Dulloo AG, Geissler CA, Horton T, Collins A, Miller DS. Normal caffeine consumption: influence on thermogenesis and daily energy expenditure in lean and postobese human volunteers. Am J Clin Nutr 49, 44-50

Abstract

Single-dose oral administration of 100 mg caffeine increased the resting metabolic rate of both lean and postobese human volunteers by 3-4% (p less than 0.02) over 150 min and improved the defective diet-induced thermogenesis observed in the postobese subjects. Measurements of energy expenditure (EE) in a room respirometer indicate that repeated caffeine administration (100 mg) at 2-h intervals over a 12-h day period increased the EE of both subject groups by 8-11% (p less than 0.01) during that period but had no influence on the subsequent 12-h night EE. The net effect was a significant increase (p less than 0.02) in daily EE of 150 kcal in the lean volunteers and 79 kcal in the postobese subjects. Caffeine at commonly consumed doses can have a significant influence on energy balance and may promote thermogenesis in the treatment of obesity.

Full text

Normal caffeine consumption: influence on thermogenesis
44 Am i C/in Nuir l989;49:44-S0. Printed in USA. C 1989 American Society for Clinical Nutrition
and daily energy expenditure in lean and postobese
human volunteers14
AG Duioo, CA Geissler, THorton, A Collins, and DSMiller
ABSTRACT Single-dose oral administration of 100 mg caffeine increased the resting met-
abolic rate of both lean and postobese human volunteers by 3-4% (p < 0.02) over 150 mm
and improved the defective diet-induced thermogenesis observed in the postobese subjects.
Measurements of energy expenditure (EE) in a room respirometer indicate that repeated
caffeine administration (100 mg) at 2-h intervals over a 12-h day period increased the EE of
both subject groups by 8-1 1% (p < 0.01) during that period but had no influence on the subse-
quent 12-h night EE. The net effect was a significant increase (p < 0.02) in daily EE of 150
kcal in the lean volunteers and 79 kcal in the postobese subjects. Caffeine at commonly con-
sumed doses can have a significant influence on energy balance and may promote thermogene-
sis in the treatment ofobesity. Am JC/in Nutr 1989;49:44-50.
KEY WORDS Obesity, thermogenesis, caffeine, energy expenditure
Introduction
The widespread use ofcaffeine in drinks, food, and nu-
merous pharmaceutical preparations, such as muscle re-
laxants, decongestants, and allergy drugs, has generated
much interest in elucidating the multitude ofeffects and
mechanisms ofaction ofthis drug ofeveryday life. With
increasing evidence pointing to a thermogenic defect as
being contributory to the etiology of obesity (1), nutri-
tionists are particularly interested in caffeine’s effects on
energy expenditure (EE), not only as an apparently safe
thermogenic drug but also as a pharmacological tool to
elucidate the mechanisms of thermogenesis and meta-
bolic differences between lean and obese people.
Although the stimulatory effect of caffeine on meta-
bolic rate in man is well established and was demon-
strated both in subjects who fasted (2-7) and in those
who did not (5, 6), most of these studies focused on
caffeine’s acute thermogenic effects when administered
at relatively large doses. There is little information about
caffeine’s influence on daily EE and the thermogenic
effects of caffeine in amounts that are generally con-
sumed at any one time (as in a cup ofcoffee or in prepara-
tions usually containing 100 mg caffeine).
We conducted studies in lean and postobese human
volunteers that examined the effect of commonly con-
sumed doses of caffeine on the resting metabolic rate
(RMR), diet-induced thermogenesis, and 24-h energy
expenditure.
Methods
Subjects
Eighteen healthy volunteers were selected from students and
staff of King’s College, London University and were allocated
to two groups by the ease with which they maintained a rela-
tively lean body weight. One group (lean group; n=9, six fe-
males, 3 males) consisted oflean subjects who claimed to main-
thin body weight without effort. The other group (postobese
group; n=9, six females, three males) comprised subjects who
admitted to a weight problem and were previously overweight
with grade I (mild to moderate) obesity; their Quetelet index
(wt/ht2) ranged from 26.1 to 29.6 with a mean value of 27.3
±0.5 SEM. The various grades ofobesity based on the Quetelet
Index (or body mass index) were described by Garrow (8).
These obese subjects can only maintain a normal body weight
by restricting their food intake, otherwise they would become
overweight again within a few months. Although they are pre-
disposed to obesity, they had maintained a normal body weight
for at least 5-6 mo before the study began.
IFrom the Department of Nutrition, King’s College, University of
London, UK, and the Department of Medicine, Harvard Medical
School, Boston, MA.
2DS Miller is deceased.
3Supported in part by a grant from the International Foundation
for the Promotion ofNutrition Research, ISFE, Zurich (AGD); a grant
from King’s College (MC); and a grant from the Ashdown Trust (TH).
4Address reprint requests to AG Dulloo, Centre Medicale Universi-
taire, Department ofPhysiology, University ofGeneva, 9 av de Cham-
pel, CH-l21 1 Geneva 4, Switzerland.
ReceivedJuly9, 1987.
Accepted for publication January 5, 1988.
by guest on July 10, 2011www.ajcn.orgDownloaded from

CAFFEINE AND ENERGY EXPENDITURE 45
At the outset ofthe study, body weight and height were mea-
sured in both subject groups and the degree of obesity (if any)
was reevaluated by using the Queletet index and by estimating
the percentage body fat. Percentage body fat was assessed by
the method ofDurnin and Womersley (9) from measurements
ofskinfold thicknesses with a Harpenden skinfold caliper (Hot-
tam Ltd, Dyfed, Wales, UK). Lean body mass was calculated
by subtracting body fat from body weight. Food was weighed
and intake was recorded for 7 d immediately before the study;
energy content was expressed as daily means. Analysis of nutri-
ent composition showed no difference between the two groups
in the proportion of metabolizable energy intake (1 ±SEM)
derived from protein (lean subjects, 17 ±1%; postobese sub-
jects, 16 ±1%), fat (lean subjects, 32 ± 3%; postobese subjects,
34 ± 2%), and carbohydrate (lean subjects, 5 1 ± 2%; postobese
subjects, 50 ± 2%). Methylxanthine intake (eg, caffeine, the-
ophylline) from beverages (coffee, tea, hot chocolate, and coke)
ranged from 250 to 500 mg/d. All subjects were thus classified
as mild to moderate consumers of methylxanthines. None of
the subjects had a familial history ofdiabetes and none engaged
in physical training, regular exercise, or sport activities.
The study was carried out in accordance with the regulations
of the Ethical Committee for Human Experimentation of
King’s College, University of London.
Acute metabolic-rate studies
The subjects’ metabolic rate while fasting and not fasting was
measured by open-circuit indirect calorimetry using 100 L ca-
pacity Douglas bags and mouthpiece connections with respira-
tory valves. Details ofthese measurements were described pre-
viously (10) and involved measurements of oxygen consump-
tion rates and calculations of heat production using the Weir
formula(l 1).
All subjects were given four randomized treatments on four
different days and with at least a 2.4 interval between treat-
ments. The four treatments were 1) a 100-mg tablet of anhy-
drous caffeine (Pro-plus, Boots Ltd. Nottingham, UK); 2) a
300-kcal liquid meal(Complan#{174}, Glaxo, Devon, UK) made up
to 200 mL with water; 3) a 300-kcai liquid meal plus a 100-mg
tablet ofcaffeine; and 4) a 200-mL glass ofH2O. The complan
powder contained (per/kg) 180 g protein, 470 g carbohydrate,
and 330 g fat with an energy value of4.4 kcal/g dry wt.
The subjects arrived at the university department at 0800
after an overnight 12-h fast. They had traveled either by auto-
mobile or by public transportation. They were requested to
walk casually, to use elevators rather than stairs, and to avoid
any burst ofphysical activity on the way to the laboratory. On
arrival a subject was seated in a comfortable armchair and
spent the duration ofthis acute study either reading or relaxing.
The pretreatment phase consisted of at least 30 mm of relax-
ation followed by three or four measurements of base-line
RMR, each lasting 5 mm with a 5-10 mm interval between
measurements. After each treatment was administered, mea-
surements of metabolic rate were performed over the next 150
mm, each lasting 5mm with intervals of 10-15 mm between
measurements.
To ensure steady-state breathing the subjects breathed for a
few minutes through the mouth piece and tubing system with
the tap for expired air closed to the Douglas bag but open to
ambient air. The tap to the bag was then opened and expired
air was collected for the next 5 mm. All subjects were familiar
with this technique for measuring metabolic rate and felt no
discomfort during the study.
Daily energy expenditure
Five lean (three females, two males) and six postobese (three
females, three males) subjects participated in this study. EE was
measured for 24 h in a human respirometer ( 10) on two sepa-
rate occasions for each subject. One measurement determined
base-line (control) EE in the absence of any methylxanthine
and the other measurement determined EE during the admin-
istration of caffeine. At least a l-wk interval was allowed be-
tween these 2 d of measurements, which were carried out in a
randomized order for all subjects. All subjects were requested
to ensure that their activity and food consumption patterns
were normal (with no unusual excesses in either direction) for
at least 2 d before the measurements were taken.
On each measurement day the subject reported to the labo-
ratory between 0800 and 0830 after an overnight fast. After at
least 30 mm of rest and relaxation in a comfortable armchair,
the subject entered the respirometer at 0900 and left at 0900
the next day. While inside the respirometer, the subject spent
time reading, studying, lying in bed, listening to the radio, or
watching television. The subject kept an activity diary (dura-
tion to the nearest minute) describing position and activity
throughout the 24 h. No methylxanthine-containing food or
beverages were provided on either day, but subjects were pro-
vided with meals and drinks with an energy content similar to
their mean customary energy intakes. On the treatment day
each subject ingested six caffeine tablets (100 mg/tablet), one
tablet every 2 h for 12 h as follows: 0900 with breakfast, 1 100,
1300 with lunch, 1500, 1700, and 1900 with dinner.
Statistics
Data were analyzed by Student’s ttest and by two-way analy-
sis ofvariance with repeated measures. Posthoc comparison be-
tween pairs of treatments were performed with the Newman-
Keul’s multiple sample comparison test after analysis of vari-
ance had established significant differences between treatments
(12). All results are presented as mean ±SEM.
Results
The physical characteristics of the subjects are out-
lined in Table 1 .Height was similar in both groups but
the mean value for body weight was ‘--9% higher in post-
obese subjects than in lean subjects; this difference was
not statistically significant. The mean value for age was
also slightly higher (NS) in the postobese group (range
20-46 y) than in the lean group (range 1 8-35 y). The
Quetelet Index, ranging from 18.5 to 23.4 in the lean
group and from 19.5 to 24.9 in the postobese group, mdi-
cates that none ofthe subjects could be classified as obese
at the start ofthe study. The percentage body fat was sim-
ilar in both groups whereas lean body mass was 8%
higher (NS) in the postobese group. Food intake, mea-
sured for 1 wk, was significantly lower in the postobese
group than in the lean group by ‘-.-25% in absolute terms
(p < 0.0 1) and by 30% per kilogram body weight (p
<0.001). The postobese group maintained their weight
on a mean energy intake of 1600 kcal, which was 500
kcal less than that ofthe lean group.
Acute metabolic-rate studies
Table 1 shows the pretreatment RMR computed over
the measurement days after an overnight 12-h fast. The
by guest on July 10, 2011www.ajcn.orgDownloaded from

46 DULLOOETAL
TABLE 1
Physical characteristics, customary food intake, and pretreatment resting metabolic rate (RMR) ofsubject groups
Age Height Body
weight Body fat Quetelet
index
Lean
body
mass
Daily
food
intake
Food
intake by
weight Pretreatment
RMR
Daily
pretreatment
RMR
ycm kg % kg/cm kg kcal/d kcal.kg’ .d kcal/min kca1.kg .d
Leansubjects 24.8± 1.6 169±3 58.3±3.1 22.1 ±2.6 20.6±0.5 45.8±3.8 2105± 117 37.5±2.4 1.046±0.069 25.8±0.96
Postobesesubjects 28.2±2.6 168±2 63.5±3.1 21.9± 1.3 22.2±0.7 49.5±3.1 1592± 91 25.6± 1.4 1.036±0.039 23.6±0.7
*I±SEM, n=9per group; six females, three males per group.
within-subject coefficient of variation (1 ±SD) during
the measurement days ranged from 1.4 ± 1 .0 to 2.7
±1.5% in the lean group and from 1.6 ± 1 .3 to 1.8
±1.4% in the postobese group. The between-day coeffi-
cient of variation (1 ± SD) was 2.4 ± 2.5% in the lean
group and 2.6 ± 2. 1% in the postobese group. The RMR
in absolute terms was not different between the two sub-
ject groups but when the data were expressed per unit
body weight to account for both between-group as well
as within-group variations in body weight, RMR was 8%
lower in the postobese group than in the lean group. This
difference nearly achieved statistical significance (p
=0.056). Similarly, the RMR per unit lean body mass
was 8% lower (p =0.06) in the postobese group than in
the lean group.
The values for metabolic rate before and after treat-
ment are shown in Table 2 and the data for the total ther-
mogenic response (both absolute and percentage in-
creases) over the 150-mm posttreatment period are pre-
sented in Table 3. The thermogenic response curves,
expressed as a percentage ofthe base-line (pretreatment)
RMR, are shown in Figure 1 for both lean and postobese
groups. The control water treatment had no effect on the
RMR of either group (Fig 1). In contrast, ingestion of
100 mg caffeine with a similar volume ofH2O increased
metabolic rate in both groups. Metabolic rates reached
peak values within 20 to 40 mm after treatment and de-
creased gradually toward base-line levels. At the end of
the study, the metabolic rate in both groups was not sig-
nificantly different from base-line values.
The thermogenic response curve followed a similar
TABLE 2
pattern in both the lean and the postobese subjects, and
the total thermogenic response integrated over the entire
150-mm study period (Table 3) was increased by 3-4%
in both subject groups. This posttreatment increase in
metabolic rate was statistically significant when com-
pared with either the corresponding pretreatment RMR
or the posttreatment RMR assessed during the con-
trol day.
Ingestion ofa 300 kcal liquid meal caused a sharp rise
in metabolic rate that reached peak levels in 30-60 mm
in both subject groups. The peak increase in metabolic
rate was significantly greater (p <0.001) in lean subjects
(+25%) than in postobese subjects (+ 15%). Both groups
maintained peak metabolic rate for about another hour,
after which the metabolic rate declined at a faster rate in
the postobese group than in the lean group. At the end
ofthe 1 50 mm, the metabolic rate ofthe lean group was
still ‘ 17% above base-line RMR (p <0.001) whereas
that ofthe postobese group was only 5% higher (NS). As
shown in Tables 2 and 3, the overall thermogenic re-
sponse ofthe postobese subjects to the 300 kcal meal was
only halfofthat measured in the lean subjects.
Administration ofa 100-mg caffeine tablet in the lean
group produced a small additional (+ 12%, NS) stimula-
tory effect on their thermogenic response to food. In the
postobese group, caffeine was more effective in augment-
ing the thermogenic effect of the meal; both the peak
metabolic rate and the total thermogenic response (inte-
grated over 150 mm) were 25-30% higher after caffeine
and a meal than after the meal alone. The 50% reduction
in diet-induced thermogenesis (DIT) observed in the
Metabolic rate before and after treatment with 100 mg caffeine a nd/or a 300-kcal liquid meal in lean and po stobese subjects*
Caffeine Meal Meal +caffeine
Before After Before After Before After
kca//min
Lean subjects
Postobese subjects 1.055 ±0.07 1 1.095 ±0.063t
I .026 ±0.043 1.059 ±0.045 I .025 ±0.064 1.236 ±0.077j
I.039 ±0.038 1.164 ±0.052t 1.055 ±0.075
I.047 ±0.034 1.292 ±0.087f
1.2 13 ±0.04Sf
*j: SEM, n=9 per group; six females, three males per group.
tp<0.02.
fp<0.00l.
§p<0.Ol.
by guest on July 10, 2011www.ajcn.orgDownloaded from

4_9 -12h)    9am..,
Q)
0
a:
C.)
0
0
30
20
0
0
0Control
.Caffeine
4)
C
4)
0.
‘C
w
4)
C
LU
p<0.02
Lean
5.0
2.5
0
0 30 60 90 20 50
NS
ITI
Lean Post-
obese
CAFFEINE AND ENERGY EXPENDITURE 47
TABLE 3
Increases in metabolic rate over 150 mm in response to 100 mg caffeine and/or a 300-kcal liquid meal in lean and postobese subjects
Absol ute increase Perce ntage increase
Meal Meal
+caffeine Posthoc +caffeine Posthoc
Caffeine [1] Meal [2] [3] comparison Caffe inc [1] Meal [2] [31 comparison
kcal/min %
Leansubjects 0.040±0.016 0.210±0.015 0.236±0.014 1 vs2,3p<0.001 4.38± 1.8 20.6±0.9 22.7± 1.1 1 vs2,3p<0.OOl
2vs3 NS 2vs3 NS
Postobese 0.032±0.009 0.125±0.018 0.165 ±0.015 1 vs2,3p<0.OOl 3.16±0.87 11.8± 1.4 15.7± 1.3 1 vs2,3p<0.OOl
subjects
Significance o fFbetween treatments
between groups
2 vs 3 p <0.05
p<0.001
p<0.001 Significance ofFbetween treatments
between groups
2 vs 3 p <0.05
p <0.001
p<0.001
*1± SEM, n=9 per groups; six females, three males per group.
postobese group was ameliorated to an extent that their
thermogenic response to food with caffeine was only 25%
below that of the lean group. Thus the subnormal DIT
ofthe postobese group was partially corrected by the ad-
ministration of caffeine. However, the thermogenic re-
sponses in both subject groups after a meal were not mea-
sured until the study ended and were thus underesti-
mated.
Daily energy expenditure
The data on 24-h EE measured in the respirometer
were divided into two 12-h periods (Fig 2, Tables 4 and
5): the 12-h day period (0- 12 h EE) during which the
caffeine tablets were administered, followed by a second
Minutes after Treatment
FIG 1.Thermogenic response of lean and postobese subjects to a
100-mg caffeine tablet administered after fasting, to a 300 kcal mixed
meal, or to a combination of caffeine and a meal. The effect of the
control water load is also shown for the lean (closed circles) and for the
postobese subjects (open circles). Vertical bars represent the SEM
n=9).
12-h night period (12-24 h EE) when no caffeine was
ingested. The mean EE (MJ/person) was 8-10% lower in
the postobese group than in the lean group (Fig 2) but
these differences were not statistically significant. How-
ever, analysis ofvariance (Table 4) shows that during the
control study and also during treatment with caffeine,
EE expressed per unit body weight was significantly
lower by 13-18% in the postobese group than in the lean
group for 0- 12-h EE, 1 2-24-h EE, and total 24-h EE.
Administration of caffeine increased the 0-12-h EE in
(8%)
p<0.05
TI
Post -
obese
FIG 2. Energy expenditure compartmented into the first 12-h d pe-
riod (0.-I 2 h) and the subsequent 12-h night period (12-24 h) in lean
(n =5) and postobese (n =6) subjects during a control study (open
bars) and during administration ofcaffeine. Vertical bars represent the
SEM values. The probability level for significant differences is for
paired data. MJ values can be converted to kcal by multiplying them
by 239.
by guest on July 10, 2011www.ajcn.orgDownloaded from

48 DULLOO ET AL
TABLE 4
Energy expenditure (kcal/kg) in respirometer during base-line (control) study and during treatment with caffeine
0-12h l2-24h 0-24h
Control Caffeine Control Caffeine Control Caffeine
Lean subjectst 17.2 ±0.8 19.2 ±0.9 13.4 ±0.6 13.2 ±0.4 30.6 ±0.9 32.3 ±0.9
Postobesesuhjectsf 14.5 ±0.8 15.7 ±0.8 1 1.3 ±0.4 11.5 ±0.5 25.8 ±0.9 27.2 ±1.2
Significance ofF
between treatments p <0.01 NS p <0.01
between groups p <0.025 p <0.02 p <0.01
*±5EM
tn=5; three females, two males.
1:n=6; three females, three males.
the lean and postobese groups to a similar extent but had a demonstrable stimulatory effect on RMR. In addition,
no significant effect on the subsequent 12-24-h EE in ci- it seems that the thermogenic effect ofcaffeine in man is
ther subject group (Fig 2, Table 5). Thus, the total 24-h dose dependent because the present data and the results
EE was significantly increased in both groups. ofother studies (2-7, 14) indicate that the magnitude of
The data on activity levels measured within the respi- response increases almost linearly with higher doses:
rometer were analyzed by computing the amount oftime over 2-2.5 h the thermogenic responses to doses of 100,
spent on various activities. Activities were categorized as 200-250, and 400-450 mg were 4-5, 10-12, and 16%,
sleeping, lying, sitting quietly (ie, minimal action, such respectively.
as reading, watching television, etc), sitting actively (eg, The current data also indicate that after fasting both
eating, writing, knitting, etc), and pottering or fidgeting the lean and the postobese subjects show similar in-
(eg, moving around the room to collect meals, perform- creases in metabolic rate after caffeine is ingested. These
ing personal toilets, etc). In the respirometer there were findings are compatible with a previous study (6) in
no differences in activity patterns between subject groups which no difference was observed between lean and
or between treatments. obese groups but are in direct conflict with another re-
port (7) in which the thermogenic response of the post-
Discussion obese subjects to caffeine is one-third less than that in
lean subjects. These discrepancies could be because in
Effect ofcaffeine on resting metabolic rate and the latter study (7) both groups received the same
diet-induced thermogenesis amount of caffeine even though the postobese subjects
This study demonstrates that caffeine increased the weighed about one-third more than the lean subjects and
RMR by 3-4% over 2.5 h at doses as low as 100 mg. therefore received a smaller dose of caffeine per unit
Because early studies (1 3, 14) indicated that RMR was body weight. This explanation is supported by data from
unaltered by lower doses of caffeine, it is likely that the the same study that indicated a lower plasma caffeine
amount ofcaffeine administered in our study represents level in postobese subjects than in lean subjects. In the
the minimum (or near minimum) dose that would allow present study the mean body weight and lean body mass
TABLE S
Changes in energy expenditure measured in a respirometer during administration of caffeine*
Absolute changes Percentage changes
0- 12 h 12-24 h 0-24 h 0-1 2 h 12-24 h 0-24 h
kcal
Leansubjectst 120±36 -11±25 109±50 11.4±3.1 -1.8±3.3 5.5±2.3
Postobesesubjectsf 74±31 6±24 78±34 7.9±3.4 1.5±3.5 4.9± 1.8
Significance ofF
1*twetn treatments p <0.01 NS p <0.02 p <0.01 NS p <0.02
between groups NS NS NS NS NS NS
*i±SEM.
tn=5; three females, two males.
j: n =6; three females, three males.
by guest on July 10, 2011www.ajcn.orgDownloaded from

CAFFEINE AND ENERGY EXPENDITURE 49
of the postobese group and the lean group did not differ
significantly and a similar increase in RMR resulted
from the administration ofthe same dose ofcaffeine; this
suggests that under fasting conditions both groups show
similar sensitivity in thermogenic response to caffeine.
In contrast, the thermogenic response to a mixed meal
differed considerably between the two subject groups.
The response of the postobese group was only half as
much as that of the lean group. These findings provide
further evidence for a subnormal thermogenic response
to food in those with a predisposition to obesity. Their
defective DIT was improved by caffeine, its thermogenic
effect being additive to that of the food. However,
caffeine produced only a small additional stimulatory
effect on DIT in the lean group. This apparently greater
thermogenic response in the postobese group, at least at
this relatively low dose, suggests that the postobese group
is more sensitive to caffeine than the lean group but only
when caffeine is taken with food. Similar greater thermo-
genic responses of postobese subjects than the lean sub-
jects in the fed state were reported during single-dose ad-
ministration of a sympathomimetic mixture of ephe-
drine and methylxanthines, although both groups also
showed identical thermogenic responses to the drugs in
the fasted state (10).
Influence ofcaffeine on daily energy expenditure
Although this study focuses primarily on the thermo-
genic responses oflean and postobese subjects to caffeine
rather than on a comparison oftheir absolute metabolic
rates, it nevertheless demonstrates that in addition to a
subnormal DIT the RMR and 24-h EE ofthe postobese
group were lower than in the lean group. These differ-
ences are statistically significant when the metabolic-rate
data are corrected for intergroup and intragroup varia-
tions in body weight. Similar differences (p <0.02) are
also apparent if the data are expressed per lean body
mass. Therefore, these findings support previous reports
(15-17) that postobese subjects tend to have a lower en-
ergy requirement for weight maintenance than do lean
subjects.
Repeated administration of caffeine increased the EE
ofboth subject groups but only during the period of drug
administration (ie, the first 12-h day period). The lack of
any residual thermogenic effect in the second 12-h night
period is probably because any residual plasma caffeine
level would have been cleared given its relatively short
half-life of 3-3.5 h. However, the similar increase in 0-
12-h EE in both subject groups contrasts with the find-
ings ofthe single-dose study indicating that caffeine had a
much smaller stimulatory effect on DIT in the lean group
than in the postobese group. There are two explanations
for this apparent discrepancy. It is possible that after re-
peated caffeine intake the additive effect of caffeine and
food on thermogenesis in the postobese group was not
sustained and that the increase in 0-12-h EE resulted
mostly from the stimulatory effect of caffeine on the
other components of EE. Alternatively, it is plausible to
suggest that repeated administration of caffeine, result-
ing in a higher plasma level ofcaffeine than that achieved
with a single dose, stimulated DIT to a similar extent in
both the postobese group and the lean group. This agrees
with previous studies in lean subjects (5, 6) indicating
that administration ofhigher doses ofcaffeine produced
an effect additive to that of food on thermogenesis.
Therefore, the current findings and previous studies
(5, 6) imply that people with a predisposition to obesity
may differ from lean people in sensitivity but not in ca-
pacity to the stimulatory effect ofcaffeine on DIT and on
daily EE.
Cqine and weight control
A main implication of this study concerns the poten-
tial use of caffeine as an apparently safe thermogenic
agent for weight control. The effect of caffeine on appe-
tite is unknown in man, but if it is assumed that there is
no compensatory increase in food intake, the increase of
,, 5% in 24-h EE after caffeine would represent an energy
deficit of75-l 10 kcal/d. These changes may be small but
over several months could accumulate and lead to sub-
stantial changes in body weight. A long-term human
study ofthe effects ofcaffeine on body fat content is long
overdue but studies in animals demonstrated that
caffeine and other methylxanthines, albeit at high doses,
reduced body weight and body fat by both anorectic and
thermogenic effects (18). Although caffeine and other
methylxanthines were ineffective in altering EE when ad-
ministered at low doses, they can markedly potentiate
the thermogenic effects of ephedrine (a sympathomi-
metic agent) and lead to a reversal or prevention of obe-
sity in some animal models (19-2 1). In man, administra-
tion ofmethylxanthines in doses (80 mg) similar to those
administered in this study (100 mg) doubles the thermo-
genic effect of ephedrine, and such mixtures completely
normalize the defective DIT found in postobese subjects
to those levels found in lean subjects (10). The ability
ofcommonly consumed amounts ofcaffeine to increase
daily EE, as demonstrated in the present investigation,
coupled with caffeine’s ability to augment the thermo-
genic effects of certain sympathetic stimulants (10, 19-
22) may have contributed to reported weight losses in
obese humans (23). The potential use ofcaffeine as a pro-
moter of thermogenesis during the treatment of obesity
warrants further study. CI
We thank all the volunteers who participated in this study.
References
1. Dulloo AG, Miller DS. Obesity: a disorder of the sympathetic
nervous system. World Rev Nutr Diet 1987;S0: 1-56.
2. Means JH, Aub JC, Du Bois EF. Clinical calorimetry. Twentieth
paper. The effect of caffeine on the heat production. Arch Intern
Med 1917; 19:832-9.
3. Grollman A. The action of alcohol, caffeine and tobacco on the
cardiac output (and its related functions) of normal man. J
Pharmacol Exp Ther l930;39:3 13-27.
4. Horst K, Wilson RJ, Smith RG. The effect of coffee and
by guest on July 10, 2011www.ajcn.orgDownloaded from

50 DULLOO ET AL
decaffeinated coffee on oxygen consumption, pulse rate and blood
pressure. J Pharmacol Exp Ther 1936; 7:294-304.
5. Miller DS, Stock MJ, Stuart JA. The effects of carnitine and
caffeine on the oxygen consumption of fed and fasted subjects.
Proc Nutr Soc 1974; 33:28A(abstr).
6. Acheson KJ, Zahorska-Markiewicz B, Pittet PH, Anantharaman
K, Jequier E. Caffeine and coffee: their influence on metabolic rate
and substrate utilization in normal weight and obese individuals.
Am J Gin Nutr 1980; 33:989-97.
7. Jung RT, Sheuy PS, James WPT, Barrand MA, Callingham BA.
Caffeine: its effect on catecholamines and metabolism in lean and
obese humans. Gin Sci 198 l;60:527-35.
8. GarrowiS. Energy balanceand obesity in man. 1st ed. Amsterdam,
North Holland: Elsevier Science Publishers, 1974.
9. Durnin JVGA, Womersley J. Body fat assessed from total body
density and its estimation from skinfold thickness measurements
of48l men and women aged 16-72 years. Br J Nutr 1974; 32:77-
97.
10. Dulloo AG, Miller DS. The thermogenic properties of ephedrine/
methylxanthine mixtures: Human studies. Int J Obes 1986; 10:
467-81.
1 1. Weir JB. New methods for calculating metabolic rate with special
reference to protein metabolism. J Physiol(Lond) 1949; 109:1-9.
12. Zar JH. Biostatistical analyses, 2nd ed. Englewood Giffs, NJ:
Prentice-Hall, 1984.
13. Boothby WM, Rowntree LG. Drugs and basal metabolism. J
Pharmacol ExpTher l924;22:99-l08.
14. Haldi J, Bachmann G, Ensor C, Wynn W. The effect of various
amounts of caffeine on the gaseous exchange and the respiratory
quotient in man. J Nutr l941;2l:307-20.
15. Miller DS, Parsonage S. Resistance to slimming. Lancet 1975; 1:
773-9.
16. Leibel RL, Hirsh J. Diminished energy requirements in reduced-
obese patients. Metabolism 1984;33: 164-70.
17. Geissler CA, Miller DS, Shah M. The daily metabolic rate of the
post-obese and the lean. Am J Clin Nutr l987;45:914-20.
18. Dulloo AG, Miller DS. Thermogenic drugs for the treatment of
obesity: sympathetic stimulants in animal models. Br J Nutr
l984;52: I79-96.
19. Dulloo AG, Miller DS. The thermogenic properties of ephedrine/
methylxanthine mixtures: animal studies. Am J Clin Nutr
l986;43:388-94.
20. Dulloo AG, Miller DS. Prevention ofgenetic fa/fa obesity with an
ephedrine-methylxanthines thermogenic mixture. Am J Physiol
I 987;252:R507- I 3.
21. Dulloo AG, Miller DS. Reversal ofobesity in genetically obese fa/
fa zucker rat with an ephedrine/methylxanthines thermogenic
mixture. J Nutr 1987; 1 17:383-9.
22. Wellman PJ, Marmon MM: Synergism between caffeine and dl-
phenylpropanolamine on BAT thermogenesis in the adult rat.
Pharmacol Biochem Behav 1985;22:78 1-5.
23. Malchow-Moller A, Larsen 5, Hey H, Stokholm KH, Juhl E,
Quaade F. Ephedrine as an anorectic: the story of the ‘Elsinore’
pill. IntJ Obes 198l;S:l83-7.
by guest on July 10, 2011www.ajcn.orgDownloaded from