How can we assess the risk for sudden cardiac death to decide for primary prophylactic implantable cardioverter-defibrillator in patients with heart failure in 2022?

This editorial refers to ‘The MADIT-ICD benefit score helps to select implantable cardioverter-defibrillator candidates in cardiac resynchronization therapy’ by J. Dauw et al., https://doi.org/10.1093/europace/euac039. It becomes increasingly evident that most patients treated with primary prophylactic implantable cardioverter-defibrillator (ICD) according to current guidelines will never gain any benefit from this treatment.1 Heart failure with left ventricular ejection fraction (LVEF) of <35% is recommended and used as a cutoff for offering patients this treatment. Left ventricular ejection fraction is a poor predictor for sudden cardiac death (SCD), and the rate of SCD seems to decline over the years,2 probably because of more effective pharmacological treatment of heart failure, with the recommendation to use several drug classes, more of which have been shown to reduce SCD, as well as better revascularization. Several risk-stratification tools have been tested for better prediction of SCD. In this issue of Europace, Dauw et al.1 present how well the MADIT-ICD Benefit Score3 can predict the outcomes of arrhythmia and non-cardiovascular (non-CV) death in a relatively contemporary population of consecutive patients treated with a cardiac resynchronization therapy pacemaker (CRT-P) or cardiac resynchronization therapy defibrillator (CRT-D). The MADIT-ICD Benefit Score identified a smaller group of patients with high arrhythmic risk and very low non-CV death risk. These are patients, in whom a relatively high chance of benefitting from primary prophylactic ICD implantation can be expected. The value of the MADIT-ICD Benefit Score was close to the multidisciplinary approach that was used in the institution to select which of the CRT-patients should be offered CRT-D based on an expected higher chance of achieving benefit from the ICD. The results are indeed of broad interest, and the authors should be commended for this work. Both risk-assessment tools seemed to identify a group of patients with high arrhythmic risk and low risk of dying from non-arrhythmic causes, and therefore a larger likelihood of experiencing benefit from being protected from SCD by the ICD. The CRT is known to reduce SCD,4 and it, therefore, is of importance and novel that this risk-assessment score can be used also in CRT-patients. Before implementing the MADIT-ICD Benefit Score for selecting who to give ICD, a few caveats need to be mentioned. Considering that the two models compared both include several parameters that are identical or alike—age, sex, comorbidities, prior myocardial infarction, and presence of non-sustained ventricular tachycardia, it is less surprising that they perform on a similar level. Recently, the Seattle Proportional Risk Model, which predicts the proportion of SCD vs. non-sudden death, was tested in the population included in the DANISH (Danish study to Assess the Efficacy of ICDs in Patients with Non-ischemic Systolic Heart Failure on mortality) trial and found valuable in the identification of patients who had a larger chance of gaining benefit from ICD implantation.5 In the same population, several other risk markers have been suggested as predicting benefit from prophylactic ICD treatment in non-ischaemic heart failure, as low N-terminal pro-brain natriuretic peptide6 and higher periodic repolarization dynamics.7 It is thus not straightforward to choose which of these risk-assessment scores or markers should be used to guide what to do in future populations. It also must be noted that there was no concordance between the two risk-stratification models investigated in the present study by Dauw et al.,1 as some patients ended up with high risk in one model and lower risk in the other model and vice versa. These scores allow identification of high-risk groups, but still leave physicians with larger groups of patients who have a low or at least lower risk of SCD and thus the chance of benefit from receiving an ICD. How to handle these patients? In many institutions, results from observational cohort studies will not be considered a sufficient basis for deciding on not to implant an ICD in some patients in whom there is a clear indication according to current guidelines, even though such a decision should be an individualized judgment including a high degree of shared decision-making with the patient. The MADIT-ICD Benefit Score was derived from the analysis of data from patients included in a randomized controlled trial, but itself never tested in prospective controlled trials. Most optimally, a randomized controlled trial should be conducted, where the benefit of ICD implantation is tested, at least in the low-to-intermediate risk for SCD groups. This is the only way to conclude firmly that ICD treatment is or is not beneficial when triaging patients by such a tool. Also, the only valid way to create a generalizable risk-assessment score. The PROFID project is a European effort towards personalized prediction and prevention of SCD after myocardial infarction, funded by the European Union over the Horizon 2020 programme (https://profid-project.eu/).8 The purpose is to first develop a clinical prediction model for the assessment of the individual risk for SCD in patients after myocardial infarction by analyzing data from a row of registries and research databases. A personalized decision-making for primary prevention ICD implantation with the application of this model will be compared with the current clinical LVEF-based practice in two large randomized clinical trials in patients with previous myocardial infarction: (i) the PROFID-Reduced trial in patients with ejection fraction (EF) ≤35% but a low predicted individual risk for SCD that will be randomized to ICD vs. non-ICD in a non-inferiority design (NCT04540354) and (ii) the PROFID-Preserved trial in patients with EF >35% and a high predicted individual risk for SCD that will be randomized to ICD vs. non-ICD in a superiority design (NCT04540289). The total number of patients in both trials will be ≈3920, and the primary endpoint in the two trials will be all-cause death. The results of the trials will document whether this personalized approach can be successfully applied across the whole range of EF for the decision whether to implant an ICD for primary prevention of SCD after myocardial infarction or not. In contrast to what was found in the MADIT-ICD Benefit Score and the institutional approach tested in the paper by Dauw et al.,1 the analysis of data on ≈225 000 patients in the PROFID risk-assessment model did not result in any model based upon baseline characteristics, biomarkers, and imaging that substantially improved the predictive performance of LVEF (https://www.escardio.org/The-ESC/Press-Office/Press-releases/Is-it-possible-to-personalise-sudden-cardiac-death-prevention-after-myocardial-infarction). An increasing amount of data indicate that cardiac magnetic resonance imaging (MRI) is valuable for risk stratification for SCD in patients with heart failure.9–11 Currently, statistical analysis including data from cardiac MRI is being performed in the PROFID dataset to see whether MRI can make a difference in risk stratification for SCD. Finally, the currently enrolling RESET-CRT trial (NCT03494933) includes patients scheduled for CRT implantation to be randomized to either CRT-P or CRT-D with respect to the primary outcome event of all-cause death. This trial is expected to answer the question of whether CRT-P is non-inferior to CRT-D. Data from this trial may add significantly to our knowledge on which CRT candidates will not benefit from ICD. While the results testing the MADIT-ICD Benefit Score are important, using this score will not provide us with a simple tool to select the patients that should receive ‘only’ CRT-P. Most appropriately, such decisions should be guided by evidence from randomized controlled trials.