Background

Hepatocellular carcinoma (HCC) represents the fifth most common cancer around the world and the third most frequent cause of cancer-related death [1]. Among the risk factors for developing HCC is the coming out of the hepatitis C virus (HCV) [2]. Egypt has one of the highest prevalences of HCV in the world, estimated nationally to be around 14 % [3].

Early diagnosis of HCC is crucial for improving the survival rate of patients. Early detection by serum alpha fetoprotein (AFP) level is limited by its low sensitivity [4], so there is an urgent demand to develop new non-invasive biomarkers with high accuracy and feasibility.

Micro-ribonucleic acids (miRNAs) represent an abundant class of endogenous small RNA molecules of 20–25 nucleotides in length capable of regulating gene expression either by direct cleavage of targeted messenger RNAs (mRNAs) or by inhibiting translation through complementarity to targeted mRNAs at the 3′-untranslated regions (UTRs) [5].

Deregulated expression of miRNA has been linked to a variety of cancers, including HCC [6]. Differential expression of miRNAs between tumor and normal tissues as well as stability of microRNAs in serum and plasma establish the role of miRNAs as non-invasive efficient biomarkers in cancer diagnostics [6].

In our study, we employed a strategy of comparing the expression level of 13 serum miRNAs from 480 individual subjects by using quantitative real-time array in order to identify a set of miRNAs; can serve as a sensitive, specific, non-invasive, and fingerprint biomarker allowing classification of inflammatory liver, liver cirrhosis, and HCC; and also could discriminate liver diseases from the healthy control, for early detection of HCC. The studied miRNAs were selected according to their role in hepatocarcinogenesis known from previous review of literature.

Methods

Study design and grouping

This was a retrospective case–control study conducted on 384 adult patients with HCV-related chronic diseases categorized into the following: 192 patients with hepatocellular carcinoma (HCC) recruited from the Multidisciplinary HCC Clinic, Tropical Medicine Department, Faculty of Medicine and National Cancer Institute (NCI) outpatient clinic, Cairo University; 96 patients with liver cirrhosis (LC) recruited from Endemic Medicine Department, Faculty of Medicine, Cairo University; and 96 chronic hepatitis C patients (CHC) recruited from Kasr El Aini Viral Hepatitis Center, Faculty of Medicine, Cairo University, in addition to 95 healthy subjects enrolled as the control group during the period from May 2012 to April 2013. The study was approved by the Investigation and Ethics Committee of NCI, and a written informed consent was obtained from all persons involved.

Patients with HCC were diagnosed by abdominal ultrasonography, triphasic CT abdomen, and serum AFP and confirmed histopathologically. They also showed no evidence of local invasion or distant metastasis. However, patients with HCV-related liver cirrhosis were diagnosed by abdominal ultrasonography and confirmed by histopathology examination. Patients with chronic hepatitis C were characterized by persistent increase of the alanine aminotransferase (ALT) values more than three times the normal for at least 6 months. The control group showed no clinical or biochemical evidence of liver disease with normal abdominal ultrasonography. All controls were negative for HBV and HCV infection as tested by ELISA and PCR. Also, they were negative for diabetes mellitus as it is being a risk factor for HCC development. Exclusion criteria were diabetic controls, patients with HBV infection, and patients who received previous treatment or antiviral therapy for HCV.

Serum sample collection

Five milliliters of venous blood was obtained and allowed to coagulate for 30 min before centrifugation at 5000 rpm for 10 min. The serum fraction was extracted and stored at −80 °C until used.

RNA extraction and cDNA synthesis

RNA was extracted from pooled serum samples; each pool consisted of four samples from the same group by using TRIzol reagent and enriched for the miRNA miRNeasy mini kit (Qiagen) cat. no. 217004. For the quality of RNA estimated by absorbance ratios of 260/280 nm greater than 2.0, A260/A230 ratio should be greater than 1.7 and concentration by A260 should be fraction using greater than 10 μg/ml small RNA by using 2000/2000c NanoDrop. The RNA should contain a single sharp peak at a low molecular weight with no smearing and no additional peaks at higher molecular weights. RT2 miRNA First Strand Kit (Qiagen) was used to prepare cDNA from 500 ng RNA.

PCR arrays

PCR arrays (SABiosciences, custom array catalog number caih0038) were used to examine relative expression levels of serum miRNA. First-strand cDNA was synthesized from 0.5 mg total RNA diluted 1:10. The reactions contained 13 ml SYBR PCR Master Mix and 2 ml diluted cDNA in a final volume of 25 ml and were amplified at 95 °C for 10 min and 95 °C for 15 s for 1 cycle and 60 °C for 40 s and 72 °C for 30 s for 40 cycles. The 13 miRNAs are compared and analyzed across multiple plates using the ΔΔCt method and normalized to snord48 (housekeeping gene).

Statistical analysis

Website analysis was performed based on applying Student’s t test of the replicate (2^ (− Delta Ct) values for each gene in the control group and treatment groups, and P values less than 0.05 are considered significant. Fold regulation represents fold change results in a biologically meaningful way. Fold change values greater than 1 indicate a positive or an upregulation, and the fold regulation is equal to the fold change. Fold change values less than 1 indicate a negative or downregulation, and the fold regulation is the negative inverse of the fold change. The area under the receiver operating characteristic curve (AUC) was used to evaluate the diagnostic performance of miRNA panels. Correlation between the variables was analyzed using Pearson’s correlation coefficient.

miRNA selection and target predicting

miR2 disease; a comprehensive resource of miRNA deregulation in various human diseases (http://www.mir2disease.org) was used to select the miRNAs in our study, while miRTaBase, the experimentally validated microRNA-target interaction database (mirtarbase.mbc.nctu.edu.tw), was used to predict downstream targets of miRNAs.

Results

Clinical features

The clinical and laboratory data of the different studied groups were shown in Table 1. There was a statistically significant difference among the studied groups regarding age, gender, albumin, diabetes mellitus, total bilirubin, ALT, AST, ascites, child score, and AFP (P < 0.001). One of the limitations of this study is that the age of the healthy participants was not matched with the age of the other patient groups since most of the old-aged healthy controls did not approve to be involved in the sample collection.

Table 1 Clinical data of the studied groups
Full size table

Analysis of differentially expressed miRNAs

Among the 13 studied miRNAs, there were 3 significantly upregulated miRNAs (miR-602 and miR-125a-5p with fold regulation > 3, P < 0.01; miR-885-5p with fold regulation > 2, P < 0.05; respectively) and 2 significantly downregulated miRNAs (miR-29b and miR-375 with fold regulation > −3, P < 0.01) when comparing CHC group to control group as shown in Table 2 and Fig. 1.

Table 2 Differentially expressed miRNAs in CHC group versus control group
Full size table
Fig. 1
figure 1

a Volcano plot shows differential expression of 13 miRNAs in CHC group (group 1) versus control group. MiR-885-5p, miR-602, and miR-125a-5p were significantly upregulated; miR-375 and miR-29b were significantly downregulated. b Volcano plot shows differential expression of 13 miRNAs in LC group (group 2) versus control group. MiR-885-5p, miR-375, miR-22, and miR-221 were significantly downregulated. c Volcano plot shows differential expression of 13 miRNAs in the HCC group (group 3) versus control group. MiR-122 and miR-885-5p were significantly upregulated; miR-29b was significantly downregulated. d Venn graph represented the differential expression of 13 miRNAs of the three groups in comparison with control group with the overlapping miRNAs. e The clustergram performs hierarchical clustering of the entire dataset to display a heat map with dendrograms indicating co-regulated genes across control, CHC, LC, and HCC groups

Full size image

Moreover, there were only four significantly downregulated miRNAs (miR-375, miR-885-5p, miR-221, and miR-22 with fold regulation > −2, P < 0.01) on comparing LC group to control group as shown in Table 3 and Fig. 1.

Table 3 Differentially expressed miRNAs in LC group versus control group
Full size table

However, on comparing HCC group to control group, there were two significantly upregulated miRNAs (miR-122 and miR-885-5p with fold regulation > 1.5, P < 0.05) and one significantly downregulated miRNA (miR-29b with fold regulation > −2, P < 0.01) as shown in Table 4 and Fig. 1.

Table 4 Differentially expressed miRNAs in HCC group versus control group
Full size table

The differential expression of 13 miRNAs in the three groups compared to control group and the heat map with dendrograms indicating co-regulated genes across all studied groups was illustrated in Fig. 1.

Among the 13 miRNAs studied in HCC and CHC groups, there were only two significantly downregulated miRNAs (miR-22 and miR-199a-3p with fold regulation > −2, P < 0.05) as shown in Table 5 and Fig. 2.

Table 5 Differentially expressed miRNAs in HCC group versus CHC group
Full size table
Fig. 2
figure 2

a Volcano plot shows differential expression of 13 miRNAs in HCC group (group 3) versus CHC group (group 1). MiR-22 and miR-199a-3p were significantly downregulated. b Volcano plot shows differential expression of 13 miRNAs in HCC group (group 3) versus LC group (group 2). MiR-885-5p, miR-221, miR-122, and miR-181b were significantly upregulated. c Venn graph represented the differential expression of 13 miRNAs in HCC group comparing to control, CHC, and LC groups. It has been shown that miR-122 and miR-885-5p were common miRNAs in early detection of HCC in LC group and control group

Full size image

On comparing HCC group to LC group, there were four significantly upregulated miRNAs (miR-122 and miR-885-5p with fold regulation > 4, P < 0.01, and miR-221 and miR-181b with fold regulation > 3, P < 0.05) as shown in Table 6 and Fig. 2. The differential expression of 13 miRNAs in HCC group compared to all studied groups was illustrated in Fig. 2. It has been shown that miR-122 and miR-885-5p were common miRNAs in early detection of HCC in LC group and control group.

Table 6 Differentially expressed miRNAs in HCC group versus LC group
Full size table

The network showing the relationship of significant miRNAs in HCC detected in our study and the relationship between miRNAs and HCV was illustrated in Fig. 3.

Fig. 3
figure 3

a Network showing the relationship of significant miRNAs in HCC detected in our study. The green and red colors of miRNAs represent negative fold and positive fold changes, respectively. Downregulation of miR-29b is associated with increasing resistance to apoptosis. Upregulation of miR-221 is associated with increasing cell proliferation. Upregulation of miR-181b and miR-199a-3p is associated with increasing cell proliferation and metastasis. The maximum divergency is shown by miR-221 which targets many genes related to apoptosis, cell proliferation, and metastasis. b Network showing the relationship between HCV and miRNAs. HCV promotes the expression of miR-221 through NF-ĸB manner and also expression of miR-122 due to its advantage. As binding of miR-122 directly to the 5′-UTR of the HCV genome stimulated HCV translation and stabilizing the HCV genome by protecting the 5′ end of the viral genome from 5′ decay mediated by 5′ exonuclease Xrn1. However, HCV suppress the expression of miR-29b and miR-199a-3p due to its advantage as these miRNAs showed antiviral activity against HCV

Full size image

ROC curve analysis of differentially expressed miRNAs

It has been revealed that miR-29b (AUC 0.766) showed higher diagnostic performance than miR-122 (AUC = 0.617) and miR-885-5p (AUC = 0.63) in discrimination of HCC group from control group. Using combined panel of miRNAs improved the diagnostic performance (AUC = 0.898) over using each of them independently. Addition of AFP to the combined panel of miRNAs offered the highest diagnostic accuracy (AUC = 1) as shown in Table 7 and Fig. 4.

Table 7 ROC curve analysis of individual miRNAs versus combined miRNAs with and without AFP between HCC group and control group
Full size table
Fig. 4
figure 4

a ROC curve analysis of combined panel of miRNAs versus combined one with AFP between HCC group and control group. b ROC curve analysis of combined panel of miRNAs versus combined one with AFP between HCC group and LC group. c ROC curve analysis of combined panel of miRNAs versus combined one with AFP between HCC group and CHC group. d ROC curve analysis of combined panel of miRNAs versus combined one with AFP between LC group and control group. e ROC curve analysis of combined panel of miRNAs versus combined one with AFP between CHC group and control group

Full size image

Moreover, miR-221 (AUC = 0.775) and miR-885-5p (AUC = 0.702) showed higher diagnostic performance than miR-122 (AUC = 0.617) and miR-885-5p (AUC = 0.679) in discrimination of HCC group from LC group. Using combined panel of miRNAs improved the diagnostic performance (AUC = 0.845) over using each of them independently. Addition of AFP to the combined panel of miRNAs offered the highest diagnostic accuracy (AUC = 0.982) as shown in Table 8 and Fig. 4.

Table 8 ROC curve analysis of individual miRNAs versus combined miRNAs with and without AFP between HCC group and LC group
Full size table

However, miR-199a-3p (AUC = 0.7) showed higher diagnostic performance than miR-22 (AUC = 0.586) in discrimination of HCC group from CHC group. Using combined panel of miRNAs unfortunately decreased the diagnostic performance (AUC = 0.668). Addition of AFP to the combined panel of miRNAs offered improvement in the diagnostic accuracy (AUC = 0.988) as shown in Table 9 and Fig. 4.

Table 9 ROC curve analysis of individual miRNAs versus combined miRNAs with and without AFP HCC group and CHC group
Full size table

Furthermore, miR-375 (AUC = 0.797) showed higher diagnostic performance than miR-221 (AUC = 0.768), miR-885-5p (AUC = 0.714), and miR-22 (AUC = 0.748) in discrimination of LC group from control group. Using combined panel of miRNAs improved the diagnostic performance (AUC = 0.882) over using each of them independently. Addition of AFP to the combined panel of miRNAs offered the highest diagnostic accuracy (AUC = 1) as shown in Table 10 and Fig. 4.

Table 10 ROC curve analysis of individual miRNAs versus combined miRNAs with and without AFP between LC group and control group
Full size table

Also, miR-125a-5p (AUC = 0.0.964) and miR-29 (AUC = 0.839) showed higher diagnostic performance than miR-375 (AUC = 0.776), miR-602 (AUC = 0.764), and miR-885-5p (AUC = 0.757) in discrimination of CHC group from control group. Using combined panel of miRNAs improved the diagnostic performance (AUC = 0.964) over using each of them independently. Addition of AFP to the combined panel of miRNAs offered the highest diagnostic accuracy (AUC = 1) as shown in Table 11 and Fig. 4.

Table 11 ROC curve analysis of individual miRNAs versus combined miRNAs with and without AFP between CHC group and control group
Full size table

Correlation analysis between studied miRNAs

Correlation analysis between studied serological markers using Pearson’s correlation coefficient within each group revealed that there was specifically fair positive correlation in the HCC group between miR-885-5p and miR-192 (r = 0.396, P value < 0.01) as rho values were in the range of 0.25–0.5 as shown in Table S1 and Fig. 5. However, there was specifically good positive correlation in the LC group between miR-885-5p and miR-375 (r = 0.631, P value < 0.01) as rho values were in the range of 0.5–0.75 as shown in Table S2 and Fig. 5. Also, there was specifically good positive correlation in the CHC group between miR-221 and miR-375 (r = 0.687, P value < 0.01) as shown in Table S3 and Fig. 5. Furthermore, there was specifically excellent positive correlation in the control group between miR-122 and miR-29b (r = 0.847, P value <0.01) as rho values were greater than 0.75 as shown in Table S4 and Fig. 5.

Fig. 5
figure 5

a Correlation graph of the studied 13 miRNAs in HCC group. b Correlation graph of the studied 13 miRNAs in LC group. c Correlation graph of the studied 13 miRNAs in CHC group. d Correlation graph of the studied 13 miRNAs in control group

Full size image

Discussion

The deregulation of miRNA expression may be a key factor in many liver diseases including hepatocellular carcinoma [6].

This study addressed the possible use of circulating miRNAs as a tool for early detecting HCC. The expression levels of the studied miRNAs were compared in HCC and control groups. Accordingly, we found that miR-122 and miR-885-5p were significantly upregulated while miR-29b was significantly downregulated.

Regarding miR-122, it is a liver-specific miRNA and its well-known function is to regulate lipid and cholesterol metabolism [7–9]. Our finding was in agreement with El-Garem et al. [10] and Varnoholt et al. [11] who reported upregulation of miR-122 expression in HCC associated with HCV compared to healthy controls. On contrary to our results, miR-122 downregulation in HCC has been reported in many previous studies suggesting its function as tumor suppressor miRNA [7, 12, 13]. miR-122 negatively regulates oncogenic genes such as disintegrin and metalloprotease 17 (ADAM17) responsible for metastasis [14], Bcl-w (anti-apoptotic gene) [15], Wnt1 responsible for cell proliferation [16], and cyclin G1 (CCGN1) responsible for cell cycle progression [17].

Machlin et al. [18] revealed that miR-122 binds directly to the 5′-UTR of the HCV genome at two adjacent sites, stimulates HCV translation [19], and promotes HCV RNA accumulation via HCV genome stabilization [20]. Binding of miR-122 to the 5′-UTR of HCV RNA with 3′ overhanging nucleotides was speculated to mask and protect the 5′ end of the viral genome from 5′ decay mediated by 5′ exonuclease Xrn1 [21]. Even the exogenous expression of miR-122 supports HCV replication in non-permissive hepatic and non-hepatic cell lines [22]. HCC associated with HCV is able to evade tumorigenic repression of miR-122. Moreover, overexpression of miR-122 significantly suppressed the interferon-stimulated response element (ISRE) which functions as an enhancer for the induction of transcription by alpha/beta interferons [23]. Consequently, HCV-infected HCC are able to evade tumorigenic repression of miR-122 [24, 25].

Our finding of significant downregulation of miR-29b in HCC group came in accordance with previous observations by Xiong et al. [26] who demonstrated that downregulation of miR-29 was associated with poor disease-free survival of HCC patients. These data suggested a potential tumor suppressive function of miR-29b.

Interestingly, the expression level of miR-29 was found downregulated in hepatocytes infected with HCV. And this was found to be one of the mechanisms of chronic HCV-induced liver fibrosis by potentiating collagen synthesis in activated hepatic stellate cells (HSCs) [27, 28].

The upregulated expression of miR-885-5p in HCC group found in the current study implicates its role as oncogenic miRNA during hepatocarcinogenesis and came in agreement with previous work by Gui et al. [29], who reported significantly upregulated expression of miR-885-5p in HCC, LC, and chronic hepatitis associated with HBV groups compared to control group, suggesting the role of miR-885-5p as a potential marker of liver injury. Interestingly, it has been shown that overexpression of miR-885-5p suppressed the ISRE [23]. On the other hand, comparing the expression level of all studied miRNAs in HCC versus LC group showed significant upregulation of miR-122, miR-885-5p, miR-181b, and miR-221 in HCC group.

Significant upregulation of miR-181b was an event that came in agreement with previous report by Wang et al. [30], assuming its function as oncogenic miRNA in HCC. It has been showed that miR-181b belonging to miR-181 family promotes growth, migration, and invasion and chemoresistance of HCC cells through targeting tumor suppressor gene “tissue inhibitor matrix metalloproteinase 3 (TIMP3),” potent inhibitor of matrix metalloproteinase 2 (MMP2), and matrix metalloproteinase 9 (MMP9) that mediated degradation of extracellular matrix (ECM) [31].

Significant upregulation of miR-221 was an event that came in agreement with previous report by Pineau et al. [32], supporting its function as oncogenic miRNA in HCC. miR-221 promotes cell cycle progression through targeting cyclin-dependent kinase inhibitors which are important regulators of G2/M transition, CDKN1B/p27, and CDKN1C/p57 [33]. In addition to the modulation of cell cycle, miR-221 contributes to the progression of HCC by the inhibition of apoptosis through targeting a pro-apoptotic member of the Bcl-2 family “Bmf” [34].

For early detection of HCC in high-risk patients, we compare the expression level of studied miRNAs in HCC group with CHC group. Our results were that miR-199a-3p and miR-22 were significantly downregulated. Our finding of significant downregulation of miR-199a-3p in HCC group came in accordance with previous observations by Murakami et al. [35] who demonstrated that downregulation of miR-199a-3p was associated with reduced time to recurrence in HCC patients. These data suggested a potential tumor suppressive function of miR-199a-3p. It has been reported that miR-199a-3p targeted mammalian target of rapamycin (mTOR).

Interestingly, miR-199a-3p has antiviral activity against HCV as it directly interacted with the stem-loop (SL) II region of HCV 5′-UTR and inhibits HCV genome replication [36]. Thus, miR-199a-3p could be deleterious to HCV, giving a possible motivation for the virus to downregulate miR-199a-3p to its advantage. So restoration of miR-199a-3p expression would lead to G1 arrest, reducing invasive capability and increasing sensitivity to chemotherapy [37] in addition to inhibition of HCV replication [36]. Our finding of significant downregulation of miR-22 in HCC group came in accordance with previous observations by Zhang et al. [38] who demonstrated that downregulation of miR-22 was associated with poor prognosis in HCC patients. These data suggested a potential tumor suppressive function of miR-22. It has been reported that miR-22 targeted histone deacetylases 4 (HDAC4) [38]. Histone deacetylases (HDACs) regulate the expression and activity of numerous proteins implicated in both cancer initiation and progression.

Comparing the three studied groups (HCC, LC, and CHC) with control group revealed that miR-885-5p was the only significant overlapping miRNA in three groups associated with overexpression in HCC and LC groups and down expression in LC group, suggesting the role of miR-885-5p as a potential marker of liver injury. Downregulation of miR-29b in HCC group was also an early event in CHC group. Downregulation of miR-375 was associated with HCV group (CHC and LC groups). Upregulation of miR-122 was an event associated with HCC development. Upregulation of miR-602 and miR-125a-5p was an event associated with inflammation process. Downregulation of miR-221 and miR-22 was an event associated with cirrhosis process.

ROC curve analysis of individual miRNAs between HCC group and control group revealed that miR-29 showed fair diagnostic accuracy as AUC was in the range of 0.7–0.79, while miR-885-5p and miR-122 showed poor diagnostic accuracy as AUC was in the range of 0.6–0.69. ROC analysis of combined miRNAs showed good diagnostic accuracy as AUC was in the range of 0.8–0.89, while combined panel of miRNAs with AFP has offered excellent diagnostic accuracy in the discrimination between HCC group and control group as AUC was in the range of 0.9–1. While the analysis between HCC group and LC group revealed that miR-885-5p and miR-221 showed fair diagnostic accuracy as AUC was in the range of 0.7–0.79. miR-122 and miR-181b showed poor diagnostic accuracy as AUC was in the range of 0.6–0.69. The combined miRNAs showed good diagnostic accuracy as AUC was in the range of 0.8–0.89, while combined panel of miRNAs with AFP has offered excellent diagnostic accuracy in the discrimination between HCC group and control group as AUC was in the range of 0.9–1. Moreover, the analysis between HCC group and CHC group revealed that miR-199a-3p showed fair diagnostic accuracy as AUC was in the range of 0.7–0.79, while miR-22 showed failed diagnostic accuracy as AUC was in the range of 0.5–0.59. The combined miRNAs showed unfortunately poor diagnostic accuracy as AUC was in the range of 0.6–0.69, while combined panel of miRNAs with AFP has offered excellent diagnostic accuracy in the discrimination between HCC group and control group as AUC was in the range of 0.9–1. Furthermore, the analysis between LC group and control group revealed that miR-885-5p, miR-22, miR-375, and miR-221 showed fair diagnostic accuracy as AUC was in the range of 0.7–0.79. The combined miRNAs showed good diagnostic accuracy as AUC was in the range of 0.8–0.89, while combined panel of miRNAs with AFP has offered excellent diagnostic accuracy in the discrimination between HCC group and control group as AUC was in the range of 0.9–1. While the analysis of individual miRNAs between CHC group and control group revealed that miR-125a-5p showed excellent diagnostic accuracy as AUC was in the range of 0.9–1, miR-29b showed good diagnostic accuracy as AUC was in the range of 0.8–89 and miR-602 and miR-375 showed fair diagnostic accuracy as AUC was in the range of 0.7–0.79. The combined miRNAs offered improved diagnostic accuracy as AUC was in the range of 0.9–1, while combined panel of miRNAs with AFP has offered the highest diagnostic accuracy (AUC = 1) in the discrimination between HCC group and control group.

Using miRNA panel of miR-122, miR-885-5p, and miR-29b with AFP provided high diagnostic accuracy (AUC = 1) for early detection of HCC in normal population, while using miRNA panel of miR-122, miR-885-5p, miR-221, and miR-22 with AFP provided high diagnostic accuracy (AUC = 0.982) for early detection of HCC in LC patients. Also, using miRNA panel of miR-22 and miR-199a-3p with AFP provided high diagnostic accuracy (AUC = 0.988) for early detection of HCC in CHC patients. Using miRNA panel of miR-22, miR-375, miR-885-5p, and miR-221 with AFP provided high diagnostic accuracy (AUC = 1) for discrimination between LC group and control group. Using miRNA panel of miR-125a-5p, miR-375, miR-885-5p, miR-29b, and miR-602 with AFP provided high diagnostic accuracy (AUC = 1) for discrimination between CHC group and control group.

Conclusion

We identified serum miRNA panels that differentiate HCC patients from patients with benign liver disease and healthy controls with a high degree of accuracy in non-invasive and inexpensive manner. Also, our serum miRNA panels have afforded considerable clinical value for early detection of HCC so that more patients, who missed the curative treatment, can benefit from therapy. However, further studies are needed to validate those serum miRNA panels in an independent cohort study.