Beta-Blocker Therapy After Myocardial Infarction
Definition
The question of whether to initiate, maintain, or discontinue beta-blocker therapy after MI is stratified by left ventricular function. In patients with reduced LVEF (<40%) or overt heart failure, beta-blockers remain a Class I cornerstone. In patients with preserved or mildly reduced EF (≥40%) who are stable without ongoing HF, recent RCT evidence challenges the necessity of indefinite continuation.
Key Concepts
LVEF-Stratified Framework
- LVEF <40% or HF (any EF): Beta-blockers remain Class I, Level A — metoprolol succinate, carvedilol, bisoprolol reduce all-cause mortality and sudden cardiac death (CAPRICORN, CIBIS-II, MERIT-HF, SENIORS). (sources/CCS-AHA-2023, rating: very high)
- LVEF ≥50%, no HF: Individual-patient-level meta-analysis (Beta-Blocker Trialists' Collaboration; NEJM 2026; n=17,801; 5 RCTs: REBOOT, REDUCE-AMI, BETAMI, DANBLOCK, CAPITAL-RCT; median follow-up 3.6 years) provides definitive evidence: no benefit of beta-blockers on composite death/MI/HF (HR 0.97; 95% CI 0.87–1.07; P=0.54). No benefit for any individual component (death HR 1.04, MI HR 0.89, HF HR 0.87), consistent across all subgroups including sex, age, MI type, and LVEF; I² = 20%. (sources/bb-mi-nejm-2026, rating: very high)
- Individual trial results (pre-IPD-MA): REDUCE-AMI (NEJM 2024, n=5,020, median 3.5yr): HR 0.96 (95% CI 0.79–1.16; P=0.64) — no benefit on death/MI; all secondary endpoints neutral; safety profiles similar (sources/bb-mi-reduceami-nejm-2024, rating: very high); REBOOT (NEJM 2025, n=7,459): HR 1.04; BETAMI+DANBLOCK (NEJM 2025, n=4,718): HR 0.85 (P=0.03); CAPITAL-RCT: HR 0.75 (P=0.20). Conflicting individual results now resolved by the IPD-MA. (sources/bb-mi-nejm-2026, rating: very high)
- LVEF 40–49% (mildly reduced): Individual patient data meta-analysis (Rossello 2025, Lancet, n=~7,000) suggests potential benefit — HR ~0.82 for MACE; however, estimates lack precision and must be considered exploratory. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
Discontinuation After ≥1 Year — SMART-DECISION Trial (2026)
- Open-label RCT; 25 South Korean centres; n=2,540; LVEF ≥40%; no HF; ≥1 year BB therapy; 1:1 discontinuation vs continuation. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Patient profile: mean age 63.2 years; 12.8% women; 56.8% STEMI; 98% received revascularisation at index MI; median 4.7 years post-MI at randomisation — a highly stabilised population.
- Most common BB: carvedilol 47.6%, bisoprolol 32.3%, nebivolol 19.9%.
- Primary endpoint (composite of all-cause death, recurrent MI, HF hospitalisation) at median 3.1 years:
- Discontinuation: 7.2% (4-year KM estimate)
- Continuation: 9.0% (4-year KM estimate)
- HR 0.80 (95% CI 0.57–1.13); upper CI 1.13 < prespecified NI margin 1.4; P=0.001 for noninferiority
- Individual components: all-cause death HR 0.71; recurrent MI HR 1.11; HF hospitalisation HR 0.82 — all non-significant; per-protocol HR 0.79. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- No significant differences in LVEF change, NT-proBNP, or quality of life (PROMIS-29). Blood pressure and heart rate increased slightly after discontinuation without clinical sequelae. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Background secondary prevention was intensive: more ezetimibe use, P2Y12 monotherapy (clopidogrel over aspirin), lower LDL achieved — likely contributing to low overall event rate.
ABYSS Trial (2024) — Noninferiority Not Met
- ABYSS (Silvain, Montalescot et al., NEJM 2024; n=3,698; LVEF ≥40%; median 2.9 years post-MI; 49 French sites; PROBE design; median follow-up 3.0 years): tested whether BB interruption was noninferior to continuation on a composite of death/MI/stroke/CV hospitalisation. (sources/bb-mi-abyss-nejm-2024, rating: high)
- Primary endpoint — noninferiority NOT met: 23.8% interruption vs 21.1% continuation; HR 1.16 (95% CI 1.01–1.33); risk difference 2.8pp (95% CI <0.1 to 5.5); upper boundary 5.5pp exceeds the 3pp noninferiority margin (P=0.44 for NI). (sources/bb-mi-abyss-nejm-2024, rating: high)
- Hard outcomes virtually identical: Death 4.1% vs 4.0%; MI 2.5% vs 2.4%; stroke 1.0% vs 1.0% — none meaningfully different. The composite failure was driven entirely by CV hospitalisation: 18.9% vs 16.6%, primarily coronary-related admissions (recurrent angina, diagnostic angiography, coronary procedures). (sources/bb-mi-abyss-nejm-2024, rating: high)
- Quality of life: no improvement — EQ-5D between-group difference 0.002 (95% CI −0.008 to 0.012); far below the 0.05 MCID. The premise that patients would feel better off beta-blockers was not supported. (sources/bb-mi-abyss-nejm-2024, rating: high)
- Interpretation caveat: CV hospitalisation in an open-label trial is susceptible to detection bias — clinicians and patients aware of drug allocation may have lower thresholds to investigate or admit. The anti-anginal effect of beta-blockers may also explain the coronary-related hospitalisation excess; this is a pharmacological withdrawal signal, not necessarily a safety signal for hard outcomes.
- Key differences from SMART-DECISION: ABYSS used a composite including soft CV hospitalisation vs hard-only endpoints in SMART-DECISION; enrolled at median 2.9 vs 4.7 years post-MI; bisoprolol-dominant (71.5%) vs carvedilol-dominant (47.6%); single country vs Korean cohort; background secondary prevention less intensive.
Practical Clinical Framework (Post-SMART-DECISION)
| Patient Profile | Beta-Blocker Strategy |
|---|---|
| LVEF <40% or HF | Continue indefinitely (Class I) |
| LVEF 40–49%, stable, no HF | Exploratory — discontinuation may be considered; insufficient data for broad recommendation |
| LVEF ≥50%, stable ≥1 year, no angina/arrhythmia/HTN indication | Initiation not supported (IPD-MA HR 0.97, P=0.54); discontinuation also reasonable — SMART-DECISION NI |
| LVEF ≥50%, atrial fibrillation | Beta-blocker often retained for rate control — excluded from SMART-DECISION |
| SCAD | COR 2b/C-LD observational benefit for recurrent SCAD prevention — retain |
Contradictions / Open Questions
- ACC/AHA Class I vs IPD-MA evidence — direct contradiction: The 2025 ACC/AHA ACS guideline maintains a Class I recommendation for beta-blockers after MI regardless of LVEF. The Beta-Blocker Trialists' Collaboration IPD-MA (NEJM 2026, n=17,801) now provides the highest available level of evidence showing no benefit in patients with LVEF ≥50% without other indications — HR 0.97 (95% CI 0.87–1.07; P=0.54) with I²=20%. The ESC (Class IIA) is the more evidence-congruent position. Guideline update is expected; ACC/AHA Class I recommendation is likely to be downgraded. (sources/bb-mi-nejm-2026, rating: very high)
- ABYSS vs SMART-DECISION discrepancy: ABYSS failed noninferiority (HR 1.16; upper CI 5.5pp); SMART-DECISION met noninferiority (HR 0.80; upper CI 1.13 < NI margin 1.4). The divergence is attributable to: (1) endpoint design — ABYSS composite included broad CV hospitalisation (soft, detection-bias-prone in open-label) vs SMART-DECISION's hard-only composite; (2) timing — 2.9 vs 4.7 years post-MI at randomisation; (3) background therapy intensity; (4) beta-blocker agent (bisoprolol dominant vs carvedilol dominant). Hard outcome data from both trials are concordant — interruption does not increase death, MI, or HF events. (sources/bb-mi-abyss-nejm-2024, rating: high; sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Mildly reduced EF (40–49%): SMART-DECISION excluded LVEF <40% but included 40–49%; subgroup is small and underpowered; Rossello 2025 Lancet IPD meta-analysis suggests possible benefit in this range. Firm conclusions on discontinuation in LVEF 40–49% should be avoided pending dedicated trials. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Early post-MI discontinuation: SMART-DECISION enrolled at median 4.7 years post-MI. Trial does not define a safe time point for discontinuation earlier after MI. Findings should not be extrapolated to the first year. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Women underrepresented: Only 12.8% women in SMART-DECISION; sex-specific treatment effects cannot be reliably estimated. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Noninferiority margin: HR 1.4 is relatively wide by contemporary standards; event rate was lower than expected, reducing precision. The trial cannot exclude a small but clinically meaningful increase in risk from discontinuation. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Korean-only population: May not generalise to populations with different cardiovascular risk profiles, revascularisation practices, or background therapy. (sources/BB-SMARTDECISION-NEJM-2026, rating: very high)
- Pharmacogenomics: GRK5 Leu41 variant (10× more prevalent in Black patients) provides intrinsic beta-1 blockade — pharmacogenomic profiling may eventually guide whether BB discontinuation is safe in specific individuals. See concepts/Pharmacogenomics-in-HF.
Connections
- Related to entities/Acute-Coronary-Syndrome — index events; short-term beta-blocker use post-ACS
- Related to entities/Chronic-Coronary-Disease — long-term BB de-escalation; COR 2b/B-NR guidance
- Related to entities/Heart-Failure — LVEF <40% subset; Class I BB use unchanged
- Related to concepts/Pharmacogenomics-in-HF — GRK5/ADRB1 polymorphisms affecting BB response