Mineralocorticoid receptor antagonists for acute myocardial infarction: a systematic review and meta-analysis of randomised controlled trials Authors: Song Peng Ang, Jia Ee Chia, Bruno Bezerra Lima, Jose Iglesias, Eunseuk Lee, Chayakrit Krittanawong, Mahboob Alam, Debabrata Mukherjee Published: 2025 ABSTRACT AbstractBackgroundThe role of mineralocorticoid receptor antagonists (MRAs) in acute myocardial infarction (MI) remains controversial, with conflicting evidence from landmark trials. We aimed to assess the impact of MRAs on mortality and cardiovascular outcomes post-acute MI.MethodsWe systematically searched PubMed, Embase and Cochrane CENTRAL databases up to February 2025 for randomised controlled trials comparing MRAs with placebo or standard care in adults experiencing acute MI. Primary outcome was all-cause mortality; secondary outcomes included cardiovascular mortality, heart failure, recurrent MI and ventricular arrhythmia. Data were pooled using random-effects models, with heterogeneity explored via subgroup analyses and meta-regression.Results15 trials (n=18 471) were included. MRA therapy demonstrated non-significant reductions in all-cause mortality (OR 0.80, 95% CI 0.55 to 1.18), cardiovascular mortality (OR 0.84, 95% CI 0.59 to 1.18), heart failure (OR 0.76, 95% CI 0.52 to 1.12), recurrent MI (OR 0.92, 95% CI 0.67 to 1.27) and ventricular arrhythmia (OR 0.83, 95% CI 0.47 to 1.47). Subgroup analyses revealed that trials with >6 months follow-up demonstrated modest cardiovascular mortality reduction (OR 0.86, 95% CI 0.75 to 0.99). Effects were consistent across MRA types, left ventricular ejection fraction categories and initiation timing. Meta-regression showed no significant effect modifiers among baseline characteristics or concomitant therapies.ConclusionsIn MI populations, MRA therapy did not significantly improve mortality or cardiovascular outcomes in the short term. However, a significant reduction in cardiovascular mortality emerged after 6 months, alongside a non-significant trend towards less heart failure, indicating potential benefits for high-risk patients with longer-term treatment rather than routine use in all acute MI cases.Keywords: Coronary Artery Disease, Heart Failure, Pharmacology, Meta-Analysis WHAT IS ALREADY KNOWN ON THIS TOPIC Mineralocorticoid receptor antagonists (MRAs) improve outcomes in chronic HF and post-MI patients with reduced ejection fraction. Mineralocorticoid receptor antagonists (MRAs) improve outcomes in chronic HF and post-MI patients with reduced ejection fraction. WHAT THIS STUDY ADDS While MRAs showed non-significant trends towards benefit in mortality and heart failure outcomes, a modest cardiovascular mortality reduction was identified in trials with follow-up exceeding 6 months, possibly suggesting late benefits. The analysis revealed consistent effects across MRA types, left ventricular function categories and timing of therapy initiation. While MRAs showed non-significant trends towards benefit in mortality and heart failure outcomes, a modest cardiovascular mortality reduction was identified in trials with follow-up exceeding 6 months, possibly suggesting late benefits. The analysis revealed consistent effects across MRA types, left ventricular function categories and timing of therapy initiation. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY In the contemporary era of optimised MI care with improved baseline outcomes, routine MRA therapy does not provide significant early mortality or cardiovascular benefits in unselected acute MI patients. The observed heart failure reduction signal, though not statistically significant, and the late cardiovascular mortality benefit suggest potential value in targeted use for high-risk individuals. Future research should focus on biomarker-guided patient selection and timing optimisation rather than universal application in all post-MI patients. In the contemporary era of optimised MI care with improved baseline outcomes, routine MRA therapy does not provide significant early mortality or cardiovascular benefits in unselected acute MI patients. The observed heart failure reduction signal, though not statistically significant, and the late cardiovascular mortality benefit suggest potential value in targeted use for high-risk individuals. Future research should focus on biomarker-guided patient selection and timing optimisation rather than universal application in all post-MI patients. Introduction Cardiovascular diseases are the leading cause of mortality worldwide, accounting for 17.9 million deaths annually, representing 31.6% of all global fatalities.1 Within this spectrum, acute myocardial infarction (MI) remains the dominant cause of mortality despite significant advancements in reperfusion strategies and secondary prevention protocols.2,7 While contemporary strategies including early revascularisation, dual antiplatelet therapy and neurohormonal blockade have significantly improved survival, residual risks of heart failure (HF), recurrent ischaemic events and arrhythmic complications persist. Among therapeutic frontiers, mineralocorticoid receptor antagonists (MRAs) have emerged as a compelling target, given their pleiotropic effects on myocardial fibrosis, inflammation and electrical remodelling.8 9 The cardioprotective role of MRAs in chronic HF and post-MI populations with reduced ejection fraction is well-established, anchored by landmark trials such as EMPHASIS-HF and EPHESUS.10,12 However, their utility in broader MI cohorts without overt HF or systolic dysfunction remains contested. Early trials like ALBATROSS demonstrated neutral effects on composite endpoints in unselected MI patients, while the REMINDER trial suggested benefits in surrogate biomarkers among ST-segment elevation MI (STEMI) populations.13 14 This was further complicated by the recent CLEAR SYNERGY trial, which showed neutral mortality findings.15 In light of evolving reperfusion strategies and contemporary secondary prevention, isolated trial-level observations, including those from EPHESUS and CLEAR, may be insufficient to guide routine post-MI MRA use. Accordingly, we conducted a systematic review and meta-analysis of randomised controlled trials to evaluate the effects of MRAs on mortality and HF outcomes after acute MI across left ventricular ejection fraction (LVEF) categories, follow-up durations, MRA agents and timing of initiation, using a conservative and standardised analytical framework. Methods Study design and registration This systematic review and meta-analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. The study was prospectively registered on PROSPERO (CRD420251000638). Data sources and search strategy A comprehensive search of PubMed, Embase and Cochrane Central Register of Controlled Trials (CENTRAL) was conducted from inception to 20 February 2025. Key search terms included: “mineralocorticoid receptor antagonists,” “spironolactone,” “eplerenone” and “myocardial infarction”. Boolean operators and Medical Subject Headings (MeSH) were used to refine results. Reference lists of identified studies were further manually screened to identify potentially relevant studies. Details of the search strategy are available in online supplemental table S1. Eligibility criteria Inclusion criteria Trials were included if they met the following criteria: (1) randomised controlled trials (RCTs) comparing MRA therapy against placebo or standard care; (2) adults aged ≥18 years with acute MI, including STEMI or non-STEMI and (3) reported outcomes for all-cause mortality, cardiovascular mortality, recurrent MI, new or worsening HF. Exclusion criteria Studies were excluded if they (1) used non-randomised designs; (2) patients without acute MI or (3) age <18 years Data extraction, quality assessment and certainty of evidence Two independent reviewers (SPA and JEC) extracted data using a standardised template, including study characteristics, patient demographics, intervention protocols and outcomes. Discrepancies were resolved by a third reviewer (DM). Risk of bias was assessed using the Cochrane Collaboration’s RoB 2.0 tool, evaluating randomisation, allocation concealment, blinding, incomplete outcome data and selective reporting.16 Overall certainty of evidence was determined using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) criteria.17 Outcomes The primary outcome was all-cause mortality at the longest available follow-up. Secondary outcomes included cardiovascular mortality, recurrent MI and new or worsening HF, as defined by trial protocols. Statistical analysis Data were synthesised using the inverse-variance method with Hartung-Knapp-Sidik-Jonkman random-effects models to account for the anticipated between-study heterogeneity given the variations in study protocols, type of MRAs and timing of initiations.18 Effect measures for categorical outcomes were expressed as ORs with corresponding 95% CIs. Statistical heterogeneity was quantified and interpreted using Cochran Q statistic (p<0.10 indicating significance), I² values (≥50% denoting substantial heterogeneity) and τ² (between-study variance).19 20 Robustness of primary findings was evaluated through multiple sensitivity analyses, including: (1) leave-one-out sequential exclusion of each trial to assess individual study influence on pooled estimates; (2) restricted analysis of STEMI cohorts. Effect modification was explored through stratified analyses using the interaction test for heterogeneity between subgroups as well as meta-regression. Key stratifying moderators included: (1) type of MRA (spironolactone vs eplerenone vs canrenoate); (2) follow-up duration (≤6 months vs >6 months); (3) LVEF and (4) timing of MRA initiation. In this study, mean LVEF of ≥50% at baseline was regarded as preserved LVEF, whereas mean LVEF <50% at baseline was considered as reduced or mildly reduced LVEF. Stratification of timing of initiation was determined using a cut-off of 72 hours, with less than 72 hours considered early while ≥72 hours considered late. Between-subgroup differences were assessed using the Q-test of interaction. Univariate meta-regression was performed using trial-level covariates including publication year, age, male sex, hypertension, hyperlipidaemia, diabetes mellitus, prior MI, anterior MI, baseline LVEF and medications including aspirin, P2Y12 receptor antagonists, statins, beta-blockers and ACE inhibitor or angiotensin receptor blockers. Small-study effects and publication bias were evaluated with funnel plots, Egger’s regression asymmetry test and, when indicated, trim-and-fill sensitivity analysis. Statistical significance was set at two-sided p<0.05 except for heterogeneity testing (p<0.10). All analyses were conducted using STATA V.17.0 (StataCorp). Results Preliminary database search identified 1588 potentially relevant citations, with an additional 27 records identified through other sources including reference lists of key publications. After removal of 899 duplicates, 716 records underwent title and abstract screening, resulting in the exclusion of 629 records that did not meet eligibility criteria. The remaining 87 full-text articles were assessed for eligibility, with 72 excluded for the following reasons: different patient population or outcomes (n=37), non-randomised study design (n=23) and review articles or meta-analyses (n=12). Finally, 15 RCTs involving 18 471 patients with acute MI were included in the qualitative and quantitative synthesis.1113,15 21 The analysis included several landmark trials: CLEAR (n=7062), EPHESUS (n=6632), ALBATROSS (n=1603) and REMINDER (n=1012).1113,15 Study populations were predominantly male (range, 71%–85%) with mean ages spanning from 52 to 62 years. Median follow-up duration was 6 (5–12) months. Eight trials evaluated spironolactone, three investigated eplerenone and three examined potassium canrenoate. Geographic distribution was diverse, with multinational recruitment in three major trials (CLEAR, EPHESUS, REMINDER), while remaining studies originated from Europe (Italy (n=3), Turkey (n=2), UK (n=2), France (n=1), Greece (n=1)), Asia (China (n=1), Japan (n=1)) and South America (Chile (n=1)). Dosing protocols varied considerably across trials: spironolactone (25–50 mg daily), eplerenone (25–50 mg daily) and potassium canrenoate (200–1000 mg loading dose followed by 25–50 mg daily maintenance). Details of study and patients characteristics are available in table 1. Risk of bias assessment using the Cochrane risk of bias tool revealed moderate concerns across included trials, primarily driven by missing outcome data (15%–20%), though most demonstrated low risk in randomisation (online supplemental figure S2 and S3). The GRADE assessment demonstrated moderate certainty evidence for all-cause mortality, cardiovascular mortality, HF and recurrent MI outcomes (downgraded for inconsistency), and low certainty for ventricular arrhythmia (downgraded for inconsistency and imprecision) (online supplemental table S2). Table 1Baseline Characteristics of included studies and patientsStudyStudy designCountrySample sizeMRA typeDose (mg)TimingFollow-up (months)Mean ageMale, %HTN, %DM, %HLD, %Prior HF, %LVEF, %Revascularisation, %Aspirin, %P2Y12 inhibitors, %Beta-blocker, %ACEi/ARB, %CLEAR, 202515Multicentre, Double-blinded, Placebo-controlled, RCT with 2×2 factorial designMulticentre International7062Spironolactone2524–48 hours3660.9/60.478.5/80.845.2/46.317.8/19.1NA/NA0.7/1.0<45/<4510096.6/96.987.5/86.677.6/78.7NA/NAMINIMIZE STEMI, 201921Multicentre, double-blinded, placebo-controlled RCTMulticentre UK70Spironolactone25–50Within 2 hours362/60.1386.8/84.434.2/34.45.3/12.528.9/46.90/054/4910094.7/93.871.1/81.3NA/NANA/NAALBATROSS, 201613Multicentre, Nationwide, Open-labelled, blinded endpoint, RCTFrance1603Spironolactone25Within 72 hours658/5881.4/82.141.8/43.916.0/15.746.8/46.01.1/0.651.25/51.2593.499.4/99.499.4/99.493.5/92.987.5/89.6REMINDER, 201414Multicentre, randomised, double-blind, placebo-controlled, two-arm, parallel groupMulticentre International1012Eplerenone25–5016.4 hours10.558.5/57.883.0/79.647.6/51.412.8/15.4NA/NA0/00/086.198.2/98.098.4/98.487.9/88.382.6/83.2Wu et al., 201322Two-arm RCTChina528Spironolactone2017.1 hours1259.8/59.973.7/72.246.9/44.019.1/19.929.0/32.0NA/NA52.05/50.1263.498.5/98.196.2/94.094.3/92.993.9/94.0Kampourides et al., 2011*23Open-label, randomised, parallel-group, non-placebo-controlled, clinical studyGreece303Eplerenone25Day 72458.6782.0/NA86.6/NA29.9/NA68.7/NANA/NA57/086.1NA/NANA/NANA/NANA/NAKayrak et al., 201024Randomised, parallel-group, non-placebo-controlled, clinical studyTurkey110Spironolactone2518.2 hours5.555.3/57.218.2/25.529.1/25.518.2/20.0NA/NANA/NA50.5/49.510098.2/98.2100/10094.5/98.292.7/90.9Weir et al., 201025Randomised, double-blinded, placebo-controlled trialUK100EplerenoneNAAfter 24 hours661/56.874.0/80.044.0/26.0NA/NA32.0/18.0NA/NA51.7/46.27494.0/98.084.0/80.096.0/90.094.0/94.0Uzunhasan et al., 200926Randomised, double-blind, placebo-controlled studyTurkey82Spironolactone50NA752/5278.0/70.739.0/34.117.1/17.143.9/51.2NA/NA47/44100100/100NA/NA100/100100/100Di Pasquale et al., 200527Randomised, double-blind, placebo-controlled studyItaly687Canreonate254–6 hours562.6/62.871.3/70.535.8/35.038.7/40.846.3/46.50/044.5/44.736.9100/100100/10036.9/36.1100/100EPHESUS, 200311Multicentre, international, randomised, double-blind, placebo-controlled trialMulticentre International6632Eplerenone25–503–14 days1664/6471.7/70.44.4/4.523.0/22.5NA/NA1.0/1.133/336.66.4/6.6NA/NA5.5/5.56.3/6.4Hayashi et al., 200328Randomised, non-placebo-controlled, clinical studyJapan134Spironolactone25Day 1164.4/62.971.0/78.526.1/32.339.1/44.631.9/41.5NA/NA46/46.5NANA/NANA/NA27.5/33.8NA/NADi Pasquale et al., 200129Randomised, double-blind, placebo-controlled studyItaly55Canrenoate25Day 1561.11/62.1874.1/75.070.4/71.433.3/39.329.6/25.00/040.11/40.3251.9100/100NA/NA48.1/42.9100/100Modena et al., 200130Randomised, placebo-controlled studyItaly46Canrenoate50Postdischarge1259/6270.8/77.350.0/54.516.7/13.633.3/31.8NA/NA47/4684.8NA/NANA/NANA/NA100/100Rodríguez et al., 199731Randomised, single-blind, placebo-controlled studyChile47Spironolactone75NA658.8/58.678.3/91.7NA/NANA/NANA/NANA/NA34.5/32.6NANA/NANA/NANA/NANA/NAVariables are expressed as MRA/control.Categorical variables are expressed as percentages while continuous variables are expressed as mean.DM: diabetes mellitus; HLD: hyperlipidemia; HTN: hypertension*Baseline available for the entire cohort only.ACEi, ACE inhibitor; ARB, angiotensin receptor blocker; LVEF, left ventricular ejection fraction ; MRA, mineralocorticoid receptor antagonist; NA, not applicable. Baseline Characteristics of included studies and patients Variables are expressed as MRA/control. Categorical variables are expressed as percentages while continuous variables are expressed as mean. DM: diabetes mellitus; HLD: hyperlipidemia; HTN: hypertension Baseline available for the entire cohort only. ACEi, ACE inhibitor; ARB, angiotensin receptor blocker; LVEF, left ventricular ejection fraction ; MRA, mineralocorticoid receptor antagonist; NA, not applicable. Primary outcome: all-cause mortality MRA therapy was associated with a non-significant 20% reduction in all-cause mortality compared with control (OR 0.80, 95% CI 0.55 to 1.18, p=0.26) (figure 1A). The CLEAR trial, contributing 19.43% of the weight, showed neutral effects (OR 0.94, 95% CI 0.76 to 1.17).15 Heterogeneity was substantial (I²=69.90%, τ²=0.18), driven by variability in smaller trials like Weir et al (OR 7.44, 95% CI 0.37 to 147.92). Forest plots of outcomes including (A) all-cause mortality, (B) cardiovascular mortality, (C) heart failure. MRA, mineralocorticoid receptor antagonist. Secondary outcomes Cardiovascular mortality Pooled data from 12 983 patients across eight trials showed a non-significant 16% reduction in cardiovascular mortality with MRAs (OR 0.84, 95% CI 0.59 to 1.18, p=0.30)(figure 1B). Heterogeneity remained high (I²=74.19%, τ²=0.17). Heart failure Across the trials reporting HF outcomes, MRA therapy was associated with a numerically lower incidence of new or worsening HF events; however, this difference did not reach statistical significance (OR 0.76, 95% CI 0.52 to 1.12, p=0.17) (figure 1C). The CLEAR trial contributed a significant 32% risk reduction (OR 0.68, 95% CI 0.49 to 0.96), whereas EPHESUS showed borderline significance (OR 0.87, 95% CI 0.74 to 1.01).10 15 Heterogeneity persisted (I²=72.13%, τ²=0.15). Recurrent MI Analysis of 10 216 patients from seven trials revealed no significant difference in recurrent MI risk (OR 0.92, 95% CI 0.67 to 1.27, p=0.63) (figure 2A). While Kampourides et al suggested a 52% risk reduction (OR 0.48, 95% CI 0.19 to 1.26), larger trials like CLEAR (OR 0.99, 95% CI 0.75 to 1.30) and EPHESUS (OR 0.97, 95% CI 0.81 to 1.18) showed neutral effects.15 23 Heterogeneity was moderate (I²=46.54%, τ²=0.07). Forest plots of outcomes including (A) recurrent myocardial infarction, (B) ventricular arrhythmia. MRA, mineralocorticoid receptor antagonist. Ventricular arrhythmia MRA therapy was associated with a non-significant numerical reduction in ventricular arrhythmia (OR 0.83, 95% CI 0.47 to 1.47, p=0.52) (figure 2B). The CLEAR trial showed a non-significant 17% risk increase (OR 1.17, 95% CI 0.61 to 2.24), whereas the EPHESUS and ALBATROSS trials showed a neutral effect.10 13 15 Overall, the between-study heterogeneity persisted (I²=72.3%, τ²=0.26). Sensitivity analyses The robustness of the primary outcome was confirmed through leave-one-out sensitivity analysis, which demonstrated consistent effect estimates across iterations. Exclusion of the CLEAR trial yielded an OR of 0.78 (95% CI 0.50 to 1.20; p=0.25) for all-cause mortality, while omitting the EPHESUS trial resulted in an OR of 0.80 (95% CI 0.51 to 1.24; p=0.31), indicating minimal influence from individual large trials.10 15 For the secondary outcomes including cardiovascular mortality, recurrent MI, HF and ventricular arrhythmia, the results after excluding one study at a time remained consistent with primary analysis whereby there was no significant difference between use of MRA and placebo (online supplemental figure S4 to S8) In 11 219 STEMI patients, MRA therapy demonstrated a 26% non-significant all-cause mortality reduction (OR: 0.74, 95% CI 0.47 to 1.18; p=0.21), with comparable effects across secondary outcomes (online supplemental figure S9 to S13). Heterogeneity remained elevated (I²=79%). Subgroup analyses Across prespecified subgroups, findings were broadly consistent. No meaningful differences were observed between spironolactone, eplerenone and canrenoate, with all agents showing neutral or non-significant trends and no evidence of differential efficacy (all P-interaction >0.60; online supplemental figures S14 to S18). Trials with longer follow-up (>6 months) suggested a modest reduction in cardiovascular mortality compared with shorter-duration studies, but interaction testing did not support a statistically significant difference, and all-cause mortality estimates were similar irrespective of follow-up length (online supplemental figures S19 to S23). Stratification by LVEF showed comparable treatment effects in preserved vs reduced/mildly reduced EF, with numerically greater benefit in reduced EF but no significant interactions; these EF findings are exploratory and based on trial-level averages (online supplemental figures S24 to S28). Similarly, early versus late initiation of MRAs yielded broadly similar effects on mortality, HF, recurrent MI and ventricular arrhythmias, without significant interaction by timing of initiation (online supplemental figures S29 to S33). Meta-regression Univariable meta-regression using trial-level covariates did not identify any significant modifiers of MRA effects on all-cause mortality, cardiovascular mortality, recurrent MI or HF (all p>0.05; online supplemental table S3). Weak, non-significant signals for greater benefit with higher P2Y12 inhibitor use and lower baseline LVEF were observed but should be considered exploratory given wide standard errors and limited aggregate data. Overall, these analyses support broadly consistent treatment effects across studied trial characteristics. Publication bias Funnel plots for all-cause mortality, cardiovascular mortality, HF, MI and ventricular arrhythmia showed minimal or no asymmetry (Egger’s test, p>0.05 for all outcomes), suggesting no evidence of publication bias (online supplemental figures S34 to S38). Discussion In this systematic review and meta-analysis of 15 randomised controlled trials including 18 471 participants with acute MI, MRA therapy did not significantly reduce all-cause mortality in unselected post-MI populations when evaluated within a conservative random-effects framework. By integrating both legacy and contemporary trials, including CLEAR, and systematically examining outcomes across LVEF strata, follow-up durations, drug classes and timing of initiation, our findings refine prior trial-specific narratives and suggest that any incremental benefit of routine MRA therapy in the modern era is likely confined to carefully selected high-risk subgroups rather than broadly applicable to all patients after MI. The traditional paradigm supporting MRA use after MI was predominantly constructed on the foundation of the EPHESUS trial, which accounted for more than 80% of weight in 2 prior meta-analyses and demonstrated significant mortality benefits with eplerenone in patients with left ventricular dysfunction and HF following MI.10 Subsequent smaller trials including REMINDER and ALBATROSS produced inconclusive and seemingly contradictory results, creating uncertainty about whether MRA benefits extended beyond the specific EPHESUS population to broader MI cohorts without established HF.13 14 Our study represented a contemporary update to previous meta-analyses with the notable addition of the recently published CLEAR trial, which offered a new perspective into the use of MRA in post-MI patients.15 32 This 2 by two factorial RCT enrolled 7062 participants across 104 centres in 14 countries, representing one of the largest investigations of MRA therapy in acute MI. Unlike EPHESUS, which targeted patients with left ventricular dysfunction and HF, CLEAR enrolled a broadly representative population with 95% experiencing STEMI and undergoing percutaneous coronary intervention within a contemporary time frame.10 15 The investigators found that spironolactone failed to significantly reduce either of its coprimary endpoints: the composite of cardiovascular death or HF, and the composite of cardiovascular death, MI, stroke or new/worsening HF. These neutral findings directly challenge the prior assumption that MRA therapy provides substantial cardiovascular benefits when initiated early after MI in patients without established HF. While an on-treatment analysis suggested potential benefits, these findings were considered exploratory and hypothesis-generating. Perhaps most importantly, the CLEAR trial highlights a fundamental shift in the contemporary landscape of post-MI care.15 As noted by the trial’s investigators, the outcomes post-MI have improved remarkably over the last two decades and it was actually more challenging to show a difference. For instance, the contemporary HF event rates post-MI were 2% or 3% whereas they were 10% or 15% 20 years ago. The implications of this transformed risk profile profoundly impact our interpretation of trial results and the evolving place of MRAs in contemporary practice. When EPHESUS was conducted, background therapy and revascularisation strategies differed substantially from current standards.10 In CLEAR, more than 96% of patients received treatment with at least one drug-eluting stent, statin therapy and dual antiplatelet therapy at discharge.15 These advances in risk stratification, revascularisation techniques, stent technology, antithrombotic therapy and guideline-directed medical therapy have fundamentally altered the natural history of acute MI.333,36 The contemporary approach to MI management has created a challenging paradox for evaluating adjunctive therapies: as baseline event rates decrease, demonstrating incremental benefit becomes increasingly difficult, requiring either substantially larger sample sizes or more potent interventions. This ‘success penalty’ manifested clearly in CLEAR, where the trial was initially designed to detect a 25% relative risk reduction based on an estimated 15% event rate at 3 years, but required sample size adjustment after interim analysis revealed substantially lower event rates of approximately 3% per patient-year. Despite the neutral primary findings, the CLEAR results contain important nuances that warrant consideration. Although we observed numerically fewer HF events with MRA therapy in several trials, these differences were not statistically significant in pooled analyses. This observation suggests that MRAs may retain specific pathophysiological effects on ventricular remodelling and HF development, even if these benefits do not translate to reductions in hard clinical endpoints in contemporary unselected populations.37 38 Furthermore, the on-treatment analyses, which excluded patients who discontinued therapy on randomisation day and censored patients 7 days after permanent discontinuation, revealed potentially meaningful trends. These analyses showed HRs of 0.79 (95% CI 0.63 to 1.00) for the first primary outcome and 0.83 (95% CI 0.69 to 1.00) for the second primary outcome. While these findings must be interpreted cautiously as hypothesis-generating rather than definitive evidence, they raise questions about whether higher adherence rates might have demonstrated more pronounced benefits. The discontinuation rate of 28% in the spironolactone group vs 24% in the placebo group exceeded the anticipated levels.15 This higher-than-expected discontinuation rate, combined with lower-than-projected event rates, reduced the trial’s statistical power to detect potentially modest treatment effects. These implementation challenges reflect real-world considerations in MRA therapy, where concerns about hyperkalaemia and renal dysfunction often limit sustained utilisation.39 Our findings align with a broader reappraisal of legacy post-MI therapies in the context of contemporary reperfusion, lipid-lowering and antiplatelet strategies. Recent data questioning the necessity of prolonged β-blocker therapy in low-risk patients and supporting shorter or individualised durations of dual antiplatelet therapy illustrate how declining baseline risk can attenuate the absolute gains from adjunctive treatments. In a similar manner, the largely neutral or attenuated overall effects of routine post-MI MRA use in our analysis support a paradigm shift toward individualised deployment in well-defined high-risk subgroups, rather than indiscriminate use in all patients after MI. Future directions Our study findings necessitate recalibration of approaches to MRA therapy after MI. Rather than supporting routine administration to all MI patients, the evidence suggests a more targeted strategy focusing on those at highest risk for adverse ventricular remodelling and HF may be more appropriate. The remarkably low event rates in modern practice highlight the importance of careful risk–benefit assessment, weighing potential benefits against known risks of hyperkalaemia and renal dysfunction. Future post-MI MRA trials should be explicitly designed to distinguish pharmacologic inefficacy from underexposure due to early discontinuation. This includes protocolised early laboratory surveillance, standardised potassium and creatinine management algorithms, and careful selection of patients at acceptable risk for hyperkalaemia and renal dysfunction. Incorporating pragmatic adherence support tools and prespecified on-treatment analyses, alongside intention-to-treat results, would provide a clearer understanding of the true effectiveness and safety profile of MRAs in patients who can tolerate them, while maintaining patient safety as a priority. The overall neutral results, juxtaposed with a potential reduction in HF events, create a compelling case for more refined patient selection strategies. Biomarker-guided approaches, potentially incorporating aldosterone levels, markers of fibrosis such as galectin-3 or ST-2, or genetic determinants of MRA response, represent promising avenues for investigation.40,43 Third, the optimal timing and duration of MRA therapy remain incompletely defined. Future trials should prospectively evaluate different timing strategies, particularly ultra-early initiation (such as within the first 24 hours) compared with more delayed approaches, with prespecified subgroup analyses based on revascularisation strategy and infarct characteristics. Additionally, the ideal duration of therapy, whether limited to the high-risk post-MI period or continued indefinitely, merits dedicated investigation. Fourth, mechanistic studies exploring the relative contributions of MRAs’ anti-fibrotic, anti-inflammatory and electrophysiological effects to clinical outcomes would enhance our understanding of their biological effects. Cardiac MRI substudies evaluating extracellular volume fraction, myocardial strain and infarct characteristics could provide valuable insights into the structural mechanisms underlying clinical benefits.44 45 This evolution in evidence exemplifies how established therapeutic paradigms require continuous reevaluation as care standards advance. The journey from EPHESUS to CLEAR represents a case study in how large, well-conducted contemporary trials can fundamentally reshape our understanding of established therapies, challenging assumptions and refining approaches to optimise patient outcomes in an ever-evolving landscape of cardiovascular care. Limitations The study’s findings should be interpreted in the context of the following limitations. First, there is moderate heterogeneity across studies for all the endpoints examined, likely a result of a combination of the variability of trial populations, type of MRA agents, timing of initiation and dosing strategies. Second, this study used trial-level data rather than individual patient data. Thus, this approach limited our ability to conduct more granular analyses of treatment effects in specific patient subgroups or to adjust consistently for potential confounding factors across studies.33 Several pivotal trials prioritised composite endpoints such as cardiovascular death plus HF hospitalisation; however, substantial heterogeneity in endpoint definitions and incomplete reporting across the broader trial set precluded construction of a consistent composite suitable for pooled analysis. We therefore elected not to meta-analyse a harmonised cardiovascular death plus HF hospitalisation endpoint, to avoid introducing bias through post hoc assumptions. A more precise evaluation of such composites will require standardised individual patient data across trials. Individual patient data meta-analyses would have allowed for more precise subgroup analyses, consistent definition of outcomes across studies, and potentially more accurate time-to-event analyses. Additionally, the substantial reduction in event rates in contemporary practice creates statistical challenges for demonstrating treatment benefits, even for therapies with real biological effects. Conclusions In this meta-analysis of 15 randomised trials, MRAs did not demonstrate a statistically significant reduction in all-cause mortality, cardiovascular mortality or HF events in unselected post-MI populations treated with contemporary guideline-directed therapy. Subgroup and exploratory analyses suggest a possible signal of benefit in patients with reduced LVEF and with longer-term follow-up, consistent with current guideline recommendations for use in patients with HF or significant systolic dysfunction. Overall, these findings support a targeted rather than routine application of MRAs after MI, emphasise the need for careful risk–benefit assessment in clinical practice and highlight the importance of future studies using refined phenotyping and standardised designs to identify patients most likely to derive net benefit. Supplementary material Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Data availability free text: All the data are available within the article and its supplementary material. Ethics approval: This study used data from previously published literature. In addition, there are no direct human subjects or use of identified personal data; thus, ethical approval was not required. Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. References Associated Data Supplementary Materials Data Availability Statement All data relevant to the study are included in the article or uploaded as supplementary information.