new england

The
journal of medicine
established in 1812 January 9, 2020 vol. 382  no. 2

Early Surgery or Conservative Care for Asymptomatic

Aortic Stenosis
Duk‑Hyun Kang, M.D., Ph.D., Sung‑Ji Park, M.D., Ph.D., Seung‑Ah Lee, M.D., Sahmin Lee, M.D., Ph.D.,
Dae‑Hee Kim, M.D., Ph.D., Hyung‑Kwan Kim, M.D., Ph.D., Sung‑Cheol Yun, Ph.D., Geu‑Ru Hong, M.D., Ph.D.,
Jong‑Min Song, M.D., Ph.D., Cheol‑Hyun Chung, M.D., Ph.D., Jae‑Kwan Song, M.D., Ph.D.,
Jae‑Won Lee, M.D., Ph.D., and Seung‑Woo Park, M.D., Ph.D.​​

a bs t r ac t

BACKGROUND
The timing and indications for surgical intervention in asymptomatic patients with From the Division of Cardiology (D.-H.
severe aortic stenosis remain controversial. Kang, S.-A.L., S.L., D.-H. Kim, J.-M.S.,
J.-K.S.) and the Departments of Cardio-
thoracic Surgery (C.-H.C., J.-W.L.) and
METHODS Biostatistics (S.-C.Y.), Asan Medical Cen-
In a multicenter trial, we randomly assigned 145 asymptomatic patients with very ter, College of Medicine, University of
severe aortic stenosis (defined as an aortic-valve area of ≤0.75 cm2 with either an Ulsan, the Division of Cardiology, Sam-
sung Medical Center, Sungkyunkwan Uni-
aortic jet velocity of ≥4.5 m per second or a mean transaortic gradient of ≥50 mm Hg) versity School of Medicine (S.-J.P., S.-W.P.),
to early surgery or to conservative care according to the recommendations of cur- the Division of Cardiology, Severance Hos-
rent guidelines. The primary end point was a composite of death during or within pital (G.-R.H.), and the Cardiovascular
Center, Seoul National University Hospi-
30 days after surgery (often called operative mortality) or death from cardiovascular tal (H.-K.K.) — all in Seoul, South Korea.
causes during the entire follow-up period. The major secondary end point was death Address reprint requests to Dr. Kang at
from any cause during follow-up. the Division of Cardiology, Asan Medical
Center, College of Medicine, University of
RESULTS Ulsan, 388-1, Poongnap-dong, Songpa-gu,
Seoul, South Korea, or at ­dhkang@​­amc​
In the early-surgery group, 69 of 73 patients (95%) underwent surgery within 2 months .­seoul​.­kr, or to Dr. S.-W. Park at the Heart
after randomization, and there was no operative mortality. In an intention-to-treat Vascular Stroke Institute, Samsung Med-
analysis, a primary end-point event occurred in 1 patient in the early-surgery group ical Center, Sungkyunkwan University
School of Medicine, 81 Irwon-ro, Gang-
(1%) and in 11 of 72 patients in the conservative-care group (15%) (hazard ratio, nam-gu, Seoul, South Korea, or at ­s​.­woo
0.09; 95% confidence interval [CI], 0.01 to 0.67; P = 0.003). Death from any cause ​.­park@​­samsung​.­com.
occurred in 5 patients in the early-surgery group (7%) and in 15 patients in the con- This article was published on November
servative-care group (21%) (hazard ratio, 0.33; 95% CI, 0.12 to 0.90). In the con- 16, 2019, at NEJM.org.
servative-care group, the cumulative incidence of sudden death was 4% at 4 years N Engl J Med 2020;382:111-9.
and 14% at 8 years. DOI: 10.1056/NEJMoa1912846
Copyright © 2019 Massachusetts Medical Society.
CONCLUSIONS
Among asymptomatic patients with very severe aortic stenosis, the incidence of the
composite of operative mortality or death from cardiovascular causes during the
follow-up period was significantly lower among those who underwent early aortic-
valve replacement surgery than among those who received conservative care. (Fund-
ed by the Korean Institute of Medicine; RECOVERY ClinicalTrials.gov number,
NCT01161732.)

n engl j med 382;2 nejm.org  January 9, 2020 111

The New England Journal of Medicine
Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

A 
ortic stenosis is the most common trial protocol (available with the full text of this
valvular disease for which surgery is in- article at NEJM.org) was designed by the principal
dicated in developed countries, and the investigator and approved by the institutional re-
prevalence of this condition is increasing because view board at each participating center. The Ko-
of the aging population.1 Aortic-valve replacement rean Institute of Medicine, which supported the
is the only effective therapy for severe symptom- trial coordinators during the first 2 years of the
atic aortic stenosis, and despite limited data from trial, had no role in the collection, analysis, or
randomized clinical trials, current guidelines rec- interpretation of the data, writing of the manu-
ommend aortic-valve replacement because of the script, or any other aspect of the trial. The trial
dismal natural history of this disorder.1,2 Although received no other external sources of funding.
one third to one half of patients with severe aortic The trial was conducted in accordance with
stenosis are asymptomatic at the time of diagno- the principles of the Declaration of Helsinki.
sis,3,4 appropriate timing of intervention for these An independent clinical-events committee ad-
patients remains controversial.5,6 On the basis of judicated all serious adverse events, and a data
a consensus opinion that the potential benefit of and safety monitoring board oversaw the safety
aortic-valve replacement to prevent sudden death of the trial. The first draft of the manuscript
in asymptomatic patients (which has an incidence was prepared by the first author and was re-
of approximately 1% per year) may not be greater viewed and edited by all the authors. All the
than the risk of death during or within 30 days authors made the decision to submit the manu-
after surgery (often called operative mortality) and script for publication and vouch for the accu-
death related to the aortic-valve prosthesis, obser- racy and completeness of the data and for the
vation is recommended for the majority of asymp- fidelity of the trial to the protocol.
tomatic patients with severe aortic stenosis, and
aortic-valve replacement is recommended once Patient Selection
symptoms develop.1,2,4 However, advances in sur- We screened consecutive patients who were 20 to
gical techniques and aortic-valve prostheses may 80 years of age and who presented with very se-
change the risk-to-benefit ratio of aortic-valve re- vere aortic stenosis, which was assessed by means
placement, especially among patients at low sur- of transthoracic echocardiography. In accordance
gical risk.7-9 with the 1998 American College of Cardiology–
The Randomized Comparison of Early Surgery American Heart Association (ACC–AHA) guide-
versus Conventional Treatment in Very Severe lines and a traditional definition of severe aortic
Aortic Stenosis (RECOVERY) trial was designed stenosis,10,11 we defined very severe aortic steno-
to compare long-term clinical outcomes of early sis as an aortic-valve area of 0.75 cm2 or less
surgical aortic-valve replacement with those of a with either a peak aortic jet velocity of 4.5 m per
conservative care strategy based on current guide- second or greater or a mean transaortic gradient
lines in asymptomatic patients with very severe of 50 mm Hg or greater.
aortic stenosis (transvalvular velocity ≥4.5 m per In accordance with the 2006 ACC–AHA guide-
second). The major hypothesis of this trial was lines on surgical indications for severe aortic
that the incidence of death from cardiovascular stenosis,12 patients were excluded if they had ex-
causes would be lower among patients who under- ertional dyspnea, syncope, presyncope or angina,
went early surgery than among those who received a left ventricular ejection fraction of less than 50%,
conservative care. or clinically significant aortic regurgitation or
mitral valve disease or if they had undergone
cardiac surgery. Exercise testing was selectively
Me thods
performed to evaluate patients with nonspecific
Trial Design and Oversight symptoms, and patients with a positive exercise
We conducted this multicenter, randomized, par- test were excluded. We also excluded patients who
allel-group, open-label trial involving asymptom- were not candidates for early surgery because of
atic patients with very severe aortic stenosis who age (>80 years) or a medical condition such as
were candidates for either early surgery or conser- cancer. All the participants provided written in-
vative care at four medical centers in Korea. The formed consent.

112 n engl j med 382;2 nejm.org  January 9, 2020

The New England Journal of Medicine

Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 Early Surgery for Asymptomatic Aortic Stenosis

Trial Procedures by a log-rank survival power analysis, we also

Eligibility was determined after each patient un- assumed that an enrollment period of 2 years
derwent a thorough evaluation of symptoms and would be needed to complete enrollment and that
medical records and results of echocardiography the rate of loss to follow-up would be 10%.
and exercise testing were reviewed. Details re- Analyses were performed on an intention-to-
garding echocardiography and exercise testing are treat basis. Differences between the treatment
provided in the Supplementary Appendix, available groups were evaluated with the use of Student’s
at NEJM.org. Patients were randomly assigned in t-test for continuous variables and Fisher’s exact
a 1:1 ratio to early surgery or conservative care test for categorical variables. Because random-
with the use of a Web-based interactive response ization was stratified according to the partici-
system. The assignment to each treatment group pating center, we analyzed the end points with
was computer-generated and stratified according the use of stratified Cox proportional-hazards re-
to the participating center by means of a permuted- gression with Firth correction. Estimates of cu-
block sequence with variable block size. mulative incidences were calculated by the Kaplan–
The protocol specified that patients assigned Meier method and were compared with the use
to the early-surgery group should undergo aortic- of the log-rank test. For the Kaplan–Meier analy-
valve replacement within 2 months after random- sis, we analyzed all clinical events according to
ization. Patients assigned to the conservative-care the time to the first event. Hazard ratios with
group received treatment according to the ACC– 95% confidence intervals were derived with the
AHA guidelines,2,12 and they were referred for use of the stratified Cox proportional-hazards
surgery if they became symptomatic during fol- model with Firth correction. The 95% confidence
low-up, if the left ventricular ejection fraction intervals have not been adjusted for multiple com-
decreased to less than 50%, or if the peak aortic parisons, and therefore inferences drawn from
jet velocity increased each year by more than 0.5 m these intervals may not be reproducible.
per second on follow-up echocardiography. Details For the primary end-point analysis, we also
regarding surgical procedures and patient follow- performed a competing-risk analysis in which
up are provided in the Supplementary Appendix. death due to a cause other than a cardiovascular
cause was considered as a competing risk, and
Trial End Points hazard ratios with 95% confidence intervals were
The primary end point was a composite of op- calculated with the use of the method of Fine and
erative mortality or death from cardiovascular Gray. The cumulative incidences of the primary
causes during the follow-up period (continuing end point were also compared between the treat-
until 4 years after the last patient was enrolled). ment groups with the use of Gray’s test. Subgroup
Prespecified secondary end points included death analyses were performed to determine whether
from any cause, repeat aortic-valve surgery, clini- the result of the primary end point was consistent
cal thromboembolic events, and hospitalization in two prespecified subgroups defined according
for heart failure during follow-up. Specific defi- to peak aortic velocity and cause of aortic stenosis.
nitions of trial end points are provided in the A per-protocol analysis of the primary end point
Supplementary Appendix. was also performed. All reported P values were
two-sided, and a P value of less than 0.05 was
Statistical Analysis considered to indicate statistical significance. We
On the basis of our previous observational study,13 used SAS software, version 9.4 (SAS Institute), for
we estimated that a sample of 144 patients would statistical analyses.
provide the trial with 80% power, at a two-sided
significance level of 0.05, to detect a significant R e sult s
difference with respect to the primary end point,
assuming that the incidence of the primary end Patients
point would be 16% in the conservative-care group From July 2010 through April 2015, a total of 273
and 2% in the early-surgery group during a follow- asymptomatic patients with very severe aortic
up period that continued until 4 years after the last stenosis were screened for participation in the
patient was enrolled. In calculating the sample size trial, and 145 were enrolled and were randomly

n engl j med 382;2 nejm.org  January 9, 2020 113

The New England Journal of Medicine
Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

1031 Patients had very severe

aortic stenosis

758 Had symptomatic stenosis

273 Had asymptomatic stenosis

128 Were excluded

11 Had left ventricular rejection fraction <50%
21 Had cancer
14 Had mitral valve disease
16 Had severe aortic regurgitation
39 Were >80 yr of age
3 Had positive exercise test
24 Declined to participate

145 Underwent randomization

72 Were assigned to conservative-care

73 Were assigned to early-surgery group
group

73 Were included in the intention-to-treat 72 Were included in the intention-to-treat

analysis analysis

2 Crossed over to early surgery

4 Crossed over to conservative
1 Underwent transcatheter
care
aortic-valve replacement

69 Were included in the per-protocol 69 Were included in the per-protocol

analysis analysis

Figure 1. Enrollment, Randomization, and Follow-up.

Very severe aortic stenosis was defined as an aortic-valve area of 0.75 cm2 or less with a peak aortic jet velocity of
at least 4.5 m per second or a mean transaortic gradient of at least 50 mm Hg. Of the four patients in the early-sur-
gery group who crossed over to conservative care and were excluded from the per-protocol analysis, three under-
went surgery later after the development of symptoms and one did not undergo surgery.

assigned (73 patients to early surgery and 72 pa- The treatment groups were generally well bal-
tients to conservative care) (Fig. 1). Baseline char- anced with regard to baseline clinical character-
acteristics of the excluded patients are listed in istics (Table 1). The mean (±SD) age of the patients
Table S1 in the Supplementary Appendix. After was 64.2±9.4 years, and 49% were men. The cause
randomization, 2 patients assigned to conservative of aortic stenosis was a bicuspid aortic valve in
care crossed over to early surgery and 4 patients 88 patients (61%), degenerative valvular disease in
assigned to early surgery crossed over to conser- 48 (33%), and rheumatic valvular disease in 9 (6%).
vative care; surgical aortic-valve replacement in The mean peak aortic jet velocity was 5.1±0.5 m
these 4 patients was later performed after devel- per second, and the mean aortic-valve area was
opment of symptoms in 3 patients and was not 0.63±0.09 cm2. Medication use at baseline was
attempted in 1. also similar in the two groups.

114 n engl j med 382;2 nejm.org  January 9, 2020

The New England Journal of Medicine

Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 Early Surgery for Asymptomatic Aortic Stenosis

Table 1. Characteristics of the Patients at Baseline.*

Characteristic Conservative Care (N = 72) Early Surgery (N = 73)

Age — yr 63.4±10.7 65.0±7.8
Male sex — no. (%) 34 (47) 37 (51)
Body-surface area — m2 1.64±0.17 1.69±0.17
Body-mass index† 24.0±2.6 24.7±3.4
Diabetes — no. (%)   7 (10) 13 (18)
Hypertension — no. (%) 39 (54) 40 (55)
Smoking — no. (%) 21 (29) 19 (26)
Hypercholesterolemia — no. (%) 42 (58) 41 (56)
Coronary artery disease — no./total no. (%)‡ 1/59 (2) 5/72 (7)
Previous PCI — no. (%) 1 (1) 3 (4)
Previous stroke — no. (%) 3 (4) 3 (4)
Peripheral vascular disease — no. (%) 2 (3) 1 (1)
Atrial fibrillation — no. (%) 6 (8) 3 (4)
Serum creatinine level — mg/dl 0.83±0.16 0.84±0.23
EuroSCORE II score — %§ 0.9±0.4 0.9±0.3
Medication — no. (%)
Angiotensin-converting–enzyme inhibitor 0 4 (5)
Angiotensin-receptor blocker 28 (39) 24 (33)
Calcium antagonist 20 (28) 19 (26)
Beta-blocker   8 (11) 13 (18)
Diuretic 17 (24) 13 (18)
Statin 32 (44) 34 (47)
Echocardiographic findings
Cause of aortic stenosis — no. (%)
Bicuspid aortic valve 39 (54) 49 (67)
Degenerative valvular disease 26 (36) 22 (30)
Rheumatic valvular disease   7 (10) 2 (3)
Peak aortic jet velocity — m/sec 5.04±0.44 5.14±0.52
Transaortic pressure gradient — mm Hg
Peak 102.5±18.4 106.9±21.9
Mean  62.7±12.4  64.3±14.4
Aortic valve
Area — cm2 0.64±0.09 0.63±0.09
2/m2
Area index — cm 0.39±0.07 0.38±0.06
Left ventricular mass index — g/m2 133.7±31.1 135.6±38.2
Left ventricular ejection fraction — % 64.8±4.1 64.8±5.2

* Plus–minus values are means ±SD. To convert values for serum creatinine to micromoles per liter, multiply by 88.4.
There were no significant between-group differences. PCI denotes percutaneous coronary intervention.
† The body-mass index is the weight in kilograms divided by the square of the height in meters.
‡ Coronary computed tomography or coronary angiography was performed before aortic-valve replacement in 59 patients
in the conservative-care group and in 72 patients in the early-surgery group.
§ Scores on the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II), which measures patient risk
at the time of cardiovascular surgery, are calculated by means of a logistic-regression equation and range from 0 to 100%,
with higher scores indicating greater risk.

n engl j med 382;2 nejm.org  January 9, 2020 115

The New England Journal of Medicine
Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 2. Primary and Secondary End Points.

Conservative Care Early Surgery Hazard Ratio

Outcome (N = 72) (N = 73) (95% CI)*

number (percent)
Primary end point: operative mortality or death from 11 (15) 1 (1) 0.09 (0.01–0.67)‡
cardiovascular causes during follow-up†
Secondary end points
Death from any cause 15 (21) 5 (7) 0.33 (0.12–0.90)§
Clinical thromboembolic event 4 (6) 1 (1) 0.30 (0.04–2.31)§
Stroke 3 1
Myocardial infarction 1 0
Repeat aortic-valve surgery 2 (3) 0 0.19 (0.10–8.00)§
Hospitalization for heart failure 8 (11) 0 0.05 (0.00–1.05)§

* The 95% confidence intervals have not been adjusted for multiple comparisons, and therefore inferences drawn from
these intervals may not be reproducible.
† Operative mortality was defined as death during or within 30 days after surgery.
‡ This hazard ratio was calculated with the use of a Fine and Gray competing-risks analysis.
§ This hazard ratio was calculated with the use of stratified Cox proportional-hazards models with Firth correction.

Aortic-Valve Replacement Procedures Follow-up and End Points

In the early-surgery group, surgical aortic-valve Data collection ended in April 2019, when the last
replacement was successfully performed in all enrolled patient had completed 4 years of follow-
72 patients in whom the procedure was attempt- up. The median follow-up was 6.2 years (interquar-
ed; 36 patients (50%) received a mechanical valve tile range, 5.0 to 7.4) in the early-surgery group
and 36 (50%) received a biologic prosthesis. All and 6.1 years (interquartile range, 4.5 to 7.3) in the
the patients except those who crossed over un- conservative-care group. No patients were lost to
derwent surgery within 2 months after random- follow-up.
ization; the median time between randomization In an intention-to-treat analysis including all
and surgery was 23 days (interquartile range, 10 to the trial patients, 1 of 73 patients assigned to early
36). There was no operative mortality in the early- surgery (1%) and 11 of 72 patients assigned to
surgery group. conservative care (15%) died from cardiovascu-
Of the 72 patients assigned to conservative care, lar causes (hazard ratio, 0.09; 95% confidence
53 patients (74%) underwent surgical aortic-valve interval [CI], 0.01 to 0.67) (Table 2). The number
replacement (52 patients) or transcatheter aortic- needed to treat to prevent one death from cardio-
valve replacement (TAVR) (1 patient) during fol- vascular causes within 4 years was 20 patients. The
low-up, mainly because of the development of cumulative incidence of the primary end point
symptoms (in 43 patients). Indications for aortic- (operative mortality or death from cardiovascu-
valve replacement are listed in Table S2. Among lar causes during the follow-up period), as cal-
these 53 patients, urgent surgery was performed culated with the use of a Kaplan–Meier analysis,
in 9 patients (17%) who were admitted from the was 1% at both 4 and 8 years in the early-surgery
emergency department. There was no operative group, as compared with 6% at 4 years and 26%
mortality among the patients who later underwent at 8 years in the conservative-care group (P = 0.003)
aortic-valve replacement; the median time from (Fig. 2A).
randomization to aortic-valve replacement was A total of 5 deaths from any cause (7% of the
700 days (interquartile range, 277 to 1469). Ad- patients) occurred in the early-surgery group
ditional information on surgical procedures and and 15 deaths from any cause (21%) occurred in
results are provided in Table S3 and the Supple- the conservative-care group (hazard ratio, 0.33;
mental Surgical Results section in the Supplemen- 95% CI, 0.12 to 0.90). The cumulative incidence
tary Appendix. of death from any cause was lower in the early-

116 n engl j med 382;2 nejm.org  January 9, 2020

The New England Journal of Medicine

Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 Early Surgery for Asymptomatic Aortic Stenosis

surgery group than in the conservative-care group

A Operative Mortality or Death from Cardiovascular Causes
(4% vs. 10% at 4 years and 10% vs. 32% at 8 years) 100 40
(Fig. 2B). Details regarding patients who died are P=0.003 by log-rank test
90 P=0.003 by Gray’s test
summarized in Table S4. In the conservative-care 30
80 26

Cumulative Incidence (%)

group, the cumulative incidence of sudden death
70 Conservative
was 4% at 4 years and 14% at 8 years. 20 care
60
The incidences of other prespecified second-
50 10
ary end points are listed in Table 2. The incidence 6
40
of hospitalization for heart failure was lower in 1 Early surgery 1
30 0
the early-surgery group than in the conservative- 0 2 4 6 8
20
care group (no cases vs. 11%). The cumulative in-
10
cidence of the composite of any secondary end
0
point or aortic-valve replacement in the conser- 0 2 4 6 8
vative-care group was 62% at 4 years and 92% at Years since Randomization
8 years (Fig. S1).
No. at Risk
The results of the analyses involving the per- Conservative care 72 68 65 36 12
protocol population were consistent with those Early surgery 73 73 70 38 13
involving the intention-to-treat population (Ta-
B Death from Any Cause
ble S5 and Fig. S2). Subgroup analyses were per-
100 40
formed to assess the consistency of the results 32
90
with regard to the primary end point and death 30 Conservative
80
from any cause in two prespecified subgroups Cumulative Incidence (%) care
70
(Table S6). 20
60
10 10
50 10
Discussion 40 4
Early surgery
The RECOVERY trial compared early surgery with 30 0
0 2 4 6 8
conservative care in asymptomatic patients with 20
very severe aortic stenosis. The trial showed a 10
lower incidence of the primary end point of opera- 0
0 2 4 6 8
tive mortality or death from cardiovascular causes
Years since Randomization
during the follow-up period among patients who
No. at Risk
underwent early surgery than among those who Conservative care 72 68 65 36 12
received conservative care. Moreover, early surgery Early surgery 73 73 70 38 13
was associated with a lower incidence of death
from any cause among such patients. Figure 2. Time-to-Event Curves for the Primary Composite End Point
The decision to perform surgery in an asymp- and Death from Any Cause.
tomatic patient requires careful weighing of the Shown are Kaplan–Meier estimates of the cumulative incidence of the pri-
mary end point of death during or within 30 days after surgery (often called
risks of early aortic-valve replacement against those operative mortality) or death from cardiovascular causes during the follow-
of observation. In patients with asymptomatic se- up period (Panel A) and of the major secondary end point of death from
vere aortic stenosis, it has generally appeared to any cause during follow-up (Panel B) among patients who were randomly
be relatively safe to follow a watchful waiting assigned to undergo early surgery or to receive conservative care. The in-
strategy and delay surgery until symptoms de- sets show the same data on an enlarged y axis.
velop.14 However, this conservative care strategy
is also associated with a risk of sudden death,
denial or late reporting of symptoms by patients, By reducing these limitations inherent to observa-
irreversible myocardial damage, and an increase tional studies,4,16 this randomized trial provides
in surgical risk while waiting for symptoms to evidence to support early aortic-valve replacement
develop.4-6 In previous observational studies, base- in asymptomatic patients with very severe aortic
line differences between the treatment groups, stenosis.
treatment-selection bias, and unmeasured con- There may be several explanations for the sig-
founders might have influenced the results.13,15 nificant difference in long-term survival between

n engl j med 382;2 nejm.org  January 9, 2020 117

The New England Journal of Medicine
Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

the two groups. First, surgical risk was substan- could have been influenced by the clinician’s
tially lower in this trial and in recent trials com- knowledge of the treatment the patient received.
paring surgical aortic-valve replacement with TAVR Fourth, exercise testing is reasonable to confirm
in low-risk patients8,9 than in previous studies; in the absence of symptoms in asymptomatic patients
our trial, the incidence of operative mortality was with severe aortic stenosis,2 but it was performed
less than 1%. Close monitoring was performed only selectively in this trial. Fifth, the small num-
after surgery, and improvements in postoperative bers of trial patients and primary end-point events
care contributed to a substantial decrease in the constitute an important limitation of this trial.
long-term risk associated with early aortic-valve However, it was the judgment of the investigators
replacement. Second, early surgery appeared to that a larger number of patients than the calcu-
prevent sudden death, because no cases of sudden lated sample size could not be enrolled for ethical
death occurred in the early-surgery group. In con- and logistic reasons. Finally, this trial included
trast, in the conservative-care group, the annual relatively young patients (as compared with pa-
risk of sudden death tended to increase during the tients enrolled in recent TAVR trials involving low-
progression of aortic stenosis before the develop- risk patients8,9), among whom the incidence of
ment of symptoms. Third, eventual aortic-valve bicuspid aortic-valve disease was high and who
replacement was almost unavoidable in the con- had normal left ventricular systolic function, few
servative-care group, and the overall risk of aortic- coexisting conditions, and low operative risk.
valve replacement appeared to increase while Thus, our trial population is quite different from
surgery was deferred until symptoms developed. the populations enrolled in TAVR trials,8,9,17-19 and
Cardiovascular events that occurred after surgery the results of our trial cannot be directly applied
were more frequently observed in the conserva- to early TAVR for asymptomatic severe aortic ste-
tive-care group, suggesting a higher long-term nosis. Because the incidence of operative mortal-
risk associated with later aortic-valve replacement. ity was very low in our trial, our results may not
Our trial has several limitations. First, the be applicable to low-volume medical centers or to
risk–benefit ratio may be shifted toward a benefit patients at high operative risk.
of early surgery in this trial involving patients In conclusion, in this randomized trial, early
with very severe aortic stenosis because the risk surgical aortic-valve replacement resulted in a sig-
of waiting increases according to the severity of nificantly lower risk of operative mortality or death
aortic stenosis.5 The benefit of early surgery may from cardiovascular causes during the follow-up
be relatively smaller in asymptomatic patients period than conservative care among asymptom-
with less severe aortic stenosis. Second, crossover atic patients with very severe aortic stenosis.
occurred in 5% of the patients in the early-surgery
A data sharing statement provided by the authors is available
group and in 3% of the patients in the conserva- with the full text of this article at NEJM.org.
tive-care group. Nevertheless, the results of the Supported by the Korean Institute of Medicine.
per-protocol analysis were similar to those of the No potential conflict of interest relevant to this article was
reported.
primary intention-to-treat analysis. Third, since Disclosure forms provided by the authors are available with
this trial was not blinded, the nonfatal outcomes the full text of this article at NEJM.org.

References
1. Baumgartner H, Falk V, Bax JJ, et al. tic stenosis during prolonged follow-up. ated on before symptom onset. Circula-
2017 ESC/EACTS guidelines for the man- Circulation 2005;​111:​3290-5. tion 2012;​126:​118-25.
agement of valvular heart disease. Eur 4. Généreux P, Stone GW, O’Gara PT, et al. 7. Brennan JM, Edwards FH, Zhao Y,
Heart J 2017;​38:​2739-91. Natural history, diagnostic approaches, O’Brien SM, Douglas PS, Peterson ED.
2. Nishimura RA, Otto CM, Bonow RO, and therapeutic strategies for patients Long-term survival after aortic valve re-
et al. 2014 AHA/ACC guideline for the with asymptomatic severe aortic stenosis. placement among high-risk elderly pa-
management of patients with valvular J Am Coll Cardiol 2016;​67:​2263-88. tients in the United States: insights from
heart disease: a report of the American 5. Carabello BA. Should severe aortic the Society of Thoracic Surgeons Adult
College of Cardiology/American Heart stenosis be operated on before symptom Cardiac Surgery Database, 1991 to 2007.
Association Task Force on Practice Guide- onset? Aortic valve replacement should be Circulation 2012;​126:​1621-9.
lines. J Am Coll Cardiol 2014;​63(22):​e57- operated on before symptom onset. Cir- 8. Mack MJ, Leon MB, Thourani VH, et al.
e185. culation 2012;​126:​112-7. Transcatheter aortic-valve replacement with
3. Pellikka PA, Sarano ME, Nishimura RA, 6. Shah PK. Should severe aortic steno- a balloon-expandable valve in low-risk
et al. Outcome of 622 adults with asymp- sis be operated on before symptom onset? patients. N Engl J Med 2019;​380:​1695-
tomatic, hemodynamically significant aor- Severe aortic stenosis should not be oper- 705.

118 n engl j med 382;2 nejm.org  January 9, 2020

The New England Journal of Medicine

Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
 Early Surgery for Asymptomatic Aortic Stenosis

9. Popma JJ, Deeb GM, Yakubov SJ, et al. tice Guidelines, Society of Cardiovascular severe aortic stenosis. J Am Coll Cardiol
Transcatheter aortic-valve replacement with Anesthesiologists, Society for Cardiovas- 2015;​66:​2827-38.
a self-expanding valve in low-risk patients. cular Angiography and Interventions, Soci- 16. Bhattacharyya S, Hayward C, Pepper J,
N Engl J Med 2019;​380:​1706-15. ety of Thoracic Surgeons. ACC/AHA 2006 Senior R. Risk stratification in asymp-
10. Oh JK, Taliercio CP, Holmes DR Jr, guidelines for the management of pa- tomatic severe aortic stenosis: a criti-
et al. Prediction of the severity of aortic tients with valvular heart disease: a report cal appraisal. Eur Heart J 2012;​33:​2377-
stenosis by Doppler aortic valve area de- of the American College of Cardiology/ 87.
termination: prospective Doppler-catheter- American Heart Association Task Force 17. Leon MB, Smith CR, Mack M, et al.
ization correlation in 100 patients. J Am on Practice Guidelines. Circulation 2006;​ Transcatheter aortic-valve implantation
Coll Cardiol 1988;​11:​1227-34. 114(5):​e84-e231. for aortic stenosis in patients who cannot
11. Bonow RO, Carabello B, de Leon AC 13. Kang DH, Park SJ, Rim JH, et al. Early undergo surgery. N Engl J Med 2010;​363:​
Jr, et al. Guidelines for the management surgery versus conventional treatment in 1597-607.
of patients with valvular heart disease: asymptomatic very severe aortic stenosis. 18. Smith CR, Leon MB, Mack MJ, et al.
executive summary: a report of the Amer- Circulation 2010;​121:​1502-9. Transcatheter versus surgical aortic-valve
ican College of Cardiology/American 14. Rosenhek R, Binder T, Porenta G, et al. replacement in high-risk patients. N Engl
Heart Association Task Force on Practice Predictors of outcome in severe, asymp- J Med 2011;​364:​2187-98.
Guidelines (Committee on Management tomatic aortic stenosis. N Engl J Med 19. Adams DH, Popma JJ, Reardon MJ,
of Patients with Valvular Heart Disease). 2000;​343:​611-7. et al. Transcatheter aortic-valve replacement
Circulation 1998;​98:​1949-84. 15. Taniguchi T, Morimoto T, Shiomi H, with a self-expanding prosthesis. N Engl
12. American College of Cardiology/Amer- et al. Initial surgical versus conservative J Med 2014;​370:​1790-8.
ican Heart Association Task Force on Prac- strategies in patients with asymptomatic Copyright © 2019 Massachusetts Medical Society.

ARTICLE METRICS NOW AVAILABLE

Visit the article page at NEJM.org and click on Metrics to view comprehensive and
cumulative article metrics compiled from multiple sources, including Altmetrics.
NEJM.org/about-nejm/article-metrics.

n engl j med 382;2 nejm.org  January 9, 2020 119

The New England Journal of Medicine
Downloaded from nejm.org on July 14, 2020. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.