JOURNALOF
APPLIED
PHYSIOLOGY
Vol. 39, No. 5, November
1975.
Pn’nted
Longitudinal
studies on adipose
its distribution S. CHIEN,
in U.S.A.
M.
in human T. PENG,
tissue and
subjects
K. P. CHEN,
T. F. HUANG,
C. CHANG,
AND
H.
S.
FANG
Department of Physiology and Institute of Public Health, Medical College of the National Taiwan University, Taipei, Taiwan, Republic of China; and Department of Physiology, College of Physicians and Surgeons, Columbia University, New York City 10032
S., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, H. S. FANG. Longitudinal studies on adipose tissw and its distribu1975.tion in human subjects. J. Appl. Physiol. 39(5): 825-830. On 27 men and 6 women, total body density and 10 skinfolds were measured 12 yr apart, with the mean age increasing from 3 1 to 43 yr. The increase in skinfold thickness was found to be related to the increase in total body adiposity, calculated from hydrostatic weighing. The external adipose tissue was calculated from the mean skinfold thickness and body surface area. Variations in total adiposity among the population studied as well as changes in total adiposity with age showed a characteristic distribution with approximately two-thirds on the surface and one-third in the interior. The essential body mass or total adipose mass determined by hydrostatic weighing was compared with the values obtained by water-immersion volumetry, total body potassium counting, and skinfold measurements. The volumetric and skinfold determinations gave better estimates of these parameters than total body potassium counting. CHIEN,
METHODS
AND
The results described in this report were obtained on the same 28 men and 6 women and at the same time as the experiments described in the accompanying paper (13). The various determinations were made on the same day over a period of less than 2 h. Because subject 32 was only 17 yr old at the time of the first study in 1954, his data were not included in the statistical calculations or any of the figures (13) . Hydrostatic
Weighing
The method of determining the total body volume density by underwater weighing is described in detail where (3, 13).
and else-
Volumetry body density; aging; body composition; adiposity; lean body mass; potassium, surface adiposity
body total
volume; body;
internal skinfolds;
A human body volumeter (1) was used to determine the total body volume and calculate total body density. The volumeter is a plywood tank which is coated with waterproof epoxy resin. Before the experiment, the tank was partially filled with water (35°C) and the water level was read from a vertical scale connected to the tank. The subject then entered the tank and stood with his head just above the water level. After a maximum expiration, he slowly immersed his head completely under water and the observer took a second reading of water level. At least three immersions were made to ensure the constancy of the reading. The body volume was calculated from the change in water level in the portion of the tank which has a constant crosssectional area. The residual lung volume was determined with the nitrogen dilution method (3) and subtracted from the total body volume.
MANY PARAMETERS of physiological functions vary with the body size. To normalize these parameters among individuals with different body sizes, the results have often been expressed in terms of per unit body weight. Since most physiological functions depend on the metabolically active tissues (essential body mass) rather than the relatively inert adipose tissue, the essential body mass or lean body mass is a much more satisfactory basis than the body weight for the quantitative expression of size-dependent functions (8, 10, 16, 20). Therefore, in the determination of normal values for physiological functions as well as in the quantification of deviations from normal in diseased states, the partitioning of body weight into adipose tissue and nonadipose essential body mass is extremely valuable. During the course of a study on the changes of blood volume with age over a period of 12 yr (13), such partitioning of body weight was made in normal subjects by the simultaneous application of several methods and the results of these different approaches are compared in this report. Since we repeated many of these measurements on the same subjects over a period of 12 yr, this investigation provides information on the longitudinal follow up of human adiposity and its distribution.
Total Body Potassium The body content of the naturally occurring 40K was determined in a Whole Body Counter by courtesy of NAMRU-2 at Taipei. The subject, wearing only the pajamas provided by the laboratory, sat for 40 min on an adjustable chair in a lead-lined chamber. During this time, the 40K was detected with an 8 x 4 in. NaI crystal connected to a gamma-spectrometer-computer system. The total body 40K activity (in countslmin) was calculated after subtracting 825
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
CHIEN ET
826 the background and correcting for a geometrical factor determined previously with the use of 42K (19). The total body potassium content was obtained from the total body 40K activity and the ratio of 40K activity to total potassium in the naturally occurring potassium samples. Skin. olds With the use of a skinfold caliper (Lange Skinfold Caliper, Cambridge Scientific Industries, Cambridge, Mass.), the at 10 sites on the thickness of skinfolds (12) was measured right side of the body: cheek, chin, upper arm, chest (second intercostal space at mammillary line), side of chest (seventh intercostal space at midaxillary line), waist (just above iliac crest), abdomen (level of the navel at mammillary line), back (just under the tip of the scapula), knee, and calf. All of the above measurements were made in 1966. In the 1954 study on the same subjects, only hydrostatic weighing and skinfold determinations were performed. RESULTS
AND
DISCUSSION
The results will be presented in two parts. The new contribution in Part A is mainly the simultaneous use of four different methods (hydrostatic weighing, volumetry, total body potassium, and skinfold measurements) for the estimation of essential body mass. The new information in Part B is the longitudinal data on changes in total adiposity and its distribution in the same subjects over 12 yr. A. Correlations Between Dzyerent Methods for Estimating Essential Body Mass Correlation between hydrostatic weighing and volumetry. Determinations of body volume by volumeter (V,) was carried out in 27 subjects in 1966, and the results are compared with the body volume by hydrostatic weighing (Vn). As shown in Fig. 1 the two .methods gave volume measurements 90. -m ti 80-; OL ul I-
; 3
I
I
0 m
/4’
/‘.
,’ /’
//
//
/ //
//
//
P YN8 /' /'. /0 0 / /‘ /0
/
I. Total body potassium and adipose tissue of subjects Males
Total body potaasium, KT , eq KT/body wt, meq/ kg Adipose tiosue mass,
/’
Female
agreement.
vv =
1
I
30
40
50
1. Relationship meter and by hydrostatic line of equality.
VOLUME
BY
ND 17.9 ’ zt3.69 (6) 167.9 ~f32.9 (6) 1.41 ho.063 (6; in parentheses
The regression
0.974 vH +
1.93 zizo.192 (4) 33.2 h1.84 (4) 25.0 zt5.38 (6) 214.8 ziz40.5 (6) 1.50 ho.089 (6) indicate
equation
1.28
the
is (4
and the coefficient of correlation is 0.998. The essential body masses ME and M Ev were calculated from the body volume measurements VH and Vv , respectively, and the total body weight, MT (13). ME = 7.353 MT
-
6.970 VH
(2)
7.353 MT
-
6.970 Vv
(3)
MEV
=
(MEV - ME) showed an averThe paired measurements age difference of 0.8 (SEM 0.46) kg, which was not statistically significant. If the difference in sign is ignored then the average deviation is I+: 1.9 kg, which is somewhat larger than the average deviation of 1.4 kg found between duplicate determinations of ME by hydrostatic weighing. Regression analysis showed that the standard error of estimate of ME from MEV was 2.6 kg, which was larger than the value of 1.6 kg for duplicate determinations of ME . Correlation between hydrostatic weighing and total body potassium. The total body potassium (KT) was determined in 25 men and 4 women in 1966 (Table 1). The values of KT showed a good correlation with the essential body mass determined by hydrostatic weighing (ME). The ratio of KT/ME averaged 65.8 meq/kg in both sexes (SD = 5.88 for men and 6.73 for women), and this agrees well with the data in the literature (2,7,9,22). Because of the small number of women studied, the present study cannot resolve the question whether KT/ME is different between men and women (9, 11). The regression equation between potassium content per unit body weight (KT/MT , in meq/kg) and the essential body mass as a fraction of body weight (ME/MT) is
This relation is shown lation is 0.8 10. Since
= 49.0 (ME/MT)
+
11.3
in Fig. 2, and the coefficient
(4) of corre-
0
MT
30 ,/”
ND
=t SEM and the numbers ND: not determined.
&/MT l
2.95
12.2 rtl.51 (28) 92.2 &IO. 7 (28) 1.61 zhO.028 (28)
excellent
1954
ztO.072 (24) 47.2 ztO.21 (24) 18.6 &1.50 (28) 155.1 zt11.4 (28) 1.68 kO.028 (28)
ND
Values given are means number of lrubjectr studied.
showing
1966
ND
kg Sum of 10 skinfolds, m m Body surface area, m2
,a' Male
Females
1954
,r"
50-
BODY FIG.
I
.d' / 1 ,P&
70-
$ 400’
I
?=0.974 X+1.28
G 60> m ; 3
I
TABLE
AL.
I
I
I
60
70
80
WEIGHING
l
90
ME+&
(5)
where MA is the mass of total adipose be written as
(liters)
between body volumes determined by voluweighing. Diagonal dotted line represents the
=
tissue, equation
KT = 49.0 ME + 11.3 (ME + MA)
4 can (6)
or KT
= 60.3 ME +
11.3 MA
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HUMAN
ADIPOSE
MASS
AND
DISTRIBUTION
AFTER
827
12 YR
Correlation between hydrostatic weighing and skinfold measurements. Skinfold measurements were made on all subjects in both 1954 and 1966 (Table 1). With the assumption that the thickness of the double layer of nonadipose skin has a combined thickness of 4 mm (3), the value of 40 mm was subtracted from the sum of 10 skinfolds (SF). Thus, the corrected sum of 10 skinfolds (SF,) is given as SF,
1
1
ESSENTIAL
2. Relationship
FIG.
body
1
20
OOI
BODY
I 40
I
MASS
I 60
wt) and essential
l
Female
0
I
1
I
body
100
weight)
potassium wt).
(in
meq/kg
These results may be interpreted to mean that the essential body mass contains a mean potassium content of 60.3 meq/ kg and that the adipose tissue has a mean potassium content of 11.3 meq/kg. The existence of a considerable amount of potassium in adipose tissue has been demonstrated by di(17, rect determinations using 40K and flame photometry 18). Forbes and Hursh (15) found that the fat-free component of the adipose tissue has a potassium content of 56 meq/kg. If it is assumed that adipose tissue contains 80 % fat (15), the potassium content in adipose tissue would be 11.2 meq/kg, a value in excellent agreement with Eq. 6. the essential body mass When Eq. 4 is used to calculate from total body potassium measurements (M&, the standard error of estimate is 5.1 kg, which is much larger than that found for the estimation of ME from MEV . Therefore, the total body potassium in the present study does not give a sufficiently precise measurement of the essential body mass, despite the reasonably good correlation observed. This lack of precision may be attributable to variations in potassium content in the body components and uncertainties in the geometric factor used for calculating the total body 40K activity (17).
400
’
I
I
I
I
1
I
I
I
I
= SF - 40
(7)
The SF, values in 1966, as in 1954 (4), showed a curvilinear relation to the adipose tissue mass as a percentage of total body weight (100 MA/MT) (Fig. 3A). The parabolic nature of the curve may be explained by the presence of MA in both the denominator and the numerator of the term MA/ MT, which may be written as M&MA + ME). When SF, is plotted directly against MA (Fig. 3B) or MA/ME , a linear relation is obtained. The regression equation for all 66 data points (including both sexes in both studies) is
I
80
(% total
between total body body mass b % body
Male
SF,
= 7.45 MA +
14
(8)
The coefficient of correlation is 0.933. When the data are separated by sex, the regression equations are for men, SF,
= 7.09 MA +
15
(9)
and for women,
SF,
= 7.53 MA + 30
w
The three thickest skinfold sites in both the male and female subjects were abdomen, waist, and back. The same three sites also showed the highest coefficients of correlation with MA in the male subjects. In the female subjects, the highest coefficients of correlation with MA were found in the skinfolds of upper arm, abdomen, and chest. Distribution of surface and internal adiposities. SF, is a onedimensional measurement, which does not take into account the other dimensions. Thus two individuals with the same SF, would have different amounts of MA if they are significantly different in height. Hence a better way to treat the skinfold data is to transform them to surface adipose mass
I
1
I - 400
- 0 A
E
F -s
‘3 300 ro r/) UJ A +z 5s 2 G t z ‘2
200 - 200
2 G z” z m z r
FIG. 3. A : sum of 10 skinfolds less 40 mm as a function of adipose mass (in y0 body wt). B: sum of 10 skinfolds as a function of adipose mass (in kg).
5
100
5 cn
w c
ADIPOSE
MASS
(% BODY
WEIGHT)
ADIPOSE
MASS
(kg)
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
CIIIEN
ET
AL.
total adipose tissue in excess of this minimal M,, is apparently distributed between the body surface and the interior with a rather constant proportion. Regression of MA8 on MA for all subjects yields the following equation MM
= 0.646 MA
-
2.95
(15)
Thus, over a wide range of MA , approximately two-thirds of the excess MA is found as MA9 (Fig. 4) and approximately one-third as MAI . 1954 Male Female
0
20
10 TOTAL
FIG. 4. Relationship
30
ADIPOSE
between
surface
MASS
A * 40
B. Longitudinal
1966 l
0
50
(kg)
adipose mass and total adi-
pose mass.
(4). The mass of surface adipose tissue ( MGS , in kg) can be estimated from the body surface area (SA, in m2), the average corrected skinfold thickness (in mm) measured at 10 sites and the density of adipose tissue (0.948 g/cm3)
NW
= 0.948(SA)(SFC/20)
WI
In this estimation it is assumed that the average skinfold thickness at 10 sites represents the weighted mean value for the entire body (4). The mass of the internal adipose tissue ( MAI) can then be calculated as
As shown in Fig. 4, the mass of the surface adipose tissue ( MA8) showed an excellent correlation with MA . The linear regression equations are for men, MA
= 1.400 MAS + 5.96
Study on Changes in Adiposity over 12 Yr
Changes in adzpose tissue and skinfalds. Hydrostatic weighing and skinfold measurements were made on each of the subjects in both 1954 and 1966. Therefore, paired comparison of the two determinations in individual subjects allows a longitudinal study of changes in total adiposity and surface adiposity with age over the 12-yr span. As shown in Fig. 5, the changes in the mass of total adipose tissue (AMA) accounted for nearly all the changes in total body weight (AM,). The regression equation is AMA
PI/IA = 1.398 MA8 + 5.99
ASF = 7.14 AMA
with coefficients of correlations of 0.943 and 0.966 and standard errors of estimate of 2.61 and 3.08 kg, for men and women, respectively. It is interesting to note that, the relation between MA and MAS is almost identical between sexes. The intercepts in Eq. 13 and 14 are statistically significant (P < 0.01). The data indicate that, when MA9 equals zero, MA has a finite value of approximately 6 kg, which may be considered as a minimal internal adipose mass in an extremely lean person without any detectable surface adiposity.’ The l This minimal MAI value in an extremely lean person might be considered as the essential lipids in the body. The value of 6 kg (or approximately lo-15y0 of the essential body mass in our subjects), however, appears to be rather high. The present calculations are based on several assumptions, including those on the partitioning of total body mass into two components, each with a constant density. Therefore, if the assumed density of the essential body mass is reduced from the value of 1.097 g/cm3 used, this minimal MAI value would decrease correspondingly and may be included as a part of essential body mass. The partitioning of total body mass into two components and the use of a constant value for the density of the essential body mass is probably an oversimplification (5, 6, 14). If the double layer of nonadipose skin has a combined thickness of less than 4 mm, this would lead to a systematic underestimation of MAa (JZq. 12) and raise the value for minimal MAI .
(16)
-
12
(17)
and the coefficient of correlation is 0.845. Changes in surface and internal adiposities. The changes in MA8 (AM,,) over 12 yr are linearly related to the changes in MA in the same subjects (Fig. 7). The coefficient of correlation is 0.894 and the regression equation is
WI W)
+ 1.2
The standard error of estimate is 2.50 kg and the coefficient of correlation is 0.933. The changes in MA over the 12-yr period also showed a linear relation with the changes in the sum of 10 skinfolds (Fig. 6). The regression equation is
AMAs
and for women,
= 0.970 AM,
= 0.650 AMA
+ 0.64
wo
Hence, the longitudinal follow-up on individual subjects provides results agreeing with the data on the population; i.e., adiposity has a relative distribution between the sur-
.p -10
t -10
0
10
20
30
CHANGE IN BODY WEIGHT (kg) FIG. 5. Change in adipose mass from 1954 to 1966 as a function the concurrent change in body weight.
of
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HUMAN
ADIPOSE
MASS
AND
DISTRIBUTION
AFTER
829
12 YR
CHEEK CHIN UPPER
ARM
CHEST BACK S I DE WAIST ABDOMEN KNEE CALF
l l
0.
l
/
l l
l 0 0
40
30
20
Male Female
w
-10
CHANGE
IN
.
? = 0.650
(kg)
as a function
of the con
X + 0.64 l
5
Male Female
-10
l
0
a
6 0
-10 CHANGE FIG. 7. Change
tion of concurrent
IN
TOTAL
10 ADIPOSE
20
30
(m-n)
2w
MASS
in the sum of 10 skinfolds in adipose mass.
10
0 THICKNESS
FIG. 8. Mean skinfold thickness at 10 body sites in 1954 (open bars) and 1966 (open plus shaded bars). The 1966 knee and calf skinfold thicknesses in the female subjects are less than the corresponding values in 1954 and are indicated by the dotted lines.
l
0
IV
ADIPOSE
FIG. 6. Change
change
0 SKINFOLD
/
current
10
20 MASS
(kgj
in surface adipose mass from 1954 to 1966 as a func- . change in total adipose mass.
face and the internal structures of approximately two-thirds versus one-third. Changes in distribution of skinfolds. In Fig. 8, the changes in the mean values of the 10 skinfolds over 12 yr are shown for the male and female subjects separately. In both sexes, the abdominal skinfold showed the greatest increase over 12 yr, and the next four sites with similar increments are the waist, side, back, and chest. Therefore, these findings confirm the well-known bulge in the middle with age. When the changes in skinfolds are correlated with the changes in total adiposity, the best correlations are found for the abdomen, cheek, and upper arm skinfolds, with coefficients of correlation approximately equal to 0.8. The authors thank Dr. Thomas H. Allen for his encouragement and advice. We are most grateful to the subjects who have made this study possible. We acknowledge the kind permission by the U.S. Navy Medical Research Unit 2 in Taipei for the use of their total body potassium counter. This investigation was supported by a research grant from the American Heart Association, National Heart and Lung Institute Research Grant HL-06139, US Army Contract DADA-l 7-72-C-2 115, and generous gifts from the Scaife Family Charitable Trusts in Pittsburgh, Pa. S. Chien held a National Council on Science Development Visiting Professor of Physiology at the National Taiwan University (JanJuly 1966). Received
for publication
20 February
1975.
REFERENCES 1. ALLEN, T. H. Measurement of human body fat: a quantitative method suited for use by aviation medical officers. Aerospace A&d. 34: 907-909, 1963. 2. ALLEN, T. H., E. C. ANDERSON, AND W. H. LANGHAM. Total body potassium and gross body composition in relation to age. J. Gerontol. 15 : 348-357, 1960. 3. ALLEN, T. H., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, AND H. S. FANG. Prediction of blood volume and adiposity in man from body weight and cube of height. Metabolism 5 : 328-345, 1956. 4. ALLEN, T. H., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, AND H. S. FANG. Prediction of total adiposity from skinfolds and the curvilinear relationship between external and internal viscosity. Metabolism 5 : 346-352, 1956. 5. ALLEN, T. H., B. E. WELCH, T. T. TRUJILLO, AND J. E. ROBERTS. Fat, water and tissue solids of the whole body less its bone mineral. J. Appl. Physiol. 14 : 1009-1012, 1959. 6. ANDERSON, E. C. Three-component body composition analysis
based
on potassium and water determinations. Ann. NY. Acad. 189-210, 1963. E. C., AND W. H. LANGHAM. Estimation of total body fat from potassium-40 content. Science 133 : 1917, 1961. BEHNKE, A. R., JR., B. G. FEEN, AND W. C. WELHAM. The specific gravity of healthy men. Body weight-volume as an index of obesity. J. Am. Med. Assoc. 118 : 495-498, 1942. BODDY, K., P. C. KING, K. HUME, AND E. WEYERS. The relation of total body potassium to height, weight and age in normal adults. J. CZin. Pathol. 25 : 5 12-5 17, 1972. BROZEK, J. (Editor). Body composition. Ann. NY. Acad. Sci. 110: 1424, 1963. BURKINSHAW, L., AND J. E. COTES. Body potassium and fat-free mass. CZin. Sci. 44 : 62 l-625, 1973. CHEN, K. P. Factor analysis of subcutaneous fat in Formosan adult women with special reference to nutritional criterion. Mem. CoZZ.
Sci. 110: 7. ANDERSON, 8.
9.
10. 11. 12.
Med.
NatZ.
Taiwan
Univ.,
Taipei
3 : I-- 12,
1953.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
830 13. CHIEN, S., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, AND H. S. FANG. Longitudinal measurements of blood volume and essential body mass in human subjects. J. Ap~l. Physiol. 39: 818824, 1975. 14. D~BELN, W. VON. Human standard and maximal metabolic rate in relation to fat-free body mass. Acta Physiol. Stand. Suppl. 126, 1956. 15. FORBES, G. B., AND J. B. HURSH. Age and sex trends in lean body mass calculated from K40 measurements: With a note on the theoretical basis for the procedure. Ann. N.Y. Acad. Sci. 110: 255263, 1963. 16. I(EYs, A., AND J. BROZEK. Body fat in adult man. Physiol. Rev. 33: 245-325, 1953. 17. KIRTON, A. H., AND A. M. PEARSON. Relationships between potassium content and body composition. Ann. N.Y. Acad. Sci. 110: 22 l-228, 1963.
CHIEN
ET
AL.
18. KULWICH,
R., L. FEINSTEIN, C. GOLUMBIC, R. L. HINER, W. R. W. R. KAUFFMAN. Relationship of gamma-ray measurements to the lean content of hams. J. Animal Sci. 20: 497-502, 196 1. 19. MILLER, C. E., AND A. P. REMENCHIK. Problems involved in accurately measuring the K content of the human body. Ann. N.Y. Acad. Sci. 110: 175-188, 1963. 20. MOORE, F. D., K. H. OLESEN, J. D. MCMURREY, H. V. PARKER, M. R. BALL, AND C. M. BOYDEN. The Body Cell and Its Supporting Environment. Philadelphia, Pa. : Saunders, 1963, p. 19-26. 21. PASCALE, L. R., M. I. GROSSMAN, H. S. SLOANE, AND T. FRANKEL. Correlation between thickness of skinfolds and body density in 88 soldiers. Human Biol. 28: 165-176, 1956. 22. WOMERSLEY, J., K. BODDY, P. C. KING, AND J. V. G. A. DURNIN. A comparison of the fat-free mass of young adults estimated by anthropometry, body density and total body potassium content. Clin. Sci. 43 : 469-475, 1972. SEYMOUR,
AND
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
APPLIED
PHYSIOLOGY
Vol. 39, No. 5, November
1975.
Pn’nted
Longitudinal
studies on adipose
its distribution S. CHIEN,
in U.S.A.
M.
in human T. PENG,
tissue and
subjects
K. P. CHEN,
T. F. HUANG,
C. CHANG,
AND
H.
S.
FANG
Department of Physiology and Institute of Public Health, Medical College of the National Taiwan University, Taipei, Taiwan, Republic of China; and Department of Physiology, College of Physicians and Surgeons, Columbia University, New York City 10032
S., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, H. S. FANG. Longitudinal studies on adipose tissw and its distribu1975.tion in human subjects. J. Appl. Physiol. 39(5): 825-830. On 27 men and 6 women, total body density and 10 skinfolds were measured 12 yr apart, with the mean age increasing from 3 1 to 43 yr. The increase in skinfold thickness was found to be related to the increase in total body adiposity, calculated from hydrostatic weighing. The external adipose tissue was calculated from the mean skinfold thickness and body surface area. Variations in total adiposity among the population studied as well as changes in total adiposity with age showed a characteristic distribution with approximately two-thirds on the surface and one-third in the interior. The essential body mass or total adipose mass determined by hydrostatic weighing was compared with the values obtained by water-immersion volumetry, total body potassium counting, and skinfold measurements. The volumetric and skinfold determinations gave better estimates of these parameters than total body potassium counting. CHIEN,
METHODS
AND
The results described in this report were obtained on the same 28 men and 6 women and at the same time as the experiments described in the accompanying paper (13). The various determinations were made on the same day over a period of less than 2 h. Because subject 32 was only 17 yr old at the time of the first study in 1954, his data were not included in the statistical calculations or any of the figures (13) . Hydrostatic
Weighing
The method of determining the total body volume density by underwater weighing is described in detail where (3, 13).
and else-
Volumetry body density; aging; body composition; adiposity; lean body mass; potassium, surface adiposity
body total
volume; body;
internal skinfolds;
A human body volumeter (1) was used to determine the total body volume and calculate total body density. The volumeter is a plywood tank which is coated with waterproof epoxy resin. Before the experiment, the tank was partially filled with water (35°C) and the water level was read from a vertical scale connected to the tank. The subject then entered the tank and stood with his head just above the water level. After a maximum expiration, he slowly immersed his head completely under water and the observer took a second reading of water level. At least three immersions were made to ensure the constancy of the reading. The body volume was calculated from the change in water level in the portion of the tank which has a constant crosssectional area. The residual lung volume was determined with the nitrogen dilution method (3) and subtracted from the total body volume.
MANY PARAMETERS of physiological functions vary with the body size. To normalize these parameters among individuals with different body sizes, the results have often been expressed in terms of per unit body weight. Since most physiological functions depend on the metabolically active tissues (essential body mass) rather than the relatively inert adipose tissue, the essential body mass or lean body mass is a much more satisfactory basis than the body weight for the quantitative expression of size-dependent functions (8, 10, 16, 20). Therefore, in the determination of normal values for physiological functions as well as in the quantification of deviations from normal in diseased states, the partitioning of body weight into adipose tissue and nonadipose essential body mass is extremely valuable. During the course of a study on the changes of blood volume with age over a period of 12 yr (13), such partitioning of body weight was made in normal subjects by the simultaneous application of several methods and the results of these different approaches are compared in this report. Since we repeated many of these measurements on the same subjects over a period of 12 yr, this investigation provides information on the longitudinal follow up of human adiposity and its distribution.
Total Body Potassium The body content of the naturally occurring 40K was determined in a Whole Body Counter by courtesy of NAMRU-2 at Taipei. The subject, wearing only the pajamas provided by the laboratory, sat for 40 min on an adjustable chair in a lead-lined chamber. During this time, the 40K was detected with an 8 x 4 in. NaI crystal connected to a gamma-spectrometer-computer system. The total body 40K activity (in countslmin) was calculated after subtracting 825
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
CHIEN ET
826 the background and correcting for a geometrical factor determined previously with the use of 42K (19). The total body potassium content was obtained from the total body 40K activity and the ratio of 40K activity to total potassium in the naturally occurring potassium samples. Skin. olds With the use of a skinfold caliper (Lange Skinfold Caliper, Cambridge Scientific Industries, Cambridge, Mass.), the at 10 sites on the thickness of skinfolds (12) was measured right side of the body: cheek, chin, upper arm, chest (second intercostal space at mammillary line), side of chest (seventh intercostal space at midaxillary line), waist (just above iliac crest), abdomen (level of the navel at mammillary line), back (just under the tip of the scapula), knee, and calf. All of the above measurements were made in 1966. In the 1954 study on the same subjects, only hydrostatic weighing and skinfold determinations were performed. RESULTS
AND
DISCUSSION
The results will be presented in two parts. The new contribution in Part A is mainly the simultaneous use of four different methods (hydrostatic weighing, volumetry, total body potassium, and skinfold measurements) for the estimation of essential body mass. The new information in Part B is the longitudinal data on changes in total adiposity and its distribution in the same subjects over 12 yr. A. Correlations Between Dzyerent Methods for Estimating Essential Body Mass Correlation between hydrostatic weighing and volumetry. Determinations of body volume by volumeter (V,) was carried out in 27 subjects in 1966, and the results are compared with the body volume by hydrostatic weighing (Vn). As shown in Fig. 1 the two .methods gave volume measurements 90. -m ti 80-; OL ul I-
; 3
I
I
0 m
/4’
/‘.
,’ /’
//
//
/ //
//
//
P YN8 /' /'. /0 0 / /‘ /0
/
I. Total body potassium and adipose tissue of subjects Males
Total body potaasium, KT , eq KT/body wt, meq/ kg Adipose tiosue mass,
/’
Female
agreement.
vv =
1
I
30
40
50
1. Relationship meter and by hydrostatic line of equality.
VOLUME
BY
ND 17.9 ’ zt3.69 (6) 167.9 ~f32.9 (6) 1.41 ho.063 (6; in parentheses
The regression
0.974 vH +
1.93 zizo.192 (4) 33.2 h1.84 (4) 25.0 zt5.38 (6) 214.8 ziz40.5 (6) 1.50 ho.089 (6) indicate
equation
1.28
the
is (4
and the coefficient of correlation is 0.998. The essential body masses ME and M Ev were calculated from the body volume measurements VH and Vv , respectively, and the total body weight, MT (13). ME = 7.353 MT
-
6.970 VH
(2)
7.353 MT
-
6.970 Vv
(3)
MEV
=
(MEV - ME) showed an averThe paired measurements age difference of 0.8 (SEM 0.46) kg, which was not statistically significant. If the difference in sign is ignored then the average deviation is I+: 1.9 kg, which is somewhat larger than the average deviation of 1.4 kg found between duplicate determinations of ME by hydrostatic weighing. Regression analysis showed that the standard error of estimate of ME from MEV was 2.6 kg, which was larger than the value of 1.6 kg for duplicate determinations of ME . Correlation between hydrostatic weighing and total body potassium. The total body potassium (KT) was determined in 25 men and 4 women in 1966 (Table 1). The values of KT showed a good correlation with the essential body mass determined by hydrostatic weighing (ME). The ratio of KT/ME averaged 65.8 meq/kg in both sexes (SD = 5.88 for men and 6.73 for women), and this agrees well with the data in the literature (2,7,9,22). Because of the small number of women studied, the present study cannot resolve the question whether KT/ME is different between men and women (9, 11). The regression equation between potassium content per unit body weight (KT/MT , in meq/kg) and the essential body mass as a fraction of body weight (ME/MT) is
This relation is shown lation is 0.8 10. Since
= 49.0 (ME/MT)
+
11.3
in Fig. 2, and the coefficient
(4) of corre-
0
MT
30 ,/”
ND
=t SEM and the numbers ND: not determined.
&/MT l
2.95
12.2 rtl.51 (28) 92.2 &IO. 7 (28) 1.61 zhO.028 (28)
excellent
1954
ztO.072 (24) 47.2 ztO.21 (24) 18.6 &1.50 (28) 155.1 zt11.4 (28) 1.68 kO.028 (28)
ND
Values given are means number of lrubjectr studied.
showing
1966
ND
kg Sum of 10 skinfolds, m m Body surface area, m2
,a' Male
Females
1954
,r"
50-
BODY FIG.
I
.d' / 1 ,P&
70-
$ 400’
I
?=0.974 X+1.28
G 60> m ; 3
I
TABLE
AL.
I
I
I
60
70
80
WEIGHING
l
90
ME+&
(5)
where MA is the mass of total adipose be written as
(liters)
between body volumes determined by voluweighing. Diagonal dotted line represents the
=
tissue, equation
KT = 49.0 ME + 11.3 (ME + MA)
4 can (6)
or KT
= 60.3 ME +
11.3 MA
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HUMAN
ADIPOSE
MASS
AND
DISTRIBUTION
AFTER
827
12 YR
Correlation between hydrostatic weighing and skinfold measurements. Skinfold measurements were made on all subjects in both 1954 and 1966 (Table 1). With the assumption that the thickness of the double layer of nonadipose skin has a combined thickness of 4 mm (3), the value of 40 mm was subtracted from the sum of 10 skinfolds (SF). Thus, the corrected sum of 10 skinfolds (SF,) is given as SF,
1
1
ESSENTIAL
2. Relationship
FIG.
body
1
20
OOI
BODY
I 40
I
MASS
I 60
wt) and essential
l
Female
0
I
1
I
body
100
weight)
potassium wt).
(in
meq/kg
These results may be interpreted to mean that the essential body mass contains a mean potassium content of 60.3 meq/ kg and that the adipose tissue has a mean potassium content of 11.3 meq/kg. The existence of a considerable amount of potassium in adipose tissue has been demonstrated by di(17, rect determinations using 40K and flame photometry 18). Forbes and Hursh (15) found that the fat-free component of the adipose tissue has a potassium content of 56 meq/kg. If it is assumed that adipose tissue contains 80 % fat (15), the potassium content in adipose tissue would be 11.2 meq/kg, a value in excellent agreement with Eq. 6. the essential body mass When Eq. 4 is used to calculate from total body potassium measurements (M&, the standard error of estimate is 5.1 kg, which is much larger than that found for the estimation of ME from MEV . Therefore, the total body potassium in the present study does not give a sufficiently precise measurement of the essential body mass, despite the reasonably good correlation observed. This lack of precision may be attributable to variations in potassium content in the body components and uncertainties in the geometric factor used for calculating the total body 40K activity (17).
400
’
I
I
I
I
1
I
I
I
I
= SF - 40
(7)
The SF, values in 1966, as in 1954 (4), showed a curvilinear relation to the adipose tissue mass as a percentage of total body weight (100 MA/MT) (Fig. 3A). The parabolic nature of the curve may be explained by the presence of MA in both the denominator and the numerator of the term MA/ MT, which may be written as M&MA + ME). When SF, is plotted directly against MA (Fig. 3B) or MA/ME , a linear relation is obtained. The regression equation for all 66 data points (including both sexes in both studies) is
I
80
(% total
between total body body mass b % body
Male
SF,
= 7.45 MA +
14
(8)
The coefficient of correlation is 0.933. When the data are separated by sex, the regression equations are for men, SF,
= 7.09 MA +
15
(9)
and for women,
SF,
= 7.53 MA + 30
w
The three thickest skinfold sites in both the male and female subjects were abdomen, waist, and back. The same three sites also showed the highest coefficients of correlation with MA in the male subjects. In the female subjects, the highest coefficients of correlation with MA were found in the skinfolds of upper arm, abdomen, and chest. Distribution of surface and internal adiposities. SF, is a onedimensional measurement, which does not take into account the other dimensions. Thus two individuals with the same SF, would have different amounts of MA if they are significantly different in height. Hence a better way to treat the skinfold data is to transform them to surface adipose mass
I
1
I - 400
- 0 A
E
F -s
‘3 300 ro r/) UJ A +z 5s 2 G t z ‘2
200 - 200
2 G z” z m z r
FIG. 3. A : sum of 10 skinfolds less 40 mm as a function of adipose mass (in y0 body wt). B: sum of 10 skinfolds as a function of adipose mass (in kg).
5
100
5 cn
w c
ADIPOSE
MASS
(% BODY
WEIGHT)
ADIPOSE
MASS
(kg)
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
CIIIEN
ET
AL.
total adipose tissue in excess of this minimal M,, is apparently distributed between the body surface and the interior with a rather constant proportion. Regression of MA8 on MA for all subjects yields the following equation MM
= 0.646 MA
-
2.95
(15)
Thus, over a wide range of MA , approximately two-thirds of the excess MA is found as MA9 (Fig. 4) and approximately one-third as MAI . 1954 Male Female
0
20
10 TOTAL
FIG. 4. Relationship
30
ADIPOSE
between
surface
MASS
A * 40
B. Longitudinal
1966 l
0
50
(kg)
adipose mass and total adi-
pose mass.
(4). The mass of surface adipose tissue ( MGS , in kg) can be estimated from the body surface area (SA, in m2), the average corrected skinfold thickness (in mm) measured at 10 sites and the density of adipose tissue (0.948 g/cm3)
NW
= 0.948(SA)(SFC/20)
WI
In this estimation it is assumed that the average skinfold thickness at 10 sites represents the weighted mean value for the entire body (4). The mass of the internal adipose tissue ( MAI) can then be calculated as
As shown in Fig. 4, the mass of the surface adipose tissue ( MA8) showed an excellent correlation with MA . The linear regression equations are for men, MA
= 1.400 MAS + 5.96
Study on Changes in Adiposity over 12 Yr
Changes in adzpose tissue and skinfalds. Hydrostatic weighing and skinfold measurements were made on each of the subjects in both 1954 and 1966. Therefore, paired comparison of the two determinations in individual subjects allows a longitudinal study of changes in total adiposity and surface adiposity with age over the 12-yr span. As shown in Fig. 5, the changes in the mass of total adipose tissue (AMA) accounted for nearly all the changes in total body weight (AM,). The regression equation is AMA
PI/IA = 1.398 MA8 + 5.99
ASF = 7.14 AMA
with coefficients of correlations of 0.943 and 0.966 and standard errors of estimate of 2.61 and 3.08 kg, for men and women, respectively. It is interesting to note that, the relation between MA and MAS is almost identical between sexes. The intercepts in Eq. 13 and 14 are statistically significant (P < 0.01). The data indicate that, when MA9 equals zero, MA has a finite value of approximately 6 kg, which may be considered as a minimal internal adipose mass in an extremely lean person without any detectable surface adiposity.’ The l This minimal MAI value in an extremely lean person might be considered as the essential lipids in the body. The value of 6 kg (or approximately lo-15y0 of the essential body mass in our subjects), however, appears to be rather high. The present calculations are based on several assumptions, including those on the partitioning of total body mass into two components, each with a constant density. Therefore, if the assumed density of the essential body mass is reduced from the value of 1.097 g/cm3 used, this minimal MAI value would decrease correspondingly and may be included as a part of essential body mass. The partitioning of total body mass into two components and the use of a constant value for the density of the essential body mass is probably an oversimplification (5, 6, 14). If the double layer of nonadipose skin has a combined thickness of less than 4 mm, this would lead to a systematic underestimation of MAa (JZq. 12) and raise the value for minimal MAI .
(16)
-
12
(17)
and the coefficient of correlation is 0.845. Changes in surface and internal adiposities. The changes in MA8 (AM,,) over 12 yr are linearly related to the changes in MA in the same subjects (Fig. 7). The coefficient of correlation is 0.894 and the regression equation is
WI W)
+ 1.2
The standard error of estimate is 2.50 kg and the coefficient of correlation is 0.933. The changes in MA over the 12-yr period also showed a linear relation with the changes in the sum of 10 skinfolds (Fig. 6). The regression equation is
AMAs
and for women,
= 0.970 AM,
= 0.650 AMA
+ 0.64
wo
Hence, the longitudinal follow-up on individual subjects provides results agreeing with the data on the population; i.e., adiposity has a relative distribution between the sur-
.p -10
t -10
0
10
20
30
CHANGE IN BODY WEIGHT (kg) FIG. 5. Change in adipose mass from 1954 to 1966 as a function the concurrent change in body weight.
of
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HUMAN
ADIPOSE
MASS
AND
DISTRIBUTION
AFTER
829
12 YR
CHEEK CHIN UPPER
ARM
CHEST BACK S I DE WAIST ABDOMEN KNEE CALF
l l
0.
l
/
l l
l 0 0
40
30
20
Male Female
w
-10
CHANGE
IN
.
? = 0.650
(kg)
as a function
of the con
X + 0.64 l
5
Male Female
-10
l
0
a
6 0
-10 CHANGE FIG. 7. Change
tion of concurrent
IN
TOTAL
10 ADIPOSE
20
30
(m-n)
2w
MASS
in the sum of 10 skinfolds in adipose mass.
10
0 THICKNESS
FIG. 8. Mean skinfold thickness at 10 body sites in 1954 (open bars) and 1966 (open plus shaded bars). The 1966 knee and calf skinfold thicknesses in the female subjects are less than the corresponding values in 1954 and are indicated by the dotted lines.
l
0
IV
ADIPOSE
FIG. 6. Change
change
0 SKINFOLD
/
current
10
20 MASS
(kgj
in surface adipose mass from 1954 to 1966 as a func- . change in total adipose mass.
face and the internal structures of approximately two-thirds versus one-third. Changes in distribution of skinfolds. In Fig. 8, the changes in the mean values of the 10 skinfolds over 12 yr are shown for the male and female subjects separately. In both sexes, the abdominal skinfold showed the greatest increase over 12 yr, and the next four sites with similar increments are the waist, side, back, and chest. Therefore, these findings confirm the well-known bulge in the middle with age. When the changes in skinfolds are correlated with the changes in total adiposity, the best correlations are found for the abdomen, cheek, and upper arm skinfolds, with coefficients of correlation approximately equal to 0.8. The authors thank Dr. Thomas H. Allen for his encouragement and advice. We are most grateful to the subjects who have made this study possible. We acknowledge the kind permission by the U.S. Navy Medical Research Unit 2 in Taipei for the use of their total body potassium counter. This investigation was supported by a research grant from the American Heart Association, National Heart and Lung Institute Research Grant HL-06139, US Army Contract DADA-l 7-72-C-2 115, and generous gifts from the Scaife Family Charitable Trusts in Pittsburgh, Pa. S. Chien held a National Council on Science Development Visiting Professor of Physiology at the National Taiwan University (JanJuly 1966). Received
for publication
20 February
1975.
REFERENCES 1. ALLEN, T. H. Measurement of human body fat: a quantitative method suited for use by aviation medical officers. Aerospace A&d. 34: 907-909, 1963. 2. ALLEN, T. H., E. C. ANDERSON, AND W. H. LANGHAM. Total body potassium and gross body composition in relation to age. J. Gerontol. 15 : 348-357, 1960. 3. ALLEN, T. H., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, AND H. S. FANG. Prediction of blood volume and adiposity in man from body weight and cube of height. Metabolism 5 : 328-345, 1956. 4. ALLEN, T. H., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, AND H. S. FANG. Prediction of total adiposity from skinfolds and the curvilinear relationship between external and internal viscosity. Metabolism 5 : 346-352, 1956. 5. ALLEN, T. H., B. E. WELCH, T. T. TRUJILLO, AND J. E. ROBERTS. Fat, water and tissue solids of the whole body less its bone mineral. J. Appl. Physiol. 14 : 1009-1012, 1959. 6. ANDERSON, E. C. Three-component body composition analysis
based
on potassium and water determinations. Ann. NY. Acad. 189-210, 1963. E. C., AND W. H. LANGHAM. Estimation of total body fat from potassium-40 content. Science 133 : 1917, 1961. BEHNKE, A. R., JR., B. G. FEEN, AND W. C. WELHAM. The specific gravity of healthy men. Body weight-volume as an index of obesity. J. Am. Med. Assoc. 118 : 495-498, 1942. BODDY, K., P. C. KING, K. HUME, AND E. WEYERS. The relation of total body potassium to height, weight and age in normal adults. J. CZin. Pathol. 25 : 5 12-5 17, 1972. BROZEK, J. (Editor). Body composition. Ann. NY. Acad. Sci. 110: 1424, 1963. BURKINSHAW, L., AND J. E. COTES. Body potassium and fat-free mass. CZin. Sci. 44 : 62 l-625, 1973. CHEN, K. P. Factor analysis of subcutaneous fat in Formosan adult women with special reference to nutritional criterion. Mem. CoZZ.
Sci. 110: 7. ANDERSON, 8.
9.
10. 11. 12.
Med.
NatZ.
Taiwan
Univ.,
Taipei
3 : I-- 12,
1953.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
830 13. CHIEN, S., M. T. PENG, K. P. CHEN, T. F. HUANG, C. CHANG, AND H. S. FANG. Longitudinal measurements of blood volume and essential body mass in human subjects. J. Ap~l. Physiol. 39: 818824, 1975. 14. D~BELN, W. VON. Human standard and maximal metabolic rate in relation to fat-free body mass. Acta Physiol. Stand. Suppl. 126, 1956. 15. FORBES, G. B., AND J. B. HURSH. Age and sex trends in lean body mass calculated from K40 measurements: With a note on the theoretical basis for the procedure. Ann. N.Y. Acad. Sci. 110: 255263, 1963. 16. I(EYs, A., AND J. BROZEK. Body fat in adult man. Physiol. Rev. 33: 245-325, 1953. 17. KIRTON, A. H., AND A. M. PEARSON. Relationships between potassium content and body composition. Ann. N.Y. Acad. Sci. 110: 22 l-228, 1963.
CHIEN
ET
AL.
18. KULWICH,
R., L. FEINSTEIN, C. GOLUMBIC, R. L. HINER, W. R. W. R. KAUFFMAN. Relationship of gamma-ray measurements to the lean content of hams. J. Animal Sci. 20: 497-502, 196 1. 19. MILLER, C. E., AND A. P. REMENCHIK. Problems involved in accurately measuring the K content of the human body. Ann. N.Y. Acad. Sci. 110: 175-188, 1963. 20. MOORE, F. D., K. H. OLESEN, J. D. MCMURREY, H. V. PARKER, M. R. BALL, AND C. M. BOYDEN. The Body Cell and Its Supporting Environment. Philadelphia, Pa. : Saunders, 1963, p. 19-26. 21. PASCALE, L. R., M. I. GROSSMAN, H. S. SLOANE, AND T. FRANKEL. Correlation between thickness of skinfolds and body density in 88 soldiers. Human Biol. 28: 165-176, 1956. 22. WOMERSLEY, J., K. BODDY, P. C. KING, AND J. V. G. A. DURNIN. A comparison of the fat-free mass of young adults estimated by anthropometry, body density and total body potassium content. Clin. Sci. 43 : 469-475, 1972. SEYMOUR,
AND
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
