Early Surgery or Conservative Care for Asymptomatic Aortic Stenosis at 10 Years

Authors, Journal, Affiliations, Type, DOI

• Duk-Hyun Kang, Sung-Ji Park, Ga Yun Kim, Sahmin Lee, Byung Joo Sun, Joon Bum Kim, Sung-Ho Jung, Hyung-Kwan Kim, Sung-Cheol Yun, Geu-Ru Hong, Jong-Min Song, Cheol-Hyun Chung
• New England Journal of Medicine, 2026; 394:1167–74
• Asan Medical Center / Samsung Medical Center / Seoul National University Hospital / Severance Hospital — University of Ulsan College of Medicine & Sungkyunkwan University School of Medicine, South Korea
• Multicentre, randomised, parallel-group, open-label RCT (extended follow-up of the RECOVERY trial); NCT01161732
• DOI: https://doi.org/10.1056/NEJMoa2511920

Overview

The RECOVERY trial enrolled 145 asymptomatic patients with very severe aortic stenosis (AVA ≤0.75 cm² + Vmax ≥4.5 m/s or mean gradient ≥50 mmHg) between July 2010 and April 2015, randomising them 1:1 to early surgical AVR vs conservative care. This paper reports the 10-year extended follow-up (data collection closed May 2025; median follow-up 144 months). Early surgery reduced the primary composite (operative mortality + CV death) from 24% → 3% (HR 0.10, p=0.002) and all-cause mortality from 32% → 15% (HR 0.42). The Kaplan-Meier curves continued to diverge throughout the 10-year period with no sign of convergence, establishing the durability of the survival benefit.

Keywords

Aortic stenosis, asymptomatic, aortic valve replacement, early surgery, conservative care, survival, cardiovascular mortality, RECOVERY trial

Key Takeaways

Trial Design

• Multicentre randomised open-label trial; 145 asymptomatic patients with very severe AS (AVA ≤0.75 cm², Vmax ≥4.5 m/s or mean gradient ≥50 mmHg); enrolled July 2010 – April 2015 at Korean centres of surgical expertise
• Exclusions: exertional dyspnea/syncope/angina, LVEF <50%, significant AR or MV disease, prior cardiac surgery, age >80 years, cancer or other condition precluding early surgery
• Early surgery arm: SAVR within 2 months of randomisation (median 23 days; 49% mechanical, 51% bioprosthesis; 0% operative mortality)
• Conservative care arm: referred for AVR if symptoms developed, LVEF fell <50%, or Vmax increased >0.5 m/s/year on serial echo
• Primary endpoint: composite of operative mortality (death during surgery or ≤30 days post-op) OR death from cardiovascular causes during entire follow-up

Primary Endpoint — CV Death or Operative Mortality

• Early surgery: 2/73 (3%); conservative care: 17/72 (24%); HR 0.10 (95% CI 0.02–0.43; P=0.002)
• 10-year cumulative incidence: 1% vs 19% (Kaplan-Meier); 5-year: 1% vs 7%
• Kaplan-Meier curves diverged continuously — no convergence throughout the 10-year follow-up
• NNT to prevent 1 CV death within 10 years = 6 patients

Secondary Endpoints

• All-cause mortality: 11/73 (15%) vs 23/72 (32%); HR 0.42 (95% CI 0.21–0.86); 10-year cumulative: 11% vs 25%; NNT = 7
• Heart failure hospitalisation: 0% (early surgery) vs 19% (conservative care) — no HF hospitalisations in early surgery arm
• Clinical thromboembolic events, repeat AVR surgery: reported (Table 2, full text)
• In conservative care group: 10 deaths (all CV) occurred before AVR; 13 deaths after AVR (7 CV); 85% eventually underwent AVR (59 SAVR + 2 TAVR); median time from randomisation to AVR = 1048 days
• Urgent surgery in 11/59 patients (19%) in conservative care who required emergency AVR

Aortic Valve Replacement in Conservative Care Group

• 85% of conservative care patients eventually underwent AVR during follow-up (median 1048 days from randomisation)
• Indications for delayed AVR: symptom development (49 patients), LVEF <50%, rapid progression (Vmax increase >0.5 m/s/year)
• 97% cumulative incidence of death or AVR at 10 years in conservative care group
• Not undergoing AVR appeared to increase CV mortality risk; early AVR (≤2 months) associated with lower CV mortality than later AVR

Sensitivity and Subgroup Analyses

• Per-protocol and as-treated analyses consistent with ITT results
• Treatment effects consistent across subgroups defined by peak aortic velocity and aetiology of AS (see Table S7)
• Competing-risk analysis (Fine and Gray) consistent with Cox primary analysis

Context Among Other Trials

• EARLY TAVR (NEJM 2025): median 46 months; no significant difference in all-cause mortality; 87% of surveillance arm underwent TAVR (median 11.1 months) — high crossover may explain similar survival
• AVATAR trial (Circ 2022 / Eur Heart J 2024 extended): 32-month primary; 63-month extended: all-cause mortality lower in early AVR group; only 44% of conservative-care arm underwent AVR — lower crossover may explain survival separation similar to RECOVERY
• RECOVERY vs AVATAR/EARLY TAVR: Long follow-up + incomplete AVR uptake in conservative care group → larger and sustained mortality separation

Limitations of the Document

• Small sample size (N=145); small number of primary endpoint events limits precision
• Open-label design: nonfatal endpoints susceptible to ascertainment bias
• Very severe AS definition (AVA ≤0.75 cm², Vmax ≥4.5 m/s) — results may not apply to less severe asymptomatic AS (Vmax <5 m/s)
• Results may not be generalisable to older, frailer patients or centres without high surgical expertise (selected Korean high-volume centres)
• No routine exercise stress testing to confirm symptom absence
• Whether TAVR provides equivalent long-term benefit as SAVR in this population remains unknown (only 2 TAVR in conservative care group; RECOVERY used SAVR exclusively)
• Part of extended follow-up conducted during COVID-19 pandemic (potential impact on outcomes)
• Trial did not include patients >80 years or those with major comorbidities — typical real-world AS population is older and frailer

Key Concepts Mentioned

• concepts/Aortic-Stenosis — primary disease under investigation; very severe asymptomatic AS management
• concepts/TAVI — mentioned as comparator context (EARLY TAVR); not used in primary RECOVERY interventions
• concepts/Structural-Valve-Deterioration — long-term bioprosthetic valve durability concern raised in discussion

Key Entities Mentioned

• entities/Bicuspid-Aortic-Valve — higher proportion in early surgery arm (imbalance noted at baseline)

Wiki Pages Updated

• wiki/sources/as-recovery-nejm-2026.md — created (this file)
• wiki/concepts/Aortic-Stenosis.md — added RECOVERY 10-year data; updated source_count 6→7; contradictions updated
• wiki/sourceindex.md — added RECOVERY entry
• wiki/wikiindex.md — updated Aortic-Stenosis entry with RECOVERY 10-year data; date 2026-05-18