The importance of resistance to direct antiviral drugs
in HCV infection in clinical practice - Review
Download The PDF here
"Insufficient data are available for determination of the importance of NS3 RAVs for re-treatment with protease inhibitor containing regimens....Interestingly, in contrast to NS3 protease RAVs, available data suggest that long term persistence of RAVs to NS5A inhibitors is more likely as viral fitness seems not to be impaired. In clinical studies persistence of NS5A RAVs over 1-2 years after treatment failure in over 85% of patients has been reported [, , ].......For the combination of a first generation NS3 protease and NS5A inhibitor with a relative low barrier to resistance the presence of baseline resistance has a great impact with bisection of SVR rates. Here, baseline resistance testing is mandatory. However, for regimens with high antiviral activities and high genetic barrier to resistance based on single DAAs or the combination of different drug classes the presence of baseline resistance leads only to a small reduction of SVR rates. Here, additional predictors of response are of importance. While no general recommendation for baseline resistance testing can be given in subgroups of patients with certain HCV geno- or subtypes, in patients with shortened treatment duration or those with liver cirrhosis resistance testing may be used to select optimal DAA regimens. Future studies have to explore whether due the high costs of DAA regimens in regions with economical restriction baseline resistance testing for initial DAA combination regimens with the aim to avoid virologic treatment failure and the need of retreatment may be cost effective.
A number of second generation DAAs are in development. Here, unmet medical needs mainly include the development of an effective treatment in genotype 3 infected patients and on the same time to ensure coverage of all different HCV geno- and subtypes. Achievement of the latter aim would facilitate treatment of chronic hepatitis C significantly but due to the large number of HCV subtypes represents a major challenge with the risk of improvement of antiviral activity in one subtype but loosing of efficacy in other HCV subtypes.
Currently, effective treatment options in patients with failure to all-oral DAA regimens are not defined yet. Despite high SVR rates around 95% given the large number of patients with chronic hepatitis C overall the number of patients with virologic treatment failure will be high. Based on available data more than 80% of these patients will harbor HCV isolates with resistance to one, two or three DAA classes. Also in patients with failure to DAA-based regimens an accumulation of other negative response factors is likely. One approach is re-treatment with the same DAAs for a longer duration. Indeed in patients who received initial therapies for 24 weeks relapse rates have been extremely low. However, preliminary data in small patient cohorts showed reduced efficacies with this approach. Alternatively, a switch of drug classes can be explored. Due to the low likeliness of selection of major RAVs against the nucleotide inhibitors this class of drugs can be re-used. Here, one limitation could be the broad use of NS5A inhibitors in first line DAA regimens together with the high likeliness of selection of NS5A RAVs and persistence of these RAVs. Thus, for patients with virologic treatment failure to all oral DAA-based regimens a switch of DAA classes, longer treatment durations, the addition of ribavirin and finally in patients with multiple resistance also a combination with pegylated interferon alfa may be explored in clinical --studies."
Article in Press
The importance of resistance to direct antiviral drugs in HCV infection in clinical practice
Journal of Hepatology Sept 24 2015
J. W. Goethe-University Hospital, Medizinische Klinik 1, Theodor-Stern-Kai 7, 60590
Frankfurt am Main, Germany
Treatment of chronic hepatitis C virus (HCV) infection with direct acting antiviral agents (DAA) is associated with high rates of sustained virologic response. Remaining factors associated with treatment failure include advanced stages of liver fibrosis, response to previous antiviral therapy and viral factors such as baseline viral load and suboptimal interaction of the DAA with the target based on viral variants. Heterogeneity within NS3, NS5A and NS5B areas interacting with DAAs exist between HCV geno- and subtypes as well as HCV isolates of the same geno- and subtype and amino acid polymorphisms associated with suboptimal efficacy of DAAs are termed resistance associated variants (RAVs). RAVs may be associated with virologic treatment failure. However, virologic treatment failure typically occurs only if other negative predictive host or viral factors are present at the same time, susceptibility to additional antiviral agents is reduced or duration of treatment is suboptimal. In this review geno- and phenotypic resistance testing as well as clinical data on the importance of RAVs for conventional triple therapies with sofosbuvir, simeprevir and daclatasvir and available interferon-free DAA combinations are discussed.
Before discovery in 1989 transmission of hepatitis C virus (HCV) mainly through medical procedures (transfusion of blood products, vaccination etc.) but also by other means like drug abuse, tattooing or sexual intercourse has led to a significant prevalence of chronic infection between 0.4-3% in the majority of countries and to a huge epidemic with a prevalence between 6-10% in some countries like Pakistan, Egypt and Mongolia [, ]. Based on a large data base the global prevalence of viraemic HCV infection recently was re-estimated between 64 and 103 million patients . In the absence of cofactors like heavy alcohol consumption or other liver diseases chronic hepatitis C is characterized by slow progression leading to liver cirrhosis and its sequelae typically after 20-30 years only . As the majority of infections took place before the discovery of the virus now in many countries an increasing number of patients who present with late stage disease is recognized . Thus, highly effective antiviral therapy is urgently required for many patients. Treatment with (pegylated) interferon alfa and ribavirin is relatively ineffective with a viral eradication rate of approx. 50% only. Furthermore, interferon alfa-based antiviral therapy is associated with a number of severe adverse events, which excluded up to 50% of patients upfront from antiviral therapy . Increasing knowledge on HCV molecular biology and establishment of robust HCV replication and cell culture systems since 1999 facilitated the development of direct-acting antiviral agents (DAAs) for treatment of HCV infection. Currently, 4 classes of DAAs targeting three HCV proteins, which are discussed in detail for their structural and functional properties in a recent review, are approved for treatment in many countries . Different to other chronic viral infections like HBV and HIV the aim of therapy in HCV infection is viral eradication which is observed already spontaneously in a significant proportion of patients (20-40%) . Patients with chronic HCV infection are characterized by different levels of immunological failure to spontaneously eradicate the virus. The underlying mechanisms are not completely understood but several surrogate host parameters like IFNL4 genotype (formerly known as IL28B), IP-10 levels and others are associated with spontaneous and treatment induced viral eradication [, ]. The individual variability of efficacy to control HCV infection also explains the huge variation of efforts required to eradicate the virus in single patients. While for some patients weak antivirals and short term treatments are sufficient others require combination therapies with several highly active antivirals for longer durations. In addition the HCV genotype and stage of liver fibrosis have been established as independent parameters associated with virologic treatment response (Figure 1). Based on the principal possibility and the individual differences of susceptibility to viral eradication in HCV infection described above it is clear that also resistance associated variants (RAVs) to DAAs will play a different role in clinical practice as in HIV and HBV infection. On the one hand in HCV infected patients with pre-existence and selection of RAVs during treatment viral eradication has been reported for example with the NS3 protease inhibitor telaprevir in combination with PEG/R and short term treatment or even as monotherapy [, ]. On the other hand viral breakthrough or relapse was observed in patients without detectable pre-existence of RAVs and treatment with highly active DAA combination therapies . In this review the importance of RAVs will be discussed in the context of influencing parameters like prevalence of RAVs, their frequencies within the HCV quasispecies, antiviral activities and barrier to resistance of DAAs, duration of antiviral therapy and host parameters like IFNL4 genotype, stage of liver fibrosis and gender.
Genotypic resistance analysis
Genotypic resistance analysis is based on DNA sequencing technologies. Below the detection limits of available assays it is unknown whether certain variants exist and persist although due to the error prone replication of HCV it is likely that all possible single and double variants are continuously generated . In the following the term pre-existence of certain variants always is related to detectability by a certain sequencing method.
In the majority of cases population sequencing of the HCV genome with a sensitivity to detect viral variants at an approx. 20% frequency within the HCV quasispecies is performed to assess the presence or absence of RAVs to a certain DAA. Clonal and deep sequencing technologies allow reliable detection of viral variants with a frequency down to 0.5-1% . However, currently it is unclear which frequency cut-off is clinically relevant for prediction of virologic treatment failure. Generally, due to the high heterogeneity of HCV isolates and methodological restrictions all sequencing technologies may miss detection of RAVs due to non-amplification based on HCV RNA secondary structures, primer selection and low frequencies within HCV quasispecies.
Mainly dependent on the replicative fitness associated with the presence of RAVs the likeliness of pre-existence of HCV isolates which confer resistance to certain DAAs is highly variable. Many RAVs to NS3 protease inhibitors are associated with a replicative impairment which explains a relatively low likeliness of detectable pre-existence as well as a relatively rapid replacement by wild-type virus after stopping NS3 PI-containing antiviral therapy. For the majority of NS3 protease inhibitors the frequency of natural occurrence of single RAVs in HCV genotype 1 infected patients is between 0.1% and 3.1% (Table 1). An exception is the Q80K variant with no loss of replicative fitness in many patients and a relative high probability of pre-existence. The Q80K variant is associated with resistance to some approved NS3 protease inhibitors only (simeprevir, asunaprevir, paritaprevir). Interestingly, this variant is almost exclusively detected in HCV subtype 1a isolates with important regional differences based on the presence of clade 1 of HCV subtype 1a in the population. While in North America up to 48% of patients with HCV genotype 1a infection harbor the Q80K variant the frequencies in South America and Europe are around 9% and 19%, respectively, with significant variation in the different countries of these continents .
For other HCV genotypes only very few data have been published on the natural occurrence of viral variants. However, reduced antiviral activity of NS3 protease inhibitors in different HCV genotypes seems to be in part associated with the natural occurrence of RAVs which present the inherent sequence in these genotypes. For example S122R associated with medium level resistance to simeprevir is present as natural variant in HCV genotype 2 isolates and D168Q which confers high level resistance to simeprevir was observed in all HCV genotype 3 infected patients [, ]. Furthermore, Q80K seems to be the natural variant in genotype 5 isolates which may explain again reduced susceptibility to simeprevir. In contrast no variants known to be associated with resistance to NS3 protease inhibitors have been observed in genotype 4 and 6 isolates and here high antiviral activities of simeprevir in clinical studies have been observed [, ]. However, the absence of known RAVs for a certain NS3 protease inhibitor does not guarantee a high antiviral activity. For example telaprevir had no efficacy in HCV genotype 3 infected patients in a clinical study although pre-existing variants known to confer resistance to telaprevir have not been observed . Here, obviously so far unknown variants obviously impair proper binding of the inhibitor.
Median times to loss of detectability of RAVs by population sequencing have been estimated for different NS3 protease inhibitors. For boceprevir, telaprevir and simeprevir in genotype 1a (1b) infected patients half-life of 14.0 (12.5), 10.6 (0.9) and 8.3 (5.5) months have been reported [, , ]. For paritaprevir analysis is restricted to HCV genotype 1a isolates. Here, NS3 RAVs have been observed in 46% and 9% of patients after 24 and 48 weeks of follow-up, respectively . Few studies investigated long-term follow-up persistence by clonal or deep-sequencing analysis. Here, only in single patients NS3 RAVs have been observed after a follow-up of 4-5 years [, ].
Overall, in comparison to the majority of NS3 RAVs, variants conferring resistance to NS5A inhibitors generally are more frequently detected as natural variants in HCV genotype 1 infected patients not exposed to DAA-based antiviral therapies. For single RAVs the rate of natural occurrence was estimated between 0.3% and 3.5% in different studies by population sequencing (Table 1). However, there are two exceptions for HCV genotype 1b isolates. L31M conferring low to medium level resistance to daclatasvir and ledipasvir was observed in 2.1-6.3% of patients and the most frequently detected RAV was Y93H with 3.8-14.1% prevalence in HCV genotype 1b isolates. This variant confers medium to high level resistance to all three approved NS5A inhibitors. If more sensitive methods are applied also the frequency of pre-existent RAVs is increasing. Analysis of more than 2000 patients enrolled in a large phase 2/3 study program by deep sequencing revealed NS5A RAVs to ledipasvir in 15.7% and 16.4% of HCV genotype 1a and 1b infected patients, respectively .
Limited data are available on the geographical distribution of naturally occurring RAVs. While no major differences have been observed for the majority of RAVs Y93H seems to be more frequently detected in European HCV genotype 1b infected patients (15.0%) in comparison to the US (9.3%) .
For other HCV genotypes only very limited data are available. Generally, in part different variants in comparison to HCV genotype 1 have been attributed for resistance in other HCV genotypes. Moreover, several of these variants were detected at high frequencies in untreated patients (Table 1). Interestingly, in contrast to NS3 protease RAVs, available data suggest that long term persistence of RAVs to NS5A inhibitors is more likely as viral fitness seems not to be impaired. In clinical studies persistence of NS5A RAVs over 1-2 years after treatment failure in over 85% of patients has been reported [, , ]. The reasons for the unexpected low natural frequency of RAVs despite a lack of impaired viral fitness are unknown but may include subtle differences, initial immunological selection or other so far unknown mechanisms.
Currently, only one non-nucleoside inhibitor of the NS5B polymerase binding to the palm I site is approved for treatment of HCV infection. Pre-existence of RAVs to dasabuvir was reported in 0.2%-3.1% of cases with HCV genotype 1 infection (Table 1). An exception is C316N which occurs only in HCV genotype 1b isolates and was observed at a frequency of 10.9 to 35.6% as naturally occurring variant. In addition, also S556G was detected more frequently in genotype 1b in comparison to 1a infected patients (7.0-25% versus 3-6%). Both variants are associated with low level resistance to dasabuvir. Interestingly, HCV genotype 1b isolates harboring C316N (32% versus 5%) and S556G (25% versus 6%) were more frequently observed in Europe in comparison to the United States .
Again, data for other genotypes are limited. However, for HCV genotypes 2, 3, 4 and 5 S556G is the natural variant with a prevalence of 97-100%. Moreover, also naturally occurring variants at other positions within the NS5B polymerase (i.e. M289I/L,C316N) were observed and this together with the S556G variant most likely explains the lack of antiviral activity of dasabuvir in non HCV genotype 1 infected patients .
Data on viral fitness of variants associated with resistance to dasabuvir and the potential of persistence are sparse. However, preliminary data suggest that at least some RAVs (i.e. M414T, S556G) may tend to persist during long term follow-up for at least one year after treatment failure. Overall, after 24 and 48 weeks in 75% and 57% of patients, respectively, RAVs to dasabuvir were still detectable. Interestingly, the persistence rate of NS5B RAVs which occur together with NS5A RAVs seems to be higher in comparison to isolated NS5B RAVs .
Finally, the natural presence of a RAV to the nucleotide analogue sofosbuvir detected by in vitro selection was not reported so far (Table 1). This is explained by a marked impairment of the replicative fitness of HCV isolates containing the S282T variant. The loss of replicative fitness of the S282T variant also explains that so far no viral break-through in patients on treatment with sofosbuvir was observed. In single patients with relapse after the end-of-treatment and detection of the S282T variant reversion to wild-type virus was reported during few weeks of follow-up [, ]. This seems to be especially the case in patients with repeated and long exposure to sofosbuvir after failure to a previous sofosbuvir-containing regimen . Whether also other variants with higher replicative fitness and higher natural occurrence (L159F, C316N, and V321A) are associated with resistance to sofosbuvir needs further investigation . In patients with relapse after SOF-based regimens L159F and V321A have been observed at increasing frequencies although in replicon based assays no resistance to sofosbuvir could be observed .
Importance of persistence of RAVs
Although it is likely that in all patients with virologic treatment failure RAVs have been selected, in a significant rate of patients with treatment failure no RAVs have been detected by sequence analysis. In patients with virologic break-through during treatment almost always RAVs have been observed while in relapse patients the detection rate of RAVs varies between 53% and 91% dependent on the duration of treatment, the DAA class and regimen [, , , , ]. Un-detectability of RAVs is explained most likely by the sensitivity of sequencing technology applied, potential rapid reversion to wild-type between end-of-treatment and the day of blood sampling for sequence analysis and a very low frequency of isolates containing RAVs within HCV quasispecies. Moreover, in patients with short duration of DAA-based antiviral therapies wild type virus may not be completely eradicated yet which also explains relapse with a predominantly wild type variant.
Long-term persistence of RAVs selected during DAA-based antiviral therapies and subsequent virologic treatment failure is known for HIV and HBV infection which both have the potential of archiving RAVs based on their replication with stable DNA intermediates. As for HCV, some RAVs are associated with high replicative fitness and frequently additional, compensatory mutations have been observed responsible for increased replicative fitness. Due to impaired replicative fitness the frequency of many isolates with RAVs within the quasispecies rapidly decline to very low levels undetectable by population sequencing after stopping DAA treatment. However, despite low frequencies within the HIV quasispecies rapid re-selection of isolates with RAVs upon re-treatment with the same drug or a drug of the same group was observed . Deep sequence analysis together with bioinformatical assessment of nucleotide sequence backbone polymorphisms showed re-selection of persistent isolates which have been archived rather than de novo selection of identical RAVs . For HCV no stable DNA genome forms exist, viral replication showed a rapid turnover (1010 to 1012 virions per day) with a short half-life of virions (2-5 hours) and the error rate during replication with 10-3 to 10-5 mutations per nucleotide per genomic replication was estimated very high in comparison to HIV and HBV . Few studies exist with re-exposure of patients to the same DAA and results are contradictory. While in some studies no evidence for long term persistence and re-selection of isolates with RAVs during re-treatment with the same drug was observed in others indirect evidence pointed to the possibility of persistence and re-selection [, , ]. Given the large number of patients who currently receive all oral DAA therapies and the extension from well-defined HCV geno- and subtypes in clinical studies to a broad heterogeneity of existing HCV subtypes and isolates even with a SVR rate of 90-95% the question on persistence, transmission and reselection of isolates with RAVs will become more important in the near future .
Most recently, a study on re-treatment of patients with failure to sofosbuvir plus ledipasvir with and without ribavirin was presented. After 24 weeks of sofosbuvir plus ledipasvir as retreatment the SVR rate was 80% in patients with 8 weeks and only 46% in patients with 12 weeks initial treatment. In patients with detectable NS5A RAVs at baseline the overall SVR rate was 60% which explains the importance of persistent NS5A RAVs for selection of effective re-treatment options .
Phenotypic resistance analysis
RAVs are typically associated with a change of the shape of the binding or interaction site of DAAs to HCV target proteins. Due to different locations within the sites of interaction and due to different chemical structures of DAAs targeting the same site on the same HCV protein not all RAVs confer the same level of resistance. To assess the level of resistance RAVs typically are introduced as single point mutations into the backbone of the HCV genome within an existing cell-culture/replicon or enzyme based assay. Isolates harboring these RAVs are then challenged by DAAs at increasing concentrations and fold changes based on EC/IC50 and EC/IC90 values are determined for inhibition of replication or enzyme activity in comparison to wild-type virus. Table 2 summarizes EC50/fold change values of different RAVs and DAAs. As for genotypic resistance analysis the majority of values exist for HCV genotype 1 and on genotype 1a and/or genotype 1b backbone isolates only.
Only few assays have been established to assess the phenotypic sensitivity to a certain DAA based on the entire HCV quasispecies of a given patient and currently it is unclear whether these assays are required to determine suitable DAAs for patients with resistance to multiple DAAs [, ].
For DAA-based antiviral therapy the level of resistance of a certain RAV is not necessarily directly associated with treatment failure. For conventional triple therapy with simeprevir, pegylated interferon alfa and ribavirin for example Q80K as a low level NS3 protease inhibitor RAV significantly influences virologic treatment outcome while for the combination therapy of simeprevir with the nucleotide inhibitor sofosbuvir the importance of the Q80K variant seems to be more limited [, , ]. In fact, for DAA combination therapies not only pre-existing RAVs but also the presence of other predictors of virologic treatment response seems to be of importance. For the combination therapy of sofosbuvir and the NS5A inhibitor ledipasvir for example only high level resistance NS5A variants together with other features seem to have an impact on response to antiviral therapy (see below) .
Antiviral activities and barrier to resistance of single DAAs
Currently, five NS3 protease inhibitors (boceprevir, telaprevir, simeprevir, asunaprevir, paritaprevir), three NS5A inhibitors (daclatasvir, ledipasvir, ombitasvir), one non-nucleoside (dasabuvir) and one nucleotide NS5B inhibitor (sofosbuvir) are approved globally for treatment of chronic hepatitis C. Additional NS3 protease- (vaniprevir, grazoprevir, sovaprevir, ABT-493), NS5A- (elbasvir, velpatasvir/GS-5816, odalasvir/ACH-3102, ABT-530; MK-8408), non-nucleoside (beclabuvir, GS-9669) and nucleos(t)ide NS5B inhibitors (IDX-21437, ACH-3422) are in phase 2-3 clinical development [, ]. Some of them are termed second generation DAAs (grazoprevir, ABT-493, velpatasvir/GS-5816, elbasvir, ABT-530, MK-8408) as they aim to overcome restrictions in terms of resistance profile of the drug class as well concerning the coverage of HCV geno- and subtypes.
While for all DAAs short term monotherapy studies have been performed in HCV genotype 1 infected patients for assessment of antiviral activities and barrier to resistance for other HCV genotypes and subtypes antiviral activities are heterogeneous and often no clinical data are available. However, inhibitory concentrations for suppression of replication of many HCV geno- and subtypes have been assessed in vitro and these data may be helpful for selection of active DAAs in patients with failure to multiple DAA-based therapies.
Genetic barrier to resistance
It has been recognized already during the clinical development of the first DAAs (boceprevir and telaprevir) that despite an identical amino acid at a certain position within the NS3 protease of HCV subtype 1a and 1b the likeliness to observe a treatment-induced mutation was highly different between these HCV subtypes [, ]. This is explained by different nucleotide triplets encoding the same amino acid. While in HCV subtype 1b isolates at codon position 155 for generation of R155K two nucleotide changes are required in HCV subtype 1a isolates one exchange is sufficient. As a consequence distinct resistance patterns are observed in HCV subtype 1a and 1b infected patients after failure to a protease inhibitor based antiviral therapy [, ].
Similar differences have been observed also for the generation of other RAVs in other HCV genes [, ].
Beside the number of nucleotide changes required for an amino acid exchange as definition of the genetic barrier to resistance also the type of exchange seems to be of importance in HCV infection. The HCV NS5B RNA polymerase was shown to favor the generation of nucleotide transitions in comparison to transversions . This may explain that some RAVs are rarely observed at all or generated only after a longer DAA exposure (for example S282T within NS5B or L31M within NS5A) [, ].
NS3 protease inhibitors
The antiviral activity of the approved dosing of boceprevir (3x800mg/die) applied as monotherapy was never assessed in a clinical trial. However, mean maximum decline of HCV RNA concentration in patients infected with HCV genotype 1 was relatively low with half of the approved daily dose (2.1 log IU/ml) .
For all other NS3 protease inhibitors high antiviral activities (3.1-4.6 log IU/ml decline) have been described in phase 1 monotherapy trials in HCV genotype 1 infected patients (Table 3). For other HCV genotypes in vitro antiviral activities are heterogeneous and only few clinical studies have been performed (Table 3). All NS3 protease inhibitors bind to the active site of the enzyme, which demonstrate a significant variation between HCV geno- and subtypes explaining heterogeneous affinities of the different PIs and the challenge to design a pan-genotypic NS3 protease inhibitor. While for paritaprevir no clinical monotherapy studies outside of HCV genotype 1 have been performed for simeprevir detailed analysis of antiviral activities and binding profiles to the different HCV geno- and subtypes is available [, ]. Here, high antiviral activities in HCV genotype 1, 4 and 6 infected patients could have been demonstrated, while for HCV genotypes 2, 3 and 5 mainly due to the natural occurrence of RAVs low to medium mean maximum viral declines in monotherapy studies were described (Table 3). For grazoprevir the antiviral activity in HCV genotype 3 infected patients was assessed in a 7 days monotherapy study . Here, for the currently used dose of 100mg once daily a moderate antiviral activity was observed (Table 3). Higher doses of grazoprevir were discontinued from further development due elevated liver enzymes in a significant proportion of patients.
ndirect evidence for antiviral activities of NS3 protease inhibitors against different HCV genotypes and subtypes can be taken from in vitro studies. Here, data from the developing pharmaceutical companies and independent research groups are available and summarized in Table 4. All currently approved NS3 protease inhibitors were optimized for binding to HCV genotype 1 and all have substantially reduced potency against HCV genotype 3. The three newer NS3 protease inhibitors, simeprevir, paritaprevir, and asunaprevir showed high antiviral activities also against HCV genotype 4a isolates in vitro. For the other HCV geno- and subtypes in vitro activities are heterogeneous (Table 4). The second generation protease inhibitor grazoprevir in principle should be active against all HCV genotypes but data on in vitro studies are only available for genotype 1, 2 and 3 isolates .
The barrier to resistance is low for all currently approved NS3 protease inhibitors with broad cross resistance between the different substances. Sequence analysis early during therapy showed rapid selection of isolates harboring RAVs and in the majority of patients after initial viral decline a viral break-through or plateau phase was observed during monotherapy studies for 3-14 days [, , , , , , , ]. A higher barrier of resistance was observed for grazoprevir due to a higher antiviral activity against typical NS3 RAVs although variants at the same positions as for first generation NS3 protease inhibitors were observed in patients with treatment failure (R155, A156, D168) .
NS5A inhibitors typically have broader antiviral activity against different HCV genotypes which is explained by a more conserved interaction site within the NS5A protein [, ]. Unfortunately, for all three approved NS5A inhibitors phase 1 monotherapy studies have been performed only in HCV genotype 1 infected patients (Table 3). However, data from in vitro and clinical studies with DAA combination therapies showed antiviral activities also for other HCV genotypes. Moreover, indirect evidence of an antiviral activity of a specific DAA against a certain HCV genotype administered in combination with other drugs results from the detectability of RAVs in patients with virologic treatment failure. The antiviral activity of the first approved NS5A inhibitor daclatasvir is broad with little differences between HCV genotypes in replicon studies . For HCV genotype 3 isolates an approx. 10-fold lower antiviral activity was observed (Table 4). Also for ledipasvir a broad genotypic coverage was observed in vitro . However, for HCV genotype 2 and 3 isolates approx. 1000-fold higher concentrations of ledipasvir were required for inhibition of replication in vitro (EC50), which indicates an insufficient antiviral activity of ledipasvir for these HCV genotypes (Table 4). Ombitasvir is characterized by a coverage of HCV genotypes 1 to 5 with high antiviral activities . However, a 10 to 100-fold lower activity against HCV genotype 6 was observed and ombitasvir is available only in a fixed dosed combination with the protease inhibitor paritaprevir which shows high antiviral activities in HCV genotype 1 and 4 isolates only (Table 4). For HCV genotypes 2 to 6 a large number of HCV subtypes exist in clinical practice with significant amino acid sequence variation and little is known about antiviral activities of NS5A inhibitors against the different HCV subtypes. An example for potentially important variations is ledipasvir. Here, a 1000-fold difference in the antiviral activity was described between HCV genotype 6a and 6e isolates in vitro (Table 4).
Also for NS5A inhibitors a low barrier to resistance and a broad cross resistance between the different approved substances with rapid selection of RAVs mainly at positions 28, 30, 31, 58 and 93 has been observed in short term monotherapy studies [, , , ].
For the second generation NS5A inhibitor elbasvir high antiviral activities in monotherapy studies in HCV genotype 1 and 3 infected patients have been demonstrated and a higher barrier to resistance in comparison to other NS5A inhibitors was reported from in vitro studies [, ].
Non-nucleoside NS5B inhibitors
Dasabuvir currently is the only approved non-nucleoside inhibitor of the NS5B polymerase with binding to the palm I site. During a clinical study with monotherapy of different doses of dasabuvir a mean maximum decline of 1.08 HCV RNA log IU/ml was observed indicating a low to medium antiviral activity . In vitro, only data for HCV subtypes 1a and 1b are available (Table 4). However, evidence from biochemical assays indicates that dasabuvir is inactive towards HCV polymerases from HCV genotypes 2, 3 and 4 . The barrier of dasabuvir to resistance is considered to be low although no sequencing studies in patients on monotherapy have been presented. However, a number of RAVs have been selected in HCV replicon studies as well as in patients with failure to combination therapies with other DAAs [, , ].
Nucleos(t)ide NS5B inhibitors
Nucleos(t)ide inhibitors interact with the active site of the HCV NS5B polymerase which is highly conserved between all HCV geno- and subtypes. Therefore, a broad coverage of all HCV isolates is likely for this class of DAAs. For the nucleotide polymerase inhibitor sofosbuvir clinical data are available with monotherapy for 2 to 12 weeks in HCV genotype 1, 2 and 3 infected patients (Table 3) [, ]. In addition, in vitro studies showed an almost equal antiviral activity against HCV genotypes 1 to 6 (Table 4) . Moreover, a high barrier to resistance was observed in clinical studies with no viral breakthrough during mono- or combination therapies with other DAAs [, , , , ]. With the exception of few cases even in patients with virologic treatment failure no major RAVs have been detected .
Viral resistance and approved combination therapies
Protease inhibitor plus pegylated interferon and ribavirin
For conventional triple therapies with boceprevir or telaprevir in combination with ribavirin the probability of pre-existing RAVs is low and no correlation with treatment response was observed in treatment naïve patients (Figure 2) [, , , , , ]. Hence, no baseline testing was recommended. However, for patients with poor interferon-responsiveness with and without other negative predictors a correlation of pre-existing RAVs with treatment response has been described [, , , , ].
In contrast, for conventional triple therapy with simeprevir baseline resistance testing is recommended by international guidelines. Here, RAVs at position 43, 122, 155, 168 and the Q80R variant were observed in a low rate of patients only (1.3%) and thus no clear correlation with virologic treatment outcome was possible. However, one additional variant (Q80K) causing medium level resistance to simeprevir (approx. 8-fold change in EC50) is associated with a much higher natural prevalence [, ]. The Q80K variant was almost exclusively detected in HCV subtype 1a isolates with a global prevalence from phase 2/3 studies of approx. 30% (see above) . SVR rates in treatment-naïve HCV genotype 1a infected patients with and without the Q80K variant were 58% versus and 84% (Figure 2). The SVR rate of 58% in patients with Q80K who received simeprevir-based triple therapy was statistically not different from 52% in patients who received pegylated interferon alfa and ribavirin alone [, ]. Similar results were obtained in previous null-responders and partial responders and re-treatment with simeprevir-based triple therapy . Therefore, administration of simeprevir-based triple therapy is not recommended in patients with detectable Q80K variant at baseline.
NS5A inhibitor plus pegylated interferon alfa and ribavirin
Conventional triple therapy with the NS5A inhibitor daclatasvir, pegylated interferon alfa and ribavirin was investigated in clinical studies in HCV genotype 1-4 infected patients. Overall, higher SVR rates were observed in comparison to dual combination therapy with pegylated interferon alfa, ribavirin and placebo. For genotype 1 infected patients an overall SVR rate of 59-60% was achieved for daclatasvir-based triple therapy with significant differences between HCV subtypes which reflects the overall lower antiviral activity of daclatasvir in HCV subtype 1a versus 1b isolates (55-57% for HCV subtype 1a versus 76-77% in HCV subtype 1b). In 22 of 365 patients (6%) baseline NS5A inhibitor RAVs (L31M/V and/or Y93H/N/S) were observed. SVR rates in patients with pre-existent RAVs were 33% for HCV subtype 1a and 80% for HCV subtype 1b . Given the low frequency and similar SVR rates in patients with and without baseline variants no clear impact of pre-existent RAVs for virologic treatment response could be demonstrated (Figure 2). For HCV genotype 2, 3, and 4 only small cohorts of patients infected with these HCV genotypes were enrolled in clinical studies (n=6-12 per treatment group) not allowing definite conclusions on the importance of baseline RAVs [, ]. However, generally similar viral variants associated with resistance were observed in comparison with HCV genotype 1. Furthermore, in HCV genotype 3 infected patients pre-existence of Y93H or A30K may increase the risk of subsequent relapse. Here, virologic treatment failure was observed in 50% (4/8) of patients with Y93H or A30K at baseline in comparison to 16% (8/43) without pre-existence of these variants .
Sofosbuvir plus pegylated interferon alfa and ribavirin
Sofosbuvir in combination with pegylated interferon alfa and ribavirin led to an SVR rate of 89% in a large phase 3 study on treatment naïve HCV genotype 1 infected patients. Interestingly, a substantial difference was observed between HCV genotype 1 subtypes with an SVR rate of 92% in HCV subtype 1a versus 82% in HCV subtype 1b . This was unexpected as for NS3 protease- and NS5A-inhibitor based conventional triple therapies always lower SVR rates in HCV subtype 1a infected patients were observed which is attributed to the lower antiviral activity of these drug classes in HCV subtype 1a. S282T as a known sofosbuvir resistance mutation from in vitro studies was not observed in any HCV isolate before initiation of antiviral therapy or at treatment failure . This is explained by the low replicative fitness of this variant. Furthermore, no other variants with a clear reduction of susceptibility to sofosbuvir in cell-culture assays are known. A potential explanation for the difference between HCV subtype 1a and 1b comes from a variant at position 316 within the HCV NS5B polymerase which is highly conserved in subtype 1a (C316) but polymorphic in subtype 1b isolates (C316N/H). In a structural assessment it was predicted that C316N/H may alter the ability of sofosbuvir to interact with the active site of the NS5B polymerase . Dependent on geographical differences, between 10% and 30% of HCV genotype 1b isolates harbor C316N as natural variant which may explain the observed differences between subtype 1a and 1b infected patients and treatment with sofosbuvir-based conventional triple therapy [, ]. However, the numbers of HCV genotype 1b infected patients in the approval trial was too small to draw definite conclusions and further studies are needed.
Sofosbuvir plus ribavirin
Due to the limited antiviral efficacy only few HCV genotype 1 infected patients have been enrolled in studies for evaluation of the nucleotide polymerase inhibitor sofosbuvir in combination with ribavirin [, ]. In contrast, for HCV genotype 2 infected patients very high SVR rates were achieved with this combination leaving only few patients with virologic failure for resistance analysis [, ]. Interestingly, in some of the few patients with HCV genotype 2 infection failure to sofosbuvir plus ribavirin may be explained by the presence of a viral chimera. Here, the structural HCV genes are encoded by HCV genotype 2 and the non-structural genes are coming from an HCV genotype 1 isolate which may explain reduced susceptibility to sofosbuvir plus ribavirin .
However, for HCV genotype 3 a relative large number of treatment failure patients could be evaluated for the potential importance of viral resistance. Based on refined selection criteria for variants potentially associated with resistance, L159F, V321A, and S282R have been observed in patients with virologic relapse. Although these variants were not associated with reduced susceptibility against sofosbuvir in cell culture, affection of the interaction of the inhibitor with the NS5B polymerase was predicted by structural bioinformatics . Most likely, in addition to baseline RAVs other factors associated with virologic treatment response like stage of fibrosis, IFNL4 genotype or previous responsiveness to interferon alfa-based antiviral therapies are of importance and large studies in HCV genotype 2 and 3 infected patients for which sofosbuvir plus ribavirin is a standard treatment are required to understand the importance of viral resistance with this regimen.
NS3 protease inhibitor and NS5A inhibitor
Asunaprevir plus Daclatasvir
The combination of the NS3 protease inhibitor asunaprevir together with the NS5A inhibitor daclatasvir was shown to be highly effective in HCV genotype 1b infected patients while for HCV genotype 1a a high rate of viral break-through was observed in the initial study . Subsequently combination therapies with these two targets were evaluated in larger studies and currently asunaprevir plus daclatasvir is approved as standard treatment in HCV genotype 1b infected patients in Japan. In the pivotal study treatment of HCV genotype 1b infected patients (n=643) with asunaprevir plus daclatasvir for 24 weeks provided SVR rates between 82 and 91% . In 13% of patients pre-existing variants associated with resistance to asunaprevir and or daclatasvir (positions D168 within NS3 and L31, Y93 within NS5A) were observed. The SVR rate of these patients was significantly reduced to 39% while in 92% of patients without baseline RAVs treatment-induced viral eradication was achieved (Figure 3).
Simeprevir plus Daclatasvir
In addition, clinical study data are available for the combination of the NS3 protease inhibitor simeprevir plus daclatasvir. Here, with treatment of a relative small number of HCV genotype 1b infected patients (n=147) for 12-24 weeks SVR rates between 65 and 95% were obtained . Unfortunately, no data on the importance of baseline RAVs have been presented so far for this study but a significant influence on the rates of viral break-through and relapse is likely. Furthermore, also a small number of HCV genotype 1a infected patients were explored in this study. While 67% of treatment-naïve patients achieved SVR, 7 out of 9 previous null-responders experienced viral break-through. Beside the lower antiviral activity of this regimen in HCV genotype 1a infected patients it is also likely that baseline RAVs are of importance for prediction of virologic treatment response.
Grazoprevir plus Elbasvir
Recently, phase 3 study results have been reported also for combination of grazoprevir and elbasvir and treatment duration of 12 weeks. RAVs in the NS3 protease had no impact on SVR. Relevant RAVs to the NS5A inhibitor elbasvir were observed in 10% and 13% at baseline in HCV genotype 1a and 1b infected patients. For HCV genotype 1b infected patients the presence of baseline NS5A RAVs had no importance for treatment response. However, in treatment-naïve HCV genotype 1a infected patients SVR rates with and without baseline RAVs were 22% and 98% . Similar results were obtained in treatment-experienced patients. Here, independent of the duration (12 vs 16 weeks) and the addition of ribavirin in HCV genotype 1a infected patients with and without baseline RAVs SVR rates of 52% and 99% were reported . Interestingly, for treatment-naïve and -experienced patients only RAVs with a >5-fold resistance level were of importance (Figure 3) [, ]. For patients with HCV genotype 4 and 6 infection number were too small for valid resistance analysis.
Taken together, the barrier to resistance of first wave protease inhibitors like asunaprevir or simeprevir in combination with an NS5A inhibitor is relatively low. Based on a relative high frequency of patients with pre-existing NS3 and / or NS5A RAVs a significant impact on virologic treatment response is obvious. Thus, HCV genotype 1a infected patients should not be treated with this regimen and baseline resistance testing should be mandatory for this DAA combination in genotype 1b infected patients. For treatment with grazoprevir plus elbasvir due to a higher barrier to resistance treatment is highly effective in genotype 1b infected patients without importance of baseline RAVs. However, baseline resistance testing seems to be required in HCV genotype 1a infected patients.
Interestingly, very recent data suggested an association of a major NS5A RAV (Y93H) with beneficial IFNL4 genotypes [, ]. If confirmed in additional studies, this would also explain the unexpected inverse correlation of SVR rates with IFNL4-rs12979860 genotype in the dual combination asunaprevir plus daclatasvir combination phase 3 study. Here, SVR rates in patients with the beneficial CC genotype were 76-89% versus 86-96% in those with the TT genotype .
NS3 protease inhibitor plus nucleos(t)ide NS5B inhibitor
The combination of simeprevir and sofosbuvir with and without the additional application of ribavirin for 12 or 24 weeks was initially investigated in a small prospective randomized phase 2 study . Here, in null-responder patients with early liver fibrosis stages as well as in treatment naïve patients and previous null-responders with advanced fibrosis or liver cirrhosis high SVR rates were reported (92%). Pre-existing RAVs are very rarely observed in HCV genotype 1b infected patients while for HCV genotype 1a the rate of a naturally occurring Q80K variant varies between 10 and 50% with important regional differences (see above) [, ]. In line with this, virologic treatment failure was not observed in HCV genotype 1b infected patients while the pre-existence of Q80K in 4 out of 6 HCV genotype 1a infected patients with virologic relapse point to an importance for treatment failure. For subsequent phase 3 studies the combination of simeprevir plus sofosbuvir without ribavirin was explored. Here in HCV genotype 1 infected patients without liver cirrhosis the overall SVR rate was 97% and 83% for 12 and 8 weeks treatment duration. While for a duration of 12 weeks the presence of Q80K at baseline in genotype 1a infected patients had no impact (96% versus 97% SVR) in the group with 8 weeks treatment SVR rates were 73% and 84% with and without Q80K at baseline (Figure 4) . Similar results were obtained in HCV genotype 1a infected patients with cirrhosis and 12 weeks treatment duration (74% versus 92% SVR with and without Q80K) (Figure 4) .
NS5A inhibitor plus nucleos(t)ide inhibitor
Daclatasvir plus Sofosbuvir
In the phase 2 study program only a limited number of HCV genotype 1 infected patients (n=167) was treated with heterogeneous regimens of sofosbuvir and daclatasvir with and without ribavirin for 12 (n=82) or 24 weeks (n=85). Furthermore, patients with liver cirrhosis were excluded and after elimination of non-virologic treatment failure patients the SVR rate was 100% . The subsequent phase 3 studies enrolled non-cirrhotic HCV genotype 1-6 infected patients with HIV coinfection and those with liver cirrhosis or after liver transplantation. Overall high SVR rates in HCV genotype 1 infected patients were obtained (82%-98%). Virologic treatment failure mainly was associated with HCV subtype 1a infection in combination with shortened treatment duration to 8 weeks (and suboptimal dosing of daclatasvir) or the presence of cirrhosis. Unfortunately, full data are not yet available on the precise importance of pre-existent resistance for different types of patients and HCV geno- / subtypes. However, the presence of baseline NS5A RAVs also seems to influence the chance of SVR only if additional negative predictors like liver cirrhosis are present (Figure 5).
Ledipasvir plus Sofosbuvir
For the combination of sofosbuvir plus ledipasvir with and without the addition of ribavirin more than 2100 patients were enrolled in different phase 2 and 3 studies [, , , ]. Sequence analysis of the NS5A gene was successfully performed in more than 99% of baseline samples. In addition and different to the standard analysis in other studies where population sequencing was used, in the vast majority of patients (89%) deep sequence analysis was performed. The rate of pre-existing NS5A RAVs in the entire HCV genotype 1 infected population was 17% with a 1% cut-off and 8% with a 20% cut-off within the HCV quasispecies. The SVR rate in patients with and without baseline NS5A resistance was only slightly reduced but due to the large sample size this result was significant (93% versus 97%, respectively). Similar data were obtained when HCV genotype 1a and 1b patients were analyzed separately . Further analysis was performed based on the level of resistance caused by the different RAVs and treatment duration. RAVs with a less than 100-fold shift of EC50ies had no influence on SVR independent of pre-treatment status and treatment duration (8, 12, 24 weeks) (Table 2). However, in patients with highly resistant RAVs (>100 fold shift EC50) and shorter treatment durations a more significant decline of SVR rates was observed. In treatment naïve patients with 8 weeks the SVR rate was 83% in comparison to 95% in patients with highly resistant RAVs and those without baseline RAVs (Figure 5). In a combined analysis of treatment-naïve and -experienced patients a reduced SVR of 87% with 12 weeks treatment was observed compared to 97% in patients without highly resistant baseline RAVs (Figure 5). SVR rates ranged between 84% and 97% in patients with RAVs at position 24, 28, 30, 31 and 93 of the NS5A protein and were similar for different frequencies of RAVs within the HCV quasispecies. In patients with liver cirrhosis also reduced SVR rates have been reported due to the presence of baseline RAVs (Figure 5) .
Concerning sofosbuvir resistant variants as described also in other studies S282T as major in vitro RAV was not observed at baseline in any patient. Furthermore, also no correlation of other variants within the NS5B polymerase including N142T, L159F, S282G, C316N, and L320F with virologic treatment response was observed. The overall prevalence of these RAVs was 2.5% and the SVR rate 100%. Finally, as to be expected RAVs to NS3 protease inhibitors had no influence on treatment response to sofosbuvir plus ledipasvir .
In 76% of patients who failed to achieve SVR NS5A RAVs were observed . In a recent presentation long term persistence of NS5A RAVs has been reported in 86% of patients after failure to different treatment regimens with ledipasvir and 96 weeks of follow-up .
Studies in non-genotype 1 patients
For treatment of patients with sofosbuvir in combination with an NS5A inhibitor like daclatasvir or ledipasvir infected with other HCV genotypes the experience is very limited. Given the high SVR rates obtained only few patients with treatment failure are available for resistance analysis so far [, ]. For sofosbuvir plus daclatasvir relapse was reported in one patient with HCV genotype 3 infection from the phase 2 study. Here, a resistance associated NS5A polymorphism (A30K) was detected at baseline and at the time of relapse . Furthermore, 16 HCV genotype 3 infected patients experienced virologic relapse after 12 weeks sofosbuvir plus daclatasvir in the phase 3 Ally 3 study (n=152) . Here, in 6 patients the resistance variant Y93H was detectable at baseline as well as at treatment failure and in all other patients NS5A RAVs (mainly Y93H and one L31I) emerged at relapse. The overall rate of pre-existing Y93H in the entire cohort of treated patients was 9% (n=13). The SVR rate in patients with Y93H was clearly reduced to 54% in comparison with 89% in the entire cohort. However, for patients with pre-existing Y93H also the additional presence of liver cirrhosis seems to be of importance. For patients without cirrhosis the SVR rate was moderately reduced to 67% (6/9 patients) while patients with cirrhosis and baseline Y93H achieved an SVR in 25% (1/4 patients) only (Figure 6). Variants at positions L31, M28 and A30 very not or rarely observed at baseline and no clear effect on virologic treatment response was visible . Resistance analysis for the combination of sofosbuvir plus ledipasvir has not been published so far.
NS3 protease-, NS5A- and non-nucleoside NS5B inhibitor
Paritaprevir/r plus Ombitasvir plus Dasabuvir
Triple DAA combination therapy with paritaprevir/r, ombitasvir and dasabuvir with and without ribavirin was approved based on a large clinical study program with more than 2500 patients. Baseline resistance analysis based on population sequencing is available from approx. 700 patients only . Here, the prevalence of RAVs within one and two targets was 18% and 0.4% in HCV subtype 1a and 33% and 3% in HCV subtype 1b infected patients, respectively. None of the patients had RAVs within all three targets. As known from other studies within the NS5A protein M28 variants (8% in HCV subtype 1a) and Y93H variants (8% in HCV subtype 1b) were observed most frequently. For RAVs to the non-nucleoside NS5A inhibitor dasabuvir mainly C316N and S556G were observed in genotype 1b (15-17%) and S556G/N/R (3%) in HCV genotype 1a infected patients. For the NS3 protease Q80K is considered a low level resistant variant to paritaprevir and other RAVs were rarely observed at baseline (<1%). Frequencies of SVR in patients with and without baseline RAVs are available from a phase 2 study only (n=391). For the entire group of HCV genotype 1 infected patients after excluding Q80K variants no difference was visible (91 versus 91%). As the vast majority of patients with virologic treatment failure (91%) was infected with HCV subtype 1a further sub-analysis in this subgroup of patients may be of interest. Here, a slightly reduced overall SVR rate in patients with versus those without baseline RAVs in the NS3, NS5A and / or the NS5B genes was reported (87% versus 92%) (Figure 7) . Moreover, in the 8 weeks treatment arm, SVR rates were reduced in patients with baseline Q80K (74% versus 87%), suggesting a relevance of this polymorphism also for this triple DAA regimen (Figure 7). So far no data are available on SVR rates in patients with RAVs to two targets and other treatment predictors like 12 versus 24 weeks of therapy and liver cirrhosis .
In 85% of patients with treatment failure RAVs to at least one target of the triple DAA therapy were observed and 58% of virologic failure patients had RAVs against all three targets. Long term follow-up data have been presented recently. While NS3 RAVs were detectable in only 9% of patients by population sequencing after approx. 1 year RAVs within NS5A and NS5B tend to persist with detectability in 96% and 57%, respectively.
Asunaprevir plus Daclatasvir plus Beclabuvir
Other triple DAA combination therapies without a nucleos(t)ide inhibitor have been investigated in clinical studies. The combination of the protease inhibitor asunaprevir, the NS5A inhibitor daclatasvir and the non-nucleoside thumb 1 inhibitor beclabuvir without the additional application of ribavirin has been administered for 12 weeks in a large phase 3 study in non-cirrhotic patients. Here, SVR rates in HCV genotype 1a infected patients with (n=34) and without (n=195) baseline NS5A RAVs were 74% versus 93% (Figure 7) . Interestingly, in patients with liver cirrhosis and evaluation of the same triple DAA regimen no apparent differences between patients with and without baseline NS5A RAVs were observed which may be explained by the addition of ribavirin in half of the patients (Figure 7) . Thus, a relative low importance and frequency of single RAVs in a regimen targeting 3 HCV proteins for virologic treatment failure may be further reduced by the addition of ribavirin. However, further data especially in difficult-to-treat HCV genotype 1a infected patients are required for understanding of the importance of pre-existing resistance.
Virologic treatment response of patients who failed to respond to triple therapies with boceprevir or telaprevir has been evaluated in clinical studies with DAA combination therapies without NS3 protease inhibitors. For both regimens, daclatasvir plus sofosbuvir as well as ledipasvir plus sofosbuvir high SVR rates have been reported (94-100%) with no influence of NS3 RAVs on virologic treatment response [, ]. Insufficient data are available for determination of the importance of NS3 RAVs for re-treatment with protease inhibitor containing regimens.
Data for patients with failure to the different all oral DAA combination therapies are sparse. In one study 14 HCV genotype 1 infected patients with failure to sofosbuvir plus ribavirin for 24 weeks were retreated with sofosbuvir plus ledipasvir for 12 weeks and all patients achieved SVR . This cohort included 7 patients with advanced fibrosis / cirrhosis. In one patient transiently early after termination of the initial therapy with sofosbuvir and ribavirin the S282T variant was detectable while in all other patients wild type NS5B polymerase sequences were observed . In another study, patients who failed 12 to 24 weeks of sofosbuvir plus ribavirin mainly infected with HCV gentoype 3 were enrolled for retreatment with either 12 weeks pegylated interferon, ribavirin and sofosbuvir or 24 weeks sofosbuvir plus ribavirin. An interim analysis of this study showed a high efficacy of conventional triple therapy (SVR12 92%) while another course of sofosbuvir plus ribavirin combination therapy led to an SVR rate of 63% only .
A single HCV genotype 1 infected patient who failed to respond to sofosbuvir plus ledipasvir for 8 weeks received successful retreatment with the same regimen in combination with ribavirin for 24 weeks despite the presence of highly resistant NS5A RAVs and the S282T variant which causes major resistance to sofosbuvir . However, in 41 patients with failure to 8-12 weeks of sofosbuvir plus ledipasvir with or without ribavirin salvage treatment with sofosbuvir plus ledipasvir for 24 weeks was initiated. The overall SVR rate was 71%. In the subgroup of patients with failure to 12 weeks previous treatment only 5/11 (45%) achieved SVR and the presence of NS5A resistance was associated with treatment outcome .
Further studies investigating salvage therapies for DAA failure patients are ongoing.
Currently, it seems that longer re-treatment with the same class of drugs may be effective in some patients although a reduced antiviral efficacy is obvious. Most likely, additional predictors like the stage of liver fibrosis, pre-treatment status, duration of initial therapy and others will be of importance to select optimal salvage therapies. Given the high rates of the presence of multiple RAVs in DAA treatment failures, the known trend to slightly lower SVR rates in patients with pre-existing RAVs, and a high probability of accumulation of negative predictive factors in DAA-failure patients it is likely that resistance analysis will be useful to determine suitable DAAs for re-treatment options. Also due to the high costs of DAAs it would be important to select the most effective re-treatment option. For the time being guidelines recommend to wait for results of clinical studies or in patients with urgent need of re-treatment if possible to switch the used class of DAAs and to take data of resistance analysis into account .
High efficacy of all-oral DAA combination therapies have been obtained in clinical studies with SVR rates above 90% in the majority of patients with chronic hepatitis C. For the combination of a first generation NS3 protease and NS5A inhibitor with a relative low barrier to resistance the presence of baseline resistance has a great impact with bisection of SVR rates. Here, baseline resistance testing is mandatory. However, for regimens with high antiviral activities and high genetic barrier to resistance based on single DAAs or the combination of different drug classes the presence of baseline resistance leads only to a small reduction of SVR rates. Here, additional predictors of response are of importance. While no general recommendation for baseline resistance testing can be given in subgroups of patients with certain HCV geno- or subtypes, in patients with shortened treatment duration or those with liver cirrhosis resistance testing may be used to select optimal DAA regimens. Future studies have to explore whether due the high costs of DAA regimens in regions with economical restriction baseline resistance testing for initial DAA combination regimens with the aim to avoid virologic treatment failure and the need of retreatment may be cost effective.
A number of second generation DAAs are in development. Here, unmet medical needs mainly include the development of an effective treatment in genotype 3 infected patients and on the same time to ensure coverage of all different HCV geno- and subtypes. Achievement of the latter aim would facilitate treatment of chronic hepatitis C significantly but due to the large number of HCV subtypes represents a major challenge with the risk of improvement of antiviral activity in one subtype but loosing of efficacy in other HCV subtypes.
Currently, effective treatment options in patients with failure to all-oral DAA regimens are not defined yet. Despite high SVR rates around 95% given the large number of patients with chronic hepatitis C overall the number of patients with virologic treatment failure will be high. Based on available data more than 80% of these patients will harbor HCV isolates with resistance to one, two or three DAA classes. Also in patients with failure to DAA-based regimens an accumulation of other negative response factors is likely. One approach is re-treatment with the same DAAs for a longer duration. Indeed in patients who received initial therapies for 24 weeks relapse rates have been extremely low. However, preliminary data in small patient cohorts showed reduced efficacies with this approach. Alternatively, a switch of drug classes can be explored. Due to the low likeliness of selection of major RAVs against the nucleotide inhibitors this class of drugs can be re-used. Here, one limitation could be the broad use of NS5A inhibitors in first line DAA regimens together with the high likeliness of selection of NS5A RAVs and persistence of these RAVs. Thus, for patients with virologic treatment failure to all oral DAA-based regimens a switch of DAA classes, longer treatment durations, the addition of ribavirin and finally in patients with multiple resistance also a combination with pegylated interferon alfa may be explored in clinical studies.