Introduction
NAFLD may present as a spectrum of disease from asymptomatic steatosis with or without elevated aminotransferases to cirrhosis with complications of liver failure and hepatocellular carcinoma [
1].
The interaction between genetic background and environmental factors plays an important component in the pathogenesis of NAFLD [
2]. With regard to the genetic factors involved, a single-nucleotide polymorphism occurring in the sequence of the human patatin-like phospholipase domain-containing 3 gene (PNPLA3), known as I148M variant, is one of the most investigated variants as it is associated with increased of developing hepatic steatosis and progressing to advanced fibrosis [
3].
Specifically, the rs738409 C > G variant of PNPLA3 gene and its encoded protein adiponutrin have been associated with a variety of roles, ranging from an acquired lipogenic function to a loss of lipolytic activities [
4]. NAFLD is closely associated with obesity, and due to the increased prevalence of metabolic syndrome among young age groups, NAFLD is to affect a substantial proportion of our population [
5].
The high prevalence of NAFLD and its low risk of progression to steatohepatitis make a large-scale screening strategy for NAFLD unfitting for the general population [
6]. Thus, focusing our screening on individuals at higher risk of progressing to more advanced liver disease is more cost-effective. Patients with prediabetes fall in this category as they are not only at high risk of developing NAFLD [
7], but also show a higher rate of progression to significant hepatic fibrosis relative to other patients with NAFLD [
8]. Furthermore, as the prevalence of NAFLD in prediabetic patients remains uncertain and effective screening strategies in this patient population are lacking, we observe a higher frequency of non-referrals, especially when transaminases are deceivingly within a normal range.
The way of managing patients with suspected liver disease has been recently modified by the advent of transient elastography (TE) [
9]. This methodology accurately quantifies liver fat by measuring the so-called controlled attenuation parameter (CAP), which identifies steatosis independently from the presence of fibrosis [
10]. Thus, the use of TE with CAP measurement in prediabetes may allow us to risk stratify these patients, to assess their progression of NAFLD and to evaluate the possible reversibility of their liver disease in the context of specific intervention strategies.
Lifestyle modifications such as dietary changes and increase in physical activity have been previously documented as interventions for control of NAFLD [
11,
12]. Exercise contributes to NASH regression by reducing insulin resistance, decreasing the synthesis of free fatty acids (FFAs) and their storage into triglycerides [
12]. Insulin sensitizers like metformin have shown improvement of liver enzymatic factors but without any effect of liver histology [
13]. Similarly, therapy with herbal drugs like silymarin, a complex of flowers and leaves of
Silybum marianum (milk thistle), has shown improvement of transaminases in the context of liver disease [
14,
15]. Nevertheless, the combined effect of nutraceuticals and diet on liver fibrosis and its possible regression in prediabetic patients remains understudied.
With these considerations in mind, we herein propose to investigate which specific genetic factors and patients’ demographics can be effectively utilized as critical variables in the risk stratification analysis of prediabetic patients with NAFLD. Finally, by using TE with CAP measurement, we specifically evaluate whether prediabetic patients who undergo lifestyle changes and use of silymarin have a significant regression of fibrosis relative to their initial baseline.
Methods
Patients
First, a case-control study was conducted; in this initial phase, we have recruited and enrolled 212 patients in prediabetic state (HbA1c 5.7–6.4% or 39–46 mmol/mol) who had not yet received pharmacological treatment and 126 healthy controls, all with BMI ranging from 25 to 29. All patients underwent clinical assessment, transient elastography with measure of liver stiffness (LS) and controlled attenuation parameter (CAP).
At the time of enrolment, a sample of blood was taken for biochemical evaluations and patients were registered for age, sex and BMI. The inclusion criteria were as follows: age ≥ 18 years and prediabetic subjects who had never been treated for more than 12 weeks with antidiabetic drugs or with diet and exercise.
The exclusion criteria were as follows: positive serology for hepatitis C virus (HCV) or hepatitis B virus (HBV); concomitance of other causes of chronic liver disease (autoimmune hepatitis, hemochromatosis, cholestatic liver disease, drug-induced damage); presence of focal hepatic lesions of suspected malignant origin; excessive consumption of alcoholic beverages (≥3 drinks per day); obesity (BMI ≥ 30 kg/m2); inability to obtain valid TE measurements.
Total cholesterol and HDL; LDL cholesterol and triglycerides were measured by enzymatic methods. Glycosylated haemoglobin (HbA1c) was analysed by high pressure chromatography. Plasma glucose concentration was measured by hexokinase (ADVIA, Siemens Healthcare, Leverkusen, Germany). Aspartate aminotransferase (AST), alanine aminotransferase (ALT) and gamma-glutamyl transferase (GGT) were measured by enzymatic methods (Advia 1800 Chemistry System, Siemens Healthcare Diagnostics, Leverkusen, Germany).
Randomized intervention study
After the case-control study phase, pre-diabetic patients were asked to participate in this open label, prospective randomized interventional study in accordance with the ethical guidelines of the Helsinki Declaration of 1975 and after approval by the local Ethics Committee (IRB n. 176/18). All consecutive adult patients with prediabetic state (HbA1c from 5.7 to 6.4%), not under pharmacological treatment, evaluated at the Diabetes Clinic of the Castelli Hospital in Verbania and at the Hepatology Clinic of the Madonna del Popolo Hospital in Omegna, in Northern Italy, between March 2017 and March 2018, were offered to participate to this study, according to the protocol approved. A written informed consent was obtained from all eligible patients (n = 212, of whom n = 157 were males) before participation. All were subjected to full history and clinical examination to assess history of hypertension, history of drug use, any chronic illness, and family history of diabetes.
All patients met, at least once, with a dietician for nutritional guidance, and were encouraged to start and maintain a low-calorie diet before study enrollment. The moderate-fat, restricted-calorie, Mediterranean diet was rich in vegetables and low in red meat, with poultry and fish replacing beef and lamb. We restricted energy intake to 1500 kcal per day for women and 1800 kcal per day for men.
At enrollment, 210 mg BID of silymarin (a flavonolignan complex isolated from Silybum marianum and Morus alba [Meda Pharma]) were added to the regimen of 108 patients (group 2) and supplementation was continued for the entirety of the study (6 months). At each visit, we enquired from the participants about any adverse events and medication compliance.
NAFLD evaluation
Abdominal ultrasonography was performed after a fasting period of 12 h, by a radiologist who was blinded about the patients’ information. All patients underwent ultrasound examination of the liver; a sagittal sonographic plane of the section demonstrating the hepatic parenchyma and right kidney echogenicity was used for determination of liver parenchyma echogenicity [
16]. With the same kidney cortex and liver parenchyma echogenicity it is evaluated as normal, no fatty liver (Grade 0). The liver steatosis was classified into 3 grades, as follows: Mild (Grade 1) - minimal diffuse increase in hepatic echogenicity, normal visualization of the diaphragm and intrahepatic vessels; Moderate (Grade 2) - medium diffuse increase in hepatic echogenicity, impaired visualization of intrahepatic vessels and diaphragm; Severe (Grade 3) - severe diffuse increase in hepatic echogenicity. Posterior segment of the right hepatic lobe is difficult to display. Failure to visualize the walls of the intrahepatic vessels and a marked decrease in the reflectivity of the hemidiaphragm.
LS was evaluated by TE (FibroScan®, Echosens, Paris, France). Measurements of LS and CAP were performed on the same day as abdominal ultrasonography after fasting for at least 12 h. The LS measurements from TE were performed on the right lobe of the liver through the intercostal space of patients in the position of dorsal decubitus with the right arm in maximum abduction.
TE was performed by a blind expert technician on patient clinical data. The median value of the successful measurements was selected as representative of the LS and CA
P values of the subjects. The CAP measured the ultrasonic attenuation at 3.5 MHz using the signals acquired by TE [
17]. As an indicator of variability, the IQR ratios of the LS and CAP values with respect to the median values (IQR/M and IQR/MCAP, respectively) were calculated. In this study, only procedures with at least ten valid measurements, a success rate of at least 60% and an IQR/M value of LS less than 0.3 were considered reliable and used for the statistical analysis.
To correct variability in the general population, LS values in prediabetic patients were compared to those observed in 126 healthy blood donors, (
n = 94 males, 75%;
p = 0.861). To define the presence of significant fibrosis, we used the cut-off value of 7.9 kPa, as proposed by others [
18]. Furthermore, in prediabetics we measured the CAP, to classify hepatic steatosis by measuring the degree of US attenuation by liver fat [
19]. CAP values range from 100 to 400 dB/m: the cut-off values we have chosen to indicate steatosis as absent, mild, moderate and severe were < 236 dB/m, ≥236 dB/m, ≥270 dB/m and ≥ 302 dB/m, respectively [
20]. Finally, the NAFLD and FIB-4 fibrosis score were calculated for each patient, as previously reported [
21,
22].
Genetic studies
Genomic DNA was extracted from whole blood or 100 μL buffy coat, using a commercial kit (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions. For each subject, 20 ng of genomic DNA was used for PNPLA3 rs738409 allelic discrimination using TaqMan SNP Genotyping Assays (Life Technologies, Carlsbad, CA, USA) on the Applied Biosystems 7900HT Fast Real-Time PCR System (Life Technologies).
The DNA was then amplified by the polymerase chain reaction (PCR). 1 μL of DNA was added to 9 μL of the master PCR mix and specific primers were used to amplify the PNPLA3. The reaction protocol consisted of: i) a denaturation cycle at 95 °C for 3 min; ii) 35 cycles at 95 °C for 30 s, annealing at 62 °C for 30 s, extension at 72 °C for 30 s iii) extension at 72 °C for 10 min in a 96-well thermal cycler (Applied Biosystems 2720, ThermoFisher Scientific, Waltham, Massachusetts, United States). At the end of the reaction, the integrity of DNA extracted was assessed by gel electrophoresis. The bands were displayed under UV using the image system, interfaced with the Quantity One program. To define the genotype of the target genes, we performed a PCR of the restriction fragment length. We used NLA-III restriction enzymes for the PNPLA3 digest. All samples were amplified twice, discordant ones were performed a third time. Random samples were confirmed by direct genotyping, which provided concordant results in all cases; controls were included in all analyzed batches, and quality controls were used to verify the reproducibility of the results. Valid genotyping data were obtained for more than 99% of the analyzed subjects.
Statistical analysis
Statistical data analysis was performed with the Stata statistical software package, version 13.1 (StataCorp LP, College Station, Texas, USA) and IBM SPSS® Statistics. The centrality and dispersion measurements of the data were either means, medians or interquartile intervals. Differences between control and prediabetics groups were assessed with Independent t-Test for continues variable and Mann-Whitney U test for ordinal variables. Paired Sample T test was utilized to analyze mean differences for continuous variables while Wilcoxon Signed-Rank Test was utilized to analyze mean differences for categorical variables. 1-wayANCOVA and Mann-Whitney U Test were run to determine the difference between the treatment with only Mediterranean diet versus silymarin plus Mediterranean diet after controlling for pre-intervention (baseline) biochemical parameters. To study strength and direction of association between categorical and continuous variables the nonparametric kendall’s (τb) tau-b correlation was run; this test was preferred over the Pearson chi-square test as the assumptions of either linearity or normality were not met. A chi-square test was used to study the association between the categorical variables PNPLA3 genotypes (CC or CG/GG) and LS (<7.9Kpa or > 7.9 Kpa); Phi (φ) and Cramer’s V measures were used to define the strength of the association.
The association of hypothetical predictors with hepatic rigidity was evaluated with a multiple linear regression model. A binomial logistic regression was also performed to ascertain the effects of gender, BMI, ALT, AST, GGT, HbA1c, cholesterol on the likelihood that participants have significant liver fibrosis defined as liver stiffness ≥7.9 kPa in prediabetic patients. Linearity of the continuous variables with respect to the logit of the dependent variable was assessed via the Box-Tidwell procedure. The significance level chosen for all statistical tests was 0.05 (two-tailed).
Discussion
NAFLD is a slowly progressive disease, where the rate of progression is influenced by the stage at initial diagnosis, as observed in patients with non-alcoholic steatohepatitis [
23]. The prevalence of NAFLD is higher among diabetics than in the general population [
24]. Furthermore, patients with prediabetes are more likely to have more severe histological forms of NAFLD [
25]. In patients with prediabetes and type 2 diabetes, liver lipogenesis is elevated while fatty acid oxidation remains decreased; at the same time, peripheral insulin resistance increases fatty acid release from adipose tissue and promotes their hepatic uptake. This milieu favors the progressive development of NAFLD in patients with prediabetes, and increases the likelihood of NAFLD to progress to NASH and cirrhosis [
26,
27]. These results support the idea that in prediabetic patients, early diagnosis and treatment of NAFLD is paramount.
Our study shows that in an unselected cohort of patients with prediabetes, not previously known to have NAFLD, the prevalence of liver disease is unexpectedly high. The clinical characteristics of the recruited patients explain most of the variability observed, with a significant contribution offered by the genetic profile. As shown in Table
4, the age and the duration of the prediabetic condition do not contribute significantly to the prevalence of NAFLD. Yet, as previously shown, we also observe that excess body weight facilitates progression of liver fibrosis (Tables
4,
5) [
28].
The reason obesity affects the pathophysiology of liver damage is likely related to the fact that leptin induces pro-inflammatory and angiogenic cytokines, thus altering the liver wound-healing response [
29‐
31].
In recent years, in addition to environmental factors and comorbidities, genetic predictors of progression of liver disease have also been identified [
32]. The strongest genetic predictor of progression of chronic liver disease is SNS rs738409 in the PNLPA3 gene, which encodes a membrane-bound triacylglycerol lipase mediating the hydrolysis of the triacylglycerol. This “missense” mutation causes a reduction in the enzymatic activity of PNPLA3, leading to the development of macrovascular steatosis [
33,
34]. In our study, we evaluate how the PNPLA3 mutation plays a part in the progression of liver disease in prediabetes. Increased risk of hepatic fibrosis in patients with DMT2 who are carriers for the mutated variant of rs738409 was previously reported [
35,
36]; yet, the estimate of liver fibrosis was based on Fibrotest rather than on liver stiffness. Herein, we show the existence of significantly increased liver stiffness in mutated heterozygous and homozygotes compared to wild-type PNPLA3 in prediabetic patients (Table
3). We can thus infer that in prediabetic patients, PNLPA3 contributes to the progression of liver disease; nevertheless, in prediabetics, unlike in the general population, the net effect may be driven by excess BMI. In fact, our results indicate that of the 22 patients who are homozygous rs738409 G/G, 13 of them had LS > 7.9 kPa and of those 13 patients all of them had BMI > 28. Instead, in the 9 patients homozygous for rs738409 G/G who had LS < 7.9 kPa only 2 of them (22%) had a BMI > 28. This observation suggests that elevated BMI index may synergistically modulate fibrosis progression in specific variants of PNLPA3. Similarly, increase in BMI augments the effect of M variant on the prevalence of cirrhosis [
37]. Therefore, a future study should be conducted on patients with BMI greater >30 to further analyze the significance of this interaction between PNPLA3 gene and BMI in the contest of prediabetes and progression of NAFLD to significant fibrosis.
Though genetic variants may be considered in the construction of prediction models aimed to detect the progression of NAFLD to fibrosis in patients with prediabetes, the use of Fibroscan in this patient population remains paramount. The prevalence of moderate to advanced steatosis scored using CAP, independently of BMI, does not significantly differ between the control group and patients with prediabetes (see Table
1, 52% vs 58%,
p = 0.3); nevertheless, when using LS as a measure of liver fibrosis, there appears to be a significant difference in advanced steatosis when comparing these two groups (Table
1, 3% vs 29%,
p < 0.001), thus indicating that prediabetic state plays a critical factor in the progression of steatosis to fibrosis. The role of TE should be particularly underscored in that group of prediabetic patients with mild fibrosis (LS 5.9–7.9 kPa) as transaminase levels are not elevated in this patient group (see Table
4); this indicates that transaminases levels are not a good indicator of mild fibrosis and can lead to a sense of false reassurance in this group of patients. Nonetheless, we show that transaminases, together with BMI, remain the principle indicators associated with significant fibrosis (Tables
4 and
5).
As aforementioned, lifestyle modifications are recommended for control of NAFLD [
11,
12]. It has been demonstrated that an intensive weight loss program led to resolution of NASH and improvement of fibrosis [
38]. Moreover, follow-up studies on morbidly obese patients that underwent bariatric surgery have also showed that the extent of weight loss correlates with the degree of NASH and fibrosis [
39]. We herein show that prediabetic patients who follow a Mediterranean diet and achieve weight loss greater than 5% experienced a significant regression of fibrosis after 6 months (mean LS is 6.70 vs. 5.90;
p < 0.001, Table
6). In recent years, the use of natural products has been considered for the treatment of hyperglycemia. Furthermore, the protective effects of silymarin as a powerful ROS scavenger, and its hepato-protective effects via attenuation of pro-inflammatory gene expression in the liver have been discussed previously [
40]. There is also recent experimental evidence that in diabetic rats treated with silymarin and in which weight reduction was achieved, markers of hepatic injury were significantly reduced and favorable histopathological changes were showed [
41]. Randomized control trials have shown that regimens with either silymarin alone or its most active biological compound silybin lead to liver aminotransferase reduction in patients with NAFLD after only 3 months [
42‐
44]. One of these studies also investigated the role of silymarin therapy in the context of PNPLA3 polymorphism and showed that PNPLA3 G-allele carriers did not receive any benefits from treatment with silymarin. Furthermore, silymarin together with lifestyle modification has an increased safety profile relative to other anti-oxidant inhibitors or anti-fibrotic agents while conserving liver-enzyme ameliorating effects in patients with NAFLD [
45,
46]. A systematic review of clinical trials conducted between 2003 and 2008 did not report any significant side effects attributable to silymarin therapy compared to the placebo-treated population and also did not show any significant interactions with other drugs [
47]. Given its safety profile, silymarin continues to be widely used in the treatment of acute or chronic hepatitis. Yet, no study has shown regression of liver fibrosis with silymarin therapy upon assessment with Fibroscan. We thus investigated whether in patients with hepatic steato-fibrosis and prediabetes the reduction of body weight > 5% and the intake of silymarin reduces progression of comorbid liver disease. We show that Mediterranean diet either alone or in conjunction with silymarin regimen leads to a significant reduction of liver fibrosis, normalization of ALT and significant reduction in HbA1c and FBS. Nevertheless, after controlling for pre-intervention (baseline) biochemical parameters, there is no difference in fibrosis regression between the treatment with only Mediterranean diet versus silymarin plus Mediterranean diet (
p = 0.696). Finally, it should be underscored that both Mediterranean diet with weight reduction greater than 5% alone and in conjunction with silymarin treatment lead to significant reduction of HbA1c (
p < 0.001, Table
6). Interestingly, diet with weight reduction greater than 5% together with silymarin regimen leads to a statistically significant reduction in HbA1c when compared to diet alone (95% CI: 37.8–38.6 vs 39.5–40.3, p < 0.001, Table
6). Interestingly, Li et al. showed the role of silybin in the prevention of palmitate-induced insulin resistance and inhibition of the IRS-1/PI3K/Akt pathway in skeletal muscle cells [
48]. These observations imply that silymarin may play an important role in glycemic control and it may be of interest to investigate its anti-glycemic properties in isolation in future studies. Among the various limitations of this study, we must note the fact that Fibroscan is not the best method to characterize NAFLD. However, compared to the magnetic resonance evaluation, the evaluation with TE is much more practicable and at lower cost in clinical practice.
In conclusion, it is critical to acknowledge that patients with prediabetic status have a high prevalence of liver disease characterized by significant fibrosis, which goes largely unnoticed in the absence of symptoms and laboratory abnormalities. We herein show how increased BMI and genetic factors represent critical variables in the risk stratification analysis of prediabetic patients with NAFLD. Lastly, we propose that Mediterranean diet together with intake of the nutraceutical silymarin contributes to the regression of liver damage and diabetes markers.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.