iconstar paper   Hepatitis C Articles (HCV)  
Back grey arrow rt.gif
Failure on voxilaprevir, velpatasvir, sofosbuvir and efficacy of rescue therapy
  Download the PDF here
April 1 2021 Jnl of Hepatology - Julia Dietz1,2,, Velia Chiara Di Maio3,, Adolfo de Salazar4,, Dolores Merino5, Johannes Vermehren1,2, Stefania Paolucci6, Andreas E. Kremer7, Magdalena Lara8, Maria Rodriguez Pardo9, Heinz Zoller10, Elisabetta Degasperi11, Kai-Henrik Peiffer1,2, Laura Sighinolfi12, Francisco Téllez13, Christiana Graf1,2, Valeria Ghisetti14, Jonas Schreiber15, Elisa Fernández-Fuertes16, Lucio Boglione17, Leopoldo Muñoz-Medina18, Rudolf Stauber19, William Gennari20, Blanca Figueruela21, Jesús Santos22, Pietro Lampertico11, Stefan Zeuzem1,2, Francesca Ceccherini-Silberstein3,a,, Federico García4,b,, Christoph Sarrazin 1,2,23,*,c,
Lay summary
The advent of direct-acting antivirals has enabled the effective cure of chronic hepatitis C in most patients. However, treatment failure occurs in some patients, who are often retreated with a combination regimen called VOX/VEL/SOF, which is associated with very high rates of cure. However, VOX/VEL/SOF retreatment also fails in some patients. Herein, we analysed samples from patients in whom VOX/VEL/SOF retreatment failed and we assessed the efficacy of different rescue therapies, showing that rescue treatment is effective in most patients (81%).
• In patients with VOX/VEL/SOF failure HCV GT3a infections and cirrhosis were overrepresented and HCC was present in 28% of patients
• Sequencing after VOX/VEL/SOF failure showed minor changes in the NS3 and NS5A RAS pattern and no NS5B RASs were observed in comparison to baseline
• Rescue treatment was conducted in 55% of patients and the main reasons were the presence of advanced liver disease and/or liver cancer
• The majority of patients (68%) received G/P+SOF ± ribavirin
• Rescue treatment was effective in the majority of patients with 81% SVR
Background & Aims
There are limited data on patients with chronic HCV infection in whom combination voxilaprevir (VOX), velpatasvir (VEL), sofosbuvir (SOF) retreatment fails. Thus, we aimed to assess treatment failure and rescue treatment options in these patients.
Samples from 40 patients with HCV genotypes (GT) 1-4 in whom VOX/VEL/SOF retreatment failed were collected within the European Resistance Study Group. Population-based resistance analyses were conducted and clinical parameters and retreatment efficacies were evaluated retrospectively in 22 patients.
Most VOX/VEL/SOF failure patients were infected with HCV GT3a (n = 18, 45%) or GT1a (n = 11, 28%) and had cirrhosis (n = 28, 70%). Previous treatments included an NS3-inhibitor (30%), an NS5A-inhibitor (100%) and SOF (85%). Baseline RAS data from a subgroup of patients before VOX/VEL/SOF retreatment (78%) showed few NS3 RASs apart from Q80K in GT1a (40%), typical NS5A RAS patterns in most patients (74%) and no S282T in NS5B. Sequencing after VOX/VEL/SOF failure was available in 98% of patients and showed only minor changes for NS3 and NS5A RASs. In 22 patients, rescue treatment was initiated with glecaprevir, pibrentasvir alone (n = 2) or with SOF±ribavirin (n = 15), VOX/VEL/SOF±ribavirin (n = 4) or VEL/SOF and ribavirin (n = 1) for 12 to 24 weeks. Sustained virologic response was achieved in 17/21 (81%) patients with a final treatment outcome. Of these, 2 GT3a-infected patients had virologic failure after rescue treatment with VEL/SOF or glecaprevir/pibrentasvir+SOF+ribavirin, and 2 patients with cirrhosis died during treatment or before reaching SVR12.
VOX/VEL/SOF failure was mainly observed in HCV GT3- and GT1a-infected patients with cirrhosis and was not associated with specific RAS patterns within NS3, NS5A or NS5B target regions. Rescue treatment with multiple targeted therapies was effective in most patients.


Chronic HCV infection is a global health challenge leading to progression of chronic liver disease, cirrhosis and hepatocellular carcinoma (HCC). Overall, high sustained virologic response (SVR) rates of more than 90-95% across all HCV genotypes and subtypes can be achieved by combining second-generation direct-acting antivirals (DAAs). However, due to the high numbers of patients who were treated since the introduction of DAAs, the number of patients who have failed on DAA treatment including an NS5A inhibitor (NS5Ai) is increasing. Resistance-associated substitutions (RASs) were selected in most of these patients and NS5A RASs were common after NS5Ai failure and persisted after end-of-treatment (EOT) because their fitness was not affected. Several real-world studies have shown that high-level resistant NS5A RASs reduce the effectiveness of certain retreatment regimens if first generation DAAs are used.
The second-generation DAA regimen consisting of the second-generation protease inhibitor (PI) voxilaprevir (VOX) plus the NS5A inhibitor velpatasvir (VEL) plus sofosbuvir (SOF) has been approved for pangentopic retreatment after DAA failure. In phase III clinical studies, 97% of DAA-experienced patients achieved SVR after retreatment with VOX/VEL/SOF and RASs had no influence on treatment outcome. In total, only 7 patients with virologic failure on VOX/VEL/SOF retreatment were observed in the approval studies and most of these patients were infected with HCV genotype (GT) 3 and had cirrhosis. Even in real-world studies, the effectiveness of VOX/VEL/SOF was ≥95% and very few patients experienced treatment failure. Furthermore, no studies are available that have investigated a rescue treatment for patients in whom a prior retreatment with VOX/VEL/SOF had failed. Therefore, this multicentric European real-world study aimed to analyse clinical and virological characteristics, RAS patterns, and the efficacy of rescue treatments in patients following VOX/VEL/SOF retreatment failure.
Clinical characteristics of patients with VOX/VEL/SOF retreatment failure
Overall, samples from 40 patients in whom retreatment with VOX/VEL/SOF failed were collected in the European Resistance Study group. The patients were infected with HCV GT3a (45%, n = 18/40), GT1a (28%, n = 11/40), GT1b (23%, n = 9/40) and GT3b or GT4d (4%, n = 2/40). Overall, 70% of patients had cirrhosis (n = 28/40), mostly classified as Child-Pugh grade A (87%, n = 20/23 with available data). The prevalence of HCC was 28% (n = 11/40) and a liver transplantation was performed in 6 individuals (15%, n = 6/40) (Table 1).
Most patients (78%, n = 31/40) had a history of 1 previous DAA treatment and the other 22% (n = 9/40) had a history of multiple DAA treatments. All patients had at least 1 previous course of a NS5Ai treatment, 30% (n = 12/40) had received protease inhibitor (PI)-based treatment and 85% (n = 34/40) had received a SOF-based regimen (Table 1). A total of 35% (n = 14/40) were pre-treated with at least 1 previous course of VEL/SOF (Table S4). NS3, NS5A and NS5B RAS analyses were conducted at baseline in ≥78% of patients and after VOX/VEL/SOF failure in ≥98% of patients (Table S5). The median time interval between the EOT of the last DAA pre-treatment and the baseline RAS test before VOX/VEL/SOF retreatment was 6.5 (0.2-31.5) months. The time interval between the baseline RAS test and VOX/VEL/SOF retreatment initiation was 5.6 (0.0-41.9) months. The time interval between the EOT of VOX/VEL/SOF and the second RAS test was 5.2 (0.0-21.5) months. In patients who received rescue treatment, the median time interval between RAS testing after VOX/VEL/SOF failure and the start of rescue therapy was 7.0 (0.5-22.0) months (Table 2, Fig. S1).
RASs in patients with VOX/VEL/SOF retreatment failure and HCV GT1a infection
A total of 11 patients were included and serum samples were available for 10 patients each at baseline and after VOX/VEL/SOF failure (Table S5). RAS testing after VOX/VEL/SOF failure showed that all patients had either NS3 or NS5A RASs (Fig. 1A).
Regarding NS3, a total of 50% of patients (n = 5/10 with sequencing data available) harboured Q80K after VOX/VEL/SOF failure. With the exception of 1 patient with Q80K, no NS3 RASs were selected in patients with matching sequencing data available. In another patient, D168E could no longer be detected after VOX/VEL/SOF failure (Fig. 1B, Table S6). The NS5A RAS pattern after VOX/VEL/SOF failure was also comparable to baseline. One patient had a change in the NS5A RAS pattern. This patient harboured Q30H, L31V at baseline and lost Q30H after VOX/VEL/SOF failure (Fig. 1C, Tables S6 and S7). Overall, the pre-existing NS3 and NS5A RAS patterns were maintained in most patients after VOX/VEL/SOF failure (78%, n = 7/9, and 89%, n = 8/9, with matching sequencing data available, respectively).
Regarding NS5B, we could not detect RASs in any patient (Table S4).
RASs in patients with VOX/VEL/SOF retreatment failure and HCV GT1b infection
Nine patients were included, and serum samples were available from 7 patients at baseline and from 9 patients after VOX/VEL/SOF failure (Table S5).
After VOX/VEL/SOF failure, individual NS3 RASs were rather rare and the overall frequencies of NS5A RASs were moderate (Fig. 2A). Interestingly, no NS3 RASs were selected upon treatment failure. The RAS profile was maintained in 86% of patients (n = 6/7 with matching sequencing data available) compared to baseline. One individual with baseline Q80R plus D168E lost D168E upon treatment failure (Fig. 2B, Tables S6 and S7).
Also, there were only few changes in the RAS profile in NS5A after VOX/VEL/SOF failure. Of 7 patients with matching sequencing data available, 1 patient lost Y93H and a second patient selected L31M in addition to pre-existing Y93H. (Fig. 2C, Tables S6 and S7). Within NS5B, L159F plus C316N were detected in 1 GT1b-infected patient both at baseline as well as after VOX/VEL/SOF failure (Table S4).
RASs in patients with VOX/VEL/SOF retreatment failure and HCV GT3a infection
Eighteen patients were followed-up and serum samples from 14 patients were investigated at baseline and 18 samples after VOX/VEL/SOF failure (Table S5).
After VOX/VEL/SOF failure, no single NS3 RASs were detected, but NS5A and NS3 plus NS5A combination RASs were relatively common (Fig. 3A). Overall, NS3 RASs were less frequent in HCV GT3a compared to other genotypes and the RAS profile was maintained in all patients (n = 10/10 with matching NS3 sequencing data available). One patient had Q168K, another had Q168R, and a third patient harboured high-level VOX-resistant Y56H plus Q168R at both time points. (Fig. 4B, Tables S4 and S7).
Changes in the NS5A RASs pattern were only observed in a few patients. Out of 13 patients with matching NS5A sequencing data available, 3 patients had a change in the RAS profile after VOX/VEL/SOF failure. One patient lost pre-existing Y93H. Two patients selected additional NS5A RASs. One patient had pre-existing Y93H and additionally selected A30K. The second patient had pre-existing A30K and additionally selected Y93H. Overall, high-level VEL-resistant Y93H was more frequent than in the other HCV genotypes (Fig 3C, Tables S6 and S7).
Regarding NS5B, A150V plus V321A were detected at baseline and after VOX/VEL/SOF failure in 1 patient (Table S4).
RASs in individual cases with VOX/VEL/SOF retreatment failure
In this group, 2 patients with HCV GT3b and GT4d were followed-up. One serum sample could be examined at baseline and 2 serum samples were available after treatment failure (Table S5).
The patient with HCV GT3b had no NS3 RASs and the NS5A A30K plus L31M double variant after VOX/VEL/SOF failure, while no serum sample was available at baseline.
The patient with HCV GT4d selected A156S to pre-existing D168V within NS3 and maintained the M31V plus Y93H double variant in NS5A after VOX/VEL/SOF failure.
NS5B RASs were not found in either patient (Tables S4, S6 and S7).
Rescue treatment in patients with VOX/VEL/SOF retreatment failure
Rescue treatment was initiated in 55% (n = 22/40) of VOX/VEL/SOF failure patients, of whom 27% (n = 6/22) had a diagnosed HCC. The main reasons for not conducting rescue treatment were the diagnosis/suspicion of HCC as well as the postponement of treatment for different reasons (Table S8).
Most patients received glecaprevir/pibrentasvir (G/P)+SOF with or without RBV as rescue treatment and the treatment duration was 12-24 weeks (68%, n = 15/22). Two patients received G/P without SOF. Another 5 individuals received a repetition of VOX/VEL/SOF±RBV or VEL/SOF+RBV for 24 weeks (Table 3). The final treatment outcome was available for 21 patients and SVR was achieved in 81% of patients (n = 17/21) (Fig. 4).
Of the 17 patients who achieved SVR, 11 had responded to G/P+SOF±RBV. Two more patients achieved SVR after G/P±RBV without SOF. Four additional patients achieved SVR after a repetition of VOX/VEL/SOF±RBV (Table 3, Fig. 4).
Of the 4 patients without SVR12, 1 patient died on treatment due to complications from decompensated cirrhosis and a second patient with cirrhosis died after reaching SVR4. Two patients with HCV GT3a had virologic failure on rescue treatment. A patient with decompensated cirrhosis and pre-existing high-level VEL-resistant Y93H, relapsed after rescue treatment with VEL/SOF+RBV for 24 weeks and the RAS profile was identical to that before treatment. Interestingly, this patient was subsequently retreated with G/P+SOF+RBV for 12 weeks and achieved SVR. The second patient relapsed to G/P+SOF+RBV for 16 weeks. While Y93H was found prior to rescue treatment, the triple variant A30K, L31F, Y93H was selected after treatment failure and this patient was not subsequently retreated. Rescue treatment was also conducted in one more patient, for whom follow-up after EOT is still pending.
Overall, 55% (n = 6/11) of patients with HCC received rescue treatment. The SVR rate in the 6 patients with HCC was 67% (n = 4/6) compared to 86% (n = 12/14) in patients without HCC (p = 0.5) (Table 3).
Data on the characteristics and RAS profiles of patients with virologic failure to a VOX/VEL/SOF retreatment is limited as only single patients could be investigated in clinical studies and data concerning the efficacies of a rescue treatment in this difficult-to-treat patient group are not available.
This real-world study characterized 40 patients from different European countries infected with HCV GT 1-4 in whom VOX/VEL/SOF retreatment failed. All patients had previously received at least 1 DAA treatment including a NS5A inhibitor and the majority of patients (>75%) maintained their NS5A RAS profile after VOX/VEL/SOF failure, which is in accordance with the phase III clinical studies.
Overall, NS3 RASs conferring a >2-fold resistance to VOX or glecaprevir were rare. One reason could be the low rate of PI-based pre-treatments. The median time of RAS testing after DAA pre-treatment was 6.5 (0.2-31.5) months after EOT. NS3 RASs disappear within a few weeks after EOT due to replication deficiency, while NS5A RASs can persist for years. Therefore, the probability of NS3 RASs being detected more than 6 months after EOT is low. Moreover, most RASs conferring resistance to other PIs did not impact VOX or glecaprevir susceptibility. That is also the case for Q80K, which was detected at higher frequencies after VOX/VEL/SOF failure compared to its natural prevalence. One reason may be that Q80K mediates viral escape from PIs with fitness advantages.
Regarding the NS5A RAS profile, we observed some differences between the HCV genotypes. In HCV GT1a, a variety of RASs appeared as previously described including high-level VEL-resistant combination RASs. In HCV GT1b, NS5A RASs were only detected at positions L31 and Y93H and detected combination RASs were shown to confer medium- to high-level VEL resistance.
The highest prevalence of NS5A RASs after VOX/VEL/SOF failure was found among patients with HCV GT3a, although the median times of RAS analyses after EOT were comparable with other GT. One reason could be that all NS5A RASs conferred at least medium-level VEL resistance in GT3a which was not the case for GT1. Especially for high-level VEL-resistant RASs, such as Y93H and A30K plus Y93H, it has been shown that they can persist for years after EOT.
Interestingly, 1 patient was infected with subtype 3b and harboured A30K plus L31M. This double variant was shown to confer >10,000-fold VEL resistance and was detected in a cohort of DAA-naïve patients infected with HCV GT3b and GT3g suggesting that these subtypes may be inherently resistant to NS5Ai. Unfortunately, this patient did not receive rescue treatment, as he returned to his home country. NS5B RASs that confer resistance to SOF have rarely been observed after DAA failure to treatment with SOF-containing regimens and the in vitro SOF resistant S282T variant had a poor replication capacity and quickly disappeared after EOT. In our study, S282T was not observed. Two SOF-pre-treated patients harboured NS5B variants (L159F plus C316N or A150V plus V321A) that have been described in association with a SOF-based treatment in other reports.
Regarding DAA pre-treatments, a VEL/SOF treatment was often conducted in our study (35%), especially in patients with HCV GT3a. One real-world study showed that SVR rates to VOX/VEL/SOF were lower in patients with previous VEL/SOF experience. The addition of just 1 new DAA (like VOX) may not be optimal. This could be particularly the case when high-level VEL-resistant NS5A RASs and comorbidities like cirrhosis/HCC occur together.
Although no DAA serum levels could be determined in this real-world study, treatment adherence can be assumed, since the type of treatment response (including a possible non-compliance) was documented retrospectively.
Unfortunately, this real-world study lacks a comparison group of patients who achieved SVR after VOX/VEL/SOF retreatment in order to identify factors associated with achievement of SVR including baseline RASs. Cirrhosis, HCV GT3 and HCC were associated with VOX/VEL/SOF treatment failure in real-world studies.
In our study, too, the proportion of patients with these comorbidities was high. This indicates that these are indeed risk factors for treatment failure, whereas RAS patterns were rather heterogeneous and, remarkably, most VOX/VEL/SOF failure patients had no change in the NS3, NS5A and NS5B RAS pattern compared to baseline (69%, n = 18/26 with matching sequencing data available). The selection of RASs in patients with VOX/VEL/SOF treatment failure was also uncommon in other studies and baseline RASs did not impact SVR rates. However, the risk of treatment failure should be minimized and the administration of ribavirin should be considered in patients with HCV GT3, cirrhosis and high-level resistant Y93H. In addition, a possible HCC should be identified in all patients with cirrhosis before treatment initiation.
Rescue treatment was successful in most of these difficult-to-treat patients with advanced liver disease. This is interesting because no patients with HCC were included in the clinical studies and data from real-world studies on this patient group are sparse. Data on the efficacy of a rescue treatment including patients with a prior VOX/VEL/SOF failure are not available. A small study showed high SVR rates (≥95%) in patients with previous DAA failure with advanced liver disease/cirrhosis when retreated with G/P+SOF, excluding patients with VOX/VEL/SOF failure. In the ongoing Magellan-3 clinical study, rescue treatment for patients with previous G/P failure using G/P+SOF+RBV for 12-16 weeks was highly effective with 96% SVR.
In our study, the SVR rate in patients who received G/P+SOF as rescue treatment was 79%. Patient 22 had post-treatment relapse to G/P+SOF+RBV, HCV GT3a, cirrhosis, and harboured the NS5A A30K, L31F, Y93H triple variant. Moreover, an HCC was first diagnosed during rescue treatment. Multiple RASs were frequent after G/P failure and conferred higher resistance levels compared to single RASs. Most RASs that conferred high-level resistance to first generation NS5Ai, maximally conferred medium-level pibrentasvir resistance. This could explain the high efficiency of G/P+SOF in patients with VOX/VEL/SOF failure. Since patient 22 only harboured non- pibrentasvir resistant Y93H before rescue treatment initiation, it is likely that the undetected and untreated HCC was the main reason for rescue treatment failure. This patient has not received any further rescue therapy.
Also, patient 32 who was infected with HCV GT3a, did not achieve SVR upon rescue treatment. This patient has decompensated cirrhosis, therefore rescue treatment was initially conducted without a PI using VEL/SOF+RBV. However, the inefficiency of repeating a NS5Ai/SOF-based rescue treatment in presence of high-level resistant NS5A RASs (as with Y93H in this case), particularly in patients with HCV GT3a, has been proven in several studies. Of note, in this patient no previously undetected HCC was detected by sonography after rescue treatment failure. Interestingly, this patient finally achieved SVR after G/P+SOF+RBV for 12 weeks.
Overall, in this study the number of patients with HCC was relatively high and it seems that HCC may be generally associated with DAA treatment failure along with other risk factors such as cirrhosis and HCV GT3a. Since the SVR rates to rescue treatment were lower in patients with HCC compared to patients without HCC (67% vs. 86%), HCC should be identified before DAA treatment initiation and the addition of ribavirin and longer treatment durations up to 24 weeks should be considered. Few patients who received a repetition of VOX/VEL/SOF or G/P alone without SOF achieved SVR. However, in this patient group, comorbidities and high-level resistant NS5A RASs were rare. Thus, repetition of DAA regimens cannot be recommended as a general approach for rescue treatment in VOX/VEL/SOF failures.
Regarding the detected RASs, there was no uniform pattern that could predict failure of rescue therapy after VOX/VEL/SOF failure. It should be noted that in addition to the major RASs detected in the blood, minor RASs and other variants outside of the target regions may be present as well as variants in other compartments (e.g. in the liver). General RAS testing before starting rescue treatment after VOX/VEL/SOF failure is not mandatory, but can be helpful for patients with risk factors and may guide retreatment in case of highly resistant RAS patterns. Furthermore, the number of patients with VOX/VEL/SOF failure is limited and data on RAS profiles for different HCV genotypes and subtypes should be collected for more valid analysis in larger patient groups. Based on results from the present study, we recommend a rescue treatment with G/P+SOF or VOX/VEL/SOF for VOX/VEL/SOF failure patients.
So far is unclear whether in the presence of advanced liver disease or other negative predictors, such as certain RAS profiles, G/P+SOF may be superior to VOX/VEL/SOF. This, and the question of whether ribavirin should be added or whether a specific treatment duration (i.e. 12, 16, 24 weeks) is required, should be addressed in further studies on the management of VOX/VEL/SOF failures. The limitations of this study are the non-controlled design, the decentralized sequencing analyses conducted at several European centres as well as the retrospective data analysis. In summary, in the present study, most patients with failure on a VOX/VEL/SOF retreatment did not select specific RAS patterns in NS3, NS5A or NS5B DAA target regions compared to baseline. HCV GT3- and GT1a-infected patients with cirrhosis were frequent in this difficult-to-treat cohort, and rescue treatment with multiple targeted therapies was effective in most patients.

  iconpaperstack View Older Articles   Back to Top   www.natap.org