P

JMa

b

c

d

e

a

ARRA

KAHAHHATS

1d

Journal of Clinical Virology 43 (2008) 169–175

Contents lists available at ScienceDirect

Journal of Clinical Virology

journa l homepage: www.e lsev ier .com/ locate / j cv

erformance evaluation of the ADVIA Centaur® anti-HBe and HBeAg assays

osef van Heldena,1, Carl Cornelyb,2, Francesco Dati c,3, H. Roma Levyd,∗, Tricia Bald,4,ary Seegere,5, Theodore Wrighte,6, Lawrence Bakere,7

Gemeinschaftspraxis fuer Laboratoriumsmedizin, Mikrobiologie und Humangenetik, Dr. Stein & Kollegen, Wallstrasse 10, D41061 Moenchengladbach, GermanyGemeinschaftspraxis für Laboratoriumsmedizin und Mikrobiologie, Dres. Cornely, Riebe und Berndt, Wallstrasse 56, D-52064 Aachen, GermanyIVD-Consulting, Pappelweg 18, D-35041 Marburg, GermanySiemens Healthcare Diagnostics, 5210 Pacific Concourse Drive, Los Angeles, CA 90045, United StatesSiemens Healthcare Diagnostics, 511 Benedict Avenue, Tarrytown, NY 10591, United States

r t i c l e i n f o

rticle history:eceived 15 January 2008eceived in revised form 24 May 2008ccepted 28 May 2008

eywords:nti-HBeBeAgDVIA Centaurepatitis B virusBVbbott AxSYM

a b s t r a c t

Background: Detection of HBeAg and anti-HBe is valuable for the evaluation and therapeutic managementof hepatitis B infection.Objectives: To determine the clinical performance of the newly CE-approveda HBeAg and anti-HBe assayson the fully automated, random access ADVIA Centaur® immunoassay system.Study design: Patient samples collected at two sites were used to compare the ADVIA Centaur assaysto Abbott AxSYMTM assays. Consensus of discordant results was reached using Roche Elecsys® assays.Additionally, two well-characterized seroconversion panels were evaluated.Results: The ADVIA Centaur HBeAg assay sensitivity was 100% and specificity was 99.5%. The ADVIA Cen-taur anti-HBe assay sensitivity was 100% and the resolved specificity was 98.2%. Fewer samples requiredretesting with the ADVIA Centaur assays than with the AxSYM. In two well-characterized seroconversionpanels, the ADVIA Centaur anti-HBe assay detected anti-HBe 20–25 days earlier than the AxSYM assay;

andem MSiemens Healthcare Diagnostics

the ADVIA Centaur and AxSYM HBeAg assays detected HBe reactivity on the same day.Conclusions: The ADVIA Centaur HBeAg and anti-HBe assays demonstrated good sensitivity and specificity,

linffe

and thus are suitable for cassays may be more cost e

1. Introduction

An estimated 360 million people are chronically infected withhepatitis B virus (HBV) worldwide.1 Infection is associated withthe development of cirrhosis and hepatocellular carcinoma. Several

Abbreviations: HBV, hepatitis B virus; HBeAg, HBe antigen; anti-HBe, antibod-ies to HBeAg; mAb, anti-HBe monoclonal antibody; RLUs, relative light units; CTS,common technical specifications.

∗ Corresponding author. Tel.: +1 310 645 8200x7028; fax: +1 310 649 1909.E-mail addresses: jvhelden@labor-stein.de (J. van Helden),

gpcr.labor@t-online.de (C. Cornely), cornely@gpcr.de (F. Dati),helene.levy@siemens.com (H.R. Levy),tricia.bal@siemens.com (T. Bal), mary.seeger@siemens.com(M. Seeger), theodore.wright@siemens.com (T. Wright),lawrence.baker@siemens.com (L. Baker).

1 Tel.: +49 2161 819 4408.

2 Tel.: +49 241 477 880.3 Tel.: + 49 6421 360181.4 Tel.: +1 310 645 8200x7065; fax: +1 310 649 1909.5 Tel.: +1 914 524 2908; fax: +1 914 524 2500.6 Tel.: +1 914 524 2596; fax: +1 914 524 2543.7 Tel.: +1 914 524 3807; fax: +1 914 524 2543.

386-6532/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.jcv.2008.05.008

ical use. Their novel algorithms require reduced retesting, suggesting thesective.

© 2008 Elsevier B.V. All rights reserved.

routine serologic tests, including HBe antigen (HBeAg) and anti-bodies to HBeAg (anti-HBe), are used to distinguish recovered orrecovering patients from those with chronic infection. Chronic HBVinfections are typically characterized by prolonged elevated HBeAgassociated with severely delayed and attenuated anti-HBe response(i.e., poor seroconversion).

Recent data suggest that HBeAg may modulate host responsethrough induction of the interleukin-1 pathway, thereby promot-ing T-helper2—like responses that lead to suppression of the hostimmune response. This suppression facilitates viral persistenceand inhibits viral clearance,2 which is characteristic of the chronicpatient.

HBeAg in patient sera indicates ongoing HBV replicationand liver disease; anti-HBe is associated with lower levels ofviremia and remission of liver disease. Additionally, HBeAg-positivepatients are at increased risk for progression to chronic active(replicative) hepatitis and cirrhosis, and are at highest risk for hep-

atocellular carcinoma (HCC): studies comparing men negative forboth HBsAg and HBeAg to men positive for either HBsAg alone orin addition to HBeAg showed that the relative risks for developingHCC were 9.6, and 60.2, respectively.3,4

170 J. van Helden et al. / Journal of Clinical Virology 43 (2008) 169–175

Table 1The study populations

Patient type Anti-HBe study n = 625 HBeAg study n = 621

Anti-HBe (+) (n = 216) Anti-HBe (−) (n = 409) HBeAg (+) (n = 218) HBeAg (−) (n = 403)

Chronic infection 216 0 196 0Acute infection 0 1 22 1a

Hospitalized/clinic patient 0 201 0 202Blood donor 0 207 0 201

HBV infection and HBeAg/anti-HBe status as originally determined using the Roche ElecSys anti-HBe and HBeAg assays.a This sample was not included in the specificity study because it was not drawn from a patient known to be infected with HBV. In accordance with CTS guidelines,

specificity is determined using samples from low-risk populations presumed to be HBV-negative.

Table 2Reactivity criteria

ADVIA anti-HBe AxSYM anti-HBe ADVIA HBeAga AxSYM HBeAg

Reactive ≥1.2 index ≤1.0 S/CO ratio ≥1.0 index ≥1.0 S/CO ratioNonreactive <0.80 index >1.0 S/CO ratio <1.0 index <1.0 S/CO ratioEquivocal ≥0.80 index/<1.2 index – – –

a ADVIA HBeAg initial results that are ≥0.80 index and <10.0 index are retested in duplicate and HBeAg status determined by two of three results using reactivity criteriaabove.

Fig. 1. Testing algorithms for the method comparison studies.

J. van Helden et al. / Journal of Clinical Virology 43 (2008) 169–175 171

Table 3Number of patient samples requiring retesting on each system

YM

NP

fcaaaAopratstHmHeai

aCDamal

ADVIA anti-HBe AxS

umber 10 219ercent of total tests 1.6 35

Anti-HBe-positive seroconversion serves as a treatment goalor HBeAg-positive patients.5 Treatment of HBeAg-positive (i.e.,hronically infected) patients can increase seroconversion tonti-HBe-positive status by 23%, which reduces cirrhosis and hep-tocellular carcinoma by 47% and 78%, respectively.6 Thus, HBeAgnd anti-HBe testing play important roles in routine monitoring.lthough most chronic infections are characterized by expressionf HBeAg, mutations in the precore region of the viral genome canrevent or down-regulate HBeAg production, even though HBVeplication is ongoing. Such mutant strains may be infrequentlycquired with the initial infection7 but more typically arise spon-aneously after seroconversion. This can occur either shortly aftereroconversion or decades later, and is characterized by consis-ently elevated levels of ALT and HBV DNA levels in the face ofBeAg-negative serology in the majority of patients, but may alsoanifest as spikes in ALT and HBV DNA levels in 30–40% of chronicBeAG− infections.8 In either case, such patients typically experi-nce a more virulent progression of the disease and worse outcome,nd a high sensitivity and specificity assay for HBeAg can play anmportant role in identifying these patients.

We present clinical performance data for the new HBeAg andnti-HBe assays on the fully automated, random access ADVIAentaur immunoassay system (Siemens Healthcare Diagnostics,

eerfield, IL, USA). These assays are CE-approveda (see footnote) and are used to aid in determination of the status and manage-ent of HBV infection by the qualitative detection of HBeAg and

nti-HBe in human serum and plasma (EDTA, sodium heparin orithium heparin).

Fig. 2. Testing algorithm used to determine concordan

anti-HBe ADVIA HBeAg AxSYM HBeAg

10 2071.6 33

2. Methods

The protocols used in the performance evaluations of theADVIA Centaur® anti-HBe and HBeAg assays conformed to therequirements of the EU Commission decision of 7 May 2002 oncommon technical specifications (CTS) for in vitro diagnostic med-ical devices.9 The primary comparator methods used were theAbbott AxSYM anti-HBe and HBeAg assays (Abbott Diagnostics,Abbott Park, IL, USA).

2.1. Population studies

Fresh and frozen samples were used in the anti-HBe (n = 625)and HBeAg (n = 622) studies. Approximately equal numbers of sam-ples were drawn at two sites from hospitalized and clinic patients,in addition to unselected donors (Table 1). Two reagent lots weretested for each assay.

The ADVIA Centaur anti-HBe assay employs a two-step competi-tive format using recombinant HBeAg, acridinium-labeled anti-HBemonoclonal antibody (mAb), and streptavidin-coated paramag-netic microparticles preformed with biotinylated mouse anti-HBemAb. The anti-HBe activity present in the patient sample isinversely related to the relative light units (RLUs) detected by the

system.

The ADVIA Centaur HBeAg assay is a two-wash sandwichimmunoassay employing biotinylated anti-HBe mAb bound tostreptavidin-coated microparticles and acridinium ester-labeledanti-HBe mAb. HBeAg reactivity is directly related to the RLUs. The

ce in the population/method comparison study.

172 J. van Helden et al. / Journal of Clinical Virology 43 (2008) 169–175

Table 4Initial specificity (a) and resolved specificity (b) of the ADVIA Centaur anti-HBe assayas determined from specimens derived from a low-risk (presumed negative) donorpopulation

aa

AxSYM

Positive Negative

ADVIA Centaur anti-HBe Positive 8 11Equivocal 0 5Negative 0 385

bb

Resolution method

Positive Negative

ADVIA Centaur anti-HBe Positive 11 7Equivocal 0 4Negative 0 385

Resolution was achieved by retesting discordant samples on the Roche ElecSys.(Equivocal results were excluded from the calculations.)

a Initial specificity = 97.22% (385/396); 95% CI = 95.08–98.61%; equivocal = 1.22%.b

c

ruc

ccrcwrbara

pmCaa(

iacwe

TSf

A

Ds

Table 6Initial sensitivity (a) and resolved sensitivity (b) of the ADVIA Centaur anti-HBe assayas determined from specimens derived from HBV-positive patients

aa

Abbott AxSYMa

Positive Negative

ADVIA Centaur anti-HBe Positive 211 1Equivocal 2 0Negative 0 2

ab

Resolution methodb

Positive Negative

ADVIA Centaur anti-HBe Positive 212 0Equivocal 2 0Negative 0 2

Resolution was achieved by retesting the discordant samples on the Roche ElecSys(equivocal results were excluded from calculations, per CTS specifications).

a Initial sensitivity = 100.00% (211/211); 95% confidence interval = 98.27–100.00%;e

c

2

tGr

3

cCap

3

tCCw

sOleft for consensus testing. An “indeterminate” sample (equivocal byCentaur, AxSYM-negative/ElecSys-positive) was also removed from

Resolved specificity = 98.21% (385/392); 95% CI = 96.36–99.28%; equivo-al = 0.98%.

esults of each test were determined according to the index val-es established with the calibrators and read from a stored masterurve (Table 2).

Systems were calibrated using high and low calibrators in tripli-ate. All runs included positive and negative controls and the studyommenced only when all controls were within the acceptableange. Assays were run according to the manufacturers’ specifi-ations. All equivocal samples were retested. In addition, samplesere retested if the ADVIA Centaur or the Abbott AxSYM assay

esults were discordant. Samples were not retested if a result byoth manufacturers was nonreactive. One sample included in thenti-HBe evaluation contained insufficient volume to allow forepeat testing and was discarded. Results for this sample werechieved for the HBeAg assay evaluation, however.

All AxSYM anti-HBe-positive samples (≤1.0 index) and HBeAg-ositive samples (≥1.0 index) required retesting in duplicate per theanufacturer’s instructions. In contrast, retesting with the ADVIA

entaur HBeAg assay is not necessary, only for results between 0.8nd 10.0 index. ADVIA Centaur anti-HBe results ≥0.8 and <1.2 indexre considered equivocal by the manufacturer and were retestedFig. 1, Table 3).

For the population/method comparison studies, repeat test-ng was performed on discordant samples: sample reactivity wasssigned by agreement (two of three results). If no determinationould be made, a third assay (Elecsys; Roche, Basel Switzerland)as run according to the manufacturer’s specifications and used to

stablish consensus (Fig. 2).

able 5pecificity of the ADVIA Centaur HBeAg assay as determined from specimens derivedrom a low-risk (presumed negative) donor population (n = 403)

AxSYM HBeAg

Positive Negative

DVIA Centaur HBeAgPositive 0 2Negative 0 401

iscordant samples were retested on the Roche ElecSys, but yielded identical results;pecificity = 99.5% (401/403); 95% CI = 98.22–99.94%.

to

TId

A

N

quivocal = 0.94%.b Resolved sensitivity = 100% (212/212); 95% CI = 98.28–100.00%; equivo-

al = 0.93%.

.2. Seroconversion panels

Two reagent lots for each ADVIA Centaur assay were tested usingwo commercial seroconversion panels (Life Diagnostics, Clarkston,A, USA). Because the commercial panels are well characterized, no

epeat testing or resolution testing was required.

. Specificity

Specificity was calculated from the HBe-negative populationomprised of random blood donors and hospitalized patients, perTS guidelines. One HBe-negative sample was excluded from thenalysis because it was drawn from a patient known to be HBV-ositive.

.1. ADVIA Centaur anti-HBe

According to the AxSYM assay, 396 samples were negative. Ofhese, 385 were negative and 11 positive according to the ADVIAentaur assay, yielding an initial specificity of 97.22% (385/396, 95%I = 95.08–98.61%). Following the CTS guideline, equivocal samplesere not included in the calculations.

The 11 discordant samples were tested again using the consen-us method (Roche Elecsys); 7 were negative and 3 were positive.ne sample was unresolved because there was not enough volume

he final analysis. Overall, these results yielded a resolved specificityf 98.21%, surpassing the CTS requirement of 98% (Table 4).

able 7nitial sensitivity of the ADVIA Centaur HBeAg assay as determined from specimenserived from HBV-positive patients

AxSYM HBeAg Totals

Positive Negative

DVIA Centaur HBeAgPositive 206 0 206Negative 0 12 12Totals 206 12 218

o resolution was required; sensitivity = 100% (206/206); 95% CI = 98.23–100.00%.

J. van Helden et al. / Journal of Clinical Virology 43 (2008) 169–175 173

Table 8Characteristics of the seroconversion panels from Life Diagnostics

Shading indicates reactive results. The first bleeds at which the anti-HBe and HBeAg assays became reactive arenoted in bold italics.

1 Clinica

3

owtr

4

HaH

4

aOri

4

icC

TRR

Tt

5

arbtpaeoa

HAatA

H5a

6

o

74 J. van Helden et al. / Journal of

.2. ADVIA Centaur HBeAg

The predicate HBeAg assay recorded 403 negative results: 401f these were negative according to the ADVIA Centaur assay and 2ere positive, yielding an initial specificity of 99.5% and surpassing

he CTS requirement (Table 5). Consensus testing of the discordantesults did not affect the initial specificity.

. Sensitivity

Sensitivity was calculated using the samples acquired fromBV-positive patients, per CTS requirements. Patients were eithercutely or chronically infected, according to HBeAg status and otherBV markers.

.1. ADVIA Centaur anti-HBe

Of 216 anti-HBe-positive samples tested, 211 were positiveccording to both the AxSYM and ADVIA Centaur anti-HBe assays.ne initially discordant sample (ADVIA-positive/AxSYM-negative)

esolved in favor of the ADVIA Centaur assay (Table 6), yielding annitial and resolved sensitivity of 100%.

.2. ADVIA Centaur HBeAg

Of 218 HBeAg-positive samples tested, 206 samples were pos-tive and 12 were negative using both manufacturers’ assays. Noonsensus testing was necessary, and the sensitivity of the ADVIAentaur HBeAg assay was 100% (Table 7).

able 9eactivity of the ADVIA and AxSYM anti-HBe assays with seroconversion panelsP-016 and RP-009

he bleeds/days at which the assays were first reactive are in bold italics and iden-ified in the table footnotes (see Table 2 for cut off values).

mafis

TRa

Tt

l Virology 43 (2008) 169–175

. Patient seroconversion panels

The anti-HBe and HBeAg assays were tested with commerciallyvailable seroconversion panels acquired from two patients as theyecovered from acute HBV infection. The panel profiles (suppliedy Life Diagnostics, Table 8) indicated that both patients proceededhrough infection and recovery as expected for acute disease. Bothatients became HBsAg positive and seroconverted from HBsAg tonti-HBs within the time frame of the seroconversion panel. Differ-nces in the panel results are attributed to the more rapid recoveryf the RP-016 patient, as is evidenced by the difference in HBsAg tonti-HBs seroconversion between the two patients.

Table 9 demonstrates that, in our study, the ADVIA Centaur anti-Be assay detected reactivity with both panels earlier than thexSYM method (RP009: day 56 [bleed 9] vs. day 81 [bleed 11];nd RP016 day 60 [bleed 8] vs. day 81 [bleed 11]), indicating thathe ADVIA Centaur anti-HBe assay may be more sensitive than thexSYM assay.

As indicated in Table 10, seroconversion panel reactivity to bothBeAg assays was identical (RP009: day 11 [bleed 3]; RP016: day7 [bleed 7]) suggesting that the ADVIA Centaur and AxSYM HBessays share similar sensitivity.

. Discussion

HBeAg and anti-HBe are important markers in the managementf chronic hepatitis B infection along with other HBV serologic and

olecular tests. Our study shows that the ADVIA Centaur HBeAg

nd anti-HBe assays meet the benchmarks for sensitivity and speci-city required by the CTS. For both assays, specificity is >98% andensitivity is 100%. Furthermore, sensitivity and specificity were

able 10eactivity of the ADVIA and AxSYM HBeAg assays with seroconversion panels RP-016nd RP-009

he bleeds/days at which the assays were first reactive are in bold italics and iden-ified in the table footnotes (see Table 2 for cut off values).

nical

10. Dati F, Denoyel G, van Helden J. European performance evaluations of theADVIA Centaur infectious disease assays: requirements for performance eval-uation according to the European directive on in vitro diagnostics. J Clin Virol

J. van Helden et al. / Journal of Cli

comparable to the predicate assay, although a few discordant sam-ples were noted using the anti-HBe assay. Earlier reactivity withcommercial third-party seroconversion panels suggests that theADVIA Centaur anti-HBe assay is more sensitive than the predicateassay, although a definitive demonstration of greater sensitivity isbeyond the scope of this current investigation.

Unlike the AxSYM assays, the ADVIA Centaur HBe assays do notrequire retesting of all positive results (in this study, 33–35% vs. 1.6%,respectively). Given the clinical use of HBe assays as a third-linemonitoring tool for HBV infections, a high proportion of HBe testresults will be positive, thus making the ADVIA Centaur HBe testingalgorithm very efficient and suggesting greater cost effectiveness.

Previous performance evaluations have reported on the qual-ity of the anti-HBs, anti-HBc total, anti-HBc IgM, HBsAg, andHBsAg confirmatory tests on the ADVIA Centaur system.10,11 Theaddition of the CE-marked anti-HBe and HBeAg assays com-pletes the comprehensive HBV immunoassay menu for thisanalyzer.

References

1. World Health Organization (WHO). Hepatitis B [monograph on the Inter-net]. Fact sheet no. 204. Geneva: WHO; revised Oct 2000. http://www.who.int/mediacentre/factsheets/fs204/en/ [cited Nov 2007].

1

Virology 43 (2008) 169–175 175

2. Yang CY, Kuo TH, Ting LP. Human hepatitis B viral e antigen interacts withcellular interleukin-1 receptor accessory protein and triggers interleukin-1response. J Biol Chem 2006;281(45):34525–36.

3. Ganem D, Prince AM. Hepatitis B virus infection—natural history and clinicalconsequences. N Engl J Med 2004;350(11):1118–29.

4. Yang HI, Lu SN, Liaw YF, You SL, Sun CA, Wang LY, et al. Taiwan Community-Based Cancer Screening Project Group. Hepatitis B e antigen and the risk ofhepatocellular carcinoma. N Engl J Med 2002;347(3):168–74.

5. Lai CL, Yuen MF. The natural history and treatment of chronic hepatitis B: acritical evaluation of standard treatment criteria and end points. Ann InternMed 2007;147(1):58–61.

6. Lin SM, Yu ML, Lee CM, Chien RN, Sheen IS, Chu CM, et al. Interferon ther-apy in HBeAg positive chronic hepatitis reduces progression to cirrhosis andhepatocellular carcinoma. J Hepatol 2007;46(1):45–52 [Epub 2006 October 20].

7. Chang MH. Hepatitis B virus mutation in children. Int J Pediatr 2006;73:803–7.

8. Saikia N, Talukdar R, Mazumder S, Khanna S, Tandon R. Management of patientswith HBeAg-negative chronic hepatitis B. Postgrad Med J 2007;83:32–9.

9. Liikanen E. Commission Decision of 7 May 2002 on common technicalspecifications for in vitro-diagnostic medical devices. Official Journal of theEuropean Communities, 16.5.2002, L131/17-30. Accessed from http://eur-lex.europa.eu/LexUriServ/site/en/oj/2002/l 131/l 13120020516en00170030.pdf[cited Aug 2007].

2004;30(Suppl. 1):S6–10.1. van Helden J, Denoyel GA. Experience with the IVDD performance evalua-

tions of the ADVIA Centaur infectious disease assays. J Clin Virol 2004;30(Suppl.1):S16–8.