Am. J. Physiol. Endocrinol. Metab.
278: E308–E315, 2000.
Metabolically active components of fat free mass
and resting energy expenditure in nonobese adults
KIRSTEN ILLNER,1,2 GISBERT BRINKMANN,2 MARTIN HELLER,2
ANJA BOSY-WESTPHAL,1 AND MANFRED J. MÜLLER1
1Institut für Humanernährung und Lebensmittelkunde und 2Klinik für Radiologische Diagnostik,
Christian-Albrechts-Universität zu Kiel, D-24105 Kiel, Germany
Illner, Kirsten, Gisbert Brinkmann, Martin Heller,
Anja Bosy-Westphal, and Manfred J. Müller. Metaboli-
cally active components of fat free mass and resting energy
expenditure in nonobese adults. Am. J. Physiol. Endocrinol.
Metab. 278: E308–E315, 2000.—Resting energy expenditure
(REE) and components of fat-free mass (FFM) were assessed
in 26 healthy nonobese adults (13 males, 13 females). De-
tailed body composition analyses were performed by the
combined use of dual-energy X-ray absorptiometry (DEXA),
magnetic resonance imaging (MRI), bioelectrical impedance
analysis (BIA), and anthropometrics. We found close correla-
tions between REE and FFMBIA (r 5 0.92), muscle massDEXA
(r 5 0.89), and sum of internal organsMRI (r 5 0.90). In a
multiple stepwise regression analysis, FFMBIA alone ex-
plained 85% of the variance in REE (standard error of the
estimate 423 kJ/day). Including the sum of internal organsMRI
into the model increased the r2 to 0.89 with a standard error
of 381 kJ/day. With respect to individual organs, only skeletal
muscleDEXA and liver massMRI significantly contributed to
REE. Prediction of REE based on 1) individual organ masses
and 2) a constant metabolic rate per kilogram organ mass was
very close to the measured REE, with a mean prediction error
of 96 kJ/day. The very close agreement between measured
and predicted REE argues against significant variations in
specific REEs of individual organs. In conclusion, the mass of
internal organs contributes significantly to the variance in
REE.
body composition; muscle mass; organ mass; dual-energy
X-ray absorptiometry; magnetic resonance imaging; bioelec-
trical impedance analysis; anthropometrics
VARIATION IN THE SIZE of fat-free mass (FFM) has been
shown to explain 65–90% of the between-subject varia-
tion in resting energy expenditure (REE) (1, 3, 9, 13, 25,
28). REE per unit FFM is not constant, and the ratio of
REE to FFM varies with body weight. REE per unit
FFM decreases with increasing body mass, which sug-
gests that subjects with higher FFM have lower REEs
per kilogram FFM, whereas the opposite is true for
individuals with lower FFM (25). This finding consigns
the utility of indexing REE to body weight or FFM in
subjects with different body masses. A potential expla-
nation for this problem is the composition of the
metabolically active FFM. Skeletal muscle and internal
The costs of publication of this article were defrayed in part by the
payment of page charges. The article must therefore be hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
E308 0193-1849/00 $5.00 Copyright r 2000 the American Physiological Society
Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.
Copyright © 2000 American Physiological Society. All rights reserved.
organ mass substantially differ with respect to their
individual rates of energy expenditure. Estimates of
REE per kilogram organ mass based mainly on in vitro
measurements vary between 837 and 1,841 kJ for
internal organs and 54 and 63 kJ for skeletal muscle
(10, 24). Thus, from a metabolic point of view, different
ratios of high vs. low energy-requiring organs may explain
an unknown part of the variation in REE (25, 29).
Techniques of computed tomography (CT), magnetic
resonance imaging (MRI), and dual-energy X-ray ab-
sorptiometry (DEXA) can be used for the in vivo
measurement of metabolically active components of
FFM. Up to now, only four studies have combined these
measurements with measurements of REE (6, 12, 21,
27). Using CT in combination with densitometry, De-
riaz et al. (6) measured REE and body composition (i.e.,
skeletal muscle mass and the sum of tissue masses) in
22 men before and after a 100-day overfeeding period.
They found significant correlations between REE and
muscle, as well as nonmuscle, compartments of FFM.
Using these two variables of body composition did not
improve the prediction of REE over that provided by
muscle mass alone. After overfeeding, body mass, lean
body mass, and skeletal muscle mass were the best
correlates of REE. While the present study was under
investigation, Sparti et al. (27) simultaneously used
CT, DEXA, and echocardiography for the assessment of
the composition of FFM in 20 females and 20 males,
respectively. These authors also concluded that the
composition of FFM could not improve the prediction of
REE compared with FFM alone. In a further prelimi-
nary report, McNeill et al. (21) also provided no evi-
dence that liver, kidney, and spleen masses (as deter-
mined by MRI in 30 women) explain any of the variance
in REE between individuals. Taken together, the re-
sults of three studies (6, 21, 27) suggest that the
variance in REE is more dominated by the energy
expenditure of the individual organs than by the vari-
ance in the internal organ size. This idea is contrary to
a very recent paper by Gallagher et al. (12). These
authors measured body cell mass (as measured by total
body potassium) and organ-tissue volumes by MRI and
echocardiography. They found strong associations be-
tween REE and individual or combined organ weights.
Moreover, calculating REE from individual organ
masses and previously reported organ metabolic rates
closely predicted measured REE [i.e., the difference
between measured and calculated REE was less than
http://www.ajpendo.org
 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE
84 kJ/day (12)]. These data suggest a constant organ-
tissue respiration rate.
After reviewing the available literature, we reached
two conclusions. First, the data base on concomitant in
vivo measurements of organ size and REE is very
small. Second, the available data on the association
between body composition and REE are in part contra-
dictory. Faced with these discrepancies and the limited
information from concomitant in vivo measurements of
organ size and REE, we reassessed the relationship of
metabolically active components of FFM to REE in a
homogeneous group of healthy adults by use of DEXA,
MRI, bioelectrical impedance analysis (BIA), and an-
thropometrics for detailed body composition analysis.
Our data support the idea that organ sizes are impor-
tant determinants of REE. The use of metabolically
active components of FFM instead of total FFM re-
duced the variance in REE prediction.
METHODS
Study population. Twenty-six healthy subjects (13 females,
13 males) participated in the study (see Table 1). The study
protocol was approved by the ethical committee of the Chris-
tian-Albrechts-Universität zu Kiel. Before participation, each
subject provided written consent. All subjects were healthy
and weight stable. None had a history of recent illness,
obesity, diabetes, hyperlipidemia, hypertension, or endocri-
nopathy. Each subject had a normal physical examination.
Dieting or physical exhaustion were avoided during the 7
days before examination.
Measurements and estimations of REE. In females, mea-
surements of REE and body composition were performed in
the follicular phase of the menstrual cycle (i.e., ,10 days
since last menstruation). REE was measured as described
elsewhere (23). Briefly, REE was measured by an open-circuit
indirect calorimeter (Deltatrac Metabolic Monitor, Datex
Instruments, Helsinki, Finland). Measurements were per-
formed between 7:00 and 8:00 AM in a metabolic ward at
constant humidity (55%) and room temperature (22°C). The
day before testing, the subjects had eaten their last evening
meal between 6:00 and 7:00 PM. For $1 h, gas exchange
measurements were done continuously. The first 20 min of
data were omitted. For the residual time of investigation (i.e.,
for a period of $40 min), data were integrated for 5-min
intervals. The means of $40 measurements were presented
as individual values. Calibrations of the gas analyzers were
performed immediately before and after the measurements.
Variation caused by the technique was calculated on the basis
of five repeated measurements of propane combustion and
was found to be ,4%. The within-day coefficent of variation of
the 5-min oxygen consumption (V̇O2) values was ,7.5%.
Intraindividual variances in REE were assessed in a sub-
group of 10 weight-stable men, who performed test-retest
measurements on three different days within a 14-day period.
The intraindividual variances were ,6%. REE was calcu-
lated as described by Weir (30): kcal/min 5 V̇O2 (l/min) 3
[3.9 1 (1.1 3 RQ)], where RQ is respiratory quotient. REE
(kcal/day) was also calculated using various prediction formu-
las as proposed by the following investigators
Cunningham (Ref. 3)
REE 5 21.6 3 FFMCalc 1 501.6
Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.
Copyright © 2000 American Physiological Society. All rights reserved.
E309
Garby et al. (Ref. 13)
REE 5 27.88 3 FFMDens 1 6.4 3 FM (2)
Elia (Ref. 10) REE 5 21.11 3 FFMANTHRO 1 405 (3)
Ravussin and Bogardus (Ref. 25)
REE 5 20.8 3 FFMDens 1 471 (4)
The indexes clarify which method was used by the investi-
gators to measure or calculate FFM (Calc, calulated from
weight and age; Dens, densitometry; ANTHRO, anthropom-
etry, from skinfold measurements). Formulas 1–4 were used
because they assume body composition (i.e., FFM 6 fat mass)
as a predictor of REE. The data were therefore compared with
our approach (i.e., calculating REE from FFM and/or indi-
vidual organ masses) to assess the differences between esti-
mated and measured values.
Body composition analysis. By use of an electronic scale,
weight was measured without shoes, with light clothing, and
after voiding, at an accuracy of 0.1 kg (SECA, Modell 709,
Vogel & Halke, Hamburg, Germany). Height was assessed to
the nearest 0.5 cm with a stadiometer. Body composition was
assessed by the use of anthropometrics (ANTHRO), BIA,
DEXA, and MRI. ANTHRO and BIA were performed immedi-
ately after gas exchange measurements. DEXA and MRI were
both performed on the following day. ANTHRO was used to
assess body fat (measurements of 4 skinfolds) and arm
muscle area (arm circumference). Skinfold thickness was
measured by the use of caliper (Lafayette Instruments,
Lafayette, IN). Fat mass was calculated according to Durnin
and Wormersley (8), and arm muscle area was calculated
according to Heymsfield et al. (17). BIA was performed as
described previously (22) as a measure of total body water
(TBW). We used a body impedance analyzer at a frequency of
50 kHz and the manufacturer’s software for data analyses
(BIA 2000-S, Data Input, Frankfurt, Germany). FFMBIA was
calculated assuming a water content of FFM of 73.2% (FFM 5
TBW/0.732). Bone mineral content and whole body and
regional lean body mass (LBM) were measured by DEXA
(Hologic QDR 4500A, Hologic, Waltham, MA). Limb lean
mass was used as a measure of muscle mass, as suggested by
Heymsfield et al. (18). DEXA scans were analyzed with the
manufacturer’s whole body Version 5.54 (Hologic).
The volume of internal organs (brain, heart, liver, kidneys,
spleen) was measured by transversal MRI images. The scans
were made by use of a 1.5-Tesla Magnetom Vision scanner
(Siemens, Erlangen, Germany). All organs were examined
native and without distance factors. The slice thickness for
the brain was 6 mm, for the heart, 7 mm, and for abdominal
organs, 8 mm. The brain and the abdominal organs of the
participants were examined by T1-weighted breathhold
FLASH sequences (repetition time, TR: 174.9 ms; echo time,
TE: 4.1 ms/echo). Because of the movement of the heart and to
prevent artifacts, ultra-short scans were made by electrocar-
diogram (ECG)-triggered, T2-weighted HASTE sequences
that were taken in breathhold technique (acquisition time: 20
ms). The volume of the organs was calculated from the sum of
cross-sectional areas as determined ‘‘by hand’’ (i.e., by exactly
drawing a line at the external limits of the organs) and
multiplied by the scan’s slice thickness. All scans were read
by the same trained observer (K.I.); the coefficient of varia-
tion, based on three test-retest measurements, was ,1%.
Volume data were transformed into organ weights by use of
the following densities: 1.036 g/cm3 for brain, 1.06 g/cm3 for
heart and liver, 1.05 g/cm3 for kidneys, and 1.054 g/cm3 for
(1) spleen (7). Total body mass was considered as the sum of
E310 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE

organ masses. Residual mass was calculated as body mass c 1 (k2 3 kg FFMBIA 1 k3 3 kg FMDEXA ) (7)
minus the sum of skeletal muscleDEXA, brainMRI, heartMRI,
liverMRI, kidneysMRI, spleenMRI, bone mineralDEXA, and fat d 1 (k4 3 kg muscle massDEXA
massDEXA (5 residual 1 in Tables 1–3). Because skeletal 1 k5 3 kg liver weightMRI ) (8)
muscleDEXA assesses appendicular, and not whole body skel-
etal muscle mass, residual mass was reanalyzed using skel- In addition, whole body REE (REEc) was calculated as the
etal muscleANTHRO (5 residual 2 in Tables 1–3). Residual sum of seven individual organs (brainMRI, liverMRI, skeletal
mass was also calculated assuming constant contributions of muscleDEXA, fat massDEXA, plus residual organs) times organ-
blood volume (i.e., 7.9% body wt), skin (3.7% body wt), tissue metabolic rates, on the basis of organ-tissue metabolic
connective tissue (2.3% body wt), lung tissue (1.4% body wt), rates reported by Elia et al. (10)

REEc 5 (1.008 3 brain massMRI) 1 (840 3 liver massMRI ) 1 (1.848 3 heart massMRI ) 1 (1.848 3 kidney massMRI )
(9)
1 (55 3 skeletal muscle massDEXA ) 1 (19 3 fat massDEXA ) 1 (50 3 residual mass 1)

intestine (1.7% body wt) to body weight, as given in Ref. 9 (5 The agreement between calculated and measured REE
residual 3 in Tables 1–3). was analyzed by plotting the difference between mea-
Methodological limitations. Calculation of organ masses sured and calculated REE vs. the average of REE mea-
from data obtained by imaging techniques is based on 1) the sured and REE calculated, as described by Bland and Alt-
sum of cross-sectional areas multiplied by the distance be- man (2).
tween scans [coefficient of variation (CV) ,1%] and 2) approxi-
mate organ densities taken from the literature (7). In vivo
organ weight was measured to the nearest 0.1 kg. This is RESULTS
within the order of the accuracy reached for radiographic
volume and mass determination by use of excised human
Body composition data. Body masses, as well as the
cadaver organs (i.e., 63–5%; Refs. 15, 16). If a standard organ sizes of the subjects, are shown in Table 1.
deviation for organ weights of #0.4 kg and a mean specific Significant sex-dictated differences were seen for weight,
energy expenditure of ,1,255 kJ/kg organ weight are as- height, total body waterBIA, FFMBIA, fat massDEXA, fat
sumed (15), the precision of the imaging techniques may
introduce a considerable error (i.e., 7–9% of REE). This may
contribute to the residual variance in REE and also limit the Table 1. Characterization of study population
value of modeling REE from metabolically active tissue mass.
Females Males
There is also evidence that our approach to assessing metaboli- n 5 13 n 5 13
cally active components of FFM left a considerable amount of
so-called ‘‘residual masses’’ unexplained. Residual mass is a Age, yr 24.8 6 2.4 26.2 6 2.1
heterogeneous component that includes intestine, pancreas, Body weight, kg 62.8 6 9.5† 72.3 6 10.2
Body height, cm 169.6 6 6.2† 185.2 6 7.7
lung, skin, blood volume, endocrine glands, and connective BMI, kg/m2 22.14 6 2.47 22.53 6 1.82
tissue. It is evident that different calculation procedures Total body waterBIA , liters
result in different amounts of residual mass (see residuals (% body wt) 32.8 6 3.1 (52.2)† 42.2 6 5.0 (58.4)
1–3, Tables 1–3). These differences are in part explained by Fat free massBIA , kg
the methods used to assess skeletal muscle mass (e.g., (% body wt) 45.9 6 4.7 (73.1)† 63.1 6 6.9 (87.2)
skeletal muscle massDEXA measures appendicular instead of Organ masses, kg
whole body muscle mass). The close association between the (% body wt)
MuscleDEXA 17.9 6 2.5 (28.5)† 28.8 6 3.0 (39.8)
different measures of residual mass and 1) FFMBIA and thus MuscleANTHRO 21.5 6 4.1 (34.2)† 31.9 6 5.5 (44.2)
2) REE suggests that residual mass indirectly reflects meta- BrainMRI 1.5 6 0.1 (2.4) 1.6 6 0.2 (2.2)
bolic active tissues (Table 2–3). HeartMRI 0.3 6 0.0 (0.5)† 0.4 6 0.1 (0.6)
Data analyses. All data were recorded in a database system LiverMRI 1.5 6 0.2 (2.4) 1.7 6 0.3 (2.4)
with a personal computer; statistical analyses were per- SpleenMRI 0.2 6 0.0 (0.3)* 0.3 6 0.1 (0.4)
formed by SPSS for Windows 5.0.2. Data are presented as KidneysMRI 0.2 6 0.0 (0.3)† 0.2 6 0.1 (0.3)
means 6 SD. The Mann-Whitney U-test or Fisher’s Exact Bone mineralDEXA 2.4 6 0.3 (3.8)† 3.2 6 0.6 (4.4)
FatDEXA 19.2 6 6.0 (30.6)† 12.9 6 4.5 (17.8)
Test was used for comparisons between groups. Pearson’s Residual 1 (muscle mass-
correlation coefficients were calculated to test for relation- DEXA ), kg (% body wt) 19.7 6 2.1 (31.7)† 28.3 6 4.2 (36.6)
ships among different parameters. In addition, a multivariate Residual 2 (muscle
stepwise regression analysis was performed using REE as massANTHRO ), kg (%
dependent variable. A probability value of 0.05 was accepted body wt) 16.2 6 3.8 (26.2)† 25.1 6 4.0 (32.8)
as the limit of significance. The specific contributions of body Residual 3 (theoretical),
weight, FFM or muscle mass, and organ masses (i.e., slopes or kg (% body wt) 10.7 6 1.6 (17.0)† 13.1 6 1.7 (17.0)
REE, MJ/day 5.74 6 0.68† 7.28 6 0.85
k values) were calculated from the individual regression
RQ 0.84 6 0.05 0.85 6 0.05
lines. The model assumes that whole body REE is the sum of
respiration of the individual tissues. Then REE was predicted Values are means 6 SD. BMI, body mass index; BIA, bioelectrical
by different formulas impedance analysis; DEXA, dual-energy X-ray absorptiometry; MRI,
magnetic resonance imaging; ANTHRO, anthropometrics; REE, rest-
a 1 (k1 3 body wt.) (5) ing energy expenditure; RQ, respiratory quotient. Residual 3 (theo-
retical) is calculated from fixed organ weights as assumed by Elia (9).
b 1 (k2 3 kg FFMBIA ) (6) Significant differences between genders: * P , 0.05; † P , 0.01.

Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.

Copyright © 2000 American Physiological Society. All rights reserved.
 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE E311

massANTHRO, and the different organs measured except

Residual 2
for brainMRI and liverMRI. The water content of LBMDEXA

0.57†
was calculated from TBWBIA and LBMDEXA to be 0.73 6
0.03 in females and 0.73 6 0.00 in males, respectively.
Residual mass accounted for 32 and 37% of body weight

Residual 1
in males and females, respectively (Table 1). Calculat-

0.87†
0.83†
ing residual mass based on the assumption of constant
contributions of blood volume, skin, intestine, connec-
tive tissue, and lungs gave residual masses of 13.1 6

Table 2. Pearson correlation coefficients among body compartment sizes in 26 healthy subjects as assessed by different methods
TBWBIA
1.7 kg (male) and 10.7 6 1.6 kg (female), respectively (5

0.95†

0.85†
0.78†
residual 3). There were significant and sex-independent
differences between residual mass 1 and 2 or 3, respec-
tively (P , 0.01 or , 0.05). We also found significant

20.28NS
20.26NS

0.22NS
FatDEXA

20.55†
differences between residual masses 2 and 3 (P , 0.01).
DEXA was used to assess the composition of different
regions of the body (i.e., trunk 5 t, legs 5 l, arms 5 a).

BoneDEXA

0.04NS
The different segments contributed to body masses by

0.84†
0.83†
0.57†
0.89†
44 and 45% (t), 39 and 36% (l), and 11 and 13% (a) in
females and males, respectively. Proportions of fat for
each region were greater in females than in males (t:

SpleenMRI

20.08NS
0.65†

0.70†
0.76†
0.47*
0.70†
23.7 6 6.7 and 15.0 6 5.2%, l: 38.1 6 4.5 and 17.6 6
4.7%, a: 36.1 6 7.3 and 15.7 6 4.0% segmental weight
for females and males, respectively; P , 0.01 for sex
differences). By contrast, the proportions of weight of

KidneysMRI

20.08NS
0.71†
0.61†

0.61†
0.69†
0.54†
0.63†
trunk, arms, and legs consisting of LBMDEXA were
increased in males (t: 74.0 6 6.6 and 82.8 6 5.0%;
l: 58.0 6 4.4 and 77.6 6 4.5; a: 59.1 6 7.0 and 79.3 6
3.9% segmental weight, in females and males, respectively, LiverMRI

0.21NS
0.44*
0.47*
0.67†

0.63†
0.68†
0.43*
0.77†
P , 0.01 for sex differences). Regional proportions of bones
were similar between sexes (data not shown).
The relationships between different body and tissue

20.34NS
HeartMRI

0.45*
0.69†
0.59†
0.71†

0.78†
0.82†
0.77†
0.68†
masses are given in Table 2. There were significant

FFM, fat-free mass; TBW, total body water. * P , 0.05; † P , 0.01; NS, not significant.
correlations between FFMBIA and all other organs
except the brain. Fat massDEXA did not reach significant
associations with body weight in normal-weight sub-
0.35NS
0.36NS
20.17NS
BrainMRI

0.40*
0.42*
0.41*

0.41*
0.54*
0.46*
0.42*
jects. Plotting the ratio of skeletal muscle massDEXA and
the sum of organ massesMRI against FFMBIA showed
significant and positive association (Fig. 1).
REE and body composition. Measured REE varied
MuscleDEXA

20.34NS
0.43*
0.81†
0.57†
0.59†
0.68†
0.83†

0.97†
0.94†
0.83†
0.83†
from 4.77 to 8.62 MJ/day. REE values were higher in
males than in females (127%; P , 0.001; Table 1).
Adjusting REE on the basis of group mean REE plus
the measured REE minus predicted REE (as predicted
20.21NS
FFMBIA

0.97†
0.40*
0.77†
0.66†
0.61†
0.72†
0.86†

0.99†
0.95†
0.76†
0.89†

from the individual FFMBIA in the linear regression

equation generated in our population) (see Fig. 2 and
Ref. 25 for details of the underlying assumptions) gave
similar values for both sexes [m, 6.45 6 0.51 vs. f,
0.21NS
0.17NS

0.22NS
0.37NS

0.10NS
0.47*
0.36†

0.63†

0.59†
0.63†
0.40*
0.40*

0.73†
BMI

6.58 6 0.30 MJ/day; not significant (NS)].

REE was significantly associated with FFMBIA (r 5
0.92), muscle massDEXA (r 5 0.89), and the sum of
0.28NS

20.16NS

0.71†
Height

0.89†
0.90†
0.44*
0.81†
0.58†
0.69†
0.71†
0.80†

0.88†
0.90†

0.86†

organsMRI (r 5 0.90) (Fig. 2, Table 3). There were no

sex-dictated differences between the slopes of the REE
regression lines when FFMBIA, muscle massDEXA, or
Body Wt

organ massMRI was used as the variable. REE

0.22NS

0.57†
0.86†
0.73†
0.89†
0.83†
0.42*
0.68†
0.77†
0.63†
0.70†
0.89†

0.85†
0.87†

1.00†

(kJ · day21 · kg FFM21) was 126, 121, and 114 for sub-
jects with different FFM values, i.e., FFM ,50 kg (n 5
10), FFM 50–60 kg (n 5 9), and FFM .60 kg (n 5 7),
MuscleDEXA

KidneysMRI

Residual 2
Residual 1

Residual 3

respectively (P , 0.01 for the difference in REE/FFM

SpleenMRI
BoneDEXA
HeartMRI
BrainMRI
Body Wt

LiverMRI

TBWBIA
FatDEXA
FFMBIA

between subjects ,50 kg FFM vs. .60 kg FFM).

Height
BMI

Plotting REE on the ratio of skeletal muscle massDEXA

to the sum of organ massesNMR resulted in a positive

Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.

Copyright © 2000 American Physiological Society. All rights reserved.
E312 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE

Table 3. Pearson correlation coefficients

between REE and age, body, or organ sizes
Resting Energy Expenditure

Females Males All

Age 20.62* 0.14NS 0.07NS

Body wt 0.88† 0.75† 0.88†
Height 0.75† 0.49*NS 0.82†
BMI 0.75† 0.63* 0.55†
FFMBIA 0.95† 0.80† 0.92†
MuscleDEXA 0.94† 0.68† 0.89†
BrainMRI 0.27NS 0.49* 0.51†
HeartMRI 0.13NS 0.46NS 0.72†
LiverMRI 0.77† 0.82† 0.74†
KidneyMRI 0.58* 0.12NS 0.52†
SpleenMRI 0.53* 0.43NS 0.67†
BoneDEXA 0.86† 0.47NS 0.79†
FatDEXA 0.66† 0.35NS 20.09NS
TBWBIA 0.81† 0.80† 0.90†
Residual 1 0.76† 0.79† 0.89†
Residual 2 0.22NS 0.38NS 0.69†
Residual 3 0.88† 0.75† 0.88†
Significance between REE and individual variables: * P , 0.05;
† P , 0.01.

Prediction of REE. Predicting whole body REE from

Fig. 1. Ratios of high energy (sum of organs 5 organ massMRI)- vs.
low energy-requiring organs (5 skeletal muscleDEXA) plotted against
FFMBIA. BIA, bioelectrical impedance analysis; DEXA, dual-energy
X-ray absorptiometry; MRI, magnetic resonance imaging; FFM,
fat-free mass.
and significant association (data not shown). Plotting
REE per kilogram FFMBIA on the ratio of skeletal
muscle massDEXA per sum of organsMRI gave a negative
and significant correlation (Fig. 3).
In a multiple stepwise regression analysis, FFMBIA
alone explained 85% of the variance in REE (SE of the
estimate 423 kJ/day). Additionally, including the sum
of internal organsMRI into the model increased the r2 to
0.89 with a SE of 381 kJ/day. In a stepwise multiple
regression analysis, only skeletal muscleDEXA and liver
massMRI significantly contributed to REE, as indicated
in the following regression equation
REE 5 359 1 29 3 skeletal muscle massDEXA
(8a)
1 319 3 liver massMRI
FFMBIA 6 FMDEXA by use of formulas 1–4 (METHODS)
resulted in a very close agreement between measured
and predicted values (mean deviations between 1155
and 2485 kJ/day). There were no systematic errors in
groups of subjects differing with respect to their body
mass index (BMI, ,21 vs. 21–23 vs. .23 kg/m2) or their
FFMBIA (,50 vs. 50–60 vs. .60 kg) (data not shown).
Calculating REE (REEc) on the basis of measured
organ masses times constant organ tissue respiration
rates, as reported in the literature (formula 9), a mean
prediction error of 96 kJ/day was observed. There was a
very close correlation between theoretically calculated
(according to formula 9) and measured REE (Fig. 4).
We found significant differences in REE and REEc
(according to formula 9: FFMBIA ,50 kg, n 5 10, vs.
FFMBIA 50–60 kg, n 5 9, vs. FFMBIA .60 kg, n 5 7;
REE, 5.4 6 0.4 vs. 6.7 6 0.5 vs. 7.8 6 0.7 MJ/day; REEc,
5.7 6 0.4 vs. 6.7 6 0.4 vs. 7.7 6 0.9 MJ/day; P , 0.01 vs.
FFMBIA 50–60 kg) between groups differing with re-
spect to FFMBIA. A Bland-Altman plot showed no signifi-
cant trend (r 5 0.19; P 5 0.357) between measured and
calculated REE difference (i.e., the difference between

Fig. 2. Metabolically active compo-

nents of FFM plotted against resting
energy expenditure (REE) by BIA,
DEXA, and MRI-DEXA.

Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.

Copyright © 2000 American Physiological Society. All rights reserved.
 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE E313

Fig. 3. REE per kg FFMBIA plotted against different ratios of high- Fig. 4. REE plotted against REE calculated according to formula 9
vs. low energy-requiring organs. (REEc). REEc was calculated from organ masses times constant
tissue respiration rates as reported from the literature (see METHODS
for details of underlying assumptions).
REE measured and REE calculated according to formula 9
vs. the average of REE measured and REE calculated;
Fig. 5), suggesting that there is no systematic error. REE. This is close to our data, as well as to the recent
results of others (12, 27), which were all based on direct
DISCUSSION and concomitant in vivo assessments of organ masses
and REE. By contrast, Deriaz et al. (6) and McNeill et
REE varies among individuals. Body size, body com-
position, age, sex, hormones, and genetic factors ex-
plain most of its variability (1, 24, 28). It has been
speculated that the relative proportion of high and low
metabolically active tissues independent of differences
in FFM significantly adds to the residual variance in
REE (24, 28, 29).
Our present knowledge regarding the contribution of
individual organs to REE in humans is mainly based on
1) in vitro measurements of tissue respiration (5, 9, 19)
and 2) postmortem analysis of body composition (10, 14,
15). These studies suggest that 1) V̇O2 per gram tissue
is relatively constant and 2) organ size is a major
determinant of REE. Tissue V̇O2 can be directly esti-
mated in vivo by measurements of the arteriovenous
(a-v) differences of O2 together with blood flow measure-
ments. Methodological problems may limit the in vivo
assessment as well as the interpretation of organ-
tissue metabolic rates (11). However, the in vivo data
suggest that the sum of regional V̇O2 exceeds whole
body REE (3, 5). The comparison of in vivo with in vitro
data is inconclusive (5). Thus our knowledge on energy
expenditure-body composition relationships in humans
is limited.
On the basis of data obtained from 1,598 autopsies
and with the assumption of a constant mass-specific
energy expenditure, Garby and co-workers (14, 15) Fig. 5. Bland-Altman Plot exploring the agreement between calcu-
calculated that the composition of FFM may explain 5% lated (according to formula 9) and measured REE. Dotted lines are
of the variation in the between-subjects variation in given for the first SE values.

Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.

Copyright © 2000 American Physiological Society. All rights reserved.
E314 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE
al. (21) provided no evidence that the composition of
FFM explains any of the variance in whole body REE.
Regarding the role of metabolically very active organs
(i.e., muscle, brain, liver) contributing to REE, the
different authors also came out with different results.
We found that skeletal muscle and liver are the major
determinants of REE in young, healthy, and nonobese
subjects (RESULTS). By contrast, Gallagher et al. (12)
found that brain and skeletal muscle were the major
determinants of REE. The discrepancy between the
results of these two studies may be explained by
differences in the database (e.g., the number and age of
the subjects differ between the two studies). In the two
other studies, only skeletal muscle mass (6) or muscle
plus fat plus heart mass (27) significantly contributed
to the prediction of REE. However, in their multiple
regression analysis, Deriaz et al. used only skeletal
muscle mass and nonmuscular LBM (which is the sum
of internal organs). Thus these authors could not
differentiate among individual organs. In addition, in
the study by Deriaz et al., only a limited number of CT
scans at nine selected sites were performed, and the
relationship between REE and FFM was poor (i.e., r 5
0.56 for REE vs. LBMCT, or r 5 0.49 for REE vs. FFM,
as determined by densiometry; Ref. 6). Compared with
Deriaz et al., Sparti et al. measured liver but not brain
by serial CT images (27). In addition, appendicular
skeletal muscle mass was measured by DEXA. With
use of simple correlation coefficients, muscle and liver
showed significant associations with REE (r 5 0.84 and
0.75, respectively), which is very close to our data (r 5
0.94 and 0.77, respectively; see Table 3). Because a
homogeneous and comparable group of subjects was
studied in both studies (27, this study) and similar
methods have been used (CT, echocardiography, DEXA
in Ref. 27; MRI, DEXA, BIA in this study), it is unclear
why muscle but not liver reached statistical signifi-
cance in Sparti’s regression analysis. It should be
mentioned that both studies (Ref. 27 and this study),
although very similar with respect to the physical
variables of the subjects, differ with respect to the
magnitude of some internal organs (i.e., kidney mass
was higher but left ventricular mass was lower in Ref.
27 compared with our data). Some of the differences in
organ masses given in the different studies (12, 27, this
study) are due to methodological problems. For ex-
ample, Sparti et al. (27) as well as Gallagher et al. (12)
assessed left ventricular mass by echocardiography,
whereas ECG-triggered MRI was used in our study. In
contrast with MRI, echocardiography measures only
left ventricular mass, which accounts for approxi-
mately two-thirds of heart weight in healthy adults.
Organ contribution to whole body energy expendi-
ture can also be assessed by direct measures of organ
energy metabolism. Regarding direct in vivo measure-
ments of muscle and liver V̇O2 obtained by use of a-v
difference techniques, both organs together contribute
,50% of REE (10, 22, 31). However, the a-v difference
technique cannot differentiate between nutritive and
nonnutritive blood flow and thus may overestimate
organ-tissue respiration (11, 22). At present, there are
Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.
Copyright © 2000 American Physiological Society. All rights reserved.
only limited in vivo data on organ-tissue V̇O2. Suitable
methods (e.g., 150 or positron emission tomography, Ref.
26) for the in vivo assessment of regional V̇O2 should be
applied in future studies. These techniques will contrib-
ute to the development of new energy expenditure-body
composition estimation models.
Body size-related variations in REE are explained by
1) the proportional contributions of different organs to
FFM, as well as 2) tissue O2 consumption. In a stepwise
multiple regression analysis, FFM alone explains 85%
of the variance in REE, leaving an SE of the estimate of
423 kJ/day (RESULTS). Calculating REE as the sum of
individual organ-tissue masses times a constant organ-
tissue respiration rate, on the basis of data published in
the literature (10), reduces the variance in REE and
results in small differences between measured and
calculated REE of 83 (12) or 96 kJ/day (this paper),
respectively. However, it should be mentioned that the
use of standard formulas for the prediction of REE also
reaches a very high precision in our homogeneous
group of young, healthy, and nonobese subjects. It is
tempting to speculate that the accuracy of prediction
may differ in a more heterogeneous sample of subjects
(e.g., in patients with chronic diseases or obese patients).
In conclusion, the proportions of metabolically active
components of REE contribute to the variance in REE
and also explain the relation between REE and FFM.
The essential findings of the present study are that 1)
the contribution of the mass of the metabolically more
active internal organs to the variance in REE is ,5%;
only skeletal muscleDEXA and liverMRI significantly con-
tribute to REE; 2) the decrease in REE per kilogram
FFM with increasing FFM is explained by the changing
proportions of metabolically active compounds of FFM;
and 3) predictions of REE on the basis of individual
organ masses were very close to measured REE. Our
data support the assumptions (3, 24, 28, 29) and the
data of some authors (12) but are contrary to the results
of others (6, 21, 27).
A preliminary report of this work was presented in abstract form
at the 16th International Congress of Nutrition, Montreal, 1997 (PW
14.3).
Address for reprint requests and other correspondence: M. J.
Müller, Institut für Humanernährung und Lebensmittelkunde, Chris-
tian-Albrechts-Universität zu Kiel, Düsternbroker Weg 17, D-24105
Kiel, Germany (E-mail: mmueller@nutrfoodsc.uni-kiel.de).
Received 4 January 1999; accepted in final form 20 September 1999.
REFERENCES
1. Astrup, A., B. Buemann, N. J. Christensen, J. Madsen, C.
Gluud, P. Bennett, and B. Svenstrup. The contribution of
body composition, substrates, and hormones to the variability in
energy expenditure and substrate utilization in premenopausal
women. J. Clin. Endocrin. Metab. 74: 279–286, 1992.
2. Bland, M. J., and D. G. Altman. Statistical methods for
assessing agreement between two methods of clinical measure-
ment. Lancet 1: 307–310, 1986.
3. Cunningham, J. C. Body composition as a determinant of
energy expenditure: a synthetic review and a proposed general
prediction equation. Am. J. Clin. Nutr. 54: 963–969, 1991.
4. Cunningham, J. J. A reanalysis of the factors influencing
metabolic rate in normal adults. Am. J. Clin. Nutr. 33: 2372–
2374, 1980.
5. Davies, M. On body size and tissue respiration. J. Cell. Comp.
Physiol. 57: 135–147, 1961.
 ORGAN-TISSUE MASS AND ENERGY EXPENDITURE
6. Deriaz, O., G. Fournier, A. Tremblay, J.-P. Despres, and C.
Bouchard. Lean-body mass composition and resting energy
expenditure before and after long-term overfeeding. Am. J. Clin.
Nutr. 56: 840–847, 1992.
7. Duck, F. A. Physical Properties of Tissue. New York: Academic,
1990.
8. Durnin, J. V. G. A., and J. Wormersley. Body fat assessed from
body density and 1st estimation from skinfold thickness: measure-
ment on 481 men and women from 16 to 72 years. Br. J. Nutr. 32:
77–97, 1974.
9. Elia, M. Energy expenditure in the whole body. In: Energy
Metabolism: Tissue Determinants and Cellular Corollaries, ed-
ited by J. M. Kinney and H. N. Tucker. New York: Raven, 1992, p.
19–59.
10. Elia, M. Organ and tissue contribution to metabolic rate. In:
Energy Metabolism: Tissue Determinants and Cellular Corollar-
ies, edited by J. M. Kinney and H. N. Tucker. New York: Raven,
1992, p. 61–77.
11. Frayn, K. N., P. Lund, and M. Walker. Interpretation of organ
and carbon dioxide exchange across tissue beds in vivo. In: Clin.
Sci. (Colch). 85: 373–384, 1993.
12. Gallagher, D., D. Belmonte, P. Deurenberg, Z. Wang, N.
Krasnow, F. X. Pi-Sunyer, and S. B. Heymsfield. Organ-
tissue mass measurement allows modeling of REE and metaboli-
cally active tissue mass. Am. J. Physiol. Endocrinol. Metab. 275:
E249–E258, 1998.
13. Garby, L., J. S. Garrow, and B. Jörgensen. Reexamination
between energy expenditure and body composition in man:
specific energy expenditure in vivo of fat and fat free tissue.
Eur. J. Clin. Nutr. 42: 301–305, 1988.
14. Garby, L., and O. Lammert. Between-subjects variation in
energy expenditure: estimation of the effect of variation in organ
size. Eur. J. Clin. Nutr. 48: 376–378, 1994.
15. Garby, L., O. Lammert, K. F. Kock, and B. Thobo-Carlsen.
Weights of brain, heart, liver, kidneys and spleen in healthy and
apparently healthy adult Danish subjects. Am. J. Hum. Biol. 5:
291–296, 1993.
16. Heymsfield, S. B., T. Fulenwider, B. Nordlinger, R. Baröow,
P. Stones, and M. Kutner. Accurate measurement of liver,
kidney, and spleen volume and mass by computerized axial
tomography. Ann. Intern. Med. 90: 185–187, 1979.
17. Heymsfield, S. B., C. McManus, J. Smith, V. Stevens, and
D. W. Nixon. Anthropometric measurement of muscle mass:
revised equations for calculating bone free arm muscle area.
Am. J. Clin. Nutr. 36, 680–690, 1982.
18. Heymsfield, S. B., R. Smith, M. Aulet, B. Bensen, S. Licht-
man, J. Wang, and R. N. Piersen, Jr. Appendicular skeletal
Downloaded from www.physiology.org/journal/ajpendo by ${individualUser.givenNames} ${individualUser.surname} (201.223.091.015) on February 4, 2018.
Copyright © 2000 American Physiological Society. All rights reserved.
E315
muscle mass. Measurement by dual photon absorptiometry.
Am. J. Clin. Nutr. 52: 214–218, 1990.
19. Holliday, M. A., D. Potter, A. Jarrah, and S. Bearg. The
relation of metabolic rate to body weight and organ size. Pediatr.
Res. 1: 185–195, 1967.
20. Krebs, H. A. Body size and tissue respiration. Biochim. Biophys.
Acta 4: 249–268, 1950.
21. McNeill, G., M. A. Foster, J. Love, and V. Antfang. Liver and
kidney volume and their relationship to metabolic rate at rest.
Proc. Nutr. Soc. 54: 151A, 1995.
22. Müller, M. J. Hepatic energy and substrate metabolism: a
possible metabolic basis for early nutritional support in cirrhotic
patients. Nutrition 14: 30–38, 1998.
23. Müller, M. J., A. von zur Mühlen, H. U. Lautz, F. W. Schmidt,
M. Daiber, and P. Hürter. Energy expenditure in children with
type 1 diabetes: evidence for increased thermogenesis. Br. Med. J.
299: 487–491, 1989.
24. Nelson, K. M., R. L. Weinsier, C. L. Long, and Y. Schutz.
Prediction of resting energy expenditure from fat-free mass and
fat mass. Am. J. Clin. Nutr. 56: 848–856, 1992.
25. Ravussin, E., and C. Bogardus. Relationship of genetics, age,
and physical fitness to daily energy expenditure and fuel utiliza-
tion. Am. J. Clin. Nutr. 49: 968–975, 1989.
26. Redies, C., L. J. Hoffer, C. Beil, E. B. Marliss, A. C. Evans, F.
Lariviere, S. Marrett, E. Meyer, M. Diksic, A. Gjedde, and
A. M. Hakim. Generalized decrease in brain glucose metabolism
during fasting in humans studied by PET. Am. J. Physiol.
Endocrinol. Metab. 256: E805–E810, 1989.
27. Sparti, A., J. P. DeLany, J. De La Bretonne, G. E. Sander,
and G. A. Bray. Relationship between resting metabolic rate
and the composition of the fat-free mass. Metabolism 46: 1225–
1230, 1997.
28. Tataranni, P. A., and E. Ravussin. Variability in metabolic
rate: biological sites of regulation. Int. J. Obes. 19, Suppl. 4:
S102–S106, 1995.
29. Weinsier, R. L., Y. Schutz, and D. Bracco. Reexamination of
the relationship of resting metabolic rate to fat-free mass and to
metabolically active components of fat free mass in humans.
Am. J. Clin. Nutr. 55: 790–794, 1992.
30. Weir, J. B., and V. De. New methods for calculating metabolic
rate with special reference to protein metabolism. J. Physiol.
(Lond.) 109: 1–9, 1949.
31. Zurlo, F., K. Larson, C. Bogardus, and E. Ravussin. Skeletal
muscle metabolism is a major determinant of resting energy
expenditure. J. Clin. Invest. 86: 1423–1427, 1990.