Impact of Body Mass and Body Composition on Circulating Levels of Natriuretic Peptides
http://www.100md.com
《循环学杂志》
the Donald W. Reynolds Cardiovascular Clinical Research Center, University of Texas Southwestern Medical Center, Dallas, Tex.
Abstract
Background— The association between higher body mass index (BMI) and lower B-type natriuretic peptide (BNP) level is thought to be mediated by expression of the natriuretic peptide clearance receptor (NPR-C) in adipose tissue. To explore this association, we tested 2 hypotheses: (1) that N-terminal (NT)-proBNP, which is not believed to bind NPR-C, would not be associated with BMI and (2) that lower BNP would be more closely associated with fat mass than with lean mass.
Methods and Results— Measurements of BNP, NT-proBNP, and body composition by direct dual energy x-ray absorptiometry (DEXA) were performed in 2707 subjects from the Dallas Heart Study. The associations between obesity and low BNP (<4 ng/L) or low NT-proBNP (lowest sex-specific quartile) were evaluated with multivariable logistic regression models stratified by sex and adjusted for age, race/ethnicity, hypertension, left ventricular mass, and end-diastolic volume. Higher BMI was independently associated with lower BNP and NT-proBNP (all P<0.001). When BMI was replaced with both DEXA-derived lean and fat mass, greater lean mass, but not fat mass, was associated with low BNP and NT-proBNP levels.
Conclusions— In a large, population-based cohort, we confirm the previously described association between higher BMI and lower BNP and demonstrate a similar inverse association between BMI and NT-proBNP. Interestingly, both BNP and NT-proBNP are more closely associated with lean mass than with fat mass. These findings do not support the hypothesis that the lower BNP levels seen in obesity are driven by enhanced BNP clearance mediated via NPR-C.
Key Words: natriuretic peptides obesity hormones physiology
Introduction
Several recent studies have examined the relationship between plasma B-type natriuretic peptide (BNP) and obesity, demonstrating an inverse relationship between body mass index (BMI) and BNP concentration in subjects with1 and without2 heart failure. One mechanism that has been suggested to explain this inverse relationship is increased expression of the natriuretic peptide clearance receptor (NPR-C) by adipose tissue, which results in lower BNP levels in obese individuals.3 To further elucidate the role of NPR-C in mediating the relationship between obesity and BNP, we examined 2 hypotheses: first, that the concentration of amino-terminal (NT)-proBNP, which is not believed to bind NPR-C,4 would be unrelated to measures of obesity, and second, that BNP would be inversely related to fat mass but not lean mass.
Methods
Study Population
The Dallas Heart Study (DHS) is a probability-based random sample of Dallas County residents. The present study is based on the 2971 subjects from DHS aged 30 to 65 years who participated in all 3 phases of data collection, including medical history, measurement of anthropometrics, collection of blood samples, MRI for the assessment of cardiac size and function, and dual-energy x-ray absorptiometry (DEXA) for body composition analysis.5 We excluded subjects with unavailable DEXA data (n=84), serum creatinine >2.0 mg/dL (n=16), a history of heart failure by self-report (n=95), unavailable BNP (n=51), and unavailable NT-pro-BNP (n=18). This left a final population of 2707 subjects for analysis. This study was approved by the Institutional Review Board of the University of Texas Southwestern Medical Center and conducted in accordance with institutional guidelines; all participants have provided informed consent. Subjects were enrolled into the study between July 2000 and September 2002, and all study-related procedures were completed by November 2002.
Demographic characteristics were determined by subject self-report at the time of study entry. BMI was calculated based on measured height and weight at the time of cardiac imaging. Cardiac MRI was performed to determine left ventricular (LV) mass and volumes.6,7 To determine interscan variability, patients were scanned once, taken off the scanner briefly, and then rescanned. Interobserver difference for LV mass was 9.2±5 g (5.8±3.5%, n=15), intraobserver difference was 10.5±8.6 g (7.1±6.0%, n=8), and interscan variability was 4.9±10.9 g (2.9±7.5%, n=8). DEXA, a well-established technique of measuring body composition that has been validated against skinfold measurements,8 hydrodensitometry,9 and bioelectrical impedance,10 was used to estimate body composition, with total body mass divided into 3 compartments: bone, fat mass, and lean mass.11 Scans were performed with a dual-beam absorption energy unit (Delphi W, Hologic) bone densitometer in array mode. Oasis software (Hologic) was used to quantify body composition. Three quality controls were monitored routinely with a lumbar spine phantom #8362: bone mineral composition (g), with a coefficient of variation (CV) of 0.5%; bone mineral density (g/cm2), with a CV of 0.3%; and area (cm2), with a CV of 0.4%.
BNP and NT-proBNP Measurement
Venous blood was collected in standard blood collection tubes containing EDTA. Samples were maintained at 4°C for 4 hours and then centrifuged (1430g for 15 minutes) at 4°C. Plasma was then removed and frozen at –70°C until assays were performed. BNP was measured on a TECAN Genesis RSP 200/8 robotic high-throughput platform (Biosite Inc),12 and NT-proBNP was measured on the Elecsys proBNP platform (Roche Diagnostics).13 The CV for the BNP assay averaged 11.2% at a concentration of 30 ng/L and 5.8% at concentrations >60 ng/L; the CV for the NT-proBNP assay was 3.3% at a concentration of 282 ng/L and 3.0% at a concentration of 6012 ng/L. Natriuretic peptide assays were performed between August and December of 2003.
Statistical Analyses
Data were analyzed with the SAS version 9.1 (SAS Corporation) statistical software package. Descriptive statistics stratified by BMI and by natriuretic peptide levels were determined. Given the marked differences in natriuretic peptide levels and body composition between men and women, subsequent analyses were performed stratified by sex. The relationship between natriuretic peptide levels and BMI was examined, with BMI divided into 3 categories: normal weight (BMI <25 kg/m2), overweight (25BMI<30 kg/m2), and obese (BMI 30 kg/m2).14 Sex-specific tests for linear trend were performed with linear regression across categories of BMI.
Linear regression of the log-transformed natriuretic peptide levels was performed with Tobit models to more precisely quantify the association between circulating natriuretic peptide levels and body fat (modeled with BMI or direct DEXA measurements). Tobit models were used to include the partial information contained in the observations censored at the detection threshold of the assay without biasing the estimates of the linear regression.15 In addition to body fat and sex, the following covariates were considered potential confounders, on the basis of prior published work and results of the bivariate analyses: age, race/ethnicity, hypertension, diabetes, hypercholesterolemia, prior myocardial infarction (MI), smoking, ejection fraction, serum creatinine, and LV mass and end-diastolic volume. The models were fit through a stepwise backward-selection algorithm. All candidate variables were entered into the initial saturated model, and then variables with probability values >0.05 in the multivariable model were removed iteratively. The final restricted multivariable model contains all variables that remained significant at the P<0.05 level. Effect modification was tested for by creating a dichotomous age variable at the median (44 years) and then sequentially adding multiplicative interaction terms for age, diabetes, or hypertension to the multivariable model. Only 1 measure of body fat was used in each model.
Because of the skewed nature of the natriuretic peptide distributions and potential violations of assumptions inherent in a least-squares model, logistic regression models were also performed in which low BNP or low NT-proBNP were considered as categorical variables (prespecified as BNP <4 ng/L, the assay detection limit,2 or NT-proBNP in the lowest sex-specific quartile (<7.6 ng/L for men, <20.4 ng/L for women). These cutpoints were selected to maximize comparability with the results of the Framingham investigators,2 who treated "low BNP" as their dichotomous outcome of choice. In the population sample in the present study, the lowest quartiles of BNP for both men and women were below the detection threshold of the assay, as described by Biosite, Inc. Therefore, this lower detection limit was chosen as the dichotomous "low BNP" end point for both men and women. For confirmatory analyses, we also used logistic regression to model for natriuretic peptide levels in the highest sex-specific quartiles (BNP: 9.1 ng/L for men, 16.1 ng/L for women; NT-proBNP: 39.2 ng/L for men, 75.9 ng/L for women). For intergroup comparisons, Student’s t test, Pearson’s 2, or ANOVA F test were used as appropriate. For all analyses, 2-tailed probability values <0.05 were considered statistically significant.
Results
Baseline Characteristics, BMI and Natriuretic Peptide Levels
Baseline characteristics for the study population are shown stratified by BMI category in Table 1. Increasing BMI was associated with an increased likelihood of being female or black; an increased likelihood of having hypertension, LV hypertrophy (LVH), or diabetes; and higher LV ejection fraction, LV mass, and LV end-diastolic volume. The distributions of natriuretic peptide levels and of body composition are presented in Table 2; baseline characteristics stratified by natriuretic peptide levels are shown in Table 3. Compared with subjects with a higher BNP, subjects with lower BNP (<4 ng/L) were younger, were more likely to be male, and were less likely to have hypertension, history of prior MI, and LVH. Compared with subjects with a higher NT-proBNP, subjects with a lower NT-proBNP (lowest sex-specific quartile: <7.6 ng/L for men, <20.4 ng/L for women) were younger, more likely to be black and to have diabetes, and less likely to have hypertension, history of prior MI, or LVH. Because the definition of lower NT-proBNP was sex-specific, there was no sex-based difference in NT-proBNP by construction.
Association Between Body Composition and Natriuretic Peptide Levels
The association between natriuretic peptide levels and BMI is illustrated in Figure 1. For both sexes, BNP and NT-proBNP levels decreased in a stepwise fashion across categories of increasing BMI (P<0.0001 for trend for all comparisons). In multivariable Tobit linear regression analysis, an inverse association was confirmed between BMI and BNP, as shown in Table 4. Of note, a similar inverse association was observed between BMI and NT-proBNP. To eliminate the assumption of a linear relationship between natriuretic peptide levels and BMI, we repeated the multivariable models using BMI as a categorical exposure variable. The results of these models with BMI treated categorically were qualitatively unchanged from those that modeled BMI as a continuous covariate. Model 1 used BMI as a measure of obesity, and model 2 replaced BMI with directly measured fat and lean mass values from DEXA body composition analysis. Both models are shown stratified by gender and adjusted for age, race/ethnicity, hypertension, LV mass, and end-diastolic volume. Other considered covariates (see statistical methods section) were not statistically significant in the final multivariable model. When BMI was replaced in the models by direct measurements of fat and lean mass from DEXA body composition analysis, only lean mass retained the independent inverse association with both BNP and NT-proBNP; fat mass was not associated with either BNP or NT-proBNP, also shown in Table 4.
Logistic regression analyses were also performed in which low BNP or low NT-proBNP were considered as categorical variables; the same covariates were evaluated in these models as in the linear regression models described above. The adjusted ORs for low BNP (<4 ng/L) and low NT-proBNP (lowest sex-specific quartile: <7.6 ng/L for men, <20.4 ng/L for women) are shown in Table 5. Again, model 1 used BMI as a measure of obesity, and model 2 replaced BMI with directly measured fat and lean mass values from DEXA body composition analysis. Qualitatively, these results are identical to the linear regression models. Higher BMI was associated with increased odds of low BNP and low NT-proBNP for both men and women. After BMI was replaced with fat and lean mass in the multivariable model, fat mass was not related to the probability of having a low BNP or NT-proBNP for either men or women. However, higher lean mass was associated with an increased probability of low BNP and NT-proBNP for both men and women. Results were similar after exclusion of subjects taking antihypertensive medication, those with LV ejection fraction <55%, those with history of MI, and those with symptoms of dyspnea or edema. Results of models for natriuretic peptide levels in the highest sex-specific quartiles (BNP: 9.1 ng/L for men, 16.1 ng/L for women; NT-proBNP: 39.2 ng/L for men, 75.9 ng/L for women) yielded qualitatively similar results. No significant interactions were observed between natriuretic peptide levels, body composition measurements, and age, hypertension, or diabetes (P>0.05 for each).
Because fat mass and lean mass were highly correlated, several additional analyses were performed to address issues of collinearity. First, the addition of fat mass to a model that contained lean mass did not qualitatively alter the relationship between lean mass and natriuretic peptide levels; increasing lean mass was associated with decreasing natriuretic peptide levels in models with and without fat mass, yielding very similar point estimates. Furthermore, other analyses performed with different constructs of adiposity (eg, percent body fat) led to the same qualitative result. Finally, we performed stratified analyses comparing subjects with discordant lean mass and fat mass values. BNP and NT-proBNP levels were significantly lower among subjects with below-median fat mass/above-median lean mass than among those with above-median fat mass/below-median lean mass (Figure 2).
Discussion
In a large, population-based cohort, we confirm the previously described association between higher BMI and lower BNP.1,2 We also make 2 novel observations that provide insight into the mechanism of reduced natriuretic peptide levels in subjects with higher BMI values. First, we demonstrate for the first time a similar inverse relationship between higher BMI and lower NT-proBNP levels. Second, using validated DEXA-derived measurements of body composition, we show that the association between BMI and BNP and NT-proBNP is mediated by lean mass rather than fat mass.
A recent article by Wang and colleagues from the Framingham Heart Study2 showed that higher BMI was associated with lower BNP levels. They postulated that this inverse relationship may be due to increased expression of NPR-C by adipose tissue resulting in increased clearance of BNP in obese subjects and suggested that this may contribute to hypertension in obese subjects. In the present study, the association between higher BMI and lower NT-proBNP suggests that nonclearance mechanisms are likely to be important, because NT-proBNP is structurally distinct from BNP and thus unlikely to be cleared via NPR-C. Of interest, in the Framingham study, NT-proANP levels were also lower in subjects with higher BMI, a finding that also suggests that decreased release of natriuretic peptides from the heart, rather than increased clearance, may be responsible for the association between higher BMI and lower natriuretic peptide levels.16
Previous studies evaluating the association between adiposity and natriuretic peptide levels used BMI as a measure of obesity. Here, when we divided body mass into its fat and lean mass components using DEXA, we found that lean mass and not fat mass was responsible for the association between higher BMI and lower natriuretic peptide levels. Because the association was similar for BNP and NT-proBNP, we postulate that a substance produced in the lean mass suppresses either synthesis or release of natriuretic peptides from cardiomyocytes. It is also possible that this effect could be mediated by sex steroid hormones that coordinately influence natriuretic peptide synthesis as well as body composition. For example, androgens, which promote the development of lean mass, may suppress natriuretic peptide release,17,18 whereas estrogens, which are associated with lower lean mass, increase natriuretic peptide levels.19 Less likely hypotheses to explain our observations include the possibility that lean mass secretes a substance that leads to BNP and NT-proBNP degradation, or that lean mass contributes directly to BNP and NT-proBNP clearance. Further studies will be needed to confirm these findings and to explore the associated mechanisms. Fully elucidating the link between obesity and low natriuretic peptide levels may prove an important step toward understanding the association between obesity and cardiovascular disease.
Acknowledgments
This article was supported in part by the Donald W. Reynolds Foundation, by USPHS GCRC grant #M01-RR00633 from NIH/NCRR-CR, and by Roche Diagnostics, Indianapolis, Ind. Reagents for NT-proBNP were provided by Roche Diagnostics and reagents for BNP by Biosite, Inc, San Diego, Calif.
Disclosure
Dr de Lemos has received grant support and consulting fees from Biosite, Inc and Roche Diagnostics, Inc.
Footnotes
The online-only Data Supplement can be found at http://circ.ahajournals.org/cgi/content/full/112/14/2163/DC1.
References
Mehra MR, Uber PA, Park MH, Scott RL, Ventura HO, Harris BC, Frohlich ED. Obesity and suppressed B-type natriuretic peptide levels in heart failure. J Am Coll Cardiol. 2004; 43: 1590–1595.
Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Wilson PW, Vasan RS. Impact of obesity on plasma natriuretic peptide levels. Circulation. 2004; 109: 594–600.
Dessi-Fulgheri P, Sarzani R, Rappelli A. The natriuretic peptide system in obesity-related hypertension: new pathophysiological aspects. J Nephrol. 1998; 11: 296–299.
Vanderheyden M, Bartunek J, Goethals M. Brain and other natriuretic peptides: molecular aspects. Eur J Heart Fail. 2004; 6: 261–268.
Deo R, Khera A, McGuire DK, Murphy SA, Meo Neto JP, Morrow DA, de Lemos JA. Association among plasma levels of monocyte chemoattractant protein-1, traditional cardiovascular risk factors, and subclinical atherosclerosis. J Am Coll Cardiol. 2004; 44: 1812–1818.
Katz J, Milliken MC, Stray-Gundersen J, Buja LM, Parkey RW, Mitchell JH, Peshock RM. Estimation of human myocardial mass with MR imaging. Radiology. 1988; 169: 495–498.
Drazner MH, Dries DL, Peshock RM, Cooper RS, Klassen C, Kazi F, Willett D, Victor RG. Left ventricular hypertrophy is more prevalent in blacks than whites in the general population: the Dallas Heart Study. Hypertension. 2005; 46: 124–129.
Avesani CM, Draibe SA, Kamimura MA, Cendoroglo M, Pedrosa A, Castro ML, Cuppari L. Assessment of body composition by dual energy X-ray absorptiometry, skinfold thickness and creatinine kinetics in chronic kidney disease patients. Nephrol Dial Transplant. 2004; 19: 2289–2295.
Johansson AG, Forslund A, Sjodin A, Mallmin H, Hambraeus L, Ljunghall S. Determination of body composition: a comparison of dual-energy x-ray absorptiometry and hydrodensitometry. Am J Clin Nutr. 1993; 57: 323–326.
Lintsi M, Kaarma H, Kull I. Comparison of hand-to-hand bioimpedance and anthropometry equations versus dual-energy X-ray absorptiometry for the assessment of body fat percentage in 17–18-year-old conscripts. Clin Physiol Funct Imaging. 2004; 24: 85–90.
Salamone LM, Fuerst T, Visser M, Kern M, Lang T, Dockrell M, Cauley JA, Nevitt M, Tylavsky F, Lohman TG. Measurement of fat mass using DEXA: a validation study in elderly adults. J Appl Physiol. 2000; 89: 345–352.
de Lemos JA, Morrow DA, Bentley JH, Omland T, Sabatine MS, McCabe CH, Hall C, Cannon CP, Braunwald E. The prognostic value of B-type natriuretic peptide in patients with acute coronary syndromes. N Engl J Med. 2001; 345: 1014–1021.
Barnes SC, Collinson PO, Galasko G, Lahiri A, Senior R. Evaluation of N-terminal pro-B type natriuretic peptide analysis on the Elecsys 1010 and 2010 analysers. Ann Clin Biochem. 2004; 41 (pt 6): 459–463.
National Heart, Lung, and Blood Institute Expert Panel. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: Evidence Report. Bethesda, Md: National Heart, Lung, and Blood Institute; June 1998. Available at: http://www.nhlbi.nih.gov/guidelines/obesity/ob_home.htm. Accessed December 21, 2004.
Tobin J. Estimation of relationships for limited dependent variables. Econometrica. 1958; 26: 24–36.
Minami J, Nishikimi T, Matsuoka H. Plasma brain natriuretic peptide and N-terminal proatrial natriuretic peptide levels in obese patients:a cause or result of hypertension Circulation. 2004; 110: e76. Letter.
Deng Y, Kaufman S. The influence of reproductive hormones on ANF release by rat atria. Life Sci. 1993; 53: 689–696.
Wu S, Weng X. Regulation of atrial natriuretic peptide, thromboxane and prostaglandin production by androgen in elderly men with coronary heart disease. Chin Med Sci J. 1993; 8: 207–209.
Maffei S, Del Ry S, Prontera C, Clerico A. Increase in circulating levels of cardiac natriuretic peptides after hormone replacement therapy in postmenopausal women. Clin Sci (Lond). 2001; 101: 447–453.(Sandeep R. Das, MD, MPH; )
Abstract
Background— The association between higher body mass index (BMI) and lower B-type natriuretic peptide (BNP) level is thought to be mediated by expression of the natriuretic peptide clearance receptor (NPR-C) in adipose tissue. To explore this association, we tested 2 hypotheses: (1) that N-terminal (NT)-proBNP, which is not believed to bind NPR-C, would not be associated with BMI and (2) that lower BNP would be more closely associated with fat mass than with lean mass.
Methods and Results— Measurements of BNP, NT-proBNP, and body composition by direct dual energy x-ray absorptiometry (DEXA) were performed in 2707 subjects from the Dallas Heart Study. The associations between obesity and low BNP (<4 ng/L) or low NT-proBNP (lowest sex-specific quartile) were evaluated with multivariable logistic regression models stratified by sex and adjusted for age, race/ethnicity, hypertension, left ventricular mass, and end-diastolic volume. Higher BMI was independently associated with lower BNP and NT-proBNP (all P<0.001). When BMI was replaced with both DEXA-derived lean and fat mass, greater lean mass, but not fat mass, was associated with low BNP and NT-proBNP levels.
Conclusions— In a large, population-based cohort, we confirm the previously described association between higher BMI and lower BNP and demonstrate a similar inverse association between BMI and NT-proBNP. Interestingly, both BNP and NT-proBNP are more closely associated with lean mass than with fat mass. These findings do not support the hypothesis that the lower BNP levels seen in obesity are driven by enhanced BNP clearance mediated via NPR-C.
Key Words: natriuretic peptides obesity hormones physiology
Introduction
Several recent studies have examined the relationship between plasma B-type natriuretic peptide (BNP) and obesity, demonstrating an inverse relationship between body mass index (BMI) and BNP concentration in subjects with1 and without2 heart failure. One mechanism that has been suggested to explain this inverse relationship is increased expression of the natriuretic peptide clearance receptor (NPR-C) by adipose tissue, which results in lower BNP levels in obese individuals.3 To further elucidate the role of NPR-C in mediating the relationship between obesity and BNP, we examined 2 hypotheses: first, that the concentration of amino-terminal (NT)-proBNP, which is not believed to bind NPR-C,4 would be unrelated to measures of obesity, and second, that BNP would be inversely related to fat mass but not lean mass.
Methods
Study Population
The Dallas Heart Study (DHS) is a probability-based random sample of Dallas County residents. The present study is based on the 2971 subjects from DHS aged 30 to 65 years who participated in all 3 phases of data collection, including medical history, measurement of anthropometrics, collection of blood samples, MRI for the assessment of cardiac size and function, and dual-energy x-ray absorptiometry (DEXA) for body composition analysis.5 We excluded subjects with unavailable DEXA data (n=84), serum creatinine >2.0 mg/dL (n=16), a history of heart failure by self-report (n=95), unavailable BNP (n=51), and unavailable NT-pro-BNP (n=18). This left a final population of 2707 subjects for analysis. This study was approved by the Institutional Review Board of the University of Texas Southwestern Medical Center and conducted in accordance with institutional guidelines; all participants have provided informed consent. Subjects were enrolled into the study between July 2000 and September 2002, and all study-related procedures were completed by November 2002.
Demographic characteristics were determined by subject self-report at the time of study entry. BMI was calculated based on measured height and weight at the time of cardiac imaging. Cardiac MRI was performed to determine left ventricular (LV) mass and volumes.6,7 To determine interscan variability, patients were scanned once, taken off the scanner briefly, and then rescanned. Interobserver difference for LV mass was 9.2±5 g (5.8±3.5%, n=15), intraobserver difference was 10.5±8.6 g (7.1±6.0%, n=8), and interscan variability was 4.9±10.9 g (2.9±7.5%, n=8). DEXA, a well-established technique of measuring body composition that has been validated against skinfold measurements,8 hydrodensitometry,9 and bioelectrical impedance,10 was used to estimate body composition, with total body mass divided into 3 compartments: bone, fat mass, and lean mass.11 Scans were performed with a dual-beam absorption energy unit (Delphi W, Hologic) bone densitometer in array mode. Oasis software (Hologic) was used to quantify body composition. Three quality controls were monitored routinely with a lumbar spine phantom #8362: bone mineral composition (g), with a coefficient of variation (CV) of 0.5%; bone mineral density (g/cm2), with a CV of 0.3%; and area (cm2), with a CV of 0.4%.
BNP and NT-proBNP Measurement
Venous blood was collected in standard blood collection tubes containing EDTA. Samples were maintained at 4°C for 4 hours and then centrifuged (1430g for 15 minutes) at 4°C. Plasma was then removed and frozen at –70°C until assays were performed. BNP was measured on a TECAN Genesis RSP 200/8 robotic high-throughput platform (Biosite Inc),12 and NT-proBNP was measured on the Elecsys proBNP platform (Roche Diagnostics).13 The CV for the BNP assay averaged 11.2% at a concentration of 30 ng/L and 5.8% at concentrations >60 ng/L; the CV for the NT-proBNP assay was 3.3% at a concentration of 282 ng/L and 3.0% at a concentration of 6012 ng/L. Natriuretic peptide assays were performed between August and December of 2003.
Statistical Analyses
Data were analyzed with the SAS version 9.1 (SAS Corporation) statistical software package. Descriptive statistics stratified by BMI and by natriuretic peptide levels were determined. Given the marked differences in natriuretic peptide levels and body composition between men and women, subsequent analyses were performed stratified by sex. The relationship between natriuretic peptide levels and BMI was examined, with BMI divided into 3 categories: normal weight (BMI <25 kg/m2), overweight (25BMI<30 kg/m2), and obese (BMI 30 kg/m2).14 Sex-specific tests for linear trend were performed with linear regression across categories of BMI.
Linear regression of the log-transformed natriuretic peptide levels was performed with Tobit models to more precisely quantify the association between circulating natriuretic peptide levels and body fat (modeled with BMI or direct DEXA measurements). Tobit models were used to include the partial information contained in the observations censored at the detection threshold of the assay without biasing the estimates of the linear regression.15 In addition to body fat and sex, the following covariates were considered potential confounders, on the basis of prior published work and results of the bivariate analyses: age, race/ethnicity, hypertension, diabetes, hypercholesterolemia, prior myocardial infarction (MI), smoking, ejection fraction, serum creatinine, and LV mass and end-diastolic volume. The models were fit through a stepwise backward-selection algorithm. All candidate variables were entered into the initial saturated model, and then variables with probability values >0.05 in the multivariable model were removed iteratively. The final restricted multivariable model contains all variables that remained significant at the P<0.05 level. Effect modification was tested for by creating a dichotomous age variable at the median (44 years) and then sequentially adding multiplicative interaction terms for age, diabetes, or hypertension to the multivariable model. Only 1 measure of body fat was used in each model.
Because of the skewed nature of the natriuretic peptide distributions and potential violations of assumptions inherent in a least-squares model, logistic regression models were also performed in which low BNP or low NT-proBNP were considered as categorical variables (prespecified as BNP <4 ng/L, the assay detection limit,2 or NT-proBNP in the lowest sex-specific quartile (<7.6 ng/L for men, <20.4 ng/L for women). These cutpoints were selected to maximize comparability with the results of the Framingham investigators,2 who treated "low BNP" as their dichotomous outcome of choice. In the population sample in the present study, the lowest quartiles of BNP for both men and women were below the detection threshold of the assay, as described by Biosite, Inc. Therefore, this lower detection limit was chosen as the dichotomous "low BNP" end point for both men and women. For confirmatory analyses, we also used logistic regression to model for natriuretic peptide levels in the highest sex-specific quartiles (BNP: 9.1 ng/L for men, 16.1 ng/L for women; NT-proBNP: 39.2 ng/L for men, 75.9 ng/L for women). For intergroup comparisons, Student’s t test, Pearson’s 2, or ANOVA F test were used as appropriate. For all analyses, 2-tailed probability values <0.05 were considered statistically significant.
Results
Baseline Characteristics, BMI and Natriuretic Peptide Levels
Baseline characteristics for the study population are shown stratified by BMI category in Table 1. Increasing BMI was associated with an increased likelihood of being female or black; an increased likelihood of having hypertension, LV hypertrophy (LVH), or diabetes; and higher LV ejection fraction, LV mass, and LV end-diastolic volume. The distributions of natriuretic peptide levels and of body composition are presented in Table 2; baseline characteristics stratified by natriuretic peptide levels are shown in Table 3. Compared with subjects with a higher BNP, subjects with lower BNP (<4 ng/L) were younger, were more likely to be male, and were less likely to have hypertension, history of prior MI, and LVH. Compared with subjects with a higher NT-proBNP, subjects with a lower NT-proBNP (lowest sex-specific quartile: <7.6 ng/L for men, <20.4 ng/L for women) were younger, more likely to be black and to have diabetes, and less likely to have hypertension, history of prior MI, or LVH. Because the definition of lower NT-proBNP was sex-specific, there was no sex-based difference in NT-proBNP by construction.
Association Between Body Composition and Natriuretic Peptide Levels
The association between natriuretic peptide levels and BMI is illustrated in Figure 1. For both sexes, BNP and NT-proBNP levels decreased in a stepwise fashion across categories of increasing BMI (P<0.0001 for trend for all comparisons). In multivariable Tobit linear regression analysis, an inverse association was confirmed between BMI and BNP, as shown in Table 4. Of note, a similar inverse association was observed between BMI and NT-proBNP. To eliminate the assumption of a linear relationship between natriuretic peptide levels and BMI, we repeated the multivariable models using BMI as a categorical exposure variable. The results of these models with BMI treated categorically were qualitatively unchanged from those that modeled BMI as a continuous covariate. Model 1 used BMI as a measure of obesity, and model 2 replaced BMI with directly measured fat and lean mass values from DEXA body composition analysis. Both models are shown stratified by gender and adjusted for age, race/ethnicity, hypertension, LV mass, and end-diastolic volume. Other considered covariates (see statistical methods section) were not statistically significant in the final multivariable model. When BMI was replaced in the models by direct measurements of fat and lean mass from DEXA body composition analysis, only lean mass retained the independent inverse association with both BNP and NT-proBNP; fat mass was not associated with either BNP or NT-proBNP, also shown in Table 4.
Logistic regression analyses were also performed in which low BNP or low NT-proBNP were considered as categorical variables; the same covariates were evaluated in these models as in the linear regression models described above. The adjusted ORs for low BNP (<4 ng/L) and low NT-proBNP (lowest sex-specific quartile: <7.6 ng/L for men, <20.4 ng/L for women) are shown in Table 5. Again, model 1 used BMI as a measure of obesity, and model 2 replaced BMI with directly measured fat and lean mass values from DEXA body composition analysis. Qualitatively, these results are identical to the linear regression models. Higher BMI was associated with increased odds of low BNP and low NT-proBNP for both men and women. After BMI was replaced with fat and lean mass in the multivariable model, fat mass was not related to the probability of having a low BNP or NT-proBNP for either men or women. However, higher lean mass was associated with an increased probability of low BNP and NT-proBNP for both men and women. Results were similar after exclusion of subjects taking antihypertensive medication, those with LV ejection fraction <55%, those with history of MI, and those with symptoms of dyspnea or edema. Results of models for natriuretic peptide levels in the highest sex-specific quartiles (BNP: 9.1 ng/L for men, 16.1 ng/L for women; NT-proBNP: 39.2 ng/L for men, 75.9 ng/L for women) yielded qualitatively similar results. No significant interactions were observed between natriuretic peptide levels, body composition measurements, and age, hypertension, or diabetes (P>0.05 for each).
Because fat mass and lean mass were highly correlated, several additional analyses were performed to address issues of collinearity. First, the addition of fat mass to a model that contained lean mass did not qualitatively alter the relationship between lean mass and natriuretic peptide levels; increasing lean mass was associated with decreasing natriuretic peptide levels in models with and without fat mass, yielding very similar point estimates. Furthermore, other analyses performed with different constructs of adiposity (eg, percent body fat) led to the same qualitative result. Finally, we performed stratified analyses comparing subjects with discordant lean mass and fat mass values. BNP and NT-proBNP levels were significantly lower among subjects with below-median fat mass/above-median lean mass than among those with above-median fat mass/below-median lean mass (Figure 2).
Discussion
In a large, population-based cohort, we confirm the previously described association between higher BMI and lower BNP.1,2 We also make 2 novel observations that provide insight into the mechanism of reduced natriuretic peptide levels in subjects with higher BMI values. First, we demonstrate for the first time a similar inverse relationship between higher BMI and lower NT-proBNP levels. Second, using validated DEXA-derived measurements of body composition, we show that the association between BMI and BNP and NT-proBNP is mediated by lean mass rather than fat mass.
A recent article by Wang and colleagues from the Framingham Heart Study2 showed that higher BMI was associated with lower BNP levels. They postulated that this inverse relationship may be due to increased expression of NPR-C by adipose tissue resulting in increased clearance of BNP in obese subjects and suggested that this may contribute to hypertension in obese subjects. In the present study, the association between higher BMI and lower NT-proBNP suggests that nonclearance mechanisms are likely to be important, because NT-proBNP is structurally distinct from BNP and thus unlikely to be cleared via NPR-C. Of interest, in the Framingham study, NT-proANP levels were also lower in subjects with higher BMI, a finding that also suggests that decreased release of natriuretic peptides from the heart, rather than increased clearance, may be responsible for the association between higher BMI and lower natriuretic peptide levels.16
Previous studies evaluating the association between adiposity and natriuretic peptide levels used BMI as a measure of obesity. Here, when we divided body mass into its fat and lean mass components using DEXA, we found that lean mass and not fat mass was responsible for the association between higher BMI and lower natriuretic peptide levels. Because the association was similar for BNP and NT-proBNP, we postulate that a substance produced in the lean mass suppresses either synthesis or release of natriuretic peptides from cardiomyocytes. It is also possible that this effect could be mediated by sex steroid hormones that coordinately influence natriuretic peptide synthesis as well as body composition. For example, androgens, which promote the development of lean mass, may suppress natriuretic peptide release,17,18 whereas estrogens, which are associated with lower lean mass, increase natriuretic peptide levels.19 Less likely hypotheses to explain our observations include the possibility that lean mass secretes a substance that leads to BNP and NT-proBNP degradation, or that lean mass contributes directly to BNP and NT-proBNP clearance. Further studies will be needed to confirm these findings and to explore the associated mechanisms. Fully elucidating the link between obesity and low natriuretic peptide levels may prove an important step toward understanding the association between obesity and cardiovascular disease.
Acknowledgments
This article was supported in part by the Donald W. Reynolds Foundation, by USPHS GCRC grant #M01-RR00633 from NIH/NCRR-CR, and by Roche Diagnostics, Indianapolis, Ind. Reagents for NT-proBNP were provided by Roche Diagnostics and reagents for BNP by Biosite, Inc, San Diego, Calif.
Disclosure
Dr de Lemos has received grant support and consulting fees from Biosite, Inc and Roche Diagnostics, Inc.
Footnotes
The online-only Data Supplement can be found at http://circ.ahajournals.org/cgi/content/full/112/14/2163/DC1.
References
Mehra MR, Uber PA, Park MH, Scott RL, Ventura HO, Harris BC, Frohlich ED. Obesity and suppressed B-type natriuretic peptide levels in heart failure. J Am Coll Cardiol. 2004; 43: 1590–1595.
Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Wilson PW, Vasan RS. Impact of obesity on plasma natriuretic peptide levels. Circulation. 2004; 109: 594–600.
Dessi-Fulgheri P, Sarzani R, Rappelli A. The natriuretic peptide system in obesity-related hypertension: new pathophysiological aspects. J Nephrol. 1998; 11: 296–299.
Vanderheyden M, Bartunek J, Goethals M. Brain and other natriuretic peptides: molecular aspects. Eur J Heart Fail. 2004; 6: 261–268.
Deo R, Khera A, McGuire DK, Murphy SA, Meo Neto JP, Morrow DA, de Lemos JA. Association among plasma levels of monocyte chemoattractant protein-1, traditional cardiovascular risk factors, and subclinical atherosclerosis. J Am Coll Cardiol. 2004; 44: 1812–1818.
Katz J, Milliken MC, Stray-Gundersen J, Buja LM, Parkey RW, Mitchell JH, Peshock RM. Estimation of human myocardial mass with MR imaging. Radiology. 1988; 169: 495–498.
Drazner MH, Dries DL, Peshock RM, Cooper RS, Klassen C, Kazi F, Willett D, Victor RG. Left ventricular hypertrophy is more prevalent in blacks than whites in the general population: the Dallas Heart Study. Hypertension. 2005; 46: 124–129.
Avesani CM, Draibe SA, Kamimura MA, Cendoroglo M, Pedrosa A, Castro ML, Cuppari L. Assessment of body composition by dual energy X-ray absorptiometry, skinfold thickness and creatinine kinetics in chronic kidney disease patients. Nephrol Dial Transplant. 2004; 19: 2289–2295.
Johansson AG, Forslund A, Sjodin A, Mallmin H, Hambraeus L, Ljunghall S. Determination of body composition: a comparison of dual-energy x-ray absorptiometry and hydrodensitometry. Am J Clin Nutr. 1993; 57: 323–326.
Lintsi M, Kaarma H, Kull I. Comparison of hand-to-hand bioimpedance and anthropometry equations versus dual-energy X-ray absorptiometry for the assessment of body fat percentage in 17–18-year-old conscripts. Clin Physiol Funct Imaging. 2004; 24: 85–90.
Salamone LM, Fuerst T, Visser M, Kern M, Lang T, Dockrell M, Cauley JA, Nevitt M, Tylavsky F, Lohman TG. Measurement of fat mass using DEXA: a validation study in elderly adults. J Appl Physiol. 2000; 89: 345–352.
de Lemos JA, Morrow DA, Bentley JH, Omland T, Sabatine MS, McCabe CH, Hall C, Cannon CP, Braunwald E. The prognostic value of B-type natriuretic peptide in patients with acute coronary syndromes. N Engl J Med. 2001; 345: 1014–1021.
Barnes SC, Collinson PO, Galasko G, Lahiri A, Senior R. Evaluation of N-terminal pro-B type natriuretic peptide analysis on the Elecsys 1010 and 2010 analysers. Ann Clin Biochem. 2004; 41 (pt 6): 459–463.
National Heart, Lung, and Blood Institute Expert Panel. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: Evidence Report. Bethesda, Md: National Heart, Lung, and Blood Institute; June 1998. Available at: http://www.nhlbi.nih.gov/guidelines/obesity/ob_home.htm. Accessed December 21, 2004.
Tobin J. Estimation of relationships for limited dependent variables. Econometrica. 1958; 26: 24–36.
Minami J, Nishikimi T, Matsuoka H. Plasma brain natriuretic peptide and N-terminal proatrial natriuretic peptide levels in obese patients:a cause or result of hypertension Circulation. 2004; 110: e76. Letter.
Deng Y, Kaufman S. The influence of reproductive hormones on ANF release by rat atria. Life Sci. 1993; 53: 689–696.
Wu S, Weng X. Regulation of atrial natriuretic peptide, thromboxane and prostaglandin production by androgen in elderly men with coronary heart disease. Chin Med Sci J. 1993; 8: 207–209.
Maffei S, Del Ry S, Prontera C, Clerico A. Increase in circulating levels of cardiac natriuretic peptides after hormone replacement therapy in postmenopausal women. Clin Sci (Lond). 2001; 101: 447–453.(Sandeep R. Das, MD, MPH; )