Nationwide Experience of Treatment with Protease Inhibitors in Chronic Hepatitis C Patients in Denmark: Identification of Viral Resistance Mutations

Background and Aims The first standard of care in treatment of chronic HCV genotype 1 infection involving directly acting antivirals was protease inhibitors telaprevir or boceprevir combined with pegylated-interferon and ribavirin (triple therapy). Phase III studies include highly selected patients. Thus, treatment response and development of viral resistance during triple therapy in a routine clinical setting needs to be determined. The aims of this study were to investigate treatment outcome and identify sequence variations after triple therapy in patients with chronic HCV genotype 1 infection in a routine clinical setting. Methods 80 patients, who initiated and completed triple therapy in Denmark between May 2011 and November 2012, were included. Demographic data and treatment response were obtained from the Danish Database for Hepatitis B and C. Direct sequencing and clonal analysis of the RT-PCR amplified NS3 protease were performed in patients without cure following triple therapy. Results 38 (47%) of the patients achieved cure, 15 (19%) discontinued treatment due to adverse events and remained infected, and 27 (34%) experienced relapse or treatment failure of whom 15 of 21 analyzed patients had well-described protease inhibitor resistance variants detected. Most frequently detected protease variants were V36M and/or R155K, and V36M, in patients with genotype 1a and 1b infection, respectively. Conclusions The cure rate after triple therapy in a routine clinical setting was 47%, which is substantially lower than in clinical trials. Resistance variants towards protease inhibitors were seen in 71% of patients failing therapy indicating that resistance could have an important role in treatment response.


Introduction
Persistent hepatitis C virus (HCV) infection is a major cause of chronic hepatitis, cirrhosis and hepatocellular carcinoma. Liver related mortality, once cirrhosis has developed, is 3% per year [1]. Until recently, standard-of-care (SOC) for chronic HCV genotype (GT) 1 infection was pegylated-interferon alfa (PEGINF) and ribavirin (RBV), curing only 40-50% of patients (Pt.) [2]. Protease inhibitors (PI's) telaprevir (TEL) and boceprevir (BOC), targeting the HCV non-structural 3 (NS3) protease, have, when added to PEGINF/RBV (triple therapy), improved cure rates up to 79% in clinical trials of highly selected, treatment-naïve and relapse HCV GT1 infected patients [3]. However, due to possible selection bias it is not yet clear if this promising treatment response can be transferred into a routine clinical setting.
In chronic HCV infection, the circulating viral quasispecies can harbour PI resistant variants, identified in 0.2 to 2.8% of treatment naïve patients [4]. Under pressure of antiviral treatment, PI resistant variants are selected in patients with no or partial response to treatment. Several major amino acid (aa) positions within the NS3 HCV protease [5][6][7] associated with different resistance levels have been identified, and most confer broad cross-resistance between TEL and BOC [1,8]. The selection, emergence and persistence of PI resistant viruses are of significant concerns, since these resistant variants could influence future treatment options with second generation PI's.
The objectives of this study were to investigate treatment response to triple therapy in HCV GT1 patients in a nationwide, routine clinical setting, and to describe the development of variations in the protease sequence after treatment failure.

Patients and treatment
Study participants recruited from The Danish Database for Hepatitis B and C (DANHEP) commenced triple therapy from May 2011 to November 2012. They were treated in accordance with Danish guidelines [9,10].
Investigators prescribed treatment following the manufacturer's protocol of the relevant PI. Outcome was defined as sustained virological response (SVR; negative HCV-RNA 24 weeks after End of Treatment (EOT)), relapse (undetectable HCV-RNA at treatment completion, but detectable during follow-up), viral breakthrough (HCV-RNA levels initially decreases during treatment (undetectable levels can be seen), followed by a clinical relevant increase while on treatment) or non-response (persistent HCV-RNA positive) [5].
Demographic data, choice of PI, RBV dose reduction and triple therapy response ( Table 1) were extracted from DANHEP. The HCV subtype was determined by RT-PCR amplification and direct sequencing [11][12][13].
HCV-RNA levels were measured with a real-time TaqMan assay (Roche Molecular systems) with a lower limit of detection of 15 IU/ml [14] or an inhouse real-time PCR method as described [15]. IL-28B genotyping was performed as described [16].

Genotype, Phylogenetic and Sequence Analysis
Viral HCV-RNA was extracted from 200 ml of plasma or serum by using High Pure Viral RNA kit from Roche. For non-SVR patients to be analyzed for protease variants, the subtype was confirmed in the pre-treatment sample by analysis of C-E1 amplicons [11]. For NS3 protease analysis, direct sequencing of RT-nested PCR amplicons was done in 21 non-SVR patients; six patients, one with HCV GT 1l and five without study post-treatment samples were not included. RT was performed with Superscript III (Invitrogen) and cDNA was treated with RNAse H and T1. First round and nested PCR were performed with standard cycling conditions, using TaqGold polymerase (Applied Biosystems) for low viral load samples and Advantage (Clontech) for high viral load samples (primers are listed in Table 3). DNA was column-purified (Wizard Genomic DNA Purification Kit, Promega) and sequenced (Macrogen Inc.), and a phylogenetic tree ( Fig. 1) was generated for intra-individual confirmation. In a subgroup of 14 patients, NS3 amplicons at baseline and post-treatment were subjected to TA cloning (TOPA TA Cloning kit, Invitrogen). NS3 sequences, obtained pre-and post-treatment, were inspected for presence of putative resistant variants by Sequencher version Phylogenetic analysis of the NS3 sequences was performed with the Neighbour-Joining method, using Kimura 2-parameter as the nucleotide substitution model (PHYLIP, v3.6). The tree was visualized with TreeView ( Fig. 1). Molecular evolution analysis of the NS3 quasispecies in each patient was performed with MEGA software (version 5) [19].

Statistical analysis
Data is presented as absolute numbers. Categorical data was analyzed with Pearson's Chi Square test and continuous data with Fisher's exact test. A significant confidence interval at 95% was used. Rstudio software was used for all statistical analysis (www.R-project.org).

Patient Characteristics
In Denmark, 80 patients with chronic HCV genotype 1 infection initiated and completed triple therapy during the study period. Demographic characteristics and treatment outcome are shown in Table 1 and Fig. 2. Fifty-three (66%) and 26 (33%) patients had HCV GT 1a and 1b, respectively; a single patient had genotype 1l. BOC was used in 42 (53%) and TEL in 38 (47%) of the patients. RBV dose reduction was seen in 32 (40%) patients. All patients with SVR received minimum 50% of the standard amount of RBV but in two patients RBV was discontinued due to anaemia. In the non-SVR group, three patients (7%) received ,50% of RBV assigned by response-guided therapy and three patients were taken off RBV during treatment.

Virological Response
Thirty-eight (47%) patients achieved SVR ( Table 1 and Fig. 2). Figure 3 shows treatment responses for TEL and BOC, expressed as the percentage of patients with undetectable HCV-RNA during treatment weeks 4 to 48.
Only 1 (2.4%) BOC treated patient had undetectable HCV-RNA at week 4, which could be explained by the 4-week lead-in phase with PEGINF/RBV. Peak TEL response (25 patients (66%) with undetectable HCV-RNA) was reached at week 8, but thereafter the number of patients with undetectable HCV-RNA decreased. For BOC, the number of patients with undetectable HCV-RNA increased to 29 (69%) at week 36. At week 48 it dropped to 52% due to either relapse or viral breakthrough. Table 4. Resistance protease variants identified in vitro in replicon-based studies and in vivo in clinical trials [1,4,7,17,18].
To determine whether changes in the protease domain of NS3 had occurred after triple therapy we compared the viral direct sequence in the pre-and posttreatment samples ( Table 5). Additionally, in a selected group of 14 patients, we   Table 6).

Viral resistance in genotype 1a
We identified NS3 protease changes in 13 of 14 patients with GT1a; changes were not detected in one patient (Pt. 14). In all patients, except one (Pt.12), showing amino acid changes in the post treatment sequence, those changes were not present in the pre-treatment sequence. Among the observed aa changes, positions V36, T54, Q80, R155 and I170 were previously involved in resistance to PI's (  Table 4) [5][6][7]20]. The combination of V36M and R155K, known to confer high level of PI resistance, was present in patients 4, 5 and 6 post-treatment. Clonal analysis of post-treatment samples from patients 4 and 5, where all clones contained the combination V36M/R155K, further supported these results. The single resistant variant R155K was present in patients 1, 3 and 9 post-treatment; clonal analysis confirmed R155K in all three patients. V36M as a single resistance variant was present post-treatment in patients 8 and 11. Patients 10 and 13 presented only changes in residues S122 or V153 post-treatment, respectively, which have not been previously reported as resistant variants against PI's (  Table 4). In patient 10, the clonal analysis did not show the change S122G but changes Y75N (20% of clones) and F184I (20% of clones). Patient 12 showed Q80Q/L, previously reported as a resistant mutation, in the direct sequence of both the pre-and post-treatment sample. In patient 2, two resistance variants, V36M and I170T were detected in the direct sequencing of the post-treatment sample; V36M appeared in all clones post-treatment, while I170T was present in only 50% of the clones. The post-treatment sample of patient 7 showed the resistance variants T54S and R155K.

Viral resistance in genotype 1b
The direct NS3 sequences of HCV recovered from the seven non-SVR patients with GT1b, revealed 2 patients (patients 18 and 21) with no changes in the protease domain sequence pre-or post-treatment ( Table 5), but the remaining patients showed changes that were not present in the pre-treatment sample. Patient 19 had a unique clone in the post-treatment sample containing the resistant variant V55A. Patients 15, 16, and 17 each had a single previously reported resistant mutation (A156T, V36A, and V36A, respectively) posttreatment; the clonal analysis agreed with the direct sequencing in all three patients. Patient 20 presented three changes post-treatment not previously involved in resistance to PI's. Table 6. Clonal analysis of the NS3 protease in 14 non-SVR patients.

AA in NS3
Patient ID-change

Quasispecies analysis
For each of the 14 non-SVR patients analyzed, nucleotide clonal sequences were divided into pre-and post-treatment groups and diverse population parameters were calculated. Intra and inter-population genetic distances were estimated for each patient ( Table 7). For 5 of 9 GT1a patients and 3 of 5 GT1b patients, genetic distances between sequences were higher in the pre-treatment population than in the post-treatment population, suggesting a decrease in diversity of the quasispecies after treatment. We observed no relationship between HCV GT or type of PI to the increase or decrease in genetic diversity of HCV. The dN/dS ratio was calculated for the 14 patients to define the direction and magnitude of natural selection acting on the protein coding genes.
Overall, the number of synonymous substitutions per synonymous site (dS) exceeded the number of non-synonymous substitutions per non-synonymous site (dN) and as previously reported [21,22], a dN/dS ratio ,1 was observed in all patients suggesting that purifying selection was acting in the NS3 protease domain ( Table 7).
We also determined the number of variants at the aa level in the quasispecies pre-and post-treatment. In GT1a patients, the number of different variants increased in the post-treatment quasispecies in four of the five TEL-treated patients. In GT1a BOC-treated patients, the number of clones with aa differences only increased post-treatment in patient 11, remained unchanged in patient 10 and decreased in patients 8 and 9. Regarding GT1b infected patients, the number  of different clonal variants decreased after treatment in three patients and increased in the remaining two (Fig. 4).

Discussion
The present study is to our knowledge the first to report SVR rates at week 24 after triple therapy as well as detection of resistance mutations in patients with chronic hepatitis C GT1 in a nationwide, routine clinical setting. The SVR rate was only 47% and well-described resistance variants were detected in the majority of non-SVR patients. An explanation for the much lower SVR rate than the SVR rate reported in clinical trials [1,23] could be that study populations in clinical trials are usually highly selected. The SVR rate in our study was similar to SVR rates reported in patients treated with PEGINF/RBV [24]. From the French Early Access Programme including 497 PEGINF/RBV treatment-experienced patients with HCV GT1 infection and compensated cirrhosis [25], response rates at post-treatment week 16 have been reported to be 67% for TEL-treated and 58% for BOC-treated patients, respectively. However, as SVR rates have not yet been reported and the patient characteristics of the French cohort differ significantly from ours, direct comparison is not possible. A recent study by Macartney et al. [26] including 58 treatment naïve and experienced patients found that 37 (63.8%) had undetectable HCV-RNA 12 weeks after EOT. However, SVR rates at week 24 have not been published and may be lower due to relapse during the follow-up period.
The poor response to triple therapy in our study could partly be explained by a significantly larger number of patients with cirrhosis in the non-SVR -versus the SVR group (23 (70%) versus 10 (30%)), as well as fewer patients with the favourable IL28-B genotype CC (4 (29%) versus 10 (71%)). In the study by Maasoumy et al., IL-28B genotype CC was present in 18% of patients which is similar to our results with 14 (21%) of 67 patients and was also associated with a better treatment outcome.
We observed that adverse events leading to early treatment discontinuation occurred in 20 (25%) patients, similar to observations from other non-clinical trials [25,27]. Poordad et al. [28] observed that RBV dose reduction throughout the course of triple therapy did not affect SVR rates, as long as the patients received at least 50% of total amount RBV assigned by response-guided RBV therapy. We observed no difference between SVR (42%) and non-SVR (38%) patients in relation to RBV reduction.
We detected well-described PI resistance variants in 15 (71%) of 21 patients analyzed. A similar rate (50-75%) has been shown in patients in phase III trials at time of triple therapy failure [18] and a rate of 81% was seen in a routine clinical setting in the study by Macartney et al. [26]. In consistency with other studies [8,18,20], the most frequently detected variants in patients with HCV GT1a infection were R155K and V36M, whereas for GT1b infection, V36A was mainly observed.
Interestingly we did not detect Q80K, the signature mutation for the recently licensed PI simeprevir, neither at baseline nor at treatment failure. This indicates that Q80K may not be as prevalent in Denmark as in the US where 20% of the population harbours this mutation at baseline [18]. This may also suggest that patients who failed 1 st . generation PI treatment may not be resistant to simeprevir.
Twelve patients showed amino acid changes in the post treatment sample that had not been previously associated with resistance to PIs. To our knowledge, from these changes, only amino acid P88 (seen in patient 8), have been previously described in the literature, and was associated with viral fitness compensation. The function of most of the additional changes reported in this manuscript is yet to be defined in future studies.
While one patient (Pt. 15) had A156T, which is known to confer the highest level of resistance to TEL and BOC [8], six of the non-SVR patients in this study had no known resistance variants detected pre-or post-treatment. We cannot discard that resistant mutations were indeed present in a small portion of the viral population, and therefore could not be detected by direct sequencing or clonal analysis of 10 clones in the indicated cases. Other methodologies such as deep sequencing could help clarifying this matter. On the other hand in four of these six patients, post-treatment analysed samples were obtained more than 10 weeks after EOT, which could imply, that in most of the circulating variants carrying resistance mutations these mutations would have reverted to the original amino acid, due to the decrease in viral fitness they produce. In this regard, studies have shown that the frequency of PI resistance variants decline over time, following drug withdrawal, but in some cases, resistance variants can be found at low-level in post-treatment samples, up to five years after treatment completion, using high resolution deep sequencing analysis [2,3,8].
PI's in phase II and III trials show extensive cross-resistance with TEL and BOC [18,20]. Patients with detectable resistance variants included in this study should be tested short-term and long-term for the persistence of resistance variants as it is currently unknown, if the found variants can re-emerge during re-treatment with 2 nd generation PI's.
Regarding the viral quasispecies analysis in non-SVR patients, no clear pattern of genetic evolution was observed. The lack of correlation with clinical data could be a result of the complexities of the forces driving evolution of NS3. The NS3 protease domain contains immunological relevant epitopes and therefore is subjected to immune selection, as seen in natural evolution of NS3 in chronically infected patients [29]. In patients treated with PI, a second evolutionary force takes action, and at last both selective pressures may act together, since the immune epitopes may overlap with PI resistant mutations [22], creating a complex interplay.
After treatment, the viral population (amino acid level) changed significantly. In the post-treatment viral population, most patients presented sequences that were different from the sequences found in the pre-treatment population. Only three patients had post-treatment sequences that were already found in the pretreatment sample. Interestingly, these patients had none or very low frequency of resistant mutations after treatment thus indicating a trend towards preservation of pre-treatment sequences through the post-treatment in the absence (or very low frequency) of PI resistant mutations. However, due to the small number of patients these findings remain to be confirmed in larger studies.
As previously observed in the analysis of NS3 [21,22], the dN/dS was less than 1, suggesting purifying selection in the protease sequence during triple therapy. This result was unexpected, since most of the patients had a dramatic change in the quasispecies, where a minor variant harbouring a resistance mutation increased its frequency in the quasispecies and became a dominant sequence after treatment, thus suggesting strong positive selection. Domingo et al. [30] have reported this phenomenon, where a few amino acid changes driving the evolution of a region are accompanied by an excess of synonymous changes, greatly affecting the results of the dN/dS ratio.
The strengths of this study is its nationwide conduction and the continuous recording of patient data and blood samples into DANHEP, making it possible to obtain data on demographic characteristics and treatment response on all patients receiving triple therapy outside clinical trials in Denmark. Our lost to follow-up was limited to one patient who discontinued treatment after 16 weeks.
There are some limitations to the present study. Denmark is a relative small country with a low prevalence of chronic HCV GT1 infected patients, affecting the number of patients possible to treat with triple therapy. Due to the DANHEP database and the close collaboration between the treating departments, the 81 patients included in this study represent the total number of patients who started triple therapy in Denmark during the study period and before treatment with 1 st generation PI's stopped due to the possibility of treating with 2 nd generation Direct Acting Antivirals (DAAs). The compliance to drug intake was reflected in the blood samples, with decreased levels of neutrophilocytes, platelets and red blood cells, in patients undergoing treatment and was rated to be very good by the treating physicians. Only 4 (5%) patients had one missed blood sample during the treatment course which reflects the very good show up and adherence to treatment at the different departments. We used clonal analysis in 14 patients, which is a recommended procedure for detection of complex mutation patterns, but the sensitivity of the procedure is highly dependent on the number of sequenced clones. Clonal sequencing can detect variants at very low frequencies, requiring a sequence rate .80 clones for a 4% detection limit at a 95% confidence level [20]. Ten clones were analyzed in this study, which is a limiting factor in detecting variants found pre-treatment at a very low level. Studies using deep sequencing indicate that pre-existing protease resistant variants are detectable at the 1% level in approximately 7-8% of patients [18]. This method could perhaps have increased the detection rate of resistance variants at baseline and could have clarified if patients with non-SVR already pre-treatment had the resistance variants found later. However, we found a very good correlation between the direct sequencing and the clonal analysis. In only two patients did clonal analysis reveal a resistance mutation not detected by direct sequencing (patients 2 and 19), and in five patients, aa changes, unrelated to resistance, arose after clonal analysis.
In conclusion, this study is the first to report, on a nationwide basis, that SVR rates after triple therapy outside clinical trials was lower than that in clinical trials and comparable to the SVR rate obtained by PEGINF/RBV therapy. The overall success rate of these new treatment regimens towards chronic HCV GT1 infection will require more data from routine clinical settings. Furthermore, we detected no major pre-treatment sequence variants associated with resistance to PI's, but during treatment, resistance variants occurred in the majority of patients with non-SVR. These data are important in relation to future treatment with 2 nd generation PI's.