Research Article

Coffee and herbal tea consumption is associated with lower

liver stiffness in the general population: The Rotterdam study
Louise J.M. Alferink1, Juliana Fittipaldi1,2, Jessica C. Kiefte-de Jong2,3, Pavel Taimr1,
Bettina E. Hansen1,4, Herold J. Metselaar1, Josje D. Schoufour2, M. Arfan Ikram2,5,6,
Harry L.A. Janssen1,4, Oscar H. Franco2, Sarwa Darwish Murad1,⇑
1
Department of Gastroenterology and Hepatology, Erasmus MC University Medical Centre, Rotterdam, The Netherlands; 2Department of
Epidemiology, Erasmus MC University Medical Centre, Rotterdam, The Netherlands; 3Leiden University College, The Hague, The Netherlands;
4
Liver Centre, Toronto Western & General Hospital, University Health Network, Toronto, Canada; 5Department of Radiology, Erasmus MC
University Medical Medical Centre, Rotterdam, The Netherlands; 6Department of Neurology, Erasmus MC University Medical Medical Centre,
Rotterdam, The Netherlands

Background & Aims: Coffee and tea have been proposed to limit Lay summary: The Rotterdam Study is a large ongoing popula-
the progression of liver fibrosis in established liver disease, but it tion study of suburban inhabitants of Rotterdam in whom data
is unknown if this is also true for subclinical fibrosis. We there- on liver stiffness, as proxy for liver fibrosis, presence of fatty liver
fore aimed to evaluate whether coffee and tea consumption are on ultrasound and detailed information on coffee and tea con-
associated with liver stiffness in the general population. sumption were obtained in 2,424 participants. The consumption
Methods: The Rotterdam Study is an ongoing prospective of herbal tea and daily consumption of three or more cups of cof-
population-based cohort. We included participants who under- fee was related to the presence of lower liver stiffness, indepen-
went transient elastography, ultrasound and completed a food dent of a great number of other lifestyle and environmental
frequency questionnaire. Coffee and tea consumption were cate- factors. Previous studies have found a protective effect of coffee
gorized into no, moderate ([0–3), or frequent (P3) intake (cups/ on established liver disease and we now show for the first time
day), and tea further into green, black and herbal tea (no/any). that this effect is already measurable in the general population.
Significant fibrosis was defined as liver stiffness measurements Ó 2017 European Association for the Study of the Liver. Published
(LSM) P8.0 kPa. We performed regression analyses relating cof- by Elsevier B.V. All rights reserved.
fee and tea intake with fibrosis, steatosis and log-transformed
LSM and adjusted for energy, sugar and creamer intake, age, gen-
der, BMI, steatosis/LSM, HOMA-IR, ALT, alcohol, smoking, soda,
healthy diet index and physical activity. Introduction
Results: We included 2,424 participants (age 66.5 ± 7.4; 43%
male) of whom 5.2% had LSM P8.0 kPa and 34.6% steatosis. Chronic liver diseases constitute a major public health problem.
Proportion of LSM P8.0 kPa decreased with higher coffee Liver cirrhosis was the 12th cause of death worldwide and the
consumption (7.8%, 6.9% and 4.1% for no, moderate and frequent sixth cause of life-years lost in the adult population in developed
respectively; Ptrend = 0.006). This inverse association was countries in 2010.1,2 Chronic liver diseases are often silent for
confirmed in multivariable regression (ORmod 0.75, 95% CI over 20 years until cirrhosis develops. Indeed, several studies
0.33–1.67; ORfreq 0.39, 95% CI 0.18–0.86; p = 0.005). Amongst have suggested that liver fibrosis may be present within unse-
tea consumers, only herbal tea consumers (36.3%) had lower lected individuals. Using transient elastography (TE) as a diagnos-
log-transformed LSM after adjustment (Beta-0.05, 95% tic tool for liver fibrosis, a prevalence of 6–7% was found in the
CI-0.08;-0.02, p = 0.001). Subtypes of tea were associated with general population3,4 and even up to 17% in those high-risk pop-
steatosis in univariate but not multivariable analysis. ulations with metabolic syndrome and type 2 diabetes.5
Conclusions: In the general population, frequent coffee and her- Lifestyle is an important factor in the pathogenesis of many
bal tea consumption were inversely related with liver stiffness liver diseases, examples of which include alcohol abuse in alco-
but not steatosis. Longitudinal analyses, as well as studies holic liver disease and high caloric diet and inactivity in non-
validating and unravelling underlying mechanisms are needed. alcoholic fatty liver disease (NAFLD). At the same time, a healthy
lifestyle, such as implementing a well-balanced diet or consump-
tion of nutraceuticals, i.e. foods or nutrients with a health benefit,
Keywords: General population; Liver stiffness; Nutraceuticals; Steatosis; Herbal
tea; Coffee.
can prevent and even attenuate liver disease.6
Received 24 October 2016; received in revised form 7 March 2017; accepted 12 March Coffee and tea are the most consumed beverages worldwide
2017 and emerging as promising nutraceuticals for liver health.7 Both
⇑ Corresponding author. Address: Office Ha-211, Postbus 2040, 3000 CA
beverages are part of well-rooted cultural traditions and also
Rotterdam, The Netherlands. Tel.: +31 628021137; fax: +31 10 436 5916.
represent the second most traded commodity on the world
E-mail address: [email protected] (S. Darwish Murad).

Journal of Hepatology 2017 vol. xxx j xxx–xxx

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
Research Article
markets.8 Consumption of these nutraceuticals has been associ- dichotomized into no (0) and any ([0) consumption. Additionally, data on con-
ated with lower all-cause and cause-specific mortality, presum- sumption of soda, alcohol, sugar and cream use in coffee or tea were obtained
and used as covariates (in g/day). Excessive alcohol consumption was defined
ably through reducing the risk of features of the metabolic as [14 units per week for women and [21 units per week for men (one unit
syndrome.9,10 equals 10 g). Furthermore, to account for confounding by overall dietary quality,
Coffee consumption was for the first time associated to liver the Dutch Healthy Diet Index (DHDI) was added to multivariable analyses.26
health, that is lower liver enzymes, almost two decades ago.11
Evidence supporting this protective effect of coffee on liver Liver stiffness measurements and hepatic steatosis
enzymes rapidly emerged henceforth.12 Coffee consumption
seemed to attenuate alcoholic liver disease13 and studies in hep- LSM were performed using TE (FibroScanÒ, EchoSens, Paris, France) by a single
atitis C showed less severe fibrosis in coffee consumers.14 Also in operator who had performed more than 1,000 examinations before the start of
the study. Practical implementation of TE has been described previously.3 The
NAFLD patients, coffee consumption was inversely associated operator obtained 10 serial measurements of stiffness, using the M- or the XL-
with fibrosis grade, but inconclusive concerning the relation to probe according to the manufacturer’s instructions. Participants were excluded
steatosis.15–17 To date, only one study examined the correlation from our analyses if: 1) LSM did not meet the reliability criteria of Boursier
between coffee and liver fibrosis in the general population, albeit et al., i.e. interquartile range (IQR)/median LSM [0.30 with median LSM
P7.1 kPa;27 2) no LSM was obtained after at least 10 shots (defined as failure)
using a surrogate serum biomarker test as a proxy for fibrosis,
and; 3) intra-cardiac devices or physical disabilities prohibited the use of TE.
and found lower odds for fibrosis in frequent coffee consumers.17 Clinically relevant fibrosis was defined as LSM P8.0 kPa and clinically rele-
The association between tea and liver health is less well estab- vant cirrhosis as LSM P13.0 kPa. At these cut-off levels, previous studies showed
lished than that of coffee. All studies are either limited to Asian high positive predictive values for the presence of liver fibrosis and cirrhosis,
populations or include only serum transaminases as a primary respectively.4,28,29
Abdominal US was carried out by a certified and experienced technician on
endpoint. In addition, results of these studies are inconclusive
Hitachi HI VISION 900. Images were stored digitally and re-evaluated by a single
regarding the presumed health benefit of tea.18–22 hepatologist with more than 10 years of experience in US (P.T.). Diagnosis of
To our knowledge, there are no studies examining whether steatosis was determined dichotomously according to the protocol of Hamaguchi
coffee and tea consumption are associated with a lower preva- et al.,30 as presence or absence of a hyper-echogenic liver parenchyma.
lence of steatosis and liver fibrosis, using reliable imaging tech-
niques, in the general population. Hence, we conducted a cross- Biochemistry
sectional analysis of individuals within a large prospective cohort
study, who completed extensive dietary questionnaires, liver Fasting blood samples were collected just before US and TE imaging. Blood lipids,
platelet count, glucose, alanine aminotransferase (ALT), aspartate aminotrans-
stiffness measurements (LSM), as proxy for fibrosis, and hepatic
ferase, gamma-glutamyltransferase (GGT), alkaline phosphatase and total biliru-
ultrasound (US) for the diagnosis of steatosis. Our aim was to bin were measured using automatic enzyme procedures (Roche Diagnostic
determine whether coffee and tea consumption were associated GmbH, Mannheim, DE). Insulin, hepatitis B surface antigen and anti-hepatitis C
with lower risk of liver fibrosis and steatosis in the general virus were measured by an automatic immunoassay (Roche Diagnostic GmbH).
Patients with viral hepatitis were excluded from the analyses.
population.

Additional covariates

Participants and methods Data concerning demographics, education level, medical history, physical activ-
ity, comorbid conditions, smoking behavior and drug use were obtained during
Study population an extensive home interview by trained interviewers. Detailed information on
medication use was obtained from automated linkage to pharmacies with which
This is a cross-sectional analysis of The Rotterdam Study, a large ongoing 98% of the participants were registered. Anthropometric measurements were per-
population-based cohort of participants aged 45 years and older living in a suburb formed by well-trained research assistants. Body mass index (BMI) was calcu-
of Rotterdam, The Netherlands. The rationale and design of this study have been lated as weight (kg)/height (m2) and waist circumference (WC) in centimeters.
described previously and a more detailed description of the design is added as Blood pressure measurements were taken as the average of two subsequent mea-
Supplementary methods.23 For the purpose of our study, all participants visiting surements on the same day in upright position. According to the Adult Treatment
the research center between January 2011 and September 2013 were included Panel III criteria,31 metabolic syndrome was diagnosed if at least three of the fol-
and underwent anthropometric assessment, abdominal US, TE and blood sam- lowing five traits were present: 1) abdominal obesity, defined as WC [102 cm
pling. For all cohorts, this was the first hepatic examination in The Rotterdam (40 inch) in men and [88 cm (35 inch) in women; 2) serum triglycerides
Study. The Rotterdam Study has been approved by the institutional review board P150 mg/dl (1.0 mmol/L) or drug treatment for elevated triglycerides; 3) serum
(Medical Ethics Committee) of the Erasmus MC University Medical Center Rotter- high-density lipoprotein cholesterol (HDL-C) \40 mg/dl (1.0 mmol/L) in men and
dam and by the review board of The Netherlands Ministry of Health, Welfare and \50 mg/dl (1.3 mmol/L) in women or drug treatment for low HDL-C; 4) blood
Sports. Written informed consent was obtained from all participants. pressure P130/85 mmHg or drug treatment for elevated blood pressure; 5) fast-
ing plasma glucose (FPG) P100 mg/dl (5.6 mmol/L) or drug treatment for ele-
vated blood glucose. Homeostasis model assessment of insulin resistance
Coffee and tea consumption (HOMA-IR) was used as proxy for insulin resistance and calculated by multiplying
fasting glucose (mmol/dl) by fasting insulin (mU/L) divided by 22.5.32
All participants completed an externally validated 389-item food frequency ques-
tionnaire (FFQ) developed for Dutch adults.24,25 The questionnaire addressed the Statistical analyses
type of food consumed over the last month, as well as frequency, portion size and
preparation methods. Incomplete or unreliable FFQs, i.e. total energy intake less
than 500 or more than 7,500 kilocalories per day, were excluded. Questions Population characteristics were described using descriptive statistics. Continuous
regarding coffee and tea consumption included: ‘‘How often did you drink coffee data were presented as mean ± standard deviation or median with interquartile
last month?” and ‘‘How often did you drink black/green/herbal tea (e.g. chamo- range (IQR) according to the distribution of the variable. Chi-square test,
mile, red bush and nettle) last month?” to which the possible answers were: Student’s t test or Wilcoxon rank-sum test were used to assess significant
‘‘1) not at all, 2) 1 or 2–3 times, or 3) 1; 2–3; 4–5 or 6–7 times a day”, and in case differences in distribution of categorical, normally distributed and not-normally
of daily consumption ‘‘1–2; 3–4; 5–6; 7–8; 9–10 or 11 or more cups per day”. Cof- distributed data, respectively. Univariate linear and logistic regression analyses
fee and tea consumption was thereafter categorized into no (0), moderate ([0–3) were performed to examine the association between coffee, overall tea and its
and frequent (P3) consumption (in cups/day, one cup equals 150 g). Tea con- subtypes herbal, green and black tea consumption and LSM as continuous and
sumption was further specified into herbal, green and black tea and subsequently dichotomous (LSM \8.0 kPa vs. LSM P8.0 kPa) variable. In all regression models,

2 Journal of Hepatology 2017 vol. xxx j xxx–xxx

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
 JOURNAL OF HEPATOLOGY
multivariable adjustments were made by including age, gender, energy intake,
RS III-1
BMI, insulin resistance, steatosis, serum ALT, excessive alcohol intake, soda con-
sumption, smoking status, cream use in coffee and sugar use in tea or coffee, - FFQ-data
DHDI, type of probe, physical activity and education level as potential covari-
2006-2008
ates.33 To evaluate whether there was interaction with gender, HOMA-IR, BMI
and steatosis on the one hand and coffee and tea consumption on the other hand,
we evaluated the interaction terms (e.g. coffee x gender) in the multivariable
model. In case of p \0.05 stratified analyses were performed. Additionally, corre- RS I-5 and RS II-3 RS III-2 Data availble for
lation and multicollinearity was tested between steatosis and liver stiffness, given N = 1,353 N = 2,086 all cohort visits
their strong relation, i.e. simple steatosis is a well-known risk factor for fibrosis. Extensive physical
Data available Data available examination
Furthermore, predicted probabilities of having LSM P8.0 kPa were calculated,
- FFQ data - Abdominal ultrasound Blood sampling
using the fully adjusted multivariable logistic regression model with LSM
- Abdominal ultrasound - FibroScan Drug use
P8.0 kPa as a dependent variable. Probabilities were graphically depicted for - FibroScan
the different coffee categories and presence of steatosis, to give more insight in Comorbidities
to these possibly interrelated covariates. Data are expressed as mean (95% CI) per- Demographics
2009-2012 2012-2014 Anthropometry
centage. Also, different cut-offs for significant fibrosis were tested as outcome
variable in univariate logistic regression, to emphasize the robustness of our find-
ings. Additionally, a cut-off of P6.2 kPa was used, indicating significant fibrosis in
participants with steatosis. This cut-off has been recently proposed to exhibit a
high sensitivity for staging F2 fibrosis in participants with NAFLD.34 Univariate Eligible participnats
and multivariable logistic regression analyses were carried out assessing the for this study
association between coffee, tea and the secondary outcome of interest, hepatic N = 3,439
steatosis. In addition to the previous mentioned potential covariates, LSM was Intracardiac device n = 27
added to the model as well. Physical disability n = 3
Several sensitivity analyses were performed to test the robustness of the data. Maintenance of
Firstly, subgroup analysis was done by stratifying participants who filled in FFQ FibroScan n = 67
FibroScan
data 5.8 years prior to liver imaging (RS III-2) and, participants who filled in
performed
FFQ simultaneously with liver imaging (RS I-5 and RS II-3). Because dietary data
n = 3,342
are known to be stable over time,35 RS III-2 was included in the total study pop-
ulation. Secondly, to ensure that the study population is a general population Failure LSM n = 162
with no known liver disease (even though participants with viral hepatitis were Unreliable LSM n = 139
excluded based on serology and participants were asked to report liver comor-
bidities during home interviewing), a sensitivity analysis was performed, exclud- Reliable LSM
ing abnormal levels of ALT ([2 x upper limit of normal, which is 80 U/L for men n = 3,041
and 60 U/L for women). A p value of \0.05 was considered statistically significant.
All analyses were performed using SPSS 21.0 (SPSS Inc., Chicago, IL, USA). Incomplete dietary
There are no further details regarding the materials used, please refer to the data n = 566
CTAT table. Unreliable FFQs n = 34

Reliable FFQs
n = 2,441
Results
Hepatitis C n = 14
Hepatitis B n = 3
Study population
Total study
population
The flowchart of the study population is depicted in Fig. 1. Of the
n = 2,424
3,439 participants visiting the research center, participant failure
of LSM was reported in 162 (4.8%) and unreliable results in 139
(4.2%) participants. In addition, 34 (1.1%) participants with unre- Fig. 1. Flowchart of the study. RS, Rotterdam Study; I-III, number of cohort; 1–3
and 5, times cohort visited.
liable FFQs and 566 (16.4%) with incomplete FFQs were excluded.
Hence, the total study population was n = 2,424. Population char-
acteristics are presented in Table 1. Fifty-seven percent of the female gender, less excessive alcohol use, less smoking, lower
participants were women, mean age was 66.5 ± 7.4 years, mean BMI and less features of the metabolic syndrome (Table S1).
BMI 27.2 ± 4.0 kg/m2 and median LSM was 4.7 kPa (3.8–5.8). A Amongst subtypes of tea, black tea was most commonly con-
total of 125 (5.2%) participants had LSM P8.0 kPa, suggesting sumed (64.2%) followed by herbal (36.3%) and green tea
presence of significant fibrosis. (26.1%). Only 32 (1.3%) participants reported neither coffee nor
tea consumption.
Data on coffee and tea consumption
Coffee consumption and liver stiffness
Dietary data are shown in Table 1. In total, 2,258 participants
(93.2%) consumed coffee with a mean consumption of Proportion of participants with LSM P8.0 kPa decreased with
2.6 ± 1.7 cups/day. Frequent coffee consumers were proportion- increasing coffee consumption (7.8% in no, 6.9% in moderate
ally more overweight or obese. Furthermore, it was associated and 4.1% in frequent coffee consumption groups, Ptrend = 0.006).
with younger age, male gender, Caucasian ethnicity, higher edu- Using no coffee consumption as a reference, univariate logistic
cation level, more current or past smoking, lower prevalence of regression showed an inverse association between coffee con-
lipid disorders and type 2 diabetes mellitus (Table S1). No differ- sumption and LSM P8.0 kPa (ORmod 0.87, 95% CI 0.46–1.64; ORfreq
ence was observed in serum ALT and HOMA-IR. Tea was con- 0.50, 95% CI 0.27–0.94; p = 0.007). This association enhanced after
sumed by 2,052 (84.7%) participants, with a median of 1.2 (0.4– adjusting for energy intake, age, gender, BMI, steatosis, HOMA-IR,
2.7) cups/day. Frequent tea consumption was associated with ALT, excessive alcohol use, current or former smoking, soda,

Journal of Hepatology 2017 vol. xxx j xxx–xxx 3

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
Research Article
Table 1. Study characteristics.

Total LSM \8.0 kPa LSM P8.0 kPa p value

N = 2,424 N = 2,299 N = 125
Demographics
Age (years) 66.5 ± 7.4 66.4 ± 7.3 68.7 ± 8.4 0.003
Female 56.6 57.7 35.2 \0.001
Caucasian 96.3 96.2 98.4 0.322
Low/intermediate/high education 45.0/29.6/25.4 45.2/29.4/25.4 42.3/34.1/23.6 0.531
Excessive alcohol use 15.9 15.9 16.0 1.000
Current/past/never smoker 11.3/52.9/35.7 11.0/52.3/36.6 16.8/63.9/19.3 \0.001
Physical activity* 46 (19–85) 46 (19–85) 42 (18–85) 0.748
Physical examination
BMI (kg/m2) 27.2 ± 4.0 27.1 ± 3.9 28.8 ± 5.3 0.001
Normal 29.7 30.1 22.8 0.003
Overweight 49.1 49.4 43.9
Obese 21.2 20.5 33.3
WC (cm)
Men 99.1 ± 10.2 98.9 ± 9.9 101.1 ± 13.1 0.152
Women 89.0 ± 11.5 88.7 ± 11.3 96.6 ± 15.7 0.002
Biochemistry¥
Aspartate transaminase (U/L) 24 (21–28) 24 (21–28) 27 (24–37) \0.001
ALT (U/L) 18 (14–24) 18 (14–23) 22 (16–33) \0.001
Bilirubin (mmol/L) 8 (6–11) 8 (6–11) 9 (6–12) 0.252
GGT (U/L) 24 (17–35) 23 (17–34) 38 (26–71) \0.001
Platelets (*109/L) 267.8 ± 63.5 268.7 ± 63.6 251.7 ± 59.7 0.004
HOMA-IR 2.5 (1.7–3.9) 2.5 (1.7–3.8) 3.7 (2.1–7.1) \0.001
Comorbidities
Metabolic syndrome 44.5 43.7 59.3 0.001
WC [88 cm ($) or [120 cm (#) 44.1 43.5 53.7 0.032
Triglycerides [150 mg/dl 38.2 38.0 41.6 0.450
HDL-C \40 mg/dl (#) or 50 mg/d ($) 33.6 33.3 40.0 0.144
Blood pressure P130/85 mmHg¥ 79.0 78.5 88.0 0.020
FPG [100 mg/dl 44.6 43.3 68.0 \0.001
Diabetes mellitus 10.5 9.2 33.6 \0.001
Hypertensiony 64.7 63.8 81.5 \0.001
Liver imaging
LSM (kPa) 4.7 (3.8–5.8) 4.6 (3.8–5.6) 9.1 (8.6–10.4) \0.001
Steatosis 34.6 33.2 60.0 \0.001
Dietary data
Coffee consumption (cups/day) 2.6 ± 1.7 2.6 ± 1.6 2.2 ± 1.7 0.019
None 6.8 6.7 10.4 0.006
Moderate ([0–3) 28.8 28.3 38.4
Frequent (P3) 64.4 65.1 51.2
Tea consumption (cups/day) 1.2 (0.4–2.7) 1.2 (0.4–2.7) 1.2 (0.2–2.3) 0.211
None 15.3 15.1 20.0 0.256
Moderate ([0–3) 56.4 56.4 56.0
Frequent (P3) 28.3 28.5 24.0
Tea subtypes (%)
Black 64.2 64.4 61.6 0.528
Green 26.1 26.3 21.6 0.242
Herbal 36.3 36.8 27.2 0.035
Other
Alcohol (units/day) 0.8 (0.1–1.8) 0.8 (0.1–1.8) 0.9 (0.1–2.1) 0.334
Soda (%) 47.7 47.5 51.2 0.462
Kilocalorie intake 2,189.7 ± 740.3 2,189.1 ± 735.5 2,199.7 ± 826.2 0.877
Cream use 46.0 45.8 50.4 0.357
Sugar use 25.5 24.9 37.6 0.002
Data is presented as mean (± SD), median (IQR) or percentage. p value is based on t test, Wilcoxon rank-sum test, Chi-square test or Fisher’s exact test.
*
Physical activity in metabolic equivalent task hours/week.
¥
Biochemistry data was available for 2,320.
y
BP data was available for 2,051 participants.

DHDI, physical activity, total tea consumption and cream and adjustment for education level, and type of probe did not affect
sugar use in coffee (ORmod 0.75, 95% CI 0.33–1.67; ORfreq 0.39, this association (data not shown). Coffee consumption was also
95% CI 0.18–0.86; p = 0.005) as shown in Table 2. Additional inversely related to log-transformed LSM as shown in a stepwise

4 Journal of Hepatology 2017 vol. xxx j xxx–xxx

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
 JOURNAL OF HEPATOLOGY
Table 2. Association between coffee, tea and LSM by 1) stepwise linear regression modelling using log-transformed LSM as dependent variable and 2) stepwise
logistic regression modelling using LSM P8 kPa as dependent variable. Non-consumers were used as reference.

log-transformed LSM LSM P8 kPa

Beta 95% CI p value OR 95% CI p value
Model 1*
Coffee
No 0 (ref) \0.001 1 (ref) 0.003
Moderate ([0–3) 0.003 0.058 ; 0.051 0.84 0.44 - 1.61
Frequent (P3) 0.055 0.106 ; 0.003 0.46 0.24 - 0.87
Herbal tea
No 0 (ref) \0.001 1 (ref) 0.024
Any 0.100 0.128 ; 0.071 0.61 0.40 - 0.94
Green tea
No 0 (ref) 0.076 1 (ref) 0.274
Any 0.028 0.059 ; 0.003 0.81 0.56 - 1.18
Black tea
No 0 (ref) 0.014 1 (ref) 0.476
Any 0.034 0.061 ; 0.007 0.85 0.54 - 1.34
Total tea
No 0 (ref) \0.001 1 (ref) 0.116
Moderate ([0–3) 0.073 0.110 ; 0.035 0.74 0.46 - 1.20
Frequent (P3) 0.106 0.147 ; 0.064 0.56 0.32 - 0.97
Model 2¥
Coffee
No 0 (ref) 0.001 1 (ref) 0.012
Moderate ([0–3) 0.016 0.069 ; 0.036 0.86 0.44 - 1.70
Frequent (P3) 0.060 0.110 ; 0.010 0.50 0.26 - 0.97
Herbal tea
No 0 (ref) \0.001 1 (ref) 0.446
Any 0.051 0.079 ; 0.023 0.84 0.54 - 1.31
Green tea
No 0 (ref) 0.304 1 (ref) 0.638
Any 0.016 0.014 ; 0.046 1.12 0.70 - 1.80
Black tea
No 0 (ref) 0.131 1 (ref) 0.652
Any 0.020 0.046 ; 0.006 0.92 0.62 - 1.35
Total tea
No 0 (ref) 0.205 1 (ref) 0.945
Moderate ([0–3) 0.035 0.071 ; 0.001 0.95 0.58 - 1.55
Frequent (P3) 0.032 0.073 ; 0.009 0.91 0.51 - 1.62
Model 3y
Coffee
No 0 (ref) 0.001 1 (ref) 0.005
Moderate ([0–3) 0.026 0.083 ; 0.032 0.75 0.33 - 1.67
Frequent (P3) 0.067 0.121 ; 0.013 0.39 0.18 - 0.86
Herbal tea
No 0 (ref) 0.001 1 (ref) 0.289
Any 0.051 0.081 ; 0.022 0.76 0.45, 1.27
Green tea
No 0 (ref) 0.146 1 (ref) 0.424
Any 0.023 0.008 ; 0.055 1.24 0.73 - 2.10
Black tea
No 0 (ref) 0.400 1 (ref) 0.615
Any 0.012 0.039 ; 0.016 0.88 0.52 - 1.47
Total tea
No 0 (ref) 0.514 1 (ref) 0.932
Moderate ([0–3) 0.023 0.061 ; 0.015 1.01 0.58 - 1.77
Frequent (P3) 0.063 0.061 ; 0.024 1.11 0.58 - 2.13
*
Model 1: adjusted for teaà or coffee and energy intake.
¥
Model 2: adjusted for teaà or coffee, energy intake, BMI, gender and age.
y
Model 3: adjusted for teaà or coffee, energy intake, BMI, gender, age, steatosis, ALT, excessive alcohol intake, current or former smoking and HOMA-IR, soda consumption,
cream and sugar use, DHDI and physical activity.
à
All regression models either contain total tea or tea subtypes as covariate.

Journal of Hepatology 2017 vol. xxx j xxx–xxx 5

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
Research Article
Coffee categories (cups/day): consumption (OR 0.61, 95% CI 0.40–0.94; p = 0.024). After further
20 No consumption (0)
adjustment for lifestyle and metabolic traits herbal tea was not
Moderate consumption (>0.3)
p <0.001
18
Frequent consumption (≥3)
independently associated with LSM P8.0 kPa, however, it
Predicted probability LSM ≥8 kPa (%)

ptrend = 0.017 ptrend= 0.034 remained inversely related to log-transformed LSM values in a
16
multivariable linear regression model as shown in Table 2
14 (p = 0.001). Interactions between covariables and herbal tea con-
sumption were tested and not significant.
12

10 Tea, coffee consumption and steatosis

8
Steatosis was found in 34.6% of the study population and was
6 higher amongst participants with LSM P8.0 kPa (60% vs. 33.2%;
4 p \0.001). Although the proportion of steatosis was similar
amongst coffee consumption groups (p = 0.656), proportion dif-
2
fered significantly in tea consumers (42.7%, 35.4% and 28.6% for
0 no, moderate and tea frequent consumers; p \0.001). Likewise,
n = 100 n = 388 n = 915 n = 43 n = 226 n = 462
tea consumption was associated with steatosis in logistic regres-
No steatosis Steatosis
n = 1403 n = 731 sion analysis, in contrast to consumption of coffee. However,
when correcting for potential confounders in a multivariable
Fig. 2. Predicted probabilities (%) of having LSM P8.0 kPa for the different logistic regression model, this association was no longer signifi-
coffee categories amongst participants with and without steatosis. Y-axis:
cant, as shown in Table 3.
predicted probabilities of LSM P8.0 kPa in percentages. Data are expressed as
mean (95% CI) prediction. X-axis: different coffee consumption categories for
participants with and without steatosis. Probabilities are adjusted for energy Sensitivity analyses
intake, age, gender, BMI, steatosis, HOMA-IR, ALT, excessive alcohol use, current
or former smoking, soda, DHDI, physical activity, total tea consumption, cream
In 1,360 participants (56.1%), FFQs were completed 5.8 years
and sugar use in coffee. The Ptrend is derived from the logistic regression analysis
stratified for steatosis with the following results: in participants without prior to TE examination. As can be seen from this sensitivity anal-
steatosis: ORmod 1.19, 95% CI 0.35–4.06; ORfreq 0.47, 95% CI 0.14–1.57; ysis presented in Table S2, results in both subgroups were consis-
p = 0.025; Ptrend 0.017; and in participants with steatosis: ORmod 0.45, 95% CI tent to that of the total study population, underlining the stable
0.15–1.40; ORfreq 0.31, 95% CI 0.11–0.92; p = 0.082; Ptrend 0.034. Interaction term
character of dietary data. Furthermore, 13 (0.5%) participants
was tested and found to be non-significant (Pinteraction = 0.527 for steatosis x
coffee). The comparison of predicted probability between the non-steatosis and
had LSM P13.0 kPa, suggestive of cirrhosis. As expected, the
steatosis group was based on the independent t test (p \0.001). association between coffee and LSM P13.0 kPa was also found
to be significant (ORmod 0.35, 95% CI 0.10–1.26; ORfreq 0.08, 95%
CI 0.02–0.35; p = 0.004). The inverse, univariate relationship
multivariable linear regression model in Table 2 (Ptrend = 0.001). between different cut-offs of LSM and coffee consumption is pre-
Interactions between covariables and coffee consumption were sented in Fig. 3. Moreover, when applying a more stringent cut-
tested in all regression models, but were found not to be signifi- off for significant fibrosis, i.e. LSM P6.2 kPa, in participants with
cant. Additionally, steatosis and log-transformed liver stiffness steatosis, similar results were found, albeit slightly attenuated
were significantly correlated (r = 0.152, p \0.001) but there was and statistically not significant (ORmod 0.62, 95% CI 0.28–1.38
no multicollinearity (VIF 1.33). Nevertheless, steatosis remains and ORfreq 0.47, 95% CI 0.22–1.02; p = 0.096). Lastly, exclusion
an important well-known risk factor for the development of of abnormal ALT levels from all analyses, to ensure results are
fibrosis. We therefore graphically depicted the adjusted predicted generalizable for the general population without known liver dis-
probabilities of having LSM P8.0 kPa for the different coffee cat- ease, did not alter the results (data not shown).
egories and presence of steatosis in Fig. 2. As expected, the prob-
ability of having LSM P8.0 kPa is higher for participants with
steatosis than for those without (p \0.001). Within both the no Discussion
steatosis and steatosis groups, the probability of LSM P8.0 kPa
for frequent coffee consumption is lower compared to no con- To the best of our knowledge, this is the first study to determine
sumption (within no steatosis: 2.5% in frequent vs. 4.4% in no cof- whether coffee and tea consumption are associated with subclin-
fee consumption; Pfreq vs.no = 0.218 and within steatosis: 6.9% in ical fibrosis and steatosis, using reliable imaging techniques, in
frequent vs. 13.1% in no coffee consumption; Pfreq vs.no = 0.036). the general population. This large population-based cohort study
shows an inverse and independent association between coffee
Tea consumption and liver stiffness and herbal tea consumption and log-transformed LSM. Frequent
coffee consumption was also independently associated with
Overall tea consumption was similar in participants with vs. lower odds and probability of significant liver fibrosis (defined
without LSM P8.0 kPa (Table 1). Likewise, tea was not associated as LSM P8.0 kPa) compared to non-consumers, independent of
with log-transformed LSM in linear regression or with LSM steatosis presence and a great number of metabolic and
P8.0 kPa in logistic regression when correcting for potential con- environmental traits.
founders (Table 2). Looking at subtypes of tea, only herbal tea was A recent meta-analysis showed that, amongst liver disease
less frequently consumed by participants with LSM P8.0 kPa patients, coffee consumers were less likely to develop liver
compared to participants with normal LSM (Table 1), this finding fibrosis and cirrhosis than no coffee consumers.36 Yet, all
remained when adjusted for energy intake and coffee studies included in this meta-analysis concerned specific,

6 Journal of Hepatology 2017 vol. xxx j xxx–xxx

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
 JOURNAL OF HEPATOLOGY
Table 3. Stepwise logistic regression models with steatosis as dependent variable using non-consumers as reference.

OR 95% CI p value
Model 1*
Coffee
No 1 (ref) 0.239
Moderate ([0–3) 1.32 0.92 - 1.91
Frequent (P3) 1.17 0.82 - 1.66
Herbal tea
No 1 (ref) 0.039
Any 0.82 0.68 - 0.99
Green tea
No 1 (ref) 0.018
Any 0.78 0.63 - 0.96
Black tea
No 1 (ref) 0.006
Any 0.78 0.66 - 0.93
Total tea
No 1 (ref) \0.001
Moderate ([0–3) 0.72 0.57 - 0.92
Frequent (P3) 0.53 0.41 - 0.69
Model 2¥
Coffee
No 1 (ref) 0.186
Moderate ([0–3) 1.31 0.87 - 1.98
Frequent (P3) 1.09 0.74 - 1.62
Herbal tea
No 1 (ref) 0.674
Any 0.96 0.77 - 1.18
Green tea
No 1 (ref) 0.297
Any 0.89 0.70 - 1.11
Black tea
No 1 (ref) 0.086
Any 0.84 0.69 - 1.03
Total tea
No 1 (ref) 0.064
Moderate ([0–3) 0.85 0.65 - 1.10
Frequent (P3) 0.70 0.52 - 0.95
Model 3y
Coffee
No 1 (ref) 0.192
Moderate ([0–3) 1.38 0.88 - 2.16
Frequent (P3) 1.15 0.75 - 1.77
Herbal tea
No 1 (ref) 0.779
Any 1.03 0.82 - 1.29
Green tea
No 1 (ref) 0.425
Any 0.91 0.71 - 1.15
Black tea
No 1 (ref) 0.169
Any 0.86 0.70 - 1.06
Total tea
No 1 (ref) 0.197
Moderate ([0–3) 0.86 0.65 - 1.14
Frequent (P3) 0.74 0.54 - 1.03
*
Model 1: adjusted for teaà or coffee and energy intake.
¥
Model 2: adjusted for teaà or coffee, energy intake, BMI, gender and age.
y
Model 3: adjusted for teaà or coffee, energy intake, BMI, gender, age, HOMA-IR, excessive alcohol intake, current or former smoking. Additional adjustment for LSM did not
affect this association.
à
All regression models either contain total tea or tea subtypes as covariate.

non-generalizable populations, i.e. patients with cirrhosis, other independently associated with lower odds (OR 0.64) of advanced
chronic liver disease or morbid obesity.14,15,37 In addition, a study histology-proven fibrosis in subjects with low insulin resis-
from a NASH consortium showed that coffee consumption was tance.16 Another study found that NASH patients with advanced

Journal of Hepatology 2017 vol. xxx j xxx–xxx 7

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
Research Article
considering the favorable lifestyle traits in frequent tea con-
sumers, a univariate association of (green) tea with liver stiffness
and steatosis could be confounded by a healthy lifestyle in
general.
1.0 Interestingly, we did find herbal tea to be independently asso-
Odds ratio for increased LSM (95% Cl)

ciated with lower liver stiffness. To the best of our knowledge, no

other study has addressed this association. Although herbal tea
encompasses a broad spectrum of different types of herbs, such
as Chamomile and Calicotome villosa, and methods of preparation,
a study of Carlsen et al.41 stated that, despite the great diversity,
‘herbal plants’ are amongst the most antioxidant-rich food items
0.1 and supplements worldwide and contribute the most to antioxi-
dant capacity as assessed by the ferric-reducing ability of plasma.
Coffee categories We therefore hypothesize that the antioxidant capacity of herbal
(cups/days):
tea exceeds that of other teas, displaying a more pronounced
Moderate consumption (>0-3)
effect on liver health. In addition, it has been previously demon-
Frequent consumption (≥3) strated that the ferric-reducing ability of plasma is reduced in
patients with steatosis and NASH.42 However, future studies are
0.01
needed to confirm this observation.
Mechanisms through which coffee and herbal tea promote
6 7 8 9 10 11 12 13 liver health are unclear. Also, it is not known if these two bever-
Cut-off increased LSM (kPa) age act through similar pathways. However, our data suggest that
both have inhibitory properties for fibrogenesis and not steatoge-
Fig. 3. Estimated odds ratio for moderate and frequent coffee consumers vs.
no coffee consumers for a grid of different cut-off values of LSM. Results of the nesis. Both coffee and tea consist of more than 100 individual
univariate logistic regression analyses of the effect of coffee consumption for the components, of which few, i.e. polyphenols and caffeine, are
endpoint of LSM greater than a specific cut-point (yes/no). Y-axis: univariate OR shared. Shim et al. showed that caffeine inhibits hepatic stellate
(95% CI) on log-scale. X-axis: different cut-off values for LSM.
cell proliferation in cirrhotic rat models and subsequently coun-
teracted fibrogenesis.43 Interestingly, caffeine in coffee had a
fibrosis, i.e. fibrosis score 2–4, consumed less coffee than patients greater effect on insulin sensitivity than caffeine in water alone.
with fibrosis score 0–1.38 But to date, a study by Zelber-Sagi et al. Moreover, another animal study found that conventional coffee,
is the only one addressing the association between coffee and decaffeinated coffee and caffeine alone all reduced serum levels
liver fibrosis in the general population, albeit using FibroTest of ALT and fibrosis scores. But, only rats fed with conventional
(i.e. a commercial biomarker test combining serum markers age coffee showed significantly less histologic injury from repeated
and gender) rather than biopsy or TE, to assess fibrosis.17 In line administration of toxins.44
with our data, an association between the prevalence of NAFLD Since caffeine alone does not seem to fully explain the bene-
and coffee consumption was not found. The authors found a ficial effect of coffee, other components are proposed to con-
lower proportion of significant fibrosis (defined as FibroTest score tribute. Both coffee and tea contain a substantial amount of
P2, borderline F1-F2 included) in participants consuming three polyphenols, which have been suggested to promote hepatopro-
or more coffee cups/day (OR 0.49), which is also comparable to tective effects independent of caffeine.45 Animal studies showed
our data. However, the authors did not include a clear reference that coffee-polyphenols, chlorogenic acid and tocopherols, atten-
group, i.e. non-consumers, in their analyses and the association uate obesity-related lipid accumulation in the liver46 and have
was only significant in univariate fashion. In contrast, we found antioxidant properties. Furthermore, experimental animal data
that this association in fact enhanced after adjustment for several suggest that tea-polyphenols attenuate fibrosis, oxidative stress
potential metabolic and environmental confounders. We hypoth- and inflammation through pivotal inflammatory transcription
esize that this can be explained by the relatively unhealthy life- factor47 and inhibition of hepatic stellate cell activation.48 In
style of coffee consumers, as they were more often heavier our population study, we could not explore the underlying mech-
drinkers, current or former smokers and obese. anistic pathway responsible for the observed effects.
Only few studies have assessed the effect of overall tea con- A major strength of this study is the inclusion of a large
sumption on liver disease and these show inconsistent number of subjects from an unselected population, correcting
results.18,37,39 Moreover, five large Asian studies specifically for a great number of environmental and metabolic traits and,
examined green tea consumption in relation to liver health,19,20 using well-validated diagnostic tools. Furthermore, we per-
of which three found a significant beneficial association between formed several sensitivity analyses, which should be interpreted
green tea and serum transaminases.21,22,40 However, this associ- in light of decreased power, showing consistency in magnitude
ation was only seen in individuals consuming more than 700 ml and direction of associations found in the overall analyses. How-
green tea per day. We could not detect a beneficial effect of green ever, some limitations need to be addressed. First, due to the
tea in our study. This may partly be explained by our predomi- cross-sectional design of this study, we are not able to draw
nantly elderly Caucasian population, in which drinking green conclusions regarding the cause-effect relationship of coffee,
tea is not customary, as reflected in the low consumption rate. herbal tea and liver stiffness. Second, the golden standard for
We speculate that the beneficial effects of tea could be dose- diagnosis of liver fibrosis remains a liver biopsy rather than
dependent and might only exert an independent effect on liver TE. However, performing an invasive liver biopsy in presumed
health when consumed above a certain threshold. Furthermore, healthy participants is unethical. TE strongly correlates with

8 Journal of Hepatology 2017 vol. xxx j xxx–xxx

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
 JOURNAL OF HEPATOLOGY
histologic stages of liver fibrosis49 and hence, LSM appears to be Authors’ contributions
a good surrogate marker for liver fibrosis in the general popula-
tion.3–5 Third, as with any questionnaire, data are subject to Louise J.M. Alferink: study concept and design, acquisition of
reporter- and recall-bias even though this FFQ has been exten- data, statistical analysis, analysis and interpretation of data,
sively validated in previous studies24,25 and unreliable FFQs drafting of the manuscript, and finalizing the article. Juliana Fitti-
were excluded. Also, none of the subjects were aware of the paldi: acquisition of data, analysis and interpretation of data, crit-
presence of fibrosis, since liver imaging had been conducted ical revision of the manuscript for important intellectual content,
after dietary assessment. Fourth, unfortunately FFQs do not pro- and approval of the final article. Jessica C. Kiefte-de Jong: analysis
vide information on the type of coffee or method of preparation. and interpretation of nutritional data, statistical analysis, and
Therefore we could not further specify type of coffee consump- approval of the final article. Pavel Taimr: technical support, and
tion. However, previous studies have been inconclusive about approval of final article. Bettina E. Hansen: analysis and interpre-
whether type of coffee is of any influence. Fifth, in light of a tation of data, statistical analysis, and approval of the final article.
healthy lifestyle, the association between herbal tea and liver Herold J. Metselaar: study concept and design, critical revision of
stiffness could be affected by residual confounding factors, even the manuscript for important intellectual content, and approval
though we adjusted for a great number of lifestyle factors. of the final article. Josje D. Schoufour: analysis and interpretation
Lastly, part of our study population completed the FFQs of nutritional data, and approval of the final article. M. Arfan
5.8 years prior to liver imaging. Since dietary data are known Ikram: study supervision, and approval of the final article. Harry
to be globally stable, we assumed that drinking habits for coffee L.A. Janssen: obtained funding, and approval of the final article.
and tea did not change significantly over time in this group Oscar H. Franco: study supervision, and approval of the final arti-
either.35 This assumption is supported by our sensitivity analy- cle. Sarwa Darwish Murad: study concept and design, principal
sis, showing that magnitude and direction of results in both investigator of the hepatology department within The Rotterdam
groups (i.e. RS I-5 and RS II-3 vs. RS III-2) was very similar to Study, analysis and interpretation of data, study supervision, crit-
the overall analyses (see Supplementary data). ical revision of the manuscript for important intellectual content,
In conclusion, we found a protective association between cof- and approval of the final article.
fee and liver stiffness that not only occurs in disease-specific set-
tings but appears also to be present in the general population Acknowledgements
with and without steatosis. Since coffee is an accessible and rel-
atively inexpensive beverage, it could be further implemented as The authors thank The Rotterdam Study participants and staff; in
a preventative strategy if future studies were to confirm our find- particular, the collaborating general practitioners and pharma-
ings. Though we did not find an association with steatosis per se, cists. The authors are also deeply grateful to Mrs. van Wijngaar-
coffee and tea consumption might still be useful to prevent pro- den (nurse ultrasonographist) for performing the abdominal
gression to more advanced stages, such as inflammation and in ultrasonography and transient elastography measurements. The
particular, fibrosis. Prospective studies are needed to establish a Rotterdam Study is supported by the Erasmus MC University
cause-effect relationship between coffee and liver fibrosis. Herbal Medical Center and Erasmus University Rotterdam, the Nether-
tea consumption was also independently related to lower liver lands Organization for Scientific Research (NWO), the Nether-
stiffness, even though consumed in small quantities. To date, very lands Organization for Health Research and Development
little is known about herbal tea and its effects on human health, (ZonMw), the Research Institute for Diseases in the Elderly
let alone, its mechanisms of action in liver disease. Future studies (RIDE), the Ministry of Education, Culture and Science, the Min-
are therefore needed to validate our data on the protective effect istry of Health, Welfare and Sports, the European Commission
of herbal tea on liver stiffness. (DG XII) and by the Municipality of Rotterdam.

Supplementary data
Financial support
Supplementary data associated with this article can be found, in
The Rotterdam Study is supported by the Erasmus MC University the online version, at http://dx.doi.org/10.1016/j.jhep.2017.03.
Medical Center and Erasmus University Rotterdam, the Nether- 013.
lands Organization for Scientific Research (NWO), the Nether-
lands Organization for Health Research and Development
(ZonMw), the Research Institute for Diseases in the Elderly References
(RIDE), the Ministry of Education, Culture and Science, the Min-
istry of Health, Welfare and Sports, the European Commission Author names in bold designate shared co-first authorship
(DG XII) and by the Municipality of Rotterdam.
[1] Lozano R, Naghavi M, Foreman K, Lim S, Shibuya K, Aboyans V, et al. Global
and regional mortality from 235 causes of death for 20 age groups in 1990
and 2010: a systematic analysis for the Global Burden of Disease Study 2010.
Lancet 2012;380:2095–2128.
Conflict of interest
[2] Murray CJ, Vos T, Lozano R, Naghavi M, Flaxman AD, Michaud C, et al.
Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21
The authors who have taken part in this study declared that they regions, 1990–2010: a systematic analysis for the Global Burden of Disease
do not have anything to disclose regarding funding or conflict of Study 2010. Lancet 2012;380:2197–2223.
interest with respect to this manuscript. [3] Koehler EM, Plompen EP, Schouten JN, Hansen BE, Darwish Murad S, Taimr
P, et al. Presence of diabetes mellitus and steatosis is associated with liver
Please refer to the accompanying ICMJE disclosure forms for
stiffness in a general population: The Rotterdam study. Hepatology
further details. 2016;63:138–147.

Journal of Hepatology 2017 vol. xxx j xxx–xxx 9

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013
Research Article
[4] Roulot D, Costes JL, Buyck JF, Warzocha U, Gambier N, Czernichow S, et al. [28] Wong VW, Vergniol J, Wong GL, Foucher J, Chan HL, Le Bail B, et al. Diagnosis
Transient elastography as a screening tool for liver fibrosis and cirrhosis in a of fibrosis and cirrhosis using liver stiffness measurement in nonalcoholic
community-based population aged over 45 years. Gut 2011;60:977–984. fatty liver disease. Hepatology 2010;51:454–462.
[5] Wong VW, Chu WC, Wong GL, Chan RS, Chim AM, Ong A, et al. Prevalence of [29] Castera L, Forns X, Alberti A. Non-invasive evaluation of liver fibrosis using
non-alcoholic fatty liver disease and advanced fibrosis in Hong Kong transient elastography. J Hepatol 2008;48:835–847.
Chinese: a population study using proton-magnetic resonance spectroscopy [30] Hamaguchi M, Kojima T, Itoh Y, Harano Y, Fujii K, Nakajima T, et al. The
and transient elastography. Gut 2012;61:409–415. severity of ultrasonographic findings in nonalcoholic fatty liver disease
[6] Serfaty L. Clinical implications of concomitant alcohol use, obesity, and viral reflects the metabolic syndrome and visceral fat accumulation. Am J
hepatitis. Gastroenterology 2016;150:1718–1722. Gastroenterol 2007;102:2708–2715.
[7] Marventano S, Salomone F, Godos J, Pluchinotta F, Del Rio D, Mistretta A, [31] Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, et al.
et al. Coffee and tea consumption in relation with non-alcoholic fatty liver Diagnosis and management of the metabolic syndrome: an American Heart
and metabolic syndrome: A systematic review and meta-analysis of Association/National Heart, Lung, and Blood Institute Scientific Statement.
observational studies. Clin Nutr 2016;35:1269–1281. Circulation 2005;112:2735–2752.
[8] Yesil A, Yilmaz Y. Review article: coffee consumption, the metabolic [32] Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC.
syndrome and non-alcoholic fatty liver disease. Aliment Pharmacol Ther Homeostasis model assessment: insulin resistance and beta-cell function
2013;38:1038–1044. from fasting plasma glucose and insulin concentrations in man. Diabetologia
[9] Eskelinen MH, Ngandu T, Tuomilehto J, Soininen H, Kivipelto M. Midlife 1985;28:412–419.
coffee and tea drinking and the risk of late-life dementia: a population-based [33] Willett WC, Reynolds RD, Cottrell-Hoehner S, Sampson L, Browne ML.
CAIDE study. J Alzheimers Dis 2009;16:85–91. Validation of a semi-quantitative food frequency questionnaire: comparison
[10] Khan N, Mukhtar H. Tea and Health: Studies in Humans. Curr Pharm Des with a 1-year diet record. J Am Diet Assoc 1987;87:43–47.
2013;19:6141–6147. [34] Cassinotto C, Boursier J, de Ledinghen V, Lebigot J, Lapuyade B, Cales P, et al.
[11] Arnesen E, Huseby N-E, Brenn T, Try K. The Tromsø Heart Study: Distribution Liver stiffness in nonalcoholic fatty liver disease: A comparison of supersonic
of, and determinants for, gamma-glutamyltransferase in a free-living shear imaging, FibroScan, and ARFI with liver biopsy. Hepatology
population. Scand J Clin Lab Invest 1986;46:63–70. 2016;63:1817–1827.
[12] Ruhl CE, Everhart JE. Coffee and caffeine consumption reduce the risk of [35] Goldbohm RA, van ’t Veer P, van den Brandt PA, van ’t Hof MA, Brants HA,
elevated serum alanine aminotransferase activity in the United States. Sturmans F, et al. Reproducibility of a food frequency questionnaire and
Gastroenterology 2005;128:24–32. stability of dietary habits determined from five annually repeated measure-
[13] Klatsky AL, Armstrong MA. Alcohol, smoking, coffee, and cirrhosis. Am J ments. Eur J Clin Nutr 1995;49:420–429.
Epidemiol 1992;136:1248–1257. [36] Liu F, Wang X, Wu G, Chen L, Hu P, Ren H, et al. Coffee consumption
[14] Modi AA, Feld JJ, Park Y, Kleiner DE, Everhart JE, Liang TJ, et al. Increased decreases risks for hepatic fibrosis and cirrhosis: a meta-analysis. PLoS One
caffeine consumption is associated with reduced hepatic fibrosis. Hepatol- 2015;10:e0142457.
ogy 2010;51:201–209. [37] Triantos C, Manolakopoulos S, Smirnidis A. Is caffeine responsible for the
[15] Anty R, Marjoux S, Iannelli A, Patouraux S, Schneck A-S, Bonnafous S, et al. hepatoprotective effect of coffee consumption in patients with chronic liver
Regular coffee but not espresso drinking is protective against fibrosis in a diseases? A multicentre study. AASLD abstract 2013;1.
cohort mainly composed of morbidly obese European women with NAFLD [38] Molloy JW, Calcagno CJ, Williams CD, Jones FJ, Torres DM, Harrison SA.
undergoing bariatric surgery. J Hepatol 2012;57:1090–1096. Association of coffee and caffeine consumption with fatty liver disease,
[16] Bambha K, Wilson LA, Unalp A, Loomba R, Neuschwander-Tetri BA, Brunt nonalcoholic steatohepatitis, and degree of hepatic fibrosis. Hepatology
EM, et al. Coffee consumption in NAFLD patients with lower insulin 2012;55:429–436.
resistance is associated with lower risk of severe fibrosis. Liver Int [39] Ruhl CE, Everhart JE. Coffee and tea consumption are associated with a lower
2014;34:1250–1258. incidence of chronic liver disease in the United States. Gastroenterology
[17] Zelber-Sagi S, Salomone F, Webb M, Lotan R, Yeshua H, Halpern Z, et al. 2005;129:1928–1936.
Coffee consumption and nonalcoholic fatty liver onset: a prospective study [40] Honjo S, Kono S, Coleman MP, Shinchi K, Sakurai Y, Todoroki I, et al. Coffee
in the general population. Transl Res 2015;165:428–436. consumption and serum aminotransferases in middle-aged Japanese men. J
[18] Klatsky AL, Morton C, Udaltsova N, Friedman GD. COffee, cirrhosis, and Clin Epidemiol 2001;54:823–829.
transaminase enzymes. Arch Intern Med 2006;166:1190–1195. [41] Carlsen MH, Halvorsen BL, Holte K, Bohn SK, Dragland S, Sampson L, et al.
[19] Tanaka K, Tokunaga S, Kono S, Tokudome S, Akamatsu T, Moriyama T, et al. The total antioxidant content of more than 3100 foods, beverages, spices,
Coffee consumption and decreased serum gamma-glutamyltransferase and herbs and supplements used worldwide. Nutr J 2010;9:3.
aminotransferase activities among male alcohol drinkers. Int J Epidemiol [42] Agrawal S, Duseja AK. Non-alcoholic fatty liver disease: east versus west. J
1998;27:438–443. Clin Exp Hepatol 2012;2:122–134.
[20] Kono S, Shinchi K, Imanishi K, Todoroki I, Hatsuse K. Coffee and serum [43] Shim SG, Jun DW, Kim EK, Saeed WK, Lee KN, Lee HL, et al. Caffeine
gamma-glutamyltransferase: a study of self-defense officials in Japan. Am J attenuates liver fibrosis via defective adhesion of hepatic stellate cells in
Epidemiol 1994;139:723–727. cirrhotic model. J Gastroenterol Hepatol 2013;28:1877–1884.
[21] Imai K, Nakachi K. Cross sectional study of effects of drinking green tea on [44] Furtado KS, Prado MG, Aguiar ESMA, Dias MC, Rivelli DP, Rodrigues MA, et al.
cardiovascular and liver diseases. BMJ 1995;310:693–696. Coffee and caffeine protect against liver injury induced by thioacetamide in
[22] Sakata R, Nakamura T, Torimura T, Ueno T, Sata M. Green tea with high- male Wistar rats. Basic Clin Pharmacol Toxicol 2012;111:339–347.
density catechins improves liver function and fat infiltration in non- [45] Vitaglione P, Morisco F, Mazzone G, Amoruso DC, Ribecco MT, Romano A,
alcoholic fatty liver disease (NAFLD) patients: a double-blind placebo- et al. Coffee reduces liver damage in a rat model of steatohepatitis: the
controlled study. Int J Mol Med 2013;32:989–994. underlying mechanisms and the role of polyphenols and melanoidins.
[23] Hofman A, Brusselle GG, Darwish Murad S, van Duijn CM, Franco OH, Hepatology 2010;52:1652–1661.
Goedegebure A, et al. The Rotterdam Study: 2016 objectives and design [46] Rodriguez de Sotillo DV, Hadley M, Sotillo JE. Insulin receptor exon 11+/ is
update. Eur J Epidemiol 2015;30:661–708. expressed in Zucker (fa/fa) rats, and chlorogenic acid modifies their plasma
[24] Goldbohm RA, van den Brandt PA, Brants HA, van’t Veer P, Al M, Sturmans F, insulin and liver protein and DNA. J Nutr Biochem 2006;17:63–71.
et al. Validation of a dietary questionnaire used in a large-scale prospective [47] Xiao J, Ho CT, Liong EC, Nanji AA, Leung TM, Lau TY, et al. Epigallocatechin
cohort study on diet and cancer. Eur J Clin Nutr 1994;48:253–265. gallate attenuates fibrosis, oxidative stress, and inflammation in non-
[25] Feunekes GI, Van Staveren WA, De Vries JH, Burema J, Hautvast JG. Relative alcoholic fatty liver disease rat model through TGF/SMAD, PI3 K/Akt/FoxO1,
and biomarker-based validity of a food-frequency questionnaire estimating and NF-kappa B pathways. Eur J Nutr 2014;53:187–199.
intake of fats and cholesterol. Am J Clin Nutr 1993;58:489–496. [48] Nakamuta M, Higashi N, Kohjima M, Fukushima M, Ohta S, Kotoh K, et al.
[26] van Lee L, Geelen A, van Huysduynen EJ, de Vries JH, van’t Veer P, Feskens EJ. Epigallocatechin-3-gallate, a polyphenol component of green tea, suppresses
The Dutch Healthy Diet index (DHD-index): an instrument to measure both collagen production and collagenase activity in hepatic stellate cells.
adherence to the Dutch Guidelines for a Healthy Diet. Nutr J 2012;11:49. Int J Mol Med 2005;16:677–681.
[27] Boursier J, Zarski JP, de Ledinghen V, Rousselet MC, Sturm N, Lebail B, et al. [49] Friedrich-Rust M, Ong MF, Martens S, Sarrazin C, Bojunga J, Zeuzem S, et al.
Determination of reliability criteria for liver stiffness evaluation by transient Performance of transient elastography for the staging of liver fibrosis: a
elastography. Hepatology 2013;57:1182–1191. meta-analysis. Gastroenterology 2008;134:960–974.

10 Journal of Hepatology 2017 vol. xxx j xxx–xxx

Please cite this article in press as: Alferink LJM et al. Coffee and herbal tea consumption is associated with lower liver stiffness in the general popula-
tion: The Rotterdam study. J Hepatol (2017), http://dx.doi.org/10.1016/j.jhep.2017.03.013