CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2013;11:14131421

Coffee Reduces Risk for Hepatocellular Carcinoma: An Updated

Meta-analysis

FRANCESCA BRAVI,*, CRISTINA BOSETTI,* ALESSANDRA TAVANI,* SILVANO GALLUS,* and CARLO LA VECCHIA*,
*Department of Epidemiology, IRCCS Istituto di Ricerche Farmacologiche Mario Negri, Milan; and Department of Clinical Sciences and Community Health,
Universit degli Studi di Milan, Milan, Italy

BACKGROUND & AIMS: Coffee consumption has been suggested to reduce the risk for hepatocellular carcinoma
(HCC). We performed a meta-analysis of epidemiological studies to provide updated
information on how coffee drinking affects HCC risk.

METHODS: We performed a PubMed/MEDLINE search of the original articles published in English

from 1966 through September 2012, on case-control or cohort studies that associated coffee
consumption with liver cancer or HCC. We calculated the summary relative risk (RR) for any,
low, and high consumption of coffee vs no consumption. The cut-off point for low vs high
consumption was set to 3 cups per day in 9 studies and 1 cup per day in 5 studies.

RESULTS: The summary RR for any coffee consumption vs no consumption was 0.60 from 16 studies,
comprising a total of 3153 HCC cases (95% condence interval [CI], 0.500.71); the RRs
were 0.56 from 8 case-control studies (95% CI, 0.420.75) and 0.64 from 8 cohort studies (95%
CI, 0.520.78). Compared with no coffee consumption, the summary RR was 0.72 (95% CI,
0.610.84) for low consumption and 0.44 (95% CI, 0.390.50) for high consumption. The
summary RR was 0.80 (95% CI, 0.770.84) for an increment of 1 cup of coffee per day. The
inverse relationship between coffee and HCC risk was consistent regardless of the subjects
sex, alcohol drinking, or history of hepatitis or liver disease.

CONCLUSIONS: From this meta-analysis, the risk of HCC is reduced by 40% for any coffee consumption vs no
consumption. The inverse association might partly or largely exist because patients with liver and
digestive diseases reduce their coffee intake. However, coffee has been shown to affect liver en-
zymes and development of cirrhosis, and therefore could protect against liver carcinogenesis.
Keywords: Chemoprevention; Epidemiology; Caffeine; Neoplasm.

condence interval [CI], 0.490.72) for coffee drinkers compared

See editorial on page 1422; see related article, Bhoo- with nondrinkers. Moreover, there was a trend in risk with dose:
Pathy N et al, on page 1486 in this issue of CGH. the summary RR was 0.70 (95% CI, 0.570.85) for low/moderate
coffee drinkers, and 0.45 (95% CI, 0.380.53) for high coffee

L iver cancer is the sixth most common cancer in the world

and the third most common cause of cancer mortality.1
In 2008, approximately 750,000 liver cancers were reported
worldwide, with approximately 700,000 deaths. Of these, more
than 80% were from low- and middle-income countries, and about
50% were from China alone. Hepatocellular carcinoma (HCC) is
the main type of liver cancer, accounting for more than 90% of
cases worldwide. HCC rates have been increasing moderately over
the past few decades in North America and Northern Europe.2,3
Chronic infections with hepatitis B and C viruses are the
main causes of HCC worldwide. Other relevant risk factors are
alcohol and alcohol-related cirrhosis, tobacco, being overweight,
and diabetes, and, in selected low-income countries, foodstuff
contamination with aatoxin. Oral contraceptives and selected
other drugs also have been related to HCC risk.36 In contrast,
coffee drinking has been related inversely to the risk of liver
cancerand specically HCCin various studies. A meta-
analysis of case-control and cohort studies published up to
February 2007,7 based on 10 studies and a total of 2260 HCC
cases, showed a summary relative risk (RR) of 0.59 (95%
drinkers vs nondrinkers, and the summary RR was 0.77 (95% CI,
0.720.82) for an increment of 1 cup of coffee per day.
Since then, 4 prospective811 and 2 case-control studies12,13
have been published on an additional 900 cases of HCC. A
Finnish prospective study of 60,323 subjects followed-up for
19 years, including 128 liver cancer cases, showed an RR of 0.32 for
drinkers of at least 8 cups of coffee per day as compared with 1 cup
or less.8 In the Japan Public Health Centerbased Prospective
Study Cohort II, which included 18,815 participants followed up
for 13 years and 110 liver cancers among subjects affected by
hepatitis C virus (HCV) or hepatitis B virus (HBV), an RR of 0.54
was found for drinkers of 3 or more cups of coffee per day
Abbreviations used in this paper: BMI, body mass index; CI, con-
dence interval; d, day; HbsAg, hepatitis B surface antigen; HBV, hepa-
titis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; OR,
odds ratio; RR, relative risk; w, week.
2013 by the AGA Institute
1542-3565/$36.00
http://dx.doi.org/10.1016/j.cgh.2013.04.039
 1414 BRAVI ET AL
Table 1. Case-Control and Cohort Studies on Coffee Consumption and HCC
Controls/cohort Follow-up period,
Study Country Cases, n size, n y (cohort studies) Adjustment Quality scorea
Case-control studies
Kuper et al,30 2000b Greece 333 360 - Age, sex 5
Gallus et al,20 2002b Italy 501 1552 - Age, sex 4
Gelatti et al,21 2005b Italy 250 500 - Age, sex, alcohol drinking, HCV, HBV 7
Ohfuji et al,26 2006b,c Japan 73 253 - Age, sex, date of rst visit, duration of liver disease, BMI, disease severity, 5
family history of liver disease, interferon therapy, tobacco smoking,
alcohol drinking, other caffeine-containing beverages
Montella et al,23 2007b Italy 185 412 - Age, sex, education, tobacco smoking, alcohol drinking, serologic 6
evidence of HCV and/or HBV infection
Tanaka, et al,27 2007d Japan 209 1253 - Age, sex, alcohol drinking, tobacco smoking 4
Kanazir et al,12 2010d Serbia 45 90 - - 5
Leung et al,13 2011c,d China 109 125 - Age, sex, tobacco smoking, alcohol drinking, tea drinking, physical activity 5
Cohort studies
Inoue et al,22 2005 Japan 334 90,452 10 Age, sex, study center, tobacco smoking, alcohol drinking, vegetable 7
consumption, tea drinking
Kurozawa et al,24 2005 Japan 258 83,966 11 Age, sex, education, history of diabetes and liver disease, tobacco 7
smoking, alcohol drinking
Shimazu et al,25 2005: Japan 70 22,404 9 Age, sex, history of liver disease, tobacco smoking, alcohol drinking 7

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 11

cohort 1
Shimazu et al,25 2005: Japan 47 38,703 6 Age, sex, history of liver disease, tobacco smoking, alcohol drinking 6
cohort 2
Hu et al,8 2008 Finland 128 60,323 30 Age, sex, study year, alcohol drinking, tobacco smoking, education, 8
diabetes, chronic liver disease, BMI
Ohishi et al,11 2008e Japan 139 472 44 - 6
Inoue et al,9 2009 Japan 110 18,815 13 Age, sex, study center, tobacco smoking, alcohol drinking, BMI, history of 8
diabetes mellitus, tea drinking, serum ALT level, HCV infection, HBV
infection
Johnson et al,10 2011 China 362 63,257 13 Age, sex, dialect group, study year, BMI, education, alcohol drinking, 8
tobacco smoking, tea drinking, history of diabetes
a
Based on the Newcastle-Ottawa Scale.16
b
Hospital-based case-control study.
c
All cases and controls had HCV.
d
Population-based case-control study.
e
Nested case-control study.
November 2013
Figure 1. Study-specic and
summary RRs of HCC for
coffee consumption vs no
consumption.
compared with never/almost never drinkers.9 In the Singapore
Chinese Health Study, a prospective cohort of 63,257 subjects
followed up for 13 years and including 362 HCC incident cases, an
RR of 0.56 was observed for drinkers of at least 3 cups per day vs
noncoffee drinkers.10 In a Japanese case-control study of 224 HCC
cases, nested within the cohort of atomic bomb survivors, an RR
of 0.40 was found for daily coffee drinkers vs nondrinkers.11 A
Serbian case-control study including 45 HCC cases showed an
odds ratio (OR) of 1.0 for coffee drinkers vs nondrinkers.12 In a
Chinese case-control study including 109 HBV-positive HCC
cases, an OR of 0.41 was found for drinkers of at least 4 cups of
coffee per week compared with noncoffee drinkers.13
To obtain an updated quantication of the association
between coffee drinking and HCC risk, we updated the previous
meta-analysis, including the results from studies published
between 2007 and 2012.
Materials and Methods
Search Strategy
We performed a PubMed/MEDLINE search of the
articles published between 1966 and September 2012, using the
terms coffee or caffeine or beverage, risk, and combi-
nations of liver or hepatocellular and carcinoma or
cancer or neoplasm, following the Meta-analysis of Obser-
vational Studies in Epidemiology.14,15 We limited the search to
studies performed in human beings. Moreover, we checked the
reference lists of the identied studies to nd any other relevant
COFFEE AND HEPATOCELLULAR CARCINOMA 1415
publication. Studies were included in the meta-analysis if they
met the following criteria: (1) provided information on the
association between coffee consumption and liver cancer,
including estimates of the RR (for cohort studies) or the OR (for
case-control studies), with the corresponding 95% CI, or fre-
quency distribution to calculate them; (2) were focused on pri-
mary liver cancer or HCC; (3) were original case-control or
cohort studies; and (4) were published as full-length articles in
English. When we found multiple articles based on the same
study population, we included only the most recent and infor-
mative one. We used the Newcastle-Ottawa Scale16 to assess the
quality of individual studies and performed a sensitivity analysis
according to the quality of each study.
From each publication, we extracted details on study design,
country, number of subjects (cases, controls, or cohort size),
duration of follow-up period (for cohort studies), frequency of
coffee consumption and/or measure of association (RR or OR)
and the corresponding 95% CI, and confounding variables
allowed for in the analyses, if any. Whenever possible, estimates
adjusted for multiple potential confounding variables were
used. When the RRor the corresponding 95% CIwas not
provided, this was derived from tabular data.
Statistical Analysis
We derived summary estimates of the RR using both
xed-effects models (ie, as weighted averages on the sum of the
inverse of the variance of the log RR/OR) and random-effects
models (ie, as weighted averages on the sum of the inverse of
1416 BRAVI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 11

Figure 2. Study-specic sum-

mary RRs of HCC for low
coffee consumption vs no
consumption.

the variance of the log RR/OR and the moment estimator Results
of the variance between studies).17 However, only the results
From the literature search we identied 76 publica-
from the latter models were presented to take into account the
tions; 2 other publications were retrieved from the references
heterogeneity of risk estimates and thus be more conservative.
of the identied articles (Supplementary Figure 1). After ex-
We carefully evaluated heterogeneity among studies through
clusions of publications not pertinent or not satisfying the
the chi-square test,18 and we quantied heterogeneity using
inclusion criteria, and further excluding 2 studies30,31 whose
the I2 statistic, which represents the percentage of the total
data were included in 2 subsequent publications20,24 and
variation across studies that is attributable to heterogeneity
1 study that had no sufcient information to obtain estimates
rather than chance.19 Furthermore, a sensitivity analysis was
of the RR,32 we considered 14 articles. Of these, 1 article re-
performed to identify which studies inuenced the most
ported data from 2 case-control studies performed in Greece
heterogeneity.
and Italy,20 which were considered separately (afterwards
We computed summary RR for any coffee consumption, and
referred to as Kuper et al, 2000 and Gallus et al, 2002), and
for low and high consumption vs no consumption (including
1 article reported results from 2 prospective cohorts (afterwards
occasional consumption). Given the differences in the intake of
referred to as Shimazu et al, 2005, cohort 1; and Shimazu et al,
coffee, different cut-off points were chosen for various study
2005, cohort 2),25 again considered as 2 separate studies. Thus,
populations: thus, the cut-off point between low and high con-
in the present meta-analysis we combined data from 16 studies
sumption was 3 cups per day in 9 studies810,13,2023 and 1 cup
(8 cohort and 8 control studies), including a total of 3153 HCC
per day in 5 studies.2427 In addition, we provided a continuous
cases (Table 1).
summary RR for an increment of 1 cup of coffee per day, using
Figure 1 shows the study-specic and summary RRs of HCC
the method proposed by Greenland and Longnecker.18
for coffee consumption vs no consumption. The summary RR
We also performed a cumulative meta-analysis to determine
from all the 16 studies813,2027 was 0.60 (95% CI, 0.500.71)
whether the association between coffee and HCC changed over
overall, 0.56 (95% CI, 0.420.75) from 8 case-control studies, and
time, and we performed separate analyses by strata of sex, alcohol
0.64 (95% CI, 0.520.78) from 8 cohort studies. Signicant
drinking, and history of hepatitis/liver diseases, including those
heterogeneity was found between studies, both among case-
studies for which such information was available.
control and cohort studies, although all studies provided risk
We showed study-specic and summary RRs for any, low, and
estimates below 1. Sensitivity analyses showed that heteroge-
high coffee consumption vs no consumption, using forest plots.
neity was mainly owing to a case-control study from Japan
Publication bias was evaluated using a funnel plot28 and was
reporting a strong inverse association27 and a large cohort study
quantied by the Eggers test.29
from China showing no signicant association.10 The summary
November 2013
Figure 3. Study-specic sum-
mary RRs of HCC for high
coffee consumption vs no
consumption.
RR of HCC for coffee consumption vs no consumption from
the 6 studies published after 2007 was 0.62 (95% CI, 0.440.87,
data not shown). The summary RR was 0.61 (95% CI, 0.500.74)
from 6 European studies8,12,20,21,23 and 0.58 (95% CI, 0.450.75)
from 10 Asian studies.911,13,22,2427
Study-specic and overall estimates of HCC for low coffee
consumption vs no consumption are shown in Figure 2. The
summary RR was 0.72 (95% CI, 0.610.84) based on 14
studies,810,13,2027 0.67 (95% CI, 0.490.92) from 7 case-control
studies, and 0.75 (95% CI, 0.620.90) from 7 cohort studies.
The corresponding estimates for high coffee consumption vs
no consumption are shown in Figure 3. The summary RR was
0.44 (95% CI, 0.390.50) overall, 0.41 (95% CI, 0.330.52) from
case-control studies, and 0.46 (95% CI, 0.380.57) from cohort
studies. The RR estimate for an increment of 1 cup of coffee per
day was 0.80 (95% CI, 0.770.84) from all the studies, 0.77 (95%
CI, 0.710.83) from case-control studies, and 0.83 (95% CI,
0.780.88) from cohort studies (data not shown).
Table 2 shows the study-specic RR of HCC and the cor-
responding CI for coffee consumption vs no consumption,
according to strata of alcohol drinking and history of hepatitis
or liver disease. Seven studies8,20,23,25,27 provided information
on coffee drinking according to strata of alcohol consumption
and gave a summary RR of 0.61 (95% CI, 0.510.72) among
never/low alcohol drinkers and of 0.60 (95% CI, 0.490.73)
among moderate/high alcohol drinkers. The summary RR was
0.59 (95% CI, 0.450.78) among subjects with a self-reported
history of hepatitis B and/or C or liver disease, and 0.52
COFFEE AND HEPATOCELLULAR CARCINOMA 1417
(95% CI, 0.390.69) among subjects with serologic evidence
of HBV and/or HCV from 9 studies.9,13,20,2326 The summary
RR in subjects with no evidence of hepatitis/liver disease
from 7 studies was 0.70 (0.560.88).20,2225 The summary RR
for coffee consumption was 0.58 (95% CI, 0.460.74) among
men, and 0.70 (95% CI, 0.550.89) among women (data not
shown).
Figure 4 shows the cumulative meta-analysis of HCC risk for
coffee consumption vs no consumption over time, from 2000 to
2011. The estimate was 0.80 (95% CI, 0.501.29) in 2000 and
decreased to 0.59 (95% CI, 0.480.72) in 2007 and was almost
stable over the past few years (RR, 0.60; 95% CI, 0.500.71).
No evidence of publication bias was found either from visual
inspection of the funnel plot (Supplementary Figure 2) or from
the Eggers test (P .379).
Quality score ranged between 4 and 7 for case-control studies
and between 6 and 8 for cohort studies (median scores: case-
control studies, 5; cohort studies, 7). The pooled RR was 0.63
(95% CI, 0.530.74) from 10 high-quality studies (ie, studies with
quality score 6), 0.58 (95% CI, 0.450.74) from 2 case-control
studies, and 0.64 (95% CI, 0.520.78) from 8 cohort studies
(data not shown).
Discussion
Since the publication of previous meta-analyses on
coffee and HCC,7,33 4 cohort811 and 2 case-control12,13 studies
have been published. The pooled estimate from the 6 studies
1418 BRAVI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 11

Table 2. Study-Specic RRs and Summary RRs, With Corresponding 95% CIs, for Coffee Consumption and HCC, According to Strata
of Alcohol Drinking and History of Hepatitis/Liver Disease
Study Strata of covariate Coffee consumption RR (95% CI)
Gallus et al, 2002; <4 drinks/d 2 vs 1 cups/d 0.9 (0.71.2)
Kuper et al, 200020 3 vs 1 cups/d 0.6 (0.40.9)
4 drinks/d 2 vs 1 cups/d 0.8 (0.51.2)
3 vs 1 cups/d 0.7 (0.51.2)
Montella et al,23 2007 Maximum lifetime alcohol intake <14 drinks/week <14 cups/week vs no coffee 0.82 (0.312.20)
1420 cups/wk vs no coffee 0.50 (0.171.48)
21 cups/wk vs no coffee 0.56 (0.211.53)
Maximum lifetime alcohol intake 14 drinks/week <14 cups/wk vs no coffee 0.86 (0.421.77)
1420 cups/wk vs no coffee 0.77 (0.371.61)
21 cups/wk vs no coffee 0.34 (0.160.72)
Tanaka et al,27 2007 <23 g alcohol/d (community controls) 1 vs <1 cups/d 0.19 (0.110.35)
23 g alcohol/d (community controls) 1 vs <1 cups/d 0.18 (0.070.44)
Shimazu et al,25 2005: cohort 1 Never alcohol drinker <1 cup/d vs never 0.69 (0.291.65)
Shimazu et al,25 2005: cohort 2 1 cup/d vs never 0.46 (0.141.52)
Former alcohol drinker <1 cup/d vs never 0.60 (0.201.78)
1 cup/d vs never 0.74 (0.232.39)
Current alcohol drinker <1 cup/d vs never 0.90 (0.471.71)
1 cup/d vs never 0.56 (0.241.29)
Hu et al,8 2008 Never alcohol drinker 23 vs 01 cups/d 0.59 (0.261.33)
45 vs 01 cups/d 0.46 (0.220.97)
67 vs 01 cups/d 0.36 (0.160.79)
8 vs 01 cups/d 0.35 (0.140.85)
Alcohol drinker 23 vs 01 cups/d 0.62 (0.271.39)
45 vs 01 cups/d 0.32 (0.130.77)
67 vs 01 cups/d 0.36 (0.150.88)
8 vs 01 cups/d 0.29 (0.110.79)
Summary estimates Never/low alcohol drinkers Moderate/high vs no/low 0.61 (0.510.72)
Moderate/high alcohol drinkers 0.60 (0.490.73)

Gallus et al,20 2002; Self-reported hepatitis 2 vs 1 cups/d 0.6 (0.31.1)

Kuper et al,30 2000 3 vs 1 cups/d 0.5 (0.21.3)
No self-reported hepatitis 2 vs 1 cups/d 0.9 (0.71.2)
3 vs 1 cups/d 0.7 (0.50.9)
Inoue et al,22 2005 HCV and HBV negative 1-2 cups/wk vs almost never 0.87 (0.411.87)
34 cups/wk vs almost never 0.57 (0.191.70)
12 cups/d vs almost never 0.50 (0.201.20)
Inoue et al,9 2009 HCV and/or HBsAg positive <1 cup vs almost never 0.55 (0.330.93)
12 cups vs almost never 0.47 (0.240.93)
3 cups vs almost never 0.61 (0.231.62)
Kurozawa et al,24 2005 Self-reported history of liver disease <1 cup/d vs nondrinker 0.94 (0.531.66)
1 cup/d vs nondrinker 0.44 (0.220.88)
No self-reported history of liver disease <1 cup/d vs nondrinker 0.79 (0.441.41)
1 cup/d vs nondrinker 0.61 (0.321.16)
Leung et al,13 2011 HBV positive Drinkers vs nondrinkers 0.54 (0.300.97)
Montella et al,23 2007 HCV and/or HBsAg positive <14 cups/wk vs nondrinkers 0.56 (0.152.15)
1420 cups/wk vs nondrinkers 0.36 (0.091.36)
21 cups/wk vs nondrinkers 0.67 (0.162.85)
HCV and/or HBsAg negative <14 cups/wk vs nondrinkers 0.55 (0.221.39)
1420 cups/wk vs nondrinkers 0.35 (0.121.01)
21 cups/wk vs nondrinkers 0.27 (0.100.73)
Ohfuji et al,26 2006 HCV positive Drinkers vs nondrinkers 0.47 (0.211.05)
Shimazu et al,25 2005: cohort 1 Self-reported history of liver disease Occasional vs never 0.51 (0.270.97)
Shimazu et al,25 2005: cohort 2 1 vs never 0.52 (0.251.07)
No self-reported history of liver disease Occasional vs never 0.98 (0.531.80)
1 vs never 0.75 (0.371.50)
Summary estimates Self-reported hepatitis/liver disease Moderate/high vs no/low 0.59 (0.450.78)
Serologic evidence of HBV or HCV 0.52 (0.390.69)
No evidence of hepatitis/liver disease 0.70 (0.560.88)
November 2013
Figure 4. Cumulative meta-
analysis of studies on coffee
consumption and HCC. RR for
coffee consumption vs no
consumption.
published after 2007 was comparable with that of previous ones.
Thus, recent studies add up to the evidence that coffee drinkers
have a 40% reduced risk of HCC compared with nondrinkers,
and high drinkers have a more than 50% risk reduction.
Coffee contains several bioactive compounds with potential
favorable effects on health,34 including minerals and antioxi-
dants, mainly phenolic compounds (eg, chlorogenic, caffeic,
ferulic, and cumaric acids), melanoidins and diterpenes (eg,
cafestol and kahweol). In particular, chlorogenic acid and
other antioxidant substances from coffee beans have been
indicated to have an inhibitory effect on liver carcinogenesis.35
Studies in animal models and cell culture systems have sug-
gested potential anticarcinogenic effects of cafestol and kah-
weol against aatoxin B1induced genotoxicity in both rats
and human beings,3638 although it is not clear if these com-
pounds reach sufcient levels for such actions in human be-
ings after coffee intake. Coffee consumption has been related
inversely to the activity of some hepatic enzymes, including
glutamyltransferase, an indicator of several liver diseases,39 and
serum alanine aminotransferase.40 Coffee components were
found to modify the xenobiotic metabolism, inducing
glutathione-S-transferase and inhibiting N-acetyltransferase.41
Moreover, coffee consumption has been associated inversely
with cirrhosis,4245 and thus it is possible that the favorable
effect of coffee on HCC is mediated by its benecial effects on
cirrhosis, suggesting the existence of a continuum of clinical
and epidemiologic evidence in the favorable effect of coffee,
which spans from liver enzymes to cirrhosis to HCC. More-
over, part of the favorable effect of coffee on HCC might be
mediated by its prevention of diabetes, a known risk factor for
HCC.4648
The present results may be affected by bias and confounding
inherent to the observational studies included in the meta-
analysis. Coffee consumption is based on patients self-reported
information. However, information on coffee consumption has
been shown to be satisfactorily reproducible and valid.4951
Moreover, the inverse relation between coffee and HCC was
observed in both retrospective (case-control) and prospective
(cohort) studies, in populations from Europe and Asia, thus
reassuring against a major role of bias. Allowance for confounding
factors varied among different studies, but the fact that the inverse
relation persisted after allowance for major risk factors for HCC,
COFFEE AND HEPATOCELLULAR CARCINOMA 1419
including evidence of hepatitis B and C infection, cirrhosis and
other liver diseases, social class indicators, alcohol drinking, and
tobacco smoking, reassures against a major role of confounding.
Moreover, the inverse relation between coffee and HCC were
consistent across different categories at increased risk of HCC,
particularly subjects with a history of HBV/HCV, liver disease, and
alcohol drinkers. The inverse association also was consistent over
time, pointing against the hypothesis of a false-positive result.52
Signicant heterogeneity between studies was found, which
mainly was owing to a case-control study from Japan27 and a
cohort study from China.10 However, because results from these
studies diverged from the pooled estimate in opposite directions,
their exclusion did not modify the results substantially. Further,
the association was of similar magnitude when only high-quality
studies were considered.
Still, despite the consistency of results across studies, time
periods, and populations, it is difcult to establish whether the
inverse relation between coffee drinking and HCC is causal on
the basis of evidence from observational studies only.53 This
inverse relation indeed may be partly attributable to the fact that
patients with a broad spectrum of digestive tract conditions,
including liver disease and cirrhosis, may reduce their coffee
intake. Thus, a reduction of coffee consumption in unhealthy
subjects cannot be ruled out, although the inverse relation be-
tween coffee and liver cancer also was present in subjects with no
history of hepatitis/liver disease. Publication bias also is possible
in meta-analyses, with selective publication of favorable results,
particularly when a relatively small number of studies is avail-
able. However, we did not nd any relevant asymmetry in the
funnel plot, and the Eggers test was not statistically signicant.
Thus, publication bias does not seem to have inuenced the
relation between coffee and HCC appreciably.
Further, it remains difcult to translate the apparent inverse
relation between coffee drinking and liver cancer risk observed
in epidemiologic studies into potential implications on a public
health level. Primary liver cancers are the most largely avoidable
neoplasms, through HBV vaccination, control of HCV trans-
mission, and reduction of alcohol drinking. These three mea-
sures can, in principle, avoid more than 90% of primary liver
cancers worldwide. It remains unclear whether coffee drinking
has an additional role in liver cancer prevention, but in any case
such a role would be limited compared with that achievable
1420 BRAVI ET AL
through control of HBV, HCV, and reduction of alcohol
intake.53
In summary, the present meta-analysis conrmed that coffee
consumption is associated with a reduction in HCC risk of
approximately 40%. The protective effect of coffee was consistent
across different populations and subgroups at increased HCC
risk. However, the issue of causality and the role of specic coffee
components remain open to discussion.
Supplementary Material
Note: To access the supplementary material accom-
panying this article, visit the online version of Clinical
Gastroenterology and Hepatology at www.cghjournal.org, and at
http://dx.doi.org/10.1016/j.cgh.2013.04.039.
References
1. Jemal A, Bray F, Center MM, et al. Global cancer statistics. CA
Cancer J Clin 2011;61:6990.
2. Bosetti C, Levi F, Boffetta P, et al. Trends in mortality from hepa-
tocellular carcinoma in Europe, 1980-2004. Hepatology 2008;
48:137145.
3. Chuang SC, La Vecchia C, Boffetta P. Liver cancer: descriptive
epidemiology and risk factors other than HBV and HCV infection.
Cancer Lett 2009;286:914.
4. Franceschi S, Montella M, Polesel J, et al. Hepatitis viruses, alcohol,
and tobacco in the etiology of hepatocellular carcinoma in Italy.
Cancer Epidemiol Biomarkers Prev 2006;15:683689.
5. La Vecchia C, Bosetti C. Oral contraceptives and neoplasms other
than breast and female genital tract. Eur J Cancer Prev 2009;
18:407411.
6. Trichopoulos D, Bamia C, Lagiou P, et al. Hepatocellular carcinoma
risk factors and disease burden in a European cohort: a nested
case-control study. J Natl Cancer Inst 2011;103:16861695.
7. Bravi F, Bosetti C, Tavani A, et al. Coffee drinking and hepatocellular
carcinoma risk: a meta-analysis. Hepatology 2007;46:430435.
8. Hu G, Tuomilehto J, Pukkala E, et al. Joint effects of coffee con-
sumption and serum gamma-glutamyltransferase on the risk of liver
cancer. Hepatology 2008;48:129136.
9. Inoue M, Kurahashi N, Iwasaki M, et al. Effect of coffee and green
tea consumption on the risk of liver cancer: cohort analysis by
hepatitis virus infection status. Cancer Epidemiol Biomarkers Prev
2009;18:17461753.
10. Johnson S, Koh WP, Wang R, et al. Coffee consumption and
reduced risk of hepatocellular carcinoma: ndings from the
Singapore Chinese Health Study. Cancer Causes Control 2011;
22:503510.
11. Ohishi W, Fujiwara S, Cologne JB, et al. Risk factors for hepatocel-
lular carcinoma in a Japanese population: a nested case-control
study. Cancer Epidemiol Biomarkers Prev 2008;17:846854.
12. Kanazir M, Boricic I, Delic D, et al. Risk factors for hepatocellular
carcinoma: a case-control study in Belgrade (Serbia). Tumori 2010;
96:911917.
13. Leung WW, Ho SC, Chan HL, et al. Moderate coffee consumption
reduces the risk of hepatocellular carcinoma in hepatitis B chronic
carriers: a case-control study. J Epidemiol Community Health 2011;
65:556558.
14. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observa-
tional studies in epidemiology: a proposal for reporting. Meta-
analysis Of Observational Studies in Epidemiology (MOOSE)
group. JAMA 2000;283:20082012.
15. Kanwal F, White D. "Systematic reviews and meta-analyses" in
Clinical Gastroenterology and Hepatology. Clin Gastroenterol Hep-
atol 2012;10:11841186.
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 11
16. Ottawa Hospital Research Institute. The Newcastel-Ottawa
Scale (NOS) for assessing the quality of nonrandomised studies
in meta-analyses. Available at: www.ohri.ca/programs/clinical_
epidemiology/oxford.asp. Accessed: March 13, 2013.
17. Greenland S. Quantitative methods in the review of epidemiologic
literature. Epidemiol Rev 1987;9:130.
18. Greenland S, Longnecker MP. Methods for trend estimation from
summarized dose-response data, with applications to meta-anal-
ysis. Am J Epidemiol 1992;135:13011309.
19. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-
analysis. Stat Med 2002;21:15391558.
20. Gallus S, Bertuzzi M, Tavani A, et al. Does coffee protect against
hepatocellular carcinoma? Br J Cancer 2002;87:956959.
21. Gelatti U, Covolo L, Franceschini M, et al. Coffee consumption
reduces the risk of hepatocellular carcinoma independently
of its aetiology: a case-control study. J Hepatol 2005;42:528534.
22. Inoue M, Yoshimi I, Sobue T, et al. Inuence of coffee drinking on
subsequent risk of hepatocellular carcinoma: a prospective study in
Japan. J Natl Cancer Inst 2005;97:293300.
23. Montella M, Polesel J, La Vecchia C, et al. Coffee and tea con-
sumption and risk of hepatocellular carcinoma in Italy. Int J Cancer
2007;120:15551559.
24. Kurozawa Y, Ogimoto I, Shibata A, et al. Coffee and risk of death
from hepatocellular carcinoma in a large cohort study in Japan. Br
J Cancer 2005;93:607610.
25. Shimazu T, Tsubono Y, Kuriyama S, et al. Coffee consumption and
the risk of primary liver cancer: pooled analysis of two prospective
studies in Japan. Int J Cancer 2005;116:150154.
26. Ohfuji S, Fukushima W, Tanaka T, et al. Coffee consumption and reduced
risk of hepatocellular carcinoma among patients with chronic type C liver
disease: a case-control study. Hepatol Res 2006;36:201208.
27. Tanaka K, Hara M, Sakamoto T, et al. Inverse association between
coffee drinking and the risk of hepatocellular carcinoma: a case-
control study in Japan. Cancer Sci 2007;98:214218.
28. Thornton A, Lee P. Publication bias in meta-analysis: its causes and
consequences. J Clin Epidemiol 2000;53:207216.
29. Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis
detected by a simple, graphical test. BMJ 1997;315:629634.
30. Kuper H, Tzonou A, Kaklamani E, et al. Tobacco smoking, alcohol
consumption and their interaction in the causation of hepatocel-
lular carcinoma. Int J Cancer 2000;85:498502.
31. Wakai K, Kurozawa Y, Shibata A, et al. Liver cancer risk, coffee, and
hepatitis C virus infection: a nested case-control study in Japan. Br J
Cancer 2007;97:426428.
32. Yu SZ, Huang XE, Koide T, et al. Hepatitis B and C viruses infection,
lifestyle and genetic polymorphisms as risk factors for hepatocellular
carcinoma in Haimen, China. Jpn J Cancer Res 2002;93:12871292.
33. Larsson SC, Wolk A. Coffee consumption and risk of liver cancer: a
meta-analysis. Gastroenterology 2007;132:17401745.
34. Ferruzzi MG. The inuence of beverage composition on delivery of
phenolic compounds from coffee and tea. Physiol Behav 2010;
100:3341.
35. Tanaka T, Nishikawa A, Shima H, et al. Inhibitory effects of chloro-
genic acid, reserpine, polyprenoic acid (E-5166), or coffee on hep-
atocarcinogenesis in rats and hamsters. Basic Life Sci 1990;
52:429440.
36. Cavin C, Holzhauser D, Constable A, et al. The coffee-specic
diterpenes cafestol and kahweol protect against aatoxin B1-
induced genotoxicity through a dual mechanism. Carcinogenesis
1998;19:13691375.
37. Cavin C, Holzhaeuser D, Scharf G, et al. Cafestol and kahweol, two
coffee specic diterpenes with anticarcinogenic activity. Food Chem
Toxicol 2002;40:11551163.
38. Majer BJ, Hofer E, Cavin C, et al. Coffee diterpenes prevent the
genotoxic effects of 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyri-
dine (PhIP) and N-nitrosodimethylamine in a human derived liver
cell line (HepG2). Food Chem Toxicol 2005;43:433441.
November 2013
39. Tanaka K, Tokunaga S, Kono S, et al. Coffee consumption and decreased
serum gamma-glutamyltransferase and aminotransferase activities
among male alcohol drinkers. Int J Epidemiol 1998;27:438443.
40. Ruhl CE, Everhart JE. Coffee and caffeine consumption reduce the
risk of elevated serum alanine aminotransferase activity in the
United States. Gastroenterology 2005;128:2432.
41. Huber WW, Parzefall W. Modication of N-acetyltransferases
and glutathione S-transferases by coffee components: possible
relevance for cancer risk. Methods Enzymol 2005;401:307341.
42. Corrao G, Lepore AR, Torchio P, et al. The effect of drinking coffee
and smoking cigarettes on the risk of cirrhosis associated with
alcohol consumption. A case-control study. Provincial Group for
the Study of Chronic Liver Disease. Eur J Epidemiol 1994;
10:657664.
43. Corrao G, Zambon A, Bagnardi V, et al. Coffee, caffeine, and the risk
of liver cirrhosis. Ann Epidemiol 2001;11:458465.
44. Gallus S, Tavani A, Negri E, et al. Does coffee protect against liver
cirrhosis? Ann Epidemiol 2002;12:202205.
45. Klatsky AL, Armstrong MA. Alcohol, smoking, coffee, and cirrhosis.
Am J Epidemiol 1992;136:12481257.
46. Huxley R, Lee CM, Barzi F, et al. Coffee, decaffeinated coffee, and
tea consumption in relation to incident type 2 diabetes mellitus: a
systematic review with meta-analysis. Arch Intern Med 2009;
169:20532063.
47. El-Serag HB, Hampel H, Javadi F. The association between diabetes
and hepatocellular carcinoma: a systematic review of epidemiologic
evidence. Clin Gastroenterol Hepatol 2006;4:369380.
48. Polesel J, Zucchetto A, Montella M, et al. The impact of obesity and
diabetes mellitus on the risk of hepatocellular carcinoma. Ann
Oncol 2009;20:353357.
COFFEE AND HEPATOCELLULAR CARCINOMA 1421
49. DAvanzo B, La Vecchia C, Katsouyanni K, et al. Reliability of in-
formation on cigarette smoking and beverage consumption pro-
vided by hospital controls. Epidemiology 1996;7:312315.
50. Ferraroni M, Tavani A, Decarli A, et al. Reproducibility and validity of
coffee and tea consumption in Italy. Eur J Clin Nutr 2004;
58:674680.
51. Jacobsen BK, Bonaa KH. The reproducibility of dietary data from a
self-administered questionnaire. The Tromso Study. Int J Epidemiol
1990;19:349353.
52. Boffetta P, McLaughlin JK, La Vecchia C, et al. False-positive results
in cancer epidemiology: a plea for epistemological modesty. J Natl
Cancer Inst 2008;100:988995.
53. La Vecchia C. Cancer and liver cancer prevention: is it a fact or just
a potential? Hepatology 2008;48:79.
Reprint requests
Address requests for reprints to: Carlo La Vecchia, MD, Department of
Epidemiology, IRCCS Istituto di Ricerche Farmacologiche Mario Negri,
Via G. La Masa 19, 20156 Milan, Italy. e-mail: carlo.lavecchia@marionegri.it;
fax: (39) 0233200231.
Acknowledgments
The authors thank Mrs Ivana Garimoldi for editorial assistance.
Conicts of interest
The authors disclose no conicts.
Funding
Supported by the Associazione Italiana per la Ricerca sul Cancro
(AIRC) (grant no. 10068).
1421.e1 BRAVI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 11

Supplementary Figure 2. Funnel plot of studies on coffee con-

sumption and HCC. RR, relative risk for coffee consumption vs no
consumption.

Supplementary Figure 1. Flow-chart of the selection of studies on

coffee consumption and HCC included in the meta-analysis.