| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1New York Obesity Research Center, St. Luke's-Roosevelt Hospital, Columbia University Institute of Human Nutrition, College of Physicians and Surgeons, New York, New York; and 2Exercise and Health Laboratory, Faculty of Human Movement-Technical University of Lisbon, Lisbon, Portugal
Submitted 23 November 2004 ; accepted in final form 17 February 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
= 0.0005, P = 0.005) than in the three other race groups. Race, sex, and age differences in fluid distribution persisted after adjusting for body composition in a subgroup (n = 994) with dual-energy X-ray absorptiometry lean soft tissue and fat measurements. A relative ECW expansion (i.e., E/I) was present with greater age in most sex-race groups, although the effect was not significantly larger in AA males (P > 0.05) compared with the other race groups, except Asians (P < 0.05). For females, a larger E/I-age effect was found in AA compared with the other race groups, but only the comparison against Hispanics was significant (P < 0.05). The ECW compartment and E/I are thus variably larger, according to race, in healthy older subjects independent of sex, lean soft tissue, and fat mass. body composition; obesity; aging; fluid distribution; ethnicity
Aging is associated with the onset of chronic diseases such as congestive heart failure (17, 31), renal insufficiency (9, 14), and other states associated with absolute or relative ECW expansion. Appropriate interpretation of ECW in these conditions requires normative values adjusted for age, weight, and the other recognized factors that influence fluid status. This information is currently lacking in a large healthy group of subjects varying widely in age.
The primary aim of the current study was to establish, in an ethnically mixed subject group, the cross-sectional relationships between ECW and age, after controlling first for other influencing factors.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Subjects were a convenience sample of healthy adults participating in other unrelated investigations. Additional details of the study group are provided in references (29, 34). Four race groups were identified using self-report: African American (AA), Asian, Caucasian, and Hispanic. Classification into racial groups required similar parent and grandparent race. Subjects with a history of high blood pressure and/or under medication treatment for high blood pressure were excluded. Body composition studies were carried out over a single day after a screening health examination established that the subject was in good health.
There were three main evaluations: total body potassium (TBK) was measured by whole body 40K counting; total body water (TBW) was measured by deuterium or tritium dilution; and lean soft tissue (LST) and fat mass (FM) were estimated in a subgroup of the sample by dual-energy X-ray absorptiometry (DXA). TBW and TBK were used to calculate ECW and intracellular water (ICW). Each subject performed all of the measurements at the Body Composition Unit of St. Luke's-Roosevelt Hospital in New York City.
To fully develop the cross-sectional relations between ECW and age, separate models were prepared for ECW, ICW, and ECW/ICW (E/I).
The Institutional Review Board of St. Luke's-Roosevelt Hospital approved the investigation.
Body Composition Measurements
TBW. Deuterium (2H2O, liter) and tritium dilution (3H2O, liter) volumes were measured with a coefficient of variation (CV) of 1.5 and 2.0%, respectively. The dilution volumes were then used to calculate TBW as reported by Schoeller (40). Specifically, both dilution volumes were converted to water mass assuming an average body temperature of 36°C. The tritium dilution space was also adjusted for proton exchange by assuming actual water volume is 96% of the measured isotope distribution volume.
TBK.
TBK was estimated from the measured 1.46 MeV 
-ray decay of naturally occurring 40K as TBK = 40K/0.000118 (33). The subject's 40K was determined by counting for 9 min in a 4
whole body counter (33). The raw count is corrected for body mass as described in Pierson et al. (35). This system has a between-measurement CV of 1.5%.
LST and FM. The LST and FM compartments were measured using a series of pencil-beam dual photon systems and related software manufactured by GE Lunar (Madison, WI). Body composition data collected on different DXA systems was translated to common values with cross calibration equations using the procedure reported by Russell-Aulet et al. (39).
Calculations and Statistical Methods
ECW and ICW.
Potassium is mainly present in ICW and ECW at stable concentrations of 4 and 152 mmol/kgH2O, respectively (7). The ICW mass can be derived from TBK and TBW (24, 45) as:
 | (1) |
 | (2) |
 | (3) |
 | (4) |
Statistical methods. All group results are expressed as the mean and standard deviation. Analyses were carried out using the statistical program SPSS v12.0 (SPSS, 2003). P < 0.05 was considered significant for individual tests and for the overall protection level against type I error when multiple comparisons were made. In the tables of baseline characteristics, one-way ANOVA was used to compare unadjusted ECW across race groups within each sex. Scheffé's adjustment for multiple comparisons was made if variances were equal. Otherwise, Dunnett's T3 adjustment was used if variances were not equal (P < 0.05). Independent sample t-tests were used to compare ECW values between males and females.
The main focus was to examine the associations between ECW and age, independent of other influencing factors. Multiple regression analysis was applied to investigate the associations of ECW with recognized influencing factors, including sex, race, age, height, weight, age squared, height squared, weight squared, and two- and three-way interactions. Variables or interactions making no significant contribution were eliminated from the final model. If there was no sex x race interaction for each sex and race group, the associations between age and ECW adjusted for weight, height, and other covariates were calculated and plotted separately. After a significant F test for age x race interaction, comparisons of age -coefficients among races were made using the Schaffer modification of Holm's step-down multiple comparison procedure to protect against inflation of type I error (41, 43). P values that failed to reach significance by this procedure are designated NS; otherwise, all P values presented in race comparisons exceed the multiple comparison threshold of P < 0.05.
During model development, homogeneity of variance and normality of residuals were tested. If variances were not equal or residuals were not normal, the dependent variable was transformed using a logarithmic function (log10). In all cases, this brought the models into, or close to, compliance with the assumptions of multiple linear regression models. In addition, in some models, a small number of cases with outlying residuals were excluded during model development to achieve normality of the residuals. Neither the transformation nor the deletion of cases with outlying residuals resulted in significant changes to the results and conclusions.
Data from the subgroup of cases with DXA body composition measurements were used to expand the model building and data explorations by adding LST and FM as potential ECW and fluid distribution explanatory variables. These additional models were developed after observing significant age x race interactions in the first phase of this investigation, which used only weight and height as body composition indicators. In this second phase, multiple regression models with ECW, ICW, and E/I as separate dependent variables were developed with LST, FM, age, sex, race, and interactions as potential explanatory variables. Variables not making significant contributions to the model were excluded from further consideration. Because a significant sex x race interaction was detected, separate models were developed for males and females to simplify interpretation. Male and female adjusted ECW values were then plotted separately against age for each race group. Adjusted values were calculated by adding the group mean value to the residuals after completing the regression analysis.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
The characteristics of the total subject group are summarized in Table 1. The characteristics of subjects who also had DXA measurements are summarized in Table 2. In general, both the main study group and DXA subgroup were similar in age (
50 yr) and body mass index (25 kg/m2).
|
|
The bivariate correlations between variables evaluated in the entire sample are presented in Table 3. Weight and height were associated with age in males, whereas for females, only height was related with age. Older males and females were shorter than their younger counterparts, and older males weighed less.
|
|
= 0.001) as indicated in Table 4, compared with the other race groups (AA vs. Hispanic: race x age: P < 0.001; AA vs. Caucasian: race x age: P < 0.001; AA vs. Asian: race x age: P = 0.001). The model also includes an age x height interaction ( = –0.004) (Table 4). The finding of race x age interactions in this sample led us to refine our analysis using only the DXA subgroup, where we could replace crude body composition indicators, weight and height, with LST and FM as explanatory variables. After adjusting for LST, FM, race, and interactions, the relationships between ICW and E/I as separate dependent variables with age were also analyzed.
The bivariate relations between ECW, age, and body composition variables in the DXA subgroup are summarized in Table 5. Lean soft tissue and FM were associated with age in males and females. Older males and females had a smaller LST mass than their younger counterparts, whereas a larger FM was present in the older males and females.
|
|
= 0.008, P < 0.001; females, = 0.010, P < 0.001) and FM (males: = 0.002, P < 0.001; females, = 0.002, P < 0.001) are both associated with more ECW. Race and age were significant only in interactions with other variables; hence their coefficients are not shown separately in the regression equations of Table 6. The largest age x race coefficient was seen in AA males, significantly larger than in Asians (AA vs. Asian, P = 0.007), although not significantly different from Caucasians or Hispanics (AA vs. Caucasian, P = 0.019, NS; AA vs. Hispanic, P = 0.858, NS), as indicated in Table 7. For females, as for males, the largest coefficient was for AA, significantly greater than Hispanics (AA vs. Hispanic, P = 0.002), although not significantly different from Asians or Caucasians (AA vs. Asian, P = 0.043, NS; AA vs. Caucasian, P = 0.298, NS). Race also moderated the relations between LST and ECW in the female groups.
|
|
ICW was log transformed (log10), and 71 cases (7.1%) with outlying residuals were excluded during model development to meet model requirements for equal variance and normality of residuals.
For males, the significant explanatory variables were race x age (P < 0.001), LST (P < 0.001), race x LST (P = 0.006), and race x fat (P = 0.018), explaining 93% of the total variance of log10ICW. For females, the variables significantly associated with log10ICW were race x age (P = 0.002), age (P < 0.001), LST (P < 0.001), FM (P < 0.001), race (P < 0.001), and LST x FM (P < 0.001), explaining 86% of the total variance.
The coefficients for males can be seen in Table 7. The negative association with age is greatest in AA males and is significantly greater than that of Asians and Caucasians but not Hispanics (AA vs. Hispanic: P = 0.130, NS), as presented in Table 7. Similarly, a greater negative association is also seen in AA females, and the difference is significant compared with all the other race groups. Once again, for illustrative purposes the relationship between adjusted untransformed ICW and age for each race group is presented in Fig. 1 for males and females. Overall, a smaller adjusted ICW is present with greater age across all the sex/race groups.
E/I
Because of unequal variances and nonnormality of residuals, E/I was log transformed (log10), and five cases (0.5%) with outlying residuals were excluded from model development. For males, the variables that were associated with the log10E/I were race x age (P < 0.001), age (P < 0.001), FM (P < 0.001), and race x LST (P = 0.043), explaining 43.1% of the total variance of log10E/I. For females, the significant explanatory variables were race x age (P = 0.022), age (P < 0.001), race x LST (P < 0.002), and LST x FM (P < 0.001), explaining 27.6% of the total variance. For males, the highest positive association between transformed E/I and age was in AA males, and the difference reached statistical significance in the comparison with Asian men, as shown in Table 7.
For females also, the largest positive association was seen in the AAs, although the comparison was statistically significant only with the Hispanics (Table 7). For illustrative purposes, the relationship between adjusted untransformed E/I and age for each race group is presented in Fig. 1. For males and females, a relative larger adjusted fluid distribution is present with greater age across the race groups, except for Asian males.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Our initial transformed ECW models suggested a small (<1 kg) but significantly larger ECW in males and smaller ECW in females after adjusting for body weight and other covariates. When we expanded our analysis to include DXA-measured LST and FM, replacing body weight in the ECW regression models, the results were more consistent across males and females: greater age was generally associated with a significantly larger ECW after controlling for covariates. Race, however, moderated the relation between ECW and age such that one of the most significant age-associated effects on ECW was observed in AAs. Similar developed models revealed a smaller ICW and larger E/I with greater age. Thus, even after controlling for LST and FM, there were differences in fluid compartment volumes and distribution across the adult life span in this cross-sectional sample of healthy adults.
ECW and Fluid Distribution
Sex and age effects. Several investigators, using different ECW markers, report that ECW relative to body weight is similar in males and females (5, 19, 21, 22, 26, 28). Our findings, however, indicate that factors other than body weight may also influence ECW and account for sex differences in fluid distribution. Notably, males had a smaller LST slope than females in the body composition ECW regression model presented in Table 6. In the regression on untransformed ECW values (not shown), these slopes (0.35 and 0.43) are nearly identical to the ECW content of LST for the male (i.e., 0.36) and female (i.e., 0.42) groups (Table 2). In contrast, the untransformed slopes for FM observed in the ECW model for males and females were nearly identical. Absolute ECW is larger in males, who in general have a greater LST mass compared with females, with LST accounting for over 53.3 and 61.1% of the amount of transformed ECW present, respectively, for males and females according to our developed models.
However, LST is not a homogenous compartment when considered from the tissue-organ body composition level (18). Fat-free skeletal muscle, adipose tissue, liver, and kidney, for example, have respective ECW/LST and E/Is of 0.28/0.58, 0.10/2.0, 0.31/0.76, and 0.45/1.4 (42). Thus the slopes of FM in our untransformed models (data not shown, 0.08–0.09) are very close to the fraction of ECW in adipose tissue of 0.10. The slope of the LST term in our models may reflect the integrated ECW fraction for all nonfat organs and tissues; because males have a larger fraction of LST as skeletal muscle (18) compared with females, this may account for the smaller LST slope in males compared with females.
With respect to age, previous results based on small samples, variable measurement methods, and different methods for size-adjusting fluid volumes are conflicting (21, 32, 37). The findings of the present study indicate that multiple factors and their interactions are associated with the volume of ECW present and may account for the variable observations reported earlier. Because aging is associated with absolute and relative loss of skeletal muscle and enlargement of the adipose tissue compartment (1, 13), ECW adjusted for body mass would be expected to remain the same or be smaller with greater age depending on the magnitude of senescence body composition effects. Moreover, as the fraction of LST as skeletal muscle is smaller with greater age (18), this may in part account for the larger observed ECW even after controlling for LST in older subjects. That is, as noted earlier, skeletal muscle has a small ECW compared with other tissues and organs. With age-related muscle cell atrophy, the fraction of LST as ECW thus would be expected to increase. This hypothesis is consistent with the current study observations.
Race effects. An unanticipated finding in our study was the detection of clear race differences in the association of ECW with age. Our models using transformed ECW indicate a 6.7 and 3.4% larger ECW in older AAs compared with younger males and females, respectively, after adjusting for other influences. Race differences were also evident in ICW and E/I models. MacFarlane et al. (23) also noted that their estimates of TBW in Bantu miners were higher than those reported in the literature for Europeans. Raman et al. (36) likewise found a higher TBW in elderly male and female AAs.
The basis for a larger increment in ECW with greater age in AAs is unknown, although two potential groups of mechanisms should be considered. First, AAs reportedly have a larger skeletal muscle mass than Caucasians after controlling for age, weight, height, and FFM (16, 20, 30). As noted earlier, a larger skeletal muscle mass after adjusting for LST and fat would be associated with a smaller ECW. This prediction is accurate for young subjects: AA males and females both had the lowest adjusted ECW at age 20 yr. However, AAs also had the largest increment in adjusted ECW with age among the four race groups. If this line of reasoning is correct, our findings suggest AAs experience a greater senescence-related lowering of skeletal muscle mass across the adult life span compared with other race groups. This hypothesis requires testing using appropriate methods in large and carefully evaluated samples as in the present report.
The second potential mechanistic basis for the observed race-related ECW differences involves physiological processes mediating fluid balance. One empirical observation is that race differences are present in the prevalence of high blood pressure (44). Freis et al. (12) reported that >50% of AAs with high blood pressure are sensitive to salt intake and that the prevalence of diuretic-sensitive hypertension approaches 75%. Salt-sensitive hypertension may be a manifestation of expanded extracellular fluid and blood volumes. According to Weir and colleagues (46), AAs have an impaired ability to excrete a salt load and develop a higher arterial pressure as a necessary physiological response to enhance natriuresis. Campese et al. (4) demonstrated slow salt excretion in hypertensive AAs after a dietary salt load. Chrysant et al. (6) report that AAs are more likely than Caucasians and other non-black hypertensive patients to have markedly low plasma renin activity. Our observation of a relatively large ECW and E/I in both AA males and females, notably with greater age, are synchronous with and extend these earlier observations.
Study Limitations
There are several limitations of the current study. First, our investigation was cross-sectional and, ideally, age-related inferences should be based on longitudinal data. Second, we identified patients with high blood pressure by history and with a screening physical examination. However, our clinical blood pressure measurements were not sufficiently accurate and standardized to directly explore the associations between ECW and blood pressure. Also, other variables, including diet and exercise habits and weight loss (25), could have occurred at different rates in the race/ethnic groups and may have contributed to the observed differences seen with aging. Our findings should, however, prove useful in guiding future research and experimental design.
In conclusion, in the present study we applied multiple regression modeling methods to evaluate ECW variation with age. Our findings in general show that ECW, particularly E/I, is larger, and ICW is smaller, with greater age after adjusting for race and body composition. We propose age-related tissue-organ body composition differences as one potential explanation for these observations. We also observed significant differences in the fluid distribution-age relations among race groups, with the largest age-related differences observed in AA subjects. Finally, the underlying mechanisms of age and race-related differences in ECW are worthy of future investigation.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Nass, S. S. Pezzoli, M. C. Oliveri, J. T. Patrie, F. E. Harrell Jr., J. L. Clasey, S. B. Heymsfield, M. A. Bach, M. L. Vance, and M. O. Thorner Effects of an Oral Ghrelin Mimetic on Body Composition and Clinical Outcomes in Healthy Older Adults: A Randomized Trial Ann Intern Med, November 4, 2008; 149(9): 601 - 611. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M Silva, S. B Heymsfield, D. Gallagher, J. Albu, X. F Pi-Sunyer, R. N Pierson Jr, J. Wang, S. Heshka, L. B Sardinha, and Z. Wang Evaluation of between-methods agreement of extracellular water measurements in adults and children Am. J. Clinical Nutrition, August 1, 2008; 88(2): 315 - 323. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
