Hepatitis C virus drug resistance and immune-driven adaptations: Relevance to new antiviral therapy


  • Potential conflict of interest: Nothing to report.


The efficacy of specifically targeted anti-viral therapy for hepatitis C virus (HCV) (STAT-C), including HCV protease and polymerase inhibitors, is limited by the presence of drug-specific viral resistance mutations within the targeted proteins. Genetic diversity within these viral proteins also evolves under selective pressures provided by host human leukocyte antigen (HLA)-restricted immune responses, which may therefore influence STAT-C treatment response. Here, the prevalence of drug resistance mutations relevant to 27 developmental STAT-C drugs, and the potential for drug and immune selective pressures to intersect at sites along the HCV genome, is explored. HCV nonstructural (NS) 3 protease or NS5B polymerase sequences and HLA assignment were obtained from study populations from Australia, Switzerland, and the United Kingdom. Four hundred five treatment-naïve individuals with chronic HCV infection were considered (259 genotype 1, 146 genotype 3), of which 38.5% were coinfected with human immunodeficiency virus (HIV). We identified preexisting STAT-C drug resistance mutations in sequences from this large cohort. The frequency of the variations varied according to individual STAT-C drug and HCV genotype/subtype. Of individuals infected with subtype 1a, 21.5% exhibited genetic variation at a known drug resistance site. Furthermore, we identified areas in HCV protease and polymerase that are under both potential HLA-driven pressure and therapy selection and identified six HLA-associated polymorphisms (P ≤ 0.05) at known drug resistance sites. Conclusion: Drug and host immune responses are likely to provide powerful selection forces that shape HCV genetic diversity and replication dynamics. Consideration of HCV viral adaptation in terms of drug resistance as well as host “immune resistance” in the STAT-C treatment era could provide important information toward an optimized and individualized therapy for chronic hepatitis C. (HEPATOLOGY 2009.)

Hepatitis C treatment has been transformed over the last decade by the use of improved therapy regimens, including pegylated interferon-alfa, as either monotherapy or combined with the nucleoside analog ribavirin.1, 2 Combination therapy is effective in obtaining a sustained virological response in most treated individuals.1, 2 However, response rates to pegylated interferon-alfa/ribavirin therapy are lower in individuals with the prevalent genotype 1 strain compared with those with genotypes 2 and 3. Also, current therapies are often poorly tolerated because of a plethora of treatment-associated side effects.1 Currently, there is no available effective alternative therapy in individuals who fail to achieve a sustained virological response to current treatment regimens.

Recent advances in molecular biology have led to the development of novel small molecules that target specific viral proteins integral to the hepatitis C virus (HCV) life cycle. These drugs, collectively termed specifically targeted antiviral therapy for HCV (STAT-C), include a range of nonstructural (NS) 3/NS4A protease and NS5B polymerase inhibitors at various stages of clinical development (Tables 1, 2). As expected for these small molecule antiviral drugs, and consistent with experience of similar approaches to HIV and hepatitis B treatment,3, 4 preliminary studies using these agents have indicated a number of drug resistance mutations (Table 1), including examples in which drug resistance mutations reduced the intrinsic replicative capacity of HCV (Table 3). The list of resistance mutations for STAT-C drugs is likely to grow, given the commencement of several large phase II/III trials and the development of new drugs, and it is anticipated that use of these agents in the future will, to some extent, drive contemporary HCV evolution at a population level.

Table 1. Drug Resistance and Related Sites for NS3/4A Serine Protease Inhibitors
Protease InhibitorsCompanyClassResistance MutationsClinical DevelopmentRef.
  • PI, protease inhibitor; Np, not published.

  • *

    Second site mutations (T72I, P88L, I71T, T312A) observed in the replicon system.20

  • Based on the replicon system.

  • Second site mutation (T260A).20

  • §

    P89L, Q86R, G162R (second site mutations) partially restore viral fitness without reducing resistance.31

BILN 2061BoehringerPIA156T*Studies canceled18–24
VX 950Vertex/Jansen Pharma./TibotecPIV36M/GPhase 215, 16, 18, 21, 23–26
SCH503034Schering-PloughPIQ41Phase 218, 23, 27, 28
ITMN 191InterMune/RochePIR155KPhase 11, 23, 24, 29
SCH6Schering-PloughPIT54AStudies canceled20, 30, 31
MK-7009MerckPINpPhase 11
BI12202BoehringerPINpPhase 11
TMC435350Tibotec/MedivirPINpPhase 11, 32
GS9132/ACH 806Gilead Sciences/Achillion PharmaceuticalsInhibits binding of NS4a to NS3NpStudies canceled2
ACH 1095Achillion PharmaceuticalsInhibits binding of NS4a to NS3NpPreclinical2
Table 2. Drug Resistance and Related Sites for NS5B Polymerase Inhibitors
Polymerase InhibitorsCompanyClassResistance MutationsClinical DevelopmentRef.
  • NI, nucleoside polymerase inhibitor; NNI, non-nucleoside inhibitors; Np, not published.

  • *

    Based on the replicon system.

  • I585T partially compensates replication defect because of P495L.

NM 283/NM107Idenix/NovartisNIS282T*Stopped1, 23, 33, 34
R 7128/PSI6130Roche/PharmasetNIS282T/R*Phase 134
R1626/R1479RocheNIS96T*Phase21, 23, 35
GSK625433GlaxoSmithKlineNNIM414TPhase 136, 37
MK-0608MerckNIS282T*Preclinical2, 34
JTK-003Japan Tobacco/Akros PharmaNNINpPhase 2/stopped2, 36
VCH 759ViroChemNNINpPhase 238, 39
XTL 2125XTLbioNNINpPhase1/stopped2
GS9190GileadNNINpPhase 136, 40
BILB 1941Boehringer Ing.NNI Phase1/hold1
HCV 796ViroPharmaNNIL314F*Phase 2/hold1, 23, 41–43
A-782759AbbottNNID55E*Preclinical22, 44, 45
A-837093AbbottNNIS368A*Preclinical46, 47
Thiophene-2-carboxylic acid NNIL419M 1, 48
Benzimidazoles (general) NNIP495S/T* 1, 49, 50
Benzothiadiazine (general) NNIK50R 49, 51, 52
Table 3. Effect of Drug-Associated Mutations on Viral Fitness and Target Protein
Drug Resistance Mutation (Wildtype/Variants) or Compensatory MutationEffect on Viral Fitness*Enzyme Resistance
  • Numbering relates to position within NS3 (protease) and NS5B (polymerase).

  • *

    Same viral fitness as wild-type sequence used in test.

  • High for 2-C-methyl-modified nucleoside analogs, moderate for 7-deaza-2-C-methyl-adenosine analogs. The category of enzyme resistance and viral fitness are based on references in Tables 1 and 2. Enzyme resistance may vary for different drugs. Increased resistance is used when category of enzyme resistance could not be determined. The system used to assess viral fitness is the replicon system and the choice of category is based on the interpretation of the data in the corresponding reference.

Protease inhibitors  
L Low
AWild-type/slightly reducedLow-moderate
I153V Low-moderate
KSlightly increasedHigh
VWild-type/reducedHigh (no effect for SCH-6, moderate for Vx-950 & SCH 503034)
Polymerase inhibitors  
N142T Low-moderate
L314FReducedIncreased resistance
I363VReducedIncreased resistance
AReducedIncreased resistance
LReducedIncreased resistance
N411SSlightly reduced (minimal)Moderate
L419MSlightly reducedHigh
I447F High
Y448HSlightly reduced (minimal)Moderate
SSlightly reducedHigh
ASlightly reducedHigh
SSlightly reducedModerate
V499ASlightly reducedLow
I585T2nd site partially restores fitness (with P495) 

Another selective force that continues to shape HCV diversity is the host's human leukocyte antigen (HLA)–restricted immune response. These immune responses are stimulated by the presentation of parts of internally processed viral peptides (epitopes) in the context of the HLA molecule, and hence the selection of HCV sequences targeted by the immune response is dependent on the HLA repertoire of the host. Many HLA-restricted epitopes have been described within the HCV genome, and there are numerous examples in which mutations within or flanking these epitopes allow the virus to evade host immune control.5, 6 Accordingly, HCV immune escape mutations occur under HLA-specific selection pressure and are likely to shape viral diversity within individuals and across host populations. Evidence for HLA-driven immune pressure shaping viral sequences at a population level has recently been shown in both HCV7, 8 and HIV.9 These viral genetic variants, often referred to as escape mutations, may also be considered as immune “resistance” mutations.

The issue of viral adaptation or immune “resistance” is particularly pertinent given the development of STAT-C drugs, in which specific viral mutations can cause drug resistance. We hypothesize that HLA-driven viral escape in proteins targeted by the small molecules could act as drug resistance mutations and impact on STAT-C treatment response. Conversely, the selection of drug resistance mutations in the same proteins that are targeted by immune responses could disrupt epitope presentation or processing and consequently impair the host's immune response. An example of this scenario is shown in Fig. 1. Accordingly, an individual's HCV immune escape and drug resistance profiles are likely to be critical influences on the outcome with specific antiviral treatments.

Figure 1.

Overlapping selection pressures from anti-HCV drugs and the host's immune response. Convergence of drug-driven and immune-driven selection may influence the outcome of therapy or host's immune response to HCV. HLA repertoire refers to the expression of the HLA types within an individual.

In this study, baseline drug resistance profiles were examined within a large cohort of treatment-naïve HCV monoinfected and HIV coinfected individuals, to determine the extent of natural variation that exists at these sites in the absence of drug pressure. Furthermore, the potential for overlap of drug and immune selective pressures at sites or areas within the HCV genome is explored. This study reveals that many of the described STAT-C drug resistance mutations are already present in circulating HCV strains, albeit at low frequency, and that regions of NS3 protease and NS5B polymerase are likely to be under HLA immune pressure and therapy selection.


HCV, hepatitis C virus; HIV, human immunodeficiency virus; HLA, human leukocyte antigen; IFN-α, interferon alpha; NS, nonstructural; PCR, polymerase chain reaction; STAT-C: specifically targeted antiviral therapy for HCV.

Patients and Methods


Individuals with chronic HCV genotype 1a (n = 205), 1b (n = 54), or 3a (n = 146) infection were recruited from Australia (1a = 90, 1b = 24, 3a = 61), Switzerland (1a = 62, 1b = 21, 3a = 37) and the United Kingdom (1a = 53, 1b = 9, 3a = 48). Of these individuals, 38.5% are coinfected with HIV (38.0%, 1a; 51.9% 1b; and 34.2%, 3a). All HCV sequences were obtained from HCV treatment-naïve individuals. Written informed consent was obtained from participants, and local Institutional Review Board approval was obtained by centers contributing to the study.

Viral Sequencing.

Viral RNA was obtained from plasma samples using the Cobas Amplicor HCV sample prep kit according to manufacturer's conditions (Roche Diagnostics). Complementary DNA conversion and first-round polymerase chain reaction (PCR) products were obtained as previously described using the SuperScript III OneStep RT-PCR System (Invitrogen).7 Overlapping second-round PCR products were amplified using Taq DNA Polymerase (Roche) that covered NS3 protease and NS5B polymerase. Genotype-specific and generic primers were used for first-round and second-round PCR (Supporting Table 1). PCR products were bulk-sequenced as previously described,7, 9 and mixtures were identified in which the secondary peak was more than 20% of the main peak. Because of the variability of the HCV genome, some samples failed to produce a PCR product, and as a result, some individuals do not have complete coverage of NS3 protease or NS5B polymerase.

HCV sequences contained the complete or partial sequence of NS3 protease or NS5B polymerase (from position 1 to 530). The mean NS3 coverage for HCV genotype 1a was 85.4% (n = 175), and mean NS5B coverage was 73.1% (n = 196). The mean NS3 coverage for HCV genotype 1b was 80.2% (n = 39), and mean NS5B coverage was 80.1% (n = 50). The mean NS3 coverage for HCV genotype 3a was 92.4% (n = 121), and mean NS5B coverage was 78.1% (n = 134).

HLA Genotyping.

HLA class I and II typing was performed as previously described (n = 185, 1a; n = 51, 1b; n = 137, 3a).7 The distribution of HLA alleles in the different cohorts is shown in Supporting Fig. 1.

Data Analysis.

HCV NS3 protease and NS5B polymerase sequences were aligned using the program ClustalW (http://www.ebi.ac.uk). A consensus sequence for each HCV genotype/subtype was constructed from the alignment based on the most common amino acid at each site. Variation from consensus was measured for each site within the two regions. Shannon entropy (accessible at http://hcv.lanl.gov) was used to measure the variation exhibited within, and between, HCV subtypes/genotypes for NS3 protease and NS5B polymerase.

Fisher's exact test was used to identify significant differences in the proportion of sequences with variation from consensus at a single site in individuals with or without a specific HLA.

HCV Drug Resistance Mutations.

A literature review of publications and conference proceedings was performed to create a list of published developmental protease and polymerase inhibitors and their resistance mutations.


Baseline Profile of Described STAT-C Drug Resistance Mutations in HCV Sequences from Treatment-Naïve Individuals with Chronic HCV Infection.

Known drug resistance mutations for the STAT-C drugs were observed in NS3 protease and NS5B polymerase sequences obtained from treatment-naïve HCV monoinfected and HIV coinfected individuals (Tables 4, 5). Alternative amino acids were also present at these sites that have not been previously observed or tested for using in vitro systems (Tables 4 and 5). In the HCV genotype 1a sequences, the frequency of described STAT-C resistance mutations was generally low, with the most frequent drug resistance mutation at NS3 protease T54S (4.4%; Table 4). Similarly, recent studies of NS3 protease and NS5B polymerase sequences from treatment-naïve individuals with chronic HCV genotype 1 infection showed a low frequency of baseline drug resistance mutations.10, 11

Table 4. Occurrence of Drug Resistance Mutations and Related Variations in NS3 Protease in Treatment-Naïve Individuals in Study Cohort
Drug Resistance MutationHCV Consensus/Variant(s)*
 Frequency of Variant in TotalFrequency of Variant in HIV-Co-infected Only Frequency of Variant in TotalFrequency of Variant in HIV-Co-infected Only Frequency of Variant in TotalFrequency of Variant in HIV-Co-infected Only
  • Some individuals do not have complete coverage of NS3, resulting in differences in the denominators for the different groups. Bold indicates difference in consensus between HCV genotypes and underlined amino acids indicate presence of tested or observed resistance mutation.

  • *

    Presence of mixture in sequence is counted as 0.5, and an unresolved site within the codon or mixtures resulting in more than two amino acids were not included in the analysis.

Table 5. Occurrence of Drug Resistance Mutations and Related Variations in NS5B Polymerase in Treatment-Naïve Individuals in Study Cohort*
Drug Resistance MutationHCV Consensus/Variant(S)*
 Frequency of Variant in TotalFrequency of Variant in HIV-Co-infected Only Frequency of Variant in TotalFrequency of Variant in HIV-Co-infected Only Frequency of Variant in TotalFrequency of Variant in HIV-Co-infected Only
  • *

    Some individuals do not have complete coverage of NS5B, resulting in differences in the denominators for the different groups. NS5B sequence covering the sites G554D, Y555C, G558R and D559G from Table 1 is not available from this cohort. Bold indicates difference in consensus between HCV genotypes and underlined amino acids indicate presence of tested or observed resistance mutation.

  • *

    Presence of mixture in sequence is counted as 0.5 and an unresolved site within the codon or mixtures resulting in more than two amino acids were not included in the analysis.


Despite the reduction in replicative fitness associated with mutations at many of the drug resistance sites (Table 3), only a fraction of these sites in both NS3 protease and NS5B polymerase exhibited complete conservation of the wild-type amino acid in the HCV genotype 1a sequences (3/11 for NS3 protease and 16/24 for NS5B polymerase; Tables 4 and 5). Note that the NS5B polymerase V138I mutation, a mutation associated with resistance to the developmental polymerase inhibitor A-782759, represented the consensus amino acid. Not surprisingly, this mutation has been shown to increase viral fitness in the replicon system (Table 3).

The amino acid composition at several of the drug resistance sites can vary between the HCV genotypes/subtypes, resulting in different consensus amino acids (NS3 proteases 36, 168, 170, and 176 and NS5B polymerase 71, 419, 447, 451, 482, and 499; Tables 4, 5). In some examples, the consensus for one HCV genotype/subtype is the reported drug resistance mutation (NS3 protease 36 and NS5B polymerase 71, 482, and 499; Tables 4, 5). Similar levels of variation at these drug resistance sites for the different genotypes/subtypes can be seen in HCV sequences obtained from the public database (Supporting Table 2). However, confounding issues such as the use of different methodologies and limited demographic and clinical history for each individual reduce the usefulness of this dataset.

In summary, 21.5% of individuals with HCV genotype 1a have a variation at one or more drug resistance sites, 44.4% for 1b and 41.8% for 3a. For those individuals with sequence in both NS3 protease and NS5B polymerase (n = 167 for 1a; 37 for 1b; 109 for 3a), 16.2% of 1a, 10.8% of 1b, and 18.3% of 3a sequences have variants at drug resistance sites in NS3 protease only, 7.2% of 1a, 35.1% of 1b, and 18.3% of 3a sequences have variants at drug resistance sites in NS5B polymerase only, and 0.6% of 1a, 2.7% of 1b, and 11% of 3a sequences have variants at drug resistance sites in both. It should be noted that at some sites the consensus and variant amino acids for genotypes 3a and 1b are different from that of genotype 1a, and this will inflate these proportions. However, these results suggest that a substantial proportion of individuals have some evidence of variation at known drug resistance sites.

Colocalization of Areas in NS3 Protease and NS5B Polymerase Under HLA-Driven Pressure and Therapy Selection.

In Figs. 2 and 3, published HLA-restricted epitopes and specific HLA-driven HCV-adaptations8 are mapped onto the NS3 protease and NS5B polymerase regions that contain most of the observed drug resistance sites. In both regions, several HLA-restricted epitopes overlap with, or are flanked by, known drug resistance sites (see A2-restricted, A68-restricted, DR4-restricted, and DR11-restricted epitopes in NS3 protease with positions 54, 153, 155, 156, 176; A24-restricted and B27-restricted epitopes in NS5B polymerase with positions 419, 423, 451). Also, epitopes for different HLA types can cover similar areas and overlap in amino acids (HLA-A2, HLA-A68, and HLA-DR4, Fig. 2A; HLA-B27 and HLA-A24, Fig. 3B). Thus, a single site may be within more than one HLA-restricted epitope and be subject to selection from more than one HLA type. Furthermore, we searched for evidence of potential HLA-driven immune pressure at drug resistance sites that exhibited variation within this cohort. Fisher's exact tests were performed at drug resistance sites with more than three sequence variations from consensus to identify associations with specific HLA types. This analysis identified HLA-associated variation (P ≤ 0.05) at drug resistance sites for genotype 1a: HLA-DRB1*0401 with position 54 and HLA-A*1101 with position 170 in NS3 protease; and with genotype 3a: HLA-C*0802 with position 176 in NS3 protease; HLA-A*0101 with position 50, HLA-A*1101 and HLA-A*0201 with position 71 in NS5B polymerase (Fig. 2).

Figure 2.

Overlap between drug resistance and immune escape profiles in NS3 protease. Amino acid diversity profile of the region between position 30-60 and 150-180 of NS3 protease in genotype 1a (A), 1b (B), and 3a (C) measured as the proportion of sequences with variation from consensus. Arrowheads indicate known protease inhibitor drug resistance sites within NS3 with position indicated. Published HLA-restricted HCV epitopes (http://hcv.lanl.gov and http://www.immuneepitope.org) are shown as black bars for the genotype described (epitope sequences are shown in brackets following the HLA allele; sequence falling within the window is underlined). HCV adaptations to immune pressure as found by Timm et al.8 are shown as open circles, and HLA-associated variation at drug-resistance sites within this cohort (P ≤ 0.05) are shown as black circles with the HLA-restriction indicated for both.

Figure 3.

Overlap between drug resistance and immune escape profiles in NS5B encoding the RNA-dependent RNA polymerase. Amino acid diversity profile of the region between position 90–150 and 400–460 of NS5B in genotype 1a (A), 1b (B), and 3a (C) measured as the proportion of sequences with variation from consensus. Arrowheads indicate known polymerase inhibitor drug resistance sites within NS5B, with the position indicated. Published HLA-restricted HCV epitopes and HCV adaptations to immune pressure8 are shown as in Fig. 2. (D) Histograms showing the proportion of HLA-B27+ (black box) and HLA-B27− (open box) individuals with variation from consensus within and flanking the published HLA-B27 epitope shown in (A+B) for genotype 1 and 3a. Subtypes 1a and 1b were combined, because the consensus sequence in this area was the same. P-values <0.005 (Fisher's exact test) are shown as an asterisk. Number of individuals expressing the HLA type and with sequence covering the region is shown for each HCV genotype.

To illustrate the effect HLA-driven pressure may have on the HCV sequence in an individual, variation within the published HLA-B27 epitope in NS5B12 for HCV genotype 1 was examined (Fig. 3D). A higher proportion of nonconsensus amino acids at sites within the epitope—including the drug resistance site 423—in those expressing HLA-B27 compared with HLA-B27–negative individuals was found (Fig. 3D). Two sites (421 and 426) exhibited a significant difference in the proportion of variation from consensus in the two groups. Position 421 has already been shown to be an HLA-driven HCV adaptation site for individuals with HLA-B27.8 A significant difference in the proportion of sequences with variation from consensus within this epitope is still observed when sites 421 and 426 are removed from the analysis (HLA-B27+ = 7/13 versus HLA-B27 = 19/201; P = 0.002). In contrast, there was almost no variation within this region in genotype 3 sequences irrespective of HLA type (Fig. 3D), which suggests that this area of NS5B is not under the same HLA-driven selection pressure among the different genotypes. To address the potential influence of a founder effect, we examined the relatedness of the NS5B sequences from the different cohorts and then specifically for the HLA-B27+ individuals (Supporting Fig. 2). The phylogenetic analysis of NS5B shows a lack of distinct cohort clusters for the genotypes and for the location of the HLA-B27+ individuals within the subtypes. The relationship between these sequences demonstrates that the link between viral polymorphism within the epitope and expression of HLA-B27 is not attributable to a founder effect.

We also used an epitope prediction program (Syfpeithi) to identify potential epitopes overlapping with drug resistance sites. This approach predicts that several drug resistance sites fall within predicted epitopes, including ones in which the emergence of a drug resistance mutation would reduce HLA-specific binding for the epitope (beneficial for virus) or uncover an HLA-restricted epitope (beneficial for the host; Table 6).

Table 6. Putative HLA-Restricted Epitopes That Overlap with Drug-Resistance Sites
A. Reduced HLA-Epitope Binding Potential Due to Drug-Resistance Mutation
ProteinPositionWT/VariantPredicted EpitopeRestricting HLABinding Score*
  • *

    Binding score obtained using the web-based program Syfpeithi (http://www.syfpeithi.de/). A score cutoff of 20 was used to indicate likely predicted epitopes. Only putative epitopes with twofold difference in binding potential because of variations at drug-resistance sites have been included in list..

B. Putative New Epitopes Uncovered with Drug-Resistance Mutation
ProteinPositionwt/variantPredicted EpitopeRestricting HLABinding Score*

Different Amino Acid Diversity Profiles for HCV Genotypes/Subtypes Suggest Differences in the Position/Type of Immune Escape and Drug Resistance Mutations.

To determine the differences in the amino acid diversity profiles of the HCV subtypes/genotypes in NS3 protease and NS5B polymerase, entropy plots (a measure of amino acid diversity) of the HCV genotypes 1a, 1b, and 3a sequences in this worldwide cohort were compared across these regions.

There are differences in the diversity observed at sites along these proteins for genotypes and, to a lesser extent, subtypes (Supporting Fig. 3). Tables 4 and 5 show that the number and type of naturally occurring variations observed within one genotype or subtype is not necessarily present for another genotype. This suggests that genotypes differ in their ability to mutate at sites along NS3 protease and NS5B polymerase and could reflect potential differences in STAT-C drug resistance and immune escape profiles. These findings are further supported by Figs. 2 and 3.


Recent reports have highlighted the existence of STAT-C drug resistance mutations in treatment-naïve individuals either monoinfected with HCV10, 13 or coinfected with HIV.11, 14 Other studies have suggested the pre-existence of quasispecies harboring drug resistance mutations at low frequencies, given their rapid appearance after commencement of drug therapy, particularly with protease inhibitors.15, 16 However, these studies have mainly focused on HCV genotype 1 or a particular class of inhibitors. Here, we analyzed NS3 protease and NS5B polymerase sequences from a large cohort of treatment-naïve individuals infected with HCV genotypes 1 and 3. The results show that variation at several of these drug resistance sites can occur among treatment-naïve individuals, generally at a frequency of between 0.5% and 5%, and in some cases the variant is the known drug resistance mutation.

The inclusion of individuals infected with genotypes 1b and 3a shows that the selection of STAT-C drug resistance may vary between HCV genotypes and possibly subtypes. These results suggest that variation at the drug resistance sites can occur in the absence of the specific drug pressure, and other selection pressures such as the host's immune response may be occurring at these sites. Furthermore, because very few individuals worldwide have been exposed to STAT-C drugs, the drug resistance variations identified here are unlikely to be attributable to the transmission of resistant viruses from individuals previously exposed to these drugs.

The selection pressures of anti-HCV therapy and the host's immune response may act in the same area and even on the same sites along the HCV genome as illustrated in Figs. 2 and 3 and by the model shown in Fig. 1. In this study, we highlighted areas of the HCV genome in which HLA and drug selection pressures overlap. Accordingly, the overlap of these selection pressures suggests that drug resistance mutations might arise through immune responses in individuals carrying the relevant HLA allele. This could potentially affect a large proportion of subjects, because the frequency of certain HLA alleles can vary from approximately 45% carrying HLA-A2 to 5% with HLA-B27 (http://www.allelefrequencies.net). In these individuals, the risk of treatment failure caused by preexisting resistance mutations would be increased. Alternatively, the selection of drug resistance mutations that abrogate HLA binding or peptide processing may impair the host's immune response and potentially alter therapy outcome. Furthermore, anti-HCV drugs are likely to provide a stronger selection force than the host's immune response, and mutations that can cause immune escape but are not normally observed in the circulating viruses because of a high fitness cost may become visible with the introduction of a specific anti-HCV drug, with subsequent effects on the host's immune response.

The overlap between the two selection pressures shown in this study is likely to be a conservative measure, because the number of HLA-restricted epitopes described in the literature is an underestimate of the true number of epitopes that exist in NS3 protease and NS5B polymerase (Table 6). Furthermore, drug and “immune” resistance profiles are likely to differ between the genotypes given the differences in the consensus and variation profile observed in the combined cohort. Hence, knowledge of an individual's HLA type, HCV genotype/subtype, and baseline resistance profile before the commencement of STAT-C therapy is likely to become relevant with respect to individual HCV treatment strategies.

The overlap between drug-driven and HLA-driven pressures may be particularly significant when immune-based interferon alpha (IFN-α)/ribavirin therapy is combined with STAT-C drugs. IFN-α–based therapy can reduce viral load in an individual and could create a bottleneck phenomena in which low-frequency viral species that contain the immune or drug-resistant strain may quickly become the dominant viral strain conferring resistance to STAT-C. These changes are advantageous to the newly arising strain, which may have a selective advantage for both immune and drug pressure. The converse also may be true, in which the two selective forces may act in the opposite direction, potentially reducing the likelihood of the appearance of the resistant/escape mutation. To what extent IFN-α therapy drives viral dynamics and HLA-driven adaptation remains a key question and needs further study. Furthermore, a detailed examination of intraindividual variation using cloning techniques may provide additional information regarding the evolution and importance of low-level drug resistance variants (<20%). Finally, the frequency measures given in this study did not differentiate between variations that are associated with “drug class” resistance versus those that may be drug-specific or less important in the context of drug resistance based on in vitro assays. However, the hierarchy of resistance mutations to STAT-C drugs has not yet been established, and the relevance of genetic variation at known (and new) drug resistance sites remains to be elucidated in subsequent clinical trials and with widespread use of STAT-C drugs.

In conclusion, the overlap of immune escape and drug resistance profiles for HCV suggests that knowledge of the host HLA type and HCV subtype/genotype may provide important information in defining an individual's drug regimen. In the field of HIV medicine, it has already been established that synergistic effects of antiretroviral drug resistance and HLA-driven HIV adaptation results in an increased frequency of the resistant viral strain in individuals expressing the relevant HLA type and undergoing HIV treatment with the specific drug.17 We contend that this issue may be even more relevant to the management of HCV infection, in which immune-modulating IFN-α therapy will likely remain the cornerstone of future combination treatment regimens. Accordingly, it may be useful to develop a “stratification of risk” for drug resistance based on host and viral genetics before the commencement of STAT-C in combination with IFN-α.


The authors thank Dr. Larry Park, Kiloshni Naidoo, Susan Herrmann, Marion McInerney, Saroj Nazareth, Maria Baccala, Silvie Miczkova, and Dr. Andrew Lucas for their input into the study. We thank the individuals and clinical staff at the different sites that have contributed to this study and the Swiss HIV Cohort Study.

Supported by the National Health and Medical Research Council of Australia (Grant numbers 334603 and 384702), the Haemophilia Foundation of Australia, the Swiss National Science Foundation (grant no. 3347-069366 and 324700-116862), NIHR Biomedical Research Centre Programme and the Medical Research Council (UK). M.L. was supported by a Hepatology/Immunology Research Fellowship from Royal Perth Hospital and Murdoch University. A.R. was supported by a Fellowship for Prospective Researchers from the Swiss National Science Foundation. The members of the Swiss HIV Cohort Study are M. Battegay, E. Bernasconi, J. Böni, H C. Bucher, Ph. Bürgisser, A. Calmy, S. Cattacin, M. Cavassini, R. Dubs, M. Egger, L. Elzi, P. Erb, M. Fischer, M. Flepp, A. Fontana, P. Francioli (President of the SHCS, Centre Hospitalier Universitaire Vaudois, CH-1011-Lausanne), H. Furrer (Chairman of the Clinical and Laboratory Committee), C. Fux, M. Gorgievski, H. Günthard (Chairman of the Scientific Board), H. Hirsch, B. Hirschel, I. Hösli, Ch. Kahlert, L. Kaiser, U. Karrer, C. Kind, Th. Klimkait, B. Ledergerber, G. Martinetti, B. Martinez, N. Müller, D. Nadal, M. Opravil, F. Paccaud, G. Pantaleo, A. Rauch, S. Regenass, M. Rickenbach (Head of Data Center), C. Rudin (Chairman of the Mother & Child Substudy), P. Schmid, D. Schultze, J. Schüpbach, R. Speck, P. Taffé, P. Tarr, A. Telenti, A. Trkola, P. Vernazza, R. Weber, S. Yerly.

The following GenBank accession numbers are for the NS3 protease and NS5B polymerase nucleotide sequences described in this manuscript: EF139658-EF139663, EF139665, EF139669-EF139684, EF139686-EF139687, EF139690, EF139692-EF139693, EF139695, EF139697-EF139699, EF139701-EF139706, EF139708-EF139710, EF139712-EF139725, EF139727-EF139739, EF139741-EF139742, EF139745-EF139757, EF139759-EF139766, EF139769-EF139771, EF139773, EU709893-EU7-10290, EU710292-EU710135, EU710137-EU710464, EU710467-EU710516.