0195-101/90/2206-072982090 MEDICINE AND SCENCE IN SPORTS AND EXERCISE Coyright ©1880 by me Arca Clepa cl Spt Medche Va. 22, No.8 Anthropometric estimation of muscle mass in men A. D. MARTIN, L. F. SPENST, D. T. DRINKWATER, and J. P. CLARYS. ‘Sport & Exercise Sciences Research Institute, The University of Manitoba, Winnipeg, CANADA R3T 2N2 ABSTRACT MARTIN, A. DL. F. SPENST, D. T. DRINKWATER, and J. P, CLARYS. Anthropometric estimation of muscle mass in men, Med. ‘Sci. Sports Exerc, Vol. 22, No. 5, pp. 729-733, 1990. Twelve male ‘cadavers (aged 50-94 y1) were subjected to comprehensive anthro- pometry, dissection, and weighing of all skeletal muscle, Correlation Coefficients of limb grths with tota skeletal muscle mass (MM) were high: forearm r= 0.96, mid-thigh r= 0.94, calfr= 0.84, and midarm F = 0.82, These increased when limb girhs were corrected (by subtracting x times the skinfold thickness) (0 estimate muscle girth, For dimensional consistency, variables in the regression analyses included the product of stature and the square of each corrected girth. For the six unembalmed cadavers, this yielded a three for MM (r= 0.93; SEE = 1,56 kg), which was then data from the embalmed cadavers. It predicted MM with an SEE of 1.58 kg and r = 0.93. Because the values of these SEESs were similar, wwe pooled the subjects from the two groups to generate the final estimation equation: MM = STAT (0.0553CTG? + 0.0987FG + 0.0331CCG') ~ 2445 (SEE = 1.53 ke, = 0.97), where STAT is stature (cm), CTG is thigh circumference corrected for the front thigh skinfold thickness (cm), FG is the uncorrected forearm circumference (em), and CCG is the calf circumference corrected for the medial calf skinfold thickness (em). Despite the limitations of the cadaver sample, the proposed equation appears to provide the best estimate of skeletal muscle mass to date, in that itis ‘the only cadaver-validated equation and it gives valves that are consistent with all known dissection data, BODY COMPOSITION, LEAN BODY MASS, CADAVER It is remarkable that, while there is a large body of literature on the estimation of body fat, there is no satisfactory method for estimating the mass of skeletal muscle in the human body. Such a technique would have many applications in sports science. It would be a useful addition to the physiological profile of elite ath- letes, since athletes in different sports may show greater variability in muscle mass than in fat mass. It would enable a coach to monitor the effects of different train- ing regimens (34) and to determine whether a decre- ment in performance has accompanied a loss of lean tissue. Also, muscle mass could be used instead of body Subd for poliion Jay 1988 ‘Accepted for petion November 1989. mass to normalize maximum oxygen uptake when comparing individuals with different levels of fatness (9). There are applications in many other disciplines. Skeletal muscle is “ideally suited for establishing how severely protein-energy malnutrition has depleted lean tissues and how far the recovery process has come” (18), an observation that applies to other forms of starvation such as anorexia nervosa. The measurement of muscle mass is important for drug dosage in phar- macology since some drugs have an affinity for lean tissue (25.) Bone and muscle both atrophy during pe- riods of immobilization, as in space flight, bed rest, or paraplegia (21). An accurate assessment of muscle mass would be useful to monitor such atrophy and to eval- uate preventive and therapeutic strategies. Estimates of muscle mass could be used to indicate biological ma- turity, as an alternative to the traditional method of hand radiographs (30). One reason for the lack of a usefull technique is the dearth of direct data on anatomical tissue masses. A search of the nineteenth century anatomical literature for data on adult males revealed only nine dissections for which both muscle mass and body mass were re- ported: Bischoff (1), Bruel (3) (cited in Welcker and Brandt (33)), Dursy (10) (cited in Vierordt (32)), Liebig (26), and Theile (31). Subsequently only two laborato- ties have reported such data: four male subjects between 1945 and 1956 (11,12,29) and 12 more males from our Brussels Cadaver Study (6,8,27). Thus, skeletal muscle masses determined by dissection are known for only 25 adult males. For these men, dissected muscle masses ranged from 15.8 kg to 40.4 kg or from 27.4% to 49.1% of the body mass. These values are confounded by adiposity. Fatter subjects, such as our older Belgians, showed lower percentages, and therefore a better indi- cator of muscularity is obtained by normalizing muscle ‘mass to the adipose tissue-free mass (ATFM). This has been done for those dissections where adipose tissue ‘mass was also reported (V = 22). Muscle mass ranged from 36.6% to 59.4% of the ATFM, with a mean of 79730 50.0% and standard deviation of 4.9%. There was little difference between our mean (52.0%) and that of the 19th century German data (49.7%), while the cadavers from the four dissections of Forbes and Mitchell showed a lower proportion of muscle in the ATFM (44.5%). IN VIVO ESTIMATION OF SKELETAL MUSCLE MASS The principle of all indirect methods in body com- position is to measure a variable other than the desired one and 10 make some quantitative assumption about the relationship between the two variables. Breakdown products of muscle, excreted in the urine, have been used to estimate muscle mass, in particular creatinine and 3-methyl histidine (5). Skeletal muscle mass has also been estimated by counting gamma ray emission from K or “N, which permits quantification of total body potassium (TBK) and total body nitrogen (TBN) with an accuracy of about 4% (4). Conversion of these chemical estimates to muscle mass must then rely on assumptions about what proportion of the body's nitro- gen or potassium resides in skeletal muscle. Radiographic and other related imaging techniques are a useful indicator of regional muscularity. The high resolution currently available with computed tomog- raphy and nuclear magnetic resonance allows clear separation of adjacent tissues, and scanners have var- ious types of software for distinguishing areas with a given pixel density range (16). Tissue volumes can be determined by using a series of slices of known thick- ness and separation. This is particularly useful for organ, volumes. When the amounts of tissues that are distrib- uted over the whole body are required, adjacent slices from head to toe must be taken. This becomes prohib- itive in both cost and radiation exposure. If only re- ‘gional slices are taken, there is the problem inherent in all imaging approaches to tissue mass determination, that of extrapolating from regional tissue mass to whole body tissue mass. ANTHROPOMETRIC INDICATORS OF MUSCLE MASS The task of estimating muscle mass from anthropo- metry is to select one or more anthropometric variables that reflect total muscle mass. Commonly, one muscle group is selected with the assumptions that (i) local anthropometry reflects the mass of that muscle group and (ji) the mass of the muscle group is directly related to total muscle mass. The scientific and historical basis for the upper arm being the site of choice has been discussed by Jelliffe and Jelliffe (23), though the lower limb has also been used (19). The most common choices of variables have been limb circumferences MEDICINE AND SCIENCE IN SPORTS AND EXERCISE corrected for subcutaneous adipose tissue and muscle cross-sectional area estimated from this corrected cir- cumference. The tissue boundaries in the limb cross- section are assumed to be circular and concentric. If the skin-plus-subcutaneous adipose tissue thickness is d and the limb circumference is Cime, then the limb muscle circumference Cy, is given by Cm = Cin — 2nd.Jf it is further assumed that the skinfold caliper reading, S, is twice the adipose tissue thickness, then Cm = Cin #8. This simple linear index has been used extensively in protein-energy malnutrition studies (13,23) and in chronic disease (2,20). Nonetheless, mus- cle mass is a three-dimensional quantity while arm circumference is unidimensional. This has prompted some workers to use estimates of muscle area (14,17,18,24), Jelliffe (22) has proposed and used a three-dimensional index consisting of arm muscle area multiplied by one-fifth of the acromial-olecranon length. These approaches have two basic weaknesses. First, just as the attempt to estimate percent body fat from one or two skinfold thicknesses leads to large errors arising from inter-subject variability in regional fat distribution, so estimation of the muscle mass of the whole body from one or two limb circumferences may also be vulnerable to variability in the body's regional distribution of muscle. Second, these approaches only provide indices of muscle mass. The only two anthropometric approaches intended to estimate absolute rather than relative muscle mass are those of Heymsfield et al. (18) and Matiegka (28). ‘The former was based on their earlier arm muscle area index, which they calibrated against urinary creatinine, assuming Ig of urinary creatinine per day to be equiv. alent to 20 kg of muscle. They specified clearly that this, was only intended to give an approximate estimate of muscle mass. Matiegka’s equation used circumferences of arm, forearm, thigh, and calf, corrected for the corresponding skinfold thicknesses, to derive a mean value for limb muscle radius. This he squared and multiplied by stature and by a constant whose value “appears to be about 6.5”. Up to now, all attempts to estimate muscle mass have had the inherent problem of not being validated in humans. The Brussels Cadaver Study has provided, for the first time, both anthropometry and tissue masses in the same subjects. We report here the relationships ‘between muscle mass and selected anthropometric vari- ables in adult males and a regression-based estimation equation for muscle mass, METHODS In the course of the Brussels Cadaver Study, 25 human cadavers were subjected to comprehensive an- thropometry prior to complete dissection and weighing, of skin, skeletal muscle, adipose tissue, bones, andANTHROPOMETRIC ESTIMATION OF MUSCLE MASS IN MEN organs. Gross tissue masses and details of the dissection have been reported elsewhere (6,8,27). Of the 25 sub- jects, 12 were embalmed (6 men, 6 women) and 13 were unembalmed (6 men, 7 women). Anthropometry included skinfold thicknesses at the following sites: triceps, subscapular, biceps, front thigh, and medial calf. Circumference measurements were taken on the arm (midway between acromiale and radiale), forearm (at maximum girth), thigh (midway between the in- guinal crease and the midpoint of the patella), and calf (at the maximum girth). Detailed descriptions of the measurements have been reported elsewhere (27). Limb muscle girths were estimated by correcting limb girths for skinfold thickness using the circular model of the limb cross-section described previously. Pearson correlation coefficients were calculated for muscle mass with limb girths, skinfold-corrected girths, and stature, Regression equations were generated with muscle mass as the dependent variable and anthropometric variables as independent variables. For dimensional consistency regression equations were derived from the product of squared circumferences and stature so that each term in the equation took the form of a length cubed. With only one set of data containing both muscle mass and anthropometry, the problem of validation of the regres- sion equation arose. Rather than use all the male ca- davers for the regression equation, we have used data from the six unembalmed cadavers to generate the equation and data from the embalmed cadavers for validation. Though six embalmed males were dissected, examination of the data revealed considerable atrophy of right leg muscle and bone in one subject. This cadaver was therefore not included in the validation sample. ‘An equation for women was not generated since the ‘women in the cadaver sample were considerably fatter than the men and this reduced the accuracy of the corrected girths, resulting in poorer correlations with muscle mass. RESULTS Cadaver analysis. Correlation coefficients of limb circumferences with total dissected muscle mass ranged from 0.82 to 0.96, remarkably high values in view of the small sample size. Forearm circumference (F° = 0.93) and mid-thigh circumference (r° = 0.89) were the best predictors of muscle mass. These values were greater still when skinfold-corrected circumferences were used (Table 1). Squaring the corrected circumfer- ences gave little or no improvement. Since forearm skinfolds are rarely measured and the correlation coef ficient of the uncorrected forearm circumference with muscle mass was very high, it was decided to use the uncorrected forearm circumference along with the cor- rected arm, thigh, and calf circumferences. For dimen- 73 TABLE 1. Corélation coos of ib gts with otal muscle mas or ix unerbaimed cadavers. ‘corte Coetick Site Se ‘Squore of lo Coreted cameted i) oa anes 0900 Foe 0960, oom on ose =—ga%0 oe Call oe tail oa sional consistency the variables used in the regression analysis were the square of each selected circumference multiplied by stature, The equation with the lowest SEE, 1.39 kg, contained only two circumference terms: MM = STAT (0.0739CTG? + 0.111FG*) ~ 2930, where MM is total skeletal muscle mass (g), STAT is stature (em), CTG is thigh circumference corrected for the front thigh skinfold thickness (cm), and FG is the uncorrected forearm circumference (cm). It was de- cided that, to reduce sample specificity of the equation, it should contain at least three circumference terms, and the following equation had the lowest SEE (1.56 kg) of all the three-term equations: ‘MM = STAT (0.0S46CTG* + 0.119FG? + 0.0256CCG") - 2980, where CCG is the calf circumference corrected for the calf skinfold thickness (cm). Validation was carried out by using this equation to estimate muscle mass for the five embalmed cadavers. The SEE was 1.58 kg, with ? = 0,93 (Fig, 1). To compare the equa- tions of Heymsfield et al. (17), Matiegka (28), and us, the errors in estimating muscle mass in the five em- balmed cadavers (estimated minus actual) were calcu- lated (Fig. 2). The similar values of these SEEs suggested that the subjects from the two different groups (six unem- balmed, five embalmed) could be combined to generate the final estimation equation, rather than using only the unembalmed cadavers. This resulted in the follow- ing equation: MM = STAT (0.0553CTG? + 0.0987FG" + 0.0331CCG") ~ 2445, with SEE = 1.53 kg and P= 0.97. DISCUSSION Cadaver data. From the adult human male data previously discussed, skeletal muscle accounts for about 50% of the adipose tissue-free mass. Variability is con- siderable, with reported values as low as 36.6% and as high as 59.4%. The Brussels subjects were all over 50 yr of age and so would probably have experienced, to varying degrees, an age-related decline in muscle mass,732 MEDICINE AND SCIENCE IN SPORTS AND EXERCISE 5 = embalmed © unembalmed 0 Estimated Estimation o ‘muscle or (kg) Q a mass (kg) 5 7 2 10 . as 10 20 30 a0 50 ‘Actual muscle mass (kg) Figure 1—Muscle mass of five embalmed cadavers estimated from ion derived from the six unembalmed cadavers. The line of| Identity is also shown. not dissect any cadaver showing signs of wast- ing, but it seems likely that emaciated individuals would have lower values, while athletes in sports requiring great strength would have higher values than the range reported here. Indirect assessment using total body nitrogen and total body potassium appears to underestimate absolute muscle mass. Of the 14 subjects over 50 yr of age for whom muscle mass has been reported (including our 11), only one had a muscle mass less than 17.3 kg, the mean value reported by Cohn et al. (7) for their men in the age range 50-79 yr. The mean for all 25 men for whom muscle mass has been reported was 25.3 ke, considerably higher even than the mean of 21.1 kg reported for the younger men of Cohn et al. (7). This discrepancy reinforces the importance of validation, preferably by cadaver dissection, for any indirect method. In view of the small sample size, the high correlation coefficients for skinfold-corrected girths with dissected muscle mass in the Brussels cadavers are encouraging. Perhaps the most striking finding was a correlation coefficient of 0.96 for uncorrected forearm girth with muscle mass. This suggests that muscle patterning throughout the body was similar in all men in our sample, despite the wide range in both age and muscle mass. Of all four limb girths, corrected arm girth, the most commonly used anthropometric indicator of mus- cularity, had the lowest correlation coefficient with muscle mass. This was a surprising finding since arm girth is often the only anthropometric variable used in clinical evaluation of protein-energy malnutrition, both in victims of famine and in hospitalized patients (15,18,23). Regression equations that predict body composition are highly sample-specific. Simply using a regression procedure to generate an equation, based on some optimization criterion such as minimum standard error of the estimate, allows the procedure to exploit the 16 1a 20 22 24 26 28 Dissected muscle mass (kg) ye 2—Muscle mass estimation error (estimated ~ actual) for five embalmed cadavers using the equations of Feymsficid and Matiegka and our equation derived from the unembalmed cadavers. uniqueness of the sample to achieve the optimum so- lution, In so doing, however, it increases the errors that, arise when the equation is applied to a different sample. Some of the potential limitations to the external validity of the equation can be overcome if certain conditions are met: (i) the equation should include enough vari- ables to account for individual differences in the tissue distribution even though the inclusion of such extra variables may have little effect on the SEE, and (ii) the equation should be dimensionally consistent (35), thus reducing errors arising from differences in body size across samples. ‘The performance of the three equations on the em- balmed sample emphasizes the importance of including more than one variable in equations that predict whole- body tissue masses. Heymsfield’s equation, which uses arm circumference alone to estimate MM, shows a consistent underestimation (Fig. 2), with deviations as large as 12 kg from dissection values. Matiegka’s pre~ dictions, based on four muscle circumferences, also underestimate all the dissection values, but to a lesser degree. This suggests that the constant he used may have been too low, a possibility that he noted in his original report. CONCLUSION Though the equation proposed here is based on regression analysis and therefore may reflect unique characteristics of the cadaver sample, it has two char- acteristics that may improve its external validity. It is dimensionally consistent, and it contains more vari- ables than would result from the optimal regression analysis. Despite the limitations of the cadaver sample, the proposed equation appears to provide the best estimate of skeletal muscle mass to date, in that it is the only cadaver-validated equation and it gives values that are consistent with all known dissection data,ANTHROPOMETRIC ESTIMATION OF MUSCLE MASS IN MEN Present address for authors: A. D. Martin and D. 7. Drinkwater, ‘Sport & Exercise Sciences Research Institute, Unversity of Manitoba; LF. Spenst, College of Physical Education, University of Saskaten- ‘ewan; J.P. Clarys, Department of Anatomy, Vrije Universitet Brussel, REFERENCES |. BISCHOFF, E. Binige Gewichts und Trockenbestimmungen der Organe des Menschilichen Korpers. Z. Rationelle Med. 3:75-85, 1863. 2. BLACKBURN, G. L, B. R. BISTRIAN, B, S, Mant, H, T, SCHLAUM, and M. F. Suitu. Nutritional and metabolic assessment of the hospitalized patient J. Parenter. Enter. Nutr. 1:11-16, 1975. 3, BRUEL, W. (cited in Welcker, H. and A. Brandt (33) 4, Buskink, E.R. Body composition analysis: the past, present and the future, Res. Q. Exere. Sport $8:1-10,1987. 5. Buskirk, E.R, and J. MENDEZ. Lean-body tissue assessment, with emphasis on skeletal muscle mass. In: Body Composition Assessments in Youth and Adults. Report of the Sixth Ross Conference on Medical Research, A. F, Roche (Ed.). Columbus, (OF: Ross Laboratories, 1985, pp. 59-65, 6. CLarvs, J. PA. D. Martin, and D. T, DrivkwareR, Gross tissue weights in the human body by cadaver dissection, Hum, Biol. 56:459-473, 1984, 7. Coun, S. H., D. Vartsky, S. YasuMURA, A. N. VASWANI, and KJ, ELLs. Indexes of body cell mass: nitrogen versus potassium, Ain. J. Physiol. 246:E305-E310, 1983. 8, Drinkwaren, D. T. An Anatomically Derived Method for the Anthropometric Estimation of Human Body Composition. Doc- {oral Dissertation, Simon Fraser University, British Columbia, ‘Canada, 1984, 9. Duanin, J. V.G. A. The use of surface area and body weight as standards of reference in studies on human energy expenditure. Br. Nutr. 13:68-T1, 1959, 10, Dunsy, E. Lehrbuch der Systematiscen Anatomie, $16, 1863 (cited by Vierrdt (32). 11, Fortes, R.M., A. R. Coorer, and H. H, Mrrcweut, The composition ofthe adult human body as determined by chemical analysis. J. Bil. Chem. 203:359-366, 1953, 12, Fornts, R. MH. H. MITCHELL, and A. R, Cooper. Further studies on the gross composition and mineral elements of the adult human body. J. Biol. Chem. 223:969-975, 1956, 13, FRisaNcHo, A. R. Triceps skinfold and upper arm muscle size norms for assessment of nutritional status, Am. J. Clin. Nutr 27:1052-1058, 1974, 14, FRisancno, A. R, New standards of weight and body composi- tion by frame size and height for assessment of nutritional status ‘of adults and the elderly. Am. J. Clin, Nur. 40:808-819, 1984 15, Hatper, M. and S. Q. Haipes. Assessment of protein-calorie malnutrition, Clin, Cem. 30:1286~1299, 1984, 16, Hevasrieip, S. B, Clinical assessment of lean tissues: future directions. In: Body Composition Assessment in Youth and ‘Adults. Repor ofthe Sixth Ross Conference on Medical Research, ‘A. F. Roche (Ed,), Columbus, OH: Ross Laboratories, 1985, pp. 53-59, 17, Heyusrievo, S. B, C. McManus, J. Sur, V. STEVENS, and D. W. NIXON. Revised equation for calculating bone free arm muscle area. Am. J. Clin Nur. 36:680-690, 1982, 733 Belgum. ‘Address for correspondence: Alan Martin, P.D., Sport & Exercise Sciences Research institute, The University of Manitoba, Winnipeg, ‘Canada RST 2N2. 20, 24 22 23. 24, 25, 26, 20 28, 29. 30, 31. 2 33. M4, 35, . HEYMSFIELD, S. B., C, McMANus, V. STEVENS, and J. Sat, Muscle mas: reliable indicator of protein energy malnutrition, Severity and outcome. Am, J. Clin. Nutr. 35:1192-1199, 1982. HEYMSrIELD, S. B., R. P. Otasson, M. H. KUTNER, and D. W. Nixon. A radiographic method of quantifying protein-calorie undernutrition. Am. J. Clin Nutr, 32:693-702, 1979, Hitt, G. L,, RL. BLackerr, I. PlekForD, etal, Malnutrition in surgical patents: an unrecognized problem. Lancet 1:689, 1977, Jaworski, Z. F. G. and H. K. UHTHOFF, Disuse osteoporosis: ‘current status and problems. In: Current Concepts of Bone Fra- silty, H. K. Unthoft (Ed.). New York: Springer-Verlag, 1986, pp. 181-194, JELLIFeE, D. B. The Assessment of Nutritional Status of the ‘Community. W.#.0. Monograph #53. Geneva: World Health Organization, 1966, JeLLirre, E. FP, and D. B. JeLLirre. The arm circumference as 4 public health index of protein-calorie malnutrition of early childhood. (Letter to the editor). J. Trop. Pediat. 15:179, 1969. Keys, A. J. BRozek, O. WENSCHEL, O, WENSCiIEL, O. MICKEL~ son, and H. L. Tavtor. The Biology of Human Starvation Minneapolis: University of Minnesota Press, 1950. LAURENCE, D. R. and P. N. BENNETTE. Clinical Pharmacology. Churchill Livingstone, 1980, pp. 371-372. Liewic, G. v. Gewichisbestimmungen der Organe des minsched. Korpers, Arch. Anat. Physiol. Wissensch. Med. 4:96-117, 1874 Marnin, A. D. An Anatomical Basis for Assessing Human Body Composition: Evidence from 25 Dissections. Doctoral Disserta- tion, Simon Fraser University, British Columbia, Canada, 1984 Mariska, J, The testing of physical efficiency. Am. J: Phys: Anthropol. 4223-230, 1921 Mircuett, H. H., T.'S. HAMmroN, F, R. STEGGERDA, and H. \W. BEAN. The chemical composition of the human body and its bearing on the biochemistry of growth. J. Biol. Chem. 158:625~ 637, 1945, ‘TaNNeR, J. M,, R. H. WurTenouse, W, A, MARSHALL, M, J. R. Heaty, and H. Gouosrein, Assessment of Skeletal Maturity and Prediction of Adult Height (TW2 Method). London: Academic Press, 1975, ‘TueiLe, F. W. Gewichtsbestimmugen zur Entwickelung des Mu- skelsystems und des Skelettes beim Menschen. Nova Acta KSI. Leop-Carol-Deutschen Akad. Natorforocher 46:170, 1884. Vienonor, H. dnatomische, Physiologische und Physikalische Daten wnd Tabellen. Jena: Fisher, 1906 \Wetcker, H. and A. BRaNpt. Gewichteswerthe det Korperor- jgane bei dem Menschen und den Thieren, Arch, Anthropol 28: 59, 1903. ‘Witaore, J. H. Body composition in sport and exercise: direc- tions for future research. Med. Sci. Sports Exerc. 15:21-23, 1983, ‘models and computer simulations of physiol itoril), Am. J. Physiol. 234:R159-R160, 1978