Risk Stratification for Sudden Death in 2008

Consider the following two cases: a 45-year-old woman is at work speaking to colleagues when suddenly, without warning, she collapses, unresponsive. Cardiopulmonary resuscitation (CPR) is begun and 911 is called. Ventricular fibrillation (VF) is documented, and she is promptly defibrillated and resuscitated. Initial electrocardiograms (ECGs), upon hospitalization, show no evidence of acute myocardial infarction or other abnormalities. Biomarkers show no evidence of acute infarction. Her left ventricular ejection fraction (EF) is normal shortly after resuscitation. Coronary angiography reveals no obstructive coronary disease. Her past medical history is negative.

A 64-year-old man was hospitalized 2 months ago with an acute myocardial infarction complicated by heart and then renal failure. He is currently making a slow recovery and reports exertional dyspnea when walking one block. His left ventricular EF is 30%. The ECG shows sinus rhythm with right bundle-branch block.

These two cases illustrate the spectrum of the problem of risk stratification for sudden cardiac death. The first case, a young woman, previously asymptomatic, has survived a cardiac arrest. Although perhaps surprising, sudden cardiac arrest is the initial manifestation of heart disease in 50% of cases.[1] Thus, fully half the time, we may not have any opportunity for risk stratification prior to cardiac arrest or sudden death by current standards. The second case represents a common problem faced by cardiologists daily. Some patients with advanced ventricular dysfunction, on the basis of myocardial infarction, may be at considerable risk for cardiac arrest, even though they have not demonstrated spontaneous ventricular arrhythmias. How should we go about determining which patients can benefit from prophylactic therapy with implantable defibrillators?

The scientific statement on noninvasive risk-stratification techniques is a comprehensive summary of the current status of noninvasive tests used to quantify risk for sudden cardiac arrest. Unfortunately, a recurring theme that appears throughout this manuscript is the fact that, for most of the techniques reviewed, there are either insufficient data to recommend routine use at this time, or the existing data suggests lack of clinical utility. We are left with the message that current standards for risk stratification lag behind our therapeutic technology, the implantable cardioverter-defibrillator (ICD), for ventricular arrhythmias. As a result, we have superb technology (the ICD), but we do not know how to use it optimally for primary prevention of sudden cardiac arrest.

What is missing from the scientific statement is an organized guideline of which tests to use, in what order, on which patients, and at what time in the course of disease. There is good reason for this omission. I am aware of only one risk-stratification protocol that has been tested in a randomized controlled trial for its ability to reduce sudden death, and that trial evaluated an invasive test: the electrophysiologic study (EPS).[2] The hundreds of references quoted in the scientific statement do not include any trials wherein a specific noninvasive risk-stratification technique or protocol is used to determine which patients receive an ICD and to assess the effect of this strategy on incidence of sudden death. This trial design, whereby a risk-stratification technique is used to determine which patients receive antiarrhythmic (ICD) therapy and compared to a control group that receives therapy guided by an alternative technique, is necessary to evaluate the clinical utility of any risk-stratification strategy. This type of trial design will be necessary in order to alter practice. Note that this is not the same as a trial design that uses a particular test (such as measurement of EF) to qualify patients for entry into a trial and then randomizes patients to receive alternate therapies such as an ICD or conventional medical therapy.[3] The latter design can evaluate efficacy of a therapy in a particular patient population, but it cannot evaluate the utility of risk-stratification tests. The latter type of trials are more difficult to design and execute, but these trials are what we now need in order to learn how best to use ICDs for primary prevention of sudden death.

I am struck by reading this summary statement that the best test available today in patients with coronary disease or nonischemic dilated cardiomyopathy is EF. Yet, the authors of the scientific statement clearly make the point that EF bears no direct relation to the mechanisms of ventricular arrhythmias responsible for the majority of episodes of cardiac arrest. Thus, the utility of EF is derived from its ability to predict total mortality, of which roughly one-half is likely due to sudden cardiac arrest initiated by ventricular tachycardia (VT) or VF. A paradox in using EF to define high and low risk is the observation that sudden unexpected cardiac arrest accounts for a smaller proportion of deaths in patients with lowest EF, and most advanced heart failure, than in patients with less severe ventricular dysfunction.[4] Another consequence of the use of low EF to guide ICD use, well stated by the authors of the scientific statement, is the fact that at least half of the episodes of cardiac arrest occur in patients that do not meet the current definition of high-risk EF (≤30%-35%).[5,6] Thus, restricting use of ICDs to only those patients with very low EF misdirects use of this therapy. How can it be that most sudden deaths occur in patients at "low risk"? The answer to this question is that use of EF to stratify risk for sudden cardiac arrest is inappropriate. The patients that experience sudden death whose EF is greater than 35%-40% are certainly not low risk! They merely had the wrong test used to stratify risk. This point is exemplified by two recent observational studies of patients with recent myocardial infarction (MI) and relatively preserved EF. The first study examined the use of T-wave alternans (TWA) in patients with EF ≥40% after recent MI. Patients with abnormal TWA tests had a 10% incidence of arrhythmic events (sudden death or resuscitated cardiac arrest) at 2 years follow-up.[7] The second study examined the use of baroreflex sensitivity (BRS) testing in patients with EF greater than 35% after recent MI.[8] The incidence of cardiovascular death at 2 years follow-up was approximately 15% in patients with an abnormal response to BRS testing. I don't believe anyone would consider patient populations with 10%-15% 2-year rates of sudden death as low risk.

What is the take home message from this well-constructed scientific statement? First, I believe we need to establish standards for the conduct and analysis of studies that examine techniques of risk stratification for sudden death. These standards should include the following: 1) Analysis of events by committees of experienced clinician investigators blinded to test results to enable classification of deaths as sudden or nonsudden. The studies should compare the ratio of sudden to nonsudden deaths in patients having positive versus negative results of the risk-stratification test in question. Many argue that the only appropriate endpoint is total mortality, because there is too much subjectivity in judging arrhythmic deaths. However, newer monitoring techniques with implanted devices permit validation of clinical observations. If we pay attention only to total mortality, we cannot possibly use therapies that affect only arrhythmic death in a cost-effective manner. Furthermore, if ICDs are used in trials, mortality reduction in ICD-treated patients is likely to principally reflect affects on arrhythmic events. 2) Test results should be analyzed at standard durations of follow-up to permit comparison. Determining the appropriate length of a study is not easy, and represents a trade off between too short a follow-up (which does not allow enough time for events to occur in at-risk patients) versus inappropriately long follow-up (eventually all survival curves converge). Because of disease progression, results of risk-stratification tests are likely to remain valid for limited time periods. The appropriate duration of studies evaluating risk-stratification techniques probably should be less than the duration of studies evaluating efficacy of therapy, such as the ICD. 3) Study populations must be clearly defined, and study design must ensure that, as far as possible, the study populations accurately reflect the entry criteria. This criterion is critical if we are to be certain that the results will be valid when applied to the general population of patients that fit the study entry criteria.

The final lesson this scientific statement conveys is the need, at this time, for prospective controlled trials designed to critically assess strategies to guide deployment of ICDs for primary prevention of sudden cardiac arrest. Observational studies, such as those described in the scientific statement, are but the initial step in reducing occurrence of sudden death. Practice guidelines will not change until controlled trials test the efficiency and efficacy of risk-stratification protocols.