ApT

Sa

b

ARRAA

KTBHP

1

aae(o(eiasswe

t7

(

0d

Journal of Virological Methods 153 (2008) 156–162

Contents lists available at ScienceDirect

Journal of Virological Methods

journa l homepage: www.e lsev ier .com/ locate / jv i romet

nalysis of HCV resistance mutations during combination therapy withrotease inhibitor boceprevir and PEG-IFN �-2b usingaqMan mismatch amplification mutation assay

tephanie Currya, Ping Qiub, Xiao Tonga,∗

Virology, Schering-Plough Research Institute, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United StatesBioinformatics, Schering-Plough Research Institute, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States

rticle history:eceived 5 May 2008eceived in revised form 17 July 2008ccepted 22 July 2008vailable online 11 September 2008

eywords:

a b s t r a c t

TaqMan Mismatch Amplification Mutation Assay (TaqMAMA) is a highly sensitive allelic discriminationmethod. The mismatch amplification mutation assay (MAMA) is based on preferential amplification ofmutant allele by the ‘MAMA’ primer, which is designed to have two mismatches with the wild-type alleleand only one mismatch with the mutant allele. In this report, the TaqMAMA method was adapted forthe detection and quantitation of minor HCV variants resistant to the protease inhibitor boceprevir (SCH503034) from clinical samples. A good correlation of mutant frequency was observed between TaqMAMAand the results of clonal sequencing. TaqMAMA detected consistently minor variants at a level as low as

aqMAMAoceprevirCV resistancerotease inhibitor

0.1%. Using TaqMAMA, it was demonstrated that resistant variants existed in the viral population beforeboceprevir treatment. The frequency of two resistant mutants (T54A and V170A) increased significantlyduring treatment with boceprevir, but was suppressed by combination treatment of PEG-IFN �-2b andboceprevir. The prevalence of both mutants decreased at the end of the two-week follow-up period. Theseresults show that TaqMAMA can be used to detect minor resistant variants in pretreatment samples and

ution

atc

ttipTnpa

to study in detail the evol

. Introduction

Resistance to targeted antiviral treatment has been recognizeds a major determinant of clinical efficacy for the treatment of hep-titis C virus (HCV) infections. In infected patients, due to the highrror rate of viral polymerase and rapid turnover rate of HCV virionsNeumann et al., 1998), HCV exists as quasispecies which consistsf a population of genetically distinct but closely related variantsCabot et al., 2001; Farci et al., 2002; Herring et al., 2005; Martellt al., 1992). Given the large pool of variants in the quasispeciesn an infected patient, resistance mutations to a specific antiviralgent are predicted to be pre-existing in the absence of drug-related

elective pressure. In fact, the A156T resistance mutation has beenhown recently to be present in a patient with chronic hepatitis Cho was never treated with anti-NS3-protease inhibitors (Cubero

t al., 2008). Treatment with an antiviral drug allows rapid selection

∗ Corresponding author at: Virology, MS4945, Schering-Plough Research Insti-ute, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States. Tel.: +1 980 740446; fax: +1 908 740 3032.

E-mail addresses: stephanie.curry@spcorp.com (S. Curry), ping.qiu@spcorp.comP. Qiu), xiao.tong@spcorp.com (X. Tong).

ecvcRs2

aa3

166-0934/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.jviromet.2008.07.020

of mutant viruses during targeted antiviral therapy.© 2008 Elsevier B.V. All rights reserved.

nd outgrowth of viral variants with reduced susceptibility to thereatment, which often results in rebound of viral load and reducedlinical efficacy.

The HCV NS3 protease, an essential component of HCV replica-ion in host cells (Kolykhalov et al., 2000), has become an importantarget for the development of anti-HCV therapies. The HCV genomes translated as a single polyprotein precursor which has to berocessed by cellular and viral proteases into functional proteins.he NS3 protease is responsible for the processing of most of theon-structural proteins (NS3, NS4A, NS4B, NS5A, and NS5B). Therotease inhibitor boceprevir (SCH 503034) binds to the enzymective site and inhibits cleavage of the viral polyprotein (Malcolmt al., 2006). It has been under clinical investigation in mono- andombination therapy regimens. Resistance mutations to bocepre-ir have been identified in the HCV replicon system as well as inlinical trials. The major resistance loci are V36, Q41, F43, T54,155, A156 and V170, which are located near the inhibitor-bindingite (Sarrazin et al., 2007; Tong et al., 2006, 2008; Zeuzem et al.,

005a,b).

As reported recently (Sarrazin et al., 2007), virological responsend safety parameters of combination of boceprevir and PEG-IFNlfa-2b (PEG-IFN �-2b) were tested in an open-label, randomized,-way crossover Phase 1b study in patients with genotype 1 HCV

ogical

iBrp

iprvoanaatdrvsptei

2

2

spl2ia�btataofj

2

QtpSfstrfPG3oGat

2

Cpsp6PmdaNc

2

sFdbmmpacd

cn

MTG-

GCGT

GTCT

94p(r

2

drr

S. Curry et al. / Journal of Virol

nfection who had failed previously to respond to PEG-IFN �-2b.y direct population sequencing analysis of the HCV NS3 protease,esistance mutations at amino acid T54 of NS3 were detected in oneatient (patient 105) during the trial (Zeuzem et al., 2005b).

With more anti-HCV agents entering clinical trials, quantify-ng minor mutant variants and analyzing their distribution mayrovide information to support selection of appropriate antivi-al regimens that may minimize the development of resistantiruses. To develop a highly sensitive method for the analysisf minor mutants, a real-time PCR based allelic discriminationssay known as TaqMAMA was adapted to detect known single-ucleotide mutations that confer resistance to specifically targetedntiviral therapies. TaqMAMA combines the TaqMan® fluorogenicssay (Heid et al., 1996) with the mismatch amplification muta-ion assay (MAMA) (Cha et al., 1992) to provide extremely sensitiveetection of known mutations (Glaab and Skopek, 1999). Recentesearch has shown that optimization of MAMA primers pro-ides significant discrimination between wild type and mutantequences (Li et al., 2004). In the present study, TaqMAMA waserformed on serum samples of patient 105, the presence ofhe T45A and other known boceprevir resistance mutations wasvaluated and compared to results obtained by clonal sequenc-ng.

. Materials and methods

.1. Treatment of patient 105 and collection of samples

It was an open-label, randomized, 3-way crossover Phase 1btudy in patients with genotype 1 HCV infection who had failedreviously to respond to PEG-IFN �-2b. The design, safety and viro-

ogical response of the study have been described (Sarrazin et al.,007). Patient 105 was randomized to receive antiviral treatment

n the following order: PEG-IFN �-2b (1.5 �g/kg weekly) monother-py for two weeks, followed by combination therapy with PEG-IFN-2b plus boceprevir (200 mg TID) for two weeks, followed byoceprevir monotherapy (200 mg TID) for one week. The first tworeatment periods were followed by two-week washout periodsnd the patient was followed up for two weeks after the finalreatment period. Serum samples were collected for sequencingnalysis on Days 1, 6, and 13 of each two-week treatment period,n Days 1 and 6 of the final treatment period, and at end of theollow-up period. Informed consent was obtained from the sub-ect.

.2. Viral RNA isolation and RT-PCR

Viral RNA was extracted from 100 �l serum samples using theIAamp viral RNA mini-kit (Qiagen) according to the manufac-

urer’s instructions. Reverse transcription of total viral RNA wasrimed with random hexamers using Superscript III First Strandynthesis System for RT-PCR (Invitrogen) according to the manu-acture’s instructions. The HCV protease region was amplified withpecific primers using Platinum PCR Supermix High Fidelity underhe following nested reaction conditions (Lodrini et al., 2003): Firstound, 94 ◦C for 3 min, then 94 ◦C for 1 min, 55 ◦C for 30 s, and 72 ◦Cor 30 s, for 40 cycles, and 72 ◦C for 5 min with first round primers1b0f, TCGTCTTTTCTGACATGGAG and P1b1r, TTGTACCCTTGGGCT-CATA. Second round: 94 ◦C for 3 min, then 94 ◦C for 1 min, 60 ◦C for

0 s, and 72 ◦C for 1 min, for 40 cycles, and 72 ◦C for 5 min with sec-nd round primers P1b2f, TCATCACCTGGGGGGCAGAC and P1b3r,TGCTCTTGCCGCTGCCAGT. All samples were gel-purified from a 2%garose gel using the QIAquick gel extraction kit (Qiagen) accordingo the manufacturer’s instructions.

smoof

Methods 153 (2008) 156–162 157

.3. Clonal sequencing

Purified RT-PCR products were cloned using the TOPO TAloning Kit (pCR 2.1TOPO vector, Invitrogen). For each serum sam-le, 96 bacteria colonies were sent to Qiagen and Genewiz forequencing, using M13 forward and reverse primers as well as tworotease specific primers (56f, GACATCATCTTGGGTCTGCCCGTCTC,5r, GTGGGAGCGTGTAGGTGGGC). Trace files were analyzed by aHRED-based algorithm developed at Schering-Plough to identifyutations and to calculate the percentage of each mutation. Stan-

ard error of the percentage of each mutation was calculated usingbootstrap procedure: SE = (

√Nx

√(M/N)x(1 − M/N))/N where

is the total number of clones sequenced and M is the number oflones containing the mutation.

.4. TaqMAMA conditions

Protease sequences obtained from clonal sequencing of serumamples of patient 105 were aligned using DNASTAR program.orward and reverse primers and probes for each mutation wereesigned based on sequence conservation and guidelines outlinedy Applied Biosystems. The wild-type primer (WT) was a perfectatch to the original sequence, whereas the mutant primer (MUT)atched the sequence carrying the resistance mutation. MAMA

rimer contained two mismatches to the original sequence, onet the 3′ ultimate base, another at the 3′ penultimate base. The spe-ific mismatch was chosen based on its greatest strength of alleliciscrimination (Li et al., 2004).

Specific wild type, mutant, and MAMA primers and probes wereonstructed as follows for V36I, T54A, and V170A. The mutateducleotides are italicized.

V36I reverse primer: WT, 5′-GAAAGATTGTGTTGCGGTGGAAAC;UT, 5′-GAAAGATTGTGTTGCGGTGGAAAT; MAMA, 5′-GAAAGAT-

GTGTTGCGGTGGAACT; forward primer: 5′-GGGCCTACTTGGTT-CATCATCA; probe: 5′-FAM-CACAGGCCGTGACAAAAACCAGGTCGA

TAMRA.T54A forward primer: WT, 5′-GACCTGTATTAACGGTGTGTGCTG-

A; MUT, 5′-GACCTGTATTAACGGTGTGTGCTGGG; MAMA, 5′-GAC-TGTATTAACGGTGTGTGCTGCG; reverse primer: 5′-CGAGGTCCT-GTCTACATTGGTGT; probe: 5′-FAM-TGGCGCCGGCTCAAAGACCT-AG-TAMRA.

V170A reverse primer: WT, 5′-TAGTGGTTTCCATAGACTCGACG-GTA; MUT, 5′-TAGTGGTTTCCATAGACTCGACGGGTG; MAMA, 5′-AGTGGTTTCCATAGACTCGACGGGAG; forward primer: 5′-GCCAGT-TCCTACTTGAAGGGTTCTT; probe: 5′-FAM-CATCTTCCGGGCTGC-GTGTGCACC-TAMRA.

TaqMan reaction conditions were as follows: 50 ◦C for 2 min,5 ◦C for 10 min, and then 95 ◦C for 15 s and 60 ◦C for 1 min, for0 cycles. Samples were tested using 5 and 50 pg of input RT-PCRroducts. Reactions were carried out with the instrument ABI 7900Applied Biosystems). The CT values are the PCR cycle numbers toeach critical threshold as specified by the manufacturer.

.5. Generation of standard curves

To determine percent mutant from a mixed population, a stan-ard curve was generated using samples with known mutant/WTatio. Bacteria clones containing mutant and wild-type proteaseegions for V36I, T54A, and V170A were identified by clonal

equencing of serum samples from patient 105. Wild type andutant DNA plasmids were extracted and quantitated by Nan-

drop spectrophotometer at OD260 (Thermo Scientific). The ratiosf mutant/WT DNA used to generate standard curve were asollows: 100/0, 80/20, 50/50, 20/80, 10/90, 5/95, 1/99, 0.5/99.5,

1 ogical Methods 153 (2008) 156–162

0waw

2

slscwam2td

3

3

iwatmm(fasmastwt

FtMwra

Fig. 2. Preferential amplification of mutant allele by MAMA primer. Templates withwild-type or mutant sequences (T54A) were amplified with WT, MUT or MAMAprimers. The CT values (the PCR cycle numbers to reach critical threshold) are plot-tCs

3s

acppMtt

58 S. Curry et al. / Journal of Virol

.1/99.9, 0.05/99.95, 0.01/99.99, and 0/100, the total input DNAas 5 and 50 pg. The CT values were plotted against log[mutant%]

nd fitted by linear regression using GraphPad PRISM soft-are.

.6. Data analysis

The percent mutant was calculated from CT values using thetandard curve. This standard curve was also used to determine theevel of detection (LOD), which is the lowest ratio of mutant/WTtill in the linear range. Results from multiple experiments wereombined to generate assay statistics. The coefficient of varianceas calculated to be ∼10–30% from multiple experiments (n = 4–6)

nd the 95% confidence interval was less than 2.5-fold over theean. The level of quantification (LOQ) was therefore defined as

.5 × LOD; so if the ratio between the mutant% of a given muta-ion and LOD was >2.5, the result was considered quantifiableata.

. Results

.1. Principles of TaqMAMA technology

The key determinant of allelic discrimination by TaqMAMAs the MAMA (mismatch amplification mutation assay) primer

hose ultimate 3′ base is complementary to the mutant allele,nd the penultimate 3′ base is changed to mismatch with bothhe wild type and mutant alleles. As a result, there is one mis-

atch between MAMA primer and the mutant sequence and twoismatches between MAMA primer and the wild-type sequence

Fig. 1). MAMA primer is then paired with another primer to be usedor PCR amplification of alleles by TaqMan assay. The one mismatcht the penultimate base between MAMA primer and the mutantequence would allow for efficient priming and amplification ofutant allele, whereas the two mismatches at the penultimate

nd ultimate bases between MAMA primer and the wild-typeequence would drastically reduce the its efficiency of amplifica-ion. Therefore the preferential amplification of mutant alleles overild type allows for the detection of minor variants in a popula-

ion.

ig. 1. Schematic of allelic discrimination by TaqMAMA. The site of T54A muta-ion (A in wild type, G in mutant sequence) is shown in red. Mismatches with the

AMA primer are shown in blue. MAMA primers are designed to have one mismatchith the mutant sequence and two mismatches with the wild-type sequence which

esults in preferential amplification of mutant alleles in the subsequent TaqManssay.

oootefiposTt

acos(ecf

ioeftpnT

ed against the amount of input DNA template. The MAMA primer shows a 10T (approximately 1000-fold) difference in amplification of mutant and wild-typeequences.

.2. TaqMAMA primer/probe optimization and generation oftandard curves

The mutation T54A (ACT to GCT as shown in Fig. 1) was used asn example to illustrate the evaluation and validation of TaqMAMAonditions. To define the relative amplification efficiency of MAMArimer, three primers, one matching the wild-type sequence (WTrimer), one matching the mutant sequence (MUT primer) andAMA primer were used to amplify the wild type and mutant

emplates. As shown in Fig. 2, MAMA primer amplified wild-typeemplate with ∼1000-fold reduced efficiency due to the presencef two mismatches. The impact of a single mismatch at the terminalr penultimate position on amplification efficiency can vary basedn primer/template sequences. In the example given in Fig. 2, wildype and mutant templates were amplified with equal efficiency byither WT or MUT primers. The mutant template was also ampli-ed efficiently by the MAMA primer. However, results from otheratients showed that amplification efficiency could be decreased byne terminal or penultimate mismatch (data not shown). Similarequence-dependent results were also observed by Li et al. (2004).he slopes of all curves were calculated to be 3.2, an indication thathe TaqMan reactions were almost 100% efficient.

To generate a standard curve to quantitate mutant alleles frommostly wild-type population, DNA plasmids (total 5 or 50 pg)

ontaining mutant and wild-type sequences were mixed at vari-us ratios (mutant/WT: 0.01–100%) and assayed by TaqMAMA. Thetandard curve was also used to determine the level of detectionLOD), which is the lowest ratio of mutant/wild type still in the lin-ar range. Data points outside the linear range were excluded fromalculation (Fig. 3). The LOD is 0.01% for the 5 pg reaction and 0.05%or the 50 pg reaction.

Given that HCV has a quasispecies nature which may resultn significant polymorphism near any resistance loci, the effectf random mismatches in the primer and probe on amplificationfficiency was evaluated. As shown in Fig. 4, one to as many asour mismatches were introduced into templates (in addition to

he designed penultimate and ultimate mismatches with MAMArimer). These additional mismatches were shown to decrease sig-ificantly the amplification efficiency (50-fold with one mismatch).hese data suggest that TaqMAMA is very sensitive to perturba-

S. Curry et al. / Journal of Virological

Fig. 3. Generation of standard curves. Total 5 and 50 pg of templates carrying wild-tap

tcp

3s

wttPltw

wlcrFtoHtaiTiwPd

3

ssototd∼10–30% from multiple experiments (Table 3) and the 95% confi-

TC

P

VQFFTAVN

Fw1

ype or mutant sequences were mixed at various ratios (0.01–100% of mutant) andssayed by TaqMAMA. The CT values are plotted against log[mutant%]. The dataoint outside the linear range is excluded from calculation.

ion in template sequences and it may be necessary to performlonal sequencing to identify conserved regions for primer androbe design to obtain accurate results.

.3. Analysis of protease resistance mutations by clonalequencing

Patient 105 (genotype 1 non-responder) received at first twoeeks of PEG-IFN �-2b treatment (1.5 �g/kg weekly) which had lit-

le impact on the HCV viral load. After a two-week wash out period,he patient received two weeks of combination treatment with

EG-IFN �-2b and 200 mg TID boceprevir, which reduced the viraload by ∼1.5 log on Day 6. After another two-week washout period,he patient received boceprevir monotherapy for one week duringhich the viral load remained unchanged. At the end of the two-

dTsw

able 1lonal sequencing results from samples obtained from patient 105

ercent mutation PEG-IFN �-2bmonotherapy

Washout Combination�-2b + bocep

Day 1 Day 6 Day 13 Day 1

36I – – – 141R 2 – – –43S 1 – – –43C – – – –54A 1 1 – –156S – – – –170A – – – –umber of clones 83 91 87 87

ig. 4. Effect of sequence polymorphism on TaqMAMA efficiency. 25 pg of templates contaas amplified in TaqMan assays. The designed penultimate and ultimate mistaches in MA/2dCT, where dCT = CT value of sample with mismatch − CT value of sample without mism

Methods 153 (2008) 156–162 159

eek follow-up period, the viral load rebounded to pretreatmentevel. Clonal sequencing of approximately 90 clones from samplesollected throughout the study detected mutations at six knownesistance loci (Table 1 and Fig. 5). Some mutations (V36I, Q41R,43S/C, T54A, and V170A) were observed at low levels (in one orwo clones per 90 clones sequenced) prior to boceprevir treatment;ne mutation, A156S, was detected only at the end of follow-up.owever, the detection of these minor variants was intermittent

hroughout the study with large estimated error (coefficient vari-nce ∼100%), which suggests that clonal sequencing of 90 cloness not sufficient to detect and quantify these low level mutations.he prevalence of two mutants, T54A and V170A, increased signif-cantly during boceprevir monotherapy (16 and 12% respectively,

ith coefficient variance <30%) while others remained at low levels.revious analysis using the less sensitive direct sequencing methodetected T54A but not V170A (Zeuzem et al., 2005b).

.4. Analysis of protease resistance mutations by TaqMAMA

The TaqMAMA technique was used to analyze three repre-entative mutations (V36I, T54A and V170A) identified by clonalequencing from patient 105. The results for T54A and V170A fromne experiment are shown in Table 2. RNA samples from variousime points were run in TaqMAMA as described above. The CT valuef each sample was used to calculate the percentage of each muta-ion based on a standard curve generated in parallel (LOD = 0.05%,ata not shown). The coefficient of variance was calculated to be

ence interval was estimated to be less than 2.5-fold over the mean.he level of quantitation (LOQ) was therefore defined as 2.5 × LOD;o if the ratio between the percentage of a given mutation and LODas >2.5, the result was considered quantifiable data (Table 2).

PEG-IFNrevir therapy

Washout Boceprevir monotherapy Follow-up

Day 6 Day 13 Day 1 Day 6

– – – – –1 2 – 1 –– – – – –– – – 1 –– 1 1 16 1– – – – 11 – 1 12 1

94 95 91 90 88

ining additional mismatches in the MAMA primer and probe region (shown in red)MA primer are shown in blue. The relative amplification efficiency is calculated as:atch.

160 S. Curry et al. / Journal of Virological Methods 153 (2008) 156–162

Table 2aTaqMAMA results for mutation T54A from one representative experiment

Samples CT Mutant% Mutant%/LOD Quantifiable data

PEG-IFN �-2b monotherapy Day 1 28.0 0.18 3.6 YesDay 6 27.9 0.19 3.8 YesDay 13 28.3 0.14 2.8 Yes

Combination PEG-IFN �-2b plus boceprevir Day 1 27.5 0.25 4.9 YesDay 6 28.2 0.16 3.1 YesDay 13 28.1 0.17 3.4 Yes

Boceprevir monotherapy Day 1 25.4 1.07 21 YesDay 6 21.6 14.6 292 Yes

Follow-up D 14 24.5 2.0 40 Yes

Table 2bTaqMAMA results for mutation V170A from one representative experiment

Samples CT Mutant% Mutant%/LOD Quantifiable data

PEG-IFN �-2b monotherapy Day 1 24.8 0.17 3.3 YesDay 6 24.9 0.16 3.1 YesDay 13 25.0 0.14 2.8 Yes

Combination PEG-IFN �-2b plus boceprevir Day 1 24.7 0.17 3.5 YesDay 6 24.9 0.15 3.0 YesDay 13 21.6 1.59 32 Yes

B 3.09.8

F 2.2

mcwsod

tTo

TS

V

T

V

oceprevir monotherapy Day 1 2Day 6 1

ollow-up D 14 2

The results from clonal sequencing and TaqMAMA are sum-arized in Fig. 5. As discussed above, clonal sequencing of ∼90

lones detected intermittently low level (1–2%) resistant variants

ith large estimated error. In contrast, TaqMAMA detected con-

istently variants at levels as low as 0.1% throughout the coursef the treatment. All 3 mutations (V36I, T54A and V170A) wereetected prior to treatment at ∼0.1% level. The prevalence of muta-

i1tC

able 3ummary of detection of V36I, T54A and V170A mutations (mutant%) by TaqMAMA

Expt PEG-IFN �-2b monotherapy Combinat�-2b + bo

Day1 Day6 Day13 Day1

36I 1 0.28 0.50 0.37 0.542 0.30 0.44 0.27 0.363 0.26 0.49 0.37 0.494 0.29 0.43 0.25 0.375 0.26 0.45 0.33 0.506 0.32 0.47 0.27 0.40Average 0.29 0.46 0.31 0.44S.D. 0.02 0.03 0.05 0.08CV% 9 6 17 17

54A 1 0.16 0.15 0.13 NA2 0.21 0.26 0.17 NA3 0.13 0.15 0.12 0.214 0.18 0.19 0.14 0.255 0.14 0.17 0.13 0.196 0.13 0.13 0.11 0.15Average 0.16 0.17 0.13 0.20S.D. 0.03 0.04 0.02 0.04CV% 19 25 16 20

170A 1 0.19 0.15 0.14 NA2 0.16 0.12 NA NA3 0.13 0.12 0.11 0.134 0.17 0.16 0.14 0.175 0.14 0.14 0.11 0.116 0.24 0.11 0.11 0.13Average 0.17 0.13 0.12 0.14S.D. 0.04 0.02 0.02 0.03CV% 23 14 15 19

0.58 12 Yes5.49 110 Yes

1.00 20 Yes

ion V36I did not change during the study (Fig. 5a). Mutation54A was selected during boceprevir monotherapy to 14% levelf the total population on Day 6, consistent with clonal sequenc-

ng results (Fig. 5b). Mutation V170A was found to increase to% level on Day 13 of combination treatment, and increase fur-her to 5% level on Day 6 during boceprevir monotherapy (Fig. 5c).lonal sequencing identified V170A mutation on Day 6 of bocepre-

ion PEG-IFNceprevir therapy

Boceprevir monotherapy Follow up

Day6 Day13 Day1 Day6

0.28 0.85 0.41 0.52 0.380.20 0.69 0.45 0.65 0.380.26 0.78 0.42 0.51 0.36NA 0.64 0.46 0.61 0.360.27 0.71 0.41 0.50 0.380.23 0.73 0.39 0.62 0.400.25 0.73 0.42 0.57 0.380.03 0.07 0.03 0.06 0.0214 10 6 11 4

0.14 NA 0.99 20.57 2.340.20 0.19 1.04 13.93 2.170.12 0.14 0.82 14.84 2.530.16 0.17 1.07 14.59 2.020.11 0.13 0.66 10.40 1.940.11 0.11 0.63 10.19 1.610.14 0.15 0.87 14.09 2.100.03 0.03 0.19 3.79 0.3224 22 22 27 15

0.16 1.85 0.65 7.04 0.900.13 1.30 0.45 5.73 0.650.10 1.40 0.48 6.19 0.860.15 1.59 0.58 5.49 1.00NA 1.21 0.43 3.59 0.660.12 1.22 0.50 3.93 0.540.13 1.42 0.52 5.33 0.770.02 0.25 0.08 1.33 0.1816 18 16 25 23

S. Curry et al. / Journal of Virological

Fig. 5. Summary of clonal sequencing and TaqMAMA results for V36I, T54A, andV170A mutations. Samples were collected for sequencing analysis on Day 1, 6 and13 during interferon monotherapy (D1i, D6i, D13i), Day 1, 6 and 13 during interferonand boceprevir combination therapy (D1c, D6c, D13c), Day 1 and 6 during boceprevirmmV

vcsp

4

qi

eaotacwcitda

MrmwrvaappauadtielTnoAmt1emairpeln

p�ditatomvwcresistance. (4) At the end of two-week follow-up period, the preva-

onotherapy (D1p, D6p), and at end of follow-up (FUP). The percentage of eachutation (with SE) as measured by TaqMAMA and clonal sequencing is shown for36I (a), T54A (b) and V170A (c).

ir monotherapy but failed to detect earlier emergence duringombination therapy (Fig. 5c). Both T54A and V170A decreasedignificantly to 1–2% level at the end of the two-week follow-uperiod.

. Discussion

Several techniques have been developed for the detection anduantitation of resistant variants. Direct population sequencing fornitial genetic characterization of mutation patterns can detect gen-

ltaw

Methods 153 (2008) 156–162 161

rally minority species of 20% or greater. If clinical outcomes (suchs on-therapy viral load rebound) cannot be explained by the resultsf direct sequencing or other factors (such as poor adherence toherapy), clonal analysis can be applied in early clinical trials or inmanageable subset of samples to allow detection of a lower per-

entage of minority species. For example, sequencing of 80 clonesould allow detection of mutants at 4% level or greater with 95%

onfidence interval (Parkin, 2008). Clonal analysis also providesnformation on the linkage of mutations when multiple muta-ions are present. However, more sensitive methods are required toetect and quantify minor variants such as those in pretreatmentnd follow-up samples.

The TaqMan mismatch amplification mutation assay (Taq-AMA) is a highly sensitive allelic discrimination method. In this

eport, the method was adapted to detect and quantify knowninor resistant variants of HCV and was shown to correlate wellith clonal sequencing results. TaqMAMA was linear over a wide

ange of mutant levels (0.01–100%), and could detect consistentlyariants at ∼0.1% level. The assay was highly reproducible, withcoefficient of variance of approximately 10–30%. However, the

ssay also has limitations related to the fact that HCV exists inatients as quasispecies which may result in significant sequenceolymorphism near the mutation site. Specific TaqMAMA primersnd probes must be developed for each mutation for each individ-al patient; and the primer design is best based on a well-conservedrea identified by clonal sequencing. In addition to TaqMAMA con-itions, assay sensitivity is dependent on input viral load. Assuminghat at least 10 copies of mutant viral genomes have to be presentn 100 �l starting material (i.e. 102 copies/ml) to ensure consistentxtraction and amplification, the total viral titer will need to be ateast 105 copies/ml to allow detection of such a mutant at 0.1% level.herefore, the assay is most applicable for detailed study of a smallumber of samples with relatively high viral load (the viral titerf patient 105 remained >105 copies/ml at all time points tested).nother factor that can influence the assay outcome is sporadicutations that arise from misincorporation by reverse transcrip-

ase and Taq DNA polymerase during RT-PCR reactions (Smith et al.,997). The error rate of retrovirus reverse transcriptase has beenstimated at 10−4 (Smith et al., 1997). The Platinum PCR Super-ix High Fidelity system (Invitrogen) used in this report containsthermostable proofreading enzyme to increase fidelity. Based on

ts error rate (1.8 × 10−6) and total cycle number (80) of nested PCReactions, the misincorporation rate was estimated to be 1.4 × 10−4

er PCR reaction. The combined error rate for RT-PCR reactions wasstimated to be 2.4 × 10−4, about 5-fold below the lowest mutantevel (0.12%) reported in this study, thus artefactual mutations wereot expected to affect the interpretation of the assay results.

Results from TaqMAMA analysis of samples from a singleatient who underwent treatment with boceprevir and PEG-IFN-2b demonstrated that: (1) low levels of resistant variants wereetected by TaqMAMA in pretreatment samples, providing exper-

mental evidence for the prediction, based on quasispecies theory,hat resistant variants pre-exist in the viral population. (2) Notll pre-existing resistant variants were selected during boceprevirherapy. The prevalence of the V36I variant did not change through-ut the treatment period. (3) The prevalence of the two otherajor mutations, T54A and V170A, did increase during bocepre-

ir monotherapy but was suppressed during combination therapyith boceprevir and PEG-IFN �-2b, providing further evidence that

ombination therapy is critical to suppression of the emergence of

ence of T54A and V170A mutants declined significantly, suggestinghat mutant viruses were out-competed by wild-type virus in thebsence of boceprevir selection. However, both mutant speciesere detected by TaqMAMA method at much higher levels than

1 ogical

pfat

A

pRt

R

C

C

C

F

G

H

H

K

L

L

M

M

N

P

S

S

T

T

Z

62 S. Curry et al. / Journal of Virol

retreatment. These results may have important implications bothor re-treatment of patients with the same class of antiviral drugsnd the possibility of transmission of resistant viruses when theherapy becomes available widely.

cknowledgements

We would like to thank Vincent Sanfiorenzo for providingatient serum samples and Drs. Robert Ralston, Eric Hughes,achael Steiner and Janelle Landau for critically reading and editinghe manuscript.

eferences

abot, B., Martell, M., Esteban, J.I., Piron, M., Otero, T., Esteban, R., Guardia, J., Gómez,J., 2001. Longitudinal evaluation of the structure of replicating and circulatinghepatitis C virus quasispecies in nonprogressive chronic hepatitis C patients. J.Virol. 75, 12005–12013.

ha, R.S., Zarbl, H., Keohavong, P., Thilly, W.G., 1992. Mismatch amplification muta-tion assay (MAMA): application to the c-H-ras gene. Genome Res. 22, 14–20.

ubero, M., Esteban, J.I., Otero, T., Sauleda, S., Bes, M., Esteban, R., Guardia, J., Quer,J., 2008. Naturally occurring NS3-protease-inhibitor resistant mutant A156T inthe liver of an untreated chronic hepatitis C patient. Virology 370, 237–245.

arci, P., Strazzera, R., Alter, H.J., Farci, S., Degioannis, D., Coiana, A., Peddis, G., Usai, F.,Serra, G., Chessa, L., Diaz, G., Balestrieri, A., Purcell, R.H., 2002. Early changes inhepatitis C viral quasispecies during interferon therapy predict the therapeuticoutcome. Proc. Natl. Acad. Sci. U.S.A. 99, 3081–3086.

laab, W.E., Skopek, T.R., 1999. A novel assay for allelic discrimination that combinesthe fluorogenic 5′ nuclease polymerase chain reaction (TaqMan®) and mismatchamplification mutation assay. Mutat. Res. 430, 1–12.

eid, C.A., Stevens, J., Livak, K.J., Williams, P.M., 1996. Real time quantitative PCR.Genome Res. 6, 986–994.

erring, B.L., Tsui, R., Peddada, L., Busch, M., Delwart, E.L., 2005. Wide range ofquasispecies diversity during primary hepatitis C virus infection. J. Virol. 79,

4340–4346.

olykhalov, A.A., Mihalik, K., Feinstone, S.M., Rice, C.M., 2000. Hepatitis Cvirus-encoded enzymatic activities and conserved RNA elements in the 3′ non-translated region are essential for virus replication in vivo. J. Virol. 74, 2046–2051.

i, B., Kadura, I., Fu, D.-J., Watson, D.E., 2004. Genotyping with TaqMAMA. Genomics83, 311–320.

Z

Methods 153 (2008) 156–162

odrini, S., Bagaglio, S., Canducci, F., de Mitri, M.S., Andreone, P., Loggi, E., Laz-zarin, A., Clementi, M., Morsica, G., 2003. Sequence analysis of NS3 proteasegene in clinical strains of hepatitis C virus. J. Biol. Regul. Homeost. Agents 17,198–204.

alcolm, B.A., Liu, R., Lahser, F., Agrawal, S., Belanger, B., Butkiewicz, N., Chase, R.,Gheyas, F., Hart, A., Hesk, D., Ingravallo, P., Jiang, C., Kong, R., Lu, J., Pichardo,J., Prongay, A., Skelton, A., Tong, X., Venkatraman, S., Xia, E., Girijavallabhan, V.,Njoroge, F.G., 2006. SCH 503034, a mechanism-based inhibitor of hepatitis Cvirus NS3 protease, suppresses polyprotein maturation and enhances the antivi-ral activity of alpha interferon in replicon cells. Antimicrob. Agents Chemother.50, 1013–1020.

artell, M., Esteban, J.I., Quer, J., Genescà, J., Weiner, A., Esteban, R., Guardia, J.,Gómez, J., 1992. Hepatitis C virus (HCV) circulates as a population of differentbut closely related genomes: quasispecies nature of HCV genome distribution.J. Virol. 66, 3225–3229.

eumann, A.U., Lam, N.P., Dahari, H., Gretch, D.R., Wiley, T.E., Layden, T.J., Perel-son, A.S., 1998. Hepatitis C viral dynamics in vivo and the antiviral efficacy ofinterferon-� therapy. Science 282, 103–107.

arkin, N., 2008. HCV resistance assessment: a proposal from the HCV DRAG (DrugResistance Advisory Group). EASL-AASLD-APASL-ALEH-IASL conference Hepati-tis B & C virus resistance to antiviral therapies. Paris, France.

arrazin, C., Rouzier, R., Wagner, F., Forestier, N., Larrey, D., Gupta, S.K., Hussain, M.,Shah, A., Cutler, D., Zhang, J., Zeuzem, S., 2007. SCH 503034, a novel hepatitisC virus protease inhibitor, plus pegylated interferon alpha-2b for genotype 1nonresponders. Gastroenterology 132, 1270–1278.

mith, D.B., McAllister, J., Casino, C., Simmonds, P., 1997. Virus ‘quasispecies’: makinga mountain out of a molehill? J. Gen. Virol. 78, 1511–1519.

ong, X., Chase, R., Skelton, A., Chen, T., Wright-Minogue, J., Malcolm, B.A., 2006.Identification and analysis of fitness of resistance mutations against the HCVprotease inhibitor SCH 503034. Antiviral Res. 70, 26–38.

ong, X., Bogen, S., Chase, R., Girijavallabhan, V., Guo, Z., Njoroge, F.G., Prongay,A., Saksena, A., Skelton, A., Xia, E., Ralston, R., 2008. Characterization of resis-tance mutations against HCV ketoamide protease inhibitors. Antiviral Res. 77,177–185.

euzem, S., Sarrarin, C., Rouzier, R., Tarral, A., Brion, N., Forestier, N., Gupta, S., Deck-man, D., Fellows, K., Hussain, M., Cutler, D., Zhang, J., 2005a. Anti-viral activity ofSCH 503034, a HCV protease inhibitor, administered as monotherapy in HCV-1patients refractory to pegylated interferon (PEG-IFN-�). Hepatology 42 (Suppl.

1), 233A.

euzem, S., Sarrarin, C., Wagner, F., Rouzier, R., Forestier, N., Gupta, S., Hussain,M., Shah, A., Cutler, D., Zhang, J., 2005b. Combination therapy with HCVprotease inhibitor SCH 503034, Plus PEG-Intron in Hepatitis C Genotype-1 PEG-Intron non-responders: phase 1b results. Hepatology 42 (Suppl. 1),276A.