Sudden cardiac death (SCD) remains one of the most devastating outcomes of non-ischemic cardiomyopathy (NICM). Although recent data suggest a decline in SCD incidence – likely reflecting advances in pharmacological therapy and heart failure management – accurately identifying patients at highest risk for malignant ventricular arrhythmias remains a major clinical challenge.1–3 The unpredictable occurrence of these events, often in patients who do not meet conventional high-risk criteria, highlights the need for improved risk stratification strategies.4

For decades, primary prevention implantable cardioverter-defibrillator (ICD) therapy has been guided almost exclusively by a left ventricular ejection fraction (LVEF) threshold of <35%, based largely on landmark clinical trials.5,6 However, this approach has important limitations. While reduced LVEF reflects heart failure severity, it lacks both sensitivity and specificity for predicting arrhythmic risk in patients with NICM.7,8

Cardiac magnetic resonance (CMR) imaging has emerged as a powerful tool to address these limitations. Beyond accurate LVEF quantification, CMR enables the detection of myocardial fibrosis using late gadolinium enhancement (LGE). In particular, midwall LGE has been identified as a strong arrhythmogenic substrate, independent of systolic function.9 Alongside imaging, genetic information has also been important in the evaluation of NICM, as specific genotypes are known to confer a high risk of arrhythmic events.10

The DERIVATE Risk Score 2.0 (DRS 2.0) was recently introduced as a multidimensional risk prediction model. By combining male sex, CMR-derived LVEF, and the presence and pattern of midwall LGE, the score demonstrated high diagnostic accuracy and prognostic value in its derivation and validation cohorts.4

However, its performance across different geographic regions and clinical settings remains uncertain. Given potential variations in genetic background, clinical practice, and referral patterns, external validation in diverse populations is essential to support broader clinical implementation.

This study aimed to evaluate the performance, discrimination, and calibration of the DRS 2.0 for predicting major arrhythmic events in a contemporary Portuguese cohort of patients with NICM.

The baseline characteristics of the 385 NICM patients are shown in Table 2.

Performance of the DRS 2.0 score

The median DERIVATE Risk Score 2.0 (DRS 2.0) was 6.0 (interquartile range: 4.0–9.0). According to predefined risk categories, 20% of patients were classified as low risk (n=76), 53% as intermediate risk (n=203), and 28% as high risk (n=106).

Over a mean follow-up of 3.0±1.4 years, 26 MAACE (6.8%) were observed, including seven sudden cardiac deaths, nine episodes of sustained ventricular tachycardia, and 10 appropriate implantable cardioverter-defibrillator therapies. Annualized MAACE rates increased progressively across risk categories, at 0.4%/year in the low-risk group, 1.0%/year in the intermediate-risk group, and 6.0%/year in the high-risk group. Patients classified as high risk by the DRS 2.0 had an eightfold higher risk of MAACE compared with those in the low- or intermediate-risk categories (hazard ratio 8.0; 95% confidence interval 3.36–19.04; p<0.001).

Event-free survival according to DRS 2.0 risk categories is illustrated in the Kaplan–Meier curves (Figure 1). At 3 years of follow-up, estimated event-free survival was 98.8%, 97.1%, and 82.1% in the low-, intermediate-, and high-risk groups, respectively.

Figure 1.

Kaplan–Meier curves and survival plot.

The DRS 2.0 demonstrated good discriminative ability, with a C-statistic of 0.76 (95% confidence interval 0.66–0.85; p<0.001) (Figure 2). Model calibration was satisfactory, as indicated by a non-significant Hosmer–Lemeshow goodness-of-fit test (χ2=0.802; p=0.371) (Figure 3).

Figure 2.

Receiver operating characteristics curve of the DRS 2.0 model for predicting MAACE. AUC: area under the curve.

Figure 3.

DRS 2.0 calibration using Hosmer–Lemeshow test.

In this contemporary cohort of patients in Portugal with non-ischemic cardiomyopathy (NICM), application of the DERIVATE Risk Score 2.0 (DRS 2.0) enabled stratification of arrhythmic risk across clinically distinct groups. A minority of patients classified as high risk accounted for most major arrhythmic cardiac events (MAACE), whereas most patients in the low- and intermediate-risk categories remained event-free during follow-up. These findings suggest that the score retains meaningful discriminatory capacity in this setting and may provide incremental clinical insight beyond conventional risk markers.

The overall incidence of arrhythmic events in this cohort was low, with an annual MAACE rate of 2.3% and a sudden cardiac death (SCD) rate of 0.6% per year. These results are consistent with contemporary data and likely reflect substantial advances in heart failure management. In the DANISH trial, for example, the annual SCD rate was 1.13% despite inclusion of patients with left ventricular ejection fraction (LVEF) <35%.2 The low event burden observed in this study is plausibly explained by high uptake of guideline-directed medical therapy (GDMT), as well as the use of cardiac resynchronization therapy in a subset of patients.

More than 80% of patients in this cohort were treated with renin–angiotensin system inhibitors and beta-blockers, while approximately 60% received mineralocorticoid receptor antagonists and sodium–glucose cotransporter 2 inhibitors. Although these figures remain slightly below those reported in randomized clinical trials such as DAPA-HF and EMPEROR-Reduced,11,12 they compare favorably with other real-world observational studies, underscoring the contemporary nature of this cohort.13,14 Notably, while 43% of patients had a CMR-derived LVEF <35%, only 26% ultimately received an implantable cardioverter-defibrillator (ICD) or CRT-D, closely mirroring the proportion of patients classified as high risk by the DRS 2.0. This discrepancy may reflect recovery of LVEF in some patients during follow-up, but it may also indicate that clinical decision-making is increasingly influenced by factors beyond LVEF alone.

The DRS 2.0 integrates simple clinical variables with advanced CMR-derived measures of ventricular function and myocardial tissue characterization. By incorporating not only the presence of late gadolinium enhancement (LGE), but also its extent and regional distribution, the score captures key features of the arrhythmogenic substrate in NICM more comprehensively than binary LGE assessment. This approach aligns with growing evidence that both the burden and anatomical pattern of myocardial fibrosis are critical determinants of arrhythmic risk.15,16

In recent years, several alternative models have been proposed to refine arrhythmic risk stratification in NICM, many of which include increasingly sophisticated LGE-based metrics, such as detailed quantification, three-dimensional remodeling indices, or measures of fibrosis heterogeneity.17,18 While these approaches are promising, their clinical applicability may be limited by technical complexity, lack of standardization, or restricted availability outside specialized centers. In this context, risk scores that balance prognostic performance with clinical feasibility remain particularly relevant.19

Importantly, validation of clinical risk scores in populations distinct from those in which they were derived is essential before widespread clinical adoption. Differences in genetic background, disease etiology, comorbidity burden, healthcare systems, referral patterns, and therapeutic strategies can all influence baseline risk and modify the performance of prognostic models. Scores developed in selected trial populations or highly specialized centers may demonstrate attenuated discrimination or miscalibration when applied to broader or different real-world cohorts.20 External validation therefore serves not only to confirm generalizability, but also to identify potential limitations, recalibration needs, or subgroups in which a given model may perform suboptimally.21

Several limitations of the present study should be acknowledged. Its single-center observational design and the relatively small number of arrhythmic events limited statistical power and precluded multivariable analysis. In addition, although these findings support the applicability of the DRS 2.0 in a Southern European cohort, further prospective validation in larger and more diverse populations is required. Future developments in risk stratification may also benefit from the integration of genotype-specific data, particularly for variants known to confer elevated arrhythmic risk in NICM.22,23

In this contemporary real-world cohort of patients with non-ischemic cardiomyopathy, the DERIVATE Risk Score 2.0 demonstrated the ability to stratify arrhythmic risk across clinically distinct groups. Despite low overall event rates in the setting of optimized guideline-directed medical therapy, the score identified a subset of patients with a disproportionate burden of major arrhythmic events. While further multicenter validation and incorporation of genetic data are warranted, these findings suggest that multiparametric risk models such as the DRS 2.0 may complement conventional LVEF-based approaches in refining arrhythmic risk assessment.

This study was performed in line with the principles of the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the study and the use of de-identified patient data. All data were anonymized prior to analysis to ensure patient confidentiality. Access to the electronic medical records was restricted to the research team. The authors declare that they have no known competing financial interests.

The authors have no conflicts of interest to declare.