Heart failure with reduced ejection fraction is a common condition associated with a poor prognosis and risk of sudden cardiac death (SCD; Fig. 1). Approximately 50% of deaths, especially in mild-to-moderate cases, are sudden.

Fig. 1
figure 1

Risk stratification for sudden cardiac death in cases of heart failure with reduced ejection fraction

Full size image

Ischemic heart disease and idiopathic dilated cardiomyopathy (DCM) are two frequent structural heart diseases associated with heart failure and SCD. Ischemic heart disease is the anatomical substrate in 80% of SCD events. Idiopathic DCM, on the other hand, accounts for 10% of SCD cases in the adult population [1], and up to 30% of deaths in patients with DCM are sudden and likely mediated by arrhythmia [2].

In these patients, multiple randomized controlled trials (with > 6000 patients) demonstrated the superiority of implantable cardioverter-defibrillators (ICD) over antiarrhythmic agents for primary prevention of SCD [3].

The only indicator shown to have an association with increased risk of SCD in the setting of ischemic or nonischemic left ventricular (LV) dysfunction is LV ejection fraction (LVEF); [4,5,6,7]. Therefore, this parameter has been used as the major criterion for ICD implantation for primary prevention of SCD, often in combination with New York Heart Association (NYHA) class, and is still listed in recent guidelines [8].

However, only 35% of patients randomized to the ICD arm in the Multicenter Automatic Defibrillator Implantation Trial II (MADIT II) received appropriate therapy during 3 years of follow-up [9]. It was reported that LVEF had limited accuracy in identifying patients at high risk for SCD.

Recent studies have evaluated novel echocardiographic parameters of myocardial deformation and cardiac magnetic resonance (CMR) with late-gadolinium enhancement (LGE) for risk stratification of patients with ischemic and idiopathic cardiomyopathy. This noninvasive imaging facilitates the characterization of arrhythmogenic substrate explaining the mechanism of ventricular arrhythmia. Moreover, it helps in the assessment of the anatomical substrate (viable myocardium and scar tissue) and transient factors such as myocardial ischemia.

In the present review, we describe the emerging role of these novel imaging parameters in identifying high-risk patients.

Left ventricular ejection fraction as a predictor of increased risk

Although LVEF has some limitations related to reproducibility, geometric assumptions, and experience, current guidelines for SCD risk stratification emphasize the use of this parameter.

The calculation of LVEF can be improved by using contrast echocardiography or a three-dimensional (3D) approach, but the gold standard for 3D quantification of left ventricular volumes and ejection fraction is CMR [10, 11].

The Marburg Cardiomyopathy Study (MACAS), including 343 patients with nonischemic heart failure, revealed that the relative risk for major arrhythmic events was 2.28 for every 10% decrease in EF, in patients with sinus rhythm.

International guidelines consider an EF ≤ 35% as a criterion for ICD implantation in primary prevention for patients with nonischemic heart failure [12]. However, most patients who suffer SCD have a preserved LVEF, and many patients with poor LVEF do not benefit from ICD prophylaxis.

Data from the Maastricht study confirm these findings and indicate that, among patients for whom EF was measured before an episode of SCD, 52% had an EF > 30% and 32% had an EF > 40% [13].

The use of the criterion of LVEF < 35% alone has limited power in predicting SCD in patients with nonischemic heart failure. In the DEFINITE trial, the use of low LVEF alone as an indicator for ICD placement was associated with both a low event rate of SCD in the control and treatment groups and a significant number of inappropriate ICD shocks (49 inappropriate versus 91 appropriate ICD shocks) in the treatment group [14].

Recently, results from the DANISH Study [15] suggest that for many patients with DCM, ICDs do not increase longevity.

It is clear from these data that there is a need for a better assessment of arrhythmic risk using other parameters for improved characterization and selection of patients for ICD implantation [16].

Mechanisms of ventricular arrhythmia

The pathophysiology of ventricular arrhythmias is complex and involves the anatomical and functional substrate as well as transient factors altering the electrophysiology of the substrate.

In patients with structural heart disease such as ischemic cardiomyopathy (ICM) or DCM, re-entry is the most frequent mechanism of severe ventricular arrhythmia.

A central area of conduction block (functional or fixed), a unidirectional conduction block, and a zone of slow conduction are substrates for re-entry. In infarcted areas, scar tissue is the most common cause of fixed conduction block. Furthermore, the interposition of bundles of fibrous tissue within layers of viable myocytes is a model of spatial heterogeneity and creates electrical dispersion and areas of unidirectional conduction block and slow conduction.

Myocardial scar is less common in DCM, bundle branch re-entrant ventricular tachycardia and focal automaticity ventricular tachycardia have been proposed as other arrhythmogenic mechanisms in DCM.

Noninvasive imaging and risk stratification

Role of magnetic resonance imaging

Cardiac magnetic resonance imaging has developed into a powerful tool that allows for a comprehensive cardiac assessment of left ventricular structure, function, perfusion, and tissue characteristics, including the presence or absence of fibrosis [17].

Contrast-enhanced MRI is the preferred imaging modality for evaluating the extent of scar after myocardial infarction, which is known to be an independent predictor of ventricular arrhythmias [18]. Contrast-enhanced MRI can detect scar areas as small as 0.16 g. The contrast agent is trapped in the extracellular matrix, which is increased in the infarct areas, and the scar appears as hyperenhanced, white areas. The extent and characteristics of the scar area on contrast-enhanced MRI have been related to increased risk of ventricular arrhythmias and cardiac death [19]. Moreover, the myocardial scar burden on contrast-enhanced MRI was superior to LVEF for prediction of ventricular arrhythmias [20].

In the study by Klem et al., including 137 patients considered for ICD placement, myocardial scar detected by cardiac MRI was an independent predictor of death or appropriate ICD discharge for sustained ventricular tachyarrhythmia. This study included a wider range of LVEF and showed that in patients with LVEF > 30%, significant scarring (≥5% LV) identifies a high-risk group similar in risk to those with LVEF ≤ 30%; by contrast, those with EF ≤ 30% and minimal scar (<5%) had similar risk to those with EF > 30% [19].

Contrast-enhanced MRI, using different signal intensity thresholds, can differentiate and quantify the core infarct zone and the peri-infarct or border zone (bundles of viable myocardium intermingling with fibrous tissue; [22]).

The infarcted myocardium can be divided into the following zones: core infarct zone; gray or peri-infarct zone; and total infarct = core + peri-infarct zones. The core and peri-infarct areas have been defined as areas with LGE signal intensity (SI) ≥ 3 SD, and 2 SD ≤ SI < 3 SD, respectively [23, 24].

In the Roes et al. study, the extent of the peri-infarct zone was the only independent predictor of appropriate ICD therapy or cardiac mortality [19].

A recent meta-analysis [25] was performed to identify the predictive accuracy of LGE-CMR for SCD risk stratification. The extent of LGE on CMR was strongly associated with the occurrence of ventricular arrhythmias in patients with reduced LVEF (relative risk estimated at 4.33 for all, 4.63 for ICM and 3.79 for nonischemic cardiomyopathy). Core scar and the gray zone are predictors of ventricular arrhythmia events with a relative risk of 3.82 (2.19–6.66) and 5.94 (2.82–12.52), respectively.

In DCM, contrast-enhanced MRI has provided important information on the relationship between myocardial scar burden, scar location, and the risk of ventricular arrhythmias [26,27,28]. Scar tissue usually involves the midwall (Fig. 2) or shows a patchy distribution.

Fig. 2
figure 2

Magnetic resonance imaging: four-chamber (a) and short-axis (b) views in a patient with dilated cardiomyopathy and a midwall area of fibrosis in the interventricular septum

Full size image

Nazarian et al. [29] demonstrated that the distribution of myocardial scar assessed with contrast-enhanced MRI was predictive of inducible sustained monomorphic ventricular tachycardia in 26 patients with DCM. In this study and after adjustment for LVEF, the presence of fibrosis covering 26–75% of the wall thickness was associated with a ninefold increase in the risk of ventricular arrhythmia according to an electrophysiological study.

In the study by Wu et al. [30], SCD or appropriate ICD discharge were detected in 22% of patients with CMR evidence of myocardial scar versus only 8% of patients without evidence of gadolinium enhancement (p = 0.03). In these patients, fibrosis of the midwall detected by LGE-CMR was associated with adverse cardiac events (hospitalization for heart failure, appropriate ICD firing, and cardiac death; [30, 31]).

Neilan et al. [32] determined the prognostic value of LGE in 162 patients with nonischemic cardiomyopathy, and found that cardiovascular death and appropriate ICD therapy were substantially higher in patients with LGE (24%) than in those without LGE (2%).

The presence and the extent of LGE have the strongest associations with cardiovascular death and appropriate ICD therapy: LGE presence, hazard ratio (HR): 14.5; p < 0.001; LGE extent, HR: 1.15 per 1% increase in volume of LGE; p < 0.0001.

However, in multivariate analyses, LGE extent was the strongest predictor of cardiovascular death and appropriate ICD therapy (a sevenfold hazard per 10% LGE extent after adjusting for patient age, sex, and LVEF; adjusted HR: 7.61; p < 0.0001).

Given the multiple small and single-center studies reporting on the prognostic data of LGE in patients with DCM, a systematic review and meta-analysis was performed. The meta-analysis [33] collected data from nine studies with a total of 1488 patients and a mean follow-up of 30 months. It was found that LGE was present in 38% of patients. Those with LGE had increased overall mortality (odds ratio, 3.27; p < 0.00001) and SCD/aborted SCD (odds ratio, 5.32; p < 0.00001) compared with those without LGE.

The largest study on this topic was recently published by Gulati et al. [34], including 472 patients with nonischemic heart failure examined with MRI and with a median follow-up of 5.3 years. Combined events of SCD and aborted SCD were observed in 29.6% patients with myocardial fibrosis and 7.0% patients without fibrosis. For this event, the presence of fibrosis represented an HR of 4.61 (95% CI, 2.75–7.74; p < 0.001).

Another recent study [21] enrolling patients with ischemic cardiomyopathy or nonischemic cardiomyopathy found that the presence of both LGE and LVEF < 30% increased the event rate of SCD or ICD discharge compared with event rates in patients with LVEF < 30% alone.

This additive prognostic value of LGE was also demonstrated in the large study by Gulati et al. [34]. After multivariate analysis with adjustment for EF and other prognostic factors, the presence of fibrosis represented an HR of 2.43 (95% CI, 1.50–3.92; p < 0.001), and the extent of fibrosis represented an HR of 1.11 (95% CI, 1.06–1.16; p < 0.001).

Disertori et al. analyzed data from 19 studies of SCD primary prevention, which included 2850 patients with 423 arrhythmic events over an average follow-up of 2.8 years. The patients had either ischemic cardiomyopathy (31%) or nonischemic cardiomyopathy and ventricular dysfunction; the composite arrhythmic outcomes included SCD, aborted SCD, ventricular tachycardia/fibrillation, and ICD therapy. Patients with negative LGE test results had a composite annualized event rate of 1.7% versus 8.6% for those positive LGE test results (p < 0.0001). In both the etiology-based and EF-based subgroups, LGE correlated with arrhythmic events. In the overall population, the pooled odds ratio was 5.62 (95% CI: 4.20–7.51; [35]).

The second recent meta-analysis, performed by Di Marco et al., included 29 studies (2948 patients). It was found that LGE was significantly associated with the arrhythmic endpoint both in the overall population (odds ratio: 4.3; p < 0.001) and when including only those studies that performed multivariate analysis (hazard ratio: 6.7; p < 0.001; [36]).

Magnetic resonance imaging can also be helpful for patients with DCM and mild or moderate reductions in LVEF (>35–40%), in which midwall LGE identifies a group of patients at increased risk of SCD. This finding is important because these patients are not currently offered ICDs for the primary prevention of SCD on the basis of guideline recommendations [36, 37].

The main studies evaluating the association between myocardial fibrosis assessed via CMR and the risk of arrhythmic and nonarrhythmic events are summarized in Table 1.

Table 1 Studies of association between myocardial fibrosis assessed via CMR and risk of arrhythmic and nonarrhythmic events in IDCM and NIDCM
Full size table

Current studies examining LGE by CMR in patients with DCM use varying definitions to define the presence and extent of LGE [42]. Different thresholds of signal intensity have been proposed to determine the presence of LGE, but there is a lack of consensus on an acceptable threshold for the diagnosis of LGE. This is more challenging in DCM where the intensity of the LGE is much more variable than in ischemic heart disease.

Assomull et al. [26] found that these patients with DCM and with LGE of ≥ 4.8% of LV mass were at higher risk of cardiovascular events than those with LGE < 4.8%.

In the study by Neilan’ et al. [32], patients with LGE involving > 6.1% of LV myocardium had the highest cardiovascular death and appropriate ICD therapy.

Different methods are also used to determine LGE extent as a percentage of LV mass or scar volume. Although current guidelines recommend using the 2‑SD method, data suggest that the use of this technique leads to an overestimation of the extent of LGE in comparison with other techniques [23]. There is so a need for homogeneity in the definition of both the presence and extent of LGE so as to improve reproducibility and standardize the technique.

The LGE border zone on CMR imaging has also been proposed as an independent predictor of ventricular arrhythmias, and a recent meta-analysis [43] found that the quantification of the LGE border zone is the strongest predictor of appropriate ICD therapy, as a surrogate for SCD, in ICM patients with primary prophylactic ICD at medium- to long-term follow-up.

Role of myocardial deformation on echocardiography

Speckle-tracking imaging is a relatively new approach for assessing myocardial deformation by detecting features on grayscale 2D images.

The assessment of LV global longitudinal strain with 2D speckle-tracking echocardiography has been shown to be an accurate marker of LV function. This technique is feasible and reproducible without geometric assumptions, and is independent of LV geometry.

Since global longitudinal strain measures pure longitudinal function, and it may provide other information than EF, which is strongly influenced by the radial motion of the myocardium [44]. Global longitudinal strain has been reported to provide superior prognostic information in the setting of ischemic heart disease [44, 45].

The Bertini et al. study was a larger one including a homogeneous population with chronic ischemic heart disease (1060 patients). In this study, global longitudinal strain was independently related to all-cause mortality (HR per 5% increase, 1.69; 95% CI, 1.33–2.15; p < 0.001) and combined end point (all-cause mortality and heart failure hospitalization; 1.64; 95% CI, 1.32–2.04; p < 0.001) and patients with an LV global longitudinal strain value of ≤ −11.5% had better outcome than those with LV global longitudinal strain > −11.5% [46].

Iacoviello et al. [47] studied a group of heart failure patients affected by ischemic or nonischemic DCM without a history of sustained ventricular arrhythmias. During a follow-up of 26 ± 13 months, 31 of 230 patients experienced ventricular tachycardia/fibrillation or SCD.

At multivariate analysis, global longitudinal strain remained significantly associated with ventricular arrhythmic events. The best global longitudinal strain cut-off value for the 1‑year occurrence of major ventricular arrhythmias was −10.0% (73% sensitivity and 61% specificity).

Longitudinal strain also adds incremental prognostic value to EF alone for the prediction of adverse outcomes in both ischemic and nonischemic cardiomyopathy [48]. This finding was also noted in a multicenter study that included 147 patients with heart failure with an LVEF ≤ 45% (ischemic in 42.8%).

Among prognostic factors obtained by echocardiography, global longitudinal strain was the best predictor of cardiac events, and a cut-off value of −7% predicted cardiac events at 12 months with high sensitivity and specificity [49].

In the study of Motoki et al. [50], which included 194 patients with chronic systolic heart failure, global longitudinal strain was an independent prognostic factor for cardiac events in heart failure regardless of age, LVEF, ischemic etiology and E/e’, and it had greater prognostic power than LVEF.

In ischemic cardiomyopathy, the functional properties of the peri-infarct zone have been evaluated with 2D speckle-tracking echocardiography (Fig. 3).

Fig. 3
figure 3

Infarcted inferior and inferoseptal myocardium (dark blue) with more preserved longitudinal strain in peri-infarct zone (pale blue)

Full size image

Usually, the infarct zone had the most impaired longitudinal strain, whereas the peri-infarct zones had more preserved longitudinal strain.

In 424 patients with ischemic cardiomyopathy considered for ICD implantation, the presence of impaired segmental longitudinal strain in the peri-infarct zone was independently associated with an increased risk of appropriate ICD therapy for ventricular tachycardia and fibrillation [51].

The relationship between the longitudinal peak systolic strain of the peri-infarct zone detected with speckle-tracking echocardiography and monomorphic ventricular tachycardia inducibility in patients with chronic ischemic cardiomyopathy was explored. Only longitudinal peak systolic strain of the peri-infarct zone was independently related to monomorphic ventricular tachycardia inducibility [52].

Mechanical dispersion, another strain parameter that reflects contraction heterogeneity, was recently used in a prospective, multicenter study of patients after myocardial infarction. This parameter predicted arrhythmic events independently of LVEF. A combination of mechanical dispersion and global longitudinal strain may improve the selection of patients after myocardial infarction for ICD therapy, particularly when LVEF is > 35% [53].

Conclusion

  • This review detailed the emerging role of the novel imaging parameters in identifying high risk patients with ischemic or nonischemic dilated cardiomyopathy.

  • Cardiac magnetic resonance imaging can be a useful technique for the risk stratification of these patients but there is a need for a homogeneous definition for both the presence and extent of late-gadolinium enhancement to improve reproducibility and to standardize the technique.

  • Strain assessment by echocardiography, particularly longitudinal strain, can also be used for cardiovascular risk stratification in patients with heart failure with greater accuracy than left ventricular ejection fraction (LVEF), but the cut-off values must be better defined.

  • Future risk stratification for arrhythmia and patient selection for implantable cardioverter-defibrillator placement may rely on a multiparametric approach using combinations of imaging modalities that may complement primary reliance on LVEF.