Metabolism

Clinical and Experimental

VOL. XXX, NO. 9 SEPTEMBER 1981

Physical Training and Adipose Tissue Fatty Acid Composition in Men

W. H. F. Sutherland, S. P. Woodhouse, and M. R. Heyworth

Four months physical training significantly raised the mean proportions (%I of lauric (+0.64), myristic (f0.52). stearic
( +2.06) and linolaic ( + 1.69) acids and lowered the mean proportions of palmitic (- 1.46) and olaic (-3.46) acids in adipose
tissue of 20 apparently normal man aged 20-55 yr. These changes ware dependent on pretraining proportions of the fatty
acids. In addition, the decreases in palmitic acid and olaic acid proportions ware positively correlated (p = 0.905, P < 0.01,
T = 0.756, P < 0.01). The pattern of significant correlations among adipose tissue fatty acids was altered as a result of
training. We suggest that preferential mobilisation of fatty acids from adipose tissue is responsible for the change in
adipose tissue fatty acid composition with increased physical training. The significant increase in adipose tissue linolaic
acid proportions may be linked with the reduced risk of coronary heart disease which has been previously associated with
increased physical activity.

E PIDEMIOLOGICAL data suggests that regular bicycle ergomete? and anthropometric parameters”’ were measured
physical activity is associated with reduced risk at the beginning of the study, at peak training 4 mo later, and after 6
mo, which corresponds with the end of the study. Samples of
of developing coronary heart disease (CHD).ld The
subcutaneous gluteal fat were obtained by needle biopsy at the three
mechanism underlying this association is not clear, sampling times. These were immediately frozen in liquid nitrogen
therefore the study of metabolic changes produced by and stored at -SO0 until analysed. Lipids were extracted” from the
increased physical training is indicated. We have adipose tissue and transesterified.” The fatty acid methyl esters
previously reported on the influence of physical train- were analysed by gas-liquid chromatography using a flame ionisa-
tion detector, on a 1.5 m x 4 mm i.d. glass column packed with 10%
ing on plasma lipoprotein lipid concentrations in men
Silar IOC on Gas Chrom Q. The initial column temperature was
from the present study.’ In the current study we have l85O which was raised to 210” immediately after the emergence of
examined the effect of increased physical training on methyl linoleate. Peak area was determined using a Varian CDS-
adipose tissue fatty acid composition in normal men 11 I integrator.
for three reasons. Firstly, because there are no longitu- Statistical analyses were performed using the paired t test,
Student’s t test, correlation analysis and analysis of variance.
dinal data available on the effect of physical training
The net correlation coefficients in Table 6 were calculated as
on adipose tissue fatty acid composition in normal previously described.13 To calculate correlation coefficients between
men. Secondly, adipose tissue linoleic acid proportions changes in adipose.tissue fatty acid proportions and to evaluate their
and levels of physical fitness were significantly higher statistical significance, the following rationale was used. Let (X,2,0,
in men aged 40 yr at lower risk of developing CHD X ,4,0, .) denote the pretraining multinomial distribution of fatty

compared to those at higher risk6,’ Thirdly, physical acids, ZX = 100%. and (Y,,:,, Y ,,,,, .) denote this distribution
after four months training. Let (P,,,,,. P,,,O, .) and (Q,2:., Q,&,,, . .)
training altered adipose tissue fatty acid composition denote the respective multinomial proportions parameters. Because
in patients with coronary artery disease.*

MATERIALS AND METHODS

Informedconsent was obtained from all participants in the study.
Twenty-three apparently normal men aged 20-55 (mean age 35 yr)
who were not regular athletes, participated in a self-regulated, but
structured, four month training program which culminated in a
marathon race. After the race, most subjects reduced their levels of
training during the final 2 mo of the study. Weekly distance run,
smoking habits and alcohol consumption during the study were
estimated by interview. Heart rate at submaximal workloads on a
From the Departments of Medicine and Preventive Medicine,
University of Otago Medical School, Dunedin, New Zealand.
Supported in part by the National Heart Foundation of New
Zealand and the Medical Research Council of New Zealand.
Received for publication July 15. 1980.
Address reprint requests to W. H. F. Sutherland, Department of
Medicine. P.O. Box 913, Dunedin. New Zealand.
0 1981 by Grune & Stratton, Inc.
0026-0495/81/3009-0001$01.00

Metabolism, Vol. 30, No. 9 (September), 1981 839

840 SUTHERLAND, WOODHOUSE, AND HEYWORTH

Table 1. Anthropometric Measurements, Exercise Parameters and Reported Alcohol Consumption in 20 Men
Participating in a Physical Training Programme in Preparation for a Marathon Race and Including a Detraining Period

n = 20 Mean + SEM

Subscapular Total +
6odV Skinfold Skinfold Reported Pulse Rate Reported
Weight Estimated Thickness Thickness Distance Run at 900 kp m/min Alcohol Consumption
kg % Body Fat mm mm mile+& beats/min alwk

Before training 72.9 + 1.9 14.2 + 0.8 11.3 + 0.7 33.2 f 2.3 17 ? 2 135 * 4 66 f 17
Immediately after
4 mo training 72.7 + 1.8 13.1 f 0.8* 10.5 f 0.9 30.3 + 2.5, 37 f 4 130+4t 58+ 17
End of 2 mo
detraining period 72.9 + 1.9 13.3 i 0.7 10.4 t 0.6 30.2 + 2.0 19 + 4 129 + 4 59 ? 14

Significance by paired t test, of the differences between mean values are as follows:
lP < 0.05, pre vs post,.
tP < 0.0 1, pre vs post, + Total skinfoldthicknessis the sum of measurements at the biceps, triceps, subscapular and suprailiac sites.

of constraints ZP = I = Xq, the expected correlation coefficient Table 1 shows that the men significantly increased
between, for example; 12:0 and 14:0 is given by:”
their weekly distance run during the training period.
p,*:o P,,:, This resulted in an increase in physical fitness indi-
J (1 - P,,:,) (1 - P14:o)
The correlation between changes in proportions of fatty acids 12:O
cated by the significant decrease in mean pulse rate at
submaximal workload after training. Thirteen men
and 14:O should be zero if and only if:
completed the marathon course, three ran approxi-
mately half the distance, three did not take part due to
-=Qw P,,:, physical injury and sickness and one person did not
1 - Qm 1- pm compete as he was unable to maintain the required
The following ratios were chosen to eliminate the constraints, to level of training. The increased physical training
measure meaningfully the changes in proportions and to satisfy altered body composition. Thus mean total skinfold
theoretical considerations concerning the uniqueness of correlation
thickness and the related percent estimated body fat
coefficients.
were significantly reduced by training (Table 1).
Y,xl(lOO~ - X14:0) However, mean body weight did not change signifi-
R (12:0, 14:O) =
&II (100% - Y,,:,) cantly throughout the study (Table 1).
To determine the correlation between changes in proportions of 12~0 Mean reported alcohol consumption did not vary
and 14:0 the correlation between R(12:0, 14:O)and R(14:0, 12:O)is significantly during the study (Table 1) and three
determined. subjects smoked tobacco but did not change their
Four correlation coefficients were calculated; (1) Pearson’s
smoking habits over this period.
(product-moment) on the original data (the X’s and Y’s), (2)
Pearson’s on the transformed data (the R’s), (3) Spearman’s p on The between chromatogram and between samples
the transformed data, and (4) Kendall’s T on the transformed data. variance for adipose tissue fatty acid proportions were
Typically, there was a remarkable consistency between statistical mostly small compared to the corresponding between-
significance of the non-parametric coefficients and Table 4 shows subject variances (Table 2). Thus variance derived
Spearman’s and Kendall’s rank correlation coefficients together
from chromatograms and samples are unlikely to
with their one-tailed significance.
influence the significant differences between mean
RESULTS values in Table 3.
Two subjects dropped out of the study at an early Table 3 shows that mean adipose tissue fatty acid
stage and their results are not included. Adipose tissue composition changed significantly during training and
was not obtained from one subject at the end of the remained altered during the post training period.
study. Mean proportions of lauric (12:0), myristic (14:0),

Table 2. Variance Estimates for the Gas Liquid Chromatographic Analysis of the Fatty Acid Composition
of Adipose Tissue Samples From Five Men*

Source of Variance 12:o l&O l&O l&107 18:O 18:lw9 18:2w6

Between chromatograms 0.0005 0.009 0.003 0.140 0.081 0.013 0.009

Between subjects 0.155 0.138 4.295 2.431 1.517 2.224 8.084
Between samples 0.007 0.065 0.029 0.097 0.245 0.696 0.245

*The sample of adipose tissue from each subject was divided into two pieces and two chromatograms were obtained for each sample.
EXERCISE AND ADIPOSE TISSUE FATTY ACIDS 841

Table 3. Adipose Tissue Fatty Acid Composition in 20 Men Table 5. Mean Change After Training in Adipose Tissue Fatty
Participating in a Physical Training Programme in Preparation for Acid Proportions in Tertiles of Pretraining Values for These
a Marathon Race and Including a Detraining Period Proportions in Men

n = 20 Mean k SEM % Tertks of Mean Change in

Pretraining Fatty Acid
Immediately After End of 2 mo
Fatty Acid Proportion
Fatty Acid Before Training 4 mo Training Detraining Perwd
Proportion Number (%)
(%I of Men t SEM
12:o 5.10 + 0.13t 5.74 * 0.11t 5.68 + 0.11
14:o 2.01 * 0.13’ 2.53 + 0.11 l 2.52 t 0.09 Laurie Acid

16:O 19.06 & 0.53’ 17.60 + 0.33’ 17.49 + 0.39 3.91 - 4.58 7 0.89 k O.lOT
16:lw7 10.16 2 0.53 9.82 + 0.32 9.74 ? 0.28 4.59 - 5.38 6 0.90 k 0.25
1810 7.30 + 0.31t 9.36 + 0.28t 9.10 ? 0.24 5.39 + 7 0.15 +_ 0.12t
18:lw9 37.14 + 0.9ot 33.68 + 0.55T 33.18 + 0.60
Myristic Acid
18:2w6 9.56 + 0.75t 11.25 + 0.57t 1 1.54 i 0.62
0.64 - 1.87 7 0.86 k 0.16’
Significance of differences by paired t test between the mean values 1.88 - 2.09 6 0.81 + 0.12
are as follows: 2.10 + 7 -0.07 + 0.21.
lP < 0.005, pre vs post,.
Palmitic Acid
tP < 0.001, pre vs post,.
14.63 - 17.88 7 0.26 + 0.67*

17.89 - 20.13 6 - 1.65 + 0.57

stearic (18:O) and linoleic ( 18:2w6) acids were signifi- 20.14 + 7 -3.05 + 0.38’
cantly raised and mean proportions of palmitic (16:O)
Stearic Acid
and oleic ( 18: 109) acids were significantly reduced at
5.22 - 6.95 7 3.26 + 0.33T
the end of the training period. 6.96 - 7.61 6 2.53 k 0.46
Several significant inter-relations between the 7.62 + 7 0.46 k 0.40t

changes in adipose tissue fattly acid proportions with Oleic Acid

training were noted (Table 4). In particular, the 30.31 - 34.47 7 1.07 i 0.62t
changes in palmitic acid and oleic acid proportions 34.48 - 39.87 6 -5.82 k 0.90
39.88 + 7 -4.73 -t 0.99t
were correlated at a high level of significance and each
was inversely correlated with the change in myristic Linoleic Acid

acid proportions. The changes in adipose tissue fatty 5.94 - 7.66 7 2.91 zk 0.85.

7.67 - 9.29 6 1.79 + 0.31

acid composition with training were not related to age,
9.30 + 7 0.46 + 0.20*
anthropometric measurements or changes in the
Significance of difference between mean values in highest and lowest
!atter.
tertiles is as follows:
The magnitudes of the changes in adipose tissue
lP < 0.01.
fatty acid proportions with training were influenced by TP < 0.001.

Table 4. Spearmans (p) and Kendalls ITI Rank Correlation the pretraining proportions of the fatty acids (Table
Coefficients Between Mathematically Transformed* Changes in
5). Thus lauric, myristic, stearic and linoleic acids
Adipose Tissue Fatty Acid Proportions After 4 mo Increased
Physical Training in 20 Men
showed significantly greater increases with training at
lowest compared to highest pretraining levels. Palmitic
Change I”
Adipose and oleic acids showed significantly greater decreases
TlSSW with training, at highest compared to lowest pretrain-
Change in Adipose Tissue Fatty Acid Proportion
Fattv Acld
1%)
ing proportions of the acids.
ix) 14:o 16:0 lB:O 18:lw9 18:2w6 Physical training altered the pretraining pattern of
12:o /I 0.611$ 0.192 0.412t 0.262 0.459t correlations between adipose tissue fatty acid propor-
T 0.495$ 0.116 0.284f 0.147 0.295t tions in men from the study group (Table 6). This
14:o p --0.463t 0.277 -0.409T 0.224
occurred mainly by the appearance of new correlations
-0.326t 0.200 -0.326T 0.147
16:0 -0.042 0.905$ 0.011
at the end of the training period rather than by the
P
7 iO.032 0.758$ -0.021 disappearance of existing correlations. For example,
18:0 P 0.084 0.597$ highly significant correlations between lauric acid and
0.053 0.41 l$ linoleic acid and between myristic acid and palmitoleic
18:lw9 p 0.283
acid became evident after training. Three interacid
T 0.189
correlations of less statistical significance also
*The method for calculating correlation coefficients between changes
appeared at this time. Four significant correlations,
in fatty acid proportions is given in the Subjects and Methods section.
Significance of correlations is as follows:
including the highly significant relationship between
tP < 0.05. palmitic acid and oleic acid proportions, remained
$P < 0.01. throughout the study. However, the levels of signifi-
642 SUTHERLAND, WOODHOUSE, AND HEYWORTH

Table 6. Net* Coefficients of Correlations among Adipose Tissue with reported changes in adipose tissue fatty acid
Fatty Acid Proportions in 20 Men Participating in e Physical
composition with training in patients with coronary
Training Programme in Preparation for a Marathon Race and
Including a Detraining Period
artery disease’ showed several differences. The
increase in mean linoleic and stearic acid proportions
Immediately End of
After 2mo after training seen in the present study were not
Before 4 mos Detraining observed in the patients. Also training induced an
Fatty Acids Training Training Period
increase in mean palmitic acid proportions in the
12:o vs 14:o 0.590% 0.503t 0.702$
patients’ whereas a decrease in the percentage of this
16:0 0.366 0.515t 0.226
16:lw7 0.494t 0.376t 0.42 1 t
acid was observed in the adipose tissue of the normal
16:O 0.129 0.22 1 0.256 men. Apart from the presence of coronary artery
16:lw9 0.356 0.322 0.062 disease, differences between the study groups in mean
16:2w6 -0.327 -0.629$ -0.506t values for age, initial adipose tissue fatty acid compo-
14:Ovs 16:0 -0.417t -0.032 -0.267 sition, diet and adiposity may be responsible.
16:lw7 0.346 0.675$ 0.619$ The magnitudes and the time course of the changes
16:0 0.245 -0.095 -0.003 in adipose tissue fatty acid composition in the present
16:lw9 0.370 -0.062 -0.300
study make it unlikely that altered diet composition is
16:2w6 -0.009 -0.240 -0.296
responsible for the changes in depot fat composition
16:0 vs 16:lw7 0.366 0.252 0.391t
with training. Previous studies have shown that feed-
16:O -0.321 -0.069 -0.333
16:lw9 0.910$ 0.919$ 0.919$
ing subjects experimental diets rich in linoleic acid for
16:2w6 0.240 -0.427t -0.367t example, could barely,” if at alLI6 produce the
changes seen in adipose tissue linoleic acid proportions
16:l vs 16:0 0.122 -0.405t -0.313
16:lw9 0.736% 0.415t 0.493t over the four months training period. Furthermore, it
16:2w6 -0.010 -0.466T -0.459t is improbable that diet compositions similar to the
16:Ovs 16:1o9 -0.276 -0.133 -0.300
experimental diets’5,‘6 were attained as result of
16:2w6 0.622$ 0.426t 0.325 increased physical activity.
The most likely explanation of the changes in
16: 1o9 vs 16:2w6 0.024 -0.317 -0.251
adipose tissue fatty acid composition with training
*Net coefficient determined from the difference between the
involves preferential mobilisation of palmitic and oleic
observed coefficient and the expected coefficient using transformation to
zeta equivalents.‘3
acids and preferential retention of stearic and linoleic
Significance levels are as follows: acids. The preference of palmitic acid for the (Yand (Y‘
tP < 0.05. positions of the human depot triglyceride molecules
iP < 0.01. and Iinoleic acid for the ,6 position,” coupled with
probable specificity of tissue lipases for ester linkages
in the LYand (Y’positions,‘8 may explain the opposite
cance of two of these correlations decreased, one
changes in mean proportions of these acids in adipose
increased and one was unchanged over the study
tissue with training. Similarly the major contribution
period. A majority of significant correlations observed
of oleic acid to the LYand LY’positions of depot triglycer-
after four months training were also found at the end
ides” could result in preferential release of this acid
of the detraining period. Thus, several correlations
from adipose tissue. Thus the changes in palmitic acid
which first appeared after training persisted to the end
and oleic acid proportions with training were highly
of the study.
significantly correlated (Table 4). Another mecha-
nism which may account for the decreased proportions
DISCUSSION of oleic acid in adipose tissue with training is preferen-
The results show that changes in adipose tissue fatty tial extraction of the acid from arterial blood by
acid composition occurred in 20 men after physical muscle tissue.” Preferential retention of stearic acid in
training and persisted for two months in which most depot fat, during increased release of fatty acids from
men reduced their level of training. this tissue with training, may result from reduced
Previously reportedI differences in skeletal muscle accessibility to tissue lipases of ester linkages involving
fatty acid composition between trained and untrained stearic acid.” However, the present results show that
normal men were similar in some respects to our for each fatty acid there is a range of pretraining
findings. Thus mean palmitic acid proportions were proportions in which the mean change in the fatty acid
significantly lower in skeletal muscle of trained proportions with training is markedly reduced, if not
compared to untrained subjects.14 reversed, compared to changes at other pretraining
A comparision of the results in the present study levels (Table 5). This suggests that factors such as
EXERCISE AND ADIPOSE TISSUE FATTY ACIDS 843

positional specificity of fatty acid distribution in depot A previous studyz2 has suggested that patterns of
triglycerides’do not influence the relative amounts of interacid correlations among adipose tissue fatty acid
fatty acids released from adipose tissue during train- proportions may vary characteristically with national
ing, at all levels of pretraining fatty acid proportions. populations and may be associated with population
Altered body composition and body weight have differences in coronary risk and diet. The pattern of
been associated with changes in adipose tissue fatty inter-acid correlations in men preparing for a mara-
acid compositions in men.” However, in the present thon run was different from those reported for Ameri-
study, mean body weight did not change significantly, can or Japanese men” and was altered by physical
and while there was a significant decrease in total training (Table 6). After training, however, the inter-
skinfold thickness with training (Table l), this change acid correlation pattern for men in the present study
did not relate to any changes in adipose tissue fatty showed an increased number of correlations in
acid composition. common with those reported for Japanese men.22 This
An increase in adipose tissue linoleic acid propor- includes the appearance of correlations between
tions with physical training may contribute to reduced proportions of palmitic acid and linoleic acid and
risk of CHD14 associated with regular physical activ- between palmitoleic acid and both stearic and linoleic
ity by predisposing towards reduced levels of blood acid proportions. The trend with training towards a
pressure.” This proposal is consistent with the findings pattern more similar to Japanese men may have impli-
of the Edinburgh-Stockholm study6*” in which adipose cations with regard to risk of developing CHD since
tissue linoleic acid proportions and levels of physical Japanese men had a low risk of developing this
fitness were significantly higher while blood pressure disease.*’
and incidence of CHD were significantly lower in It seems therefore that high level exercise can alter
Stockholm men compared to men from Edinburgh. the fatty acid composition and interrelations of
The proposed relevance of specifically linoleic acid to adipose tissue in a direction which may favour reduced
factors such as blood pressure and risk of developing incidence of CHD. Further study of the possible links
CHD may relate to the synthesis from this acid of between physical activity, adipose tissue linoleic acid
vasoactive substances such as prostaglandins. proportions, prostaglandins, blood pressure and risk of
Using the mean value for adipose tissue Iinoleic acid developing CHD are warranted.
proportions for Edinburgh men (7.3%) and our results
in Table 5, it can be shown that this mean proportion
of linoleic acid could be raised by physical training to ACKNOWLEDGMENT
nearly the mean value for Stockholm men (I 1.3%).
The excellent technical assistance of S. Williamson and B. Smith
This demonstrates that the magnitude of the mean is gratefully acknowledged. The authors also thank the participants
change in adipose tissue linoleic acid proportions with in the study as well as Associate Professor E. R. Nye for helpful
training has potential physiological significance. comments.

REFERENCES
I. Epstein L. Miller GH, Stitt FW, et al: Vigorous exercise in tissue and increased serum cholesterol of coronary patients follow-
leisuretime, coronary risk factors and resting electrocardiograms in ing training. Canad Med Assoc J 109:194-197. 1973
middle-aged civil servants. Br Heart J 38:403-409. 1976 9. Astrand P-O, Ryhming I: A nomogram for calculation of
2. Kannel WB, Gordon T. Sorlie P, et al: Physical activity and aerobic capacity (physical fitness) from pulse rate during submaxi-
coronary vulnerability: the Framingham Study. Cardiol Dig 6:28- mal work. J Appl Physiol 7:218-221, 1954
40,197l
10. Durnin JVGA, Womersley J: Body fat assessed from total
3. Morris JH, Chave SPW, et al: Vigorous exercise in leisuretime
body density and its estimation from skinfold thickness measure-
and the incidence of coronary heart disease. Lancet 1:333-339,
ments on 481 men and women aged from 16 to 72 years. Br J Nutr
1973
32177-97, 1974
4. Paffenbarger RS, Hale WE: Work activity and coronary heart
mortality. N Engl J Med 292545-580, 1975
11. Folch JM, Lees M, Sloan-Stanley GH: A simple method for
the isolation and purification of total lipids from animal tissues. J
5. Sutherland WHF, Woodhouse SP: Physical activity and
Biol Chem 226:497-509, 1957
plasma lipoprotein lipid concentrations in men. Atherosclerosis
37~285-292. 1980 12. Kingsbury KJ, Morgan DML, Brett CG: Gas-liquid chroma-
6. Logan Rt, Riermersma RA, Thomson M, et al: Edinburgh- tographic analysis of the fatty acid composition of the plasma
Stockholm. Study of risk factors for ischaemic heart disease in cholesterol esters and its relevance to atherosclerosis. Biochem Med
healthy men. Lancet 1:949-954,1978 18:37-51, 1977
7. Oliver MF. Nimmo IA, Cook M, et al: Ischaemic heart disease 13. insuli W, Bartsch GE: Fatty acid composition of human
and associated risk factors in 40 year old men in Edinburgh and adipose tissue related to age, sex and race. Am J Clin Nutr
Stockholm. Europ J Clin Invest 5:507-5 14,1975 20: 13323, 1967
8. Allard C, Alteresco M, Ferguson RJ, et al: Changes in adipose 14. Thomas TR, Londeree BR, Gerhardt KO, et al: Fatty acid
 844 SUTHERLAND, WOODHOUSE, AND HEYWORTH

profile and cholesterol in skeletal muscle of trained and untrained 19. Tancredi RG, Dagenais CR, Zierler KH: Free fatty acid in
men. J Appl Physiol43:709-7 13, 1977 the forearm at rest: muscle uptake and adipose tissue release of free
15. Hirsch J, Farquar JW, Ahrens Jr EH, et al: Studies of fatty acids. The Johns Hopkins Med J 138:167-176, 1976
adipose tissue in man-a micro-technic for sampling and analysis. 20. McDonald I, Barry PJC: Changes in fatty acid pattern of
Am J Clin Nutr 8:499-511,196O adipose tissue of obese subjects while on a reducing regimen. Am J
16. Dayton S, Pearce ML, Hashimoto S, et al: A controlled Clin Nutr 15: 158-I 60, I964
clinical trial of a diet high in unsaturated fat in preventing complica-
tions of atherosclerosis. Circulation (suppl. II) 40:1-63, 1969 21. Oster P, Arab L, Schellenberg B, et al: Blood pressure and
17. Brockerhoff H: Stereospecific analysis of triglycerides: An adipose tissue linoleic acid levels in man. Res Exp Med (Berlin)
analysis of human depot fat. Arch Biochem Biophys 110586-592, 175:287-291, 1979
1965 22. Insull W, Lang PD, Hsi BP, et al: Studies of arteriosclerosis
18. Bjorntorp P, Furman RH: Lipolytic activity in rat epididy- in Japanese and American men. I Comparison of fatty acid compo-
ma1 fat pads. Am J Physiol203:316-322, 1962 sition of adipose tissue. J Clin Invest 48:1313-1327, 1969