This study aimed to investigate the validity of using segmental bioelectrical impedance (BI) analysis for estimating skeletal muscle volume (MV) in the trunk, defined as the body segment from the acromion process to the greater trochanter. Using a magnetic resonance imaging (MRI) method, the trunk MV was determined in 28 men (19∼34 yr), divided into validation (n = 20) and cross-validation (n = 8) groups, and used as a reference (MVMRI). For BI measurements of the trunk, the source electrodes were placed at the dorsal surface of the third metacarpal bone of both hands and the dorsal surface of the third metatarsal bone of both feet, and the detector electrodes were placed at the acromion process of both shoulders and the greater trochanter of both femurs. Using this arrangement, the BI values of five parts of the trunk, both sides of the upper region, the middle region, and both sides of the lower region, were obtained and then used to calculate the whole trunk BI value and BI index (BI indexTR). In the validation group, a simple regression analysis of the relationship between BI indexTR and MVMRI showed a significant correlation between the two variables (r = 0.884, P < 0.05) and produced a prediction equation with a SE of estimation of 1,020.3 cm3 (8.5%). In the validation and cross-validation groups, there were no significant differences between the measured and estimated MV without systematic errors. These findings indicate that the segmental BI analysis employed in the present study can be used to estimate trunk MV.
- human body composition
- magnetic resonance imaging
- muscle distribution
- cross validation
the trunk, defined as the part of the body from the acromion process to the greater trochanter, contains about one-half of the body's mass and lean tissue mass (6, 22). Skeletal muscles located in the trunk play an important role in moving and stabilizing the upper body during various movements in daily life or sport (17, 18, 23). Moreover, these muscles are sensitive to inactivity (1) and aging (16, 19), and their quantitative profiles are assumed to be associated with the occurrence of lower back pain (14, 25). Therefore, establishing a method for determining accurately and conveniently the muscularity of the trunk would help us to evaluate physical resources in relation to physical performance in daily life and/or sporting activities.
At present, computerized axial tomography and magnetic resonance imaging (MRI) of multiple sections of the body are widely used to determine human skeletal muscle volume (MV). On the other hand, there is increasing interest in the use of bioelectrical impedance (BI) analysis to assess body composition. Since Organ et al. (22) developed various electrode combinations for determining the BI of every body segment, several attempts have been made to examine the applicability of segmental BI analysis for estimating limb MV through comparison with MV determined by MRI (3, 20, 21). These previous studies reported that the use of segmental BI analysis enabled estimates of limb MV with an accuracy of 6.1∼10.4% in terms of the SE of the estimate (SEE) (3, 20, 21). However, no study has tried to investigate the validity of using BI analysis to predict trunk MV.
The studies cited above have employed segmental BI analysis to estimate limb MV by assuming the limbs to be one cylindrical conductor with a uniform cross-sectional area (CSA). The distribution of the tissues making up the trunk, however, is not so simple (12). Chumlea et al. (12) suggested that the complexity of the internal structure of the trunk would lead to a highly specific resistivity in this part of the body. Baumgartner et al. (5) examined the relationship between MV and bioelectrical resistance in every body segment and indicated that the BI determined for the trunk contributed less to the equation for predicting total body MV, despite a large mass in this region. On the other hand, Organ et al. (22) and Cornish et al. (13) arranged the positions of electrodes so as to divide the trunk into regions by taking the structural complexity of the trunk into account and tried to measure the BI of each region. However, neither study examined whether this approach is applicable for predicting trunk MV.
From the findings of Abe et al. (2), the distribution of the skeletal muscle CSA in the trunk is not as simple as that in the limbs for either gender. In their study, the skeletal muscle CSA measurements per slice in the trunk showed three peaks and troughs. In addition, the trunk involves visceral tissue such as the heart, lung, gastrointestinal tract, and urinary bladder. In the present study, therefore, we assumed the trunk to be a collection of five cylinders, each of which corresponds to a region having a predominant distribution of skeletal muscle or visceral tissues. On the basis of this assumption, the present study tried to determine the BI value of each of the five regions with the use of a segmental BI analysis. We hypothesized that the method used in the present study could measure BI reflecting the distribution of trunk skeletal muscle and so predict trunk MV with an accuracy similar to that previously reported for limb MV. The main purpose of the present study was to investigate whether this approach can be validated and cross-validated by comparing the data obtained using MRI.
Twenty-eight healthy men (19∼34 yr) voluntarily participated in this study. Thirteen of the subjects were athletes (8 American football players, 3 power lifters, 1 weight lifter, and 1 triathlete) who had participated in competitive meets in their own events at the college level within a year preceding the measurements. The rest were either sedentary or mildly active, but none was currently involved in any type of exercise program (≥30 min/day, ≥2 days/wk). To confirm the cross validity of the predicting equation, the subjects were randomly separated into a validation group (n = 20) and a cross-validation group (n = 8), in which the percentage of the number of athletes to the total number of subjects was almost the same: 10 athletes in the validation group and 3 athletes in the cross-validation group. Physical characteristics of each subject group are listed in Table 1. Data for the athletes were collected during preseason training. Therefore, none of the athletes was dehydrated to control his body mass for competition. All measurements for the athletes were performed >40 h after completion of a training session. This study was approved by the ethics committee of the Department of Life Sciences, Graduate School of Arts and Sciences, University of Tokyo, and was consistent with their requirements for human experimentation. The subjects were fully informed about the procedures and the purpose of this study. Written, informed consent was obtained from all participants.
Body height was measured to the nearest 0.1 cm on a standard physician's scale. Body mass was measured to the nearest 0.1 kg on a calibrated electric scale.
With the use of MRI scans with a body coil (Airis, Hitachi Medco, Japan), a series of transverse images from the acromion process of both shoulders to the greater trochanter of both femurs were obtained. The distance between the acromion process of the right shoulder and the greater trochanter of the right femur was defined as the length of the trunk (LTR). The 0% LTR corresponds to the level of the acromion process, and 100% LTR to the greater trochanter. The image condition was T1 weighted, spin-echo, multislice sequences, with a slice thickness of 10 mm and a slice interval of 20 mm, with a repetition time of 200 ms and an echo time of 20 ms. Each subject lay supine in the body coil with his arms and legs extended and relaxed. From each cross-sectional image, outlines of tissues (skeletal muscle, subcutaneous fat, bone, and others) were traced and digitized by personal computer (Power Macintosh G4, Apple) to calculate the anatomical CSA of every tissue. Adipose and tendinous tissues, which were imaged in different tones from the muscle tissue, were excluded when digitizing. By summing the anatomical skeletal muscle CSA and then multiplying the sum by the interval of 20 mm, MV was determined and referred to as MVMRI. As described in a prior study (5), the skeletal muscle in the trunk was separated from limbs by using slices between specific landmarks: the acromion process of the shoulder and the greater trochanter of the femur. Therefore, some muscles located in the shoulders or gluteal (i.e., triangular or gluteal muscle) were partially analyzed as part of MVMRI. The test-retest variability of MVMRI was assessed with 10 men (22∼26 yr) on 2 separate days. The intraclass correlation coefficient for the test-retest measurements was 0.990, and the coefficient of variation (%CV) was 1.8%. There was no significant difference between the mean values of the two tests. Again, the intraobserver reproducibility was assessed by analyzing the trunk MRI images of five men (22∼26 yr) two times. The intraclass correlation coefficient and the %CV of MVMRI values from the two trials were 0.951 and 2.9%, respectively. There was no significant difference between the mean values of the two trials.
A BI acquisition system (Muscle α, Art Haven 9) and disposable electrodes (Red Dot 2330, 3M) were used to determine the BI value of the trunk. The measured BI value was referred to as Z. This system applies a constant current of 500 μA and frequency of 50 kHz through the body. The trunk Z measurements were performed on different days from the MRI measurements with an interval of 1 or 2 days. The subjects refrained from vigorous exercise and alcohol intake for 24 h and from taking a meal for 4 h preceding the experiments. All Z measurements were carried out in the supine position, the arms relaxed at the side but not touching the body, and the legs separated at least 25.0 cm at the ankles so that there was no contact between the thighs. Subjects were instructed to keep breathing quietly because the respiratory cycle affected Z (8). During the measurements, the room temperature was usually kept at 23°C (10).
The source electrodes were placed at the dorsal surface of the third metacarpal bone of both hands and the dorsal surface of the third metatarsal bone of both feet. The detector electrodes were placed at the acromion process of both shoulders and the greater trochanter of both femurs. The combination of electrodes used in this study makes it possible to separate the trunk into five parts and to determine the Z of each of the parts (13, 22). In a pilot study, using one adult man aged 26 yr, we confirmed the dividing point of each voltage measurement area by measuring the electric potential pattern in the trunk. The positions of the source electrodes were fixed while the positions of the detector electrodes were crisscrossed every 5 cm. Figure 1 shows a schematic representation of the electrical potential pattern in each of the anterior (Fig. 1A) and posterior (Fig. 1B) trunk. The gray areas indicate two equipotential zones of the acromion process and the greater trochanter, respectively, generated with each zero potential at each detector electrode. In both the anterior and posterior trunk, the former two equipotential zones cross each other in the upper part in the center line of the trunk, just above the xiphoid process (∼30% LTR) and the latter ones, in the lower part in the center line of the trunk, just below the iliac crest (∼90% LTR). From this result, we considered that the arrangement of the electrodes used in the present study was able to separate the trunk BI network into five regions, as shown in Fig. 1C: upper right (ZTRur) and upper left trunk Z (ZTRul) at both sides of the upper region, middle trunk Z (ZTRm), in the middle region, and lower right (ZTRlr) and lower left trunk Z (ZTRll) at both sides of the lower region.
The whole trunk Z (ZTR whole) can be calculated by the following equation using each Z measurement: The BI index (BI indexTR) was calculated as follows: The test-retest variability of the Z and BI index was assessed with 13 men (22∼30 yr) on 2 separate days. The intraclass correlation coefficients were 0.839∼0.922 for each Z and 0.929 for BI indexTR. The %CV was 2.4∼2.9% for each Z and 3.0% for BI indexTR. There were no significant differences in each Z and BI indexTR between the two tests.
Descriptive values were presented as means (SDs). In the validation group, first, a simple regression analysis was applied to develop a prediction equation for MV in the trunk with BI indexTR as an independent variable. The estimated MV was referred to as MVBI. Second, it was confirmed that the regression slope and intercept for the relationship between the MVMRI and MVBI values did not significantly differ from 1 and 0, respectively. The SEE was calculated to evaluate the accuracy of MVBI. The SEE was expressed as an absolute value and relative to the mean of MVMRI. Third, the difference between MVMRI and MVBI was plotted against the mean MV of the two methods to examine for systematic error, as described by Bland and Altman (7). When the three conditions mentioned above were satisfied, the predicted values of MV were calculated for individuals in the cross-validation group by using the equation derived from the validation group. In the cross-validation group, the significance of the difference between MVMRI and MVBI and the existence of systematic error were tested by using Student's paired t-test and a Bland-Altman plot (7), respectively. The probability level for statistical significance was set at P < 0.05.
Baseline characteristics of the validation and cross-validation groups.
Table 1 shows the descriptive data on physical characteristics and MRI-measured tissue volumes in the validation and cross-validation groups. There were no significant differences between the two groups in any variables except for body mass. Moreover, no significant differences were found between the groups in the measured Z and BI indexTR (Table 2).
Figure 2 shows the distribution of the measured tissue volume along LTR, calculated in each of 10 divisions of LTR (0–10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 71–80, 81–90, and 91–100% LTR). The skeletal MVs were significantly greater at ∼20% LTR and ∼81% LTR than at the other slice levels. The visceral tissue volume was significantly greater at 21∼60% LTR than at the other slice levels. The bone, other tissue, and subcutaneous fat volumes were greatest at 91–100, 0–10, and 91–100% LTR, respectively. At ∼20% LTR and ∼61% LTR, skeletal MVs were significantly larger than the other tissue volumes analyzed. At the slice levels of 21∼50% LTR, however, the visceral tissue volume was significantly greater than the skeletal MV. On the other hand, subcutaneous fat volume became significantly greater than visceral tissue, bone, and other tissue volumes at 71–80% LTR.
Prediction equation derived from the validation group.
BI indexTR was significantly correlated to MVMRI (r = 0.844, P < 0.05, Fig. 3) in the validation group. This relationship produced an equation, MVMRI = 108.2 × BI indexTR − 402.2, with R2 and SEE values of 0.713 and 1,020.3 cm3 (8.5%), respectively. Regression analysis indicated that the slope and intercept of the regression equation for the relationship between MVMRI and MVBI were not significantly different from 1 and 0, respectively (Fig. 4A). There was no significant difference between MVMRI [11,949.5 cm3 (SD 2,255.8)] and MVBI [11,946.9 cm3 (SD 1,908.5)]. Moreover, no significant systematic error was found in the Bland-Altman plot (Fig. 4B).
Cross validation of the prediction equation.
The prediction equation derived from the validation group was used to estimate MVMRI in the cross-validation group. There was no significant difference between MVMRI [10,733.3 cm3 (SD 1,284.0)] and MVBI [10,566.5 cm3 (SD 1,245.1)]. In addition, no significant systematic error was found in the Bland-Altman plot (Fig. 5).
In a comparison with MRI data, the segmental BI analysis used in this study was validated and cross validated for estimating trunk MV. Some studies have already tried to determine the Z value of the trunk and used it to develop prediction equations for the lean tissue mass or MV of the total body (4, 8, 11, 12). However, some problems exist with the previous approaches used to determine trunk Z. First, the trunk Z has been determined together with limb Z using the network circuit model in which the trunk was connected in series with the limbs (11). As a result, trunk Z was affected by limb composition, because a limb has a smaller CSA and simpler composition than the trunk (6, 8). Furthermore, the previous studies assumed the trunk to be one cylinder (4, 8, 12), regardless of the complexity of the tissue composition in this region. These points have been considered to be the reasons why the trunk Z contributes less to the prediction equation of the lean tissue mass or MV of the total body (4, 8, 11, 12), despite the fact that the trunk contains about one-half of the body's mass and lean tissue mass (6, 22). To solve these problems, therefore, we took the trunk to be a collection of five cylinders, each of which represents a region having a predominant distribution of skeletal muscle or visceral tissues and determined Z in each of them, i.e., ZTRur, ZTRul, ZTRm, ZTRlr, and ZTRll, separated from the limbs. The use of BI indexTR as an independent variable produced a prediction equation for MVMRI with a SEE of 8.5% in the validation group. The observed SEE value was similar to that (6.1∼10.4%) reported in previous studies, which estimated MV in the limbs by BI analysis (3, 20, 21). This implies that, with a similar accuracy in the prediction of limb MV, the segmental BI analysis used in the present study enables us to estimate trunk MV.
Although the present result supports the applicability of segmental BI analysis for predicting trunk MV, we should comment on the limitations of the approach used here. First, the voltage measurement area of each of the measured Z values does not always reflect the distribution of the skeletal MV in the trunk, especially that in the lower parts. As shown in Fig. 1, the Z measurement areas were summarized to the upper (ZTRur and ZTRul), middle (ZTRm), and lower (ZTRlr and ZTRll) regions, based on the lines indicating equipotential zones crossover to each other at upper (∼30% LTR) and lower (∼90% LTR) parts in the center line of the trunk. In the distribution of tissue volume along LTR (Fig. 2), the skeletal MV was significantly greater at the slice levels of ∼20% LTR and ∼81% LTR than at the other slice levels. In the two ranges of the slice levels, skeletal MV was significantly larger than the other tissue volumes. There were no significant differences between the skeletal MVs at slice levels of 41∼80% LTR. Hence, the two equipotential zones that cross over in the upper part of the trunk at ∼30% LTR can be considered valid for reflecting the difference in the distribution of skeletal MV between the slice levels of ∼20% LTR and ∼41% LTR to the Z measurements. However, the Z measurement of the lower region does not involve the slice level of 81–90% LTR, at which skeletal MV showed a second peak among the slice levels. The reason for the discrepancy is unknown but might be related to the anisotropic effects that skeletal muscle fibers can have on Z (5). Namely, it is speculated that the morphological and architectural profiles of each skeletal muscle group located in the lower part of the trunk influence the determination of the Z measurement area of this region. In any case, we have no data to examine this assumption. Further study is warranted to elucidate the voltage measurement area, determined by the electrode combinations used here, with relation to not only the distribution but also morphological and architectural profiles of individual skeletal muscle groups located in the trunk.
The second of the limitations is that the approach used in the present study cannot exclude the influence of the visceral tissue volume on the MV estimates. Among the slice levels involved in the upper region, visceral tissue volume was significantly greater than skeletal MV at 21–30% LTR. However, it should be noted that most of the visceral tissue involved at the slice levels corresponding to the upper region is lung, which is a dielectric tissue (6). Hence the influence of the visceral tissue volume in this region on the MV estimate can be ignored. Meanwhile, the visceral tissue involved in the slice levels corresponding to the middle region is mainly gastrointestinal tract, which is made up of smooth muscles and contains much water, which has high conductivity (6). Considering these points, it is reasonable to assume that the visceral tissue involved at slice levels of >31–40% LTR has an electrically different effect on the Z measurement from that at slice levels of <21–30% LTR. Notably, the visceral tissue volumes at 31∼50% LTR were significantly greater than the skeletal MVs at the corresponding levels. This makes it difficult to separate skeletal muscle from visceral tissue electrically. In the data obtained from the validation group, if one examines the relationship between the percentage of the visceral tissue volume to the skeletal MV at every slice level and the residuals of MV estimates, expressed as a percentage of MVMRI, a low but significant negative correlation (r = −0.447, P < 0.05) is found in the corresponding value at 41–50% LTR (Fig. 6). This implies that the influence of visceral tissue volume on the accuracy of predicting MV in the trunk is not negligible.
Furthermore, another factor that might affect the accuracy of predicting trunk MV is the volume of fat in this region. Fat tissue has a high resistivity in general (9). However, Baugartner et al. (5) reported that the ratio of subcutaneous fat volume to skeletal MV has a slight but significant effect on Z measurements in obese women. In addition, Tagliabue et al. (24) indicated that fat volume had a significant effect on the accuracy of predicting lean tissue mass by BI analysis. The subjects examined here were young and nonobese. Therefore, a slice level at which subcutaneous fat tissue volume was greater than skeletal MV was not found. In addition, there was no slice level at which the percentage of subcutaneous fat tissue volume to skeletal MV was significantly correlated to the percentage of residuals of the MV estimates. However, Bracco et al. (8) reported that the trunk acts as a reservoir for fat accumulation in overweight individuals. In addition, it is known that visceral fat shows a greater increase than subcutaneous fat with aging (15). To apply the BI analysis used in this study to individuals with a high percentage of body fat and/or the elderly, therefore, the influence of visceral fat volume on MV estimates must be established.
In summary, the findings obtained here indicate that segmental BI analysis can be used to estimate trunk MV. However, the present study examined only young and nonobese men. In addition to the development of the best approach for reducing the effects of visceral tissues on the estimation of trunk MV, further investigation is needed to examine the application of the BI analysis used in this study to women, the elderly, the obese, and/or individuals hospitalized long term.
We thank Dr. Masamitsu Ito for help with MRI measurements.
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.
- Copyright © 2006 the American Physiological Society