up the same criticisms of prior work, reporting 624 TAVI patients (-101 patients) compared to the 537 remaining SAVR patients (-141 patients). The authors now present both percentages and absolute values, with TAVI vs. SAVR showing all-cause mortality of 6.3% vs. 8.3%, cardiovascular mortality of 4.1% vs. 5.6%, stroke of 7.4% vs. 6.6% (disabling: 2.3% vs. 3.4%), rehospitalization of 7.4% vs. 9.2%, and prosthetic thrombosis of 0.7% vs. 0.6%.

Many follow-up losses could be attributed to paravalvular leakage or pacemaker rates during the first year (19.4% vs. 6.7%), impacting the results and needing distinction in responsibility between groups. This almost significant difference the authors found in the three-year composite event likely results from pressure by the sponsor (Medtronic®) and their affiliates, aiming to extend the indication for TAVI in low-risk patients.

For all the reasons mentioned, it will be difficult to justify any present or, especially, future conclusions drawn from this study. The issue is that it will influence future meta-evidence results as well as clinical recommendations, as seen in the recent 2020 joint guidelines by the AHA/ACC. Although it may not be possible to contain the pressure exerted, a call for responsibility, transparency, and avoiding confusion in published data is, at the very least, necessary. As the saying goes, “noise does no good, as good does no noise.”

REFERENCE:

Forrest JK, Deeb GM, Yakubov SJ, Gada H, Mumtaz MA, Ramlawi B, et al.; Low Risk Trial Investigators. 3-Year Outcomes After Transcatheter or Surgical Aortic Valve Replacement in Low-Risk Patients With Aortic Stenosis. J Am Coll Cardiol. 2023 May 2;81(17):1663-1674. doi: 10.1016/j.jacc.2023.02.017.

Forrest JK, Deeb GM, Yakubov SJ, Rovin JD, Mumtaz M, Gada H, et al. 2-Year Outcomes After Transcatheter Versus Surgical Aortic Valve Replacement in Low-Risk Patients. J Am Coll Cardiol. 2022 Mar 8;79(9):882-896. doi: 10.1016/j.jacc.2021.11.062.

Popma JJ, Deeb GM, Yakubov SJ, Mumtaz M, Gada H, O’Hair D, et al.; Evolut Low Risk Trial Investigators. Transcatheter Aortic-Valve Replacement with a Self-Expanding Valve in Low-Risk Patients. N Engl J Med. 2019 May 2;380(18):1706-1715. doi: 10.1056/NEJMoa1816885.