Sequence variation in hepatitis C virus nonstructural protein 5A predicts clinical outcome of pegylated interferon/ribavirin combination therapy

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

A substantial proportion of hepatitis C virus (HCV)-1b–infected patients still do not respond to interferon-based therapy. This study aims to explore a predictive marker for the ultimate virological response of HCV-1b–infected patients treated with pegylated interferon/ribavirin (PEG-IFN/RBV) combination therapy. Nonstructural protein 5A (NS5A) sequences of HCV in the pretreated sera of 45 patients infected with HCV-1b were analyzed. The mean number of mutations in the variable region 3 (V3) plus its upstream flanking region of NS5A (amino acid 2334–2379), referred to as IFN/RBV resistance-determining region (IRRDR), was significantly higher for HCV isolates obtained from patients who later achieved sustained virological response (SVR) by PEG-IFN/RBV than for those in patients undergoing non-SVR. The receiver operating characteristic curve analysis estimated six mutations in IRRDR as the optimal threshold for SVR prediction. Indeed, 16 (76%) of 21 SVR, but only 2 (8%) of 24 non-SVR, had HCV with six or more mutations in IRRDR (IRRDR ≥ 6) (P < 0.0001). All of 18 patients infected with HCV of IRRDR of 6 or greater examined showed a significant (≥1 log) reduction or disappearance of serum HCV core antigen titers within 24 hours after initial dose of PEG-IFN/RBV, whereas 10 (37%) of 27 patients with HCV of IRRDR of 5 or less did (P < 0.0001). The positive predictive value of IRRDR of 6 or greater for SVR was 89% (16/18; P = 0.0007), with its negative predictive value for non-SVR being 81% (22/27; P = 0.0008). Conclusion: A high degree (≥6) of sequence variation in IRRDR would be a useful marker for predicting SVR, whereas a less diverse (≤5) IRRDR sequence predicts non-SVR. (HEPATOLOGY 2008.)

Hepatitis C virus (HCV) infection is the major cause of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma in industrialized countries. However, HCV infection is curable, and its complications can be prevented by antiviral therapy.1, 2 Currently, the most effective treatment of chronic HCV infection is based on a combination of pegylated interferon (PEG-IFN) and ribavirin (RBV).3 Even with this treatment regimen, however, sustained virological response (SVR) rates for those infected with the most resistant genotypes, HCV-1a and HCV-1b, still hover at approximately 50%.3, 4 Considering the high cost and the significant side effects associated with this combination therapy, it is worthy to identify patients most likely to benefit from therapy.5 Predictors of IFN-based therapy can be classified into two categories, pretreatment and on-treatment factors. Pretreatment factors comprise host factors, such as age, sex, obesity, ethanol consumption, hepatic iron overload, fibrosis, immune responses, and coinfection with other viruses, and viral factors, which mainly include viral genotypes and viral load. On-treatment factors are mainly related to the viral kinetics within the first few weeks of treatment.6 Because the HCV genotype is one of the major factors affecting IFN-based therapy response, IFN resistance is, at least partly, genetically encoded by HCV itself.7 In this context, nonstructural protein 5A (NS5A), one of the HCV nonstructural proteins, has been widely discussed for its correlation with IFN responsiveness. Enomoto et al.8, 9 proposed that the sequence variations within a region in NS5A, called the IFN sensitivity-determining region (ISDR), is correlated with IFN responsiveness. It was further demonstrated that ISDR and its adjacent sequence was able to bind to double-stranded RNA-activated protein kinase (PKR), one of the important antiviral proteins of the host cell, to inhibit its enzymatic activity and, therefore, the combined region is called PKR-binding domain (PKR-BD).10, 11 A significant correlation between sequence variation in PKR-BD and IFN responsiveness was also reported.12 In addition, there are some reports that showed a correlation between IFN responsiveness and the sequence diversity of the variable region 3 (V3) [amino acids (aa) 2356 to 2379] or its surrounding regions near the carboxy terminus of NS5A.12–20

We have recently reported that a high degree of sequence variations in the V3 and the pre-V3 regions (aa 2334–2355) of NS5A, which we collectively refer to as IFN/RBV resistance-determining region (IRRDR) (aa 2334–2379), was closely correlated with early virological response (EVR) by week 16 in HCV-1b–infected patients treated with PEG-IFN and RBV.21 In the current study, we aimed to follow up our previous observations to investigate whether the degree of sequence variation in IRRDR could also correlate with SVR on PEG-IFN/RBV combination therapy.

Abbreviations

aa, amino acid; CI, confidence interval; CNR, complete nonresponse; ETR, end-of-treatment response; EVR, early virological response; HCV, hepatitis C virus; IFN, interferon; IRRDR, interferon/ribavirin resistance-determining region; ISDR, interferon sensitivity-determining region; NS5A, nonstructural protein 5A; nt, nucleotide; PEG, pegylated; PKR-BD, double-stranded RNA-activated protein kinase-binding domain; RBV, ribavirin; RT-PCR, reverse transcription polymerase chain reaction; SVR, sustained virological response; V3, variable 3.

Patients and Methods

Patients.

A total of 45 patients chronically infected with HCV-1b, whose diagnoses had been made based on anti-HCV antibody detection, HCV subtype determination according to the method by Okamoto et al.,22 and clinical follow-up, were treated with PEG-IFN α-2b (1.5 μg/kg body weight, once weekly, subcutaneously) and RBV (600–800 mg daily, per os), according to a standard treatment protocol for Japanese patients established by a hepatitis study group of the Ministry of Health, Labour, and Welfare, Japan, at Kobe Asahi Hospital, Hyogo Prefecture, Japan. All the patients were confirmed negative for hepatitis B surface antigen using chemiluminescent immunoassay (Abbott Japan Co., Ltd., Tokyo, Japan). Serum samples were collected from the patients at intervals of 4 weeks before, during, and after the treatment, and tested for HCV RNA by reverse transcription polymerase chain reaction (RT-PCR), as reported previously.21 The quantification of serum HCV RNA titers was performed by RT-PCR with an internal RNA standard derived from the 5′ noncoding region of HCV (Amplicor HCV Monitor test, version 2.0, Roche Diagnostics, Tokyo, Japan). The thresholds of the low-range and high-range measurements of this assay were 50 and 600 IU/mL, respectively. HCV core antigen in the sera was also quantitatively measured by chemiluminescent immunoassay (Abbott Japan Co., Ltd., Tokyo, Japan). The threshold of this assay is 20 fmol/L.

The study protocol was approved beforehand by the Ethic Committee in Kobe Asahi Hospital, and written informed consent was obtained from each patient before the treatment.

NS5A Sequence Analysis.

HCV RNA was extracted from 140 μL serum using a commercially available kit (QIAmp viral RNA kit; QIAGEN, Tokyo, Japan). For amplification of the NS5A region of the HCV genome, the extracted RNA was reverse transcribed and amplified for full-length NS5A using SuperScript One-step RT-PCR for long templates (Invitrogen, Tokyo, Japan) and a set of primers, NS5A-F1 [5′-TACTCCCTGCCATCCTCTCTCCTG-3′; sense, nucleotides (nt) 5974–5997] and NS5A-F2 (5′-CTCCTTGAGCACGTCCCGGT-3′; antisense, nt 7777–7796). The resultant RT-PCR product was subjected to a second-round PCR by using Platinum Taq DNA polymerase (Invitrogen) and an inner set of primers, NS5A-F3 (5′TCTCCAGCCTTACCATCACYCA-3′; sense, nt 6172–6193) and NS5A-F4 (5′-CGGTARTGRTCGTCCAGGAC-3′; antisense, nt 7761–7780). The samples that were not amplifiable (nos. 3, 23, 47, 61, 65, and 69) using the aforementioned primers were amplified using primer sets reported previously.23 Reverse transcription was performed at 45°C for 30 minutes and terminated at 94°C for 2 minutes, followed by the first-round PCR over 35 cycles, with each cycle consisting of denaturation at 94°C for 30 seconds, annealing at 55°C for 30 seconds and extension at 68°C for 90 seconds. The second-round PCR was performed under the same condition. The amplified fragments were purified with QIA quick PCR purification kit (QIAGEN), and visualized by agarose gel electrophoresis and ethidium bromide staining. The sequences of the amplified fragments were determined by direct sequencing without subcloning using Big Dye Deoxy Terminator cycle sequencing kit and ABI 337 DNA sequencer (Applied Biosystems, Inc, Japan). The aa sequences were deduced and aligned using GENETYX Win software version.7.0 (GENETYX Corp., Tokyo, Japan). Numbering of aa throughout the complete manuscript is according to the poly protein of HCV genotype 1b prototype HCV-J.24

Statistical Analysis.

Statistical difference in the parameters, including all available patients' demographic, biochemical, hematological, and virological data as well as IRRDR sequence variations factors, was determined between different patients' groups by Student t test for numerical variables, and Fisher's exact probability test for categorical variables. In the case of multiple comparisons for various regions of NS5A, P values were adjusted by the Bonferroni method to reduce the probability of erroneously classifying nonsignificant hypothesis as significant. Although there are five regions of comparison (full-NS5A, N-half, ISDR, PKR-BD and IRRDR), the ISDR is entirely within the PKR-BD, and all the regions fall within the full-NS5A. Therefore, it would be reasonable to adjust the P values for three regions of comparison. Accordingly, the P value for a test was multiplied by 3. To evaluate the optimal threshold of IRRDR mutations for SVR prediction, the receiver operating characteristic curve was constructed and the area under the curve as well as the sensitivity and specificity were calculated. Subsequently, univariate and multivariate logistic analyses were performed to identify variables that independently predict SVR. The odds ratios and 95% confidence intervals (95% CI) were also calculated. Kaplan-Meier HCV survival curve analysis was performed based on serum HCV-RNA positivity data during treatment period (48 weeks) according to the number of IRRDR mutations (IRRDR ≥ 6 and IRRDR ≤ 5). The HCV death event was estimated as the first time point of HCV-RNA clearance after initiation of the treatment. The data obtained were evaluated by the log-rank test. Positive and negative predictive values of SVR predictors were computed, and their significance levels were evaluated using the sign test. All statistical analyses were performed using the SPSS version 16 software (SPSS Inc., Chicago, IL). Unless otherwise stated, a P value of less than 0.05 was considered statistically significant.

Nucleotide Sequence Accession Numbers.

The sequence data reported in this article have been deposited in the DDBJ/EMBL/GenBank nucleotide sequence databases with the accession numbers AB285035 through AB285081, and AB354116 through AB354118.

Results

Virological Responses of the Patients Treated with PEG-IFN and RBV.

Proportions of various virological responses of the patients treated with PEG-IFN/RBV combination therapy are shown in Table 1. Of 45 patients enrolled in this study, 23 (51%), 31 (69%), and 21 (47%) patients, respectively, achieved EVR by week 12 [EVR(12w)], end-of-treatment response (ETR), and sustained virological response (SVR). Among 23 patients with EVR(12w), 22 (96%) and 18 patients (78%) achieved ETR and SVR, respectively. This indicates that EVR(12w) was significantly correlated with ETR and SVR (P < 0.0001). A total of 24 patients (53%) failed to achieve SVR, and they were referred to as non-SVR. Non-SVR can be divided into two categories: (i) complete nonresponse (CNR), which is defined by continued presence of serum HCV RNA up to the end of the treatment, and (ii) relapse, which is defined by transient disappearance of HCV RNA at a certain time point followed by reappearance of HCV RNA either before or after the end of the treatment. CNR represented 24% (11/45) of all cases and 46% (11/24) of non-SVR. Thirteen (29%) of 45 patients underwent relapse. Among 13 relapsers, 3 (23%) patients had rebound in HCV viremia before the end of the treatment and, hence, were defined as undergoing viral breakthrough, whereas 10 (77%) patients had rebound in HCV viremia after the end of the treatment, defined as ETR-relapsers.

Table 1. Proportions of Various Virological Responses of Patients Treated With PEG-IFN/RBV
Virological ResponseProportion
  • *

    No. of patients/no. of total.

  • Abbreviations: PEG-IFN/RBV, pegylated-interferon/ribavirin; EVR, early virological response; ETR, end-of-treatment response; SVR, sustained virological response; CNR, complete nonresponse.

EVR(12w)51% (23/45)*
ETR69% (31/45)
SVR47% (21/45)
Non-SVR53% (24/45)
CNR24% (11/45)
Relapse29% (13/45)
ETR-relapse22% (10/45)
Viral breakthrough7% (3/45)

Demographic characteristics of patients with SVR, non-SVR, CNR, and relapse are summarized in Table 2. Age, sex, body weight, hemoglobin levels, or gamma guanosine triphosphate titers did not significantly differ between SVR and non-SVR or CNR. However, patients with SVR showed a trend toward having significantly higher platelet counts than those with non-SVR and CNR. Also, the mean initial titers of HCV core antigen for non-SVR and CNR, respectively, were 1.6 times and 2.3 times higher than that for SVR, although the difference was not statistically significant. HCV RNA titers were almost the same among them.

Table 2. Demographic Characteristics of Patients With SVR, Non-SVR, CNR, and Relapse
FactorSVRNon-SVRCNRRelapseP Value
SVR versus Non-SVRSVR versus CNRSVR versus Relapse
  • *

    Mean ± SD.

  • Not significant.

  • Student t test.

  • Abbreviations: SVR, sustained virological response; CNR, complete nonresponse; γ-GTP, gamma guanosine triphosphate.

Age56.5 ± 8.0*59.9 ± 10.659.4 ± 10.060.3 ± 11.5NSNSNS
Sex (male/female)12/913/116/57/6NSNSNS
Body weight (kg)58.5 ± 9.459 ± 13.261.0 ± 10.857.8 ± 15.3NSNSNS
Platelets (× 104/mm3)18.3 ± 4.415.0 ± 4.912.3 ± 3.916.8 ± 4.90.020.001NS
Hemoglobin (g/dL)14.1 ± 1.313.7 ± 1.414.4 ± 1.314.3 ± 1.5NSNSNS
γ-GTP (IU/L)43.5 ± 28.751.6 ± 35.762.8 ± 40.343.8 ± 30.5NSNSNS
HCV-RNA (KIU/mL)1326 ± 12561667 ± 13111488 ± 12281818 ± 1408NSNSNS
HCV core antigen (fmol/L)6183 ± 68949830 ± 121414,033 ± 17,0896481 ± 4023NSNSNS

Correlation Between Virological Responses and the Sequence Variation of IRRDR of HCV NS5A Obtained from the Pretreated Sera.

The entire NS5A region of the HCV genome was amplified from the pretreated sera and the aa sequences deduced. We compared each NS5A sequence with a consensus sequence inferred from aligning the previously published NS5A-1b sequences.8 In this connection, the consensus sequence for IRRDR differs from the corresponding sequence of a prototype strain of IFN resistance HCV-1b (HCV-J; DDBJ/EMBL/Genbank accession no. D90208) by a single residue at position 2367 (Ala instead of Gly). Because Ala2367 was conserved in 95% of the reported sequences, we used the IRRDR consensus sequence in this study. As shown in Table 3, the mean number of aa substitutions in the entire NS5A obtained from patients with SVR was significantly greater compared with non-SVR and relapse. There was no difference in the number of mutations in an N-terminal half of NS5A (aa 1972–2208), the ISDR (aa 2209–2248) or the PKR-BD (aa 2209–2274) between the different patients' groups. Conversely, we found a more obvious significant difference in the mean numbers of aa mutations within a region consisting of the pre-V3 and V3 regions, which we refer to as IRRDR, between SVR and other patients' groups (Table 3).

Table 3. Average Numbers of aa Mutations Within Different Regions of HCV NS5A Obtained From Pretreated Sera of Patients With SVR, Non-SVR, CNR, and Relapse
NS5A RegionNo. of Mutations*P Value
SVRNon-SVRCNRRelapseSVR versus Non-SVRSVR versus CNRSVR versus Relapse
  • *

    Mean ± SD.

  • The P values obtained with Student t test were adjusted using the Bonferroni method (see Materials and Methods).

  • Not significant.

  • Abbreviations: SVR, sustained virological response; CNR, complete nonresponse; aa, amino acid; ISDR: interferon sensitivity-determining region; PKR-BD, double-stranded RNA-activated protein kinase-binding domain; IRRDR, interferon/ribavirin resistance-determining region.

Full-NS5A (aa 1972–2419)24.9 ± 6.1*19.7 ± 4.320.1 ± 5.219.4 ± 3.50.0120.1440.03
N-half (aa 1972–2208)9.2 ± 1.98.6 ± 1.99.0 ± 2.48.2 ± 1.2NSNSNS
ISDR (aa 2209–2248)2.1 ± 2.81.2 ± 1.11.7 ± 1.40.8 ± 0.7NSNSNS
PKR-BD (aa 2209–2274)3.8 ± 3.42.5 ± 2.02.9 ± 2.42.1 ± 1.5NSNSNS
IRRDR (aa 2334–2379)6.1 ± 2.13.9 ± 1.43.7 ± 0.94.0 ± 1.80.00060.0030.018

To estimate a cutoff number of mutations in IRRDR predicting SVR, the receiver operating characteristics analysis was performed. The result revealed that six mutations were an optimal number of mutations to predict SVR, because it achieved the highest sensitivity (76%) combined with the highest specificity (92%) and yielded an area under the curve of 0.81 (Fig. 1).

Figure 1.

The receiver operating characteristic curve analysis of IRRDR sequence variation for SVR prediction. The curve depicted with solid line shows an area under the curve of 0.81. Solid circles with numerals plotted on the curve represent different numbers of IRRDR mutations analyzed. The dashed line in the upper left corner indicates the optimal number of IRRDR mutations for SVR prediction, which yields the highest sensitivity (76%) and the highest specificity (92%).

Indeed, only 2 (8%) of 24 patients with non-SVR, in contrast to 16 (76%) of 21 patients with SVR, had HCV with IRRDR of 6 or greater, with the difference between the two groups being statistically significant (P < 0.0001) (Table 4). Furthermore, none of 11 patients with CNR had HCV of IRRDR of 6 or greater, and the difference between SVR and CNR was statistically significant (P < 0.0001). Similarly, only 2 (15%) of 13 relapsers (10 ETR-relapsers + 3 patients with viral breakthrough) had HCV of IRRDR greater than or equal to 6, with the result demonstrating significant difference between SVR and relapse (P = 0.001).

Table 4. Correlation Between NS5A Sequence Variation and Virological Responses of the Patients
CriteriaNo. of Subjects/no. of Total*P Value
SVRNon-SVRCNRRelapseSVR versus Non-SVRSVR versus CNRSVR versus Relapse
  • *

    Total no. of SVR, Non-SVR, CNR, or relapse.

  • Fisher's exact test.

  • Not significant.

  • Abbreviations: SVR, sustained virological response; CNR, complete nonresponse; IRRDR, interferon/ribavirin resistance-determining region; Ala2360, alanine at position 2360; Thr2378, threonine at position 2378.

IRRDR ≥ 616/21 (76%)2/24 (8%)0/11 (0%)2/13 (15%)< 0.0001< 0.00010.001
Ala236010/21 (48%)3/24 (13%)0/11 (0%)3/13 (23%)0.020.006NS
Thr23789/21 (43%)3/24 (13%)0/11 (0%)3/13 (23%)0.040.01NS

When the IRRDR sequences obtained from all 45 patients were aligned along with the consensus sequence (Fig. 2), we noticed that 10 (48%) of 21 patients with SVR had alanine at position 2360 (Ala2360), whereas only 3 (13%) of 24 patients with non-SVR and none of 11 patients with CNR did (P = 0.02 and 0.006, respectively) (Table 4). Similarly, 9 (43%) of 21 patients with SVR had threonine at position 2378 (Thr2378), whereas only 3 (13%) of 24 patients with non-SVR and none of 11 patients with CNR did (P = 0.04 and 0.01, respectively).

Figure 2.

Sequence alignment of IRRDR (Pre-V3 plus V3 regions) of NS5A of HCV-1b obtained from the pretreatment sera. The consensus sequence (Cons) is shown on the top. The Cons sequence differs from a prototype sequence of IFN-resistant HCV-1b (HCV-J; DDBJ/EMBL/GenBank accession no. D90208) by a single residue at position 2367 (alanine instead of glycine). The numbers along the sequence indicate aa positions. Dots indicate residues identical to those of the Cons sequence. Ala2360 and Thr2378are written in boldface.

To identify significant independent SVR predictors, we first entered all available baseline patients' features and IRRDR sequence variations data in univariate logistic analysis. As had been expected, this analysis yielded four factors significantly associated with SVR: IRRDR mutations, either continuous variable (P < 0.0001) or dichotomized at 6 (P < 0.0001), Ala2360 (P = 0.002), Thr2378 (P = 0.019), and platelet count (P = 0.017). Subsequently, we analyzed these four factors by multivariate logistic analysis. When the IRRDR mutations were dichotomized at 6, the multivariate analysis identified only the IRRDR of 6 or greater criterion as the independent predictor of SVR (odds ratio = 16.0; CI, 2.4–104.3; P = 0.004) (Table 5). However, when the IRRDR mutations were analyzed as a continuous variable, the multivariate analysis yielded IRRDR mutations (odds ratio = 1.8; CI, 1.1–3.1; P = 0.02) and Ala2360 (odds ratio = 9.3; CI, 1.1–78.8; P = 0.04) as independent SVR predictors.

Table 5. Multivariate Logistic Regression Analysis to Identify Independent SVR Predictors
FactorOdds (95% CI)P value
  1. Only factors that were significantly associated with SVR in univariate analysis were included in multivariate logistic regression analysis.

  2. Abbreviations: IRRDR, interferon/ribavirin resistance-determining region; Ala2360, alanine at position 2360; Thr2378, threonine at position 2378, CI; confidence interval.

Multivariate analysis 1  
 IRRDR ≥ 616.0 (2.4–104.3)0.004
 Ala23607.1 (0.8–66.8)0.09
 Thr23784.1 (0.5–30.9)0.17
 Platelets1.1 (0.9–1.4)0.27
Multivariate analysis 2  
 IRRDR mutations as a continuous variable1.8 (1.1–3.1)0.02
 Ala23609.3 (1.1–78.8)0.04
 Thr23784.9 (0.7–33.3)0.1
 Platelets1.2 (1.0–1.5)0.08

Figure 3A shows the viral clearance rates of patients infected with HCV of IRRDR of 6 or greater and those with IRRDR of 5 or less at 4-week intervals during the whole observation period (72 weeks). All of 18 patients infected with HCV of IRRDR 6 or greater cleared the virus by week 16 and remained free of viremia thereafter until the end of the PEG-IFN/RBV treatment (week 48). Within 4 weeks after the cessation of the combination therapy, however, 2 (11%) of the 18 patients underwent relapse (ETR relapse). Conversely, 16 (59%) of the 27 patients with HCV of IRRDR of 5 or less cleared the virus by week 32. Of the 16 patients who once cleared the virus, 3 (19%) and 8 (50%) underwent relapse to become viral breakthrough and ETR relapsers, respectively.

Figure 3.

Time course of HCV clearance during IFN/RBV treatment and follow-up period. (A) The viral clearance rates of patients infected with HCV isolates with six or more mutations in IRRDR (IRRDR ≥ 6) or five or fewer mutations (IRRDR ≤ 5) at 4-week intervals during the whole observation period. *P < 0.01; †, P < 0.001 (Fisher's exact probability test). (B) Kaplan-Meier HCV survival curve analysis based on serum HCV-RNA positivity during IFN/RBV treatment course (48 weeks) for HCV isolates with IRRDR of 6 or greater and IRRDR of 5 or fewer. P < 0.0001 (log-rank test).

Kaplan-Meier HCV survival curve analysis confirmed that, after the initiation of the IFN/RBV treatment, HCV clearance was achieved significantly more rapidly in patients infected with HCV isolates with IRRDR of 6 or greater than those with IRRDR of 5 or less, with the difference between the two groups being statistically significant (P < 0.0001) (Fig. 3B).

Sequence Analysis of ISDR and PKR-BD of HCV NS5A Obtained from Pretreated Sera.

As described, there was no difference in the mean number of mutations in ISDR or PKR-BD between SVR and non-SVR or CNR (Table 3). Only four patients had HCV with four or more mutations in ISDR (data not shown), the criterion for IFN-sensitive HCV strains according to Enomoto et al.8, 9 Although there appeared to be a trend for patients with HCV having four or more mutations in ISDR toward SVR (3 of 4), the difference was not statistically significant. Also, the prevalence of HCV with four or more mutations in ISDR was not significantly different between SVR (3 of 21; 14.3%) and non-SVR (1 of 24; 4.2%). It would be interesting to note, however, that all three HCV strains with four or more mutations in ISDR obtained from SVR (nos. 10, 65, and 72) had HCV of IRRDR of 6 or greater, whereas the only one strain with four or more mutations in ISDR from non-SVR (no. 13) had three mutations in IRRDR (data not shown). It is thus possible that the IRRDR sequence variation is associated with PEG-IFN/RBV responsiveness more closely than is the ISDR variation.

Correlation Between Rapid Reduction of HCV Core Antigen Titers and the Sequence Variation in IRRDR of HCV NS5A Obtained from the Pretreated Sera.

As stated before, there was no significant difference in the mean values of initial HCV core antigen titers between patients with SVR and those with non-SVR (Table 2). However, we observed a strong association of SVR with rapid reduction of HCV core antigen titers during the very early stages of treatment, that is, 24 hours and 1, 2, and 4 weeks after the initiation of treatment (data not shown). Therefore, we analyzed whether the degree of sequence variation in IRRDR correlated with the very rapid reduction (24 hours after the first dose of PEG-IFN/RBV) of HCV core antigen titers. The result obtained clearly demonstrated a significant correlation between IRRDR of 6 or greater and the very rapid reduction of HCV core antigen titers 24 hours and 1, 2, and 4 weeks after the initiation of treatment (Table 6). Most notably, all 18 patients infected with HCV isolates of IRRDR of 6 or greater achieved significant (≥1 log) reduction or disappearance of serum HCV core antigen titers 24 hours after the first dose of PEG-IFN/RBV.

Table 6. Significant Correlation Between the Rapid Reduction of HCV Core Antigen Titers and IRRDR Sequence Variations
CriteriaNo. of Patients With Significant Reduction of HCV Core Antigen Titers/No. of Total
24 Hours* (≥ 1 log)P Value1 Week* (≥ 1 log)P Value2 Weeks* (≥ 1.5 log)P Value4 Weeks* (≥ 2 log)P Value
  • *

    Period after initiation of IFN/RBV therapy.

  • Criteria of the significant reduction of HCV core antigen titers. Two (both at 24 hours and 1 week) and three patients (both at 2 and 4 weeks) who achieved disappearance of serum HCV core antigen were also considered to meet these criteria.

  • Fisher's exact test.

  • Abbreviations: IRRDR, interferon/ribavirin resistance-determining region; IFN/RBV, interferon/ribavirin.

IRRDR ≥ 618/18< 0.000115/180.00213/180.01616/180.0007
IRRDR ≤ 510/27 9/27 9/27 10/27 

Proposed Markers for Prediction of Various Virological Responses During PEG-IFN/RBV Combination Therapy.

As described, IRRDR of 6 or greater and Ala2360 were statistically selected as independent SVR predictors. Therefore, we aimed to assess their predictability, in terms of positive and negative predictive values, for various virological responses to PEG-IFN/RBV combination therapy (Table 7). IRRDR greater than or equal to 6 could predict EVR(12w), ETR, and SVR with the positive predictive values of 78% (P = 0.01), 100% (P = 0.000007), and 89% (P = 0.0007), respectively. Moreover, the negative predictive value of IRRDR of 6 or greater for non-SVR was 81% (P = 0.0008). Thus, IRRDR of 6 or greater would be useful to predict not only SVR but also non-SVR. Similarly, Ala2360 could also predict ETR and SVR with positive predictive values of 92% (P = 0.002) and 77% (P = 0.046), respectively.

Table 7. Positive Predictive Value, Negative Predictive Value, Sensitivity and Specificity of IRRDR ≥ 6 and Ala2360 on the Likelihood of Achieving Various Virological Responses
Virological ResponseIRRDR ≥ 6Ala2360
PPVNPVSensitivitySpecificityPPVNPVSensitivitySpecificity
  1. Abbreviations: IRRDR, interferon/ribavirin resistance-determining region; Ala2360, alanine at position 2360; EVR, early virological response; ETR, end-of-treatment response; SVR, sustained virological response; PPV, positive predictive value; NPV, negative predictive value.

EVR (12W)78% (14/18)67% (18/27)61% (14/23)82% (18/22)69% (9/13)56% (18/32)39% (9/23)82% (18/22)
ETR100% (18/18)56% (15/27)58% (18/31)100% (14/14)92% (12/13)41% (13/32)39% (12/31)93% (13/14)
SVR89% (16/18)81% (22/27)76% (16/21)92% (22/24)77% (10/13)66% (21/32)48% (10/21)88% (21/24)

Discussion

A substantial proportion of HCV-1b–infected patients do not respond to IFN/RBV combination therapy. Given the significant side effects and high cost associated with this combination therapy, it would be of great utility if clinicians could predict, either before or during the treatment, which patients would, or would not, achieve SVR. Useful predictors of SVR must have a high positive predictive value; conversely, useful predictors of non-SVR must have high negative predictive value.5 Most recent studies have focused on the possible correlation between the likelihood of achieving SVR and viral clearance kinetics during the first few months of the treatment.25, 26 Conversely, some studies dealt with the possible correlation between SVR and sequence variation within a part of NS5A, especially the V3 region.12–20

We previously reported that a high degree of sequence variation (≥6 mutations) in IRRDR was significantly correlated with the EVR by week 16 in HCV-1b–infected patients treated with PEG-IFN/RBV combination therapy.21 In the current follow-up study, we aimed to investigate whether the IRRDR sequence variation is correlated also with SVR. By using different statistical approaches, the results obtained clearly demonstrated that the high degree of sequence variation in IRRDR (IRRDR ≥6) significantly correlated with SVR, whereas the low degree of sequence variation in this region (IRRDR ≤5) correlated with non-SVR. Nearly two-thirds of patients with SVR had HCV of IRRDR of 6 or greater, whereas only 2 (8%) of 24 patients with non-SVR did (P < 0.0001) (Table 4). Notably, 16 of the 18 patients infected with HCV of IRRDR of 6 or greater achieved SVR. Accordingly, the positive predictive value and negative predictive value of IRRDR greater than or equal to 6 for SVR and non-SVR were 89% (P = 0.0007) and 81% (P = 0.0008), respectively (Table 7). Our current results thus strongly suggest that IRRDR greater than or equal to 6 would be a useful marker for prediction of SVR.

It was reported that the determination of HCV core antigen levels in the serum was an accurate and reliable alternative to monitor HCV RNA titers and that rapid reduction of HCV core antigen levels within a few weeks after the initiation of the therapy could predict treatment outcome in patients receiving PEG-IFN/RBV combination therapy.27–29 Indeed, we found a strong association between the likelihood of achieving SVR and rapid reduction of HCV core antigen during the first 4 weeks of PEG-IFN/RBV combination therapy. More importantly, we found a significant correlation between the rapid reduction of HCV core antigen titers and the degree of sequence variation in IRRDR. Notably, all the patients infected with HCV of IRRDR greater than or equal to 6 showed a significant (≥1 log) reduction or disappearance of serum HCV core antigen titers 24 hours after the first dose of PEG-IFN/RBV (Table 6). This, in particular, suggests a possible influence of IRRDR of 6 or greater on HCV replication kinetics during IFN-based therapy because the direct effect of IFN begins a few hours after the first dose. Moreover, IRRDR greater than or equal to 6 was significantly associated with rapid clearance of serum HCV RNA as early as week 8 during PEG-IFN/RBV combination therapy (Fig. 3). These results collectively reinforce the possible correlation between the sequence variation in IRRDR and HCV clearance by the IFN-based therapy.

We also examined whether the criterion of IRRDR of 6 or greater was applicable to previously reported studies, for which information on both treatment outcome (responder versus nonresponder) and IRRDR sequences are available.8, 13 As shown in Table 8, the average numbers of amino acid variations from the same consensus sequence used in the current study were significantly larger for SVR than for non-SVR in a study with Japanese patients (P = 0.003)8 and hovered at nearly a significant level in a study with European patients (P = 0.06).13 More importantly, the criterion of IRRDR greater than or equal to 6 could significantly differentiate between responders and nonresponders in the Japanese study (P = 0.003) and also hovered at nearly a significant level in the European study (P = 0.058). It should be noted that, in the latter study, there were only three patients who had HCV with IRRDR of 6 or greater, all of whom became SVR. Taken together, these results suggest the useful application of IRRDR of 6 or greater as an SVR marker even in different geographical regions, although the prevalence of HCV with IRRDR of 6 or greater may vary with different regions of the world.

Table 8. Comparative Analysis of the Mean Numbers of aa Mutations in IRRDR in Previously Reported Japanese and European Studies
StudyFactorSVRNon-SVRP Value
  • NOTE. Same consensus sequence was used in this comparative analysis.

  • *

    Mean ± SD.

  • Student t test.

  • Fisher's exact test.

  • Abbreviations: aa, amino acid; IRRDR, interferon/ribavirin resistance-determining region; SVR, sustained virological response.

Enomoto et al.8No. of IRRDR mutations7.6 ± 2.6*3.7 ± 2.00.003
 No. of patients with IRRDR ≥ 6810.003
 IRRDR ≤ 518 
Duverlie et al.13No. of IRRDR mutations5.6 ± 2.34.2 ± 0.80.06
 No. of patients with IRRDR ≥ 6300.056
 IRRDR ≤ 5511 
This studyNo. of IRRDR mutations6.1 ± 2.13.9 ± 1.40.0006
 No. of patients with IRRDR ≥ 6162< 0.0001
 IRRDR ≤ 5522 

Although we observed significant correlation between the overall number of mutations in IRRDR and PEG-IFN/RBV responsiveness, we also found particular amino acid mutations, Ala2360 and Thr2378, that were significantly associated with SVR (Table 4 and Fig. 2). In particular, Ala2360 was identified as an independent SVR marker. In this connection, it should be noted that four of five HCV isolates of IRRDR of 5 or less obtained from patients with SVR had either Ala2360 or Thr2378. Furthermore, 20 (95%) of 21 HCV isolates obtained from patients with SVR had either one of the three factors (IRRDR ≥6, Ala2360, or Thr2378). To our knowledge, there is no known CD4 or CD8 epitope(s) in IRRDR so far reported. Interestingly, however, Neumann-Haefelin et al.30 recently identified an HLA-A26 CD8+ T-cell epitope located at position 2416, 37 aa distant from IRRDR. This epitope was shown to be targeted in all patients with acute resolving HCV infection examined. Further studies are needed to elucidate the role(s) for the distal carboxy terminal region of NS5A, including IRRDR, in both IFN/RBV responsiveness and T cell–mediated virus clearance.

In conclusion, our results suggest that a high degree of sequence variation in IRRDR (IRRDR ≥6), and a particular aa mutation (Ala2360) to a lesser extent, would be a useful marker to predict SVR.

Ancillary