Concern spread through the October 24 session at the TCT Congress in San Francisco when Martin B. Leon presented the 5-year results of the PARTNER 3 study. Survival curves, which showed a crossover trend favoring surgery at 4 years in the previous year, now indicate superior outcomes for surgery (10% vs. 8.2%, HR 1.23). Although not reaching statistical significance, the trend indicates a likely advantage for surgery, which would be statistically significant if we consider the survival curve from two years onward when initial surgical mortality (2.5% vs. 1%) was balanced.
The same day, the NEJM published the complete study results, forming the basis of this blog entry. With 91.6% follow-up of the initial sample (496 patients in the TAVI arm and 453 in the surgical arm), updated 5-year results are shown. First, it is essential to highlight a greater follow-up loss in the surgical arm (454 patients; 88.3% of the initial sample) compared to the TAVI arm (469 patients; 94.6%). The primary reason for patient loss in the surgical group was voluntary withdrawal (45 participants), compared to only 21 patients in the TAVI arm. Regarding stroke outcomes, there remains a technical tie, with a less steep slope for surgery, which the TAVI arm has now overlapped: 6.4% for surgery vs. 5.8% for TAVI. This trend also reflects a balancing of the initial penalty suffered by the surgical option (3.1% vs. 1.2% at one-year results). Finally, the readmission rate, primarily responsible for significant differences in the composite event at one year, continues the initial 4% disadvantage of surgery, marked early postoperatively. However, for the first time, the composite event combining all three outcomes loses statistical significance between the two options.
This, combined with the technical tie in clinical events (endocarditis <2%, new coronary events, or need for revascularization, and even the need for pacemaker implantation at 5-6% when not considering the postprocedural period), hemodynamic performance (for gradients and structural degeneration, 3.3% vs. 3.8%), and quality of life measured by the KCCQ-OS score, makes it possible, in Martin B. Leon’s words, “to inform our low-risk aortic stenosis patients eligible for either treatment that, after five years, they have more than an 85% chance of being alive, doing well, and with a prosthesis free from structural degeneration.”
Such comments call for a review of past literature, as the same forum predicted TAVI’s superiority in low-risk profiles with one- and two-year results. While explanations currently focus on most of the developed mortality being non-cardiac (the cardiac origin difference being only 0.4% between procedures), it has been repeatedly discussed on this blog that mortality classification methods may not be entirely accurate. Notably, more deaths occurred in the TAVI group due to cancer (9 vs. 5), COVID-19 (3 vs. 1), or sepsis (4 vs. 1), alongside the mentioned imbalance in follow-up losses.
The study’s conclusion is that, in patients with severe symptomatic aortic stenosis undergoing TAVI vs. surgery, no significant differences exist between groups for primary outcomes at five-year follow-up.
COMMENTARY 1:
Reactions to the assumed result followed swiftly. Comments like, “Surgery has set a high standard for TAVI in low-risk patients” (Kendra Grubb, Emory University, Atlanta), “the age threshold reduction should be postponed… reducing to 65 years in the US may have been too bold a step” (B. Pendergast, Cleveland Clinic, London) echoed throughout the session. As expected, Michael Reardon’s analysis (Methodist DeBakey, Houston), lead investigator for the PARTNER 3 equivalent Evolut study, noted that “TAVI has an early advantage over surgery in mortality, stroke, and recovery… but the real question is whether this is sustainable.”
At this same session, an update from the Evolut-Low Risk study results was presented, showing an opposite trend to that of PARTNER 3. Briefly, with 94.7% of the initial sample, four-year outcomes showed no significant mortality differences between surgery and TAVI (12.1% vs. 9%) and none for stroke (3.8% vs. 2.9%) or valve-related reintervention (1.7% vs. 1.3%). These results were accompanied by a JACC publication promising five-year outcomes in the next issue.
With this analysis, the focus remains on clarifying why there is a discrepancy between studies and ensuring transparency on underemphasized aspects.
Regarding the first point, the initial reaction was to avoid extrapolating results and, above all, refrain from asserting that one TAVI prosthesis is superior to another. Survival results correlate closely with patient selection, and such trials typically involve highly selected populations. In fact, PARTNER 3’s selection criteria were stricter concerning TAVI implantation technique, using the Sapien 3 prosthesis versus self-expanding devices and the technology gap of the Evolut Low-Risk. Moreover, teams involved, surgical prosthesis types, etc., are not comparable, thus neither are the results. Hopefully, as this represents the first setback in the TAVI field, these points will be considered in meta-analyses. However, when preprocedural variables are evaluated, both studies involve similar low-risk surgical patients (mean STS score 1.9%) and comparable ages, ensuring minimal comorbidity. Only COPD showed a higher proportion (15-17% vs. 5-6%) in PARTNER 3. While minor, one-year mortality differences of 2.5% vs. 1% for SAVR vs. TAVI in PARTNER 3 and 3% vs. 2.4% in Evolut Low-Risk, and stroke (4.3% vs. 4.1% in Evolut Low-Risk and 3.1% vs. 1.2% in PARTNER 3) are crucial in understanding presented results. We must remember significant differences in associated procedures in the surgical group (up to a quarter of SAVR patients in both studies) compared to the interventional arm, which has affected outcomes from the start. These differences persist over Evolut Low-Risk study follow-up, while trends in PARTNER 3 have crossed. This reflects the prognosis of the surgery group cohort, appearing either early (PARTNER 3, now balanced) or later in follow-up (Evolut Low-Risk, not yet balanced), biasing studies and resulting in disparate findings.
The second point involves the absence of paravalvular leakage results in the 5-year PARTNER 3 report. Evolut-Low Risk’s four-year JACC update indicates 98.4% freedom from leakage in the surgical group vs. 84.7% for TAVI (p < 0.05), with a worrisome leakage rate of mild or higher: 38% in the first year, 27.5% in the second, 22.5% in the third, and 16.6% in the fourth, with complete disappearance of severe leakage by the second year (shown only at TCT). Minimal attention was given to prosthetic thrombosis in PARTNER 3, which was over 10 times higher for TAVI (2.5% vs. 0.2%). In Evolut Low-Risk, pacemaker need remains at 24.6% vs. 9.9% (p < 0.001) at four years; a prognostic factor previously analyzed on this blog.
Finally, the durability of TAVI’s superior early results depends on adequate prosthetic durability. Four- and five-year studies have shown excellent durability for both prosthesis types under VARC-3 criteria. However, durability’s importance in low-risk patients was underscored by NOTION’s 10-year findings. This all-comers intermediate-risk study demonstrated the bias consequences in its design: >30% Abbott Trifecta and Livanova Mitroflow bioprostheses, both now withdrawn, and a biased structural deterioration definition using VARC criteria based on fixed gradients >20 mmHg when >40% of implants were size 21 mm or less.
The PARTNER 3 five-year publication has served as a “scare” for the TAVI world, emphasizing that extending low-risk indications, especially in younger patients, requires patience. The current study’s conclusion is that the “treat” is to celebrate the “draw” that real aortic valve replacement now enjoys, contrary to predictions of doom based on one- and two-year results.
REFERENCE:
Mack MJ, Leon MB, Thourani VH, Pibarot P, Hahn RT, Genereux P, et al.; PARTNER 3 Investigators. Transcatheter Aortic-Valve Replacement in Low-Risk Patients at Five Years. N Engl J Med. 2023 Oct 24. doi: 10.1056/NEJMoa2307447.
COMMENTARY 2:
Given these findings, the STS and EACTS issued a joint statement on October 30 reiterating the results and arguments in this blog. They remind us that real-world SAVR survival rates, such as those from the STS, covering over 42,000 patients, are excellent: 92.9% at five years and 90% at eight years; with 95% at eight years for patients under 75 with an STS score <1%. Therefore, both societies urge caution and consensus on expanding TAVI indications, particularly for younger, low-risk patients.
Partial results in this and other interventional areas cannot be taken as definitive conclusions. As stated in the past by EACTS President Patrick Perier, such results should be seen as hypothesis-generating rather than as set dogma. Confirmation by further studies with adequate follow-up and comparisons against real-world registry outcomes is mandatory. Surgery has evolved technically over decades, progressing slowly and steadily to reach the level of excellence it holds today. It has responded to setbacks, removing poorly performing prostheses, reinventing itself through the resurgence of minimally invasive techniques initially proposed decades ago, and innovating with new devices like sutureless prostheses.
We are grateful for the publication of these recent study results, something that has not been consistently available for intermediate-risk studies such as PARTNER 2, SURTAVI and others, apart from NOTION, which offers nearly a decade of follow-up data. It seems the “nightmare before the next guidelines” is upon us, and we hope that clinical and scientific responsibility will prevail.
REFERENCE:
