INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

BODY COMPOSITION IS IMPORTANT in assessing nutritional status, disease risk, and physical fitness. The most common indirect laboratory body composition assessment methods are based on a two-component model, which assumes that the body is comprised of fat and fat-free components and that the fat-free mass (FFM) has constant composition (4, 23). Densitometry, long considered the most valid indirect method based on a two-component model, assumes that the density of the FFM (D_FFM) is 1.1 g/cm³, which, in turn, is based on the assumption that the relative proportions and densities of the water, mineral, and protein constituents are uniform across individuals (23). On the basis of cadaver data, these constituents average 73.8, 6.8, and 19.4%, respectively, of the FFM with the accepted densities of 0.9937, 3.038, and 1.34 g/cm³ for water, mineral, and protein, respectively (5). Variability in D_FFM is the primary factor limiting the accuracy of body composition estimates from body density (23).

In recent years, multicomponent models, in which two or more components of the FFM are measured, have been used to provide more accurate estimates of body composition. Technological advances, such as dual-energy X-ray absorptiometry (DXA) for measuring total body bone mineral combined with tracer dilution techniques for measuring total body water, have led to the development of a four-component (fat, water, mineral, protein) model. With the use of the four-component model to estimate body composition, the two most variable compartments of FFM, body water and bone mineral, are measured, which provides an estimate of body composition based on fewer assumptions, thereby providing a more accurate assessment of fat and FFM (13).

Deviations in the mineral fraction of the FFM (M_FFM) from the 6.8% assumed by cadaver studies and, in turn, alterations in the D_FFM from 1.1 g/cm³ are often inferred from measures of bone mineral density (BMD) or bone mineral content (BMC). In theory, high bone mineral would be expected to increase M_FFM and D_FFM because the BMD is ~2.982 g/cm³, well above the D_FFM (5). A compromised skeleton would have the opposite effect on M_FFM and D_FFM. For example, men have been shown to have higher BMC (11) and have been hypothesized to have higher D_FFM compared with women (14). Blacks have been shown to have a higher BMC when normalized to bone length (16), and it has been hypothesized that blacks have higher D_FFM than whites (22, 24). Cote and Adams (8) determined that young adult black women had a significantly higher BMD than young adult white women. Because BMD explained nearly half of the discrepancies between body fat estimates by a four-component model and the Siri equation, the authors suggest that racial differences in D_FFM are due, in part, to alterations in BMD. In addition, it has been determined that athletes have higher BMC and BMD than nonathletes (7, 17). It has been implied, without supporting data, that individual differences in BMC or BMD result in deviations in D_FFM from the assumed value of 1.1 g/cm³ via the impact on the M_FFM (6, 14, 24).

In contrast to these hypothesized group differences, in a large cohort (n = 703), Visser et al. (25) determined that, although blacks had a higher BMC and M_FFM than whites within groups of men and women, D_FFM was similar. Furthermore, similar to others (3, 17, 26), Visser et al. determined that, across age, race, and gender subgroups, the variation in the water fraction of FFM explained the majority of variation in D_FFM, whereas only a moderate relationship existed between M_FFM and D_FFM.

Thus the aim of the present study was to determine whether BMD and BMC were related to M_FFM and D_FFM in a heterogeneous group of young adults and, if so, to explore whether these relationships were affected by gender, race, or athletic status. We hypothesized that BMD and BMC would be related to M_FFM and D_FFM. Furthermore, we predicted that the relationships between BMD and BMC and M_FFM and D_FFM would be different in men and women, blacks and whites, and athletes and nonathletes.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects. Two-hundred sixteen young men (n = 115) and women (n = 101), including whites (n = 155) and blacks (n = 61) and collegiate athletes (n = 132) and nonathletes (n = 84), participated in the study. The collegiate athletes were recruited from the University of Georgia varsity football, basketball, volleyball, gymnastics, swimming, and track and field teams. Men and women who were recreationally active but did not participate in structured exercise >20 min three times per week were recruited from the University of Georgia student population as nonathletes. Partial data from the 216 volunteers were included in previous papers regarding the effects of athletic status on the density of FFM (2, 20).

Protocol. Physical characteristics, anthropometric measurements, body density, total body water, and total body bone mineral were measured during a single test session (~3.5 h). Subjects were asked to arrive at the laboratory well hydrated after 8-12 h without food and beverages except water. Vigorous physical activity was not performed 24 h before the test session or during the 8- to 12-h fast. In addition, no food or beverages, including water, were consumed during the testing session. Hydration was normal, as indicated by urine specific gravity (1.020 ± 0.0008 g/cm³). In addition, all subjects were within ±2 (SD) of the mean urine specific gravity.

Densitometry. Body density was measured by using underwater weighing and Archimedes' principle to determine body volume. Body weight in air was determined by using an electronic scale to the nearest 0.01 kg. Body weight under water was determined by using a Chatillion autopsy scale to the nearest 0.25 kg with residual lung volume measured simultaneously by using a closed-circuit oxygen-rebreathing, nitrogen-dilution technique modified from Goldman and Buskirk (9). Volume of gas in the gastrointestinal tract was assumed to be 0.1 liter. The within-subjects SD of replicate measurements of body density on 2 days approximately 7 days apart was 0.0016 g/cm³ (17).

Body water. Total body water was measured using deuterium oxide (²H₂O) dilution as previously described (20). Briefly, after a baseline blood sample was taken, subjects ingested a known quantity of ²H₂O [0.31 ± 0.01 (SD) g ²H₂O/kg body mass] in 100 ml of distilled water. Postequilibration (~3 h), another blood sample was taken. Plasma samples were purified by diffusion and analyzed by using a single-beam infrared spectrophotometer (Miran 1FF, Foxboro, Foxboro, MA). Readings from the analyzer were taken at 10 Hz for 1 min with the use of Labtech Notebook data aquisition software (v.8.0, Cambridge, MA) after the sample had equilibrated for 5 min. Total body water was corrected for ²H₂O lost in urine and decreased 4% to account for hydrogen exchange with protein and carbohydrate during the 3-h equilibrium period (21). The within-subjects SD of duplicate measurements of body water on 2 days approximately 7 days apart was 0.75 liter. The within-subjects SD of two to four measurements of body water on the basis of the same blood sample on a given day was 0.71 liter (17), which was considered the technical error associated with the measurement of total body water.

Bone mineral. Bone mineral mass and areal BMD were determined from whole body scans by DXA (Hologic QDR-1000W, enhanced whole body analysis software, version 5.71, Waltham, MA). Hologic DXA is calibrated to assess the mass of bone mineral with X-ray attenuation properties of hydroxyapatite. Bone mineral differs slightly from hydroxyapatite and contains extra water of crystallization and bicarbonate (5). Ho et al. (10) determined that mineral in lumbar vertebrae assessed with DXA was more closely related to ashed weight than to dry bone weight. We assumed that the mineral measured by DXA approximates bone ash, which is the total bone mineral minus the volatile components lost in ashing (i.e., water of crystallization and carbon dioxide from carbonate). Total bone mineral was estimated from the bone mineral content from DXA (ash) by multiplying by 1.04, assuming that 4% of the bone mineral is lost during ashing (5, 15). Total nonosseous mineral, which includes the nonbone sodium, potassium, chloride, phosphate, magnesium, and bicarbonate in cells and the extracellular fluid, was estimated by assuming it was 23% of bone mineral ash (5). Total BMC was calculated by summing osseous and nonosseous mineral.

Body composition calculations. The percentage of body-mass fat estimated from body density, total body water, and total body mineral (%Fat_d,w,m) was determined by using the following equation of Lohman (12)

where D_b is body density, w is total body water measured by ²H₂O dilution expressed relative to body mass, and m is total body mineral estimated from bone mineral ash expressed relative to body mass. FFM was calculated from %Fat_d,w,m and body mass. Protein mass was calculated by difference as FFM  water mass  mineral mass.

Statistical analysis. Statistical analyses were performed with SPSS for Windows version 7.0 (SPSS, Chicago, IL). Means and SD for all variables of interest were calculated for the complete sample and by gender, race, and athletic status subgroups. A t-test for independent samples was used to determine the significance of differences between men and women, blacks and whites, and athletes and nonathletes on relevant body composition measures. Means for D_FFM, and the water (W_FFM), mineral (M_FFM) and protein (P_FFM) fractions of the FFM for the total group and subgroups were compared with assumed values based on cadaver analyses (5) by using a one-sample t-test. Relations between variables were analyzed by using linear regression. An experimentwise alpha level of 0.05 was used for all significance tests. Significance values were corrected to control for familywise error rate for multiple comparisons by using the Bonferroni method.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject characteristics are presented in Table 1. There were no significant differences in age between men and women, blacks and whites, or athletes and nonathletes. Men were significantly taller, had significantly greater body mass and FFM, and had significantly lower %Fat and fat mass than women. Blacks were not significantly different from whites in height, %Fat, or fat mass; however, they possessed significantly more body mass and FFM than whites. Athletes were significantly taller, had significantly more body mass, less %Fat, and possessed less fat mass and more FFM than nonathletes.

As expected, based on previous research, men had significantly greater BMD and BMC than women, blacks had significantly greater BMD and BMC than whites, and athletes had significantly greater BMD and BMC than nonathletes (Table 1). However, women had significantly greater M_FFM than men, and nonathletes had significantly greater M_FFM than athletes. There was no significant difference in M_FFM between whites and blacks (Table 2). The male, female, white, black, athlete, and nonathlete subgroups all possessed M_FFM that was significantly lower than the cadaver data summarized by Brozek et al. (5). In addition, women had significantly higher D_FFM than men, blacks had significantly higher D_FFM than whites, and nonathletes had significantly higher D_FFM than athletes. Male, white, and athlete subgroups had a D_FFM that was significantly lower than assumed values from cadaver data (5).

There was no relation between BMD and M_FFM (r = 0.01; Fig. 1) in the total sample. However, when the effects of gender, race, athletic status, and their interactions were held constant with the use of multiple regression, BMD contributed significantly to the prediction of M_FFM, which explained 16.8% of the variance in M_FFM (Table 3). For men and women, there were significant positive, yet markedly different, relations between BMD and M_FFM (in men: y = 0.60x + 5.02, r = 0.24, SE = 0.48%; in women: y = 3.03x + 2.83, r = 0.54, SE = 0.50%). The slope from the regression equation, indicating the change in M_FFM associated with a unit change in BMD, was fivefold higher in women than in men. There was no significant relation between BMD and M_FFM in whites (r = 0.06) or blacks (r = 0.19) or for athletes (r = 0.08), whereas a significant positive relation existed in the nonathlete group (y = 1.17x + 4.89, r = 0.25, SE = 0.54%).

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Relation of bone mineral density (BMD) to mineral content of the fat-free mass (MFFM). , White, female, control; , white, female, athlete; , black, female, control; , black, female, athlete; , white, male, control; , white, male, athlete;  = black, male, control; , black, male, athlete. Regression equation: y = 0.018x + 6.05, r = 0.01, SE = 0.59 g/cm3.

There was no significant relation between BMC and M_FFM (r = 0.11; Fig. 2) in all subjects. Similar to BMD, BMC contributed significantly to the prediction of M_FFM with the effects of gender, race, athletic status, and their interactions held constant, which explained 10.5% of the variance (Table 3). Within groups of men and women, significant positive relations existed, but the slope from the regression equation predicting M_FFM from BMC was sixfold greater in women (y = 0.66x + 4.82, r = 0.40, SE = 0.55%) than in men (y = 0.12x + 5.44, r = 0.19, SE = 0.49%). Within groups of black and whites, no significant relation existed between BMC and M_FFM for whites (r = 0.09) or blacks (r = 0.25, P = 0.06). Similarly, no significant relations were found between BMC and M_FFM within athlete (r = 0.06) and nonathlete (r = 0.10) groups.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Relation of bone mineral content (BMC) to MFFM. Symbols are the same as in Fig. 1. Regression equation: y = 0.075x + 6.29, r = 0.11, SE = 0.59 kg.

There was no significant relation between BMD and D_FFM (r = 0.06; Fig. 3) in the total sample. When the effects of gender, race, athletic status, and their interactions were held constant (Table 3), BMD contributed significantly to the prediction of D_FFM, but it explained only 1.6% of the variance. There was no significant relation between BMD and D_FFM in men (r = 0.14); however, there was a significant positive relation in women (y = 0.04x + 1.053, r = 0.38, SE = 0.011 g/cm³). There was no significant relation between BMD and D_FFM in whites (r = 0.02), whereas, in blacks, a significant negative relation was found (y = 0.026x + 1.14, r = 0.51, SE = 0.0084 g/cm³). Although there was no significant relation between BMD and D_FFM in athletes (r = 0.08), a significant positive relation existed in nonathletes (y = 0.025x + 1.07, r = 0.30, SE = 0.010 g/cm³).

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Relation of BMD to density of the fat-free mass (DFFM). Symbols are the same as in Fig. 1. Regression equation: y = 0.003x + 1.10, r = 0.06, SE = 0.01 g/cm3.

Although there was a small but significant negative relation between BMC and D_FFM (r = 0.14, P = 0.04; Fig. 4) in the total sample, BMC did not contribute significantly to the prediction of D_FFM when the effects of gender, race, athletic status, and their interactions were held constant (Table 3). There was no significant relation between BMC and D_FFM (r = 0.15) in men, whereas there was a significant positive relation in women (y = 0.007x + 1.08, r = 0.23, SE = 0.011 g/cm³). No significant relation was found between BMC and D_FFM in whites (r = 0.08), whereas, in blacks, a significant negative relation existed (y = 0.006x + 1.12, r = 0.50, SE = 0.009 g/cm³). No significant relation was found between BMC and D_FFM in the athlete (r = 0.17, P = 0.06) or nonathlete (r = 0.21, P = 0.05) groups.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Relation of BMD to DFFM. Symbols are the same as in Fig. 1. Regression equation: y = 0.002x + 1.10, r = 0.14, SE = 0.01 g/cm3.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The primary findings of this study were 1) there was no significant relation of BMD to M_FFM or D_FFM or of BMC to M_FFM and a weak negative relation of BMC to D_FFM in a large heterogeneous sample of young adults; and 2) significant low to moderate relationships of BMD or BMC to M_FFM or D_FFM were found within some gender-race-athletic status subgroups or when the effects of gender, race, and athletic status were held constant by using multiple regression, but BMD and BMC explained only 10-17% of the variance in M_FFM and 0-2% of the variance in D_FFM in addition to that explained by the demographic variables. These results conflict with studies that have inferred a positive relation of BMD and BMC to M_FFM and D_FFM (6, 8, 24). In agreement with past research, we found that men had a greater BMD and BMC than women (11), blacks had greater BMD and BMC than whites (1, 8, 19, 25), and athletes had greater BMD and BMC than nonathletes (7, 14, 17), but these differences did not uniformly translate into hypothesized differences in M_FFM or D_FFM. Our results demonstrate that BMC and BMD should not be used to infer differences in M_FFM or D_FFM.

The lack of relation of BMD to D_FFM is, in part, because DXA measures the mass of bone per unit area (g · cm²), technically called areal or planar density. This is not the density of bone mineral (g/cm³) that was used to estimate D_FFM from chemical studies of bone (5). Thus variation in BMD, which can be large (present study's SD = 0.2 g/cm²) and varies with the heterogeneity of the sample studied, is not necessarily indicative of variations in the density of bone mineral, which is thought to be relatively small and stable.

The impact of BMD on M_FFM and D_FFM is through its effect on BMC. As anticipated, the relation between BMD and BMC in the present study was quite strong (r = 0.97; data not shown). Variation in BMC affects M_FFM and D_FFM only when BMC comprises a disproportionately high or low percentage of the FFM. Comparison of average values for BMC, M_FFM, and D_FFM in the subgroups of subjects in the present study indicates that M_FFM and D_FFM cannot be inferred from absolute BMC. Women had lower BMC but greater M_FFM and D_FFM than men. Blacks had higher BMC, the same M_FFM, and higher D_FFM than whites. Athletes had higher BMC but lower M_FFM and D_FFM than nonathletes. Relationships between BMC, M_FFM, and D_FFM within the groups also were inconsistent and often did not conform to theoretical expectations. For example, although there was a significant relation between BMC and D_FFM in blacks, it occurred without a significant relation between BMC and M_FFM, and, more importantly, it was a negative association. Likewise, although there was a significant relation between BMC and M_FFM in men, no significant relation existed between BMC and D_FFM.

There were stronger relations between BMD and BMC with M_FFM and D_FFM in women than men (Figs. 1-4). The correlations were approximately twice as large and the slopes from regression equations predicting M_FFM from BMD and BMC were five- to sixfold greater in women than in men. The cause of this gender difference is not clear. It may be partially explained by different distributions of values. The ranges of BMD and BMC values for women were relatively small compared with those of men (0.48 vs. 0.82 g/cm² and 1.72 vs. 2.83 kg, respectively), whereas the ranges of M_FFM and D_FFM values were nearly identical for women and men (2.97 vs. 2.69% and 0.05 vs. 0.05 g/cm³, respectively). The greater range of BMD and BMC values in men reflects the greater variation in body size of the men compared with the women in the current sample (SD for FFM = 16.9 and 6.8 kg for men and women, respectively).

The gender differences in the relations of BMD and BMC to M_FFM and D_FFM may also be because of other factors affecting M_FFM, such as differences in W_FFM or P_FFM, because the percentages of water, protein, and mineral sum to 100% of the FFM. In this sample, there were no statistically significant differences between men and women in W_FFM or P_FFM, although men had higher mean W_FFM than women. On the basis of correlations calculated within the groups of men and women, W_FFM explained a similar percentage of the variance in D_FFM (90.1 and 93.9%), but there were substantial differences in the percentages of D_FFM explained by M_FFM (34.6 and 43.5%) and P_FFM (82.0 and 69.8%), suggesting that the relative amounts of FFM constituents other than the M_FFM may have impacted the D_FFM and contributed to the gender difference in the relations of BMD and BMC to M_FFM and D_FFM.

In this study, variability in D_FFM was more strongly related to variance in W_FFM and P_FFM than in M_FFM. Considered separately, water explained 92%, protein explained 75%, and mineral explained 40% of the variance in D_FFM. This finding is similar to those of others (17, 25, 26) and supports the view that, of the various chemical components of the FFM, variability in water has the greatest effect on D_FFM despite the fact that mineral has greater density (3.038 g/cm³) than water (0.9937 g/cm³). The greater effect of water is because 1) the variability in W_FFM is much larger than the M_FFM (Table 2) and 2) it occupies an ~11× greater proportion of the FFM than mineral. On a relative basis, any small change in water will have a greater effect on D_FFM because of the high proportion of FFM that is water compared with mineral.

Although we found significant differences in D_FFM of 0.004 g/cm³ between men and women, blacks and whites, and athletes and nonathletes, differences of this magnitude would affect an estimation of body fatness from body density using the Siri equation by only ~1.7 %Fat, only slightly greater than measurement error. None of these differences could be primarily attributed to differences in BMD or BMC. These differences were because of small differences in water, protein, and mineral composition of the FFM, with the differences in water being of most importance. Our results do not provide strong support for the conclusion that there are gender or racial differences in D_FFM. It is likely that the differences reflected sampling variation. Our results support those of Visser et al. (25), who found no effects of gender or race on D_FFM in a very large sample of adults. Some groups of athletes, such as weight lifters and football players, appear to have a D_FFM that is lower than 1.1 g/cm³ (17, 20) and others, such as gymnasts, may have a D_FFM that is greater than 1.1 g/cm³ (20). However, more data on larger samples are needed to confirm those conclusions.

In summary, we found no relation of BMD to M_FFM or D_FFM or of BMC to M_FFM, and only a weak negative relation of BMC to D_FFM in a large heterogeneous sample of young adults. Significant low to moderate relationships of BMD or BMC to M_FFM or D_FFM were found within some gender-race-athletic status subgroups or when the effects of gender, race, and athletic status were held constant with the use of multiple regression, but BMD and BMC explained only 10-17% of the variance in M_FFM and 0-2% of the variance in D_FFM in addition to that explained by the demographic variables. We conclude that there is not a significant positive relation of BMD and BMC to M_FFM or D_FFM in young adults and that BMC and BMD should not be used to infer differences in M_FFM or D_FFM.