Heymsfield et al. [1] first showed in adults that ALM, the lean tissue mass in the arms and legs, when measured using radioisotopebased dualphoton absorptiometry (DPA) was strongly related to both total body potassium, estimated by wholebody counting of ^{40}K, and to total body nitrogen measured by neutron activation. It followed that DPA could form the basis of a technique for the measurement of SMM, the component of muscle tissue that is used to effect locomotion and to maintain posture, especially since it was known that ALM constitutes between 73% and 75% of SMM. The technique was validated [2] when wholebody SMM was determined using MRI in a large group of healthy adults and compared with ALM measured by dualenergy Xray absorptiometry (DXA). The derived relationship was wholebody SMM = (1.19 × ALM) − 1.01, and ALM explained 96% of the variance in wholebody SMM. This relationship was refined by accounting for the intermuscular adipose tissue that had been included previously so that the predictive relationship in adults was adjusted to become SMM = (1.19 × ALM) − 1.65 [3]. Again, 96% of the variance in wholebody SMM was explained by the variance in ALM.
The technique was extended to children by comparing MRI measures of SMM with DXA measures of appendicular lean tissue mass [4]. It was first shown that for adolescents at Tanner stage 5 and beyond, the adult predictive equation applied. However, for younger children, the adult equation was inadequate. Using regression techniques, a prediction model was developed in 65 children who were less than Tanner stage 5. The predictive equation for such children was SMM = (1.115 × ALM) − 1.135 [4]. This equation was tested in an independent sample of 18 children in whom it was found that ALM was still the strongest predictor of SMM accounting for 98% of the variance. Addition of weight and height as predictor variables added a small but significant improvement in SMM prediction.
The purpose of our study was to apply the above predictive equations in a population of normal Canadian children and adolescents in order to establish ranges for expected values of SMM. We also evaluated the reproducibility of SMM estimations from repeated measures of ALM in a separate group of children.
(1) 
(2) 
(3) 
(4) 
A separate group of 32 normal children and adolescents (18 females and 14 males) aged between 3.7 and 17.7 years underwent three consecutive wholebody DXA scans with repositioning between scans. These scans, which were reanalyzed by the same investigator to assess reproducibility of DXAbased estimates of SMM, had been used previously for the assessment of reproducibility of wholebody BMD and body composition in children and adolescents [7]. Ninetyfive percent confidence intervals for reproducibility were calculated assuming a chisquare distribution for the individual variances found for each child [8].
A 
B 
C 
D 
E 
F 


Female 
7.0 
6.5 
0.55 
13.0 
0.75 
11.5 
Male 
12.4 
11.0 
0.45 
19.7 
0.85 
13.7 
The results for SMM reproducibility were considered separately for children above and below the age of 10 years. There were 15 younger children (nine females and six males). Their mean age (±standard deviation), height, and weight were 8.1 ± 1.6 years, 129.5 ± 10.7 cm, and 27.4 ± 6.1 kg, respectively. The precision for the younger group was 149 g with a 95% confidence interval of 119–199 g. The older group consisted of 17 children (nine females and eight males). Their mean age, height, and weight were 12.6 ± 2.0 years, 154.7 ± 15.3 cm, and 45.0 ± 12.5 kg, respectively. The precision for the older group was 170 g with a 95% confidence interval of 138–223 g. If all children were considered as a single group, the reproducibility was 161 g with a 95% confidence interval of 137–194 g.
G 
H 


Female 
0.10 
0.16 
Male 
0.005 
0.325 
A number of methods for the measurement of SMM in children have been evaluated with the objective of providing accurate, precise results in the clinical setting. Poortmans et al. [9] evaluated the merits of estimating SMM based on readily accessible variables in nonobese subjects. They measured SMM using the wholebody DXA technique in 39 children and 20 adults and examined regression relationships between the measured value and a predicted value based upon a group of variables that included height, age, and sex as well as skinfold thickness and limb circumference at each of the midarm, midthigh, and midcalf sites; a coefficient of correlation (r ^{2} value) of 0.966 was observed. When the DXAbased SMM assessments were correlated with 24 h urine creatinine excretions in the same group of subjects, the r ^{2} value fell to 0.73. The predictive equations were not tested in an independent group of subjects.
Wang et al. [10] explored the correlation between MRI determined SMM and total body potassium determined from wholebody counting of the naturally occurring radioisotope ^{40}K in 116 healthy children aged 5 to 17 years. SMM in children was shown to be a smaller fraction of total body potassium than in adults; in adults, the fraction is constant and independent of age [11]. Factors in addition to total body potassium that slightly improved the prediction of SMM in children were weight, height, and race.
The technique for the derivation of total body SMM from the mass of lean tissue in the arms and legs as measured by DPA was developed in adults [1] and was then extended to children [4]. The use of DXA scanning to measure ALM and hence to estimate SMM was validated by comparison with direct measures of SMM using wholebody MRI. DXAbased estimates of SMM are likely to be more acceptable than estimates of SMM based on wholebody MRI scans because of: (1) the difficulty of access to MRI scanners in comparison to DXA scanners; (2) the expense associated with MRI scanning; and (3) the need for image processing to segment MRI images.
The DXAbased technique for estimation of SMM has a reproducibility of 161 g. Expressed as a coefficient of variation, the reproducibility is 1.4%. However, it should be noted that expression of reproducibility in terms of a coefficient of variation is inappropriate, especially in children. When assessed for children below age 10, the reproducibility was 149 g; for children over age 10, the reproducibility was 170 g. When expressed as a CV, reproducibility in the younger children appeared to be worse at 1.8% compared to a value of 1.0% in the older children. However, despite an overlap of 95% confidence intervals for the standard deviations in the two groups of children, numerically, precision was better in the younger children. For a child aged 8, the ageexpected value of SMM is about 8–10 kg. With a precision of 149 g, the uncertainty in the difference between two successive measurements will be (149^{2} + 149^{2})^{1/2} or 211 g. To be 95% confident that a change had occurred between the two measurements, the difference would have to exceed (1.97 × 211) or 413 g.
(5) 
A limitation of our study is that Eqs. 1 and 2 were developed using wholebody scans obtained from a Lunar Densitometer [10]. The wholebody DXA scans for our normal subjects were obtained from a Hologic Densitometer. Software differences between the two manufacturers will likely mean that the values of the slope and intercept in Eqs. 1 and 2 need to be adjusted for Hologic equipment. However, this will have little, if any, effect on the Zscores calculated from Eq. 5 since the measured SMM and the SMM predicted for age will both include the same systematic error. Since the difference between these two variables is required for the Zscore calculation, the influence of the systematic error is minimized. This conclusion needs to be validated by comparing, in a group of children and adolescents, absolute SMM measured with wholebody MRI and SMM values determined from ALM measured on an Hologic densitometer. Another criticism might be that Tanner stages for our normal population were unknown. The influence of this effect can be estimated. If, for example, a 13yearold female had a measured ALM of 17 kg, our procedure would have estimated her SMM from Eq. 2 as 17.82 kg. Since she was, in fact, at Tanner stage 5, Eq. 1 should have been used which would have yielded a SMM of 18.58 kg. The difference between these values is somewhat greater than the smallest detectable change for the DXA technique of SMM measurement but is very much smaller than the normal ranges established for boys and girls. Finally, it must be stressed that the prediction equations presented here apply only to children and adolescents below age 20 and do not account for the expected increases in SMM beyond that age.
This work has provided a means of interpreting DXAbased measurements of SMM in children and adolescents in terms of expectedforage values. Our results also permit the interpretation of the significance of the differences between consecutive measurements of SMM.