Serum microRNA-122 level correlates with virologic ... - PNAS

2 downloads 0 Views 170KB Size Report
Apr 23, 2013 - and the drug (miravirsen) is now in phase II trial; however, in a recent study, Hsu et al. demonstrated the deletion of mouse miR-. 122 resulted ...
Serum microRNA-122 level correlates with virologic responses to pegylated interferon therapy in chronic hepatitis C Tung-Hung Sua,b,c, Chen-Hua Liua,b,c, Chun-Jen Liua,b,c, Chi-Ling Chena, Te-Tien Tingd, Tai-Chung Tsenga,e, Pei-Jer Chena,b,c, Jia-Horng Kaoa,b,c,f,1, and Ding-Shinn Chena,b,c,1 a Graduate Institute of Clinical Medicine, National Taiwan University College of Medicine, Taipei 10002, Taiwan; bDivision of Gastroenterology and Hepatology, Department of Internal Medicine, National Taiwan University Hospital, Taipei 10002, Taiwan; cHepatitis Research Center, National Taiwan University Hospital, Taipei 10002, Taiwan; dGraduate Institute of Epidemiology and Preventive Medicine, National Taiwan University College of Public Health, Taipei 10002, Taiwan; eDivision of Gastroenterology, Department of Internal Medicine, Buddhist Tzu Chi General Hospital Taipei Branch, Taipei 23142, Taiwan; and fDepartment of Medical Research, National Taiwan University Hospital, Taipei 10002, Taiwan

MicroRNA-122 (miR-122) facilitates hepatitis C virus replication in vitro. Serum miR-122 has been implicated as a biomarker for various liver diseases; however, its role in chronic hepatitis C remains unclear. To address this issue, 126 patients with chronic hepatitis C who completed pegylated IFN plus ribavirin therapy with sustained virologic response (SVR) or nonresponse (NR) were retrospectively included, and their pretreatment clinical profiles and treatment responses were collected. Serum miR-122 was quantified before and during treatment. Another 51 patients in SVR and NR groups were prospectively enrolled for validation. Serum miR-122 was found to be a surrogate for hepatic miR-122 and positively correlated with hepatic necroinflammation. Patients who showed complete early virologic response and SVR had significantly higher pretreatment serum miR-122 levels than those with NR (P = 0.001 and P = 0.008, respectively), especially in subgroups of patients with hepatitis C virus genotype 2 and IL-28B rs8099917 TT genotype. Patients with IL-28B TT genotype had significantly better treatment responses and higher pretreatment serum miR-122 level than those with GT or GG genotypes. Univariate analysis showed that pretreatment body mass index, γ-glutamyl transpeptidase, triglyceride, IL28B TT genotype, and serum miR-122 are predictors for SVR. Multivariate analysis specifically in IL-28B TT genotype demonstrated that pretreatment serum miR-122 independently predicted SVR. The validation cohort confirmed a significantly greater pretreatment serum miR-122 level in patients with SVR compared with NR (P = 0.025). In conclusion, serum miR-122 may serve as a surrogate of hepatic miR122, and a higher pretreatment serum miR-122 level can help predict virologic responses to pegylated IFN plus ribavirin therapy.

H

epatitis C virus (HCV) infection is one of the leading causes of chronic liver disease, cirrhosis, and hepatocellular carcinoma (HCC) (1), affecting more than 170 million individuals worldwide. Currently, pegylated IFN plus ribavirin (pegIFN/ RBV) is the mainstay of therapy, with overall sustained virologic response (SVR) rates of 36% to 79% in patients with HCV genotype 1 and 80% to 95% in patients with HCV genotype 2/3 (2– 4). The introduction of direct antiviral agent further improves SVR and may shorten treatment duration in patients with genotype 1 (5, 6). Nevertheless, pegIFN/RBV is still the backbone of current triple therapy, despite many unpleasant adverse effects (7). It is therefore prudent to identify novel therapeutic targets and outcome predictors. MicroRNA (miRNA) is a single-stranded RNA with 21 to 23 nt, which can regulate gene expression by base pairing with their complementary mRNA to inhibit mRNA translation or induce mRNA degradation (8). miRNA-122 (miR-122) accounts for ∼70% of all miRNAs cloned from the liver, and is undetectable in other organs (9). MiR-122 has been found to facilitate HCV RNA synthesis and translation through its interaction with the 5′ UTR of HCV genome (10–12), which is highly conserved across major

www.pnas.org/cgi/doi/10.1073/pnas.1306138110

HCV genotypes. Furthermore, IFN-β down-regulates miR-122 in vitro and in vivo (13), possibly acting as a mechanism to inhibit HCV replication. A previous study indicated that hepatic miR-122 level was lower in treatment nonresponders and higher in those with good early virologic response to pegIFN/RBV therapy (14). Recently, Lanford et al. demonstrated that the modified anti–miR122 molecules exerted a potent anti-HCV effect in HCV-infected chimpanzees (15). Collectively, hepatic miR-122 may serve as a potential therapeutic target and outcome predictor in the management of chronic hepatitis C (CHC). In clinical practice, monitoring hepatic miR-122 is inconvenient because of the invasiveness of liver biopsy. Recent studies convincingly showed that miRNAs stably present in human serum or plasma samples by the carriers of exosomes or argonaute2 (Ago2) complex (16, 17), and may be used as diagnostic markers in several cancers (18, 19). Serum miR-122 has been implicated as a potential biomarker in drug-, alcohol-, hepatitis B virus-, or HCV-related liver diseases and HCC (20–24). However, the role of serum miR-122 in the management of CHC remains largely unknown and deserves further investigations. To this end, we aimed to evaluate the clinical usefulness of serum miR-122 in the treatment of CHC. Results Correlation of miR-122 Level Among Serum, Plasma, and Liver Samples. Because a serum sample is easier to obtain than

a plasma sample in clinical practice, we evaluated the correlation of miR-122 levels between serum and plasma samples from six healthy volunteers. A strong correlation was found between plasma and serum miR-122 (r = 0.943, P = 0.0048; Fig. S1), indicating that serum and plasma miR-122 levels were suitable for measurement of blood-based biomarkers. MiR-122 is mainly synthesized by hepatocytes; therefore, it is important to investigate whether serum miR-122 could reflect its amount in the liver. In our cohort, 20 patients had residual pretreatment liver tissue specimens, and a positive correlation was found between serum and hepatic miR-122 expression (r = 0.51, P = 0.02; Fig. S2). Serum miR-122 Level in Patients with CHC. The clinical characteristics of 126 treatment-naive patients with CHC are shown in Table

Author contributions: T.-H.S., J.-H.K., and D.-S.C. designed research; T.-H.S., C.-H.L., C.-J.L., T.-C.T., P.-J.C., J.-H.K., and D.-S.C. performed research; T.-H.S., C.-L.C., T.-T.T., J.-H.K., and D.-S.C. analyzed data; and T.-H.S., J.-H.K., and D.-S.C. wrote the paper. The authors declare no conflict of interest. 1

To whom correspondence may be addressed. E-mail: [email protected] or chends@ntu. edu.tw.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1306138110/-/DCSupplemental.

PNAS Early Edition | 1 of 6

MEDICAL SCIENCES

Contributed by Ding-Shinn Chen, April 1, 2013 (sent for review September 25, 2012)

1. In total, 118 patients had genomic DNA samples for IL-28B rs8099917 genotyping and a predominance of T alleles (90%) was found: major homozygotes (TT) in 82%, heterozygotes (GT) in 16%, and minor homozygotes (GG) in 2%. The mean serum miR122 level was 9.47 log10 copies per milliliter and had been verified by two individual experiments with good reproducibility (R2 = 0.99, P < 0.0001; Fig. S3). The compliance of pegIFN/RBV therapy was good in our patients, with a mean administration rate of >95% of expected dosage. Before treatment, serum miR-122 level significantly correlated with hepatic histology activity index (HAI) score (r = 0.23, P = 0.012) and serum alanine aminotransferase (ALT) levels (r = 0.24, P = 0.008; Fig. S4 A and B), suggesting a positive association of serum miR-122 level with severity of hepatic necroinflammation. In contrast, no correlation was found between serum miR-122 level and HCV viral load (r = −0.15, P = 0.093), liver fibrosis (Fig. S4 C and D), and lipid profiles (data not shown). Serum miR-122 Correlates with Virologic Responses. The variables of patients with different treatment responses to pegIFN/RBV therapy (SVR, n = 98; NR, n = 28) are shown in Table 2. Patients with SVR had significantly lower body mass index (BMI), γ-glutamyl transpeptidase (GGT) level, triglyceride level, and Metavir score, and a greater percentage of IL-28B rs8099917 TT genotype, compared with those with NR. Of note, patients who showed complete early virologic response (cEVR) or SVR had a significantly higher pretreatment serum miR-122 level than those with poor virologic responses (cEVR+ vs. cEVR−, 9.55 vs. 9.27 log10 copies per milliliter ; P = 0.001; and SVR vs. NR, 9.52 vs. 9.29 log10 copies per milliliter ; P = 0.008), but this was not the case for rapid virologic response (RVR; RVR+ vs. RVR−, 9.53 vs. 9.39 log10 copies per milliliter ; P = 0.070; Fig. 1A). Subgroup analysis showed this difference between SVR and NR was more significant among patients with genotype 2 (9.60 vs. 9.25 log10 copies per milliliter ; P = 0.006) but not patients with genotype 1 (9.43 vs. 9.31 log10 copies per milliliter ; P = 0.269; Fig. 1B). Regarding host genetic variations, patients with SVR harbored a predominance of IL-28B rs8099917 T allele (TT/GT genotype; 100%), whereas the G allele frequency increased in patients with NR (GG/GT genotype; 67%). The IL-28B rs8099917 TT genotype was significantly associated with SVR (P < 0.001). In addition, patients with IL-28B TT genotype had a significantly higher baseline serum miR-122 level

than those with GT or GG genotype (9.52 vs. 9.35 vs. 8.78 log10 copies per milliliter ; P for trend = 0.014; Fig. 2A). A significantly higher pretreatment serum miR-122 level was found among patients with cEVR (cEVR+ vs. cEVR−, 9.56 vs. 9.24 log10 copies per milliliter ; P = 0.004) and those with SVR (SVR vs. NR, 9.55 vs. 9.29 log10 copies per milliliter ; P = 0.030) specifically in subgroup of IL-28B TT genotype, but not in GT or GG genotype (cEVR+ vs. cEVR−, 9.20 vs. 9.31 log10 copies per milliliter ; P = 0.688; and SVR vs. NR, 9.20 vs. 9.31 log10 copies per milliliter ; P = 0.688; Fig. 2B). Univariate analysis showed pretreatment BMI, GGT, triglyceride, IL-28B TT genotype, and serum miR-122 were predictors for SVR. Multivariate analysis showed IL-28B TT genotype was the only prominent predictor for SVR [odds ratio (OR), 63.60; 95% CI, 11.08–365]. The OR of serum miR-122 increased to 4.12 (95% CI, 0.64–26.58). When limited to IL-28B TT genotype patients (n = 97), multivariate analysis showed that pretreatment serum miR-122 was an independent predictor for SVR (OR, 22.86; 95% CI, 1.31–398; Table 3). If stratified by different HCV genotypes, serum miR-122 independently predicted SVR specifically in genotype 2 patients (OR, 385; 95% CI, 5.01–29,650; Table 3). To find the dynamic change of serum miR-122 during pegIFN/ RBV treatment, we selected 10 patients from the SVR group and 15 from the NR group with serial serum samples (week 0, 4, 8, 12, 24, 48, or 72 of treatment) for miR-122 quantification (Table S1). The repeated-measures mixed model for longitudinal analysis showed that the predicted values of serum miR-122 of the SVR group were significantly higher than in the NR group (P < 0.0001); however, there was no dynamic change of on-treatment serum miR-122 (Fig. 3 A and B). Validation Cohort. The baseline characteristics of 51 patients with CHC (42 SVRs and nine NRs by treatment responses and 43 TT, five GT, and three GG genotype by IL-28B analysis) in the validation cohort are shown in Table S2. The patients with SVR had a significantly greater pretreatment serum miR-122 levels compared with those with NR (9.60 vs. 9.20 log10 copies per milliliter ; P = 0.025). We also found a trend of decreasing serum miR-122 level among patients with IL-28B GT or GG genotype (TT vs. GT vs. GG, 9.57 vs. 9.40 vs. 9.28 log10 copies per millliter).

Table 1. Baseline characteristics of 126 patients with CHC who completed pegIFN/RBV therapy Genotype Characteristic Age, y Sex, male:female BMI, kg/m2 ALT, U/L GGT, U/L Total cholesterol, mg/dL Triglyceride, mg/dL LDL, mg/dL HCV RNA, log10 IU/mL HAI score Metavir score, F0:F1:F2:F3:F4 IL-28B genotype, TT:GT+GG Serum miR-122, log10 copies per milliliter SVR, % Dosage of pegIFN, % Dosage of RBV, %

All pts. (n = 126)

1 (n = 64)

2 (n = 62)

P value (genotype 1 vs. 2)

52 (12) 71:55 25.6 (3.5) 137 (102) 59 (53) 173 (34) 99 (42) 101 (36) 5.74 (0.90) 8 (3) 2:35:46:21:22 97:21 9.47 (0.44) 78 100 (2) 95 (10)

54 (12) 37:27 25.2 (3.5) 125 (82) 63 (62) 172 (34) 97 (42) 98 (33) 6.20 (0.67) 8 (3) 0:17:20:7:20 44:15* 9.40 (0.50) 72 100 (2) 94 (10)

51 (12) 34:28 26.0 (3.5) 150 (118) 54 (42) 175 (34) 102 (41) 106 (38) 5.26 (0.87) 9 (3) 2:18:26:14:2 53:6† 9.55 (0.37) 84 100 (2) 96 (9)

0.250 0.738 0.194 0.210 0.962 0.584 0.346 0.428 90% to 97%) are carried by the Ago2 protein (16, 17). These ribonucleoprotein complexes are quite resistant to the RNase degradation in the plasma for at least 2 mo (16, 17). There are two possible origins of extracellular (i.e., circulating) miR-122: a secreted product from hepatocyte or a byproduct from dead hepatocytes. The Ago2/ miRNA complex may serve for intercellular communication or have a paracrine effect; however, the recipient cells/receptors of circulating miR-122 remain unclear. Because the origin and effector target of serum miR-122 as well as its long-lasting existence in the serum remain unclear, the dynamic change of miR-122 after IFN treatment would provide additional information. SarasinFilipowicz et al. found no dynamic change of hepatic miR-122 after IFN treatment (14). Pedersen et al. showed only a transient decrease of miR-122 after IFN administration in vitro (13). In our study, we did not find a significant trend of on-treatment serum miR-122 during time-series analysis, indicating the complexity of dynamic change of serum miR-122. This is a potential limitation for the clinical usefulness of on-treatment serum miR-122. The difference in serum miR-122 between patients with NR and SVR was 1.7- to 2.2-fold, which is smaller than the fourfold difference in hepatic miR-122 found by Sarasin-Filipowicz et al. (14). This fact Su et al.

again indicates that serum miR-122 is an indirect positive marker to the liver, and that the life cycle of serum miR-122 is rather complex. Previous studies suggested a low pretreatment hepatic miR-122 level may predict NR to IFN-based therapy (14, 26) irrespective of serum or hepatic HCV RNA level (14, 25). In this study, we consistently found no correlation between hepatic or serum miR-122 and serum HCV RNA level. Of particular note, we found higher pretreatment serum miR-122 levels among patients with CHC who showed cEVR and SVR after pegIFN/RBV therapy, especially patients with IL-28B TT genotype or HCV genotype 2. As shown in previous studies (27–29), we found lower pretreatment BMI, GGT, and triglyceride levels and Metavir score, as well as higher serum miR-122 level, to predict SVR in univariate analysis. Because of the dominant predictability of IL-28B TT genotype and small sample size, the risk estimate of serum miR-122 was statistically insignificant to predict SVR (OR, 4.12; 95% CI, 0.64–26.58). We therefore restricted the analysis to IL-28B TT genotype and found that higher serum miR-122 level still predicted SVR independently (OR, 22.86; 95% CI, 1.31–398), indicating that, in this IL-28B genotype, a higher pretreatment serum miR-122 level might further predict SVR. Previous studies showed that the binding site of miR-122 is highly conserved across major HCV genotypes (11), and the miR-122 antagonism against HCV was genotype-independent (30). Our findings suggest that factors other than miR-122 are involved to predict therapeutic responses. Recently, genetic variations near the IL-28B gene have been reported to be highly associated with spontaneous and treatmentinduced HCV clearance, suggesting a regulatory role of endogenous IFN responses (31–34). The preactivation of the endogenous IFN system, but with reduced induction of IFN-stimulated genes (ISGs) during pegIFN/RBV treatment, has been found to be associated with poor virologic responses (35). IL-28B gene encodes IFN-λ3 and induces intrahepatic ISG expression via the JAKSTAT pathway: patients with unfavorable IL-28B genotype will have higher ISG expression (26, 36, 37). Sarasin-Filipowicz et al. first reported an inverse correlation between hepatic miR-122 and ISG, showing low pretreatment hepatic miR-122 levels and high ISG expression in nonresponders (14). A recent study found an ISG (NT5C3) as an miR-122 target, efficiently inhibiting miR-122 by binding and sequestering miR-122 with the 3′-UTR of its mRNA, indicating the involvement of ISGs in IFN-mediated PNAS Early Edition | 3 of 6

MEDICAL SCIENCES

Univariate analysis

miR-122 (39). The seemingly contradictory effect of miR-122 inhibition in both studies may be explained by the difference in the duration of miR-122 inhibition, which is transient during therapeutic use, but chronic in the KO mouse model. Continuous monitoring the effect and side effects of chronic miR-122 inhibition in human is warranted to solve this interesting and important issue. The present study has a few limitations. First, although our findings were interesting with potential clinical implications, the sample size of the study was relatively small. Further large-scale studies are needed for validation. Second, we used serum rather than plasma for the quantification of miR-122, and the possibility of peripheral blood mononuclear cell (PBMC) contamination cannot be excluded. However, the risk of PBMC contamination for miR-122 should be minimal because miR-122 does not exist in PBMCs (40), which was also verified in our pilot study. On the contrary, our results illustrate the potential role of serum miR-122 in the mechanistic study of host/viral interaction of HCV infection. In addition to pretreatment prediction, serum miR-122 might be

Fig. 1. Pretreatment serum miR-122 level is associated with treatment response. (A) A significantly higher pretreatment serum miR-122 level was found among patients who showed cEVR and SVR, but not RVR. (B) The pretreatment serum miR-122 level was significantly higher in patients of genotype 2 compared with genotype 1 in terms of cEVR and SVR. Box plot (25th–50th–75th quartile) with whiskers (lower and upper extremes) and outliers demonstrated.

miR-122 suppression (38). Therefore, we hypothesized that those nonresponders with unfavorable IL-28B genotypes may have a higher expression of pretreatment ISG, which sequesters and down-regulates hepatic miR-122, and thus leads to a low level of serum miR-122. Our findings that a significant trend of decreasing expression of serum miR-122 level among IL-28B unfavorable genotypes (rs8099917 GT and GG) gave support to this hypothesis. The IL-28B variation is weakly associated with treatment responses in genotype 2 patients, as well as patients receiving direct antiviral agents (37). Therefore, a non–IL-28B predictor such as pretreatment serum miR-122 level is still required to identify patients with poor responses to IFN-based therapies or even IFNfree regimens. Besides, the majority of Asian patients with CHC have favorable IL-28B genotypes; serum miR-122 may provide additional benefit for the prediction of treatment outcomes. Whether patients with favorable pretreatment miR-122 levels might benefit from a truncated therapy remains to be investigated. Anti–miR-122 oligonucleotide exerts potential anti-HCV efficacy, and the drug (miravirsen) is now in phase II trial; however, in a recent study, Hsu et al. demonstrated the deletion of mouse miR122 resulted in hepatosteatosis, hepatitis, and the development of tumors resembling HCC, suggesting a tumor-suppressor effect of 4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1306138110

Fig. 2. Association of pretreatment serum miR-122 level with IL-28B genotypes. (A) A significant trend of higher miR-122 level was found among patients with IL-28B TT genotype compared with GT and GG genotypes (9.52 vs. 9.35 vs. 8.78 log10 copies per milliliter ; P = 0.014 for trend). (B) A significantly higher serum miR-122 level was found between cEVR and SVR patients of IL-28B TT genotype but not in GT/GG genotypes. Box plot (25th– 50th–75th quartile) with whiskers (lower and upper extremes) and outliers demonstrated.

Su et al.

Table 3. Multivariate analysis of baseline predictors for SVR to pegIFN/RBV therapy in subgroups of patients

Characteristic Age >40 y vs. ≤40 y Male vs. female Genotype 1 vs. 2 BMI >25 vs. ≤25 kg/m2 GGT >52 vs. ≤52 U/L LDL >160 vs. ≤160 mg/dL Triglyceride >200 vs. ≤200 mg/dL HCV RNA (per log10 IU/mL increase) Metavir score (per unit increase) Serum miR-122 (per log10 copies per milliliter increase)

IL28B TT genotype

HCV genotype 1

HCV genotype 2

OR

95% CI

OR

95% CI

OR

95% CI

0.68 2.38 1.53 0.10 0.29 0.28 0.03 2.03 0.30 22.86

0.02–26.94 0.38–14.90 0.09–25.57 0.08–1.22 0.05–1.85 0.02–5.22 0.001–1.31 0.48–8.68 0.09–1.03 1.31–398

0.14 0.59 — 1.03 0.16 0.48 0.14 1.32 0.73 3.18

0.01–3.71 0.14–2.59 — 0.25–4.26 0.04–0.70 0.03–9.12 0.01–3.83 0.48–3.61 0.40–1.34 0.60–16.93

40.04 5.59 — * 0.70 3.18 0.35 0.74 1.39 385

1.28–1248 0.74–42.44 — — 0.09–5.31 0.08–127 0.01–24.70 0.21–2.54 0.36–5.37 5.01–29650

Reference values are as follows: GGT ≤ 52 U/L, LDL ≤ 160 mg/dL, triglyceride ≤ 200 mg/dL. *Because BMI >25 completely predicts non-SVR in this cell, BMI was omitted.

serum miR-122 level can help predict early responses and SVRs to pegIFN/RBV therapy in patients with CHC, especially in IL-28B rs8099917 TT genotype and patients with HCV genotype 2. Materials and Methods Study Population. We retrospectively included 126 patients with CHC who received pegIFN/RBV therapy at the outpatient clinics of the National Taiwan University Hospital with SVR or NR before October 2009. Another 51 patients with CHC with SVR or NR to pegIFN/RBV therapy during 2009 to 2012 were prospectively included as the validation cohort. Virologic relapsers after completion of therapy were excluded. All patients fulfilled the following criteria: age >18 y, presence of anti-HCV antibody and serum HCV RNA for >6 mo, and available pretreatment serum samples. They did not have hepatitis B virus or HIV coinfection, autoimmune liver disease, hemochromatosis, Wilson disease, neoplastic disease, organ transplantation, or a history of excess alcohol intake (>20 g/d). All patients received weekly s.c. injection of pegIFN α-2a 180 μg (Hoffmann-La Roche) or pegIFN α-2b 1.5 μg/kg (Schering-Plough) and daily weight-based oral RBV 800 to 1,200 mg (Hoffmann-La Roche or ScheringPlough) for 48 wk for HCV genotype 1 and 24 wk for genotype 2. They were then followed for an additional 24 wk off therapy. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethics Committee of the National Taiwan University Hospital. Written informed consent was obtained from each participant. Demographic data, biochemical and virologic profiles, as well as treatment responses were retrieved from the medical records. Blood samples were collected after 8 h fasting and centrifuged at 3,000 × g at 4 °C for 10 min within 1 h after venipuncture for serum collection. HCV genotype was determined by a reverse hybridization assay (Inno-LiPA HCV II; Innogenetics). Serum HCV RNA level was quantified by real-time PCR assay (Cobas TaqMan HCV Test, version 2.0; limit of detection, 25 IU/mL; Roche Diagnostics). All residual serum samples were stored at −80 °C until use. Liver biopsy was performed before treatment, and the specimen was divided into two parts. The first part was assessed by an independent pathologist who was blinded to the clinical data of patients. The second part was immediately stored in liquid nitrogen until use. Hepatic fibrosis was graded according to the Metavir system (41). The HAI was recorded according to the Knodell system (42). Definitions of Treatment Response. RVR and cEVR were defined as an undetectable serum HCV RNA level at weeks 4 and 12 of therapy, respectively. NR was defined as a persistently detectable HCV RNA level at the end of treatment. SVR was defined as persistently undetectable serum HCV RNA level at the end of treatment and 6 mo after treatment.

Fig. 3. Longitudinal analysis of serum miR-122 level during pegIFN/RBV therapy. (A) Line plot of individual serum miR-122 profiles from week 0, 4, 8, 12, 24, 48, or 72 by treatment responses: SVR vs. NR. (B) Predicted values (with 95% CI) of serum miR-122 levels during therapy by repeated-measures mixed-model analysis (P < 0.0001, SVR vs. NR). No significant dynamic change was observed after therapy.

Su et al.

Relative Quantification of Serum miR-122. Total serum RNA was extracted using the TRIzol LS reagent (Invitrogen) and the synthetic miR-39 of Caenorhabditis elegans was spiked into the denatured serum as our exogenous internal control (18). The RNA samples were mixed with the miRNA-specific stem-loop primers (TaqMan MicroRNA Assay; Applied Biosystems) and the TaqMan MicroRNA reverse transcription (RT) Kit (Applied Biosystems) and followed by RT. The RT product was relatively quantified by real-time PCR with a StepOnePlus Real-Time PCR System (Applied Biosystems).

PNAS Early Edition | 5 of 6

MEDICAL SCIENCES

used to monitor treatment responses during anti–miR-122–based therapies in the future. In summary, serum miR-122 level correlates with hepatic miR122 expression as well as liver necroinflammation, and may serve as a biomarker for liver diseases. In addition, a higher pretreatment

In our pilot study, a good linearity between the Ct value and concentration of synthetic miR-39 of C. elegans was demonstrated (R2 = 0.9997; P < 0.001); therefore, we generated a regression formula for the relative quantification of serum miR-122 level (Fig. S5). Additional information is provided in SI Materials and Methods.

included in multivariate analysis. A repeated-measures mixed model was adopted to analyze the longitudinal change of serum miR-122 levels between NR and SVR groups with the unstructured variance/covariance matrix. P values were two-tailed, and those