Association of serum ferritin with non-alcoholic fatty liver disease: a meta-analysis
Du et al. Lipids in Health and Disease
Association of serum ferritin with non- alcoholic fatty liver disease: a meta-analysis
Shui-Xian Du 1 2
Lin-Lin Lu 0 6
Ning Geng 1
David W. Victor 3
Li-Zhen Chen 1 5
Cong Wang 1 4
Hai-Yan Yue 1 4
Yong-Ning Xin 0 1 2
Shi-Ying Xuan 0 1 2
Wen-wen Jin 1
0 Digestive Disease Key Laboratory of Qingdao , Qingdao 266071 , China
1 Department of Gastroenterology, Qingdao Municipal Hospital , 1 Jiaozhou Road, Qingdao, Shandong Province 266011 , China
2 Medical College of Qingdao University , Qingdao 266071 , China
3 Hepatology & Transplant Medicine, Department of Medicine, Houston Methodist Hospital , Houston , USA
4 Department of Gastroenterology, Dalian Medical University , Qingdao 266011 , China
5 College of Medicine and Pharmaceutics, Ocean University of China , Qingdao 266003 , China
6 Central Laboratories, Qingdao Municipal Hospital , Qingdao 266071 , China
Background: A growing number of studies reported the connection between the level of serum ferritin (SFL) and non-alcoholic fatty liver disease (NAFLD). However, such connection was still disputable. The aim of our meta-analysis was to estimate SFL between the groups as below: patients with NAFLD against control group; non-alcoholic steatohepatitis (NASH) patients against control group; non-alcoholic fatty liver (NAFL) patients against a control group and NASH patients vs NAFL patients. Methods: We screened the studies in PubMed, EMBASE, the Cochrane Database and the Cochrane Central register controlled trials from the beginning to July 10, 2016 to find the studies indicated the connection between SFL and NAFLD (NAFL and/or NASH). Fourteen published studies which evaluate the SFL in NAFLD patients were selected. Results: Higher SFL was noticed in NAFLD patients against control group (standardized mean difference [SMD] 1.01; 95% CI 0.89, 1.13), NASH patients against control group (SMD 1.21; 95% CI 1.00, 1.42), NAFL patients against control group (SMD 0.51; 95% CI 0.24, 0.79) and NASH patients against NAFL patients (SMD 0.63; 95% CI 0.52, 0. 75). These results remained unaltered actually after the elimination of studies which were focused on paediatric or adolescent populations. Higher SFL was presented in NAFLD patients against the control group (SMD 1.08; 95% CI 0.95, 1.20) in adults and NASH patients against NAFL patients in adults (SMD 0.74; 95% CI 0.62, 0.87). The connection between SFL and NASH against NAFL group in paediatric or adolescent populations was observed inconsistently (SMD 0.10; 95% CI -0.18, 0.38). Conclusions: The level of SFL was elevated in patients with NAFLD (NAFL and/or NASH) compared with the controls. Compared with NAFL, The level of SFL was increased in NASH. The result remained unaltered actually after the elimination of studies focused on paediatric or adolescent populations.
Non-alcoholic fatty liver disease (NAFLD); Meta-analysis; Serum ferritin (SFL); Non-alcoholic steatohepatitis (NASH); Non-alcoholic fatty liver (NAFL)
Non-alcoholic fatty liver disease (NAFLD) is the most
prevalent chronic liver disease worldwide. The prevalence
of it was 25.24% of the overall population [
comprises of a wide spectrum of liver damage, including
non-alcoholic fatty liver (NAFL) and non-alcoholic
steatohepatitis (NASH), as well as cirrhosis and fibrosis which
can be complicated by hepatocellular carcinoma and liver
failure . NAFLD is frequently associated with insulin
resistance (IR) and metabolic syndrome (MS) and it is
typically manifested as type 2 diabetes mellitus (T2DM),
dyslipidemia, obesity, as well as hypertension [
Therefore, the diagnose of NAFLD at very early stage is
Liver biopsy is considered to be a principle procedure
for the diagnosis of patients with NAFLD [
], however, it
is invasive [
]. NAFLD may be recognised only after the
elimination of the other liver disorders during the image
]. There are several researches which use
magnetic resonance imaging (MRI) proton density-fat
fraction to diagnose NASH. It is a non-invasive method
to assess and quantify hepatic steatosis in NAFLD patients
]. Nonetheless, its precise cut-off value has not been
estimated. In addition, the tackle for MRI is not widely
available because it is expensive. Thus, scientists are actively
looking for cheap and non-invasive biological markers
which may be helpful in the diagnosis of NAFLD and the
prognosis of NAFLD.
Serum ferritin (SFL) is a protein expressed in an acute
phase, so its level is elevated in the case of liver necrosis,
]. Some recent investigations stated that
the level of SFL can be an irrespective indicator to assess
the progression of hepatic fibrosis in the patients with
NAFLD because of its association with hepatic iron
storage and hepatic inflammation. Researchers came to a
conclusion that SFL is higher in patients with NAFLD
that might be linked with insulin resistance and
hepatocyte damage [
]. However, some empirical evidences
showed that SFL can not indicate the stage of NAFLD
. These connections are still disputed.
According to our research, no previous meta-analysis
had been done to estimate the connection between SFL
and NAFLD (NAFL and/or NASH). The purpose of
this meta-analysis was to investigate the quantitative
connection between the SFL and NAFLD (NAFL and/
or NASH) and to estimate the influence factors of this
relationship. The other aim was to evaluate whether
SFL can be treated as a potentially effective and
lessinvasive biological marker in patients with NAFLD
(NAFL and/or NASH).
According to the PRISMA directions [
], the published
studies through a systematic screening of PubMed,
EMBASE and Cochrane Database from the beginning to
July 10, 2016 were found. Keywords for the search were
as follows: (“Non-alcoholic Fatty Liver Disease”,
“Nonalcoholic Fatty Liver Disease”, “NAFLD”, “Non-alcoholic
Fatty Liver Disease”, “Fatty Liver”, “Non-alcoholic”, “Fatty
Livers”, “Non-alcoholic”, “NASH”, “Liver”, “Non-alcoholic
Fatty”, “Livers”, “Non-alcoholic Fatty”, “Non-alcoholic
Fatty Liver”, “Non-alcoholic Fatty Liver”,
“Steatohepatitides”, “Non-alcoholic”, “Steatohepatitis”,
“Nonalcoholic”) and (“Iron”,“Ferritin”).
Two authors (DU SX and LU LL) irrespectively screened
the suitable records studies: 1)are published in English;
2) the original observations including population of any
sex or ethnicity; 3) provide input related to SFL and
NAFLD (NAFL and/or NASH); 4) include the comparison
of SFL between NAFLD (NAFL and/or NASH) patients
and controls; 5) include the comparison of SFL between
NAFL and NASH patients as well.
Two researchers (DU SX and LU LL) irrespectively
elicited the data from particular eligible articles. Recorded
input consisted of: first author’s name, venue of study,
year of publication, design of study, number of patients as
well as controls and their gender, the histological degree
of NAFLD (if provided), method of NAFLD assessment,
additional information, mean values and standard
deviation (SD) of SFL.
The methodological value of studies was estimated by the
NOS (Ottawa Hospital Research Institute, Ottawa, ON,
] by two reviewers (DU SX and XIN YN) who
involved in our study.
Categories of NAFLD
In accordance to the benchmarks of NAFLD activity
score (NAS). NAS of >5 correlated with a definition of
NASH and NAS < 5 was defined as NAFL [
The primary result of this meta-analysis was the
standardise mean difference (SMD) of SFL among NAFLD patients
and control groups. NAFLD patients were categorised as
NAFL or NASH based on NOS [
]. Afterwards, we
performed a comparison of SFL among the following
] NAFLD patients against control group; [
NAFL patients against control group; [
] NASH patients
against a control group and [
] NASH patients against
Our meta-analysis used SFL as basic result. SFL was
described as the standard mean difference (SMD)
displaying 95% confidence intervals (CI). The variety of the
statistical results were estimated by the Cochran Q test
and the I2 statistic. Heterogeneity was recognised as
significant when the Cochran Q test was p < 0.05 or I2 was
more than 50% [
]. Depending on the absence or
presence of heterogeneity, different types of models
including fixed-effects and random model were used. All
subgroups were subjected to analyse. We investigated all
related articles on the SFL individually of different types
of studies (including case-control studies, prospective
studies and cross-sectional). In order to explore if the
level of SFL can effect the progression of NAFLD, we
also investigated the SFL among NAFL patients
compared with NASH patients separately.
Furthermore, we increased a sensitivity analysis through
the elimination of studies focused on adolescent/paediatric
population. Next, the impact of each study on the pooled
measures was evaluated by ignoring one in each turn and
then the summarised SMDs of the rest subjects were
]. We used Funnel plots to estimate the publication
bias at first [
] and later this bias was corroborated by
using Begg’s [
] and Egger’s tests [
]. Our meta-analysis
was performed using Stata Statistical Software (ver. 12.0;
StataCorp LP, College Station, TX).
Figure 1 presents the selection process of the studies
and literature search results in this meta-analysis. After
the initial search, we obtained 563 results. We screened
titles and abstracts, 494 of them were excluded due to
plenty of reasons, including lack of primary data (reviews
and meta-analysis), inappropriate topics, non-human
studies, negligible population (alcoholic fatty liver
disorder) and liver disease other than NAFLD. At last, 14
studies in total were chosen for further analysis after
reviewing full texts
Characteristics of the included studies
The major features of these trials were summarised in
Table 1. After the whole presented workflow, 14 studies
were admitted to our meta-analysis [
for NAFLD patients was carried only when a control group
was not included in the study (i.e. in the situation when
there was a comparison of SFL only between NAFL and
NASH patients). Therefore, it was impossible to compare
NAFLD patients. Five studies were performed in Europe,
five in Asia and four in North America. Studies in the
meta-analysis included one cross-sectional study, nine
casecontrol studies and four prospective studies. NAFLD
(NAFL and/or NASH) was confirmed by hepatic
ultrasonography in two studies and liver biopsy in twelve studies.
The outcome measure of each study was presented in
Table 2. Two studies consisted of all groups (controls,
NAFL, NASH patients) [
]. Three studies compared
SFL between NASH patients and controls [
studies compared SFL between NAFL patients and controls
]. Three studies compared SFL between NAFLD
patients and controls, but they didn’t carry independent
evidence on both NAFL and NASH [
9, 21, 28
studies compared SFL between NAFL and NASH
10, 11, 20, 22, 23, 25–27, 29, 30
] (Table 2).
Following comparative data were provided: three studies,
NAFLD patients (n = 519) against control group (n = 748),
two studies, NAFL patients (n = 107) against control group
(n = 108), three studies, NASH patients (n = 178) against
control group (n = 198) and ten studies, NAFL (n = 561)
against NASH patients (n = 871).
Quality of included studies
In accordance to NOS, Table 1 shows the value of
included studies. Two studies scored 7, seven studies scored
6, four studies scored 5 and one study scored 4 (mean ±
SD 6.15 ± 0.97). No study was eliminated due to the low
NOS (score ≤ 2).
Higher SFL was noticed in the following groups: (1)
NAFLD patients against controls; (2) NAFL patients against
control; (3) NASH patients against control and (4) NASH
against NAFL patients (Table 2; Figs. 2, 3, 4, 5 and 6). The
variety amongst the studies was mild-to-severe in the case
of all juxtapositions (I2 ranged from 0% to 88.4%; Fig. 2).
iit rm se se
d o e e
d f A A b A A A A A A A A b A A
A in N N o N N N N N N N N o N N
iito iob iob sno iob iob iob iob iob iob iob iob sno iob iob
ifenD lirve lirve lrtaU ilrve irLve lirve lirve lirve lirve lirve lirve lrtau lirve lirve
l]e )re )1 ,44 )3 ) ) )
l/aaFeem (SunbHm )(/1315 ,(/99905 (/12716 ,(/49615 ,(/312644 ()50 ()52 )(/3730 ,(/171311 ,(/482013
9 1 0
M A 8 8 A 9 1 7 A 1 2 5 7 A 0 8
[ N 2 1 N 2 1 5 N 1 6 1 6 N 3 6
N ) )
f r ) 6 ) ) )
so eb ) ,51 ,24 ,49 ) ,11 ,12
irteeog (LFunm (/3723 (/2341 (/5426 (/4347 ()46 ()48 (/1721 (/5119 (/3019
a A 0 4 A 1 0 A A 7 4 5 8 A 0 9
C N 6 6 N 1 9 N N 1 2 3 3 N 7 4
) ) )
)]rsae ,4679 ,)630 .,)425 ,9453 ,4023 )2
sse (yaeg lrto (/8231 (/7412 (/8254 (/0611 (/0132 )0 ,/3317 lliaaeb
lya an no A 87 76 A 00 45 31 (3 0 A A (01 A A av
n C N 5 1 N 1 2 5 8 2 N N 5 N N to
a m n
a e A
m e ) )
1 8 )
e /F ,5 ,4 6
th le 1 ) 6 ) ) ,4
a 3 3 2 8 4 5
in [M /1 ,33 /2 01 ,46 /9
r 2 2 9 / 2 1
d e 2 / 7 2 / 5
e b (1 (20 (1 (9 (31 (3
in 3 5 0 6 5 0 7 ic
a um sae 8 5 3 0 0 7 4 8 6 0 0 0 0 1 l
t N c 8 2 4 4 2 5 4 2 8 5 1 3 1 1 oh
ieud rtyun aand anp ianw ian A ly irskh ian an an itaa rkye ly A H
fts oC aC Ja aT hC SU Ita Tu hC Jap Jap roC SU Tu Ita SU SAN
) ) a
) ) 1 2 p
,34 )1 ,21 ,12 ,19 eoh
8 5 1 3 3 t
1 / 1 / / a
/ 4 / 8 8 e
23 (5 91 (6 (7 ts
studies. The signs of publication bias were not observed
(p > 0.05 for all comparisons, Table 3).
There was no meaningful bias in any collation (p > 0.05 for
all comparisons; Table 3 and Fig. 1).
In the sensitivity analysis, after the elimination of
paediatric/adolescent studies, there were only little
alterations among groups (Figs. 7 and 8, Table 2). The estimated
heterogeneity of NAFLD and control group was 63.4%
and the heterogeneity in NAFL and NASH group was
still 76.5%. Based on the different types of studies,
subgroup showed that NASH patients showed 0.78 ng/mL
higher level of SFL compared with NAFL (95% CI: 0.59,
0.97 ng/mL) (I2 = 82.5%, p < 0.001) in four case-control
10, 20, 23, 26
], while the SMD of SFL was
0.71 ng/mL (95% CI, 0.54, 0.89 ng/mL) (I2 = 75.8%, p =
0.006) in four prospective studies [
11, 22, 25, 27
] (Figs. 7
and 8) after the elimination of paediatric/adolescent
After performing this meta-analysis, we concluded that
higher SFLcan be linked with the severity of NAFLD since
the controls showed lower SFL compared with NAFL,
NASH or NAFLD patients and NAFL patients showed
lower SFL compared with NASH patients. The sensitivity
analyses and subgroup analyses did not essentially influence
or alter these conclusions. As such, SFL can be as a
lessinvasive and effective biological marker to prognosticate
the progression of NAFLD.
In terms of the hypothetical mechanisms linking SFL
and NAFLD, SFL displayed strong biological plausibility,
thus it can be used as a marker in the determination of
NAFLD. Existing two-hit theory which takes the
progression to NASH and fibrosis into account, is the most
common mechanism regarding the pathogenesis of NAFLD
]. In this assumption, the first “hit” is IR which related
with visceral obesity, resulting in free fatty acids and
elevated circulating hepatic steatosis. On the other hand, the
second “hit” might be induced by the additional factors
which may result in inflammation of the liver and elevated
oxidative stress and ultimately lead to tissue injury,
steatohepatitis and fibrosis [
]. Few researches indicated that the
elevated deposition of iron was an important factor in
catalysing the production of reactive oxygen species through
the Fenton reaction, which was suggested to be the second
hit. Besides the production of reactive oxygen species [
iron may play a role in a number of different disastrous
pathways, including changed insulin signalling and lipid
metabolism. In the liver, where the majority of extra body
iron is retained. SFL is the main iron-storage protein. It can
be increased secondary due to the steatohepatitis, obesity,
histiocytic neoplasm, chronic consumption of alcohol as
well as chronic inflammation including viral hepatitis [
Together with the elevated level of ferritin concentration,
the risk of serious liver disease is increasing constantly.
Manousou P, et al. [
] reported that the elevated SFL may
reflect the occurrence of hepatitic failure and metabolic
syndrome because of the activation of inflammatory
cytokines in NAFLD patients. What’s more, Nelson JE, et
al. reported that hepatic iron accumulation is correlated
with hepatic fibrosis in NAFLD subjects, what is confirmed
in a large number of studies focused on the
pathophysiological point of view [
]. Valenti L, et al. reported that the
accumulation of hepatic iron may contribute to the
production of inflammatory cytokines, what might lead to the
hepatic fibrosis [
]. According to the research performed by
Kowdley et al. [
], the histological characteristics,
including fibrosis of NAFLD, steatosis and hepatocellular
ballooning,were more serious in the case of patients with higher
SFL. They reported that SFL may be linked with the
aggravated histological function and hepatic iron exemption
among patients with NAFLD.
There are many benefits resulting from the presented
study. As we know, this is the first meta-analysis which
evaluates the connection between the SFL and NAFLD
based on the extensive search. As NAFLD consists of a
wide spectrum of disorders, our meta-analysis was carried
out in order to uncover changed SFL in NAFL and NASH,
comparing to the healthy controls. In addition, we also
performed the evaluation of NAFL and NASH patients in
order to examine whether SFL was related with the
severity of NAFLD. On the other hand, the analysis was
revealed the connection between SFL and NAFLD in adults
and paediatric or adolescent populations separately.
However, there are some significant restraints concerning
this meta-analysis. First, the majority of original studies did
not match the potential confounders, such as
hyperlipidemia, IR, liver enzymes and body mass index. We did not
manage to confirm that SFL poses an independent risk
factor for NAFLD. Second, the evaluation of liver enzymes
was relatively insensitive to detect NAFLD, what may be
the result of possible wrong categorization of patients with
NAFLD as unaffected controls. Third, the veracity of the
results was restrained due to the variety of between-study,
which should be exclusively commentated in the reference
to dissimilarities of BMI between compared groups. Four,
we eliminated unpublished studies or abstracts from
conferences, which may lead to the bias. However, such
elimination is crucial in order to refrain the low-quality
input, because its value cannot be evaluated in total
]. Five, because of the lack of corroborated quality
assessment instrument for cross-sectional studies, NOS,
the most prevalent ratio for observational studies was
used in order to eliminate low-value studies. Six, since
only four studies specified that the controls constitute the
same pool as the subjects do, the number of valuable
case-control studies’ might also be restrained. Seven, we
did not evaluate SFL in case of inflammation, fibrosis stage
or steatosis individually because of the lack of the available
histological lesions data. It was significantly limited by the
division of groups and different histological
interpretations. Finally, we did not manage to perform subgroup
and sensitivity analyses in order to reveal the effects of
other potential factors, such as the definition of NAFLD,
gender and race, and the way of testing the SFL, due to an
inadequate number of data.
This meta-analysis explored that NAFLD patients showed
a higher SFL, what can be related with the severity of
NAFLD. These results are consistent with the hypothesis
that the elevated SFL is related with IR and hepatocyte
damage and it also plays a fibrotic and pro-inflammatory
role during the progression of the disease. The further
studies also be needed to reveal the causal role of SFL in
the progression of NAFLD and the mechanism of the
pathogenesis of NAFLD.
MRI: Magnetic resonance imaging; MS: Metabolic syndrome; NAFL:
Nonalcoholic fatty liver; NAFLD: Non-alcoholic fatty liver disease; NASH:
Nonalcoholic steatohepatitis; SFL: Serum ferritin; T2DM: Type 2 diabetes mellitus
We thank Qingdao Medical University, Qingdao Municipal Hospital, Digestive
Disease Key Laboratory of Qingdao and all the participants in the our study.
This study was supported by the Key Research Project of Shandong
Province(2016GSF201217)、Medical and Health Technology Development
Project of Shandong Province (2015WS0321)、Qingdao, Shinan District Science
and Technology Development Project Fund (2015–6-014-YY、2016–3-016-YY).
Availability of data and materials
All BioMed Central journals require a data availability statement to be present in
the article. Data availability statements provide a statement about where data
supporting the results reported in a published article can be found. More
guidance and a list of appropriate text for this statement can be found in our
Data Availability Guidance for Authors and Editors.
SXD, LLL and YNX were accountable for the concept and design of the study.
SXD and LLL acquired data. Data were interpreted by SXD, CW, HYY and YNX and
statistical analysis were performed by LZC and NG. The manuscript was drafted by
SXD, WWW, DWV and SYX and critically reviewed for important intellectual
content by all authors. The final version of the manuscript was approved for
publishing by all authors. SXD and LLL are the guarantors of this work.
Ethics approval and consent to participate
This study was approved by the ethics committee on human research of
Qingdao municipal hospital (Qingdao, China). This study was performed in
accordance with the principles of the declaration of Helsinki and its appendices.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
1. Chalasani N , Younossi Z , Lavine JE , et al. The diagnosis and management of non-alcoholic fatty liver disease: practice guideline by the American Gastroenterological Association, American Association for the Study of Liver Diseases, and American College of Gastroenterology . Gastroenterology. 2012 ; 142 : 1592 - 609 .
2. Rinella M , Charlton M. The globalisation of non-alcoholic fatty liver disease: prevalence and impact on world health . Hepatology . 2016 ; 64 ( 1 ): 19 - 22 .
3. Dai Y , Zhu J , Meng D , et al. Association of homocysteine level with biopsyproven non-alcoholic fatty liver disease: a meta-analysis . J Clin Biochem Nutr . 2016 ; 58 ( 1 ): 76 - 83 .
4. Méndez-Sánchez N , Arrese M , Zamora-Valdés D , et al. Current concepts in the pathogenesis of nonalcoholic fatty liver disease . Liver Int . 2007 ; 27 : 423 - 33 .
5. Angulo P , Hui JM , Marchesini G , et al. The NAFLD fibrosis score: a noninvasive system that identifies liver fibrosis in patients with NAFLD . Hepatology . 2007 ; 45 : 846 - 54 .
6. Na JH , Park SW , Kang Y , et al. The clinical significance of serum ferritin in pediatric non-alcoholic fatty liver disease . Pediatr Gastroenterol Hepatol Nutr . 2014 ; 17 ( 4 ): 248 - 56 .
7. Permutt Z , Le TA , Peterson MR , et al. Correlation between liver histology and novel magnetic resonance imaging in adult patients with non-alcoholic fatty liver disease - MRI accurately quantifies hepatic steatosis in NAFLD . Aliment Pharmacol Ther . 2012 ; 36 : 22 - 9 .
8. Bell H , Skinningsrud A , Raknerud N , et al. Serum ferritin and transferrin saturation in patients with chronic alcoholic and non-alcoholic liver diseases . J Intern Med . 1994 ; 236 : 315 - 22 .
9. Manousou P , Kalambokis G , Grillo F , et al. Serum ferritin is a discriminant marker for both fibrosis and inflammation in histologically proven nonalcoholic fatty liver disease patients . Liver Int . 2011 ; 31 ( 5 ): 730 - 9 .
10. Yoneda M , Nozaki Y , Endo H , et al. Serum ferritin is a clinical biomarker in Japanese patients with nonalcoholic steatohepatitis (NASH) independent of HFE gene mutation . Dig Dis Sci . 2010 ; 55 ( 3 ): 808 - 14 .
11. Chandok N , Minuk G , Wengiel M , et al. Serum ferritin levels do not predict the stage of underlying non-alcoholic fatty liver disease . J Gastrointestin Liver Dis . 2012 ; 21 ( 1 ): 53 - 8 .
12. Liberati A , Altman DG , Tetzlaff J , et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration . J Clin Epidemiol . 2009 ; 62 : e1 - e34 .
13. von Elm E , Altman DG , Egger M , et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies . Ann Intern Med . 2007 ; 147 : 573 - 7 .
14. Kleiner DE , Brunt EM , Van Natta M , et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease . Hepatology . 2005 ; 41 : 1313 - 21 .
15. Higgins JP , Thompson SG . Quantifying heterogeneity in a meta-analysis . Stat Med . 2002 ; 21 : 1539 - 58 .
16. Higgins JP , Thompson SG , Deeks JJ , et al. Measuring inconsistency in metaanalyses . BMJ . 2003 ; 327 : 557 - 60 .
17. Thornton A , Lee P . Publication bias in meta-analysis: its causes and consequences . J Clin Epidemiol . 2000 ; 53 : 207 - 16 .
18. Begg CB , Mazumdar M. Operating characteristics of a rank correlation test for publication bias . Biometrics . 1994 ; 50 : 1088 - 101 .
19. Egger M , Davey Smith G , Schneider M , et al. Bias in meta-analysis detected by a simple, graphical test . BMJ . 1997 ; 315 : 629 - 34 .
20. Hotta K , Yoneda M , Hyogo H , et al. Association of the rs738409 polymorphism in PNPLA3 with liver damage and the development of nonalcoholic fatty liver disease . BMC Med Genet . 2010 Dec 22 ; 11 : 172 .
21. Hsiao TJ , Chen JC , Wang JD . Insulin resistance and ferritin as major determinants of nonalcoholic fatty liver disease in apparently healthy obese patients . Int J Obes Relat Metab Disord . 2004 ; 28 ( 1 ): 167 - 72 .
22. Goh GB , Issa D , Lopez R , et al. The development of a non-invasive model to predict the presence of non-alcoholic steatohepatitis in patients with nonalcoholic fatty liver disease . J Gastroenterol Hepatol . 2016 ; 31 ( 5 ): 995 - 1000 .
23. Di Rosa M , Mangano K , De Gregorio C , et al. Association of chitotriosidase genotype with the development of non-alcoholic fatty liver disease . Hepatol Res . 2013 ; 43 ( 3 ): 267 - 75 .
24. Sazci A , Akpinar G , Aygun C , et al. Association of apolipoprotein E polymorphisms in patients with non-alcoholic steatohepatitis . Dig Dis Sci . 2008 ; 53 ( 12 ): 3218 - 24 .
25. Jiang Y , Zeng J , Chen B . Hemoglobin combined with triglyceride and ferritin in predicting non-alcoholic fatty liver . J Gastroenterol Hepatol . 2014 ; 29 ( 7 ): 1508 - 14 .
26. Tsuchiya H , Ashla AA , Hoshikawa Y , et al. Iron state in association with retinoid metabolism in non-alcoholic fatty liver disease . Hepatol Res . 2010 ; 40 ( 12 ): 1227 - 38 .
27. Canbakan B , Senturk H , Tahan V , et al. Clinical, biochemical and histological correlations in a group of non-drinker subjects with non-alcoholic fatty liver disease . Acta Gastroenterol Belg . 2007 ; 70 ( 3 ): 277 - 84 .
28. Demircioğlu F , Görünmez G , Dağıstan E , et al. Serum hepcidin levels and iron metabolism in obese children with and without fatty liver: case-control study . Eur J Pediatr . 2014 ; 173 ( 7 ): 947 - 51 .
29. Nobili V , Siotto M , Bedogni G , et al. Levels of serum ceruloplasmin associate with pediatric nonalcoholic fatty liver disease . J Pediatr Gastroenterol Nutr . 2013 ; 56 ( 4 ): 370 - 5 .
30. Alkhouri N , Alisi A , Okwu V , et al. Circulating soluble Fas and Fas ligand levels are elevated in children with non-alcoholic steatohepatitis . Dig Dis Sci . 2015 ; 60 ( 8 ): 2353 - 9 .
31. Dowman JK , Tomlinson JW , Newsome PN . Pathogenesis of non-alcoholic fatty liver disease . QJM . 2010 ; 103 : 71 - 83 .
32. Nelson JE , Klintworth H , Kowdley KV . Iron metabolism in non-alcoholic fatty liver disease . Curr Gastroenterol Rep . 2012 ; 14 ( 1 ): 8 - 16 .
33. Kowdley KV , Belt P , Wilson LA , et al. NASH clinical research network. Serum ferritin is an independent predictor of histologic severity and advanced fibrosis in patients with nonalcoholic fatty liver disease . Hepatology . 2012 ; 55 : 77 - 85 .
34. Nelson JE , Wilson L , Brunt EM , et al. Non-alcoholic steatohepatitis clinical research network. Relationship between the pattern of hepatic iron deposition and histological severity in nonalcoholic fatty liver disease . Hepatology . 2011 Feb; 53 ( 2 ): 448 - 57 .
35. Valenti L , Dongiovanni P , Fargion S. Diagnostic and therapeutic implications of the association between ferritin level and severity of nonalcoholic fatty liver disease . World J Gastroenterol . 2012 ; 18 : 3782 - 6 .
36. Egger M , Juni P , Bartlett C , et al. How important are comprehensive literature searches and the assessment of trial quality in systematic reviews? Empirical study . Health Technol Assess . 2003 ; 7 : 1 - 76 .