Why revisit ICD indications after MI

More than two decades after landmark trials established the role of the implantable cardioverter defibrillator, the epidemiology and management of post-infarction arrhythmic risk have evolved. Contemporary percutaneous revascularization, high-intensity lipid lowering, mineralocorticoid receptor antagonists, and sodium-glucose cotransporter-2 inhibitors have collectively reduced adverse events. As a result, the absolute benefit of an ICD for primary prevention likely varies more widely than when earlier trials were conducted. Meanwhile, device-related complications, lead failure, and inappropriate shocks remain nontrivial. This combination of lower baseline risk and persistent harms invites a reappraisal of who should receive a device and who might safely defer implantation.

Traditional selection has centered on reduced left ventricular ejection fraction, a coarse surrogate for substrate and trigger mechanisms that lead to lethal ventricular arrhythmias. Yet a substantial proportion of sudden cardiac deaths occur in individuals with moderately preserved systolic function, while many with severely reduced function never experience malignant arrhythmias. This discordance supports a broader lens that integrates scar burden, conduction heterogeneity, and autonomic balance. It also underscores the importance of time since infarction and the dynamic recovery of structure and function. A precision approach aims to quantify risk on a continuous scale and align interventions accordingly.

Beyond epidemiology, health system realities sharpen the question. Device implantation consumes procedural resources and entails long-term surveillance, generator replacements, and possible lead extraction. Policymakers and payers seek evidence that devices are concentrated where net benefit is maximized, particularly as competing technologies like wearable or subcutaneous systems expand choices. Patients, for their part, value longevity but also freedom from shocks and procedural complications. The clinical community therefore faces a dual mandate: refine prognostication and incorporate patient priorities into decisions about primary prevention ICDs.

In this setting, the PROFID EHRA randomized clinical trial tests whether risk-guided selection can deliver better alignment between patient risk and device benefit after infarction. Rather than presuming uniform benefit below or above an arbitrary ejection fraction threshold, the approach operationalizes multiparametric prediction to inform implant decisions. The objective is pragmatic and patient-centered: reduce under-treatment among those at high risk for arrhythmic death and avoid over-treatment among those with low predicted risk. If successful, the strategy could shift practice patterns, payer policies, and guideline language toward individualized thresholds.

Limitations of LVEF-only selection

Ejection fraction captures global pump performance but incompletely reflects arrhythmogenic substrate. Two patients with the same ejection fraction may have profoundly different scar geometry, conduction channels, and repolarization reserve. A purely structural threshold also overlooks triggers such as myocardial ischemia, electrolyte disturbance, and autonomic fluctuations. Moreover, reliance on a single cutoff marginalizes time-varying recovery, including reverse remodeling and scar maturation after revascularization. These nuances help explain both non-events in low ejection fraction cohorts and catastrophic events among those not typically considered for devices.

Clinical experience reinforces these gaps. Not all post-infarction ventricular arrhythmias arise from transmural scar, and nonischemic substrates may coexist in some individuals. Heterogeneity in infarct territory, microvascular obstruction, and intramural scar corridors influences reentry propensity beyond global function. The physiologic readouts we often use at the bedside, such as QRS duration or mitral regurgitation grade, are imperfect correlates of the underlying 3-dimensional substrate. A precision strategy calls for integrating more direct signals of substrate and triggers where feasible and validated.

Toward multivariable risk assessment

Risk models can synthesize predictors across domains. Examples include demographic and clinical factors, medication exposure, coronary anatomy and completeness of revascularization, and electrophysiologic signals. Imaging of myocardial scar and border zone characteristics offers structural context, while arrhythmic burden and conduction heterogeneity provide functional context. The statistical architecture can be flexible, but transportability, calibration, and interpretability are essential. For clinicians, the most valuable model is one that estimates absolute risk within a relevant timeframe and meaningfully reclassifies decisions compared with standard care.

Because device harms are not trivial, model thresholds must be tied to net benefit rather than event risk alone. Even a well-calibrated predictor may not change care if the implied benefit of the device is small relative to procedural risk and long-term complications. Conversely, if a model identifies truly high-risk individuals outside conventional criteria, a risk-guided strategy could reduce preventable deaths otherwise missed by ejection fraction thresholds. Confidence intervals around predicted risk, sensitivity analyses across subgroups, and planned updating strategies should be part of the implementation blueprint.

Inside the PROFID EHRA trial

The PROFID EHRA randomized clinical trial, publicly indexed on PubMed, is designed to evaluate an individualized strategy for primary prevention ICD use after infarction. The core premise is to randomize patients based on predicted arrhythmic risk rather than solely on ejection fraction. The trial sets out clinically relevant endpoints that reflect both mortality and device-related outcomes in contemporary practice. It is anchored in the recognition that modern pharmacotherapy and revascularization alter baseline risks compared with historical cohorts used to define current indications.

Patient selection targets a population with prior infarction in whom the question of device benefit is equipoise when judged by standard criteria. A multivariable model is used to estimate absolute risk of arrhythmic death within a prespecified horizon that is meaningful to patients and clinicians. Those predicted to be high risk yet not routinely eligible, and those predicted to be low risk despite conventional eligibility, are candidates for randomized evaluation of ICD implantation versus no ICD. This design probes the value of risk-guided reclassification in both directions.

The investigative network leverages the expertise of the European Heart Rhythm Association to standardize device programming, follow-up, and adjudication. Outcome definitions encompass sudden cardiac death, all-cause mortality, appropriate and inappropriate therapies, and major device complications. Patient-reported outcomes and health economics are incorporated to reflect real-world trade-offs. The statistical plan considers both superiority and noninferiority hypotheses depending on the randomization stratum, aligning analytical frameworks with the clinical questions posed.

Importantly, the protocol emphasizes safety oversight and predefined stopping rules. A data and safety monitoring structure can respond to imbalances in death or serious adverse events across arms. The trial also delineates pragmatic pathways for crossover, particularly in response to clinical deterioration or new risk information. By embedding flexible yet prespecified management rules, the design aims to preserve internal validity while accommodating the realities of longitudinal cardiovascular care.

Risk components and modalities

Although final model composition is trial-specific, several domains are biologically and clinically plausible contributors to predicted risk. Structural substrate can be characterized with cardiac magnetic resonance, including quantification of late gadolinium enhancement and border zone features. Functional signals include premature ventricular complexes, non-sustained tachycardia, and markers of dispersion or alternans. Conduction and repolarization heterogeneity can be summarized by electrocardiographic markers such as QRS duration, QTc, and T-wave morphology parameters. Clinical context encompasses age, comorbidities, renal function, ischemic burden, and medication exposure.

Such multimodal integration supports individualized estimates of arrhythmic death, distinct from competing risks like progressive pump failure. Because post-infarction mortality arises from multiple pathways, modeling approaches must explicitly handle competing risks to avoid overstating arrhythmic hazards. External validation and recalibration across centers strengthen generalizability. For bedside application, interpretability and workflow compatibility matter as much as discrimination, arguing for parsimonious predictors that can be measured reliably and repeated over time.

Randomization and endpoints

The trial randomizes eligible patients to ICD implantation or no ICD within predefined strata based on predicted risk and conventional candidacy. In strata where conventional guidelines would recommend an ICD but the model indicates low predicted risk, the design asks whether forgoing the device is noninferior for arrhythmic and all-cause mortality while reducing device-related harms. In strata where conventional guidelines would not typically recommend an ICD but the model indicates high predicted risk, the design asks whether the device confers superiority on arrhythmic endpoints. This bidirectional test of reclassification is a notable innovation.

Primary outcomes prioritize events that matter most to patients and clinicians. Sudden cardiac death and all-cause mortality offer complementary views of benefit, while appropriate shock rates capture device action on life-threatening arrhythmias. Prespecified safety outcomes include infection, lead dislodgement, pocket complications, inappropriate therapy, and psychological distress related to shocks. Secondary outcomes encompass hospitalization, quality of life, and cost-effectiveness. Event adjudication is blinded and standardized to ensure consistent assignment across sites.

Device programming and follow-up

Standardized programming reduces variability that could confound efficacy and safety comparisons. Longer detection intervals and higher rate cutoffs can reduce inappropriate therapies while maintaining protection from fast ventricular arrhythmias. Remote monitoring protocols enable early detection of lead issues and arrhythmic events, aiding timely clinical responses. Follow-up schedules align with contemporary recommendations and ensure capture of both early procedural risks and longer-term outcomes. Harmonization across centers helps isolate the effect of the selection strategy itself.

Balanced management of antiarrhythmic drugs, beta-blockade, and ischemia-directed therapy is equally important. These treatments influence arrhythmic substrate and triggers and may interact with device efficacy. By protocolizing background therapy to the extent feasible, the trial strives to maintain generalizability while minimizing confounding. The result is a platform that tests not simply a device but a risk-guided pathway for deploying it in contemporary care.

Statistical and ethical guardrails

The analytical plan accounts for clustering by site and prespecified covariates, with sensitivity analyses exploring robustness to model calibration drift. Missing data strategies and predefined per-protocol and intention-to-treat analyses are included to handle real-world deviations. Interim analyses focus on safety first, preserving power for final efficacy testing. Subgroup analyses probe heterogeneity of treatment effect across age, sex, infarct territory, and time since index infarction, while guarding against false discovery.

Ethically, equipoise is central. Patients randomized to forgo a device in low predicted risk strata are monitored closely, with rescue pathways if risk changes materially. Conversely, those offered a device in high predicted risk strata undergo thorough counseling about procedural and long-term risks. In both cases, shared decision making is embedded, recognizing that preferences about shocks, procedures, and quality of life differ. The trial thereby aligns statistical rigor with respect for patient values.

Clinical and system implications

Should the risk-guided strategy demonstrate favorable net benefit, implications would extend beyond electrophysiology. Clinical pathways could incorporate risk estimation at set intervals after infarction, perhaps at the time of discharge and again after an optimization period for guideline-directed therapy. A validated model could be integrated into electronic records to surface actionable risk to clinicians. The result would be a workflow that normalizes precision assessment rather than relying on ejection fraction snapshots alone. Importantly, such pathways would need to preserve flexibility for clinician judgment in exceptional scenarios.

Guideline committees may consider conditional recommendations that tie device implantation to model-derived thresholds of predicted arrhythmic death and estimated net benefit. This approach mirrors the evolution seen in lipid management, anticoagulation, and valve disease, where multivariable risk, patient values, and absolute benefit guide decisions. For reimbursement, payers could recognize validated risk tools as medical necessity criteria, supporting coverage where benefit is likely and avoiding misaligned spending. Implementation science will be key to ensuring equitable access and avoiding unintended disparities in who gets evaluated and who receives devices.

In parallel, professional education would need to emphasize risk communication and interpretation. Clinicians should be prepared to explain absolute risk, uncertainty, and how devices influence outcomes under different scenarios. Decision aids can make trade-offs more transparent and ensure that the device choice aligns with patient preferences. For centers, operational adjustments may include dedicated post-infarction clinics or virtual pathways that combine clinical, electrocardiographic, and imaging inputs into a consistent risk assessment. These steps can translate trial findings into sustained practice change.

Future research could refine and extend the framework. Incorporating dynamic predictors, such as changes in arrhythmic burden or remodeling, may improve calibration over time. Pragmatic studies can test model updating and transportability across regions and healthcare systems. The interaction between antiarrhythmic drugs, ischemia control, and device efficacy warrants careful analysis to optimize comprehensive care. Finally, qualitative research on patient perceptions of risk and device therapy can illuminate barriers and facilitators to adoption in everyday practice.

Navigating uncertainty and next steps

While the rationale for risk-guided selection is compelling, several uncertainties remain. First, the magnitude of risk separation achievable in routine care will influence effect sizes and the feasibility of noninferiority or superiority conclusions. Second, model performance may vary across institutions and populations, necessitating recalibration and governance for updates. Third, as therapies and technologies evolve, so too will background risks, challenging static thresholds. These realities argue for a living approach to risk assessment embedded in quality improvement cycles.

For now, the PROFID EHRA trial offers a rigorously conceived test of a strategy already gaining intuitive traction. It operationalizes personalized risk to guide high-stakes decisions in a domain where harms and benefits are both meaningful. By aligning clinical, statistical, and ethical frameworks, the trial sets a high bar for evidence that could reshape how ICDs are deployed after infarction. Whatever the results, the field will gain clarity on how far beyond ejection fraction we can and should go in targeting protection from lethal arrhythmias.

In synthesis, risk-guided ICD implantation after infarction aims to match therapy intensity to individualized need, maximize net clinical benefit, and steward device resources responsibly. The PROFID EHRA trial provides the crucible to test this approach under contemporary conditions, with endpoints that matter to patients and health systems alike. If validated, risk stratification could become a standardized step in post-infarction care, complementing guideline-directed therapy and improving outcome precision. If not, the work will still delineate limits and inform the next generation of models and trials.