| |
Natural Prevalence of Hepatitis C Virus Variants with Decreased Sensitivity to NS3·4A Protease Inhibitors (Vertex's telaprevir) in Treatment-Naive Subjects
|
| |
| |
The Journal of Infectious Diseases Sept 15 2008;198:800-807
Doug J. Bartels,a Yi Zhou,a Eileen Z. Zhang, Michelle Marcial, Randal A. Byrn, Thomas Pfeiffer, Ann M. Tigges, Bambang S. Adiwijaya, Chao Lin, Ann D. Kwong, and Tara L. Kieffer
Department of Infectious Diseases, Vertex Pharmaceuticals, Inc., Cambridge, Massachusetts
"In conclusion, the present study suggests that naturally occurring variants with decreased sensitivity to HCV protease inhibitors were uncommon (detected at <1% each) in HCV treatment-naive patients. Treatment with TVR/PR efficiently inhibited the V36M and R109K variants. Furthermore, TVR contributed additional antiviral activity against the R155K variant over that provided by PR alone. As new HCV agents are evaluated in clinical trials, it will be important to monitor the effect of baseline variants on sensitivity."
ABSTRACT
Background. The prevalence and clinical implications of naturally occurring variants that are resistant to hepatitis C virus (HCV) protease inhibitors in treatment-naive patients has not been reported. We report here the prevalence of such variants and their effect on clinical response.
Methods. Population sequence analysis of the NS3·4A protease was conducted in 570 treatment-naive subjects.
Results. Most subjects (98%) had wild-type virus. The remaining subjects had the following variants present in significant proportions (100%): V36M, 0.9%; R155K, 0.7%; V170A, 0.2%; and R109K, 0.2%. The V36M, R109K, and V170A substitutions confer low-level resistance (<7-fold) to protease inhibitors in replicon cells. The R155K substitution confers low-level resistance to telaprevir (TVR) and boceprevir and confers high-level resistance (>70-fold) to BILN 2061 and ITMN-191. Five subjects with the V36M or R109K variant were treated with 8-24 weeks of TVR and peginterferon-α2a (P) with or without ribavirin (R). Four achieved a sustained viral response, and 1 was lost to follow-up. In subjects with the R155K variant, TVR/PR provided greater antiviral activity than PR alone; however, the antiviral response was lower than that observed in subjects with wild-type virus.
Conclusion. High levels of naturally occurring protease inhibitor-resistant variants were uncommon (<1% each) in HCV treatment-naive patients. TVR/PR efficiently inhibited V36M and R109K variants and contributed partial antiviral activity against the R155K variant. As new HCV agents are evaluated in clinical trials, it will be important to monitor the effect of baseline variants on sensitivity.
Hepatitis C virus (HCV) NS5B polymerase and NS3·4A protease are essential for viral replication, and compounds that inhibit these enzymes comprise a new class of drugs known as gspecifically targeted antiviral therapy for HCVh (STAT-C) [1]. HCV has extremely high sequence diversity [2], allowing for the viral quasispecies present in subjects before treatment to exist as a mixed population of predominantly wild-type (WT) virus and low levels of variants with varying degrees of resistance to STAT-C drugs. Thus, the natural presence of resistant variants at baseline offers the potential for their rapid selection in subjects treated with STAT-C. The prevalence of a resistant variant at baseline is determined by its replicative fitness compared with that of other variants. In general, amino acid changes in or near the highly conserved active site of an enzyme are poorly tolerated, resulting in diminished replicative capacity and less-fit variants. In contrast, changes conferring resistance to allosteric, non-active-site inhibitors tend to be better tolerated and more fit, potentially resulting in a higher baseline prevalence compared with that of active-site variants.
Numerous drug resistance substitutions have been shown to develop in vitro in the presence of STAT-C. HCV polymerase inhibitors can be divided into nucleoside (active-site) inhibitors and nonnucleoside (allosteric-site) inhibitors. The S282T amino acid change in the NS5B gene has been shown to confer resistance to the HCV polymerase nucleoside inhibitors NM283 [3, 4], MK-0608 [5], and R7128 (a prodrug of PSI-6130) [6], and the S96T and S96T/N142T changes have been shown to confer resistance to R1626 (a prodrug of R1479) [7]. Amino acid substitutions in NS5B that confer resistance in vitro to allosteric inhibitors include P495L/A (JTK-109/benzimidazole) [8, 9]; C316F/Y and S365T/A (HCV-796) [10]; and H95Q, N411S, M414L/T, and Y448H (A-782759) [11]. Amino acid changes in NS3 that confer decreased sensitivity to linear HCV protease inhibitors in replicon cells include A156V/T and R109K (SCH6) [12]; T54A, A156S/T, and V170A (SCH 503034, or boceprevir) [13]; and A156V/T/S (VX-950, or telaprevir [TVR]) [14, 15]. Amino acid changes in the NS3 gene that confer decreased sensitivity to macrocyclic inhibitors in HCV replicon cells include R155Q, A156V/T, and D168V/A (BILN 2061) [11, 14, 16] and D168A/V/E, A156S/V, F43S, Q41R, and S138T (ITMN-191) [17]. Decreased sensitivity to ITMN-191 is also conferred by changes in the NS3 helicase (S489L) and NS4A (V23A) regions.
In clinical trials, sequence analysis in patients with viral rebound has identified variants resistant to a number of polymerase inhibitors, such as C316Y (HCV-796) [18], and protease inhibitors, such as T54A, V36M, R155K/T/I, A156S/V/T, and V36M+R155K (TVR) [19] and V36M, T54A, R155K, A156S, V170A/T, and V36M+R155K (boceprevir) [20, 21]. In vitro studies with interferon (IFN)-_ and ribavirin (RBV) in HCV replicons containing mutations conferring resistance to HCV polymerase and protease inhibitors have revealed that these variants remain sensitive to both of these compounds [11, 21-24]. Several clinical studies have revealed that the depth and durability of the antiviral response to STAT-C polymerase or protease inhibitors is increased when combined with peginterferon-α2a or 2b and RBV [25-30]. These results suggest that the addition of these compounds increases the antiviral effect both by decreasing the replication of WT as well as STAT-C-resistant variants and by boosting the host immune system to eventually clear the virus [31].
Recently, the first report of a single protease inhibitor-naive patient harboring R155K as the dominant quasispecies [32] emphasizes the need to assess the natural prevalence of HCV NS3 variants carrying mutations previously associated with decreased sensitivity to protease inhibitors. We report here the low prevalence of these variants as the majority quasispecies in 570 treatment-naive patients and their effect on clinical response in individuals who are chronically infected with genotype 1 virus.
Discussion
Successful treatment of viral diseases can be limited by drug resistance, which can arise from 1 of 2 sources: (1) incomplete viral suppression during the course of therapy and (2) preexisting viral variants before therapy is initiated. HCV replication is highly efficient but utilizes an error-prone polymerase, which generates a diverse population of viral variants. Therefore, the speed and degree to which viral replication can be inhibited during therapy with an inhibitor is critical to limit the generation of new resistant variants. Other factors that affect the development of resistance during therapy include the genetic barrier, dosage, duration, and compliance.
In treatment-experienced HIV- and HBV-infected patients, the development of resistance during treatment can limit subsequent treatment options. In treatment-naive patients, naturally occurring polymorphisms that confer resistance to therapy can similarly affect response [34, 35]. In this regard, allosteric inhibitors are usually more sensitive to the effects of polymorphisms that confer resistance, because these amino acid changes do not typically impair fitness as much as substitutions located in the active site of an enzyme and therefore tend to be more prevalent. Polymorphisms conferring different levels of resistance to direct-acting HIV inhibitors have been observed in treatment-naive patients [36-38]. Naturally occurring variants with substitutions conferring decreased sensitivity to direct-acting HCV polymerase inhibitors have also been reported in treatment-naive patients. For example, the M414T substitution in the HCV polymerase, which confers high-level resistance against benzothiadiazine inhibitors (such as A-782759), was found in the serum of many treatment-naive HCV-infected patients at a range of 0.11% to 0.60% [39]. In addition, the C316Y substitution, which resulted in high-level resistance against another benzothiadiazine inhibitor, HCV-796 (A-837093), was observed in a chimpanzee that received A-837093 [40] and in HCV-infected patients treated with HCV-796 alone [18] or in combination with peginterferon [27]. There is significant polymorphism at residue 316 in the genotype 1 HCV polymerase, with the low-resistance variant C316N observed in almost 20% of HCV isolates in GenBank and C316Y reported in 1 isolate.
Here, we report for the first time the natural prevalence of variants with decreased sensitivity to HCV protease inhibitors and their effect on clinical response. Population sequence analysis of the full-length NS3·4A protease was conducted in 570 treatment-naive subjects (before administration of drug) in two phase 2 clinical studies of TVR combination therapy. Because TVR is an active-site HCV protease inhibitor and because previously described resistance mutations have been shown to have reduced fitness in vitro and in vivo, the vast majority of the baseline sequences were expected to be primarily WT. As predicted, the majority of subjects (98%) in this large sample had WT virus at baseline. Additionally, a very small number of patients (0.9%) had predominantly the V36M variant at baseline. Perhaps more unexpectedly was the detection of a smaller number of subjects (0.7%) with high levels of the R155K variant at baseline. Previously we have shown WT virus reemerging in patients in whom the resistant R155K variant had developed during a short course of TVR monotherapy, and both in vitro and in vivo studies have shown that the R155K variant had impaired fitness compared with WT [19, 22]. Although it is unclear why the R155K variant is dominant in these patients, it is possible that the R155K substitution is present in conjunction with other substitutions that improve fitness and allows replication at a level similar to that of WT virus.
In addition to understanding the frequency of a resistant variant or resistance polymorphism in the treatment-naive population, it is critical to address the potential impact on treatment outcome. In the HCV replicon system, the V36M, R109K, and V170A substitutions conferred low-level resistance to the representative linear and macrocyclic HCV protease inhibitors studied (TVR, boceprevir, BILN 2061, and ITMN-191) and are fully sensitivity to IFN-α, RBV, and representative HCV polymerase nucleoside and nonnucleoside inhibitors [11, 21-24]. Subjects in whom the V36M or R109K variant was predominant at baseline and who were treated with TVR/PR had a rapid decline in plasma HCV RNA level that was similar to the decline observed in subjects with WT virus receiving the same regimen. Although the number of patients is small, it is encouraging that all of these patients (except for one whose status is unknown because of discontinuation from the study) achieved an SVR, suggesting that the combination of TVR and PR is sufficient to suppress and clear the V36M variant. Thus, patients with the V36M variant may have treatment outcomes similar to those of patients with WT virus.
In the HCV replicon system, the R155K substitution also conferred a low-level decrease in sensitivity to linear HCV protease inhibitors (TVR and boceprevir). In contrast, R155K conferred high-level resistance to macrocyclic inhibitors (BILN 2061 and ITMN-191). The R155K variant showed no decrease in sensitivity to IFN-α, RBV, or HCV polymerase inhibitors, compared with WT [11, 21-24]. Subjects in whom the R155K variant was predominant at baseline and who were treated with TVR/PR had a different antiviral response than did the subjects in whom WT virus or the V36M variant was predominant. In subjects with R155K treated with TVR/PR, the initial decline in plasma HCV RNA level was more rapid and of greater magnitude than the average decline on PR treatment for either the 75 subjects with WT-dominant quasispecies or the single R155K-dominant quasispecies. However, the overall decline in HCV RNA level was slower in subjects with the R155K variant than in subjects with WT virus or the V36M variant, suggesting that the combination of TVR/PR is more advantageous than PR alone but is less effective in suppressing the R155K variant than WT virus. It is interesting that there is a distinct difference in clinical response to TVR/PR between the R155K variant and the V36M variant, given that their in vitro sensitivity to TVR is similar. The reason for this difference is unclear; however, HCV replication dynamics in patients is undoubtedly more complex than a noninfectious, subgenomic HCV replicon system in Huh7 cells. The R155K substitution does not decrease sensitivity to IFN-α, RBV, or HCV polymerase inhibitors and, hence, should not be deleterious to the use of these agents in combination.
In conclusion, the present study suggests that naturally occurring variants with decreased sensitivity to HCV protease inhibitors were uncommon (detected at <1% each) in HCV treatment-naive patients. Treatment with TVR/PR efficiently inhibited the V36M and R109K variants. Furthermore, TVR contributed additional antiviral activity against the R155K variant over that provided by PR alone. As new HCV agents are evaluated in clinical trials, it will be important to monitor the effect of baseline variants on sensitivity.
Results
Baseline sequence analysis. Population-based sequencing of the NS3 protease catalytic domain from baseline samples was successful in 570 of 573 subjects. The majority of subjects (559/570 [98%]) had WT virus at baseline. Of the remaining 11 subjects (all with genotype 1a), 5 had the V36M substitution (0.9%), 4 had the R155K substitution (0.7%), 1 had the V170A substitution (0.2%), and 1 had the R109K substitution (0.2%) present in significant proportions (100%) at baseline. As shown in table 1, these variants have been rarely observed (<0.05%) in the public HCV sequence databases. This is consistent with a previous finding that the R155K variant had a decreased fitness compared with WT virus in the absence of drug [19].
Therefore, the treatment histories of the subjects with resistant variants at baseline were thoroughly reviewed with the investigators, and no evidence for prior exposure to HCV protease inhibitors or any other investigational antiviral agents could be elicited.
Sensitivity of resistant variants to HCV protease inhibitors. To evaluate the potential effect of the resistance substitutions in the HCV NS3 protease domain, HCV subgenomic replicon cell lines containing the individual mutations were created. The sensitivity of WT, V36M, R109K, R155K, and V170A replicon variants to a variety of linear and macrocyclic protease inhibitors (TVR, boceprevir, BILN 2061, and ITMN-191) was determined (figure 1). These experiments showed that the V36M, R109K, and V170A substitutions confer a low level of resistance (defined as a FC value of <10) to the linear protease inhibitors TVR and boceprevir and no or little resistance to the macrocyclic inhibitors BILN 2061 and ITMN-191. Similarly, the R155K substitution resulted in a low level of resistance to the linear protease inhibitors. In contrast, the R155K substitution conferred a high level of resistance to the macrocyclic inhibitors (250-fold increase in resistance to BILN 2061 and 62-fold increase in resistance to ITMN-191). All variants remained fully sensitivity to IFN-α and RBV (data not shown) and had a slightly reduced in vitro replication capacity, consistent with their previously described diminished in vivo fitness [19, 22].
Antiviral response to TVR/PR in subjects with resistant variants predominant at baseline. Eleven subjects with V36M, R109K, R155K, or V170A variants predominant at baseline were treated with TVR/PR, TVR/P (no RBV), or PR alone. Five of 6 subjects with either the V36M or the R109K variant at baseline were treated with TVR/PR or TVR/P, and all of these subjects exhibited a rapid decline in plasma HCV RNA to undetectable levels (<10 IU/mL), similar to the one observed in most subjects with WT virus at baseline receiving the same regimen. One subject with the V36M variant discontinued treatment after 10 weeks of TVR/PR and had an undetectable HCV RNA level at that time point but was subsequently lost to follow-up (data not shown). The remaining 4 of these 5 subjects achieved a sustained viral response (SVR; defined as an HCV RNA level of <10 IU/mL 24 weeks after the end of treatment) and are shown in figure 2: subject A (V36M) completed 8 weeks of TVR/PR before discontinuing therapy early and achieved an SVR (figure 2A); subject B (V36M) completed 12 weeks of TVR/P and achieved an SVR (figure 2B); subject C (V36M) completed 12 weeks of TVR/PR followed by 12 additional weeks of PR and achieved an SVR (figure 2C); and subject D (R109K) completed 12 weeks of TVR/PR and achieved an SVR (figure 2D). Subject E (V36M) was in the control arm and had an undetectable HCV RNA level after 36 weeks of PR (figure 2E). Subject F had the V170A variant, received PR, and had an undetectable HCV RNA level at week 36 (figure 2F).
Of the 4 subjects with the R155K variant at baseline, 3 were treated with TVR/PR and exhibited an initial decline in plasma HCV RNA level (figure 3). Subject A received 8 weeks of TVR/PR and reached an HCV RNA nadir of 5030 IU/mL before discontinuing therapy (figure 3A). Subject B received 12 weeks of TVR/PR followed by 12 additional weeks of PR and had a viral breakthrough with the V36M/R155K variant at week 4, but the HCV RNA level declined to 291 IU/mL while on PR (figure 3B). Subject C received 12 weeks of TVR/PR followed by 36 additional weeks of PR and experienced a decline in HCV RNA to an undetectable level (<10 IU/mL) by week 20, followed by a viral breakthrough with R155K at week 36 and a subsequent decline to 42 IU/mL by the end of therapy (figure 3C). Subject D was in the control arm and received 14.5 weeks of PR before discontinuing therapy. In this subject, the HCV RNA level declined 1 log10 from baseline by week 5, and WT virus became predominant over the R155K variant by week 12 and continued to be dominant through the follow-up sample collected at week 16 (1.5 weeks after the end of therapy) (figure 3D).
A viral dynamic model was used to estimate the relative contribution of TVR and PR to the antiviral inhibition of the R155K variant. The decline in plasma HCV RNA level from baseline at the transition between 2 phases of HCV RNA decline slopes represents the extent of inhibition [33]. This transition occurred between days 1 and 3. In the 3 subjects treated with TVR/PR, the log10 declines in HCV RNA level on day 3 were -0.79, -1.11, and -1.16 (mean ± SD, -1.02 ± 0.20). The mean TVR/PR HCV RNA decline on day 3 was 0.31 log10 greater than the median decline of -0.71 observed in subjects receiving PR (n=75), suggesting that TVR provided some antiviral contribution in subjects with R155K-dominant quasispecies. In comparison, the log10 decline in plasma HCV RNA level on day 3 in subjects with WT-dominated quasispecies who were treated with TVR/PR was a mean ± SD of -3.86 ± 0.80 (n=175), suggesting reduced telaprevir inhibition in subjects with R155K-dominant quasispecies.
Methods
Subject population. The study included 573 treatment-naive subjects who had chronic genotype 1 HCV infection and who were enrolled in studies VX05-950-104 or VX05-950-104EU. The study was conducted in full compliance with good clinical practice guidelines and those of the World Medical Assembly Declaration of Helsinki. Before study initiation, the protocol and informed consent form were reviewed and approved by the institutional review boards at each site. All patients provided written informed consent before participating in any study-related activity. All subjects were between 18 and 65 years of age, had a detectable baseline plasma HCV RNA level, and were hepatitis B surface antigen and HIV antibody negative. Plasma HCV RNA levels were determined using the COBAS TaqMan HCV/HPS assay (Roche Molecular Systems). The lower limit of quantitation for the HCV RNA assay was 30 IU/mL, and the limit of detection (LOD) was 10 IU/mL.
Subjects were randomized to receive either (1) 750 mg of TVR every 8 h, peginterferon-α2a (P) at 180 μg/week, and RBV (R) at 1000-1200 mg/day for 12 weeks followed by 0, 12, or 36 weeks of PR or (2) TVR/P (no RBV) for 12 weeks. The control arm received 48 weeks of placebo/PR.
Amplification and sequencing of HCV from subject plasma. Population sequence analysis of the full-length NS3·4A protease was conducted in 570 treatment-naive subjects infected with genotype 1 HCV (389 with genotype 1a and 241 with genotype 1b) before administration of drugs (day 1). All patients were enrolled in the study and randomized as described above to receive TVR/PR or only PR for varying durations. Results are presented here for patients containing known protease variants on day 1, for whom sequencing was done at specified time points during treatment and throughout the study. A 4-mL blood sample was collected from subjects by venipuncture of a forearm vein into tubes containing EDTA (K2) anticoagulant. Plasma was separated by 10 min of centrifugation, aliquoted, and stored at -80°C. Sequence analysis of HCV was done by nested reverse-transcriptase polymerase chain reaction (PCR) amplification of an 9-kb HCV RNA fragment spanning the HCV polyprotein coding region (H77; nt 286-9277). The DNA from this PCR was purified using the QIAquick 96 PCR purification kit (Qiagen) and was analyzed on an agarose gel. The quality and quantity of the purified PCR product were measured by use of an EnVision Multilabel Reader (PerkinElmer). Purified DNA was sent to Agencourt Biosciences for sequencing of the NS3·4A protease region. Oligonucleotide primers for sequencing of the NS3 protease domain were 1a3278F (5'-GGAGACCAAGCTCATCACGTGG-3'), 1a3316F (5'-GCGTGCGGTGACATCAT-3'), 1a3710R (5'-GCAGGGTGTCAATGAGCGGG-3'), and 1a3928R (5'-ATAAAGTCCACCGCCTTAGCCACTCC-3') or 1b3315F (5'-GCGTGTGGGGACATCAT-3'), 1b3756F (5'-TCATTCCGGTGCGCC), 1b3772R (5'-GGCGCACCGGAATGA-3'), and 1b4204R (5'-CCCGTGGTGATGGTCCT-3'). The sequencing assay was successful in samples containing >1000 IU/mL HCV RNA.
Sequence alignment and analysis. Sequences were aligned and analyzed for the presence of previously identified resistance substitutions in the NS3 protease catalytic domain by means of Mutational Surveyor software (version 3.2; SoftGenetics). The GenBank accession numbers for the baseline sequences of the NS3 protease for the 10 patients shown in figure 2A-2F and figure 3A-3D of the present study are as follows: 2A, EU853661; 2B, EU853662; 2C, EU853663; 2D, EU853664; 2E, EU853665; 2F, EU853666; 3A, EU853667; 3B, EU853668; 3C, EU853669; and 3D, EU853670.
Determination of replicon IC50 values in HCV replicon cells with NS3 protease mutations. The V36M, R109K, R155K, and V170A mutations were introduced into HCV replicons, and stable cells were selected as described elsewhere [14]. The IC50 values for HCV protease inhibitors (TVR, boceprevir, BILN 2061, and ITMN-191), HCV polymerase inhibitors, IFN-α, and RBV were determined in a 48-h assay using HCV replicon cells, as described elsewhere [14]. Multiple independent assays were conducted for each replicon variant, and the mean and SD of the replicon IC50 values were calculated. The fold change (FC) in IC50 for the HCV protease inhibitors was calculated by dividing the mean IC50 for the protease inhibitor against each variant by the average IC50 for the same HCV protease inhibitor against the WT (mADE) replicon cells. Variants were considered to have a low level of resistance if the increase in FC value was <10 and a high level of resistance if the increase was >50.
|
|
| |
| |
|
|
|
