Incidence of virological failure and emergence of resistance with twice-daily vs every 8-h administration of telaprevir in the OPTIMIZE study

Summary: The OPTIMIZE study demonstrated noninferior efficacy between telaprevir (TVR) twice daily (bid) vs every 8‐h (q8h) administration. This analysis compared the selective pressure of both dosing regimens by characterisation of the hepatitis C virus (HCV) variants emerging in genotype 1 (G1) HCV‐infected patients who did not achieve sustained virological response (SVR). HCV NS3•4A population sequencing was performed at baseline and time of failure (viral breakthrough, stopping rule or relapse). TVR‐resistant variants were classified by fold change in inhibitory concentration (IC50). Baseline TVR‐resistance was low (<5%) and did not preclude achieving SVR in either arm. The proportion of patients with TVR‐resistant variants at time of failure was similar in the bid (15%) and q8h (17%) dosing arms. The majority of variants and virological failures occurred in G1a patients, and mutations V36M, R155K and R155T (G1a), and V36A, T54A and A156S (G1b) were significantly enriched in both treatment arms. The number and type of emerging TVR‐resistant variants in non‐SVR patients were comparable between treatment arms and were consistent with previous observations. No differences in viral resistance profiles were observed between TVR‐based treatment arms in non‐SVR patients, indicating a similar selective pressure of TVR bid and q8h dosing.

genotype 1; hepatitis C virus; resistance; telaprevir; twice‐daily dosing; viral variants

Abbreviations

DAA direct‐acting antiviral

G1 genotype 1

HCV hepatitis C virus

IL28B interleukin 28B

pEOT planned end of treatment

P pegylated interferon

R ribavirin

SVR sustained virological response

TVR telaprevir

VF virological failure

The hepatitis C virus (HCV) NS3•4A protease is the target of several direct‐acting antiviral (DAA) drugs, including telaprevir (TVR) [1] . Phase III clinical studies in adult patients with genotype 1 (G1) chronic HCV infection demonstrated significantly higher rates of sustained virological response (SVR12) with 12 weeks of treatment with the NS3•4A protease inhibitor TVR in combination with either 24 or 48 weeks of pegylated interferon (P) plus ribavirin (R) (pegylated interferon/ribavirin, PR) than with 48 weeks of PR alone, both in treatment‐naïve and treatment‐experienced patients [2] , [3] , [4] , [5] .

The high HCV replication rate, in addition to poor polymerase fidelity, can lead to the generation of numerous viral variants with a high degree of sequence variation [1] , [6] . Therefore, HCV variants with differing levels of susceptibility to DAAs may exist naturally in the absence of drug pressure; however, these natural variants exist at low levels [1] . In patients not responding to TVR‐based therapy, selection of HCV variants with decreased sensitivity to TVR has been seen, as observed with other DAAs [2] , [3] , [7] , [8] , and a consistent, subtype‐dependent resistance profile has been identified from a comprehensive analysis of patients who did not achieve SVR in the TVR Phase II and Phase III studies [9] . However, changes in the frequency of substitutions in the viral population after the end of TVR dosing have shown an inverse relationship between in vivo viral fitness and resistance [10] . After treatment, viral populations became predominantly wild type after a median of 10 months for G1a patients and 3 weeks for G1b patients [11] .

The Phase III, randomised OPTIMIZE study in G1 HCV‐infected, treatment‐naïve patients recently demonstrated that the TVR 1125 mg twice‐daily (bid) regimen was noninferior to a 750 mg every 8‐h (q8h) regimen for the primary efficacy endpoint of SVR12 (74% and 73% for TVR bid and TVR q8h, respectively; difference 1.5%; [95% confidence interval [CI]: 4.9–12.0]) [12] . These results support the use of TVR bid dosing in patients with G1 chronic HCV infection, including those with cirrhosis [13] , [14] . Given previous observations of resistant variants in other clinical studies of TVR, the aims of the present analysis were to characterise the viral variants emerging in patients who did not achieve SVR after TVR‐based treatment in the OPTIMIZE study and to compare the selective pressure of the TVR bid and q8h dosing regimens.

OPTIMIZE was a Phase III, randomised, open‐label, multicentre study. The study methodology has been reported in detail elsewhere [12] . Briefly, treatment‐naïve patients with chronic HCV G1 infection were randomised 1:1 to receive 12‐week TVR 1125 mg bid or 750 mg q8h, each in combination with peginterferon alfa‐2a (180 μg/week) and weight‐based ribavirin (1000–1200 mg/day) for either 24 or 48 weeks based on response at week 4 (patients with HCV RNA <25 IU/mL, target not detected at week 4 [defined as rapid virological response, RVR] received 24 rather than 48‐week treatment in total). Patients were stratified by fibrosis stage and IL28B genotype. Futility rules were applied to all patients to minimise the risk of viral resistance in patients without an adequate antiviral response. HCV RNA values were monitored and all treatments were stopped if HCV RNA levels were >1000 IU/mL at week 4 or ≥25 IU/mL at weeks 12, 24, 32 or 40.

The study protocol was reviewed and approved by the appropriate review boards or institutional ethics committees and health authorities and was conducted in accordance with the Declaration of Helsinki, the Good Clinical Practice guidelines and applicable regulatory requirements. Written informed consent was obtained from all patients. The study was registered with ClinicalTrials.gov (NCT01241760).

The primary efficacy endpoint of the study was SVR12 (HCV RNA <25 IU/mL 12 weeks after the last planned dose of study drug) [12] . Patients not achieving SVR12 were included in this analysis and classified according to the reason for treatment failure: on‐treatment virological failure (VF), relapse or other reasons.

On‐treatment VF was defined as patients who met a virological stopping rule and/or had viral breakthrough either during the TVR treatment phase or PR treatment phase (after TVR treatment). Viral breakthrough was defined as confirmed >1 log10 increase in HCV RNA from nadir, or confirmed on‐treatment HCV RNA >100 IU/mL if HCV RNA was previously <25 IU/mL, from week 4 onwards. Relapse was defined as all non‐SVR12 patients with HCV RNA ≥25 IU/mL during follow‐up after previous HCV RNA <25 IU/mL at the planned end of treatment (pEOT).

The ‘other’ category included patients with HCV RNA ≥25 IU/mL at their pEOT but who did not have viral breakthrough, patients with HCV RNA <25 IU/mL at pEOT and at least one HCV RNA ≥25 IU/mL result thereafter, and those with a missing HCV RNA assessment during follow‐up.

Plasma HCV RNA values were measured using the High Pure System COBAS® TaqMan® HCV assay (v2.0; Roche, Basel, Switzerland) with a lower limit of quantification (LLOQ) of 25 IU/mL. Values below the LLOQ were reported as ‘<25 IU/mL, target detected’ if an HCV signal was detected, but were reported as ‘<25 IU/mL, target not detected’ if no HCV target was detected. An NS3‐based genotyping method was used for genotype and subtype determination in this virological analysis.

NS3•4A viral sequencing samples were taken at baseline, weeks 4, 8, 12, 24, 32, 40 and EOT (i.e., week 24, week 48 or at early discontinuation) and 4 weeks (safety follow‐up visit), after the last dose of study drug. Follow‐up samples were taken 12 and 24 weeks after the last planned dose of study drugs and 72 weeks after the start of study drugs, and in case of early discontinuation also at the pEOT visit.

HCV NS3•4A population sequencing analysis was performed for all patients at baseline and during treatment and/or follow up in those who did not achieve SVR12. Population sequencing methods have been presented elsewhere [15] . Briefly, HCV RNA was extracted from plasma virions under denaturing conditions, and viral RNA was isolated on a standard commercial silica gel membrane using a modified QIAamp Virus BioRobot 9604 method (Qiagen, Valencia, CA, USA). The NS3•4A protease regions were amplified using a nested reverse‐transcriptase polymerase chain reaction assay. The resulting DNA was purified and sequenced directly. Sequence analysis of the full‐length NS3•4A region was performed for samples with sufficient levels of HCV RNA (lower limit of detection of the sequencing assay HCV RNA ~1000 IU/mL), and population sequencing can typically detect down to 20–25% of the viral population. The sequence analysis focused on G1 amino acid substitutions in the NS3•4A region that were identified in patients who did not achieve SVR in TVR Phase II and III clinical studies [9] . TVR‐resistant variants confer lower level (3‐ to 25‐fold increase in 50% inhibitory concentration [IC50]: single change V36A/G/M, T54A/S, R155G/K/M/T and A156S) or higher level (>25‐fold increase in IC50: single changes A156F/N/T/V; double changes V36M + R155K, V36A + R155K/T, V36M + R155T, V36A/M + A156T, T54A + A156S, and T54S + A156S/T) in vitro resistance to TVR. Substitutions V36I/L, I132V (G1a only) and D168N, which were also observed in patients who did not achieve SVR in TVR clinical studies do not change the sensitivity to TVR in the replicon system (<3 fold increase in IC50). A VF sequence time point was determined for each patient who did not achieve SVR, at which the viral population was considered to be representative of the viral population present at the time of failure. For patients experiencing VF during TVR treatment, this was the last time point on TVR with sequence data available. For patients experiencing VF after the end of TVR dosing, this was the first time point with sequence data available.

Statistical analysis comparing pretreatment and VF sequences was based on Fisher's exact test for unpaired sequences and Liddell's exact test for paired sequences (statistical significance defined as P < 0.05 in both tests), after a Bonferroni correction for multiple comparison. The position‐specific, pretreatment distribution of amino acid states was estimated from baseline or predose sequences in Phase II and III TVR studies [9] . The VF distribution was computed from the sequences at the VF sequence time point from all patients failing treatment in OPTIMIZE. Potential resistance‐associated mutations were identified as amino acid states occurring at significantly higher frequencies in VF vs pretreatment sequences. No statistical analyses were performed to compare the bid and q8h arms in subgroups where the sample size was too low.

A total of 740 treatment‐naïve patients were randomised and treated with TVR bid (n = 369) or q8h (n = 371) plus PR [12] . Baseline characteristics were balanced between treatment arms and have been described elsewhere [12] . Overall, 57% and 43% of patients had HCV G1a and G1b infection, respectively, and 14% of patients had cirrhosis.

Treatment outcomes were comparable between both dosing regimens for SVR12, on‐treatment VF (patients who met a virological stopping rule and/or viral breakthrough) and relapse rates (Table [NaN] ). SVR12 rates were 74% in the TVR bid arm and 73% in the TVR q8h arm. Patients who did not achieve SVR12 failed treatment due to relapse, on‐treatment VF or for other reasons (including patients with HCV RNA ≥25 IU/mL at pEOT but who did not have viral breakthrough and patients with a missing HCV RNA assessment during planned follow‐up). On‐treatment VF occurred in 10% of patients in both treatment arms, while relapse was reported in 8% and 6% of patients in the TVR bid and q8h arm, respectively. The majority of on‐treatment VF occurred in G1a patients (12% and 15% vs 8% and 3% for G1b patients in the TVR bid and TVR q8h arms, respectively). Of those patients who did not achieve SVR12 for reasons other than VF and relapse, the most common reason (5%) was having HCV RNA ≥25 IU/mL at least once after an HCV RNA result <25 IU/mL at end of treatment (12 patients in the bid arm and 23 patients in the q8h arm). Overall, 9 (1%) patients had HCV RNA ≥25 IU/mL at EOT and 19 (3%) patients never had HCV RNA ≥25 IU/mL after HCV RNA <25 IU/mL at EOT. Seventeen patients (2%) could not be assigned to the previously defined ‘other’ categories and were lost to follow‐up or discontinued from the study.

Categories of treatment outcomes according to treatment arm and genotype

1 EOT, end of treatment; N = number of patients with data; n = number of patients with that observation; PR, peginterferon/ribavirin; SVR12, sustained virological response (HCV RNA <25 IU/mL, 12 weeks after the last planned dose of study drug) TVR, telaprevir; VF, virological failure. *For four patients (two in each treatment group), no subtype could be generated by NS3 genotyping. †The denominator includes patients with HCV RNA <25 IU/mL at pEOT or who have a missing HCV RNA result at pEOT but HCV RNA <25 IU/mL afterwards. ‡Based on the actual time of meeting a stopping rule as derived from disposition and exposure information. §Patients with viral breakthrough that did not meet a stopping rule. ¶These patients could not be assigned to any of the previously defined ‘other’ categories and were lost to follow up or discontinued from the study.

Population‐based sequencing of the NS3•4A protease domain was successful for 99% (735/740) of patients. Of these patients, 95% (701/735) had wild‐type virus (no TVR‐resistant variants) at baseline. The prevalence of TVR‐resistant variants at baseline was low (34 [5%] patients including 25 [6%] G1a and 9 [3%] G1b patients) and was comparable between the TVR bid and q8h arms (5% and 4%, respectively). Across genotypes, the observed variants were T54S (2.3%), V36L (1.4%) and V36M (0.3%) alone, with V36I, V36L/M, I132V, R155K and V36I + T54S detected in one patient (0.1%) each. All of these baseline variants had lower levels (IC50 FC < 25) or no (IC50 FC < 3) resistance to TVR [9] . Treatment outcomes following TVR‐based therapy in patients with baseline variants are shown in Table [NaN] . Of the 34 patients with TVR‐resistant variants at baseline, 25 (74%) achieved SVR12, with similar SVR12 rates in both treatment arms for patients with or without pretreatment TVR‐resistant variants (Table [NaN] ). By genotype, SVR12 rates for patients who were G1a and who had TVR‐resistant variants at baseline were 75% (9/12) and 77% (10/13) for the bid and q8h dosing arms, respectively; by contrast, SVR12 rates for G1b patients were 71% (5/7) and 50% (1/2), respectively.

Treatment outcomes of patients with TVR‐resistant variants at baseline, overall and by treatment arm

2 SVR12, sustained virological response (HCV RNA <25 IU/mL 12 weeks after the last planned dose of study drug); TVR, telaprevir; VF, virological failure.

In total, 196 (26%) of the 740 patients included in the study did not achieve SVR12. HCV NS3•4A sequence data at the time of treatment failure (available for 163 patients) showed that V36M, R155K and R155T (G1a patients), and V36A, T54A and A156S (G1b patients) were significantly enriched in patients who did not achieve SVR12, relative to patients who had never been exposed to TVR (P < 0.05) in Phase II and III TVR studies. No new statistically significant changes in amino acid substitutions were detected, and there was no difference in significantly enriched mutations between treatment arms.

The proportion of patients with TVR‐resistant variants at time of failure was comparable between both treatment groups (Fig. [NaN] ). In total, 15% (54/369) vs 17% (62/371) in the TVR bid and TVR q8h arms, respectively, had TVR‐resistant variants. TVR‐resistant variants occurred more frequently in patients with G1a (20%; 83/419) vs patients with G1b (10%; 33/317).

Telaprevir‐resistant variants were detected in the majority of patients with available sequencing data who did not achieve SVR12 in both the TVR bid (54/77 patients; 70%) and q8h (62/86; 72%) treatment arms. The proportions of patients with on‐treatment VF during the TVR treatment phase (4% and 6%) or PR phase (6% and 4%) were comparable between the TVR bid and TVR q8h treatment arms. The number of patients with on‐treatment VF with resistant variants and the type of variants was generally similar between the treatment arms (Fig. [NaN] ).

Virological failure during the TVR treatment phase was predominantly associated with higher level TVR‐resistant variants (32/37 patients; 87%). This was evident particularly for G1a patients (8/10, 80%, for TVR bid; 20/22, 91%, for TVR q8h), with the most frequently detected variant combination being V36M + R155K (detected in seven patients for TVR bid and 20 patients for TVR q8h; Fig. [NaN] a). There were no G1b patients in the q8h arm with VF during the TVR treatment phase. VF during the PR phase was associated with a mixture of higher‐level resistant variants (15/34 patients; 44%), lower level resistant variants (9/34; 26%) and wild‐type virus (10/34; 29%). For G1a patients, VF was most commonly associated with the higher level variant combination of V36M + R155K, while the majority of G1b patients harboured wild‐type virus at the time of failure (Fig. [NaN] b).

For patients with relapse, the number of patients with TVR‐resistant variants and type of TVR‐resistant variants was similar between both treatment arms (Fig. [NaN] ). Relapse was mostly associated with lower level TVR‐resistant variants or wild‐type virus, especially in G1b patients. All patients who received <4 weeks of TVR‐based therapy failed to achieve SVR and were found to have wild‐type virus.

The introduction of DAAs has significantly improved SVR rates to HCV therapy for both treatment‐naïve and treatment‐experienced patients [2] , [3] , [4] , [5] . However, regimen convenience is also an important consideration for physicians and patients. Twice‐daily TVR administration combined with PR has been demonstrated to be noninferior to the TVR q8h regimen for the primary efficacy endpoint of SVR12, suggesting that this regimen could improve convenience without compromising efficacy [12] . Our results extend this finding by showing no difference in the viral resistance profiles between the TVR bid and q8h arms in patients not achieving SVR12, indicating similar selective pressure between the two regimens.

The proportion of patients with TVR‐resistant variants at time of failure was similar between both treatment groups, and similar on‐treatment VF or relapse rates were noted between the different dosing regimens. TVR‐resistant variants were present in the majority of patients who did not achieve SVR12. Moreover, the number of patients with variants and the types of variants did not differ between treatment arms. In addition, the viral variants significantly enriched at the time of treatment failure (V36M, R155K and R155T in G1a patients, and V36A, T54A and A156S in G1b patients) were similar in both treatment groups. They were also consistent with previously described mutations associated with reduced susceptibility to TVR [9] , thus minimising concerns from a clinical perspective in terms of the emergence of new variants with the TVR bid regimen.

The resistance profile observed in patients not achieving SVR in OPTIMIZE confirmed the previously described profile based on other Phase II and Phase III studies [9] . In both treatment groups, the majority of patients with on‐treatment VF during the TVR treatment phase had higher level TVR‐resistant variants; on‐treatment VF during the PR treatment phase was associated with higher (G1a) and lower level (G1a/1b) TVR‐resistant variants or wild‐type virus (G1b) in both treatment groups, and relapse was mostly associated with lower level TVR‐resistant variants or wild‐type virus in both treatment groups, especially in G1b patients.

An important observation when comparing both dosing regimens was that in patients with on‐treatment VF, there was no increase in the number of lower level TVR‐resistant variants in the TVR bid arm compared with the TVR q8h arm, indicating that TVR bid dosing resulted in a comparable selective pressure on lower level TVR‐resistant variants to TVR q8h dosing. In contrast, in the Phase III ADVANCE study 8 weeks of a TVR‐based regimen resulted in a slightly higher on‐treatment VF rate during the PR treatment phase (7%) than in the 12 weeks TVR treatment arm (4%); with more frequent emergence of wild‐type or lower level TVR‐resistant variants in the 8 weeks arm [2] . These earlier data suggested that 8 weeks of TVR‐based treatment was insufficient in some patients to fully clear lower level variants. Our results indicate that TVR bid dosing or q8h administration resulted in the selection of a similar type and frequency of resistant variants.

In this analysis, the low (<5%) prevalence of baseline TVR‐resistant variants previously observed in other Phase II and Phase III TVR studies [9] was also observed. The presence of pretreatment TVR‐resistant variants did not necessarily lead to treatment failure, with similar SVR rates in both treatment groups for patients with or without baseline TVR‐resistant variants. These findings provide further support for use of TVR bid administration by demonstrating that simplification of the regimen is unlikely to compromise resistance profiles. In addition, consistent with previous observations, an analysis has shown that resistance variants in the OPTIMIZE study were present in a similar fraction of patients who failed to achieve SVR in the TVR bid and TVR q8h dosing groups, regardless of liver disease stage (data not shown) [16] .

Due to the limitation of the population sequencing method used, low‐frequency mutations (<20%) were not detected in this study. Another limitation of our analysis was that sampling for viral sequencing was infrequent, and sequencing results may vary depending on the time the sample was obtained as the viral population is no longer under selective pressure after the end of TVR treatment. Additionally, the limited follow‐up period following treatment failure in the OPTIMIZE study precluded the collection of long‐term data from the analysis, although previous studies have shown that the virus shifts back towards a predominantly wild‐type population following treatment discontinuation [4] .

In summary, our findings indicate no difference in the viral resistance profiles selected in patients treated with TVR bid or q8h. TVR‐resistant variants emerging at the time of treatment failure were consistent with those previously observed regardless of TVR regimen. Taken together with the noninferior efficacy of TVR bid and q8h [12] , this finding further supports the use of the simplified, twice‐daily TVR dosing regimen for G1 HCV‐infected patients.

M. Buti has been a clinical investigator, speaker and/or consultant for Boehringer Ingelheim, Bristol‐Myers Squibb, Gilead Sciences, Janssen Pharmaceuticals, Merck Sharp & Dohme, Novartis and Vertex Pharmaceuticals Incorporated. A. Ghys, J. Witek, D. Luo, K. Janssen, B. Daems, G. Picchio and S. De Meyer are employees of Janssen and may be Johnson & Johnson stockholders. At the time of this study analysis, I. Dierynck was an employee of Janssen.

We express our gratitude to the patients who participated in the study, study centre staff, Janssen Pharmaceuticals personnel and Vertex Pharmaceuticals personnel. We acknowledge Sally Gray (Medical Writer, Gardiner‐Caldwell Communications, Macclesfield, UK) for assistance in drafting the manuscript and collating author contributions, which was funded by Janssen Pharmaceuticals.

This work was supported by Janssen Pharmaceuticals and Vertex Pharmaceuticals Incorporated.

Some data in this manuscript have been presented at the International Workshop on HIV & Hepatitis Virus Drug Resistance and Curative strategies, Toronto, Canada, June 4–8, 2013. However, these data have not been published in a paper and this manuscript is not being considered for publication elsewhere.

Graph: Proportion of all patients with TVR ‐resistant variants by treatment arm, grouped (A) overall, and by genotype: (B) G1a, (C) G1b. FC, fold change; G, genotype; SVR12, sustained virological response (HCV RNA <25 IU/ mL 12 weeks after the last planned dose of study drug); TVR, telaprevir.

Graph: image_n/jvh12347-fig-0001.png

Graph: TVR ‐resistant variants in patients with on‐treatment virological failure ( VF ) (A) during the TVR treatment phase or (B) during the peginterferon/ribavirin treatment phase by treatment arm and genotype. *V36M + R155K: the most frequently observed higher level combination. G, genotype.

Graph: image_n/jvh12347-fig-0002.png

Graph: TVR ‐resistant variants in patients with relapse by treatment group and genotype. G, genotype.

Graph: image_n/jvh12347-fig-0003.png