Biomarker MicroRNAs for Diagnosis, Prognosis and Treatment of Hepatocellular Carcinoma: A Functional Survey and Comparison

Hepatocellular Carcinoma (HCC) is one of the most common malignant tumors with high incidence and mortality rate. Precision and effective biomarkers are therefore urgently needed for the early diagnosis and prognostic estimation. MicroRNAs (miRNAs) are important regulators which play functions in various cellular processes and biological activities. Accumulating evidence indicated that the abnormal expression of miRNAs are closely associated with HCC initiation and progression. Recently, many biomarker miRNAs for HCC have been identified from blood or tissues samples, however, the universality and specificity on clinicopathological features of them are less investigated. In this review, we comprehensively surveyed and compared the diagnostic, prognostic, and therapeutic roles of HCC biomarker miRNAs in blood and tissues based on the cancer hallmarks, etiological factors as well as ethnic groups, which will be helpful to the understanding of the pathogenesis of biomarker miRNAs in HCC development and further provide accurate clinical decisions for HCC diagnosis and treatment.


Functional characterization of HCC biomarker miRNAs based on cancer hallmarks. The func-
tional characterization of HCC biomarker miRNAs are summarized from the primary references and classified into 12 categories as shown in Fig. 1. It indicates that the biomarker miRNAs are associated with all aspects of hallmarks of cancers and all the hallmarks lead to the cancer. Therefore, the personalized biomarkers are needed to precision diagnosis, prognosis and treatment of the complex HCC. The functions of the biomarker miRNAs are summarized as follows.
Insensitivity to Antigrowth Signals. Although it is unclear for the units and interconnections between the different kinds of antigrowth and differentiation-including signals and the core cell cycle machinery, an antigrowth signaling must be exist to circumvent developing HCC 31 . MiR-125b-5p and miR-15b-5p were the circulating diagnostic miRNA biomarkers associated with insensitivity to antigrowth signals and all of them were up-regulated and highly expressed in early-stage HCC cases 33 . Liu et al. combined miR-15b-5p and miR-130b-3p as a classifier for HCC detection, yielding a receiver operating characteristic curve area of 0.98 in their validation study, the same was found in tissue samples, miR-15-5p was also reported highly expressed 34 . As for prognostic biomarkers, three miRNAs related to insensitivity to antigrowth signals in the tissue samples were identified, including miR-137, miR-185-5p and miR-26a-5p. All of them were down-regulated in poor prognostic group which had a lower survival rate and shorter time to recurrence [35][36][37] .
Scientific RepoRts | 6:38311 | DOI: 10.1038/srep38311 Resisting Cell Death. Cancer cells evolve various ways to circumvent or restrict apoptosis. The diversity of apoptosis-avoiding machinery and program reflects the multiplicity of apoptosis-including signals that tumor cell populations experienced while their evolution to the malignant state 32 . In tissues, miR-101-3p, miR-224-5p and miR-483-5p were associated with resisting cell death. Among them, miR-101-3p was down-regulated whereas the remaining two were reported to be up-regulated [38][39][40] . Resisting cell death was significantly associated with lower expression of miR-101-3p, miR-16-5p, miR-195-5p, miR-203a-3p and miR-221-3p in blood samples 38,[41][42][43] . Increased miR-221-3p, miR-224-5p, miR-483-5p and miR-122-5p expression were also detected in blood of HCC patients 40,44 . These above diagnostic biomarkers as classifiers for HCC detection, yielding a receiver operating characteristic curve area of 0.635 to 0.884 (see Tables 1 and 2). On the other hand, miR-155-5p, miR-206, miR-21-5p and miR-212-3p could be recognized as biomarkers for HCC prognosis in tissues. The expression levels of miR-155-5p and miR-21-5p were up-regulated whereas others were down-regulated [45][46][47][48] . Circulating miR-122-5p and miR-16-5p could be used as putative biomarkers for HCC. Among them, miR-122-5p and miR-16-5p were shown to be up and down-regulated, respectively 49,50 . Avoiding Immune Destruction. According to the long-standing theory of immune surveillance proposes, most of solid tumors such as HCC appeared to have somehow controlled to avoid detection by the different kinds of arms of the immune system or could limit the extent of immunological killing, thus they could evade eradication by immune system 32 . Motawi and his colleagues overviewed that serum miR-146p-5p was up-regulated in HCC and showed the clinical value for HCV-related HCC diagnosis. This circulatory biomarker miRNA was reported to exerted negative effects on anti-tumor immune response 42 .
Tumor Promoted Inflammation. Inflammation has been proved to be existed at the earliest stage of tumor processes and to be capable of fostering the progression of incipient neoplasia into advanced tumors 79 . Besides chemicals, particularly reactive oxygen species were positively mutagenic for adjacent cancer cells, accelerating their genetic evolution towards the high malignant carcinoma 80 . In blood, the increased expression of miR-30c-5p could be used as a new classifier for HCV-positive HCC in early-stage 81 . In addition, hepatic necroinflammatory activity was associated with the high expression of miR-122-5p in plasma. The over expression of circulating miR-122-5p was a prognostic biomarker predicting the poor survival rate of patients underwent radio frequency ablation 49 .

Comparison of HCC biomarker miRNAs based on etiological factors and ethnic groups.
Recently, accumulating evidence indicated that the occurrence and development of HCC are closely associated with etiological factors as well as ethnic groups. The differentiation between HCC and liver cirrhosis, for instance,  Biomarker miRNAs for classifying of HCC and liver cirrhosis. After manually searching for citations in PubMed, a total of 13 miRNA biomarkers for liver cirrhosis diagnosis were collected (see Table S1). We then compared them with HCC diagnostic miRNA biomarkers in order to screen key signatures for HCC early detection. As shown in Fig. 2, eight miRNAs, i.e. miR-106b-5p, miR-122-5p, miR-141-3p, miR-146a-5p, miR-181b-5p, miR-18a-5p, miR-19a-3p and miR-21-5p, were shared by cirrhosis and HCC. Interestingly, three of them (miR-106b-5p, miR-18a-5p and miR-21-5p) showed inverse expression patterns in cirrhosis and HCC groups. For example, the expression of miR-106b-5p (miR-106b) was down in cirrhosis samples 115 whereas it turned out to be up-regulated in the blood of HCC patients 10 . In addition, miR-19a-3p (miR-19a) was reported as a useful molecular marker for monitoring the progression of liver fibrosis to cirrhosis and finally, to HCC 42 .
The remaining 5 and 49 miRNAs, respectively, were specific to cirrhosis and HCC, which could be served as independent factors for classifying of cirrhosis and HCC. For example, miR-29c-3p showed significant positive correlations with the level of serum cholinesterase (CHE) and albumin (ALB) in liver cirrhosis patients, suggesting that the miRNA played functional roles in the establishment of liver cirrhosis 116 . Han et al. found that two miRNAs, i.e. miR-224 (miR-224-5p) and miR-214 (miR-214-3p), were significantly up-and down-regulated in HCC tissue samples respectively, which provided novel biomarker signatures for HCC diagnosis and treatment 39 .
It can be concluded that biomarker miRNAs revealed the pathogenesis of cirrhosis and HCC at the post-transcriptional level and could help deeply understand the differentiation between cirrhosis and HCC. From the perspective of precision medicine, HCC miRNA biomarkers, especially those specific to HCC, were indicators for capturing the early diagnostic signatures at the time of HCC initiation. Biomarker miRNAs for monitoring the development of HBV/HCV-related HCC. It has been widely acknowledged that the progression of HCC is closely affected by the infection of etiological factors, such as HBV, HCV, etc. On the other hand, miRNAs are reported to play crucial roles in HBV/HCV replication and pathogenesis [117][118][119] , i.e. they regulated HBV by directly binding to HBV transcripts or changing HBV gene expression at the transcriptional level 118 . For better investigating the influence of HBV/HCV on HCC development, miRNA biomarkers for HBV/HCV-related HCC were extracted from our collected dataset. As illustrated in Fig. 3, several miRNAs, i.e. miR-122-5p, miR-126-3p, miR-143-3p, miR-192-5p, etc., were functional in both HBV-and HCV-related HCC evolutionary progression. For example, Tan et al. found that serum miR-122-5p could be used as the diagnostic biomarker for detecting HBV-related HCC. Both the area under the receiver operating characteristic curve (AUC) and logistic regression model convinced the predictive power 86 . Meanwhile, the miRNA was also turned out to be effective for early detection of HCC on top HCV infection. Using the miRNA panel where miR-122-5p included, HCC patients could be classified from healthy controls and liver cirrhosis patients with high diagnostic accuracy 120 .
There is still a large number of biomarker miRNAs that could be specifically used for monitoring the development of HBV/HCV-related HCC. Chen et al. analyzed the plasma samples from 242 individuals and uncovered that the expression of miR-125b-5p (miR-125b) was significantly down-regulated in HBV-induced HCC (HBV-HCC) patients compared to healthy controls as well as HBV groups without HCC 121 . Moreover, the low plasma level of miR-125b-5p also reflected the higher possibility of metastasis. Therefore, the miRNA held promise as a valuable diagnostic biomarker for HBV-HCC and HBV-infected patients with high HCC risks could be early detected by dynamically monitoring the changes of this miRNA. Liu et al. demonstrated that the expression levels of miR-30c-5p (miR-30c) and miR-203a-3p (miR-203a) were crucial indicators for predicting the poor prognosis of HCV-related HCC because the core protein of HCV could down-regulate the expression of miR-30c-5p and miR-203a-3p, resulting in the activation of epithelial-mesenchymal transition in normal hepatocytes as well as HCC tumor cells. As reported before, the activation process may contribute to the carcinogenesis of HCC 105 .
Understanding the pathogenesis of miRNA biomarkers in HBV/HCV-related HCC provided insights to evaluate the potential effects of HBV/HCV on HCC development, which will be helpful to the early and personalized detection of HCC.
HCC miRNA biomarkers within different ethnic groups. Genomic profiling of HCC tumors showed that HCC patients in different geographic regions tended to have specific recurrent molecular aberrations 122 . Asians, on the whole, achieved the highest HCC incidence according to the report by Wong et al. 123 . In terms of prognosis, the overall survival rate was also disparate among different ethnic groups 124 . Here we reorganized HCC miRNA biomarkers based on the ethnicity of patients described in each citation. As illustrated in Fig. 4a, most of the reported HCC miRNA biomarkers were related to Chinese population, which indirectly indicated the high risk or high incidence of HCC in China. For further exploring the ethnic specificity of HCC miRNA biomarkers, we then partitioned miRNAs into two categories based on the patient race, i.e. Asian-related (Chinese, Japanese, South Korean, Indian and Iranian) and non-Asian-related (Egyptian, American, Turk and German) HCC miRNA biomarkers. As shown in Fig. 4b, the number of Asian-specific HCC miRNA biomarkers is far more than that of non-Asian. We noticed that some miRNAs were reported to be functional in both Asian and non-Asian group. However, the expression pattern of them was sometimes quite different when they were involved in different pathogenic processes or belonged to different ethnic groups. For example, miR-125b-5p was associated with the biological behavior of HCC and had the diagnostic value of HCC for both Turks and Chinese. As in plasma samples of Chinese patients, it was found to be down-regulated 121 whereas in Turks samples, its expression level was  33 . For comparison of Egyptian and Chinese, the down-regulation of miR-146a-5p was correlated with HCC carcinogenesis and deterioration in Chinese population 103 , but in samples of Egyptian patients, it was inverse 42 .

Reported ID
This ethnic difference may be caused by the heterogeneous pathogenesis, lifestyles and various factors including the diet, environmental exposures, etc. Moreover, the incidence of HBV/HCV infection in different countries is also inconsistent. Therefore, more in-depth researches on ethnically specific miRNA biomarkers is of clinical significance, which would provide personalized strategies for HCC diagnosis and treatment in the era of precision medicine.
Pathway enrichment analysis for targets of HCC miRNA biomarkers. We performed the pathway enrichment analysis for targets of different types of reported miRNA biomarkers using IPA program. Here the targets of miRNA biomarkers originated from seven publicly available miRNA-target databases, including four experimentally validated databases and three computationally predicted databases (see Methods). For the three categories, i.e. the diagnostic, prognostic and therapeutic biomarker miRNAs, the top 10 significantly enriched pathways (p-value < 0.01) were chosen and shown in Fig. 5. The common enriched pathways among them were Molecular Mechanisms of Cancer, Glucocorticoid Receptor Signaling, HGF Signaling, NGF Signaling, p53 Signaling etc. Most of them are well-studied cancer associated pathways. Das et al. reported that the pathway Molecular Mechanisms of Cancer was potentially associated with recurrent HCC secondary to HCV following liver transplantation 125 . Glucocorticoids are involved in controlling many essential biological processes that are related to energy supply and growth control. The Glucocorticoid Receptor often functions as a cofactor of transcription factor STAT5 for growth hormone induced genes and Glucocorticoid Receptor Signaling has been turned out to be important in body growth, steatosis and metabolic liver cancer development 126  Signaling pathway was significantly dysregulated in HCC and it could reflect the development and progression of HCC 129 . Moreover, a number of genes participated in regulating human HCC by interacting with p53 Signaling pathway. For instance, the key gene RASSF10, which is located on chromosome 11p15.2, could suppress the growth of HCC via activating p53 Signaling pathway 130 . EGR1 is one of the key components in p53 Signaling, the re-expression of gene BCL6B in HCC cells could increase its expression and finally contribute to the activation of p53 Signaling 131 .

Discussion
In this review, we made comprehensive functional survey and comparison of HCC diagnostic, prognostic and therapeutic miRNAs in blood and tissues. The number of diagnostic miRNA biomarkers in blood is approximately twice as much as those in tissues and meanwhile, the number of prognostic miRNA biomarkers in tissues is twice as much as those in blood. The reason for the statistical difference may be that many studies are inclined to investigate the noninvasive diagnostic miRNA biomarkers and researchers tend to use relatively stable hepatogenic biomarkers as prognostic indicators because miRNAs may be released into the blood selectively 132,133 . Most of the diagnostic, prognostic and therapeutic miRNA biomarkers are associated with one or two clinic pathological features in blood and tissues. A great number of prognostic biomarkers with high expression levels were detected in patients with shorter overall survival. Since the etiological factors as well as ethnic groups are closely associated with HCC carcinogenesis, we analyzed miRNA biomarkers by taking the HBV/HCV infection as well as regional variations into account in order to provide better clues for HCC pathogenic research. We mainly selected miRNAs which were explicitly reported as HCC markers/biomarkers in our current study. Besides, several miRNAs are still common and important during HCC development. For example, miR-142-3p was functional in HCC tumorigenesis and played a key role in regulating human RAC1 gene. The upregulation of miR-142-3p inhibited the expression level of RAC1 mRNA, suppressing the migration and invasion of HCC cells 134 . Interferon regulatory factor-1 (IRF-1) is a tumor-suppressor in HCC and its down-expression would help HCC tumors evade death. Yan et al. found that miR-23a was a negative regulator of IRF-1in HCC, which highlighted its importance in HCC initiation and progression 135 . Zhang et al. demonstrated that miR-99a could directly regulate AGO2 and control tumor growth in HCC, indicating the potential strategies for HCC treatment 136 .
HCC is a complex disease which is difficult for early diagnosis and treatment. The death rate of HCC remains high due to its poor prognosis. To some extent, miRNAs are effective biomarkers for HCC because of the noninvasive detection, good specificity and sensitivity. More systematic investigations and clinical experiments need to be done for better understanding the role and function of miRNA biomarkers in HCC pathogenesis [137][138][139] .