Do obese individuals gain weight than nonobese individuals? Gilbert
B Forbes
ABSTRACT who
A review
were
overfed
under
mass,
body
fatter weight
people than
fat,
fat,
more
percent to
do those
lean,
body more
to
who
are
thin.
tend tissue,
which
on human
to put whereas
bigger
a given
cost
of weight
gain
should
and of
controlled
A likely
explanation
is
assays,
on a large
proportion
of gain
by each investigator, 4) a calculated 12 kcal/g (the theoretical value
Nutr
of
people
tend
Am
tissue.
to
J Clin
be greater
mdi-
in obese
To test this prediction, a review was made ofstudies of overfeeding in adolescent and adult subjects that met the following criteria: 1) 2 wk ofdefined overfeeding with the subjects in a
amount
thin
is a low-energy
the energy viduals.
rang-
cost of induced and oflean body
Therefore,
gain
subjects
for periods
energy weight
fat.
eat
individuals
a high-energy
done conditions
shows that the ofinitial body
and
need
obese
of studies
controlled
ing from 2 to 12 wk weight gain is a function
that
more easily
environment,
3) a range
2) performance
ofinitial
body
of body
fat content
composition
for subjects
studied
energy cost ofweight for fat), and 5) a weight
gain
gain
1 kg.
1990;52:224-7.
Methods KEY WORDS
Weight
gain,
energy
need,
body
fat, obesity, Data of body
thinness
together
Introduction In discussing fronted
with
matters
with
the
ofweight
opinion
remarkable
ease,
uals
will gain data gain
weight
thin
there
at least
overweight
food;
one
the corn-
show that excess energy
interindividual
controlled of
more
answer
can
overweight
to the answer
with
when
be found
over-
in an analwho
together
with
anorexia
nervosa
is to be found
tend
to put
Because
the energy
about
six times
greater
kcal/g
protein)
X (0.205
mass (LBM)] (8, individuals would
individuals
and
rest
were
were
studies
of
under
to gain
on
relatively
cost
than
in the composition
individuals that thin whereas
the
large
of gaining cost
g protein/g
lean)
of
of fat
1 g fat (12 kcal)
of gaining =
1 g lean 1 .78
is
[(8.68
kcal/g
lean
9), one could reasonably predict that have to eat more than would nonobese
a given
amount
of weight.
In other
words,
Am J C/in Nutr
1990;52:224-7.
Downloaded from https://academic.oup.com/ajcn/article-abstract/52/2/224/4651386 by Washington University in St. Louis user on 22 May 2018
type
of diet
that
provided
period.
The
subjects
of Bandini
young
and
et al (12)
middle-aged
the
were
with
extra
adolescents;
adults.
At the
energy
anorexia
ner-
all of the time
of study
the condition of the patients with anorexia nervosa had been stabilized: all had normal values for blood hemoglobin, serum proteins, and electrolytes, and diet was carefully monitored on the metabolic ward. As noted in Table 1 the excess food consisted ofcarbohydrate for 2 1 subjects, offat for 4 subjects, and ofmixed diet for 38 subjects. composition
status
was
estimated
in 24 subjects, by densitometry dilution in 12 subjects. The period
ranged
eral studies. In each reported during overfeeding author was taken ence
when they are individuals gain those with larger amounts
those
Body
conditions.
fat stores
vosa
counting oxygen-l8
vari-
individuals
conditions, patients
readily
the
,
individ-
indeed
experiments to the total
weight
The and
(5-7).
224
and
fed excess
to gain
repletion
A clue
body obese
to have
weight
yet doesn’t gain weight. (1) and Passmore and
thin
tissue gained by thin and overweight overfed. A recent compilation shows relatively large amounts oflean tissue body
gain
.
the overfeeding
feeding tends of normal
under
controlled
seems
is considerable
or the obese?
ysis of studies nutrition
both
when
although
ability (4). Which group
overfed
that
one is apt to be con-
individuals
from several controlled is directly proportional
consumed,
fed-the
obese
to eat to excess and Newburgh
(2, 3) showed
bined weight
control,
that
and everyone
thin friend who claims Many years ago Wiley colleagues
with
during
are listed in Table 1 Included are the ranges LBM, and body fat for each group of subjects
sources weight,
was
from
14 to 83 d (i25
series the difference and the maintenance as the excess energy
subtracted
for
by
fecal
potassium-40
in 27 subjects, and by duration of the overd) among
the
sev-
between energy intake intake as stated by the intake; 5% of the differ-
losses.
However,
a different
method was used for the nine subjects with anorexia nervosa. Once the patients had been stabilized, basal metabolic rate (BMR) was determined and a potassium-40 assay was done.
I From the Departments of Pediatrics and Biophysics, University Rochester School ofMedicine and Dentistry, Rochester, NY. 2 Supported by NIH grants 5 K06 HD 18454 and RR 00044. 3
Address
Box 777,
reprint requests to GB Forbes, Department University of Rochester Medical Center,
of PediatricsRochester, NY
14642.
ReceivedJune 19, 1989. Accepted for publication Printed
in USA.
January
31, 1990.
© 1990 American
Society
of
for Clinical
Nutrition
ENERGY
TABLE Subject
OF
WEIGHT
225
GAIN
1 characteristics,
Number,
diet
sex
weight
Body
Body fat
LBM
kg
kg
kg 13 M
The number
ofsubjects
Type
ofexcess
18
7-33
50-85
Fat [4], carbohydrate
4-54 4-72
49-83 35-76
Mixed diet [8] Carbohydrate(12J
30-93
2-25
28-77
Mixed diet [24]
46- 107
6-43
34-76
Mixed
for each type given
subjects excluded. subject excluded. § One male and eight females with anorexia
Reference
food*
54- 1 15 44-142
59-I
6 M, 2 Ft 8M,4Fj 3 M, 2 1 F 4 M. 2 F II *
COST
[9]
Goldman
et al (10)
Katzeffet al ( 1 1) Bandinietal(l2) Forbes et al (4, 7) Webb and Annis (13)
diet [6]
in brackets.
t Four additional :1:One additional
II Five additional
subjects
nervosa,
all others
normal.
excluded.
vealed
These assays were made twice weekly during nutrition recovery, and excess energy intake was calculated as the difference between actual intake and 1 .5 X BMR; due account was taken ofthe gradual rise in BMR that occurred. Fecal losses were assumed to be 5% ofexcess energy intake. The total excess energy consumed during the overfeeding period ranged from I 6 000 to 80 000 kcal (1 28 500 kcal) for the various studies, and the gain in weight ranged from 1.8 to 14.7 kg. The energy cost of weight gain is determined by dividing the former by the latter and is expressed as kcal/g gain for each subject. Statistical analyses were done on the C/info Data Analysis System (BBN Software Products Corp, Cambridge, MA).
it
Figure
1 is a plot ofexcess
(kcal/g)
against
initial
energy body
consumed
weight
unit
per
rithm
Indeed,
A further sion
the heaviest
as much slope
food
people
as the lightest
analysis
ofthe
for males
does
data
need
to gain
is shown
not differ
over, a comparison of those sisted solely of carbohydrate
on average ones from
that
kcal/g
for
were gain
slopes
are
carbohydrate
done
diet/
on all subjects,
against
initial
LBM,
along
the x-axis
is symmetrical
whereas
those
gain
is greater
for
fatter
people
analysis
than
revealed
for those
the following
who
are
(fat and
LBM in kg): kcal/g
=
5.427
+ 0.0630
fat (±0.017)
+ 0.0126
LBM
(±0.0
18)
(1)
1.7
‘
weight. 2. The regres-
for females.
assays
regress
thin. Multiple-regression
1 g body
in Table
of values
weight
The
to eat
to
regression
except
for weight and fat are skewed. Table 2 shows that energy cost is related to both ofthese body components and to relative fat content, the relationship being somewhat stronger for fat than for LBM. The effect of relative body fatness is shown in Figure 2. The slope of the regression line is significantly different from zero (p = 0.0001). Although there is some scatter of points about the regression line, the trend is evident: the energy cost of
of the regression line is significantly different from zero (p = 0.0002). Despite the rather wide scatter of points about the regression line, the trend is obvious: the energy cost of weight gain is greater for heavy people than for those who are lighter.
zero
body fat, against percent fat, and against the logafat. This last index was chosen because the distri-
ofbody
bution
slope
times
initial
from
composition
body
possible
against
weight
for all 63 subjects.
was
(All of the various
different 0.057.)
=
Because
Results
gain
no difference.
significantly weight, p
The correlation coefficient is 0.490. It is apparent is a more important determinant of the energy gain than is LBM.
More-
subjects whose excess energy conwith those given mixed diets re-
that body fat cost of weight
Discussion Cu
0
0
10
The prediction
#{149}
energy
0
E
who S
Cu (2
0
S
Males
0
Females
40
‘
o0
‘
80
‘
ic;o
Initial Weight, FIG 1. Plot ofenergy weight (kg). Females(O), + 0.039x(±0.OlO), r
=
‘
120
‘
140
cost ofweight gain (kcal/g) against initial males(#{149}).Calculated regression line: v 0.45.
‘
be answered
=
body 4.46
Downloaded from https://academic.oup.com/ajcn/article-abstract/52/2/224/4651386 by Washington University in St. Louis user on 22 May 2018
be greater
ofbody
namely in those
fat than
that
the
individuals
in those
who are
.
.
.
.
.
in the negative. earlier, the reason for the higher energy cost and fatter people is that weight gains induced by overfeeding are composed of a relatively large proportion of fat, whereas thin people tend to gain more lean (5). Two examples from that report, in which the relative contributions of lean As mentioned for overweight
kg
in the introduction, would
have a large component
.
y=4.46+0.039x
,
gain
viduals actually gain weight more easily than do obese individuals. The question posed in the title ofthis paper can therefore
Cl)
Cl) 4) C.)
set forth
ofweight
lean, is confirmed by the data presented. When subjects are studied under controlled conditions, it turns out that thin mdi-
#{149}
S
.
cost
226
FORBES
TABLE
2
Regression
equations
for energy
cost ofweight
gain (kcal/g)
vs weight,
LBM,
Allsubjects(n Weight, kg LBM, kg Fat,kg log fat, kg Percent fat Females(n
fat, log fat, and percent =
fat
63)
4.46 + 0.039 (±0.OlO)* 5.05 + 0.040 (±0.0 18) 6.01 +0.068(±0.016) 4.25 + 2.66 (±0.6 1 7) 5.42 + 0.080 (±0.019)
r = 0.45 r = 0.27 r=0.48 r = 0.48 r = 0.46
< 0.01 p < 0.05 p