Integrated Resistance Analysis of CERTAIN-1 and CERTAIN-2 Studies in Hepatitis C Virus-Infected Patients Receiving Glecaprevir and Pibrentasvir in Japan

ABSTRACT Glecaprevir and pibrentasvir are hepatitis C virus (HCV) pangenotypic inhibitors targeting NS3/4A protease and NS5A, respectively. This once-daily, fixed-dose combination regimen demonstrated high sustained virologic response 12 weeks postdosing (SVR12) rates in CERTAIN-1 and CERTAIN-2 studies in Japanese HCV-infected patients, with a low virologic failure rate (1.2%). There were no virologic failures among direct-acting antiviral (DAA)-treatment-naive genotype 1a (GT1a) (n = 4)-, GT1b (n = 128)-, and GT2 (n = 97)-infected noncirrhotic patients treated for 8 weeks or among GT1b (n = 38)- or GT2 (n = 20)-infected patients with compensated cirrhosis treated for 12 weeks. Two of 33 DAA-experienced and 2 of 12 GT3-infected patients treated for 12 weeks experienced virologic failure. Pooled resistance analysis, grouped by HCV subtype, treatment duration, prior treatment experience, and cirrhosis status, was conducted. Among DAA-naive GT1b-infected patients, the baseline prevalence of NS3-D168E was 1.2%, that of NS5A-L31M was 3.6%, and that of NS5A-Y93H was 17.6%. Baseline polymorphisms in NS3 or NS5A were less prevalent in GT2, with the exception of the common L/M31 polymorphism in NS5A. Among DAA-experienced GT1b-infected patients (30/32 daclatasvir plus asunaprevir-experienced patients), the baseline prevalence of NS3-D168E/T/V was 48.4%, that of NS5A-L31F/I/M/V was 81.3%, that of the NS5A P32deletion was 6.3%, and that of NS5A-Y93H was 59.4%. Common baseline polymorphisms in NS3 and/or NS5A had no impact on treatment outcomes in GT1- and GT2-infected patients; the impact on GT3-infected patients could not be assessed due to the enrollment of patients infected with diverse subtypes and the limited number of patients. The glecaprevir-pibrentasvir combination regimen allows a simplified treatment option without the need for HCV subtyping or baseline resistance testing for DAA-naive GT1- or GT2-infected patients. (The CERTAIN-1 and CERTAIN-2 studies have been registered at ClinicalTrials.gov under identifiers NCT02707952 and NCT02723084, respectively.)

very few HCV-infected patients (Ͻ1% of GT1-infected patients) are infected with GT1a in Japan (8), and the subtype diversity in GT2 is mostly limited to GT2a and GT2b (2).
Therapy for HCV was improved considerably with the availability of several interferon (IFN)-free direct-acting antiviral (DAA) regimens. In Japan, IFN-free DAA regimens, including daclatasvir plus asunaprevir with or without beclabuvir, ledipasvir-sofosbuvir, ombitasvir-paritaprevir-ritonavir, and elbasvir plus grazoprevir with or without ribavirin (RBV), are available for HCV GT1 treatment (9)(10)(11). Sofosbuvir plus RBV and ombitasvirparitaprevir-ritonavir plus RBV were the IFN-free regimens available for the treatment of patients infected with HCV GT2, while sofosbuvir plus RBV was recommended for the treatment of HCV GT3 to GT6 in Japan according to Japan Society of Hepatology (JSH) 2016 guidelines for the management of hepatitis C virus infection (10,12,13). The approved and recommended regimens were not equally effective across all HCV genotypes and subpopulations. Additional limitations of several of the above-listed approved regimens included the requirement for the inclusion of RBV in certain populations, significant drug-drug interactions, limited options for patients with renal insufficiency, reduced efficacy in patients with baseline amino acid polymorphisms associated with reduced susceptibility to HCV nonstructural viral protein 3/4A (NS3/4A) protease inhibitors (PIs) or NS5A inhibitors, and limited options for patients who had failed DAA-containing treatment regimens (10).
The glecaprevir-pibrentasvir regimen is approved in the United States and Canada for the treatment of PI-experienced patients without prior experience with an NS5A inhibitor (12-week treatment duration) or NS5A inhibitor-experienced patients without prior experience with a PI (16-week treatment duration). The SVR 12 rates among GT1-infected patients treated with the glecaprevir-pibrentasvir regimen who were PI experienced/NS5A inhibitor naive or NS5A inhibitor experienced/PI naive were 92% (23/25) and 94% (16/17), respectively (17).
The safety and efficacy of glecaprevir-pibrentasvir in HCV-infected patients in Japan was evaluated in two phase 3 clinical studies, CERTAIN-1 and CERTAIN-2 (19)(20)(21). Treatment-naive or (peg)IFN with or without RBV-experienced (but DAA treatmentnaive) GT1-and GT2-infected noncirrhotic patients received glecaprevir-pibrentasvir for 8 weeks, and patients with compensated cirrhosis received 12 weeks of treatment. GT3to GT6-infected patients and DAA treatment-experienced GT1-or GT2-infected patients without cirrhosis or with compensated cirrhosis received 12 weeks of treatment. High SVR 12 rates were observed for all patient populations and genotypes (Tables 1 and 2). The glecaprevir-pibrentasvir regimen recently received marketing approval in Japan for a treatment duration of 8 weeks in DAA treatment-naive GT1-and GT2-infected patients without cirrhosis, including those with chronic kidney disease, and for a treatment duration of 12 weeks for patients infected with GT3 to GT6, patients with compensated cirrhosis, and those not cured with a previous DAA treatment.
Pooled data from efficacy and resistance analyses of the CERTAIN-1 and CERTAIN-2 studies is presented. Baseline sequencing of HCV NS3/4A and NS5A was conducted by next-generation sequencing (NGS) on all available baseline samples to assess the prevalence and impact of baseline polymorphisms on treatment outcomes. Analyses were integrated across the two studies by HCV subtype, prior DAA treatment experience, treatment duration, and cirrhosis status. Baseline polymorphisms and treatmentemergent substitutions in patients experiencing virologic failure were also evaluated. The prevalences of baseline polymorphisms were compared by geographic region, using sequence data from non-Japan (referred to as overseas regions) phase 2 and 3 studies with the glecaprevir-pibrentasvir regimen.

RESULTS
Efficacy of glecaprevir-pibrentasvir in analyses of pooled data from the CERTAIN-1 and CERTAIN-2 clinical studies. The SVR 12 rates by HCV subtype, prior treatment experience, treatment duration, and cirrhosis status are shown in Tables 1  and 2. Among DAA treatment-naive GT1-and GT2-infected patients without cirrhosis a HCV subtype was determined by phylogenetic analysis of NS3/4A and/or NS5A sequences. In the absence of phylogenetic data (4 treatment-naive GT1b-and 2 treatment-naive GT2*-infected patients), subtype designation was done by LiPA 2.0 assay. b Includes 2 GT1b-infected, 1 GT2a-infected, 2 GT2b-infected, and 2 GT2 (with an undetermined subtype)infected SVR 12 -achieving patients with severe renal impairment. c Includes 1 GT1b-and 2 GT2a-infected SVR 12 -achieving patients with severe renal impairment. d Patients not achieving SVR 12 were lost to follow-up or prematurely discontinued treatment. e Includes 2 GT2b-infected SVR 12 -achieving patients with severe renal impairment. f X indicates that patients with these HCV genotypes were excluded per enrollment criteria. * indicates that the HCV subtype was unable to be determined. IFN-TE, treatment experienced with an interferon-containing regimen but DAA naive; TN, DAA treatment naive. treated for 8 weeks, including 10 patients with severe renal impairment, there were no virologic failures, and the SVR 12 rates were 99.2% (131/132; 1 patient was lost to follow-up) and 97.9% (95/97; 1 patient prematurely discontinued the study drug, and 1 patient was lost to follow-up) for GT1-and GT2-infected patients, respectively (Table  1). SVR 12 rates of 100% were observed among GT1 (n ϭ 38)-and GT2 (n ϭ 20)-infected patients with compensated cirrhosis treated for 12 weeks, including 2 GT2b-infected patients with severe renal impairment. SVR 12 was achieved in 83.3% (10/12; 2 patients experienced posttreatment relapse) of patients with GT3 HCV infection treated for 12 weeks. The CERTAIN-1 study enrolled 33 patients with HCV GT1 or GT2 infection who failed at least 1 previous DAA treatment, and SVR 12 was achieved in 93.9% (31/33) of patients in this population ( Table 2). None of the patients enrolled in the two CERTAIN studies were infected with HCV GT4, GT5, or GT6. Analysis of baseline polymorphisms in NS3 and NS5A in DAA-naive patients. The prevalence of baseline polymorphisms relative to the subtype-specific reference sequence was evaluated for 4 GT1a-, 162 GT1b-, 78 GT2a-, 37 GT2b-, 7 GT3a-, 4 GT3b-, and 1 GT3k-infected DAA-naive patients grouped by HCV subtype, prior treatment experience, and cirrhosis status ( Table 3). The prevalences of baseline polymorphisms were similar across treatment-naive and (peg)IFN (with or without RBV)-experienced patients without cirrhosis or with compensated cirrhosis for all genotypes. Among GT1b-infected patients with an available baseline sequence (n ϭ 162), the prevalences of polymorphisms at amino acid position 56 or 80 (mostly Q80L) in NS3 were high, 38.9% and 18.5%, respectively, while polymorphisms at position 168 were rare. In NS5A, the prevalence of Y93H was 18.0%, and the prevalence of polymorphisms at position 24, 28, 30, 58, 92, or 93 ranged between 7.5% and 11.8%, while L31M was less frequent.
In studies across overseas regions, among DAA treatment-naive GT1b-infected patients grouped by North America (n ϭ 74), Europe (n ϭ 266), or ROW (n ϭ 126), the prevalences of baseline polymorphisms at position 36, 54, 55, or 56 in NS3 were similar across the 3 regions and Japan, while the prevalence of Q80L was numerically higher in Japan and ROW than in North America and Europe (see Fig. S1 in the supplemental material). The prevalence of baseline polymorphisms in NS5A at position 24, 28, 30, or 93, specifically Q24K, L28M, R30H/Q, or Y93H, was higher in Japan than in any of the other 3 regions (Fig. S1).
Only two HCV GT2 subtypes, GT2a (n ϭ 78) and GT2b (n ϭ 37), were identified by phylogenetic analysis in CERTAIN-1 and CERTAIN-2 studies; baseline sequences from two of the GT2-infected patients were not available for HCV subtyping by phylogenetic analysis. The majority of GT2-infected patients (67.8%; n ϭ 78) were treatment naive without cirrhosis. Baseline polymorphisms in NS3 were detected in 6.3% of the patients across both GT2 subtypes. In GT2a-infected patients, polymorphisms at positions 24, 28, 58, and 92 in NS5A were detected at a low prevalence, while the M31 polymorphism was detected in 92.3% of the patients. In GT2b-infected patients, L28F and the M31 polymorphism in NS5A were detected at prevalences of 16.2% and 83.8%, respectively. L31 in NS5A in GT2b-infected patients was detected at a much higher prevalence in overseas regions (67.9%) than in Japan (10.8%) (Fig. S2).
In GT3a-infected patients (n ϭ 7), A166S/T in NS3 was detected in 2 patients, and A30K and Y93H in NS5A were each detected in one patient. Baseline polymorphisms at other amino acid positions important for the PI class were not detected in NS3, and V31M in NS5A was detected in the 4 GT3b-infected patients. The single GT3k-infected patient had G92E in NS5A, while the NS3 gene could not be sequenced due to technical difficulties.
None of the single amino acid polymorphisms observed at baseline in DAA-naive patients that were tested in replicon assays conferred resistance to glecaprevir or pibrentasvir for any genotype ( Table 4).
Analysis of baseline substitutions in NS3 and NS5A in DAA-experienced patients. Of the 32 DAA-experienced GT1b-infected patients enrolled, 30 had most recently failed a regimen containing daclatasvir plus asunaprevir (some had failed multiple DAA regimens), and 2 had previous experience with a PI-containing regimen (simeprevir) but not an NS5A inhibitor (see Table S1 in the supplemental material). One additional GT2a-infected patient that had previous experience with the sofosbuvir plus RBV regimen was also enrolled; this patient had T24A and M31 in NS5A at baseline.
Baseline substitutions detected in NS3 and NS5A in the GT1b-infected patients are shown in Table 5. A list of baseline substitutions in each patient is shown in Table S1 None     (Tables 6 and 7).
Impact of baseline polymorphisms/substitutions on treatment outcome. Among DAA-naive GT1-and GT2-infected patients, there were no virologic failures; therefore, baseline polymorphisms had no impact on treatment outcomes. One GT3binfected patient and one GT3k-infected patient experienced virologic failure; the GT3b-infected patient had unusually low plasma concentrations of glecaprevirpibrentasvir (21). Given the low number of GT3-infected patients enrolled in this study  (n ϭ 12), the impact of baseline polymorphisms on treatment outcomes in GT3-infected patients could not be meaningfully assessed. Among DAA-experienced GT1b-infected patients, baseline substitutions in NS3 and/or NS5A, including those at amino acid position 168 in NS3 or at position 31 or 93 in NS5A, had no impact on SVR 12 (see Table S2 in the supplemental material). Two patients with a P32deletion in NS5A experienced virologic failure. This is consistent with the resistance profile of glecaprevir and pibrentasvir, as none of the baseline substitutions, with the exception of the P32deletion in NS5A, conferred resistance to glecaprevir or pibrentasvir (Tables 6 and 7).
Resistance analysis for patients experiencing virologic failure. Baseline and postbaseline variants in NS3 and NS5A among the 4 patients experiencing virologic failure are shown in Table 8. Of the 2 GT1b-infected PI-plus NS5A inhibitor-experienced patients with virologic failure, treatment-emergent substitution A156D/V in NS3 was detected in 1 patient (D168V was also present at baseline and postbaseline in this patient). In NS5A, P32deletion (P32L was detected at the 2% but not at the 15% detection threshold) and L31F plus P32deletion were present at both baseline and postbaseline in 1 patient each. D168V plus A156V in NS3 and P32deletion alone or in combination with L31F in NS5A conferred high levels of resistance to glecaprevir and pibrentasvir, respectively. A P32deletion in NS5A also confers high levels of resistance to all the currently marketed NS5A inhibitors, including velpatasvir, ledipasvir, ombitasvir, elbasvir, and daclatasvir (Table 9). Of the 2 GT3-infected patients with virologic  failure, treatment-emergent substitutions L28F and/or Y93H was detected in NS5A; NS3 sequences were not available for analysis.
Baseline polymorphisms at amino acid positions important for the PI class (position 155, 156, or 168) were rarely detected. In NS5A, the baseline prevalence of Q24K (8%), L28M (8%), R30Q (11%), or Y93H (18%) was higher in Japanese GT1b-infected patients than in patients from other geographic regions. The NS5A Y93H polymorphism in GT1b confers resistance to NS5A inhibitors ombitasvir (77-fold), daclatasvir (24-fold), elbasvir (17-fold), and ledipasvir (1,807-fold) (22)(23)(24)(25)(26). According to JSH guidelines for the management of hepatitis C virus infection, ombitasvir-paritaprevir-ritonavir or daclatasvir plus asunaprevir is not recommended for use in GT1b-infected patients with a baseline NS5A Y93H polymorphism (10). In Japanese clinical studies with elbasvir plus grazoprevir in GT1b-infected patients for 12 weeks, SVR 12 rates in patients with and without baseline NS5A polymorphisms were 93.1% (54/58) and 98.9% (260/263), respectively, and the SVR 12 rate in patients with Y93 variants was 93% (40/43) (11). Additionally, a recent real-life multiple-cohort study in Japan showed a similar negative impact of these baseline polymorphisms on GT1b-infected patients with cirrhosis (treatment naive, prior IFN/RBV failure, or prior PI failure) receiving sofosbuvir-ledipasvir; baseline NS5A polymorphisms at amino acid positions 31 and/or 93 were associated with a reduced SVR 12 rate of 90.7% (49/54), compared to an SVR 12 rate of 99.4% (154/155) in patients without baseline polymorphisms (27). Thus, in Japanese GT1binfected patients, the preexistence of NS5A polymorphisms, especially Y93H, had a negative impact on SVR 12 rates in patients receiving regimens containing any of the these NS5A inhibitors. Pibrentasvir is a next-generation NS5A inhibitor that retains its activity against Y93H in NS5A in GT1 replicons (14). Consistent with the high barrier to resistance in vitro, baseline polymorphisms in NS5A had no impact on treatment outcome with the glecaprevir-pibrentasvir regimen, and there were no virologic failures among DAA-naive GT1-infected patients without cirrhosis (8-week treatment duration) or with compensated cirrhosis (12-week treatment duration) in the CERTAIN-1 study. Very few HCV-infected patients (Ͻ1% of GT1-infected patients) are infected with GT1a in Japan (8), and consistent with the epidemiology in Japan, only 4 DAA-naive GT1ainfected patients were enrolled in CERTAIN-1. Although the data set for the analysis of baseline polymorphisms in GT1a-infected patients was limited in this study, 2 of the 4 patients had M28V, Q30H, or Y93F in NS5A at baseline, and all 4 achieved SVR 12 . In global studies with the glecaprevir-pibrentasvir regimen, baseline polymorphisms in NS3 at amino acid position 155, 156, or 168 in NS3 were detected in 2.1% (8/384) of DAA treatment-naive GT1a-infected patients, and those in NS5A at amino acid position 24, 28, 30, 31, 58, 92, or 93 in NS5A were detected in 20.5% (78/380) of DAA treatment-naive GT1a-infected patients (16). Baseline polymorphisms had no impact on treatment outcomes in GT1a-infected patients irrespective of the treatment duration or cirrhosis status (16,17).
The NS5A M31 polymorphism was predominant in patients infected with GT2a (92.3%) and GT2b (83.8%) in Japan, while in other geographic regions, M31 was predominant in GT2a-infected patients but was detected in only 31.6% of the GT2binfected patients. First-generation DAA regimens that were effective against GT1 in Japan have been less successful in treating GT2-infected patients because the NS5A M31 polymorphism reduces susceptibility to first-generation NS5A inhibitors such as daclatasvir (Ͼ1,000-fold) or ledipasvir (12-fold) and reduces the barrier to resistance for ombitasvir (23,25,28). Glecaprevir and pibrentasvir have potent activity against both GT2a and GT2b replicons irrespective of the presence of common polymorphisms and have a higher barrier to resistance than the first-generation PIs or NS5A inhibitors (14,15). Consistent with their respective in vitro profiles, baseline polymorphisms in NS3 or NS5A had no impact on treatment outcomes with the glecaprevir-pibrentasvir regimen, and there were no virologic failures among DAA-naive GT2-infected patients without cirrhosis or with compensated cirrhosis in the CERTAIN-1 and CERTAIN-2 studies.
The incidence of HCV GT3 is low in Japan, and the only approved DAA-based treatment option (prior to the approval of glecaprevir-pibrentasvir) included sofosbuvir plus RBV for 24 weeks, which had an SVR rate of 85% (29). In the CERTAIN-1 study, the SVR 12 rate in GT3-infected patients was 83.3% (10/12) with the 12-week glecaprevirpibrentasvir regimen; 1 of 4 GT3b-infected and 1 of 1 GT3k-infected patients experienced virologic failure, and all 7 GT3a-infected patients achieved SVR 12 . In global studies with glecaprevir-pibrentasvir, the resistance analysis data set included data for 627 GT3a-, 6 GT3b-, and 2 GT3i-infected subjects. Overall, baseline NS3 and most of the NS5A polymorphisms had no impact on the SVR 12 rate; A30K in NS5A was associated with lower SVR 12 rates in GT3a-infected pegIFN/RBV treatment-experienced noncirrhotic patients receiving 12 weeks of treatment (16). In the CERTAIN-1 study, A30K was detected in 1 GT3a-infected subject; however, K30 in NS5A is present in the GT3b and GT3k reference sequences and thus was not counted as a polymorphism. The overall prevalence of K30 in NS5A among the 12 GT3-infected patients in CERTAIN-1 was 50% (6/12), and 2 of these patients experienced virologic failure. Data on the in vitro activity of pibrentasvir against GT3b and GT3k replicons containing K30 with or without the treatment-emergent Y93H substitution are not available. Additionally, the GT3binfected patient experiencing virologic failure had received prior IFN-based treatment, and the plasma concentrations were substantially low for both glecaprevir and pibrentasvir during weekly treatment visits for this patient (21). Therefore, it is unclear how much of an impact, if any, K30 in NS5A had on treatment outcomes in the GT3b-and GT3k-infected patients experiencing virologic failure.
For GT1-infected patients who had failed previous PI-NS5A inhibitor therapy, sofosbuvir-ledipasvir was the only IFN-free DAA regimen recommended in the 2016 JSH guidelines for the management of hepatitis C virus infection and only if the virus did not harbor multiple polymorphisms at positions 31 plus 93 in NS5A (10). Due to insufficient evidence on the impact of DAA resistance in sofosbuvir-ledipasvir failures on future treatment, the 2016 JSH guidelines also considered waiting for new drug development as an option. In 2014, the daclatasvir plus asunaprevir regimen was the first IFN-free regimen approved in Japan, leading to the significant usage of this regimen prior to the approval of subsequent DAA regimens. As baseline L31 and/or Y93 polymorphisms had a negative impact on treatment outcome in patients administered daclatasvir plus asunaprevir, the overall the SVR 12 rate in a phase 3 clinical study was only 84.7% in GT1-infected patients (30). In addition, patients experiencing virologic failure were left with D168 substitutions in NS3 and L31 and/or Y93 substitutions in NS5A (30). A recent real-life study of daclatasvir plus asunaprevir-experienced patients retreated with sofosbuvir-ledipasvir demonstrated a high virologic failure rate of around 36% (31)(32)(33). Recent approval of the glecaprevir-pibrentasvir regimen in Japan will help address retreatment options for prior DAA-experienced patients, including those with experience with an NS5A inhibitor alone or a PI plus an NS5A inhibitor.
The majority of the DAA-experienced GT1b-infected patients enrolled in CERTAIN-1 (30/32) most recently had experience with a daclatasvir plus asunaprevir-containing regimen, and 2/32 patients had experience with a simeprevir plus IFN-containing regimen. One GT2a-infected patient had experience with sofosbuvir plus RBV. In daclatasvir plus asunaprevir-experienced patients, baseline Q80R and D168E/T/V substitutions in NS3 and Q24K, L28I/M/T/V, R30H/L/M, L31F/I/M/V, A92K/T, and Y93F/H/S substitutions in NS5A were enriched relative to the prevalence observed in DAA-naive patients in the study. Overall, 97% (32/33) of patients had baseline substitutions in NS3 and/or NS5A. D168E/T/V (15/30 patients) in NS3 and L31F/I/M/V (26/30) and Y93F/H/S (21/30) in NS5A were the most common baseline substitutions detected in previously daclatasvir plus asunaprevir-experienced patients. D168E/T/V in NS3 was detected in combination with L31F/I/M/V and/or Y93F/H/S in NS5A in 15 patients; all 19 patients with NS5A Y93H also had L31 substitutions. Baseline substitutions observed in these DAA-experienced patients are representative of substitutions commonly selected in patients experiencing virologic failure with daclatasvir plus asunaprevir with or without beclabuvir; ombitasvir-paritaprevir-ritonavir; sofosbuvir-ledipasvir; sofosbuvir-velpatasvir; or elbasvir plus grazoprevir (30,(34)(35)(36)(37). These amino acid substitutions, alone or in combination, remained susceptible to glecaprevir or pibrentasvir in in vitro replicon assays (14,15). Consistent with this high barrier to resistance, the common and prevalent baseline substitutions in NS3 and/or NS5A had no impact on treatment outcomes. The only 2 patients experiencing virologic failure both had a P32deletion in NS5A, which had a baseline prevalence of 6.7% (2/30) in daclatasvir plus asunaprevirexperienced patients in this study. A P32deletion in NS5A has been observed as an uncommon treatment-emergent substitution in patients who have failed treatment with a daclatasvir-containing regimen (38)(39)(40) and confers Ͼ1,000-fold resistance to pibrentasvir, accounting for the negative impact of this substitution on treatment outcome. The P32deletion also confers Ͼ10,000-fold resistance to all of the other currently marketed NS5A inhibitors, daclatasvir, ledipasvir, velpatasvir, elbasvir, or ombitasvir (Table 9).
In summary, the CERTAIN-1 and CERTAIN-2 studies in Japan evaluated a glecaprevirpibrentasvir regimen in 332 treatment-naive, IFN-experienced, and DAA-experienced patients, including those without cirrhosis and with compensated cirrhosis and those with and without severe renal impairment. A high overall SVR 12 rate of 98.3% was observed across DAA-naive patients treated with glecaprevir-pibrentasvir. In addition, a high SVR 12 rate of 93.9% was observed in patients with previous DAA experience. In both DAA-naive and DAA-experienced patients, the treatment outcome was not affected by baseline substitutions in NS3 and/or NS5A that are commonly selected by currently available PIs and/or NS5A inhibitors, including those at amino acid positions 168 in NS3 and 31 or 93 in NS5A. Unlike the first-generation HCV pegIFN/RBV-free DAA regimens, the glecaprevir-pibrentasvir regimen allows a simplified treatment algorithm without the need for baseline resistance testing or HCV subtyping in DAA-naive GT1and GT2-infected patients in Japan.

MATERIALS AND METHODS
Study design. CERTAIN-1 and CERTAIN-2 study designs, randomization procedures, and efficacy and safety analyses were previously described (19)(20)(21). Briefly, CERTAIN-1 (ClinicalTrials.gov identifier NCT02707952) is a phase 3, open-label, multicenter study assessing the safety and efficacy of glecaprevirpibrentasvir (300 mg/120 mg) once daily in Japanese patients with HCV infection. The study was composed of two substudies: in substudy 1, DAA-naive patients with GT1 HCV infection without cirrhosis and without the Y93H polymorphism were randomized to 8 weeks of treatment with glecaprevirpibrentasvir (arm A) or 12 weeks of treatment with ombitasvir-paritaprevir-ritonavir (arm B). Patients without the NS5A Y93H polymorphism were randomized 2:1 to arm A or B, while patients with the Y93H polymorphism were assigned to arm A only. HCV NS5A population sequencing was performed by a central laboratory at the screening visit for substudy 1 patients to detect the presence or absence of the Y93H polymorphism (approximately a 15% detection threshold for Y93H). In substudy 2, DAA-naive patients with GT1 HCV infection and with compensated cirrhosis were assigned to 12 weeks of glecaprevir-pibrentasvir (300 mg/120 mg) once daily. The remainder of the patients in substudy 2 belonged to four special populations: (i) GT1-or GT2-infected patients who failed prior DAA treatment, including patients with compensated cirrhosis; (ii) GT1-or GT2-infected patients with severe renal impairment and compensated cirrhosis; (iii) GT3-infected patients who were treatment naive or IFN treatment experienced; and (iv) GT1-or GT2-infected patients with severe renal impairment without cirrhosis. The first 3 cohorts received glecaprevir-pibrentasvir (300 mg/120 mg) for 12 weeks, and the fourth cohort received treatment for 8 weeks. Although substudy 2 of CERTAIN-1 was open to patients infected with GT1 to GT6, no subjects with GT4, GT5, or GT6 infection enrolled, due to the very low prevalence of these genotypes in Japan. Patients with severe renal impairment included patients with an estimated glomerular filtration rate (eGFR) of Ͻ30 ml/min/1.73 m 2 (using the MDRD method modified for Japanese population, where eGFR J ϭ 194 ϫ serum creatinine Ϫ1.094 ϫ age Ϫ0.287 ϫ 0.739 [if female]) and patients with end-stage renal disease requiring treatment with chronic intermittent hemodialysis. Patients enrolled in CERTAIN-2 (ClinicalTrials.gov identifier NCT02723084) had GT2 infection without cirrhosis and were randomized 2:1 to 8 weeks of treatment with glecaprevir-pibrentasvir (arm A) or 12 weeks of treatment with SOF (400 mg QD) plus RBV (600 to 1,000 mg, weight based, divided twice a day [BID]) (arm B). Patients in arm B of CERTAIN-1 and CERTAIN-2 were excluded from the resistance analyses.
All patients provided written, informed consent to participate, and the study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki and the International Conference on Harmonization good clinical practice guidelines. The study was approved by the institutional review board of each study site prior to the initiation of any screening or study-specific procedures.
Sample processing. The Versant HCV Genotype Inno-LiPA assay v2.0 (LiPA 2.0), conducted by the Central Lab, was used to determine HCV genotypes for the enrollment of patients into clinical studies. However, for non-HCV GT1, the LiPA 2.0 assay is unable to accurately identify the viral subtype. The viral subtype was therefore determined by phylogenetic analysis of a 329-nucleotide region of HCV NS5B (44,45). Results from the NS5B phylogenetic analysis were used to assign a preliminary subtype to each HCV-infected patient sample, which then determined the subtype-specific reverse transcriptase PCR (RT-PCR) and nested PCR primer sets used for the independent amplification of NS3/4A and NS5A genes from baseline samples. Subtype-specific RT-PCR, nested PCR, and sequencing primers were designed based on the alignments of available sequences in GenBank and the European HCV database. The primers were designed in conserved regions specific to the gene of interest, and nucleotide degeneracies were incorporated in positions where significant variability existed among the sequences. Only samples with Ն1,000 IU/ml HCV RNA were amplified in order to reduce the chances of oversampling bias. For samples with Յ50,000 IU/ml HCV RNA, RT-PCRs were done in triplicate, and the products were pooled prior to their use as a template for nested PCR. Nested PCR products encompassing the genes encoding full-length NS3/4A or NS5A were analyzed by next-generation sequencing (NGS) analysis performed by the DDL Diagnostic Laboratory (Rijswijk, Netherlands). NGS analysis parameters. PCR amplicons were purified by using AMPure XP beads (Beckman Coulter Genomics, Danvers, MA) and quantified by using the Quant-iT PicoGreen double-stranded DNA (dsDNA) kit (Life Technologies, Carlsbad, CA). The DNA was then fragmented and tagged by using the Nextera XT sample preparation kit (Illumina, San Diego, CA) according to the manufacturer's instructions. Index primers were added by limited-cycle PCR using the Nextera XT Index kit (Illumina, San Diego, CA), and samples were normalized by using beads with maximum binding capacity according to instructions provided with the Nextera XT sample preparation kit. Multiplexed paired-end sequencing was conducted on the Illumina MiSeq platform using the MiSeq v2 sequencing kit with 300 cycles (Illumina, San Diego, CA). Demultiplexed FASTQ files were then mapped against the reference sequence by using CLC Genomics Workbench software (CLCbio, Denmark). Sequences were trimmed to remove nucleotides with a quality score of Ͻ30. The attempted minimum coverage is 5,000 sequencing reads per nucleotide position. An amino acid variant report relative to a subtype-specific reference sequence was generated with Athena pipeline proprietary software. The threshold for the detection of nucleotide polymorphisms by NGS was set at 2%.
HCV genotype and subtype classification. For each sample analyzed by NGS, a consensus sequence was generated for each target gene from the NGS nucleotide sequences, with an ambiguity setting of 0.25. Phylogenetic analyses were subsequently conducted by using the available full-length HCV NS3/4A and NS5A consensus nucleotide sequences in order to confirm the subtype assignment. Nucleotide sequences for NS3/4A and NS5A were aligned by using the MAFFT sequence alignment method (46). Phylogenetic trees were constructed by using the neighbor-joining tree-building method with the HKY85 nucleotide substitution model (47,48). The reliability of the tree topology was examined by using bootstrap analysis, and 1,000 bootstrapping replicates were utilized to generate a consensus tree with a 50% threshold cutoff. Nucleotide alignments and phylogenetic trees were generated by using Geneious software (Biomatters Ltd., Auckland, New Zealand). The final HCV subtype assignment was determined by consensus between NS3/4A and NS5A phylogenetic analysis results. If sequences were not available for phylogenetic analyses, subtype assignment by the LiPA 2.0 assay was utilized.
Resistance analyses of data from CERTAIN-1 and CERTAIN-2. Data were grouped by HCV subtype, treatment duration, prior HCV treatment experience, or cirrhosis status and broadly included the following analyses: (i) prevalence of polymorphisms at baseline at amino acid positions important for the NS3 protease and NS5A inhibitor classes at a 15% NGS detection threshold relative to the subtype-specific reference sequence for GT1a-H77 (GenBank accession number NC_004102), GT1b-Con1 (GenBank accession number AJ238799), GT2a-JFH-1 (GenBank accession number AB047639), GT2b-HC-J8 (GenBank accession number D10988), GT3a-S52 (GenBank accession number GU814263), GT3b-HCV-Tr (GenBank accession number D49374), or GT3k-JK049 (GenBank accession number D63821); (i) impact of baseline polymor-phisms on the treatment response (SVR 12 ); and (iii) analysis of baseline polymorphisms and treatmentemergent substitutions in patients experiencing virologic failure (at a 2% detection threshold). The detection threshold of 15% for the analysis of baseline polymorphisms was chosen because lower detection thresholds will likely increase the reported prevalences of baseline polymorphisms, leading to an underestimation of the impact of polymorphisms on treatment outcome. For patients experiencing virologic failure, it is important to understand whether a substitution is enriched at the time of failure relative to its prevalence at baseline as a result of drug pressure; therefore, a more conservative detection threshold of 2% was utilized to ensure that any enrichment was captured. In DAA-naive patients, polymorphism was defined as a baseline amino acid difference relative to the appropriate subtypespecific reference sequence. In DAA-experienced patients, a variant(s) relative to the appropriate subtype-specific reference sequence could have resulted from a prior treatment regimen and is therefore referred to as a baseline substitution(s).
Antiviral activity against a panel of NS3 or NS5A variants. The methods for assessing the effects of individual amino acid substitutions on the activity of an inhibitor in HCV replicon cell culture assays were described previously (14,15). NS3 and NS5A substitutions were each introduced into the subtypespecific subgenomic replicon plasmid by using the Change-IT Multiple Mutation site-directed mutagenesis kit (Affymetrix, Santa Clara, CA), or synthetic DNA constructs encoding NS3 and NS5A substitutions (Integrated DNA Technologies, Coralville, IA) were inserted into a subtype-specific subgenomic replicon plasmid. In a transient assay, the replicon RNA containing the substitutions was transfected via electroporation into a Huh7 cell line. Glecaprevir, pibrentasvir, ombitasvir, and daclatasvir were synthesized at AbbVie. Elbasvir was purchased from ACME Bioscience (Palo Alto, CA), ledipasvir was purchased from Cayman Chemical (Ann Arbor, MI), and velpatasvir was purchased from eNovation Chemicals (Bridgewater, NJ). The luciferase activity in the cells was measured by using an EnVision multilabel plate reader (Perkin-Elmer, Waltham, MA). The 50% effective concentrations (EC 50 s) were calculated by using nonlinear regression curve fitting to the 4-parameter logistic equation in Prism5 software (GraphPad Software Inc., La Jolla, CA). Mean EC 50 s and standard deviations were calculated from at least 3 independent experiments.

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AAC .02217-17.