Not so very long ago, it was probably easier to win the lottery than qualify to get an implantable cardioverter-defibrillator (ICD). ICDs were considered extreme measures for the most desperate cases. As such, patients had to have survived not one but two documented episodes of sudden cardiac arrest, be drug refractory, but also be strong enough to be able to undergo a thoracotomy to implant the device. Somehow, some patients qualified and received devices!
However, the life-saving potential of ICDs was apparent to many physicians. The ICD world changed when a few committed physicians got together to develop a suite of clinical studies that would explore the implantation of this radical device into arrhythmia-rich populations. The most famous of these studies turned out to be the Multicenter Automatic Defibrillator Implantation Trial (MADIT) and MADIT II.
Both MADIT and MADIT II looked at heart attack survivors with impaired left-ventricular function. In the first MADIT trial (published in 1996), patients had to be inducible in an EP study. The study was based on the notion that potentially dangerous ventricular arrhythmias occurred more frequently in post myocardial infarction (MI) patients. A positive EP study was utilized to help further confirm the fact that the patient possessed the necessary arrhythmic substrates to sustain a ventricular arrhythmia. The ICD turned out to reduce all-cause mortality by 54% in the ICD group when compared to the control group.
But what if these same patients had not undergone the EP study? MADIT II (published in 2002) dispensed with the EP study and found ICDs still reduced the risk of all-cause mortality by 31% versus the control group (Figure 1).

While it is very tempting to view MADIT and MADIT II as two versions of the same study, they were actually set up using two different statistical designs. Thus, it is not really accurate to make head-to-head comparisons of the studies. Both showed a significant mortality benefit for ICDs. Eliminating a positive EP study was also a major step; as we know, EP studies can be difficult procedures for some patients, particularly highly symptomatic, debilitated heart failure patients.
MADIT and MADIT II prompted a new indication for ICDs and created a new word in our vocabulary: the primary prevention patient. Primary prevention patients were those who would get an ICD to prevent the first instance of sudden cardiac arrest.
This marked a huge departure from our previous appreciation of ICDs. Now an ICD could be prescribed for a prior-MI patient without documented arrhythmias and without an EP study, even if the patient was responded well to drugs. (Figure 2)

But while the MADIT studies caused physicians to re-think how to manage MI survivors, other investigators were wondering about patients who had not had a heart attack. Would ICDs confer a mortality benefit on nonischemic patients at high risk for arrhythmic activity? That question was answered by the the Defibrillators in Non-Ischemic Cardiomyopathy Treatment Evaluation (DEFINITE) study, which enrolled patients with nonischemic dilated cardiomyopathy and a LVEF < 35% with documented evidence of arrhythmic activity. The DEFINITE patients did not have to undergo an EP study. Instead, there had to be evidence of at least 10 PVCs an hour or runs of nonsustained ventricular tachycardia (3-15 beats above the rate of 120 bpm) on Holter monitoring.
DEFINITE patients were randomized into an ICD and a non-ICD group. The results, published in 2003, showed that ICDs reduced all-cause mortality even in nonischemic patients.
This raised another question in the cascading series of questions about who might benefit from device-based treatment. Would there be benefit from an ICD in patients who had poor left-ventricular function, even if they never had a heart attack, had no documented history of arrhythmias, and did not undergo EP testing?
Clearly, compromised left-ventricular function was emerging as an important factor in identifying potential primary-prevention patients. MADIT required patients to have a left ventricular ejection fraction of 35% or lower, while MADIT II lowered that to 30% or below. DEFINITE made a LVEF < 35% an inclusion criterion. However, MADIT and MADIT II had no inclusion criteria about NYHA Heart Failure Classification. (The articles about these studies sometimes provide NYHA Classification data on the patient population of MADIT and MADIT II, but technically, there were no inclusion criteria for NYHA Class.) Subsequent studies would often use NYHA class to help define patient populations.
Arrhythmias are not uncommon in heart failure patients, and it has been observed that Class II patients (more so than the sicker Class IV patients) are at higher risk of sudden cardiac death. It s important to clarify that patients in Class III or IV heart failure are at greater risk of overall mortality, in particular, what might be called non-sudden cardiac death. However, sudden cardiac arrest is more likely to strike healthier heart failure patients.
Managing arrhythmias in heart failure patients has always been challenging. Anti-arrhythmic drugs are contraindicated in heart failure patients, with the exception of amiodarone. The largest randomized ICD trial in history (2,520 patients) compared ICDs to amiodarone in a population of Class II and III heart failure patients with ischemic (52%) and non-ischemic (48%) dilated cardiomyopathy. There were no inclusion criteria requiring documented arrhythmias, inducibility, or prior MI. These were simply heart failure patients with dilated cardiomyopathy and poor LVEF scores.
The Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT), published in 2004, randomized patients into three arms: those who got an ICD, those who got amiodarone, and those who got neither an ICD nor amiodarone. All patients were given state-of-the-art pharmacological treatment for heart failure.
SCD-HeFT was set up so that each study site had two investigators: one electrophysiologist and one heart failure specialist. This model of electrophysiology and heart failure expertise (devices and drugs) is exemplary and reflects our current understanding of heart failure as a disease that has both a neurohormonal component (managed by drug therapy) and an electrical component (managed electrically by devices).
SCD-HeFT found that ICDs significantly reduced all-cause mortality in Class II or III heart failure patients with a LVEF ? 35% by 23% compared to placebo. Amiodarone did not offer any significant mortality benefits versus the placebo. (Figure 3)

In a subgroup analysis of SCD-HeFT, it was found that benefits were most pronounced in Class II patients, which is what most clinicians thought intuitively. What surprised many observers was that amiodarone did not show a clear and significant mortality benefit. That does not mean that amiodarone has no role in managing arrhythmias for heart failure patients amiodarone suppresses ambient arrhythmias. For patients who experience lots of arrhythmic activity, amiodarone likely suppresses some of their ventricular tachyarrhythmias as well. But amiodarone does not suppress every arrhythmia, and this is where the rescue function of an ICD comes in. SCD-HeFT demonstrated how significant this rescue function can be.
For patients who experience lots of arrhythmic events, the use of amiodarone plus an ICD may be a good course of action. Amiodarone will suppress many arrhythmias, sparing the patient many shocks. Most patients experience shocks as being unpleasant, to say the least, so minimizing their occurrence can vastly improve a patient s quality of life. However, since amiodarone does not suppress 100% of arrhythmias all of the time, the ICD can rescue the patient if a malignant arrhythmia emerges.
SCD-HeFT expanded draft coverage for ICDs in the U.S. to include patients with documented congestive heart failure (Class II or III) for more than three months and a LVEF less than 35%. There is no requirement that the patient have had a prior MI, an EP study, coronary artery disease, or a long QRS duration. Patients no longer had to be drug-refractory. In fact, today we recognize that it is never a matter of drugs or devices, but drugs first and then devices added for certain patients.
Long QRS duration (typically anything over 120 milliseconds) sometimes is mentioned when talking about expanding device indications, but this typically refers to the use of CRT-D systems. A wide QRS is considered a good (but not perfect) marker for ventricular dyssynchrony. For heart failure patients with a long QRS, a cardiac resynchronization therapy (CRT) system with defibrillation capability marries the benefits of cardiac resynchronization with the rescue function of defibrillation.
The Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure (COMPANION) study was the first large, randomized clinical trial to look at mortality for both CRT and CRT-D devices (CRT systems were stimulation devices, while CRT-D devices offered defibrillation as well as CRT). COMPANION was the first large-scale randomized clinical trial of CRT systems.
COMPANION was a large (1,520) study that included Class III and IV heart failure patients the most symptomatic. Inclusion criteria required patients to also have a QRS duration of 120 ms or greater on ECG, a PR interval of over 150 ms, and a LVEF ? 35%. COMPANION determined there was a significant 20% risk reduction in its combined endpoint (all-cause mortality or first all-cause hospitalization) for the CRT or CRT-D patients compared to drug therapy alone. For patients in the CRT-D group, there was a 36% risk reduction in all-cause mortality (secondary endpoint). While COMPANION was the first trial to show a mortality benefit for CRT and CRT-D systems, this benefit was not-quite significant (p = 0.059). Furthermore, COMPANION was not designed to compare CRT and CRT-D function, so the study does not show whether CRT has mortality benefits apart from defibrillation.
The recent Cardiac Resynchronization in Heart Failure (CARE-HF) study (published in 2005) evaluated only low-voltage CRT devices (no defibrillation capability) in Class III and IV heart failure patients with compromised LV function (LVEF < 35%) and long QRS (> 120 ms with some other criteria applicable to those with a QRS between 120 and 150 ms).
The CARE-HF study showed a 37% risk reduction for the primary endpoint (all-cause mortality or first hospitalization for a cardiovascular cause). CRT patients showed a 36% reduction in all-cause mortality alone (secondary endpoint) compared to patients who only received drug therapy. CARE-HF is the first trial to show morbidity and mortality benefits for CRT apart from defibrillation. Although more study is warranted, it appears that CRT works by retarding or even undoing some of the cardiac dysfunction that heart failure causes. Resynchronization therapy is associated with reverse remodeling, reduced symptoms, and now it appears to reduce the occurrence of lethal arrhythmias. (Figure 4)

While CARE-HF evaluated only low-voltage CRT devices without defibrillation, it makes sense to believe that adding the rescue benefit of defibrillation in a CRT-D system would be even more beneficial. In fact, many clinicians are already selecting devices for their patients with LV dysfunction based on the notion that CRT restores cardiac function and defibrillation can rescue even if a dangerous arrhythmia breaks through.
Clearly, things have changed a lot in the past 15 years in terms of understanding who can benefit from device-based treatments. We can anticipate that our understanding of the electrical management of heart disease will change as well.
One of the big unanswered questions for rhythm management devices involves risk stratification. The first MADIT study used a positive EP study as a risk stratifier. That was effective, but EP studies are costly and invasive procedures that can be very hard on debilitated patients. It makes sense for some people to have EP studies, but it would waste resources (and tax patients) to give EP studies to large groups of primary-prevention patients.
While LV dysfunction appears to be an important stratifier, it is clear that not all patients with impaired systolic function benefit from devices. LV dysfunction seems to be part of the puzzle, but not the whole picture. For instance, some patients with very severe LV dysfunction (Class IV) are at lower risk of arrhythmic death than patients with more moderate LV dysfunction (Class II).
Some clinicians use a long QRS duration as a good marker for ventricular dyssynchrony, but we know that some patients with ventricular dyssynchrony may not exhibit this particular marker. A long QRS duration is not a very elegant marker, but it has a lot of utility, in particular, because it is easy and inexpensive measurement to obtain. The problem arises in that some patients with narrow QRS complexes may also have ventricular dyssynchrony.
In short, we have learned and continue to learn more and more about device-based therapy for heart disease, but we still don t have the total picture. We know many of the factors that come into play: a low LVEF score, a prior MI, and long QRS. We have looked at other things too: signal-averaged ECGs, T-wave alternans, positive EP studies, documented history of arrhythmias. Personally, I believe that we are close to finding the formula of the simple, inexpensive, easy way to stratify primary-prevention patients.
The very notion of primary prevention would have sounded strange to the doctors, nurses, and technicians who worked on the first generation of ICDs. Today, more and more people will get devices in anticipation of a first event. As we better learn to stratify the primary-prevention population, we can extend the life-saving properties of device-based therapy to those for whom it is literally a matter of life and death.