Nonalcoholic fatty liver disease (NAFLD) is becoming an epidemic in Western populations, with prevalence ranging from approximately 20% to 30% depending on the country and the different reports [
A growing body of evidence is demonstrating that genetic susceptibility plays an important role in NAFLD development and evolution [
Based on this background, in the present study, we aimed to verify if combining different SNPs in a genetic score could increase the predictive power for more severe forms of NAFLD in adult patients. To address this, we performed a case-control study in a well-characterized NAFLD population and tested the same SNPs which had been previously assembled to provide the genetic score in the pediatric population [
Patients followed at the Hepatology Unit of the University Hospital Campus Bio-Medico of Rome, at the Hepatology outpatient clinic of the Colleferro Hospital (Rome, Italy), and at the Liver Disease Unit of the San Camillo-Forlanini Hospital of Rome and those assessed at the Laboratory of Pathology of the National Institute for Infectious Diseases Lazzaro Spallanzani of Rome and recorded in the respective clinical databases with the diagnosis of NASH cirrhosis were extracted. When reviewing the clinical records for the purpose of the present study, the etiological diagnosis of cirrhosis as post-NASH was considered definite and patients eligible for the study only after having verified that all the other causes of liver disease had been excluded and that at least one of the two criteria was satisfied, (1) histological diagnosis of NASH cirrhosis; (2) clinical diagnosis of cirrhosis in patients with a previous documented history of steatosis (by abdominal ultrasonography, computed tomography scan, or magnetic resonance imaging) and at least one of the following diagnoses precede that of cirrhosis: type-2 diabetes mellitus (T2DM), obesity with a body mass index (BMI) > 30 kg/m2, and metabolic syndrome. Secondly, the group of patients who had received the diagnosis of noncirrhotic NAFLD (simple steatosis or NASH) by liver biopsy performed at the same Centers was extracted from the clinical databases. Also in this case, it was confirmed that all the other causes of liver disease had been excluded before or after liver biopsy. More specifically, it was verified that none of the patients with NASH cirrhosis and noncirrhotic NAFLD had history of alcoholic intake >20 gr/day if woman and >30 gr/day if man or of use of drugs known to induce liver steatosis. Moreover, it was verified that the following results had been obtained from all the patients: negative anti-hepatitis C virus (anti-HCV) antibodies and hepatitis B surface antigen (HBsAg); anti-nuclear antibodies (ANA) ≤1 : 80 and negative anti-mitochondrial (AMA), anti-smooth muscle (ASMA), and anti-liver and kidney microsomal (anti-LKM) antibodies; normal transferrin saturation and serum levels of ceruloplasmin and alpha-1 antitrypsin. Finally, a third group to represent the general population was recruited among consecutive subjects undergoing general medicine company checkups at the Campus Bio-Medico Hospital. Upper abdomen ultrasonography is among the examination included in every checkup, and ultrasonographic steatosis was considered the only exclusion criterion.
Patients with NASH cirrhosis and noncirrhotic NAFLD individuated according to the aforementioned criteria were contacted and invited to participate to the study, as well as all consecutive subjects attending company checkups. All subjects enrolled in the present study agreed to participate and signed an informed consent. The protocol of the study was approved by the Ethics Committee of the University Campus Bio-Medico of Rome.
Epidemiological, anthropometric, and clinical data at the time of cirrhosis diagnosis of liver biopsy or of the medical checkup were collected for patients with NASH cirrhosis and noncirrhotic NAFLD and for subjects of the control group, respectively. All subjects enrolled in the study underwent blood drawing and venous blood was stored at −80° until examination.
Genomic DNA was extracted from whole blood using the kit DNA-Sorb-B (Sacace Biotechnologies, Como, Italy), as instructed by the manufacturer. The analysis of the five polymorphisms was done by real-time polymerase chain reaction (rt-PCR) using a commercial kit (FLT PLUS Fatty Liver Test; Orga Bio Human, Rome, Italy), as instructed by the manufacturer. For each SNP, the presence of two TaqMan® allows allelic discrimination (ancestral allele, variant allele, or both).
Abdominal ultrasound was performed with an Acuson S3000 ultrasound system (Siemens®, Munich, Germany), using both a 1.5–6 MHz curved array probe and a 5.5–18 MHz high frequency linear array probe, the latter for the evaluation of the left liver surface. All patients were studied with subcostal, right intercostal and transaxial epigastric scans, and ultrasonography was operated by a single experienced liver sonographer blinded to the patients’ clinical data. Liver was defined bright when abnormally intense, high-level echoes arose from hepatic parenchyma and/or when a liver-kidney difference in echo amplitude was detected [
Liver biopsy was performed under ultrasonographic guidance, and specimens (at least 1.5 cm in length) were fixed in formalin for evaluation. Tissue sections were stained with hematoxylin and eosin and with Sirius red. Liver biopsies were reviewed by two expert liver pathologists (A. B. and S. C.), who were unaware of clinical and genetic data. Histological cases of NASH cirrhosis were reevaluated in order to confirm diagnosis, while cases of noncirrhotic NAFLD were diagnosed as simple steatosis or NASH [
Results are reported as mean ± standard deviation (SD), number (%), and OR with 95% confidence intervals (CI), as appropriate. Differences between groups of patients with NASH cirrhosis and noncirrhotic NAFLD and healthy controls were tested using Kruskal-Wallis test for continuous variables and
All subjects included in the present study were of Caucasian race and of Italian nationality. One hundred and seventy-five patients with the diagnosis of NASH cirrhosis were extracted from the four clinical databases. Among these patients, 35 had died, 17 refused to participate to the study, and other 16 were discarded, 12 because they did not meet the aforementioned criteria to be included as NASH cirrhosis cases and 4 because of an excessive alcohol intake. Finally, 107 patients with NASH cirrhosis were enrolled: 39 had received histological diagnosis of NASH cirrhosis; 68 had received the clinical diagnosis of cirrhosis (clinical, biochemical, and ultrasonographic ± upper endoscopic ± transient elastographic findings) and, according to inclusion criteria, had a previous documented history of steatosis (mainly ultrasonographic) together with at least one among T2DM (63 patients), obesity (53 patients), and metabolic syndrome (60 patients) preceding the diagnosis of cirrhosis. One hundred and nineteen patients who had received the histological diagnosis of noncirrhotic NAFLD were extracted from the four clinical databases. Five patients had died, 13 of them refused to participate in the study, and 8 were discarded, 5 because of an excessive alcohol intake and three because not all the alternative etiologies had been excluded. Finally, 93 patients with noncirrhotic NAFLD were enrolled in the study. According to Brunt et al. [
As reported in Table
Epidemiologic and clinical data and frequency of the PNPLA3 rs738409 G, TM6SF2 rs58542926 T, KLF6 rs3750861 T, SOD2 rs4880 T, and LPIN1 rs13412852 T alleles in the study population, divided according to subgroups.
All | Healthy Subjects | Noncirrhotic NAFLD | NASH cirrhosis | |
---|---|---|---|---|
| 290 | 90 | 93 | 107 |
Age (years) | 54.4 (15.1) | 43.1 (11.9) | 51.6 (13.2) | 66.2 (9.8) |
Sex (male) | 53.5% | 46.6% | 60.2% | 53.3% |
BMI (Kg/m2) | 28.8 (5.3) | 25.2 (4.6) | 29.5 (4.1) | 31.2 (5.2) |
T2DM | 41.6% | 7% | 45.2% | 66.4% |
PNPLA3 G allele frequency | 0.46 | 0.24 | 0.48 | 0.61 |
TM6SF2 T allele frequency | 0.11 | 0.06 | 0.11 | 0.14 |
KLF6 T allele frequency | 0.08 | 0.06 | 0.07 | 0.12 |
SOD2 T allele frequency | 0.5 | 0.48 | 0.53 | 0.49 |
LPIN1 T allele frequency | 0.32 | 0.35 | 0.32 | 0.31 |
Data are shown as mean (standard deviation) or percentages.
The frequency of the PNPLA3 rs738409 minor (G) allele was significantly higher in patients with NASH cirrhosis than in the other two study populations, as well as in noncirrhotic NAFLD patients with respect to healthy controls. Concerning the TM6SF2 rs58542926 polymorphism, an increase of the T allele frequency was observed in patients with NASH cirrhosis with respect to healthy subjects, while the comparisons between the other groups did not yield statistically significant differences. Similarly to the latter, the increase in the frequency of the T allele at the KLF6 rs3750861
Table
Genotype distribution for the polymorphisms of interest at the five different
Comparison of PNPLA3 rs738409, TM6SF2 rs58542926, KLF6 rs3750861, SOD2 rs4880, and LPIN1 rs13412852 polymorphisms prevalence, according to study subgroups.
Healthy Subjects | Noncirrhotic NAFLD | NASH cirrhosis | | |
---|---|---|---|---|
PNPLA3 | ||||
CC | 53 (58.9%) | 30 (32.3%) | 20 (18.7%) | <0.001 |
CG | 30 (33.3%) | 36 (38.7%) | 44 (41.1%) | |
GG | 7 (7.8%) | 27 (29%) | 43 (40.2%) | |
TM6SF2 | ||||
CC | 79 (87.8%) | 74 (79.6%) | 79 (73.8%) | 0.105 |
CT | 11 (12.2%) | 17 (18.3%) | 27 (25.2%) | |
TT | 0 (0%) | 2 (2.2%) | 1 (0.9%) | |
TM6SF2 dichotomized | ||||
CC | 79 (87.8%) | 74 (79.6%) | 79 (73.8%) | 0.06 |
CT/TT | 11 (12.2%) | 19 (20.4%) | 28 (26.2%) | |
KLF6 | ||||
CC | 79 (87.8%) | 80 (86%) | 83 (77.6%) | 0.224 |
CT | 11 (12.2%) | 13 (14%) | 23 (21.5%) | |
TT | 0 (0%) | 0 (0%) | 1 (0.9%) | |
KLF6 dichotomized | ||||
CC | 79 (87.8%) | 80 (86%) | 83 (77.6%) | 0.113 |
CT/TT | 11 (12.2%) | 13 (14%) | 24 (22.4%) | |
SOD2 | ||||
CC | 26 (28.9%) | 20 (21.5%) | 30 (28%) | 0.795 |
CT | 42 (46.7%) | 48 (51.6%) | 49 (45.8%) | |
TT | 22 (24.4%) | 25 (26.9%) | 28 (26.2%) | |
LPIN1 | ||||
CC | 38 (42.2%) | 45 (48.4%) | 49 (45.8%) | 0.672 |
CT | 41 (45.6%) | 37 (39.8%) | 50 (46.7%) | |
TT | 11 (12.2%) | 11 (11.8%) | 8 (7.5%) |
Data are shown as absolute number and column percentages.
Epidemiological, anthropometric, and genetic data of the noncirrhotic NAFLD group divided into two subgroups according to the degree of fibrosis: absent/mild (F0-F1) and advanced (F2-3).
F0-1 | F2-3 | | |
---|---|---|---|
| 51 | 42 | |
Age (years) | 46.7 (13.1) | 57.5 (11) | <0.001 |
Sex (male) | 64.7% | 54.8% | 0.332 |
BMI (Kg/m2) | 28.5 (3.8) | 30.8 (4.2) | 0.012 |
T2DM | 31.4% | 61.9% | 0.006 |
PNPLA3 | |||
G allele frequency | 0.44 | 0.54 | 0.203 |
CC | 19 (37.3%) | 11 (26.2%) | 0.491 |
CG | 19 (37.3%) | 17 (40.5%) | |
GG | 13 (25.5%) | 14 (33.3%) | |
TM6SF2 | |||
T allele frequency | 0.09 | 0.14 | 0.375 |
CC | 42 (82.4%) | 32 (76.2%) | 0.278 |
CT | 9 (17.6%) | 8 (19%) | |
TT | 0 (0%) | 2 (4.8%) | |
KLF6 | |||
T allele frequency | 0.06 | 0.08 | 0.781 |
CC | 45 (88.2%) | 35 (83.3%) | 0.705 |
CT | 6 (11.8%) | 7 (16.7%) | |
TT | 0 (0%) | 0 (0%) | |
SOD2 | |||
T allele frequency | 0.56 | 0.49 | 0.395 |
CC | 9 (17.6%) | 11 (26.2%) | 0.58 |
CT | 27 (52.9%) | 21 (50%) | |
TT | 15 (29.4%) | 10 (23.8%) | |
LPIN1 | |||
T allele frequency | 0.31 | 0.32 | 1 |
CC | 26 (51%) | 19 (45.2%) | 0.585 |
CT | 18 (35.3%) | 19 (45.2%) | |
TT | 7 (13.7%) | 4 (9.5%) | |
SCORE (linear) | 1.2 (0.9) | 1.5 (1) | 0.139 |
Data are shown as mean (standard deviation), percentages, or absolute number and column percentages. NAFLD: nonalcoholic fatty liver disease; NASH: nonalcoholic steatohepatitis; BMI: body mass index; T2DM: type 2 diabetes mellitus; PNPLA3: patatin-like phospholipase domain containing protein 3; TM6SF2: transmembrane 6 superfamily member 2; KLF6: Kruppel-like factor 6; SOD2: superoxide dismutase 2; LPIN1: lipin 1.
The PNPLA3 rs738409, TM6SF2 rs58542926, and KLF6 rs3750861 polymorphisms were then considered for the development of the genetic risk score. Consistent with previous literature [
The genetic risk score was significantly higher in noncirrhotic NAFLD patients with respect to healthy subjects [1.3 (0.9) versus 0.8 (0.9),
Prevalence of different results in the genetic risk score according to the three study populations.
Table
Uni- and multivariable regression models for the association of the genetic score (modelled as numeric or as factor variable) and presence of NASH cirrhosis or noncirrhotic NAFLD.
NASH cirrhosis versus | Noncirrhotic NAFLD versus | NASH cirrhosis versus | ||||
---|---|---|---|---|---|---|
OR | aOR | OR | aOR | OR | aOR | |
Score | 1.54 (1.14–2.1) | 1.51 (1.04–2.22) | 2.05 (1.46–2.93) | 2.22 (1.40–3.65) | 3.01 (2.14–4.38) | 3.62 (1.63–9.69) |
Score = 0 | ref | ref | ref | ref | ref | ref |
Score = 1 | 1.62 (0.69–3.9) | 2.07 (0.69–6.45) | 3.40 (1.63–7.28) | 4.59 (1.74–12.91) | 5.50 (2.46–13.00) | 15.40 (2.32–149) |
Score = 2 | 2.18 (0.93–5.29) | 2.5 (0.84–7.79) | 3.33 (1.55–7.36) | 4.38 (1.58–13.05) | 7.28 (3.24–17.35) | 18.54 (2.78–176) |
Score = 3 | 3.83 (1.4–11.14) | 3.66 (1.04–13.75) | 23.00 (4.01–436.62) | 21.98 (2.73–496.88) | 88.17 (16.11–1658) | 125.37 (4.22–19336) |
OR: odds ratio; aOR: adjusted odds ratio. aOR are adjusted for age, sex, BMI, and diabetes mellitus.
This is the first study aimed to verify if combining the presence of different polymorphisms of risk could increase the predictive power for NAFLD and, mainly, for its evolution to cirrhosis. The present results are proof of concept that the effects determined by disease-associated variants at different
A growing body of evidence indicates that NAFLD develops as a result of a complex process in which genetic susceptibility plays an important role [
American and European NAFLD guidelines recognize the disease modifying role of the PNPLA3 variant and of both the PNPLA3 and the TM6SF2 SNPs, respectively, but testing for these genetic variants in routine clinical practice is still not recommended [
A pioneering study aimed to combine the effect of different SNPs in a genetic risk score able to predict NASH has been carried out by Nobili et al. in the pediatric population [
Based on this background, we designed this case-control study in a well-characterized NAFLD population from 4 different hepatological tertiary centers from Rome and its province, by recruiting three groups: NASH cirrhotics, noncirrhotic NAFLD, and controls. This subdivision may have the limit to render the noncirrhotic NAFLD population heterogeneous, since it was constituted by patients with both mild (F0-1) and advanced fibrosis (F2-3). Moreover, since NAFLD patients with milder fibrosis are younger than those with more advanced disease, the possibility that they can progress to advanced fibrosis and even cirrhosis in the following years cannot be ruled out. Actually, all the previously relevant studies on this same topic are affected by this same possible bias [v.b.], which can limited only by a fully adjusted statistics, as that applied in our case. Moreover, whatever cut-off is applied for fibrosis in a cross-sectional study, F0 versus F1-4 [
In order to avoid the risk of missing cases of early cirrhosis, noncirrhotic NAFLD patients were included rigorously based on the result of a liver biopsy. Conversely, in agreement with previous studies on NASH cirrhosis and NAFLD-related HCC [
Epidemiological and anthropometric data confirmed that age, BMI, and T2DM are strong risk factors for NAFLD and for its fibrotic evolution. Concerning genetics, as expectable, the most significant intergroup differences were observed for the PNPLA3 variant. A clear trend towards an increased prevalence of the minor T allele when moving from healthy subjects, through noncirrhotic NAFLD patients, to NASH cirrhotics, was observed also for the TM6SF2 SNP, for which the difference between NASH cirrhotics and healthy controls was actually significant, and for the KLF6 variant, for which the difference between these same two populations was only at the limit of significance. Conversely, concerning the other two SNPs, the frequency of the minor allele in the three different study populations was almost identical. No significant differences in genetics were observed between noncirrhotic NAFLD patients with absent/mild and those with advanced fibrosis, but the limited number in this subanalysis should be considered. Only the PNPLA3, TM6SF2, and KLF6 SNPs were then selected to constitute the genetic risk score, participating according to the model which best fitted with their association to the risk of NASH cirrhosis. The genetic risk score was significantly associated with the risk of both noncirrhotic NAFLD with healthy subjects as reference group and with the risk of NASH cirrhosis with noncirrhotic NAFLD as reference. The more impressive results were obtained, to our opinion, when the score was expressed categorically; notably, for example, with respect to a score of 0, after adjusting for age, sex, diabetes, and BMI, a score > 2 was associated with about 20-fold increased risk of NAFLD development in healthy subjects and almost 4-fold increased risk of progression to cirrhosis in NAFLD patients.
The strengths of this study are the following: it is carried-out in a well-characterized NAFLD population from hepatological tertiary centers; it includes a control group to represent genetics in the general population; it is the first evident proof of concept that the more the information from genetics the higher possibility to predict the risk of NAFLD and its progression to cirrhosis. This study has also some limitations: it is a case-control study with a transversal design, carrying the possible biases which have been addressed previously; the inclusion of some variants with respect to others which are known to be associated with NAFLD, such as the rs641738 SNP in MBOAT7 [
In conclusion, the present study suggests that combining different disease-associated variants may probably represent the way for genetics to keep strength and consent in NAFLD diagnostics. Further studies on larger series are eagerly awaited in order to individuate the best set of variants to be included and the most powerful model and also to verify if combining the best genetic model with clinical predictors would lead to further and meaningful advantages.
The authors declare that there are no conflicts of interest regarding the publication of this article.
The authors thank Rita Marini and Barbara Giannetti for their precious technical assistance in laboratory procedures (DNA extraction and genotyping).