Beta-Blocker Interruption or Continuation after Myocardial Infarction — ABYSS Trial

Authors, Journal, Affiliations, Type, DOI

• Johanne Silvain, Guillaume Cayla, Emile Ferrari, et al.; for the ABYSS Investigators of the ACTION Study Group
• N Engl J Med 2024;391:1277-1286
• Sorbonne Université / Pitié–Salpêtrière Hospital, Paris; ACTION Study Group; 49 sites across France
• Multicenter, open-label, randomised noninferiority trial (PROBE design)
• DOI: https://doi.org/10.1056/NEJMoa2404204
• Funded by French Ministry of Health (PHRC-2015 grant) and ACTION Study Group

Overview

The ABYSS trial tested whether interruption of long-term beta-blocker therapy was noninferior to continuation in 3,698 patients with a history of MI and LVEF ≥40%, randomised at a median of 2.9 years after the index event. Noninferiority was not demonstrated: the composite primary endpoint (death/MI/stroke/CV hospitalisation) occurred in 23.8% vs 21.1% (HR 1.16; 95% CI 1.01–1.33; P=0.44 for noninferiority; upper CI 5.5pp exceeds the 3pp margin). However, the difference was driven entirely by CV hospitalisation (18.9% vs 16.6%); hard outcomes (death, MI, stroke) were virtually identical between groups. Beta-blocker interruption did not improve quality of life. This trial is concordant with REDUCE-AMI on hard endpoints but diverges on the softer composite owing to more coronary-related hospitalisations in the interruption arm, plausibly reflecting anti-anginal drug withdrawal effects or open-label detection bias.

Keywords

Beta-blocker, myocardial infarction, secondary prevention, interruption, continuation, noninferiority, LVEF preserved, quality of life, cardiovascular hospitalization, ABYSS

Key Takeaways

Background and Rationale

• Beta-blocker benefit in post-MI patients was established in pre-reperfusion era trials (pre-1990s); modern reperfusion with primary PCI has markedly reduced post-MI LVEF impairment and HF risk, calling into question the necessity of indefinite therapy in those with preserved EF
• No prior contemporary large RCT had specifically evaluated late discontinuation (>1 year post-MI) of beta-blockers in patients without HF or LV dysfunction
• Large registry data suggested no long-term mortality benefit in preserved-EF post-MI patients, but data were inconsistent
• Beta-blockers prescribed lifelong to >90% of post-MI patients in Western registries; drug side effects (fatigue, sexual dysfunction, exercise intolerance) are commonly cited as reasons for poor adherence and physician inclination to discontinue

Study Design and Patients

• Multicenter open-label RCT (PROBE: prospective, randomised, open-label, blinded endpoint); 49 French sites; August 2018–September 2022; includes COVID-19 pandemic period
• Eligibility: MI history ≥6 months before randomisation; current beta-blocker therapy (any agent/dose); LVEF ≥40%; no cardiovascular event in prior 6 months; no other primary indication for beta-blocker (no arrhythmia, migraine, uncontrolled hypertension)
• N=3,698: 1,846 interruption / 1,852 continuation
• Mean age 63.5±11 years; 17.2% women; 63% had prior STEMI; 7.8% had >1 prior MI; 95% revascularised (96.9% by PCI)
• Median time from index MI to randomisation: 2.9 years (IQR 1.2–6.4)
• Median resting HR at randomisation: 63 bpm (IQR 57–71) — indicating appropriate beta-blocker use at baseline
• Most common beta-blockers: bisoprolol 71.5%, acebutolol 10.8%, atenolol 8.7%, nebivolol 5.9%
• Background therapy: guideline-directed secondary prevention; 22.8% on ticagrelor or prasugrel at randomisation
• Crossover: 8.6% interruption → continuation; 2.8% continuation → interruption
• Median follow-up: 3.0 years (IQR 2.0–4.0)
• Noninferiority margin: upper boundary of 95% CI for risk difference <3 percentage points (corresponds to ~25% relative risk increase)

Primary Endpoint

• Composite of death, nonfatal MI, nonfatal stroke, or hospitalisation for any cardiovascular reason:
  • 432/1812 (23.8%) interruption vs 384/1821 (21.1%) continuation
  • Risk difference: 2.8pp (95% CI <0.1 to 5.5)
  • HR 1.16 (95% CI 1.01–1.33; P=0.44 for noninferiority)
  • Upper CI boundary 5.5pp exceeds the 3pp noninferiority margin — NONINFERIORITY NOT MET
  • Kaplan–Meier 4-year estimates: 21.8% interruption vs 19.3% continuation

Individual Components of the Primary Endpoint

• Hard outcomes (death/MI/stroke) were virtually identical; CV hospitalisation drove the composite difference — primarily coronary-related admissions (recurrent angina, diagnostic angiography, coronary procedures)

Secondary Endpoints

• Hard composite (death/MI/stroke): 7.2% interruption vs 6.8% continuation — NS
• Death/MI/stroke/HF hospitalisation: 8.4% vs 7.6% — NS
• Quality of life (EQ-5D): Mean score change 0.033 vs 0.032; between-group difference 0.002 (95% CI −0.008 to 0.012) — far below the 0.05 MCID; no QoL improvement from beta-blocker interruption
• Results consistent across prespecified subgroups

Interpretation and Context

• The signal driving noninferiority failure is CV hospitalisation, not hard events — and this endpoint is susceptible to open-label detection bias (clinicians more likely to admit or investigate patients who have stopped their beta-blocker)
• CV hospitalisation included admissions for recurrent angina, diagnostic angiography without revascularisation — not MI or death — consistent with removal of beta-blocker's anti-anginal effect rather than a true safety signal
• Authors acknowledge Sir James Black's original rationale for beta-blockers: anti-anginal properties; this is precisely what ABYSS may be capturing
• Concordant with REDUCE-AMI (Yndigegn/Lindahl, NEJM 2024; n=5,020; LVEF ≥50%): no benefit on death/MI (HR 0.96; NS) — REDUCE-AMI did not assess CV hospitalisation
• SMART-DECISION (NEJM 2026; Korean; n=2,540; median 4.7 years post-MI): BB discontinuation met noninferiority for hard composite (death/MI/HF hosp; HR 0.80; NI margin HR <1.4; P=0.001 NI) — key differences: exclusively hard endpoints; later-phase more-stabilised population; more intensive background therapy

Limitations

• Open-label design (PROBE): CV hospitalisation endpoint subject to detection and clinical decision-making bias — clinicians aware of drug allocation may be more likely to admit or investigate symptomatic patients who stopped their beta-blocker
• CV hospitalisation is a soft endpoint with heterogeneous components including diagnostic angiography without revascularisation and non-acute presentations
• Single-country trial (France): May not generalise to health systems with different prescribing practices, patient populations, or access to care
• Broad noninferiority margin (3pp, ~25% RR): chosen based on anticipated 12% event rate in continuation arm at 3 years; actual event rate was 21.1%, reflecting higher baseline risk; the original margin may underpower detection of a clinically meaningful difference on soft endpoints
• Patients younger and lower-risk than real-world post-MI populations: 17.2% women; relatively young (63.5 years); 95% revascularised — may not represent the typical long-term beta-blocker user
• No blinding: Patient-reported outcomes (QoL) and physician hospitalisation decisions both potentially influenced; however, all hard endpoints adjudicated by blinded committee

Key Concepts Mentioned

• concepts/Beta-Blocker-Post-MI — primary topic; ABYSS adds dedicated source evidence for the interruption question; contrasts with REDUCE-AMI and SMART-DECISION

Key Entities Mentioned

• entities/Acute-Coronary-Syndrome — index event population
• entities/Chronic-Coronary-Disease — long-term management context

Wiki Pages Updated

• wiki/sources/bb-mi-abyss-nejm-2024.md — created
• wiki/concepts/Beta-Blocker-Post-MI.md — ABYSS section expanded; source added; source_count updated
• wiki/sourceindex.md — entry added
• wiki/wikiindex.md — Beta-Blocker-Post-MI entry updated