Hindawi Publishing Corporation

Journal of Nutrition and Metabolism

Volume 2012, Article ID 207426, 9 pages
doi:10.1155/2012/207426

Research Article
Effects of 16-Week Consumption of Caffeinated and
Decaffeinated Instant Coffee on Glucose Metabolism in
a Randomized Controlled Trial

Keizo Ohnaka,1 Mizuko Ikeda,2 Takako Maki,2 Tomoko Okada,3 Takao Shimazoe,3
Masahiro Adachi,4 Masatoshi Nomura,4 Ryoichi Takayanagi,4 and Suminori Kono2
1 Department of Geriatric Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka 812-8582, Japan
2 Department of Preventive Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka 812-8582, Japan
3 Department of Clinical Pharmacology, Graduate School of Pharmaceutical Sciences, Kyushu University, Fukuoka 812-8582, Japan
4 Department of Medicine and Bioregulatory Science, Graduate School of Medical Sciences, Higashi-ku, Fukuoka 812-8582, Japan

Correspondence should be addressed to Suminori Kono, skono@phealth.med.kyushu-u.ac.jp

Received 4 April 2012; Revised 9 October 2012; Accepted 9 October 2012

Academic Editor: Cindy Davis

Copyright © 2012 Keizo Ohnaka et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Objective. Observational studies have shown a protective association between coffee consumption and type 2 diabetes mellitus
whereas caffeine or caffeinated coffee acutely deteriorates glucose tolerance. We investigated the effects of chronic drinking of
instant coffee on glucose and insulin concentrations during a 75 g oral glucose tolerance test. Methods. Overweight men with a
mild-to-moderate elevation of fasting plasma glucose were randomly allocated to a 16-week intervention of consuming 5 cups
of caffeinated (n = 17) or decaffeinated (n = 15) instant coffee per day or no coffee (n = 13). Results. The caffeinated coffee
group showed statistically significant decreases in the 2-hour concentrations and the area under the curve of glucose while neither
decaffeinated coffee nor coffee group showed such a change. Waist circumstance decreased in the caffeinated coffee group, increased
in the decaffeinated coffee group, and did not change in the noncoffee group (P = 0.002). With adjustment for the change in waist
circumference, caffeinated and decaffeinated coffee consumption were associated with a modest decrease in the postload glucose
levels. Conclusion. Both caffeinated and decaffeinated coffee may be protective against deterioration of glucose tolerance.

1. Introduction glucose metabolism. Chlorogenic acids are such bioactive

compounds of a potent antioxidant property [13, 14].
Observational studies have consistently shown a protective Alternatively, habitual coffee consumption may be protective
association between coffee consumption and type 2 dia- against type 2 diabetes mellitus by mechanisms other than
betes mellitus [1, 2]. Both caffeinated and decaffeinated postulated for the acute effect of caffeine.
coffee seem to be associated with a decreased risk of type Few trials have addressed the long-term effect of caffeine
2 diabetes mellitus [3, 4], although a prospective study or caffeinated coffee on glucose metabolism. Participants in
suggested a stronger association with decaffeinated coffee the intervention group in these studies consumed caffeine
[5]. On the other hand, it is well documented that the (400 mg/day) for one week [15]; filtered coffee (900 mL/day)
ingestion of caffeine or caffeinated coffee acutely deteriorates or caffeine (870 mg/day) for 2 weeks [16]; filtered coffee
glucose tolerance, as assessed by postprandial plasma glu- (1 L/day) for 4 weeks [16]; caffeinated or decaffeinated
cose concentrations [6–10]. Caffeine administration is also instant coffee (5 cups/day) for 8 weeks [17]. In a one-arm
shown to decrease the insulin-mediated glucose disposal in trial [18], the subjects consumed 4 cups of filtered coffee
the hyperinsulinemic-euglycemic glucose clamp procedure per day for 1 month and 8 cups of filtered coffee per
in humans [11, 12]. It is thus hypothesized that coffee day for another month. None of these studies showed a
compounds other than caffeine exert a beneficial effect on clear effect on either fasting [15, 16] or postload [17, 18]
2 Journal of Nutrition and Metabolism

glucose concentrations, while some suggested an elevation of Randomization was done by permuted block design using
fasting insulin concentrations after one-week use of caffeine blocks of different sizes (multiples of three) with each block
[15] and 4-week consumption of filtered coffee [16]. In a consisting of equal numbers of the three assignments. The
randomized controlled trial reported here, we investigated block size was adjusted to the number of subjects entering
the effects of drinking caffeinated and decaffeinated instant the intervention at different points of time. Each participant
coffee (5 cups/day) for 16 weeks on glucose and insulin drew one tip from among the bag containing chips numbered
concentrations during a 75 g oral glucose tolerance test sequentially and was allocated to an intervention group
(OGTT). We also evaluated the effects of coffee consumption specified for the sequential number. Remaining chips were
on plasma adiponectin concentrations and serum C-reactive retained in the bag to which a new block of chips were
protein (CRP), because coffee consumption was reported added so that the bag always contained at least one for
to be associated with these inflammation-related biological each intervention group. Study participants received $30 at
markers in some [19, 20], but not all [21], human studies. screening, $100 at 0 and 8 weeks each, and $200 at 16 weeks
($1 = 100 Japanese yen).
2. Methods
2.3. Coffee Preparation. Subjects were instructed to prepare
2.1. Participants. Participants were men aged 40–64 years one cup/glass of coffee using one spoonful of instant coffee
who had body mass index of 25–30 kg/m2 and fasting (approximately 1.2–1.3 g). Either hot or ice coffee was
plasma glucose of 100–140 mg/dL in the past year (or 90– permitted, but coffee was drunk without sugar, milk, or any
140 mg/dL at screening in the case of no measurement in the other additives. A required amount of instant coffee and two
past year). Exclusion criteria were disease under mediation, spoons of the same size were supplied to each subject in the
life-limiting disease without medication, a prior history of caffeinated or decaffeinated coffee group.
gastrectomy, being unable to drink coffee, and daily use of
coffee. Participants were recruited by means of articles in
2.4. Measurements. Subjects attended the screening at 10:00
nationwide newspapers, advertisements in local newspapers
hours, and those without a recorded measurement of fasting
and magazines, posters at workplaces, and personal contact.
plasma glucose were instructed to be fast at least for 10
The trial was approved by the ethics committee of Kyushu
hours. Height and body weight were measured with subjects
University Graduate School of Medical Sciences, and each
wearing light cloths without shoes. On scheduled visits
participant gave written informed consent.
at weeks 0, 8, and 16 of intervention, participants were
During the period from March 2008 to April 2009, a total
instructed to be fast from 21:00 hours on the previous day
of 276 men contacted us about the trial. After a prescreening
and to be present at the study site at 09:30 hour. Body weight,
by telephone and a screening examination, a total of 49 men
waist circumference, and blood pressure were measured.
entered the study.
Waist circumference was measured at the horizontal plane of
the umbilical level with subjects wearing an undergarment
2.2. Protocol. Eligible consenting participants entered a 2- at the expiratory phase of usual abdominal respiration.
week run-in period, followed by a 16-week intervention A 75 g OGTT was started at 10:00 hours, and venous
period. The run-in period was employed for washout of blood was drawn at 0, 30, 60, 90, and 120 min after the
caffeine and for ensuring tolerability to caffeine depletion. glucose challenge. Plasma and serum were separated after
After completion of the run-in period, subjects were ran- centrifugation within two hours. Plasma samples for glucose
domly allocated to one of the following three treatments and insulin measurements were stored at −20◦ C until the
of 16 weeks, that is, 5 cups of caffeinated instant coffee assay within 2 working days. Serum samples for caffeine,
per day, 5 cups of decaffeinated instant coffee per day, CRP, and adiponectin measurements were stored at −80◦ C
and no coffee. During the run-in and intervention periods, until the completion of the trial.
subjects were not allowed to consume caffeine-contained All laboratory measurements except for caffeine were car-
beverages and foods (coffee, tea, cola, cocoa, chocolate, ried out at an external laboratory (SRL, Tokyo). Plasma glu-
and caffeine-contained supplements) except supplied coffee. cose and insulin concentrations were determined by hexoki-
They were instructed to maintain the current diet and nase method and chemiluminescent enzyme immunoassay,
physical activity during the study period. Mineral water respectively. Serum concentrations of high-sensitivity CRP
(Coca-Cola MINAQUA) of an amount corresponding to were measured by using a latex-enhanced immunoneph-
two 500-mL bottles per day was supplied to all the subjects elometric assay on a BN II analyzer (Siemens Healthcare
during the run-in period and to those in the noncoffee group Diagnostics, Marburg, Germany) [22]. Total adiponectin
during the intervention period. Subjects in the coffee groups concentrations were measured by the ELISA method [23],
were supplied with caffeinated or decaffeinated instant coffee and high-molecular weight (HMW) adiponectin concentra-
(NESCAFE GOLDBLEND) of the same lot number and with tions were assayed by the two-step sandwich ELISA method
mineral water of one 500-mL bottle per day during the [24]. For one subject in the decaffeinated coffee group, blood
intervention period. sampling was missed at 90 min during the OGTT at 8 weeks,
Subjects were invited to a study site at the medical and the glucose and insulin concentrations were imputed
campus of Kyushu University prior to the run-in period and by averaging the 60 min and 120 min values. One in the
at weeks 0, 8, and 16 of intervention for 75-g OGGT. caffeinated coffee group had CRP below the detection limit
Journal of Nutrition and Metabolism 3

(0.05 mg/L) at the baseline and 16 weeks, and a value of covariance was used to obtain adjusted mean percent changes
0.05 mg/L was imputed. In addition, CRP was greater than in glucose metabolism parameters with control for change
10 mg/L at 16 weeks for one in the noncoffee group, and the in waist circumference and to assess the overall difference
value was discarded because such a high value is generally among the three groups. Linear regression coefficients for the
indicative of acute inflammation. indicator variables corresponding to the treatment categories
As for saliva collection (see below), 3 cotton balls of 1 cm (caffeinated and decaffeinated coffee) were used for the
in diameter were placed in the sublingual and cheek pouch between-group difference as compared with the noncoffee
after mouth rinse with a cup of water, and approximately group. Statistical significance was declared if two-sided
1 mL of saliva was collected into a tube by pressing wet cotton P value was less than 0.05. All statistical analyses were
balls in a 5 mL syringe. Saliva solution was stored at −80◦ C. performed by Stata Statistical Software version 10 (Stata
Pretreatment of saliva samples was performed by the method Corporation, College Station, TX).
described elsewhere [25]. Serum and salivary caffeine con-
centrations were determined by high-performance liquid
chromatography at Kyushu University Graduate School of 3. Results
Pharmaceutical Sciences [26]. Standard caffeine solutions of
3.1. Baseline Characteristics. As shown in Figure 1, a total
1, 2, 5, 10, 25, 50, and 100 μM were used for calibration. Thus
of 49 men were randomly allocated to one of the three
the detection limit of caffeine concentration was 1 μM.
groups: caffeinated coffee (n = 17), decaffeinated coffee
(n = 16), and noncoffee (n = 16). After the baseline OGTT,
2.5. Compliance. On each visit to the study site, a self- 4 men withdrew from the trial for diabetes mellitus requiring
administered questionnaire ascertained smoking, alcohol treatment (decaffeinated coffee group, n = 1; noncoffee
drinking, use of caffeine-containing foods and beverages, group, n = 1) and hospitalization for acute diseases
and drug use during the between-visit period. The ques- (noncoffee group, n = 2). Furthermore, after the 8-week
tionnaires at 8 and 16 weeks of intervention inquired OGTT, 2 men withdrew for diabetes mellitus requiring
about any changes in diet and physical activity. Serum and treatment (decaffeinated coffee group, n = 1) and treatment
salivary caffeine concentrations were determined at weeks of hypertension (decaffeinated coffee group, n = 1).
0, 8, and 16 of intervention. Subjects were visited at their Withdrawal of the 3 men for treatment of diabetes mellitus
homes or workplaces without appointment twice during the was decided on the basis of the OGTT in the study. Thus the
intervention period, that is, each in the first 8 weeks and study subjects were 45 men with the 8-week measurements,
second 8 weeks. and the analysis on the changes at 16 weeks was confined to
43 men with the 16-week measurements.
Age of the subjects ranged from 40 to 64 years with
2.6. Statistical Analyses. Areas under the curve (AUC) for
a mean of 52.7 years (SD 7.9 years). Smokers numbered
glucose and insulin during the OGTT were calculated
11 (24.4%), and median amounts of coffee and tea (green,
using the trapezoidal method. Composite insulin sensitivity
black, and oolong tea combined) consumption were 2 cups
index, which represents the whole-body insulin disposal,
per week (IQR 0.5–4.0) and 5.5 cups per week (IQR 2–
was calculated by the method proposed by Matsuda and
21), respectively. There was no appreciable difference among
DeFronzo [27]. Homeostasis model assessment of insulin
the three intervention groups with respect to age (P =
resistance (HOMA-IR) was calculated [28].
0.46), smoking (P = 0.69), coffee use (P = 0.55), and tea
The distributions of laboratory measurements were
consumption (P = 0.69).
mostly skewed to the right side, and median and interquartile
None of the glucose and insulin parameters showed
range (IQR) was used to present the measurements at
a measurable variation among the three groups (Table 1).
baseline. Analysis of variance, χ 2 -test, and Kruskal-Wallis
Adiponectin and CRP concentrations were also similar in the
test were used for the three-group comparisons of means,
three groups.
proportions, and medians, respectively. As for the changes
in laboratory measurements after the intervention, we used
mean and standard deviation (SD) of the change for ease 3.2. Changes during the Intervention. Mean changes of the
of presentation with nonparametric methods for statistical laboratory parameters at 8 weeks and 16 weeks of the
tests (Kruskal-Wallis test and Wilcoxon rank-sum test for treatment are summarized in Tables 2 and 3, respectively.
the between-group comparison and Wilcoxon singed-rank The caffeinated coffee group showed statistically significant
test for the within-group comparison). We also performed decreases in the 2-hour glucose and AUC glucose at 16 weeks
the analysis by using the values at different points of time (Table 3), but not at 8 weeks (Table 2), as compared with the
transformed to natural logarithms and assessed the mean baseline values. Neither decaffeinated coffee nor noncoffee
and 95% confidence interval (CI) of the percent change from group showed such decreases. These decrease at 16 weeks
the baseline by analysis of variance. Exponentiation of the among the caffeinated coffee group also statistically signifi-
mean difference in the log-scale necessarily corresponds to cantly differed from the changes observed in the noncoffee
the average percent change from the baseline. Pearson corre- group. The average percent decreases were 13.1% (95% CI
lation coefficients were calculated to evaluate how changes 1.6–23.2) for 2-hour glucose and 7.5% (95% CI 1.1–13.5)
in body weight and waist circumference were correlated for AUC glucose after a 16-week consumption of caffeinated
with changes in glucose metabolism parameters. Analysis of coffee. Insulin parameters including the composite ISI and
4 Journal of Nutrition and Metabolism

276 applicants

200 ineligible

76 subjects invited to screening

60 subjects visited screening

10 ineligible and 1 withdrew

49 subjects randomized

Caffeinated coffee Decaffeinated coffee No coffee

n = 17 n = 16 n = 16

1 withdrew 3 withdrew

8-week visit 8-week visit 8-week visit

n = 17 n = 15 n = 13

2 withdrew

16-week visit 16-week visit 16-week visit

n = 17 n = 13 n = 13

Figure 1: Trial profile.

Table 1: Anthropometric measures, glucose metabolism parameters, and serum adiponectin and C-reactive protein at baseline.

Parameter (unit) Caffeinated coffee (n = 17) Decaffeinated coffee (n = 15) No coffee (n = 13) P value∗
Body weight (kg) 77 (70–84) 75 (71–80) 79 (78–83) 0.22
Waist circumference (cm) 93 (89–99) 91 (88–95) 95 (90–97) 0.46
Plasma glucose (mmol/L)
Fasting 6.0 (5.3–6.2) 6.1 (5.3–7.4) 5.8 (5.4–6.3) 0.68
2-hour 8.5 (7.3–9.9) 10.1 (7.7–12.8) 9.2 (7.4–12.9) 0.35
AUC† 18.3 (15.9–20.1) 20.6 (17.5–23.7) 20.8 (18.5–23.2) 0.16
Plasma insulin (pmol/L)
Fasting 57 (38–66) 34 (27–71) 58 (52–66) 0.53
2-hour 551 (347–868) 314 (187–565) 472 (312–854) 0.24
AUC† 999 (676–1289) 570 (376–978) 881 (629–1057) 0.13
Composite ISI 3.6 (2.3–4.6) 4.8 (2.4–7.9) 3.4 (2.1–5.0) 0.38
HOMA-IR 2.0 (1.3–2.5) 1.2 (0.9–3.3) 2.0 (1.8–2.9) 0.51
Total adiponectin (μg/mL) 4.9 (4.1–8.2) 5.8 (5.6–6.3) 4.8 (3.9–6.5) 0.66
HMW adiponectin (μg/mL) 2.6 (1.5–4.9) 3.1 (2.2–4.4) 2.2 (1.7–3.6) 0.73
C-reactive protein (mg/L) 0.49 (0.26–0.85) 0.23 (0.19–0.83) 0.42 (0.29–0.72) 0.24
AUC: area under the curve; HMW: high-molecular weight; HOMA-IR: homeostasis model assessment of insulin resistance; ISI: insulin sensitivity index.
Values are medians and interquartile ranges (25th and 75th percentiles) in parentheses.
∗ Overall difference based on Kruskal-Wallis test.
† Hour was used for the time scale in the oral glucose tolerance test.

HOMA-IR did not change materially during the intervention the mean percent change showed that total adiponectin at
in any treatment groups and showed no between-group 8 and 16 weeks and HMW adiponectin at 16 weeks increased
difference in the change. statistically significantly, as compared with the baseline, in
Although the nonparametric analysis showed no statis- the caffeinated coffee group while these increases did not
tically significant change in total or HMW adiponectin in differ from the changes observed in the noncoffee group. The
any treatment groups (Tables 2 and 3), the analysis using mean percent increases of total adiponectin were 6.0% (95%
Journal of Nutrition and Metabolism 5

CI 0.2–12.0) at 8 weeks and 8.9% (95% CI 1.8–16.4) at 16 The first 8 weeks

30
weeks, and the mean percent increase of HMW adiponectin
at 16 weeks was 13.2% (95% CI 0.8–27.2).
Body weight and waist circumference did not change 25
in any of the three groups after 8 weeks of intervention
(data not shown). At 16 weeks, however, waist circumference
decreased by 1.5 cm (95% CI 0.6–2.5) in the caffeinated 20
coffee group and increased by 1.3 cm (95% CI 0.2–2.4) in
the decaffeinated coffee group while a small decrease of
15
0.6 cm (95% CI −0.5 to 1.7) was observed in the noncoffee
group (overall P = 0.002). Body weight at 16 weeks also
showed a similar, but less prominent, pattern; the changes 10
from the baseline were −1.1 kg (95% CI −2.0 to −0.1) in
the caffeinated coffee group, 0.5 kg (95% CI −0.6 to 1.6) in
the decaffeinated coffee group, and −0.6 kg (95% CI −1.7 5
to 0.5) in the noncoffee group (overall P = 0.10). The
16-week change in waist circumference was fairly strongly
correlated with the changes in the log-scale of the 2-hour 0
glucose (correlation coefficient r = 0.403) and AUC glucose Caffeinated Decaffeinated No coffee
(r = 0.399), but the correlation coefficients for the other
(a)
parameters were relatively small: fasting glucose 0.06, fasting
insulin 0.13, 2-hour insulin 0.19, AUC insulin 0.14, ISI The second 8 weeks
30
−0.21, HOMA-IR 0.12, total adiponectin −0.21, HMW
adiponectin −0.13, and CRP 0.17.
When statistical adjustment was made for the 16-week 25
change in waist circumference, the percent changes in 2-hour (95)
and AUC values of glucose concentrations were attenuated
in the caffeinated coffee group and were accentuated in 20
the decaffeinated coffee group. The adjusted mean percent
changes of 2-hour glucose at 16 weeks were −8.2% (95% CI
15
−18.8 to 3.9) in the caffeinated coffee group, −7.7% (95%
CI −20.4 to 6.9) in the decaffeinated coffee group, and 9.0%
(95% CI −4.5 to 24.5) in the noncoffee group (overall P = 10
0.11); a pooled average percent decrease in the caffeinated
and decaffeinated coffee group was significantly different as
compared with the noncoffee group (P = 0.04). The adjusted 5
mean percent changes of AUC glucose were −4.5% (95% CI
−10.5 to 2.0) for caffeinated coffee, −7.6% (95% CI −14.6
0
to −0.1) for decaffeinated coffee, and 2.2% (95% CI −4.8 to
9.7) for control (overall P = 0.14); a pooled percent decrease Caffeinated Decaffeinated No coffee
in the two coffee groups was nearly significant as compared (b)
with the noncoffee group (P = 0.053).
Figure 2: Caffeine concentrations (μM) in saliva collected at
unannounced visits in the first 8 weeks (a) and the second 8 weeks
3.3. Compliance. At the baseline after the 2-week abstinence (b) during the intervention.
from caffeine, 36 (80%) of the 45 men were negative for
serum caffeine; 8 (18.6%) had serum caffeine concentrations
of 1.2–3.3 μM, and one (2.3%) had 10.3 μM of serum
caffeine. Caffeine concentrations in saliva collected by unan-
nounced visits are shown in Figure 2. Median concentrations visit. High concentrations of serum caffeine (>5 μM) were
of salivary caffeine in the caffeinated coffee group were detected for 3 subjects in the decaffeinated coffee group
9.5 μM (IQR 5.6–13.4 μM) during the first 8 weeks and and for one subject in the noncoffee group at either of the
9.3 μM (IQR 5.2–12.1 μM) during the second 8 weeks. Few scheduled visits during the intervention period.
men in the decaffeinated coffee group (n = 3) and noncoffee Two men in the caffeinated coffee group and 3 men in
group (n = 1) had salivary caffeine concentrations of >5 μM the decaffeinated coffee group reported changes in physical
at either occasion. activity at weeks 8 or 16. In the former group, one reported a
In the caffeinated coffee group, median concentrations lowered physical activity and the other reported an increase
of serum caffeine were 6.9 μM (IQR 3.2–9.9 μM) at the 8- in physical activity. All of the 3 men in the decaffeinated
week visit and 8.2 μM (IQR 5.3–12.4 μM) at the 16-week coffee group reported a decrease in physical activity. The
6 Journal of Nutrition and Metabolism

Table 2: Mean changes of glucose metabolism parameters and serum adiponectin and C-reactive protein at 8 weeks of intervention as
compared with the baseline values.

Parameter Caffeinated coffee (n = 17) Decaffeinated coffee (n = 15) No coffee (n = 13) P value∗
Plasma glucose (mmol/L)
Fasting 0.0 (0.4) 0.1 (0.4) −0.0 (0.2) 0.55
2-hour −1.1 (2.4) −0.5 (2.2) −0.3 (1.6) 0.51
AUC† −0.7 (2.3) −0.3 (1.8) −0.6 (1.2) 0.62
Plasma insulin (pmol/L)
Fasting 3 (21) 5 (19) −7 (24) 0.53
2-hour −44 (315) −6 (192) 18 (319) 0.99
AUC† 7 (234) −5 (291) −12 (214) 0.82
Composite ISI −0.2 (1.5) −0.1 (1.9) 0.2 (0.8) 0.46
HOMA-IR 0.1 (0.9) 0.2 (0.8) −0.3 (1.0) 0.27
Total adiponectin (μg/mL)‡ 0.2 (0.9) −0.1 (0.5) −0.0 (0.7) 0.15
HMW adiponectin (μg/mL)‡ 0.1 (0.6) −0.1 (0.6) 0.0 (0.7) 0.48
C-reactive protein (mg/L)† −0.39 (1.54) 0.18 (0.62) −0.12 (0.57) 0.80
AUC: area under the curve; HMW: high-molecular weight; HOMA-IR: homeostasis model assessment of insulin resistance; ISI: insulin sensitivity index.
Values in parentheses are standard deviations.
∗ Overall difference based on Kruskal-Wallis test.
† Hour was used for the time scale in the oral glucose tolerance test.
‡ Number of the subjects in the decaffeinated coffee group was 13.

Table 3: Mean changes of glucose metabolism parameters and serum adiponectin and C-reactive protein at 16 weeks of intervention as
compared with the baseline values.

Parameter Caffeinated coffee (n = 17) Decaffeinated coffee (n = 13) No coffee (n = 13) P value∗
Plasma glucose (mmol/L)
Fasting −0.0 (0.5) 0.0 (0.7) −0.0 (0.5) 0.88
2-hour −1.3 (2.2)† 0.1 (2.6) 0.7 (1.4) 0.07
AUC‡ −1.4 (2.6)† −0.3 (3.3) 0.3 (1.6) 0.08
Plasma insulin (pmol/L)
Fasting 4 (23) 5 (20) 5 (16) 0.94
2-hour −2 (515) 102 (896) 90 (260) 0.40
AUC‡ 115 (526) 63 (771) 36 (252) 0.94
Composite ISI −0.2 (1.7) −0.4 (1.5) −0.3 (1.0) 0.85
HOMA-IR 0.1 (0.9) 0.2 (0.8) 0.1 (0.7) 0.86
Total adiponectin (μg/mL) 0.4 (0.9) 0.0 (0.5) 0.4 (1.1) 0.45
HMW adiponectin (μg/mL) 0.3 (0.8) −0.0 (0.5) 0.3 (0.7) 0.40
C-reactive protein (mg/L) −0.43 (1.33) 0.42 (1.16) −0.21 (0.66) 0.63
AUC: area under the curve; HMW: high-molecular weight; HOMA-IR: homeostasis model assessment of insulin resistance; ISI: insulin sensitivity index.
Values in parentheses are standard deviations.
∗ Overall difference based on Kruskal-Wallis test.
† P < 0.05 as compared with the baseline and as compared with the noncoffee group.
‡ Hour was used for the time scale in the oral glucose tolerance test.

exclusion of these 5 men slightly attenuated the between- caffeinated instant coffee per day. A consumption of decaf-
group difference in the change of waist circumference; mean feinated coffee did not show such decreases. However, waist
changes at 16 weeks were −1.3 cm (95% CI −2.3 to −0.2 cm) circumference decreased in the caffeinated coffee group and
in the caffeinated coffee group; 0.9 cm (95% CI −0.4 to increased in the decaffeinated coffee group. With allowance
2.2 cm) in the decaffeinated coffee group; −0.6 cm (95% CI for the change in waist circumference, the postload glucose
−1.7 to 0.5 cm) in the noncoffee group (overall P = 0.04). levels seemed to be lowered after a 16-week consumption of
caffeinated or decaffeinated coffee.
4. Discussion The differential change in waist circumstance between
the caffeinated and decaffeinated groups was indeed prob-
Modest decreases in 2-hour glucose and AUC glucose lematic. An apparent increase in waist circumference after a
were observed after a 16-week consumption of 5 cups of 16-week consumption of decaffeinated coffee was ascribed
Journal of Nutrition and Metabolism 7

partly to a decrease in physical activity as reported by 3 There was a fairly large variation in serum and salivary
men. In the caffeinated coffee group, physical activity did caffeine concentrations in the caffeinated coffee group.
not reportedly change to such an extent, but slight decreases Caffeine is metabolized almost exclusively by CYP1A2 in
in body weight and waist circumference may have been the liver. Functional genetic polymorphisms are known in
due to an unreported change in diet or physical activity. the CYP1A2 gene, and the between-subject variation in
The decreases in body weight and waist circumference in caffeine metabolism is well known [40]. Thus low caffeine
the caffeinated coffee group may be regarded as compatible concentrations do not necessarily indicate poor compliance
with caffeine’s effects of increasing thermogenesis and fat in the caffeinated coffee group.
oxidation [29]. In a meta-analysis of chamber studies on The 16-week intervention period, use of the standard test
humans [30], a dose of 300 mg/day of caffeine was associated for glucose tolerance, and saliva collection without appoint-
with an 80 kcal increase of energy expenditure per day. ment were advantages in the present studies. However, there
Caffeine enhances the release of epinephrine and free fatty were several limitations. We did not have direct information
acids in a fasting condition [6–10], and these physiological as to how stable the dietary intake and physical activity were
effects may be linked to the increase in energy expenditure during the intervention period. Statistical adjustment was
[29]. However, it remains uncertain whether a long-term made for the change in waist circumstance, but this measure
use of caffeine is beneficial in maintaining body weight or alone probably did not capture the changes in physical activ-
decreasing body fat, while an increase in caffeine intake was ity and diet which would have affected glucose metabolism.
reported to be associated with a small reduction in long-term The treatment was not blind to either the participants or
weight in an observational study [31]. the investigators, but this lack of blindness did not affect
The present study suggested that both caffeinated and the laboratory measurements. Coffee was consumed without
decaffeinated coffee were associated with a modest improve- any additives in the present study, and the findings may
ment in the 2-hour glucose concentrations. The present not be applicable to coffee drinking with sugar and/or milk.
findings are consistent with the results from observational Finally, we did not measure chemical compounds contained
studies [32, 33]. These studies consistently showed that in caffeinated and decaffeinated coffee. It was previously
coffee consumption was more strongly associated with lower reported that decaffeinated instant coffee contained a slightly
concentrations of 2-hour glucose than of fasting glucose lower amount of chlorogenic acids than caffeinated instant
during a 75 g OGTT [32, 33]. The present study adds to coffee of the same brand [17].
evidence that coffee compounds other than caffeine are
protective in glucose metabolism. Chlorogenic acids and
5. Conclusion
other noncaffeine coffee compounds may exert protective
effects by decreasing hepatic glucose production through In overweight men with a mild-to-moderate elevation of
inhibition of hepatic glucose-6-phosphate translocase [34], fasting glucose concentrations, the 2-hour glucose and AUC
delaying intestinal glucose absorption as suggested by an glucose during a 75 g OGTT decreased after a 16-week
altered profile of plasma concentrations of gastrointestinal consumption of 5 cups of caffeinated instant coffee per
hormones [35], and increasing whole-body glucose disposal day. However, with adjustment for the change in waist
or insulin sensitivity [36]. Coffee polyphenols were also circumference, both caffeinated and decaffeinated coffee
shown to be protective against the damage of pancreatic islet seemed to be associated with lowered levels of the postload
caused by oxidative stress [37]. glucose. Habitual use of both caffeinated and decaffeinated
There is no doubt that caffeine or caffeinated coffee coffee may be protective against deterioration of glucose
deteriorates glucose tolerance when administered prior to tolerance.
glucose load or meal. This adverse effect was observed not
only in healthy subjects [6, 10] but also in patients with type
2 diabetes mellitus [7, 38, 39], with a habitual consumption Acknowledgments
of caffeinated coffee of different types. The caffeine’s acute
The investigators are grateful to Neslé Japan, Co., Ltd., for
effect on glucose tolerance does not seem to diminish or
provision of instant coffee and to Coca-Cola West Japan,
weaken with a habitual consumption of caffeinated coffee
Co., Ltd., for support as to purchase of mineral water. The
[38]. Further studies are needed to elucidate a mechanism
study was financially supported by the All Japan Coffee
or mechanisms for a possible protective effect of a habitual
Association.
use of caffeinated coffee as well as of decaffeinated coffee in
glucose metabolism.
The insulin parameters did not change by either caf- References
feinated or decaffeinated coffee in the present study. The
[1] R. M. Van Dam and F. B. Hu, “Coffee consumption and risk of
changes in the insulin-related parameters generally showed
type 2 diabetes: a systematic review,” Journal of the American
a large between-subject variation, and it may have been Medical Association, vol. 294, no. 1, pp. 97–104, 2005.
difficult to detect a possible effect of coffee on the insulin [2] R. Huxley, C. M. Y. Lee, F. Barzi et al., “Coffee, decaffeinated
parameters. A suggestive increase in adiponectin concentra- coffee, and tea consumption in relation to incident type 2
tions observed for the caffeinated coffee group is consistent diabetes mellitus: a systematic review with meta-analysis,”
with the previous observation [17], but the increase did not Archives of Internal Medicine, vol. 169, no. 22, pp. 2053–2063,
differ from the change in the noncoffee group. 2009.
8 Journal of Nutrition and Metabolism

[3] E. Salazar-Martinez, W. C. Willett, A. Ascherio et al., “Coffee for type 2 diabetes: a clinical trial,” American Journal of Clinical
consumption and risk for type 2 diabetes mellitus,” Annals of Nutrition, vol. 91, no. 4, pp. 950–957, 2010.
Internal Medicine, vol. 140, no. 1, pp. 1–8, 2004. [19] E. Lopez-Garcia, R. M. Van Dam, L. Qi, and F. B. Hu, “Coffee
[4] R. M. Van Dam, W. C. Willett, J. E. Manson, and F. B. Hu, consumption and markers of inflammation and endothelial
“Coffee, caffeine, and risk of type 2 diabetes: a prospective dysfunction in healthy and diabetic women,” American Journal
cohort study in younger and middle-aged U.S. women,” of Clinical Nutrition, vol. 84, no. 4, pp. 888–893, 2006.
Diabetes Care, vol. 29, no. 2, pp. 398–403, 2006. [20] C. J. Williams, J. L. Fargnoli, J. J. Hwang et al., “Coffee
[5] M. A. Pereira, E. D. Parker, and A. R. Folsom, “Coffee con- consumption is associated with higher plasma adiponectin
sumption and risk of type 2 diabetes mellitus: an 11-year concentrations in women with or without type 2 diabetes: a
prospective study of 28 812 postmenopausal women,” Archives prospective cohort study,” Diabetes Care, vol. 31, no. 3, pp.
of Internal Medicine, vol. 166, no. 12, pp. 1311–1316, 2006. 504–507, 2008.
[6] T. E. Graham, P. Sathasivam, M. Rowland, N. Marko, F. Greer, [21] A. Zampelas, D. B. Panagiotakos, C. Pitsavos, C. Chrysohoou,
and D. Battram, “Caffeine ingestion elevates plasma insulin and C. Stefanadis, “Associations between coffee consumption
response in humans during an oral glucose tolerance test,” and inflammatory markers in healthy persons: the ATTICA
Canadian Journal of Physiology and Pharmacology, vol. 79, no. study,” American Journal of Clinical Nutrition, vol. 80, no. 4,
7, pp. 559–565, 2001. pp. 862–867, 2004.
[7] L. E. Robinson, S. Savani, D. S. Battram, D. H. McLaren, P. [22] E. J. Erlandsen and E. Randers, “Reference interval for serum
Sathasivam, and T. E. Graham, “Caffeine ingestion before an C-reactive protein in healthy blood donors using the Dade
oral glucose tolerance test impairs blood glucose management Behring N Latex CRP mono assay,” Scandinavian Journal of
in men with type 2 diabetes,” Journal of Nutrition, vol. 134, no. Clinical and Laboratory Investigation, vol. 60, no. 1, pp. 37–44,
10, pp. 2528–2533, 2004. 2000.
[8] H. J. Petrie, S. E. Chown, L. M. Belfie et al., “Caffeine ingestion [23] Y. Arita, S. Kihara, N. Ouchi et al., “Paradoxical decrease of an
increases the insulin response to an oral-glucose-tolerance test adipose-specific protein, adiponectin, in obesity,” Biochemical
in obese men before and after weight loss,” American Journal and Biophysical Research Communications, vol. 257, no. 1, pp.
of Clinical Nutrition, vol. 80, no. 1, pp. 22–28, 2004. 79–83, 1999.
[9] D. S. Battram, R. Arthur, A. Weekes, and T. E. Graham, “The [24] Y. Nakano, S. Tajima, A. Yoshimi et al., “A novel enzyme-
glucose intolerance induced by caffeinated coffee ingestion is linked immunosorbent assay specific for high-molecular-
less pronounced than that due to alkaloid caffeine in men,” weight adiponectin,” Journal of Lipid Research, vol. 47, no. 7,
Journal of Nutrition, vol. 136, no. 5, pp. 1276–1280, 2006. pp. 1572–1582, 2006.
[10] L. L. Moisey, S. Kacker, A. C. Bickerton, L. E. Robinson, and [25] O. O. Akinyinka, A. Sowunmi, R. Honeywell, and A. G.
T. E. Graham, “Caffeinated coffee consumption impairs blood Renwick, “The pharmacokinetics of caffeine in Nigerian
glucose homeostasis in response to high and low glycemic children suffering from malaria and kwashiorkor,” European
index meals in healthy men,” American Journal of Clinical Journal of Clinical Pharmacology, vol. 56, no. 2, pp. 153–158,
Nutrition, vol. 87, no. 5, pp. 1254–1261, 2008. 2000.
[11] F. Greer, R. Hudson, R. Ross, and T. Graham, “Caffeine inges- [26] T. Okada, D. Kobayashi, S. Kono, and T. Shimazoe, “Clinical
tion decreases glucose disposal during a hyperinsulinemic- research on improvement of glucose metabolic marker level
euglycemic clamp in sedentary humans,” Diabetes, vol. 50, no. by coffee drinking—validity of saliva caffeine concentration
10, pp. 2349–2354, 2001. measurement,” Yakugaku Zasshi, vol. 130, no. 5, pp. 713–715,
[12] G. B. Keijzers, B. E. De Galan, C. J. Tack, and P. Smits, “Caffeine 2010.
can decrease insulin sensitivity in humans,” Diabetes Care, vol. [27] M. Matsuda and R. A. DeFronzo, “Insulin sensitivity indices
25, no. 2, pp. 364–369, 2002. obtained from oral glucose tolerance testing: comparison with
[13] Y. Yoshida, M. Hayakawa, and E. Niki, “Evaluation of the the euglycemic insulin clamp,” Diabetes Care, vol. 22, no. 9, pp.
antioxidant effects of coffee and its components using the 1462–1470, 1999.
biomarkers hydroxyoctadecadienoic acid and isoprostane,” [28] D. R. Matthews, J. P. Hosker, and A. S. Rudenski, “Homeostasis
Journal of Oleo Science, vol. 57, no. 12, pp. 691–697, 2008. model assessment: insulin resistance and β-cell function from
[14] E. M. Marinova, A. Toneva, and N. Yanishlieva, “Comparison fasting plasma glucose and insulin concentrations in man,”
of the antioxidative properties of caffeic and chlorogenic Diabetologia, vol. 28, no. 7, pp. 412–419, 1985.
acids,” Food Chemistry, vol. 114, no. 4, pp. 1498–1502, 2009. [29] J. A. Greenberg, C. N. Boozer, and A. Geliebter, “Coffee,
[15] T. MacKenzie, R. Comi, P. Sluss et al., “Metabolic and hor- diabetes, and weight control,” American Journal of Clinical
monal effects of caffeine: randomized, double-blind, placebo- Nutrition, vol. 84, no. 4, pp. 682–693, 2006.
controlled crossover trial,” Metabolism, vol. 56, no. 12, pp. [30] S. Rudelle, M. G. Ferruzzi, I. Cristiani et al., “Effect of
1694–1698, 2007. a thermogenic beverage on 24-hour energy metabolism in
[16] R. M. Van Dam, W. J. Pasman, and P. Verhoef, “Effects of coffee humans,” Obesity, vol. 15, no. 2, pp. 349–355, 2007.
consumption on fasting blood glucose and insulin concen- [31] E. Lopez-Garcia, R. M. Van Dam, S. Rajpathak, W. C. Willett, J.
trations: randomized controlled trials in healthy volunteers,” E. Manson, and F. B. Hu, “Changes in caffeine intake and long-
Diabetes Care, vol. 27, no. 12, pp. 2990–2992, 2004. term weight change in men and women,” American Journal of
[17] N. M. Wedick, A. M. Brennan, Q. Sun, F. B. Hu, C. S. Clinical Nutrition, vol. 83, no. 3, pp. 674–680, 2006.
Mantzoros, and R. M. van Dam, “Effects of caffeinated and [32] T. Yamaji, T. Mizoue, S. Tabata et al., “Coffee consumption
decaffeinated coffee on biological risk factors for type 2 and glucose tolerance status in middle-aged Japanese men,”
diabetes: a randomized controlled trial,” Nutrition Journal, vol. Diabetologia, vol. 47, no. 12, pp. 2145–2151, 2004.
10, no. 93, pp. 1–9, 2011. [33] R. M. Van Dam, J. M. Dekker, G. Nijpels, C. D. A. Stehouwer,
[18] K. Kempf, C. Herder, I. Erlund et al., “Effects of coffee con- L. M. Bouter, and R. J. Heine, “Coffee consumption and inci-
sumption on subclinical inflammation and other risk factors dence of impaired fasting glucose, impaired glucose tolerance,
Journal of Nutrition and Metabolism 9

and type 2 diabetes: the Hoorn Study,” Diabetologia, vol. 47,

no. 12, pp. 2152–2159, 2004.
[34] H. Hemmerle, H. J. Burger, P. Below et al., “Chlorogenic acid
and synthetic chlorogenic acid derivatives: novel inhibitors of
hepatic glucose-6-phosphate translocase,” Journal of Medicinal
Chemistry, vol. 40, no. 2, pp. 137–145, 1997.
[35] K. L. Johnston, M. N. Clifford, and L. M. Morgan, “Coffee
acutely modifies gastrointestinal hormone secretion and glu-
cose tolerance in humans: glycemic effects of chlorogenic acid
and caffeine,” American Journal of Clinical Nutrition, vol. 78,
no. 4, pp. 728–733, 2003.
[36] J. Shearer, E. A. Sellars, A. Farah, T. E. Graham, and D.
H. Wasserman, “Effects of chronic coffee consumption on
glucose kinetics in the conscious rat,” Canadian Journal of
Physiology and Pharmacology, vol. 85, no. 8, pp. 823–830, 2007.
[37] B. Cheng, X. Liu, H. Gong et al., “Coffee components inhibit
amyloid formation of human islet amyloid polypeptide in
vitro: possible link between coffee consumption and diabetes
mellitus,” Journal of Agricultural and Food Chemistry, vol. 59,
no. 24, pp. 13147–13155, 2011.
[38] J. D. Lane, M. N. Feinglos, and R. S. Surwit, “Caffeine
increases ambulatory glucose and postprandial responses in
coffee drinkers with type 2 diabetes,” Diabetes Care, vol. 31,
no. 2, pp. 221–222, 2008.
[39] J. D. Krebs, A. Parry-Strong, M. Weatherall, R. W. Carroll, and
M. Downie, “A cross-over study of the acute effects of espresso
coffee on glucose tolerance and insulin sensitivity in people
with type 2 diabetes mellitus,” Metabolism, vol. 61, no. 9, pp.
1231–1237, 2012.
[40] X. M. Han, D. S. Ou-Yang, P. X. Lu et al., “Plasma caffeine
metabolite ratio (17X/137X) in vivo associated with G-2964A
and C734A polymorphisms of human CYP1A2,” Pharmacoge-
netics, vol. 11, no. 5, pp. 429–435, 2001.
Copyright of Journal of Nutrition & Metabolism is the property of Hindawi Publishing Corporation and its
content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.