Left Heart Chamber Quantification in Obese Patients: How Does Larger Body Size Affect Echocardiographic Measurements?

Left Heart Chamber Quantification in Obese Patients: How Does Larger Body Size Affect Echocardiographic Measurements? Pu Zong, MD, Lili Zhang, MD, Nada M. Shaban, MD, Jessica Pe~ na, MD, MPH, Leng Jiang, MD, and Cynthia C. Taub, MD, Bronx, New York; Springfield, Massachusetts

Background: Accurate normalization of cardiac chamber size in the obese population is a challenge. The aim of this study was to develop and assess the performance of allometric models for scaling left heart chamber sizes, including left atrial anteroposterior dimension (LADAP), left atrial volume (LAV), left ventricular enddiastolic volume (LVEDV), and left ventricular end-diastolic dimension (LVEDD), in an obese population. Methods: To normalize left heart chamber measurements (Y: LADAP, LAV, LVEDV, and LVEDD) to body size variables (X: height, weight, body mass index, and body surface area), both isometric models (Y = aX) and optimal allometric models (Y = aXb) were tested. A logarithmic transformation (LnY = Lna + b  LnX) and ordinary least squares linear regression was performed to estimate the allometric scaling exponents. Pearson’s correlation coefficients were obtained for measured and indexed left chamber sizes using both isometric and allometric models against body size variables. Gender-specific allometric models were also derived as sensitivity analyses. Results: A total of 717 healthy obese subjects were included in the analysis. The mean body surface area and body mass index were 2.3 m2 and 42.2 kg/m2, respectively. Measured LADAP, LAV, LVEDD, and LVEDV were positively correlated with body size variables. Allometric scaling of LADAP, LAV, LVEDD, and LVEDV showed stronger correlation with measured chamber sizes compared with isometric scaling. The overcorrection caused by isometric scaling significantly improved after allometric models were used. The sensitivity analysis showed no significant differences in scaling exponents between men and women. Conclusions: Normalizing cardiac chamber measurements with allometric scaling methods is superior to the use of isometric methods in removing the effects of body size and minimizing overcorrection in the obese population. Using an allometric model with height provides the most accurate results. (J Am Soc Echocardiogr 2014;-:---.) Keywords: Echocardiography, Left heart, Obesity, Isometric scaling, Allometric scaling

Obesity is epidemic in the United States, affecting >60 million adults, among whom 5 million to 10 million individuals are morbidly obese (body mass index [BMI] $ 40 kg/m2).1 It is known to be associated with the development of various cardiac conditions, including hypertension, heart failure, and arrhythmias. Echocardiography plays a major role in quantifying left heart chamber dimensions, which are thought to increase proportionally with increasing body size.2 To accurately differentiate pathologic cardiac conditions from normal From the Department of Medicine, James J. Peters VA Medical Center, Bronx, New York (P.Z.); Department of Medicine, Jacobi Medical Center, Bronx, New York (L.Z.); Department of Cardiology, Montefiore Medical Center, Bronx, New York (N.M.S., J.P., C.C.T.); Department of Cardiology, Baystate Medical Center, Springfield, Massachusetts (L.J.). Drs Zong and Zhang are co–first authors. Reprint requests: Cynthia C. Taub, MD, Montefiore Medical Center, Albert Einstein College of Medicine, 1845 Eastchester Road, Bronx, NY 10461 (E-mail: ctaub@ montefiore.org). 0894-7317/$36.00 Copyright 2014 by the American Society of Echocardiography. http://dx.doi.org/10.1016/j.echo.2014.07.015

obesity-related increases in cardiac dimensions, it is important to normalize chamber size with an appropriate body size variable. However, the accurate scaling of left heart chamber size in the obese population remains challenging. The American Society of Echocardiography suggests an isometric scaling method to normalize the measured cardiac chamber sizes.3 This method is based on the assumption of a linear relationship between the cardiac chamber size variable (Y, the dependent variable) and the body size variable (X, the independent variable) such as body surface area (BSA) or height, resulting in Y = aX, where a is a scaling factor.4 However, previous studies have suggested that the relationship between cardiac dimensions and body size may be exponential rather than linear.5,6 An allometric model, on the other hand, would allow for a nonlinear relationship between the cardiac chamber size variable (Y) and the body size variable (X). This would assume the form Y = aXb, where b is a scaling exponent. An allometric approach using an optimal scaling exponent for normalization could effectively remove the effect of body size on cardiac chamber sizes.7,8 A few studies have shown the value of allometric scaling for cardiac dimensions. Neilan et al.9 showed that an allometric model using body 1

2 Zong et al

weight as the scaling variable would remove the effect of BMI = Body mass index body size and yielded more accurate results in normal middleBSA = Body surface area aged adults. Further supporting LADAP = Left atrial the use of an allometric model, anteroposterior dimension Yao et al.10 demonstrated, in an obese population (n = 266), LAV = Left atrial volume that applying the same allometric LVEDD = Left ventricular model using the variable of end-diastolic dimension height provided more consistent LVEDV = Left ventricular endresults in scaling left atrial anterodiastolic volume posterior dimension (LADAP). However, the use of allometric models has yet to be fully adopted, mainly because of their mathematical complexity and a lack of supportive evidence from larger studies. In this study, we aimed to develop allometric models for scaling left heart chamber dimensions and to assess the performance of these models in a large obese population. We hypothesized that allometric models would be more accurate than conventional isometric models in normalizing left heart chamber sizes with body size variables in obese individuals. Abbreviations

METHODS Study Population A total of 717 obese (BMI $ 30 kg/m2) adults >18 and <40 years of age with normal echocardiographic results obtained between 2010 and 2012 at Montefiore Medical Center (Bronx, NY) were included in the study. The catchment area of Montefiore Medical Center, with nearly 1,500 inpatient beds and approximately 50 primary care outpatient sites, represents one of the most diverse populations in the New York metropolitan area and in the nation. To ensure a healthy study population, adults with preexisting diagnoses of hypertension, valvular heart disease, myocardial infarction, cardiomyopathy, atrial arrhythmia, pericardial disease, or congenital heart disease, ascertained by International Classification of Diseases, Ninth Revision, codes from emergency room, outpatient, and inpatient visits were excluded. We also reviewed and assessed the electronic medical record of each patient to ensure that the subjects were free of the aforementioned cardiac conditions. The institutional review board approved this study. Body Size Measurements and Echocardiographic Assessments To calculate BMI, body weight (in kilograms) was divided by the square of height (in meters), while BSA was calculated using both the Du Bois and Du Bois formula (BSA [m2] = 0.007184  height [cm]0.725  weight [kg]0.425) and the more recently used Haycock formula (0.024265  height [cm]0.3964  weight [kg]0.5378). The Haycock formula was shown to provide the best fit for most echocardiographic measurements among seven BSA formulas in infants and toddlers.11 Left heart chamber dimensions were calculated in accordance with the American Society of Echocardiography’s chamber quantifications guidelines.3 LADAP was measured from the parasternal longaxis view at the level of the aortic valve. A measurement was taken from the trailing edge of the posterior aortic root to the leading edge

Journal of the American Society of Echocardiography - 2014

of the posterior left atrial wall. Left atrial volume (LAV) was measured using the biplane method of disks (modified Simpson’s rule) using apical four-chamber and apical two-chamber views at ventricular end-systole (maximum left atrial size). Left ventricular end-diastolic dimension (LVEDD) measurements were made from the two-dimensional parasternal long-axis acoustic window at the level of the left ventricular minor axis, approximately at the level of the mitral valve leaflet tips. A linear measurement was taken at end-diastole (maximum left ventricular size) from the leading edge of the septal wall to the leading edge of the posterior wall. Left ventricular end-diastolic volume (LVEDV) was calculated using the biplane method of disks (modified Simpson’s rule) in the apical four-chamber and apical two-chamber views at ventricular enddiastole. Papillary muscles were excluded from the cavity in the tracing. Data Analysis and Statistics To normalize left heart chamber measurements (including LADAP, LAV, LVEDD, and LVEDV) to body size, we tested both isometric and allometric models. In the isometric models, a proportional relationship between left chamber measurements (Y: LADAP, LAV, LVEDD, and LVEDV) and the body size variables (X: height, weight, Du Bois and Du Bois BSA, Haycock BSA, and BMI) was assumed, resulting in Y = aX. Hence, the indexed chamber sizes were equal to chamber measurements/body size (Y/X). In the allometric models, indexed left chamber measurements (Y: LADAP, LAV, LVEDD, and LVEDV) were obtained using each of the body size variables (X: height, weight, Du Bois and Du Bois BSA, Haycock BSA, and BMI) by applying the equation Y = aXb. The indexed chamber sizes were equal to Y/Xb, where b is the scaling exponent. In the allometric models, the scaling factor a and the scaling exponent b were determined using the body size and echocardiographic measurements in our patient population. A commonality of scaling exponents between men and women was assumed. A logarithmic transformation of the above equation was used: LnY = Lna + b  LnX. Ordinary least squares linear regression was then performed to estimate the allometric scaling (a) factors and scaling exponents (b). The homoscedasticity and normality of residual variance were tested using the Breusch-Pagan/CookWeisberg tests and the Shapiro-Wilk tests, respectively. The indexed left chamber measurements using the allometric models and scaling exponents are presented in Table 2. For LADAP, the scaling factors and scaling exponents derived by Neilan et al.9 were also applied. To evaluate the performance of indexing models, Pearson’s correlation coefficients were obtained for measured and indexed left chamber sizes using both isometric and allometric models against each body size variable (Tables 3–6). The ratio of unindexed Y to indexed Y for any individual represents the degree of deviation of left chamber sizes from that expected on the basis of change in body size alone. Two essential criteria had to be met for the scaling model and body size variables for normalizing left heart chamber measurements to be accurate. First, the indexed echocardiographic value had to closely correlate with the measured value (ideally r = 1.00); second, the correlation of the indexed echocardiographic value with body size variables had to be removed (ideally r = 0; negative values indicate overcorrection). We also examined gender-specific allometric models as our sensitivity analyses. The gender-specific scaling exponents were derived and graphically illustrated. The left heart chamber measurements

Zong et al 3

Journal of the American Society of Echocardiography Volume - Number -

RESULTS

Table 1 Baseline clinical and echocardiographic characteristics of study population

Subject Characteristics

Study population (n = 717)

Variable

Mean 6 SD or Percentage

Age (y) Men Non-Hispanic white Non-Hispanic black Hispanic Other Height (m) Weight (kg) Du Bois and Du Bois BSA (m2) Haycock BSA (m2) BMI (kg/m2) LADAP (mm) (n = 627) LAV (mm3) (n = 361) LVEDD (mm) (n = 648) LVEDV (mm3) (n = 277)

30.7 6 6.3 46.6 % 7.3% 36.5% 48.4% 7.8% 1.72 6 0.1 124.0 6 21.4 2.3 6 0.2 2.5 6 0.2 42.2 6 7.9 3.7 6 0.5 56.9 6 16.8 4.9 6 0.5 122.4 6 33.3

Range

18–40

1.4–2.1 102.2–222.3 1.8–3.3 1.8–3.3 30.2–68.5 2.3–6.0 17.2–2.1 3.5–7.0 51.5–5242

Table 2 Summary of allometric models for left heart chamber measurements Scaling exponents Allometric models

LADAP Height Weight Du Bois BSA Haycock BSA BMI LAV Height Weight Du Bois BSA Haycock BSA BMI LVEDD Height Weight

b

95% CI

P

LADAP/height0.20 LADAP/weight0.07 LADAP/BSA0.16 LADAP/BSA0.14 LADAP/BMI0.01

0.20 0.03 to 0.37 0.07 0.003 to 0.13 0.16 0.05 to 0.28 0.14 0.03 to 0.25 0.01 0.05 to 0.06

.018 .04 .005 .013 .818

LAV/height0.86 LAV/weight0.24 LAV/BSA0.68 LAV/BSA0.54 LAV/BMI 0.03

0.86 0.24 0.68 0.54 0.03

0.32 to 1.39 0.02 to 0.45 0.29 to 1.08 0.17 to 0.92 0.21 to 0.15

.002 .033 .001 .004 .757

0.47 0.09

0.34 to 0.60 0.04 to 0.14

<.001 <.001

0.30 0.23 0.03

0.21 to 0.38 <.001 0.14 to 0.31 <.001 0.08 to 0.01 .171

1.44 0.51 1.24 1.06 0.01

0.93 to 1.96 0.29 to 0.73 0.87 to 1.61 0.70 to 1.42 0.18 to 0.21

LVEDD/height0.47 LVEDD/ weight0.09 Du Bois BSA LVEDD/BSA0.30 Haycock BSA LVEDD/BSA0.23 BMI LVEDD/BMI 0.03 LVEDV Height LVEDV/height1.44 Weight LVEDV/weight0.51 Du Bois BSA LVEDV/BSA1.24 Haycock BSA LVEDV/BSA1.06 BMI LVEDV/BMI0.01

<.001 <.001 <.001 <.001 .8686

before and after normalization were compared between men and women. Analyses were performed with Stata version 11 (StataCorp LP, College Station, TX) and Excel 2007 (Microsoft Corporation, Redmond, WA).

Baseline characteristics of our study population as well as left heart chamber dimensions are shown in Table 1. In total, 717 subjects were included in the analyses, with 46.6% being male. The population was highly diverse and included 7.3% non-Hispanic whites, 36.5% non-Hispanic blacks, 48.4% Hispanics, and 7.8% of other ethnicities. The mean height of this population was 1.72 m, and the mean weight was 124.0 kg, resulting in a mean BSA (Du Bois and Du Bois) of 2.3 m2 and a mean BMI of 42.2 kg/m2. The average measurements for LADAP, LAV, LVEDD, and LVEDV were 3.7 mm, 56.9 mm3, 4.9 mm, and 122.4 mm3, respectively. Allometric Models We developed allometric models for LADAP, LAV, LVEDD, and LVEDV on the basis of our obese population’s data. The formulas used are as follows: LADAP/(height0.20); LADAP/(weight0.07); LADAP/(BSA0.16) (Du Bois and Du Bois); LADAP/(BSA0.14) (Haycock); LADAP/(BMI0.01) LAV/(height0.86); LAV/(weight0.24); LAV/(BSA0.68) (Du Bois and Du Bois); LAV/(BSA0.54) (Haycock); LAV/(BMI 0.03) LVEDD/(height0.47); LVEDD/(weight0.09); LVEDD/(BSA0.30) (Du Bois and Du Bois); LVEDD/(BSA0.23) (Haycock); LVEDD/ (BMI 0.03) LVEDV/(height1.44); LVEDV/(weight0.51); LVEDV/(BSA1.24) (Du Bois and Du Bois); LVEDV/(BSA1.06) (Haycock); LVEDV/ (BMI0.01) Table 2 summarizes the allometric formulas and scaling exponents (b) for the left heart chamber measurements. The majority of the scaling exponents were statistically significant (P < .05), except for the models that used BMI. Isometric Models versus Allometric Models Figure 1 summarizes the findings of the Pearson’s correlation coefficients (r) of indexed LADAP, LAV, LVEDD, and LVEDV to measured LADAP, LAV, LVEDD, and LVEDV. In the isometric models, Pearson’s correlation coefficients were modest (r = 0.48–0.980), whereas in the allometric models, Pearson’s correlation coefficients were consistently closer to 1 (r = 0.92–1.0). This suggested that indexing left heart chamber measurements using allometric models correlated better with initial measurements compared with using isometric models. The Du Bois and Du Bois and the Haycock BSAs performed very similarly in normalizing left heart chamber measurements. Allometric models using Haycock BSA correlated with initial echocardiographic measurements slightly better (r = 0.94–0.99), compared with allometric models using Du Bois and Du Bois BSA (r = 0.92–0.98). Correlation with Body Size Measurements Tables 3 to 6 summarize the Pearson’s correlation coefficients of indexed LADAP, LAV, LVEDD, and LVEDV to body size measurements (height, weight, Du Bois and Du Bois BSA, Haycock BSA, and BMI). Measured LADAP, LAV, LVEDD, and LVEDV were positively correlated with body size variables, which indicated that left-sided chamber measurements increased with body size. Volumetric measurements (LAV, r = 0.0004 to 0.18; LVEDV, r = 0.03 to 0.38) correlated better with body size than did

4 Zong et al

Journal of the American Society of Echocardiography - 2014

Table 3 Pearson’s correlation coefficients of normalized LADAP Variable

LADAP

Height

Weight

Du Bois BSA

Haycock BSA

BMI

LADAP Isometric models LADAP/height LADAP/weight LADAP/BSA (Dubois) LADAP/BSA (Haycock) LADAP/BMI Allometric models LADAP/(height0.20)* LADAP/(weight0.07)* LADAP/(BSA0.16) (Du Bois)* LADAP/(BSA0.14) (Haycock)* LADAP/(BMI0.01)* Allometric models (Neilan et al.9) LADAP/(25.66height0.43)† LADAP/(10.52weight0.26)† LADAP/(24.63BSA0.45)† LADAP/(13.65BMI0.27)†

1.00

0.10

0.10

0.12

0.11

0.01

0.90 0.61 0.81 0.79 0.58

0.35 0.09 0.30 0.19 0.48

0.00 0.72 0.42 0.50 0.53

0.17 0.61 0.48 0.49 0.19

0.09 0.69 0.47 0.51 0.38

0.22 0.59 0.20 0.34 0.78

1.00 1.00 0.99 0.99 1.00

0.001 0.08 0.03 0.06 0.10

0.08 0.02 0.001 0.002 0.10

0.06 0.06 0.01 0.03 0.12

0.07 0.04 0.01 0.01 0.11

0.06 0.05 0.03 0.05 0.01

0.98 0.95 0.95 0.93

0.10 0.03 0.10 0.27

0.05 0.22 0.17 0.17

0.007 0.16 0.18 0.01

0.03 0.19 0.18 0.10

0.10 0.23 0.10 0.34

*LADAP allometric models derived from our patient population. † LADAP allometric models derived from Neilan et al.9

Table 4 Pearson’s correlation coefficients of normalized LAV Variable

LAV

Height

LAV Isometric models LAV/height LAV/weight LAV/BSA (Du Bois) LAV/BSA (Haycock) LAV/BMI Allometric models LAV/(height0.86) LAV/(weight0.24) LAV/(BSA0.68) (Dubois) LAV/(BSA0.54) (Haycock) LAV/(BMI 0.03)

1.00

0.16

0.12

0.18

0.15

0.0004

0.98 0.89 0.96 0.96 0.85

0.04 0.10 0.01 0.05 0.43

0.10 0.32 0.11 0.15 0.26

0.06 0.21 0.09 0.10 0.01

0.08 0.28 0.11 0.13 0.14

0.10 0.34 0.09 0.16 0.49

0.98 0.99 0.98 0.99 1.00

0.01 0.15 0.04 0.10 0.15

0.10 0.005 0.04 0.03 0.13

0.07 0.07 0.01 0.03 0.18

0.09 0.03 0.03 0.005 0.16

0.09 0.10 0.06 0.09 0.02

dimensional measurements (LADAP, r = 0.01 to 0.12; LVEDD, r = 0.04 to 0.28), and ventricular measurements correlated better with body size than did atrial measurements. Isometric models did not decrease the dependence of echocardiographic measurements on body size effectively, as many correlation coefficients were not close to 0 (Tables 3–6, rows 3–7). Isometric models also tended to overcorrect or underestimate the left heart chamber measurements, as most of the correlation coefficients were negative (Tables 3–6, rows 3–7). Isometric scaling using weight and BMI resulted in more overcorrection compared with using height and BSA. For example, weight-indexed LADAP using the isometric model resulted in a strong overcorrection (r = 0.72 to 0.09; Table 3, row 4), whereas height-indexed LADAP led to only a modest overcorrection (r = 0.35 to 0.22; Table 3, row 3). In comparison, allometric scaling of LADAP, LAV, LVEDD, and LVEDV with body size variables improved scaling in two ways: (1) by effectively eliminating the body size effect (correlation coefficients

Weight

Du Bois BSA

Haycock BSA

BMI

close to 0) and (2) by avoiding overcorrection (less negative correlation). For example, in Table 3, height-indexed LADAP using allometric models resulted in less overcorrection (r = 0.001 to 0.08; Table 3, row 8) than height-indexed LADAP using the isometric models (r = 0.35 to 0.22; Table 3, row 3). In the allometric method, height and BSA (Du Bois and Du Bois and Haycock) indexed cardiac dimensions performed better than weight and BMI. In the analyses of LADAP, we also compared the allometric models previously derived by Neilan et al.9 (LADAP/ 25.66height0.43, LADAP/10.52weight0.26, LADAP/24.63BSA0.45, LADAP/13.65BMI0.27; Table 3, rows 13–16) with our own models. We found that our own patient-specific models performed better in improving the correlation of indexed LADAP with measured LADAP (r = 1.00, 1.00, 0.99, and 1.00) compared with Neilan et al.’s models (r = 0.98, 0.95, 0.95, and 0.93). The effect of body size variables was also removed more effectively, with much less overcorrection. For example, weight-indexed LADAP using Neilan et al.’s

Zong et al 5

Journal of the American Society of Echocardiography Volume - Number -

Table 5 Pearson’s correlation coefficients of normalized LVEDD Variable

LVEDD

Height

Weight

Du Bois BSA

Haycock BSA

LVEDD Isometric models LVEDD/height LVEDD/weight LVEDD/BSA (Du Bois) LVEDD/BSA (Haycock) LVEDD/BMI Allometric models LVEDD/(height0.47) LVEDD/(weight0.09) LVEDD/(BSA0.30) (Du Bois) LVEDD/(BSA0.23) (Haycock) LVEDD/(BMI 0.03)

1.00

0.28

0.83 0.48 0.70 0.69 0.54 0.96 0.99 0.97 0.98 1.00

BMI

0.15

0.25

0.21

0.04

0.30 0.01 0.23 0.09 0.57

0.04 0.78 0.50 0.58 0.53

0.11 0.61 0.51 0.50 0.15

0.03 0.72 0.52 0.56 0.36

0.23 0.71 0.31 0.47 0.84

0.004 0.25 0.12 0.19 0.25

0.10 0.01 0.07 0.05 0.19

0.08 0.13 0.002 0.06 0.27

0.10 0.07 0.04 0.003 0.24

0.09 0.15 0.14 0.17 0.01

Table 6 Pearson’s correlation coefficients of normalized LVEDV Variable

LVEDV Isometric models LVEDV/height LVEDV/weight LVEDV/BSA (Du Bois) LVEDV/BSA (Haycock) LVEDV/BMI Allometric models LVEDV/(height1.44) LVEDV/(weight0.51) LVEDV/(BSA1.24) (Du Bois) LVEDV/(BSA1.06) (Haycock) LVEDV/(BMI0.01)

LVEDV

Height

Weight

0.32

0.28

0.38

0.33

0.03

0.98 0.87 0.95 0.94 0.85

0.12 0.22 0.12 0.18 0.54

0.24 0.22 0.001 0.04 0.13

0.25 0.06 0.07 0.06 0.18

0.25 0.17 0.04 0.01 0.01

0.14 0.34 0.07 0.16 0.47

0.95 0.96 0.92 0.94 1.00

0.02 0.28 0.07 0.17 0.33

0.23 0.01 0.06 0.06 0.27

0.18 0.15 0.01 0.04 0.37

0.20 0.06 0.03 0.008 0.33

0.19 0.18 0.10 0.17 0.02

model (LADAP/weight0.26) resulted in a negative correlation with weight (r = 0.22), whereas weight-indexed LADAP using our own model (LADAP/weight0.07) improved the correlation to r = 0.02. Sensitivity Analysis We examined the gender-specific allometric models for left heart chamber measurements and illustrate the gender-specific scaling exponents of height and BSA (Du Bois and Du Bois) in Figure 2. There were no significant differences in scaling exponents between men and women, as the 95% confidence intervals of scaling exponents overlapped. In addition, scaling exponents of LADAP and LVEDD were not statistically significant, likely because of a lack of power. Table 7 presents the gender-specific left heart chamber measurements before and after normalization using both isometric and allometric models for height and BSA (Du Bois and Du Bois). Before normalization, men had much greater left heart chamber measurements than women. After indexing with BSA or height, except for LAV, statistically significant differences still existed. DISCUSSION In a total of 717 healthy obese individuals from a diverse urban population, we demonstrated that left heart chamber measurements

Du Bois BSA

Haycock BSA

BMI

(LADAP, LAV, LVEDD, and LVEDV) increased with body size variables (height, weight, BSA [Du Bois and Du Bois and Haycock], and BMI). Second, we developed allometric models for left heart dimensions, including LADAP, LAV, LVEDD, and LVEDV. We found that allometric scaling was superior to conventional isometric scaling in our obese population, manifested by better correlation with measured chamber size, improved elimination of body size effects, and less overcorrection and underestimation. In addition, we compared a previously derived allometric model of LADAP with a model developed using our unique population and found that our model had better correlation with measured LADAP and less overcorrection of body size variables. Echocardiography plays a major role in quantifying left heart chamber dimensions, and accurate normalization of left heart chamber size in the obese population remains a challenge. The role of normalization is to facilitate inter- and intragroup comparisons of cardiac dimensions while eliminating the effects of differences in patients’ size. This allows the accurate determination of dimensions and helps reduce the number of falsely elevated or decreased cardiac measurements due to body size. In this way, a range of normal values can be defined, and the presence or absence of disease can be ascertained with confidence. To determine an accurate normalization model, two elements need to be selected: (1) a mathematical model (isometric model vs allometric) and (2) the body size measurement involved in the model, such as height, weight, BSA, and BMI.

6 Zong et al

Journal of the American Society of Echocardiography - 2014

Table 7 Comparison of left heart chamber measurements before and after normalization using isometric models and allometric models (height and BSA), by gender Men

Figure 1 Pearson’s correlation coefficients of indexed LADAP, LAV, LVEDD, and LVEDV using isometric and allometric models with measured LADAP, LAV, LVEDD, and LVEDV.

Women

Variable

Mean

SD

Mean

SD

P

LADAP (mm) LADAP/height LADAP/BSA* LADAP/(height0.20) LADAP/(BSA0.16) LAV (mm3) LAV/height LAV/BSA LAV/(height0.86) LAV/(BSA0.68) LVEDD (mm) LVEDD/height LVEDD/BSA LVEDD/(height0.47) LVEDD/(BSA0.30) LVEDV (mm3) LVEDV/height LVEDV/BSA LVEDV/(height1.44) LVEDV/(BSA1.24)

3.80 2.12 1.59 3.38 3.30 58.29 32.54 24.56 35.39 32.45 5.05 2.82 2.11 3.84 3.89 133.56 74.46 55.90 57.78 45.53

0.5 0.3 0.2 0.5 0.4 17.2 9.6 7.2 10.4 9.4 0.5 0.3 0.3 0.4 0.4 35.9 19.8 14.6 15.5 12.0

3.66 2.21 1.64 3.31 3.22 55.66 33.63 24.84 36.13 32.18 4.85 2.93 2.17 3.83 3.81 110.46 66.39 49.21 53.85 41.12

0.4 0.3 0.2 0.4 0.4 16.4 9.8 7.3 10.5 9.4 0.5 0.3 0.2 0.4 0.4 29.7 17.8 12.7 13.0 9.0

<.001 <.001 .004 .022 .006 .14 .29 .71 .50 .79 <.001 <.001 .001 .66 .01 <.001 <.001 <.001 .02 .001

*Body surface area was calculated using the Du Bois and Du Bois formula.

Figure 2 Comparison of scaling exponents of allometric models between men and women. *BSA is calculated using the Du Bois and Du Bois formula. Solid diamonds represent scaling exponents. Vertical bars represent 95% confidence intervals.

Conventional isometric methods, based on an assumption of a linear relationship between cardiac dimensions and body size measurements, have been criticized because of their incompatibility with geometric relationships (two-dimensional body size vs threedimensional cardiac chamber size),12 and few studies have shown if their normalization process effectively removed the correlation of their scaled data with the size variable used.13-15 Nevertheless daily practice and most cardiac studies rely on this method of linear scaling for normalization. The allometric method based on nonlinear assumptions has been suggested to be more appropriate in normalizing cardiac dimensions.7-10 A small number of studies have investigated the use of allometric scaling in the single measurement of LADAP in the normal adult and pediatric populations.9,12,16,17 Neilan et al.9 showed that allometric modeling of LADAP using body size variables optimally removed the effects of patient size in a group of normal adults with a mean BMI of 25 kg/m2. In a total of 266 overweight and obese

subjects, Yao et al.10 demonstrated that an allometric method of LADAP using height provides more consistent results than models using body weight or body weight–containing variables, which tended to underestimate LA size in overweight and obese populations.10,18 George et al.19 studied scaling cardiac dimensions in normal athletes and similarly found that initial linearity checks suggested that most of the relationships between the body size variables and cardiac dimensions were nonlinear, thus precluding the simple ratio standard approach to scaling. Allometric scaling, however, remains understudied, especially in the obese population, and has not been applied to normalize other cardiac dimensions. Choosing the appropriate body size dimension to normalize cardiac dimensions is also challenging. Ideally, scaling cardiovascular dimensions should use a fat-free variable, such as lean body mass. In the obese subjects, the contribution of adipose tissue to body weight is very large, constituting >35% of total weight. As a result, it is likely that overcorrection (underestimation) of cardiac dimensions by scaling to weight or BMI occurs in patients with obesity. It has been postulated that using height, a fat-free variable, would reduce the chance of overlooking obesity related enlargement of cardiac dimensions. Specifically, Nagarajarao et al.20 and Gerdts et al.,21 in separate studies, examined height-indexed LADAP and found that it was a powerful predictor of cardiovascular events and all-cause mortality. Our study also showed that height had the best correlation with left heart measurements and less overcorrection in normalization. In a large group of obese individuals, we confirmed that allometric scaling is superior to conventional isometric scaling, but, more important, our findings expand our current knowledge in several aspects. First, we tested the allometric scaling method in a large multiethnic obese population. Compared with previous studies conducted in

Zong et al 7

Journal of the American Society of Echocardiography Volume - Number -

white-dominant populations with smaller sample sizes, our results can be easily and reliably generalized to other diverse populations. Second, we developed allometric models to normalize LADAP on the basis of our population’s unique data and compared it with the previous allometric models derived by Neilan et al.9 Our LADAP models endorse improved performance in terms of better correlation with measured LADAP and independence with body size variables. This finding suggests that allometric scaling parameters should be population dependent. To ensure the best performance, institutions will need to develop specific scaling parameters of their own populations. A multicenter study with large sample size is warranted to ensure internal and external validation of allometric models. To overcome the mathematical challenge, a software module will need to be developed, validated, and integrated into echocardiographic reporting systems. Third, to the best of our knowledge, our study is the first to investigate the allometric scaling method for other cardiac dimensions including LAV, LVEDD, and LVEDV. We showed that the allometric method is superior to the conventional isometric method in normalizing to these three left heart measurements as well. Fourth, consistent with other studies, the allometric models using height, among all body size variables, correlated best with left heart chamber measurements. We found that the exponent factors were the greatest in the height-indexing models (0.20 for LADAP, 0.86 for LAV, 0.47 for LVEDD, and 1.44 for LVEDV). Closely after height, BSA (Du Bois and Du Bois) also demonstrated acceptable correlation with left heart measurements, with exponent factors of 0.16 for LADAP, 0.68 for LAV, 0.30 for LVEDD, and 1.24 for LVEDV. Fifth, it is also of interest to note that volumetric measurements (LAV and LVEDV) had better correlations with height or BSA than dimensional measurements (LADAP and LVEDD). Furthermore, ventricular measurements (LVEDD and LVEDV) had better correlations with height or BSA than atrial measurements (LADAP and LAV). Finally, in the gender-specific allometric models, there were no significant differences in scaling exponents between men and women, suggesting a common model can be used for both. Limitations The results of the present study should be interpreted with some limitations in mind. First, we applied the Du Bois and Du Bois formula to calculate BSA.22 This formula has been consistently and widely used in the existing literature, though it has been criticized, as it was developed >90 years ago, when the population was less overweight and less diverse. For this reason, we also applied the more recently used Haycock formula in our study.23,24 The Haycock formula has been shown to provide the best fit for most echocardiographic measurements among seven BSA formulas in infants and toddlers.11 In our analysis, Du Bois and Du Bois and Haycock BSAs performed very similarly in normalizing cardiac dimensions. Second, to ensure a healthy obese population, we selected patients <40 years of age and free of preexisting cardiac conditions on the basis of International Classification of Diseases, Ninth Revision, codes and chart review. However, given that this was a hospital-based retrospective study, we realize that these patients underwent echocardiography for certain clinical indications. As a result, one may argue that the ‘‘normal obese subjects’’ in our study represented a relative, rather than a community-based, ‘‘normal population.’’ However, echocardiography, in contrast to screening tools, is always performed for a clinical indication. Therefore, we believe we studied the question in a common setting, in which this method can be generally applied.

Third, despite the known and accepted limitations that are posed by isometric scaling, the allometric method, although proposed in many studies, has not been widely adopted. This is likely due to its mathematical complexity. To improve the applicability of this process, a user-friendly software module can be established and integrated into echocardiographic analysis systems.

CONCLUSIONS In a large group of healthy obese individuals from a diverse urban population, we found that normalizing left heart chamber measurements (LADAP, LAV, LVEDD, and LVEDV) using the allometric scaling method is superior to using the isometric method in removing the effect of body size variables. The allometric model using height provides most accurate and consistent results. Our study proposes a very important and highly clinical relevant method, which can be used in differentiating physiologic and pathologic variations of left heart chamber quantification in obese patients.

REFERENCES 1. Flegal KM, Carroll MD, Ogden CL, Johnson CL. Prevalence and trends in obesity among US adults, 1999-2000. JAMA 2002;288:1723-7. 2. Alpert MA, Terry BE, Kelly DL. Effect of weight loss on cardiac chamber size, wall thickness and left ventricular function in morbid obesity. Am J Cardiol 1985;55:783-6. 3. Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, et al, Chamber Quantification Writing Group, American Society of Echocardiography’s Guidelines and Standards Committee, European Association of Echocardiography. Recommendations for chamber quantification: a report from the American Society of Echocardiography’s Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 2005;18:1440-63. 4. Henry WL, Ware J, Gardin JM, Hepner SI, McKay J, Weiner M. Echocardiographic measurements in normal subjects. Growth-related changes that occur between infancy and early adulthood. Circulation 1978;57:278-85. 5. Tanner JM. Fallacy of per-weight and per-surface area standards, and their relation to spurious correlation. J Appl Physiol 1949;2:1-15. 6. Gutgesell HP, Rembold CM. Growth of the human heart relative to body surface area. Am J Cardiol 1990;65:662-8. 7. Lauer MS, Anderson KM, Larson MG, Levy D. A new method for indexing left ventricular mass for differences in body size. Am J Cardiol 1994;74: 487-91. 8. de Simone G, Daniels SR, Devereux RB, Meyer RA, Roman MJ, de Divitiis O, et al. Left ventricular mass and body size in normotensive children and adults: assessment of allometric relations and impact of overweight. J Am Coll Cardiol 1992;20:1251-60. 9. Neilan TG, Pradhan AD, Weyman AE. Derivation of a size-independent variable for scaling of cardiac dimensions in a normal adult population. J Am Soc Echocardiogr 2008;21:779-85. 10. Yao GH, Vallurupalli N, Cui J, Hiser WL, Cook JR, Jiang L. Allometric model improves scaling of left atrial size in obese population: the use of body weight containing variables is challenged. Echocardiography 2011; 28:253-60. 11. Cantinotti M, Scalese M, Murzi B, Assanta N, Spadoni I, Festa P, et al. Echocardiographic nomograms for ventricular, valvular and arterial dimensions in caucasian children with a special focus on neonates, infants and toddlers. J Am Soc Echocardiogr 2014;27:179-91. 12. Sluysmans T, Colan SD. Theoretical and empirical derivation of cardiovascular allometric relationships in children. J Appl Physiol 2005;99:445-57.

8 Zong et al

13. Moller JE, Hillis GS, Oh JK, Seward JB, Reeder GS, Wright RS, et al. Left atrial volume: a powerful predictor of survival after acute myocardial infarction. Circulation 2003;107:2207-12. 14. Popescu BA, Popescu AC, Antonini-Canterin F, Rubin D, Cappelletti P, Piazza R, et al. Prognostic role of left atrial volume in elderly patients with symptomatic stable chronic heart failure: comparison with left ventricular diastolic dysfunction and B-type natriuretic peptide. Echocardiography 2007;24:1035-43. 15. Dewey FE, Rosenthal D, Murphy DJ Jr., Froelicher VF, Ashley EA. Does size matter? Clinical applications of scaling cardiac size and function for body size. Circulation 2008;117:2279-87. 16. Kaski JP, Daubeney PE. Normalization of echocardiographically derived paediatric cardiac dimensions to body surface area: time for a standardized approach. Eur J Echocardiogr 2009;10:44-5. 17. Neilan TG, Pradhan AD, King ME, Weyman AE. Derivation of a sizeindependent variable for scaling of cardiac dimensions in a normal paediatric population. Eur J Echocardiogr 2009;10:50-5. 18. Yao GH, Cui J, Hiser WL, Camarano GP, Slawsky MT, Cook JR, et al. Height is the preferred index for left atrial volume measurement in overweight and obese adults. Circulation 2008;118:S_1154.

Journal of the American Society of Echocardiography - 2014

19. George K, Sharma S, Batterham A, Whyte G, McKenna W. Allometric analysis of the association between cardiac dimensions and body size variables in 464 junior athletes. Clin Sci (Lond) 2001;100: 47-54. 20. Nagarajarao HS, Penman AD, Taylor HA, Mosley TH, Butler K, Skelton TN, et al. The predictive value of left atrial size for incident ischemic stroke and all-cause mortality in African Americans: the Atherosclerosis Risk in Communities (ARIC) Study. Stroke 2008;39: 2701-6. 21. Gerdts E, Wachtell K, Omvik P, Otterstad JE, Oikarinen L, Boman K, et al. Left atrial size and risk of major cardiovascular events during antihypertensive treatment: losartan intervention for endpoint reduction in hypertension trial. Hypertension 2007;49:311-6. 22. Dubois D, Dubois E. A formula to estimate the approximate surface area if height and weight be known. Arch Intern Med 1916;17:863-71. 23. Haycock GB, Schwartz GJ, Wisotsky DH. Geometric method for measuring body surface area: a height weight formula validated in infants, children and adults. J Pediatr 1978;93:62-6. 24. Gehan EA, George SL. Estimation of human body surface area from height and weight. Cancer Chemother Rep 1970;54:225-35.