Multiclass HCV Resistance to Direct-acting Antiviral Failure in Real-life Patients Advocates for Tailored Second-line Therapies

Velia C. Di Maio; Valeria Cento; Ilaria Lenci; Marianna Aragri; Piera Rossi; Silvia Barbaliscia; Michela Melis; Gabriella Verucchi; Carlo F. Magni; Elisabetta Teti; Ada Bertoli; FrancescoPaolo Antonucci; Maria C. Bellocchi; Valeria Micheli; Chiara Masetti; Simona Landonio; Simona Francioso; Francesco Santopaolo; Adriano M. Pellicelli; Vincenza Calvaruso; Laura Gianserra; Massimo Siciliano; Dante Romagnoli; Raffaele Cozzolongo; Antonio Grieco; Jacopo Vecchiet; Filomena Morisco; Manuela Merli; Giuseppina Brancaccio; Antonio Di Biagio; Elisabetta Loggi; Claudio M. Mastroianni; Valeria Pace Palitti; Pierluigi Tarquini; Massimo Puoti; Gloria Taliani; Loredana Sarmati; Antonino Picciotto; Vincenzo Vullo; Nicola Caporaso; Maurizio Paoloni; Caterina Pasquazzi; Giuliano Rizzardini; Giustino Parruti; Antonio Craxì; Sergio Babudieri; Massimo Andreoni; Mario Angelico; Carlo F. Perno; Francesca Ceccherini-Silberstein


Liver International. 2017;37(4):514-528. 

In This Article


Despite the great improvement in HCV clearance rates endorsed by modern DAA therapies, around 5%-10% of patients fail to achieve viral eradication, usually in association with the development of RASs.[8] Few real-life data are currently available on DAA failures, and an exhaustive description of clinical characteristics of failing patients, along with their resistance profile, is still largely missing. For this reason, we aimed at defining the pattern of clinically relevant RASs associated with DAA failures, as well as the role of resistance testing in this setting, in a large real-life population of DAA-treated patients.

Overall, a total of 200 virological failures from 197 patients who failed a DAA-containing regimen, with or without IFN, were analysed. As compared to clinical trials, our population was much more heterogeneous and all with an advanced liver disease (according to Italian rules of treatment): with a high prevalence of cirrhosis (65.5%), previous treatment experience (56.8%), different HCV genotype/subtype (non-HCV-1: 45.9% in the DAA IFN-free group) and with different types of DAA regimen failure.

Out of 197 patients, 111 failed a DAA IFN-free regimen. In particular, 47.8% failed a recommended regimen according to the international recommendations,[24] while 52.2% of patients failed a suboptimal/not-recommended IFN-free regimen. Interestingly, although the majority of virological failures to IFN-free regimens reported in literature are relapses,[8,26] our population included also around 20% of breakthrough/non-responses. One possible explanation is a misleading genotype assignment by commercial genotyping assays that, in our population, led to a wrong treatment indication in at least 4/7 (57.1%) non-responders to a DAA IFN-free regimen. As we have previously reported,[27] in view also of the outstanding cost of DAA-based regimens, a precise and definite assessment of HCV genotype and subtype is mandatory in order to avoid mistreating during the first DAA course. In this context, HCV sequencing should be considered as the best method for subtype/genotype assignment.[1,28] Another possible explanation for our unusual high rate of breakthrough/non-responders is the possible presence of pretreatment natural RASs. Unfortunately, baseline resistance testing was largely unavailable to support this hypothesis, although two of the eight breakthrough/non-responders tested actually had natural RASs. Overall, however, baseline resistance was found in 20.0% of patients and it was confirmed always at virological failure, and in cases of patients failing a NS3- or NS5A-containing regimen, new additional RASs were observed. Lastly, another possible explanation, that could justify this uncommon presence of breakthrough/non-response to IFN-free DAA regimen, is a potential poor adherence and/or presence of other host factors, which could result with RAS selection for an inappropriate viral suppression, but in such a case RASs cannot be considered as a major reason of non-response.

A high prevalence of RASs at failure was observed particularly in patients failing an IFN-containing regimen (73.0% vs 49.5% in IFN-free), concordantly with previous results on TVR or BOC failures.[29,30]

The lower prevalence of RASs in IFN-free treated patients can be explained by the fact that most of these patients (N=51, 45.9%) were treated with only SOF+RBV. The majority of them failed as relapse and without resistance (RASs only in 13.7%). This result is in line with the fact that SOF is a DAA with a high barrier to resistance[31] and the presence of RASs at SOF failures was reported to be not frequent.[32] On the contrary, RAS prevalence was higher in NNI-failing patients (28.6%), and even more in NS3 (70.5%) and NS5A-failing patients (96.1%). This high prevalence of NS5A-RASs found in NS5A-failing patients is supported by the low barrier to resistance of the current NS5A inhibitors, despite their high potency and broad genotypic activity.

The NS5A-RASs are currently acquiring an increasing clinical relevance, since they tend to be long-term persistent (over 1–2 years after treatment failure)[13,14,33] and may thus affect future retreatment strategies. Moreover, not all NS5A-RASs have the same impact according to the inhibitors used and also to the different HCV genotype.[15,23] In our dataset, among the major clinically relevant NS5A-RASs, Y93H was the most frequent at failure (69.2%), although with different prevalence among HCV genotypes (89%-100% in HCV-1b-HCV-3 and 0% in HCV-2), followed by the Q30R in HCV-1a (80%). Furthermore, 11/26 patients (42.3%) presented >2 NS5A-RASs and 2/26 (7.8%) three NS5A-RASs. In particular, complex RAS patterns were observed more frequently in HCV-1b failures (with Y93H+L31M/F/I), while all the HCV-3 NS5A failures showed only the Y93H. Such a RAS alone is able to confer a high level of resistance (fold-change >1000) to the majority of NS5A inhibitors in HCV-1a genotype, while in HCV-1b confers high-level of resistance (fold-change >1000) only to LDV. Interestingly, HCV-1b, when Y93H is associated with additional RASs in position 31, is able to confer a high level of resistance (fold-change >1000) to different NS5A inhibitors, much higher than single RASs.[23]

Considering all these aspects, today patients that failed a NS5A inhibitor-containing regimen are considered "difficult-to-retreat", and NS5A-free retreatment strategies are generally encouraged,[2,24,25] or retreatments should be based on triple or quadruple DAA regimens, including a drug with a high barrier to resistance (currently, SOF), plus a combination of NS3+NS5A inhibitors.[1,2]

Additionally, since 2015, the AASLD recommendations suggest to perform NS3 and NS5A resistance testing in cirrhotic/advanced patients with previous NS5A failure[34] prior to deciding on a new treatment strategy. Differently, whether HCV resistance testing prior to retreatment is helpful to make a decision is still an open issue in recently published EASL guidelines, as well as which therapeutic decision should be made based on this result. If reliable resistance testing is performed, retreatment can be guided by probabilities of response according to the resistance profile observed in the context of an experienced multidisciplinary team.[1]

In the presence of only NS5A-RASs, the best option should be the change in the DAA class, by using the NS5B+NS3-containing regimen, however, the most difficult-to-cure patients may need more than 12 weeks of treatment and/or the addition of RBV.[21]

Interestingly, in our real-life population we observed a higher presence of SOF-RASs among any SOF-failing patients (19/93, 20.4%) than previously reported. In particular, the major SOF-RAS S282T which was rarely (<1%) observed among SOF-failing patients in clinical trials and real life[35–37] was found in three of four SOF-breakthrough patients, with an overall prevalence of 3% of SOF failures. Moreover the presence of other SOF-RASs (L159F, C316N and V321A), recently described as associated with virological failure to SOF,[38,39] was also evaluated in our population. The L159F was found (frequently in combination with C316N) in 14.6% of SOF relapsers in the IFN-free group (10/29, 34.5% HCV-1b and 3/19, 15.8% HCV-3) and in 80% of HCV-1b patients who failed a SOF+PegIFN+RBV. Interestingly, L159F was present also in 10% of patients before starting any treatment. However, since the baseline sample was not available for all SOF-failing patients, we cannot evaluate if these variants emerged at virological failure or were, at least in some cases, already present at baseline as natural polymorphisms. Differently, the V321A RAS was never detected in our population, neither at failure nor at baseline.

Interestingly, in the IFN-free group, the distribution of RASs at failure was higher in patients with unfavourable IL-28 CT/TT genotype compared to patients with favourable IL-28 CC genotype, suggesting that the host immune system can still play a role in the IFN-free DAA era. However, only 28 patients had IL-28 data available, therefore, this observation should be confirmed by analysing a larger cohort of patients.

Another critical point raised by our study was the frequent detection of RASs in multiple targets. Indeed, 47.4% of patients treated with ≥2 DAA classes showed RASs on ≥2 DAA targets at failure, including 100% NS3-NS5A failures. Moreover, around 9.0% of failing patients presented "extra-target" RASs, probably because of natural resistance. Patients who show RASs at three DAA classes, or in both NS3 and NS5A targets are currently the most difficult to retreat, indeed they have no chances of retreatment with commercially available IFN-free regimens and could be managed only by multiple DAA combinations targeting nearly all replication steps.[1,25,40,41]

In conclusion, in this real-life setting, RAS prevalence at failure was remarkably high in all genes tested (with a partial exception for NS5B, whose limited resistance is still higher than previously reported). These results underline the importance of performing a HCV resistance test at failure in order to select the best DAA regimen retreatment option. Indeed, other preliminary data in a small number of patients suggest that retreatment can be optimized based on RAS testing.[22]

In particular, because of the frequent natural presence of RASs, HCV resistance test at failure should be recommended for all three genes (including NS5B), independently of the administered regimen. Our results advocate for tailored retreatments, sometimes based upon unconventional, more aggressive/prolonged regimens which are able to overcome the natural/acquired resistance and to warrant high success rates.