Sudden Cardiac Death Risk Stratification in Heart Failure -The Potential Role of Biomarkers Part 1


Although there has been significant recent progress in the management of heart failure its associated mortality remains high. A large proportion of these patients die suddenly, termed sudden cardiac death (SCD), mostly from potentially reversible malignant cardiac arrhythmias. Despite the availability of a highly effective treatment in the form of an implantable cardioverter defibrillator (ICD), SCD in the heart failure population is still a significant problem. One important reason for this is the difficulty in identifying which patients are at highest risk of SCD and would benefit from an ICD. A number of tests are currently available to risk stratify heart failure patients at risk of SCD. However, used alone or in combination these are not sufficiently accurate and there is significant need for better risk stratification tools.

Multiple studies have demonstrated that serum biomarkers can accurately predict adverse outcomes in patients with heart failure of both ischaemic and non-ischaemic aetiology. A range of biomarkers predict both the occurrence of SCD in patients without ICDs and the occurrence of malignant arrhythmias in patients with devices, and in these studies individual biomarkers are at least as accurate as the current best markers of SCD risk. The pathophysiology of SCD is a complex process with a range of electrophysiological and molecular alterations contributing to arrhythmogenesis in the failing heart. By providing an assessment of these various processes, serum biomarkers may improve prediction of SCD in heart failure and help guide ICD use. Furthermore, it is likely that optimal SCD risk stratification will require the combination of multiple tests that reflect these diverse upstream processes. As such the greatest potential benefit of biomarkers may be in measuring multiple complementary markers that assess distinct aspects of arrhythmic risk.


There has been significant recent progress in the management of heart failure with advances in neurohormonal blockade and the advent of device therapy. In spite of this the mortality associated with heart failure remains high – 80% of men and 70% of women under the age 65 with heart failure will die within 8 years [1]. A large proportion of these patients die suddenly, termed sudden cardiac death (SCD), mostly from potentially reversible malignant arrhythmias. Despite the availability of a highly effective treatment, in the form of an implantable cardioverter defibrillator (ICD), SCD in the heart failure population is still a significant problem. One important reason for this is the difficulty in identifying which patients are at highest risk of SCD and would benefit from an ICD. In this topic we review the importance and pathophysiology of SCD in heart failure, detail the currently available tools for SCD risk stratification, and consider the potential role of biomarkers.

The Impact of Sudden Cardiac Death in Heart Failure

Cardiac death soon after symptom onset - termed sudden cardiac death – is a major health problem. It is the commonest mode of death in the developed world and causes approximately 100,000 adult deaths per year in the United Kingdom and four times that in the United States [2-4]. In patients who die within an hour of the onset of symptoms or during sleep, more than 90% will be due to cardiac arrhythmias [5], and most of these events are likely to be caused by potentially reversible ventricular tachyarrhythmias [6].

SCD is a major cause of mortality in heart failure irrespective of its aetiology. Early data concerning the importance of SCD in heart failure came from epidemiological studies. Among 652 members of the Framingham Heart Study who developed congestive heart failure, 5-year survival rates after disease onset were 25% in men and 38% in women, and up to half of these deaths were sudden [7,8]. These findings still hold despite contemporary management. Mozaffarian et al assessed the mode of death in 10,538 ambulatory patients with New York Heart Association class II-IV heart failure enrolled in 6 randomised trials or registries [9]. Ischaemic heart disease accounted for 62% of cases. During 16,735 person-years of follow-up, 2014 deaths occurred, including 1014 sudden deaths and 684 pump-failure deaths. Though overall sudden death was the commonest mode of death, pump-failure was more frequent in advanced heart failure. Solomon et al studied 14,609 patients with asymptomatic left ventricular dysfunction or heart failure after myocardial infarction [10]. Over a median follow-up of 180 days there were 1067 cardiac arrests, 903 leading to death, which accounted for approximately a third of all deaths.

The Pathophysiology of Sudden Cardiac Death in Heart Failure

Most cases of SCD in heart failure result from a malignant ventricular arrhythmia, either ventricular fibrillation or ventricular tachycardia [11]. This is supported by data from patients dying suddenly while undergoing Holter recording. In 157 episodes of SCD in ambulatory patients undergoing monitoring, 84% were secondary to ventricular arrhythmias, most commonly ventricular fibrillation (62%), while bradycardias accounted for only 16% [12]. Though these were not exclusively patients with heart failure, it is probable that the mechanisms in heart failure are similar.

The underlying electrophysiological and molecular processes that lead to these malignant arrhythmias are incompletely understood. However there is likely to be a complex interplay between acquired abnormalities of cardiac structure and function, and genetic predisposition. The acquired changes include alterations in myocardial repolarisation, calcium homeostasis and neurohormonal signalling [13-15]. Two of the more important processes are action potential prolongation, due to changes in ion channel expression, and alterations in neurohormonal signalling.

Action Potential Prolongation and Ion Channel Expression

Prolongation of the action potential (AP) is a consistent finding in the ventricular myocardium of failing hearts irrespective of the cause [16]. The underlying physiological basis of the changes in AP duration is alteration in the functional expression of ion channel proteins, including potassium and sodium channels. The ventricular myocardium has a number of distinct classes of voltage-gated potassium ion channels. The most consistent finding in human and animal heart failure models is the downregulation of the Ito protein, but changes in the potassium channels IKr and IK have also been noted [17-19]. Furthermore the importance of different potassium channels may vary depending on the aetiology of the heart failure [20]. Changes in sodium channels, which are important in the maintenance of the plateau phase of the action potential, have also been implicated [21].

The AP prolongation that occurs as a result of these changes in ion channel expression is inhomogeneous, leading to spatial and temporal heterogeneity in ventricular repolarisation [22]. It is this dispersion of repolarisation that may provide the substrate for the occurrence of malignant ventricular arrhythmias that lead to SCD [23]. These changes in repolarisation can be detected on the surface electrocardiogram (ECG), and form the basis of the risk stratification test Microvolt T-wave Alternans described below [24].

Altered Neurohormonal Signalling

Abnormal neurohormonal activation plays an integral role in the genesis of ventricular arrhythmias. Although the exact details of altered neurohormonal signalling are debated there is widespread acceptance of the importance of the autonomic nervous system and the renin-angiotensin-aldosterone (RAAS) system. Modulation of these neurohormonal systems have been shown to improve prognosis in patients with heart failure, including sudden death, and therapies that target them are now the mainstay of treatment for heart failure [25]. Further evidence of the importance of the sympathetic nervous system comes from the observation that there is a circadian variation in the frequency of SCD [26].

Myocardial infarction leads to sympathetic dennervation in the infarct zone [27]. This may be followed by neurilemma cell proliferation and axonal regeneration (nerve sprouting) leading to increased sympathetic nerve density or hyperinnervation in some areas of the myocardium [28]. In the normal human ventricle sympathetic activation causes a reduction in the action potential duration and a decrease in the dispersion of repolarisation [29]. In the failing heart the juxtaposition of dennervated and hyperinnervated myocardium may lead to spatial heterogeneity in ventricular repolarisation during sympathetic activation, predisposing to ventricular arrhythmogenesis [28]. Measuring these alterations in autonomic function has been demonstrated to be predictive of SCD, though such tests are not currently in widespread clinical use.

The RAAS system, through its two main effectors angiotensin II and aldosterone, has a range of effects on the myocardium that may predispose to malignant arrhythmias. These include induction of myocardial hypertrophy, increased collagen synthesis, promotion of inflammation and thrombosis, and modulation of active membrane properties [13].

Genetic Predisposition

Evidence for genetic predisposition to SCD comes from epidemiological data. Jouven et al assessed the occurrence of SCD in 7746 middle-aged men in the Paris Prospective Study. The risk of sudden death was increased by 80% in men who had a parental history of SCD, and nearly 9 times with a history in both parents [30].

It is well established that mutations in genes coding for cardiac ion channels underlie a range of heritable conditions that predispose to ventricular arrhythmias and SCD, including Long QT and Brugada syndrome [31]. It is also becoming clear that some gene polymorphisms, while not causing monogenic inherited arrhythmogenic syndromes, can increase susceptibility to proarrhythmic drugs by reducing "repolarisation reserve" [32]. It may be that specific polymorphisms in cardiac ion channel genes similarly predispose patients with heart failure to arrhythmias.

Implantable Cardioverter Defibrillators

Since their introduction in the 1980s ICDs have revolutionised the management of patients at high risk of SCD. Multiple large randomised controlled trials have demonstrated that ICDs reduce mortality from SCD in high risk patients [33]. They are currently given to two groups of patients: survivors of life-threatening arrhythmias (secondary prevention) and patients at high risk for developing a life-threatening arrhythmia (primary prevention). In both of these settings they are both highly efficacious and cost effective [11,34].

Despite considerable effort to improve the results of out-of-hospital cardiac arrest, survival remains relatively low. Annual survival rates to hospital discharge of out-of-hospital cardiac arrest secondary to ventricular fibrillation are between 24% and 33% [35]. The use of ICDs for primary prevention of SCD is therefore of paramount importance in reducing overall SCD rates. In this respect the key issue is risk stratifying patients for SCD to identify which groups are at highest risk. While selecting patients for a secondary prevention ICD is relatively straightforward, identifying patients for primary prevention device therapy is more difficult.

Traditional Risk Stratification Tools to Guide Primary Prevention ICD use

Risk stratification has been studied primarily in patients with congestive heart failure (CHF) or asymptomatic left ventricular dysfunction, as these groups are well known to be at increased risk of SCD. A large number of tests have been evaluated. These include tests of left ventricular function, autonomic function, ventricular repolarisation, and the presence or absence of spontaneous or inducible ventricular arrhythmias. The diverse nature of these tests reflects the complex underlying pathophysiology of ventricular arrhythmogenesis. The clinically relevant risk stratification tests are:

Left Ventricular Ejection Fraction (LVEF)

Depressed LVEF, as measured by echocardiography, contrast and radionuclide ventriculography, or magnetic resonance imaging, has long been recognised to be the most important determinant of all-cause mortality in patients with IHD [36,37]. More recently, a reduced LVEF has been demonstrated to be consistently the strongest predictor of SCD in both ischaemic and non-ischaemic cardiomyopathy.

In 14,609 post-MI patients enrolled in the VALIANT trial, depressed LVEF was the most powerful predictor of SCD [10]. In the first 30 days following MI each decrease in 5 percentage points in LVEF was associated with a 21 percent increase in the risk of sudden death or cardiac arrest with resuscitation. In a prospective study of 343 patients with idiopathic dilated cardiomyopathy, LVEF was the only significant predictor of arrhythmic events in multivariate analysis, with a relative risk of 2.3 per 10% decrease in ejection fraction [38].

As a result of this robust data a depressed LVEF has been the main entry criterion used in the randomised controlled clinical trials of primary prevention ICD therapy in heart failure [39,40].

Ambulatory Monitoring

A number of studies have suggested an association between the presence of non-sustained ventricular tachycardia (NSVT) on ambulatory monitoring and SCD in both ischaemic and non-ischaemic cardiomyopathy [41-43] However, although it is used as an important determinant in the latest UK NICE guidance on ICD use, more recent evidence has cast doubt on its predictive accuracy in the modern era [34,44].

Electrophysiological Studies (EPS)

Following the finding that post-MI patients with inducible ventricular arrhythmias had a significantly increased risk of SCD, EPS was for a long time considered the "gold standard" SCD risk stratification test in IHD patients [45-47]. However more recent studies have suggested that non-inducible patients are still at high risk of SCD, casting doubt on the prognostic value of EPS in IHD [48,49]. EPS has no significant prognostic role in non-ischaemic cardiomyopathy [50,51].

Microvolt T-wave Alternans (MTWA)

The electrocardiogram, or ECG, is a surface recording of the electrical activity of the heart. It records both ventricular depolarisation (the QRS complex) and repolarisation (the T-wave). Abnormalities in ventricular repolarisation, which are integral to arrhythmogenesis, are reflected in changes in the shape and size of the T-wave.

MTWA, which is a change in the size or shape of the T-wave on alternate beats, can be detected by complex computerised techniques. Multiple trials have demonstrated that MTWA testing is predictive of malignant arrhythmias. A meta-analysis of 19 studies, evaluating MTWA in 2608 patients over an average of 21 months follow-up, found a positive predictive value of 19.3% and negative predictive value of 97.2% [52]. There was no difference in predictive value between ischaemic and nonischaemic heart failure subgroups. However, patients with an indeterminate result were excluded from the analysis, and the high proportion of such patients (20-40%) is a significant limitation of MTWA. In addition there are currently a lack of prospective trials in which MTWA has been used to guide ICD use, and both of these issues will need addressing before MTWA is in routine clinical use [53].

Other Tests

In addition a number of other risk stratification tests have some predictive ability though they are not in widespread clinical use. These include tests of autonomic function, the signal-averaged ECG, and changes in the ECG QT segment [6].

Overall LVEF is consistently the strongest and most widely used predictor of SCD and the role of additional tests is currently unclear. Most contemporary guidelines suggest that heart failure patients with severely depressed LVEF (<30-35%) should be considered for an ICD without the need for additional testing, while patients with higher ejection fractions may benefit from further evaluation with additional risk stratification tests prior to ICD implantation [11,34].

The Limitations of Current Risk Stratification Systems

Despite their proven benefits and universal recommendation in national and international guidelines [11,34,54], uptake of ICDs has been variable, and the majority of patients who might benefit from a device for ‘primary prevention’ of SCD do not receive one [55-58]. The reasons for this under-use are likely multifactorial. Firstly, implanted ICDs are often unused. Four year follow-up in two large trials, MADIT-II and SCD-HeFT, which used contemporary risk stratification tools to direct device use, showed under 40% of patients with ICDs received appropriate anti-tachycardic therapy [39,40]. Secondly, serious device-associated complications such as inappropriate device therapy and infection, though uncommon in trials, are increasingly recognised in routine practice [55,59,60]. Thirdly, at an estimated cost of £20102 per device, ICDs are an expensive technology [61,62].

The development of more accurate risk stratification systems would enable better targeting of ICD use. This would ensure devices are used in patients most likely to benefit and avoided in those who are unlikely to benefit but may still have complications. There is therefore significant value in developing improved risk stratification systems using existing and/or novel markers of SCD.

Serum Biomarkers in Cardiac Disease

There has been a wealth of interest over the last decade in the use of biomarkers in cardiac disease. Many individual biomarkers have demonstrated associations with adverse cardiovascular outcomes, including C-reactive protein (CRP), interleukin-6, fibrinogen, d-dimer, albuminuruia, and plasminogen activator inhibitor type 1 [63-67]. Supported by systematic reviews confirming their value and consensus recommendations supporting their use, two specific serum markers, cardiac troponin (cTn) and brain natriuretic peptide (BNP), are now in widespread clinic use [68-71].

There is some evidence combining multiple cardiac biomarkers improves outcome prediction [72-74], though the magnitude of benefit is uncertain. For example, Wang et al studied 10 biomarkers, including CRP and BNP, in 3209 people in the Framingham Heart Study over 7 years and reported high "multimarker" scores increased the risks of death (hazard ratio 4.08) and major cardiovascular events (hazard ratio 1.84) [72]. However, they also noted that adding multimarker scores to conventional risk factors delivered only small increases in risk classification.

Serum Biomarkers in Heart Failure

Evidence of the value of serum biomarkers to predict SCD in heart failure comes from two types of study. Firstly, studies that have evaluated the relationship of biomarker levels to overall mortality or sudden cardiac death in heart failure. Secondly, studies that have evaluated biomarkers in patients with ICDs, using malignant ventricular arrhythmias as surrogate markers of SCD.

Serum Biomarkers to Predict Overall Mortality in Heart Failure

Heart failure is a clinical syndrome associated with complex molecular, endocrine and inflammatory changes [75]. The prognostic value of numerous serum biomarkers that reflect these underlying pathophysiological processes have been evaluated. Markers of neurohormonal activation, myocyte injury, myocardial stretch, and inflammation have all shown to be predictive of adverse outcomes [76].

Table 1. Studies evaluating the association of serum biomarkers with sudden cardiac death in patients with heart failure or left ventricular dysfunction.



No. of patients

Aetiology of heart disease



Berger et al. [83]




BNP, NT-BNP NT-ANP, big endothelin

All 4 biomarkers predictive of SCD in univariate analysis On multivariate analysis only BNP predictive

Tapanainen et al. [84]





All 3 biomarkers predictive of SCD in univariate analysis On multivariate analysis only ANP and BNP predictive

IHD, ischaemic heart disease; NICM, non-ischaemic cardiomyopathy; BNP, brain natriuretic peptide; NT-BNP, N-terminal brain natriuretic peptide; NT-ANP, N-terminal atrial natriuretic peptide; ANP, atrial natriuretic peptide.

Multiple studies have demonstrated that levels of serum inflammatory cytokines predict long-term heart failure mortality [77-79]. Rauchaus et al prospectively evaluated the predictive value of inflammatory cytokine levels in 152 patients with heart failure (121 patients in NYHA class II-III) [78]. During a mean 34 months follow-up there were 62 deaths. In univariate analyses tumour necrosis factor-alpha (TNF-a) and soluble TNF-receptors 1 and 2 (sTNF-R1/sTNFR2) (p<0.0001), interleukin-6 (p=0.005), and soluble CD14 receptors (p=0.0007) were all predictive of death. In multivariate analysis the strongest predictor was sTNF-R2 (p<0.001), which proved better than depressed LVEF. Serum cardiac troponin (cTn) is also an independent predictor of adverse outcomes, including mortality, in both stable and decompensated heart failure [80-82].

The majority of these studies were small and evaluated the relationship of biomarkers to overall mortality rather than SCD. However the commonest mode of death in all but the most advanced heart failure is sudden death [9]. Therefore it is probable that these biomarkers predict SCD as well as overall mortality. This is supported by data from the VEST trial [79].

Deswal et al analysed circulating levels of two inflammatory cytokines (TNF and IL-6) and their cognate receptors in 1200 patients enrolled in a multicentre placebo-controlled trial of Vesnarinone, an inotropic drug, in advanced heart failure [79]. All patients were NYHA class III-IV and the aetiology of heart failure in the majority was IHD (58%). In the placebo group (384 patients) there were 65 deaths, 31 each due to SCD and pump failure. Data from these 384 patients demonstrated serum levels of tumor necrosis factor (p=0.02), IL-6 (p=0.002), sTNF-R1 (p=0.0001), and sTNF-R2 (p=0.0001) were all independent predictors of overall mortality in multivariate analysis. Although the predictive relationship of biomarkers to SCD was not specifically evaluated, levels of TNF and IL-6 were not significantly different between the SCD and pump failure groups.

Next post:

Previous post: