Whole-body skeletal muscle mass: development and validation of total-body potassium prediction models1-3

ZiMian Wang, Shankuan Zhu, Jack Wang, Richard N Pierson Jr, and Steven B Heymsfield
ABSTRACT Background: A substantial proportion of total body potassium (TBK) in humans is found in skeletal muscle (SM), thus affording a means of predicting total-body SM from whole-body countermeasured 40K. There are now > 30 whole-body counters worldwide that have large cross-sectional and longitudinal TBK databases. Objective: We explored 2 SM prediction approaches, one based on the assumption that the ratio of TBK to SM is stable in healthy adults and the other on a multiple regression TBK-SM prediction equation. Design: Healthy subjects aged 20 y were recruited for bodycomposition evaluation. TBK and SM were measured by wholebody 40K counting and multislice magnetic resonance imaging, respectively. A conceptual model with empirically derived data was developed to link TBK and adipose tissuefree SM as the ratio of TBK to SM. Results: A total of 300 subjects (139 men and 161 women) of various ethnicities with a mean ( SD) body mass index (in kg/m2) of 25.1 5.4 met the study entry criteria. The mean conceptual modelderived TBK-SM ratio was 122 mmol/kg, which was comparable to the measurement-derived TBK-SM ratios in men and women (119.9 6.7 and 118.7 8.4 mmol/kg, respectively), although the ratio tended to be lower in subjects aged 70 y. A strong linear correlation was observed between TBK and SM (r = 0.98, P < 0.001), with sex, race, and age as small but significant prediction model covariates. Conclusions: Two different types of prediction models were developed that provide validated approaches for estimating SM mass from 40K measurements by whole-body counting. These methods afford an opportunity to predict SM mass from TBK data collected in healthy adults. Am J Clin Nutr 2003;77:7682. KEY WORDS Body composition, nutritional assessment, whole-body counting, total body potassium, skeletal muscle, prediction models, ratio of total body potassium to skeletal muscle INTRODUCTION Skeletal muscle (SM), the largest component of the tissueorgan body-composition level in adult humans, plays an important role in physiologic processes and energy metabolism. Wholebody SM mass is influenced by several modifying biological factors such as age, sex, race, physical activity, and disease (1). Although of growing research and clinical interest (2), SM mass remains a difficult or impractical body component to accurately quantify in living humans. 76 At present, the most accurate in vivo methods of measuring whole-body SM are multislice magnetic resonance imaging (MRI) and computed axial tomography (CT) (2). Although MRI and CT are often used as criterion methods for estimating SM, their application is limited because of expense, lack of instrument access, and unavailability of resources for image analysis. The CT method also exposes subjects to radiation, and CT for research purposes is not often approved by institutional review boards, especially in evaluating healthy children and premenopausal women. Two available field methods for estimating SM are the use of anthropometric measurements and bioelectrical impedance analysis (25). Although noninvasive and inexpensive, these methods are not sufficiently accurate to evaluate individuals or to monitor small changes in muscle mass (4, 5). Two urine collectionbased laboratory methods, one relying on urinary creatinine excretion and the other on urinary 3-methylhistidine excretion, ideally include a 1-wk meat-free diet protocol and 3 consecutive 24-h urine collections (1, 68). The between-day CV for the urinary marker methods approaches 5%, even with the rigorous conditions available in a metabolic ward (6). The limitations of these SM estimation methods led us in the present study to seek an in vivo method for measuring total-body SM mass. Potassium is a measurable element in vivo, and the use of total body potassium (TBK) as an index of body composition has a long history in nutritional research. Body potassium is distributed entirely in the fat-free mass (FFM) compartment, and a large proportion exists in SM. For example, 60% of TBK in reference man is found in SM (9). The ratio of TBK to FFM is 5459 mmol/kg for healthy females and 5962 mmol/kg for healthy males (1, 10), and this relation provides a means of estimating FFM from measured TBK (1, 11). Similarly, SM is distributed in FFM, and the mean ( SD) ratio of SM to FFM is 0.473 0.037 for healthy females and 0.528 0.036 for healthy males (1214). These 2 observations led us to hypothesize that the ratio of TBK to SM may be relatively stable in healthy adults, thus providing the basis for an SM prediction method.

Downloaded from www.ajcn.org by on November 26, 2009

1 From the Obesity Research Center, St Lukes-Roosevelt Hospital, Columbia University College of Physicians and Surgeons, New York. 2 Supported by National Institutes of Health grant PO1-DK42618. 3 Reprints not available. Address correspondence to ZM Wang, Weight Control Unit, 1090 Amsterdam Avenue, 14th Floor, New York, NY 10025. E-mail: zw28@columbia.edu. Received November 9, 2001. Accepted for publication April 9, 2002.

Am J Clin Nutr 2003;77:7682. Printed in USA. 2003 American Society for Clinical Nutrition

MUSCLE MEASUREMENT A similar concept was advanced by Forbes (1), who rst measured TBK by whole-body counting and then calculated FFM according to the assumed constant ratio of TBK to FFM (ie, 87 mmol/kg). Total-body SM was next calculated on the basis of the assumption that SM makes up, on average, 49% of FFM in adult humans (15). However, Forbes was unable to assess the accuracy of TBK-predicted SM estimates because a reference SM measurement method, such as MRI or CT, was unavailable 3 decades ago. Whole-body 40K counting is available at over 30 body-composition centers throughout the world that have large cross-sectional and longitudinal TBK databases. This led us in the present study to develop and validate models associating TBK and SM mass, with the specific aim of providing an approach to the measurement of total-body SM mass.

77

BCM that consists of SM cells and a is the fraction of BCM that consists of ICW. Similarly, SM ECF is a portion of whole-body ECF and can be expressed as SM ECF = n ECF = n ECW/b, where n is the fraction of whole-body ECF that consists of SM ECF and b is the fraction of ECF that consists of ECW. A TBK/AT-free SM model is thus derived as TBK/AT-free SM = (KICF + KECF)/ (SM cells + SM ECF) = ([K]ICW ICW + [K]ECW ECW)/(m ICW/a + n ECW/b)

(4)

Because total body water is made up of ICW and ECW, ECW can be expressed as a function of ICW: ECW = E/I ICW, where E/I is the ratio of ECW to ICW. Equation 4 can be converted and simplified as TBK/AT-free SM = ([K]ICW ICW + [K]ECW ICW E/I)/[m ICW/a + n ICW (E/I)/b] = ([K]ICW + [K]ECW E/I)/(m/a + n/b E/I)

SUBJECTS AND METHODS Protocol Our strategy was to provide 2 alternative means to predict totalbody SM mass from TBK. A conceptual cellular-level TBK/SM model was first developed and then used, with available experimental data, to explore the magnitude of the TBK-SM ratio. We then specifically tested the model predictions by evaluating TBK and SM in a large sample of healthy adults. Under ideal circumstances a stable TBK-SM ratio (k) could be used to predict totalbody SM as SM = 1/k TBK (1)

(5)

When MRI is applied, the observed SM image area in each slice contains a small amount of interstitial AT that cannot be entirely separated out by analysis software. The MRI-measured SM, or anatomical SM, contains AT-free SM and this small amount of interstitial AT within muscle. Equation 5 is thus modified for MRI-measured SM as TBK/MRI-measured SM = (1 f) [([K]ICW + [K]ECW E/I)/(m/a + n/b E/I)] (6)

Downloaded from www.ajcn.org by on November 26, 2009

A second modeling strategy was also applied because we assumed at the outset that the TBK-SM ratio might have influencing factors of unknown magnitude such as age, sex, and race. Accordingly, we developed an SM prediction multiple regression function ( F ) in which total-body SM was set as the dependent variable and TBK as the main predictor variable, SM = F(TBK, age, sex, race) (2)

where f is the fraction of MRI-measured SM that consists of interstitial AT. As shown in Equation 6, the TBK-SM ratio observed by MRI is determined by 8 factors. The mean magnitudes for the 8 determinants are described below. Determinants [K]ICW, [K]ECW, a, b, and E/I In our previous studies, we discussed the physiologic aspects of these 5 determinants (11, 18). The same magnitude and variation range for each determinant was applied in the present study. The cellular and extracellular potassium concentrations are physiologically stable in mammals. Previous studies reported similar [K]ICW and [K]ECW in mammals: 150160, 150 7.2 (SD), and 159 mmol/kg H2O for [K]ICW and 46 mmol/kg H2O for [K]ECW (11). In healthy subjects the mean [K]ICW and [K] ECW are thus assumed to be 155 and 5 mmol/kg H2O, respectively. Determinant a is cellular hydration (ie, ICW/BCM), and determinant b is ECF hydration (ie, ECW/ECF). Cellular hydration is tightly regulated and thus should be minimally variable in healthy adults. We discussed the physiologic basis for the means of these 2 determinants (ie, a = 0.70 with a variation range of 0.690.71 and b = 0.98 with a variation range of 0.970.99) in an earlier report (11). E/I is the ratio of ECW to ICW. We measured E/I by combining 3H2O and sodium bromide dilution and observed means ( SDs) of 0.79 0.13 and 1.03 0.19 for men and women, respectively (18). Determinants m and n Determinant m is the fraction of BCM that consists of SM cells, and determinant n is the fraction of whole-body ECF that consists of SM ECF. Information is limited regarding m and n because of the difculty in quantifying cell mass and ECF within individual tissues in vivo. Therefore, we based the estimates of m and n on data available for reference man (9). In reference man, the amount of TBK is 3580 mmol

where age, sex, and race were entered as potential independent predictor variables after controlling for TBK. TBK/SM model Because almost all body potassium exists within intracellular fluid (ICF) and extracellular fluid (ECF), TBK is equal to the sum of the potassium masses within ICF (KICF) and ECF (KECF), TBK = KICF + KECF (3)

KICF can be expressed as the product of intracellular water (ICW) and ICW potassium concentration ([K]ICW; KICF = ICW [K]ICW). Similarly, KECF is the product of extracellular water (ECW) and ECW potassium concentration ([K]ECW; KECF = ECW [K]ECW). The cellular body-composition level is composed of 3 compartments: cells, ECF, and extracellular solids (16). The cellular compartment can be further divided into fat and body cell mass (BCM), which is defined as a component of body composition containing the oxygen-exchanging, potassium-rich, glucose-oxidizing, work-performing tissue (17). In body-composition studies, 2 SM terms are usually used: adipose tissue (AT)free SM and anatomical SM. AT-free SM is composed of 2 main compartments: SM cells and SM ECF. SM cell mass is the largest contributor to BCM and can be expressed as SM cells = m BCM = m ICW/a, where m is the fraction of

78

WANG ET AL Equations 7 and 8 indicate that the mean magnitude of the TBK-SM ratio should be identical in men and women if there is no large difference in the f value. On the basis of Equations 7 and 8 , f values and the corresponding TBK-SM ratios are shown in Table 1 . As the f value increases from 0 to 0.05, the MRImeasured TBK-SM ratio gradually decreases from 122 to 116 mmol/kg. For AT-free SM (ie, f = 0), a single SM prediction model from TBK can be developed for healthy adults, AT-free SM = 1/122 TBK = 0.0082 TBK Subjects Healthy adult subjects aged 20 y were recruited into this study. To be included in the study, subjects had to be ambulatory and have no orthopedic problems or physical handicaps. Each subject completed a medical history, physical examination, and routine blood studies to exclude the presence of underlying diseases. The Institutional Review Board of St Lukes-Roosevelt Hospital Center approved the study protocol, and all subjects gave informed written consent before participation. About 60% of the subject data came from our earlier body-composition studies (5). The subjects body mass was measured to the nearest 0.1 kg while they were fasting and wearing minimal clothing. Height was measured with a stadiometer to the nearest 0.1 cm. Total body fat and FFM were estimated with the use of dual-energy X-ray absorptiometry (DXA; Lunar DPX, software version 3.6; Lunar Corp, Madison, WI). The between-measurement technical error at our center for FFM measured by DXA is 1.2%. Whole-body counting The St Lukes 4 whole-body counter was used to detect the natural 1.46 MeV -ray of 40K. The 40K raw counts collected over 9 min were adjusted for body size on the basis of an experimental 42K calibration equation (19). The correction equations for men and women are corrected body K (from 42K) = 0.85 body K (from 40K) + 7.6 body mass + 250 (R2 = 0.90; SD = 183 mmol) and corrected body K (from 42K) = 0.88 body K (from 40K) + 6.4 body mass + 64.2 (R2 = 0.88; SD = 143 mmol), respectively, where K is in mmol and body mass is in kg. TBK was calculated as 40K/0.000118 (1). The current technical error in our laboratory for repeated phantom 40K counting is 2.4% (20). Magnetic resonance imaging Total-body SM was measured by using multislice MRI (21). Subjects were placed on the MRI scanner (1.5 T 6X Horizon; General Electric, Milwaukee) platform with their arms extended above their heads. One-centimeter-thick images were taken from the gap between lumbar vertebrae 4 and 5 to the tip of the toes and ngers with a 4.0cm gap between scans. The protocol involved the acquisition of 3545 axial images, depending on height, over the whole body. All MRI scans were analyzed with the use of VECT image analysis software (Tomovision, Montreal) by a group of highly trained observers. The mean ( SD) technical error for betweenday measurements of the same scan by the same observer of MRIderived SM volume is 0.34 1.1% (22). Total-body SM mass was calculated as SM (kg) = 0.00104 [5 (A1 +A2)/2] TBK/SM = (1 f) [(155 + 5 1.03)/ (0.55/0.70 + 0.50 1.03/0.98)] = 122 (1 f) (mmol/kg; for women) (10) (9)

TABLE 1 Relation between determinant f and the MRI-measured ratio of TBK to SM1 Determinant f 0 0.01 0.02 0.03 0.04 0.05
1

MRI-measured TBK-SM ratio mmol/kg 122.0 120.8 119.6 118.3 117.1 115.9

SM prediction model SM = 0.0082 TBK SM = 0.0083 TBK SM = 0.0084 TBK SM = 0.0085 TBK SM = 0.0085 TBK SM = 0.0086 TBK

MRI, magnetic resonance imaging; f, fraction of MRI-measured skeletal muscle that consists of interstitial adipose tissue; SM, whole-body skeletal muscle measured by using multislice MRI; TBK, total body potassium measured by whole-body counting.

(140 g) and the amount of potassium in SM is 2148 mmol (84 g). On the basis of the assumption that all potassium exists within cells, wholebody BCM is 29.8 kg (ie, 0.00833 kg/mmol 3580 mmol) and SM cell mass is 17.9 kg (ie, 0.00833 kg/mmol 2148 mmol) (17). The determinant m is thus 17.9/29.8 = 0.60 for reference man. The total-body SM mass in reference man is 28.0 kg, and the SM ECF mass is equal to the difference between the SM mass and the SM cell mass (ie, 28.0 kg 17.9 kg = 10.1 kg). In reference man, the total-body ECW mass is 18.0 kg, which is equivalent to 18.4 kg of ECF (ie, 18.0 kg/0.98). The determinant n is thus 10.1/18.4 = 0.55 for reference man. The mean m and n values in women should be slightly smaller than those in men. This is because women have a smaller ratio of SM to ATfree mass (ie, 0.470.49 for women compared with 0.530.57 for men) (12, 14). In the present study, we assumed mean values of 0.55 and 0.50 for m and n, respectively, in women. Determinant f Determinant f is the fraction of MRI-measured SM that consists of interstitial AT. The f value is assumed to be < 0.03 in healthy, nonobese subjects. In reference man, for example, the f value is 0.022 (9). For obese subjects, the f value may reach 0.05. Although we usually try to remove interstitial AT pixels from the muscle cross-sectional area when analyzing MRI images, some AT pixels are below the visible detection threshold. However, the remaining interstitial AT pixels must be much smaller relative to the whole muscle area. Magnitude of TBK-SM ratio The mean values of each of the determinants described above are as follows: [K] ICW = 155 mmol/kg, [K] ECW = 5 mmol/kg, a = 0.70, b = 0.98, E/I = 0.79 for men and 1.03 for women, m = 0.60 for men and 0.55 for women, and n = 0.55 for men and 0.50 for women. According to Equation 6, the mean magnitude of the TBK-SM ratio can be calculated as TBK/SM = (1 f) [(155 + 5 0.79)/ (0.60/0.70 + 0.55 0.79/0.98)] = 122 (1 f) (mmol/kg; for men) and

Downloaded from www.ajcn.org by on November 26, 2009

(7)

(8)

where 0.00104 is SM density (kg/cm3) (9), A1 and A2 are SM areas (in cm2) in adjacent scans, and 5 is the distance (in cm) between adjacent scans.

MUSCLE MEASUREMENT
TABLE 2 Baseline characteristics of the subjects (n = 300)1 Men (n = 139) Age (y) Body mass (kg) Height (cm) BMI (kg/m2) Fat (%) 1 x SD.
2,3

79

Women (n = 161) 45.2 16.42 66.5 14.63 162.1 7.53 25.3 5.1 33.2 10.23

39.0 14.6 80.4 13.4 176.5 11.2 25.6 3.9 20.2 8.0

acteristics of the subjects are presented in Table 2. The men were significantly younger (P < 0.01), heavier (P < 0.001), and taller (P < 0.001) than the women. There was no significant difference in body mass index between the sexes, although the men had a significantly lower percentage of fat than did the women (P < 0.001). There were also no significant differences in baseline characteristics between the model-development and cross-validation groups. TBK-SM ratio The men had significantly more TBK and MRI-measured SM than did the women ( P < 0.001 for both). However, there was no significant difference ( P = 0.17) in the TBK-SM ratio between the men and the women (119.9 6.7 compared with 118.7 8.4 mmol/kg, respectively), and both ratios were minimally (< 3%) smaller than the model-derived value of 122 mmol/ kg for AT-free SM (Table 3 ). The TBK-SM ratio was inversely correlated (r = 0.22, P < 0.001; SEE = 7.5 mmol/kg) with age (n = 300), TBK/SM (mmol/kg) = 123.7 0.11 age (11)
Downloaded from www.ajcn.org by on November 26, 2009

Signicantly different from men (Students t test): 2 P < 0.01, 3 P < 0.001.

Statistical analysis Data are expressed as means ( SDs). Statistical comparisons of physical characteristics and body composition between men and women were made by Students t test. Age-related differences in TBK, MRI-measured SM, and the TBK-SM ratio were examined across age and sex groups by one-way analysis of variance. Simple linear regression analysis was used to investigate the associations between TBK and MRI-measured SM. All subjects were randomly separated into 2 groups, a modeldevelopment group and a cross-validation group. Empirical models for predicting MRI-measured SM were developed in the modeldevelopment group by using multiple-regression analysis. TBK served as the major predictor variable, whereas age, sex, race, and the interactions of these variables with TBK were evaluated as covariates in the model. Dummy variables were used for the 4 ethnic groups. A forward-backward stepwise selection procedure was applied to determine significant variables and derive prediction equation models. The prediction equation obtained from the modeldevelopment group was then validated in the cross-validation group by using the approach suggested by Bland and Altman (23). Twotailed tests of signicance (P < 0.05) were used. All analyses were carried out by using SPSS (version 10.0; SPSS Inc, Chicago).

TBK, SM, and TBK-SM ratio values are shown for subjects within 10-y age groups (Table 3). Both TBK and SM were negatively associated with age in the men and the women (P for trend < 0.050.001). In the men there was no significant difference in TBK-SM ratios across the age groups. In the women, one-way analysis of variance showed significant differences in TBK-SM ratios across the age groups (P for trend < 0.01). SM prediction model TBK was the strongest SM predictor identified ( r = 0.98, P < 0.001), explaining 95.9% of the observed between-individual variation (Figure 1 ). A simple prediction equation was thus derived from the 300 subjects. SM (kg) = 0.0085 TBK (mmol) (12)

RESULTS Baseline characteristics A total of 300 healthy adult subjects, of whom 139 were men (35 African Americans, 17 Asians, 63 whites, and 24 Hispanics) and 161 were women (56 African Americans, 18 Asians, 68 whites, and 19 Hispanics), met the study entry criteria. The baseline char-

The SEE of MRI-measured SM estimation by TBK alone was 1.60 kg. The regression line slope (0.0085) is similar to the reciprocal of our model-derived TBK-SM ratio (ie, 1/122 = 0.0082). When age, sex, race, and the interactions of these variables with TBK were entered along with TBK into multiple regression SM prediction equations in the model-building group (n = 150), TBK, sex, race, and age were selected by the stepwise procedure (r2 = 0.97, P < 0.001; SEE = 1.45 kg),

TABLE 3 Total body potassium (TBK), total-body skeletal muscle (SM), and ratio of TBK to SM by sex and age1 Age (y) TBK2 mmol Men SM3 kg TBK/SM mmol/kg TBK4 mmol Women SM3 kg TBK/SM3 mmol/kg 121.3 7.0 119.1 9.0 120.4 8.8 117.4 8.2 117.5 7.9 112.6 6.1 118.7 8.4 < 0.01

2029 (n = 46 M, 30 F) 4110 567 34.3 5.8 120.6 7.1 2405 324 19.9 3.3 3039 (n = 38 M, 41 F) 4141 527 34.9 5.0 119.1 5.3 2504 360 21.2 3.6 4049 (n = 25 M, 32 F) 3877 641 32.4 6.6 120.6 6.9 2599 348 21.7 3.5 5059 (n = 17 M, 28 F) 3782 542 31.5 5.0 120.4 7.9 2442 295 20.9 3.2 6069 (n = 5 M, 14 F) 3490 457 29.3 4.0 119.3 3.6 2187 267 18.7 2.3 70 (n = 8 M, 16 F) 3343 388 28.9 4.4 116.6 8.2 2130 306 19.0 2.8 Total (n = 139 M, 161 F) 3970 592 33.3 5.7 119.9 6.7 2429 354 20.6 3.4 P for trend < 0.001 < 0.01 NS < 0.001 < 0.05 1 x SD. SM mass was measured by using multislice magnetic resonance imaging; TBK was measured by whole-body counting.
24

Signicant differences across the 6 age groups (one-way ANOVA): 2 P < 0.01, 3 P < 0.05, 4 P < 0.001.

80

WANG ET AL TBK-SM ratio model and the other based on an empirical multiple regression model. Both approaches appear to provide a good means of estimating total-body SM in healthy adults. Ratio model This approach for predicting SM mass is based on the concept that TBK is measurable as 40K and the assumption of a stable and known TBK-SM ratio in healthy adults. When combined with empirically derived data, our conceptual model suggests a sexindependent TBK-SM ratio of 122 mmol/kg. AT-free SM can thus be estimated as 0.0082 TBK. This modeling approach is similar to the use of 2 classic ratios in predicting FFM: total body water/FFM and TBK/FFM (11, 13). Our derived TBK-SM ratio of 122 mmol/kg differs by < 3% from the observed mean values of 120.1 and 119.4 mmol/kg for men and women, respectively, aged < 70 y. If we assume for discussion purposes that the fraction of SM that consists of interstitial AT is 0.03, the model-derived TBKSM ratio would be equal to 119.6 mmol/kg, almost identical to the observed values. The small difference in the measured TBK-SM ratio between men and women (ie, 0.5%) is probably due to sex differences in the content of interstitial AT within MRI-measured SM (Table 1). These observations suggest the appropriateness of 2 SM prediction models in subjects aged < 70 y: AT-free SM = 0.0082 TBK, and MRI-observed SM = 0.0083 TBK. Our observed results indicating a relatively stable TBK-SM ratio for women aged < 70 y (119.4 8.3 mmol/kg) were accompanied by a significantly lower ratio in women aged 70 y (112.6 6.1 mmol/kg; P < 0.01). There are at least 2 biological observations that might explain a lower TBK-SM ratio in elderly subjects. First, the amount of interstitial AT is well known to increase with age (24, 25). Second, skeletal muscle per se atrophies with age, leading to a loss of potassium-rich myofiber mass and a relative expansion of connective tissue and ECF (26). Both a relative increase in intramuscular AT and a lowering of SM cell mass would reduce the overall TBK-SM ratio. Additional studies are needed to test this hypothesis. Whatever the mechanism, the simple SM prediction model based on the TBK-SM ratio would need modication for subjects aged 70 y. Appropriate TBK/SM coefficients are presented in Table 3 by age group. Several investigators (27, 28) made a similar observation of an age-related lowering of the TBK-FFM ratio. There are 2 error sources to consider with the ratio model approach. These are measurement error and model error. Measurement error TBK assessment is the only source of measurement error in the ratio model method. The error caused by TBK assessment can be evaluated in the subjects in the present study by assuming a mean body composition as shown in Table 2 and a TBK measurement precision of 2.4% as described in Subjects and Methods, SM = 0.0082 3970 0.024 = 0.78 kg (for men) and SM = 0.0082 2429 0.024 = 0.48 kg (for women) (15) The SM estimation error in this approach is thus 0.51.0 kg in healthy adults. Model error The suggested constancy of the TBK-SM ratio is based on 3 assumptions. First, we assume that the ratio of SM to other potassium-rich tissues and organs is stable. For instance, a large (14)

FIGURE 1. Total-body skeletal muscle (SM) measured by magnetic resonance imaging plotted against total body potassium (TBK) measured by whole-body 40K counting. SM (kg) = 0.0085 TBK (mmol) (r = 0.98, P < 0.001; SEE = 1.60 kg; n = 300).

SM (kg) = 0.0093 TBK (mmol) 1.31 sex + 0.59 black + 0.024 age 3.21

Downloaded from www.ajcn.org by on November 26, 2009

(13)

where sex is 0 for women and 1 for men, and black is 1 for African Americans and 0 for the other 3 ethnic groups. When Equation 13 was applied in the cross-validation group (n = 150), a significant correlation was observed between the measured and predicted SM (r2 = 0.96, P < 0.001; SEE = 1.52 kg). The predicted SM for the cross-validation group was 27.0 7.8 kg (95% CI: 11.4, 42.6 kg), which was very similar to the MRI-measured SM value of 26.8 7.9 kg (95% CI: 11.1, 42.5 kg). Bland-Altman analysis (23) failed to disclose a significant bias in prediction for the crossvalidation group (r2 = 0.003, P = 0.52) (Figure 2).

DISCUSSION In the present study we examined 2 different strategies for predicting total-body SM mass from measured 40K, one based on a

FIGURE 2. Bland-Altman (23) analysis of the cross-validation group. The difference between skeletal muscle (SM) measured by magnetic resonance imaging (MRI) and SM predicted by Equation 13 plotted against the MRImeasured and predicted SM mean. The line describing the data is expressed by the following equation: y = 0.0102x 0.54 (r = 0.052, P = 0.52; n = 150). The 95% CI is indicated by the upper and lower lines.

MUSCLE MEASUREMENT
TABLE 4 Comparison of different methods of predicting whole-body skeletal muscle mass1 Method Main predictor SEE kg Anthropometric method (n = 244) BIA (n = 388) Anthropometic method (n = 244) 3-MH (n = 10) Creatinine (n = 12) DXA (n = 414) Whole-body 40K counting (n = 300) Body mass and height Height2/R Skinfold thicknesses and circumferences 24-h Urinary 3-methylhistidine 24-h Urinary creatinine Appendicular lean soft tissue TBK 2.8 2.7 2.2 2.3 1.9 1.6 1.5 0.86 0.86 0.91 0.77 0.85 0.96 0.96 MRI MRI MRI CT CT MRI MRI r2 Criterion

81

Reference

5 4 5 8 7 29 Present study

1 3-MH, 3-methylhistidine; BIA, bioelectrical impedance analysis; CT, computed axial tomography; DXA, dual-energy X-ray absorptiometry; MRI, magnetic resonance imaging; R, resistance; TBK, total body potassium.

portion of body potassium exists within SM (ie, 60% for reference man), and the remaining 40% of potassium is present in non-SM components including organs, skin, AT, and skeleton (9). Second, we assume that the ratio of SM cells to SM ECF is stable. Third, we assume that the potassium concentration (ie, 155 mmol/kg H2O) is stable in SM cells and in other tissue and organ cells. Any variation in the 3 assumed stable ratios between subjects will cause a corresponding change in the TBK-SM ratio. Our observations across age groups indicate that the TBK-SM ratio is relatively stable up to an age of 70 y. Although our sample aged 70 y was small, the lower observed TBK-SM ratio suggests that 1 of the 3 assumptions may be invalid in the elderly. Empirical model The second approach was to link SM with TBK via a multiple regression prediction model and then cross-validate the model in another group. Our in vivo measurements showed that TBK is the strongest predictor and that alone it explains 95.9% of the observed between-individual variation of MRI-measured SM mass (ie, Equation 12). Of the easily acquired biological factors, age, sex, and race also had a small influence on SM prediction, after TBK was controlled for rst. Accordingly, a multiple regression equation (Equation 13) was derived for predicting SM mass on the basis of TBK and these other easily acquired prediction variables. Adding age, sex, and race to TBK, Equation 13 explained 97% of the observed between-individual variation in MRI-measured SM mass. The SEE for the TBK-SM multiple regression prediction method (ie, 1.45 kg) was lower than that for Equation 12 (ie, 1.60 kg), which predicts SM by TBK alone. As with the ratio model, the simple linear regression formula (Equation 12 ) should perform well in subjects aged < 70 y, although the complete validated model (Equation 13) with a lower SEE is preferred. The multiple regression model is also preferred when evaluating SM in subjects aged 70 y, but further studies with larger numbers of elderly subjects are needed for crossvalidation purposes. Comparison between SM prediction methods MRI and CT are now applied as the criterion methods for measuring total-body SM mass. Other alternative methods include neutron activation analysis, anthropometric methods, bioelectrical impedance analysis, 24-h urinary creatinine and 3methylhistidine excretion, and DXA-measured appendicular lean soft tissue mass (Table 4). On the basis of potassium-to-nitrogen ratios of SM and non-SM lean tissue, Burkinshaw et al (30) suggested a model for estimating total-body SM mass. However, this

method underestimates SM by an average of 6.9 kg or 20.1% (P = 0.0001) in healthy men (31). The remaining 7 alternative methods can be empirically organized into 3 groups. Group I, which consists of potential field methods, includes bioelectrical impedance analysis (4) and 2 anthropometric methods (5), one based on body mass and height and the other on limb skinfold thicknesses and circumferences. Each of these 3 methods is simple and can be readily applied in field settings. However, the SEEs of these methods are large (2.22.8 kg), indicating low estimation accuracy for individual subjects. The methods in group II, which are urinary marker methods, are based on 24-h urinary creatinine and 3-methylhistidine excretion (7, 8). These 2 methods have intermediate SEEs (1.9 kg). However, both methods are impractical because they require a meat-free diet for 7 d and complete 24-h urine collections during the final 3 d of the diet. These 2 methods are therefore not suitable for routine measurement and field studies. Group III, which consists of nonurinary laboratory methods, include a DXA method (29) and the currently suggested TBK method. Both the DXA and TBK methods have similar low SEEs (1.51.6 kg) and high r2 values (0.96). No radiation is administered with the whole-body counting TBK method, whereas DXA involves a small radiation dose, 110% of that of a chest X-ray. Both methods are thus convenient and safe for most subjects and are routinely approved by institutional review boards for both women and children. An additional advantage of the DXA method is that it may provide regional SM estimates (32). About 30 whole-body counters are now available worldwide, and 11 of them are in the United States. The measurement of TBK by whole-body counting is rapid, convenient, and simple to carry out. This method can be applied in subjects of any age because no radiation is involved and no active subject participation is necessary. In addition to predicting whole-body SM, TBK measurements can also be applied to predict BCM, an important body component at the cellular level (17). Although the 4 whole-body counting facilities are very heavy and cannot be easily moved, the so-called shadow shield counter is much lighter than completely shielded facilities and can be applied in eld studies (1). Conclusion The present research was initially prompted by the need for improved SM estimation methods that are safe and practical. The long-considered potential of whole-body 40K counting in evaluating total-body SM mass was critically examined, and a theoretical model was fitted with existing data, suggesting a sex-independent TBK-SM ratio of 122 mmol/kg. This ratio and its variability were

Downloaded from www.ajcn.org by on November 26, 2009

82

WANG ET AL
15. Forbes GB, Bruining GJ. Urinary creatinine excretion and lean body mass. Am J Clin Nutr 1976;29:135966. 16. Wang ZM, Pierson RN Jr, Heymsfield SB. The five-level model: a new approach to organizing body composition research. Am J Clin Nutr 1992;56:1928. 17. Moore FD, Olesen KH, McMurray JD, Parker HV, Ball MR, Boyden CM. The cell mass and its supporting environment. Philadelphia: Saunders, 1963. 18. Wang ZM, Deurenberg P, Wang W, Pietrobelli A, Baumgartner RN, Heymsfield SB. Hydration of fat-free body mass: new physiologic modeling approach. Am J Physiol 1999;276:E9951003. 19. Pierson RN Jr, Wang J, Thornton JC, Van Itallie TB, Colt EW. Body potassium by four-pi 40K counting: an anthropometric correction. Am J Physiol 1984;246:F2349. 20. Gallagher D, Belmonte D, Deurenberg P, et al. Organ-tissue mass measurement allows modeling of REE and metabolically active tissue mass. Am J Physiol 1998;275:E24958. 21. Ross R, Rissanen J, Pedwell H, Clifford J, Shragge P. Influence of diet and exercise on skeletal muscle and visceral adipose tissue in men. J Appl Physiol 1996;81:244555. 22. Mitsiopoulos N, Baumgartner RN, Heymsfield SB, Lyons W, Ross R. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J Appl Physiol 1998; 85:11522. 23. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods for clinical measurement. Lancet 1986;8: 30710. 24. Lexell J. Human aging, muscle mass, and fiber type composition. J Gerontol 1995;50:116. 25. Lexell J, Taylor CC, Sjstrm M. What is the cause of the ageing atrophy? J Neurol Sci 1988;84:27594. 26. Tomlinson BE, Walton JN, Rebeiz JJ. The effects of ageing and of cachexia on skeletal muscle: a histopathological study. J Neurol Sci 1969;9:32146. 27. Kehayias JJ, Fiatarone MA, Zhuang H, Roubenoff R. Total body potassium and body fat: relevance to aging. Am J Clin Nutr 1997;66: 90410. 28. Pierson RN Jr, Lin DHY, Phillips RA. Total-body potassium in health: effects of age, sex, height, and fat. Am J Physiol 1974;226: 20612. 29. Kim J, Wang Z, Heymsfield SB, Baumgartner RN, Gallagher D. Total-body skeletal muscle mass: estimation by a new dual-energy X-ray absoptiometry method. Am J Clin Nutr 2002;76:37883. 30. Burkinshaw L, Hill GL, Morgan DB. Assessment of the distribution of protein in the human body by in vivo neutron activation analysis. In: International symposium on nuclear activation techniques in the life sciences. Vienna: International Atomic Energy Association, 1978: 78798. 31. Wang ZM, Visser M, Ma R, et al. Skeletal muscle mass: evaluation of neutron activation and dual-energy X-ray absorptiometry methods. J Appl Physiol 1996;80:82431. 32. Wang W, Wang ZM, Faith MS, Kotler D, Shih R, Heymsfield SB. Regional skeletal muscle measurement: evaluation of new dualenergy X-ray absorptiometry model. J Appl Physiol 1999;87: 116371.

explored as a prelude to developing a TBK-SM prediction formula that includes easily acquired variables such as age, sex, and race. The present study did not include children or patients with underlying diseases, and thus future validation studies are needed to evaluate the relation between TBK and SM in populations other than healthy adults.

REFERENCES
1. Forbes GB. Human body composition. New York: Springer-Verlag, 1987. 2. Lukaski H. Estimation of muscle mass. In: Roche AF, Heymsfield SB, Lohman TG, eds. Human body composition. Champaign, IL: Human Kinetics, 1996:10928. 3. Elia M, Fuller NJ, Hardingham CR, et al. Modeling leg sections by bioelectrical impedance analysis, dual-energy X-ray absorptiometry, and anthropometry: assessing segmental muscle volume using magnetic resonance imaging as a reference. Ann N Y Acad Sci 2000;904: 298305. 4. Janssen I, Heymsfield SB, Baumgartner RN, Ross R. Estimation of skeletal muscle mass by bioelectrical impedance analysis. J Appl Physiol 2000;89:46571. 5. Lee RC, Wang ZM, Heo M, Ross R, Janssen I, Heymsfield SB. Totalbody skeletal muscle mass: development and cross-validation of anthropometric prediction models. Am J Clin Nutr 2000;72:796803. 6. Heymsfield SB, Arteaga C, McManus C, Smith J, Moffitt S. Measurement of muscle mass in humans: validity of the 24-h urinary creatinine method. Am J Clin Nutr 1983;37:47894. 7. Wang ZM, Gallagher D, Nelson M, Matthews DE, Heymsfield SB. Total-body skeletal muscle mass: evaluation of 24-h urinary creatinine excretion by computerized axial tomography. Am J Clin Nutr 1996;63:8639. 8. Wang ZM, Deurenberg P, Matthews DE, Heymsfield SB. Urinary 3-methylhistidine excretion: association with total-body skeletal muscle mass by computerized axial tomography. JPEN J Parenter Enteral Nutr 1998;22:826. 9. International Commission on Radiological Protection. Report of the Task Group on Reference Man. Oxford, United Kingdom: Pergamon Press, 1975. 10. Ellis KJ. Whole-body counting and neutron activation analysis. In: Roche AF, Heymsfield SB, Lohman TG, eds. Human body composition. Champaign, IL: Human Kinetics, 1996:4562. 11. Wang ZM, Pi-Sunyer FX, Kotler DP, Wang J, Pierson RN Jr, Heymsfield SB. Magnitude and variation of total-body potassium to fat-free mass ratio: a cellular level modeling study. Am J Physiol 2001;281:E17. 12. Clarys JP, Martin AD, Drinkwater DT. Gross tissue weights in the human body by cadaver dissection. Hum Biol 1985;56:45973. 13. Wang ZM, Deurenberg P, Wang W, Heymsfield SB. Proportion of adipose tissuefree body mass as skeletal muscle: magnitude and constancy in men. Am J Hum Biol 1997;9:48792. 14. Wang ZM, Heo M, Lee RC, Kotler DP, Withers RT, Heymsfield SB. Muscularity in adult humans: proportion of adipose tissuefree body mass as skeletal muscle. Am J Hum Biol 2001;13:6129.

Downloaded from www.ajcn.org by on November 26, 2009