Total body lipid and triglyceride response to energy
deficit: relevance to body composition models
RENEE COMIZIO,1 ANGELO PIETROBELLI,1 YAN XIU TAN,1 ZIMIAN WANG,1
ROBERT T. WITHERS,2 STEVEN B. HEYMSFIELD,1 AND CAROL N. BOOZER1
1Obesity Research Center, St. Luke’s-Roosevelt Hospital, Columbia University College of Physicians
and Surgeons, New York, New York 10025; and 2Exercise Physiology Laboratory, School of
Education, The Flinders University of South Australia, Adelaide 5000, Australia
total body fat; adipose tissue; animal models
LIPIDS represent a diverse group of chemical compounds
that in most mammalian species comprise a large
fraction of body weight (14). Lipids are usually classified according to physical or chemical properties, and
the main recognized groups include triglycerides, phospholipids, sphingolipids, cholesterol, and waxes (8, 14).
An alternative classification is to divide lipids into two
groups, those that are essential for sustaining life and
those that are nonessential (8, 14). Essential or ‘‘structural’’ lipids include phospholipids, sphingolipids, cholesterol, and other lipid species that are necessary for
maintaining cellular integrity, cell membrane fluidity
and function, and other indispensable processes. The
nonessential lipids in rodents, humans, and other
E860
mammals are made up almost entirely of triglycerides
(8, 14).
Lipids all share in common solubility in nonaqueous
solvents such as diethyl ether, petroleum ether, acetone, and methanol (8). Nonpolar lipids such as triglycerides are usually bound in tissues by weak Van der
Waals forces or hydrophobic bonds and can be extracted
with ether and other nonpolar solvents (14). More polar
lipids, such as phospholipids, may in part be bound to
proteins by hydrogen bonds and electrostatic associations that require polar solvents such as acetone and
methanol for disruption and extraction. The nature of
tissue-lipid isolates is therefore directly related to the
employed solvents and associated tissue preparation
procedures.
Substantial confusion surrounds the concept of lipid
as a human body composition component (14). Various
terms are applied to the lipid-solvent extractable material observed in vivo, including ‘‘lipids,’’ ‘‘fat,’’ and
‘‘essential’’ and ‘‘nonessential lipids.’’ Another common
problem is that the term fat is often used in reference to
two different but closely related components, lipids and
adipose tissue.
Several widely used body composition methods rely
on lipid-related assumptions, and the prevailing confusion creates problems ranging from inaccurate body
composition models to incorrect terminology in research reports. For example, the classic molecular-level
body composition model includes five components: fat,
protein, water, mineral, and glycogen (16). It is not
always clear in this model whether fat represents
triglycerides or total lipid. If triglycerides are considered fat, then what provision is made in the model for
nontriglyceride lipids?
In the classic studies by Pace and Rathbun (22, 25) of
total body water and other components in guinea pigs,
fat was the total petroleum ether-extractable material
from whole carcass. The ‘‘fat-free body mass’’ in this
study therefore contained some lipids such as sphingomyelin, a compound that is soluble in hot absolute
alcohol and insoluble in ether, acetone, and water (14).
The extraction method employed by Rathbun and Pace
thus does not allow us to precisely ascertain whether
the classic ratio of total body water to fat-free body
mass of 0.732, compiled by the investigators for several
species in addition to the guinea pig, is the ratio of total
body water to triglyceride or lipid-free mass (28).
As another example of the prevailing confusion, the
classic underwater weighing method is based on the
two-compartment molecular-level model that consists
of fat and fat-free body mass. This method assumes
0193-1849/98 $5.00 Copyright r 1998 the American Physiological Society
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
Comizio, Renee, Angelo Pietrobelli, Yan Xiu Tan,
Zimian Wang, Robert T. Withers, Steven B. Heymsfield,
and Carol N. Boozer. Total body lipid and triglyceride
response to energy deficit: relevance to body composition
models. Am. J. Physiol. 274 (Endocrinol. Metab. 37): E860–
E866, 1998.—Although the study of human body composition
is advancing rapidly, confusion still prevails regarding the
molecular-level lipid component. Most molecular-level body
composition models are presently based on the overall hypothesis that nontriglyceride lipids constitute an insignificant
proportion of total body lipid. A single lipid or ‘‘fat’’ component
consisting of triglycerides is thus assumed in most molecularlevel body composition models. To test this hypothesis, the
present study, carried out in adult rats, was designed to
examine two questions: 1) What is the proportion of total
lipids as triglycerides? and 2) Is this proportion constant or
does it change with negative energy balance and weight loss
produced by calorie restriction and increased exercise? Results indicated that with negative energy balance and weight
loss there were progressive losses of total body triglyceride
and lipid. The proportion of total lipids as triglyceride was
0.83 6 0.08 (SD) in control animals, with reductions at 2 and
9 wk of energy restriction [0.82 6 0.04 (P 5 NS vs. control)
and 0.70 6 0.15 (P 5 0.05)] and at 9 wk for energy restriction
plus exercise [0.67 6 0.09 (P 5 0.003)]. Nontriglyceride lipids
comprised 2.8% of carcass weight at baseline and decreased to
2.2% by 9 wk of energy restriction and exercise (P 5 NS).
Substantial differences were observed between body composition ratios expressed as percentages of the lipid-free body
mass (LFM) and triglyceride-free body mass (TGFM); (e.g.,
total body water/LFM and TGFM in controls 5 72.7 6 0.7 and
70.4 6 2.2, respectively; P 5 0.02). These observations
strongly support the existence and importance of nontriglyceride lipids as a body composition component that responds
independently from storage triglycerides, with negative energy balance produced by food restriction and exercise.
BODY COMPOSITION LIPIDS
E861
METHODS
Study Hypothesis
Two specific questions were examined in the present study:
1) What is the proportion of total lipids as triglycerides in the
adult female rat? and 2) Is this proportion constant or does it
change with negative energy balance and weight loss produced by calorie restriction and increased exercise? These
questions are based on the overall hypothesis that nontriglyceride lipids constitute an insignificant proportion of total
body lipid. If the hypothesis is inaccurate and there is a large
nontriglyceride lipid fraction, and if this component behaves
with interventions in a manner distinct from triglycerides, it
would suggest an expanded role for these lipids in molecularlevel body composition models and in future body composition
studies.
The two main lipid-related questions were explored in
experimental animals before and after energy deficits that
varied in duration and magnitude. Other body composition
components were also estimated to explore their relationship
to lipid components during the experimental protocol phases
with varying levels of negative energy balance.
Protocol
The experimental procedure involved two phases, an 8-wk
weight gain phase followed by a 9-wk food restriction-weight
loss phase (Fig. 1). During the weight gain phase, rats were
fed a defined diet that provided 45% of total energy as fat. At
the end of the 8-wk weight gain phase, rats were divided into
four weight-matched groups. One group was euthanized to
serve as the baseline control group (n 5 7). During the weight
loss phase, the remaining three groups of animals were food
restricted by placing them on an energy intake equal to 75%
of the energy intake at the end of the ad libitum period. The
diet was low in fat (12% of total energy) and meals were
provided twice daily. The three groups of animals were
randomly assigned to the restricted diet for 2 (n 5 5) or 9 (n 5
8) wk and an additional group to restricted diet plus exercise
(n 5 7). The exercise group swam three times a week at 1 PM
for up to 35 min.
Female Sprague-Dawley rats were housed individually in a
temperature- and humidity-controlled room with a 12:12-h
light-dark cycle. Rats were fed twice per day, at 9 AM and 4
PM. Body weight was recorded three times per week. Animals
were euthanized at the end of 2- or 9-wk protocols, and
carcass weights were recorded after removal of hair and
gastrointestinal contents.
Fig. 1. Study experimental design and protocol.
Analytic Methods
Tissue preparation. Euthanized carcasses were frozen at
210°C. At study completion, 500 ml of distilled water were
added to the frozen carcasses, which were placed in a
preweighed beaker and heated to 125°C for 60 min. The
cooled carcasses were homogenized using a large-bore polytron (PT6000; Brinkman Instruments, Westbury, NY) for
7–10 min, and a 45-ml aliquot was then stored at 210°C.
Total body lipid. The following procedure for total carcass
lipid purification is based on previously established methods
of lipid purification (3, 24). Approximately 1 g of carcass
homogenate and 27 ml of methanol were added to an Erlenmeyer flask. The sample was stirred, 14 ml of chloroform were
added at 30 min, and the mixture was set aside overnight.
Samples were run in triplicate.
Clean conical Teflon-lined glass capped tubes were heated
to 90°C and weighed after cooling to room temperature. The
homogenate mixture was filtered into the preweighed tubes,
and the flask was rinsed with 2:1 chloroform-methanol. The
volume of liquid in the tubes was then compared with a
preprepared capped tube of known volume (21 ml). The
volume of sample was then made up with 2:1 chloroformmethanol, and 4.2 ml of 0.88% KCl solution were added as the
wash (3, 14, 24, 31). The liquid mixture was homogenized for
,20 s, and samples were then centrifuged at 1,500 rpm for 10
min. The top layer was aspirated and discarded, and the
tubes were placed in a 55–60°C water bath until volume
decreased to ,4 ml. Tubes were then dried in an oven at 90°C
for several hours until only an oily lipid layer remained.
Tubes were then cooled, and lipid weight per sample was
obtained by calculating the difference between the conical
tube weight with and without lipid. Triplicate values of total
lipid weight per gram of homogenate were averaged and
multiplied by carcass weight to obtain total body lipid.
Total body triglyceride. To prevent autoxidation of the
lipids, a solution of 15 mg of butylated hydroxytoluene (BHT),
an antioxidant, in 200 ml chloroform was prepared and used
as the storage solvent (7, 14, 31). Four milliliters of BHT in
chloroform solution were added to the extracted lipid, and the
solution was then poured into glass vials. Nitrogen gas was
infused into the vials, which were then capped (7, 14, 31).
The procedure for triglyceride quantification was adapted
from the colorimetric serum triglyceride assay kit of Diagnostic Chemicals (Oxford, CT). The kit contained a triglyceride
standard of 2 mM triolein (177.08 mg/dl). Two additional
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
that fat has a density of 0.9000 g/ml, which is based on
ethyl ether-extractable material from adipose tissue
(10, 21). The material obtained from adipose tissue
extraction with ethyl ether as a solvent is almost
entirely triglyceride, and, accordingly, nontriglyceride
lipids are not included in the fat component. The
underwater weighing method of fat-free body mass
makes no provision for nontriglyceride lipids and
thereby assumes their amount is negligible or that they
do not contribute measurably to body density.
The current study is the first in a series aimed at
expanding knowledge of the lipid portion of body mass
with the purpose of clarifying some of the abovementioned ambiguities. We selected the rodent model
as one that allows complete carcass analysis and an
opportunity for experimentally manipulating energy
and lipid balance by dietary and exercise interventions.
E862
BODY COMPOSITION LIPIDS
Fig. 2. Weight curves for the 3 experimental groups
over the 2- and 9-wk protocols. Vertical bars are SEs.
Statistical Methods
The study hypothesis was tested by examining the triglyceride-to-total body lipid ratio with progressive levels of energy deficit. Specifically, the hypothesis states that no statistically significant change in the triglyceride-to-total lipid ratio
would be observed with energy restriction. Three depletion
levels were examined (2 wk, 9 wk, and 9 wk 1 exercise), and,
using the Bonferroni test for multiple comparisons (17), we
accepted P # 0.0167 (i.e., 0.05/3) as statistically significant.
Student t-tests were used to compare each time point with the
baseline triglyceride-to-total lipid ratio.
Other statistical tests used in the study were of a more
exploratory and descriptive nature. Student t-tests were used
to compare the most extreme energy-deficit group (i.e., 9-wk
low-energy diet plus exercise) with the control group. These
statistical analyses are provided in Tables 1 and 2 but are not
discussed in the text.
Results are expressed in the text and Tables 1 and 2 as
group means 6 SD and in Fig. 2 as group means 6 SE of the
estimate. All statistical calculations were performed using
the SAS statistical software (Carey, NC) package for personal
computers.
RESULTS
Body Weight Kinetics
Weight curves for the three chow-restricted groups
are presented in Fig. 2. The 2- and 9-wk weight losses
for the low-energy diet groups were 17 6 5.4 and 34 6
9 g, respectively. Expressed as a percentage of prediet
weight, the two low-energy diet groups lost 5.5 6 0.3
and 10.9 6 0.3% of initial weight, respectively. More
weight was lost by the 9-wk low-energy diet plus
exercise group (43 6 10.3 g; 14.1 6 0.1%; P , 0.001 vs.
control) than the comparable 9-wk nonexercise group.
Lipid Changes
The results of total lipid and triglyceride analyses for
the four groups of rats are presented in Table 1. In the
control group, total lipid and triglyceride constituted
17.1 and 14.3% of carcass weight, respectively. Of the
total amount of lipid present, 83% was triglyceride.
Nontriglyceride lipid, the difference between total lipids and triglyceride, was 7.6 6 4.2 g or 2.8% of body
weight.
Both total body lipid and triglyceride decreased with
lengthening energy restriction (9 wk . 2 wk) and
energy deficit (9-wk exercise . 9-wk nonexercise). At 9
wk of energy restriction in the nonexercise group,
weight loss, total lipids, and triglycerides were reduced
by 10.9, 50.4, and 57.7%, respectively, compared with
the control group. In the greater-energy-deficit 9-wk
exercise group, the corresponding reductions were 14.1,
67.1, and 72.0%, respectively.
With progressive energy deficit, the relative reduction in triglyceride exceeded the lowering of total lipids,
and there was a decline in the triglyceride-to-total lipid
ratio (Table 1) from control (0.83 6 0.08) to 2 wk of
energy restriction (0.82 6 0.04; P 5 0.80), 9 wk of
energy restriction (0.70 6 0.15; P 5 0.05), and energy
restriction with exercise (0.67 6 0.09; P 5 0.003;
significant with Bonferroni correction) (17). Absolute
nontriglyceride lipids also decreased with energy restriction and were 2.7% of carcass weight in the 9-wk
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
standards were prepared by using corn oil purified by first
mixing with Zeolite (Sigma Chemical, St. Louis, MO). The
high standard contained 3.31 mM (293.3 mg/dl) purified corn
oil, and the low standard contained 0.99 mM (87.73 mg/dl)
purified corn oil. Samples for the triglyceride assay were
prepared in dilutions so that the estimated triglyceride
concentration (,80% of total lipids) would be close to that of
the standard solution.
Other body composition components. Total body water was
determined by drying duplicate 1-g samples of homogenate
overnight at 90°C to stable weight. Total body nitrogen was
estimated by adjusting homogenates to pH 2 with hydrochloric acid and then measuring nitrogen with an adaptation of
the Kjeldahl method (23). Protein was calculated by assuming a nitrogen-to-protein ratio of 0.16 (19). Total body ash was
measured by placing ,3 g of well-stirred carcass homogenate
in crucibles. The crucibles were then heated to 90–95°C
overnight in a drying oven, placed into a muffle furnace for
,4 h at 600°C, and cooled, and the ash weight of the sample
was then determined. Water, protein, and ash contents per
gram of homogenate were multiplied by carcass weight to
obtain total body water, protein, and ash, respectively.
E863
BODY COMPOSITION LIPIDS
Table 1. Results of body composition studies
Group
(n)
Carcass
Mass, g
Total
Lipid, g
TG, g
TG/Total
Lipid, %
Total
Lipid/
BW, %
TG/BW, %
273.8 6 19.2 47.4 6 14.8 39.7 6 14.6 0.83 6 0.08 17.1 6 4.6 14.3 6 4.5
260.9 6 11.2 30.9 6 6.5 25.2 6 5.2 0.82 6 0.04 12.8 6 1.6 9.6 6 1.7
t* 5 0.25
P 5 0.80
Restricted 244.3 6 13.4 23.5 6 9.1 16.8 6 7.8 0.70 6 0.15 9.5 6 3.2 6.8 6 2.8
9 wk (8)
t 5 2.05
P 5 0.05
Restricted 232.0 6 11.6 15.6 6 6.0 11.1 6 5.2 0.67 6 0.09 6.9 6 2.4 4.7 6 2.1
9 wk 1 ex- t* 5 4.93
t 5 5.27
t 5 4.88
t 5 3.52
t 5 5.20
t 5 5.12
ercise (7) P 5 0.0003 P 5 0.0002 P 5 0.0004 P† 5 0.003 P 5 0.0002 P 5 0.0003
Control (7)
Restricted
2 wk (5)
Non-TG
Lipid, g
Non-TG
Lipid/
Total
Total Body
BW, % Protein, g Water, g
Mineral
Ash, g
7.6 6 4.2 2.8 6 1.5 51.5 6 2.8 164.5 6 4.8 5.5 6 1.6
8.4 6 0.9 3.2 6 0.3 52.5 6 2.1 166.0 6 4.5 5.4 6 0.8
Residual
Mass, g
4.8 6 7.2
3.3 6 2.8
6.7 6 3.6 2.7 6 1.4 49.3 6 2.4 166.3 6 6.5 6.2 6 1.7 21.0 6 2.7
5.1 6 1.5 2.2 6 0.6 45.9 6 4.7 162.6 6 5.4 6.2 6 1.1
t 5 1.48 t 5 0.98 t 5 0.73
t 5 0.69
t 5 20.95
P 5 0.16 P 5 0.35 P 5 0.47
P 5 0.49 P 5 0.35
1.1 6 3.9
t 5 1.2
P 5 0.26
Values are means 6 SD. BW, body weight; TG, triglyceride. t *, Student’s t-tests comparing control group vs. experimental groups. † Statistically
significant with Bonferroni correction for multiple comparisons.
Other Component Changes
The results of other component analyses are presented in Tables 1 and 2. The last column in Table 1
represents residual mass, which is the difference between body weight and all measured components. This
unmeasured component represents mainly glycogen,
although small amounts of other unmeasured substances are also included.
Lipid-free mass (i.e., body weight 2 total lipids) and
TGFM both decreased at 9 wk of energy restriction
relative to the control group, without and with exercise.
Compared with losses of lipid, the relative lowering of
lipid-free mass and TGFM with energy restriction was
substantially less. For example, compared with the
control group, lipid-free mass decreased by 2.5% (P 5
0.19) and 4.7% (P 5 0.02) at 9 wk of energy restriction
without and with exercise, respectively.
Hydration, defined as the ratio (in percent) of total
body water (TBW) to lipid-free mass and TGFM, differed between denominators (i.e., lipid-free mass and
TGFM) and also changed progressively with energy
restriction. Mean hydration was 72.7% relative to
lipid-free mass and 70.4% relative to TGFM in the
control group, a significant (P 5 0.02) difference that
persisted throughout energy restriction. For both forms
of hydration expression, relative water content increased as animals lost weight with energy depletion.
For example, relative to lipid-free mass, hydration
increased from a mean of 72.7% in the control group to
75.4% with energy restriction plus exercise (P , 0.0001).
As with hydration, other components also changed
relative to TGFM with energy restriction (Table 2). The
general pattern was for water and mineral ash to
increase and for protein, nontriglyceride lipids, and
residual mass to decrease relative to TGFM when
control and energy restriction groups were compared.
The relative contributions of the various components to
total TGFM were substantially different with extreme
weight loss from the control animals.
DISCUSSION
To our knowledge, this is the first study to systematically explore both the total lipid and triglyceride components of body weight in either rodents or humans under
various dietary and exercise conditions. Our results
clearly show that triglycerides are not all of total lipids
and that triglycerides do not maintain a constant
proportionality to total lipids with energy restriction
produced by food deprivation and exercise.
Table 2. Results of relative changes in lipid and triglyceride-free mass with energy depletion
Group
LFM, g
TGFM, g
TBW/LFM, %
TBW/TGFM, %
Total Protein/
TGFM, %
Non-TG Lipid/
TGFM, %
Mineral Ash/
TGFM, %
Residual Mass/
TGFM, %
Control
Restricted 2 wk
Restricted 9 wk
Restricted 9 wk
1 exercise
226.4 6 7.6
230.3 6 7.1
220.8 6 8.1
215.8 6 7.5
t* 5 2.63
P 5 0.02
234.0 6 10.1
235.7 6 7.5
227.6 6 9.5
220.9 6 7.7
t 5 2.72
P 5 0.01
72.7 6 0.7
72.1 6 0.9
75.3 6 1.1
75.4 6 1.0
t 5 25.85
P , 0.0001
70.4 6 2.2
70.5 6 1.0
73.1 6 1.6
73.6 6 1.4
t 5 23.25
P 5 0.007
22.0 6 1.2
22.3 6 0.5
21.7 6 1.4
20.8 6 1.8
t 5 1.47
P 5 0.16
3.3 6 1.8
3.6 6 0.3
2.9 6 1.5
2.3 6 0.7
t 5 1.37
P 5 0.19
2.4 6 0.6
2.3 6 0.4
2.7 6 0.5
2.8 6 0.4
t 5 21.47
P 5 0.16
2.0 6 3.0
1.4 6 1.2
20.5 6 1.2
0.5 6 1.8
t 5 1.13
P 5 0.28
Values are means 6 SD. LFM, lipid-free mass; TGFM, fat-free mass; TBW, total body water. t *, Student’s t-tests comparing control group vs.
experimental groups.
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
nonexercise group and 2.2% (P 5 NS) in the exercise
group compared with 2.8% in the control group. Finally,
a trend analysis (17) across the four experimental
conditions yielded a significant linear decrease (P 5
0.005) in the triglyceride-to-total lipid ratio.
Nontriglyceride lipids are part of the triglyceridefree body mass (TGFM), which is defined as the difference between body weight and triglyceride. Nontriglyceride lipids were 3.3% of TGFM (Table 2) in the control
group and increased with energy restriction at 2 wk
(3.6%, P 5 NS) and decreased to 2.9% (P 5 NS) at 9 wk
of energy restriction without exercise. With exercise
added to energy restriction, the fraction of TGFM as
nontriglyceride lipids declined further to 2.3% (P 5
0.01).
E864
BODY COMPOSITION LIPIDS
Linkage with Earlier Studies
The definition of lipid or fat components in the field of
body composition research has hitherto been vague and
inconsistent. These inconsistencies are spread across
the widely used methods of estimating body fatness,
including the total body water, underwater weighing,
total body potassium, in vivo neutron activation, and
imaging methods (26).
Specifically, Pace and Rathbun (22) first introduced
the total body water method of estimating fat and
‘‘fat-free body weight’’ or ‘‘lean body mass’’ in 1945,
when they demonstrated a mean TBW-to-lean body
mass ratio (TBW/lean body mass) of 0.724 in 50 guinea
pigs that was similar to the TBW/lean body mass of
0.732 found in a compilation of five other animal
species ranging in size from the rat to the monkey (22).
Fat in this classic study was extracted from the guinea
pigs with petroleum ether, and the residual mass thus
includes an unspecified amount of phospholipids, sphingolipids, and other polar lipid classes (25). In a thorough composite review of the literature on three male
human cadavers, Brozek et al. (4) estimated mean
TBW-to-fat-free body mass ratio as 0.737. Fat in the
three evaluated cadavers was extracted from whole
body homogenates by use of ether extraction methods
similar to those employed by Pace and Rathbun. These
classic studies therefore provided primarily triglyceride estimates as the fat component. In the present
study we observed in control animals a TBW-to-lipidfree mass of 0.727 and TBW/TGFM of 0.704. Hence, our
comparable estimate, the ratio of TBW to TGFM, is
lower than ,0.72–0.74 as observed in earlier studies.
Nevertheless, our findings emphasize that hydration
estimates are critically sensitive to the employed extraction method and quantified lipid components.
Variation in lipid terminology can also be found in
relation to the underwater weighing method. Specific
gravity as a measure of fatness was first suggested by
Bull in 1896 (5) and 1897 (6). Tester (32) in 1940
reported specific gravity as a means of quantifying the
ether (i.e., oil) extract of Pacific herring. Behnke and
co-workers (1, 2) extended this work to humans in 1942
and suggested a two-component model consisting of fat
and lean body mass. Lean body mass was defined by
Behnke as including an ‘‘undetermined and probably
constant percentage (2–3%) of essential lipids in bone
marrow, the central nervous system, and other organs’’
(1). This definition is consistent with the study of
Fidanza et al. (10), in which the density of fat was
measured on the ether extract of surgically harvested
human adipose tissue samples. Very little nontriglyceride lipid would be present in ether extracts of adipose
tissue. Fidanza included this human data along with
comparable animal data to arrive at a fat density of
0.900 g/ml at 37°C (10). The analysis by Brozek et al. (4)
of the three human cadavers, however, provided a
density estimate of lipid-free mass (i.e., sum of water,
proteins, and minerals) of 1.100 g/ml. They used Fidanza’s fat density to develop their two-component model,
and therefore no consideration was given to nontriglyceride lipids in the underwater weighing method (4).
Later workers such as Siri (28a) used the original
density estimates developed by Fidanza et al. in developing their own body composition models. Varying
terminology is thus applied to the underwater weighing
method, and the models now in use make no clear
provision for nontriglyceride lipids.
Forbes et al. (11) first reported the total body potassium method of evaluating body composition in 1961.
This method was based on the assumed potassium
constancy of lean body mass. According to Forbes et al.,
lean body mass is a ‘‘term taken to mean body weight
minus chemically determined neutral fat’’ (11). The
tissues from four cadavers cited by Forbes and Lewis
(12) were extracted using ether, and these authors’
reference to neutral fat is indeed consistent with tissue
extraction of mainly triglycerides. Lean body mass as
originally defined by Forbes is thus largely TGFM. The
total body potassium approach thus appears to quantify total body triglycerides and TGFM.
There are a number of in vivo neutron activation
analysis models in current use (18, 20), but the most
widely cited model is the one proposed by Cohn et al.
(9), in which total body fat is calculated as the difference between body weight and the sum of total body
water, proteins, and minerals. With the assumption
that glycogen is a minor component in the fasting
subject, Cohn’s model estimates total body lipids and
lipid-free mass (9).
Imaging methods such as computed tomography and
magnetic resonance imaging (MRI) provide whole body
adipose tissue estimates (13, 15, 27). Although technically the difference between body weight and total body
adipose tissue is adipose tissue-free body mass, Sjostrom (29) prefers to term this compartment lean body
mass. Adipose tissue-free body mass should, under
most circumstances, be roughly equivalent to triglyceride-free body mass (the latter includes water from
adipose tissue). The main problem with use of the term
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
Specifically, our findings indicate that, in animals fed
a defined ‘‘weight gain’’ chow, triglycerides represented
83% of total lipids, whereas the remaining 17% presumably represented phospholipids, sphingolipids, and other
nontriglyceride lipid classes. Although absolute amounts
of both triglycerides and nontriglyceride lipids decreased with energy restriction, there was a lowering of
the triglyceride-to-total lipid ratio in the animals from
control (0.83 6 0.08) to 9 wk of energy restriction
without (0.70 6 0.15) and with exercise (0.67 6 0.09).
Hence, with severe weight loss brought about by tissue
energy-depleting methods, approximately one-third of
total body lipid was in the form of nontriglyceride
chemical moieties. Although the precise mechanisms
accounting for the differential rate of lipid loss are
unknown, we presume that triglycerides are mobilized
with energy deficit at a relatively greater rate than the
essential nontriglyceride lipid compartment. Regardless of the underlying mechanisms accounting for the
differing rates of lipid loss with energy restriction, our
findings indicate that the proportion of total lipids as
triglycerides cannot be assumed constant.
BODY COMPOSITION LIPIDS
Body mass 5 triglyceride (nonessential lipids)
1 nontriglyceride (essential) lipids 1 water 1 protein
1 glycogen 1 minerals
On the basis of this suggestion, the equation can be
simplified to a two-compartment model consisting of fat
and fat-free body mass, the latter including nontriglyceride or essential lipids. According to this model, fatfree body mass and triglyceride-free mass are for
practical purposes the same component. Total body
water and potassium methods, as defined by this
model, would provide fat and fat-free body mass estimates, whereas the Cohn neutron activation method
provides an estimate of lipid (i.e., sum of triglyceride
and nontriglyceride lipids) and lipid-free mass (g). The
two-compartment underwater weighing method would
require additional consideration of the nontriglyceride
Fig. 3. Current molecular-level body composition model (left) with
suggested alternative lipid components (right). Triglycerides (TG)
are assumed in model to represent the ether extract of tissue and to
be synonymous with the term ‘‘fat’’.
lipids that are not now included in the model. Imaging
methods, such as MRI, would provide estimates of
adipose tissue and adipose tissue-free mass.
Nonlipid Energy-Restriction Effects
The present study results suggest other relative
component changes with energy restriction, and exercise. At 2 wk of energy restriction, the weight of the
animals was reduced by 5.5%, which reflected mainly
lipid loss. Lipid-free mass, total body water, and mineral ash either remained unchanged or increased compared with levels observed in the control animals.
With prolonged energy restriction and exercise, there
were absolute reductions in most lean components with
the exception of mineral ash, which continued to increase throughout the 9-wk protocol. Moreover, there
was a relative increase in hydration (e.g., TBW/lipidfree mass 5 0.727 in control vs. 0.754 in 9-wk exercise
group, P 5 0.3) and lowering of protein (e.g., protein/
TGFM 5 0.220 in control vs. 0.208 in 9-wk exercise
group, P 5 0.16). Although the mechanisms leading to
these extreme relative changes in protein energymalnourished animals are undoubtedly complex, they
do suggest a lack of ‘‘constancy’’ in assumed component
ratios such as TBW/lipid-free mass. These findings
indicate that animal models may be useful as a means
of investigating body composition relationships, because the required studies are extremely complex and
difficult to interpret in humans, who often lose weight
involuntarily with diseases that may alter water balance.
Future Directions
This study was based on a relatively small sample of
27 animals. The exploratory nature of the present
study dictated the selected sample size, and future
large-scale studies are needed to test the statistical
significance of some observed body composition trends.
Our results strongly support the need for improved
molecular-level body composition models. At some future time, investigators should develop a consensus on
terminology of the various lipid components. As a
minimum, research reports should carefully document
the nature of evaluated lipids.
An important question arising from our results is
how the essential lipid component might be estimated
in vivo. Our findings indicate that essential lipids are
not a constant fraction of total lipids and are thus not
easily estimated from total lipid mass. Moreover, we
explored in our initial analyses (not presented in this
report) other potential relations, such as the essential
lipid-to-protein, -ash, -water, and -body weight ratios,
and these also appeared unstable with energy depletion. These findings suggest that it would be difficult to
infer essential lipid mass from these other components
that can be measured in vivo. Hence, estimating lipid
fractions in vivo poses an important challenge for
future investigators.
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
lean body mass is its application, noted earlier, as a
measure of both lipid-free and fat-free body mass.
Terminology used in the body composition field is
thus both inconsistent and in some cases inaccurate
with respect to lipid and its subcomponents. The initial
results reported in this study strongly support inclusion of a nontriglyceride lipid component in applied
models. This component was not inconsequential (2.8%
of control body weight; about one-third of total lipids
after 9 wk of food restriction and exercise) and varied
independently from other components with negative
energy balance.
The present study results, combined with the confusing terminology applied to lipids found in previous
reports, prompted us to explore the definition of fat.
The large majority of our sources defined fat as neutral
lipids or triglycerides. Although there is no consensus
on this point, applying the term fat or nonessential
lipids to the ether extract of biological tissues has the
advantage of making most of the earlier studies and
models consistent. Accordingly, we suggest the following molecular-level model (Fig. 3)
E865
E866
BODY COMPOSITION LIPIDS
Conclusion
This study firmly established the existence of at least
two separate lipid body composition components that
respond differently with energy restriction and exercise. As a minimum, body composition models should be
cautiously labeled for specific lipid fractions, thereby
allowing investigators to fully understand the nature of
measured components. A need exists in future studies
to develop estimation methods for specific lipid body
composition components.
Received 3 September 1997; accepted in final form 4 February 1998.
REFERENCES
1. Behnke, A. R. Anthropometric evaluation of body composition
throughout life. Ann. NY Acad. Sci. 110: 450–464, 1963.
2. Behnke, A. R., B. G. Feen, and W. C. Welham. The specific
gravity of healthy men. JAMA 118: 495–498, 1942.
3. Bligh, E. G., and W. J. Dyer. A rapid method of total lipid
extraction and purification. Can. J. Biochem. Physiol. 37: 911–
917, 1959.
4. Brozek, J., F. Grande, J. T. Anderson, and A. Keys. Densitometric analysis of body composition: revision of some quantitative assumptions. Ann. NY Acad. Sci. 110: 113–140, 1963.
5. Bull, H. Norsk Finkeritidende 15: 549–557, 1896.
6. Bull, H. Norsk Finkeritidende 16: 294–300, 1897.
7. Cabrini, L., L. Landi, C. Stefanelli, V. Barzanti, and A. M.
Sechi. Extraction of lipids and lipophilic antioxidants from fish
tissues: a comparison among different methods. Comp. Biochem.
Physiol. B Biochem. Mol. Biol. 101: 383–386, 1992.
8. Christie, W. W. Lipid Analysis (2nd ed.). Oxford: Pergamon,
1982.
9. Cohn, S. H., A. N. Vaswani, S. Yasumura, K. Yuen, and K. J.
Ellis. Improved models for determination of body fat by in vivo
neutron activation. Am. J. Clin. Nutr. 40: 255–259, 1984.
10. Fidanza, F., A. Keys, and J. T. Anderson. Density of body fat
in man and other mammals. J. Appl. Physiol. 6: 252–256, 1953.
11. Forbes, G. B., J. Gallup, and J. B. Hursh. Estimation of total
body fat from potassium-40 content. Science 133: 101–102, 1961.
12. Forbes, G. B., and A. Lewis. Total sodium, potassium, and
chloride in adult man. J. Clin. Invest. 35: 596–600, 1956.
13. Foster, M. A., J. M. S. Hutchinson, J. R. Mallard, and M.
Fuller. Nuclear magnetic resonance pulse sequence and discrimination of high and low tissues. Mag. Res. Imaging 2: 187–192,
1984.
14. Gurr, M. I., and J. L. Harwood. Lipid Biochemistry (4th ed.).
London: Chapman and Hall, 1991.
15. Heymsfield, S. B., R. P. Olafson, M. H. Kutner, and D. W.
Nixon. A radiographic method of quantifying protein-calorie
undernutrition. Am. J. Clin. Nutr. 32: 693–702, 1979.
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 15, 2017
We express our appreciation to Drs. Myles S. Faith, Moonseong
Heo, and John Thornton for their helpful comments during the
preparation of this manuscript.
This study was supported by National Institute of Diabetes and
Digestive and Kidney Diseases Grant PO1-NIDDK-42618.
Address for reprint requests: A. Pietrobelli, Obesity Research
Center, 1090 Amsterdam Ave., New York, NY 10025.
16. Heymsfield, S. B., and M. Waki. Body composition in humans:
advances in the development of multicompartment models. Nutr.
Rev. 49: 97–108, 1991.
17. Judd, C. M., and M. McClelland. Data Analysis. A ModelComparison Approach. Orlando, FL: Harcourt Brace Jovanovich,
1989.
18. Kehayias, J. J., S. B. Heymsfield, A. F. LoMonte, J. Wang,
and R. N. Pierson, Jr. In vivo determination of body fat by
measuring total body carbon. Am. J. Clin. Nutr. 53: 1339–1344,
1991.
19. Knight, G. S., A. H. Beddoe, S. J. Streat, and G. L. Hill. Body
composition of two human cadavers by neutron activation and
chemical analysis. Am. J. Physiol. 250 (Endocrinol. Metab. 13):
E179–E185, 1986.
20. Kyere, K., B. Oldroyd, C. B. Oxby, L. Burkinshaw, R. E.
Ellis, and G. L. Hill. The feasibility of measuring total body
carbon by counting neutron inelastic scatter gamma rays. Phys.
Med. Biol. 27: 805–817, 1982.
21. Martin, A. D., M. Z. Daniel, D. T. Drinkwater, and J. P.
Clarys. Adipose tissue density, estimated adipose lipid fraction
and whole body adiposity in male cadavers. Int. J. Obes. 18:
79–83, 1994.
22. Pace, N., and E. N. Rathbun. Studies on body composition. III.
The body water and chemically combined nitrogen content in
relation to fat content. J. Biol. Chem. 158: 685–691, 1945.
23. Peters, J. P., and D. D. Van Slyke. Quantitative Clinical
Chemistry. Methods. Baltimore, MD: Williams & Wilkins, vol. 2,
1942.
24. Radin, N. S. Methods. In: Enzymology, edited by J. M. Lowenstein. New York: Academic, 1969, p. 245–254.
25. Rathbun, E. N., and N. Pace. Studies on body composition. I.
The determination of total body fat by means of the body specific
gravity. J. Biol. Chem. 158: 667–676, 1945.
26. Roche, A. F., S. B. Heymsfield, and T. G. Lohman (Editors).
Human Body Composition. Champaign, IL: Human Kinetics,
1996.
27. Seidell, J. C., C. J. C. Bakker, and K. Van Der Kooy. Imaging
techniques for measuring adipose-tissue distribution—a comparison between computed tomography and 1.5-T magnetic resonance. Am. J. Clin. Nutr. 51: 953–957, 1990.
28. Sheng, H. P., and R. A. Huggins. A review of body composition
studies with emphasis on total body water and fat. Am. J. Clin.
Nutr. 32: 630–647, 1979.
28a.Siri, W. E. Body composition from fluid spaces and density:
analysis of methods. In: Techniques for Measuring of Body
Composition, edited by J. Brozek and A. Henschel. Washington,
DC: Natl. Acad. Sci., 1961, p. 223–244.
29. Sjostrom, L. A computer-tomography based multi-compartment
body composition technique and anthropometric predictions of
lean body mass, total and subcutaneous adipose tissue. Int. J.
Obes. 15: 19–30, 1991.
30. St. John, L. C., and F. P. Bell. Extraction and fractionation of
lipids from biological tissues, cells, organelles, and fluids. Biotechniques 7: 476–481, 1989.
31. Story, J. A. Lipid Research Methodology. New York: Liss, 1984.
32. Tester, A. L. A specific gravity method for determining fatness
(condition) in herring (Clupea pallosii). J. Fish. Res. Board Can.
4: 461–471, 1940.
33. Wang, Z. M., R. N. Pierson, Jr., and S. B. Heymsfield. The
five-level model: a new approach to organizing body composition
research. Am. J. Clin. Nutr. 56: 19–28, 1992.