We evaluated the relation between components of body composition and mortality in patients with cardiovascular disease (CVD). Dual x-ray absorptiometry body composition data from the National Health and Nutrition Examination Survey 1999 to 2004 was linked to total and CVD mortality data 1999 to 2006 in 6,451 patients with CVD. Kaplan-Meier survival analysis for the end points of total and CVD mortality was plotted by quartiles of muscle mass, fat mass, and categories of body mass index (BMI). Subjects were stratified into 4 groups (low muscle/low fat mass, low muscle/high fat mass, high muscle/low fat mass, and high muscle/high fat mass). Adjusted Cox proportional hazards regression determined hazard ratios for total and CVD mortality. Rates of cardiovascular/total mortality were lower in higher quartiles of muscle mass, fat mass, and higher categories of BMI (p <0.001). The high muscle/low fat mass group had a lower risk of CVD and total mortality (risk-adjusted hazard ratios of 0.32, 95% confidence interval 0.14 to 0.73 and 0.38, 95% confidence interval 0.22 to 0.68, for CVD and total mortality, respectively). Thus, increasing fat mass, muscle mass, and BMI were all correlated with improved survival. The specific subgroup of high muscle and low fat mass had the lowest mortality risk compared with other body composition subtypes. This suggests the importance of body composition assessment in the prediction of cardiovascular and total mortality in patients with CVD.
The concept of the “obesity paradox”—the unexpected association between higher body mass index (BMI) and lower mortality—has now been described in multiple cardiovascular disease (CVD) states, including heart failure, hypertension, coronary artery disease, stroke, and peripheral vascular disease. In most studies describing the “obesity paradox” in CVD populations, BMI has been the methodology used to measure obesity. However, BMI—weight in kg/(height in m) 2 —is highly correlated with both fat and muscle mass, and thus, BMI may not adequately reflect variations in the level of fat mass versus muscle mass. Initial studies suggest that the fat-muscle mass balance may be the key to understand the underlying pathophysiology of the “obesity paradox.” In this investigation, we used dual-emission x-ray absorptiometry (DXA) body composition data to investigate cardiovascular death and death from natural causes in subjects with CVD. We postulated that the better outcomes associated with higher BMI in CVD populations may potentially be related to higher muscle mass.
Methods
We used DXA data from National Health and Nutrition Examination Survey (NHANES) 1999 to 2004, linked to mortality data from 1999 to 2006. NHANES uses a stratified, multistage, probability cluster design to evaluate the general, noninstitutionalized population of the United States.
Of 14,200 subjects, more than 20 years old with DXA body composition data, we restricted our analysis to 6,451 participants who had prevalent CVD, as defined by confirmation of coronary artery disease, cerebrovascular disease, hypertension, or congestive heart failure on self-report questions. Hypertension was defined by the presence of ≥1 of the following 3 conditions : (1) systolic blood pressure >140 mm Hg, (2) diastolic blood pressure >90 mm Hg, or (3) self-report.
Whole-body DXA examinations were obtained from a QDR 4500A fan beam densitometer by procedures recommended by the manufacturer (Hologic, Inc., Bedford, Massachusetts). Detailed description of the raw DXA body composition data are provided on the NHANES Website.
The NHANES data sets contained whole-body DXA measurements of fat mass (grams) and muscle mass (grams) for each subject. The main index of adiposity used for this study was trunk fat mass index (TRFI) (trunk fat mass/height 2 ), and the main index of muscle mass used for this study was appendicular muscle mass index (AMMI) (appendicular muscle mass (arms + legs)/height 2 ). We chose trunk fat as it reflects visceral fat levels (compared with total body fat) and, thus, is the most metabolically significant reservoir of fat in the human body and is well known to correlate with both metabolic syndrome and cardiovascular risk. We chose appendicular muscle mass as it has a strong, inverse association with insulin resistance and metabolic syndrome, both of which are positively associated with cardiovascular risk. Gender-specific quartiles of the 2 indices were created.
Additionally, 4 gender-specific body composition categories were created:
low muscle/low fat mass: AMMI < median and TRFI < median,
low muscle/high fat mass: AMMI < median and TRFI ≥ median,
high muscle/low fat mass: AMMI ≥ median and TRFI < median, and
high muscle/high fat mass: AMMI ≥ median and TRFI ≥ median.
This approach of creating 4 groups, based on an above and below median split in biologic variables, has been used to assess the joint impacts of lean and fat mass on cardiovascular outcomes in previously published works.
Lastly, BMI was divided into 4 groups based on the World Health Organization obesity categories definition.
Normal weight = BMI ≤25 kg/m 2 ; overweight = 25 < BMI ≤30 kg/m 2 ; obese = 30 < BMI ≤35 kg/m 2 ; and morbidly obese = BMI >35 kg/m 2 .
NHANES 1999 to 2004 participants were assessed for cardiovascular mortality from National Health Interview Survey interview through December 31, 2006, by the National Center for Health Statistics Research Data Center. Participants whose deaths resulted from accidents, suicide, homicide, firearms, and war were treated as not having died and censored at the recorded time of death. Hence, the total mortality end point is defined in this study as death from natural causes.
A full description of covariates is available on the NHANES Website. With respect to demographics (age, gender, race/ethnicity), this data was obtained from questionnaires. Smoking status was ascertained from the questions regarding previous or current smoking status. Cancer status was ascertained from the question regarding diagnosis of cancer of any organ other than skin cancer.
Diabetes mellitus was defined by the presence of one or more of the following 3 conditions: (1) glycosylated hemoglobin (HbA1C) ≥6.5%, (2) fasting glucose ≥7 mmol/L (126 mg/dl), or (3) self-report. HbA1c was measured using an ion-exchange high-performance liquid chromatography method with a Diamat Analyzer System.
Descriptive statistics (mean, SD, and percentage) were generated for demographic and baseline clinical information for the analytic sample and the complete NHANES 1999 to 2004 sample to characterize the study population.
Missing DXA data, because of no scan or an invalid scan, were nonrandom (i.e., more frequent with greater age, BMI, weight, and height) which could lead to biased population estimates if simply excluded from analyses. Therefore, missing DXA data were imputed by the National Center for Health Statistics using multiple-imputation methodology. Multiple imputation was done before data was made publically available for analysis. Of the 21,230 eligible DXA participants aged ≥8 years who participated in the MEC examinations in 1999 to 2004, scans with 100% nonmissing data were obtained from 16,973 or 80.0%. Of the analytic sample of 6,451, 3,526 subjects had no missing/imputed DXA data. Of the 2,925 subjects in the analytic sample with imputed DXA data, 981 had 1 imputed value and 9 had all DXA values imputed. Five versions of imputed values were generated randomly and independently, resulting in 5 complete data sets of measured and imputed values. We analyzed each of the 5 data sets separately, and the results were combined according to Rubin’s rules. We examined rates of both cardiovascular death and death from natural causes (number of deaths per 10,000 person-months) by gender-specific quartiles of AMMI, TRFI, WHO categories of BMI, and additionally by the 4 body composition categories described earlier. Kaplan-Meier plots were constructed. Log-rank test was used to compare death from natural causes and cardiovascular mortality among groups.
Cox proportional hazards regression analysis was performed (with person-time censored at December 31, 2006, for those who were still alive at that date) adjusting for demographics (age, gender, race/ethnicity), smoking status, and diabetes when assessing risk of cardiovascular mortality. When examining the risk of death from natural causes, we adjusted for demographics (age, gender, race/ethnicity), smoking status, diabetes, and history of cancer.
To make the results representative of the US population, we used NHANES weights (with robust standard error estimation), and to account for the NHANES survey design, we modeled clustering at the NHANES geographic (primary) sampling units using the Wei-Lin approach in Cox proportional hazards regressions.
We checked for interactions in the Cox proportional hazards regression analysis, between the 4 body composition types and age, smoking, diabetes, and race. SAS, version 9.2 (SAS Institute Inc., Cary, NC), was used for all the analyses.
Results
The study sample (n = 6,451) was defined by the presence of CVD, and in most anthropometric characteristics, including appendicular muscle mass/trunk fat mass, was significantly different compared with the complete NHANES sample (n = 14,200) ( Table 1 ).
| Variable | Analytic Sample (n = 6451) | NHANES 1999– 2004 (n = 14200) ∗ | p Value |
|---|---|---|---|
| Age(years) | 56.7 ±16.0 | 46.7 ±18.1 | <.001 |
| Black | 12.6%(N=814) | 10.8% (N=1534) | <.001 |
| Hispanic | 4.5% (N=292) | 7.1% (N= 1008) | <.001 |
| White | 74.3% (N=4795) | 71.9% (N= 10209) | <.001 |
| Appendicular Muscle mass (kg) | 7.8 ±1.8 | 7.6 ±1.9 | <.001 |
| Trunk Fat mass ( kg) | 5.9 ±2.6 | 5.0 ±2.8 | <.001 |
| Body Mass Index ( kg/m 2 ) | 30.0 ±7.0 | 28.1±7.2 | <.001 |
| Men | 47.9% (N= 3090) | 48.7% (N= 6915) | .06 |
| Women | 52.2% (N= 3367) | 51.3% (N=7285) | .06 |
| Fasting glucose (mmol/L) | 6.1 ±2.2 | 5.7 ±1.9 | <.001 |
| Hemoglobin A1c (%) | 5.7 ±1.1 | 5.5 ±1.0 | <.001 |
| C-reactive protein (mg/dl) | 0.55 ±1.0 | 0.43 ±0.94 | <.001 |
| Total Cholesterol (mg/dl) | 207.7 ±46.7 | 202.6 ±50.1 | .03 |
| Diabetes mellitus II | 31.6% (N= 2038) | 18.4% (N=2613) | <.001 |
| Cancer | 10.4% (N=671) | 6.2% (N=880) | <.001 |
| Current Smokers | 19.8% (N=1277) | 24.9% (N= 3536) | <.001 |
| Former smokers | 32.9% (N=2122) | 25.4% (N= 3607) | <.001 |
| Coronary Artery disease | 17.5% (N=1129) | 8% (N=1129) | |
| Congestive Heart Failure | 6.3% (N=406) | 2.9% (N=406) | |
| Stroke | 6.8% (N=439) | 3.1% (N=439) | |
| Hypertension | 92.9% (N= 5993) | 42.2% (N=5993) | |
| Vigorous activity † | 19.5% (N=1258) | 24.7% (N=3507) | <.001 |
∗ Those in the NHANES 1999–2004 sample, greater than 20 years old with completed DXA assessment.
† Individuals reporting that they must climb stairs/hill and/or lift heavy load often daily.
The body composition type groups also showed significant differences in all demographic characteristics ( Table 2 ).
| Variable | Low Muscle/Low Fat (N = 1839) | Low Muscle/High Fat (N = 822) | High Muscle/Low Fat (N = 658) | High Muscle/ High Fat (N = 1615) | p Value |
|---|---|---|---|---|---|
| Age (years) | 60.6±17.0 | 64.1±12.6 | 48.8 ±16.4 | 52.1±9.5 | <.001 |
| Black | 15.3% (N=281) | 4.4% (N=36 ) | 48.5% (N=319) | 27.6% (N=508) | <.001 |
| Hispanic | 22.5% (N=413) | 34.2% (N=281) | 14.9% (N=69) | 22.7% (N=367) | <.001 |
| White | 58.1% (N=1068) | 59.6% (N=490 ) | 33.4% (N=220) | 47.0% (N=759) | <.001 |
| Men | 48.0% (N=883) | 59.0% (N=485) | 51.7% (N=340) | 43.8% (N=707) | <.001 |
| Women | 52.0% (N=956) | 41.0% (N=337) | 48.3% (N=318) | 56.2% (N=908) | <.001 |
| Body Mass Index (kg/m 2 ) | 24.3±2.8 | 29.4±2.4 | 28.5±2.5 | 36.6±6.4 | <.001 |
| Appendicular Muscle mass (kg) | 6.47±1.13 | 7.0±0.92 | 8.5±1.3 | 9.1±1.7 | <.001 |
| Trunk Fat mass ( kg) | 3.99±1.23 | 6.4±1.2 | 4.3±1.2 | 8.0±2.5 | <.001 |
| Fasting Glucose (mmol/l) | 5.7±1.5 | 6.7 ±2.8 | 5.7±2.2 | 6.4±2.5 | <.001 |
| C-reactive protein(mg/dl) | 0.45±1.2 | 0.55±0.68 | 0.34±0.59 | 0.7±1.2 | <.001 |
| Total Cholesterol (mg/dl) | 208.6±45.3 | 209.1±42.8 | 204.3±45.1 | 207.9±50.9 | .59 |
| Diabetes Mellitus II | 28.5% (N=524) | 43.5% (N=358) | 27.8% (N=183) | 46.5% (N=751) | <.001 |
| Cancer | 13.0% (N= 239) | 13.0% (N=107) | 9.3% (N=61) | 7.7% (N=124) | <.001 |
| Current Smokers | 20.0% (N=368) | 11.3% (N=93) | 22.1% (N=145) | 16.3% (N=263) | <.001 |
| Former smokers | 34.5% (N=634 ) | 45.0% (N=370) | 25.7% (N=169) | 33.4% (N=539) | <.001 |
| Coronary Artery disease | 21.9% (N=402) | 25.1% (N=206) | 11.8% (N=78) | 15.7% (N=254 ) | <.001 |
| Congestive Heart Failure | 8.3% (N=153 ) | 9.6% (N=79) | 4.0% (N=26) | 6.9% (N= 111 ) | <.001 |
| Stroke | 9.26% (N=170 ) | 8.9% (N=73) | 5.0% (N=33) | 6.94% (N=112) | <.001 |
| Hypertension | 90.5% (N= 1664) | 90.9% (N=747) | 95.0% (N=625 ) | 95.9% (N=1548) | <.001 |
| Vigorous activity ∗ | 15.3% (N=281 ) | 12.9% (N=106 ) | 23.7% (N=156) | 16.9% (N=273 ) | <.001 |
∗ Individuals reporting that they must climb stairs/hill and/or lift heavy load often daily.
Median length of follow-up in the study sample was 48 person-months, and there were 925 deaths. Median age at death from natural causes was 75 years. Median age at the end of follow-up (or censoring) for those not known to have died was 55 years. Table 3 (footnote) lists cutoffs for quartiles (Q) of AMMI, TRFI, and BMI groups used for analyses.
| Appendicular Muscle mass index ∗ | Trunk Fat mass index † | BMI (kg/m 2 ) | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Quartiles | 1 | 2 | 3 | 4 | p ‡ | 1 | 2 | 3 | 4 | p ‡ | ≤25 | 26-30 | 31-35 | >35 | p ‡ |
| Unadjusted cardiovascular mortality rate | 11.8 | 6.5 | 3.3 | 3.1 | <.001 | 9.1 | 6.5 | 4.5 | 4.1 | <.001 | 10.5 | 5.8 | 3.7 | 4.4 | <.001 |
| Unadjusted total mortality rate | 31.6 | 14.4 | 11.4 | 8.4 | <.001 | 21.8 | 15.3 | 13.2 | 14.4 | <.001 | 26.0 | 14.2 | 13.8 | 11.5 | <.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
