Characterization of Resistance to the Protease Inhibitor GS-9451 in Hepatitis C Virus-Infected Patients

Abstract

GS-9451, a novel hepatitis C virus (HCV) nonstructural 3/4a (NS3/4a) protease inhibitor, is highly active in patients infected with HCV genotype 1 (GT 1). The aim of this study is to characterize the clinical resistance profile of GS-9451 in GT 1 HCV-infected patients in a phase 1, 3-day monotherapy study. The full-length NS3/4A gene was population sequenced at baseline, on the final treatment day, and at follow-up time points. NS3 protease domains from patient isolates with emerging mutations were cloned into an NS3 shuttle vector, and their susceptibilities to GS-9451 and other HCV inhibitors were determined using a transient replication assay. No resistance mutations at NS3 position 155, 156, or 168 were detected in any of the baseline samples or in viruses from patients treated with 60 mg of GS-9451 once daily. Among patients who received 200 mg and 400 mg of GS-9451, viruses with mutations at position D168 (D168E/G/V) and R155 (R155K), which confer high-level resistance to GS-9451, were detected in those with GT 1b and GT 1a virus, respectively. Viruses with D168 mutations were no longer detected in any GT 1b patient at day 14 and subsequent time points. In GT 1a patients, R155K mutants were replaced by the wild type in 57% of patients at week 24. These NS3 clinical mutants were sensitive to NS5B and NS5A inhibitors, as well as alpha interferon (IFN-α) and ribavirin. The lack of cross-resistance between GS-9451 and other classes of HCV inhibitors supports the utility of combination therapy.

INTRODUCTION

Hepatitis C virus (HCV) infects an estimated 170 million people around the world (2, 19, 21). Infection can lead to cirrhosis and, sometimes, to hepatocellular carcinoma (1, 13). Prior to May 2011, when the two protease inhibitors (PIs) telaprevir and boceprevir were approved, treatment of chronic HCV infection included a combination of pegylated interferon (PEG-IFN) and ribavirin (RBV) (5, 13, 21). This treatment is associated with significant side effects, such as fever, fatigue, anemia, leucopenia, thrombocytopenia, and depression (3, 14, 24), and results in sustained virologic response (SVR) in only 42% to 53% of patients with HCV genotype 1 (GT 1) and GT 4, respectively, and up to 78% to 82% of patients infected with HCV GT 2 or 3 (5, 13).

Direct-acting antivirals (DAAs), including the HCV nonstructural (NS) 3/4a serine protease inhibitors (NS3 PIs), have demonstrated antiviral activity in HCV-infected patients (6, 8). Among NS3 PIs, telaprevir and boceprevir have recently been approved for genotype 1 infections. There are more than 9 other NS3 PIs in different stages of clinical development (TMC-435, danoprevir, vaniprevir, BI201335, narlaprevir, MK-5172, asunaprevir, BMS-791325, ABT-450, ACH-1625, GS-9451, and GS-9256). The approved NS3 PIs have demonstrated increased SVR rates in patients when combined with PEG-IFN plus RBV. During the phase 2b PROVE2 study, genotype 1 (GT 1)-infected individuals treated with 12 weeks of telaprevir plus PEG-IFN plus RBV followed by 12 additional weeks of PEG-IFN plus RBV had SVR rates of 60%, compared to 46% in the standard-of-care-alone arm (6). Similarly, the boceprevir SPRINT-1 trial reported a 75% SVR rate for patients treated with a 4-week PEG-IFN-plus-RBV lead-in followed by boceprevir plus PEG-IFN plus RBV for 44 weeks, compared to 38% SVR in the PEG-IFN-plus-RBV-alone arm (8). Thus, proof-of-concept for the addition of an HCV NS3 protease inhibitor to PEG-IFN plus RBV for GT 1 HCV-infected patients has been established. However, because of the short half-lives of telaprevir and boceprevir, these agents require frequent dosing (every 8 h) with large numbers of pills (6 and 12 per day, respectively), which may adversely impact adherence. Telaprevir and boceprevir have been associated with adverse events, such as rash, pruritus, anemia, and dysgeusia. Furthermore, these PIs have also been found to select for viral resistance during monotherapy or combination studies in chronic HCV patients. Telaprevir selected multiple NS3 mutations in the clinic, including V36A/M, T54A, R155K/T, and A156S/V/T (18). Boceprevir selected NS3 mutations T54A and V170A during phase 1 studies (28, 29). Viral variants with amino acid changes at one or more of the amino acid positions 80, 155, and/or 168 of NS3 were detected in each of the patients treated with the macrocyclic protease inhibitor TMC435 (17). Furthermore, substitutions at NS3 positions 155 and 168 have been reported to be related to viral rebound in a 14-day multiple ascending dose trial of the HCV protease inhibitor ITMN-191 (danoprevir) (4).

GS-9451 (Fig. 1) is a novel, reversible, noncovalent inhibitor of the HCV NS3 serine protease with a 50% effective concentration (EC₅₀) of 7 to 10 nM and a 50% cytotoxic concentration (CC₅₀) of >50,000 nM in replicon cell assays (20, 25). In biochemical assays, GS-9451 has a K_i of 0.41 nM against GT 1 NS3 protease. In clinical studies, GS-9451 was well tolerated (9). In addition, a QD (once-daily) dosing of GS-9451 has shown highly effective antiviral activity in GT 1-infected patients in monotherapy (9) and is currently being evaluated in combination with other DAAs and RBV with or without PEG-IFN.

Fig 1.

GS-9451 structure.

This study characterizes the viral profile of resistance to GS-9451 in patients treated with multiple ascending doses of GS-9451 for 3 days.

MATERIALS AND METHODS

Compounds.

IFN-αA human and ribavirin (1-β-d-ribofuranosyl-1,2,4-triazole-3-carboxamide) were purchased from Sigma-Aldrich (St. Louis, MO). Telaprevir and boceprevir were purchased from Acme Biosciences (Belmont, CA). All other compounds (GS-9451, GS-9256, GS-6620, GS-5885, GS-9190, GS-9669, danoprevir, and TMC-435) were synthesized by Gilead Sciences (Foster City, CA).

Patient population and study design.

Forty patients were enrolled in a randomized, double-blind, placebo-controlled multiple ascending dose study designed to investigate the safety, tolerability, pharmacokinetics, and antiviral activity of GS-9451 in 4 cohorts of HCV-infected patients: 3 cohorts with HCV genotype 1a and 1 cohort with genotype 1b. In all cohorts, oral tablets of GS-9451 or matching placebo were administered once daily on days 1 to 3. Among these patients, GS-9451 was administered at 60 mg QD (n = 8), 200 mg QD (n = 9), and 400 mg QD (n = 8) in GT 1a patients and at 200 mg QD (n = 7) in GT 1b patients. All patients received a capsule formulation or matching placebo (n = 8). Blood samples for determining plasma HCV RNA levels were collected prior to study drug dosing on days 1, 2, and 3. Additional samples were collected at 12 (day 3), 24 (day 4), 48 (day 5), and 96 (day 7) hours after the last dose of study drug. Plasma HCV RNA was analyzed by real-time PCR (RT-PCR) using the COBAS TaqMan RT-PCR HCV assay, version 2.0, with the High Pure system (quantitation range, 25 IU/ml to 300 × 10⁶ IU/ml; Roche Molecular Systems, Inc., Branchburg, NJ).

All patients had a chronic infection with subtype 1a or 1b HCV with plasma HCV RNA levels of ≥10⁵ IU/ml and no evidence of coinfection with hepatitis B virus (HBV), hepatitis delta virus (HDV), or human immunodeficiency virus (HIV). Written informed consent was obtained from each patient in accordance with the Declaration of Helsinki.

For NS3/4A sequencing and NS3 protease phenotypic analyses, plasma samples were collected from all patients before dosing on day 1 (baseline) and on days 4 and 14 and subsequently stored at −80°C. Plasma samples were collected for those patients who returned for follow-up evaluation visits at weeks 12 and 24.

Amplification and sequencing of the HCV NS3/4A.

Full-length HCV NS3/4A was amplified by RT-PCR and population sequenced using dideoxy sequencing-based technology on the Applied Biosystems 3100 platform (Applied Biosystems, Foster City, CA). The QIAamp viral RNA minikit (Qiagen, Inc., Valencia, CA) was used to isolate HCV RNA. Genotype-specific primers (for 1a, 1a4a3′5735 [5′-TTG GCT AGT GGT TAG TGG GCT GG-3′], and for 1b, 1b4a3′5650 [5′-GTG GAC AAG CCT GCT AAG TAC TGT ATC CCG C-3′]) were used to synthesize cDNAs. Reverse transcription using a SuperScript III kit (Invitrogen, NY) was run on an MJ Research PTC-100 thermal cycler (Bio-Rad Laboratories, Hercules, CA). A nested PCR strategy was used with genotype-specific primers to amplify the NS3/4A gene, which was subsequently used as a template for sequencing. GT 1a NS3/4A was amplified using primers 1a3-5′3181 (5′-ATC AAG TTA GGG GCG CTT ACT GGC AC-3′) and 1a4a-3′5735 (5′-TTG GCT AGT GGT TAG TGG GCT GG-3′) for the first-round PCR and 1a3-5′3298 (5′-ATG GAG ACC AAG CTC ATC ACG TG-3′) and 1a4a-3′5712 (5′-CTG GTG ACA GCA GCT GTA AAA GCC ATC-3′) for nested PCR. GT 1b NS3/4A was amplified using primers 1b3-5′3150 (5′-GTC GCT GGG GGT CAT TAT GTC CAA ATG G-3′) and 1b4a-3′5650 (5′-GTG GAC AAG CCT GCT AAG TAC TGT ATC CCG C-3′) for the first-round PCR, followed by primers 1b3-5′3279 (5′-GAG CCC GTC GTC TTC TCT GAC ATG G-3′) and 1b4a-3′5529 (5′ GTT TGC AGC AAC CCG AGC GCC TTC TG-3′) for nested PCR. PCR parameters for both rounds of PCR were as follows: 94°C for 2 min, 35 cycles of 94°C for 30 s, 60°C for 30 s for 1a or 65°C for 30 s for 1b, 72°C for 3 min, 72°C for 7 min. An aliquot of each PCR product was run on an agarose gel to confirm amplification of target amplicons, and the remainder of each reaction mixture was purified using a QIAquick PCR purification kit.

Viral population sequencing was performed on plasma samples for all patients at baseline, day 4, and day 14 and for those patients who returned for follow-up evaluation visits at week 12 and week 24. Sequences were not available from two subjects on day 14 (one dosed with 60 mg and the other dosed with 400 mg) and nine subjects on day 4 (four dosed with 200 mg and five dosed with 400 mg), due to sequencing failure.

Sequence alignment and data analysis.

Sequencher 4.0 (Gene Codes Corporation, Ann Arbor, MI) was used to assemble and analyze nucleotide sequences and to translate them into amino acid sequences. Full-length NS3/4A sequences were aligned against the reference sequences of strain H77 (GenBank accession no. AF009606) for subtype 1a and Con1 (GenBank no. AJ238799) for subtype 1b to identify differences between patient isolate and reference amino acid sequences. Analyses of the emerging amino acid changes at day 4 and day 14 compared to the sequences on day 1 (baseline) were conducted and are reported. In addition, amino acid substitutions were cross-referenced against an amino acid frequency database of 397 GT 1a and 541 1b HCV NS3 protease gene sequences obtained from the EU databases. Amino acid sequence analyses were performed for amino acids 1 to 631 of the NS3 gene and amino acids 1 to 54 of the NS4A gene for both GT 1a and GT 1b.

Generation of chimeric replicons carrying the NS3 protease gene from patient isolates.

HCV GT 1b-PI-luc, a bicistronic replicon, and cured Huh7 cells (Huh-Lunet) were obtained from ReBLikon (Mainz, Germany) (11). To clone the NS3 protease gene from patient isolates, two unique restriction sites (ClaI and AscI) were created in the 1b-PI-luc construct (16).

PCR products generated for NS3 sequencing were used as the template to generate gene cassettes encoding cloning sites at both ends. For GT 1b samples, the protease gene was amplified using the forward primer 1b PCR NS3-4A F2 (5′-ATTAGTCAATCGATACCATGGCGCCYATCACGGCCTACTCCCAACAGACGCG-3′) and the reverse primer 1b PCR Prot R2 (E1202G) (5′ATATGCTCAGGCGCGCCGTTGTCYGTGAAGACCGGRGACCGCATRGTRGTTCCCAT-3′). For GT 1a samples, the forward primer 1a PCR NS3-4A F2 (5′-ATTAGTCAATCGATACCATGGCGCCCATCACGGCGTACGCCCAGCAGAC-3′) and the reverse primer 1a PCR Prot R2 (E1202G) (5′-ATATGCTCAGGCGCGCCGTTGTCCGTGAACACCGGGGACCTCATGGTTGTCCCTAGG-3′) were used. All nested PCRs were performed with a High Fidelity PCR master kit (Roche Applied Science, Indianapolis, IN) as directed by the manufacturer. Final PCR products were purified and digested with ClaI and AscI. Shuttle vector DNA was similarly digested and then ligated using a DNA ligation kit (TaKaRa Bio, Inc., Madison, WI), followed by transformation into E. coli by electroporation using XL-Gold ultracompetent cells (Agilent Technologies, Santa Clara, CA). Ten percent of each transformation mixture was plated on antibiotic selection plates to determine transformation efficiency, and the remaining 90% of each transformation mixture was expanded in liquid culture to propagate pools of NS3 quasispecies. Plasmid DNA was first prepared for RNA transcription by linearization with ScaI, and then RNA was synthesized using a T7 Megascript RNA synthesis kit (Ambion, Austin, TX).

Transient transfection of replicon RNA into Huh7 Cells and EC₅₀ determination.

Replicon RNA was transfected into Huh7-lunet cells following the method of Lohmann et al. (11). Briefly, cells were trypsinized and washed twice with phosphate-buffered saline (PBS). A suspension of 4 × 10⁶ cells in 400 μl of PBS was mixed with 1 to 5 μg of replicon RNA and subjected to electroporation using settings of 960 μF and 270 V. Cells were quickly transferred into 25 ml of culture medium, seeded into 96-well plates at 100 μl/well, and allowed to attach overnight. For EC₅₀ determinations, compounds were serially diluted in 100% dimethyl sulfoxide (DMSO) and then added to the cells at a 1:200 dilution, achieving final concentrations of 0.5% DMSO in total volumes of 200 μl per well. Cells were cultured for 3 days at 37°C, after which culture medium was removed and Renilla luciferase activity was measured using the Renilla luciferase assay system (Promega, Madison, WI) with a Victor luminometer (PerkinElmer, Waltham, MA).

Data analysis.

EC₅₀s were calculated as the compound concentration at which a 50% reduction in the level of Renilla reporter activity was observed compared with the activity in control samples with DMSO. Dose response curves and EC₅₀s were generated by nonlinear regression analysis using the GraphPad Prism software package (GraphPad Software, La Jolla, CA). The replication levels of either reference strain (1b-Con1 or 1a-H77) or of chimera replicons derived transiently from clinical isolates were determined as the ratio of the Renilla luciferase signal at day 4 to that at 4 h postelectroporation, to normalize for transfection efficiency. The replication capacity of each chimera replicon derived from clinical isolates was expressed as its normalized replication efficiency compared with that of the reference strain (1b-Con1 or 1a-H77) within the same experiment. For the relationship between the emergence of resistance and viral response, the maximal viral load reduction derived from each individual during treatment and emergent resistance mutations were compared in a statistical analysis using a two-tailed unpaired t test at the 95% confidence interval.

RESULTS

Antiviral response to GS-9451.

The samples analyzed in this phase 1 study were obtained from 25 GT 1a patients and 7 GT 1b patients who were dosed with GS-9451 for 3 days, as well as 8 patients (7 GT 1a and 1 GT 1b) who received placebo for 3 days. The 60-mg QD dose of GS-9451 for 3 days resulted in a mean maximal HCV RNA reduction of −0.91 log₁₀ IU/ml in HCV GT 1a patients (Table 1). The mean maximal reductions in HCV RNA were −3.16 log₁₀ IU/ml for 200 mg QD for GT 1a, −3.26 log₁₀ IU/ml for 200 mg QD for GT 1b, and −3.77 log₁₀ IU/ml for 400 mg QD for GT 1a compared to a −0.2 log₁₀ IU/ml reduction for placebo patients. Comparison of the maximum levels of viral reduction of the two subtypes at the 200-mg QD dose showed that there was not a statistically significant difference (P = 0.7).

Table 1.

Antiviral response to GS-9451 monotherapy

GS-9451 dose, HCV genotype of patients	No. of patients in group	Mean maximal HCV RNA reduction ± SD (range) (log10 IU/ml)a
Placebo	8	−0.20 ± 0.16 (0 to −0.45)
60 mg QD, GT 1a	8	−0.91 ± 0.41 (−0.28 to −1.54)
200 mg QD, GT 1a	9	−3.16 ± 0.51 (−2.51 to −4.15)
400 mg QD, GT 1a	8	−3.77 ± 0.53 (−2.97 to −4.66)
200 mg QD, GT 1b	7	−3.26 ± 0.46 (−2.29 to −3.58)

^aMean maximal viral load reduction at any time point during the first 7 days.

Population sequencing analysis.

Population sequencing of HCV NS3 did not detect any of the previously identified HCV PI resistance mutations at positions 155, 156, and 168 in viruses from any of the patients at baseline. Drug resistance mutations were also not detected by population sequencing in viruses from patients receiving placebo or 60 mg of GS-9451 for 3 days (Table 2). However, NS3 substitutions at residue R155 (R155K or R155K/R) were observed at day 4, day 14, or week 12 in viruses from 13 of 17 GT 1a patients who received multiple doses of GS-9451 at 200 mg QD or 400 mg QD (Table 2). By week 24, most of the R155K variants were replaced by the wild type (WT), and the variant was detected in 43% (3/7) of patients who came for follow-up visits at week 24 (Table 3). Similarly, NS3 substitutions at residue D168 (D168G, D168D/E, or D168V) were observed at day 4 in viruses from 4 of 7 GT 1b patients who received 200-mg doses of GS-9451 QD (Table 2). For these GT 1b patients, D168E/V/G variants were no longer detected by population sequencing in any patient at day 14 or at the week 12 and week 24 follow-up time points (Table 3).

Table 2.

Summary of NS3 drug resistance mutations detected^a

Mutation in NS3 at day 4, day 14, or week 12b	No. of patientsc in indicated group					Reference amino acidd
	Placebo (n = 8)	60 mg QD, GT 1a (n = 8)	200 mg QD, GT 1a (n = 9)	400 mg QD, GT 1a (n = 8)	200 mg QD, GT 1b (n = 7)
R155K	0	0	7	6	0	R155 (99% in GT 1a, 99% in GT 1b)
D168E	0	0	0	0	2	D168 (99% in GT 1a, 99% in GT 1b)
D168V	0	0	0	0	1
D168G	0	0	0	0	1
Total	0	0	7	6	4

^aAny patient who had mutant virus at day 4 or/and day 14 or/and day 12 was counted once.

^bDetected as a full (when WT is not detected by population sequencing) or mixed mutation.

^cThe number of placebo-dosed patients from all 4 cohorts or the number of dosed patients in each cohort (shown as dose, HCV genotype) is shown in parentheses.

^dFrequency analyses include a total of 397 GT 1a and 541 1b HCV NS3 protease gene sequences obtained from the EU databases. Percent indicates that 99% of the sequences contain WT amino acid residue.

Table 3.

Summary of NS3 drug resistance mutations observed during posttreatment follow-up

Patient	HCV genotype	Dose (mg)	Mutation(s) observed ata:
			Day		Week
			4	14	12	24
BH	1a	200	Unable to sequence	R155K	R155K/R	R155K/R
BJ	1a	200	R155K	S86P/S R155K K244K/R I/V288I A/T402T F418F/Y I/T433T T449I/T S459S/T P470P/S T586I V615I	S86P/S R155K K244K/R I/V288I A/T402T I/T433T T449F/I/S/T S459S/T V609I/V V615I/V	S86P/S R155K K244K/R I/V288I A/T402T I/T433T T449S/T S459S/T V609I V615I/V
CG	1a	400	Unable to sequence	Unable to sequence	R155K/R S/T343S K/R469R	A68A/S T87A/T R155K/R K213K/R
BB	1a	200	Unable to sequence	No change from baseline	V71I/V R155K/R K/R213R G314G/R	No change from baseline
CD	1a	400	R155K E/V183V V609I/V	R155K/R E/V183V	R155K/R E/V183V	—
CK	1a	400	A/S147A/L/S/V R155K/R L/P574P	R155K/R	A/S147A/L/S/V	A/S147A/L/S/V
DB	1a	200	A87A/S R155K/R I170I/T I/V329V V358E/V A/T477T A/T497A	T91A/T R155K/R I/V329V I347I/V A/T477T A/T497A F/L557L	A/S40A T91A/T F/L557L I586I/T	A/S40A I287I/L F/L557L I586I/T
CC	1a	400	Unable to sequence	R155K/R I/V329I I386I/V A/T459T I/T586T	—	—
CF	1a	400	Unable to sequence	T72I/T R155K	—	—
BE	1a	200	R155K	No change from baseline	No change from baseline	—
BF	1a	200	R155K/R V490I/V	No change from baseline	No change from baseline	—
BK	1a	200	R155K/R G237A/G N/S251D/G/N/S A/V306V V329I/V F391F/Y F418F/Y	G237A/G I248I/V N/S251D/G/N/S A/V306V V329I/V F391F/Y F418F/Y	A/V306V F418F/Y M/T505T	—
CB	1a	400	K/Q80I/K/L/Q R155K/R	A87A/S	K/Q80Q A87A/S I/V386V	K/Q80Q A87A/S A/G220A K244K/R I/V386V
DJ	1b	200	F/L14F D168D/E I344I/T	P574L/P I610I/T	F/L14F	F/L14F
DD	1b	200	D168G A/T358T	A/T358T	A/T358T	A/T358T
DH	1b	200	D168V D/E357D T358I	T358A/T	T358A/T	—
DF	1b	200	D168D/E T402S/T	No change from baseline	—	—

^a—, plasma sample was not available due to loss to follow-up.

Other substitutions in the NS3 protease gene (amino acids 1 to 181).

In addition to the resistance mutations at amino acids 155, 156, and 168, there were 12 positions in the NS3 protease with amino acid changes at day 4 and/or 14 (positions 14, 18, 54, 72, 80, 86, 87, 91, 114, 147, 170, and 174). Seven of these 12 substitutions were observed with R155 or D168 mutants (see Table S1 in the supplemental material). Cross referencing the amino acid substitutions at these residues against an amino acid frequency database of 938 NS3 protease amino acid sequences of GT 1a and 1b collected from the EU databases revealed that these substitutions appear to be at highly polymorphic sites and may therefore be natural variation (>1%) of the WT HCV population. Substitutions at positions 80 and 147 that were observed as mixtures of 2 to 4 amino acids (K/Q80I/K/L/Q and A/S147A/L/S/V) do not appear to be associated with GS-9451 selection, and most of these amino acids are possible substitutions observed in WT HCV. One substitution at position 170 (I170T) was changed to an amino acid observed with <1% frequency in the databases (I170I/T, I = 58%, T = 0.1%, V = 42%). Virus with this substitution was observed in only one patient.

Phenotypic analysis.

To determine if the sequence changes described above are associated with reduced susceptibility, phenotypic analyses were performed for samples from patients with amino acid substitutions detected at R155 or D168 by population sequencing. Given the sensitivity limitations of the phenotypic assay in detection of WT/mutant mixtures, patients with full mutants (when WT is not detected by population sequencing) detected were selected for phenotypic analysis. The phenotypic analyses were also performed for the corresponding baseline samples for use as individual comparators. GS-9451 EC₅₀s were obtained for viruses from six patients both at baseline and either day 4 or day 14. The results for GS-9451 susceptibility are summarized in Table 4. Samples from GT 1a patients with full R155K mutants (patients BE, BJ, CD, and CF) had increases in the GS-9451 EC₅₀ of ≥595-fold compared to their baseline values (Table 4). As with replicons with single mutations (Table 5), all patient isolates had reduced replication capacities compared to that of the wild-type replicon. Both patient isolates and the R155K replicon showed high levels of resistance to GS-9451.

Table 4.

Levels of GS-9451 susceptibility and replication capacity of patient isolates with resistance mutations

Patient	Genotype	Dose (mg)	Mutation(s) in NS3 proteasea	Time pointb	Mean replication capacity ± SDc,d	Mean EC50 ± SD (nM)d	Mean EC50 change from baseline (fold)d
BE	1a	200	R155K	BL	45.3 ± 17.2	2.5 ± 1.1	>595
				D4	2.7 ± 1.5	>1,458
BJ	1a	200	R155K	BL	2.8 ± 0.2	3.9 ± 3.6	>1,279
				D4	2.7 ± 1.5	>5,000
CD	1a	400	R155K, E/V183V, V609I/V	BL	56.0 ± 12.7	1.7 ± 0.4	>3,007
				D4	8.9 ± 2.9	>5,000
CF	1a	400	T721I/T, R155K	BL	12.7 ± 1.1	1.4 ± 1.0	>874
				D14	5.5 ± 1.44	>1,250
DH	1b	200	D168V, N357D, T358I	BL	37.8 ± 1.6	2.5 ± 0.34	>2,016
				D4	30.6 ± 2.9	>5,000
DD	1b	200	D168G, A/T358T	BL	5.3 ± 0.4	2.2 ± 0.2	>152
				D4	7.7 ± 2.6	>336

^aBoldface indicates a GS-9451 resistance mutation.

^bBL, baseline; D4, day 4; D14, day 14.

^cReplication capacity of clinical isolates was determined as the ratio of luciferase activity in untreated cells at 96 h relative to that at the 4-h input time point.

^dValues are the results of at least 2 independent experiments.

Table 5.

Levels of replication capacity and EC₅₀ fold change to GS-9451 of NS3 mutants with mutations in the replicon backbone

Mutation	Mean replication capacity ± SDa,b	Mean fold change ± SDb,c
R155K (1a)	56.9 ± 13.44	>150
D168E (1b)	86.1 ± 53.5	82 ± 28.3
D168G (1b)	64.6 ± 15.3	85 ± 35.2
D168V (1b)	85.6 ± 33.2	>1,000

^aReplication capacity relative to that of WT GT 1a or GT 1b replicon was determined as the ratio of luciferase activity in untreated cells at 96 h relative to the activity at the 4-h input time point.

^bValues are the results of at least 2 independent experiments.

^cMean of the fold change in the EC₅₀ for the mutant replicon compared to that for the wild-type replicon determined in each experiment.

Amino acid changes at position D168 were observed in four GT 1b patients. Samples from patients with full D168V (patient with identification code DH) and D168G (DD) mutants were analyzed for their levels of susceptibility to GS-9451. The day 4 samples from both patients DH and DD displayed >152-fold reduced susceptibilities to GS-9451 compared to the levels of susceptibility of the baseline samples (Table 4). Interestingly, the full D168V mutant from patient DH displayed higher levels of resistance to GS-9451 than the full D168G mutant from patient DD, which correlates with the resistance levels of the replicon mutants (Table 5).

Cross-resistance analyses.

It is anticipated that GS-9451 may be used in combination with PEG-IFN and RBV, as well as other DAAs, in the future. Therefore, the susceptibilities of GS-9451-resistant variants to other HCV inhibitors were determined using a transient replication assay. Chimeric replicons carrying the NS3 protease gene from patient isolates determined to carry known PI-resistant mutations, together with their corresponding baseline samples, were used for cross-resistance analyses. Table 6 shows cross-resistance data from a set of patient isolates harboring full mutations at positions 155 or 168 to represent the spectrum of HCV mutants observed. Baseline isolates for all these patients (BE, BJ, CD, CF, DH, and DD) were sensitive to all nine HCV DAAs tested. Day 4 or day 14 GT 1a patient isolates with mutations at R155 in HCV protease displayed moderate to high levels of resistance to the HCV PIs GS-9256, danoprevir, and TMC-435 and low to moderate levels of resistance to telaprevir and boceprevir. The D168V or D168G mutants displayed high levels of resistance to the protease inhibitor GS-9256; however, they were fully susceptible to telaprevir and boceprevir. One GT 1b patient isolate (DH) with a full D168V mutation displayed high levels of cross-resistance to TMC-435 (>174-fold). However, an isolate with a full D168G mutation from patient DD was susceptible to TMC-435. All of these GS-9451-resistant mutants were fully susceptible to the nonnucleoside polymerase inhibitors GS-9190 and GS-9669, the nucleoside polymerase inhibitor GS-6620, and the NS5A inhibitor GS-5885, as well as IFN-α and RBV, indicating no cross-resistance of GS-9451 resistance mutations with these inhibitors (Table 6).

Table 6.

Susceptibility of GS-9451-resistant isolates to other HCV inhibitors

Compound (descriptiona)	Target protein	Mean EC50 change from baseline (fold)b
		BE, D4, R155Kc	BJ, D4, R155K	CD, D4, R155K	CF, D14, R155K	DH, D4, D168V	DD, D4, D168G
GS-9256 (PI)	NS3 protease	>265	>147	>688	>758	>114	>268
Danoprevir (PI)	NS3 protease	118.1	65.6	64.2	97.2	8.9	9.1
TMC-435 (PI)	NS3 protease	16.9	11.8	39.2	8.5	174.3	0.6
Telaprevir (PI)	NS3 protease	10.3	46.7	23.8	4.5	0.5	0.8
Boceprevir (PI)	NS3 protease	4.7	3.9	3.2	2.9	0.6	0.4
GS-9190 (NNI)	NS5B	1.7	1.4	1.1	1.5	0.8	1.8
GS-9669 (NNI)	NS5B	0.9	0.9	1.0	1.1	0.8	1.3
GS-6620 (NI)	NS5B	1.3	1.3	0.8	1.0	0.8	1.4
GS-5885 (NS5Ai)	NS5A	1.0	1.0	1.2	1.1	0.6	0.5
IFN-α (SOC)		2.1	0.4	1.1	0.9	1.2	1.2
Ribavirin (SOC)		1.2	0.9	1.5	1.3	1.0	1.7

^aPI, protease inhibitor; NNI, nonnucleoside inhibitor; NI, nucleoside inhibitor; NS5Ai, NS5A inhibitor; SOC, standard of care.

^bResults of at least 2 independent experiments are shown, calculated as the fold change in the compound's EC₅₀ on day 4 or 14 relative to the compound's EC₅₀ at baseline.

^cPatient identification code, day (D4, day 4; D14, day 14), NS3 resistance mutation of isolate.

The relationship between emergence of resistance and viral response.

There were 32 patients that received GS-9451. HCV NS3 sequence data were obtained from 23 of the 32 patients on day 4, but not from the remaining 9 patients, most likely because of low viral load. Of these 23 patients, 12 had a known PI resistance mutation detected. The mean maximal viral load reduction for patients with viruses with identified resistance mutations was −3.33 log₁₀, compared to −1.37 log₁₀ for those patients without viruses with resistance mutations observed (Fig. 2). The P value for the comparison of viral load reductions between the two groups was <0.0001 (two-tailed t test).

Fig 2.

Maximal viral load reduction in 23 patients with and without resistance mutations who received GS-9451 and had sequence data on day 4 (GT 1a and GT 1b). Colors show different doses: green for 60 mg, pink for 200 mg, and purple for 400 mg. Means ± standard deviations are shown. P value compares viral load reductions between the two groups (two-tailed t test). + and −, patients with and without resistance mutations, respectively, as detected by population sequencing on day 4.

DISCUSSION

This study analyzed drug-resistant HCV variants selected in patients dosed with GS-9451, an HCV protease inhibitor in clinical development. The NS3 mutation R155K is the predominant mutation observed in viruses from GT 1a patients, while NS3 substitutions at residue D168 (D168G, D168E, or D168V) were commonly observed in viruses from GT 1b patients who received GS-9451. No patient had a mutant virus bearing 2 amino acid substitutions (155 and 168) as detected by population sequencing. In vitro, the R155K isolates confer high levels of resistance (>50-fold) to GS-9451 and GS-9256, moderate to high levels of resistance to TMC-435 and danoprevir, and low to moderate levels of resistance to telaprevir and boceprevir. Interestingly, one patient isolate with a full D168V mutation displayed a high level of cross-resistance to TMC-435 compared to that of another patient isolate with a full D168G mutation. Our observations support previously reported data that showed that D168V conferred higher levels of resistance to TMC435 than D168G (10). The D168 isolates were fully susceptible to telaprevir and boceprevir, suggesting the option of treatment with these approved PIs for these patients. These results are in agreement with other previous in vitro resistance studies and clinical observations. For example, R155 in GT 1a virus and D168 in GT 1b virus, conferring cross-resistance to other macrocyclic PIs, are also commonly seen in patients treated with vaniprevir, danoprevir, BI 201335, and TMC-435 (7, 12, 15, 26). Clinical studies with telaprevir identified V36A/M, T54A, R155T/K, and A156S/V/T mutations in NS3. Phenotypic analysis showed low- to intermediate-level (V36, T54, R155, and A156S) and high-level (A156V/T) resistance to telaprevir. Double mutations that were detected on the same isolate (V36A/M and R155K/T) conferred high levels of resistance (18). In patients treated with boceprevir, viruses with V36M/A, T54A/S, R155K/T, A156S, V170A, and V55A (less common) mutations were observed. Genotypic analysis showed a largely overlapping cross-resistance profile for boceprevir and telaprevir. However, V170A and V55A mutations conferred higher-level resistance to boceprevir than to telaprevir (22). Variants at positions 36, 54, 55, 156, and 170 were not detected in this study by population sequencing after 3 days of administration of GS-9451. In vitro studies showed that viruses with T54A/S, V36M, and R155T mutations are sensitive to GS-9451, while viruses with mutations at A156 were cross-resistant to GS-9451 (data not shown).

Interestingly, no NS3 mutations were observed by population sequencing in viruses from patients who received GS-9451 at 60 mg QD for 3 days. We observed that the emergence of resistant variants in this study correlated with the degree of GS-9451 selective pressure, i.e., higher dose and greater HCV viral load reductions. This observation suggests that the mutant viruses preexisted at low levels prior to treatment, and once the WT population was sufficiently inhibited by GS-9451, the resistant variants were detectable by population sequencing. The levels of preexisting mutations at baseline are too low to be detected by population sequencing (detection limit is 20%) or deep sequencing (detection limit of 1%); however, it has been proposed that mutants with single mutations (at positions 155, 156, and 168) preexist in infected subjects with an estimated average of 0.025% and 0.015% for genotype 1a and genotype 1b, respectively, using a back-calculation estimate from mutant frequencies observed at day 2 and/or 4 (23). A more substantial suppression of WT virus consequently resulted in more frequent detection of resistance mutations. In this case, the rapid emergence of the resistant variants indicates a greater inhibition of WT HCV variants.

Sequence analysis of the follow-up time points demonstrated that viruses with mutations at position D168 were no longer detected at day 14, compared to the R155K variant that was detected at week 24 in 43% of the patients (3/7). Viruses with the resistant D168 variants (D168G, D168E, or D168V) are replaced more rapidly by WT virus than virus with the R155K variant. In addition, the D168 mutants were not observed in GT 1a patients, although these mutations also confer high levels of resistance to GS-9451 and require only one nucleotide change, similar to the R155K mutation in GT 1a. The difference in the persistence of the R155K and D168G/E/V mutants may indicate that the in vivo replication fitness of R155K is higher than that of D168 mutants. In addition, two nucleotide changes are required to generate an amino acid change in position 155 in subtype 1b R155K (CGG-AAG), while only one change (AGG-AAG) is needed for subtype 1a. That can result in a lower frequency of the preexisting R155K mutant in 1b patients than in GT 1a patients and, hence, its lack of detection upon suppression of WT virus in 1b patients. The duration of the persistence of mutants may also depend on the degree of the WT clearance and the duration of treatment.

The replication capacity of the R155K replicon was slightly lower than that of the D168 mutants; however, the backbones of the R155K (1a replicon) and D168 (1b replicon) mutants were different. For all patient isolates, the replication capacities were lower than that of the 1b replicon, which suggests incompatibility between the patient isolates and the replicon subtype backbone or other accumulated mutations that affect RNA replication. Interestingly, the replication capacities at day 4 or 14 were comparable to or lower than the baseline replication capacities of isolates from the same subjects. That may also suggest that most of the virus populations at these time points are mutants with lower replication capacities.

Previous in vitro replicon studies identified the A156T/G mutation as conferring resistance to GS-9451 (27). The reason that this variant was not detected in patients could be lower frequencies of the preexisting A156 mutants than of the R155K mutant in GT 1a and D168 mutants in GT 1b at baseline and/or reduced fitness of the virus in patients.

Substitutions at positions 80 and 170 that previously were shown to confer resistance to other PIs (10) were observed in only one patient for each substitution and will be further investigated.

The sequences of the C-terminal helicase domain of NS3 and the cofactor NS4A were also analyzed for their changes from baseline. The substitutions detected in this region appear to be natural variations of the HCV WT population that are present at >1% in the EU databases. In this study, the GS-9451 resistance-associated mutations all mapped to the NS3 protease domain.

It is clear that monotherapy with the majority of DAAs is ineffective in curing HCV, because resistant mutants are detected after a few days of treatment, resulting in virologic rebound and treatment failure. Combination therapy would be required for sustained viral suppression and prevention of viral resistance. Cross-resistance analysis showed that patient isolates with reduced susceptibility to GS-9451 maintained WT sensitivity to IFN-α, RBV, GS-9190 (nonnucleoside site III/IV inhibitor), GS-9669 (nonnucleoside site II inhibitor), GS-6620 (nucleoside inhibitor), and GS-5885 (NS5A inhibitor). These results support the use of GS-9451 in combination with these anti-HCV agents in GT-1-infected patients.

In summary, highly effective inhibition of WT HCV by the NS3 protease inhibitor GS-9451 revealed mutants with the resistance mutations R155K in GT 1a and D168E/G/V in GT 1b patients. The R155K mutants persisted longer than the D168 mutants, suggesting their greater relative fitness. The lack of cross-resistance between GS-9451 and other classes of DAAs, IFN, and RBV supports the combination of GS-9451 with these agents.