Address reprint requests to: Ratna B. Ray, Department of Pathology, Saint Louis University, DRC 207, 1100 South Grand Boulevard, St. Louis, MO 63104. E-mail: firstname.lastname@example.org fax: 314-771-3816
Chronic hepatitis C virus (HCV) infection is one of the major causes of liver fibrosis and liver transplantation in the United States. Circulating microRNAs (miRNAs) in the blood are emerging as biomarkers for pathological conditions. In the present study we performed a systematic screening approach to identify up-regulated miRNAs in the plasma/serum of HCV-infected patients with different stages of hepatic histological disease severity. We initially screened serum samples of HCV-infected patients with fibrosis and compared them with sera of healthy volunteers using serum miRNA array profiling and identified a group of modulated miRNAs. Subsequent study demonstrated that miR-20a and miR-92a in HCV-infected fibrosis patients sera were significantly up-regulated when compared with that of healthy volunteers or non-HCV-associated liver disease. We have also observed an increase of plasma miR-20a and miR-92a in acute and chronic HCV-infected patients as compared to that of healthy volunteers. However, there was no correlation between the plasma/serum levels of any of these miRNAs with HCV viral loads. We next investigated longitudinal plasma samples from HCV-infected patients. Our results suggested that miR-20a and miR-92a remained unaltered in HCV-infected patients who progressed from acute to chronic infection. On the other hand, miR-92a expression was reduced in acute to resolved individuals. These data provide evidence that plasma/serum levels of miR-20a and miR-92a have potential as sensitive and cost-effective biomarkers for early detection of HCV infection. Conclusion: Circulating miR-20a may serve as a potential for predictive biomarker in HCV-mediated fibrosis. (Hepatology 2013;53:863–871)
Chronic hepatitis C virus (HCV) infection-associated liver disease is an important public health problem worldwide. An estimated 200 million people worldwide and 4 million people in the United States are infected with HCV. In the U.S., HCV genotypes 1a and 1b are predominant in patients with chronic infection. Approximately 20% of chronically infected patients develop endstage liver disease. Chronic HCV infection is characterized by high levels of interindividual variation in disease progression. Although several chronically infected individuals never develop cirrhosis, some may develop severe fibrosis. A number of cellular factors, demographic, and clinical characteristics, as well as viral factors, have been associated with the development of HCV-related liver fibrosis.
MicroRNAs (miRNAs) are a class of small, single-stranded noncoding RNA of 22 nucleotides with a characteristic hairpin secondary structure.[5, 6] They regulate gene silencing either by targeting messenger RNA (mRNA) directly into degradation or by inhibiting translation. Aberrant expression of miRNAs has been linked to variety of cancers, including hepatocellular carcinoma.[7, 8]
Several groups have reported the presence of miRNAs in human serum and plasma, called circulating miRNAs.[9, 10] These miRNAs are not affected by endogenous RNases in the blood. In addition, circulating miRNAs display consistent profiles between healthy individuals and significantly altered levels in disease conditions.[11, 12] These characteristics of circulating miRNAs established their potential value as biomarkers for changes in physiological and pathological conditions. For example, miR-25 and miR-223 are shown to be serum biomarkers for lung cancer, miR-184 for squamous cell carcinoma, and miR-92a for leukemia. Circulating miR-122 and miR-155 were identified as inflammation biomarkers in different forms of liver injuries.[16-19] miR-141 and miR-375 were the most promising markers correlated with prostate tumor progression. The circulating miRNAs can also be used to predict the clinical outcomes of nonsmall-cell lung cancer patients.
In our present study, circulating miRNAs, miR-20a and miR-92a, were identified as possible predictive biomarkers for HCV-mediated liver disease. Our data show that an increase in circulating miR-20a correlate with HCV-mediated liver fibrosis severity, which may serve as predictor for liver disease progression. We also observed that miR-20a and miR-92a are up-regulated in acute and chronic stages of HCV infection. To our knowledge this is the first report describing a group of miRNAs up-regulated in HCV infection which could be used as a potential predictive biomarker.
Materials and Methods
Study Design and Patient Samples
Our study was approved by the Saint Louis University and Massachusetts General Hospital Institutional Review Board and written informed consent was obtained from all subjects. A total of 86 sera samples including 44 HCV-infected patients with different stages of fibrosis, 20 non-HCV-associated patients with liver fibrosis, and 22 healthy volunteers were included in this study. The liver fibrosis stage was evaluated according to Batts and Ludwig scoring system in patients with chronic hepatitis C, including 33 (F0-F2) early-stage and 11 (F3-F4) late-stage fibrosis. We included plasma samples from patients with different clinical stages of HCV infection, including acute (4-24 weeks from time of exposure), chronic (33-163 weeks from time of exposure), and resolved samples from Massachusetts General Hospital, Harvard Medical School. All subjects in the control group had normal aminotransferase activities, no history of liver disease or alcohol abuse, and were tested negative for HBV, HCV, and human immunodeficiency virus (HIV) infections. Table 1 shows the characteristics of patients infected with HCV with different clinical stages and those with non-HCV-associated liver disease included in this study.
Table 1. Patient Characteristics
Patients With Acute HCV
Patients With Chronic Hepatitis
Patients With Resolved HCV Infection
Patients With HCV Mediated Fibrosis
Patients With Non-HCV Associated Liver Disease
*Mean ± SD; †Median (range); ‡HCV viral load in log10 ± SEM; N/A, not applicable; ND, not done; UD, undetermined.
10 / 19
9 / 9
1 / 10
29 / 15
13 / 7
HCV viral load (log10 copies / ml)‡
Stage of fibrosis (F0-F2 / F3-F4)
33 / 11
9 / 11
Serum miRNA Expression Profiling
We used the Human Serum and Plasma miScript miRNA PCR (polymerase chain reaction) Array (MIHS-106Z, Qiagen, Chatsworth, CA) that profiles the expression of 84 miRNAs detectable and differentially expressed in serum, plasma, and other bodily fluids. The Human Serum miScript miRNA PCR Array was used for miRNA profiling in serum samples (n = 4 from each group) of healthy volunteers, early-stage (F0-F2), and late-stage (F3-F4) HCV-infected fibrosis patients. In brief, RNA was reverse-transcribed to complementary DNA (cDNA) using the miScript Reverse Transcription kit (Qiagen) according to the manufacturer's instructions. Real-time qPCR was performed using the miScriptSYBR Green PCR kit (Qiagen) with the manufacturer-provided miScript Universal primer. Array data were analyzed using free Web-based software (http://pcrdataanalysis.sabiosciences.com/mirna/arrayanalysis.php) and automatically performed all ΔΔCt fold change calculations.
miRNA-Specific Quantitative Real-Time RT-PCR
Total RNA was isolated from 200 μL of plasma/serum with the miRVana PARIS kit (Ambion, Austin, TX), according to the manufacturer's instructions. Synthetic spiked-in Caenorhabditis elegans miR-39 was added to the plasma/serum and cell culture supernatant samples prior to RNA extraction as an internal control. There is no consensus on the use of housekeeping miRNAs and it was reported that frequently used reference genes like U6 small nuclear RNA (RNU6B) and 5S ribosomal RNA are easily degraded in plasma/serum samples. In addition, a large variation of serum U6 levels was reported in several studies. We used TaqMan qRT-PCR assays to examine the expression of miRNAs in plasma/serum RNA of all samples. All reagents, primers, and probe were purchased from Applied Biosystems. Real-time PCR was performed using an ABI 7500 Sequence Detection System and fold changes in gene expression were calculated using the 2−ΔΔCt method. The mean miRNA level of the three real-time quantitative PCR experiments was calculated for each case.
Cell Culture and HCV Infection
Immortalized human hepatocytes (IHH) were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, 100 U/mL of penicillin G, and 100 μg/mL of streptomycin at 37°C in a 5% CO2 atmosphere. We have grown HCV genotypes 1a (clone H77) in IHH as described.
Data were analyzed by nonparametric tests using the Wilcoxon test for comparison of paired samples, and Mann-Whitney U test for two nonparametric groups. Receiver operating characteristic (ROC) curves were constructed and the area under the curve (AUC) was calculated to evaluate specificity and sensitivity of predictive value or feasibility of using serum/plasma miRNA as a marker for liver disease progression. P < 0.05 was considered statistically significant. All statistical analyses were performed and graphs were generated using GraphPad Prism 5.0 (GraphPad Software, CA).
Profiling of Serum miRNA Levels in HCV-Infected Patients With Liver Fibrosis
To generate a comprehensive set of miRNA expression profiles in HCV-infected fibrosis patients, we analyzed expression of 84 miRNAs from four samples of each group: healthy volunteers (control group), HCV-infected patients with early-stage (F0-F2) liver fibrosis (Group 1), and late-stage (F3-F4) liver fibrosis (Group 2). The miRNA expression profile was generated using Qiagen miRNA Array Web-based software. Clustering analysis revealed that HCV-infected liver disease patients expressed distinct pattern of miRNAs (data not shown). In all, 36 miRNAs were differentially expressed in HCV-infected patients compared to healthy volunteers and normalized with spiked-in cel-miR-39. We chose the following six miRNAs, which displayed significant differences in fold change between HCV-infected patients and healthy controls: miR-20a, miR-25, miR-27a, miR-92a, miR-148a, and miR-195 to validate miRNA array results by real-time qPCR. We also observed that the miR-574-3p expression level was similar in all categories of samples in miRNA array analysis. Since no significant difference was found in the levels of miR-574-3p between controls, non-HCV-associated samples, and HCV-infected samples, we used miR-574-3p as an additional endogenous control to normalize the samples in further analysis.
Among these dysregulated serum miRNAs, miR-21 and miR-122 were also up-regulated in our array, which were recently reported to be highly expressed in HCV-infected patients. Although there is some information about circulating serum miR-21 in chronic HCV, these reports are conflicting.[25, 26] Further, miR-21 may be a nonspecific biomarker for chronic HCV, since other studies have suggested its potential as a biomarker for other cancers. miR-122 has also been shown to be differentially regulated in liver injury irrespective of etiology.[16, 17, 27] We chose not to include these two miRNAs as predictors for HCV-mediated liver disease progression.
Candidate miRNAs, miR-20a and miR-92a, Are Highly Expressed in Serum of HCV-Infected Patients With Liver Fibrosis
We further validated the six miRNAs individually in a small cohort of sera from healthy volunteers, non-HCV, and HCV-infected samples. After validation, we chose to examine the expression levels of miR-20a and miR-92a in subsequent studies. We observed that miR-20a and miR-92a expression were significantly up-regulated in HCV-infected patients with fibrosis as compared to healthy volunteers and non-HCV related liver disease patients with fibrosis (Fig. 1A,B). We also observed that expression levels of miR-20a in serum is gradually increased from early-stage to late-stage fibrosis in HCV-infected patients (Fig. 2A). However, the level of miR-92a decreased in late-stage fibrosis in HCV-infected patient sera (Fig. 2B). We next determined the predictive value of these miRNAs in identifying the HCV-mediated liver fibrosis progression in a cohort of 22 healthy controls, 20 non-HCV liver disease patients, and 44 patients with HCV. The levels of two miRNAs in these serum samples were measured and ROC analysis was performed on individual miRNAs. miR-20a had an AUC of 0.704 ± 0.067 (95% confidence interval [CI] = 0.571-0.836) with a sensitivity of 61.4% and specificity of 81.8%, and miR-92a had an AUC of 0.787 ± 0.058 (95% CI = 0.672-0.901) with sensitivity of 70.5% and specificity of 77.3% in separating the healthy controls from HCV-infected patients (Fig. 3A). Further, miR-20a displayed an AUC of 0.679 ± 0.070 (95% CI = 0.542-0.817) with sensitivity of 61.4% and specificity of 65%, and miR-92a displayed an AUC of 0.684 ± 0.069 (95% CI = 0.548-0.819) with sensitivity of 70.5% and specificity of 70% (Fig. 3B) in separating non-HCV-infected fibrosis patients from HCV-infected fibrosis patients.
Expression Level of miR-20a and miR-92a in HCV-Infected Patients With Different Clinical Status
We next examined the expression level of these miRNAs in HCV-infected patients with acute, chronic, or resolved samples. The plasma levels of these two miRNAs in HCV-infected patients during different stages of HCV infection (29 with acute hepatitis, 18 with chronic hepatitis, and 11 with resolved infection) were analyzed. Our data indicated that in the acute and chronic stage of HCV infection, miR-20a and miR-92a levels are significantly elevated when compared with healthy volunteers (Fig. 4A,B). On the other hand, the expression level of miR-20a and miR-92a declines in resolved infection. Thus, an increase in expression of miR-20a and miR-92a in plasma may correlate with an early stage of HCV infection. We also performed ROC analysis to determine the predictive value of these miRNAs for detection of the early stage of HCV infection. Our analysis showed that miR-20a had an AUC of 0.883 ± 0.058 (95% CI = 0.769-0.996) with sensitivity of 89.6% and specificity of 80%, and miR-92a had an AUC of 0.889 ± 0.057 (95% CI = 0.778-1.001) with a sensitivity of 89.6% and specificity of 90% in separating the healthy volunteers from acutely infected HCV patients (Fig. 5A). miR-20a also displayed an AUC of 0.983 ± 0.019 (95% CI = 0.944-1.022) with a sensitivity of 100% and specificity of 80%, and miR-92a had an AUC of 0.989 ± 0.015 (95% CI = 0.960-1.018) with a sensitivity of 100% and specificity of 80% in separating healthy volunteers from chronic HCV-infected patients with no fibrosis (Fig. 5B). We further analyzed the longitudinal samples for status of plasma miRNAs in acute to chronic HCV-infected patients (18 pairs). Our results suggested that both miR-20a and miR-92a levels in plasma remained unchanged in acute to chronic pairs of HCV-infected patients (Fig. 6A). Interestingly, when we analyzed the longitudinal samples of acute to resolved groups (11 pairs), only miR-92a expression is significantly reduced (Fig. 6B). There was no significant change in miR-20a expression in the acute to resolved group. While our longitudinal sample data are promising, a larger sample group in a future study will help to confirm the use of these miRNAs as potential biomarkers in the early stage of infection.
Up-Regulation of Candidate miRNAs in In Vitro HCV-Infected Culture Supernatants
To further verify that these identified miRNAs are indeed up-regulated from HCV infection, we determined their status in the HCV-infected culture supernatants as compared to that of mock-infected culture supernatants. We found that miR-20a and miR-92a were highly up-regulated in HCV-infected culture supernatants in comparison to that of mock-infected control, whereas miR-574-3p expression was similar between mock-treated and HCV-infected culture supernatants (Fig. 7).
The search for noninvasive biomarkers for diagnosis of diseases has become a rapidly growing area of clinical research. Unlike screening for large numbers of mRNAs, a small group of miRNAs or even one specific miRNA might be sufficient to differentiate patients from healthy individuals. In this study we demonstrated that up-regulation of selected miRNAs were associated with progression of liver fibrosis in HCV-infected patients. We identified two miRNAs (miR-20a and miR-92a) in association with HCV infection and liver fibrosis. We also observed that expression levels of miR-20a and miR-92a follow an increasing trend in acute and chronic hepatitis. Interestingly, lack of a significant differential pattern of plasma levels of miR-20a in acute to chronic hepatitis (longitudinal samples) but significantly elevated expression in fibrosis stage of HCV infection suggested that miR-20a may be a good predictive biomarker for HCV-mediated liver disease progression.
miR-17-92 cluster is a proto-oncogenic cluster (also called oncomir-1) consisting of six miRNAs which include miR-20a and miR-92a. Increased expression of miR-20a was found in the plasma of chronic lymphocytic leukemia (CLL) patients and in the serum of individuals with gastric cancer. The serum miR-92a level was increased in epithelial ovarian cancer. The circulating miR-92a level was also up-regulated in patients with colorectal cancer (CRC), and advanced adenomas as compared to that of controls, suggesting that circulating miR-92a may serve as a biomarker on early detection of benign lesions before neoplastic formation of CRC. Apart from the oncogenic potential, the miR17-92 cluster is involved in regulation of fibrosis in rodents and human liver. We observed that miR-92a is up-regulated in acute and chronic HCV-infected sera and reduced in resolved samples, suggesting its potential as an early detection marker. Interestingly, plasma miR-92a expression was higher in acute to chronic HCV-infected patients with the highest AUC value. We also observed the presence of these miRNAs in HCV-infected culture supernatants as compared to mock-infected hepatocytes. Thus, we believe that HCV infection in hepatocytes enhances expression of these miRNAs, which, in turn, releases in circulation. Future work is needed to understand the role of these miRNAs in HCV infection.
Chronic HCV infection is associated with liver fibrosis, and eventually develops endstage liver disease. The progression of liver fibrosis varies in HCV-infected patients; therefore, identifying a predictive biomarker will help in developing treatment strategy. Further, current follow-up for fibrosis is liver biopsy or measurement of liver stiffness, and these procedures have limitations. Therefore, a minimally invasive serological marker may be a good alternative for assessment of liver disease progression. Our study demonstrated an up-regulation of miR-20a in HCV-infected patients which positively correlate with progression of liver fibrosis. On the other hand, the circulatory miR-92a level is inversely correlated with fibrosis stage. There are reports suggesting that miRNAs get secreted in the extracellular milieu in response to inflammation or injury to hepatocytes. miR-21 is positively correlated with HCV-mediated fibrosis. miR-21 is known to target SMAD7, and thereby enhances transforming growth factor beta (TGF-β)-mediated fibrosis. Increased levels of miR-21 in association with necroinflammation and drug-induced liver injury are also reported.[36, 37] While our article was in preparation, Trebicka et al. reported that circulating miR-122 levels is inversely correlated with fibrosis stages. The molecular mechanisms of HCV-mediated liver fibrosis are different than that of nonalcoholic steatohepatitis (NASH) or nonalcoholic fatty liver disease (NAFLD). Therefore, it is possible that the HCV-specific circulatory miRNAs play a role in the promotion of liver fibrosis. Indeed, further studies are necessary to elucidate the underlying mechanism.
In conclusion, our study demonstrated that miR-20a and miR-92a are expressed at higher levels in serum/plasma of patients with HCV infection as compared with healthy individuals, suggesting that these serum/plasma miRNAs may serve as potential biomarkers of HCV infection. We have further demonstrated that serum miR-20a expression is elevated with early to late fibrosis stage of HCV-infected patients, but not in non-HCV-infected patients, suggesting its potential as a predictive biomarker for HCV-mediated liver disease progression.
We thank Patricia Osmack for help with the serum samples.