The co-existence of NS5A and NS5B resistance-associated substitutions is associated with virologic failure in Hepatitis C Virus genotype 1 patients treated with sofosbuvir and ledipasvir

Objective The present study aimed to reveal the factors associated with virologic failure in sofosbuvir and ledipasvir (SOF/LDV)-treated patients, and identify baseline NS5A or NS5B resistance-associated substitutions (RASs). Methods Four hundred ninety-three patients with Hepatitis C Virus (HCV) genotype 1b infection were treated with SOF/LDV; 31 had a history of interferon (IFN)-free treatment with daclatasvir and asunaprevir. The effect of baseline RASs on the response to SOF/LDV therapy was analyzed. Results Overall, a sustained virologic response at 12 weeks (SVR12) was achieved in 476 patients (96.6%). The SVR12 rates in the patients with IFN-free treatment-naïve and retreatment were 97.6% and 80.6%, respectively. HCV elimination was not achieved in 17 patients, 11 (including 5 with IFN-free retreatment) of whom had virologic failure. Eight patients had coexisting NS5A RASs of Q24, L28 and/or R30, L31, or Y93 and one patient had coexisting NS5A RASs of P32L and A92K. Interestingly, 10 and 8 patients had NS5B A218S and C316N RAS respectively. According to a multivariate analysis, coexisting NS5A RASs, NS5A P32 RAS, NS5B A218 and/or C316 RASs, and γ-glutamyltranspeptidase were associated with virologic failure. In the naïve patients, all patients without NS5B A218 and/or C316 RAS achieved an SVR12. Notably, the SVR12 rates of patients with coexisting NS5A and NS5B RASs were significantly lower (83.3%). Conclusions Although SOF/LDV therapy resulted in a high SVR12 rate, coexisting NS5A and NS5B RASs were associated with virologic failure. These results might indicate that the coexisting baseline RASs influence the therapeutic effects of SOF/LDV.


Introduction
Direct acting antiviral (DAA) therapy is associated with a high sustained virologic response (SVR) rate in patients with chronic hepatitis type C in comparison to conventional interferon (IFN) therapy. In July 2014, daclatasvir (DCV) (a NS5A inhibitor) and asunaprevir (ASV) (an NS3-4A protease inhibitor) were authorized for the treatment of chronic hepatitis or compensated cirrhosis in patients with hepatitis C virus (HCV) genotype 1 in Japan. It was reported that baseline resistance-associated substitutions (RASs) of NS3 D168, NS5A L31 and Y93 predicted the effects of DAA treatment in patients undergoing DCV/ASV combination therapy [1]. We previously reported that in patients without NS5A L31 or Y93 RASs and a history of protease inhibitor therapy, the virologic effect of DCV+ASV therapy was associated with the presence of NS5A Q24, L28, and/or R30 RASs and concomitant F37 and Q54 RASs before treatment [2].
In July, 2015, sofosbuvir (SOF) (an NS5B inhibitor) and ledipasvir (LDV) (an NS5A inhibitor) combination therapy were authorized for the treatment of chronic hepatitis or compensated cirrhosis in patients with HCV genotype 1 in Japan. An open-label phase 3 clinical trial in Japan showed that an SVR was achieved by 100% of patients [3]. The presence of baseline NS5A RASs did not affect the treatment outcome [4]. However, in the real world, some patients experience virologic failure. We therefore aimed to reveal the factors associated with virologic failure in patients treated with SOF/LDV, and to identify the baseline NS5A and NS5B RASs.

The study population
A total of 493 patients with HCV genotype 1 infection who were treated with SOF (90 mg/day) and LDV (400 mg/day; Harboni; Gilead Sciences, Tokyo, Japan) for 12 weeks were followed for an additional 12 weeks after treatment. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by Kagoshima University Hospital Clinical Research Ethics Committee (approval number: 170199). Written informed consent was obtained from each patient. The HCV RNA concentration was measured by a TaqMan PCR, which has a lower quantitation limit of 1.2 log IU/mL. The FIB4 index, a surrogate marker of liver fibrosis, was calculated based on the methods of previous studies [5].

The detection of HCV RASs
As described in detail previously [2], we investigated the viral genome sequence by direct sequencing. Total RNA was extracted from 140 μL of each serum sample using a QIAamp Viral RNA Mini Kit (QIAGEN, Tokyo, Japan). The extracted RNA was reverse-transcribed using a PrimeScript RT Reagent Kit attached Random 6 mers and Oligo dT Primer (TaKaRa Bio, Kusatsu, Japan). The target HCV genome was amplified by a nested PCR using PrimeS-TAR GXL DNA Polymerase or LA Taq with GC Buffer (TaKaRa Bio). The primers used to amplify the NS5A and NS5B regions have been reported previously [6][7][8][9] (S1 Table). Reverse transcription was performed at 37˚C for 15 minutes and terminated at 85˚C for 5 seconds, followed by the first-round PCR over 30 cycles, with each cycle consisting of denaturation at 98˚C for 10 seconds, annealing at 60˚C for 15 seconds and extension at 68˚C for 60 seconds. The second-round PCR was performed under the same condition. After purification with a GenElute PCR Clean-Up Kit (Sigma-Aldrich, Saint Louis, MO, USA), the nucleotide sequences of the second amplicons were determined using a BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific, Waltham, MA, USA) and Sanger sequencing. The sequences of the NS5A and NS5B RAS positions in the HCV gene were determined using HCV-Con1 (accession no. AJ238799) as a reference, as previously described and the package insert of Harboni 1 [10][11][12][13][14][15][16][17][18]

Statistical analysis
Statistical analyses were performed using the Statistical Package for Social Sciences software program (version 18, SPSS, Inc., Chicago, IL, USA). Categorical data were compared using the chi-squared test and Fisher's exact test, as appropriate. Continuous variables were analyzed using the Mann-Whitney U test. P values of <0.05 were considered to indicate statistical significance. Factors associated with virologic failure were determined using a multivariable logistic regression analysis with forward selection using p<0.15 as a cutoff for inclusion in the model. SVR was estimated in an intention-to-treat analysis.

The baseline characteristics of the patients
The demographic baseline characteristics of the 493 patients are shown in Table 1. The age of the patients ranged from 26 to 90 years (mean, 68 years). There were 295 female patients (60%) and 105 (21%) patients with cirrhosis. Thirty-one patients had previously been treated with DCV and ASV. Sixty-three patients (13%) had previously been treated for hepatocellular carcinoma (HCC). NS5B A218 RAS (79.1%) had concomitant NS5B A218 and C316 RASs (It was not possible to detect A218 or C316 RASs in 11 patients).

The virologic response rate
Among the 462 IFN-free treatment-naïve patients who received SOF/LDF, 452 patients (97.8%) completed treatment. Overall, 95.0% of the patients (363/450) had undetectable HCV RNA levels at the fourth week of treatment and 99.8% (450/451) had undetectable HCV RNA levels at the end of treatment. An SVR at 12 weeks (SVR12) was achieved in 451 patients (97.6%) in the intention-to-treat analysis (Fig 1). On the other hand, among the 31 patients who received SOF+LDF therapy in IFN-free retreatment, 30 patients (96.8%) completed treatment. Overall, 63.3% of patients (19/30) had undetectable HCV RNA levels at week 4 of treatment and 100% (30/30) had undetectable HCV RNA levels at the end of treatment. An SVR12 was achieved in 25 patients (80.6%) in the intention-to-treat analysis (Fig 1).

The RASs in patients with virologic failure
Eleven (2.2%) patients (IFN-free treatment-naïve, n = 6; IFN-free retreatment, n = 5) were considered to have virologic failure ( Table 2). All patients had relapsed disease. We examined the baseline RASs of these 11 patients. Eight of 11 patients (72.7%) had NS5A Y93 RASs, 5 of 11 patients (45.5%) had NS5A Q24, L28, and/or R30 RASs, and 4 of 11 patients (36.4%) had NS5A L31 RASs. Three patients had coexisting NS5A L31 and Y93 RASs at the baseline. In brief, 8 of 11 patients (72.2%) had coexisting NS5A Q24, L28 and/or R30 RASs, L31 RAS or Y93 RAS at the baseline. One patient who underwent IFN-free retreatment and who did not have NS5A L31 or Y93 RASs, had baseline NS5A P32L and A92K RASs. In the analysis of post-treatment RASs, the patients who had coexisting Q24, L28, and/or R30, L31 or Y93 RASs at the baseline showed no change in their NS5A RAS profile. However, two patients acquired NS5A L31 and Y93 RASs. The patient who had baseline NS5A P32L and A92K RASs did not acquire NS5A L31 or Y93 RASs. We analyzed 7 NS5B RASs. Interestingly, 10 of 11 patients had an NS5B A218 RAS, 8 of 11 patients had an NS5B C316 RAS, and 5 of 11 patients had an NS5B A207 RAS at the baseline. No patients had NS5B L159, S282, L320 or V321 RASs at the baseline. In particular, the patients who had NS5B C316 RAS also had coexisting A218 RAS. In addition, no patients acquired these RASs.

Factors associated with virologic failure
We analyzed the factors associated with virologic failure among the patients who completed treatment. Among the blood biochemistry data, significant differences were observed in the platelet count and γ-glutamyltransferase (GGT) levels between the SVR12 and no-SVR12 groups; however, there were no differences between the IFN-free treatment-naïve and IFNfree retreatment groups (S3 Table). NS5A and NS5B resistance-associated substitutions associates virologic failure Table 3 showed the SVR rates according to the various categories. A univariate analysis in all cases showed that HCV RNA !6.5 log IU/mL (p = 0.016), platelet count <10×10 4 /μL (p = 0.041), and GGT !37IU/L (p = 0.022) were associated with an SVR12. Among the RASs of NS5A, RASs of NS5A: Q24, L28, and/or R30 (p = 0.014), L31 (p = 0.006), P32 (p<0.001), and Y93 (p = 0.001), were associated with SVR12. In particular, coexisting NS5A RASs (Q24, L28, and/or R30, L31, or Y93 RAS) were associated with a lower SVR12 rate in comparison to no RASs or a single RAS. Interestingly, among the RASs of NS5B, RASs of A218 (p = 0.013) and C316 (p = 0.042) were associated with an SVR12. Similarly, NS5B A218 and/or C316 were significantly associated with an SVR12 (p = 0.015).

SVR12 rate in patients with coexisting NS5A and NS5B RASs
We analyzed the SVR12 rate in patients with coexisting NS5A and NS5B RASs. Among the IFN-free treatment-naïve patients, all of the patients without NS5B A218 and/or C316 RASs achieved an SVR12 (Fig 2). In particular, the SVR12 rate of the patients who had more than two NS5A RASs (Q24, L28 and/or R30 RAS, L31 RAS, or Y93 RAS) and NS5B A218 and/or C316 RASs was 83.3%. On the other hand, in the IFN-free retreatment patients, all patients with virologic failure had coexisting NS5A RASs (Q24, L28 and/or R30 RAS, L31 RAS, or Y93 RAS) with the exception of one case that had NS5A P32L and A92K RASs; however these RASs did not have a significant difference on the SVR12 rate.

Discontinuation of therapy
Eleven patients (2.2%) discontinued SOF/LDV therapy (Table 5). All patients were >65 years of age. Three patients discontinued therapy due to fatigue, and 2 patients each discontinued therapy due to renal injury or skin itching. Five patients discontinued therapy due to other reasons (hyperbilirubinemia, diarrhea, brain hemorrhage, and unknown). The mean time to discontinuation was 4.5 weeks. All patients receiving SOF/LDV for more than 8 weeks achieved an SVR12.

Discussion
We revealed, for the first time, that most SOF/LDV-treated patients with virologic failure had RASs of NS5B A218 and/or C316, and these RAS was significantly associated with virologic failure. Donaldson et al. reported that NS5B C316 was associated with virologic failure [17]; however while many studies have reported the analysis of large study populations [19][20][21][22][23][24][25], no reports have indicated that NS5B A218 RAS is associated with virologic failure in SOF/LDVtreated patients. The reason why we paid attention to NS5B A218 and/or C316 RAS was that 8 of the 11 patients with virologic failure had coexisting NS5B A218 and C316 RAS (  Table). Further, the prevalence of NS5B C316 in the IFN-free treatment-naïve and IFN-free retreatment patients in this cohort was 42.7% (193/452) and 56.7% (17/30) respectively (S2 Table). Ito et al. reported that the prevalence of NS5B C316N in 96 Japanese patients was 46.9% [9]. However, no studies have reported the prevalence of NS5B A218 RAS. Especially, 99.0% of patients who had NS5B C316 RAS had coexisting A218 RAS. The prevalence needs to be confirmed in other populations.
SOF is a nucleotide inhibitor directed at the activity of NS5B, the protein of which is essential for HCV replication. SOF, which prevents HCV RNA synthesis by acting as a "chain terminator", has broader genotype coverage and a higher barrier to viral resistance [13]. Some clinical researchers reported that RASs of NS5B L159, S282, C316, L320, or V321 were associated with virologic failure [14,15,17]. However, with the exception of S282, these RASs were associated with low-level resistance in the replicon systems [13,14,16,26]. In this cohort, no cases had RASs of S282, L320 before treatment or at the time that treatment was deemed unsuccessful, and all of the cases with RASs of L159 and V321 achieved an SVR (Table 3). There have been no reports about the RAS of NS5B A218 or the coexistence of A218 and C316 in vitro. Thus, it is important to confirm the effects of SOF in cases involving the coexistence of NS5B A218 and/or C316 RASs using a replicon system or a human hepatocyte chimera mouse model as soon as possible.
SOF/LDV therapy was associated with good therapeutic outcomes, even in patients with NS5A Y93 RAS, with 100% of the patients in a phase III trial in Japan showing an SVR [3]. Some reports indicate that NS5A Y93 RAS is not related to the therapeutic effect [4,19,24]. The American Association for the Study of Liver Diseases/Infectious Diseases Society of America [27] and the European Association for the Study of the Liver guidelines [28] have not mentioned any need for the routine investigation of NS5A RAS. On the other hand, other reports suggest that it is related to the therapeutic effect [20,21]. In the present study, the SVR12 rate was 97.6%; thus the routine investigation of NS5A RAS might not be necessary. However, most cases of virologic failure involved coexisting NS5A RAS, and this was related to the therapeutic effect (Table 4). It is noteworthy that the SVR rate of the IFN free-naïve patients with a single substitution of Q24, L28, and/or R30, L31, or Y93 were significantly higher than those with more than two substitution (>99% vs 88.2%) ( Table 3), and most cases of virologic failure involved the coexistence of these RASs (Table 2). On the other hand, an SVR was achieved in all cases without NS5B A218 and/or C316 RASs. In contrast, the SVR12 rates of the patients who had more than two NS5A RASs and NS5B A218 and/or C316 RAS was 83.3% (Fig 2). It was suggested that the coexistence of NS5A and NS5B RASs was highly associated with virologic failure; thus, investigating the baseline NS5A and NS5B RAS status, is useful for in efforts to increase the SVR rate to 100%. Therefore, the existence of baseline RAS should be considered at the choice of treatment regimen in order to eradicate HCV.
Overall, HCV elimination is not achieved in 10-15% of the patients who receive DCV/ASV [1,2,29]. This represented a major problem for HCV treatment because such patients had coexisting RASs in the NS3 and NS5A regions and these RASs had cross-resistance [30]. Upon DCV/ASV treatment failure in human hepatocyte chimeric mice, NS5A L31V and Y93H RAS were relatively resistant to ledipasvir and NS5B polymerase nucleotide inhibitor [31]. In previous studies, the SVR rates of IFN-free retreatment patients who received LDV/SOF (+ribavirin) were 71-86.7% [20,[32][33][34]. However, it was unclear which RASs were associated with virologic failure in these patients. In the present study, most patients with a single RAS of NS5A were achieved an SVR; however the result did not reach statistical significance, because of the small number of cases. Furthermore, NS5A P32 RAS was observed in 3 patients (one patient was IFN-free treatment-naïve, two patients received IFN-free retreatment). IFN-free treatment-naïve patients with NS5A P32 RAS achieved an SVR, one of two IFN-free retreatment patients showed virologic failure and had coexisting NS5A A92K RAS. These RASs were very rare; however, they have been reported to emerge after failure to respond to DCV/ASV [2,29,35]. NS5A A92K RAS is reported to have over 1000-fold resistance to LDV in comparison to the wild-type; in addition, P32L RAS have 2.5-10.0-fold resistance [26]. The coexistence of NS5A P32L and A92K RASs might be associated with the therapeutic effect. Meanwhile, there were no patients with NS5A P32 deletion in this cohort. In Japan, Glecaprevir/Pibrentasvir therapy was approved in September 2017, following a study in which the SVR rate in patients with DAA retreatment was 93.9%; however, two patients with virologic failure had an NS5A P32 deletion [36]. In the future, it is necessary to clarify the factors associated with the therapeutic effects of IFN-free retreatment patients in a large study population.
In the present study, GGT elevation was associated with virologic failure. Abe et al. reported that GGT elevation was associated with IL28B polymorphism [37]. The therapeutic effect of SOF/LDV might be related to host factors in IFN-free retreatment especially.
The present study was associated with several limitations. First, only 11 (2.2%) of 482 patients had virologic failure; thus, the range of the 95% CI was wide, despite the fact that a statistically significant difference was observed. Second, it was not possible to detect all RASs in some cases. Third, since the number of cases was smaller than that of previous studies, a larger cohort should be analyzed in a future study.
In conclusion, SOF/LDV therapy showed a high SVR12 rate. However, the coexistence of NS5A and NS5B RASs were associated with virologic failure. These results might indicate that the coexistence of baseline RASs influence the therapeutic effects of SOF/LDV. Supporting information S1