Hepatitis C virus (HCV) is the most common blood-borne pathogen of chronic liver diseases in Japan andU.S.A. (17). It establishes persistent infection thatoften results in chronic hepatitis followed by liver cir-rhosis, from which hepatocellular carcinoma arises(18). HCV is an RNA virus, which belongs to the Fla-viviridae family. Based on the nucleotide sequencediversity of the viral genome, HCV is classified into sixgenotypes, HCV-1 to -6, with each genotype being fur-ther subdivided into a number of subtypes, such as

HCV-1a and -1b (38, 39). Because of a considerable lag time between the onset

of infection and clinical manifestation, the prevalence ofHCV-related liver disease is still increasing (17) . Thus,effective treatment is essential. There are three inde-pendent factors for successful treatment of chronicHCV infection: (i) the virus must be sensitive to treat-ment, (ii) all infected cells must be eliminated, and (iii)the patients must comply with therapy regimens (7). Inthis context, global consensus obtains that a combina-

Editor-Communicated Paper

Prediction of Efficient Virological Response toPegylated Interferon/Ribavirin Combination Therapyby NS5A Sequences of Hepatitis C Virus and Anti-NS5A Antibodies in Pre-Treatment Sera

Ahmed El-Shamy1, Mikiko Sasayama1, Motoko Nagano-Fujii1, Noriko Sasase2, Susumu Imoto2, Soo Ryang Kim2, and Hak Hotta*,1

1Division of Microbiology, Kobe University Graduate School of Medicine, Kobe, Hyogo 650–0017, Japan, and 2Division ofGastroenterology, Kobe Asahi Hospital, Kobe, Hyogo 653–0801, Japan

Communicated by Dr. Yoshiharu Matsuura: Received February 13, 2007. Accepted February 17, 2007

Abstract: A considerable number of patients infected with Hepatitis C virus subtype 1b (HCV-1b) do notrespond to pegylated interferon/ribavirin combination therapy. In this study we explored a useful factor(s)to predict treatment outcome. A total of 47 HCV-1b-infected patients were treated with pegylated inter-feron/ribavirin for 48 weeks. Sera of the patients were examined for the entire NS5A sequence of theHCV genome, HCV RNA titers and anti-NS5A antibodies. According to their responses, the patients weredivided into two groups, early viral responders who cleared the virus by week 16 (EVR[16w]) and thosewho did not (Non-EVR[16w]). The mean number of mutations in the V3 region (aa 2356 to 2379) or that inthe V3 region plus its N-terminally flanking region, which we refer to as interferon/ribavirin resistance-determining region (IRRDR; aa 2334 to 2379), of NS5A obtained from the pretreatment sera was signifi-cantly larger for EVR(16w) compared with Non-EVR(16w). Moreover, HCV-1b isolates with ≥5 mutationsin V3 or those with ≥6 mutations in IRRDR were almost exclusively found in EVR(16w). Also, the presenceof detectable levels of anti-NS5A antibodies in the pretreatment sera was closely associated withEVR(16w). In conclusion, a high degree of sequence variation in V3 (≥5) or IRRDR (≥6) and the presenceof detectable levels of anti-NS5A antibodies in the pretreatment sera would be useful factors to predictEVR(16w). On the other hand, a less diverse sequence in V3 (≤4) or IRRDR (≤5) together with theabsence of detectable anti-NS5A antibodies could be a predictive factor for Non-EVR(16w).

Key words: Early viral responder (EVR), Non-EVR, Variable (V) 3 region, IFN/RBV resistance-deter-mining region (IRRDR)

471

Microbiol. Immunol., 51(4), 471–482, 2007

*Address correspondence to Dr. Hak Hotta, Division ofMicrobiology, Kobe University Graduate School of Medicine,7–5–1 Kusunoki-cho, Chuo-ku, Kobe, Hyogo 650–0017, Japan.Fax: �81–78–382–5519. E-mail: hotta@kobe-u.ac.jp

Abbreviations: aa, amino acid; EVR, early viral responder;HCV, hepatitis C virus; IFN, interferon; IRRDR, interferon/rib-avirin resistance-determining region; NCR, non coding region;NS, non-structural region; PBS, phosphate buffered saline; PEG,pegylated; RBV, ribavirin; V3, variable region 3.

tion of interferon (IFN) or pegylated IFN (PEG-IFN)plus ribavirin (RBV) is the treatment of choice forchronic hepatitis C. Even with this treatment regimen,however, sustained virological response rates for thoseinfected with the most resistant genotypes, HCV-1a and-1b, still hover at ~50% (11, 22). Thus, it is worthy toidentify the predictive factors that allow the selectionof patients who will achieve the eradication of HCVRNA either before or during treatment, becauseIFN/RBV combination therapy is more expensive andhas several side effects (29).

Predictors of IFN-based therapy can be classifiedinto two categories, pre-treatment and on-treatment fac-tors. Pre-treatment factors comprise of host factors,such as age, gender, obesity, ethanol consumption,hepatic iron overload, fibrosis, immune responses andco-infection with other viruses, and viral factors thatmainly include viral genotypes, particular amino acid(aa) sequence variation within a given genotype andviral load. On-treatment factors are mainly related tothe viral kinetics within the first few weeks of treatment(7).

As for the particular aa sequence variation within agiven HCV genotype, the IFN sensitivity-determiningregion (ISDR) in the nonstructural protein 5A (NS5A)was first proposed by Enomoto et al. (5, 6). It was fur-ther demonstrated that ISDR and its adjacent sequencewas able to bind to double-stranded RNA-activatedprotein kinase (PKR), one of the important antiviralproteins of the host cell, to inhibit its enzymatic activityand, therefore, the combined region is called PKR-bind-ing domain (PKR-BD) (12, 13). A significant correla-tion between sequence variation in PKR-BD and IFNresponsiveness was also reported (31). However, todate, there is a big controversy surrounding the ISDRtheory, as some studies supported it (1, 10, 26) and oth-ers did not (4, 27, 34). We have demonstrated in clinicaland experimental settings that NS5A plays a certainrole(s) in determining IFN resistance and HCV RNAtiters in both ISDR-dependent and -independent man-ners (21, 40, 41). Also, there are some reports thatshowed a correlation between IFN responsiveness andthe sequence diversity of the V3 region or its surround-ing regions near the carboxy terminus of NS5A (4, 19,27, 31, 35, 37, 44). Moreover, a significant correlationbetween IFN responsiveness and the presence of anti-HCV NS5A antibodies in the pre-treatment sera wasobserved for patients chronically infected with HCV (9,10); however, this observation is also controversial (36,44). Thus, despite the uncertainties and still unknownimportance of this viral protein in determining the IFNsensitivity, NS5A does warrant investigation to betterunderstand the possible mechanism of virus-mediated

IFN resistance and also to find a predictive factor(s) forefficient viral clearance by IFN. In this study we aimedto define molecular markers within different regions ofNS5A that can predict efficient viral clearance duringPEG-IFN/RBV therapy. We also examined whetheranti-NS5A antibodies in the pre-treatment sera couldpredict efficient viral clearance.

Materials and Methods

Patients. A total of 47 patients chronically infectedwith HCV-1b, whose diagnosis had been made basedon anti-HCV antibody detection, HCV subtype deter-mination according to the method by Okamoto et al.(32) and clinical follow-up, were treated with PEG-IFNα-2b (1.5 µg per kilogram body weight, once weekly,subcutaneously) and RBV (600–800 mg daily, per os),according to a standard treatment protocol for Japanesepatients established by a hepatitis study group of theMinistry of Health, Labour and Welfare, at Kobe AsahiHospital, Hyogo Prefecture, Japan. All the patientswere confirmed negative for HBsAg using chemilumi-nescent immunoassay (Abbott Japan Co., Ltd., Tokyo).Before and during the treatment, serum samples werecollected from the patients and tested for HCV RNA byreverse transcription (RT)-PCR at intervals of 4 weeks.The quantification of serum HCV RNA titers was per-formed by RT-PCR with an internal RNA standardderived from the 5' non-coding region (5' NCR) ofHCV (Amplicor HCV Monitor test, version 2.0, RocheDiagnostics, Tokyo). The thresholds of the low-rangeand high-range measurements of this assay were 50 and600 IU/ml, respectively. HCV core antigen in the serawas also quantitatively measured by chemiluminescentimmunoassay (Abbott Japan Co., Ltd.).

The study protocol was approved beforehand by theEthic Committee in Kobe Asahi Hospital, and writteninformed consent was obtained from each patient priorto the treatment.

NS5A sequence analysis. HCV RNA was extractedfrom 140 µl of serum using a commercially available kit(QIAmp viral RNA kit; QIAGEN, Tokyo). For amplifi-cation of the NS5A region of the HCV genome, theextracted RNA was reverse transcribed and amplifiedfor full-length NS5A using SuperScript One-step RT-PCR for long templates (Invitrogen, Tokyo) and a set ofprimers, NS5A-F1 (5'-TACTCCCTGCCATCCCTCTC-CTG-3'; sense, nucleotides [nt] 5974 to 5997) andNS5A-F2 (5'-CTCCTTGAGCACGTCCCGGT-3'; anti-sense, nt 7796 to 7777). The resultant RT-PCR productwas subjected to a second-round PCR by using Plat-inum Taq DNA polymerase (Invitrogen) and an innerset of primers, NS5A-F3 (5'-TCTCCAGCCTTACCAT-

472 A. EL-SHAMY ET AL

CACYCA-3'; sense, nt 6172 to 6193) and NS5A-F4(5'-CGGTARTGRTCGTCCAGGAC-3'; antisense, nt7780 to 7761). The samples that were not amplifiable(nos. 3, 23, 47, 61, 65 and 69) using the aforementionedprimers were amplified using primer sets reported pre-viously (21). Reverse transcription was performed at45 C for 30 min and terminated at 94 C for 2 min, fol-lowed by the first-round PCR over 35 cycles, with eachcycle consisting of denaturation at 94 C for 30 sec,annealing at 55 C for 30 sec and extension at 68 C for90 sec. The second-round PCR was performed underthe same condition. The amplified fragments werepurified with QIA quick PCR purification kit (QIA-GEN), and visualized by agarose gel electrophoresisand ethidium bromide staining. Nucleotide sequencesof the amplified fragments were determined by directsequencing without subcloning using Big Dye DeoxyTerminator cycle sequencing kit and ABI 337 DNAsequencer (Applied Biosystems, Inc., Japan). The aasequences were deduced and aligned using GENETYX.Win software version 7.0 (GENETYX Corp., Tokyo).

Detection of anti-NS5A antibodies. Detection ofanti-NS5A antibodies in the pre-treated sera was per-formed using Western blot analysis. Lysates of Huh-7cells harboring an HCV-1b subgenomic RNA replicon(pFK5B2884Gly; a kind gift from Dr. R. Barten-schlager, University of Heidelberg, Heidelberg, Ger-many) (20) were prepared in a buffer containing 50 mM

Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate (SDS),10% glycerol, 100 mM dithiothreitol and 0.1% bro-mophenol blue, and were subjected to 10% SDS-poly-acrylamide gel electrophoresis (1�105 cells equivalentper each lane). Lysates of the parental Huh-7 cellsserved as a control. The samples were then transferredto nitrocellulose membranes. After being blocked withskimmed milk for 1 hr at room temperature followedby washing with phosphate-buffered saline (PBS) con-taining 0.05% Tween 20 (PBS-T), the membranes wereincubated with each of the patients’ sera (1:100 diluted)for 1 hr, washed three times with PBS-T, and incubatedwith peroxidase-conjugated goat anti-human IgG(1:1,000 diluted; MBL, Nagoya, Japan) at room tem-perature for another 1 hr. After being washed threetimes with PBS-T, the positive bands were visualizedby using ECL detection system (Amersham PharmaciaBiotech) according to the manufacturer’s instruction.

Statistical analysis. The data obtained were statisti-cally analyzed by Student’s t test, χ2 test and the test forthe proportion. A P value of �0.05 was considered sta-tistically significant.

Nucleotide sequence accession numbers. Thenucleotide sequence data reported in this paper willappear in the DDBJ/EMBL/GenBank nucleotide

sequence databases with the accession numbersAB285035 through AB285081.

Results

Time Course of Viral Clearance and Determination ofEarly Virological Responders (EVR)

Therapeutic responses of PEG-IFN/RBV-treatedpatients were monitored at intervals of 4 weeks. Thenumber of patients who cleared HCV from the seraincreased with time and reached nearly a plateau level atweek 16 after the initiation of the PEG-IFN/RBV thera-py (Table 1). Based on this viral clearance kinetics, weprovisionally chose week 16 for determination of effi-cient viral clearance and, accordingly, the patients wereclassified into early virological responders at week 16(EVR[16w]) and Non-EVR(16w).

Demographic characteristics of EVR(16w) and Non-EVR(16w) are summarized in Table 2. The mean titerof HCV core antigen was 1.9-times higher for Non-EVR(16w) than that for EVR(16w) although the differ-ence was not statistically significant (P�0.17). HCVRNA titers measured by RT-PCR were almost the samebetween the two groups. Also, age, gender, bodyweight, platelet counts, hemoglobin levels, ALT or γ-GTP titers did not significantly differ betweenEVR(16w) and Non-EVR(16w).

Sequence Analysis of the V3 Region and Its AdjacentRegions of NS5A Obtained from Pre-Treated Sera

The entire NS5A region of the HCV genome wasamplified from the pre-treated sera, and the aasequences deduced. We first determined the consensussequence by aligning the sequences obtained from all 47patients and then compared each NS5A sequence to theconsensus sequence. A tendency was noted that thenumber of aa substitutions in the entire NS5A obtainedfrom EVR(16w) was greater than that from Non-EVR(16w) (Table 3). There was no difference in thenumber of mutations in an N-terminal half of NS5A (aa

473HCV NS5A SEQUENCE, ANTI-NS5A AND IFN RESPONSIVENESS

Table 1. Time course of the virological response of the patientsduring PEG-IFN/RBV combination therapy

Weeka) No. of respondersb) / no. of total (%)

4 6 / 47 (12.8)8 13 / 47 (27.7) 12 23 / 47 (48.9)16 31 / 47 (66.0) 20 31 / 47 (66.0) 24 34 / 47 (72.3)

a) Weeks after initiation of PEG-IFN/RBV combination therapy.b) Responders are those who cleared HCV from the sera.

1972 to 2208), the ISDR (aa 2209 to 2248) or the PKR-binding domain (aa 2209 to 2274) between EVR(16w)and Non-EVR(16w). Instead, a significant differencewas observed in the V3 region (aa 2356 to 2379), withthe mean � S.D. for EVR(16w) being 4.4 � 1.5 whilethat for Non-EVR(16w) 3.0 � 1.5 (P�0.01). We alsofound a significant difference between the two groups ina V3-upstream region, which we refer to as the pre-V3region (aa 2334 to 2355). When these two regionswere combined, the difference between EVR(16w) andNon-EVR(16w) became more apparent, with the mean� S.D. being 5.5 � 2.0 and 3.3 � 1.6 for EVR(16w)and Non-EVR(16w), respectively (P�0.001). Wehereafter refer to this combined region as IFN/RBVresistance-determining region (IRRDR).

Only 2 (12.5%) of 16 Non-EVR(16w) had HCVwith 5 or more (�5) mutations in the V3 region where-as 15 (48.4%) of 31 EVR(16w) did (Table 4). The dif-ference between EVR(16w) and Non-EVR(16w) wasstatistically significant (P�0.05). Similarly, only one(6.3%) of 16 Non-EVR(16w) and 17 (54.8%) of 31EVR(16w) had HCV with �6 mutations in IRRDR,with the difference between the two groups being statis-tically significant (P�0.01). It should be noted that, atlater time points as well, such as weeks 20 and 24,strong correlation was observed between the virusclearance rate and the sequence variation in the V3region and IRRDR (data not shown). No correlationwas observed between the initial viral load and thenumber of mutations in the V3 region or IRRDR.

474 A. EL-SHAMY ET AL

Table 2. Demographic characteristics of EVR(16w) and Non-EVR(16w) patients

Factor EVR(16w) Non-EVR(16w) P value

Age 56.0 � 10.5a) 59.6 �9.6 NSb)

Sex (male/female) 17 / 14 12 / 4 NSBody weight (kg) 58.3 � 11.6 63.7 � 10.9 NSPlatelets (�104/mm3) 17.3 � 4.9 14.4 � 5.4 NSHemoglobin (g/dl) 14.1 � 1.5 14.0 � 1.6 NSALT (IU/liter) 35.3 � 25.2 44.6 � 34.2 NSγ-GTP (IU/liter) 39.0 � 27.7 53.4 � 36.1 NSHCV-RNA (kIU/ml) 1,455 � 1,195 1,689 � 1,407 NSHCV core antigen (fmol/liter) 6,349 � 6,066 11,879 � 14,902 NS

a) Mean � S.D.b) NS, not significant.

Table 3. The number of aa mutations within different regions of NS5A of the HCV genome obtained from pre-treated seraof EVR(16w) and Non-EVR(16w)

NS5A region EVR (16w) Non-EVR (16W) P value

Full-NS5A (aa 1972–2419) 22.9 � 5.4a) 19.6 � 4.5 �0.05N-half (aa 1972–2208) 8.7 �2.0 9.1 � 2.6 NSb)

ISDR (aa 2209–2248) 1.7 � 2.4 1.3 � 1.3 NSPKR-BD (aa 2209–2274) 3.3 � 3.0 2.8 � 2.1 NSPre-V3 (aa 2334–2355) 1.1 � 1.2 0.3 � 0.5 �0.05V3 (aa 2356–2379) 4.4 � 1.5 3.0 � 1.5 �0.01IRRDR (aa 2334–2379) 5.5 � 2.0 3.3 � 1.6 �0.001

a) Mean � S.D.b) NS, not significant.

Table 4. Correlation between NS5A sequence variation, Anti-NS5A antibodies and EVR(16w)

CriteriaNo. of subjects / no. of subtotal a) (%)

P valueEVR(16w) Non-EVR(16w)

V3 mutation �5 15 / 31 (48.4) 2 / 16 (12.5) �0.05IRRDR mutation �6 17 / 31 (54.8) 1 / 16 (6.3) �0.01Thr2378 12 / 31 (38.7) 0 / 16 (0.0) �0.01Anti-NS5A (�) 22 / 31(71.0) 2 / 16 (12.5) �0.001

a) Subtotal no. of EVR(16w) or Non-EVR(16w).

475HCV NS5A SEQUENCE, ANTI-NS5A AND IFN RESPONSIVENESS

Fig. 1. Sequence alignment of IRRDR (pre-V3 and V3 regions) of NS5A of HCV-1b obtained from the pre-treatmentsera. The consensus sequence is shown on the top (Cons). The numbers along the sequence indicate aa positions. Dotsindicate residues identical to those of the consensus sequence. Thr2378 is written in boldface Result of anti-NS5A anti-bodies detection is also shown on the right.

The IRRDR sequences obtained from all 47 patientswere aligned along with the consensus sequence (Fig.1). We noticed that 12 (38.7%) of 31 EVR(16w) hadthreonine at position 2378 (Thr2378) whereas none of 16Non-EVR(16w) did (P�0.01) (Table 4).

Sequence Analysis of ISDR and PKR-BD of N55AObtained from Pre-Treated Sera

As described above, there was no difference in themean number of mutations in ISDR or PKR-BDbetween EVR(16w) and Non-EVR(16w) (Table 3).Only 4 patients had HCV with �4 mutations in ISDR(Fig. 2), the criteria for IFN-sensitive HCV strainsaccording to the criteria reported by Enomoto et al. (5,6). Although there appeared to be a trend for patientswith HCV having �4 mutations in ISDR towardEVR(16w) (3 of 4 patients), the difference was not sta-tistically significant. Also, the prevalence of HCV with�4 mutations in ISDR was not significantly differentbetween EVR(16w) (3 of 31 patients; 9.7%) and Non-EVR(16w) (1 of 16 patients; 6.3%). It would be inter-esting to note, however, that all 3 HCV strains with �4mutations in ISDR obtained from EVR(16w) (nos. 10,65 and 72) had �6 mutations in IRRDR, while the onlyone strain with �4 mutations in ISDR from Non-EVR(16w) (no. 13) had only 3 mutations in IRRDR(Figs. 1 and 2).

Detection of Anti-NS5A Antibodies in the Pre-Treat-ment Sera

It was previously reported that the presence of anti-NS5A antibodies in the pre-treatment sera was corre-lated well with IFN responsiveness (9, 10); however,this observation is controversial (36, 44). Therefore,we examined whether or not the patients had detectablelevels of anti-NS5A antibodies in the sera before PEG-IFN/RBV treatment. Consistent with the previousobservations, 22 (71.0%) of 31 EVR(16w) patients haddetectable levels of anti-NS5A antibodies whereas only2 (12.5%) of 16 Non-EVR(16w) did, with the differ-ence between the two groups being statistically signifi-cant (P�0.001) (Table 4). Representative results ofanti-NS5A-positive and -negative sera are shown inFig. 3. Thus, there was a strong correlation betweenthe presence of anti-NS5A antibodies and PEG-IFN/RBV responsiveness assessed by EVR(16w).

Proposed Predictive Factors for EVR and Non-EVRTable 5 summarizes proposed predictive factors for

EVR(16w) or Non-EVR(16w). The positive predictivevalue (PPV) of a factor is the proportion (%) of patientswith the factor who have achieved the outcome of inter-est (either EVR[16w] or Non-EVR[16w]) while the

negative predictive value (NPV) is the proportion (%) ofpatients without the same factor who have not achievedthe outcome of interest. High degrees of sequence vari-ation in the V3 region (�5) and IRRDR (�6) of HCVNS5A could predict EVR(16w) with PPV of 88.2%(P�0.01) and 94.4% (P�0.0001), respectively. Simi-larly, the presence of anti-NS5A antibodies could be auseful predictive factor for EVR(16w), with PPV being91.7% (P�0.0001).

It should be noted that the NPV for the low degree ofsequence variation in V3 (�4) and IRRDR (�5) aloneor the absence of detectable anti-NS5A antibodies alonedid not reach a statistically significant level and, there-fore, cannot be used as a predictive factor for Non-EVR(16w). On the other hand, the absence ofdetectable levels of anti-NS5A antibodies together withthe low degree of sequence variation in either V3 (�4)or IRRDR (�5) could be a useful predictive factor forNon-EVR(16w), with PPV being 81.3% (P�0.05) and87.5% (P�0.01), respectively (Table 5).

Viral clearance kinetics at intervals of 4 weeks forHCV-1b isolates with and without the predictive factorsis shown in Fig. 4A (�5 mutations in V3), 4B (�6mutations in IRRDR), 4C (the presence of anti-NS5Aantibodies) and 4D (the absence of anti-NS5A antibod-ies plus �4 mutations in V3 or �5 mutations inIRRDR). High degrees of sequence variation in V3(�5) and IRRDR (�6) were significantly associatedwith early viral clearance not only at week 16 but also atweeks 12, 20 and 24. Similarly, significant correlationbetween the presence of anti-NS5A antibodies andearly viral clearance at week 16 and thereafter wasobserved (Fig. 4C). On the other hand, the absence ofanti-NS5A antibodies together with lower degrees ofmutation in V3 (�4) or IRRDR (�5) was significantlyassociated with persistent HCV viremia at week 16 andthereafter (Fig. 4D).

Discussion

The introduction of IFN/RBV combination therapywas an important breakthrough in the treatment ofchronic HCV infection (2, 23). However, the rate ofsustained virological response is still unsatisfactory(~50%), particularly in patients infected with HCVgenotype 1 (11, 22). Given the significant side effectsand high cost associated with the IFN/RBV therapy fora period of 48 weeks (29), it would be beneficial if onecould predict ahead of time which patients would, orwould not, respond to the therapy. Most of the recentstudies have focused on the possible correlationbetween EVR and viral clearance kinetics during thefirst few months of the treatment (8, 14, 15, 24). On the

476 A. EL-SHAMY ET AL

477HCV NS5A SEQUENCE, ANTI-NS5A AND IFN RESPONSIVENESS

Fig. 2. Sequence alignment of ISDR and PKR-BD of NS5A of HCV-1b obtained from the pre-treatment sera. Theconsensus sequence of HCV-1b isolates obtained in this study is shown on the top (Cons). The ISDR sequence is thesame between the consensus sequence and that of HCV-J (42), a representative IFN-resistant strain (9, 10). Theremaining carboxy-terminal sequence in PKR-BD differs between them by four residues (I2259, V2262, I2268 and P2271 forHCV-J). The corresponding residues in our consensus sequence (L2259, E2262, V2268 and A2271) have been reported as aconsensus for HCV-1b in other studies (17, 21, 25). The numbers along the sequence indicate aa positions. Dots indi-cate residues identical to those of the consensus sequence. *, The HCV isolates from patient no. 72 had a unique inser-tion (FKECW) at position 2249, which has not been shown in this figure in order to avoid disturbance of the mostcommon sequence alignment pattern.

other hand, some studies dealt with the possible correla-tion between EVR and sequence variation of part ofHCV NS5A, especially the V3 region and its surround-ing regions (4, 19, 27, 31, 35, 37, 44). As NS5A hasalways been among the focus of attention, we carriedout sequence analysis over the entire NS5A in the hopethat we could find possible correlation between NS5Asequence variation and EVR. Interestingly, we foundthat a high degree of sequence variation in the V3region (�5) or IRRDR (�6) correlated well with viralclearance by week 16 and, therefore, would be a usefulfactor to predict EVR(16w) (Tables 3 to 5). Similarly,Thr2378 could be used as a secondary predictive factorfor EVR(16w) (Table 4 and Fig. 1). It should also benoted, however, that nearly a half of the patients infect-ed with HCV isolates without these markers stillbecame EVR(16w).

In this study, we provisionally chose week 16 as thetime point for determination of EVR since the ratio ofvirological responders reached nearly a plateau level at

week 16 (Table 1). Most of the previous studies adopt-ed weeks 12 (3, 7, 8, 23, 30) and 24 (7, 25) as the timepoint for EVR determination. As was the case withEVR(16w), we observed a strong correlation betweenEVR(24w) and the sequence variation in V3 or IRRDR(Fig. 4A and B). Also, there was a weaker but signifi-cant correlation between EVR(12w) and the sequencevariation in V3 or IRRDR. As we observed a consider-able number of patients responding to the PEG-IFN/RBV therapy between weeks 12 and 16, we rec-ommend that a proper time point be checked in eachstudy, depending upon certain differences in the charac-teristics of the virus and the baseline health/genetic con-ditions of the patients in the cohort, by careful monitor-ing of the EVR rate at different time points, especiallywhen the researchers want to choose an earlier timepoint, such as weeks 12 and 16, for EVR determination.Moreover, how many of EVR(12w), EVR(16w) andEVR(24w) can achieve the end of treatment responseafter 48 weeks of treatment and, eventually, the sus-

478 A. EL-SHAMY ET AL

Fig. 3. Detection of antibodies against NS5A and NS3 proteins of HCV. Pre-treatment sera of EVR(16w) and Non-EVR(16w) were examined by Western blot analysis using lysates of Huh-7 cells harboring HCV subgenomic RNAreplicon (lanes 1 to 9) or the HCV RNA-negative control Huh-7 cells (lane 10). Lanes 1 to 4, EVR(16w) nos. 14, 21,24 and 39; lanes 5 to 8, Non-EVR(16w) nos. 8, 11, 16 and 31; lane 9, uninfected individual; lane 10, EVR(16w) test-ed with the control cell lysates.

Table 5. Proposed predictive factors for EVR(16w) and Non-EVR(16w)

Predictive factorNo. of patients

PPV (P value) NPV (P value)Total EVR Non-EVR

V3 �5 17 15 2 88.2a) (�0.01)V3 �4 30 16 14 46.7b) (NSc))

IRRDR �6 18 17 1 94.4a) (�0.0001)IRRDR �5 29 14 15 51.7b) (NS)

Anti-NS5A (�) 24 22 2 91.7a) (�0.0001)Anti-NS5A (�) 23 9 14 60.9b) (NS)

V3 �4 plus anti-NS5A (�) 16 3 13 81.3d) (�0.05)Others 31 28 3 90.3e) (�0.0001)

IRRDR �5 plus anti-NS5A (�) 16 2 14 87.5d) (�0.01)Others 31 29 2 93.5e) (�0.0001)a) PPV for EVR(16w).b) NPV for EVR(16w).c) NS, not significant.d) PPV for Non-EVR(16w).e) NPV for Non-EVR(16w).

tained virological response 24 weeks after the cessationof the treatment needs to be clarified in the follow-upstudy. Nonetheless, the clinical importance of viralclearance, either EVR(12w) or EVR(16w), should beemphasized as it was reported that even patients havingexperienced transient disappearance of HCV RNA dur-ing the course of IFN treatment showed a lower inci-dence of hepatocellular carcinoma compared with non-responders having sustained HCV viremia (16, 43).Only 4 (8.5%) of 47 HCV-1b isolates examined had �4

mutations in ISDR, 3 of which were found inEVR(16w), with the remaining one in Non-EVR(16w).No significant correlation was observed between ISDRmutations and EVR(16w) (Table 3 and Fig. 2). Thislack of significant difference is due possibly to thesmall number of samples tested. In a geographic regionwhere the majority of HCV-1b strains have �4 muta-tions in ISDR, the ISDR theory proposed by Enomoto etal. (1995, 1996) may be applied to predict EVR. Infact, it was recently reported that there was a close cor-

479HCV NS5A SEQUENCE, ANTI-NS5A AND IFN RESPONSIVENESS

Fig. 4. Viral clearance kinetics of HCV-1b isolates classified on the basis of the number of mutations in the V3region (A) and IRRDR (B), the presence of anti-NS5A antibodies (C) and the combination of the absence of anti-NS5A antibodies plus �4 mutations in V3 or �5 mutations in IRRDR (D) in the pre-treated sera. *, P�0.05; †, P�0.01; ‡, P�0.001.

relation between the number of mutations in ISDR andresponse to IFN/RBV combination therapy (28, 33). Inthis connection, we found that all 3 HCV isolates with�4 mutations in ISDR obtained from EVR(16w) (nos.10, 65 and 72) had �5 and �6 mutations in the V3region and IRRDR, respectively, while the only isolatewith �4 mutations in ISDR from Non-EVR(16w) (no.13) having only 2 and 3 mutations in V3 and IRRDR,respectively (see Figs. 1 and 2). It is possible that theV3 region or IRRDR is associated with PEG-IFN/RBVresponsiveness more closely than is ISDR.

While cell-mediated immunity plays a pivotal role inrecovery from virus infections, including HCV (42),antibodies in general do not significantly contribute toclearance of the virus unless they have neutralizingactivities that recognize certain viral epitopes on thesurface of the viral particle. In the present study, weobserved that the presence of antibodies in pre-treatedsera against NS5A was correlated with IFN/RBVresponsiveness (Tables 4 and 5). Since NS5A, a non-structural protein, is not present on HCV particles, anti-NS5A antibodies would not neutralize them. There aretwo possible implications for the observed correlationbetween the presence of anti-NS5A antibodies andIFN/RBV responsiveness: (i) the degree of anti-NS5Aantibody responses parallels that of cell-mediatedimmunity against NS5A, the latter of which plays animportant role in eradicating the virus in cooperationwith IFN, and (ii) NS5A of PEG-IFN/RBV-sensitiveHCV strains is antigenically closer to that of the HCVsubgenomic RNA replicon used in the assay than that ofthe resistant strains and, therefore, antibodies againstNS5A of the sensitive strains were more efficientlydetected by the present method. In this connection, wedid not find so far any particular aa residues orsequences that are common among HCV isolatesobtained from EVR(16w). This, however, does notnecessarily exclude the latter possibility since we havenot yet completely dissected all the antigenic epitopes inNS5A. Our result is consistent with a previous observa-tion that, in a clinical setting with IFN monotherapy,95% of sustained viral responders, but only 13% ofnon-responders, had detectable levels of anti-NS5Aantibodies in pre-treatment sera (9, 10). On the otherhand, apparently conflicting observations were reportedthat detection of anti-NS5A antibodies in pre-treatmentsera did not correlate significantly with virologicalresponse to IFN treatment (36, 44). However, the num-ber of patients examined in those studies might be toosmall to find a significant correlation and, more impor-tantly, depending on the sensitivity of the assay, quanti-tative assessment of the antibody titers might be neededto find a difference between responders and non-

responders. Further studies are needed to elucidate theissue.

In conclusion, our present results suggest that a highdegree of sequence variation in the V3 region (�5mutations) or IRRDR (�6 mutations) and the presenceof detectable levels of anti-NS5A antibodies in the pre-treatment sera would be useful factors to predictEVR(16w). On the other hand, a less diverse sequencein V3 (�4 mutations) or IRRDR (�5 mutations)together with the absence of detectable anti-NS5A anti-bodies could be a predictive factor for Non-EVR(16w).These findings emphasize the need for further investiga-tion of NS5A, especially IRRDR, with regard to theunderlying molecular mechanism of the inhibition ofantiviral effect(s) of IFN and/or RBV.

The authors are grateful to Dr. R. Bartenschlager (Universityof Heidelberg, Heidelberg, Germany) for providing an HCVsubgenomic RNA replicon (pFK5B2884Gly). This study wassupported in part by grants-in-aid for Scientific Research fromthe Japan Society for the Promotion of Science, the Ministry ofEducation, Culture, Sports, Science and Technology, and Min-istry of Health, Labour and Welfare, Japan. This study was alsocarried out as part of the 21st Century COE Program at KobeUniversity Graduate School of Medicine.

References

1) Chayama, K., Tsubota, A., Kobayashi, M., Okamoto, K.,Hashimoto, M., Miyano, Y., Koike, H., Kobayashi, M.,Koida, I., Arase, Y., Saitoh, S., Suzuki, Y., Murashima, N.,Ikeda, K., and Kumada, H. 1997. Pretreatment virus loadand multiple amino acid substitutions in the interferon sen-sitivity-determining region predict the outcome of interferontreatment in patients with chronic genotype 1b hepatitis Cvirus infection. Hepatology 25: 745–749.

2) Davis, G.L., Esteban-Mur, R., Rustgi, V., Hoefs, J., Gor-don, S.C., Trepo, C., Shiffman, M.L., Zeuzem, S., Craxi,A., Ling, M.H., and Albrecht, J. 1998. Interferon alfa-2balone or in combination with ribavirin for the treatment ofrelapse of chronic hepatitis C. International Hepatitis Inter-ventional Therapy Group. N. Engl. J. Med. 339:1493–1499.

3) Davis, G.L., Wong, J.B., McHutchison, J.G., Manns, M.P.,Harvey, J., and Albrecht, J. 2003. Early virologic responseto treatment with peginterferon alfa-2b plus ribavirin inpatients with chronic hepatitis C. Hepatology 38: 645–652.

4) Duverlie, G., Khorsi, H., Castelain, S., Jaillon, O., Izopet, J.,Lunel, F., Eb, F., Penin, F., and Wychowski, C. 1998.Sequence analysis of the NS5A protein of European hepati-tis C virus 1b isolates and relation to interferon sensitivity. J.Gen. Virol. 79: 1373–1381.

5) Enomoto, N., Sakuma, I., Asahina, Y., Kurosaki, M.,Murakami, T., Yamamoto, C., Izumi, N., Marumo, F., andSato, C. 1995. Comparison of full-length sequences ofinterferon-sensitive and resistant hepatitis C virus 1b. Sensi-tivity to interferon is conferred by amino acid substitutions

480 A. EL-SHAMY ET AL

in the NS5A region. J. Clin. Invest. 96: 224–230.6) Enomoto, N., Sakuma, I., Asahina, Y., Kurosaki, M.,

Murakami, T., Yamamoto, C., Ogura, Y., Izumi, N., Maru-mo, F., and Sato, C. 1996. Mutations in the nonstructuralprotein 5A gene and response to interferon in patients withchronic hepatitis C virus 1b infection. N. Engl. J. Med. 334:77–81.

7) Ferenci, P. 2004. Predictors of response to therapy forchronic hepatitis C. Semin. Liver Dis. 24: 25–31.

8) Ferenci, P., Fried, M.W., Shiffman, M.L., Smith, C.I., Mari-nos, G., Goncales, F.L., Jr., Haussinger, D., Diago, M.,Carosi, G., Dhumeaux, D., Craxi, A., Chaneac, M., andReddy, K.R. 2005. Predicting sustained virological respons-es in chronic hepatitis C patients treated with peginterferonalfa-2a (40 KD)/ribavirin. J. Hepatol. 43: 425–433.

9) Frangeul, L., Cresta, P., Perrin, M., Duverlie, G., Khorsi,H., Musset, L., Opolon, P., Huraux, J.M., and Lunel, F.1998. Pattern of HCV antibodies with special reference toNS5A reactivity in HCV-infected patients: relation to viralgenotype, cryoglobulinemia and response to interferon. J.Hepatol. 28: 538–543.

10) Frangeul, L., Cresta, P., Perrin, M., Lunel, F., Opolon, P.,Agut, H., and Huraux, J.M. 1998. Mutations in NS5Aregion of hepatitis C virus genome correlate with presenceof NS5A antibodies and response to interferon therapy formost common European hepatitis C virus genotypes. Hepa-tology 28: 1674–1679.

11) Fried, M.W., Shiffman, M.L., Reddy, K.R., Smith, C.,Marinos, G., Goncales, F.L., Jr., Haussinger, D., Diago, M.,Carosi, G., Dhumeaux, D., Craxi, A., Lin, A., Hoffman, J.,and Yu, J. 2002. Peginterferon alfa-2a plus ribavirin forchronic hepatitis C virus infection. N. Engl. J. Med. 347:975–982.

12) Gale, M.J., Jr., Korth, M.J., and Katze, M.G. 1998. Repres-sion of the PKR protein kinase by the hepatitis C virusNS5A protein: a potential mechanism of interferon resis-tance. Clin. Diagn. Virol. 10: 157–162.

13) Gale, M.J., Jr., Korth, M.J., Tang, N.M., Tan, S.L., Hop-kins, D.A., Dever, T.E., Polyak, S.J., Gretch, D.R., andKatze, M.G. 1997. Evidence that hepatitis C virus resis-tance to interferon is mediated through repression of thePKR protein kinase by the nonstructural 5A protein. Virolo-gy 230: 217–227.

14) Iino, S., Tomita, E., Kumada, H., Suzuki, H., Toyota, J.,Kiyosawa, K., Tanikawa, K., Sata, M., Hayashi, N., Kaku-mu, S., Matsushima, T., and Ohno, T. 2004. Prediction oftreatment outcome with daily high-dose IFN alpha-2b plusribavirin in patients with chronic hepatitis C with genotype1b and high HCV RNA levels: relationship of baseline virallevels and viral dynamics during and after therapy. Hepatol.Res. 30: 63–70.

15) Jensen, D.M., Morgan, T.R., Marcellin, P., Pockros, P.J.,Reddy, K.R., Hadziyannis, S.J., Ferenci, P., Ackrill, A.M.,and Willems, B. 2006. Early identification of HCV genotype1 patients responding to 24 weeks peginterferon alpha-2a(40 kd)/ribavirin therapy. Hepatology 43: 954–960.

16) Kasahara, A., Hayashi, N., Mochizuki, K., Takayanagi, M.,Yoshioka, K., Kakumu, S., Iijima, A., Urushihara, A.,Kiyosawa, K., Okuda, M., Hino, K., and Okita, K. 1998.

Risk factors for hepatocellular carcinoma and its incidenceafter interferon treatment in patients with chronic hepatitisC. Osaka Liver Disease Study Group. Hepatology 27:1394–1402.

17) Kim, W.R. 2002. The burden of hepatitis C in the UnitedStates. Hepatology 36: 30–34.

18) Lauer, G.M., and Walker, B.D. 2001. Hepatitis C infection.N. Engl. J. Med. 345: 41–52.

19) Layden-Almer, J.E., Kuiken, C., Ribeiro, R.M., Kunstman,K.J., Perelson, A.S., Layden, T.J., and Wolinsky, S.M.2005. Hepatitis C virus genotype 1a NS5A pretreatmentsequence variation and viral kinetics in African Americanand white patients. J. Infect. Dis. 192: 1078–1087.

20) Lohmann, V., Korner, F., Koch, J., Herian, U., Theilmann,L., and Bartenschlager, R. 1999. Replication of subgenomichepatitis C virus RNAs in a hepatoma cell line. Science285: 110–113.

21) Lusida, M.I., Nagano-Fujii, M., Nidom, C.A., Soetjipto,Handajani, R., Fujita, T., Oka, K., and Hotta, H. 2001. Cor-relation between mutations in the interferon sensitivity-determining region of NS5A protein and viral load ofhepatitis C virus subtypes 1b, 1c, and 2a. J. Clin. Microbiol.39: 3858–3864.

22) Manns, M.P., McHutchison, J.G., Gordon, S.C., Rustgi,V.K., Shiffman, M., Reindollar, R., Goodman, Z.D., Koury,K., Ling, M., and Albrecht, J.K. 2001. Peginterferon alfa-2bplus ribavirin compared with interferon alfa-2b plus rib-avirin for initial treatment of chronic hepatitis C: a ran-domised trial. Lancet 358: 958–965.

23) McHutchison, J.G., Gordon, S.C., Schiff, E.R., Shiffman,M.L., Lee, W.M., Rustgi, V.K., Goodman, Z.D., Ling,M.H., Cort, S., and Albrecht, J.K. 1998. Interferon alfa-2balone or in combination with ribavirin as initial treatmentfor chronic hepatitis C. Hepatitis Interventional TherapyGroup. N. Engl. J. Med. 339: 1485–1492.

24) Mihm, U., Herrmann, E., Sarrazin, C., and Zeuzem, S.2006. Review article: predicting response in hepatitis Cvirus therapy. Aliment Pharmacol. Ther. 23: 1043–1054.

25) Moreno, A., Barcena, R., Garcia-Garzon, S., Muriel, A.,Quereda, C., Moreno, L., Mateos, M.L., Fortun, J., Martin-Davila, P., Garcia, M., Blesa, C., Oton, E., Moreno, A., andMoreno, S. 2005. HCV clearance and treatment outcome ingenotype 1 HCV-monoinfected, HIV-coinfected and livertransplanted patients on peg-IFN-alpha-2b/ribavirin. J.Hepatol. 43: 783–790.

26) Murashima, S., Ide, T., Miyajima, I., Kumashiro, R., Ueno,T., Sakisaka, S., and Sata, M. 1999. Mutations in the NS5Agene predict response to interferon therapy in Japanesepatients with chronic hepatitis C and cirrhosis. Scand. J.Infect. Dis. 31: 27–32.

27) Murphy, M.D., Rosen, H.R., Marousek, G.I., and Chou, S.2002. Analysis of sequence configurations of the ISDR,PKR-binding domain, and V3 region as predictors ofresponse to induction interferon-alpha and ribavirin therapyin chronic hepatitis C infection. Dig. Dis. Sci. 47:1195–1205.

28) Nagaki, M., Imose, M., Naiki, T., Kimura, K., Hayashi, H.,Shimizu, M., Ohnishi, H., Tomita, E., Sugihara, J., Amano,K., Sakai, T., Kojima, T., Katsumura, N., Kondo, Y., Fuji-

481HCV NS5A SEQUENCE, ANTI-NS5A AND IFN RESPONSIVENESS

moto, M., and Moriwaki, H. 2005. Prospective study onearly virologic response to treatment with interferon alpha-2b plus ribavirin in patients with chronic hepatitis C geno-type 1b. Hepatol. Res. 33: 285–291.

29) Nakamura, H. 2005. Early prediction of sustained viralresponder and non-responder during interferon and rib-avirin combination therapy in chronic hepatitis C. Hepatol.Res. 33: 269–271.

30) NIH Consensus Statement on Management of Hepatitis C.2002. NIH Consens. Stat. Sci. Statements 19: 1–46.

31) Nousbaum, J., Polyak, S.J., Ray, S.C., Sullivan, D.G., Lar-son, A.M., Carithers, R.L., Jr., and Gretch, D.R. 2000.Prospective characterization of full-length hepatitis C virusNS5A quasispecies during induction and combinationantiviral therapy. J. Virol. 74: 9028–9038.

32) Okamoto, H., Sugiyama, Y., Okada, S., Kurai, K., Akahane,Y., Sugai, Y., Tanaka, T., Sato, K., Tsuda, F., andMiyakawa, Y. 1992. Typing hepatitis C virus by poly-merase chain reaction with type-specific primers: applica-tion to clinical surveys and tracing infectious sources. J.Gen. Virol. 73: 673–679.

33) Pascu, M., Martus, P., Hohne, M., Wiedenmann, B., Hopf,U., Schreier, E., and Berg, T. 2004. Sustained virologicalresponse in hepatitis C virus type 1b infected patients ispredicted by the number of mutations within the NS5A-ISDR: a meta-analysis focused on geographical differences.Gut 53: 1345–1351.

34) Paterson, M., Laxton, C.D., Thomas, H.C., Ackrill, A.M.,and Foster, G.R. 1999. Hepatitis C virus NS5A proteininhibits interferon antiviral activity, but the effects do notcorrelate with clinical response. Gastroenterology 117:1187–1197.

35) Puig-Basagoiti, F., Forns, X., Furcic, I., Ampurdanes, S.,Gimenez-Barcons, M., Franco, S., Sanchez-Tapias, J.M.,and Saiz, J.C. 2005. Dynamics of hepatitis C virus NS5Aquasispecies during interferon and ribavirin therapy inresponder and non-responder patients with genotype 1bchronic hepatitis C. J. Gen. Virol. 86: 1067–1075.

36) Sarrazin, C., Berg, T., Lee, J.H., Ruster, B., Kronenberger,B., Roth, W.K., and Zeuzem, S. 2000. Mutations in the pro-tein kinase-binding domain of the NS5A protein in patientsinfected with hepatitis C virus type 1a are associated withtreatment response. J. Infect. Dis. 181: 432–441.

37) Sarrazin, C., Herrmann, E., Bruch, K., and Zeuzem, S.

2002. Hepatitis C virus nonstructural 5A protein and inter-feron resistance: a new model for testing the reliability ofmutational analyses. J. Virol. 76: 11079–11090.

38) Simmonds, P., Bukh, J., Combet, C., Deleage, G., Enomoto,N., Feinstone, S., Halfon, P., Inchauspe, G., Kuiken, C.,Maertens, G., Mizokami, M., Murphy, D.G., Okamoto, H.,Pawlotsky, J.M., Penin, F., Sablon, E., Shin, I.T., Stuyver,L.J., Thiel, H.J., Viazov, S., Weiner, A.J., and Widell, A.2005. Consensus proposals for a unified system of nomen-clature of hepatitis C virus genotypes. Hepatology 42:962–973.

39) Simmonds, P., Holmes, E.C., Cha, T.A., Chan, S.W.,Mcomish, F., Irvine, B., Beall, E., Yap, P.L., Kolberg, J.,and Urdea, M.S. 1993. Classification of hepatitis C virusinto six major genotypes and a series of subtypes by phylo-genetic analysis of the NS-5 region. J. Gen. Virol. 74:2391–2399.

40) Song, J., Fujii, M., Wang, F., Itoh, M., and Hotta, H. 1999.The NS5A protein of hepatitis C virus partially inhibits theantiviral activity of interferon. J. Gen. Virol. 80: 879–886.

41) Taguchi, T., Nagano-Fujii, M., Akutsu, M., Kadoya, H.,Ohgimoto, S., Ishido, S., and Hotta, H. 2004. Hepatitis Cvirus NS5A protein interacts with 2',5'-oligoadenylate syn-thetase and inhibits antiviral activity of IFN in an IFN sen-sitivity-determining region-independent manner. J. Gen.Virol. 85: 959–969.

42) Takaki, A., Wiese, M., Maertens, G., Depla, E., Seifert, U.,Liebetrau, A., Miller, J.L., Manns, M.P., and Rehermann,B. 2000. Cellular immune responses persist and humoralresponses decrease two decades after recovery from a sin-gle-source outbreak of hepatitis C. Nat. Med. 6: 578–582.

43) Tanaka, H., Tsukuma, H., Kasahara, A., Hayashi, N.,Yoshihara, H., Masuzawa, M., Kanda, T., Kashiwagi, T.,Inoue, A., Kato, M., Oshima, A., Kinoshita, Y., and Kama-da, T. 2000. Effect of interferon therapy on the incidence ofhepatocellular carcinoma and mortality of patients withchronic hepatitis C: a retrospective cohort study of 738patients. Int. J. Cancer 87: 741–749.

44) Vuillermoz, I., Khattab, E., Sablon, E., Ottevaere, I., Duran-tel, D., Vieux, C., Trepo, C., and Zoulim, F. 2004. Geneticvariability of hepatitis C virus in chronically infectedpatients with viral breakthrough during interferon-ribavirintherapy. J. Med. Virol. 74: 41–53.

482 A. EL-SHAMY ET AL