Author + information
- Received October 1, 2005
- Revision received January 20, 2006
- Accepted January 25, 2006
- Published online August 1, 2006.
- Jay N. Cohn, MD⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Jay N. Cohn, Cardiovascular Division, Mayo Mail Code 508, University of Minnesota Medical School, 420 Delaware Street SE, Minneapolis, Minnesota 55455.
Mortality and morbid events are insensitive guides to the efficacy and safety of interventions in chronic cardiovascular disease (CVD). To enhance the ability to find new and effective long-term treatments, especially for the early stages of CVD, a revised strategy for clinical trials should emphasize efficacy on disease progression while monitoring symptoms and quality of life as guides to clinical benefit. Mortality, which is uncommon except in acute or advanced disease, provides at best a crude guide to net efficacy and safety. It must be monitored to support demonstrated efficacy on disease progression without adverse safety effects. This revised approach, made possible by our enhanced ability to monitor the progression of disease, should make it possible to study earlier disease and to improve cardiovascular health while reducing health care costs.
Clinical trials are designed to assess the efficacy and safety of interventions to steer management. In recent years, all-cause mortality has emerged as the primary end point in most cardiovascular morbid conditions because it is a noncontroversial outcome that appropriately is viewed as the most serious unwanted consequence of the disease. Indeed, the net effect on mortality has been used as a major factor in regulatory approval of a drug and in choosing one regimen over another in diseases such as heart failure (HF) and acute myocardial infarction (MI). In some trials, nonfatal morbid events, such as hospitalizations or MIs, have been added to deaths to enhance the power of the end point to detect a treatment effect.
This focus on mortality or defined morbid events as the end point for intervention trials is more common in cardiovascular diseases (CVDs) than in other clinical conditions. Trials in infectious diseases, arthritis, renal disease, gastrointestinal disorders, or anemias could never choose a mortality or morbid event end point. Even cancer therapy has usually focused on nonmortality end points. This preoccupation with mortality or morbid events reflects not only the recognition that CVD is the world’s major cause of death but also the perception that there are inadequate objective criteria for its progression and its response to therapy. This perception has justified unwieldy megatrials of painfully long duration to establish efficacy of therapy. It is time to re-evaluate this perception in the light of new insights into the mechanisms of the progression of CVD.
Prolongation of life is not the sole therapeutic goal of interventions in CVD. Symptoms, impaired quality of life, burden on family and friends, and utilization of expensive health care resources are perceived by many patients as of more importance in their daily lives (1). The goals of CVD treatment might be divided into 3 categories: 1) relieve symptoms that adversely affect quality of life; 2) slow disease progression that will ultimately lead to morbid events, a worsening quality of life, and greater emotional and financial burden; and 3) delay deaths and morbid events, some of which may be complications of disease progression and some of mechanistically distinct origin.
Therapy may favorably affect 1 or more of these therapeutic targets, but not necessarily all. Furthermore, some therapies may result in unwanted side effects that can adversely affect 1 or more of these targets. For example, beta-blockers may produce short-term aggravation of symptoms in HF but may slow progression of the disease and prolong life (2). Complications of surgery or angioplasty may cause short-term mortality but may lead to long-term benefit (3). Positive inotropic drugs may relieve symptoms but may increase the risk of arrhythmic deaths (4). Type I antiarrhythmics may relieve symptomatic palpitations but may do nothing for disease progression and may increase the risk of sudden death (5). Implanted cardiac-defibrillators can prevent sudden death but may induce new symptoms and not slow disease progression (6). In planning therapeutic strategies and therapeutic trials, it would seem appropriate to target specifically 1 or more of these goals rather than a global “outcome” that does not distinguish the mechanistic effect.
Prolonging survival must be a long-term goal of all interventions aimed at interfering with the progression of chronic CVD. Mortality reduction during the follow-up period of a time-limited clinical trial has served us well to show the remarkable efficacy of angiotensin-converting enzyme inhibitors (7) and beta-blockers (8) for HF, thrombolytic therapy for acute MI (9), and statins for atherosclerosis (10). However, mortality as an end point, powerful as it can be, suffers from several major deficiencies: 1) mortality reduction is neither a sensitive nor a specific guide to the efficacy of therapy on the disease process; 2) mortality cannot distinguish between efficacy and safety; 3) a mortality end point mandates the study of patients with advanced disease likely to die; and 4) a long follow-up period is usually required to achieve the number of events required for adequate power to detect a benefit. Each of these issues is examined in more detail.
Delaying death is certainly a long-term goal of treatment of chronic progressive CVD. But is mortality reduction a sensitive and specific guide to efficacy in a clinical trial? Although death is a noncontroversial event, the relationship of the event to the disease process being treated often is controversial. All-cause mortality has served as the primary end point in most large trials in CVD because attempts to distinguish cardiovascular causes of death are sometimes flawed. Did the driver’s accident stem from a heart attack? Would pneumonia have led to death in the absence of heart disease? Death or morbid events resulting from noncardiovascular causes can thus dilute the measurement of efficacy. Furthermore, some cardiovascular events may be unresponsive to targeted therapy. For example, a treatment for HF may not prevent MIs or strokes. Even if one could clearly define the deaths related to the disease process being treated, it is intuitive that death is a late manifestation of disease progression and that it impacts only a small proportion of the study population during the course of a trial. Therefore, at best, all-cause mortality and even cardiovascular morbidity are neither specific nor sensitive as a guide to the favorable effect of an intervention on a specific CVD process.
Efficacy versus safety
Mortality alone does not allow for a distinction between efficacy and safety. The importance of this distinction may be critical. The absence of efficacy on any of the three therapeutic goals renders a therapy useless, but efficacy counterbalanced by adverse safety effects could lead to specific measures to reduce risk. For example, adverse hypokalemic effects of diuretics can be counteracted by potassium supplementation, and consequences of arrhythmogenic effects could be neutralized by implanting a defibrillator. If safety concerns are confined to an identifiable subgroup, then such a group could be excluded from the therapy. These measures could therefore allow the efficacy effects of an intervention on one of the therapeutic goals to be expressed. Relying on overall mortality in a trial does not allow this distinction to become apparent.
Mortality and morbidity end point trials are usually event-driven so that the power of the study to detect a favorable effect of the therapy is dependent on achieving a prestudy estimated required number of events. This design mandates that the trial recruit patients with a high likelihood of experiencing a morbid event. The result, of course, is that entrance criteria often include markers for advanced disease, e.g., a very low left ventricular (LV) ejection fraction in HF (11), LV failure complicating acute MI (12), LV hypertrophy or diabetes complicating hypertension (13). A favorable effect of a therapy in this high-risk population is often advocated, however, for patients without these high-risk attributes. For example, although hypertension trials usually target uniquely high-risk individuals, the therapy is recommended for everyone with high blood pressure (14).
Devotion to a mortality end point precludes the study of early disease not likely to result in death during the limited follow-up period of all clinical trials. Indeed, the challenge to prevent advanced disease, which is the most effective way to reduce burgeoning health care costs, cannot be met if efficacy can only be assessed in trials powered for mortality reduction.
The creativity of the medical profession has led to a remarkable rate of development of new and potentially exciting therapies. Under the current guidelines, each of these experimental therapies must be subjected to large, long, and expensive trials to prove efficacy. Not only do these studies consume a large proportion of limited research funds, but they also subject patients to long-term treatment with experimental drugs or placebos and many deaths before efficacy or safety has been established. Does an earlier efficacy end point not dependent on counting bodies exist? If efficacy of an intervention on disease progression could be established early in a trial, if mortality and morbidity tracked with disease progression, and if the adverse event profile was acceptable, then the control therapy could be replaced with the effective therapy before mortality reduction had reached the magical p value associated with “significance.” Although more robust safety data are needed to identify infrequent but potentially serious adverse effects of an intervention, such low-incidence events are rarely divulged before marketing and should mandate intensive postmarketing surveillance.
Nonmortality end points
Chronic CVD is a progressive process. The premature deaths that complicate the disease are the result of its progression, often interrupted by sudden events such as arrhythmias or clots. Although the addition of morbidity end points (e.g., nonfatal MI or stroke, hospitalization for HF) to augment mortality as a guide to efficacy may improve sensitivity, the end result still is that the majority of patients entered into the trial do not contribute to the end point and the nonfatal end points require adjudication that can always be questioned.
Heart failure is a progressive disease in which structural changes in the LV advance (remodeling), neurohormonal stimulation increases, quality of life decreases, there is a need for hospitalization for worsening HF, and life expectancy shortens (15). Mortality may be the hardest of these end points, but it is hardly the most sensitive. Furthermore, sudden death may result from adverse effects of drug interventions (digitalis , milrinone , vesnarinone ) without necessarily impacting on some other potential efficacy parameter for the therapy. Any of the above markers that characterize advancing disease could serve as a guide to the slowing of disease progression, but the most mechanistic and powerful has been a marker for chronic structural remodeling of the LV, including chamber volume or dimension and ejection fraction measured at least 3 to 6 months after institution of the experimental therapy (18–22). Structural changes in the LV are usually accompanied by comparable changes in the plasma level of B-type natriuretic peptide (23).
How can a measure of disease progression be used in an overall assessment of efficacy and safety? It is first necessary to establish that the markers for disease progression are sensitive and specific mechanistic guides to adverse outcomes directly related to the disease under study. Considerable supportive data relating LV structural remodeling to outcome in HF already exist (18–22). Collection of confirmatory data in future clinical trials is of critical importance. Vascular and cardiac structural changes and urinary albumin in hypertension and endothelial dysfunction in atherosclerosis are candidate markers whose sensitivity and specificity as guides to risk of events and to therapeutic effects on disease progression need to be documented (24,25). These potential markers for disease progression are distinct from risk factors that are merely statistically related to disease events (26), and some previously used so-called surrogates, such as ventricular premature beats, which proved to be misleading (5). These cardiac and vascular structural/functional measurements are not surrogates for the disease, but they in fact appear to represent the CVD itself.
Once we are convinced that these mechanistic guides can track progression of disease, we should be in a position to administer therapy to achieve a clear end point that can be tracked in all patients. Relief of symptoms and improved quality of life are appropriate clinical targets, but slowing disease progression in an effort to delay morbid events is probably even more important. Efficacy and safety can be confirmed by tracking morbid events and deaths, which should trend with the data on disease progression. Rather than powering trials for highly significant reductions in events, we must become comfortable with powering trials, depending on the therapeutic goal, for highly significant benefits on softer clinical outcomes buttressed by evidence for slowed disease progression with efficacy and safety confirmed by favorable directional changes in morbidity and mortality. Composite end points that include adverse events as well as quality of life and even markers for disease progression may be a useful strategy (27,28). Then it should be possible to carry out smaller and more efficient therapeutic trials, and of utmost importance, trials at earlier stages of disease to preserve health rather than only to salvage those with advanced disease.
Our enhanced understanding of the progressive nature of CVD now provides us with the opportunity to track disease in addition to events. Cardiovascular medicine must now focus its attention on early detection and effective intervention to slow disease progression to delay costly and life-terminating events until our patients’ rewarding lives have ended. Such a strategy should both improve societal health and reduce health care costs. A change in trial design and regulatory philosophy is necessary to achieve this goal.
- Abbreviations and Acronyms
- cardiovascular disease
- heart failure
- left ventricle/ventricular
- myocardial infarction
- Received October 1, 2005.
- Revision received January 20, 2006.
- Accepted January 25, 2006.
- American College of Cardiology Foundation
- Chobanian A.V.,
- Bakris G.L.,
- Black H.R.,
- et al.,
- Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure,
- National Heart, Lung, and Blood Institute,
- National High Blood Pressure Education Program Coordinating Committee
- Wong M.,
- Staszewsky L.,
- Latini R.,
- et al.
- Cohn J.N.,
- Anand I.,
- Taylor A.,
- Tam S.W.
- Latini R.,
- Masson S.,
- Anand I.,
- et al.