Our large and comprehensive screening revealed that hepatic miRNA expression can be associated with a patient's drug response. There are several reports that miRNAs are closely related to innate immunity, and in this study, we found that several miRNAs had the potential to recognize immuno-related genes as target candidates [24
]. For example, the following hypothetical candidate genes of miR-378, miR-18a, miR-27b, miR-34b, and miR-145 each identified as target genes, Interferon Response Factor (IRF) 1, IRF2, IRF4, IRF6, and IRF7, respectively (additional file 1
). Past reports show that miR-422b was related to the B cell differentiation [27
]. When an immuno-reaction induces aberrant expression of miRNA, the expression level of miR-34b significantly decreased in H69 cells following IFN-γ stimulation [28
]. Bcl-6 positively directs follicular helper T cell differentiation, through combined repression of miR-18a and miR-27b and transcription factors [29
In our study, there was significant difference in the fold change of the expression level of miRNA based on the drug response, however, the absolute value of the fold change was not so significant (Table ). Usually one miRNA can regulate many genes including immuno-related gene (additional file 1
), and these genes in turn can synergistically affect immune activity. In our preliminary study (additional file 2
) BCL-2, RARA, and SMAD2 can be regulated by miR-143, miR-18a, and miR-27b, respectively. Considering that the expression level of several miRNAs changed these minute changes taken together can have a significant impact on a patient's drug-response and innate immunity.
Aberrant expression of miRNA can modify the replication of HCV. According to Vita algorithm, several miRNAs, related to drug response, can recognize HCV genotype 1b sequence as a target (additional file 3
]. For example, miR-199a* is able to target the HCV genome and inhibit viral replication [12
]. IFN has the ability to modulate expression of certain miRNAs that may either target the HCV RNA genome (miR-196 or miR-448) or markedly enhance its replication (miR-122) [10
]. The low expression level of miR-122 in the subjects shown in the NR group is in accordance with our results, however, after miRNA expression profile with FDR correction, the expression level of miR-122 was not significantly different between SVR and NR groups [8
]. One reason why this difference is that their study comprised of patients infected with HCV genotype from 1 to 4 while this study consisted of HCV genotype 1b patients only.
The expression pattern of mRNA in HCV infected liver tissue is different from that of healthy tissue [15
]. The expression pattern of the IFN-related genes in liver tissue before administering of IFN therapy, also differs according to the drug response [15
]. The amount of plasmacytoid dendritic cell (pDC), which are the most potent secretors of antiviral Type-I IFN, has been shown to decrease in the peripheral blood of patients, however, pDC tend to become trapped in the liver tissue if HCV infection is present [31
]. Taken together, it is possible that the variation in the miRNA expression pattern according to the drug response existed even before therapy.
Previously, large randomized controlled trials of IFN therapy for CHC, identified at pre-treatment stage several possible factors which are associated with the final virological response. These factors include: genotype, amount of HCV RNA in peripheral blood, degree of fibrosis, age, body weight, ethnicity, and steatosis [33
]. Viral genome mutation in the ISDR region and the substitutions of amino acid in the HCV core region also served as predictors of early, as well as end-of treatment response [13
]. The miRNA expression obtained from the therapeutic response, can be applied to the prediction of drug response. The advantages of using miRNA for the microarray analysis include the following; (i) It was relatively easy to analyze because fewer probes were installed compared with the usual cDNA array. (ii) The change in each manifestation of a miRNA was low, in fact, in most miRNA, standard deviation was twice or less in average value (data not shown). The expression levels of miR-34b and 422b in the early response phase and final responses to treatment were consistently and significantly high and low in non-responders, respectively. Therefore these two miRNAs may be useful markers for early-to-final drug response to the IFN treatment.
Further studies are indeed needed to clarify the connection between miRNA expression and patient response to CHC combination therapy. Because information on miRNA is regularly being updated, we are planning to performed more analysis using the latest microarray and a larger sample in the future. However, in the meantime, as we have shown in Figure , the bigger the size of the training set, the higher the prediction performance that is achieved. This combined with the results of our Monte Carlo cross validation provided a strong based to verify the concepts in this report. We believe that our results have three advantages (i) the prediction methods used were quite reasonable, (ii) the prediction performance can later be improved if more patients' data become available and (iii) obtaining miRNA profile (not specific miRNAs) is useful for predicting the drug response. While current therapy is based on positive selection with HCV genotype or negative selection with IL28B SNP, and is limited to only some cases, our methods are applicable to all patients [13