The effects of coffee consumption on lipid peroxidation and plasma total homocysteine concentrations: a clinical trial

Free Radical Biology & Medicine 38 (2005) 527 – 534
www.elsevier.com/locate/freeradbiomed
Original Contribution
The effects of coffee consumption on lipid peroxidation and plasma total
homocysteine concentrations: a clinical trial$
Jaakko Mursua,*, Sari Voutilainena, Tarja Nurmia, Georg Alfthanb, Jyrki K. Virtanena,
Tiina H. Rissanena,c, Pertti Happonena, Kristiina Nyyssfnena, Jari Kaikkonena,d,
Riitta Salonena, Jukka T. Salonena,d
aResearch Institute of Public Health, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland
bBiomarker Laboratory, National Public Health Institute, Helsinki, Finland
cDepartment of Public Health and General Practice, University of Kuopio, Kuopio, Finland
dJurilab Ltd, Kuopio, Finland
Received 14 September 2004; revised 16 November 2004; accepted 16 November 2004
Available online 8 December 2004
Abstract
Despite extensive research, the cardiovascular effects of coffee consumption in humans remain controversial. Our aim was to investigate
the excretion of coffee phenols and the effects of filtered coffee consumption on oxidative stress and plasma homocysteine (tHcy)
concentration in humans. The study consisted of a multiple-dose clinical supplementation trial and a single-dose study. In the long-term trial,
43 healthy nonsmoking men optionally consumed daily either no coffee, 3 cups (450 mL), or 6 cups (900 mL) of filtered coffee for 3 weeks,
while in the short-term study 35 subjects consumed a single dose of 0, 1 (150 mL), or 2 cups (300 mL) of coffee. Long-term consumption of
coffee increased the urinary excretion of caffeic and ferulic acid. The change in the total excretion of phenolic acids in 3 and 6 cups groups
represented 3.8 and 2.5% of the amount ingested daily. Plasma tHcy concentrations increased nonsignificantly, but the consumption of coffee
had neither short-nor long-term effects on lipid peroxidation or the activity of measured antioxidant enzymes. In conclusion, the consumption
of filtered coffee does not have any detectable effects on lipid peroxidation in healthy nonsmoking men. The effect of coffee consumption on
tHcy concentrations needs further investigation.
D 2004 Elsevier Inc. All rights reserved.
Keywords: Coffee; Phenols; Lipid peroxidation; Plasma total homocysteine; Humans; Free radicals
Introduction
Coffee drinking is among the most widespread habits in
the world and thus its effects on health may be highly
relevant for public health. During the past three decades, a
Abbreviations: ALAT, alanine aminotransferase; ASAT, aspartate
wealth of epidemiological studies addressing the role of
aminotransferase; BMI, body mass index; CHD, coronary heart disease;
coffee drinking in coronary heart disease (CHD) etiology
g-GT, gamma-glutamyltransferase; GPX, plasma glutathione peroxidase
has been published with controversial results from different
activity; HCL, hydrochloric acid; MeOH, methanol; PBS, phosphate-
buffered saline; PON, serum paraoxonase; tHcy, total homocysteine; TRAP,
populations [1]. Coffee drinking is one of many behavioral
total peroxyl radical trapping potential.
traits that are associated with each other, making it difficult
$ The present study was supported by grants from the Institute for
to differentiate the effects of coffee from other dietary and
Scientific Information on Coffee, ISIC, Physiological Effects of Coffee
behavioral factors.
Committee, PEC, Juho Vainio Foundation (J.M.), Yrjf Jahnsson Founda-
The current opinion is that the oxidative modification
tion (J.M.) and Academy of Finland (S.V., K.N., J.K., J.T.S.).
* Corresponding author. Fax: +358 17 162936.
of LDL plays an important role in the pathogenesis of
E-mail address: jaakko.mursu@uku.fi (J. Mursu).
atherosclerosis [2–4]. Recently, coffee has aroused scien-
0891-5849/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.freeradbiomed.2004.11.025

---

528
J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
tific interest because it is a rich source of a number of
Study design
phenolic compounds with antioxidant effects in vitro
[5,6]. Main polyphenols in coffee are phenolic acids such
Long-term study
as chlorogenic and caffeic acid. Caffeine and its
A 2-week run-in period preceded the long-term study and
metabolites, di-and monomethylxanthines, have also been
during this period the use of coffee, tea, red wine, cocoa, and
shown to have antioxidative properties [7]. Recent studies
chocolate was forbidden. In addition the intake of fruit-and
also suggest that coffee phenols are absorbed and
berry-derived juices was restricted to a maximum of 2 glasses
ingestion of single dose of coffee may have antioxidative
(300 mL) per day. Subjects were given caffeine tablets to be
effects in humans [8–10].
used if necessary to deal with withdrawal symptoms. The
In addition to the theoretical beneficial effects, coffee
maximum daily amount of caffeine was the amount
has recently aroused interest also because in several
comparable to that obtained from the daily study bolus in
supplementation studies the consumption of coffee has
the long-term study (0, 300, or 600 mg). After the run-in
increased the concentration of plasma total homocysteine
period subjects were free to choose whether they wanted to
(tHcy) in humans [11–13]. Elevated plasma tHcy
consume 0, 3 (450 mL), or 6 cups (900 mL) of coffee during
concentrations have been associated with increased lipid
the 3-week supplementation period. In order to ensure
peroxidation [14] and it is also suggested to be an
compliance, the study was not randomized. The ingested
independent risk factor for cardiovascular disease [15].
amounts of phenolic acids through coffee were 0, 364, and
Chlorogenic acid and caffeine have been identified to be
728 mg/d in the 0, 3, and 6 cup groups, corresponding to
mainly responsible for the increase of tHcy after coffee
molar amounts of 0, 1122, and 2243 Amol/day.
consumption [11,12].
The subjects were advised to avoid the use of alcohol and
To this date there are only scarce data about the systemic
analgesics 3 days and vigorous physical activity 1 day before
availability of coffee phenols, their antioxidant effects, and
the study visits. The subjects collected a 24-h urine sample
the effects of coffee consumption on plasma tHcy concen-
before study visits and blood samples were drawn with
trations in humans. The aim of the present study was to
Venoject vacuum tubes (Terumo) after an overnight fast
study the short-and long-term effects of coffee consumption
(10 h). A 4-day food record was collected before and during
on serum lipid peroxidation, on the activity of antioxidant
the last week of the intervention period to control for
enzymes, and on plasma tHcy concentrations in healthy
possible confounding factors and to check compliance with
men.
the dietary instructions. The instructions for the food records
were given, checked with the subjects, and analyzed by a
nutritionist using Nutrica software, Social Insurance Insti-
Materials and methods
tution, Helsinki, Finland (version 2.5).
The coffee used in this study was finely ground coffee
Subjects
packed in 500 g packages. Subjects were instructed to
measure the daily amount of coffee (7–8 g of grounds
Forty-five nonsmoking volunteer men with a mean age of
per one 150-mL cup), prepare the coffee by filtering
26 F 6 years were recruited from the Kuopio area in eastern
through paper, and to consume the total daily amount in
Finland through advertisement via e-mail and intranet in the
three separate portions. The coffee used in the study was
University of Kuopio. Potential participants were screened
delivered by Oy Paulig Ab, Helsinki, Finland.
in an interview for the following eligibility criteria: no
severe obesity (Body Mass Index, BMI b 32 kg/m2), no
Short-term study
regular use of any drugs or supplements with antioxidative
The short-term study was conducted directly after
or lipid-lowering properties, no chronic diseases such as
obtaining the blood samples for the supplementation
diabetes, CHD, claudication, cerebrovascular disease, hypo-
period of the long-term study. The subjects remained in
thyroidism, or other major illness, willingness to abstain
the same group as in the long-term study, but consumed
from coffee drinking or to consume 3 or 6 cups of coffee for
a single dose of 1/3 of the total daily dose consumed in
3 weeks. Of these subjects, 35 men participated also in the
the long-term study (0, 1, or 2 cups, 0, 150, or 300 mL,
short-term study of the acute effects of coffee drinking in
respectively). They continued to fast (intake of tap water
which they consumed a single dose of 0, 1, or 2 cups of
was allowed). A blood sample was taken 1.5 h after
coffee. To enhance compliance, volunteers were given the
coffee ingestion. From the blood samples, the following
choice to participate in one or both of the studies. All of the
markers of lipid peroxidation were analyzed: concentra-
mentioned criteria were ascertained prior to study entry.
tion of serum LDL-conjugated dienes, plasma hydroxy
Written informed consent was obtained in writing from all
fatty acids, and plasma F2-isoprostanes.
participants after they had read a description of the
experimental procedures. The study protocol was approved
Resistance of serum lipids to oxidation
by the Research Ethics Committee, Hospital District of
The resistance of serum lipids to oxidation was measured
Northern Savo.
as described previously [16]. Briefly, serum was diluted to a

---

J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
529
concentration of 0.67% in 0.02 mol/L phosphate-buffered
Plasma total homocysteine
saline (PBS), pH 7.4. Oxidation was initiated by addition of
Plasma tHcy concentration was analyzed by high
100 Al of 1 mmol/L CuCl2into 2 mL of diluted, prewarmed
performance liquid chromatography (HPLC) at the National
(308C) serum. The formation of conjugated dienes was
Public Health Institute, Helsinki, Finland, as described
followed by monitoring the change in 234 nm absorbance at
previously [21]. The results represent total plasma homo-
308C on a Beckman DU-6401 spectrophotometer (Full-
cysteine and are referred to as plasma tHcy. The coef-
erton, CA) equipped with a six-position automatic sample
ficients of variation (CV) between batches (n = 30) for two
changer. The change in absorbance was recorded every 5
pooled plasma samples were 5.7% (7.3 Amol/L) and 7.1%
min for 4 h. The time required from the start to the maximal
(10.5 Amol/L).
rate of the reaction (lag time) was determined.
Other measurements
Serum LDL-conjugated dienes
Serum cholesterol (Konelab, Espoo, Finland) and tri-
The oxidation of LDL in vivo was assessed as the
acylglycerides (Roche Diagnostics, Mannheim, Germany)
amount of conjugated dienes as described previously [17].
were determined with enzymatic colorimetric tests. Serum
In brief, serum LDL was isolated by precipitation with
LDL cholesterol was determined with a direct measurement
buffered heparin. The precipitate was resuspended in PBS.
without precipitation (Konelab). Serum HDL cholesterol
Cholesterol concentration was determined and the rest of the
was measured from the supernatant after magnesium
suspension was used for conjugated diene measurement.
chloride dextran sulfate precipitation. Blood cell profile,
Lipids were extracted from the LDL by mixture of chloro-
including erythrocyte, leukocyte, and thrombocyte counts
form and methanol (3:1), dried under nitrogen, and
and hemoglobin, was measured by a blood cell counter
redissolved in cyclohexane, and the amount of conjugated
(Advia 60, Bayer, Tarrytown, NY). Serum aspartate amino-
dienes was measured spectrofotometrically at 234 and 300
transferase (ASAT) and alanine aminotransferase (ALAT)
nm. Absorbance at 300 nm was subtracted from that at 234
were measured with a clinical chemistry analyzer (Konelab).
nm. The conjugated diene concentration was calculated per
The activity of serum gamma-glutamyltransferase (g-GT)
cholesterol concentration in LDL.
was measured with the International Federation of Clinical
Chemistry method [22]. Serum fatty acids were analyzed
Plasma hydroxy fatty acids
after extraction with chloroform–methanol and methylation
Plasma C18 hydroxy fatty acids were measured by gas
with sulphuric acid–methanol by a gas chromatograph
chromatography/mass spectrometry (GC/MS) method [18].
(Hewlett Packard 5890, Avondale, PA) equipped with a
Plasma fatty acids and fatty acid hydroperoxides were first
flame ionization detector and a NB-351 capillary column
stabilized by hydrogenation using platinum as a catalyst,
(HNU-Nordion, Helsinki, Finland). Plasma folate and
saponified, esterified by diazomethane, and finally, to
vitamin B12 were measured simultaneously by radioimmu-
separate plasma hydroxy fatty acids from fatty acids,
noassay (Quantaphase II, Bio-Rad, Hercules, CA).
extracted by solid-phase minicolumns. Prior to analysis,
hydroxy groups were methylated with tetramethylammo-
Phenolic acid analyses of coffee and urine
nium hydroxide. Concentrations of different methoxy
Phenolic acid analyses were carried out with HPLC using
monohydroxy fatty acid methyl esters were determined by
a coulometric electrode array detector. Compounds were
electron impact mass spectroscopy. Monohydroxy fatty
separated with gradient elution using end-capped C 18
acids C17 and C19 were used as internal standards.
column Inertsil ODS-3 (150 Â 3 mm) packed with 3-Am
particles and C 18 guard column (10 Â 3 mm, 5 Am
Plasma F2-isoprostanes
particles). Mobile phase consisted of eluent (A) 50 mM
A deuterated prostaglandin F2a internal standard was
KH2PO4/H3PO4 buffer pH 2.3:MeOH 90:10 (v/v) and (B)
added to plasma, and F2-isoprostanes were extracted with
50 mM KH2PO4/H3PO4 buffer pH 2.3:MeOH:ACN
C18 and silica minicolumns. Compounds were converted to
40:40:20 (v/v/v). Chlorogenic, caffeic, ferulic, p-, m-, o-
pentafluorobenzyl ester trimethylsilylether derivates and
coumaric, sinapinic, protocatechuic, and gallic acid were
analyzed by a GC/MS assay [19].
measured from coffee and from urine; additionally two
metabolites 3,4-dihydroxyphenyl and m-hydroxyphenylace-
Activity of antioxidant enzymes
tic acids were measured. Coffee analysis was carried out
Plasma glutathione peroxidase activity (GPX) was
after the combination of the enzyme and base hydrolysis.
determined by a commercial kit (Ransel RS 505, Randox
Twenty-four-hour urine samples were hydrolyzed with h-
Laboratories, San Diego, CA) by using Konelab 20
glucuronidase and sulfatase obtained from Helix pomatia.
Analyzer (Thermo Clinical Labsystems, Vantaa, Finland).
Hydrolyzed samples were extracted with diethylether,
Serum paraoxonase (PON) activity was measured from
evaporated under N2 flow, and dissolved in MeOH. Samples
serum based on its capacity to hydrolyze paraoxon. The
were diluted with eluent prior to HPLC run.
formation of p-nitrophenol was monitored at 405 nm in
The study coffee contained 80.9 F 3.3 mg/100 mL of
Tris-HCl buffer, pH 8.0, in the presence of Ca2+ [20].
phenolic acids; the major compound was chlorogenic acid,

---

530
J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
Table 1
was used to estimate the contribution of changes in the
Baseline characteristics
concentration of plasma tHcy. SPSS for Windows version
Daily coffee intake
10.0 was used for statistical analyses.
0 cups
3 cups
6 cups
P
(n = 15)
(n = 14)
(n = 14)
(ANOVA)*
Age (years)
23.7 F 3.8
25.4 F 6.0
29.7 F 6.8
0.020
Results
BMI (kg/m2)
23.5 F 3.6
22.8 F 3.5
26.1 F 3.4
0.042
Total
4.1 F 1.0
4.2 F 0.7
4.7 F 0.7
0.148
Out of 45 men recruited, 43 completed the long-term
cholesterol
study; and of these, 35 participated in the short-term study.
(mmol/L)
ASAT (U/L)
22.8 F 6.1
24.1 F 6.3
45.7 F 61.1
0.163
One subject dropped out during the 2-week washout period
ALAT (U/L)
26.1 F 11.0
26.7 F 16.9
51.1 F 42.8
0.027
due to abstinence symptoms from coffee drinking, and one
g-GT (U/L)
17.5 F 7.0
16.9 F 6.4
27.9 F 12.4
0.003
was excluded due to dizziness during the process of
ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; g-GT,
drawing the blood samples. The dropped subjects were not
gamma-glutamyltransferase.
replaced.
* P for the differences in changes between the groups (one-way ANOVA).
Baseline characteristics differed between study groups
(Table 1). Age, BMI, and the activities of ALAT and g-GT
representing approximately 90% of the quantified phenolic
enzymes were significantly higher in the 6 cup group at
acid total. Another abundant phenolic acid was ferulic acid
study baseline when compared with 0 cup and 3 cup groups.
representing 8% of the quantified phenolic acids. Small
However, the difference in the activity of ALAT in the 6 cup
amounts of p-coumaric, caffeic, protocatechuic, and sina-
group was due to a high concentration in two persons (149
pinic acid were also detected.
and 140 U/L) and the activities were not different between
the groups at baseline if these subjects were excluded from
Statistical analyses
the analyzes ( P = 0.237). During the run-in period, 19 men
reported withdrawal symptoms (no specific symptoms were
Means between study groups at baseline were compared
studied), but no adverse effects because of coffee con-
by analysis of variance (ANOVA). The changes of means
sumption during the supplementation period were reported
between study groups were compared by multivariate
or detected in the safety measurements (ALAT, ASAT, or g-
analysis of variance (MANOVA). Because of the differ-
GT). According to two 4-day food recordings, the intake of
ence in the age and BMI between study groups in the
nutrients did not differ between the study groups during the
baseline, age, BMI, and baseline values of the parameter
study.
tested were used as covariates. The post hoc Tukey test
The consumption of coffee increased the concentration of
was used whenever statistically significant heterogeneity
coffee phenols and their metabolites in urine (Table 2).
between groups was shown by the MANOVA. The results
Average total excretion of phenolic acids was 123 (0 cups),
are displayed as means and standard deviations (SD).
109 (3 cups), and 101 (6 cups) Amol/day at baseline and
Differences with a P value of 0.05 or less were considered
126, 152, and 157 Amol/day after the 3-week ingestion of
statistically significant. Stepwise linear regression analysis
coffee, respectively. The increases in the concentrations of
Table 2
Urinary excretion of coffee phenols and metabolites at baseline and change after 3-week supplementation perioda
Daily coffee intake
P (MANOVA)b
0 cups (n = 15)
3 cups (n = 14)
6 cups (n = 14)
Baseline
Change
Baseline
Change
Baseline
Change
Caffeic acid
4.1 F 2.4
0.0 F 1.8
3.3 F 2.0
14.4 F 9.2
2.5 F 1.1
23.1 F 8.7
b0.001
(Amol/day)
Ferulic acid
24.8 F 8.0
À1.8 F 9.7
24.7 F 13.9
20.2 F 18.7
22.8 F 12.8
41.5 F 21.3
b0.001
(Amol/day)
p-Coumaric acid
2.7 F 3.7
À0.7 F 3.6
2.8 F 3.3
À0.4 F 2.3
1.8 F 1.5
À0.1 F 1.5
0.474
(Amol/day)
Protocatechuic acid
6.1 F 4.2
À0.1 F 4.2
4.4 F 1.7 (13)
5.4 F 3.7 (13)
4.2 F 2.4 (13)
9.7 F 3.2 (13)
b0.001
(Amol/day)
3,4-Dihydroxyphenylacetic
25.7 F 4.9
0.1 F 4.3
22.8 F 7.5 (12)
12.3 F 15.3 (12)
21.9 F 7.4 (9)
7.1 F 6.8 (9)
0.012
acid (Amol/day)
m-Hydroxyphenylacetic
59.8 F 28.4
5.3 F 33.1
51.7 F 33.1
À3.8 F 38.3
48.4 F 23.0
À16.4 F 17.0
0.061
acid (Amol/day)
a Mean F SD; number of subjects except where otherwise indicated in parentheses.
b In the MANOVA age, BMI at baseline and the baseline value of the urinary concentration of phenolic acid tested were used as covariates. P for differences in
the changes between the groups.

---

J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
531
caffeic, ferulic, protocatechuic, and 3,4 dihydroxyphenyl-
subjects in the 3 cup group and five subjects in the 6
acetic acids were significantly different between the study
cup group excreted after 3 weeks some unknown analyte,
groups ( P b 0.001 for the caffeic, ferulic, and protocate-
which coeluted with 3,4-dihydroxyphenylacetic acid and
chuic acids and P b 0.05 for the 3,4-dihydroxyphenolacetate
therefore it was not possible to quantify that metabolite
acids). The change in the total excretion of phenolic acids in
from all the subjects.
3 and 6 cups groups represented only 3.8 and 2.5% of the
The consumption of filtered coffee did not have long-
daily-ingested amounts of phenolic acids.
term effects on serum lipids or short-or long-term effects on
In both coffee groups the average increase in the caffeic
lipid peroxidation or on the activity of antioxidant enzymes
and ferulic acid excretion corresponded well to the
(Tables 3 and 4). Plasma tHcy concentration increased
increase in the average excretion of the total phenolic
nonsignificantly by 5, 16, and 26% in the 0, 3, and 6 cup
acids. In the 3 and 6 cup groups, on average 29 and 21%
groups, respectively ( P = 0.102) (Fig. 1). However, because
of the daily ingested ferulic acid were excreted into urine.
the mean age and BMI were different between study groups
The amount of m-hydroxyphenylacetic acid decreased in
at the study baseline, we adjusted the change in the plasma
the supplementation groups, but increased among those not
tHcy for age, BMI, and the baseline concentration of tHcy in
consuming any coffee. The amount of 3,4-dihydroxyphe-
the MANOVA. After adjustment the trend in the difference
nylacetic acid increased in both coffee groups. Two
of plasma tHcy change between study groups disappeared
Table 3
Concentrations of serum lipids, oxidation products of lipids, activity of antioxidant enzymes, and plasma total homocysteine (tHcy) at baseline and change after
3-week supplementation perioda
Daily coffee intake
P (MANOVA)b
0 cups (n = 15)
3 cups (n = 14)
6 cups (n = 14)
Baseline
Change
Baseline
Change
Baseline
Change
Serum LDL
2.2 F 0.8
À0.0 F 0.3
2.3 F 0.6
À0.0 F 0.4
2.6 F 0.5
À0.0 F 0.5
0.948
cholesterol
(mmol/L)
Serum HDL
1.0 F 0.1
À0.0 F 0.1
1.1 F 0.2
0.0 F 0.1
1.1 F 0.2
0.0 F 0.1
0.324
cholesterol
(mmol/L)
Serum
1.1 F 0.7
À0.0 F 0.6
1.1 F 0.6
À0.1 F 0.3
1.3 F 0.6
0.1 F 0.7
0.496
triacylglycerol
(mmol/L)
Serum lipid
210 F 23.0
À13.7 F 28.8
199.6 F 42.7
6.1 F 45.8
210.4 F 53.6
À14.6 F 51.6
0.602
peroxidation
resistance
(lagtime, min)
Serum
14.8 F 3.6
1.3 F 3.8
16.5 F 4.5
À0.3 F 4.3
16.0 F 2.7
À0.8 F 3.8
0.597
LDL-conjugated
dienes
(Amol/mmol chol)
Plasma
32.0 F 7.3
À0.1 F 4.9
30.1 F 3.0
1.7 F 7.4
31.6 F 9.4
À0.8 F 7.5
0.853
F2-isoprostanes
(pg/mL)
Plasma hydroxy
0.70 F 0.18
À0.01 F 0.25
0.77 F 0.18
À0.04 F 0.21
0.75 F 0.17
0.09 F 0.18
0.374
fatty acids
(Amol/L)
Plasma GPX
830.4 F 134.2
23.3 F 83.0
920.4 F 124.6
20.4 F 88.1
862.5 F 115.6
61.4 F 109.1
0.664
(U/L)c
Serum PON
105.3 F 63.2
À1.0 F 7.2
114.5 F 82.3
À0.6 F 7.9
114.3 F 82.6 (12)
2.4 F 13.0 (12) 0.542
(U/L)c
Plasma folate
6.7 F 2.1 (14)
À0.4 F 1.3 (14)
7.8 F 2.2 (14)
0.1 F 3.9 (14)
6.5 F 1.6 (13)
À0.4 F 0.8 (13)
0.593
(nmol/L)
Plasma
368.9 F 242.9
28.6 F 210.8
436.9 F 154.5
29.1 F 116.2
369.3 F 88.1 (13) À21.1 F 75.2 (13) 0.955
vitamin B12
(pmol/L)
Plasma tHcy
11.4 F 9.4
0.5 F 1.2
8.6 F 2.2
1.4 F 1.9
11.7 F 3.5 (12)
3.0 F 5.0 (12)
0.494
(Amol/L)c
a Mean F SD; Number of subjects except where otherwise indicated in parentheses.
b In the MANOVA age, BMI at baseline and the baseline value of the parameter tested were used as a covariate. P for differences in the changes between the
groups.
c GPX, glutathione peroxidase; PON, paraoxonase; tHcy, total homocysteine.

---

532
J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
Table 4
Oxidation products of lipids before and 1.5 h after consumption of coffeea
Daily coffee intake
P (MANOVA)b
0 cups (n = 10)
1 cup (n = 12)
2 cups (n = 13)
Baseline
Change
Baseline
Change
Baseline
Change
Serum LDL-conjugated dienes
16.2 F 4.8
À1.1 F 3.7
16.2 F 5.0
À1.4 F 3.8 (11)
15.2 F 2.9
À0.4 F 1.5
0.630
(Amol/mmol chol)
Plasma hydroxy fatty acids (Amol/L)
0.69 F 0.19
0.01 F 0.18
0.73 F 0.28
0.07 F 0.23
0.84 F 0.25
0.07 F 0.25
0.075
Plasma F2-isoprostanes (pg/mL)
31.9 F 9.0
0.9 F 5.0
31.8 F 8.1
1.5 F 6.1
30.8 F 9.4
2.5 F 2.5
0.853
a Mean F SD; number of subjects except where otherwise indicated in parentheses.
b In the MANOVA age, BMI at baseline and the baseline value of the paremeter tested were used as covariates. P for differences in the changes between the
groups.
( P = 0.494). The consumption of coffee did not have an
This was despite the fact that during the study the intake of
effect on plasma concentrations of folate or B12 vitamin.
coffee (other than used in the study), tea, red wine,
Even though the changes in the tHcy concentration were
chocolate, and cocoa was forbidden and only moderate
not significantly different between the study groups, we
consumption (b300 mL/day) of fruit-and berry-derived
analyzed the contribution of the different variables on the
juices was allowed.
increase in the plasma tHcy concentrations with a linear
In previous studies both chlorogenic and caffeic acids
regression model. The strongest associations with the change,
have been shown to be absorbed relatively effectively (33
selected by stepwise analysis ( P in 0.010, P out 0.15), were
and 95%, respectively) in the small intestine of ileostomy
the change in urinary caffeic acid (standardized coefficient
subjects and caffeic acid has also been found in human
0.65, P b 0.001), urinary 3,4-dihydroxyphenylacetic acid
plasma (peak at 1 h) and 11% of the caffeic acid has been
(–0.39, P = 0.013), and the change in serum PON activity
found in urine [8,9]. However, only traces of chlorogenic
(0.29, P = 0.081) (adjusted R 2 for the model 0.38, P = 0.001).
acid have been detected in human plasma and urine. This is
We also analyzed the correlation coefficient between the
because chlorogenic acid is effectively hydrolyzed by
changes in the plasma tHcy concentration and urine
colonic microflora into quinic acid and caffeic acid and
concentrations of phenolic acids. The simple correlation
further decarboxylated to benzoic acid [23–25]. The quinic
coefficient for the association between changes in plasma
acid moiety is further dehydroxylated into cyclohexane
tHcy and urinary caffeic acid was 0.44 (P = 0.004), between
carboxylic acid and then aromatized into benzoic acid by
plasma tHcy and urinary ferulic acid 0.39 (P = 0.012) and
microflora and excreted as hippuric acid. Hippuric acid can
between changes in plasma tHcy and urinary p-coumaric
be formed also from caffeic acid.
acid 0.30 ( P = 0.060). Other correlations were weak and
In our study the consumption of coffee did not have
nonsignificant.
any detectable short-or long-term effect on different
measurements of lipid peroxidation or on the activity of
antioxidant enzymes in healthy men. The amount of
Discussion
coffee used in our short-term study was 150 or 300 mL
and in the long-term study 450 or 900 mL per day and
Coffee is the most consumed beverage in the world and
thus its effects on health may be highly relevant for public
health. Despite extensive research the role of coffee in
human health remains equivocal [1]. The aim of the present
study was to test the excretion of coffee phenols and the
short-and long-term effects of coffee consumption on lipid
peroxidation, activity of antioxidant enzymes, and plasma
tHcy concentration in healthy men.
We found that the consumption of filtered coffee
increased the 24-h urinary excretion of caffeic, ferulic,
protocatechuic, and 3,4-dihydroxyphenylacetic acids. The
major changes of excretion were observed for caffeic and
ferulic acids, but no chlorogenic acid was detected from the
urine, even though it represents 90% of the coffee phenols.
In general, coffee consumption affected the excretion of the
phenolic compounds only modestly as the consumption of 6
cups of coffee (~700 mg of phenolic compunds) per day led
Fig. 1. The effects of long-term coffee supplementation on plasma total
only to a 55% increase in the excretion of total phenolics.
homocysteine.

---

J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
533
thus delivered 120–700 mg of phenolic acids. We assume
In previous studies, the consumption of 450–1000 mL of
that this amount should have been sufficient to induce a
coffee has increased the plasma tHcy concentration by 19%
change in the lipid peroxidation, if any clinically
at 4 h and 11–22% after 2 to 4 weeks [12,13], while in a
important effect would exist.
recent short-term study consumption of coffee did not
We measured lipid peroxidation using a wide variety
increase the concentration of tHcy [10]. It has been
of methods: the susceptibility of whole serum lipids to
previously found that chlorogenic acid and caffeine are
oxidation, concentration of serum LDL baseline-conju-
mostly responsible for the increase in tHcy, but the
gated dienes (in vivo), plasma hydroxy fatty acids, and
mechanism behind the increase is not known [11,12]. The
plasma F2-isoprostanes. The last three measurements are
tHcy increasing effect of caffeine was not known at the time
in vivo markers of lipid peroxidation and of these,
when we conducted this study (February 2002) and coffee
plasma F2-isoprostanes are often considered to be the
groups, but not the 0 cup group, were given caffeine in order
most reliable marker of lipid peroxidation in humans
to ensure compliance. This may have affected the baseline
[19,26,27]. The short-term study concerned the effects on
concentrations of tHcy, even though the baseline values
lipid peroxidation after 3 weeks of coffee supplementa-
were not different between the study groups (Table 3).
tion and we did not measure the susceptibility of whole
According to previous studies and confirmed by our study,
serum lipids to oxidation or the activity of antioxidant
coffee consumption does not have an effect on the
enzymes (GPX and PON). Therefore it is possible that
concentrations of folate, B6 or B12 vitamins [11–13].
some of the short-term effects could have remained
Elevated plasma tHcy concentrations have been associ-
undetected.
ated with an increased risk of cardiovascular diseases in
Previous data on the effects of coffee consumption on
many case-control studies and in some, but not all
lipid peroxidation are very limited. The results of a recent
prospective cohort studies [15]. Even though coffee
study suggest that coffee might increase the antioxidant
consumption would increase the tHcy concentration, there
potential for a few hours in humans [10]. In that study the
is no convincing evidence that coffee consumption would
ingestion of 200 mL of coffee increased the antioxidant
increase the risk of CVD [1]. The role of tHcy in CVD
capacity by 6 and 7% at 1 h when measured as a crocin test
remains controversial and the relevance of the increase in
and total radical antioxidant potential (TRAP), respectively.
the concentration of tHcy needs to be studied further.
We did not use measurements of antioxidant capacity in our
A weakness of the current study was that the study
study and it is possible that coffee consumption could have
was not randomized and double-blinded. Self-selection of
had detectable effects in these measurements at least in the
study groups was used because people who normally
short-term study.
consume coffee daily were not willing to discontinue the
The effects of coffee consumption on the in vivo markers
use and vice versa, those who do not habitually consume
of lipid peroxidation have, at least to our knowledge, not
coffee were not willing to drink daily as high an amount
been studied previously. The effects of other phenolic-rich
as 6 cups of coffee. Selection of study group according
foodstufts, such as tea, on lipid peroxidation have, however,
to the previous coffee drinking habits resulted in differ-
been studied more extensively. A substantial number of
ences in the baseline characteristics as the men in the 6
supplementation studies have demonstrated that tea
cup group were older and their BMI as well as ALAT
increases antioxidant capacity for ~1 h, but the effects on
and g-GT concentrations were higher (Table 1). This may
oxidative damage are inconsistent [28]. Studies which have
have had an effect on our results, even though we
included in vivo markers of oxidative stress such as F2-
included age, BMI, and the baseline values of the
isoprostanes have not found any effect [29,30]. Coffee
particular variable tested as covariates in our statistical
contains similar or slightly higher total amounts of phenolic
models. There were, however, no differences between the
compounds as tea (coffee, 70–350 mg/cup; and tea, 12–160
study groups in the baseline levels of oxidation markers
mg/cup), even though coffee is rich in simple phenols such
or tHcy, and we do think that we should have detected
as chlorogenic and caffeic, whereas tea is rich is poly-
the effect in those parameters if a true effect had existed.
phenolic compounds such as catechins and teaflavins
We conclude that in healthy nonsmoking men, con-
[5,6,31]. It is possible that phenolic compounds in coffee
sumption of fair amounts of filtered coffee does not have
may possess similar short-term antioxidant capacity raising
any detectable short-or long-term effects on lipid perox-
effects as tea.
idation or on the activity of the main antioxidative enzymes.
In our study tHcy concentrations increased by 5–26% but
The effect of coffee consumption on the concentration of
these changes were not significantly different between the
plasma tHcy needs further investigation.
study groups. The mean changes were mainly results of
large increases in tHcy concentration in few participants as
can be seen from Fig. 1. Therefore our results do not
Acknowledgments
provide any unequivocal evidence that coffee would
increase tHcy concentrations, but an increase in the tHcy
The authors thank the volunteers for their participation,
concentration in plasma cannot be ruled out.
our lab staff at the Research Institute of Public Health,

---

534
J. Mursu et al. / Free Radical Biology & Medicine 38 (2005) 527–534
University of Kuopio, Finland, and Oy Jurilab Ltd
[17] Ahotupa, M.; Marniemi, J.; Lehtimaki, T.; Talvinen, K.; Raitakari,
(www.jurilab.com) for performing blood analyses.
O. T.; Vasankari, T.; Viikari, J.; Luoma, J.; Yla-Herttuala, S.
Baseline diene conjugation in LDL lipids as a direct measure of in
vivo LDL oxidation. Clin. Biochem. 31:257 – 261; 1998.
[18] Wilson, R.; Smith, R.; Wilson, P.; Shepherd, M. J.; Riemersma, R. A.
References
Quantitative gas chromatography-mass spectrometry isomer-specific
measurement of hydroxy fatty acids in biological samples and food
[1] Thelle, D. S. Coffee, tea and coronary heart disease. Curr. Opin.
as a marker of lipid peroxidation. Anal. Biochem. 248 76 – 85;
Lipidol. 6:25 – 27; 1995.
1997.
[2] Steinberg, D.; Parthasarathy, S.; Carew, T. E.; Witztum, J. L.
[19] Morrow, J. D.; Roberts, L. J. Mass spectrometry of prostanoids: F2-
Modifications of low-density lipoprotein that increase its atheroge-
isoprostanes produced by non-cyclooxygenase free radical-catalyzed
necity. N. Engl. J. Med. 320:915 – 924; 1989.
mechanism. Methods Enzymol. 233:163 – 174; 1994.
[3] Salonen, J. T.; Yl7-Herttuala, S.; Yamamoto, R.; Butler, S.; Korpela,
[20] Mackness, M. I.; Harty, D.; Bhatnagar, D.; Winocour, P. H.; Arrol, S.;
H.; Salonen, R.; Nyyssonen, K.; Palinski, W.; Witztum, J. L.
Ishola, M.; Durrington, P. N. Serum paraoxonase activity in familial
Autoantibody against oxidised LDL and progression of carotid
hypercholesterolaemia and insulin-dependent diabetes mellitus. Athe-
atherosclerosis. Lancet 339:883 – 887; 1992.
rosclerosis 86:193 – 199; 1991.
[4] Salonen, J. T.; Nyyssfnen, K.; Salonen, R.; Porkkala-Sarataho, E.;
[21] Schwab, U.; Tfrrfnen, A.; Toppinen, L.; Alfthan, G.; Saarinen, M.;
Tuomainen, T. P.; Diczfalusy, U.; Bjorkhem, I. Lipoprotein oxidation
Aro, A.; Uusitupa, M. Betaine supplementation with a low calorie diet
and progression of carotid atherosclerosis. Circulation 95:840 – 845;
decreases plasma homocysteine concentrations but does not affect
1997.
body weight, body composition, resting energy expenditure or serum
[5] Scalbert, A.; Williamson, G. Dietary intake and bioavailability of
lipid concentrations. Am. J. Clin. Nutr. 76:961 – 967; 2002.
polyphenols. J. Nutr. 130:2073 – 2085; 2000.
[22] Shaw, L. M.; Stromme, J. H.; London, J. L.; Theodorsen, L.
[6] Mazur, W. M.; W7h7l7, K.; Rasku, S.; Salakka, A.; Hase, T.;
International Federation of Clinical Chemistry. Scientific Committee,
Adlercreutz, H. Lignan and isoflavonoid concentrations in tea and
Analytical Section. Expert Panel on Enzymes. IFCC methods for
coffee. Br. J. Nutr. 79:37 – 45; 1998.
measurement of enzymes. Part 4. IFCC methods for gamma-
[7] Lee, C. Antioxidant ability of caffeine and its metabolites based on the
glutamyltransferase [(gamma-glutamyl)-peptide: amino acid gamma-
study of oxygen radical absorbing capacity and inhibition of LDL
glutamyltransferase, EC 2.3.2.2]. IFCC Document, Stage 2, Draft 2,
peroxidation. Clin. Chim. Acta 295:141 – 154; 2000.
1983-01 with a view to an IFCC Recommendation. J. Clin. Chem.
[8] Olthof, M. R.; Hollman, P. C.; Katan, M. B. Chlorogenic acid and
Clin. Biochem. 10:633 – 646; 1983.
caffeic acid are absorbed in humans. J. Nutr. 131:66 – 71; 2001.
[23] Choudhury, R.; Srai, S. K.; Debnam, E.; Rice-Evans, C. A.
[9] Nardini, M.; Cirillo, E.; Natella, F.; Scaccini, C. Absorption of
Urinary excretion of hydroxycinnamates and flavonoids after
phenolic acids in humans after coffee consumption. J. Agric. Food.
oral and intravenous administration. Free Radic. Biol. Med. 27
Chem. 50:5735 – 5741; 2002.
278 – 286; 1999.
[10] Natella, F.; Nardini, M.; Giannetti, I.; Dattilo, C.; Scaccini, C. Coffee
[24] Olthof, M. R.; Hollman, P. C.; Buijsman, M. N.; van Amelsvoort,
drinking influences plasma antioxidant capacity in humans. J. Agric.
J. M.; Katan, M. B. Chlorogenic acid, quercetin-3-rutinoside and
Food Chem. 50:6211 – 6216; 2002.
black tea phenols are extensively metabolized in humans. J. Nutr. 133:
[11] Olthof, M. R.; Hollman, P. C.; Zock, P. L.; Katan, M. B. Consumption
1806 – 1814; 2003.
of high doses of chlorogenic acid, present in coffee, or of black tea
[25] Gonthier, M. P.; Verny, M. A.; Besson, C.; Remesy, C.; Scalbert, A.
increases plasma total homocysteine concentrations in humans. Am. J.
Chlorogenic acid bioavailability largely depends on its metabolism by
Clin. Nutr. 73:532 – 538; 2001.
the gut microflora in rats. J. Nutr. 133:1853 – 1859; 2003.
[12] Verhoef, P.; Pasman, W. J.; Van Vliet, T.; Urgert, R.; Katan, M. B.
[26] Morrow, J. D.; Frei, B.; Longmire, A. W.; Gaziano, J. M.; Lynch, S. M.;
Contribution of caffeine to the homocysteine-raising effect of
Shyr, Y.; Strauss, W. E.; Oates, J. A.; Roberts, L. J., II. Increase in
coffee: a randomized controlled trial in humans. Am. J. Clin. Nutr.
circulating products of lipid peroxidation (F2-isoprostanes) in smokers.
76:1244 – 1248; 2002.
N. Engl. J. Med. 332:1198 – 1203; 1995.
[13] Urgert, R.; van Vliet, T.; Zock, P. L.; Katan, M. B. Heavy coffee
[27] Salonen, J. T. Markers of oxidative damage and antioxidant protection:
consumption and plasma homocysteine: a randomized controlled trial
assessment of LDL oxidation. Free Radic. Res. 33:41 – 46; 2000.
in healthy volunteers. Am. J. Clin. Nutr. 72:1107 – 1110; 2000.
[28] Rietveld, A.; Wiseman, S. Antioxidant effect of tea: Evidence from
[14] Voutilainen, S.; Alfthan, G.; Nyyssonen, K.; Salonen, R.; Salonen,
human clinical trials. J. Nutr. 133:3285 – 3292; 2003.
J. T. Association between elevated plasma total homocysteine and
[29] O’Reilly, J. D.; Mallet, A. I.; McAnlis, G. T.; Young, I. S.; Halliwell,
increased common carotid artery wall thickness. Ann. Med. 30:
B.; Sanders, T. A.; Wiseman, H. Consumption of flavonoids in onions
300 – 306; 1998.
and black tea: lack of effect on F2-isoprostanes and autoantibodies to
[15] Wald, D. S.; Law, M.; Morris, J. K. Homocysteine and
oxidized LDL in healthy humans. Am. J. Clin. Nutr. 73:1040 – 1044;
cardiovascular disease: evidence on causality from a meta-analysis.
2001.
BMJ 325:1202 – 1206; 2002.
[30] Hodgson, J. M.; Croft, K. D.; Mori, T. A.; Burke, V.; Beilin, L. J.;
[16] Porkkala-Sarataho, E. K.; Nyyssfnen, K. M.; Kaikkonen, J. E.;
Puddey, I. B. Regular ingestion of tea does not inhibit in vivo lipid
Poulsen, H. E.; Hayn, E. M.; Salonen, R. M.; Salonen, J. T. A
peroxidation in humans. J. Nutr. 131:55 – 58; 2002.
randomized, single-blind, placebo controlled trial of the effects of 200
[31] Manach, C.; Scalbert, A.; Morand, C.; Remesy, C.; Jimenez, L.
mg a-tocopherol on the oxidation resistance of atherogenic lip-
Polyphenols: food sources and bioavailability. Am. J. Clin. Nutr.
oproteins. Am. J. Clin. Nutr. 68:1034 – 1041; 1998.
79:727 – 747; 2004.

---

Document Outline

• The effects of coffee consumption on lipid peroxidation and plasma total homocysteine concentrations: a clinical trial
  • Introduction
  • Materials and methods
    • Subjects
    • Study design
      • Long-term study
      • Short-term study
      • Resistance of serum lipids to oxidation
      • Serum LDL-conjugated dienes
      • Plasma hydroxy fatty acids
      • Plasma F2-isoprostanes
      • Activity of antioxidant enzymes
      • Plasma total homocysteine
      • Other measurements
      • Phenolic acid analyses of coffee and urine
    • Statistical analyses
  • Results
  • Discussion
  • Acknowledgments
  • References

---