Automated screening of blood donations for hepatitis C virus RNA using the Qiagen BioRobot 9604 and the Roche COBAS HCV Amplicor assay

169

Vox Sanguinis (2002) 82, 169–176

ORIGINAL PAPER © 2002 Blackwell Science

Blackwell Science, LtdAutomated screening of blood donations for hepatitis C virus RNA using the Qiagen BioRobot 9604 and the Roche COBAS HCV Amplicor assayP. R. Grant,1 C. M. Sims,2 F. Krieg-Schneider,3 E. M. Love,4 R. Eglin5 & R. S. Tedder1

1Department of Virology, Royal Free and University College Medical School, The Windeyer Institute, London, UK 2Technical Division, Bio Products Laboratory, Elstree, Herts., UK 3Qiagen GmbH, Hilden, Germany 4National Blood Service, Manchester Blood Centre, Plymouth Grove, Manchester, UK 5National Blood Service, North London Blood Centre, London, UK

Background and Objectives In order to reduce the potential for transmission ofhepatitis C virus (HCV) from an RNA-positive, anti-HCV-negative blood donation, theNational Blood Service (NBS) introduced nucleic acid amplification technology (NAT)testing for HCV in England and Wales. The objective of this study was to develop anautomated assay using commercial components for the detection of HCV RNA in blooddonations for transfusion.

Materials and Methods The Qiagen QIAamp 96 ‘Viral RNA’ and ‘Virus’ BioRobot kitsfor HCV RNA extraction, and the Roche COBAS HCV Amplicor v2·0 and AmpliScreenv2·0 assays for polymerase chain reaction (PCR) amplification and detection, wereinvestigated.

Results QIAamp technology and the BioRobot 9604 allow automation of theviral RNA extraction process. By combining the automated silica-membrane basedQIAamp 96 Virus extraction and automated reverse transcription–polymerase chainreaction (RT–PCR) set-up with COBAS HCV AmpliScreen v2·0 amplification anddetection it is possible to achieve a 95% detection level for HCV of 12·8 IU/ml. Cross-contamination studies have shown that use of the BioRobot 9604 does not pose adetectable contamination risk. Between 1999 and 2001, ≈ 6·8 × 106 donations weretested in England and Wales, of which only four were found to contain RNA withoutanti-HCV.

Conclusions This combination of methods results in an assay with a high samplethroughput, little ‘hands-on’ time and fast turnaround time that is also sufficientlysensitive to allow testing of pools of up to 96 samples at a time. These methods havebeen successfully introduced into routine use within the NBS for release of bloodcomponents with a shelf-life of longer than 24 h.

Key words: automation, blood screening, hepatitis C virus (HCV).

Received: 4 September 2001, revised 31 October 2002, accepted 2 February 2002

Introduction

The introduction of, and improvements in, screening of theblood supply for markers of viral infection have dramaticallyreduced the incidence of post-transfusion infection. However,there remains a period between the donor being infected

Correspondence: Paul R. Grant, Department of Virology, Royal Free and University College Medical School, The Windeyer Institute, 46 Cleveland Street, London, UK E-mail: [email protected]

170 P. R. Grant et al.

© 2002 Blackwell Science Ltd. Vox Sanguinis (2002) 82, 169–176

and the development of detectable serological markers: thisis known as the ‘window phase’ of infection.

The window phase for hepatitis C virus (HCV) has beenestimated to average 82 days for second-generation enzyme-linked immunosorbent assays (ELISAs) [1], which is reducedto 66 days for third-generation ELISAs [2,3]. Hepatitis Bvirus (HBV) and human immunodeficiency virus (HIV) havewindow phases of 59 and 22 days, respectively [1]. Thus,even with proficient serological screening there is a residualinfection risk, and cases of viral infection in patients whoreceived blood or blood products from donors in the windowphase of the infection have been documented [4].

Reverse transcription–polymerase chain reaction (RT–PCR) is able to detect the presence of HCV RNA in plasma asearly as 6 days after infection [5], and HIV RNA at 11 daysafter infection. PCR can also detect HBV DNA 34 days afterinfection [1] and in the preantigenaemic phase of acuteinfection [6]. The greater potential reduction of the windowphase in HCV compared with the other viruses is one of thereasons that HCV has been the first virus to be screened forby nucleic acid amplification technology (NAT).

The European Committee for Proprietary Medicinal Products(CPMP) now requires all plasma used for the manufactureof medicinal products to be tested for HCV RNA by NAT [7].This requirement relates solely to plasma for fractionationrather than cellular components for transfusion. The NationalBlood Service (NBS) of England and Wales embarked onimplementing an HCV NAT blood-screening programme inorder to safeguard the supply of plasma for fractionation.Although British plasma is no longer fractionated in the UKbecause of the theoretical risk of transmission of variantCreutzfeld–Jakob disease (vCJD) [8], NAT has recently beenintroduced to screen blood components for transfusion priorto release.

The rapid initial ramp up of HCV viraemia and the resultinghigh level allows the number of tests requested to be greatlyreduced by pooling the donations, with positive pools beingresolved down to the single, positive donation [9]. Assaysfor testing the pooled donations need to be highly sensitive,operationally robust and able to retain a high throughput.Automation and electronic sample tracking are obvious require-ments. Finally, commercially available components are thepreferred choice, as much of the quality control and batchrelease testing would be performed by the manufacturer.

At the time this project was started, the choice of com-ponents was limited. It was therefore decided initially toinvestigate the QIAamp 96 ‘Viral RNA’ kit (Qiagen, Hilden,Germany) and the Qiagen BioRobot 9604 (Qiagen) forautomating RNA extraction and PCR set-up, and the RocheAmplicor v2·0 assay (Roche Molecular Systems Inc., PleasantonCA) and the COBAS analyser (Roche) [10] for amplificationand detection of HCV RNA. During the introduction of theseassays into routine use within the NBS, new assays were

released by both Qiagen and Roche. Thus, the combinationof the new QIAamp 96 ‘Virus’ kit for the BioRobot 9604and the new AmpliScreen v2·0 assay for the COBAS wereevaluated and have now replaced the older assays withinthe NBS.

Materials and methods

Pooling

A dedicated blood sample is taken into a dry EDTA tube atthe time of donation [11]. The samples are sent to the poolinglaboratory where they are separated and 50 µl of plasmafrom each is added to create a minipool using a Tecan Genesis200 Robotic Sample Processor (RSP; Tecan Group Ltd., Zurich,Switzerland). A 1-ml aliquot of each sample is also archivedin a deep 96-well plate at this time. Anti-HCV-positive dona-tions are not withheld from the minipool as pooling occursbefore the serology results are available. The minipool sizewas originally 96 donations when NAT screening was usedonly for release of components with a shelf-life of longerthan 35 days. The pool size has now been reduced to 48donations with the introduction of NAT screening for releaseof all components with a shelf-life of longer than 24 h.

HCV RNA extraction using the QIAamp96 Viral RNA assay on the Qiagen BioRobot 9604

The manufacturer’s instructions are as follows: 200 µl ofplasma is added to 800 µl of viral lysis buffer containingcarrier RNA and incubated at room temperature for 10 minafter which 800 µl of 100% ethanol is added. This mixture ispassed through the 96-well QIAamp plate by vacuum andwashed twice. The RNA is eluted from the matrix in 60 µl ofRNAse-free water. The manufacturer’s protocol was modifiedto allow compatibility with the Roche Amplicor v2·0 assay.The changes were as follows:• the concentration of carrier RNA in the viral lysis buffer

was decreased from 7·1 µg/ml to 0·825 µg/ml;• the Roche Amplicor v2·0 internal control was added to the

viral lysis buffer (AVL) at a concentration of 4·44 µl/ml ofAVL; and

• the elution volume was increased to 70 µl to allow ade-quate eluate for addition to the Amplicor assay, as ≈ 20 µlis retained in the silica matrix of the QIAamp plate.

The modifications listed above are referred to as the standardprotocol.

HCV RNA extraction using the QIAamp96 Virus assay on the Qiagen BioRobot 9604

The manufacturer’s instructions are as follows: 200 µl of plasmais added to 200 µl of viral lysis buffer containing carrier RNA


Automated HCV blood screening 171

and incubated at 56 °C for 10 min after which 230 µl of 100%ethanol is added. This mixture is passed through the 96-wellQIAamp plate by vacuum and washed twice. The RNA iseluted from the matrix in 86 µl of RNAse-free water.

The Roche AmpliScreen v2·0 internal control was addedto the viral lysis buffer (AL) at a concentration of 8·33 µl/mlof AL.

HCV RNA amplification and detection using Roche HCV Amplicor v2·0 and HCV AmpliScreen v2·0

Fifty microlitres of the HCV RNA eluate from the BioRobot9604 were added to 50 µl of the Roche activated mastermix. All subsequent stages were performed according to themanufacturer’s instructions, either by the manual microwellplate assay or by using a COBAS analyser [10].

Automated addition of master mix and nucleic acid to COBAS A-rings

With the COBAS automated version of the assay, the BioRobotRNA extract is transferred to the COBAS A-ring using theBioRobot. The master mix is made up according to themanufacturer’s protocol. The activated master mix, BioRobotnucleic acid extract and two COBAS A-rings in an adaptorsupplied by Qiagen are placed onto the BioRobot which thenpipettes the master mix and nucleic acid extraction into theA-rings. The barcodes on the extract block and the A-ringsare scanned by the user therefore maintaining electronicsample tracking throughout the process.

Resolution of HCV RNA-reactive pools

If a pool is found to contain HCV RNA, cross-pools of eightor 12 donations are made from the rows and columns of the96-well archive plate by the Tecan Genesis RSP (Tecan GroupLtd.). The 20 cross-pools are then tested and the positionof the positive sample is identified from the intersect ofthe positive row and column cross-pools. The individualsample is taken from the well and tested separately as a finalconfirmation.

HCV RNA standards and diluent

The following HCV RNA standards were used in these studies:National Institute for Biological Standards and Control(NIBSC, South Mimms, UK) HCV RNA working reagent96/586 (710 IU/ml) and the Paul Ehrlich Institute (PEI,Langen, Germany) HCV RNA reference preparation no. 75/98(25 000 IU/ml). These standards have been calibrated, in aninternational study, against the World Health Organization(WHO) HCV RNA international standard (97/790; NIBSC) to

give values in International Units per millilitre (IU/ml) [12].The standards described above were diluted in pooledEDTA plasma that was confirmed as HCV RNA negative byRT–PCR. In order to obtain a sufficiently large volume ofEDTA plasma, donors were bled into custom EDTA bloodpacks. Nine of these donations were tested for HCV RNAand, when found to be negative, were pooled and frozen inaliquots.

Statistical analysis

The 95% limits of detection for the various methods werecalculated using probit analysis with the Arcus QuickStatversion 1·0 statistical software package. Fisher’s exact testwas used for comparison of the proportion of positive resultsobtained.

Results

Roche Amplicor v2·0

The sensitivity of the version 2·0 Amplicor kit used accord-ing to the manufacturer’s instructions (i.e. including themanual extraction) was assessed using a dilution series of theNIBSC HCV RNA working reagent (96/586). The 95% limit ofdetection calculated by probit analysis was 22·4 IU/ml [95%confidence interval (CI) = 14·2–105·9 IU/ml].

Sensitivity of Amplicor v2·0 using the standard BioRobot 9604 protocol for extraction of HCV RNA

The BioRobot 9604 with the QIAamp 96 Viral RNA kit wasused instead of the Amplicor manual RNA extraction. Thestandard QIAamp protocol was used with the modificationsto allow compatibility with the Amplicor assay, as describedabove in the Materials and methods. Use of the BioRobot9604 for the extraction of HCV RNA with amplificationand detection using the HCV Amplicor v2·0 assay initiallyresulted in a 95% limit of detection of 27·1 IU/ml (95%CI = 18·1–54·1 IU/ml) (Table 1).

Modification of the BioRobot 9604 protocol

Several modifications were made to the standard BioRobotprotocol, including increasing the AW2 wash volume andnumber of washes, increasing the elution volume, extendingthe drying spin time and introducing a short spin betweenwashes (Table 2). The greatest increase in sensitivity wasachieved by increasing the volume of the AW2 wash from900 to 1250 µl and increasing the volume of the elution bufferfrom 70 to 80 µl. This modified protocol was then used todetermine the HCV sensitivity by testing a dilution series of theNIBSC HCV RNA working standard (Table 1). Although the



95% limit of detection using this modified protocol was onlyslightly improved to 26·7 IU/ml (95% CI = 19·7–47·2 IU/ml),the 50% limit of detection was more than twice as sensitive,at 2·25 IU/ml (95% CI = 0·0–6·29 IU/ml) vs. 6·04 IU/ml (95%CI = 4·22–8·16 IU/ml) for the standard protocol.

Specificity of BioRobot 9604 and Roche Amplicor v2·0 hybrid assay

In order to assess the ability of the assay to detect the variousgenotypes of HCV, four samples of genotypes 1, 2, 3 and 4,previously genotyped by InnoLiPA (Innogenetics, Ghent,Belgium) [13] and quantified by the Quantiplex v2·0 (bDNA)assay (Chiron, Emeryville, CA) [14], were tested at end-pointdilution. A 1 : 100 dilution of the NIBSC HCV RNA workingreagent (genotype 3a) was run at the same time. The HCV RNAconcentration of the four genotyped samples was calculatedusing the Poisson distribution method [15]. In this methodthe HCV RNA concentration in the sample is calculated usingan assumed reverse transcription (RT) efficiency, which isusually considered to be 10%. However, in this case the NIBSCstandard was assayed by the same method and so a moreaccurate figure for the RT efficiency was calculated andapplied to the quantification of the other samples. None ofthe values of viral concentration differed from the viral loadestimated by bDNA by more than 0·5 log10 (Table 3), indicat-ing reliable detection of a range of genotypes.

To address any concerns that cross-contamination of HCV-negative samples by HCV-positive samples may occur duringthe robotic viral RNA extraction, a grid of HCV-positive sampleswith high viral concentration interspersed with negativesamples was extracted by the BioRobot 9604 and tested bythe Amplicor v2·0 assay. Sixteen plasma samples containingHCV at a concentration of 4 × 105 IU/ml were arranged in agrid surrounded by 48 negative samples. After testing withthe Amplicor v2·0 assay, all 48 of the negative samples werestill negative for HCV RNA.

Assay reliability

To determine the reliability of detection, three assays of 96replicates of a plasma sample containing HCV at 5 × 105 IU/ml were extracted using the BioRobot 9604 and tested usingthe HCV Amplicor v2·0 assay. All 288 were found to containHCV RNA.

Table 1 Hepatitis C virus (HCV) detection sensitivity using the BioRobot

9604 QIAamp 96 viral RNA extraction with Roche HCV Amplicor v2·0

amplification and detection

HCV levela (IU/ml)

Standard protocolb Modified protocolc

Number positive/tested (%)

Number positive/tested (%)

35·5 24/24 (100%) 24/24 (100%)

17·8 21/24 (87·5%) 19/24 (79·2%)

7·1 12/24 (50·0%) 32/48 (66·6%)

1·8 3/24 (12·5%) 11/24 (45·8%)

0 0/8 (0·0%) 0/8 (0·0%)

aBased on dilutions of the National Institute of Biological Standards and

Control (NIBSC) hepatitis C virus (HCV) RNA working standard (96/576) in

normal human plasma.bThe standard protocol is outlined in the Materials and methods section and

is based on the manufacturer’s protocol with a number of changes required

to allow compatibility with the Amplicor system.cThe modified protocol comprises the optimized protocol. This was

determined by a number of studies, as identified in the Results section.

The results from which the modified protocol was selected are shown

in Table 2.

Table 2 Effect of modifications of the BioRobot 9604 QIAamp 96 viral RNA

protocol on detection sensitivity with Amplicor HCV v2·0

MethodNumber positive/testeda

Percentage positivea

Single 900-µl AW2 wash 18/40 45%

70-µl elution

Double 1250-µl AW2 wash 4/16 25%

70-µl elution

Single 1250-µl AW2 wash 32/48 66·6%

80-µl elution

Double 1250-µl AW2 wash 24/48 50%

80-µl elution

Single 1250-µl AW2 wash 19/48 39·6%

20-min drying, 80-µl elution

Double 1250-µl AW2 wash 21/48 43·8%

1-min spin, 80-µl elution

aNumber/percentage positive at a 1 : 100 dilution of the National Institute

for Biological Standards and Control (NIBSC) hepatitis C virus (HCV) RNA

working standard (96/576).

Table 3 Detection of hepatitis C virus (HCV) genotypes 1–4 using the

BioRobot 9604 QIAamp 96 viral RNA protocol with Amplicor HCV v2·0 assay

(Qiagen-Roche)

Genotype

Quantiplex viral load (copies/ml)

Qiagen-Roche estimated viral load (copies/ml)a

Type 1 6·5 × 106 2·7 × 106

Type 2 5·9 × 106 6·2 × 106

Type 3 7·4 × 106 12·8 × 106

Type 4 7·5 × 106 12·8 × 106

aEstimated using the Poisson end-point distribution model.



Comparison of microwell plate (MWP) and COBAS Amplicor with BioRobot extraction

The sensitivity of HCV detection using the COBAS Amplicorautomated amplification and detection system was comparedto the MWP version with BioRobot HCV RNA extraction.Dilutions of the PEI HCV standard were extracted using theQIAamp 96 Viral RNA BioRobot protocol incorporating themodifications described previously with the additional omis-sion of the Air Pore tape during the 10-min drying spin. Thischange was to allow better evaporation of residual ethanol,which could contaminate the eluate and lead to PCR inhibi-tion (C. Sims, personal communication). RNA from each runwas then amplified and detected using either the MWP orCOBAS version of the Amplicor v2·0 assay (Table 4). The 95%limit of detection was 26·4 IU/ml (95% CI = 19·7–48·9 IU/ml)for the MWP assay and 23·3 IU/ml (95% CI = 17·8–41·3 IU/ml)for the COBAS assay. No significant difference (using Fisher’sexact test) was found between the MWP and COBAS assaysat any dilution.

Sensitivity of the Roche AmpliScreen assay with the QIAamp 96 virus assay

After the NBS HCV NAT screening had begun, the manu-facturers released new assay versions. Qiagen introducedthe QIAamp96 Virus assay, which incorporated some of themodifications described above. Roche released the HCVAmpliScreen v2·0 assay, which is the blood screening versionof Amplicor. The sensitivity of the combination of these twonew assays was assessed by testing the PEI standard in 24replicates per dilution, as described above (Table 5). Thesensitivity of this combination (as calculated using probitanalysis) gave a 95% limit of detection of 12·8 IU/ml (95%CI = 8·8–37·0 IU/ml).

Introduction of NAT assays into the NBS

In the first instance, the combination of the modified QIAampviral RNA assay with the COBAS Amplicor v2·0 assay (asdescribed above) was introduced into the NBS in April 1999for release of long shelf-life (> 35 days) blood components.The assay was performed with pools of 96 donations, asdescribed above in the Materials and methods. Subsequently,the combination of the QIAamp 96 virus assay and theAmpliScreen v2·0 assay was introduced into two of the threetesting sites from June 2001 for screening of all blood com-ponents with a shelf-life of longer than 24 h. The pool sizewas reduced from 96 to 48 donations as a result of this intro-duction. The third testing site retained manual processing.

Sensitivity monitoring

Initially the assay was run with 100 IU/ml or 71 IU/ml, 50 IU/mland 3·5 IU/ml run controls based on dilutions of a standardsupplied by the NIBSC. The 100 IU/ml and 3·5 IU/ml controlswere added to every A-ring. The 100 IU/ml control wasregarded as a ‘go or no-go’ control, i.e. if this control failedthen the samples in that A-ring were repeated; the 3·5 IU/mlcontrol was a monitoring control which had no bearing onthe validity of the assay but its frequency of detection wasmonitored over time.

The performance of the internal control was also monitoredand any sample with an internal control value of < 1·000optical density (OD) was deemed to require repeat testing.This cut-off had been re-set from the manufacturer’s definedcut-off of 0·150 as it was found that the performance of theinternal control was linked to the sensitivity of the assay. Ina study of 698 analyses of the 3·5 IU/ml control sample, 455were negative and 243 were positive. When the internal control(IC) values obtained from these analyses were compared itwas noted that the IC OD was significantly lower in thenegative assays than in the positive assays (P ≤ 0·0001,normal distribution Z-test), indicating that low IC OD can

Table 4 Comparison of the sensitivity of the modified BioRobot 9604

QIAamp 96 RNA protocol with COBAS and microwell plate (MWP) Amplicor

HCV v2·0

HCV levela (IU/ml)

MWP Number positive/tested (%)

COBAS Number positive/tested (%)

150 24/24 (100%) 24/24 (100%)

50 24/24 (100%) 24/24 (100%)

37·5 23/24 (95·8%) 24/24 (100%)

20 23/24 (95·8%) 22/24 (91·7%)

10 19/24 (79·2%) 15/24 (62·5%)

4 4/12 (33·3%) 6/12 (50·0%)

0 0/12 (0·0%) 0/12 (0·0%)

aValues are based on dilution of the Paul Ehrlich Institute (PEI) working

standard (no. 75/98) in normal human plasma.

Table 5 Hepatitis C virus (HCV) detection sensitivity using the BioRobot

9604 QIAamp 96 virus assay and COBAS AmpliScreen HCV v2·0

HCV levela (IU/ml)Number positive/tested (%)

150 24/24 (100%)

50 24/24 (100%)

20 24/24 (100%)

10 21/24 (87·5%)

4 15/24 (62·5%)

0 0/24 (0·0%)

aValues are based on dilution of the Paul Ehrlich Institute (PEI) working

standard (no. 75/98) in normal human plasma.



signify a loss of HCV sensitivity of the assay. Only 17 of the698 results analysed had IC levels between 0·150 and 1·000(2·4%); of these, two were HCV positive (11·8%) compared to239 (36·7%) HCV positives of the 652 with IC OD values of> 1·000 (P = 0·026 by Fishers’ exact test). Of the 29 controlsthat had an IC OD of < 0·150, two were HCV positive (6·9%).

Assay performance in routine use

The detection rate of the 100 IU/ml control averaged 96·2%for the 6 months before introduction of the QIAamp 96 Virusassay and the COBAS AmpliScreen assay combination. Inthe 6 months after this introduction the average detectionrose to 97·2%, and detection of the 3·5 IU/ml control hadrisen from 34·1 to 41·9%. Similarly, the internal controldetection rate had risen from 97·8 to 99·2%. These increasesin performance represent a reduction in the number of repeatassays from 9·6 to 5·2%.

Results of NAT testing in the NBS

To date (end of 2001) 6·8 × 106 blood donations have beenscreened in England and Wales by HCV NAT using themethods described above. Four confirmed HCV RNA-positive,anti-HCV-negative donations have been detected. Componentswere transfused from the first case, as at this time NAT wasonly used for the release of frozen components with a shelf-life of > 35 days. However, the recipient was already HCVinfected from previous multiple transfusions. No componentsfrom the other three cases were transfused. The first threedonors have since seroconverted; the most recent donor inOctober 2001 remained seronegative when retested 9 daysafter the initial testing.

Discussion

The automated assay for HCV RNA described here utilizescomponents from two different manufacturers and brings themtogether to form a system with good sensitivity and a highthroughput, which provides a facility for electronic sampletracking via barcodes from start to finish.

NAT testing has three separate phases: extraction, ampli-fication and detection. At present, the amplification anddetection steps of the Roche assay can be automated usingthe COBAS Amplicor system [10]; however, the extractionremains manual for this assay. The BioRobot 9604 has ageneric protocol which can be used to extract any nucleicacid, DNA or RNA from plasma, which may be used in placeof the manual procedure for extraction of HCV RNA andRoche internal control. The BioRobot 9604 can also be usedto set up the amplification and add the purified RNA and themaster mix to the COBAS amplification rings. The BioRobotautomated extraction, in combination with the COBAS

Amplicor system for amplification and detection, allows theentire testing process to be automated. This was consideredto be of vital importance to the NBS given the large numberof blood donations that require testing. Importantly, thisautomation also allows bar code information to be read foreach sample and passed on to the COBAS Amplicor, givingelectronic sample tracking throughout the whole process.Such positive sample identification is an important aspect ofavoiding transposition errors in donor screening.

The BioRobot extraction takes ≈ 2 h and the amplificationand detection in the COBAS takes 4 h. With time for setting upthe instruments and transferring RNA to the COBAS A-rings,the whole assay takes ≈ 7 h; however, only ≈ 1 h of this is‘hands-on’ time for the operator (spent preparing reagentsand setting up the automated instruments).

The CPMP guideline requires a run control of 100 HCV IU/ml to be included and detected in HCV screening assays forplasma intended for fractionation [7]. At present there isno Europe-wide legislation regarding the HCV RNA detec-tion limit in blood for transfusion. However, the PEI (theregulatory authority in Germany) has set a 95% detectionlimit of 5000 HCV IU/ml per donation [16]. This means thatany dilution of donations resulting from the strategy oftesting minipools has to be compensated for by a correspond-ing increase in the sensitivity at the limit of detection. Forexample, testing a minipool of 50 donations would requirean assay detection sensitivity of 100 IU/ml; a pool of 100donations would require an assay sensitivity of 50 IU/ml, andso on. The assays described here have 95% limits of detectionof 23·3 IU/ml and 12·8 IU/ml, both of which are significantlybelow the PEI limit of 52 IU/ml using pools of 96 donations.The Qiagen and Roche combination assay was shown todetect HCV samples of genotypes 1–4 with equal efficiency.The standards used in this study were of Type 1 (PEI) andType 3 (NIBSC), and samples with HCV types 2 and 4, whichhad been quantified using the Chiron Quantiplex v2·0 bDNAassay [14].

The BioRobot extraction did not generate false positives asa result of cross-contamination, an important finding as theNBS had decided to leave anti-HCV-positive donations inthe minipool. This decision was made for two reasons. Firstthe time requirement for release of labile components was suchthat waiting for the serology results on samples beforegenerating minipools for NAT testing was not feasible. Second,retaining seropositive samples allowed the system for resolu-tion of an HCV RNA positive minipool down to an individualpositive donation to be routinely practised. No evidence wasfound of false negatives as a result of instrument error usingmultireplicate testing of a known HCV RNA-positive sample,demonstrating that the assay was robust.

At the time this project started, the only 96-well automatednucleic acid extractor was the Qiagen BioRobot 9604. Earlydevelopment of the testing protocols involved the use of the



Amplicor microwell plate assay. However, coupling this extrac-tion to the Roche Amplicor COBAS analyser effectively gavean assay which has a high-throughput capacity, positive sampleidentification and considerable hands-off time.

The first combination (BioRobot QIAamp viral RNA assayand COBAS Amplicor v2·0) was introduced into the NBS in1999 and initially used to screen pools of 96 donations forrelease of components with a long shelf-life (> 35 days).Upgrading of both the Qiagen extraction (QIAamp virusprotocol) and Roche amplification and detection (COBASAmpliScreen v2·0) components of the assay have resulted ina more sensitive and reliable assay. The NAT test results arenow used for release of all blood components with a shelf lifeof > 24 h, which includes platelets. The pool size has alsobeen reduced from 96 to 48 donations so that fewer donationsare quarantined, whilst a positive minipool is being resolveddown to single HCV RNA-containing donations.

A number of other published methods for HCV RNAscreening of donated blood have also used a modified COBASAmplicor v2·0 method. In the Netherlands [17] and Germany[18], sensitivities of 8 IU/ml and 10 IU/ml, respectively,have been achieved using the NucliSens Extractor (OrganonTeknika, Boxtel, the Netherlands). The NucliSens extractor isalso used by the Scottish National Blood Transfusion Service(for donations from Scotland, Northern Ireland and theRepublic of Ireland) with an in-house RT–PCR assay achiev-ing a sensitivity of 7·3 IU/ml [19]. These methods all haveslightly better sensitivity than that described here (13·8 IU/ml) because the NucliSens extractor analyses 2 ml of plasmawhereas the Qiagen BioRobot 9604 extracts nucleic acidfrom only 200 µl of plasma. However, the advantages of theBioRobot 9604 over the NucliSens Extractor are that ofthroughput and electronic sample tracking. The NucliSensextractor can only process 12 samples at a time compared to96 using the BioRobot, and the NucliSens RNA eluate mustbe added to the COBAS A-rings by hand, introducing thepotential for human error. Another report describes a modi-fied COBAS Amplicor v1·0 assay with manual Qiagen spincolumn extraction [20], with a reported sensitivity of 2000genomes/ml (≈ 400 IU/ml). An alternative NAT technology,Transcription Mediated Amplification (TMA) (Gen-Probe Inc.;San Diego, CA), has been described. This uses oligo capture fornucleic acid purification and tests for both HCV and HIV-1in multiplex with a capacity of up to 100 tests per run andhas an HCV sensitivity of < 100 copies/ml (≈ 20 IU/ml) [21].

The rates of HCV window-phase donations in other Europeancountries are similar to our current rate of 1 in 1·7 million(four cases per 6·8 × 106). In the Red Cross blood centres inGermany, the window-phase donation rate is 1 in 1·2 × 106

(11 cases per 12·7 × 106) [22]. In the Netherlands no window-phase donations were found after screening 1·5 × 106 donations[17]. However, other countries have reported higher rates. InJapan, using an automated extraction system developed by

Roche coupled with the ABI 7700 (Applied Biosystems, FosterCity, CA), the rate has been reported as 1 in 480 000 (15 casesper 7·2 × 106) [23]. In North America a rate of 1 in 263 000(62 cases per 16·3 × 106) has been found [24].

The reported NAT screening methods used worldwide haveessentially similar sensitivities and the differences in rates ofwindow-phase donations, effectively a measure of incidenceof acute HCV infection, are probably a reflection of the preval-ence of carriers.

The NBS has successfully incorporated molecular techno-logy into the routine performance of blood transfusionpractice. The first target to be introduced has been HCV RNA,although whether the enhanced safety of this testing is costeffective for the low prevalence and low incidence seen in theUK is a matter for discussion. Extension of NAT screening toother agents could easily be undertaken given the genericextraction potential of the BioRobot protocol. The NBS systemhas the required flexibility to extend testing, but at presentthis would not appear to be warranted in the UK although thematter is under review.

Acknowledgements

We gratefully acknowledge the financial support of theNational Blood Authority. QIAamp BioRobot kits for evalu-ation and validation were generously donated by QiagenGmbH. Roche Amplicor HCV kits for the initial investigationswere generously donated by Roche Molecular Systems. HCVstandards were kindly provided by the UK National Institutefor Biological Standards and Controls.

The authors would like to thank Keith Alderman, TonyMurphy, Peter Garwood and Liz Caffrey for help with collec-tion of the EDTA plasma used as a diluent.

References

1 Schreiber GB, Busch MP, Kleinman SH, Korelitz JJ: The riskof transfusion-transmitted viral infections. The RetrovirusEpidemiology Donor Study. N Engl J Med 1996; 334:1685–1690

2 Barrera JM, Francis B, Ercilla G, Nelles M, Achord D, Darner J,Lee SR: Improved detection of anti-HCV in post-transfusionhepatitis by a third-generation ELISA. Vox Sang 1995; 68:15–18

3 Soldan K, Barbara J: The risks of infection transmission by bloodtransfusion in England. J Clin Pathol 1999; 52:405–408

4 Williamson LM, Lowe S, Love EM, Cohen H, Soldan K,McClelland DB, Skacel P, Barbara JA: Serious hazards of trans-fusion (SHOT) initiative: analysis of the first two annual reports.BMJ 1999; 319:16–19

5 Busch MP, Rawal BD, Nowicki M, Operskalski EA, Mosley JW.HCV RNA and ALT patterns in seroconverting transfusionrecipients, and correlation with donor ALT and RNA. Transfusion1997; 37:S443–S443



6 Zuckerman M, Wardle J, Mutton K, Tedder R: Hepatitis Bimmunisation in chronically transfused recipients. Lancet 1995;346:1565–1566

7 CPMP: The Introduction of Nucleic Acid Amplification Techno-logy (NAT) for the Detection of Hepatitis C Virus RNA in PlasmaPools. 1998; CPMP/BWP/390/97. London, EMEA, 1998

8 DOH: Further Precautionary Measures on Blood ProductsAnnounced. 1998, 98/076. London, Department of Health, 1998

9 Mortimer J: Intersecting pools and their potential application intesting donated blood for viral genomes. Vox Sang 1997; 73:93–96

10 DiDomenico N, Link H, Knobel R, Caratsch T, Weschler W,Loewy ZG, Rosenstraus M: COBAS AMPLICOR. Fully automatedRNA and DNA amplification and detection system for routinediagnostic PCR. Clin Chem 1996; 42:1915–1923

11 Grant PR, Kitchen A, Barbara JA, Hewitt P, Sims CM, Garson JA,Tedder RS: Effects of handling and storage of blood on thestability of hepatitis C virus RNA. Implications for NAT testingin transfusion practice. Vox Sang 2000; 78:137–142

12 Saldanha J, Heath A, Lelie N, Pisani G, Nubling M, Yu M:Calibration of HCV working reagents for NAT assays againstthe HCV international standard. The Collaborative Study Group.Vox Sang 2000; 78:217–224

13 Stuyver L, Wyseur A, van Arnhem W, Hernandez F, Maertens G:Second-generation line probe assay for hepatitis C virus geno-typing. J Clin Microbiol 1996; 34:2259–2266

14 Detmer J, Lagier R, Flynn J, Zayati C, Kolberg J, Collins M,Urdea M, Sanchez-Pescador R: Accurate quantification ofhepatitis C virus (HCV) RNA from all HCV genotypes by usingbranched-DNA technology. J Clin Microbiol 1996; 34:901–907

15 Simmonds P, Balfe P, Peutherer JF, Ludlam CA, Bishop JO,Brown AJ: Human immunodeficiency virus-infected individualscontain provirus in small numbers of peripheral mononuclearcells and at low copy numbers. J Virol 1990; 64:864–872

16 Paul-Ehrlich-Institut: Results of the phased plan to reduce

the infection risk in hepatitis B, C, and HIV in recipients ofpacked red cell transfusions. Bundesanzeiger 1998; 53:3835–3836

17 Cuijpers HT, Molijn MH, Bos HJ, Peeters AP, Der Poel CL, Lelie PN:Validation of the NucliSens Extractor in combination with thehepatitis C virus Cobas Amplicor 2·0 assay in four laboratoriesin the Netherlands utilizing nucleic acid amplification technologyfor blood screening. Vox Sang 2001; 81:12–20

18 Cardoso MS, Koerner K, Hinz W, Lenz C, Prokein S, Rau B,Schwandt A, Kubanek B: Experiences in HCV-NAT screeningprior to releasing cellular components by the German Red CrossBlood Transfusion Service of Baden-Wurttemberg. Biologicals1999; 27:281–284

19 Jarvis L, Cleland A, Simmonds P, Dow B, Munro M, Jordan A,Prowse C, Yap PL: Screening blood donations for hepatitis Cvirus by polymerase chain reaction (letter). Vox Sang 2000;78:57–58

20 Roth WK, Weber M, Seifried E: Feasibility and efficacy of routinePCR screening of blood donations for hepatitis C virus, hepatitisB virus, and HIV-1 in a blood-bank setting. Lancet 1999;353:359–363

21 McDonough SH, Giachetti C, Yang Y, Kolk DP, Billyard E, Mimms L:High throughput assay for the simultaneous or separate detec-tion of human immunodeficiency virus (HIV) and hepatitistype C virus (HCV). Infusionsther Transfusionsmed 1998;25:164–169

22 Seifried E, Roth WK: Prospective survey of HCV, HBV, HIV-1NAT screening of more than 10 million blood donors in the RedCross blood service centres in Germany. Transfus Clin Biol 2001;8:17s (Abstract S09–001)

23 Japanese Red Cross: NAT screening research group: nationwideNAT screening of HBV, HCV and HIV-1 for blood transfusion inJapan. Transfus Clin Biol 2001; 8:56s (Abstract PDS3–003)

24 Stramer SL, Caglioti S, Strong DM: NAT of the United States andCanadian blood supply. Transfusion 2000; 40:1165–1168

Documents

Automated screening of blood donations for hepatitis C virus RNA using the Qiagen BioRobot 9604 and the Roche COBAS HCV Amplicor assay