#aortic-stenosis #TAVR #small-aortic-annulus #prosthesis-patient-mismatch #bioprosthetic-valve-dysfunction #valve-platform #randomized-controlled-trial

Self-Expanding or Balloon-Expandable TAVR in Patients with a Small Aortic Annulus (SMART)

Authors, Journal, Affiliations, Type, DOI

Authors: H.C. Herrmann, R. Mehran, D.J. Blackman, S. Bailey, H. Möllmann, M. Abdel-Wahab, W. Ben Ali, P.D. Mahoney, H. Ruge, D.A. Wood, S. Bleiziffer, B. Ramlawi, H. Gada, A.S. Petronio, C.D. Resor, W. Merhi, B. Garcia del Blanco, G.F. Attizzani, W.B. Batchelor, L.D. Gillam, M. Guerrero, T. Rogers, J.D. Rovin, M. Szerlip, B. Whisenant, G.M. Deeb, K.J. Grubb, R. Padang, M.T. Fan, A.D. Althouse, and D. Tchétché, for the SMART Trial Investigators
Journal: New England Journal of Medicine, 2024;390(21):1959-1971 (published online April 7, 2024)
Affiliations: University of Pennsylvania Perelman School of Medicine, Philadelphia (coordinating centre), with 83 sites in Canada, Europe, the Middle East and the United States
Type: International, prospective, post-market, randomised, controlled trial. Device assignment could not be blinded; an independent clinical-events committee adjudicated end points and an independent echocardiographic core laboratory (Mayo Clinic) read the echo end points without knowledge of clinical outcomes. Medtronic funded the trial and was responsible for site selection, data monitoring and statistical analyses; a Medtronic employee prepared earlier versions of the figures and tables and provided editorial assistance.
DOI: https://doi.org/10.1056/NEJMoa2312573

Overview

SMART randomised patients with symptomatic severe aortic stenosis and an aortic annulus area of 430 mm² or less to a self-expanding supraannular valve (Evolut PRO/PRO+/FX) or a balloon-expandable valve (SAPIEN 3/3 Ultra) — a population that is about 87% women and historically under-represented in device-versus-device comparisons. At 12 months the self-expanding valve was non-inferior for the clinical composite of death, disabling stroke or heart-failure rehospitalisation (9.4% vs 10.6%), and superior for a composite of bioprosthetic-valve dysfunction (9.4% vs 41.6%). The difference in that second end point is almost entirely haemodynamic — a mean gradient of 7.7 vs 15.7 mmHg — so the trial establishes better valve performance in small annuli at one year, not better clinical outcomes or better long-term durability.

Keywords

Aortic stenosis; small aortic annulus; transcatheter aortic-valve replacement; TAVR; self-expanding valve; balloon-expandable valve; prosthesis–patient mismatch; bioprosthetic-valve dysfunction; valve haemodynamics

Key Takeaways

Background

TAVR is the dominant treatment for most patients with symptomatic severe aortic stenosis, yet the different prostheses have rarely been compared in randomised trials; the prior comparisons — CHOICE (n=241, high risk, older-generation devices) and SOLVE-TAVI (any annulus size, high risk) — showed no clinical difference.
Patients with a small aortic annulus account for up to one third of patients in TAVR randomised trials, with a marked preponderance of women. They are at greater risk of high residual gradients, prosthesis–patient mismatch, reduced exercise capacity and impaired prosthesis durability, and observational data link a mean gradient above 22.5 mmHg with excess 5-year mortality.
SMART was designed specifically for this group, powered for both a clinical composite and a haemodynamic/valve-dysfunction composite.

Methods

Design: international, prospective, post-market RCT at 83 sites in 13 countries, April 2021 to September 2022; 1:1 randomisation stratified by site and sex; all treated patients followed for 5 years.
Population: symptomatic severe aortic stenosis with an aortic-valve annulus area ≤430 mm² on multidetector CT and anatomy suitable for transfemoral TAVR with either platform. Bicuspid valves were eligible. An independent case-planning committee confirmed annular sizing before randomisation.
Devices: the supraannular self-expanding Evolut PRO/PRO+/FX (Medtronic) versus the balloon-expandable SAPIEN 3/3 Ultra (Edwards Lifesciences).
First coprimary end point: death from any cause, disabling stroke or rehospitalisation for heart failure through 12 months; assessed in the as-treated population; non-inferiority with a margin of 8 percentage points, one-sided alpha 0.05, reported with a 90% confidence interval.
Second coprimary end point: bioprosthetic-valve dysfunction through 12 months — a composite of haemodynamic structural valve dysfunction (mean gradient ≥20 mmHg), nonstructural valve dysfunction (severe prosthesis–patient mismatch or at least moderate total aortic regurgitation), clinical valve thrombosis, endocarditis or aortic-valve reintervention; assessed in the population with implantation; superiority, one-sided alpha 0.025, 95% confidence interval.
Secondary end points were tested hierarchically in a prespecified order: mean gradient and effective orifice area at 12 months, haemodynamic structural valve dysfunction, bioprosthetic-valve dysfunction in women, and moderate or severe prosthesis–patient mismatch at 30 days.

Results

Participants

737 patients randomised (366 self-expanding, 371 balloon-expandable); 716 treated; as-treated population 355 vs 361; population with implantation 350 vs 365; 642 of 655 living patients (98%) completed the 12-month visit.
Mean age 80 years; 86.7% women; mean STS-PROM 3.3%; 52.1% low, 34.2% intermediate, 13.7% high or prohibitive surgical risk; mean annulus area 381.9 ± 34.1 mm².
Most self-expanding implants were 26 mm (68.9%) or 29 mm (28.9%); most balloon-expandable implants were 23 mm (90.1%) or 20 mm (7.9%).

Coprimary End Points

End point	Self-expanding	Balloon-expandable	Difference (CI)
Death, disabling stroke or HF rehospitalisation at 12 mo	9.4%	10.6%	−1.2 pp (90% CI −4.9 to 2.5); P<0.001 for non-inferiority; HR 0.90 (0.56–1.43)
— Death	5.1%	5.9%	−0.7 pp (−3.5 to 2.1)
— Disabling stroke	3.1%	2.6%	0.6 pp (−1.5 to 2.7)
— HF rehospitalisation	3.8%	3.5%	0.4 pp (−2.0 to 2.7)
Bioprosthetic-valve dysfunction at 12 mo	9.4%	41.6%	−32.2 pp (95% CI −38.7 to −25.6); P<0.001 for superiority
— Haemodynamic structural valve dysfunction	3.2%	32.2%	−29.1 pp (−34.6 to −23.5)
— Nonstructural valve dysfunction	5.9%	18.2%	−12.3 pp (−17.6 to −7.0)
— Clinical valve thrombosis	0.3%	0.3%	0.0 pp (−0.8 to 0.8)
— Endocarditis	0.6%	2.3%	−1.7 pp (−3.5 to 0.1)
— Aortic-valve reintervention	0.9%	0.6%	0.3 pp (−1.0 to 1.6)

Both results were consistent under multiple imputation for missing data (first coprimary −1.0 pp, 90% CI −4.8 to 2.8; second coprimary −31.8 pp, 95% CI −38.5 to −25.1).

Secondary and Additional End Points

End point	Self-expanding	Balloon-expandable	Difference (95% CI)
Mean gradient at 12 mo	7.7 mmHg	15.7 mmHg	−8.0 (−8.9 to −7.1)
Effective orifice area at 12 mo	1.99 cm²	1.50 cm²	0.49 (0.42 to 0.56)
Haemodynamic structural valve dysfunction	3.5%	32.8%	−29.3 pp (−34.7 to −23.8)
Bioprosthetic-valve dysfunction in women	10.2%	43.3%	−33.0 pp (−40.2 to −25.9)
Moderate or severe prosthesis–patient mismatch at 30 d	11.2%	35.3%	−24.1 pp (−30.5 to −17.7)

Severe prosthesis–patient mismatch at 30 days 1.8% vs 7.1% (−5.3 pp, −8.6 to −1.9); mild or greater total aortic regurgitation at 12 months 14.1% vs 20.3% (−6.2 pp, −12.3 to −0.2); Doppler velocity index 0.63 vs 0.44.
KCCQ-improved at 12 months 74.9% vs 67.8% (difference 7.2 pp, 0.3 to 14.1).

Safety

Device success at 30 days: 85.2% vs 59.2% by VARC-2 criteria (difference 26.0 pp, 19.2 to 32.7) and 94.5% vs 86.6% by VARC-3 criteria (difference 7.9 pp, 3.5 to 12.4).
Death or disabling stroke through 12 months 6.8% vs 7.5% (HR 0.92, 0.53 to 1.59).
New permanent pacemaker at 30 days 12.1% vs 7.8% (HR 1.61, 95% CI 0.98 to 2.65) — numerically higher with the self-expanding valve, confidence interval crossing 1.
No apparent difference in coronary obstruction, conversion to surgery, major bleeding, acute kidney injury or hospital readmission at 30 days, or in cardiovascular death, reintervention, valve thrombosis or readmission at 12 months.

Subgroups

Effect direction was consistent across prespecified subgroups; bioprosthetic-valve dysfunction hazard ratios ranged from 0.09 to 0.36 across age, sex, STS-PROM, LVEF, atrial fibrillation and conduction-disease strata, and from 0.17 to 0.35 across the same strata for the clinical composite.
Confidence intervals were not adjusted for multiplicity and are presented for description only.

Discussion

The authors position SMART as a head-to-head, real-world comparison of the two most-used TAVR platforms, deliberately enriched for small annuli and including bicuspid valves, and argue that the haemodynamic advantage of the supraannular self-expanding design translates into less prosthesis–patient mismatch and less bioprosthetic-valve dysfunction at one year. They note that no safety difference was found, with pacemaker implantation numerically but not significantly higher.
They acknowledge that the dysfunction end point was built on the European standard definition of structural deterioration, later updated to add the VARC-3 prosthesis–patient mismatch definition and aortic reintervention, and report that results held across alternative definitions of bioprosthetic-valve dysfunction.

Limitations of the document

The second coprimary end point is a haemodynamic surrogate at 12 months, not valve failure. It is driven almost entirely by a gradient threshold (≥20 mmHg) plus prosthesis–patient mismatch and aortic regurgitation; the components that represent true clinical valve failure — thrombosis, endocarditis and reintervention — were rare and similar between groups. Whether a 1-year gradient difference predicts long-term structural deterioration is unproven.
Definition dependence. The 41.6% balloon-expandable dysfunction rate is a product of the chosen composite, which counts measured prosthesis–patient mismatch and moderate regurgitation; other definitions of dysfunction would move the absolute number substantially.
Populations differ between the two coprimary analyses (as-treated for the clinical composite, population with implantation for the dysfunction composite), and 21 randomised patients never reached a procedure.
Industry sponsorship reached into the analysis. Medtronic funded the trial, selected sites, monitored data and performed the statistical analyses.
One-year follow-up only. All patients are followed to 5 years; durability, reintervention and mortality differences are not addressed here.
Results apply only to the two platforms and the small-annulus population studied — they cannot be generalised to other TAVR devices or to larger annuli.
Pacemaker signal not excluded. The 12.1% vs 7.8% difference with a confidence interval touching 1 leaves the known conduction-risk disadvantage of the self-expanding platform unresolved rather than refuted.
Echo end points came from a core laboratory using mean gradient, which the authors note may differ from invasive haemodynamic measurement; both approaches have pitfalls.

Key Concepts Mentioned

Key Entities Mentioned

entities/Bicuspid-Aortic-Valve

Wiki Pages Updated

log.md carries the ingest entry.