Becoming obese is an anabolic event. Beside the obvious accumulation of excessive body fat, the obese child is characterized by an increase in lean body mass, acceleration of linear growth, enhanced skeletal maturation, and advanced sexual development (Forbes, 1977). An expanded circulatory system reflects this somatic growth, with increased plasma volume, hypertrophy of myocardial fibers, and cardiac chamber enlargement.

Studies in adults
Early autopsy studies of morbidly obese adults, many suffering from congestive heart failure, indicated a marked increase in heart weight, with biventricular enlargement and eccentric wall hypertrophy. Subsequent investigations utilizing echocardiography revealed that these findings 1) could be observed independent of coronary artery disease and systemic hypertension, and 2) were not simply confined to those with marked obesity but instead were evident across the range of overweight individuals (see Alpert and Alexander, 1998, for review). More specifically, measurements of left ventricular mass, chamber dimensions, and wall thickness have been observed to relate directly to both severity and duration of obesity. For example, among a group of adults with morbid obesity, Alpert et al. found an increase in average left ventricular end diastolic dimension from 4.8 cm in those with obesity of 5 years' duration to 6.5 cm after 20 years (Alpert et al., 1995). Subsequent information from magnetic resonance imaging (MRI) studies has substantiated these findings (Danias et al., 2003).

Further confirming this anabolic effect of obesity, heart mass and chamber size are observed to diminish following therapeutic weight reduction. Decreases in left ventricular mass, wall thickness, and chamber diastolic dimension have been reported not only in the morbidly obese following bariatric surgery (Alpert et al., 1994) but also after dietary weight reduction programs for those who are initially mild to moderately overweight (MacMahon et al., 1986).

Studies in youth
Echocardiographic studies, supplemented by recent MRI investigations, have consistently indicated similar anatomic features in obese children and adolescents. Larger, thicker hearts are seen in obese subjects compared to non-obese youth (Friberg et al., 2004; Koehler et al., 1997; Mehta et al., 2004; Rabbia et al., 2003). In these studies, those with mild-moderate obesity typically demonstrate approximately 15-20% greater values of cardiac mass (related to body height or surface area) than lean youth. Although comparisons are treacherous, the magnitude of these differences is somewhat less than the 20-40% described in studies of obese adults (presumably reflecting differences in duration of obesity) (Alpert and Alexander, 1998).

Positive cross-sectional associations between body fat content and left ventricular mass, wall thickness, and chamber dimensions have been evident among cohorts of obese subjects (Humphries et al., 2002) as well as in studies in the general pediatric population (Chinali et al., 2006; Gutin et al., 1998; Kono et al., 1994; Mensah et al., 1999; Paparassiliou et al., 1996; Pfleiger et al., 1994; Yoshinga et al., 1995). As in adults, these reports describe a positive relationship between severity of obesity in youth and left ventricular size. Kono et al., 1994 reported a correlation coefficient of r=0.60 between adiposity and left ventricular mass normalized for height in 6-year old males, and Rowland and Dunbar, 2007 found a correlation of r = 0.59 between body mass index and left ventricular end diastolic dimension in a group of early adolescent females with a BMI range of 14-63 kg·m^-2. The average dimension was 44 mm in those with a BMI of approximately 20 kg·m^-2 compared to 52 mm with a BMI of 42 kg·m^-2. In the only study which has assessed cardiac features relative to duration of obesity, Rabbia et al., 2003 found that a group of 13 year old obese children had a greater average heart mass than nonobese. However, they could find no significant difference in left ventricular mass among three groups of the obese children with <4, 4-7, and >7 years duration.

Longitudinal data are limited. In the Bogalusa Heart Study, left ventricular mass (by echocardiography) and body composition measures were recorded four years apart in 67 healthy children initially 9-22 years old (Urbina et al., 1995). Stepwise multivariate regression analysis indicated that initial skinfold thickness predicted final left ventricular mass (relative to body height^2.7). However, change in skinfold thickness was not related to change in cardiac mass.

Relationship to body composition
From this information it is clear that the obese condition in both adults and children is characterized by cardiac enlargement, and the extent of increased heart size is directly related to the severity of obesity. At the same time, abundant evidence indicates that this enhancement of cardiac mass and chamber size does not actually reflect the amount of body fat but rather the excess of lean body mass that accompanies the obese state. That is, the obese individual has a larger heart than the lean not because he or she is carrying excessive fat tissue but rather because of his or her extra lean body mass. This relationship is intuitively reasonable, since lean body mass, being much more metabolically active than adipose tissue, would be expected to be closely related to the dimensions of its circulatory support. It is tempting, as well, to then speculate that heart enlargement and greater lean body mass (particularly skeletal muscle) of the obese share a common mechanistic etiology.

Using echocardiography and dual-energy X-ray absorptiometry, Whalley et al., 1997 demonstrated this cardiac-lean body mass link in a group of 106 adults, showing that fat-free mass was the only independent predictor of left ventricular mass. Similar analyses in the pediatric age group have reached the same conclusion. Daniels et al., 1995 studied 201 subjects ages 6 to 17 years using the same measurement techniques. In multiple regression analysis, lean body mass by itself explained 75% of the variance of left ventricular mass, while fat mass and systolic blood pressure accounted for only 1.5% and 0.5%, respectively.

In the Muscatine Study of 124 children 8-12 years of age, fat-free mass and sum of skinfolds in the boys accounted for 50% and 15%, respectively, of variance in left ventricular mass (Janz et al., 1995). Among the girls, fat-free mass explained 62% of the variance in ventricular mass, with no significant contribution of skinfold thickness. In a similar assessment of 62 children ages 7 to 13 years, Gutin et al., 1998 found that fat free mass was responsible for 72% of the variance in left ventricular mass. When fat mass was entered into the regression, R2 rose to 0.78.

These reports substantiate the concept that increased cardiac size in the obese parallels the increase in their lean body mass rather than amount of excessive adipose tissue. Recognizing this link bears importance for 1) understanding etiologic mechanisms which stimulate heart growth in the obese state, and 2) uncovering the most appropriate means of expressing anatomic and physiologic variables in respect to body size and composition in overweight individuals.

Etiology
While the cardiac enlargement in obesity can be couched in terms of response to a chronically increased hemodynamic load (Chinali et al., 2006), there is need for a more precise mechanistic explanation for this expansion of the cardiovascular system. Most attention has focused on the role of the anabolic effects of hyperinsulinemia, a reflection of the insulin resistance commonly observed in obese individuals (Gidding et al., 2004; Giordano et al., 2003; Sasson et al., 1993; Wong et al., 2004). This model is particularly attractive since insulin is recognized to increase both cardiac and skeletal muscle mass through insulin-like growth factor-1 receptors (IGF-1) (Hill and Milner, 1985). Studies attempting to identify the etiologic role of these anabolic effects of insulin on myocardial hypertrophy are hampered, however, by the fact that insulin resistance and serum insulin levels are closely linked to the obese state itself (Weiss et al., 2004).

The association of insulin resistance and hyperinsulinemia with heart mass is well-documented in adults. For example, among 40 normotensive nondiabetic obese subjects, Sasson et al., 1993 found that markers of insulin resistance accounted for 50% of the variance of left ventricular mass. Univariate correlation coefficients between indices of insulin resistance and ventricular mass ranged from r = 0.44 to 0.51. The authors suggested that hyperinsulinemia might augment heart size by a muscle growth-stimulating effect or by increasing blood volume (via of changes in renal sodium reabsorption change). In 109 obese adults, Wong et al., 2004 found insulin levels were significantly associated with both indexed left ventricular mass (r = 0.24) and wall thickness ( r = 0.29).

Suggestive evidence for an etiologic role of insulin in promoting myocardial growth has been described in studies of both obese and nonobese youth. In a general population, Gutin et al., 1998 found a significant partial correlation of r = 0.25 between insulin levels and height-adjusted left ventricular mass. Giordano et al., 2003 reported a significant relationship (r = 0.53) between height-indexed left ventricular mass and insulin response to an oral glucose tolerance test. In a group of 48 morbidly obese youth (BMI > 40 gm·m^-2) Gidding et al., 2004 found a mean left ventricular mass at the 80^th percentile of normal and elevated fasting insulin concentrations, with only 13 in the normal range.

"Normalizing" anatomic and physiologic variables with obesity.
Cardiovascular anatomic and physiologic variables are clearly related to body size and composition. In assessing the cardiac effects of obesity, then, it is of prime importance to adjust measurements for these influences. Only then can one make appropriate comparisons between obese and non-obese individuals, for example, or examine relationships of cardiac features to severity of obesity. Although this is critical to reaching appropriate conclusions, the best means of accomplishing such variable adjustments in the obese subject is problematic--particularly as it may be affected by variables such as gender, age, athleticism-- and the issue is not easily resolved.

For "normalizing" cardiac mass to body size, it has been considered most appropriate to make adjustments for inter-individual differences in body stature or skeletal dimensions (which will make no allowance for adiposity). The cardiac mass of a 12 year old boy is considerably more than that of a 6-year old who has the same body composition (consider examining the latter with a magnifying glass). Consider Z to be a cardiac variable (ie., heart mass), and A the correction variable that will appropriately adjust for body stature/skeletal size. A greater value of Z/A in a group of obese children compared to non-obese is consistent with the conclusion that the obese state is responsible for a greater value of Z compared to the non-obese. (Other factors such as body muscularization, athleticism, gender, habitual activity, disease confounders would also have to be considered.)

Utilizing an allometric analysis, de Simone et al., 1992 concluded that for Z= cardiac mass, height^2.7 was the most appropriate value for A that controls for the influence of body fat in assessing heart mass. Most studies addressing left ventricular mass in obese subjects have used this normalizing factor (Chinali et al., 2006; Gutin et al., 1998; Humphries et al., 2002; Kono et al., 1994; Mensah et al., 1999; Papavassiliou et al., 1996; Pflieger et al., 1994). Other height exponents that might serve as normalizing factors for left ventricular mass have been suggested (Daniels et al., 1988). Yoshinaga et al., 1995 found that among 12-year-old Japanese children, height^3.1 and height^1.9 in boys and girls, respectively, were optimal for adjusting heart mass. They considered that racial differences might account for differences from those reported by de Simone et al., 1992.

Alternatively, one could consider whether the degree of obesity accounts for the difference of Z between the two groups. This can be assessed by utilizing another normalizing factor B, which will eliminate the effect of excessive body fat. Following the model above, if Z/B is not significantly different between the two groups, then it could be concluded that the differences in Z in the groups reflects the effect of obesity. One might readily conclude that the optimal B should be body fat content, or even body mass. However, B depends on the variable being studied. For instance, in respect to heart mass, the data outlined above indicate that values are most closely related to lean body mass, which would serve as an appropriate B. Since excessive body fat in the obese individual is greater than the excess of lean body mass (Forbes and Welle, 1983), use of body fat content or body mass as B in this case would result in a spurious low value of adjusted heart mass. This pitfall will be observed later particularly in the case of normalization of values of maximal oxygen uptake (VO₂max).

Given the positive effects on heart size in the obese, it is not surprising that these individuals demonstrate greater absolute values of resting stroke volume and cardiac output than lean subjects. This is evident in studies of both children and adults, which have utilized a variety of measurement techniques (thermodilution, dye dilution, echocardiography). The review of data published in adult subjects by Alexander and Alpert indicates consistent findings of increased oxygen uptake, blood volume, cardiac output, and stroke volume which are directly related to severity of obesity (Alexander and Alpert, 1998). Among those with morbid obesity, for example, Alexander et al., 1962 reported a resting mean cardiac output of approximately 6 L·min^-1 in subjects who were 60 kg overweight , while those 100 kg overweight had an average of 10 L·min^-1. A decline in these hemodynamic variables is seen following weight reduction (Reisin et al., 1983).

In young adolescent females with an average BMI of 34 ± 13 kg·m^-2 Rowland and Dunbar, 2007 found a significant direct association of BMI and resting cardiac output (r = 0.46) and stroke volume (r = 0.36). Giordano et al., 2003 reported resting values for cardiac output of 7.3 ± 1.9 and 5.7 ± 1.2 L·min^-1 (p < 0.05) in obese and normal weight children, respectively. Similar findings were described in 6-15 year old subjects by Pflieger et al., 1994. Chinali et al., 2006 found a direct relationship between severity of obesity and both cardiac output and stroke volume in 14-20 year old subjects (BMI range 16 to 57kg·m^-2) using Doppler echocardiography. Normal weight, overweight (BMI 85-95^th percentile), and obese subjects (>95^th percentile) had mean resting cardiac outputs of 4.82 ± 0.91, 5.14 ± 0.96, and 5.31 ± 1.12 L·min^-1, respectively. Average values for stroke volume were 73 ± 10, 77+11, and 80 ± 13 ml.

Despite these indicators of augmented heart size and output, obese subjects often demonstrate evidence of diminished myocardial function, which is directly related to the severity and duration of their adiposity. This initially came to light in early studies of adults with longstanding morbid obesity, who demonstrated signs and symptoms of frank congestive heart failure. The hearts of these subjects were characterized by reduced left ventricular ejection fraction, chamber dilatation, and elevated end-diastolic pressures, which were often independent of effects of hypertension or coexistent coronary artery disease (Alpert, 2001).

Clinical manifestations of this obesity of cardiomyopathy occur in about 10% of adult patients with a body weight >75% ideal or BMI >40 kg·m^-2, and usually in those whose duration of obesity exceeds 10 years Alpert, 2001). More recent echocardiographic evidence indicates, however, that subclinical evidence of depressed myocardial function, particularly diastolic, is often observed even in adults with mild-to-moderate obesity (Alpert et al., 1998;Chakko et al., 1998). The report of Wong et al., 2004 in 142 middle-aged men and women illustrated these trends. Subjects were divided into groups of referent (BMI<25), overweight (BMI 25-30), mildly obese (BMI 30-35), and severe obesity (BMI >35). While ventricular ejection fraction was similar in all groups, systolic dysfunction was indicated by progressive declines in myocardial systolic peak velocity (as measured by tissue Doppler imaging) with increasing obesity (r=-0.59 versus BMI). Similar trends of indices of diastolic function were observed with greater obesity. Isovolumetric relaxation time and tissue Doppler e' (early diastolic tissue velocity) fell while the ratio of mitral inflow velocity E to e' rose with increasing BMI, suggesting depressed myocardial relaxation properties and increased left ventricular filling pressures.

Recent studies indicate that this trend of subclinical depression in left ventricular function among obese subjects is observed in the pediatric years. Gutin et al., 1998 reported that among 62 children ages 7 to 13 years, percent body fat correlated negatively with lower midwall ventricular shortening fraction (r = -0.37). Mensah et al. (1999) found a significant negative association between midwall shortening fraction and central adiposity in black (but not white) subjects in a group of 15-year old subjects. Chinali et al., 2006 found a significantly lower left ventricular ejection fraction in 14-20 year old obese subjects (BMI >95^th percentile) compared to a non-obese group.

Rowland and Dunbar, 2007 found a progressive decline in left ventricular shortening fraction with increasing BMI in 39 young adolescent females (r=-0.47). Between a BMI of 20 and 60 kg·m^-2, mean shortening fraction fell from 40 to 33 percent. It is important to note that in this study, however, no subject demonstrated a shorting fraction below the lower limit of normal (28 percent). Naylor et al., 2006 found an elevated tissue Doppler E/e' in obese 12 year olds (BMI 30.8 ± 2.6) compared to non obese (8.16 ± 0.26 versus 6.86 ± 0.20 cm·sec^-1, respectively), suggesting mildly elevated left ventricular filling pressure from diastolic dysfunction.

Mehta et al. (2004) compared echocardiographic findings in 10-18 year old overweight and obese children with those of normal weight. No differences in left ventricular shortening fraction or ejection fraction were observed between the groups. The ratio of E/e' was also similar, suggesting no increase in left ventricular filling pressures in the obese. However, some abnormalities in diastolic function were observed by tissue Doppler imaging among the overweight subjects, particularly a lower septal e' (indicative of impaired myocardial relaxation).

Several factors have been proposed to account for this depression of systolic and diastolic function, which appears to be clearly related to both severity and duration of excess adiposity. Myocardial fatigue in the setting of obesity may be the consequence of chronic volume overwork. This model of "high output" congestive heart failure is observed in other conditions characterized by depressed myocardial function and chamber dilatation as a consequence of longstanding elevations of cardiac output (anemia, arterial venous malformations, thyrotoxicosis).

Alternatively, metabolic derangements associated with the obese state may contribute to myocardial dysfunction. Insulin resistance can modify myocardial substrate utilization, causing an increase in myocardial fatty acid oxidation and oxygen consumption and leading to a decrease in cardiac work efficiency (Peterson et al., 2004). Buchanan et al., 2005 found in a mouse model that these metabolic changes and diminished efficiency precede the development of hyperglycemia. Leptin produced by adipose cells has been demonstrated to cause myocardial dysfunction in rats (Nichola et al., 2000).

Cardiac autonomic activity is altered by obesity in both children and adults, but the patterns of these changes have not been consistent in the research literature (Liatis et al., 2004). These variations, however, might play a role in altering myocardial function.