Risk stratification for sudden cardiac death: show me the money!

This editorial refers to ‘Arrhythmic risk stratification in post-myocardial infarction patients with preserved ejection fraction: the PRESERVE EF study’†, by K.A. Gatzoulis et al., on page 2940. Despite great advances in the therapeutic realm regarding addressing the major public health problem represented by sudden cardiac death (SCD), there has been little substantive progress in the implementation of novel diagnostic approaches to optimize prevention and treatment. It has been well established that while low left ventricular ejection fraction (LVEF) is an important risk stratifier for SCD and is the most prominent tool used to guide implantation of an implantable cardioverter defibrillator (ICD),1 this discriminator lacks sensitivity and specificity.2 , 3 Specifically, there are many individuals who will suffer SCD, but do not meet the currently accepted levels of marked to severe depression of LVEF prior to the event to be considered for an ICD. In the Oregon Sudden Unexpected Death Study, a minority (30%) of victims of sudden cardiac death exhibited severe left ventricular dysfunction.4 Similarly, a community-based study in the Netherlands showed that only 35% of SCD cases had LVEF <40%.5 More recently, the POstmortem Systematic InvesTigation of Sudden Cardiac Death (POST SCD) study demonstrated that 57% of SCD victims had no prior history of cardiac disease.6 The burning question is: how do we identify these individuals in an efficient manner? That is, the less severe the underlying heart disease, the lower the risk for SCD, making it more difficult to identify the individuals among this population who will succumb to SCD. However, the absolute number of individuals with this ‘lower-risk’ form of heart disease is vast, much greater than those with more severe depression of LVEF, accounting for the overwhelming majority of SCDs arising from the lower-risk subgroups of the population. The Myerburg paradigm, originally reported in 1992, captures this concept.7 Two fundamental approaches can be conceived to address this question. Ideally, a ‘perfect’ diagnostic test would classify individuals with high precision and facilitate clinical decision-making among available treatment options. Much effort has been expended in developing such a test, but none of these tests has proved to be the ‘holy grail’ of risk stratification and none are in general clinical use. The next option employs a tiered approach. Interestingly, in 2001, Bailey et al.2 reported the theoretical implementation of a three-stage process of risk stratification: LVEF and signal-averaged electrocardiogram (ECG) results in stage 1, ventricular ectopic activity and heart rate variability in stage 2, and invasive programmed ventricular stimulation (PVS) to induce ventricular arrhythmias in stage 3. In both stage 1 and 2, patients were classified as either low risk (both tests negative) or high risk (both tests positive), or moved on to the next stage (one test positive). Based on existing data in the literature, they calculated the predictive value of each of these tests and then modelled the results of their three-stage risk stratification. In this theoretical exercise, 11.8% of patients were classified as high risk with a predicted 2 year major arrhythmic event (MAE) rate of 41.4%. It has taken 18 years for this concept to be validated in a clinical study. Gatzoulis et al.8 should be congratulated for validating this critical concept, which will surely provide the platform for extending risk stratification to the post-myocardial infarction (MI) population with minimal to moderately reduced LVEF. Gatzoulis et al.8 performed a multicentre, prospective, observational cohort study that examined post-MI ischaemia-free patients, with LVEF ≥ 40%. They had a two-stage risk-stratification process. In the first stage, they assessed a broad array of non-invasive risk factors (NIRFs) derived from the signal-averaged ECG and ambulatory 24 h monitoring. If at least one NIRF was identified, patients were then directed to PVS. A positive PVS led to ICD therapy. The high-risk group—positive NIRF and positive PVS—comprised ∼7–8% of the study population with an annual MAE rate of 8.2%. Though there are differences in implementation and actual outcomes, Gatzoulis et al.8 have validated the concept of staged evaluation of risk for SCD with NIRFs followed by PVS, as initially proposed by Bailey et al.2 It is important to note that the study population was a low-risk population with an overall MAE rate of 1.5% at a mean follow-up of 32 months. The demonstration that risk stratification for SCD can be accomplished in this challenging population should also embolden us to pursue other areas that are in critical need of optimal risk-stratification approaches for SCD. First and foremost, we need more-effective risk stratification in patients with LVEF ≤35%. The appropriate annual shock rate in current reports varies from 4 to 8%.9 , 10 Thus, many patients who receive an ICD never use it. This may be because there is a very low-risk patient subgroup within this overall high -risk population. Additionally, there is likely a subgroup with a high burden of competing risks. Risk stratification in patients who are currently indicated to receive an ICD may be helpful to identify patients at both ends of this risk spectrum who may not benefit from an ICD. The other major challenge regarding the prevention of SCD is that 37–57% of events represent the first manifestation of cardiac disease.4 , 5 A different approach is therefore required to identify these at-risk patients. Population-based risk scores have recently been reported with C-statistics upto 0.82 that could certainly be incorporated into a staged risk-stratification approach.11 Why has it taken 18 years to generate the clinical data in the PRESERVE EF study? This is a patient group that is relatively young (the mean age in this study was 57 years) with less competing risks for cardiac death than the patients with severe left ventricular dysfunction, in whom our efforts have mostly been focused. These patients deserve our dedication, commitment, and resources to figure out how to prevent/treat their SCD events. While there are many potential explanations for our collective failure to address this in a timely fashion, we will highlight one: the lack of adequate research funding, particularly for studies focusing on diagnostic algorithms and strategies for risk stratification for SCD. The National Academies of Sciences, Engineering, and Medicine published a report on cardiac arrest in 201512 in which they estimated the total amount of National Institutes of Health (NIH) support for cardiac arrest research as ∼$107.7 million, of which $67.4 million was for cardiac death and resuscitation sciences. Using NIH data13 and an estimated number of deaths attributable to cardiac arrest of 503 000,14 Figure 1 demonstrates that there is markedly low investment in this field based on dollars spent per death attributable to the indicated disease categories. When considering funding for research in this field, multiple funding sources could, in theory, be accessed. Both government and industry funding have been critical in establishing the efficacy of ICD therapy in the high-risk groups in which it is currently indicated.1 Yet, the establishment of efficacy does not imply that optimal criteria have been identified. Although better refinement of criteria can produce more cost-effective care at lower cost, investment in these types of studies has been limited because they are costly and the return on investment, while likely to be positive, would not accrue to the government or industry sponsors of this research, creating a misalignment of interests. Randomized control studies of risk-stratification strategies and treatment in the lower-risk population would also be costly because of the large population size needed in studies of lower-risk subgroups. Furthermore, they are high-risk ventures that may not yield positive results. With limited funding, it is no surprise that funds are expended on other less-risky and less-costly research. It has been proposed that there may be other potential funding sources whose financial interests align more directly with the goals of prediction and prevention initiatives, such as the insurance industry.15 There are tremendous costs associated with SCD and its subsequent care. Developing the science of risk stratification for SCD could provide more cost-effective care that benefits both the public and the insurance companies. While there is a real opportunity to align both the costs of research in risk stratification and the return on investment, this is a real challenge in our current funding model where the organizations responsible for these two aspects are disparate. Ultimately, the medical community and the general public need to develop a public health mandate to align the costs of research with the potential financial gain. The annual public research funding per death for each of the indicated disease categories, based on data and analysis by the National Academies of Sciences, Engineering, and Medicine12 and the National Institutes of Health.13 Despite the landmark status so deserving of this study, the authors highlight some limitations. We and other readers might also find other ‘limitations’. Yet, in the context of a poor research pipeline in this area of risk stratification, we propose that these ‘limitations’ be considered as challenges and opportunities to address important and clinically relevant issues. Clearly, there is opportunity for further refinement of the NIRFs used. Other testing, such as late gadolinium enhancement on magnetic resonance imaging could also substantially improve discrimination.16 The structure of the stages, number of stages, and their order also needs to be optimized. The optimal set of tests needs to be established. While the findings from this observational study are tantalizing, they do need to be assessed in a randomized clinical trial. It is clear that ICD shocks are not a surrogate for SCD and whether the extended time programming in this study has improved the relationship of shocks to SCD needs to be established. The risk assessment approach of this study will also need to be validated in other countries to examine how regional differences in risk factor prevalence affects its performance. For instance, the population studied had a 50% smoking rate. In contrast, in the past year, the number of smokers in the post-MI population in the US has been in decline, approaching 32%.17 Gatzoulis et al.8 have conducted a breakthrough study that should catalyse the field of risk stratification. SCD is the most common mode of cardiac death and still represents the number one cause of death in the developed world.18 If our public health mandate is to reduce the burden of SCD, then substantial investment is required to support co-ordinated approaches to apply the techniques that are already available to us to generate the data needed to successfully implement risk-stratification strategies. PRESERVE EF8 has provided the giant first step. Conflict of interest: Dr Goldberger is Director of the Path to Improved Risk Stratification, NFP, a not-for-profit think tank that has received unrestricted educational grants from Biotronik, Boston Scientific, GE Medical, Gilead, Medtronic, and St. Jude Medical. The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology.