`
`
`CENTER FOR DRUG EVALUATION AND
`RESEARCH
`
`
`APPLICATION NUMBER:
`205834Orig1s000
`
`MICROBIOLOGY / VIROLOGY REVIEW(S)
`
`
`
`
`
`
`
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`Reviewer: Eric F. Donaldson, Ph.D.
`Date Submitted: 02/10/14
`Date Received: 02/10/14
`
`Date Assigned: 02/10/14
`
`Sponsor: Gilead Sciences, Inc.
` 333 Lakeside Drive
` Foster City, CA, 94404
`
`Initial Submission Dates:
`Correspondence Date: February 10, 2014
`CDER Receipt Date: February 10, 2014
`Assigned Date: February 10, 2014
`Review Complete Date: July 10, 2014
`PDUFA Date: October 10, 2014
`
`Amendments:
`SDN
`002
`
`Date Submitted
`02/10/2014
`
`Date Received
`02/10/2014
`
`Date Assigned
`02/10/2014
`
`Related/Supporting Documents: IND115268, IND106739, NDA204671
`
`Product
`Names
`Structures
`
`Sofosbuvir (GS-7977)
`
`Ledipasvir (GS-5885)
`
`Chemical
`Names
`
`(S)- Isopropyl 2-((S)-
`(((2R,3R,4R,5R)-5-(2,4-dioxo-
`3,4-dihydropyrimidin-1(2H)-yl)-4-
`fluoro-3-hydroxy-4-
`methyltetrahydrofuran-2-
`yl)methoxy)(phenoxy)
`phosphorylamino) propanoate
`
`Molecular
`formula
`Molecular
`weight
`
`C22H29FN3O9P
`
`529.46
`
`Drug category: Antiviral
`
`Methyl [(2S)-1-{(6S)-6-[5-(9,9-
`difluoro-7-{2-[(1R,3S,4S)-2-{(2S)-2-
`[(methoxycarbonyl) amino]-3-
`methylbutanoyl}-2-
`azabicyclo[2.2.1]hept-3-yl]-1H-
`benzimidazol-6-yl}-9H-fluoren-2-yl)-
`1H-imidazol-2-yl]-5-
`azaspiro[2.4]hept-5-yl}-3-methyl-1-
`oxobutan-2-yl]carbamate
`C49H54F2N8O6
`
`889.00 Da
`
`Indication: Fixed-dose combination of ledipasvir, a hepatitis C virus (HCV) NS5A inhibitor and sofosbuvir, an
`HCV uridine nucleotide analog NS5B polymerase inhibitor, which is indicated for the treatment of chronic
`hepatitis C virus genotype 1 infection
`
`1
`
`Reference ID: 3540431
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`Dosage Form/Route of administration: Oral
`
`Dispensed: Rx
`
`Abbreviations: BL, baseline; DAA, direct acting antiviral; EC50, effective concentration at 50%; FC, fold-
`change; FDA, Food and Drug Administration; GT, genotype; HCV, hepatitis C virus; HSA, human serum
`albumin; IC50, inhibitory concentration at 50%; IFN, recombinant human interferon ; mt, mitochondria; NGS,
`next generation sequencing; NAPI, nucleos(t)ide analog polymerase inhibitor; NNAPI, non-nucleoside analog
`polymerase inhibitor; NRTIs, nucleoside reverse transcriptase inhibitors; PBL, peripheral blood lymphocytes;
`PDVF, protocol defined virologic failure; PEG, pegylated human interferon; PR, protease; P/R, pegylated
`interferon/ribavirin; RAV, resistance-associated variant; RBV, ribavirin; SDM, site-directed mutants; SOF,
`sofosbuvir; SVR, sustained virologic response; WT, wild-type.
`Table of Contents
`EXECUTIVE SUMMARY ........................................................................................................................................................3
`BACKGROUND AND SUMMARY ......................................................................................................................................... 4
`Rationale for Requesting and Analyzing NGS Data............................................................................................................ 6
`NGS Data Analysis Pipeline................................................................................................................................................ 6
`NGS Analysis Parameters and Overview of Data Analysis ................................................................................................ 7
`NGS Analysis Pipeline Output............................................................................................................................................. 7
`NGS Data Comparison........................................................................................................................................................ 8
`CLINICAL STUDIES...............................................................................................................................................................9
`REVIEW OF PHASE 2 TRIALS.........................................................................................................................................10
`GS-US-337-0118 (LONESTAR) ....................................................................................................................................10
`GS-US-334-0118 Baseline Sequence Data ..................................................................................................................10
`GS-US-334-0118 Resistance Analyses in Subjects Experiencing Virologic Failure.....................................................12
`GS-US-337-0118 Phenotype Analyses .........................................................................................................................14
`GS-US-334-0118 Resistance Analyses Conclusions....................................................................................................15
`REVIEW OF P7977-0523 (FUSION).................................................................................................................................15
`P7977-0523 Baseline Sequence Data...........................................................................................................................16
`P7977-0523 Resistance Analyses in Subjects Experiencing Virologic Failure .............................................................16
`P7977-0523 Phenotype Analyses .................................................................................................................................18
`P7977-0523 Resistance Analyses Conclusions ............................................................................................................18
`REVIEW OF PHASE 3 CLINICAL TRIALS .......................................................................................................................18
`REVIEW OF GS-US-337-0102 (ION-1).........................................................................................................................18
`GS-US-337-0102 Baseline Sequence Data ..................................................................................................................18
`GS-US-337-0102 Resistance Analyses in Subjects Experiencing Virologic Failure.....................................................19
`GS-US-337-0102 Phenotype Analyses .........................................................................................................................19
`GS-US-337-0102 Resistance Analyses Conclusions....................................................................................................20
`REVIEW OF GS-US-337-0109 (ION-2) ............................................................................................................................20
`GS-US-337-0109 Baseline Sequence Data ..................................................................................................................20
`GS-US-337-0109 Resistance Analyses in Subjects Experiencing Virologic Failure.....................................................21
`GS-US-337-0109 Phenotype Analyses .........................................................................................................................23
`GS-US-337-0109 Resistance Analyses Conclusions....................................................................................................23
`REVIEW OF GS-US-337-0108 (ION-3) ............................................................................................................................24
`GS-US-337-0108 Baseline Sequence Data ..................................................................................................................24
`GS-US-337-0108 Resistance Analyses in Subjects Experiencing Virologic Failure.....................................................25
`GS-US-337-0108 Phenotype Analyses .........................................................................................................................27
`GS-US-337-0108 Resistance Analyses Conclusions....................................................................................................28
`
`Reference ID: 3540431
`
`2
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`COMBINED RESISTANCE ANALYSIS ............................................................................................................................28
`Combined Resistance Analysis Conclusions.................................................................................................................29
`METHODS ............................................................................................................................................................................30
`CONCLUSIONS....................................................................................................................................................................32
`POST MARKETING RECOMMENDATIONS.......................................................................................................................32
`ADMINISTRATIVE................................................................................................................................................................32
`Reviewer’s Signature(s) ....................................................................................................................................................32
`Concurrence ......................................................................................................................................................................32
`APPENDICES.......................................................................................................................................................................33
`APPENDIX 1: P7977-0523................................................................................................................................................33
`APPENDIX 2: GS-US-337-0109........................................................................................................................................34
`
`EXECUTIVE SUMMARY
`
`This review focused on the next generation sequencing (NGS) data provided in support of NDA 205834 for the
`fixed-dose combination (FDC) of ledipasvir (LDV) and sofosbuvir (SOF; LDV/SOF) indicated for the treatment
`of hepatitis C virus (HCV) genotype (GT) 1 infection. Overall, assessment of the NGS data by the Division of
`Antiviral Products (DAVP) indicated that the data and analysis provided by the sponsor, Gilead Sciences (GSI),
`was acceptable and this NDA is approvable with respect to virology.
`
`SOF (NDA 204671; approved December 2013) is a nucleotide prodrug of 2’-deoxy-2’-fluoro-2’-C-methyluridine
`monophosphate that is converted to the active uridine triphosphate form (GS-461203) within hepatocytes. It is
`an inhibitor of the NS5B RNA dependent RNA polymerase. In HCV replicon assays, the EC50 values of
`sofosbuvir against full-length replicons from genotype 1a, 1b, 2a, 3a and 4a, and chimeric 1b replicons
`encoding NS5B from genotype 2b, 5a or 6a ranged from 0.014 to 0.11 µM. The median EC50 value of
`sofosbuvir against chimeric replicons encoding NS5B sequences from clinical isolates was 0.062 µM for
`genotype 1a (range 0.029-0.128 µM; N=67), 0.102 µM for genotype 1b (range 0.045-0.170 µM; N=29), 0.029
`µM for genotype 2 (range 0.014-0.081 µM; N=15) and 0.081 µM for genotype 3a (range 0.024-0.181 µM;
`N=106). In infectivity assays, the EC50 values of sofosbuvir against genotype 1a and 2a viruses were 0.03 µM
`and 0.02 µM, respectively.
`
`LDV is a new molecular entity that inhibits HCV replication by interfering with the viral NS5A protein. It has
`antiviral activity against HCV genotype 1a and 1b replicons, with EC50 values of 0.031 nM and 0.004 nM,
`respectively. In addition, LDV has EC50 values ranging from 0.15 to 530 nM against genotypes 2 to 6
`replicons. LDV has an EC50 value of 21 nM against the GT2a JFH-1 replicon with L31 in NS5A, but has a
`reduced activity with an EC50 value of 249 nM against the GT2a J6 HCV strain with M31, a common
`resistance-associated substitution in GT 1. LDV has less antiviral activity compared to GT1 against genotypes
`4a, 5a, and 6a, with EC50 values of 0.39 nM, 0.15 nM and 1.1 nM, respectively. LDV has substantially lower
`activity against genotypes 3a and 6e with EC50 values of 168 nM and 264 nM, respectively.
`Data from three phase 3 studies, including Study GS-US-337-0102 (ION-1; treatment-naïve subjects), Study
`GS-US-337-0108 (ION-3; treatment-naïve non-cirrhotic subjects), and Study GS-US-337-0109 (ION-2;
`treatment-experienced subjects) and two phase 2 studies, including Study P7977-0532 (ELECTRON) and
`Study GS-US-337-0118 (LONESTAR) were submitted for resistance analyses. Cell culture selection
`experiments were performed using the HCV GT1a and GT1b replicon systems to identify resistance-
`associated substitutions that emerged in NS5A in response to LDV. These experiments, along with phenotypic
`assessments, showed that Q30E and Y93H were associated with resistance to LDV in the HCV GT1a replicon
`and Y93H was the predominant resistance-associated substitution in the GT1b replicon (see the review of
`Clinical Virology Reviewer Lisa Naeger, Ph.D. for complete details). In the phase 2 and phase 3 clinical trials,
`additional resistance-associated substitutions were identified and phenotyped by the sponsor. According to
`their analyses, NS5A resistance-associated substitutions Y93H, Y93N, Y93C, M28A, or H58D that emerge in
`3
`
`Reference ID: 3540431
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`HCV GT1a and A92K or Y93H that emerges in HCV GT1b confer >1,000-fold reductions in susceptibility to
`LDV in cell culture. The L31M, L31I, L31V, Q30H, Q30R, Q30G, or P32L substitutions that emerge in HCV
`GT1a and the P58D substitution that emerges in HCV GT1b are in the 100- to 1,000-fold resistance category.
`The K24R, K24G, K24N, M28T, Q30L, Q30T, S38F, A92T, or Y93F substitutions that emerge in GT1a and the
`L31M, L31V, L31I, or P32L substitutions that emerge in GT1b are in the <100-fold resistance category. Based
`on these results, the LDV resistance analysis focused on, but was not limited to, these NS5A positions. NS5A
`polymorphisms at amino acid positions K24, M28, Q30, L31, P32L, H58, A92, and Y93 were analyzed in the
`FDA virology resistance analysis. Substitutions or mixtures of substitutions at these NS5A positions were
`detected at baseline in 23% (370/1615) of the subjects in the phase 3 studies (ION-1, ION-2, and ION-3).
`
`For the virology analyses, relapse rates were used as the measure of efficacy outcome for the three phase 3
`studies and the two phase 2 studies. The overall relapse rate was 2.7% in all the studies submitted. In GT1a
`subjects, the relapse rate was 3% (41/1378). In GT1b subjects, the relapse rate was 1.7% (7/411). When the
`effect of individual baseline NS5A polymorphisms on relapse rates was examined, the highest relapse rates
`were seen in subjects with baseline polymorphisms at positions Q30, L31, and Y93 where relapse rates were
`6.6% (5/76), 10% (5/50), and 15% (8/54), respectively. Relapse rates for subjects with one baseline NS5A
`resistance-associated polymorphism were 3.6%, but were higher for subjects with 2 or 3 baseline NS5A
`resistance-associated polymorphisms with relapse rates of 9.5% and 9%, respectively.
`
`There were a total of 50 subjects (GT1a=42 and GT1b=8) who failed treatment with the FDC of LDV/SOF and
`who comprised the resistance analysis population that was analyzed by next generation sequencing. The most
`common substitutions associated with resistance to LDV (as determined comparing three variant detection
`algorithms and only counting those detected by two) were at positions Y93 (n=19; GT1a=15 and GT1b=4),
`Q30 (n=14; GT1a=14 and GT1b=0), M28 (n=10; GT1a=9 and GT1b=1), L31 (n=6; GT1a=6 and GT1b=0), and
`H58 (n=3, GT1a=3 and GT1b=0).
`
`For SOF resistance, there were several substitutions associated with resistance that had been identified in the
`review of SOF (NDA 204671), including positions S62 (n=23; GT1a=23 and GT1b=0), D61 (n=9; GT1a=9 and
`GT1b=0), E440 (n=8; GT1a=1 and GT1b=7), V321 (n=3; GT1a=2 and GT1b=1), L159 (n=1; GT1a=1 and
`GT1b=0), S282 (n=1; GT1a=1 and GT1b=0), and L320 (n=1; GT1a=1 and GT1b=0) that emerged in these
`studies. In addition, two additional amino acid positions had substitutions that were treatment emergent,
`including A112 (n=3, GT1a=3 and GT1b=0) and E237 (n=2; GT1a=2 and GT1b=0). Of note, in this dataset,
`substitutions at positions HCV GT1a NS5B_S62 and HCV GT1b NS5B_E440 appeared to be polymorphic and
`not associated with resistance as compared to the SOF dataset that was used to identify these positions (SOF
`NDA 204671, original review and addendum). However, the D61G substitution was treatment emergent and
`detected in the NS5B HCV protein of several subjects infected with HCV GT1a who failed treatment with the
`LDV/SOF FDC. This same substitution was detected and associated with treatment failure among subjects
`infected with HCV GT1a in the Liver Pre-Transplant Study P7977-2025 (reviewed in SOF NDA 204671
`addendum). Additional substitutions that should be phenotypically evaluated for SOF resistance include,
`NS5B_A112T, NS5B_E237G, and NS5B_S473T.
`
`BACKGROUND AND SUMMARY
`
`Sofosbuvir is a nucleotide prodrug of 2’-deoxy-2’-fluoro-2’-C-methyluridine monophosphate that is converted to
`the active uridine triphosphate form (GS-461203) within the hepatocyte. In HCV replicon assays, the EC50
`values of sofosbuvir against full-length replicons from genotype 1a, 1b, 2a, 3a and 4a, and chimeric 1b
`replicons encoding NS5B from genotype 2b, 5a or 6a ranged from 0.014 to 0.11 µM. The median EC50 values
`of sofosbuvir against chimeric replicons encoding NS5B sequences from clinical isolates were 0.062 µM for
`genotype 1a (range 0.029-0.128 µM; N=67), 0.102 µM for genotype 1b (range 0.045-0.170 µM; N=29), 0.029
`µM for genotype 2 (range 0.014-0.081 µM; N=15) and 0.081 µM for genotype 3a (range 0.024-0.181 µM;
`
`Reference ID: 3540431
`
`4
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`N=106). In infectivity assays, the EC50 values of sofosbuvir against genotype 1a and 2a viruses were 0.03 µM
`and 0.02 µM, respectively.
`
`Ledipasvir inhibits HCV replication by interfering with the viral NS5A protein. It has antiviral activity against
`HCV genotypes 1a and 1b replicons, with EC50 values of 0.031 nM and 0.004 nM, respectively. In addition,
`LDV has EC50 values ranging from 0.15 to 530 nM against genotypes 2 to 6 replicons. LDV has an EC50 value
`of 21 nM against the GT2a JFH-1 replicon with L31 in NS5A, but has a reduced activity with an EC50 value of
`249 nM against the GT2a J6 HCV strain expressing M31, a common resistance substitution. LDV has less
`antiviral activity compared to GT1 against genotypes 4a, 5a, and 6a, with EC50 values of 0.39 nM, 0.15 nM and
`1.1 nM, respectively. LDV has substantially lower activity against genotypes 3a and 6e with EC50 values of
`168 nM and 264 nM, respectively.
`
`Data from three phase 3 studies, including Study GS-US-337-0102 (ION-1; treatment-naïve subjects), Study
`GS-US-337-0108 (ION-3; treatment-naïve non-cirrhotic subjects), and Study GS-US-337-0109 (ION-2;
`treatment-experienced subjects), and two phase 2 studies, including Study P7977-0532 (ELECTRON) and
`Study GS-US-337-0118 (LONESTAR) were submitted for resistance analyses. The sponsor provided next
`generation sequencing (NGS) data that were used in the resistance analysis of the five clinical trials (Table 1).
`
`Table 1. Phase 2 and 3 LDV/SOF and LDV + SOF studies analyzed for resistance by NGS (Table 1, page 15,
`Integrated Phase 2&3 Virology Study Report).
`
`The sponsor provided the NGS data on a hard drive and the dataset included: 1) frequency tables showing
`amino acid variation that occurred at each position of 3 viral proteins (NS3/4A, NS5A, and NS5B) for each
`failure sample that was successfully sequenced using Illumina; 2) raw sequence data in fastq format for all
`samples that were deep sequenced; 3) summary resistance data for each study; and 4) cross study
`comparisons of resistance data.
`Given that next generation sequencing is an emerging technology with no current standards for analysis, the
`division requested raw data so that an independent analysis could be performed on the NGS data. The
`sponsor’s summary NGS data were compared to the results generated by DAVP following these criteria:
`1. The sponsor’s frequency tables were used to generate a summary and do a direct comparison of
`the results reported by the sponsor;
`2. Frequency tables were generated by DAVP using an independent mapping of reads to a reference
`for each sample and using two independent variant detection algorithms and the results were
`compared with those reported by the sponsor and those generated using the sponsor’s frequency
`table; and
`3. The conclusions from the NGS data were compared to the results reported by the sponsor using
`Sanger population sequence analysis when applicable.
`5
`
`Reference ID: 3540431
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`Rationale for Requesting and Analyzing NGS Data
`In general, the FDA does not analyze raw nucleotide sequence data in conjunction with new drug applications
`(NDAs); however, when the technology used to generate the data is relatively new, it is necessary to perform
`independent assessments of the data to confirm that the review division understands how the data are
`interpreted by the sponsor. NGS is an emerging technology that presents many potential data integrity issues
`that must be considered upon careful review:
`1. There are currently multiple sequencing platforms available for resistance analysis by NGS (454,
`Illumina, Ion Torrent, PacBio), and these technologies are continuously emerging. Each platform
`has different error rates and chemistries that contribute to unique types of base calling errors.
`2. There are currently no standardized analysis pipelines with which to analyze NGS data and more
`than 200 algorithms can be used to generate an assembly of small reads, with each algorithm
`employing unique strategies and using unique parameters. Comparison of different platforms and
`algorithms has shown that often differences in data interpretation are attributed to the bioinformatics
`analysis and not the sequencing platform.
`3. To date, each sponsor submitting NGS data has generated data with unique NGS analysis
`pipelines that use internal scripts and programs that are not currently available in the public domain.
`
`Providing accurate resistance information is imperative for protecting public health to prevent emergence of
`novel resistant and cross-resistant viral variants that have the potential to infect others and cause major
`outbreaks of disease that cannot be controlled by approved drugs. In addition, the resistance information
`provides important guidance for health care professionals who oversee the use of these therapeutics and is
`included in the drug product information approved by DAVP.
`Because it determines the sequence for all RNAs or DNAs in a clinical sample, NGS adds complexity to the
`resistance analysis process while reducing sequencing costs. In contrast to Sanger DNA sequencing which
`provides an average sequence of the virus population, NGS provides nucleotide sequence information for
`individual viruses within a virus population, potentially providing millions of short sequences per sample. The
`complexity of the data makes it challenging for virology reviewers to analyze and validate the sequence
`information, which is complicated by the fact, as mentioned above, that there are currently no standard
`bioinformatics analysis approaches for analyzing these large datasets. Moreover, nearly every sponsor
`performing NGS has developed their own proprietary bioinformatics analysis pipeline. Given that there are over
`two hundred assembly algorithms alone, it is expected that each pipeline will provide a unique interpretation of
`the data.
`
`Currently, industry is rapidly adopting the use of NGS technology in support of product development and
`application submissions. This has created unique review challenges for CDER where no NGS data
`analysis/review capabilities had previously existed. To address this gap in the review process which could
`have a significant impact on public health, DAVP teamed up with CDER’s Computational Science Center to
`develop an independent NGS analysis pipeline that would allow virology reviewers to perform a robust and
`independent analysis of NGS resistance datasets submitted in support of antiviral drugs in development.
`
`NGS Data Analysis Pipeline
`DAVP worked with the Office of Scientific Computing within CDER to acquire the resources to analyze NGS
`data for review purposes. The CLC Genomics Workbench was installed for use on the High Performance
`Computer at CDRH and was used to establish an analysis pipeline for independently analyzing NGS data.
`CLC Genomics was used to evaluate each of the sequence runs, trim and filter the sequences prior to
`mapping, and to map the sequences to HCV GT1a and GT1b reference sequences. Two independent variant
`detection algorithms were used to call variants from each mapping, and the variant tables were exported from
`CLC Genomics Workbench and combined to generate frequency tables and resistance summary tables
`(Figure 1).
`
`Reference ID: 3540431
`
`6
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`Figure 1. An overview of the NGS analysis Pipeline using CLC Genomics Workbench.
`
`NGS Analysis Parameters and Overview of Data Analysis
`Each step of the analysis process is briefly described below. For a more detailed description, please see the
`SOF NDA review (NDA 20467 SDN 004).
`
`1. Processing fastq files with CLC Genomics Workbench. Data were received on a portable hard drive,
`which included fastq files for each subject and timepoint that was sequenced using the Illumina platform.
`The sequences were uploaded via the CLC Genomics interface, using the Illumina specific criteria. Failed
`reads were removed, read names were discarded, and Quality scores were calculated using the
`NCBI/Sanger (Illumina Pipeline 1.8) option.
`2. Segregating sequences by HCV genotype and trimming the sequence reads prior to mapping. The
`fastq files were separated by genotype and subtype and the NS5A and NS5B genes for HCV GT1a (H77)
`and GT1b (Con1) were imported and annotated as coding sequences to be used as reference sequences
`for mapping. The individual reads from each fastq file were subjected to trimming using the default
`parameters for CLC Genomics Workbench.
`3. Mapping reads to the appropriate reference sequence for each HCV genotype/subtype. The reads
`from each fastq file were aligned to the appropriate reference sequence to generate a mapping for each
`timepoint. The mapping contained the target of interest (the NS5A and NS5B gene sequences) and was
`used to generate a consensus sequence for each sequence run. The consensus sequences were
`conceptually translated to amino acid sequences to compare changes that occurred at the amino acid
`level. In general, the mappings were assessed to determine the depth of coverage at each nucleotide
`position and to evaluate read directionality (ratio of forward to reverse reads) to identify regions of bias.
`
`NGS Analysis Pipeline Output
`4. Generating frequency tables of amino acid substitutions. From the read mappings, two algorithms
`were used to call variants based on independent criteria, and variant tables were generated for each
`sequence run and variant detection method. The variant tables included the following column headers:
`Reference Position, Type, Length Reference, Allele Linkage, Zygosity, Count Coverage, Frequency,
`Forward/reverse balance, Average quality, Overlapping annotations, Coding region change, and Amino
`acid change. The two variant detection systems employed different strategies for calling variants, and the
`variant detection parameters were relaxed from default to maximize the number of variants called, given
`that true variants would likely be identified in multiple subjects, allowing those that were of low quality or
`probability to be filtered out at the analysis stage. The two detection methods were:
`
`1. Probabilistic Variant Detection (PVD75) – calls variants from a read mapping using a probabilistic
`model (combines a Bayesian model and a Maximum Likelihood approach to calculate prior and error
`probabilities). Parameters are calculated on the mapped reads without considering the reference
`7
`
`Reference ID: 3540431
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`sequence. The variant probability parameter was reduced from a default value of 90 to 75 to increase
`the number of variant calls, given that false calls would likely be filtered during data analysis.
`
`2. Quality-based Variant Detection (QbVD) - based on the Neighborhood Quality Standard algorithm, it
`uses a combination of quality filters and user-specified thresholds for coverage and frequency to call
`variants covered by aligned reads.
`
`Frequency tables were generated by exporting the variant tables for both variant detection
`methods (PVD75 and QBVD) for each mapping and then reformatting the data to reflect
`variation at the amino acid level with these pertinent changes:
`1. The variant tables were combined by genotype/subtype and study
`2. The variant tables were filtered to remove synonymous substitutions
`3. The variant tables were reformatted to be directly comparable to the frequency tables
`submitted by the sponsor
`
`5. Generating resistance analysis tables. An ETL/Kettle script was used to convert the frequency
`tables into resistance analysis tables, allowing the resistance tables to be populated using different
`frequency thresholds. For example, the frequency tables generated from CLC Genomics
`Workbench output or submitted by the sponsor contained all variants with a frequency greater than
`or equal to 1%, and this tool allowed resistance analysis tables to be generated showing variants at
`different levels of sensitivity (5%, 15%, 25%, etc.) as defined by the user.
`6. Conducting independent resistance analysis. The frequency tables and resistance analysis
`tables were then analyzed to identify substitutions that occurred above a defined frequency
`threshold of 10%, using the following criteria:
`a. SUBS10 criteria – Identified all substitutions that were not detected at baseline (<0.01
`frequency) but were detected at a frequency of 0.10 or greater at later timepoints or
`detected at baseline at a frequency of 0.10 and not detected at later timepoints.
`
`NGS Data Comparison
`7. Comparing results to those submitted by the sponsor. The remainder of this review provides
`details on how the NGS data submitted by the sponsor were independently evaluated using the
`above described NGS analysis pipeline. In general, the NGS data analysis was performed using
`data generated in this pipeline and provided by the sponsor, and the results were compared as
`follows:
`a. Frequency and resistance analysis tables were compared directly and major differences
`were noted
`b. Amino acid substitutions were identified by the three algorithms (the sponsor’s algorithm
`(GIL) and QbVD and PVD75 used by DAVP) and major differences between algorithms
`were reported
`c. Novel resistance-associated amino acid substitutions reported by different NGS analysis
`approaches were compared and major differences were reported
`d. NGS analysis results were compared to results obtained and reported by the sponsor
`using Sanger population sequencing when applicable
`e. Novel resistance-associated substitutions identified by the independent analysis were
`noted and discussed with the review team for potential labeling/post-marketing actions
`
`Reference ID: 3540431
`
`8
`
`
`
`DIVISION OF ANTIVIRAL PRODUCTS (HFD-530)
`VIROLOGY REVIEW: Eric F. Donaldson, Ph.D.
`NDA#: 205834 SDN 002 DATE REVIEWED: 06/27/2014
`
`CLINICAL STUDIES
`
`The sponso