Serum MicroRNAs as Potential Biomarkers for Early Diagnosis of Hepatitis C Virus-Related Hepatocellular Carcinoma in Egyptian Patients

Circulating microRNAs are deregulated in liver fibrosis and hepatocellular carcinoma (HCC) and are candidate biomarkers. This study investigated the potential of serum microRNAs; miR-19a, miR-296, miR-130a, miR-195, miR-192, miR-34a, and miR-146a as early diagnostic biomarkers for hepatitis C virus (HCV)-related HCC. As how these microRNAs change during liver fibrosis progression is not clear, we explored their serum levels during fibrosis progression in HCV-associated chronic liver disease (CLD) and if they could serve as non-invasive biomarkers for fibrosis progression to HCC. 112 Egyptian HCV-HCC patients, 125 non-malignant HCV-CLD patients, and 42 healthy controls were included. CLD patients were subdivided according to Metavir fibrosis-scoring. Serum microRNAs were measured by qRT-PCR custom array. Serum microRNAs were deregulated in HCC versus controls, and except miR-130a, they were differentially expressed between HCC and CLD or late fibrosis (F3-F4) subgroup. Serum microRNAs were not significantly different between individual fibrosis-stages or between F1-F2 (early/moderate fibrosis) and F3-F4. Only miR-19a was significantly downregulated from liver fibrosis (F1-F3) to cirrhosis (F4) to HCC. Individual microRNAs discriminated HCC from controls, and except miR-130a, they distinguished HCC from CLD or F3-F4 patients by receiver-operating-characteristic analysis. Multivariate logistic analysis revealed a panel of four microRNAs (miR-19a, miR-195, miR-192, and miR-146a) with high diagnostic accuracy for HCC (AUC = 0.946). The microRNA panel also discriminated HCC from controls (AUC = 0.949), CLD (AUC = 0.945), and F3-F4 (AUC = 0.955). Studied microRNAs were positively correlated in HCC group. miR-19a and miR-34a were correlated with portal vein thrombosis and HCC staging scores, respectively. In conclusion, studied microRNAs, but not miR-130a, could serve as potential early biomarkers for HCC in high-risk groups, with miR-19a as a biomarker for liver fibrosis progression to cirrhosis to HCC. We identified a panel of four serum microRNAs with high accuracy in HCC diagnosis. Additional studies are required to confirm this panel and test its prognostic significance.


Introduction
Ambion, the RNA Company-one step RT-PCR Kit, USA). All HCC patients were on top of HCV cirrhosis and HCC diagnosis was made upon the presence of hepatic focal lesions diagnosed by abdominal ultrasound and confirmed by computed tomography (CT) and/or magnetic resonance imaging according to European Association of the Study of the Liver (EASL) guidelines (2012). Liver disease severity in HCC patients was assessed by Child-Puph grade and HCC staging was done using Barcelona Clinic Liver Cancer (BCLC) staging system [15].
Diagnosis of HCV-associated CLD was based on standard clinical, biochemical, serological, and ultrasonographic criteria, as well as the histopathological data obtained at liver biopsy. Metavir scoring was used to stage fibrosis on a five-point scale: F0, no fibrosis; F1, portal fibrosis alone; F2, portal fibrosis with rare septae; F3, portal fibrosis with many septae (bridging fibrosis); and F4, cirrhosis [16]. All CLD patients were of significant fibrosis from F1 to F4, with no patients with F0 stage. CLD patients were further subdivided according to their fibrosis stage into 2 subgroups; early fibrosis (mild/moderate fibrosis; F1-F2, n = 75) and late fibrosis (severe fibrosis/cirrhosis; F3-F4, n = 50). F4 patients were Child A (n = 24) and Child B (n = 8). History taking, clinical examination, and routine laboratory investigations were made for every patient. Patients with chronic hepatitis B virus (HBV) or any other identifiable cause for chronic hepatitis other than HCV, any associated malignancies other than HCC, or who received previous treatment for HCC or antiviral therapy for HCV were excluded.
In addition, a total of 42 gender (males, n = 28 and females, n = 14) and age (mean±SD, 44±12.5) matched healthy volunteers were evaluated as controls; all have normal liver function tests, normal AFP levels (<10 ng/ml), normal hepatic ultrasound, and negative for hepatitis B surface antigen (HBsAg), hepatitis B core antibodies (HBc-Ab), and HCV RNA by PCR. Written informed consent was obtained from all patients and controls for gene analysis. The study protocol and informed consent were approved by the ethics committee of the Faculty of Pharmacy, Cairo University and conformed to the ethical guidelines of the Helsinki Declaration.

Laboratory tests
Fasting venous blood samples were collected from all patients for routine workup, including complete blood picture, liver function tests, prothrombin concentration and prothrombininternational normalized ratio, AFP, anti-HCV titer, HBsAg, and HBc-Ab using commercially available assays.

Rationale of miRNAs selection and serum miRNAs assay by qRT-PCR
We assayed serum samples for 7 miRNA candidates (miR-19a, miR-296, miR-130a, miR-195, miR-192, miR-34a, and miR-146a) having a known association with HCC. Each of these miR-NAs have been linked mechanistically to HCC or have predicted oncogenic or tumor suppressor target genes [6,[17][18][19][20]. miR-19a, a member of the angiogenic MYC-induced 17-92 miRNA cluster [20], targets PTEN, SOX4 [21], and the proapoptotic APAF1, Beclin1, and PAK6 [22]. miR-296 was linked to tumor angiogensis by regulating vascular endothelial growth factor (VEGF) receptors overexpression [23] and may also contribute to carcinogenesis by dysregulating p53 [24]. miR-130a was predicted to target genes involved in mitotic cell cycle, protein modification process, and cell differentiation [19]. miR-34a and miR-192 are p53-responsive tumor suppressor miRNAs that induce apoptosis and cell cycle arrest [6,18]. miR-34a was also predicted to target oncogenes such as PDGFRA, E2F5, Foxp1, c-Met, Wnt1, and Notch1 [6]. miR-195 is a tumor suppressor by regulating the G1/S transition in HCC [17] and targeting the oncogenic fibroblast growth factor 7 (FGF7) and growth hormone receptor (GHR) [6]. miR-146a was reported as a tumor suppressor and a negative regulator of toll-like receptor signaling and NF-κB via TRAF6 and IRAK-1 [25]. Importantly, miR-296 [26], miR-130a, miR-146a, miR-195 [27,28], and miR-34a [29][30][31] were deregulated in HCC tissues versus healthy livers. However, the invasive nature of liver biopsy may limit the use of these miRNAs as HCC biomarkers. In addition, compared with normal, we have previously demonstrated that serum levels of these 7 miRNAs were deregulated in chronic HCV patients; these miRNAs were interferon-related, and five of them were associated with the outcome of standard interferon therapy [32]. Moreover, serum miR-19a and miR-192 were also proposed as biomarkers for HBVassociated HCC [33]. Therefore, we hypothesized that these 7 serum miRNAs would represent a broad variety of functions in HCC and had high potential for release into the circulation. However, the clinical relevance of using these miRNAs as potential non-invasive serum biomarkers for HCV-related HCC detection is not yet known.
RNA extraction and reverse transcription. Total RNA, including miRNAs was extracted by miRNeasy extraction kit (Qiagen, Valencia, CA) using QIAzol lysis reagent according to the manufacturer's instructions. The quality of RNA was determined using NanoDrop2000 (Thermo scientific, USA). Reverse transcription (RT) was carried out on 100 ng of total RNA in a final volume 20 μl RT reactions (incubated for 60 min at 37°C and 5 min at 95°C) using miScript II RT Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions.
Quantitative real-time PCR. Serum expression levels of mature miRNAs, hsa-miR-19a-3p, hsa-miR-296-5p, hsa-miR-130a-3p, hsa-miR-195-5p, hsa-miR-192-5p, hsa-miR-34a-5p, and hsa-miR-146a-5p was evaluated using miScript miRNA PCR custom array (Qiagen, Valencia, CA) according to the manufacturer's protocol. The housekeeping miScript PCR control, miRNA SNORD68 was used as internal control, as earliarly reported [34,35]. Our previous experience also suggests that SNORD68 is a stable normalization control and possible for use in miRNA PCR analysis [32]. For real-time PCR of each miRNA, 2.5 μl diluted RT products was mixed with 5.5 μl RNase-free water, 10 μl QuantiTect SYBR Green PCR Master Mix and 2 μl miScript universal primer (reverse primer), and then added to a custom Rotor-Disc 100 miRNA PCR array which contains miRNA-specific miScript primer assays (Qiagen, Valencia, CA). The Rotor-Disc was sealed with optical thin wall strips. Real-time PCR was performed using Rotor gene Q Real-Time PCR System (Qiagen, Valencia, CA) with the following conditions: 95°C for 30 min, followed by 40 cycles at 94°C for 15 s, 55°C for 30 s, and 70°C for 30 s. The cycle threshold (Ct) is the number of cycles required for the fluorescent signal to cross the threshold in real-time PCR. Fold change of miRNA expression levels was calculated using the formula 2 -ΔΔCt using healthy controls as calibrator, where ΔΔCt = [Ct (target, test)-Ct (reference, test)]-[Ct (target, calibrator)-Ct (reference, calibrator)] [36].

Statistical analysis
Values were expressed as mean±standard deviation (SD), median (25%-75% percentiles) or number (percentage) when appropriate. Clinical data of three independent groups were compared using one way ANOVA and Tukey Kramer's multiple comparisons test. AFP levels were log transformed to enable parametric statistical tests. Categorical data were compared by Chi square (X 2 ) test. The non-parametric Mann-Whitney U-test was used for comparison of miRNA data (2 -ΔΔCt ) from independent samples from 2 groups as miRNA data were not normally distributed. Serum miRNAs in individual fibrosis stages (F1 to F4) were compared by Kruskal-Wallis test. The diagnostic accuracy of miRNAs was evaluated by receiver-operatingcharacteristic (ROC) analysis and the area under the curve (AUC) was calculated. The end point is that miRNA could discriminate HCC from late fibrosis (F3-F4) subgroup. Logistic regression analysis was done to identify predictor miRNAs associated with the risk of HCC. Univariate analysis was conducted to determine the relationship between expression level of each miRNA and presence/absence of HCC (HCC vs. CLD + healthy controls). Significant predictor miRNAs in the univariate analysis were included in a stepwise forward multivariate analysis (P<0.05 for entering the model and P<0.1 for removal from the model) to determine the final predictor miRNAs for the probability of being diagnosed with HCC. The predicted probability of being diagnosed with HCC was used as a surrogate marker to construct the ROC curve. AUC was used as an accuracy index for evaluating the diagnostic performance of the selected miRNA panel. Correlations between parameters were determined by Spearman or Pearson correlation when appropriate. P<0.05 was considered significant, with a 95% confidence interval (CI). Statistical analyses were performed using computer program Statistical Package for the Social Science (SPSS, Chicago, IL) software version-15 for Microsoft Windows and GraphPad Prism-5.0 (GraphPad Software, CA).

Demographic and clinical features of HCC and non-malignant CLD patients
The clinical features of HCC and non-malignant CLD patients are shown in Table 1. Studied patients showed a significant trend of elder age with progression of liver disease from F1-F2 to F3-F4 fibrosis to HCC (P<0.0001). Gender was not significantly different between studied groups (P = 0.79). However, there was a male predominance in HCV-related liver disease patients in the three groups representing 76%, 76%, and 82.14% in F1-F2, F3-F4, and HCC groups, respectively. Serum levels of AST, ALP, AFP, total bilirubin (P<0.0001, for each), and direct bilirubin (P = 0.0009) were significantly higher, whereas hemoglobin levels were significantly lower (P<0.0001) in HCC patients versus the other two groups. Serum ALT was only significantly different between HCC and F1-F2 group (P = 0.001). Hepatic synthetic functions; albumin and prothrombin concentration tended to decrease significantly (P = 0.0001, P<0.0001, respectively) during liver disease progression among studied groups. HCC and F3-F4 patients showed significantly lower platelet count versus F1-F2 patients. Total leucocyte count was not significantly different between studied groups (P = 0.46).

Clinicopathological characteristics of HCC patients
The clinicopathological characteristics of HCC patients are shown in Table 2. HCC patients produced a wide range of AFP values from normal to 46870 ng/ml (mean±SD, 4405±10470 ng/ml). Normal AFP levels (up to 10 ng/ml) are present in as many as 14% of patients, with 75% having AFP >20 ng/ml and 46% with AFP>400 ng/ml at time of diagnosis. 92.86% of HCC patients had Child-Pugh grade A and B, 75% had performance status 0, and 71.43% had stage B in BCLC staging. Regarding CT imaging, hepatic focal lesions were single in 50% of patients, arising from the right lobe (64.3%), 57.15% were of size >5cm, with 21.43% of patients having portal vein (PV) thrombosis.  1). Comparing to CLD, miR-192, miR-34a, and miR-146a showed significant fold increase (P = 0.004, P<0.0001, and P = 0.002, respectively), whereas miR-19a, miR-296, and miR-195 showed significant fold decrease (P<0.0001, P = 0.024, and P<0.0001, respectively) in HCC. miR-130a relative expression was not significantly different between HCC and CLD (P>0.05) (Fig 2).

Signature of serum miRNAs during HCV-related liver disease progression
In a further analysis, we investigated serum miRNA profiles during the progression signature from F1-F2 to F3-F4 to HCC. We examined if serum miRNA levels would change from F1-F2 to F3-F4 or from F3-F4 to HCC using Mann-Whitney U-test. We found no significant difference in miRNA levels between F1-F2 and F3-F4, and except miR-130a, all studied miRNAs were significantly different when HCC and F3-F4 were compared (P<0.05) (Fig 3A). In a detailed analysis, we investigated serum miRNA levels in patients with individual fibrosis stages (F1 to F4) using the non-parametric Kruskal-Wallis test, with no significant difference was obtained (P>0.05) (Fig 3B). In a sub-analysis, we examined if studied miRNAs could be deregulated during progression from F1-F3 (liver fibrosis) to F4 (cirrhosis) or from F4 to HCC using Mann-Whitney U-test (Fig 3B). We found that miR-19a was significantly downregulated in F4 versus F1-F3 (P = 0.033). Moreover, miR-19a was further downregulated in HCC versus F4 (P = 0.04). In addition, miR-296, miR-192, and miR-146a levels were significantly different between HCC and F4 (P<0.05), however, they were not differentially expressed between F1-F3 versus F4. On the other hand, miR-195 and miR-34a were not significantly different neither for F1-F3 versus F4 nor for HCC versus F4 (P>0.05).

Logistic regression analysis of miRNAs
Univariate and multivariate logistic regression analyses were performed to select the predictor miRNAs associated with HCV-related HCC diagnosis (Table 4). Expression levels of miR-19a, miR-195, miR-192, and miR-146a were selected as significant predictors associated with the chances of HCC diagnosis in the univariate analysis. In a stepwise forward multivariate analysis, these four miRNAs turned out to be significant predictors of the risk of being diagnosed with HCC. The predicted probability of being diagnosed with HCC from the logit model based on the 4-miRNA panel (Table 4), Logit(P) = -0.24-1.114 Ã mir-19a-0.077 Ã miR-195+0.571 Ã miR-192+0.048 Ã miR-146a, was used to construct the ROC curve.

Discussion
Early diagnosis of HCC presents a challenge due to lack of reliable biomarkers, hence stressing the need for new early diagnostic tools. The present study revealed that all studied miRNAs were differentially expressed in serum of HCV-related HCC patients versus controls, and except miR-130a, they were differentially expressed between HCC and non-malignant CLD, implicating these miRNAs as surrogate biomarkers of HCC. In addition, studied miRNAs distinguished HCC from controls, and except miR-130a, they distinguished HCC from CLD, and in particular from late fibrosis subgroup. Multivariate analysis revealed a serum miRNA panel of miR-19a, miR-146a, miR-195, and miR-192 with high diagnostic accuracy for HCC diagnosis. Moreover, the miRNA panel demonstrated significantly higher diagnostic accuracy than AFP in late fibrosis patients, which at much high-risk to develop HCV-related HCC. However, this panel has higher diagnostic accuracy than previously reported for AFP (sensitivity:39-65%; specificity: 76-94%) [9]. Nearly one third of early HCC patients are missed by AFP analysis and it is also elevated in patients with chronic hepatitis and cirrhosis [9]. These results implicate these miRNAs as reliable early biomarkers and possible therapeutic tools or targets for HCC treatment. Perhaps addition of this miRNA panel to serological tumor markers may improve the diagnostic accuracy for early HCC detection.
The current study revealed that the tested miRNAs were positively correlated in HCC group, suggesting their concordant deregulation in HCC. We have previously shown that serum expression levels of these miRNAs were correlated in chronic HCV patients and perhaps implicated in HCV pathogenesis [32]. Together, these results may relate at least some of the investigated miRNAs to the hepatocarcinogenic effect of HCV, which need further investigation. Notably, HCV proteins were shown to affect apoptosis and tumor cell behavior by altering miRNAs expression in HCC cells expressing full length-HCV [8]. Serum miR-19a and miR-195 were downregulated in HCC versus controls or CLD patients. Similarly, serum miR-19a was downregulated in HCC versus chronic HBV [33]. miR-195 was also downregulated in serum of HCC patients [37] and in HCC tissues versus healthy livers [27,28,38]. Downregulation of miR-195 promotes carcinogenesis by targeting FGF7 and GHR, both induce evading apoptosis, tissue invasion, and metastasis [6]. The present study also demonstrated that higher levels of serum miR-19a were associated with PV thrombosis in HCC patients, implying a role of miR-19a in HCC invasion and progression. miR-19a was also associated with HCC satellite nodules [39], HCC recurrence and predicted survival after liver transplantation [21]. Together, these results may emphasize the prognostic value of miR-19a in HCC.
In the current study, serum levels of 2 out of 6 studied miRNAs were matched with hepatic tissue levels in HCC versus controls [27][28][29]31], however, to the best of our knowledge, miR-19a levels in HCC tissues versus normal liver are not known. Circulating and tissue miRNA Serum MicroRNAs as Early Biomarkers of HCV-HCC levels are not always consistent; miR-122 was downregulated in HCC tissues and cancer cell lines [50], but upregulated in serum of HCC patients [51,52]. This inverse relationship suggests that secreted miRNAs from cells may be an important component of circulating miRNA expression [9]. The liver secretes circulating exosomes during injury with increase in serum miR-122 and miR-192, and a corresponding decrease in their hepatic expression [53]. On the other hand, it was reported that miRNAs that were elevated in the liver during injury displayed decreased levels in serum [54]. The variable levels of miRNAs in HCC between different studies may also reflect the counter mechanisms that regulate their expression. For example, miR-34a is activated by p53 and inactivated by DNA methylation; liver tumors that still retain an active p53 may upregulate miR-34a [31]. miR-146a is downregulated by DNA methylation [46] and upregulated by aberrantly activated STAT3 in HCC cells [49]. The discrepancies between different studies may also arise from variability in technical procedures from sampling to Serum MicroRNAs as Early Biomarkers of HCV-HCC detection method and data analysis, or the use of different normalization controls or control tissues used for normalization (healthy liver or adjacent non-tumor tissue) [55]. HCC etiology should also be considered, for example, HBV-HCC and HCV-HCC exhibited different miRNA dysregulation patterns and distinct molecular mechanisms [19].
Progression towards HCC involves multiple steps and the clinical pathway of most chronic HCV cases comprises fibrosis progression to cirrhosis, and eventually HCC [11]. We demonstrated reduced hepatic synthetic function (albumin and prothrombin) during liver disease progression among studied groups, but neither was correlated with miRNAs. Hepatic miRNA expression is deregulated in liver fibrosis, and thus circulating miRNAs are expected to be affected by fibrosis progression [13]. miR-19a was upregulated in CLD patients versus controls (4.5 fold) and detailed analysis of its levels within CLD group revealed reduced levels during progression from liver fibrosis to cirrhosis, then further downregulation in HCC. Serum miRNA levels may initially rise following release from inflamed hepatocytes in HCV followed by a drop in the levels with fibrosis progression due to hepatocyte loss and accumulation of extracellular matrix [56]. Activation of hepatic stellate cells, a key driver of fibrosis, is also associated with a specific miRNA deregulation regulating various fibrogenic signaling pathways [57]. It seems probable that miR-19a may be implicated in inflammatory and fibrogenic processes in CLD progression, ultimately to HCC. Other studied miRNAs were fibrosis-stage independent and differential expression only occurs from late fibrosis to HCC. Conversely, miR-34a was overexpressed during progression from normal liver through cirrhosis to HCC [58], with serum miR-34a being increased during fibrosis progression in chronic HCV [13]. These conflicting results may render miR-34a an inappropriate marker, when exclusively used for Serum MicroRNAs as Early Biomarkers of HCV-HCC interpretation of fibrosis progression. We demonstrated that serum miR-34a was positively correlated with child stage and BCLC score, a staging system which defines HCC patients for different treatment strategies. This correlation may delineate the usefulness of serum miR-34a in assessing HCC severity, staging, and patient classification for therapeutic interventions.
Previous studies have evaluated circulating miRNAs as HCC biomarkers in comparison to healthy controls and chronic hepatitis or cirrhosis [35,37,40,43,51,59,60], with few miRNAs revealed as candidate biomarkers. Our study revealed a four miRNA set with high accuracy for HCC and could be of clinical value in diagnosis of HCV-related HCC. In addition, most of these studies have addressed HCC of HBV etiology or mixed etiologies, including HCV, with few solely studied HCV-related HCC. Another limitation is that neither study has evaluated circulating miRNAs as potential biomarkers during CLD progression to HCC. Our study is unique as we demonstrated serum miRNA signatures during stepwise progression form liver fibrosis to HCC. We hypothesized that the miRNA that changes significantly from early fibrosis to late fibrosis to HCC would be a biomarker of disease progression. Perhaps our prominent result is that miR-19a levels were negatively associated with fibrosis progression to cirrhosis, finally to HCC. However, our study is limited by the relatively small number of patients and additional studies are required to validate the miRNA panel in an independent larger patient cohort and test its prognostic significance in HCC outcome and the implications of these miR-NAs as therapeutic tools or targets in HCC treatment. Longitudinal samples should be also considered in monitoring liver fibrosis progression, eventually to HCC.

Conclusion
Studied miRNAs, but not miR-130a, could distinguish HCV-related HCC from HCV-associated CLD and in particular, from late fibrosis (F3-F4) subgroup, suggesting the potential usefulness of these miRNAs as early biomarkers of HCC detection in high-risk groups. We identified a serum miRNA panel of miR-19a, miR-146a, miR-192, and miR-195 with high accuracy for HCC and could be of clinical value in diagnosis of HCV-related HCC. Additional studies are needed to confirm this panel in a larger patient cohort and test its prognostic significance. Serum miRNAs were fibrosis-stage independent, however, miR-19a was significantly downregulated during disease progression from liver fibrosis (F1-F3) to cirrhosis (F4) to HCC, suggesting that serum miR-19a monitoring could be of clinical relevance as a potential diagnosis tool of fibrosis progression to HCC. The investigated miRNAs were significantly correlated in HCC group, suggesting their concordant deregulation in HCC. The correlations between miR-19a and miR-34a with PV thrombosis and BCLC staging scores, respectively, may emphasize their prognostic significance in HCC.
Supporting Information S1