Genotype-Independent Near Whole Genome Next Generation Assays for HCV Resistance Evaluation
Anita Howe, Ph.D. Centre for Excellence in HIV/AIDS British Columbia, Canada
How do we test for HCV RAVs? A Technology Based Presentation
Objectives
1. Overview of key virology assays
2. Sanger population sequencing and RECall
3. Near Whole Genome HCV Next Generation Sequencing (NGS)
4. Random Primer NGS assay for Mixed Infection
5. Probe Enrichment
Key HCV Assays for Resistance Testing
Genotyping Line-Probe Hybridization VERSANT® HCV GENOTYPE 2.0 [LIPA] (Innogenetics)
RT-PCR ABBOTT REAL-TIME HCV GENOTYPE II (Abbott) Direct Sequencing TRUGENE DIRECT DNA SEQUENCING (Bayer/Siemens) Serotyping MUREX HCV SEROTYPING (Abbott) Ø 5’UTR/Core/NS5B Ø GT1 – 6 Ø Limited subtype
information
Sequencing
Sanger Population Sequencing High throughput LOD ~20% Clonal Sequencing Labor intensive Linkage of mutations Allele-specific Real-time PCR Limit to known mutations Next Generation Sequencing Illumina, Ion Torrent, 454 (Roche), PyroMark (Qiagen), ABI SOLiD, SMRT (Pac Bio) Ø Sensitive Ø Medium-high throughput Ø High costs
Phenotypic Assays
F.L. Stable Replicons • GT1a_H77 • GT1b_con1 • GT 2a_JFH1 • GT 3a_S52 • GT 4a_ED43 • GT 5a_SA1 • GT 6a_consensus
Chimeric/Transient Replicons in • GT1a_H77 • GT1b_con1 • GT2a_JFH1
Infectious HCV • GT1a_H77 • GT 2a_JFH1 Reporter Assay SEAP for NS3 Enzymatic assays NS3, NS5B
Viral Load Branched DNA VERSANT® HCV RNA 3.0 branched DNA (Bayer/Siemens) RT-PCR • ABBOTT REAL-
TIME HCV RT-PCR (Abbott)
• HCV SUPERQUANT (National Genetics Institute) • COBAS AmpliPrep/COBAS TaqMan HCV TEST (Roche Molecular Systems) Transcription-Mediated Amplification VERSANT HCV RNA (Siemens) Ø 5’UTR/Core Ø 9.6 – 108 IU/mL
RECall Web Based Sequence Analysis
http://pssm.cfenet.ubc.ca/account/login Woods et al. (2012) JCM 50.6:1936-1942
Sanger Population Sequencing
Genotype Switch in Treatment Naïve PWIDS Observed Between 1999 – 2004 (mean = 4.2 yr)
1.0
Proportion of infections in British Columbia, Canada that displayed reduced susceptibility to simeprevir
# Te
sts
per m
onth
Uptake of Q80K Screening (Feb 2014 – March 2015)
Joy et al. (2016) 5th Can Hep C Symposium. Montreal. Canada
Illumina Next Generation Sequencing
Host RNAs and DNAs
HCV RNA
http://www.illumina.com/technology/next-generation-sequencing/paired-end-sequencing_assay.html
Target N.A. amplification
Cluster Amplification on a Flow Cell
Sequencing by Synthesis with FL-NTPs Image Paired-end sequences
Library Preparation
Data Analysis
HCV GT/Subtype References from Genbank
Sample-specific Consensus
Active Seed References
1a
1b
3a
4d
Preliminary “seeding”
Re-map sample sequences using the sample-specific consensus as references
1a
4d
1b
3a
Illumina Paired-end reads in FASTQ
Remove poor consensus sequences and seed references
MiCall Pipeline
Muta%on Report FDA references e.g. 1b_R30H
Illumina Paired-end reads in FASTQ
Genotype-Independent Near Whole Genome HCV NGS Assays
WG-‐1 amplicon 288 9250
Oligo dA20 RT-primer
WG-‐2 amplicon 288 8640
Limitations of Assays used for HCV Resistance: • Genotype-dependent; prior GT knowledge is needed to generate sequences • One target gene at a Time =>$$$ and long turn around time • Inaccurate genotype/subtype identification
Accuracy Whole Genome NGS vs. Sanger Sequencing
• Overall mean nucleotide, amino acid and Q80K concordance are 98.8%, 99.6% and 100%, respectively.
NGS Basecall S
ange
r Bas
ecal
l
Limit of Quantification
100 plasma samples. All sequenced positions within HCV
Coefficient of variation
CV dramatically increases in minority variants with <0.1% mean prevalence
2 pure plasmid clones at conserved positions
The lower limit of quantification is 0.5%
variants with 0.5% prevalence = noise
Error Rate = 100% - frequency of the most common a.a. at that position
How many reads do we need for detecting variants with a 2% prevalence?
Read Coverage Read Coverage
A A
Freq
uenc
y Fo
ld-D
iffer
ence
from
Mea
n
Variant Frequency < 0.5% 0.5% < Variant Frequency < 2%
2% < Variant Frequency < 20% Variant Frequency > 20%
Sequence Coverage of HCV Genes
Genotype Spectrum GT1 - 6 plasma samples with viral load from 3.5 to >7 IU/mL
Genotype∆ # Samples ABempted GT Subtypes NS3
# Passing (%) NS5a
# Passing (%) NS5b
# Passing (%)
1 78 1a 72 (92%) 72 (92%) 72 (92%) 11 1b 11 (100%) 11 (100%) 11 (100%) 3 1e 2 (67%) 2 (67%) 2 (67%)
2 20 2, 2a, 2b, 2c 19 (95%) 19 (95%) 19 (95%)
3 76* 3, 3a 62 (82%) 62 (82%) 60 (79%)
4 20* 4, 4a, 4d, 4e, 4r, 4N, 4t 16 (80%) 13 (65%) 8 (40%)
5 3 5a 3 (100%) 3 (100%) 3 (100%)
6 21 6a, 6e, 6h, 6k, 6l, 6t 21 (100%) 20 (95%) 18 (86%)
*~50% of the samples did not have viral load information
Viral Load
log10 HCV RNA (IU/mL) NS3 NS5A NS5B
>7 25/25 (100%) 25/25 (100%) 25/25 (100%)
6.6 – 7.0 33/34 (97%) 33/34 (97%) 33/34 (97%)
6.1 – 6.5 28/28 (100%) 28/28 (100%) 28/28 (100%)
5.6 – 6.0 26/26 (100%) 26/26 (100%) 26/26 (100%)
5.1 – 5.5 21/25 (84%) 21/25 (84%) 16/25 (64%)
3.5 -‐ 5.0 8/18 (44%) 9/18 (50%) 8/18 (44%)
GT1 – GT6 Clinical Samples
Success rates for serum samples or samples with multiple freezing and thawing cycles might be lower
Receiver Operator Characteristic
• Threshold for variant calls• RAVs selected for analysis • EC50 fold-shifts etc.
Optimal Conditions for RAV Analysis
Test (threshold)
disease (e.g. non-SVR) without disease (e.g. SVR)
Non-‐SVR
SVR
+RAV TP FP
-‐RAV FN TN
Non-SVR SVR
TP /
(TP
+ FN
)
TN / (FN+TN)
Any polymorphisms at resistance loci
Adding polymorphic variants with no resistance
Condition 1: NS5A RAVs: M28A/G/T/V, Q30D/E/G/H/R/L, L31I/M/F/V, H58D and Y93any Condition 2. as condition 1 but skip M28V, Q30H/L, L31M Condition 3: any polymorphisms from GT1a_H77 i.e. M28Any, Q30any, L31any, H58any, Y93any Condition 4: as condition 1 but add K24any, A92any, R44any and R78any
Removing impactful RAVs
How does the choice of RAVs affect ROC?
Clinical Samples
• 97 samples were examined between September 2015 – March 2016 - 63 GT1a, 13 GT1b, 7 GT2, 12 GT3, 2 GT4 and 1 GT6
• Treatment status unknown but majority were likely from virologic failures • >60% of the samples had NS5A RAVs; 80% of which were resistant to ALL
approved NS5A drugs
Cunningham, E. B. et al. (2015) Mixed HCV infection and reinfection in people who inject drugs—impact on therapy Nat. Rev. Gastroenterol. Hepatol. doi:10.1038/nrgastro.2015.36
Mixed Infection Among PWIDs
Random Priming Next Generation Assay
Random Primer
• The Random Priming NGS assay accurately identified mixed-infected samples containing GT1 - 6 genotypes/subtypes
• Linearity of detection was observed in mixed-infected samples with 10:90, 50:50 and 90:10 ratios
• Good reproducibility: the average difference between replicates was <1%
Ratio was expressed in the same order of the GT mixes e.g. 10:90 for 1a vs.1b
Genotypes and Reproducibility Random Priming NGS
Mixtures of plasma samples at 10:90, 50:50 and 90:10 nominal ratios. Viral load ~5 log10 IU/mL 2 replicates from the mixtures at 10:90 were evaluated for reproducibility
A Natural Mixed Infection Case Detected by Random Priming NGS
Background: This is a patient who was thought to be infected with GT1b as determined by LiPA and Sanger Population Sequencing GT2b was identified by Sanger Population Sequencing at virologic failure after DAA treatment Subsequent sequencing using Random Priming NGS showed that this subject in fact had a mixed infection at baseline. GT1b was cleared but GT2b remained
GT1b GT 2b
GT2b
Temporal viral dynamics in a Subject with mixed infection Random Priming NGS
GT1a GT3a
GT1a
GT3a
GT1a
GT3a
GT3a GT1a
GT1a
Use of Capture Probes to Enrich Target Sequences
sequencing library of randomly primed cDNA
= human & others = HCV
HCV-specific probes (biotinylated)
magnetic beads, streptavidin coated
2. bind beads to probes
3. capture probe-bound beads with magnet
1. hybridize probes to target
sequences
Enrichment of HCV Sequences
% of reads mapping to
reference genomes
3x4−HC
V_S
12
2x6−HC
V_S
11
61518A−H
CV
_S10
61525A−H
CV
_S9
5x6−HC
V_S
8
4x6−HC
V_S
7
4x5−HC
V_S
6
pool3−HC
V_S
5
pool2−HC
V_S
4
pool1−HC
V_S
3
61516A−H
CV
_S2
61515A−H
CV
_S1
0
25
50
75
100
proportion of total reads hitting references
phiX174
Ecoli
Pacneshg38H
CV
HIV
1G
BvirusC
not_referenced
15_Dec_02_checkM
iseq proportion of total reads hitting references
3x4−HCV_S12
2x6−HCV_S11
61518A−HCV_S10
61525A−HCV_S9
5x6−HCV_S8
4x6−HCV_S7
4x5−HCV_S6
pool3−HCV_S5
pool2−HCV_S4
pool1−HCV_S3
61516A−HCV_S2
61515A−HCV_S1
0 25 50 75 100
proportion of total reads hitting references
phiX174EcoliPacneshg38HCVHIV1GBvirusCnot_referenced
15_Dec_02_checkMiseq proportion of total reads hitting references
phiX174
E. coli Propionibacterium acnes (skin bacterium)
human (hg38) HCV HIV GB virus C not referenced
56
58
7A−
HC
V_
S2
56
58
5A−
HC
V_
S1
0
25
50
75
100
proportion of total reads hitting references
ph
iX
17
4E
co
li
Pd
en
itrifica
ns
Pa
cn
es
hg
38
HC
Vn
ot_
re
fe
re
nce
d
15_Ju
l_24_S
1S
2_ch
eckM
iseq
p
ro
po
rtio
n o
f to
tal read
s h
ittin
g referen
ces
56587A−
HC
V_S
2
56585A−
HC
V_S
10
25
50
75
100
proportion of total reads hitting references
phiX
174
Ecoli
Pdenitrificans
Pacnes
hg38
HC
Vnot_referenced
15_Ju
l_24_S
1S
2_ch
eckM
iseq
p
ro
po
rtio
n o
f to
tal read
s h
ittin
g referen
ces
reference genomes
1 samples: 2 3 4 5 6 9 7 10 8 11 12 13 14 0
25
50
75
100
using HCV capture probes without probes
3x4−HCV_S12
2x6−HCV_S11
61518A−HCV_S10
61525A−HCV_S9
5x6−HCV_S8
4x6−HCV_S7
4x5−HCV_S6
pool3−HCV_S5
pool2−HCV_S4
pool1−HCV_S3
61516A−HCV_S2
61515A−HCV_S1
0 25 50 75 100
proportion of total reads hitting references
phiX174EcoliPacneshg38HCVHIV1GBvirusCnot_referenced
15_Dec_02_checkMiseq proportion of total reads hitting references
56587A−HCV_S2
56585A−HCV_S1
0 25 50 75 100
proportion of total reads hitting references
phiX174EcoliPdenitrificansPacneshg38HCVnot_referenced
15_Jul_24_S1S2_checkMiseq proportion of total reads hitting references
Paracoccus denitrificans (probable reagent contaminant)
Coverage in HCV NS5A with vs. without Probe Enrichment GT1a and GT3a in 90:10 mix
Rea
d D
epth
Total NS5A Reads 4,377,073 797,111 742,769 66,116 Fold Increase 5.5 11.2 Mean Reads 9,814 1,787 1,651 147 Maxium Reads 18129 3190 4046 221 Minium Reads 3286 945 568 59
NS5A Amino Acid Position
GT1a + Probe GT3a + Probe GT1a GT3a
• The near Whole Genome NGS assay represents an efficient tool for the evaluation of HCV resistance in all genotypes
• The lower limit of minority variant detection (threshold) is 0.5%
• Viral titer and quality of the samples are important factors that determine sequencing success
• ROC can be used to optimize the thresholds/RAVs used for prediction of treatment outcomes
• Random Primer NGS provides an agnostic approach to interrogate potential mixed-infections
• Sensitivity of the Random Primer NGS assay may be improved with the use of capture probes
Summary
Acknowledgement
Centre for Excellence in HIV/AIDS • Chanson Brumme • Winnie Dong • Celia Chui • Weiyan Dong • Jeff Joy • Art Poon • Don Kirby • Vera Tai • Conan Woods • Richard Harrigan
Merck Research Laboratory • Ping Qiu • Wei Bo • Rick Stevens
Vancouver VIDUS Cohort Patients and Investigators