There is probably no more exciting time to be both an echocardiographer and an epidemiologist. Echocardiographic left ventricular (LV) mass (LVM) assessment occupies a unique place in cardiovascular risk assessment; increased LVM is both a consequence of chronic risk exposure (hypertension and obesity) and a potent predictor, independent of those risk exposures, of cardiovascular outcomes. Since the initial observation in the Framingham study that LVM predicted cardiovascular outcomes, population-based research has routinely incorporated echocardiography into regular examinations. Fruits of this research include the inclusion of LV geometry (the ratio of ventricular wall thickness to cavity dimension) into risk algorithms, the extension of observations on heart size to include assessment of left atrial size as a predictor of outcomes, the design of clinical trials of antihypertensive treatment to test if regression of LVM improves outcomes independent of blood pressure lowering, and the incorporation of LVM into hypertension treatment algorithms in children. Recently in the Coronary Artery Risk Development in Young Adults (CARDIA) study, various measures of LV function performed at 23 to 35 years of age have been shown to predict incident heart failure in African Americans over 20 years of observation. The possibility of linking LVM to newer measures of LV function such as speckle tracking looms on the horizon.

Accurate measurement of LVM has been an important part of LVM research since the inception of the field. Here the critical first observation was establishing the relationship of LVM measured echocardiographically to actual heart weight; these studies comparing LVM measurements in living individuals with those in hearts obtained at autopsy later in the course of a patient’s illness allowed the development of the current formula used in American Society of Echocardiography guidelines. Epidemiologic studies suggest that this formula has a coefficient of variation in individuals of about 10%, on the basis of quality control procedures used in the large population-based studies.

Because LVM is strongly related to body size, a natural question arises as to the proper interpretation of raw LVM: does mass need to be indexed to body size for proper interpretation? What cutoff values can be considered to indicate hypertrophy? This issue is further complicated by the fact that not all increases in LVM are pathologic; for example, slower resting heart rate and exercise training both are associated with higher LVM. Conversely, is the increase in LVM related to obesity and hypertension “adaptive” or pathologic? If it is pathologic, indexing may mask critical risk information; that is, finding the best biologic fit of LVM to body size may obscure pathologic hypertrophy. To provide an overly simplistic example, high-density lipoprotein cholesterol, triglycerides, and blood pressure are not “adjusted” for obesity. This indexing question has led to a substantive (and statistically complex) literature arguing for various methods of indexing. Issues considered in the debate include identifying indexing tools that most accurately reflect appropriate LVM for lean body mass; strategies that account for age, gender, and development; and control for biases that statistical assumptions may provide in analyses.

More recently, and more appropriately, the indexing question has been linked to outcomes; that is, what method of indexing LVM is best associated with cardiovascular morbidity and mortality? Whereas earlier literature tended to report outcomes related to only one method of indexing, efforts are now being made to compare different indexing methods and their associations with cardiovascular endpoints. In this issue of JASE , Ristow et al use a cohort with defined cardiovascular morbidity and LVM assessed at study entry to determine whether a specific method of indexing LVM better identifies adverse cardiovascular outcomes. In fact, all indexing methods performed relatively well. In a secondary analysis, the authors also show that there was a very high correlation among the different indexing methods, suggesting that the individuals identified as at high risk were roughly the same, regardless of indexing method.

The history of attempts to define abnormal levels of cardiovascular risk factors supports the concept that the definition of LV hypertrophy should be linked to outcomes. Relying on distributions of LVM in the population and establishing arbitrary thresholds for the definition may not be appropriate. The definitions of elevated cholesterol and blood pressure have drifted downward as recognition that limiting cardiovascular disease prevention to those who can be defined as at very high risk will provide aggressive treatment for only a few individuals, whereas most events occur in the general population with modest increases in risk factors. This is most clear in secondary prevention, in which any reduction in established risk factors (unless already quite low) has benefit. Thus, the paper of Ristow et al represents a further step in identifying thresholds of LVM that are associated with adverse outcomes. For indexing by height ^2.7 , a level of 51 g/m ^2.7 has been established. A limitation of this indexing method and some others such as body surface area is the complexity of the indexing calculation. The difficulty of calculating the indexed value limits use of the measure.

An important limitation of Ristow et al’s paper is that the index patients in this study all had previously identified coronary artery disease. Thus, the conclusions may not be generalizable to primary prevention populations. Data from large population-based studies, such as CARDIA, the Multi-Ethnic Study of Atherosclerosis (MESA), and Atherosclerosis Risk in Communities (ARIC), will be needed before a firm definition of indexed LV hypertrophy can be established that can be used for the purposes of both secondary and primary prevention. In the future, it will be important for investigators from these large studies to analyze data using several indexing strategies, rather than a single strategy, so that the optimal indexing method can be identified. These studies should have sufficient statistical power to discriminate clinically meaningful differences among the various indexing strategies. A further goal of such research should be determining if the relationship of LVM to outcomes is continuous and graded or if there is a threshold above which LVM is harmful. It is probably overly simplistic to assume that the lower the LVM, the better. Inadequate LVM may have adverse consequences as well. For example, cardiomyopathy secondary to cancer chemotherapy is secondary to cardiomyocyte loss. Healthy cardiac adaptation, such as occurs with fitness training, increases LVM.

Interestingly, in Ristow et al’s study, LVM in the absence of indexing was also highly predictive of adverse outcomes. From a clinical standpoint, if one is following LVM serially, indexing may be less important. If LVM increases without change in height, this is probably adverse, while any decrease in raw LVM is likely to be beneficial.

An emerging consideration in the assessment of LVM is the idea that the heart is a continuously developing organ. Historically, cardiac development was considered to take place in the first 12 weeks or so of gestation, and thereafter its growth was determined largely by the physiologic strains incurred in fetal life, childhood, and adulthood. Research into the apoptosis of cardiac cells, recognition that cardiac function changes with aging and that the heart becomes more fibrosed with aging, and, most interestingly, the documentation of significant changes in heart size and geometry with aging between 50 and 90 years all suggest the heart is a dynamic organ. Another recent interesting observation is that leukocyte telomere length, a marker for aging whereby telomere length shortens with age, was positively associated with LVM; because both shorter telomere length and increased LVM are associated with cardiovascular events, an inverse relationship was expected. It is likely that cardiovascular factors such as hypertension, obesity, tobacco use, sedentary lifestyle, and diabetes affect heart function chronically, perhaps accelerating this aging process. Just as atherosclerosis is considered a chronic illness evolving over decades until an acute event occurs, myocardial dysfunction may be the consequence not only of acute events, such as myocardial infarction, but chronic risk exposure as well. Echocardiography provides a window onto this process, and measurements of not only LVM but other parameters, such as LV geometry, left atrial size and function, and speckle tracking, may provide useful ways of assessing chronic myocardial injury, the benefits of a healthy lifestyle, and success of interventions. Appropriate indexing of heart size to body size will be an important consideration in these analyses, and one can theorize that different methods of indexing may be more appropriate at different stages of life.

Thus, a new role for echocardiography in clinical practice may be emerging, as an important tool in cardiovascular risk assessment. Epidemiologic research over the past 20 years has provided the groundwork for this role. LVM is already incorporated into pediatric guidelines for hypertension management, in which the presence of LV hypertrophy is considered as a rationale for intensification of treatment. As we learn more about the relationship of LVM to outcomes and about age-dependent changes in cardiac function, echocardiography may become much more than a tool to provide an acute assessment of cardiac function; it may also serve as a window to effective cardiovascular prevention.