JOURNAL

Research Article OF HEPATOLOGY

Cancer

Statin use and the risk of hepatocellular carcinoma

in patients at high risk: A nationwide nested
case-control study

Graphical abstract Authors

Gyuri Kim, Suk-Yong Jang, Chung Mo Nam,

General population
Eun Seok Kang

LC DM
Correspondence
cmnam@yuhs.ac (C.M. Nam)
Statin
edgo@yuhs.ac (E.S. Kang)

Lay summary
HCC In this longitudinal nationwide population-
development based nested case-control study, the asso-
ciation between statin use and the risk of
HCC was investigated in Asian populations.
Herein, we noted a beneficial effect of
Highlights statin use on the development of HCC in

Statin use was significantly associated with a reduced risk of the general population and individuals at
HCC compared with nonusers. high risk of HCC (i.e. those with diabetes or
liver cirrhosis).

The beneficial effect of statins was shown in patients with
diabetes mellitus and relatively good glycemic control.

Statin significantly reduced the risk of HCC in subjects with

cirrhosis or diabetes mellitus, a high risk factor for HCC.

http://dx.doi.org/10.1016/j.jhep.2017.10.018
Ó 2017 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. J. Hepatol. 2018, 68, 476–484
Research Article JOURNAL
Cancer OF HEPATOLOGY

Statin use and the risk of hepatocellular carcinoma in patients at

high risk: A nationwide nested case-control study
Gyuri Kim1,2,y, Suk-Yong Jang3,y, Chung Mo Nam4,5,⇑, Eun Seok Kang1,6,⇑
1
Department of Internal Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea; 2Department of Medicine, Samsung Medical
Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea; 3Department of Preventive Medicine, Eulji University School of
Medicine, Daejeon, Republic of Korea; 4Department of Preventive Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea;
5
Institute of Health Services Research, Yonsei University College of Medicine, Seoul, Republic of Korea; 6Institute of Endocrine Research, Yonsei
University College of Medicine, Seoul, Republic of Korea

Background & Aims: Statins are widely used to treat hyperc-
holesterolemia. Statins may prevent hepatocellular carcinoma
(HCC), but have not yet been fully studied, particularly in
patients at high risk. Therefore, we investigated the risk of
HCC after statin use in the whole general population and evalu-
ated the effects of preexisting diabetes mellitus (DM) and liver
cirrhosis (LC) on that risk.
Methods: A nationwide, nested case-control study was con-
ducted with data from the National Health Insurance Service
Physical Health Examination Cohort 2002–2013 in the Republic
of Korea. Individuals diagnosed with HCC were matched to con-
trols based on the time of the follow-up, sex, and age at index
date. Odds ratios (ORs) and 95% confidence intervals (CIs) for
HCC associated with statin use were analyzed by multivariable
conditional logistic regression analyses. In total, 1,642 HCC
cases were matched to 8,210 control individuals from 514,866
participants.
Results: Statin use was associated with reduced risk of HCC
development (adjusted OR [AOR] 0.44; 95% CI 0.33–0.58) com-
pared with nonusers. The reduction in risk was significant in
the presence (AOR 0.28; 95% CI 0.17–0.46) and absence of DM
(AOR 0.53; 95% CI 0.39–0.73) and in the presence (AOR 0.39;
95% CI 0.26–0.60) and absence of LC (AOR 0.42; 95% CI 0.32–
0.57). Statin use also significantly reduced the risk of HCC
among patients with DM, without chronic complications (AOR
0.19; 95% CI 0.08–0.46) or with chronic complications (AOR
0.34; 95% CI 0.19–0.64), compared to nonusers.
Conclusions: Statin use may have a beneficial inhibitory effect
on HCC development, particularly in patients with DM or LC,
at high risk of HCC.
Lay summary: In this longitudinal nationwide population-
based nested case-control study, the association between statin
use and the risk of HCC was investigated in Asian populations.
Keywords: Statin; Hepatocellular carcinoma; Liver cirrhosis; Diabetes; Cohort study.
Received 19 March 2017; received in revised form 22 September 2017; accepted 16
October 2017; available online 26 October 2017
⇑ Corresponding authors. Address: Department of Preventive Medicine, Yonsei
University College of Medicine, 50-1, Yonsei-ro, Seodaemun-gu, Seoul 03722,
Republic of Korea. Tel.: +82 361 5387; fax: +82 2 392 8133 (C.M. Nam) or Division
of Endocrinology and Metabolism, Department of Internal Medicine, Yonsei Univer-
sity College of Medicine, 50-1, Yonsei-ro, Seodaemun-gu, Seoul 03722, Republic of
Korea. Tel.: +82 2 2228 1968; fax: +82 2 393 6884 (E.S. Kang).
E-mail addresses: cmnam@yuhs.ac (C.M. Nam), edgo@yuhs.ac (E.S. Kang).
y
These authors contributed equally to the study.
Herein, we noted a beneficial effect of statin use on the develop-
ment of HCC in the general population and individuals at high
risk of HCC (i.e. those with diabetes or liver cirrhosis).
Ó 2017 European Association for the Study of the Liver. Published by
Elsevier B.V. All rights reserved.
Introduction
Statins, 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA)
reductase inhibitors, are used to lower blood cholesterol levels
and to prevent cardiovascular disease and related mortality.1
Statin use has been suggested to have a beneficial effect on mul-
tiple cancer types, including colon cancer, breast cancer, and
prostate cancer.2,3 HMG-CoA reductase is involved in the rate
limiting step of cholesterol synthesis. Inhibition of this enzyme
by statins controls the production of mevalonate and down-
stream metabolites that were recently identified to be involved
in growth and apoptosis in many cancer types.4 Statins may
have a potential use for the prevention of hepatocellular carci-
noma (HCC), as the liver is its target organ.
HCC is a growing health problem worldwide, causing high
mortality, and having a very poor prognosis.5 Known risk factors
for HCC are chronic infections of hepatitis B virus (HBV) or C
virus (HCV), liver cirrhosis (LC), excessive alcohol consumption,
autoimmune hepatitis, and non-alcoholic fatty liver disease
(NAFLD).6 In East Asia and Africa, the prevalence of HCC is the
highest, due to chronic HBV infection.7 Recently, mainly attribu-
table to the increased number of patients with NAFLD or chronic
HCV infection, the incidence of HCC is expected to peak in 2020
in the United States and Europe.8,9 Although the safety of statin
use has been reported for patients with liver disease, due to the
prescribing bias, it is still unclear whether statin use is effective
for the prevention of HCC in patients with liver disease at high
risk.10
Diabetes mellitus (DM) was also associated with a twofold
increase in the risk of HCC.11 Regarding insulin resistance, DM
is strongly associated with NAFLD and subsequently with
HCC.12,13 Patients with DM commonly use statins because the
prevalence of dyslipidemia is reported to be higher than in
patients without DM.14,15 However, the relationship between
statins and HCC among patients with and without DM remains
unclear. Moreover, there is a lack of studies regarding the asso-
ciation between statin use and HCC among patients with DM in
the Asian population, the highest incident HCC population.7

Journal of Hepatology 2018 vol. 68 j 476–484


Statins may prevent HCC, but this association has not been
fully studied. More data are also needed on the effects of statins
on HCC among individuals with common risk factors for HCC.
Therefore, in the present study, we aimed to investigate the
association between statin use and the risk of HCC among the
whole general population and individuals in the presence or
absence of underlying DM or LC, using a nationwide sample
from longitudinal nested case-control data in an Asian
population.
Materials and methods
Design
We used a 2002–2013 data set from the National Health Insur-
ance Service Physical Health Examination Cohort (NHIS-PHEC)
in the Republic of Korea.16 The NHIS-PHEC was comprised of
514,866 examinees over the age of 40 years, who were ran-
domly sampled from the five million examinees who received
physical health examinations provided by the Korea National
Health Insurance Service in 2002 or 2003. The information in
the data set included all inpatient and outpatient medical
JOURNAL
OF HEPATOLOGY
claims data, including prescription drug use, diagnostic and
treatment codes, and primary and secondary diagnosis codes.
Measurement of body mass index (BMI) was obtained from
the data of physical health examinations. All the individuals
included in the NHIS-PHEC were followed until 2013 unless
there was a death or disqualification for National Health Insur-
ance, such as emigration. The protocol of this study was
approved by the Institutional Review Board ([IRB] No: 4–
2015–0636) at Severance Hospital, Seoul, Republic of Korea.
Written informed consent was not required by the IRB because
the researchers only accessed the database for analysis purposes
and personal information was not used.
Case and control selection criteria
We used the International Statistical Classification of Diseases
and Related Health Problems, 10th Revision (ICD-10) to identify
case patients. Among the NHIS-PHEC cases, eligibility criteria
for the HCC incidence patients were as follows: 1) first-time
diagnosis of HCC (ICD C22.0) with at least a three-year cancer
free period defined by the absence of any cancer diagnosis codes
(wash out period); 2) admission with a primary diagnosis code

Participants of the National Health Insurance Service Physical Health Examination Cohort
(N = 514,866)

Diagnosis code of HCC 2002–2013 No diagnosis code of HCC 2002–2013

(n = 4,282) (n = 510,584)

First diagnosis code of HCC during 2002-2004

(n = 1,209)

Diagnosis codes of any cancer prior to the date

of HCC diagnosis
(n = 736)
No supporting clinical procedure or death
codes for incidence of HCC
(n = 490)
Beneficiairies of the Medical Aid Program Beneficiaries of the Medical Aid Program
(n = 123) (n = 15,479)

Incidence density sampling up to 1:300

Matching on the year of birth, sex, and follow-up time

Candidated for cases Candidates for controls

(n = 1,724) (n = 513,936)

Diagnosis codes of any cancer prior to

the index date
(n = 9,520)

Prescription of statin during 2002-2003 Prescription of statin during 2002-2003

(n = 51) (n = 25,354)

Patients with missing data Subjects with missing data

(n = 31) (n = 6,085)

Patients within the stratum for

excluded case patients
(n = 20,141)

Simple random sampling up to 1:5

Final incident HCC case with patients
(n = 1,642) Final control subjects (n = 8,210)

Fig. 1. Flow of case and control selection.

Journal of Hepatology 2018 vol. 68 j 476–484 477

Research Article Cancer

Table 1. Baseline characteristics of hepatocellular carcinoma cases and matched controls.

Characteristics Cases Controls Crude
(n = 1,642) (n = 8,210)
n (%) n (%) OR (95% CI)
Age at index date*
40–49 175 (10.6) 875 (10.7) –
50–59 632 (38.5) 3,160 (38.5) –
60–69 507 (30.9) 2,535 (30.9) –
70–79 287 (17.5) 1,435 (17.5) –
≥80 41 (2.5) 205 (2.5) –
Mean (SD) 61.8 (9.2) 61.8 (9.2)
Sex*
Male 1,372 (83.6) 6,860 (83.6) –
Female 270 (16.4) 1,350 (16.4) –
Elevated liver enzymes 1,032 (62.9) 1,373 (16.7) 8.6 (7.6–9.7)
Chronic viral hepatitis 755 (46.0) 232 (2.8) 31.0 (25.4–37.9)
Liver cirrhosis 563 (34.2) 94 (1.1) 48.3 (36.7–63.6)
Diabetes mellitus
None 1,324 (80.6) 7,291 (88.8) 1.00
Uncomplicated 156 (9.5) 494 (6.0) 1.74 (1.44–2.11)
Complicated 162 (9.9) 425 (5.2) 2.13 (1.76–2.59)
Fasting blood glucose
<100 mg/dl 963 (58.7) 5,424 (66.1) 1.00
100–125 mg/dl 426 (25.9) 2,089 (25.4) 1.15 (1.02–1.30)
≥126 mg/dl 253 (15.4) 697 (8.5) 2.06 (1.76–2.43)
Mean (SD), mg/dl 107.9 (50.3) 99.1 (33.1) 1.01 (1.00–1.01)
CCI
0 877 (53.4) 4,230 (51.5) 1.00
1 520 (31.7) 2,592 (31.6) 0.96 (0.85–1.08)
2 168 (10.2) 957 (11.7) 0.83 (0.69–1.00)
≥3 77 (4.7) 431 (5.3) 0.83 (0.89–1.00)
BMI
<23 kg/m2 620 (37.8) 3,078 (37.5) 1.00
23–25 kg/m2 448 (27.3) 2,329 (28.4) 0.96 (0.84–1.09)
≥25 kg/m2 574 (35.0) 2,803 (34.1) 1.02 (0.90–1.15)
Mean (SD) 24.0 (3.1) 23.9 (2.9) 1.01 (1.00–1.03)
Alcohol-related disorders 277 (16.9) 418 (5.1) 3.90 (3.30–4.61)
Smoking
Non-smoker 808 (49.2) 4,371 (53.2) 1.00
Past smoker 205 (12.5) 1,092 (13.3) 1.05 (0.88–1.25)
Current smoker 629 (38.3) 2,747 (33.5) 1.28 (1.13–1.45)
Aspirin 290 (17.7) 1,676 (20.4) 0.82 (0.71–0.95)
Household income
Low 227 (13.8) 1,067 (13.0) 1.00
Mid-low 230 (14.0) 1,088 (13.3) 0.99 (0.83–1.22)
Middle 272 (16.6) 1,316 (16.0) 0.97 (0.80–1.18)
Mid-high 363 (22.1) 1,765 (21.5) 0.97 (0.80–1.16)
High 550 (33.5) 2,974 (36.2) 0.87 (0.73–1.03)
Residential area
Metropolitan 703 (42.8) 3,674 (44.8) 1.00
Non–metropolitan 939 (57.2) 4,536 (55.3) 1.08 (0.97–1.20)
Type of statin
Simvastatin 57 (3.5) 617 (7.5) 0.44 (0.33–0.58)
Atorvastatin 56 (3.4) 536 (6.5) 0.49 (0.37–0.65)
Lovastatin 7 (0.4) 102 (1.2) 0.34 (0.16–0.73)
Fluvastatin 12 (0.7) 56 (0.7) 1.07 (0.58–2.01)
Pitavastatin 5 (0.3) 80 (1.0) 0.31 (0.13–0.77)
Pravastatin 18 (1.1) 91 (1.1) 0.99 (0.59–1.65)
Rosuvastatin 13 (0.8) 117 (1.4) 0.56 (0.31–0.98)
The index date was defined as the date one year before the date of hepatocellular carcinoma diagnosis.
BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; OR, odds ratio; SD, standard deviation.
*
Matched on year of birth, sex, follow-up time.

of HCC; 3) no prior diagnosis of any cancer in order to ensure or
exclude recurrent and metastatic liver cancer;17 and 4) exclu-
sion of individuals without supporting clinical codes indicating
the presence of HCC including any liver diagnostic tests (biopsy
or arteriography of hepatic artery), any treatments of the liver
(hepatectomy, liver transplantation, radiofrequency ablation,
arterial embolization, radiotherapy, or chemotherapy), or death
due to HCC (death code for liver cancer) (Fig. 1). The date of a
cancer diagnosis was the first date of outpatient or inpatient
records with a primary diagnosis of cancer. The index date

478 Journal of Hepatology 2018 vol. 68 j 476–484

 JOURNAL
OF HEPATOLOGY
Table 2. Relationship between statin use and hepatocellular carcinoma.
Characteristics Cases Controls Crude Adjusted*
(n = 1,642) (n = 8,210)
n (%) n (%) OR (95% CI) OR (95% CI)
Statin use
Never use 1,531 (93.2) 7,163 (87.3) 1.00 1.00
Ever use 111 (6.8) 1,047 (12.8) 0.47 (0.38–0.58) 0.44 (0.33–0.58)
Cumulative dose of use
Never use 1,531 (93.2) 7,163 (87.3) 1.00 1.00
Ever use
<180 cDDDs 61 (3.7) 520 (6.3) 0.53 (0.40–0.70) 0.45 (0.32–0.64)
180–365 cDDDs 19 (1.2) 180 (2.2) 0.46 (0.29–0.75) 0.56 (0.31–1.00)
365–720 cDDDs 19 (1.2) 187 (2.3) 0.45 (0.28–0.72) 0.41 (0.22–0.76)
≥720 cDDDs 12 (0.7) 160 (2.0) 0.32 (0.18–0.59) 0.30 (0.14–0.63)
Cumulative dose of use
Never use 1,531 (93.2) 7,163 (87.3) 1.00 1.00
Ever use
Q1 (<60 cDDDs) 27 (1.6) 247 (3.0) 0.50 (0.33–0.74) 0.42 (0.25–0.69)
Q2 (60–179 cDDDs) 34 (2.1) 271 (3.3) 0.56 (0.39–0.81) 0.49 (0.31–0.79)
Q3 (180–528 cDDDs) 27 (1.6) 263 (3.2) 0.45 (0.30–0.68) 0.51 (0.30–0.85)
Q4 (≥528 cDDDs) 23 (1.4) 266 (3.2) 0.38 (0.24–0.58) 0.35 (0.20–0.61)
cDDDs, cumulative defined daily doses; CI, confidence interval; OR, odds ratio, Q, quartile.
*
Adjusted for diabetes mellitus, liver cirrhosis, Charlson comorbidity index, body mass index, alcohol related disorders, smoking, aspirin use, use of antidiabetic
medications, household income level, and residential area.

was defined as the date one year prior the date of HCC diagnosis.
Controls were randomly selected at a 1:5 ratio from individuals
who were at risk of becoming case patients at the time when
case patients were selected. Individuals excluded during case
patient selection were also excluded from the risk set. Case
patients and control patients were matched based on the time
of follow-up, sex, and age at index date. As case patients, control
patients also had to have a three-year cancer free period. In
addition to the main analyses, subgroup analyses were con-
ducted within additionally matched subgroups based on the
presence or absence of DM, with or without chronic complica-
tions, according to diagnosis code, liver enzyme abnormalities,
or fasting blood glucose (≥154 mg/dl or <154 mg/dl).18 Liver
enzyme abnormalities were defined as any above-normal value
of aspartate transaminase (AST) ≥51 IU/L, alanine transaminase
(ALT) ≥46 IU/L, or ≥78 IU/L in males or ≥46 IU/L in females for
gamma-glutamyltransferase (c-GTP) which could reflect
NAFLD, chronic hepatitis, or another etiology of chronic liver
disease.19–22 Fasting blood glucose (≥154 mg/dl or <154 mg/dl)
in patients with DM indicated glycemic control status of gly-
cated hemoglobin ≥7% or <7%, respectively.23 For the main anal-
ysis, 8,210 control subjects were matched to 1,642 case patients
at a 1:5 ratio. For the DM-matched analysis, 8,180 control sub-
jects were matched to 1,641 case patients with DM. No control
subject was matched to one case patients with DM and 12 case
patients with DM did not match to control subjects at a 1:5 ratio
(five case patients with a 1:4, one case patient with a 1:3, and
six case patients with a 1:2 ratio). For the LC-matched analysis,
1,574 control subjects were matched to 513 case patients with
LC. No control subject was matched to 50 case patients with LC
and 380 case patients with LC did not match to control subjects
with a 1:5 ratio (70 case patients with a 1:4 ratio, 109 case
patients with a 1:3 ratio, 101 case patients with a 1:2 ratio,
and 100 case patients with a 1:1 ratio). For the liver enzyme
abnormalities-matched analysis, 8,210 control subjects were
matched to 1,642 case patients with 1:5 ratio. Table S1 shows
the characteristics of case patients with LC who did not match
to control subjects.
Use of statins
We defined the index date as one year before the date of HCC
diagnosis. Statin use was defined as prescription of more than
30 days during admissions and outpatient visits from 2002 to
the index date. This definition was applied for statins and other
drugs including antidiabetic agents such as insulin, sulfony-
lurea, metformin, and thiazolidinedione as they may confound
the data. To ensure that all study subjects had at least two
statin-free years, subjects who were prescribed statins between
2002–2003 were excluded. We also calculated cumulative doses
of exposure for statins with the sum of the doses for all the pre-
scribed days, which was expressed as the cumulative defined
daily dose (cDDD) according to the World Health Organization
definition.
Measurements
From the health examinations, BMI was calculated as body
weight divided by the square of height in meters. Obesity was
defined as ≥25 kg/m2 of BMI for the Asian population enrolled
in the present study. Smoking status was categorized as non-
smoker, past smoker, or current smoker. The presence of comor-
bidities was defined by at least two outpatient visits or one
admission upon the primary or the first secondary diagnosis.
After overnight fasting, blood samples were obtained, and
serum glucose and liver enzymes were measured. Low density
lipoprotein (LDL) cholesterol results could be obtained after
2008 as part of the health examinations.
Statistical analyses
Conditional logistic regression analysis was performed to eval-
uate the association between statin use and the risk of HCC.
Adjusted odds ratios (AORs) and 95% confidence intervals (CIs)
were determined. Other matching variables for the nested
case-control study design, confounders that we adjusted for
were DM, alcohol-related disorders (including alcoholic liver
diseases and alcohol use disorders), LC, liver enzyme abnormal-
ities, alcoholic liver disease, smoking status, obesity, Charlson
Comorbidity Index (which was scored excluding the comorbid

Journal of Hepatology 2018 vol. 68 j 476–484 479

Research Article Cancer

diseases listed above), aspirin use, duration of DM, use of antidi-
abetic agents including insulin, sulfonylurea, metformin, thiazo-
lidinedione, or dipeptidyl peptidase-4 inhibitors, household
income, and residential area, as appropriate. Because LDL
cholesterol levels were available after 2008 from NHIS-PHEC,
an additional study dataset after 2008 was selected and ana-
lyzed with the adjustment for LDL cholesterol levels. All of the
confounders listed above were identified on the index date.
The p value <0.05 was considered significant. All statistical anal-
yses were performed using SAS software, version 9.4 (Cary, NC,
USA).
For further details regarding the materials used, please refer
to the Supplementary material and the CTAT table.
Results
Among 514,866 participants in the NHIS-PHEC, 4,282 were
diagnosed with HCC and 2,558 were excluded because they
did not meet the inclusion criteria. Additionally, 82 patients
were excluded due to the prescription of statin during 2002–
2003 or missing data. The final number of cases was 1,642
and the mean follow-up time was 7.5 years (Fig. 1). Table 1
shows the baseline characteristics of the 1,642 HCC case
patients and 8,210 control cases. The two groups had even dis-
tributions in the matching variables, including the time of
follow-up, sex, and age at index date. Subjects were predomi-
nantly male (83.6%) and most subjects (69.4%) were aged 50–
69 years. Compared with the control group, HCC case patients
had more elevated liver enzymes, LC, chronic viral hepatitis,
and alcohol-related disorders. Compared to control subjects,
HCC case patients had less frequent aspirin use (OR 0.82; 95%
CI 0.71–0.95) and statin use, including simvastatin (OR 0.44;
95% CI 0.33–0.58), atorvastatin (OR 0.49; 95% CI 0.37–0.65),
lovastatin (OR 0.34; 95% CI 0.16–0.73), rosuvastatin (OR 0.56;
95% CI 0.31–0.98), and pitavastatin (OR 0.31; 95% CI 0.13–0.77).
Statin use was significantly associated with a reduced risk of
HCC incidence compared to that of the nonusers (AOR 0.44;
95% CI 0.33–0.58; Table 2). A trend toward risk reduction with
cumulative doses of statin use was shown as AORs of 0.45
(95% CI 0.32–0.64), 0.56 (95% CI 0.31–1.00), 0.41 (95% CI 0.22–
0.76), and 0.30 (95% CI 0.14–0.63) for patients with statin use
of <180, 180–365, 365–720, and ≥720 cDDDs, respectively.
The results of study subjects who had LDL cholesterol results
after year 2008 from NHIS-PHEC data are provided (Table S2).
Among 469 cases and 2,344 control subjects, statin use was sig-
nificantly associated with AOR of 0.43 (95% CI 0.29–0.65) even
after additionally adjusting for LDL cholesterol levels. The great-
est risk reduction was shown in statin users with doses of ≥720
cDDDs with an AOR of 0.25 (95% CI 0.09–0.65), compared with
that of nonusers.

Table 3. Relationship between statin use and hepatocellular carcinoma according to diabetes mellitus.
With diabetes mellitus Without diabetes mellitus
Characteristics Cases Controls Crude Adjusted Cases Controls Crude Adjusted
(n = 317) (n = 1,560) (n = 1,324) (n = 6,620)
n (%) n (%) OR (95% CI) OR (95% CI)* n (%) n (%) OR (95% CI) OR (95% CI)**
Statin use
Never use 278 1,054 1.00 1.00 1,252 5,937 1.00 1.00
(87.7) (67.6) (94.6) (89.7)
Ever use 39 506 0.27 0.28 72 683 0.48 0.53
(12.3) (32.4) (0.19–0.39) (0.17–0.46) (5.4) (10.3) (0.38–0.63) (0.39–0.73)
Cumulative dose of use
Never use 278 1,054 1.00 1.00 1,252 5,937 1.00 1.00
(87.7) (67.6) (94.6) (89.7)
Ever use
<180 cDDDs 17 198 0.32 0.23 44 372 0.55 0.59
(5.4) (12.7) (0.19–0.54) (0.11–0.46) (3.3) (5.6) (0.40–0.76) (0.40–0.88)
180–365 cDDDs 8 93 0.31 0.43 11 124 0.41 0.50
(2.5) (6.0) (0.15–0.66) (0.18–1.05) (0.8) (1.9) (0.22–0.76) (0.23–1.08)
365–720 cDDDs 9 107 0.28 0.51 10 89 0.50 0.55
(2.8) (6.9) (0.14–0.57) (0.20–1.30) (0.8) (1.3) (0.26–0.97) (0.24–1.27)
≥720 cDDDs 5 108 0.15 0.12 7 98 0.32 0.33
(1.6) (6.9) (0.06–0.37) (0.03–0.44) (0.5) (1.5) (0.15–0.69) (0.14–0.80)
Cumulative dose of use
Never use 278 1,054 1.00 1.00 1,252 5,937 1.00 1.00
(87.7) (67.6) (94.6) (89.7)
Ever use
Q1(<60 cDDDs) 8 81 0.38 0.18 19 181 0.49 0.60
(2.5) (5.2) (0.18–0.80) (0.06–0.50) (1.4) (2.7) (0.30–0.79) (0.34–1.03)
Q2 (60–179 cDDDs) 9 117 0.29 0.28 25 190 0.60 0.59
(2.8) (7.5) (0.14–0.57 (0.11–0.71) (1.9) (2.9) (0.40–0.92 (0.34–1.02)
Q3 (180–527 cDDDs) 9 132 0.24 0.40 18 170 0.48 0.53
(2.8) (8.5) (0.12–0.49) (0.18–0.90) (1.4) (2.6) (0.30–0.79) (0.28–1.00)
Q4 (≥528 cDDDs) 13 176 0.24 0.27 10 142 0.31 0.37
(4.1) (11.3) (0.13–0.44) (0.12–0.65) (0.8) (2.2) (0.16–0.60) (0.18–0.77)
cDDDs, cumulative defined daily doses; CI, confidence interval; OR, odds ratio; Q, quartile.
*
Adjusted for liver cirrhosis, liver enzyme abnormality, Charlson comorbidity score, obesity, alcoholic liver disease, smoking, aspirin use, household income level, residential
area, duration of diabetes mellitus, and use of antidiabetic agents including insulin, sulfonylurea, metformin, thiazolidinedione, and dipeptidyl peptidase-4 inhibitors.
**
Adjusted for liver cirrhosis, liver enzyme abnormality, Charlson comorbidity score, obesity, alcoholic liver disease, smoking, aspirin use, household income level, and
residential area.

480 Journal of Hepatology 2018 vol. 68 j 476–484

 JOURNAL
OF HEPATOLOGY
Table 4. Relationship between statin use and hepatocellular carcinoma according to the chronic complications of diabetes mellitus.
Characteristics Cases Controls Crude Adjusted
(n = 317) (n = 1,560)
n (%) n (%) OR (95% CI) OR (95% CI)*
Without complication
Never use 143 (91.7) 558 (72.7) 1.00 1.00
Ever use 13 (8.3) 210 (27.3) 0.22 (0.12–0.40) 0.19 (0.08–0.46)
With complication
Never use 135 (83.9) 484 (61.1) 1.00 1.00
Ever use 26 (16.2) 308 (38.9) 0.29 (0.18–0.45) 0.34 (0.19–0.64)
CI, confidence interval; OR, odds ratio.
*
Adjusted for liver cirrhosis, liver enzyme abnormality, Charlson comorbidity score, obesity, alcoholic liver disease, smoking, aspirin use, household income level,
residential area, duration of diabetes mellitus, and use of antidiabetic agents including insulin, sulfonylurea, metformin, thiazolidinedione, and dipeptidyl peptidase-4
inhibitors.

The relationship between statin use and HCC in patients with
and without DM is shown (Table 3). In patients with DM, statin
use was significantly associated with reduced risk (AOR 0.28;
95% CI 0.17–0.46), compared to that of nonusers. For statin
users, patients without DM also showed a significant risk reduc-
tion of HCC incidence (AOR 0.53, 95% CI 0.39–0.73), compared to
the nonusers, and this effect was dose-dependent (AOR 0.59
[95% CI 0.40–0.88], 0.50 [95% CI 0.23–1.08], 0.55 [95% CI 0.24–
1.27], and 0.33 [95% CI 0.14–0.80]) for subjects who had a cDDD
of statin of <180, 180–365, 365–720, and ≥720, respectively. The
risk reduction for HCC in patients using statins was less in
patients without DM, relative to patients with DM. Furthermore,
we investigated the association between statin use and incident
HCC among patients with DM according to the presence or
absence of chronic DM complications (Table 4). Patients having
DM without chronic complications (AOR 0.19; 95% CI 0.08–
0.46), or patients with chronic complications (AOR 0.34; 95%
CI 0.19–0.64) had a lower risk of incident HCC, when compared
to that of statin nonusers. Furthermore, among patients with
DM, patients with fasting blood glucose <154 mg/dl, who
demonstrated relatively good glycemic control, showed a signif-
icantly reduced AOR of 0.43 (95% CI 0.21–0.89) for incident HCC
among statin users, compared to statin nonusers, while those
with fasting blood glucose ≥154 mg/dl showed an AOR of 0.56
(95% CI 0.23–1.33), which was not significant (Table S3).
The relationship between statins and HCC among individuals
with and without LC is shown (Table 5). Among patients with
LC, statin use was significantly associated with a reduction of

Table 5. Relationship of statin use to hepatocellular carcinoma according to liver cirrhosis.

With liver cirrhosis Without liver cirrhosis
Characteristics Cases Controls Crude Adjusted Cases Controls Crude Adjusted
(n = 513) (n = 1,574) (n = 513) (n = 1,574)
n (%) n (%) OR (95% CI) OR (95% CI)* n (%) n (%) OR (95% CI) OR (95% CI)*
Statin use
Never use 480 1,318 1.00 1.00 1,004 4,704 1.00 1.00
(93.6) (83.7) (93.1) (87.2)
Ever use 33 256 0.34 0.39 75 691 0.49 0.42
(6.4) (16.3) (0.22–0.50) (0.26–0.60) (7.0) (12.8) (0.38–0.63) (0.32–0.57)
Cumulative dose of use
Never use 480 1,318 1.00 1.00 1,004 4,704 1.00 1.00
(93.6) (83.7) (93.1) (87.2)
Ever use
<180 cDDDs 20 154 0.37 0.42 39 348 0.51 0.50
(3.9) (9.8) (0.23–0.61) (0.25–0.71) (3.6) (6.5) (0.37–0.72) (0.35–0.73)
180–365 cDDDs 3 44 0.12 0.14 16 123 0.58 0.43
(0.6) (2.8) (0.03–0.49) (0.03–0.59) (1.5) (2.3) (0.34–0.99) (0.24–0.79)
365–720 cDDDs 7 30 0.48 0.68 12 124 0.42 0.32
(1.4) (1.9) (0.20–1.15) (0.27–1.76) (1.1) (2.3) (0.23–0.77) (0.16–0.62)
≥720 cDDDs 3 28 0.33 0.37 8 96 0.36 0.29
(0.6) (1.8) (0.10–1.12) (0.11–1.28) (0.7) (1.8) (0.18–0.75) (0.13–0.63)
Cumulative dose of use
Never use 480 1,318 1.00 1.00 1,004 4,704 1.00 1.00
(93.6) (83.7) (93.1) (87.2)
Ever use
Q1 (<60 cDDDs) 8 68 0.31 0.33 18 173 0.47 0.46
(1.6) (4.3) (0.14–0.68) (0.15–0.76) (1.7) (3.2) (0.29–0.77) (0.27–0.78)
Q2 (60–179 cDDDs) 12 86 0.42 0.49 21 174 0.56 0.55
(2.3) (5.5) (0.23–0.78) (0.26–0.94) (2.0) (3.2) (0.35–0.88) (0.33–0.92)
Q3 (180–527cDDDs) 6 53 0.25 0.29 21 179 0.52 0.37
(1.2) (3.4) (0.10–0.63) (0.11–0.77) (2.0) (3.3) (0.33–0.83) (0.22–0.61)
Q4 (≥528 cDDDs) 7 49 0.33 0.41 15 165 0.40 0.33
(1.4) (3.1) (0.14–0.77) (0.17–1.01) (1.4) (3.1) (0.23–0.68) (0.18–0.60)
cDDDs, cumulative defined daily doses; CI, confidence interval; OR, odds ratio; Q, quartile.
*
Adjusted for liver enzyme abnormality, diabetes, Charlson comorbidity score, obesity, alcoholic liver disease, smoking, aspirin use, household income level, and residential
area.

Journal of Hepatology 2018 vol. 68 j 476–484 481

Research Article Cancer

HCC incidence (AOR 0.39; 95% CI 0.26–0.70) and the association and pitavastatin. The hydrophilic statins are known to be more
was also observed in patients without LC (AOR 0.42; 95% liver-specific,29 but significantly, both types of statins showed
CI 0.32–0.57), compared to nonusers. In addition, we evaluated beneficial effects.28
the relationship between statin use and HCC development Statins are approved as cholesterol-lowering agents and have
according to the presence or absence of elevated liver enzymes been safely prescribed worldwide for decades for patients with
(Table S4). Statin use was significantly associated with reduced hypercholesterolemia. Recently, the mevalonate pathway inhib-
AORs of 0.34 (95% CI 0.25–0.44) and 0.51 (95% CI 0.36–0.72) for ited by statins in the liver was suggested to be an essential
incident HCC in subjects with and without liver enzyme abnor- metabolic pathway involving metabolites that are integral to
malities, compared to statin nonusers, respectively. tumor growth and progression.4 Farnesyl diphosphate, geranyl-
geranyl diphosphate isoprenylate, and small GTPases such as
Ras and Rho are involved in tumorigenesis and cancer cell sur-
Discussion
vival.30,31 In addition, statins also ameliorated hepatic steatosis
In a nationwide longitudinal nested case-control study, we eval-
and inflammation, which were strongly associated with DM and
uated 1,642 HCC cases and 8,210 control cases with matching
liver disease, related to the development of HCC.32–34 However,
variables including age, sex, and the time of the follow-up at a
how statins have protective effects in patients with DM or liver
1:5 ratio, from the NHIS-PHEC, including the follow-up data
disease is poorly understood. Further investigations regarding
from 2002–2013 of 514,866 participants over the age of 40
the specific mechanisms of statins’ anticancer properties, cou-
years. We found that statin use was significantly associated
pled with DM and liver disease relationships should be
with a reduced risk of HCC. The beneficial effect of statins on
elucidated.
incident HCC was also exhibited in individuals with risk factors,
The present study has several strengths. First, we used well
including DM, LC, or elevated liver enzymes. Although patients
established and validated national longitudinal data, sourced
with DM and chronic complications were associated with a high
from the NHIS-PHEC, and included follow-up data from 2002–
risk of HCC, statin use also significantly reduced risk of HCC
2013 on 514,866 participants. Using a nested case-control study
among those patients. Furthermore, the beneficial effect of sta-
design, control patients were matched to case patients at a ratio
tin use on HCC was greater in patients with DM or LC than in
of 1:5 based on sex, age at index date, and the time of the
patients without DM or LC.
follow-up with strict inclusion criteria for the study population.
Previously, using a data source of 229 HCC case patients and
We firstly used matched laboratory data in the analyses to
1,145 control cases, from patients with type 2 DM, statin initia-
ascertain the effects of statin on incident HCC in relation to liver
tion was reported to reduce the risk of incident HCC with an
enzymes, fasting blood glucose, and LDL cholesterol. In addition,
AOR of 0.36 (95% CI 0.22–0.60).24 In the present study, we
adjustments for various potential confounding parameters
included the general population and examined the effect of sta-
including laboratory data were adopted in the analyses to pro-
tin use for HCC from the whole general population. In the DM-
vide accurate results. There has been a lack of research into
matched analysis, consistent with the previous results, the ben-
the relationship between statin use and incident HCC in the
eficial effect of statin use on HCC was greater in patients with
Asian population, in which the prevalence of HBV infection is
DM, compared to subjects without DM, although DM alone
high and the number of people with obesity or NAFLD is grow-
was an increasing risk factor of HCC. DM was reported to be
ing due to changes in lifestyle and diet. The results from a large
related to a twofold increased risk of HCC, independent of
Asian population study could have the potential to provide
underlying liver disease,11 and the risk was correlated with a
strong evidence to link these factors. Moreover, we investigated
long duration of DM.25 In the current study, we showed that
the association between statin use and incident HCC in sub-
HCC case patients had more DM without chronic complications
groups including DM, LC, or liver enzyme abnormalities as
and DM with chronic complications, compared to individuals
high-risk groups of HCC, as well as all the general population.
without DM. However, statin use significantly reduced the risk
We also investigated the association between statin use and
of incident HCC among patients with DM, with or without
the risk of HCC in patients having DM, according to the presence
chronic complications. Statin use had the greatest impact on
of chronic complications or glycemic control status. Further-
patients with DM. Additionally, we analyzed patients with DM
more, multiple types of statins were also evaluated including
according to glycemic control by fasting glucose level, and
both hydrophilic and lipophilic statins.
found that, for statin users, the risk of incident HCC was signif-
There were also limitations in this study which should be
icantly reduced in patients with DM and relatively good glyce-
addressed in further studies. A case-control study has several
mic control, compared to statin nonusers. Moreover, previous
limitations, particularly selection bias considering its nature as
studies showed that statin use was significantly related to a
an observational study. However, in the present study, the case
reduced risk of HCC among patients with liver disease, including
and control subjects were selected from a population-based
individuals with HBV or HCV infection.26,27 McGlynn et al.
cohort, the NHIS-PHEC. In general, when cases and controls
reported that statin use reduced the risk of HCC in the presence
are selected from the same source, the likelihood of selection
(AOR 0.32, 95% CI 0.17–0.57) and absence of chronic liver dis-
bias tends to be diminished in comparison with a traditional
ease (AOR 0.65, 95% CI 0.52–0.81).28 In a similar context, we
case-based case-control study.35 In addition, we could not
showed that statin use significantly reduced the risk of HCC in
assess other interventions that could be related with hyperc-
patients with LC or elevated liver enzymes, with a larger reduc-
holesterolemia, such as the use of other lipid-lowering medica-
tion than that observed in patients without LC or elevated liver
tions including fibrate, bile acid sequestrants, nicotinic acid,
enzymes.
and/or ezetimibe. The subgroup analysis according to the use
The relationships regarding both hydrophilic and lipophilic
of those medications was not available due to a low prescription
statins and incident HCC were also investigated. Hydrophilic
rate of 3.4% for those medications, which may have a relatively
statins included pravastatin and rosuvastatin and lipophilic sta-
low power in the present study.36 This should be evaluated in
tins included simvastatin, atorvastatin, lovastatin, fluvastatin,

482 Journal of Hepatology 2018 vol. 68 j 476–484


future studies. Moreover, the effect of statins on HCC develop-
ment according to other etiologies of chronic liver disease was
not evaluated, because of difficulty in diagnosing other etiolo-
gies by diagnostic code. Therefore, we used laboratory results
of liver enzymes and analyzed the effect of statin use on HCC
incidence in relation with liver enzyme abnormalities. Further-
more, although we conducted multivariable-adjusted analysis
after adjustment of LDL cholesterol level, the significant associ-
ation between statin use and the reduced risk of HCC develop-
ment remained. However, in this subset of data with LDL
cholesterol levels after 2008, the majority of the LDL levels
available were obtained after statin use, large portions of LDL
cholesterol level data prior to statin use were unavailable. We
could only analyze a single point LDL cholesterol level per study
subject after 2008, regardless of the time of statin use. There-
fore, it might be difficult to interpret the results in the subgroup
analysis according to LDL cholesterol level and bias could exist.
We were unable to confirm the association between LDL choles-
terol level and the risk of HCC development. The mechanism of
statin use and HCC development should be addressed in con-
nection with LDL cholesterol levels in animal studies and
human clinical trials.
In conclusion, statin use may have a beneficial inhibitory
effect on HCC development. The effect may be stronger in
patients with DM or LC, who are at a higher risk of HCC than
those without DM or LC.
Financial support
The study was funded by the Bio & Medical Technology Devel-
opment Program of the NRF, Korea, MSIP (2016R1A2B4013029
to E.S. Kang).
Conflict of interest
The authors declare no conflicts of interest that pertain to this
work.
Please refer to the accompanying ICMJE disclosure forms for
further details.
Authors’ contributions
E.S.K. and C.M.N. had the idea for and designed the study, super-
vised the study intervention, revised drafts of article, and
obtained funding; G.K. designed the study, analyzed, inter-
preted the data, wrote the first draft, and revised the article;
S-Y.J. designed the study, analyzed, performed the statistical
analysis, interpreted the data, and revised the article. All
authors approved the submitted manuscript.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jhep.2017.10.018.
References
Author names in bold designate shared co-first authorship.
[1] Baigent C, Blackwell L, Emberson J, Holland LE, Reith C, Bhala N, et al.
Efficacy and safety of more intensive lowering of LDL cholesterol: a
meta-analysis of data from 170,000 participants in 26 randomised trials.
Lancet 2010;376:1670–1681.
Journal of Hepatology 2018 vol. 68 j 476–484
JOURNAL
OF HEPATOLOGY
[2] Demierre MF, Higgins PD, Gruber SB, Hawk E, Lippman SM. Statins and
cancer prevention. Nat Rev Cancer 2005;5:930–942.
[3] Moon do C, Lee HS, Lee YI, Chung MJ, Park JY, Park SW, et al.
Concomitant statin use has a favorable effect on gemcitabine-erlotinib
combination chemotherapy for advanced pancreatic cancer. Yonsei Med
J 2016;57:1124–1130.
[4] Mullen PJ, Yu R, Longo J, Archer MC, Penn LZ. The interplay between cell
signalling and the mevalonate pathway in cancer. Nat Rev Cancer
2016;16:718–731.
[5] Flores A, Marrero JA. Emerging trends in hepatocellular carcinoma: focus
on diagnosis and therapeutics. Clin Med Insights Oncol 2014;8:71–76.
[6] Bosetti C, Turati F, La Vecchia C. Hepatocellular carcinoma epidemiology.
Best Pract Res Clin Gastroenterol 2014;28:753–770.
[7] Bosch FX, Ribes J, Diaz M, Cleries R. Primary liver cancer: worldwide
incidence and trends. Gastroenterology 2004;127:S5–S16.
[8] El-Serag HB, Davila JA, Petersen NJ, McGlynn KA. The continuing increase
in the incidence of hepatocellular carcinoma in the United States: an
update. Ann Intern Med 2003;139:817–823.
[9] Sherman M. Hepatocellular carcinoma: epidemiology, surveillance, and
diagnosis. Semin Liver Dis 2010;30:3–16.
[10] Lund JL, Montomoli J. Common flaws in pharmacoepidemiologic study
design and analysis. J Clin Oncol 2013;31:4161–4162.
[11] El-Serag HB, Tran T, Everhart JE. Diabetes increases the risk of chronic
liver disease and hepatocellular carcinoma. Gastroenterology
2004;126:460–468.
[12] Oda K, Uto H, Mawatari S, Ido A. Clinical features of hepatocellular
carcinoma associated with nonalcoholic fatty liver disease: a review of
human studies. Clin J Gastroenterol 2015;8:1–9.
[13] Moon YA. The SCAP/SREBP pathway: a mediator of hepatic steatosis.
Endocrinol Metab 2017;32:6–10.
[14] Kearney PM, Blackwell L, Collins R, Keech A, Simes J, Peto R, et al. Efficacy
of cholesterol-lowering therapy in 18,686 people with diabetes in 14
randomised trials of statins: a meta-analysis. Lancet 2008;371:117–125.
[15] Koo BK. Statin for the primary prevention of cardiovascular disease in
patients with diabetes mellitus. Diabetes Metab J 2014;38:32–34.
[16] Lee J, Lee JS, Park SH, Shin SA, Kim K. Cohort profile: the national health
insurance service-national sample cohort (NHIS-NSC), South Korea. Int J
Epidemiol 2017;46:e15.
[17] Sahasrabuddhe VV, Gunja MZ, Graubard BI, Trabert B, Schwartz LM, Park
Y, et al. Nonsteroidal anti-inflammatory drug use, chronic liver disease,
and hepatocellular carcinoma. J Natl Cancer Inst 2012;104:1808–1814.
[18] Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al.
Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10
administrative data. Med Care 2005;43:1130–1139.
[19] Marchesini G, Brizi M, Bianchi G, Tomassetti S, Bugianesi E, Lenzi M, et al.
Nonalcoholic fatty liver disease: a feature of the metabolic syndrome.
Diabetes 2001;50:1844–1850.
[20] Yu AS, Keeffe EB. Elevated AST or ALT to nonalcoholic fatty liver disease:
accurate predictor of disease prevalence? Am J Gastroenterol
2003;98:955–956.
[21] Ekstedt M, Franzen LE, Mathiesen UL, Thorelius L, Holmqvist M, Bodemar
G, et al. Long-term follow-up of patients with NAFLD and elevated liver
enzymes. Hepatology 2006;44:865–873.
[22] Yu JH, Kim JS, Lee MR, Yoon SY, Cho SY, Yoo SH, et al. Risks of borderline
liver enzyme abnormalities to the incidence of impaired fasting glucose
and diabetes mellitus: a 7 year follow up study of workers. Ann Occup
Environ Med 2016;28:18.
[23] Rohlfing CL, Wiedmeyer HM, Little RR, England JD, Tennill A, Goldstein
DE. Defining the relationship between plasma glucose and HbA(1c):
analysis of glucose profiles and HbA(1c) in the Diabetes Control and
Complications Trial. Diabetes Care 2002;25:275–278.
[24] Kim G, Jang SY, Han E, Lee YH, Park SY, Nam CM, et al. Effect of statin on
hepatocellular carcinoma in patients with type 2 diabetes: A nationwide
nested case-control study. Int J Cancer 2017;140:798–806.
[25] Hassan MM, Curley SA, Li D, Kaseb A, Davila M, Abdalla EK, et al.
Association of diabetes duration and diabetes treatment with the risk of
hepatocellular carcinoma. Cancer 2010;116:1938–1946.
[26] Tsan YT, Lee CH, Ho WC, Lin MH, Wang JD, Chen PC. Statins and the risk
of hepatocellular carcinoma in patients with hepatitis C virus infection. J
Clin Oncol 2013;31:1514–1521.
[27] Tsan YT, Lee CH, Wang JD, Chen PC. Statins and the risk of hepatocellular
carcinoma in patients with hepatitis B virus infection. J Clin Oncol
2012;30:623–630.
[28] McGlynn KA, Hagberg K, Chen J, Graubard BI, London WT, Jick S, et al.
Statin use and risk of primary liver cancer in the Clinical Practice
Research Datalink. J Natl Cancer Inst 2015;107.
483
Research Article Cancer

[29] Nezasa K, Higaki K, Matsumura T, Inazawa K, Hasegawa H, Nakano M,
et al. Liver-specific distribution of rosuvastatin in rats: comparison with
pravastatin and simvastatin. Drug Metab Dispos 2002;30:1158–1163.
[30] Kang S, Kim ES, Moon A. Simvastatin and lovastatin inhibit breast cell
invasion induced by H-Ras. Oncol Rep 2009;21:1317–1322.
[31] Pandyra A, Mullen PJ, Kalkat M, Yu R, Pong JT, Li Z, et al. Immediate
utility of two approved agents to target both the metabolic mevalonate
pathway and its restorative feedback loop. Cancer Res
2014;74:4772–4782.
[32] Ekstedt M, Franzen LE, Mathiesen UL, Holmqvist M, Bodemar G,
Kechagias S. Statins in non-alcoholic fatty liver disease and chronically
elevated liver enzymes: a histopathological follow-up study. J Hepatol
2007;47:135–141.
[33] Huang YW, Lee CL, Yang SS, Fu SC, Chen YY, Wang TC, et al. Statins
reduce the risk of cirrhosis and its decompensation in chronic hepatitis B
patients: a nationwide cohort study. Am J Gastroenterol
2016;111:976–985.
[34] Park HS, Jang JE, Ko MS, Woo SH, Kim BJ, Kim HS, et al. Statins increase
mitochondrial and peroxisomal fatty acid oxidation in the liver and
prevent non-alcoholic steatohepatitis in mice. Diabetes Metab J
2016;40:376–385.
[35] Szklo M, Nieto J. Basic study designs in analytic epidemiology epidemi-
ology beyond the basics. 3rd ed. Jones & Bartlett Publishers; 2014. p.
27–30.
[36] Petrie A, Sabin C. Errors in hypothesis testing. Medical Statistics at a
Glance. 3rd ed. Wiley-Blackwell; 2009. p. 52–53.

484 Journal of Hepatology 2018 vol. 68 j 476–484