NU32CH15-Hall ARI 13 April 2012 14:31

V I E W
E Review in Advance first posted online
R

on April 23, 2012. (Changes may

S
still occur before final publication
online and in print.)
C E
I N

A
D V A

Modeling Metabolic
Adaptations and Energy
Regulation in Humans∗
Kevin D. Hall
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

Laboratory of Biological Modeling, National Institute of Diabetes and Digestive and

Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892;
by Brown University on 06/04/12. For personal use only.

email: kevinh@niddk.nih.gov

Annu. Rev. Nutr. 2012. 32:15.1–15.20 Keywords

The Annual Review of Nutrition is online at mathematical model, body weight, body composition, obesity, energy
nutr.annualreviews.org
balance, macronutrient balance
This article’s doi:
10.1146/annurev-nutr-071811-150705 Abstract
0199-9885/12/0821-0001$20.00 Mathematical modeling of human energy regulation and body weight
∗
This is a work of the U.S. Government and is change has recently reached the level of sophistication required for
not subject to copyright protection in the accurate predictions. Mathematical models are beginning to pro-
United States.
vide a quantitative framework for integrating experimental data in
humans and thereby help us better understand the dynamic imbalances
of energy and macronutrients that give rise to changes in body weight
and composition. This review provides an overview of the various ap-
proaches that have been used to model body weight dynamics and en-
ergy regulation in humans, highlights several insights that these models
have provided, and suggests how mathematical models can serve as a
guide for future experimental research.

15.1

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31

loss cures, and every bookshop has an extensive

Contents selection of best-selling diet books. Weight loss
has even become a form of entertainment, with
INTRODUCTION . . . . . . . . . . . . . . . . . . 15.2
a proliferation of reality television programs de-
BODY WEIGHT AND BODY
picting dramatic weight loss interventions.
COMPOSITION . . . . . . . . . . . . . . . . . 15.3
But despite increased public awareness and
ENERGY BALANCE . . . . . . . . . . . . . . . . 15.4
investment in obesity research, a great deal of
Conversion Between Mass and
misinformation and confusion remains about
Metabolizable Energy . . . . . . . . . . . 15.4
the relationship between weight change, nu-
Energy Partition Models . . . . . . . . . . . 15.5
trition, and energy metabolism. Unfortunately,
MACRONUTRIENT BALANCE . . . . 15.6
these misconceptions are not limited to the
Modeling Macronutrient Balance . . . 15.7
general public but are also widespread among
ENERGY EXPENDITURE
the professional nutrition and metabolism com-
DYNAMICS . . . . . . . . . . . . . . . . . . . . . . 15.8
munities. For example, the so-called 3,500-
Thermic Effect of Feeding . . . . . . . . . 15.9
Calorie-per-pound rule has almost ubiqui-
Resting Metabolic Rate . . . . . . . . . . . . 15.9
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

tously been misused to imply that cutting

Physical Activity Expenditure . . . . . . .15.10
500 kcal/d from the daily diet will result in con-
by Brown University on 06/04/12. For personal use only.

Adaptive Thermogenesis and

Metabolic Adaptation . . . . . . . . . . .15.10
INSIGHTS OBTAINED FROM
MATHEMATICAL MODELS . . . .15.11
Is a Calorie a Calorie? The Effect
of Dietary Macronutrients
on Body Composition . . . . . . . . . . .15.11
Behavioral Adaptations to Energy
Imbalance . . . . . . . . . . . . . . . . . . . . . .15.12
Weight Change Variability Due to
the Uncertainty in Baseline
Energy Requirements . . . . . . . . . . .15.13
Greater Weight and Body Fat Gain
in Obesity for the Same
Increment in Energy Intake . . . . .15.14
CONCLUSIONS AND FUTURE
DIRECTIONS . . . . . . . . . . . . . . . . . . . .15.15
INTRODUCTION
The emergence of the worldwide obesity epi-
demic (91) and its public health consequences
(101) emphasizes the critical importance of un-
derstanding human energy regulation and its
effect on body weight and composition. In-
tense public interest in weight control has led
to a proliferation of diet and exercise prod-
ucts, with Americans alone spending tens of bil-
lions of dollars per year on these products (14).
Media advertisements promise miracle weight
stant weight loss rate of about one pound per
week (44). Such prescriptions have been em-
braced by official health and nutrition organi-
zations around the world (20, 69–71) and have
been erroneously used for individual weight
loss counseling as well as predicting the po-
tential impact of policy changes on population
obesity prevalence (25, 32, 88, 89). Although
it is generally acknowledged that weight loss
will gradually slow over time as a result of
the decreasing energy requirements, the mag-
nitude and time course of the slowing were
difficult to determine since this is a dynamic
process.
To help address these issues, dynamic math-
ematical models of human energy regula-
tion and weight change have been developed.
Although such modeling began in the 1970s
(1, 4, 74, 75), only within the past several
years have mathematical models become suffi-
ciently sophisticated to adequately capture the
most relevant physiology, guide experimental
research, and make accurate weight change pre-
dictions (33, 38, 43, 44, 94, 95). This review
provides an overview of different approaches to
mathematical modeling of body weight change
and energy regulation in humans, highlights
several insights that these models have pro-
vided, and suggests how mathematical models
can serve as a guide for future experimental
research.

15.2 Hall

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31

BODY WEIGHT AND BODY Body fat, protein, and glycogen comprise
COMPOSITION the stored energy of the body, and these stores
must be mobilized when the diet is insufficient
Body weight is an easily measured quantity that
to meet the body’s energy requirements.
has often been the primary variable of interest
Figure 1b illustrates the composition of the
in mathematical models of human energy regu-
body in terms of its energy content, with
lation (4, 12, 13, 56, 66). However, body weight
fat stored in adipose tissue providing the
is less important than body composition when
overwhelming majority of the available stored
it comes to health consequences. Thus, many
energy, especially in obesity. Despite dietary
models of human energy regulation have fo-
carbohydrate providing the majority of the
cused on predicting body composition change
body’s energy demands on a daily basis, glyco-
at some level of detail (1–3, 33, 38, 39, 43, 44,
gen represents a relatively insignificant store of
57, 74, 75, 94, 95, 102, 107).
energy (∼2,000 kcal). Body protein represents
Figure 1a illustrates the body composition
a substantial amount of energy, but in humans
in terms of body fat and fat-free mass in a typ-
it is not a storage pool in the same sense as adi-
ical obese man and lean man. Obesity is char-
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

pose tissue triglyceride. Rather, body proteins

acterized by a greatly expanded body fat mass
are functionally important and cannot be de-
but also an increased amount of fat-free mass.
by Brown University on 06/04/12. For personal use only.

pleted by a significant fraction without serious

Figure 1a also illustrates the chemical com-
complications and death. In contrast, fat stores
position of the fat-free mass, with water being
represent a considerable energy reserve, and
its greatest component. The absolute masses
body fat can be depleted to very low levels with-
of body protein and bone mineral are also in-
out substantial functional impairments (30, 61).
creased in obesity. Glycogen and cellular solids,
Many mathematical models of human en-
such as potassium and nucleic acids, contribute
ergy regulation have represented body fat as
a very small fraction of the fat-free mass.
well as fat-free mass components (1–3, 33, 38,

Figure 1
(a) The chemical composition of an obese 120 kg man (left) differs from that of a lean 70 kg man (right),
primarily as a result of the increased body fat mass. The fat-free mass (FFM) of the body is mostly water,
with a significant contribution of protein and bone mineral. Cellular solids and glycogen make up a very
small part of the chemical composition of the body. (b) Body energy stores are greatly expanded in the obese
man, primarily as a result of the increased body fat mass.

www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans 15.3

Changes may still occur before final publication online and in print
 NU32CH15-Hall
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
by Brown University on 06/04/12. For personal use only.
ARI 13 April 2012 14:31
39, 43, 44, 57, 74, 75, 94, 95, 102, 107). A
recent model also includes rapid body water
changes as a component of the fat-free mass that
can be important contributors to early weight
change (44). Only relatively complex compu-
tational models have represented the detailed
chemical composition of the body (33, 38).
ENERGY BALANCE
Most mathematical models of human energy
regulation make the assumption that weight
change is solely determined by an imbalance
between dietary energy intake and the energy
expended by the body to maintain life and per-
form physical work. The theoretical underpin-
ning of this energy balance concept is the first
law of thermodynamics, and all energy balance
models can be mathematically described by the
following equation:
d earliest mathematical model of human weight
(ρBW ) = EI − EE, (1)
dt
where the left side of the equation is the rate
of change of body energy, with BW being the
body weight and ρ being an energy density con-
verting between units of metabolizable energy
and mass. The right side is the energy imbalance
between the body’s energy intake rate, EI, and
energy expenditure rate, EE. Any of the terms
in the above equation can depend on time, t, as
well as other parameters. A conceptual repre-
sentation of energy balance models is depicted
in Figure 2.
A popular, but erroneous, application of the
energy balance concept involves the assump-
tion that the energy expended by the body
EI BW EE
Figure 2
Schematic depiction of the energy balance concept,
in which changes in body weight, BW, are assumed
to result solely from an imbalance between energy
intake, EI, and energy expenditure, EE. Translating
an energy imbalance to a change in weight requires
an assumption of the energy density of the weight
change.
15.4 Hall
Changes may still occur before final publication online and in print
remains relatively unchanged when energy in-
take deviates from an energy-balanced diet by
an amount EI. In that case, the energy balance
Equation 1 gives the following equation for
the rate of weight change, assuming a constant
value for ρ:
dBW EI
= . (2)
dt ρ
In other words, the rate of weight change is a
constant and depends only on the magnitude
of the diet change, EI, and the energy den-
sity of the weight change, ρ. With the choice of
ρ = 3,500 kcal/lb, this erroneous equation en-
capsulates the static 3,500-Calorie-per-pound
rule that has been ubiquitously misused to
predict weight change (44).
The most serious error of Equation 2 is that
the energy expenditure rate does not stay con-
stant but rather dynamically changes. Even the
change recognized that dynamic changes in en-
ergy expenditure must be taken into account
(4), and it is curious that the persistent erro-
neous use of Equation 2 remains so popular to
the present day. The assumption of a constant
ρ = 3,500 kcal/lb is also an oversimplification
that will be dealt with forthwith.
Conversion Between Mass
and Metabolizable Energy
A simple translation between the energy imbal-
ance and the rate of weight change, dBW/dt,
occurs only if the energy density of the weight
change, ρ, is a constant parameter. However,
as described above, the body is composed of
a variety of chemical constituents with widely
varying energy densities. For example, fat has
an energy density of about 9.4 kcal/g, whereas
protein and carbohydrate have energy densi-
ties of about 4.7 kcal/g and 4.2 kcal/g, respec-
tively (64). Other major chemical constituents
of the body, water being the most sizable, have
metabolizable energy densities of zero. Thus,
translating a given energy imbalance to a rate of
weight change requires additional assumptions
 NU32CH15-Hall ARI 13 April 2012 14:31
about the chemical composition of the weight
change.
The historical basis of the common ρ =
3,500 kcal/lb assumption can be traced to the
idea that body weight changes are primarily due
to loss or gain of adipose tissue, which is com-
posed of about 87% triglyceride (110). But de-
spite the simplicity of this idea, we know that
body water significantly contributes to overall
weight change, especially over the first weeks
following a reduced-calorie diet (49). Because
body water has no metabolizable energy con-
tent, early weight changes have an energy den-
sity substantially lower than 3,500 kcal/lb (48).
The mathematical models of Hall et al. (38, 44)
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
are the only models to account for these early
body water changes in terms of both intracel-
by Brown University on 06/04/12. For personal use only.
lular and extracellular fluids.
Longer-term changes in body fat are accom-
panied by changes in lean tissue mass whose me-
tabolizable energy density is significantly less
than that of body fat (35). To model these
longer-term body composition changes, Forbes
(28, 29) hypothesized that the proportion of
weight change resulting from lean versus fat
tissue is a nonlinear function of body fat. The
Forbes hypothesis has since been extended and
validated for longitudinal data implying that ρ
is a nonlinear function of the body composition
(34, 43, 93). Nevertheless, for small changes of
weight from an initial baseline it is valid to ap-
proximate the nonlinear Forbes curve with a
line, and the resulting slope gives a value for
ρ. For moderately overweight and obese indi-
viduals, the ρ = 3,500 kcal/lb is a reasonable
approximation, but this value significantly over-
states ρ for lean individuals (35).
Energy Partition Models
To explicitly model body composition change,
the energy balance Equation 1 can be written
as a pair of equations for the changes in both
lean tissue, L, and body fat, F:
dL
ρL = P (E I − E E)
dt . (3)
dF
ρF = (1 − P )(E I − E E)
dt
www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans
Changes may still occur before final publication online and in print
F L
(1-P) P
EI Δ EE
Figure 3
Schematic depiction of the energy partition concept,
where any energy imbalance, , is partitioned
between energy stored in body fat, F, or lean tissues,
L. The energy partition ratio, P, specifies the
fraction of the energy imbalance to/from lean tissue.
The energy densities of changes in L and F
substantially differ. EE, energy expenditure;
EI, energy intake.
Based on assumptions about the chemical com-
position of lean and fat changes, the energy
densities have been estimated to be about
ρ L = 1.8 kcal/g and ρ F = 9.4 kcal/g, respec-
tively (35). The energy partition ratio, P, ranges
between 0 and 1 and determines the proportion
of an energy imbalance directed to and from
lean versus fat mass. The energy partition con-
cept is illustrated in Figure 3.
The seminal energy partition model of
Payne and Dugdale referred to P as the p-ratio,
indicating that this factor determines the pro-
portion of an energy imbalance accounted for
by body protein changes since they ignored the
contribution of glycogen (74, 75). Payne and
Dugdale hypothesized that P was a constant
parameter whose value could vary between peo-
ple and could thereby explain variable interindi-
vidual weight change for equivalent changes of
diet.
Applying the Forbes hypothesis to the
energy partition model reveals that P is a
nonlinear function of F: P = C/(C+F),
where C = 10.4 kg × ρ L /ρ F , as depicted in
Figure 4 (11, 34). The initial value of P can
be quite diverse for individuals with very differ-
ent fat mass, in agreement with the Payne and
Dugdale hypothesis. However, the nonlinear-
ity of the Forbes body composition curve sug-
gests that P is not a fixed parameter because it
depends continuously on the fat mass, which
can change considerably with large weight
changes.
15.5
 NU32CH15-Hall
0.3
Energy partition ratio (P )
0.2
0.1
Figure 4
The Forbes hypothesis for the energy partition ratio, P, as a nonlinear function
of the body fat mass.
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
by Brown University on 06/04/12. For personal use only.
ARI 13 April 2012 14:31
0
0 25 50 75 100
Body fat (kg)
Weinsier et al. (102) were the first to use the
nonlinear Forbes relationship in an energy par-
tition model. A recent energy partition model
also used the Forbes relationship to describe
body fat and lean tissue changes but added the
effect of diet on extracellular fluid mass along
with changes of glycogen and its associated
intracellular water (44). These modifications
substantially improved the model’s ability to
simulate rapid changes of total body water that
underlie most of the initial weight loss fol-
lowing the induction of low-calorie, and espe-
cially low-carbohydrate and low-sodium, diets
(49).
More similar to the original Payne and
Dugdale hypothesis, Kozusko (57) used a
nonlinear function of the initial body fat to
determine a constant value of P for each
dynamic simulation. Westerterp et al. (107)
considered a more complex algorithm for de-
termining P based on body fat as well as energy
intake. Thomas et al. (93–95) determined P
by generating several polynomial equations to
fit cross-sectional body composition data from
men and women of different racial groups.
Despite the increased complexity of their pro-
posed equations for calculating P, these authors
showed that the simple Forbes equation pro-
vided more accurate predictions of longitudinal
body composition change for all cases except
15.6 Hall
Changes may still occur before final publication online and in print
those in which the body fat approaches zero
(93).
MACRONUTRIENT BALANCE
In his early twentieth-century textbook,
Food and the Principles of Dietetics, Robert
Hutchison (50) compared the human body
to a steam engine, noting that “The building
material of food corresponds to the metal of
which the engine is constructed, the energy-
producers to the fuel which is used to heat
the boiler. Where the body differs from the
engine is that it is able to use part of the
material of its construction for fuel also” (50).
A more modern engineering analogy might
be an automobile that can run on an arbitrary
mixture of different fuels (37). Such a flex-fuel
vehicle would allow the driver to fill the tank
with whatever fuel was cheaper or more readily
available, regardless of what mixture is already
in the tank. Although designing a flex-fuel
vehicle would be a significant engineering
challenge, imagine the additional complexity
if the vehicle could have no fuel tank. Rather,
the vehicle itself must be composed of its
fuel and must continually break down and
reconstruct its components. Furthermore,
despite the daily turnover of its components
and fluctuations of fuel delivery, the compo-
sition of the vehicle must remain relatively
stable and maintain similar performance
characteristics.
Exactly this remarkable engineering feat
is accomplished by the human body through
its use of the three dietary macronutrients
(carbohydrate, fat, and protein) to both fuel
metabolism and provide substrates for body
constituents. These macronutrients are ob-
tained from the diet, with about 50% of the
energy derived from carbohydrate, 35% from
fat, and 15% from protein (5). However, these
average diet proportions can vary widely from
person to person and also from day to day.
Complex physiological mechanisms maintain
normal functioning of the body despite marked
fluctuations of diet quantity and composition.
 NU32CH15-Hall ARI 13 April 2012 14:31
FI EI-FI
F L
FatOx EE-FatOx
Figure 5
Schematic depiction of the simple macronutrient
balance concept, where changes in body fat, F, are
determined by an imbalance between fat intake, FI,
and fat oxidation, FatOx, and changes in lean mass,
L, are determined by the remaining energy
imbalance. EE, energy expenditure; EI, energy
intake.
Modeling Macronutrient Balance
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
Although the molecular, cellular, and phys-
by Brown University on 06/04/12. For personal use only.
iological mechanisms underlying the regula-
tion of human macronutrient metabolism are
exceedingly complex, the whole-body system
obeys thermodynamic laws that constrain its
dynamics in ways that make the overall system
amenable to mathematical modeling (11). For
example, macronutrient imbalances between
dietary intake and metabolic utilization under-
lie changes of stored fat, glycogen, and protein
and result in changes in the chemical compo-
sition of the body. Thus, models of macronu-
trient balance calculate dynamic changes in the
chemical composition of the body as a result
of imbalances between intake and utilization
of macronutrients (33, 38). The overall body
weight change is just the sum of the individual
changes in body constituents.
The simplest mathematical model of
macronutrient balance and its relationship to
body composition change can be expressed as:
dL
ρL = (EI − FI) − (EE − FatOx)
dt
, (4)
dF
ρF = FI − FatOx
dt
where FI is the energy intake from dietary fat
and FatOx is the net energy derived from fat
oxidation. The simple macronutrient balance
model is conceptually illustrated in Figure 5.
Both energy balance models and energy par-
tition models described by Equations 1 and 3
www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans
Changes may still occur before final publication online and in print
are special cases of the more general macronu-
trient balance model (11, 39). For example,
when the simple macronutrient balance model
is constrained so that L and F follow along the
Forbes nonlinear body composition curve, the
resulting model is equivalent to the energy par-
tition model described by Equation 3 with the
energy partition ratio P shown in Figure 4 (11).
Although it is more general than the energy bal-
ance and energy partition models, the macronu-
trient balance model also obeys the first law of
thermodynamics.
The equivalence of Equations 3 and 4
under the constraint of a specified body com-
position relationship provides a quantitative ex-
planation of how interactions between diet,
energy expenditure, and fat oxidation are con-
nected to changes of body composition (11, 39).
Hall et al. (39) thereby derived an equation
that accurately predicted the observed changes
of body composition and metabolic fuel se-
lection during both experimental under- and
overfeeding in adult humans when the mea-
sured food intake and total energy expenditure
were provided as inputs to the model. A similar
approach was used to calculate metabolic fuel
selection during normal human infant growth
and provided the first dynamic picture of how
metabolism adapts in concert with changes of
diet and energy expenditure to give rise to nor-
mal tissue deposition over the first two years of
life (51).
Equation 4 represents a simplified version
of a more comprehensive computational model
of human macronutrient balance and its rela-
tionship to body composition change (33, 38).
The computational model quantitatively tracks
the metabolism of all three dietary macronu-
trients and simulates how diet changes result in
adaptations of whole-body energy expenditure,
metabolic fuel selection, and alterations in
the major whole-body fluxes contributing to
macronutrient balance. The macronutrient
balance model is conceptually depicted in
Figure 6 and mathematically represented by
the following equations describing changes in
the body’s energy stores of glycogen (G), fat
15.7
 NU32CH15-Hall
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
by Brown University on 06/04/12. For personal use only.
ARI 13 April 2012 14:31
(F ), and protein (P):
dG
ρC = CI − DNL + GNGp
dt
+ GNGf − G3P − CarbOx
dF
ρF = FI + εd DNL − KUexcr , (5)
dt
− (1 − εk )KTG − FatOx
dP
ρP = PI − GNGp − ProtOx
dt
where ρ C , ρ F , and ρ P are the energy densities
of carbohydrate, fat, and protein, respectively.
The macronutrient intake rates, CI, FI, and
PI, refer to the metabolizable energy intake
rates of dietary carbohydrate, fat, and protein,
respectively. The rates of gluconeogenesis
from amino acids and glycerol are indicated
by GNGp and GNGf , respectively. The ef-
ficiencies of de novo lipogenesis, DNL, and
ketogenesis, KTG, were represented by the
parameters εd and εk , respectively. When the
ketogenic rate increases, ketones are excreted
in the urine at the rate KUexcr . Some flux of
carbohydrates is provided for the production
of glycerol 3-phosphate, G3P, which is used
in the synthesis of triglyceride. The oxidation
rates CarbOx, FatOx, and ProtOx, summed
to the energy expenditure rate, EE, less the
small amount heat produced via flux through
ketogenic and de novo lipogenic pathways.
Turnover of glycogen, fat, and protein and the
corresponding energy costs are also included
in the model. To account for body water
shifts with diets that vary in macronutrient as
well as sodium content, the model also tracks
intracellular and extracellular fluid changes.
The main model assumptions are that
changes of the body’s energy stores are given
by the sum of metabolic fluxes entering the
pools minus the fluxes exiting the pools. Hence,
the computational model obeys the first law
of thermodynamics, and the most recent ver-
sion was developed using published human data
from over 50 experimental studies and was val-
idated by comparing model predictions with
the results of several controlled feeding stud-
ies not used for model development (38). To
date, this is the only mathematical model of
15.8 Hall
Changes may still occur before final publication online and in print
FI CI PI
DNL
GNGp
F G P
GNGf
FatOx CarbOx ProtOx
Figure 6
Schematic depiction of the macronutrient balance
model, in which changes in body glycogen, G, fat,
F, and protein, P, are determined by the imbalances
between various macronutrient fluxes (arrows).
Contributions of exchange fluxes, such as
gluconeogenesis, GNG, and de novo lipogenesis,
DNL, contribute to imbalances between the intake
and oxidation rates of carbohydrate, fat, and protein.
CarbOx, carbohydrate oxidation; CI, carbohydrate
intake; FatOx, fat oxidation; FI, fat intake,
PI, protein intake; ProtOx, protein oxidation.
human metabolism to consider all three di-
etary macronutrients and accurately simulate
the metabolic and body composition changes in
response to diet and physical activity changes in
a wide variety of subject groups.
ENERGY EXPENDITURE
DYNAMICS
When diet or physical activities are changed,
the body’s metabolic fuel selection and the en-
ergy expenditure rate dynamically adapt. All
mathematical models designed to accurately
predict body weight and composition change
must somehow account for these energy ex-
penditure dynamics to avoid the same fatal flaw
of the static 3,500-Calorie-per-pound weight
loss rule. Figure 7 depicts the energy expen-
diture components for examples of relatively
sedentary lean and obese men. The obese man
requires several hundred additional kcal/d to
maintain his increased weight compared to the
lean man. Although this general dependence
of weight on energy expenditure may be cap-
tured by a simple model of body weight alone,
more realistic models of energy expenditure in-
clude its multiple physiological components:
the thermic effect of feeding, resting metabolic
rate, and physical activity.
 NU32CH15-Hall ARI 13 April 2012 14:31
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
Figure 7
by Brown University on 06/04/12. For personal use only.
The total energy expenditure in relatively sedentary
obese and lean men, comprising the thermic effect
of feeding, resting metabolic rate, and the physical
activity.
Thermic Effect of Feeding
The smallest component of the total energy
expenditure rate in humans is the thermic
effect of feeding (also called diet-induced ther-
mogenesis or specific dynamic action), defined
as the increase of metabolic rate observed for
several hours following the ingestion of a meal.
The thermic effect of feeding is believed to
represent the energy cost of digestion and
absorption as well as the storage and metabolic
fate of dietary macronutrients (106). Although
the precise mechanisms underlying the thermic
effect of feeding are not fully understood, there
is a clear dietary macronutrient hierarchy in
the magnitude of the metabolic rate increase
after feeding, with protein causing a greater
increment than carbohydrate, which is greater
than that of fat. The computational model
of macronutrient balance model accounts for
this hierarchy (33, 38), but all other energy
balance and energy partition models ignore
the macronutrient effect and assume that
the thermic effect of feeding is given by an
overall proportion of energy intake, typically
∼7%–14%.
www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans
Changes may still occur before final publication online and in print
Resting Metabolic Rate
The resting metabolic rate (RMR) corresponds
to the energy expended by the body when not
performing physical work and is typically the
largest contribution to the total energy expen-
diture rate. Contrary to popular belief, obese
people generally have a higher absolute rest-
ing metabolic rate compared to lean people
(Figure 7). Readily available clinical measures
(e.g., sex, height, weight, and age) have been
used along with RMR measurements to gen-
erate empirical equations with RMR being an
increasing function of body weight, commonly
a linear or a power law relationship. Although
several mathematical models of body weight dy-
namics have used this simplified approach to
modeling RMR (4, 94, 95), it has long been rec-
ognized that the main contributor to the RMR
is the fat-free mass because it comprises the
metabolically active tissues of the body (15).
Fat-free mass is elevated in obesity along
with the increased body fat mass, which also
contributes to increased resting metabolic rate
in obesity. The linear relationship between
resting metabolic rate and fat-free mass is iden-
tical in obese and lean people (15, 108). This
means that the elevated resting metabolic rate
in obesity is generally in line with what is
expected for the body composition of obese
people.
Although fat-free mass and, to a lesser ex-
tent, fat mass are good predictors of RMR, such
models explain only about 70% of interindi-
vidual RMR variability, such that for a given
body composition the RMR standard deviation
is about 300 kcal/d (15, 108). Since there is a
large range of specific metabolic rates among
various organs that contribute to the fat-free
mass (21), some of this residual RMR variabil-
ity may be due to differences in organ masses.
Magnetic resonance imaging methodologies
have been used to quantify organ sizes and, us-
ing assumptions regarding the organ-specific
metabolic rates, RMR prediction equations that
sum the individual metabolic rates of various
organs explain about 80% of the RMR vari-
ability (31, 67, 68). Thus, increasingly detailed
15.9
 NU32CH15-Hall
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
by Brown University on 06/04/12. For personal use only.
ARI 13 April 2012 14:31
knowledge of body composition may improve
RMR predictions.
The Payne and Dugdale energy partition
model was the first to discriminate between
“fast” and “slow” lean tissue contributions
to RMR (74, 75). More recently, the com-
putational model of macronutrient balance
(38) incorporated how changes in the sizes of
various organs affect RMR, assuming linear
relationships between changes of fat-free
mass and various organ sizes based on cross-
sectional data from 110 men and women with
body mass index between 18 and 37 kg/m2
(D. Gallagher, personal communication). Of
course, longitudinal organ mass changes with
weight gain and loss need not follow the
cross-sectional relationships; this possibility
requires experimental investigation.
Another potentially important contributor
to RMR dynamics may involve flux changes
through various energy-requiring metabolic
pathways. The major macronutrient fluxes
of gluconeogenesis, de novo lipogenesis,
triglyceride synthesis, and protein turnover
all require energy, and these flux rates can be
significantly influenced by the energy content
of the diet as well as its composition. For
example, the breakdown and resynthesis of
body fat requires eight molecules of adenosine
triphosphate (ATP) per molecule of triglyc-
eride (22), and the flux through this pathway is
strongly influenced by dietary carbohydrate via
insulin’s inhibition of lipolysis. Similarly, pro-
tein synthesis requires four ATP per peptide
bond plus one ATP for amino acid transport
(6). Such energy-requiring metabolic fluxes
have been incorporated into computational
models of macronutrient balance (33, 38) and
may explain the observed energy cost of tissue
deposition that is especially important during
growth and weight gain (36).
Physical Activity Expenditure
The physical activities of humans typically
involve locomotion, and the energy costs are
determined by the duration and intensity of
physical activity in proportion to the overall
15.10 Hall
Changes may still occur before final publication online and in print
body weight (98). Thus, obese and lean people
can have similar daily energy costs for physical
activity despite obese people typically being
less active. With weight loss, it costs less energy
to perform most physical activities, and there-
fore the physical activity expenditure typically
decreases unless the quantity or intensity of
physical activity increases to compensate.
All previous mathematical models of human
energy expenditure include the body weight
effect on physical activity expenditure. Some
models further subdivide physical activity ex-
penditure into volitional activities (e.g., exer-
cise) and low-intensity spontaneous physical
activity or nonexercise activity thermogenesis
(38, 94, 95).
Adaptive Thermogenesis and
Metabolic Adaptation
During active weight loss, both RMR and en-
ergy expenditure have been observed to de-
crease to an extent greater than expected based
on the measured body weight and composition
change (18, 19, 46, 60, 83). Furthermore, this
improved energy efficiency appears to persist
once energy balance is established at a lower
body weight (82), although the magnitude of
this persistent effect is smaller than during ac-
tive weight loss, and its existence has been
controversial (26, 104). Conversely, overfeed-
ing and weight gain can result in highly vari-
able increases of energy expenditure that can
be greater than expected based on the observed
weight gain (62, 63). Collectively, these phe-
nomena have been called adaptive thermogen-
esis (65, 83) or metabolic adaptation (58, 80).
In the case of overfeeding, the term luxuskon-
sumption has been used to describe adaptive
increases in energy expenditure (79).
Most mathematical models of energy reg-
ulation in humans have ignored such adaptive
changes in energy efficiency, although several
have attempted to account for such effects (33,
38, 41, 43, 44, 56, 57, 72, 94, 95, 107). The
energy partition models of Hall et al. (41, 43,
44) incorporated adaptive thermogenesis as an
additive term that was a linear function of the
 NU32CH15-Hall ARI 13 April 2012 14:31
energy intake change. In these models, the
value of the adaptive thermogenesis parameter
was chosen to match changes in overall energy
expenditure measured before and after approx-
imately stable weight loss (43). When active
weight loss is followed by subsequent weight
stabilization, the change in energy intake
required for the weight loss phase is greater
than that required for weight stabilization at
the reduced weight. Thus, modeling adaptive
thermogenesis as a function of energy intake
change has the natural consequence of increas-
ing its magnitude in situations of active weight
loss, in agreement with observations (104).
A similar approach was used by Westerterp
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
et al. (107), who added an energy expenditure
reduction term to account for the effect of
by Brown University on 06/04/12. For personal use only.
the degree of energy imbalance on metabolic
rate. In contrast, the energy partition models
of Thomas et al. (94, 95) used a constant
parameter to model the metabolic adaptation
of RMR with weight loss and did not distin-
guish between active weight loss and weight
stabilization. Kozusko (56, 57) considered a
set-point model of adaptive thermogenesis as
a function of the body weight itself.
Experimental quantification of the adaptive
thermogenesis magnitude depends on the
definition of the expected values for RMR
and total energy expenditure. Typically,
cross-sectional regression equations are used
to calculate the expected values, using for RMR
and total energy expenditure measurements
derived either from baseline body composition
data in the same subjects (46, 60, 80) or from
a separate group of similar subjects (18, 19).
But such expected values for RMR and energy
expenditure ignore the possible changes in
organ size distribution as well as changes in
fluxes through energy-requiring metabolic
pathways during over- or underfeeding
(described above). Whether such consider-
ations can explain the observed changes in
energy efficiency is unclear.
Novotny & Rumpler (72) used a mathe-
matical model to investigate the impact of a
disproportionate reduction of high-metabolic-
rate organs during weight loss and found
www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans
Changes may still occur before final publication online and in print
that such changes could potentially explain
the observed reduction of RMR, but the
organ sizes were not measured. Conversely,
the computational model of macronutri-
ent balance that accounts for alterations of
energy-requiring metabolic fluxes as well as
organ mass changes required an adaptive
thermogenesis model variable to explain the
observed average decrease in both RMR and
total energy expenditure with weight loss (33,
38). The adaptive thermogenesis variable was
modeled as a linear function of the reduction in
energy intake below baseline and was used to
suppress the metabolic rate of all organs as well
as reduce the energy expended in spontaneous
physical activity. The mechanistic basis of such
a metabolic adaptation is unclear but may be
related to reduced sympathetic drive or blunted
thyroid activity, possibly as a result of decreased
circulating leptin (58, 81, 83–85, 103).
Interestingly, the adaptive thermogenesis
effect was not required to accurately simulate
overfeeding and weight gain in the compu-
tational model of macronutrient balance (33,
38). Similarly, the energy partition models of
Westerterp et al. (107) and Thomas et al.
(94, 95) required a metabolic adaptation param-
eter to explain changes of energy expenditure
with weight loss but not weight gain. If cor-
rect, such an asymmetry in energy regulation
suggests that the body is neutral to overfeed-
ing and weight gain but actively resists weight
loss by improving its energy efficiency during
underfeeding—much to the dismay of over-
weight and obese people wishing to lose weight.
INSIGHTS OBTAINED FROM
MATHEMATICAL MODELS
Is a Calorie a Calorie? The Effect
of Dietary Macronutrients
on Body Composition
A topic of great popular interest is the rela-
tive effectiveness of weight loss diets varying
in macronutrient composition (92). While fully
complying with the laws of thermodynamics,
macronutrient balance models allow for the
15.11
 NU32CH15-Hall
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
by Brown University on 06/04/12. For personal use only.
ARI 13 April 2012 14:31
possibility that body weight may depend on
the diet composition because the energy stored
per unit mass of carbohydrate, fat, and protein
varies considerably, especially when account-
ing for the intracellular water associated with
stored glycogen and protein. Furthermore, di-
etary carbohydrates have an impact on renal
sodium excretion via insulin (16), which re-
sults in concomitant changes in extracellular
fluid volume. Therefore, when the composi-
tion of the diet is altered, transient changes
in macronutrient stores and body fluid shifts
will result in an expected body weight change
even when the energy content of the diet is held
constant (99).
A more important consideration is whether
body fat mass depends on the macronutrient
composition of the diet. Proponents of low-
carbohydrate diets for weight loss emphasize
the ability of dietary carbohydrate to influ-
ence adipose tissue fat storage via circulating
insulin (92). Therefore, it is claimed that
body fat changes depend primarily on dietary
carbohydrate and not the energy content of
the diet. Macronutrient balance models allow
for this possibility, but energy balance and
energy partition models assume that “a calorie
is a calorie” (8) and therefore all equivalently
reduced energy diets should lead to identical
body fat loss regardless of their macronutrient
composition. However, this assumption is
not an obvious or necessary consequence of
the first law of thermodynamics because the
more general macronutrient balance models
allow for an effect of diet composition while
also obeying the law of energy conservation.
In fact, to achieve an independence of body
fat on the macronutrient content of the diet
requires a robust physiological control system
to precisely adapt metabolic fuel selection to
the diet composition (37). Nevertheless, most
inpatient studies with adequately controlled
diets have shown little impact of diet composi-
tion on body weight and fat mass changes (55,
59, 100, 105, 111, 112), but there are notable
exceptions (52, 77, 78). Thus, it remains to be
determined whether the physiology regulating
metabolic fuel selection can fully adapt to
15.12 Hall
Changes may still occur before final publication online and in print
diet composition over extended time periods
without resulting in body fat mass changes.
Another argument against the concept that
a calorie is a calorie is that the diet compo-
sition can have a significant impact on the
flux pattern through various energy-requiring
metabolic pathways and may thereby affect
the body’s energy expenditure rate (23, 24).
Despite the attractive theoretical possibility of
a significant “metabolic advantage” of one diet
over another, simulations using a computa-
tional model that includes the contributions
of energy-requiring metabolic fluxes (38) sug-
gest that the overall effect of diet composition
on energy expenditure appears to be relatively
modest, especially when dietary protein is un-
changed. Specifically, the model predicts that
large isocaloric exchanges of dietary carbohy-
drate and fat will result in energy expenditure
changes of around 100 kcal/d. These model re-
sults agree with experimental observations (86),
and therefore the assumption that a “calorie is a
calorie” may be a reasonable first approximation
over relatively short time periods. However,
even small differences in energy expenditure
and macronutrient balance can theoretically
lead to significant differences of body weight
and composition if the diets are maintained over
long periods. A 100 kcal/d difference in energy
expenditure alone could lead to an initial body
fat imbalance of about 10 g/d. Using current
body composition methods, it would require a
sustained period of about 100 days to detect
such a difference in body fat. Nevertheless, this
possibility requires further investigation.
Behavioral Adaptations
to Energy Imbalance
Energy imbalance and weight change can
influence behaviors that directly affect energy
intake and/or expenditure. For example,
outpatient weight loss interventions geared
to reducing energy intake typically result in a
period of weight loss that plateaus after about
six to eight months, followed by slow weight
regain (47, 90). The well-known deficiencies
in assessing free-living energy intake (97, 109)
 NU32CH15-Hall ARI 13 April 2012 14:31
make it extraordinarily difficult to interpret
such results, and mathematical models have
recently been proposed to address this difficulty
by estimating changes of free-living energy
intake using longitudinal measurements of
body weight (38, 40, 96). Such models have
shown that the typical weight loss, plateau, and
regain trajectory was likely due to short-lived
adherence to the prescribed diet that was pro-
gressively relaxed over the first year to return
to the preintervention level (38, 44). Thus, an
energy imbalance resulting in transient weight
loss leads to an eventual adaptation of behavior
to return to the original lifestyle.
Another example of a behavioral adapta-
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
tion to energy imbalance is the compensatory
changes in energy intake when a physical ac-
by Brown University on 06/04/12. For personal use only.
tivity program is added. In particular, engaging
in many weeks of supervised exercise leads to a
wide range of individual weight changes, with
some people gaining weight and others having
greater-than-expected weight loss (7, 10, 53,
54). These results imply that volitional phys-
ical activity can have a wide range of effects on
energy intake and/or other components of total
energy expenditure.
Under- and overfeeding may also signifi-
cantly influence physical activity expenditure,
especially nonvolitional spontaneous physical
activity (62, 63). This effect is very difficult
to model because of the highly variable in-
terindividual response. Thomas et al. (94, 95)
attempted to account for the average change in
spontaneous physical activity in the presence
of this variability. However, the mathematical
equation used to model this effect has two
constraints on its use. First, the model is
not applicable during prolonged substantial
underfeeding because the equation allows for
the possibility of negative values for the spon-
taneous physical activity expenditure that is not
physiological. Second, the model requires that
changes in any other component of total energy
expenditure are positively correlated to changes
in spontaneous physical activity. For example,
increased volitional exercise in the Thomas
model necessarily results in an increased spon-
taneous physical activity. Such an effect would
www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans
Changes may still occur before final publication online and in print
not apply to people where added exercise results
in fatigue and a corresponding reduction in sub-
sequent spontaneous physical activity. Hence,
application of the model of Thomas et al. re-
quires a priori knowledge of the highly variable
response characteristics of the individual.
The highly variable behavioral adaptations
to energy imbalance pose a significant challenge
for modeling individual weight changes. Rather
than attempting to directly simulate these be-
havior changes, models that start by assum-
ing no effect might be used to help estimate
the magnitude of any behavioral adaptations by
measuring the difference between the measured
and model-predicted weight change. Further-
more, models that can quantitatively integrate
physiological data collected during the inter-
vention may be used to better characterize be-
havioral adaptations in the overall context of
energy and macronutrient balance.
Weight Change Variability
Due to the Uncertainty in Baseline
Energy Requirements
Calculating the energy imbalance generated
by a given diet requires knowing the energy
requirement to maintain the baseline body
weight. Unfortunately, even the specialized and
expensive doubly labeled water method can-
not measure the initial energy requirements of
a free-living individual with a precision bet-
ter than ∼5% (87). The uncertainty of the
baseline energy requirements translates to an
expected interindividual variability of weight
change even if adherence to a prescribed diet
is perfect (44). This is a fundamental limita-
tion on a model’s ability to precisely predict the
body weight time course of an individual. For
example, assuming a ± 150 kcal/d uncertainty
in the initial energy expenditure requirements,
the dashed curves in Figure 8a illustrate the
minimum expected weight gain variability dur-
ing a 500 kcal/d overfeeding study. In this ex-
ample, the energy partition model of Hall et al.
(44) was used to simulate multiple runs of the
same 70-kg “virtual study subject,” assuming
perfect diet adherence and only differing in the
15.13
 NU32CH15-Hall ARI 13 April 2012 14:31

a illustrates the predicted average weight change

75
for a 500 kcal/d increase of daily energy intake
74 in both a 120 kg man and a 70 kg man using an
Body weight (kg)

energy partition model (44). Although the av-

73 erage weight gain over the first year is compara-
ble, it takes longer for the 120 kg man to reach
72
half of the maximum weight change compared
71 with the 70 kg man, and both men require sev-
eral years to begin to stabilize at a new weight
70 plateau. Mechanistically, these different aver-
0 2 4 6 8 age weight change predictions result from the
Time (weeks) nonlinear relationship between body fat and the
b fraction of weight change accounted for by in-
creased lean tissue. Thus, persons with higher
35 initial body fat have a greater fraction of their
Initial BW = 120 kg
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

30 weight change attributable to increased body

Weight gain (kg)

25 fat versus lean tissue. Because body fat con-

by Brown University on 06/04/12. For personal use only.

Initial BW = 70 kg
20
15
10
5 (37, 43, 44).
0 This finding of increased weight gain in
0 2 4 6 8 10
Time (years)
Figure 8
(a) Model-predicted body weight gain (solid curve) and its minimum expected
range of variability (dotted curves) in response to 500 kcal/d of overfeeding
assuming perfect adherence. (b) Greater model-predicted weight gain in
response to 500 kcal/d of overfeeding in an initially obese 120 kg man (dashed
curve) compared with an initially lean 70 kg man (solid curve). BW, body weight.
estimate of the baseline energy expenditure.
The variability of weight gain in a real overfeed-
ing study would likely be substantially greater
given that people may have variable changes in
spontaneous physical activity (62, 63) and dif-
ferent physiological responses to the diet as well
as different levels of diet adherence.
Greater Weight and Body Fat Gain
in Obesity for the Same Increment
in Energy Intake
Another source of weight change variability
results from physiological differences that are
captured by mathematical models of human
energy regulation. For example, Figure 8b
tributes less than lean tissue to RMR and over-
all energy expenditure, the person with higher
initial body fat will gain a greater amount of
weight to achieve a new state of energy balance
obese versus lean individuals for the same in-
crement of energy intake is shared by all math-
ematical models that include nonlinear body
composition changes as well as the differential
effect of lean versus fat tissue on RMR (37, 43,
44, 102). It is also a feature of models where
energy expenditure increases sublinearly with
body weight (4, 94, 95) because the same weight
change at a higher initial body weight results
in a smaller change of energy expenditure. In
other words, a sublinear increase of energy ex-
penditure with body weight indirectly repro-
duces the nonlinear body composition effect
on RMR and thereby also results in increased
predicted weight gain for the same increment
in energy intake for people with greater initial
weight.
These mathematical model predictions that
the same increment of energy intake leads to
increased weight gain in overweight and obese
people may help explain the observed increased
positive skewness of the U.S. population’s body
mass index (BMI) distribution over time (27). In
other words, if the same average increase in en-
ergy intake were added to all individuals across

15.14 Hall

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31
BMI categories, then mathematical models pre-
dict that more weight would be gained in peo-
ple with larger BMIs, thereby pushing out the
upper tail of the BMI distribution with time.
Whether this effect is sufficient to explain the
evolving shape of the BMI distribution is an
intriguing question.
CONCLUSIONS AND FUTURE
DIRECTIONS
In the physical sciences, there is a long history
of developing mathematical models that quan-
titatively describe past data and predict the re-
sults of key new experiments. Such quantitative
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org
models have been relatively rare in the biomedi-
cal sciences, possibly because biological systems
by Brown University on 06/04/12. For personal use only.
are highly complex, and defining the important
variables is often difficult. Fortunately, models
of human energy regulation and macronutri-
ent metabolism are constrained by conservation
principles and knowledge of the main metabolic
pathways that contribute to whole-body
imbalances.
A guiding principle when developing a
mathematical model is to target its complex-
ity to the class of phenomena that the model
is intended to address. Mathematical models of
human energy regulation and body weight dy-
namics have ranged in complexity from detailed
computational models that accurately simulate
dynamic changes in macronutrient metabolism
and body composition to overly simplified static
weight loss models that fail to capture the most
basic features of body weight dynamics. Most
other models fall between these extremes and
have been used to provide important insights
regarding human energy regulation and body
weight change.
Despite significant progress, much work re-
mains for improving and expanding the existing
mathematical models of human energy regu-
lation and body weight change. For example,
DISCLOSURE STATEMENT
The author is not aware of any affiliations, memberships, funding, or financial holdings that might
be perceived as affecting the objectivity of this review.
www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans
Changes may still occur before final publication online and in print
the anatomical location of body fat, especially
in visceral adipose tissue, has profound clini-
cal importance, and mathematical models have
only just begun to capture the relationship be-
tween body fat changes in various depots dur-
ing weight loss and gain (42, 45). Furthermore,
even the most detailed computational models of
macronutrient metabolism implicitly represent
the effect of hormones such as insulin, but an
explicit representation of organ systems along
with concentrations of hormones and metabo-
lites would be desirable—especially on shorter
time scales so that the response to individual
meals could be simulated (17, 73). Conversely,
capturing the dynamics of energy regulation
and body composition change during aging as
well as during childhood and adolescent growth
(9) will require extending current mathemat-
ical models to operate on much longer time
scales.
An editorial describing the future of
biomedical research remarked that “formula-
tion of a mathematical model is the ultimate
test of understanding . . . If the model repro-
duces the behavior of the system under a range
of conditions and predicts the consequences
of . . . modifications in any component, one can
be relatively confident about understanding
the system” (76). As highlighted in this re-
view, much progress has been made on inte-
grating past research on human nutrition and
metabolism into self-consistent and predictive
mathematical models of energy regulation and
body composition change. Such models can
highlight knowledge gaps, integrate metabolic
data within a broader context of knowledge, and
make testable predictions, thereby helping de-
sign new experiments. Indeed, the best mathe-
matical models will never replace experimental
research but rather will be used to help design
the key experiments that, in turn, will help im-
prove the mathematical models and our under-
standing of the overall system.
15.15
 NU32CH15-Hall ARI 13 April 2012 14:31

ACKNOWLEDGMENTS
The author thanks Caron C. Chow for his critical reading of the manuscript and the suggested
improvements. The author’s research was supported by the Intramural Research Program of the
NIH, National Institute of Diabetes and Digestive and Kidney Diseases.

LITERATURE CITED
1. Alpert SS. 1979. A two-reservoir energy model of the human body. Am. J. Clin. Nutr. 32:1710–18
2. Alpert SS. 1990. Growth, thermogenesis, and hyperphagia. Am. J. Clin. Nutr. 52:784–92
3. Alpert SS. 2005. A limit on the energy transfer rate from the human fat store in hypophagia. J. Theor.
Biol. 233:1–13
4. Antonetti VW. 1973. The equations governing weight change in human beings. Am. J. Clin. Nutr.
26:64–71
5. Austin GL, Ogden CL, Hill JO. 2011. Trends in carbohydrate, fat, and protein intakes and association
with energy intake in normal-weight, overweight, and obese individuals: 1971–2006. Am. J. Clin. Nutr.
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

93:836–43
6. Blaxter K. 1989. Energy Metabolism in Animals and Man. London: Cambridge Univ. Press
by Brown University on 06/04/12. For personal use only.

7. Blundell JE, Stubbs RJ, Hughes DA, Whybrow S, King NA. 2003. Cross talk between physical activity
and appetite control: Does physical activity stimulate appetite? Proc. Nutr. Soc. 62:651–61
8. Buchholz AC, Schoeller DA. 2004. Is a calorie a calorie? Am. J. Clin. Nutr. 79:899–906S
9. Butte NF, Christiansen E, Sorensen TI. 2007. Energy imbalance underlying the development of child-
hood obesity. Obesity (Silver Spring) 15:3056–66
10. Caudwell P, Gibbons C, Hopkins M, Naslund E, King N, et al. 2011. The influence of physical activity
on appetite control: an experimental system to understand the relationship between exercise-induced
energy expenditure and energy intake. Proc. Nutr. Soc. 70:171–80
11. Chow CC, Hall KD. 2008. The dynamics of human body weight change. PLoS Comput. Biol. 4:e1000045
12. Christiansen E, Garby L. 2002. Prediction of body weight changes caused by changes in energy balance.
Eur. J. Clin. Invest. 32:826–30
13. Christiansen E, Garby L, Sorensen TI. 2005. Quantitative analysis of the energy requirements for
development of obesity. J. Theor. Biol. 234:99–106
14. Cleland R, Graybill DC, Hubbard V, Kettel Khan L, Stern JS, et al. 1997. Commercial Weight Loss
Products and Programs: What Consumers Stand To Gain and Lose. Washington, DC: Bur. Consum. Protect.
Fed. Trade Commiss.
15. Cunningham JJ. 1991. Body composition as a determinant of energy expenditure: a synthetic review and
a proposed general prediction equation. Am. J. Clin. Nutr. 54:963–69
16. DeFronzo RA. 1981. The effect of insulin on renal sodium metabolism. A review with clinical implica-
tions. Diabetologia 21:165–71
17. de Graaf AA, Freidig AP, De Roos B, Jamshidi N, Heinemann M, et al. 2009. Nutritional systems
biology modeling: from molecular mechanisms to physiology. PLoS Comput. Biol. 5:e1000554
18. Doucet E, Imbeault P, St-Pierre S, Almeras N, Mauriege P, et al. 2003. Greater than predicted decrease
in energy expenditure during exercise after body weight loss in obese men. Clin. Sci. (Lond.) 105:89–95
19. Doucet E, St-Pierre S, Almeras N, Despres JP, Bouchard C, Tremblay A. 2001. Evidence for the existence
of adaptive thermogenesis during weight loss. Br. J. Nutr. 85:715–23
20. Duyff RL. 2006. American Dietetic Association Complete Food and Nutrition Guide. Hoboken, NJ: Wiley.
3rd ed.
21. Elia M. 1992. Organ and tissue contribution to metabolic rate. In Energy Metabolism: Tissue Determinants
and Cellular Corollaries, ed. JM Kinney, HN Tucker, pp. 61–79. New York: Raven
22. Elia M, Zed C, Neale G, Livesey G. 1987. The energy cost of triglyceride-fatty acid recycling in nonobese
subjects after an overnight fast and four days of starvation. Metabolism 36:251–55
23. Feinman RD, Fine EJ. 2003. Thermodynamics and metabolic advantage of weight loss diets. Metab.
Syndr. Relat. Disord. 1:209–19

15.16 Hall

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31

24. Fine EJ, Feinman RD. 2004. Thermodynamics of weight loss diets. Nutr. Metab. (Lond.) 1:15
25. Finkelstein EA, Zhen C, Nonnemaker J, Todd JE. 2010. Impact of targeted beverage taxes on higher-
and lower-income households. Arch. Intern. Med. 170:2028–34
26. Flatt JP. 2007. Exaggerated claim about adaptive thermogenesis. Int. J. Obes. (Lond.) 31:1626; author
reply 1627–28
27. Flegal KM, Troiano RP. 2000. Changes in the distribution of body mass index of adults and children in
the US population. Int. J. Obes. Relat. Metab. Disord. 24:807–18
28. Forbes GB. 1987. Lean body mass-body fat interrelationships in humans. Nutr. Rev. 45:225–31
29. Forbes GB. 2000. Body fat content influences the body composition response to nutrition and exercise.
Ann. N. Y. Acad. Sci. 904:359–65
30. Friedl KE, Moore RJ, Martinez-Lopez LE, Vogel JA, Askew EW, et al. 1994. Lower limit of body fat
in healthy active men. J. Appl. Physiol. 77:933–40
31. Gallagher D, Belmonte D, Deurenberg P, Wang Z, Krasnow N, et al. 1998. Organ-tissue mass mea-
surement allows modeling of REE and metabolically active tissue mass. Am. J. Physiol. 275:E249–58
32. Gustavsen GW, Rickertsen K. 2011. The effects of taxes on purchases of sugar-sweetened carbonated
soft drinks: a quantile regression approach. Appl. Econ. 43:707–16
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

33. Hall KD. 2006. Computational model of in vivo human energy metabolism during semistarvation and
refeeding. Am. J. Physiol. Endocrinol. Metab. 291:E23–37
by Brown University on 06/04/12. For personal use only.

34. Hall KD. 2007. Body fat and fat-free mass inter-relationships: Forbes’s theory revisited. Br. J. Nutr.
97:1059–63
35. Hall KD. 2008. What is the required energy deficit per unit weight loss? Int. J. Obes. (Lond.) 32:573–76
36. Hall KD. 2010. Mathematical modelling of energy expenditure during tissue deposition. Br. J. Nutr.
104:4–7
37. Hall KD. 2010. Mechanisms of metabolic fuel selection: modeling human metabolism and body-weight
change. IEEE Eng. Med. Biol. Mag. 29:36–41
38. Hall KD. 2010. Predicting metabolic adaptation, body weight change, and energy intake in humans.
Am. J. Physiol. Endocrinol. Metab. 298:E449–66
39. Hall KD, Bain HL, Chow CC. 2007. How adaptations of substrate utilization regulate body composition.
Int. J. Obes. (Lond.) 31:1378–83
40. Hall KD, Chow CC. 2011. Estimating changes in free-living energy intake and its confidence interval.
Am. J. Clin. Nutr. 94:66–74
41. Hall KD, Guo J, Dore M, Chow CC. 2009. The progressive increase of food waste in America and its
environmental impact. PLoS One 4:e7940
42. Hall KD, Hallgreen CE. 2008. Increasing weight loss attenuates the preferential loss of visceral compared
with subcutaneous fat: a predicted result of an allometric model. Int. J. Obes. (Lond.) 32:722
43. Hall KD, Jordan PN. 2008. Modeling weight-loss maintenance to help prevent body weight regain.
Am. J. Clin. Nutr. 88:1495–503
44. Hall KD, Sacks G, Chandramohan D, Chow CC, Wang YC, et al. 2011. Quantification of the effect of
energy imbalance on bodyweight. Lancet 378:826–37
45. Hallgreen CE, Hall KD. 2008. Allometric relationship between changes of visceral fat and total fat mass.
Int. J. Obes. (Lond.) 32:845–52
46. Heilbronn LK, de Jonge L, Frisard MI, DeLany JP, Larson-Meyer DE, et al. 2006. Effect of 6-month
calorie restriction on biomarkers of longevity, metabolic adaptation, and oxidative stress in overweight
individuals: a randomized controlled trial. JAMA 295:1539–48
47. Heymsfield SB, Harp JB, Reitman ML, Beetsch JW, Schoeller DA, et al. 2007. Why do obese patients
not lose more weight when treated with low-calorie diets? A mechanistic perspective. Am. J. Clin. Nutr.
85:346–54
48. Heymsfield SB, Thomas D, Martin CK, Redman LM, Strauss B, et al. 2012. Energy content of weight
loss: kinetic features during voluntary caloric restriction. Metabolism. Epub ahead of print
49. Heymsfield SB, Thomas D, Nguyen AM, Peng JZ, Martin C, et al. 2011. Voluntary weight loss: sys-
tematic review of early phase body composition changes. Obes. Rev. e348–61
50. Hutchison R. 1911. Food and the Principles of Dietetics. New York: William Wood

www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans 15.17

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31

51. Jordan PN, Hall KD. 2008. Dynamic coordination of macronutrient balance during infant growth:
insights from a mathematical model. Am. J. Clin. Nutr. 87:692–703
52. Kekwick A, Pawan GL. 1956. Calorie intake in relation to body-weight changes in the obese. Lancet
271:155–61
53. King NA, Caudwell P, Hopkins M, Byrne NM, Colley R, et al. 2007. Metabolic and behavioral com-
pensatory responses to exercise interventions: barriers to weight loss. Obesity (Silver Spring) 15:1373–83
54. King NA, Hopkins M, Caudwell P, Stubbs RJ, Blundell JE. 2008. Individual variability following
12 weeks of supervised exercise: identification and characterization of compensation for exercise-induced
weight loss. Int. J. Obes. (Lond.) 32:177–84
55. Kinsell LW, Gunning B, Michaels GD, Richardson J, Cox SE, Lemon C. 1964. Calories do count.
Metabolism 13:195–204
56. Kozusko FP. 2001. Body weight setpoint, metabolic adaption and human starvation. Bull. Math. Biol.
63:393–403
57. Kozusko FP. 2002. The effects of body composition on setpoint based weight loss. Math. Comp. Model.
29:1–7
58. Lecoultre V, Ravussin E, Redman LM. 2011. The fall in leptin concentration is a major determinant of
the metabolic adaptation induced by caloric restriction independently of the changes in leptin circadian
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

rhythms. J. Clin. Endocrinol. Metab. 96:E1512–16

59. Leibel RL, Hirsch J, Appel BE, Checani GC. 1992. Energy intake required to maintain body weight is
by Brown University on 06/04/12. For personal use only.

not affected by wide variation in diet composition. Am. J. Clin. Nutr. 55:350–55
60. Leibel RL, Rosenbaum M, Hirsch J. 1995. Changes in energy expenditure resulting from altered body
weight. N. Engl. J. Med. 332:621–28
61. Leiter LA, Marliss EB. 1982. Survival during fasting may depend on fat as well as protein stores. JAMA
248:2306–7
62. Levine JA, Eberhardt NL, Jensen MD. 1999. Role of nonexercise activity thermogenesis in resistance
to fat gain in humans. Science 283:212–14
63. Levine JA, Lanningham-Foster LM, McCrady SK, Krizan AC, Olson LR, et al. 2005. Interindividual
variation in posture allocation: possible role in human obesity. Science 307:584–86
64. Livesey G, Elia M. 1988. Estimation of energy expenditure, net carbohydrate utilization, and net fat
oxidation and synthesis by indirect calorimetry: evaluation of errors with special reference to the detailed
composition of fuels. Am. J. Clin. Nutr. 47:608–28
65. Major GC, Doucet E, Trayhurn P, Astrup A, Tremblay A. 2007. Clinical significance of adaptive ther-
mogenesis. Int. J. Obes. (Lond.) 31:204–12
66. Mickens RE, Brewley DN, Russell ML. 1998. A model of dieting. SIAM Rev. 40:667–72
67. Muller MJ, Bosy-Westphal A, Kutzner D, Heller M. 2002. Metabolically active components of fat-free
mass and resting energy expenditure in humans: recent lessons from imaging technologies. Obes. Rev.
3:113–22
68. Muller MJ, Bosy-Westphal A, Later W, Haas V, Heller M. 2009. Functional body composition: insights
into the regulation of energy metabolism and some clinical applications. Eur. J. Clin. Nutr. 63:1045–56
69. Natl. Heart Lung Blood Inst. 2005. Aim for a Healthy Weight. Rep. 05–5213. Bethesda, MD: Natl. Inst.
Health, Natl. Heart Lung Blood Inst.
70. Natl. Heart Lung Blood Inst. Obesity Educ. Initiat. Expert Panel Identific. Eval. Treat. Overweight
Obes. Adults. 2000. The Practical Guide: Identification, Evaluation, and Treatment of Overweight and Obesity
in Adults. Bethesda, MD: Natl. Inst. Health, Natl. Heart Lung Blood Inst.
71. Natl. Health Serv. 2006. Your Weight, Your Health: How to Take Control of Your Weight. Rep. 274537.
London: Natl. Health Serv., Dept. Health
72. Novotny JA, Rumpler WV. 1998. Modeling of energy expenditure and resting metabolic rate during
weight loss in humans. Adv. Exp. Med. Biol. 445:293–302
73. Pattaranit R, van den Berg HA. 2008. Mathematical models of energy homeostasis. J. R. Soc. Interface
5:1119–35
74. Payne PR, Dugdale AE. 1977. Mechanisms for the control of body-weight. Lancet 1:583–86
75. Payne PR, Dugdale AE. 1977. A model for the prediction of energy balance and body weight. Ann. Hum.
Biol. 4:525–35

15.18 Hall

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31

76. Pollard TD. 2002. The future of biomedical research: from the inventory of genes to understanding
physiology and the molecular basis of disease. JAMA 287:1725–27
77. Rabast U, Schonborn J, Kasper H. 1979. Dietetic treatment of obesity with low- and high-carbohydrate
diets: comparative studies and clinical results. Int. J. Obes. 3:201–11
78. Rabast U, Vornberger KH, Ehl M. 1981. Loss of weight, sodium and water in obese persons consuming
a high- or low-carbohydrate diet. Ann. Nutr. Metab. 25:341–49
79. Ravussin E, Schutz Y, Acheson KJ, Dusmet M, Bourquin L, Jequier E. 1985. Short-term, mixed-diet
overfeeding in man: no evidence for “luxuskonsumption.” Am. J. Physiol. 249:E470–77
80. Redman LM, Heilbronn LK, Martin CK, de Jonge L, Williamson DA, et al. 2009. Metabolic and
behavioral compensations in response to caloric restriction: implications for the maintenance of weight
loss. PLoS ONE 4:e4377
81. Rosenbaum M, Goldsmith R, Bloomfield D, Magnano A, Weimer L, et al. 2005. Low-dose leptin reverses
skeletal muscle, autonomic, and neuroendocrine adaptations to maintenance of reduced weight. J. Clin.
Invest. 115:3579–86
82. Rosenbaum M, Hirsch J, Gallagher DA, Leibel RL. 2008. Long-term persistence of adaptive thermo-
genesis in subjects who have maintained a reduced body weight. Am. J. Clin. Nutr. 88:906–12
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

83. Rosenbaum M, Leibel RL. 2010. Adaptive thermogenesis in humans. Int. J. Obes. (Lond.) 34(Suppl.
1):S47–55
by Brown University on 06/04/12. For personal use only.

84. Rosenbaum M, Murphy EM, Heymsfield SB, Matthews DE, Leibel RL. 2002. Low-dose leptin admin-
istration reverses effects of sustained weight-reduction on energy expenditure and circulating concen-
trations of thyroid hormones. J. Clin. Endocrinol. Metab. 87:2391–94
85. Rosenbaum M, Nicolson M, Hirsch J, Murphy E, Chu F, Leibel RL. 1997. Effects of weight change on
plasma leptin concentrations and energy expenditure. J. Clin. Endocrinol. Metab. 82:3647–54
86. Schoeller DA, Buchholz AC. 2005. Energetics of obesity and weight control: Does diet composition
matter? J. Am. Diet. Assoc. 105:S24–28
87. Schoeller DA, Fjeld CR. 1991. Human energy metabolism: What have we learned from the doubly
labeled water method? Annu. Rev. Nutr. 11:355–73
88. Simon P, Jarosz CJ, Kuo T, Fielding JE. 2008. Menu Labeling as a Potential Strategy for Combating the
Obesity Epidemic: A Health Impact Assessment. Los Angeles: County L.A. Dept. Public Health
89. Smith TA, Bing-Hwan L, Jong-Ying L. 2010. Taxing Caloric Sweetened Beverages: Potential Effects on
Beverage Consumption, Calorie Intake, and Obesity. Washington, DC: Econ. Res. Serv., U.S. Dept. Agric.
90. Svetkey LP, Stevens VJ, Brantley PJ, Appel LJ, Hollis JF, et al. 2008. Comparison of strategies for
sustaining weight loss: the Weight Loss Maintenance randomized controlled trial. JAMA 299:1139–48
91. Swinburn BA, Sacks G, Hall KD, McPherson K, Finegood DT, et al. 2011. The global obesity pandemic:
shaped by global drivers and local environments. Lancet 378:804–14
92. Taubes G. 2011. Why We Get Fat and What to Do About It. New York: Knopf
93. Thomas D, Das SK, Levine JA, Martin CK, Mayer L, et al. 2010. New fat free mass–fat mass model for
use in physiological energy balance equations. Nutr. Metab. (Lond.) 7:39
94. Thomas DM, Ciesla A, Levine JA, Stevens JG, Martin CK. 2009. A mathematical model of weight
change with adaptation. Math. Biosci. Eng. 6:873–87
95. Thomas DM, Martin CK, Heymsfield S, Redman LM, Schoeller DA, Levine JA. 2011. A simple model
predicting individual weight change in humans. J. Biol. Dyn. 5:579–99
96. Thomas DM, Schoeller DA, Redman LA, Martin CK, Levine JA, Heymsfield SB. 2010. A computational
model to determine energy intake during weight loss. Am. J. Clin. Nutr. 92:1326–31
97. Trabulsi J, Schoeller DA. 2001. Evaluation of dietary assessment instruments against doubly labeled
water, a biomarker of habitual energy intake. Am. J. Physiol. Endocrinol. Metab. 281:E891–99
98. van der Walt WH, Wyndham CH. 1973. An equation for prediction of energy expenditure of walking
and running. J. Appl. Physiol. 34:559–63
99. van Stratum P, Lussenburg RN, van Wezel LA, Vergroesen AJ, Cremer HD. 1978. The effect of dietary
carbohydrate:fat ratio on energy intake by adult women. Am. J. Clin. Nutr. 31:206–12
100. Vazquez JA, Adibi SA. 1992. Protein sparing during treatment of obesity: ketogenic versus nonketogenic
very low calorie diet. Metabolism 41:406–14

www.annualreviews.org • Modeling Metabolic Adaptations and Energy Regulation in Humans 15.19

Changes may still occur before final publication online and in print
 NU32CH15-Hall ARI 13 April 2012 14:31

101. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. 2011. Health and economic burden of
the projected obesity trends in the USA and the UK. Lancet 378:815–25
102. Weinsier RL, Bracco D, Schutz Y. 1993. Predicted effects of small decreases in energy expenditure on
weight gain in adult women. Int. J. Obes. Relat. Metab. Disord. 17:693–700
103. Weinsier RL, Hunter GR, Zuckerman PA, Redden DT, Darnell BE, et al. 2000. Energy expenditure
and free-living physical activity in black and white women: comparison before and after weight loss.
Am. J. Clin. Nutr. 71:1138–46
104. Weinsier RL, Nagy TR, Hunter GR, Darnell BE, Hensrud DD, Weiss HL. 2000. Do adaptive changes
in metabolic rate favor weight regain in weight-reduced individuals? An examination of the set-point
theory. Am. J. Clin. Nutr. 72:1088–94
105. Werner SC. 1955. Comparison between weight reduction on a high-calorie, high-fat diet and on an
isocaloric regimen high in carbohydrate. N. Engl. J. Med. 252:661–65
106. Westerterp KR. 2004. Diet induced thermogenesis. Nutr. Metab. (Lond.) 1:5
107. Westerterp KR, Donkers JH, Fredrix EW, Boekhoudt P. 1995. Energy intake, physical activity and body
weight: a simulation model. Br. J. Nutr. 73:337–47
108. Weyer C, Snitker S, Rising R, Bogardus C, Ravussin E. 1999. Determinants of energy expenditure and
Annu. Rev. Nutr. 2012.32. Downloaded from www.annualreviews.org

fuel utilization in man: effects of body composition, age, sex, ethnicity and glucose tolerance in 916
subjects. Int. J. Obes. Relat. Metab. Disord. 23:715–22
by Brown University on 06/04/12. For personal use only.

109. Winkler JT. 2005. The fundamental flaw in obesity research. Obes. Rev. 6:199–202
110. Wishnofsky M. 1958. Caloric equivalents of gained or lost weight. Am. J. Clin. Nutr. 6:542–46
111. Yang MU, Barbosa-Saldivar JL, Pi-Sunyer FX, Van Itallie TB. 1981. Metabolic effects of substituting
carbohydrate for protein in a low-calorie diet: a prolonged study in obese patients. Int. J. Obes. 5:231–36
112. Yang MU, Van Itallie TB. 1976. Composition of weight lost during short-term weight reduction.
Metabolic responses of obese subjects to starvation and low-calorie ketogenic and nonketogenic diets.
J. Clin. Invest. 58:722–30

15.20 Hall

Changes may still occur before final publication online and in print