C H A P T E R

45
Coffee Consumption and Serum Lipids:
A Review of Epidemiological Studies
and Experimental Studies in Humans
Defu Ma1, Li Cai2, Xiaolong Han3, Peiyu Wang1
1Department of Social Medicine and Health Education, School of Public Health, Peking University Health Science
Center, Beijing, China; 2Department of Nutrition and Food Hygiene, School of Public Health, Peking University Health
Science Center, Beijing, China; 3Department of Space Food and Nutrition, China Astronaut Research and Training
Center, Beijing, China

List of Abbreviations cardiovascular disorders and some forms of cancer.6,7 In

1981, an analysis of available data in favor of or against
CHD  Coronary heart disease
HDL-C  High density lipoprotein cholesterol coffee use concluded that there were no strong argu-
LDL-C  Low density lipoprotein cholesterol ments supporting advice to abandon drinking coffee.8
RCT  Randomized controlled trial In recent years, however, the possibility that coffee use
TC  Total cholesterol is associated with an increased incidence of coronary
TG Triglycerides
heart disease (CHD) has gained renewed interest. CHD
remains the major cause of mortality and morbidity in
the western countries. Although epidemiological studies
45.1 INTRODUCTION have failed to show consistently that coffee drinking is a
factor in the development of CHD, reports suggest that
Coffee was introduced as an economic crop during the coffee consumption may alter blood lipoprotein patterns
fifteenth century. Until now, more than 70 countries culti- and contribute to CHD by this mechanism.9 Consider-
vate this plant; Brazil, Colombia, Ethiopia, and India are ing the large amount of coffee consumption particularly
the leading producers. Coffee consumption is increasing, in the Western world, North American regions, and
and people from all over the world consume approxi- some Asian countries, even small health effects of cof-
mately 500 billion cups annually.1 Nowadays, coffee is fee could have considerable public health consequences.
the leading worldwide beverage after water and its trade Therefore, demonstration of the benefits and hazards
exceeds US $10 billion worldwide.2 Drinking coffee is associated with coffee consumption on serum lipids is
very common in Western society. In the United States, very important. Therefore, this review article can act as a
for example, 52% of all persons aged 10 years or older comprehensive treatise regarding the benefits and risks
drink coffee. Fifty-six percent of the adult US population of coffee consumption on serum lipids.
consumes an average of 3.4 cups of coffee per day.3
Although coffee is consumed worldwide, there has
been debate in the medical literature about whether 45.2  EPIDEMIOLOGICAL STUDIES
coffee is beneficial or troublesome for human health ON COFFEE CONSUMPTION AND SERUM
for more than three decades. Its consumption has been LIPIDS
associated with a momentous decrease in chronic dis-
eases such as parkinsonism, diabetes mellitus, and Heavy consumption of coffee has long been suspected
several cancer lines.4,5 On the contrary, there are also to have a cholesterol-raising effect. In 1966, a significant
some research studies reporting its promotion role in correlation between coffee consumption and serum

Coffee in Health and Disease Prevention 409

http://dx.doi.org/10.1016/B978-0-12-409517-5.00045-0 © 2015 Elsevier Inc. All rights reserved.
410 45.  COFFEE CONSUMPTION AND SERUM LIPIDS

lipid concentrations was reported in men with CHD.10 7368 men aged 20–54 years, and found that coffee con-
This possible association was subsequently studied in sumption was positively associated with levels of TC
many cross-sectional observational studies. However, and TG in both sexes and was inversely associated with
cross-sectional studies investigating the associations levels of HDL-C in women. After adjustment for all
have yielded conflicting results. Several cross-sectional covariates including age, logarithm of body mass index,
studies reported that coffee consumption is associated physical activity in leisure time, cigarette smoking, and
with elevated levels of total serum cholesterol.11,12 In alcohol consumption, the TC level was 5.56 mmol/l in
the contrary, other studies did not show an association men drinking less than one cup of coffee a day, compar-
between coffee consumption and serum cholesterol lev- ing with 6.23 mmol/l in those consuming more than nine
els.13,14 In total, most of early observational studies did cups a day. The corresponding figures for women were
report a relationship between coffee intake and serum 5.32 and 5.92 mmol/l.19 It is noteworthy that boiled cof-
cholesterol, but many were limited by failure to control fee was the predominant type consumed by the Tromso
for important confounding variables such as smoking sample. In an extensive meta-analysis of cross sectional
and alcohol drinking.15,16 For example, one study by studies, Bak showed a positive relationship between cof-
Klatsky et al. reporting that coffee consumption was fee and serum TC with statistical significant findings for
positively correlated with serum cholesterol levels did boiled coffee, espresso coffee, and Turkish coffee. The
not adjust tea consumption, which raised the question of common feature of these brewing methods is that no
whether caffeine or some other component is related to paper filter is used during preparation of the coffee.20
the observed effects.15 Williams et al. also showed a posi-
tive correlation between coffee consumption and eleva-
tions of serum apolipoprotein B, low-density lipoprotein 45.3  EXPERIMENTAL STUDIES ON
cholesterol (LDL-C), and total cholesterol (TC), but the COFFEE CONSUMPTION AND SERUM
cross-sectional nature of the design limits inferences LIPIDS
about causality.16 On the other hand, other observational
studies have shown no relationship.12,17 For example, The effect of coffee consumption on cholesterol has
Dawber et al. reported that there was no correlation also been studied in randomized controlled trials. The
between reported level of coffee consumption and the first experiment on coffee consumption and blood cho-
serum total cholesterol level.12 lesterol in hypercholesterolemic subjects was described
In 1987, Thelle et al. reviewed 22 cross-sectional by Egede-Nissen in 1970, and the author suggested that
studies involving 130,000 persons from eight different subjects abstaining from coffee could reduce their TC
countries.9 Results of these reports displayed a variety level by 17%.21 However, subsequent trials indicated
of trends in the association between coffee intake and that this effect was only seen with boiled but not filter-
serum cholesterol concentrations: 8 (36%) studies dem- brewed or instant coffee. In the trials conducted by the
onstrated a significant positive association in both sexes, Tromso, a Norway group, they had consistently found
and 5 (23%) studies showed no association in men or a positive correlation between coffee consumption and
women. In three other reports where both sexes were serum cholesterol levels.22 However, these studies did
included, significant positive association was observed not control dietary and smoking factors. Moreover, the
only in women. The remaining six investigations exam- method of preparation of coffee used, predominantly
ined only men with 4 (18%) reporting a significant cor- boiled, was different from the methods in general used
relation between coffee and cholesterol. Thelle et al. in the United States today. In addition, these studies
concluded that different brewing method and differ- permitted tea consumption, although there was dis-
ent population may be important factors in the choles- agreement in the literature as to whether caffeinated tea
terol-raising effect of coffee. Similarly, Rosmarin et al. causes the same results. Other clinical trials using differ-
reviewed 25 cross-sectional studies and found that 13 ent methods for preparing the coffee have failed to show
observational studies have reported a definite associa- an effect of coffee consumption on lipids.
tion between coffee consumption and elevated lipids, six Clinical trials used various intake levels of coffee with
studies have shown an association only in certain sub- varying types of coffee, and different protocols. To pre-
groups, and six have shown no association.18 Subsequent cisely evaluate the effects of coffee intake on serum lipids
cross-sectional studies and experimental studies have and to clarify the active factors of coffee, Jee et al. per-
proved that boiled coffee exerted significant increasing formed a meta-analyses of nine randomized controlled
effects on serum lipids comparing with that of filtered trials (RCTs) published prior to December 1998 and
coffee.19,20 The Tromso Heart Study examined the rela- found that there was a dose–response relation between
tion between coffee consumption and levels of serum coffee consumption and both TC and LDL-C (p < 0.01).23
TC, high-density lipoprotein cholesterol (HDL-C), and Increases in serum lipids were greater in studies of
triglycerides (TG) in a population of 7213 women and patients with hyperlipidemia and in trials of caffeinated

II.  EFFECTS OF COFFEE CONSUMPTION

 45.4  Boiled Coffee and Filter Coffee 411
or boiled coffee. Trials using filtered coffee demonstrated coffee, the results were quite uniform, with linear trend
very little increase in serum cholesterol. Consumption of coefficients ranging from 0.022 to 0.044 mmol/l per cup
unfiltered coffee increases serum levels of TC and LDL- with significance. Experimental studies confirmed Bak’s
C.23 However, this meta-analysis failed to include two observation and excluded any relationship between fil-
important RCTs, which included 345 participants.24,25 In tered coffee consumption and serum cholesterol level.
addition, the active factors such as caffeine and coffee Bak performed a 12-week randomized trial involving
oils which were responsible for the plasma lipoprotein- young adult subjects with normal serum cholesterol
increasing effect of coffee were not determined. levels to determine the effects of coffee prepared by
In 2012, our research group repeated a meta-analy- two common brewing methods (filtering and boiling).31
sis of 12 RCTs with a total of 1017 subjects.26 Table 45.1 They found that the serum TC level increased after the
described the characteristics of the trials included in consumption of boiled coffee by 0.48 mmol/l, and the
the meta-analysis. In the 12 studies, caffeine coffee was LDL-C increased by 0.39 mmol/l. However, there was
used in the total of 12 studies,18,22,24,25,27–34 and decaffein- no significant difference in the change in TC and LDL-C
ated coffee was used in three studies. Unfiltered coffee in the filtered-coffee group and no coffee group. In the
was used as intervention in 12 comparisons and filtered meta-analysis of RCTs, Jee et al. found that trials using
coffee was used in other comparisons. The duration of filtered coffee demonstrated very little increase in serum
treatment varied widely, ranging from 14 days to 79 days cholesterol, while consumption of unfiltered coffee
(average, 45 days). Coffee intake varied from 2.4 to 8.0 increased serum levels of TC and LDL-C significantly.23
cups/day in the various treatment groups. Six of these In the subgroup analysis of our meta-analysis in 2012,
12 studies were focused on men and the other studies we also found that unfiltered coffee exerted significant
included both sexes. In five comparisons, the partici- increasing effects on TC (12.9 mg/dl, 95% CI 6.8–18.9),
pants had hyperlipidemia at baseline according to the LDL-C (11.9 mg/dl, 95% CI 3.2–20.6), and TG (18.8 mg/
definition of the original study. Our meta-analyses con- dl, 95% CI 4.8–32.7) comparing with that of filtered cof-
cluded that drinking coffee for 45 days was associated fee (Table 45.2).26
with an average increase of 8.1 mg/dl (95% confidence The observation that boiled coffee has a more potent
interval (CI) 4.5–11.6; p < 0.001) for TC, 5.4 mg/dl (95% effect than filtered coffee has led to speculation that
CI 1.4–9.5; p < 0.009) for LDL-C, and 12.6 mg/dl (95% the active substance(s) may be removed by adsorp-
CI 3.5–12.6; p < 0.007) for TG (Figure 45.1). In addition, tion or filtration when a paper filter is used or is better
similar with the conclusion from observational studies, extracted by the higher temperatures used in boiled-
the two meta-analyses also found that the effect of coffee coffee preparation. The boiling method requires 10 or
consumption on serum lipids may be influenced by dif- more minutes of direct contact between ground coffee
ferent brewing method, different coffee oil and caffeine and boiling water. On the contrary, the methods based
content. The related influence factors will be discussed on filtration require briefer contact between coffee and
in the next sections. hot water (20–100 s). The factor in coffee responsible for
the increase of blood cholesterol should be sought in the
lipid-rich soluble fraction of the boiled coffee. Boiled
45.4  BOILED COFFEE AND FILTER coffee has a higher concentration of coffee oils because
COFFEE of the higher temperatures used during its preparation
and the longer contact time between the coffee grounds
In one trial that compared the effect of boiled coffee, and water.32 Van Dusseldorp et al. showed that paper
filtered coffee, and no coffee, all the subjects also drank filter can retain more than 80% of the lipid-soluble sub-
tea. It showed that boiled coffee (plus tea) led to increases stances that presented in boiled coffee.29 The diterpenes
in TC and LDL-C levels, compared with filtered coffee cafestol and kahweol, which are present in nonfiltered
(plus tea), and compared with tea alone. The hypoth- coffee, including boiled coffee, cafetiere coffee, and Turk-
esis that the brewing method of coffee might have an ish coffee, appeared to be responsible for the increase in
effect on this association had been discussed by Bak in plasma cholesterol. In order to identify the cholesterol-
an extensive meta-analysis of cross-sectional studies.20 raising factor of coffee oil, Weusten-Van et al. performed
In the meta-analysis, regression coefficients of serum TC an interventional study and found that in 15 volunteers
were positively signed, which indicated a rise in serum who ingested 0.75 g/day of a non-TG fraction from coffee
TC with increased coffee consumption. However, sub- oil for 4 weeks, mean cholesterol increased by 48 mg/dl
group analysis based on coffee brewing method showed (1.2 mmol/l) relative to placebo. In contrast, a coffee oil
that statistical significance seemed to be dependent on stripped of the non-TG lipids cafestol and kahweol had
the brewing method used. All studies on filtered coffee no effect. In three volunteers, purified cafestol (73 mg/
showed a small positive effect on serum TC, statisti- day) plus kahweol (58 mg/day) increased cholesterol by
cally significant in only one of seven studies. For boiled 66 mg/dl (1.7 mmol/l) after 6 weeks. Oil from Robusta

II.  EFFECTS OF COFFEE CONSUMPTION

 412
TABLE 45.1  Characteristics of Study Design, Type of Intervention, and Study Populations in the Included Studies26
Sample Study Type of Coffee/ Duration Coffee Dose Indexes
Author (Year) Sizea Designb Maskingc Controld (days) (cup/day) % Male Mean Age % Hypercholes­terolemia Reported

45.  COFFEE CONSUMPTION AND SERUM LIPIDS

II.  EFFECTS OF COFFEE CONSUMPTION

Grubben (2000) 64/64 × NA R/N 14 6.7 100 43 0 TC, TG

D’amicis (1996a) 28/28 P NA R/T 42 3.1 100 27 0 TC, LDL-C,

HDL-C, TG

D’amicis (1996b) 28/28 P NA R/T 42 2.8 100 27 0 TC, LDL-C,

HDL-C, TG

Burr (1995) 261/261 × NA I/N 42 5.0 48 39 0 TC, HDL-C

Fried (1992a) 25/25 P Open D-F/N 56 4.8 100 44 0 TC, LDL-C,

HDL-C

Fried (1992b) 25/25 P Open R-F/N 56 2.4 100 44 0 TC, LDL-C,

HDL-C

Fried (1992c) 25/25 P Open R-F/N 56 4.8 100 44 0 TC, LDL-C,

HDL-C

Superko (1991a) 62/58 P Double R-F/N 56 4.0 100 46 0 TC, LDL-C,

HDL-C, TG

Superko (1991b) 61/58 P Double D-F/N 56 4.0 100 46 0 TC, LDL-C,

HDL-C, TG

Van (1991a) 22/21 P NA R/N 79 6.0 51 39 0 TC, LDL-C,

HDL-C, TG

Van (1991b) 21/21 P NA R-F/N 79 6.0 52 39 0 TC, LDL-C,

HDL-C, TG

Rosmarin (1990) 21/21 × NA R-F/N 60 3.6 100 35 0 TC, LDL-C,

HDL-C, TG

Burr (1989a) 54/54 × Single R/N 28 6.5 65 35 0 TC, HDL-C

Burr (1989b) 54/54 × Single D/N 28 6.3 65 35 0 TC, HDL-C

 Bark (1989a) 34/34 P Single R-F/N 63 5.0 53 26 0 TC, LDL-C,
HDL-C

Bark (1989b) 33/34 P Single R/N 63 5.0 54 26 0 TC, LDL-C,

HDL-C

Aro (1987a) 42/42 × NA R/T 28 8.0 50 49 100 TC, LDL-C,

HDL-C, TG

Aro (1987b) 42/42 × NA R-F/T 28 8.0 50 49 100 TC, LDL-C,

HDL-C, TG

Forde (1985a) 8/9 P NA R/N 70 7.3 100 45 100 TC

Forde (1985b) 8/9 P NA R/N 70 7.3 100 45 100 TC

Forde (1985c) 8/9 P NA R-F/N 70 7.3 100 45 100 TC

Aro (1985) 12/12 × NA I/T 21 8.0 50 39 0 TC, LDL-C, TG

TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TG, triglyceride.
II.  EFFECTS OF COFFEE CONSUMPTION

aNumber of treatment group/number of control group.

45.4  BOILED COFFEE AND FILTER COFFEE

bP, parallel; ×, crossover.
cNA, information on masking not available; double, double blind; single, single blind.
dR, regular coffee; D, decaffeinated coffee; F, filtered coffee; N, no coffee; T, tea.

Reprinted from Ref. 26, with permission from the Nature Publishing Group.

413
 $
6WXG\ 
,' :0'&, :HLJKW

*UXEEHQ  

'
DPLFLD ± 
'
DPLFLE ± 
%XUU  
)ULHGD ± 
)ULHGE ±± 
)ULHGF  
6XSHUNRD ± 
6XSHUNRE ± 
9DQD  
9DQE ± 
5RVPDULQ ± 
%XUUD ± 
%XUUE ± 
%DND ± 
%DNE  
$URD  
$URE ± 
)RUGHD  
)RUGHE  
)RUGHF ± 
$UR ± 
2YHUDOO,VTXDUHG S   

127(:HLJKWVDUHIURPUDQGRPHIIHFWVDQDO\VLV

±   

% 6WXG\ 
,' :0'&, :HLJKW

'
DPLFLD ± 
'
DPLFLE ± 
)ULHGD ± 
)ULHGE ±± 
)ULHGF ± 
6XSHUNRD ± 
6XSHUNRE  
9DQD  
9DQE ± 
5RVPDULQ ±± 
%DND ±± 
%DNE ± 
$URD  
$URE ± 
$UR ± 
2YHUDOO,VTXDUHG S   

127(:HLJKWVDUHIURPUDQGRPHIIHFWVDQDO\VLV

±   

FIGURE 45.1  Meta-analysis of the effect of coffee consumption on TC (A), LDL-C (B), HDL-C (C), and TG (D) compared with control.26
WMD, weighted mean difference; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol;
TG, triglyceride. Reprinted from Ref. 26, with permission from the Nature Publishing Group.
 45.4  BOILED COFFEE AND FILTER COFFEE 415

&
6WXG\ 
,' :0'&, :HLJKW

'
DPLFLD ±± 
'
DPLFLE ±± 
%XUU ± 
)ULHGD ± 
)ULHGE ± 
)ULHGF ± 
6XSHUNRD ± 
6XSHUNRE ± 
9DQD ±±± 
9DQE ±± 
5RVPDULQ ± 
%XUUD ±± 
%XUUE ±± 
%DND ± 
%DNE ± 
$URD ±± 
$URE ± 
2YHUDOO,VTXDUHG S  ±± 

± ±   

'
6WXG\ 

,' :0'&, :HLJKW

*UXEEHQ  

'
DPLFLD ± 

'
DPLFLE ± 

6XSHUNRD ±± 

6XSHUNRE ± 

9DQD  

9DQE ± 

5RVPDULQ ± 

$URD  

$URE ± 

$UR ± 

2YHUDOO,VTXDUHG S   

127(:HLJKWVDUHIURPUDQGRPHIIHFWVDQDO\VLV

±    

FIGURE 45.1  (Cont’d)

II.  EFFECTS OF COFFEE CONSUMPTION

 416
TABLE 45.2  Subgroup Analyses of TC, LDL-C, and TG in Stratified by Study Design, Type of Intervention, and Study Populations26

TC (mg/dl) LDL-C (mg/dl) TG (mg/dl)

Heterogeneityb Heterogeneityb Heterogeneityb

Net Change Net Change Net Change
Variables No.a (95% CI) p I2 No.a (95% CI) p I2 No.a (95% CI) p I2

Overall 22 8.1 <0.001 67.3% 15 5.4 (1.4, 9.5) 0.002 58.4% 11 12.6 (3.5, 21.6) 0.001 66.4%
(4.5, 11.6)

Study design

 Parallel 15 7.2 0.001 61.1% 11 4.5 (0.6, 8.5) 0.018 53.4% 6 7.3 (−2.5, 17.1) 0.008 68.1%

45.  COFFEE CONSUMPTION AND SERUM LIPIDS

II.  EFFECTS OF COFFEE CONSUMPTION

(3.3, 11.0)

 Cross- 7 10.5 <0.001 78.6% 4 11.5 0.009 74.1% 5 24.1 (10.3, 37.8) 0.247 26.1%
over (1.7, 19.2) (−4.6, 27.5)

Decaffeinated

 Yes 3 3.5 0.449 0.0% 2 6.3 0.295 8.7% 1 3.5 (−10.6, 17.7) – –
(−1.1, 8.1) (−0.8, 13.4)

 No 19 9.2 <0.001 70.7% 13 5.5 0.001 62.8% 10 13.8 (3.7, 24.0) 0.001 68.7%
(5.0, 13.4) (0.8, 10.2)

Filtered

 Yes 10 3.6 (0.6, 6.6) 0.524 0.0% 9 2.3 0.349 10.4% 5 3.7 (−4.2, 11.7) 0.432 0.0%
(−1.1, 5.6)

 No 12 12.9 <0.001 79.3% 6 11.9 0.002 73.3% 6 18.8 (4.8, 32.7) 0.001 77.1%
(6.8, 18.9) (3.2, 20.6)

Duration

  <8 weeks 11 9.2 <0.001 69.6% 5 9.3 0.028 63.2% 6 15.8 (1.8, 29.8) 0.006 69.3%
(3.8, 14.5) (0.9, 17.8)

  ≥8 weeks 11 7.3 0.001 68.0% 10 4.0 0.009 59.1% 5 9.7 (−3.8, 23.2) 0.009 70.2%
(2.1, 12.6) (−0.8, 8.9)

Amount of coffee

  <6 cups 11 4.2 (1.3, 7.1) 0.109 36.3% 10 2.7 0.419 2.2% 5 0.7 (−5.5, 6.9) 0.884 0.0%
(−0.1, 5.4)

  ≥6 cups 11 15.7 <0.001 75.2% 5 15.3 0.022 65.0% 6 25.3 (17.0, 33.7) 0.367 7.8%
(7.6, 23.9) (4.5, 26.0)
 TC (mg/dl) LDL-C (mg/dl) TG (mg/dl)

Heterogeneityb Heterogeneityb Heterogeneityb

Net Change Net Change Net Change
Variables No.a (95% CI) p I2 No.a (95% CI) p I2 No.a (95% CI) p I2

Age (years)

 <40 11 5.6 (2.0, 9.2) 0.025 51.3% 8 5.4 0.030 54.9% 6 11.5 (1.3, 21.6) 0.033 58.7%
(−0.0, 10.8)

  ≥40 11 12.4 <0.001 76.0% 7 5.9 0.007 66.4% 5 15.8 (−3.4, 35.0) 0.001 77.3%
(5.1, 19.8) (−0.8, 12.7)

Sex

 Male 11 4.5 (0.2, 8.8) 0.033 49.1% 8 3.0 (0.2, 5.7) 0.445 0.0% 5 0.7 (−5.5, 6.9) 0.884 0.0%

 Male and 11 11.4 <0.001 75.8% 7 10.7 0.003 69.3% 6 25.3 (17.0, 33.7) 0.367 7.8%
female (5.7, 17.2) (0.8, 20.7)
II.  EFFECTS OF COFFEE CONSUMPTION

Hyperlipidemic

45.4  BOILED COFFEE AND FILTER COFFEE

 Yes 5 28.9 0.037 61.0% 2 21.8 0.045 75.2% 2 30.2 (−4.4, 64.7) 0.220 33.5%
(12.3, 45.5) (−3.5, 47.2)

 No 17 5.6 (2.7, 8.5) 0.006 52.8% 13 4.2 (0.6, 7.7) 0.034 46.4% 9 11.1 (1.7, 20.4) 0.001 69.7%

TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; TG, triglyceride.

aNumber of comparisons.
bp for heterogeneity from Q statistics.

Reprinted from Ref. 26, with permission from the Nature Publishing Group.

417
418 45.  COFFEE CONSUMPTION AND SERUM LIPIDS

beans, which contains cafestol but negligible kahweol, men, regardless of source of caffeine in a cross-sectional
also raised serum cholesterol.35 These findings showed study.41 In another cross-sectional study, after adjusting
that serum cholesterol is raised by cafestol and pos- for age and adiposity, the mean serum cholesterol level
sibly also kahweol, both natural components of coffee was 11 mg/dl higher for women consuming 200 mg or
beans. In addition, Weusten-Van et al. found that cafestol more of caffeine per day compared with those consum-
increased serum TG and alanine amino-transferase and ing less.42 Nevertheless, one study found that serum
depressed serum creatinine and γ-glutamyl-transferase. caffeine concentrations were closely positively related
After withdrawal, γ-glutamyl-transferase activity rose to TG in caffeine-drug users, but no associations were
above baseline.35 Therefore, they concluded that cafes- found between serum caffeine concentrations with TC
tol increased the cholesterol level by alterations in liver or LDL-C levels either in caffeine/drug users or nonus-
function enzymes. In addition, some researchers have ers.43 In addition, Du et al. concluded that chronic intake
reported that cafestol and kahweol increased the syn- of caffeine (exclusively from diet source) might slightly
thesis of cholesterol by decreasing excretion of bile acids increase HDL-C concentrations in women.43
and neutral sterols.36 Superko et al. found an increase of LDL-C and apoli-
poprotein B in subjects consuming decaffeinated filtered
coffee, without any increase among drinkers of normal
45.5  CAFFEINATED COFFEE AND coffee, which suggested that a coffee component other
DECAFFEINATED COFFEE than caffeine was responsible for the LDL-C, apolipopro-
tein B, and lipase activity changes.28 Nevertheless, Fried
As far as the composition of coffee is concerned, caf- et al. reported a significant positive association between
feine is no doubt considered as its major and active consumption of filtered normal coffee and serum lipids.
ingredient. Caffeine is white crystalline powder having a They observed an increase of TC in the group consum-
bitter taste. It was first isolated from coffee in 1820. Often, ing 720 ml per day of filtered normal coffee, while no
coffee is consumed for its stimulatory effects owing to its increase was observed in the group drinking 720 ml of
rich phytochemistry among which caffeine is the most filtered decaffeinated coffee, and in the group drinking
prominent. Coffee is the richest source of caffeine and 320 ml of filtered normal coffee.27 In our subgroup meta-
240 ml instant coffee contains approximately 100 mg of analysis, caffeinated coffee had significant effects on TC,
caffeine. Frary et al. conducted a survey regarding caf- LDL-C, and TG but decaffeinated coffee had not (Table
feine intake from different sources and reported that 45.2). However, these results should be interpreted cau-
70% of caffeine comes from coffee while soft drinks tiously, as only three RCTs involving decaffeinated cof-
and tea contribute 16% and 12%, respectively.37 Ameri- fee were identified in our meta-analysis and two of them
cans annually consume about 139 billion cups of coffee used filtered coffee.26
(20% decaffeinated) with a total caffeine consumption Regarding the lack of an effect of decaffeinated coffee
equivalent to a per capita intake of 125 mg caffeine/day on lipoprotein cholesterol levels, it may be speculated
from coffee, suggesting that caffeine in coffee is the most that the decaffeination process also extracts the active
commonly ingested drug in the United States.38 Because substance(s) or that the coffee beans used to manufacture
of the large number of people consuming coffee, even decaffeinated coffee lack the active substances. A typical
relatively small effects of caffeine on serum lipids might roasted Arabica coffee is composed of 9% protein, 30.4%
eventually have large public health consequence. carbohydrate, 13% lipids, 0.4% volatile acids, 4.2% non-
In 1965, Bellet et al. reported a clinical trial testing volatile acids, 1.6% alkaloids, 4% ash, 2.5% water, and
the effect of caffeine on the level of fatty acids and did 35% browning compounds, which include phenolic com-
not find positive relationship.39 Since then, a number of pounds.44 Robusta coffee has a much higher phenolic
investigations in humans have been published. In most content than coffee Arabica. In addition, the content of
cross-sectional studies on caffeine and serum lipids, only cafestol and kahweol is also highly dependent on the ratio
caffeine intake from coffee is positively associated with of Arabica/Robusta coffee species, as beans from coffee
serum cholesterol. Curb et al. found a positive relation- Arabica contain more of cafestol and kahweol than those
ship between coffee consumption and serum cholesterol, from coffee Robusta. It has been reported that Robusta
but there was no significant relationship between tea or bean is used to a greater extent in decaffeinated coffees. In
cola, the other major caffeine contributors to the diet, the section of boiled coffee and filter coffee, we have stated
and serum cholesterol. Thus, this analysis concluded that the diterpenes cafestol and kahweol may be respon-
that there was a significant positive relationship between sible for the increase in plasma cholesterol. Therefore, the
caffeine from coffee and serum cholesterol which is not cholesterol-increasing effect difference between boiled
present with other sources of caffeine.40 Mathais et al. coffee and filtered coffee which have the similar content
found a significant dose–response relationship between of caffeine indicated that a coffee component other than
caffeine consumption and TC in women but not in caffeine was responsible for the cholesterol-increasing

II.  EFFECTS OF COFFEE CONSUMPTION

 References 419
effect of coffee. Nevertheless, caffeine has been shown to Robusta bean that contain less cafestol and kahweol
exhibit several biological effects, such as increased fat oxi- is used to a greater extent in decaffeinated coffees.
dation and mobilization of glycogen in muscle, increased In total, the cholesterol-increasing effect difference
lipolysis, and decreased body fat.45 Further confirmation between boiled coffee and filtered coffee which have
on caffeine and serum lipids is required in large prospec- the similar content of caffeine indicated that a coffee
tive studies, especially RCTs. component other than caffeine was responsible for the
cholesterol-increasing effect of coffee.

45.6 CONCLUSIONS References
Epidemiological studies and experimental studies 1.  Prakash N, Combes M, Somanna N, Lashermes P. AFLP analysis
of introgression in coffee cultivars (Coffea arabica L.) derived from a
investigating the associations of coffee consumption and
natural interspecific hybrid. Euphytica 2002;3:265–71.
serum lipids have yielded conflicting results. To sum, 2.  Butt MS, Sultan MT. Coffee and its consumption: benefits and
boiled coffee has a more potent cholesterol-increasing risks. Crit Rev Food Sci Nutr 2011;51:363–73.
effect than filtered coffee. Diterpenes cafestol and kah- 3.  Nygard O, Refsum H, Ueland PM, Stensvold I, Nordrehaug JE,
weol, which are present in boiled coffee but largely Kvale G, et al. Coffee consumption and plasma total homocysteine:
the Hordaland Homocysteine Study. Am J Clin Nutr 1997;65:136–43.
removed by filtering coffee, appeared to be responsible
4.  Cavin C, Marin-Kuan M, Langouet S, Bezencon C, Guignard G,
for the increase in plasma cholesterol. However, more Verguet C, et al. Induction of Nrf2-mediated cellular defenses and
studies need to be conducted on the effect of coffee espe- alteration of phase I activities as mechanisms of chemoprotective
cially caffeine on serum lipids. In order to understand effects of coffee in the liver. Food Chem Toxicol 2008;46:1239–48.
the molecular mechanisms behind the biological effects 5.  Li G, Ma D, Zhang Y, Zheng W, Wang P. Coffee consumption and
risk of colorectal cancer: a meta-analysis of observational studies.
of coffee, it is necessary to identify the active components
Public Health Nutr 2013;16:346–57.
in the coffee. Although many biologically active com- 6.  Montella M, Polesel J, La Vecchia C, Dal Maso L, Crispo A,­
pounds have been identified, coffee contains hundreds Crovatto M, et al. Coffee and tea consumption and risk of hepato-
of unidentified chemical compounds with potential bio- cellular ­carcinoma in Italy. Int J Cancer 2007;120:1555–9.
activity. Furthermore, in order to understand how cof- 7.  Mukamal KJ, Hallqvist J, Hammar N, Ljung R, Gemes K,
fee can affect the risk of CVD by cholesterol-increasing Ahlbom A, et al. Coffee consumption and mortality after acute
myocardial infarction: the Stockholm Heart Epidemiology Pro-
mechanism, the molecular changes that are associated gram. Am Heart J 2009;157:495–501.
with the development of the disease must be identified. 8.  Anonymous. Coffee: should we stop drinking it? Lancet 1981;1:256.
9.  Thelle DS, Heyden S, Fodor JG. Coffee and cholesterol in epidemi-
ological and experimental studies. Atherosclerosis 1987;67:97–103.
10. Little JA, Shanoff HM, Csima A, Toronto MA, Yano R. Coffee and
45.7  SUMMARY POINTS serum-lipids in coronary heart-disease. Lancet 1966;1:732–4.
11. Davis BR, Curb JD, Borhani NO, Prineas RJ, Molteni A. Coffee con-
• E pidemiological studies investigating the sumption and serum cholesterol in the hypertension detection and
associations of coffee consumption and serum follow-up program. Am J Epidemiol 1988;128:124–36.
lipids have yielded conflicting results. In most 12. Pietinen P, Geboers J, Kesteloot H. Coffee consumption and serum
cholesterol: an epidemiological study in Belgium. Int J Epidemiol
epidemiological studies that reported significant
1988;17:98–104.
increasing effects on serum lipids of coffee 13. van Stiphout WA, Hofman A, de Bruijn AM, Valkenburg HA.
consumption, coffee is boiled. Distributions and determinants of total and high-density lipo-
• Experimental studies have also yield conflicting protein cholesterol in Dutch children and young adults. Prev Med
results because of the various intake levels of coffee 1985;14:169–80.
14. Salonen JT, Happonen P, Salonen R, Korhonen H, Nissinen A,
with varying types of coffee and different protocols.
Puska P, et al. Interdependence of associations of physical ac-
However, the cholesterol-increasing effect is often tivity, smoking, and alcohol and coffee consumption with se-
seen with boiled but not filter-brewed coffee in rum high-density lipoprotein and non-high-density lipoprotein
interventional studies. cholesterol – a population study in eastern Finland. Prev Med
• The coffee-brewing method might modify the 1987;16:647–58.
15. Klatsky AL, Petitti DB, Armstrong MA, Friedman GD. Coffee, tea
cholesterol-increasing effect of coffee had been
and cholesterol. Am J Cardiol 1985;55:577–8.
proved in a lot of studies. The diterpenes cafestol 16. Williams PT, Wood PD, Vranizan KM, Albers JJ, Garay SC, Taylor
and kahweol, which are present in boiled coffee but CB. Coffee intake and elevated cholesterol and apolipoprotein B
largely removed by filtering coffee, appeared to be levels in men. JAMA 1985;253:1407–11.
responsible for the increase in plasma cholesterol. 17. Dawber TR, Kannel WB, Gordon T. Coffee and cardiovascular
disease. Observations from the framingham study. N Engl J Med
• In light of decaffeinated coffee’s lack of an effect on
1974;291:871–4.
lipoprotein cholesterol levels, it is speculated that the 18. Rosmarin PC, Applegate WB, Somes GW. Coffee consumption
coffee beans used to manufacture decaffeinated coffee and serum lipids: a randomized, crossover clinical trial. Am J Med
lack the active substances. It has been reported that 1990;88:349–56.

II.  EFFECTS OF COFFEE CONSUMPTION

420 45.  COFFEE CONSUMPTION AND SERUM LIPIDS

19. Thelle DS, Arnesen E, Forde OH. The Tromso heart study. Does 33. Aro A, Kostiainen E, Huttunen JK, Seppala E, Vapaatalo H. Effects
coffee raise serum cholesterol? N Engl J Med 1983;308:1454–7. of coffee and tea on lipoproteins and prostanoids. Atherosclerosis
20. Bak A. Coffee and cardiovascular risk: an epidemiological study. 1985;57:123–8.
­Rotterdam: Erasmus Universiteit; 1990. 34. Grubben MJ, Boers GH, Blom HJ, Broekhuizen R, de Jong R,
21. Eegde-Nissen A. Kolesterol og kaffe. En observasjon fra praksis. Tidsskr ­van Rijt L, et al. Unfiltered coffee increases plasma homocysteine
Nor Lægeforening; 1970. p. 1506–7. concentrations in healthy volunteers: a randomized trial. Am J Clin
22. Forde OH, Knutsen SF, Arnesen E, Thelle DS. The Tromso heart Nutr 2000;71:480–4.
study: coffee consumption and serum lipid concentrations in men 35. Weusten-Van DWM, Katan MB, Viani R, Huggett AC, Liardon R,
with hypercholesterolaemia: an randomised intervention study. Br Liardon R, et al. Identity of the cholesterol-raising factor from
Med J (Clin Res Ed) 1985;290:893–5. boiled coffee and its effects on liver function enzymes. J Lipid Res
23. Jee SH, He J, Appel LJ, Whelton PK, Suh I, Klag MJ. Coffee con- 1994;35:721–33.
sumption and serum lipids: a meta-analysis of randomized con- 36. Bjelke E. Letter: colon cancer and blood-cholesterol. Lancet
trolled clinical trials. Am J Epidemiol 2001;153:353–62. 1974;1:1116–7.
24. D’Amicis A, Scaccini C, Tomassi G, Anaclerio M, Stornelli R, 37. Frary CD, Johnson RK, Wang MQ. Food sources and intakes of caf-
­Bernini A. Italian style brewed coffee: effect on serum cholesterol feine in the diets of persons in the United States. J Am Diet Assoc
in young men. Int J Epidemiol 1996;25:513–20. 2005;105:110–3.
25. Burr ML, Limb ES, Sweetnam PM, Fehily AM, Amarah L, 38. Shahidi F, McDonald J, Chandrasekara A, Zhong Y. Phytochemi-
­Hutchings A. Instant coffee and cholesterol: a randomised con- cals of foods, beverages and fruit vinegars: chemistry and health
trolled trial. Eur J Clin Nutr 1995;49:779–84. effects. Asia Pac J Clin Nutr 2008;1(Suppl. 17):380–2.
26. Cai L, Ma D, Zhang Y, Liu Z, Wang P. The effect of coffee consump- 39. Bellet S, Kershbaum A, Aspe J. The effect of caffeine on free fatty
tion on serum lipids: a meta-analysis of randomized controlled tri- acids. Arch Intern Med 1965;116:750–2.
als. Eur J Clin Nutr 2012;66:872–7. 40. Curb JD, Reed DM, Kautz JA, Yano K. Coffee, caffeine, and se-
27. Fried RE, Levine DM, Kwiterovich PO, Diamond EL, Wilder rum cholesterol in Japanese men in Hawaii. Am J Epidemiol
LB, Moy TF, et al. The effect of filtered-coffee consumption on 1986;123:648–55.
plasma lipid levels. Results of a randomized clinical trial. JAMA 41. Mathias S, Garland C, Barrett-Connor E, Wingard DL. Coffee, plas-
1992;267:811–5. ma cholesterol, and lipoproteins. A population study in an adult
28. Superko HR, Bortz WJ, Williams PT, Albers JJ, Wood PD. community. Am J Epidemiol 1985;121:896–905.
Caffeinated and decaffeinated coffee effects on plasma lipoprotein 42. Shirlow MJ, Mathers CD. Caffeine consumption and serum choles-
cholesterol, apolipoproteins, and lipase activity: a controlled, ran- terol levels. Int J Epidemiol 1984;13:422–7.
domized trial. Am J Clin Nutr 1991;54:599–605. 43. Du Y, Melchert HU, Knopf H, Braemer-Hauth M, Gerding B, Pabel E.
29. van Dusseldorp M, Katan MB, van Vliet T, Demacker PN, Association of serum caffeine concentrations with blood lipids in
­Stalenhoef AF. Cholesterol-raising factor from boiled coffee does caffeine-drug users and nonusers – results of German National
not pass a paper filter. Arterioscler Thromb 1991;11:586–93. Health Surveys from 1984 to 1999. Eur J Epidemiol 2005;20:311–6.
30. Burr ML, Gallacher JE, Butland BK, Bolton CH, Downs LG. Coffee, 44. Anonymous. The methylxanthine beverages and foods: chemis-
blood pressure and plasma lipids: a randomized controlled trial. try, consumption, and health effects. Prog Clin Biol Res 1984;158:
Eur J Clin Nutr 1989;43:477–83. 1–413.
31. Bak AA, Grobbee DE. The effect on serum cholesterol levels of cof- 45. Ranheim T, Halvorsen B. Coffee consumption and human health–
fee brewed by filtering or boiling. N Engl J Med 1989;321:1432–7. beneficial or detrimental?–Mechanisms for effects of coffee con-
32. Aro A, Tuomilehto J, Kostiainen E, Uusitalo U, Pietinen P. Boiled sumption on different risk factors for cardiovascular disease and
coffee increases serum low density lipoprotein concentration. type 2 diabetes mellitus. Mol Nutr Food Res 2005;49:274–84.
­Metabolism 1987;36:1027–30.

II.  EFFECTS OF COFFEE CONSUMPTION