Comparison of different HCV viral load and genotyping assays

Jonathan C. Anderson a,1, Josephine Simonetti b,1, Dennis G. Fisher c,James Williams b, Yasuhiro Yamamura d, Nayra Rodriguez d, Daniel

G. Sullivan e, David R. Gretch e, Brian McMahon b, Kandace J. Williams f,*a Biomedical Program, University of Alaska Anchorage, Anchorage, AK, USA

b Viral Hepatitis Program, Alaska Native Medical Center, Anchorage, AK, USAc Center for Behavioral Research and Services, California State University at Long Beach, Long Beach, CA, USA

d AIDS Research Program, Ponce School of Medicine, Ponce, PR, USAe Department of Laboratory Medicine, University of Washington Medical Center, Seattle, WA, USA

f Department of Biochemistry and Molecular Biology, Medical College of Ohio, 3035 Arlington Avenue, Toledo, OH 43614, USA

Received 16 May 2002; received in revised form 20 September 2002; accepted 11 October 2002

Abstract

Background: We report an interlaboratory comparison of methods for the determination of hepatitis C virus (HCV)

serum load and genotype between a recently, established molecular laboratory at the Alaska Native Medical Center

(ANMC) and two independent laboratories using different assays. At ANMC, a Real-time quantitative RT-PCR

amplification methodology (QPCR) has been developed in which HCV viral loads are determined by interpolation of

QPCR results to those of standards calibrated to the World Health Organization (WHO) First International Standard

for HCV. HCV genotype is subsequently determined by direct sequencing of the DNA fragment generated from the

QPCR assay. Objectives and Study Design: The above methods were statistically compared to results obtained for the

same patient sera by two independent laboratories using different commercially available viral load assays;

QuantiplexTM HCV RNA (Bayer Diagnostics) and AmplicorTM HCV MonitorTM (v 2.0) (Roche Molecular Systems),

as well as two different genotyping assays; restriction fragment length polymorphism (RFLP) and INNO-LiPA HCV II

(Innogenetics). Results: ANMC’s Real-time QPCR HCV viral load results compared moderately well with those

obtained by the QuantiplexTM HCV RNA method (R2�/0.3813), and compared quite well with recent lot numbers of

AmplicorTM HCV MonitorTM in which viral loads are derived in IU/ml (R2�/0.6408), but compared poorly with earlier

lot numbers of AmplicorTM HCV MonitorTM in which viral loads were derived in copies/ml (R2�/0.0913). The ANMC

direct sequencing method for genotype determination compared moderately to very well with both the RFLP (84�/86%)

and INNO-LiPA (85�/97.5%) methods. Conclusions: These viral load comparisons highlight the discrepancies that may

Abbreviations: HCV, hepatitis C virus; WHO, World Health Organization; QPCR, quantitative polymerase chain reaction; RT,

reverse transcriptase; RFLP, restriction fragment length polymorphism; IU, international units; ANMC, Viral Hepatitis Program,

Alaska Native Medical Center; UW, University of Washington School of Medicine; PR, AIDS Research Program, Ponce School of

Medicine, Ponce, Puerto Rico.

* Corresponding author. Tel.: �/1-419-383-4135; fax: �/1-419-383-6228

E-mail address: kanwilliams@mco.edu (K.J. Williams).1 Both authors contributed equally.

Journal of Clinical Virology 28 (2003) 27�/37

www.elsevier.com/locate/jcv

1386-6532/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.

PII: S 1 3 8 6 - 6 5 3 2 ( 0 2 ) 0 0 2 3 5 - 4

occur when patient HCV viral loads are monitored using different types of assays. Comparison of HCV genotype by

different methods is more reliable statistically and important clinically for predicting probability of response to antiviral

therapy. However, viral loads are important for monitoring response once therapy has begun.

# 2002 Elsevier Science B.V. All rights reserved.

Keywords: Hepatitis C virus; Serum viral load; Genotype; Interlaboratory comparison

1. Introduction

Hepatitis C virus (HCV) is a flavivirus that

currently infects an estimated 1.8% of the US

population (Alter et al., 1999). It is estimated that

only 20% of infected individuals will recover from

this viral infection, while the rest become chroni-

cally infected (Cohen, 1999; Management of

Hepatitis C, NIH Consensus Statement Online).

While the majority of chronically infected indivi-

duals never exhibit symptoms, approximately 10�/

30% of these patients will eventually develop

cirrhosis or hepatocellular carcinoma, both of

which are associated with significant morbidity

and mortality (Di Bisceglie et al., 1991; Tibor et

al., 1999). HCV infection is currently estimated to

cause 40�/60% of chronic liver disease and as a

consequence is recognized as the leading cause for

liver transplantation in the United States (Arens,

2001; Cohen, 1999; Hepatitis C Fact Sheet, CDC

Website).

To date, the most effective treatment regimen

for HCV consists of combination therapy with

Interferon-a-2b and Ribavirin; a treatment that

has varied effectiveness and adverse side effects

(Thomas and Lemon, 2000). However, in recent

clinical trials both pretreatment serum HCV RNA

levels (viral load) and viral genotype were found to

be the two best indicators of the potential for

treatment success (McHutchinson et al., 1998;

Poynard et al., 1998). Therefore, once an indivi-

dual has tested positive for HCV, current recom-

mendations are that viral load and genotype tests

be performed to establish patient suitability for

treatment and subsequent length of treatment

(Carithers et al., 2000; EASL International Con-

sensus Conference on Hepatitis C, 1999).

The most commonly used methods for the

determination of serum HCV viral load involve

amplification of either (i) a direct HCV genome

target (QPCR) or (ii) an associated signal

(branched chain or bDNA assays) (reviewed in

Morishima and Gretch, 1999). Many laboratories

have developed ‘home-brew’ assays based on these

methodologies. HCV quantitative assays are also

commercially available as the QPCR-based Am-

plicorTM HCV MonitorTM test from Roche Mole-

cular Systems and the bDNA-based QuantiplexTM

HCV RNA assay from Bayer Diagnostics.

Because each commercially available viral load

assay has been developed using proprietary HCV

RNA standards, the units of measure in each assay

are particular to that method, making comparison

of results between different assays difficult. As a

consequence, a patient’s HCV viral load can be

accurately compared from one time point to

another only if the same assay is used for analysis

each time. Recently, however, a WHO Interna-

tional Standard, expressed in international units

(IU), was established for HCV RNA viral load

assays and has led to the development of a HCV

RNA quantification panel (Jorgenson and Neu-

wald, 2001; Saldanha, 1999; Saldanha et al., 1999).

The NAPTM HCV�/RNA nucleic acid panel pro-

vides a commercially available standard for the

comparison of HCV RNA viral loads between

different viral load assays and fulfills the recom-

mendation that all HCV RNA quantitative assays

report results as IU/ml (Pawlotsky et al., 2000).

For genotype analysis, the three major meth-

odologies are (i) direct sequencing of PCR pro-

ducts (Arens, 2001), (ii) restriction fragment length

polymorphism (RFLP) of PCR products (David-

son et al., 1995), and (iii) reverse hybridization of

PCR products to genotype-specific probes, such as

the line probe assay available commercially as

INNO-LiPA from Innogenetics (Le Pogam et al.,

1998). Although the recognized gold standard for

genotype determination is direct DNA sequencing,

this methodology has been criticized as too

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/3728

expensive and labor intensive for routine clinicaluse (Thomas and Lemon, 2000). However, recent

advances, such as one step RT-PCR and auto-

mated sequencing technology, have made this

method less cumbersome and more cost effective,

and thus a viable and excellent alternative to the

less specific commercial methods for routine

genotype testing.

The Viral Hepatitis Program at the AlaskaNative Medical Center (ANMC) recently estab-

lished a molecular biology laboratory for both

epidemiological research and patient monitoring

purposes which has HCV viral load and genotyp-

ing capabilities. This is the first facility of this type

to be established in Alaska, thus negating the need

to transport specimens elsewhere for analyses. The

methodologies employed by ANMC consist of aone step viral load determination by Real-time

quantitative PCR and subsequent genotype deter-

mination by direct sequencing of the same ampli-

fied fragment. The portion of the HCV genome

targeted is a 250 base pair (bp) fragment of the 5

prime non-coding region (5? NCR). The 5? NCR

portion of HCV is widely used in assays of this

type because it is conserved enough betweengenotypes to allow consistent amplification but

sufficiently different at the nucleotide level to

distinguish individual genotypes from one another

(Zein, 2000). Here we report validation of the

ANMC methodology through the statistical com-

parison of ANMC genotype and viral load results

to those obtained by two independent laboratories

using different commercially available assays.Ninety serum samples were compared by ANMC

in all; 40 samples were provided by the Ponce

School of Medicine AIDS Research Program in

Ponce, Puerto Rico (PR) and 50 samples were

provided by the University of Washington School

of Medicine, Seattle, WA (UW).

2. Materials and methods

2.1. Specimen acquisition and processing

At the originating laboratory, each serum was

labeled with an individual code number and no

other identification. All sera were transported on

dry ice and subsequently stored at �/70 8C. HCVserum viral load and genotype results correspond-

ing to the individual serum code numbers were

obtained from the originating laboratory only

after all sera were processed by ANMC, thus these

interlaboratory comparisons were masked from

the data from the originating sites.

2.2. ANMC viral load and genotype determination

2.2.1. Viral RNA isolation

Viral RNA was extracted from 100 ml of serum

using the Roche High Pure Viral RNA kit (Roche

Diagnostics Corporation) as per the manufac-turer’s instructions with the following exception:

the viral RNA was eluted in 50 ml of nuclease free

water containing a 5? NCR reverse PCR primer

(5?-TACCACAAGGCCTTTCGCGACCCAA-

CACTACTC-3?) (Chen et al., 1996). The reverse

PCR primer was added in an amount sufficient to

produce a final concentration of 0.3 mM in the RT-

PCR reaction mix (described below). Immediatelybefore addition to the RT-PCR reaction solution,

the sample RNA/reverse primer mixture was

denatured at 70 8C for 3 min and quenched on ice.

2.2.2. Viral load determination by Real-time

QPCR

A 250 bp region of the HCV 5? NCR was

amplified in the presence of a dual-labeled fluoro-

genic probe specific for the cDNA region between

the forward and reverse primers. When in the

intact form, the probe contains a fluorescentreporter and a quencher dye whose physical

proximity to each other suppress any light emitted

by the reporter. During PCR the 5?0/3? nuclease

activity of the DNA polymerase releases the

reporter dye from the probe and results in an

increase in fluorescence that is then measured by

the GeneAmp† 5700 sequence detection system

(Applied Biosystems). The 5? NCR HCV primersand probe were designed using Primer ExpressTM

version 2.0 software (Applied Biosystems) and

information from the references Chen et al.

(1996), Kawai et al. (1999), Martell et al. (1999),

Takeuchi et al. (1999). All synthetic primers were

purchased from Operon Technologies, Inc.

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/37 29

Reverse transcription and QPCR amplificationswere carried out in a single step, in triplicate using

96-well plates and the SuperscriptTM 1-Step RT-

PCR Kit with Platinum† Taq (Invitrogen). Each

50 ml reaction mixture contained 1X SuperscriptTM

reaction mix, 2.8 mM MgSO4, 0.3 mM forward

primer 1 (5?-CACTCCCCTGTGAGGAACTAC-

TGTCT-3?), 0.3 mM forward primer 2 (5?-CT-

GATGGGGGCGACACTCCACCATGAA-3?),0.15 mM Fluorogenic probe (5?-FAM-TGTACT-

CACCGGTTCCGCAGACCA-TAMRA-3?) and

1 ml of the RT/Platinum† Taq enzyme mix.

Finally, 10 ml of the primed RNA/reverse primer

template was added to each reaction. Reaction

mixtures were initially heated to 50 8C for 60 min

to facilitate cDNA synthesis. An incubation of 10

min at 95 8C activated the Taq polymerase anddenatured the DNA template and reverse tran-

scriptase. Forty-five cycles of denaturation (95 8C/

30 s), annealing (65 8C/15 s) and extension (72 8C/

1 min) were then carried out allowing the con-

current amplification and quantification of the

HCV 5? NCR fragments.

Quantities of HCV RNA, in IU/ml, were

determined by comparing the results for theunknown serum samples to those on a standard

curve (NAPTM HCV panel ranging in concentra-

tion from 500 to 2 000 000 IU/ml [Acrometrix

Corporation] generated from HCV samples cali-

brated to the World Health Organization (WHO)

First International Standard for HCV RNA

(Saldanha et al., 1999). In addition to negative

controls, a 100 ml aliquot of each of the standardswas extracted, amplified and detected in separate

reactions that were run in parallel with the samples

being tested. A standard curve was generated at

the end of each run by plotting the threshold cycle

(CT) (the cycle at which a fluorescence signal

above baseline can be detected) against log10(N ),

where N is the initial concentration of the stan-

dard in IU/ml. HCV RNA concentrations for allunknown samples were subsequently calculated by

interpolation of the standard curve.

2.2.3. Genotyping

The genotype of the infecting HCV virus in each

serum sample was determined by direct sequencing

of the 5? NCR PCR product generated in the one

step Real-time QPCR viral load assay. Briefly, 6 mlof each PCR product was treated with 2 ml of

ExoSAP-IT, as described by the manufacturer

(USB Corporation), to remove excess deoxynu-

cleotides and primers. Next, cycle sequencing was

carried out using the BigDyeTM Terminator v3.0

cycle sequencing reaction kit as per manufacturer’s

instructions (Applied Biosystems). Briefly, the

products of the ExoSAP-IT reaction, 8 ml of theBigDyeTM Terminator Ready Reaction mix and 40

pmols of reverse PCR primer were combined. The

sequencing reaction was carried out for 24 cycles

of denaturation (96 8C/10 s) and annealing/exten-

sion (60 8C/4 min). Excess dye terminators were

removed using the Qiaquick-8† PCR Purification

kit as per manufacturer’s instructions (Qiagen

Inc.). Purified reaction mixtures were vacuumdried and resuspended in 6 ml of loading buffer

(3.2 mM deionized formamide, 5 mM EDTA/10

mg/ml Blue dextran) (Sigma Chemical Co.). The

sequencing reactions were analyzed on an Applied

Biosystems model 373A automated sequencer with

373XL-translation processor upgrade. The se-

quences obtained were aligned with sequences

corresponding to HCV genotypes 1a, 1b, 1c, 2a,2b, 2c, 3a and 3b obtained from GenBank

(Accession numbers: M62321; D31601;

AJ238799; D31602; D14853; D00944; D31604;

AF238485; D10988; D31606; AF238486; D50409;

D17763; AF046866; D28917; D49374, respec-

tively). The SequencherTM 3.0 software program

(GeneCodes Corp.) was then used to determine

genotype identity by sequence similarity.

2.2.4. UW Viral load determination and genotyping

Viral loads were obtained using the bDNA

signal amplification method as described by the

manufacturer (QuantiplexTM HCV RNA; Bayer

Diagnostics). Genotyping was accomplished by

RFLP analysis of 5? NCR HCV RT-PCR products

from the serum samples as described previously(Davidson et al., 1995).

2.2.5. PR Viral load determination and genotyping

Viral load quantification was performed using

earlier lot numbers of AmplicorTM HCV Mon-

itorTM (v 2.0) kit in which viral loads were derived

as copies/ml (Roche Molecular Diagnostics Sys-

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/3730

tems) for serum samples 1�/26, and later lotnumbers of the same assay in which viral loads

were derived as IU/ml for serum samples 27�/40.

This assay also quantifies the copy number of 5?NCR HCV in the patient serum samples by reverse

transcription and QPCR. Genotyping was per-

formed by reverse hybridization of genotype-

specific primers to the above 5? NCR HCV

QPCR products using the line probe assay kit asdescribed by the manufacturer (INNO-LiPA HCV

II, Innogenetics).

2.2.6. Statistical analyses

All analyses were performed in SAS for Win-

dows 2000 version 8.1 (Cary, 1999). Each of the

analytical results has been presented as two

different pairwise comparisons. One comparisonis between the ANMC Viral Hepatitis Program,

and UW. Briefly, HCV viral load and genotype

results determined by the Real-time QPCR and

DNA sequencing methods developed by ANMC

were compared to UW results obtained for the

same 50 specimens using bDNA and RFLP

methodology. Similarly, in the second comparison

ANMC’s HCV viral load and genotype results forthe 40 samples provided by the Ponce School of

Medicine (PR) were compared to the results

obtained by PR using the AmplicorTM HCV

Monitor kit and the line probe assay. The UW

bDNA and RFLP methods were not compared to

PR’s AmplicorTM HCV MonitorTM kit or the line

probe assay. Viral loads were compared by linear

regression of the log10 transformed data in theunits of measure established for each test (Jorgen-

son and Neuwald, 2001). Genotype comparisons

were analyzed and presented as both percent

agreement and either Cohen’s kappa (Cohen et

al., 1960) or Adjusted Rand Statistic (Fisher and

Hoffman, 1988; Hubert and Arabie, 1985; Morey

and Agresti, 1984; Rand, 1971) for each pairwise

comparison.

3. Results

HCV RNA was detected in 46 out of 50 UW

samples analyzed by ANMC’s Real-time QPCR

methodology but was detected in only 42 of the

same samples when analyzed by UW’s Quanti-plexTM HCV bDNA assay, although 2 of these

samples were subsequently genotyped by UVs.

This is likely due to the fact that several serum

samples were below the 200 000 Eq/ml detection

limit of the less sensitive bDNA methodology but

not below the detection limit of the more sensitive

Real-time QPCR assay (Jorgenson and Neuwald,

2001). Viral RNA was detected in all 40 PRsamples analyzed by both ANMC’s Real-time

QPCR assay and either version of the AmplicorTM

HCV MonitorTM assay used by PR.

Because Jorgenson and Neuwald (2001) have

recently published a similar evaluation of different

commercial and research methods for the determi-

nation of HCV viral loads, we determined whether

our statistical comparison results were signifi-cantly different from theirs. However, it should

be noted that we compared results for patient sera

obtained from several clinical laboratories using

different methods, while Jorgenson and Neuwald

compared results only for the NAPTM HCV

standard panel performed within the facilities of

each assay manufacturer.

The viral load data from the Quantiplex HCVbDNA procedure used at UW compared moder-

ately well with the Real-time QPCR HCV RNA

procedure used at ANMC (R2�/0.3813; Fig. 1).

The regression coefficient was 0.41047 and the

intercept was �/3.93362 (resulting in a regression

equation of log10 Eq/ml�/0.41047 log10 IU/ml�/

3.93362). Our regression coefficient of 0.41047

was significantly different from the Jorgensenand Neuwald coefficient of 0.9802 (P�/0.0001)

when comparing the same methodologies (Jorgen-

son and Neuwald, 2001). Similarly our intercept of

�/3.93362 was significantly different from theirs of

�/0.9157 (P�/0.0001) for this pairwise compar-

ison.

Although PR used version 2.0 of AmplicorTM

HCV MonitorTM assay for all samples, earlier lotnumbers determined viral loads as copies/ml, while

later lot numbers determined viral loads as IU/ml.

The viral load data from earlier lot numbers used

by PR compared poorly with the Real-time QPCR

HCV RNA procedure used at ANMC (R2�/

0.0913; Fig. 2). The regression coefficient was

0.23463 and the intercept was �/4.54684 (resulting

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/37 31

in a regression equation of log10 copies/ml�/

0.23463 log10 IU/ml�/4.54684). The later lot num-

bers reporting in IU/ml used by PR performed

much better than the earlier lot numbers reporting

in copies/ml as compared to ANMC results (R2�/

0.6408; Fig. 3). The regression equation obtained

was log10 PR IU/ml�/0.41791 log10 IU/ml�/

3.4386. Our regression coefficient of 0.41791 for

this comparison was still significantly different

from Jorgensen and Neuwald of 0.9081 (P�/

0.0003) (Jorgenson and Neuwald, 2001). Our

intercept of �/3.4386 was also significantly differ-

ent from theirs of 0.4335 (P�/0.0001). Clearly, the

above regression coefficients derived from this

interlaboratory comparison study would result in

a high degree of variability when converting

patient viral loads into the different units of

measurement used by each clinical laboratory.

Overall, these results are significantly different

from Jorgenson and Neuwald’s viral load compar-

ison study which used only the NAPTM HCV

standard panel (Jorgenson and Neuwald, 2001).

Two different sets of HCV genotype data have

been analyzed between ANMC and UW, as well as

between ANMC and PR. The first set is the major

genotype including subtype. The second set is just

the major genotype. The results of each compar-

ison are presented both as percent agreement, and

as Cohen’s kappa or Adjusted Rand Statistic

(Tables 1�/4). Cohen’s kappa (k ) is the amount

of agreement that is corrected for the amount of

agreement to be expected due to chance alone

(Cohen et al., 1960). The Adjusted Rand Statistic

is a measure of agreement when there are an

unequal number of categories (Fisher and Hoff-

man, 1988; Hubert and Arabie, 1985; Morey and

Agresti, 1984; Rand, 1971).

There were several instances of mixtures of

different subtypes at all three sites (note: all

genotyping data indicated that the subtype could

be one or the other, rather than a mixture of

subtypes) (Table 5). Because we were unable to

statistically compare subtype mixtures, obvious

patterns between the sites were used to adjust only

Fig. 1. Viral load comparisons. Quantiplex HCV bDNA at UW compared to Real-time QPCR HCV RNA at ANMC.

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/3732

subtype mixtures by adherence to the following

decisions to collapse mixed subtypes. Analysis of

ANMC�/UW comparisons demonstrated that

whenever UW had genotype 1A (or 2A), ANMC

had 1A/1C mix. In addition, when ANMC�/PR

comparisons were made, whenever PR had 1A or

1A/1B mix, ANMC had 1A/1C mix. Therefore, all

ANMC 1A/1C mixes and PR 1A/1B mixes were

set to 1A. Whenever UW had genotype 1B,

ANMC had 1B or 1A/1B mix. Therefore all

ANMC 1A/1B mixes were set to 1B (except in

one instance when ANMC had 1 and UW had 1B,

in which case the UW result was conservatively set

to 1 by deletion of the subtype). Conversely,

whenever UW had a 1A/1B mix, ANMC had

1A, therefore all UW 1A/1B mixes were set to 1A.

A single sample comparison in which ANMC had

2/2C and UW had 2B resulted in both genotypes

conservatively set to 2. When comparing ANMC�/

PR genotypes, whenever ANMC had 2B, PR had

only the major genotype of 2, therefore each PR

genotype 2 was set to 2B. In no case was there any

change in major genotype within any of the three

sets of data (Table 5). Table 2 (ANMC and UW)

and Table 4 (ANMC and PR) contain the final

genotype distribution comparisons that were used

for the analyses described below.

The ANMC�/UW comparison of major geno-

type with subtype showed a fair to good 86%

agreement and a k�/0.6139 (95% CI, 0.4634�/

0.7644) (Table 1). The ANMC�/UW comparison

of major genotype alone showed 84% agreement

with a k�/0.6983 (95% CI, 0.5086�/0.8881) (Table

1).

For the ANMC�/PR comparison of major

genotype with subtype we were not able to

calculate a k because ANMC used four categories

and PR used six categories. Therefore we calcu-

lated an Adjusted Rand Statistic which showed

85% agreement with an adjusted Rand of 0.358.

The ANMC�/PR comparison of major genotype

alone showed a high agreement of 97.5% with a

k�/0.9471 (95% CI, 0.8444�/1.0498).

Fig. 2. Viral load comparisons. Earlier lot numbers of Amplicor HCV MonitorTM (v 2.0) assay reporting in copies/ml at PR compared

to Real-time QPCR HCV RNA at ANMC.

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/37 33

4. Discussion

We report the interlaboratory comparative ana-

lyses of HCV serum viral load and genotypemethodologies between a recently established

laboratory (ANMC) and two other laboratories

(UW and PR) at distant locations. The methodol-

ogies employed by ANMC allow both viral load

and genotype determination from a single PCR

amplification product. The Real-time QPCR am-

plification and detection steps are performed in a

single tube thus reducing the potential for carry-

over contamination and eliminating post-PCR

product handling. In addition, as the ANMC

Real-time QPCR method uses standards cali-

brated to the WHO First International Standard

for HCV, patient viral loads are reported in IU/ml,

the recently established common unit of measure-

ment for HCV (Jorgenson and Neuwald, 2001;

Saldanha, 1999; Saldanha et al., 1999).

Fig. 3. Viral load comparisons. Later lot numbers of Amplicor HCV MonitorTM (v 2.0) assay reporting in IU/ml at PR compared to

Real-time QPCR HCV RNA at ANMC.

Table 1

Statistical comparisons of HCV genotype between ANMC and

UW

Genotype category % Agreement k 95% CI

Major genotype�/

subtype

86 0.6139 0.4634�/0.7644

Major genotype alone 84 0.6983 0.5086�/0.8881

Table 2

ANMC and UW genotype and subtype distribution

Genotype ANMC (% of total) UW (% of total)

1A 28 (56) 22 (44)

1B 7 (14) 7 (14)

1 1 (2) 1 (2)

2A 0 4 (8)

2B 3 (6) 3 (6)

2 1 (2) 1 (2)

3A 6 (12) 7 (14)

Negative 4 5

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/3734

Overall results from this interlaboratory com-

parison, while moderately good, clearly indicate

that alterations in viral load over time may be

most accurately determined when the same meth-

odology and standards are used. Significantly, the

analyses within this report have demonstrated that

comparison of viral loads using different units,

methods, and standards is difficult to assess, with

different methods demonstrating different levels of

correlation and sensitivity of detection. As an

example, we have failed to replicate Jorgensen

and Neuwald’s statistical results (Jorgenson and

Neuwald, 2001) under our conditions approximat-

ing in vivo circumstances, i.e., real patient sera

analyzed using different assays at different clinical

locations. Despite this, ANMC’s Real-time QPCR

viral load comparisons with UW’s Quantiplex

bDNA method and the newer lot numbers of

AmplicorTM HCV MonitorTM (v 2.0) used by PR

could be considered moderately good statistically,

but only after undergoing logarithmic conversion

for statistical analyses. It follows that interlabora-

tory conversion of results from one methodology

to another for the purpose of monitoring patient

viral load over time would not be highly accurate.

Therefore clinicians should determine treatment

protocols and monitor patient viral loads using the

same assay at each time point. At ANMC, prior to

establishment of the molecular biology laboratory,

specimens for HCV viral loads were sent to a

contract laboratory which, over the course of time

used different subcontract laboratories, each using

different methodologies and reporting viral load

results in different units of measurement. This

resulted in instances where serial specimens from

the same patient were tested using different

methodologies in different laboratories, resulting

in confusion for the patients’ provider. This

phenomenon is not uncommon for hospitals using

bidding processes to select laboratory vendors.

Additionally, based on our results, we would

discourage clinicians from attempting to convert

viral load results from different methods that are

currently performed by different clinical labora-

tories.

Interestingly, genotype comparisons between

ANMC (direct sequencing) and either UW

(RFLP) or PR (reverse hybridization) were fairly

similar (86 and 85%, respectively), when compar-

ing major genotype and subtype together. The

percent agreement of genotype comparisons in-

creased significantly when comparing only major

genotypes between ANMC and PR (97.5%), but

did not change significantly when comparing

ANMC to UW (84%). The discrepancies in

genotype and subtype determination between the

three assays may be due, in part, to the decreased

efficiency with which the indirect methods (RFLP

and reverse hybridization) identify some genotypes

and subtypes (Hawkins et al., 1997; Mellor et al.,

1999; Zein, 2000). The indirect methods can only

detect mutations at certain locations within the

genome and have been found to underestimate the

presence of certain genotypes and subtypes (Bukh

Table 3

Statistical comparisons of HCV genotype between ANMC and PR

Genotype category % Agreement Adjusted Rand Stat. k 95% CI

Major genotype�/subtype 85 0.358 n/a n/a

Major genotype alone 97.5 n/a 0.9471 0.8444�/1.0498

Table 4

ANMC and PR genotype and subtype distribution

Genotype ANMC (% of total) PR (% of total)

1A 21 (52) 19 (48)

1B 6 (8) 7 (17)

1 0 2 (5)

2B 9 (23) 8 (20)

3A 4 (10) 3 (8)

3B 0 1 (2)

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/37 35

et al., 1995; Simmonds, 1995). In contrast, directDNA sequencing of the amplified fragment from

the comparatively stable 5? NCR region of the

HCV genome delivers a larger amount of sequence

information for more accurate genotype and

subtype determination. Although less economical

than the indirect methods, genotype identification

by direct DNA sequencing has become increas-

ingly necessary as new HCV subtypes have beenclassified which cannot be identified by less

specific methods. In addition, the precise identifi-

cation of additional HCV genotype and subtypes

will have increasing epidemiological and research

value, particularly in tracing sources of infection

(Zein, 2000). Furthermore, the one step Real-time

QPCR and genotyping methodology utilized by

ANMC for the study of HCV can easily beadapted for the study and routine diagnosis of

many other viral and bacterial pathogens. Indeed,

immediate future plans at ANMC include the

establishment of Hepatitis B, HIV and TB testing

capabilities using this same technology.

In conclusion, the results of this comparative

study between ANMC, UW and PR clearly show

that one step Real-time QPCR and direct DNAsequencing are rapid, accurate and sensitive meth-

ods for the determination of HCV viral load and

genotype in a clinical laboratory setting. Further-

more, this comparison study clearly indicates that

although different methods of testing patient’s sera

are somewhat comparable to each other when

statistically analyzed, each assay produces suffi-

ciently different results such that no methodshould be substituted for another during a pa-

tient’s treatment, nor for epidemiological, re-

search, or other purposes.

Acknowledgements

This work was supported by the Alaska Scienceand Technology Foundation; Grant Agreement

Number: 99-4-113 (D.G.F., J.W., K.J.W.), and the

Viral Hepatitis Program at the Alaska Native

Medical Center (J.S., J.W., B.M.), and University

of Washington NIH grant 5U19A1048124

(D.G.S., D.R.G.).

Table 5

Original genotypes from each site

ANMC vs UWa ANMC vs PRa

1 1B NEGb 1A/1C 1A

2 1A/1C 2Ab 1A 1A

3 1B 1B 1A 1B

4 NEG NEG 1A/1C 1A/1B

5 1A 2Ab 3A 3A

6 1A 2Ab 1A/1C 1

7 1A 1A 1A 1A

8 1A 1A 1A/1B 1B

9 1A 1A/1B 1A/1C 1A

10 2/2C 2B 1A/1B 1

11 1A/1C NEGb 1A/1C 1A

12 2B 2B 2B 2B

13 1B 1B 3A 3A

14 3A 3A 1A 1A/1B

15 1B 1B 2B 2B

16 2B 2B 3A 3B

17 1A 1A 1B 1B

18 1A 1A 1A 1A/1B

19 1A/1C 1A 1A/1C 1A/1B

20 1A 1A 1A/1C 1A/1B

21 1 1B 2B 2B

22 NEG NEG 2A/2B 2B

23 1A 1A 1A 1A

24 1A 1A 2B 2B

25 1A/1B 1B 1A 1A

26 3A 3A 1A/1B 1B

27 1A 1A 1A 1A

28 1A 1A 3A 3A

29 1A/1B 1B 1A 1A/1B

30 2B 2B 1B 1B

31 1A 1A 1A 1A

32 1A 1A 2Bb 1A

33 3A 3A 2B 2B

34 1A 1A 1A/1B 1B

35 1A 1A 1A 1A

36 1A 1A 1A/1C 1B

37 1A 1A 2B 2B

38 3A 3A 2B 2B

39 1A 1A 1A 1A

40 1A 1A 1A 1A/1B

41 1A 1A

42 1A 1A

43 3A 3A

44 1A 2Ab

45 1A 1A

46 1A/1B 1B

47 NEG 1Bb

48 NEG NEG

49 3A 3A

50 1A 3Ab

a Each genotype in bold was chosen for statistical analysis.b Major genotype disagreement.

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/3736

References

Alter MJ, Kruszon-Moran D, Nainan OV, McQuillan GM,

Gao F, Moyer LA, et al. The epidemiology of hepatitis C

virus infection in the United States, 1988 through 1994. N

Engl J Med 1999;341:556�/62.

Arens M. Clinically relevant sequence-based genotyping of

HBV, HCV, CMV and HIV. J Clin Virol 2001;22:11�/29.

Bukh J, Miller RH, Purcell RH. Genetic heterogeneity of

hepatitis C virus; quasispecies and genotypes. Semin Liver

Dis 1995;15:41�/63.

Carithers RL, Marquarot A, Gretch DR. Diagnostic testing for

hepatitis C. Semin Liver Dis 2000;20:159�/71.

Cary NC. SAS/STAT User’s Guide, Version 8. SAS Institute,

Inc.; 1999.

Chen Y, Cooper DL, Ehrlich GD. Comparative analysis of

three nucleic acid-based detection systems for hepatitis C

virus RNA in plasma from liver transplant recipients. Mol

Cell Probes 1996;10:331�/6.

Cohen J. A coefficient of agreement for nominal scales. Educ

Psychol Measurement 1960;20:37�/46.

Cohen J. The scientific challenge of hepatitis C. Science

1999;285:26�/30.

Davidson F, Simmonds P, Ferguson JC, Jarvis LM, Dow BC,

Follett EA, et al. Survey of major genotypes and subtypes of

hepatitis C virus using RFLP of sequences amplified from

the 5? non-coding region. J Gen Virol 1995;76:1197�/204.

Di Bisceglie AM, Order SE, Klein JL, Waggoner JG, Sjogren

MH, Kuo G, et al. The role of chronic viral hepatitis in

hepatocellular carcinoma in the United States. Am J

Gastroenterol 1991;86:335�/8.

EASL International Consensus Conference on Hepatitis C.

Concensus Statement. J. Hepatol. 1999;30:956�/61.

Fisher DG, Hoffman P. The adjusted Rand statistic: a SAS

macro. Psychometrika 1988;53:417�/23.

Hawkins A, Davidson F, Simmonds P. Comparison of plasma

virus loads among individuals infected with hepatitis C virus

(HCV) genotypes 1, 2, and 3 by quantiplex HCV RNA

assay version 1 and 2, Roche Monitor Assay, and an in-

house limiting dilution method. J Clin Microbiol

1997;35:187�/92.

Hepatitis C Fact Sheet, CDC Website.

Hubert L, Arabie P. Camparing partitions. J Classification

1985;2:193�/218.

Jorgenson PA, Neuwald PD. Standardized hepatitis C virus

RNA panels for nucleic acid testing assays. J Clin Virol

2001;20:35�/40.

Kawai S, Yokosuka O, Kanda T, Imazeki F, Maru Y, Saisho

H. Quantification of hepatitis C virus by Taqman PCR:

comparison with HCV Amplicor Monitor assay. J Med

Virol 1999;58:121�/6.

Le Pogam S, Dubois F, Christen R, Raby C, Cavicchini A,

Goudeau A. Comparison of DNA enzyme immunoassay

and line probe assays (Inno-LiPA HCV I and II) for

hepatitis C virus genotyping. J Clin Microbiol

1998;36:1461�/3.

Management of Hepatitis C, NIH Consensus Statement Online.

Martell M, Gomez J, Esteban JI, Sauleda S, Quer J, Cabot B, et

al. High-throughput Real-Time reverse transcription-PCR

quantitation of hepatitis C virus RNA. J Clin Microbiol

1999;37:327�/32.

McHutchinson JG, Gordon SC, Schiff ER, Shiffman ML, Lee

WM, Rustgi UK, et al. Interferon alpha-2b alone or in

combination with Ribavirin as initial treatment for chronic

hepatitis C. N Engl J Med 1998;339:1485�/92.

Mellor J, Hawkins A, Simmonds P. Genotype dependence of

Hepatitis C virus load measurements in commercially

available quantitative assays. J Clin Microbiol 1999:37.

Morey L, Agresti A. The measurement of classification agree-

ment: an adjustment to the Rand statistic for chance

agreement. Educ Psychol Measurement 1984;44:33�/7.

Morishima C, Gretch DR. Clinical use of hepatitis C virus tests

for diagnosis and monitoring during therapy. In: Keefe EB,

editor. Treatment of chronic hepatitis C. Philadelphia:

Saunders, WB, 1999:717�/40.

Pawlotsky J-M, Bouvier-Alias M, Hesode C, Darthuy F,

Remire J, Dhumeaux D. Standardization of hepatitis C

virus quantification. Hepatology 2000;32:654�/9.

Poynard T, Marcellin P, Lee SS, Niederau C, Minuk GS, Ideo

G, et al. Randomized trial of Interferon alpha-2b plus

Ribavirin for 48 weeks or 24 weeks versus Interferon alpha-

2b plus placebo for 48 weeks for treatment of

chronic infection with hepatitis C virus. Lancet

1998;352:1426�/32.

Rand WM. Objective criteria for the evaluation of clustering

methods. J Am Stat Assoc 1971;66:846�/50.

Saldanha J. Standardization: a progress report. Biologicals

1999;27:285�/9.

WHO Collaborative Study Group, Saldanha J, Lelie N, Heath

A. Establishment of the First International Standard for

Nucleic Acid Amplification Technology (NAT) Assays for

HCV RNA. Vox Sang 1999;76:149�/58.

Simmonds P. Variability of hepatitis C virus. Hepatology

1995;21:570�/83.

Takeuchi T, Katsume A, Tanaka T, Abe A, Inoue K,

Tsukiyama-Kohara K, et al. Real-Time detection system

for quantification of hepatitis C virus genome. Gastroenter-

ology 1999;116:636�/42.

Thomas DL, Lemon SM. Hepatitis C. In: Mandell GL, Bennett

JF, Dolin R, editors. Principles and practice of disease,

2000:1736�/60.

Tibor L, Funk E, Beller M. Hepatitis C. State Alaska

Epidemiol Bull 1999:3.

Zein NN. Clinical significance of hepatitis C virus genotypes.

Clin Microbiol Rev 2000;13:223�/35.

J.C. Anderson et al. / Journal of Clinical Virology 28 (2003) 27�/37 37