Nejmoa 2307447

The n e w e ng l a n d j o u r na l of m e dic i n e
Original Article
Transcatheter Aortic-Valve Replacement

in Low-Risk Patients at Five Years
M.J. Mack, M.B. Leon, V.H. Thourani, P. Pibarot, R.T. Hahn,
P. Genereux, S.K. Kodali, S.R. Kapadia, D.J. Cohen, S.J. Pocock, M. Lu,
R. White, M. Szerlip, J. Ternacle, S.C. Malaisrie, H.C. Herrmann, W.Y. Szeto,
M.J. Russo, V. Babaliaros, C.R. Smith, P. Blanke, J.G. Webb, and R. Makkar,
for the PARTNER 3 Investigators*
A BS T R AC T
BACKGROUND
A previous analysis in this trial showed that among patients with severe, symptom- The authors’ full names, academic de-
atic aortic stenosis who were at low surgical risk, the rate of the composite end point grees, and affiliations are listed in the Ap-
pendix. Dr. Mack can be contacted at
of death, stroke, or rehospitalization at 1 year was significantly lower with trans- michael.mack@bswhealth.org or at 1100
catheter aortic-valve replacement (TAVR) than with surgical aortic-valve replacement. Allied Dr., Plano, TX 75093.
Longer-term outcomes are unknown. *A complete list of the PARTNER 3 Inves-
tigators is provided in the Supplemen-
METHODS tary Appendix, available at NEJM.org.
We randomly assigned patients with severe, symptomatic aortic stenosis and low This article was published on October 24,
surgical risk to undergo either TAVR or surgery. The first primary end point was 2023, at NEJM.org.
a composite of death, stroke, or rehospitalization related to the valve, the proce- N Engl J Med 2023;389:1949-60.
dure, or heart failure. The second primary end point was a hierarchical composite DOI: 10.1056/NEJMoa2307447
that included death, disabling stroke, nondisabling stroke, and the number of re- Copyright © 2023 Massachusetts Medical Society.
hospitalization days, analyzed with the use of a win ratio analysis. Clinical, echo-
cardiographic, and health-status outcomes were assessed through 5 years.
RESULTS
A total of 1000 patients underwent randomization: 503 patients were assigned to
undergo TAVR, and 497 to undergo surgery. A component of the first primary end
point occurred in 111 of 496 patients in the TAVR group and in 117 of 454 patients
in the surgery group (Kaplan–Meier estimates, 22.8% in the TAVR group and
27.2% in the surgery group; difference, −4.3 percentage points; 95% confidence
interval [CI], −9.9 to 1.3; P = 0.07). The win ratio for the second primary end point
was 1.17 (95% CI, 0.90 to 1.51; P = 0.25). The Kaplan–Meier estimates for the com-
ponents of the first primary end point were as follows: death, 10.0% in the TAVR
group and 8.2% in the surgery group; stroke, 5.8% and 6.4%, respectively; and
rehospitalization, 13.7% and 17.4%. The hemodynamic performance of the valve,
assessed according to the mean (±SD) valve gradient, was 12.8±6.5 mm Hg in the
TAVR group and 11.7±5.6 mm Hg in the surgery group. Bioprosthetic-valve failure
occurred in 3.3% of the patients in the TAVR group and in 3.8% of those in the
surgery group.
CONCLUSIONS
Among low-risk patients with severe, symptomatic aortic stenosis who underwent
TAVR or surgery, there was no significant between-group difference in the two
primary composite outcomes. (Funded by Edwards Lifesciences; PARTNER 3
ClinicalTrials.gov number, NCT02675114.)
n engl j med 389;21 nejm.org november 23, 2023 1949

The New England Journal of Medicine
Downloaded from nejm.org by Jose Carvalho on November 28, 2023. For personal use only. No other uses without permission.
Copyright © 2023 Massachusetts Medical Society. All rights reserved.
T
ranscatheter aortic-valve replace- Mortality (STS-PROM) score of less than 4%
ment (TAVR) has been increasingly used (with scores ranging from 0 to 100% and higher
as an alternative to surgery for treating scores indicating a greater risk of death within
patients with severe, symptomatic aortic steno- 30 days after the procedure) and on the basis of
A Quick Take is
sis.1,2 Randomized trials of both balloon-expand- assessment by the heart team. Patients also had
available at able and self-expanding TAVR valves have shown to be eligible for TAVR through transfemoral
NEJM.org that in patients at intermediate or high risk for access. The eligibility of all the patients was re-
death by 30 days after surgery, TAVR was either viewed and approved by a case review board. Key
noninferior or superior to surgical aortic-valve anatomical and clinical exclusions have been
replacement at 5 years of follow-up.3-11 In two reported previously and are provided in the Sup-
randomized trials involving younger patients who plementary Appendix, available at NEJM.org.13
were at low surgical risk, TAVR was either non- Details about the representativeness of the pa-
inferior or superior to surgery at 2 or 3 years.12-15 tients in the trial are also provided in the Sup-
The Placement of Aortic Transcatheter Valves plementary Appendix.
(PARTNER) 3 trial showed that the rate of the
composite end point of death, stroke, or rehos- Randomization, Procedures, and Follow-up
pitalization at 1 and 2 years was significantly Patients were assigned in a 1:1 ratio to undergo
lower with TAVR than with surgery.13-15 Here, we either TAVR with a SAPIEN 3 valve or surgical
report the 5-year outcomes in this trial. aortic-valve replacement with a commercially avail-
able bioprosthetic valve. The SAPIEN 3 system
and the procedures for TAVR and surgery have
Me thods
been described previously.13 Clinical outcomes and
Trial Design and Oversight transthoracic echocardiography data were assessed
This multicenter, randomized trial compared the at baseline, after the implantation procedure, at
use of the SAPIEN 3 transcatheter heart valve hospital discharge, 30 days, 6 months, 1 year, and
(Edwards Lifesciences) with surgical aortic-valve then annually to 5 years.
replacement in patients with severe, symptom-
atic aortic stenosis who were at low surgical risk. End Points
The trial design, details regarding oversight, and The original primary end point, assessed at 1 year,
results at 1 and 2 years have been published was a nonhierarchical composite of death from
previously.13,15 The trial protocol (available with any cause, stroke, or rehospitalization related to
the full text of this article at NEJM.org) was the procedure, the valve, or heart failure (see the
designed by the sponsor (Edwards Lifesciences), Supplementary Appendix). A time-to-first-event
with input from the trial steering committee and analysis was used to evaluate this end point.
the Food and Drug Administration, and was ap- However, some patients had more than one end-
proved by the institutional review board at each point event or more than one rehospitalization
site. The sponsor funded all trial-related activi- over the 5-year period. To better reflect the patient
ties and participated in site selection, data col- outcomes through 5 years, two primary end
lection, monitoring, and statistical analysis. The points were prespecified in the 5-year extension
trial leadership had unrestricted access to all the statistical analysis plan: the original nonhierar-
data, prepared all the drafts of the manuscript, chical composite end point and a hierarchical
and vouch for the accuracy and completeness of composite end point that included death from
the data and for the fidelity of the trial to the any cause, disabling stroke, nondisabling stroke,
protocol. and the number of rehospitalization days (see the
Supplementary Appendix). Secondary end points
Patients of interest at 5 years were death or disabling
Patients were eligible for inclusion if they had se- stroke, new-onset atrial fibrillation, aortic-valve
vere, symptomatic aortic stenosis and were consid- reintervention, endocarditis, and clinically signifi-
ered to be at low surgical risk on the basis of cant valve thrombosis; definitions are provided
clinical and anatomical assessment, including a in the Supplementary Appendix. Valve thrombosis
Society of Thoracic Surgeons Predicted Risk of was defined according to Valve Academic Re-
1950 n engl j med 389;21 nejm.org november 23, 2023

TAVR in Low-Risk Patients
search Consortium 3 (VARC-3) criteria as clini- use of the win ratio method (see the Supplemen-
cally significant bioprosthetic-valve dysfunction tary Appendix).18 The type I error was controlled
as assessed with echocardiography or contrast- between the two primary end points with the
enhanced computed tomography with either no use of the Hochberg method.19
(stage 1), moderate (stage 2), or severe (stage 3) Time-to-event analyses from baseline to 1 year,
hemodynamic valve deterioration.16 A clinical 1 to 5 years (landmark analysis), and baseline to
events committee adjudicated key 5-year clinical 5 years were performed, and hazard ratios and
outcomes, including all components of the pri- 95% confidence intervals were calculated for the
mary end points, valve thrombosis, and valve clinical end points (see the Supplementary Ap-
reintervention. Other secondary end points in- pendix). If there was clear evidence of nonpro-
cluded functional status and quality of life as portionality of hazards, the odds ratio and 95%
assessed with the Kansas City Cardiomyopathy confidence interval from the time-adjusted lo-
Questionnaire–Overall Summary (KCCQ-OS). gistic-regression model were also reported.20 For
KCCQ-OS scores range from 0 to 100, with continuous variables, the means and the differ-
higher scores indicating better health status. The ence between the means, along with the 95%
secondary end point of alive with a KCCQ-OS confidence intervals, were reported. For categor-
score of 75 or higher indicated the status of be- ical variables, the percentage of patients in each
ing alive and well. trial group, the difference in the percentages,
and the 95% confidence intervals were reported.
Echocardiographic Assessments The widths of the confidence intervals for con-
All echocardiograms were assessed by a core tinuous and categorical variables have not been
laboratory with the use of standard hemody- adjusted for multiplicity and should not be used
namic measures. Total aortic regurgitation and to infer definitive treatment effects. Additional
paravalvular aortic regurgitation were assessed methods are described in the Supplementary Ap-
with the use of a multiparametric integrative appendix.
proach.16 Valve durability was assessed with the All clinical end-point analyses were performed
use of the VARC-3 definition of bioprosthetic- in the as-treated population, which included the
valve failure, which includes the occurrence of patients who had undergone randomization and
valve reintervention, valve-related death, or dete- in whom the index procedure was initiated (see
rioration in hemodynamic valve function between the Supplementary Appendix). Echocardiographic
the day 30 and follow-up echocardiograms. All end-point analyses were performed in the valve-
potential cases of bioprosthetic-valve failure were implant population, which included the patients
adjudicated by a group of three experts for con- in whom the intended valve was implanted. All
firmation of the presence, stage, and cause of statistical analyses were performed with the use
valve failure.16 of SAS software, version 9.4 (SAS Institute).
Statistical Analysis
R e sult s
For the first primary end point (a nonhierarchi-
cal composite of death from any cause, stroke, Patients, Procedures, and Follow-up
or rehospitalization), we used the Wald test17 to A total of 1000 patients underwent randomiza-
determine the superiority of TAVR to surgery; tion at 71 clinical sites: 503 patients were assigned
the percentage of patients with an event in each to undergo transfemoral TAVR and 497 to undergo
group at 5 years was estimated with the Kaplan– surgery. The as-treated population included 496
Meier method, and Greenwood’s formula was used patients in the TAVR group and 454 in the sur-
to estimate standard errors. The odds ratio and gery group. A total of 948 patients (495 in the
95% confidence interval from the time-adjusted TAVR group and 453 in the surgery group) received
logistic-regression model were also calculated. the intended valve. Details regarding the implanted
The second primary end point (a hierarchal com- valve sizes and surgical valve types were pub-
posite that included death from any cause, dis- lished previously21 and are provided in Figure S1
abling stroke, nondisabling stroke, and the num- and Table S1 in the Supplementary Appendix. The
ber of rehospitalization days) was tested with the mean age of the patients was 73 years, 69.3% of

1000 Patients underwent randomization
503 Were assigned to undergo transcatheter

497 Were assigned to undergo surgery
aortic-valve replacement
7 Did not receive intervention 43 Did not receive intervention

1 Was found to have met 8 Were found to have met
exclusion criteria after exclusion criteria after
randomization randomization
6 Withdrew 35 Withdrew
496 Were included in the as-treated population 454 Were included in the as-treated population
11 Withdrew
1 Withdrew
1 Was lost to follow-up
495 (99.8%) Were included in the 442 (97.4%) Were included in the
1-yr follow-up 1-yr follow-up
12 Withdrew
3 Withdrew
3 Were lost to follow-up
3 Were lost to follow-up 2 Were lost to follow-up
Figure 1. Randomization and Follow-up.

Patients who met the composite primary end point of death, stroke, or rehospitalization related to the valve, the
procedure, or heart failure but who withdrew or were lost to follow-up are considered to have completed follow-up
because they had already died, had a stroke, or had rehospitalization before trial exit. Data from the vital-status
sweep are not shown.
the patients were men, and the mean STS-PROM shown in Figure 1. Follow-up data through 5 years
score was 1.9% (Table S2). Details regarding ran- were available for 91.6% of the patients, with a
domization and follow-up through 5 years are disproportional loss to follow-up in the surgery

A Death from Any Cause, Stroke, or Rehospitalization B Death from Any Cause
100 30 Hazard ratio, 0.79 (95% CI, 0.61–1.02) 100 30 Hazard ratio, 1.23 (95% CI, 0.79–1.90)
27.2
90 25 P=0.07 90 25
Surgery 22.8
80 20 80 20
Percentage of Patients
70 TAVR 70
15 15
60 10 60 10 TAVR 10.0
8.2
50 5 50 5
Surgery
40 0 40 0
30 0 12 24 36 48 60 30 0 12 24 36 48 60
20 20
10 10
0 0
0 12 24 36 48 60 0 12 24 36 48 60
Months since Procedure Months since Procedure
No. at Risk No. at Risk
Surgery 454 372 349 328 309 276 Surgery 454 427 409 394 379 346
TAVR 496 453 434 415 391 353 TAVR 496 490 478 460 438 405
C Stroke D Rehospitalization
100 30 Hazard ratio, 0.87 (95% CI, 0.51–1.48) 100 30 Hazard ratio, 0.75 (95% CI, 0.54–1.05)
90 25 90 25
80 20 80 20 Surgery
17.4
70 15 70 15 13.7
60 10 60 10
Surgery TAVR
50 6.4 50
5 5.8 5
TAVR
40 0 40 0
30 0 12 24 36 48 60 30 0 12 24 36 48 60
20 20
10 10
0 0
0 12 24 36 48 60 0 12 24 36 48 60
Surgery 454 416 397 378 361 329 Surgery 454 381 359 339 321 289
TAVR 496 486 468 450 428 391 TAVR 496 455 439 419 396 361
Figure 2. Kaplan–Meier Curves for the First Primary End Point and Its Components.
Panel A shows the Kaplan–Meier estimates of the first composite primary end point of death from any cause, stroke, or rehospitaliza-
tion, and Panels B, C, and D show the estimates for the components. Rehospitalization was defined as rehospitalization related to the
procedure, the valve, or heart failure. According to the statistical analysis plan, the analysis of the composite primary end point involved
the difference in the Kaplan–Meier estimates between the transcatheter aortic-valve replacement (TAVR) group and the surgery group,
calculated on the basis of the Wald test (difference, −4.3 percentage points; 95% CI, −9.9 to 1.3; P = 0.07). The odds ratio and 95% con
fidence interval for death from any cause were calculated because there was evidence of nonproportionality of hazards from baseline to
5 years (odds ratio, 1.24; 95% CI, 0.79 to 1.97). The inset in each panel shows the same data on an enlarged y axis.
group; follow-up data were available for 469 of Primary End Points
496 patients (94.6%) in the TAVR group and 401 The composite of death, stroke, or rehospitaliza-
of 454 (88.3%) in the surgery group. A vital-status tion related to the valve, the procedure, or heart
sweep yielded data for 66 of 95 patients who had failure (the first primary end point) occurred in
been lost to follow-up or had withdrawn from 111 of 496 patients in the TAVR group and in
the trial (21 patients assigned to the TAVR group 117 of 454 patients in the surgery group. The
and 45 assigned to the surgery group) (Fig. S2). Kaplan–Meier estimates were 22.8% in the TAVR
Therefore, vital status could be determined for 486 group and 27.2% in the surgery group (difference,
of 496 patients (98.0%) in the TAVR group and −4.3 percentage points; 95% confidence interval
441 of 454 patients (97.1%) in the surgery group. [CI], −9.9 to 1.3; P = 0.07; hazard ratio, 0.79; 95%

1954
Table 1. Key Clinical End Points.*
End Point Baseline to 5 Years 1 Year to 5 Years
TAVR Surgery Hazard Ratio TAVR Surgery Hazard Ratio

(N = 496) (N = 454) (95% CI) (N = 490) (N = 427) (95% CI)
no. of patients with event no. of patients with event

(Kaplan–Meier estimate, %) (Kaplan–Meier estimate, %)
Death, stroke, or rehospitalization† 111 (22.8) 117 (27.2) 0.79 (0.61–1.02)‡ 69 (15.7) 47 (13.7) 1.17 (0.81–1.70)
Death from any cause 48 (10.0) 34 (8.2) 1.23 (0.79–1.90)§ 43 (9.1) 23 (5.9) 1.61 (0.97–2.67)
Death from cardiovascular causes 26 (5.5) 21 (5.1) 1.08 (0.61–1.92)§ 22 (4.7) 12 (3.1) 1.58 (0.78–3.19)
Death from noncardiovascular causes 22 (4.8) 13 (3.3) 1.46 (0.74–2.90)§ 21 (4.6) 11 (2.8) 1.64 (0.79–3.41)
Stroke 27 (5.8) 27 (6.4) 0.87 (0.51–1.48) 21 (4.6) 12 (3.2) 1.49 (0.73–3.02)
The
Disabling stroke 13 (2.9) 11 (2.7) 1.03 (0.46–2.30) 12 (2.7) 6 (1.6) 1.73 (0.65–4.61)
Nondisabling stroke 15 (3.2) 16 (3.7) 0.82 (0.40–1.65) 10 (2.2) 6 (1.5) 1.41 (0.51–3.89)
Death or disabling stroke 55 (11.5) 41 (9.8) 1.17 (0.78–1.75)§ 50 (10.6) 27 (6.9) 1.60 (1.00–2.55)
Rehospitalization† 65 (13.7) 74 (17.4) 0.75 (0.54–1.05) 29 (6.9) 24 (6.9) 0.98 (0.57–1.69)
Aortic-valve reintervention 12 (2.6) 12 (3.0) 0.86 (0.39–1.92) 9 (2.0) 10 (2.6) 0.77 (0.31–1.90)
n engl j med 389;21

Endocarditis 6 (1.3) 8 (2.0) 0.65 (0.23–1.87) 5 (1.1) 6 (1.5) 0.72 (0.22–2.35)
Valve thrombosis¶ 12 (2.5) 1 (0.2) 10.52 (1.37–80.93) 10 (2.1) 1 (0.2) 8.72 (1.12–68.12)
New-onset atrial fibrillation‖** 55 (13.7) 155 (42.4) 0.25 (0.19–0.34) 21 (6.0) 5 (2.6) 2.30 (0.87–6.10)
New pacemaker‖** 63 (13.5) 43 (10.4) 1.33 (0.90–1.96) 25 (6.1) 18 (4.9) 1.22 (0.67–2.24)
nejm.org
Serious bleeding‖ 49 (10.2) 64 (14.8) 0.65 (0.45–0.95) 25 (5.6) 18 (5.1) 1.15 (0.63–2.11)
n e w e ng l a n d j o u r na l
Myocardial infarction‖ 10 (2.1) 18 (4.4) 0.48 (0.22–1.05) 6 (1.3) 10 (2.6) 0.51 (0.19–1.41)

of
Revascularization‖ 17 (3.7) 25 (6.0) 0.59 (0.32–1.09) 12 (2.7) 12 (3.2) 0.85 (0.38–1.88)

Percutaneous coronary intervention 16 (3.5) 20 (4.9) 0.69 (0.36–1.34) 11 (2.5) 12 (3.2) 0.78 (0.35–1.78)
Coronary-artery bypass grafting 2 (0.5) 5 (1.1) 0.36 (0.07–1.85) 1 (0.2) 0 —
november 23, 2023

* The total number of patients in each column header represents the number of patients at risk for death at the beginning of the interval. TAVR denotes transcatheter aortic-valve re-
m e dic i n e
placement.
† Rehospitalization was defined as rehospitalization related to the valve, the procedure, or heart failure.
‡ According to the statistical analysis plan, the analysis of the two composite primary end points involved the difference in the Kaplan–Meier estimates between the TAVR group and the
surgery group, calculated on the basis of the Wald test (difference, −4.3 percentage points; 95% CI, −9.9 to 1.3; P = 0.07).
§ The following odds ratios with 95% confidence intervals were calculated for end points that showed evidence of nonproportionality of hazards from baseline to year 5: odds ratio for
death from any cause, 1.24 (95% CI, 0.79 to 1.97); odds ratio for death from cardiovascular causes, 1.08 (95% CI, 0.60 to 1.95); odds ratio for death from noncardiovascular causes,
1.48 (95% CI, 0.74 to 2.97); and odds ratio for death or disabling stroke, 1.18 (95% CI, 0.77 to 1.81).
¶ Valve thrombosis was adjudicated according to Valve Academic Research Consortium 3 criteria.
‖ The outcome was reported by the trial site. Serious bleeding included events that led to death or another serious event; resulted in life-threatening illness, injury, or permanent impair-
ment; resulted in medical or surgical intervention; or resulted in hospitalization or prolongation of existing hospitalization.
** These categories exclude atrial fibrillation and pacemakers that were present at baseline.
CI, 0.61 to 1.02) (Fig. 2A and Table 1). These sistent with those of the primary analysis (Table
findings appeared to be consistent in all major S3). In a landmark analysis of years 1 to 5, a total
subgroups (Fig. S3). The win ratio for the second of 69 of 453 patients in the TAVR group and 47
primary end point (a hierarchical composite that of 372 patients in the surgery group had died or
included death, disabling stroke, nondisabling had had a stroke or rehospitalization. The Kaplan–
stroke, and the number of rehospitalization days) Meier estimates were 15.7% in the TAVR group
was 1.17 (95% CI, 0.90 to 1.51; P = 0.25) (Fig. 3). and 13.7% in the surgery group (hazard ratio,
The results of a sensitivity analysis that used mul- 1.17; 95% CI, 0.81 to 1.70) (Table 1 and Fig. S4).
tiple imputation for missing data and was adjusted The restricted mean event-free survival in the
for nonproportional hazards seemed to be con- analysis of the first primary end point at 5 years
TAVR Surgery
(N=496) (N=454)
496×454=225,184 Patient pairs
TAVR Wins Ties Surgery Wins
Hierarchical
Components:
1. Death from Any
6.9% 84.7% 8.4%
Cause
2. Disabling Stroke 1.4% 82.1% 1.2%
3. Nondisabling Stroke 2.9% 77.2% 2.0%
4. Rehospitalization
10.9% 58.9% 7.4%
Days
Total Wins 22.1% 19.0%
22.1
Win Ratio= = 1.17 (95% CI, 0.90–1.51)
19.0
P=0.25
Figure 3. Win Ratio Diagram for the Second Primary End Point.
Shown are the results of the win ratio analysis of the second primary end point (a hierarchical composite that in-
cluded death, disabling stroke, nondisabling stroke, and the number of rehospitalization days related to the valve,
the procedure, or heart failure).

was longer by 103 days (95% CI, 26 to 180) with group with thrombosis, hemodynamic valve de-
TAVR than with surgery (Table S4). terioration was absent (stage 1) in 4 patients, was
With respect to the individual components of moderate (stage 2) in 5 patients, and was severe
the first primary end point at 5 years, the Kaplan– (stage 3) in 3 patients (Table S11). Of the 13 pa-
Meier estimates were as follows: death from any tients with thrombosis, 7 had shortness of breath
cause, 10.0% in the TAVR group and 8.2% in the or dyspnea on exertion, 3 had a stroke (1 disabling
surgery group (odds ratio, 1.24; 95% CI, 0.79 to and 2 nondisabling), and 3 had no symptoms.
1.97); stroke, 5.8% and 6.4%, respectively (haz- The patient with thrombosis in the surgery group
ard ratio, 0.87; 95% CI, 0.51 to 1.48); and rehos- had no hemodynamic valve deterioration (stage
pitalization, 13.7% and 17.4% (hazard ratio, 0.75; 1) and had dyspnea on exertion. The percentages
95% CI, 0.54 to 1.05) (Fig. 2B, 2C, and 2D and of patients who received anticoagulation therapy
Table 1). The Kaplan–Meier estimates at 1 year are are provided in Table S12.
provided in Table S5. There were 82 deaths through
5 years of follow-up: 48 in the TAVR group (26 Echocardiographic Findings
from cardiovascular causes and 22 from noncar- At 5 years, the mean (±SD) aortic-valve gradient
diovascular causes) and 34 in the surgery group according to echocardiography was 12.8±6.5
(21 from cardiovascular causes and 13 from non- mm Hg in the TAVR group and 11.7±5.6 mm Hg
cardiovascular causes) (Fig. S5 and Tables S6 and in the surgery group; the mean aortic-valve area
S7). Three patients in the TAVR group and 1 pa- was 1.9±0.5 cm2 and 1.8±0.5 cm2 in the two
tient in the surgery group died from coronavirus groups, respectively (Fig. 4A and 4B). At 5 years,
disease 2019 (Covid-19). The 5-year mortality was aortic regurgitation of mild or greater severity
10.2% in the TAVR group and 9.0% in the sur- was present in 81 of 331 patients (24.5%) in the
gery group when additional patient information TAVR group and in 18 of 284 patients (6.3%) in
obtained from the vital-status sweep was includ- the surgery group; paravalvular aortic regurgita-
ed. Figure S6 shows the Kaplan–Meier curves for tion of mild or greater severity was present in 69
the landmark analysis at 1 year for death from any of 331 patients (20.8%) in the TAVR group and
cause, death from any cause with the inclusion of in 9 of 283 patients (3.2%) in the surgery group
data from the vital-status sweep, death from (Fig. S9). In the TAVR group, 5-year mortality
cardiovascular causes, and death from noncardio- was 9.1% among patients with no or trace para-
vascular causes. Additional 5-year data — includ- valvular aortic regurgitation at 30 days after the
ing data for stroke, disabling stroke, death, and procedure and 11.1% among those who had
rehospitalization — are provided in Table 1, Ta- mild paravalvular aortic regurgitation at 30 days
bles S5 through S8, and Figures S6, S7, and S8. after the procedure (hazard ratio, 0.78; 95% CI,
0.42 to 1.45) (Fig. S10). The Kaplan–Meier esti-
Secondary End Points mates of bioprosthetic-valve failure of any cause
Data regarding aortic-valve reintervention and were 3.3% in the TAVR group and 3.8% in the
endocarditis are provided in Table 1 and Table surgery group. The estimates of irreversible
S10. New-onset atrial fibrillation occurred in 55 stage 3 (severe) structural or hemodynamic valve
patients in the TAVR group and in 155 patients deterioration were 1.1% in the TAVR group and
in the surgery group (Kaplan–Meier estimates, 1.0% in the surgery group. The estimates of
13.7% and 42.4%, respectively). Serious bleeding aortic-valve reintervention were 2.2% and 2.6%,
occurred in 49 patients in the TAVR group and respectively. The estimates of valve-related death
in 64 patients in the surgery group. A new per- were 0.0% in the TAVR group and 0.2% in the
manent pacemaker was implanted in 13.5% of surgery group (Fig. 4C and 4D). The incidence of
the patients in the TAVR group and in 10.4% of bioprosthetic-valve failure related to structural
those in the surgery group (Table 1). Clinically valve deterioration was 1.4% in the TAVR group
significant valve thrombosis, according to VARC-3 and 2.0% in the surgery group (Table S13). At 5
criteria, occurred in 12 patients (2.5%) in the TAVR years, 392 of 454 patients (86.3%) in the TAVR
group and in 1 patient (0.2%) in the surgery group and 334 of 382 patients (87.4%) in the
group (Table 1). None of the patients with valve surgery group were alive and had a normally
thrombosis died. Of the 12 patients in the TAVR functioning valve.

Functional and Health Status tients.3-11 With respect to low-risk patients, out-
Functional outcomes appeared to be similar in comes from the PARTNER 3 trial were reported
the two groups. A total of 84.4% of the patients at 1 year and 2 years, and outcomes from a trial
in the TAVR group and 86.0% of those in the of TAVR with a self-expanding valve as com-
surgery group were alive and had New York Heart pared with surgery were reported at 1 and 3
Association (NYHA) class I or II heart failure at years.12-15 Those reports showed that TAVR re-
5 years (Fig. S11). Disease-specific health status sulted in similar or better early outcomes as
appeared to be similar in the two groups, with compared with surgery. Because low-risk pa-
a mean KCCQ-OS score of 86.2 in the TAVR tients are typically younger than high-risk pa-
group and 85.9 in the surgery group (Fig. 4E). At tients, longer-term results are critical to inform
5 years, 284 of 400 patients (71.0%) in the TAVR clinical decision making. We report the longer-
group and 238 of 331 patients (71.9%) in the term follow-up of low-risk patients undergoing
surgery group were alive with a KCCQ-OS score TAVR or surgery, with adjudicated clinical and
of 75 or higher (Fig. 4F). echocardiographic outcomes.
After the first year, there was an attenuation
of the differences between the TAVR group and
Discussion
the surgery group with respect to the nonhierar-
In this 5-year follow-up of the PARTNER 3 trial, chical composite primary end point, which had
the incidence of the composite end point of previously favored TAVR. There was a greater
death, stroke, or rehospitalization was similar in number of deaths among patients assigned to
the TAVR group and the surgery group; the inci- TAVR than among those assigned to surgery
dence of the individual components of the pri- from year 1 to year 5; these deaths were due to
mary end points (including death from any cause, both cardiovascular and noncardiovascular causes
disabling stroke, nondisabling stroke, and re- (Tables S6 and S7). Whether follow-up during the
hospitalization) was also similar in the two groups. Covid-19 pandemic disproportionately affected
The restricted mean event-free survival time over adverse outcomes could not be definitively deter-
5 years was longer in the TAVR group than in mined. The incidence of stroke at 5 years ap-
the surgery group, a result driven mainly by the peared to be similar in the two groups, as was
between-group difference in rehospitalization. the incidence of disabling and nondisabling
Aortic-valve durability according to VARC-3 defi- strokes, with most strokes being ischemic in
nitions of bioprosthetic-valve failure appeared to origin. Although the incidence of stroke at 5 years
be similar in the two groups at 5 years. Among was low, stroke remains one of the most serious
the secondary end points, atrial fibrillation and complications of aortic-valve replacement.24,25
bleeding appeared to be less frequent in the Valve durability is of critical importance, es-
TAVR group than in the surgery group, whereas pecially in younger patients. Hemodynamic valve
paravalvular aortic regurgitation, valve thrombo- performance of both TAVR and surgical valves
sis, and pacemaker implantation appeared to be seemed to be similar to that reported previously
less frequent in the surgery group. Functional at 2 years.26 The incidence of bioprosthetic-valve
and health-status outcomes assessed according failure and of the need for reintervention was
to NYHA class, KCCQ-OS score, and the percent- similar in the two groups at 5 years; these results
age of patients who were alive and well at 5 years are consistent with reported findings in inter-
appeared to be similar in the two groups. mediate-risk patients.27,28 A higher percentage of
TAVR has been widely adopted over the past patients in the TAVR group than in the surgery
decade largely owing to an abundance of clinical group had paravalvular aortic regurgitation of
evidence from randomized trials, resulting in mild or greater severity; however, mild aortic
twice as many patients with severe aortic steno- regurgitation was not associated with higher
sis being treated as compared with a decade mortality at 5 years in the TAVR group.29,30
ago.1,2,22,23 Comparative outcomes between TAVR Observed improvements in functional status
and surgery among patients who were followed and quality of life in the first year were greater
for 5 years and beyond have shown similar find- in the TAVR group than in the surgery group, a
ings in high-risk and intermediate-risk pa- finding most likely attributable to the more in-

A Aortic-Valve Gradient B Aortic-Valve Area

75 2.5
TAVR 1.9
Mean Gradient (mm Hg)
2.0 1.8
1.8 1.8 1.7 1.8
Mean Area (cm2)

50 49.4 1.7 1.7
1.8 1.8 1.8
48.3 1.7
1.5 Surgery
1.0 0.8
25
TAVR
13.7 13.6 13.4 12.7 0.8
12.7 12.8 0.5
11.2 11.6 11.8 11.8 11.3 11.7
Surgery
0 0.0
01 12 24 36 48 60 01 12 24 36 48 60
TAVR 483 492 474 437 372 348 329 TAVR 458 482 450 416 347 334 320
Surgery 442 432 391 360 304 305 282 Surgery 424 415 371 342 289 295 275
C Bioprosthetic-Valve Failure D Bioprosthetic-Valve Failure and Components at 5 Yr

100 15 TAVR Surgery
Hazard ratio, 0.86 (95% CI, 0.42–1.77) 100
90
80 10 10
70
60 8
5 Surgery 3.8
50 3.3
TAVR 6
40 0
30 0 12 24 36 48 60 3.8
4 3.3
20 2.6
2.2
10 2
1.1 1.0
0 0.2
0 12 24 36 48 60 0
Bioprosthetic- Irreversible Aortic-Valve Valve-
Months since Procedure Valve Failure Stage 3 Reintervention Related
No. at Risk from Any Hemodynamic Death
TAVR 496 489 475 454 430 392 Cause Valve
Surgery 454 426 407 390 369 334 Deterioration
E KCCQ-OS Scores F Patients Who Were Alive with KCCQ-OS Score ≥75
100 TAVR
88.9 89.9 89.5 89.1 87.4
90 86.2 100
88.1 87.9 88.2 87.2 85.9
80 70.3 90
Surgery
70 72.8 80
Mean Score
60 70.1 70
50 60
40 50 71.0 71.9
30 40
20 30
10 20
0 10
01 12 24 36 48 60
0
Months since Procedure TAVR Surgery
(N=400) (N=331)
No. at Risk
TAVR 493 491 481 444 406 381 354
Surgery 448 433 403 367 340 321 301

Figure 4 (facing page). Echocardiographic Outcomes,

in this trial will continue to be followed for 10
Bioprosthetic-Valve Failure, and Quality-of-Life Out- years to shed further light on the durability of
comes. both the transcatheter and surgical bioprosthetic
The mean aortic-valve gradients, shown in Panel A, valves.
and the mean aortic-valve areas, shown in Panel B, The main limitations of this trial have been
were assessed by an echocardiography core laborato- discussed previously.13,15 This report addresses
ry. I bars indicate standard deviations. Kaplan–Meier
estimates for bioprosthetic-valve failure, adjudicated
some of those limitations by focusing on longer-
according to Valve Academic Research Consortium 3 term clinical outcomes and valve durability. How-
criteria, are shown in Panel C. The inset in Panel C ever, other limitations remain, including the
shows the same data on an enlarged y axis. The com- constraints of a carefully defined trial popula-
ponents of bioprosthetic-valve failure at 5 years are tion, which excluded patients with poor trans-
shown in Panel D. The mean Kansas City Cardiomyop-
athy Questionnaire–Overall Summary (KCCQ-OS)
femoral access, bicuspid aortic valves, or other
scores are shown in Panel E, and the percentage of anatomical or clinical factors that increased the
patients who were alive with a KCCQ-OS score of risk of complications associated with either TAVR
75 or higher are shown in Panel F. KCCQ-OS scores or surgery. It is important to note, as reported
range from 0 to 100, with higher scores indicating previously, that more patients who underwent
better health status.
surgery than who underwent TAVR withdrew
from the trial, which potentially biased the find-
vasive nature of surgery and the longer recovery ings. To help address missing vital-status data, a
time. By 1 year, both groups had similar im- vital-status sweep was conducted to obtain in-
provements in NYHA functional class and mean formation about the patients who withdrew or
KCCQ-OS scores that were sustained to 5 years. were lost to follow-up; the data from this sweep
Furthermore, the percentage of patients who were reduced the mortality difference between the
alive with a KCCQ-OS score of 75 or higher (in- two groups. However, these data cannot correct
dicative of being well) appeared to be similar in for possible bias in underreporting of nonfatal
the two groups. events. Last, missing data regarding NYHA class,
Clinically significant valve thrombosis was KCCQ-OS score, and follow-up echocardiogra-
rare but occurred in more patients in the TAVR phy could not be fully accounted for with multiple
group than in the surgery group over the course imputation.
of 5 years. The reasons for the greater incidence Among patients with severe, symptomatic aor-
of valve thrombosis among TAVR patients re- tic stenosis at low surgical risk who underwent
main speculative, but this event did not appear TAVR or surgery, the incidence of the two primary
to affect valve durability at 5 years. It is possible composite end points appeared to be similar in
that the differences between the groups in the the two groups at 5 years of follow-up.
use of anticoagulation therapies during the first Supported by Edwards Lifesciences.
years after the procedure may have contributed Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
to the higher incidence of thrombosis in the A data sharing statement provided by the authors is available
TAVR group, but this is also unknown. Patients with the full text of this article at NEJM.org.
Appendix
The authors’ full names and academic degrees are as follows: Michael J. Mack, M.D., Martin B. Leon, M.D., Vinod H. Thourani, M.D.,
Philippe Pibarot, D.V.M., Ph.D., Rebecca T. Hahn, M.D., Philippe Genereux, M.D., Susheel K. Kodali, M.D., Samir R. Kapadia, M.D.,
David J. Cohen, M.D., Stuart J. Pocock, Ph.D., Michael Lu, Ph.D., Roseann White, Ph.D., Molly Szerlip, M.D., Julien Ternacle, M.D.,
S. Chris Malaisrie, M.D., Howard C. Herrmann, M.D., Wilson Y. Szeto, M.D., Mark J. Russo, M.D., Vasilis Babaliaros, M.D., Craig R.
Smith, M.D., Philipp Blanke, M.D., John G. Webb, M.D., and Raj Makkar, M.D.
The authors’ affiliations are as follows: Baylor Scott and White Health, Plano, TX (M.J.M., M.S.); Columbia University (M.B.L.,
R.T.H., S.H.K., C.R.S.) and the Cardiovascular Research Foundation (M.B.L., R.T.H., S.H.K., D.J.C., C.R.S.), New York, and St. Francis
Hospital and Heart Center, Roslyn (D.J.C.) — all in New York; Marcus Heart Valve Center, Piedmont Heart Institute (V.H.T.), and
Emory University (V.B.) — both in Atlanta; Laval University, Quebec, QC (P.P.), and St. Paul’s Hospital, University of British Columbia,
Vancouver (P.B., J.G.W.) — both in Canada; Morristown Medical Center, Morristown (P.G.), and Robert Wood Johnson University
Hospital, New Brunswick (M.J.R.) — both in New Jersey; Cleveland Clinic, Cleveland (S.R.K.); London School of Hygiene and Tropical
Medicine, London (S.J.P.); Edwards Lifesciences, Irvine (M.L., R.W.), and Cedars–Sinai Medical Center, Los Angeles (R.M.) — both in
California; Heart Valve Unit, Haut-Lévêque Cardiological Hospital, Bordeaux University, Pessac, France (J.T.); Northwestern University,
Chicago (S.C.M.); and the University of Pennsylvania, Philadelphia (H.C.H., W.Y.S.).

References
1. Otto CM, Nishimura RA, Bonow RO, transcatheter versus surgical aortic valve 21. Pibarot P, Salaun E, Dahou A, et al.
et al. 2020 ACC/AHA guideline for the replacement in high-risk patients. J Am Echocardiographic results of transcathe-
management of patients with valvular Coll Cardiol 2018;72:2687-96. ter versus surgical aortic valve replace-
heart disease: a report of the American 11. Makkar RR, Thourani VH, Mack MJ, ment in low-risk patients: the PARTNER 3
College of Cardiology/American Heart et al. Five-year outcomes of transcatheter trial. Circulation 2020;141:1527-37.
Association Joint Committee on Clinical or surgical aortic-valve replacement. N Engl 22. Carroll JD, Mack MJ, Vemulapalli S,
Practice Guidelines. Circulation 2021; J Med 2020;382:799-809. et al. STS-ACC TVT Registry of transcath-
143(5):e72-e227. 12. Popma JJ, Deeb GM, Yakubov SJ, et al. eter aortic valve replacement. J Am Coll
2. Vahanian A, Beyersdorf F, Praz F, Transcatheter aortic-valve replacement with Cardiol 2020;76:2492-516.
et al. 2021 ESC/EACTS guidelines for the a self-expanding valve in low-risk patients. 23. Sharma T, Krishnan AM, Lahoud R,
management of valvular heart disease. N Engl J Med 2019;380:1706-15. Polomsky M, Dauerman HL. National
Eur Heart J 2022;43:561-632. 13. Mack MJ, Leon MB, Thourani VH, et al. trends in TAVR and SAVR for patients
3. Leon MB, Smith CR, Mack M, et al. Transcatheter aortic-valve replacement with with severe isolated aortic stenosis. J Am
Transcatheter aortic-valve implantation a balloon-expandable valve in low-risk pa- Coll Cardiol 2022;80:2054-6.
for aortic stenosis in patients who cannot tients. N Engl J Med 2019;380:1695-705. 24. Kapadia SR, Makkar R, Leon M, et al.
undergo surgery. N Engl J Med 2010;363: 14. Forrest JK, Deeb GM, Yakubov SJ, et al. Cerebral embolic protection during trans-
1597-607. 3-Year outcomes after transcatheter or catheter aortic-valve replacement. N Engl
4. Smith CR, Leon MB, Mack MJ, et al. surgical aortic valve replacement in low- J Med 2022;387:1253-63.
Transcatheter versus surgical aortic-valve risk patients with aortic stenosis. J Am 25. Whitlock RP, Belley-Cote EP, Paparel-
replacement in high-risk patients. N Engl Coll Cardiol 2023;81:1663-74. la D, et al. Left atrial appendage occlusion
J Med 2011;364:2187-98. 15. Leon MB, Mack MJ, Hahn RT, et al. during cardiac surgery to prevent stroke.
5. Adams DH, Popma JJ, Reardon MJ. Outcomes 2 years after transcatheter aor- N Engl J Med 2021;384:2081-91.
Transcatheter aortic-valve replacement with tic valve replacement in patients at low 26. Sathananthan J, Hensey M, Landes U,
a self-expanding prosthesis. N Engl J Med surgical risk. J Am Coll Cardiol 2021;77: et al. Long-term durability of transcathe-
2014;371:967-8. 1149-61. ter heart valves: insights from bench test-
6. Leon MB, Smith CR, Mack MJ, et al. 16. VARC-3 Writing Committee. Valve Aca- ing to 25 years. JACC Cardiovasc Interv
Transcatheter or surgical aortic-valve re- demic Research Consortium 3: updated 2020;13:235-49.
placement in intermediate-risk patients. endpoint definitions for aortic valve clini- 27. Pibarot P, Ternacle J, Jaber WA, et al.
N Engl J Med 2016;374:1609-20. cal research. J Am Coll Cardiol 2021;77: Structural deterioration of transcatheter
7. Reardon MJ, Van Mieghem NM, Pop- 2717-46. versus surgical aortic valve bioprostheses
ma JJ, et al. Surgical or transcatheter 17. Zee J, Xie SX. The Kaplan-Meier in the PARTNER-2 trial. J Am Coll Cardiol
aortic-valve replacement in intermediate- method for estimating and comparing 2020;76:1830-43.
risk patients. N Engl J Med 2017;376:1321- proportions in a randomized controlled 28. Madhavan MV, Kodali SK, Thourani
31. trial with dropouts. Biostat Epidemiol VH, et al. Outcomes of SAPIEN 3 trans-
8. Mack MJ, Leon MB, Smith CR, et al. 2018;2:23-33. catheter aortic valve replacement com-
5-Year outcomes of transcatheter aortic 18. Pocock SJ, Ariti CA, Collier TJ, Wang pared with surgical valve replacement in
valve replacement or surgical aortic valve D. The win ratio: a new approach to the intermediate-risk patients. J Am Coll Car-
replacement for high surgical risk pa- analysis of composite endpoints in clini- diol 2023;82:109-23.
tients with aortic stenosis (PARTNER 1): cal trials based on clinical priorities. Eur 29. Okuno T, Tomii D, Heg D, et al. Five-
a randomised controlled trial. Lancet 2015; Heart J 2012;33:176-82. year outcomes of mild paravalvular regur-
385:2477-84. 19. Quan H, Luo X, Capizzi T. Multiplicity gitation after transcatheter aortic valve
9. Kapadia SR, Leon MB, Makkar RR, adjustment for multiple endpoints in clin- implantation. EuroIntervention 2022;18:
et al. 5-Year outcomes of transcatheter ical trials with multiple doses of an active 33-42.
aortic valve replacement compared with treatment. Stat Med 2005;24:2151-70. 30. Pibarot P, Hahn RT, Weissman NJ,
standard treatment for patients with inop- 20. Grambsch PM, Therneau TM. Propor- et al. Association of paravalvular regurgi-
erable aortic stenosis (PARTNER 1): a ran- tional hazards tests and diagnostics tation with 1-year outcomes after trans-
domised controlled trial. Lancet 2015;385: based on weighted residuals. Biometrika catheter aortic valve replacement with the
2485-91. 1994;81:515-26 (https://academic.oup.com/ SAPIEN 3 valve. JAMA Cardiol 2017; 2:
10. Gleason TG, Reardon MJ, Popma JJ, biomet/a rticle-abstract/81/3/515/ 1208-16.
et al. 5-Year outcomes of self-expanding 257037?redirectedFrom=f ulltext). Copyright © 2023 Massachusetts Medical Society.


Nejmoa 2307447

Uploaded by

Copyright:

Available Formats

Nejmoa 2307447

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nejmoa 2307447

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e

Transcatheter Aortic-Valve Replacement

n engl j med 389;21 nejm.org november 23, 2023 1949

1950 n engl j med 389;21 nejm.org november 23, 2023

The New England Journal of Medicine

n engl j med 389;21 nejm.org november 23, 2023 1951

1000 Patients underwent randomization

503 Were assigned to undergo transcatheter

7 Did not receive intervention 43 Did not receive intervention

Figure 1. Randomization and Follow-up.

1952 n engl j med 389;21 nejm.org november 23, 2023

The New England Journal of Medicine

n engl j med 389;21 nejm.org november 23, 2023 1953

End Point Baseline to 5 Years 1 Year to 5 Years

TAVR Surgery Hazard Ratio TAVR Surgery Hazard Ratio

no. of patients with event no. of patients with event

n engl j med 389;21

The New England Journal of Medicine

Revascularization‖ 17 (3.7) 25 (6.0) 0.59 (0.32–1.09) 12 (2.7) 12 (3.2) 0.85 (0.38–1.88)

november 23, 2023

Copyright © 2023 Massachusetts Medical Society. All rights reserved.

496×454=225,184 Patient pairs

TAVR Wins Ties Surgery Wins

2. Disabling Stroke 1.4% 82.1% 1.2%

3. Nondisabling Stroke 2.9% 77.2% 2.0%

Total Wins 22.1% 19.0%

n engl j med 389;21 nejm.org november 23, 2023 1955

1956 n engl j med 389;21 nejm.org november 23, 2023

The New England Journal of Medicine

n engl j med 389;21 nejm.org november 23, 2023 1957

A Aortic-Valve Gradient B Aortic-Valve Area

Mean Area (cm2)

C Bioprosthetic-Valve Failure D Bioprosthetic-Valve Failure and Components at 5 Yr

1958 n engl j med 389;21 nejm.org november 23, 2023

The New England Journal of Medicine

Figure 4 (facing page). Echocardiographic Outcomes,

n engl j med 389;21 nejm.org november 23, 2023 1959

1960 n engl j med 389;21 nejm.org november 23, 2023

The New England Journal of Medicine

You might also like