Automated Extraction of Viral-Pathogen RNA and DNA for High-Throughput Quantitative Real-Time PCR
http://www.100md.com
微生物临床杂志 2005年第11期
Laboratory of Molecular Diagnostics, University Hospital Gasthuisberg, Katholieke Universiteit Leuven, Leuven, Belgium
ABSTRACT
The performance of the m1000 system (Abbott Laboratories, Illinois) as a front-end extraction system for high-throughput "in-house" quantitative real-time PCR assays was analyzed and compared to that of manual extraction of plasma and serum samples (hepatitis C virus [HCV] and hepatitis B virus [HBV]) and EDTA-blood samples (cytomegalovirus [CMV] and Epstein-Barr virus [EBV]). Linearity of extraction was tested on dilution series of HCV and HBV reference materials. The correlation coefficient for standard curves based on repeated extraction runs was 0.97 ± 0.06 for HCV and 0.97 ± 0.03 for HBV, indicating a linear extraction from 100 to 1.0 x 105 HCV IU/ml and from 100 to 1.0 x 106 HBV IU/ml. Intra- and interrun variability was below 0.23 log10 IU/ml for 2.98 to 5.28 log10 HCV IU/ml and 2.70 to 5.20 log10 HBV IU/ml. Correlation between automated and manual extraction was very good. For HCV, the correlation coefficient was 0.91 and the mean difference in viral load was 0.13 log10 HCV IU/ml. For HBV, the correlation coefficient was 0.98 and the mean difference in viral load 0.61 log10 HBV IU/ml. For CMV and EBV, the correlation coefficient was 0.98 and the mean difference in viral load 0.33 log10 copies/ml. Accuracy was confirmed with a reference panel (QCMD, Glasgow, Scotland) for all four assays. No cross-contamination was observed when extracting strongly positive polyomavirus samples (8.10 log10 copies/ml) interspersed with polyomavirus-negative samples. Automated extraction via the m1000 system offers a high reliability of extraction and resulted in a strong reduction of the required extraction hands-on time for high-throughput PCR compared to manual extraction protocols.
INTRODUCTION
Real-time thermocyclers have greatly decreased the amount of hands-on time in nucleic acid (NA)-based diagnostics for pathogens. The use of commercially available universal master mixes that contain all reagents except the pathogen-specific primer-probe combination has further decreased the number of pipetting steps and thus reduced labor and the possibility of errors.
NA extraction has now become the most critical and labor-intensive step in NA-based diagnostic assays. The overall sensitivity of the assay is determined by the NA yield, its purity, and the amount of sample equivalents that can be transferred into the amplification reaction. Conventional manual sample preparation methods are labor intensive and susceptible to contamination, handling variations, or errors (3, 9, 12).
Since both the pathogen range and the number of different sample types are expanding and since multiplex downstream testing is becoming standard practice, there is a need for a generic extraction method (2, 5, 10). Ideally, the NA extraction procedure should yield pure NA from different pathogens and from a broad range of sample types.
Recently, Abbott introduced m1000, an automated generic RNA and DNA extraction system using magnetic microparticle processing (17, 18). The m1000 system provides the necessary features for full automation of NA extraction of up to 48 samples in 2 h or less, depending on the protocol. It uses ready-to-use reagents and hands-off operation from the moment of loading of the patient specimen in primary tubes onto the m1000 instrument up to completed NA extract.
Both the RNA and DNA extraction chemistries of the m1000 system are based on lysis and nuclease inactivation with guanidinium isothiocyanate. The RNA extraction chemistry further uses uncoated iron particles as the capture medium. The DNA extraction chemistry uses silica-coated magnetic particles as the solid phase.
The processes of sample and reagent pipetting, heating and incubation, magnetic capture and washing, and the final elution are optimized in ready-to-use m1000 protocols. These protocols are provided by Abbott and can be altered in accordance with the specifications and needs of the user.
In this study, the performance of m1000 as a front-end extraction system for high-throughput "in-house" real-time virological PCR assays (m1000 "open mode") was evaluated. We verified m1000 protocols for use with the following quantitative tests: hepatitis C virus (HCV) RNA, hepatitis B virus (HBV) DNA, cytomegalovirus (CMV) DNA, and Epstein-Barr virus (EBV) DNA.
The linearity for automated extraction preceding the HCV RNA and HBV DNA PCR was determined using dilution series of reference materials, while intra- and interrun variations were determined using three dilutions of an HCV- or HBV-positive sample. Accuracy was tested by proficiency panels for these two assays, and the HCV RNA and HBV DNA viral load results for automated and manual extraction were compared for clinical plasma and serum samples. For the CMV DNA and EBV DNA quantitative PCR, the correlation of viral load results for EDTA blood samples and two proficiency panels by the automated and manual methods were assessed.
The possible risk of sample cross-contamination on m1000 was evaluated through an "in-house" real-time polyomavirus (JC virus [JCV]/BK virus [BKV]) PCR assay with urine samples.
MATERIALS AND METHODS
Manual extraction protocols. Manual extraction of HCV RNA and HBV DNA starting from 1 ml serum or plasma was done according to the protocol of the QIAamp Ultrasens virus kit (QIAGEN, Hilden, Germany). Internal control (6 x 109 copies; IC1, see "Viral load determinations") were added together with the carrier RNA to the sample. Elution was performed in a 50-μl volume.
All other manual extractions were performed according to the protocol of the QIAamp DNA minikit. IC1 (6 x 109 copies) were added to the sample during the first step of the protocol together with proteinase K. The elution volume was 100 μl.
Automated extraction protocols. For fully automated sample preparation, all extraction protocols were provided by Abbott. RNA extractions were done using the Sample Preparation System kit (Abbott) after addition of 1.8 x 1011 copies of IC1 to a bottle of lysis buffer. The Sample Preparation System-kit DNA (Abbott) was used for DNA extractions after addition of 1.8 x 1012 copies of IC1 to a bottle of lysis buffer. The RNA-plasma-BA-500-110-v1 extraction protocol was used to extract HCV RNA from 0.5 ml serum or plasma, resulting in a final elution volume of 110 μl. HBV DNA was extracted from 0.5 ml serum or plasma with the DNA-plasma-500-110-v1 extraction protocol, resulting in an elution volume of 110 μl. Automated extractions for CMV and EBV DNA were performed with the DNA-Blood-BA-300-150-v1 protocol, starting from 0.3 ml of sample and eluting in a 150-μl volume. For the cross-contamination study with the BKV positive and negative urine samples, the DNA-urine-LL-400-70-v2 protocol was used to extract DNA from 0.4 ml in a end volume of 70 μl.
Viral load determinations. All "in-house" real-time virological PCR assays were performed on an ABI 7900 real-time thermocycler (Applied Biosystems, Foster City, California) in duplicate: one reaction well for a single PCR for the viral target of interest and the second reaction well for a multiplex reaction for the viral target and IC1. IC1 is a heterologous 66-bp oligonucleotide added to the sample during the extraction procedures. In the absence of inhibition of the reaction, IC1 gives a constant result. Reported results are based on the results of the single reactions. The multiplex reaction is used as a control for inhibition (result IC1) and as a confirmation of the result of the single reaction (viral target).
The real-time one-step reverse transcriptase (RT)-PCR for the amplification and detection of HCV RNA was carried out using the QIAGEN OneStep RT-PCR kit with addition of 0.15 μM of in-house carboxy-X-rhodamine dye PRTM (6-carboxyfluorescein [FAM]-5'-TATCTAGT-3' noncoding region) as a passive reference. A 262-bp gene fragment in the HCV 5' carboxy-X-rhodamine was amplified with 0.5 μM of each biotinylated primer: HCV1FB (5'-Biotin-CTAGCCATGGCGTTAGTATGAGTGT-3') and HCV2RB (Biotin-5'-GGTGCACGGTCTACGAGACCT-3'; Biotin). Primers were biotinylated so that the amplification product could be used directly in the InnoLiPA HCV II assay (Innogenetics, Belgium) (16). The single one-step RT-PCR further contained a final concentration of 1x OneStep RT-PCR buffer, 400 μM of each deoxynucleoside triphosphate, and 0.4 μM probe HCPr96FTM (FAM-5'-CACTCGCAAGCACCCTATCAGGCAGT-3'-TAMRA) (16). RT-PCR after manual RNA extraction was carried out in a reaction volume of 40 μl containing 10 μl of the manually extracted RNA and 1.6 μl OneStep RT-PCR enzyme mix. RT-PCR after automated extraction was carried out in a reaction volume of 50 μl containing 25 μl of the m1000-extracted RNA and 2.0 μl OneStep RT-PCR enzyme mix. The multiplex one-step RT-PCR was performed in the same way as the single reaction with the addition of 0.2 μM (each) primers IC1P1 (5'-TCATGGTCAAATGCTGGATGA-3') and IC1P2 (5'-CTCTGCAATCAGCTCACGAAAC-3') and 0.1 μM ICPTM (VIC-5'-TCCATGGAGACAGCCGCACCA-3'-TAMRA), i.e., the primers and probe to detect IC1. The reactions were carried out with an initial reverse transcription step at 48°C for 30 min, followed by PCR activation at 95°C for 10 min and 50 cycles of amplification (95°C for 15 s, 60°C for 1 min). Real-time data collection was performed at 60°C. Quantitative results are reported from 2.0 log10 HCV IU/ml to 6.7 log10 HCV IU/ml. All real-time DNA PCRs were carried out in a reaction volume of 40 μl containing 10 μl of the DNA extract, 20 μl 2x Universal Mastermix (Applied Biosystems, Foster City, California), and the primers/probe specific for the viral target: 0.15 μM HBV3F (5'-TCCCCGTCTGTKCCTTCTC-3'), 0.15 μM HBV4R (5'-GCGTTCAYGGTGGTYKCCAT-3'), and 0.10 μM HBV1T (FAM-5'-CCGTGTGCACTTCGCTTCACCTCTGC-3'-TAMRA) (13) for HBV; 0.25 μM CMCP11 (5'-CGTAACGTGGACCTGACGTTT-3'), 0.25 μM CMCP12 (5'-CACGGTCCCGGTTTAGCA-3'), and 0.20 μM CMCP3TM (FAM-5'-TATCTGCCCGAGGATCGCGGTTACA-3') for CMV; 0.25 μM EBBKT-1 (5'-TTTGGACCCGAAATCTGACACT-3'), 0.25 μM EBBKT2 (5'-GCCAACCATAGACCCGCTTC-3'), and 0.20 μM EBBKT-F (FAM-5'-CCATTTTGTCCCCACGCGCG-3'-TAMRA) for EBV; 0.5 μM PEP-3 (5'-GGAAAGTCTTTAGGGTCTTCTACCTTT-3'), 0.25 μM PEP-5 (5'-GATGAAGATTTATTYTGCCATGARGA-3'), 0.25 μM PEP-6 (5'-GAAGACCTGTTTTGCCATGAAGA-3'), and 0.20 μM HuPo-1 (FAM-5'-ATCACTGGCAAACAT-MGB-3') for polyomaviruses JCV and BKV. The multiplex DNA PCRs were performed in the same way as the single reactions with the addition of 0.075 μM (each) primers IC1P1 and IC1P2 and 0.15 μM ICPTM. The reactions were carried out with an initial uracyl-N-glycosylase incubation step at 50°C for 2 min, followed by PCR activation at 95°C for 10 min and 45 cycles of amplification (95°C for 15 s, 60°C for 1 min). Real-time data collection was performed at 60°C. Quantitative results for HBV are reported from 2.0 log10 HBV IU/ml to 6.7 log10 HBV IU/ml, for CMV and EBV from 2.0 log10 copies/ml to 6.7 log10 copies/ml, and for polyomavirus from 2.0 log10 copies/ml to 7.70 log10 copies/ml.
All "in-house" real-time virological PCR assays with the manual sample preparation protocol have been validated using commercial standards. For CMV and EBV, a quantitated CMV and EBV DNA control (Advanced Biotechnologies, Inc., Columbia, Maryland) was used, and for HCV and HBV, a quantitated HCV and HBV serum control (Acrometrix, Benica, California) was used. The performance of each assay with the manual sample preparation protocol has been verified with a virus-specific proficiency panel (QCMD, Glasgow, Scotland). Each of the four proficiency panels contained eight samples, and all results were quantified within the range of acceptation (geometric mean ± 1 standard deviation [SD] of all reported quantitative data sets) for this panel.
Study design. In the first part of the study, the linearity and accuracy of the in-house real-time HCV and HBV PCR with the automated sample preparation protocol was determined with dilutions of commercially available HCV and HBV controls (NAP-HCV007 and NAP-HBV006; Acrometrix) and the samples of the HCV and HBV 2004 QCMD proficiency program panels.
In the second part, the intrarun and interrun variations for the HCV and HBV assays with automated sample preparation were determined using three different dilutions of a positive sample for both assays. All dilutions were tested five times in each of four runs. At the same time, the performance of both assays with automated sample preparation was compared to the performance of the assays with the manual extraction method. A total of 40 clinical plasma/serum samples were investigated for HCV and 40 clinical plasma/serum samples for HBV.
In the third part, the correlation and agreement of CMV and EBV viral load results for 60 EDTA blood samples and the samples of the CMV and EBV 2004 QCMD proficiency program panels, extracted via the automated and the manual extraction method, were assessed.
In a last and fourth part, the possible risk of cross-contamination with m1000 was tested with the DNA-urine-LL-400-70-v2 protocol as a front-end for the "in-house" real-time polyomavirus (JCV/BKV) PCR assay. Two milliliters of a highly positive urine sample for BKV (1.8 x 109 copies/ml) was diluted to 25 ml with a viral transport medium. This diluted urine sample was loaded onto m1000 to obtain a checkerboard pattern with the pure viral transport medium.
RESULTS
Linearity. Four dilutions of the commercially available HCV control (NAP-HCV007) (2 x 106 HCV IU/ml), which contained 1.0 x 105, 1.0 x 104, 1.0 x 103, and 100 HCV IU/ml, were each extracted five to seven times in as many separate extraction runs together with the HCV-negative diluent. Five dilutions of the commercial HBV control (NAP-HBV006) (2 x 107 HBV IU/ml), which contained 1.0 x 106, 1.0 x 105, 1.0 x 104, 1.0 x 103, and 100 HBV IU/ml, were extracted two to eight times, in as many different extraction runs, together with the HBV-negative diluent. None of the negative specimens yielded a positive result in either reaction, while 100 HCV IU/ml and 100 HBV IU/ml were detected in all replicates.
The correlation coefficient for the standard curves based on the different extraction runs of these dilutions was 0.970 ± 0.060 for HCV and 0.971 ± 0.025 for HBV, indicating a linear response for 100 to 1.0 x 105 HCV IU/ml and for 100 to 1.0 x 106 HBV IU/ml. The mean PCR efficiencies calculated from these standard curves was 83% for HCV and 101% for HBV.
Accuracy. Samples of the HCV 2004 QCMD proficiency program panel were tested with automated extraction and the in-house real-time HCV-PCR. All results were found to be within 0.5 log10 HCV IU/ml of the expected result (Table 1). One weakly positive sample with an expected result of 2.23 log10 HCV IU/ml had a result (1.90 log10 HCV IU/ml) below the quantification range of our assay and was reported as "< 2.0 log10 HCV IU/ml." Automated extraction on m1000 followed by in-house real-time HBV-PCR performed on the samples of the HBV 2004 QCMD proficiency program panel resulted in an HBV viral load difference with the expected results, varying from +0.12 to +1.22 log10 HBV IU/ml for all panel members (Table 1). Our in-house HBV PCR results are expressed in HBV IU/ml because we used a commercially available reference material calibrated against the international standard to generate a standard curve. The expected HBV results were expressed in HBV copies/ml and were divided by 5 (– 0.70 log HBV IU/ml) (14) to obtain the reference results expressed in HBV IU/ml.
Interassay and intra-assay variations. Determination of interassay and intra-assay variations with automated sample preparation was performed with three dilutions of a sample positive for either HCV or HBV. These dilutions had viral loads, determined with manual extraction, of 5.28, 3.38, and 2.98 log10 HCV IU/ml for HCV and 5.20, 3.20 and 2.70 log10 HBV IU/ml for HBV. All dilutions were tested five times in each of four runs. The mean results and the variations for each of the dilutions are given in Table 2. For HBV, the two weakest samples both had one negative result (<2.0 log10 HBV IU/ml). Because the IC1 signal for these two samples was stronger than the IC1 result for all other samples in this study, these two results were removed as outliers. PCR inhibition normally results in a weakened or negative IC1 signal. The cause of this stronger IC1 signal remains unclear to us.
Correlation with manual extraction. Extraction with the m1000 system followed by real-time HCV PCR for 40 clinical samples and comparison with the results obtained with the routine manual extraction resulted in a positive result for all 34 samples which were also positive with the manual extraction. The total correlation for these 34 samples between the two methods was good (r = 0.91), as shown in Fig. 1A. The mean difference in viral load was 0.13 log10 HCV IU/ml. The 95% confidence interval (CI) (± 2 standard deviations) for the mean difference was –0.75 and +1.01 log10 HCV IU/ml. The three samples with a log10 difference of more than 0.71 between the two extractions were repeated by both extraction methods on the same sample aliquot and analyzed again. When these three reanalysis results were taken into account, a final mean difference in viral load of 0.04 log10 HCV IU/ml and a 95% CI for the mean difference of –0.70 to +0.79 log10 HCV IU/ml were obtained.
Extraction with the m1000 system followed by real-time HBV PCR for 40 clinical samples and comparison with the results obtained during the routine clinical diagnosis resulted in a positive result for 25 samples which were also positive with the manual extraction. The total correlation for the 25 samples positive by both methods was very good (r = 0.98), as shown in Fig. 1B. The mean difference in viral load is 0.61 log10 HBV IU/ml. The 95% CI for the mean difference was –0.44 and+1.66 log10 HBV IU/ml. The three samples with a log10 difference of more than 1.0 between the two extractions were repeated by both extraction methods on the same sample aliquot and analyzed again. When these three reanalysis results were taken into account, a final mean difference in viral load of 0.128 log10 HBV IU/ml and a 95% CI for the mean difference of –0.75 to +1.01 log10 HBV IU/ml were obtained. Nine samples gave a different result after automated extraction compared to the routine manual extraction: six samples that had been negative with manual extraction gave a positive result with m1000, and three weakly positive samples with manual extraction (2.20, 2.09 and 2.24 log10 HBV IU/ml) were negative with m1000. Only four out of these nine samples had sufficient fluid left to repeat both extraction methods on the same sample aliquot. These four samples that were positive with automated extraction and negative with manual extraction were now positive with both extraction methods, with a mean viral load ranging from 2.40 to 3.38 log10 HBV IU/ml and a mean log difference of 0.25 log10 HBV IU/ml between the two methods.
Sixty EDTA-blood samples were analyzed using automated extraction on m1000 prior to real-time CMV or EBV PCR. Results were compared with the results obtained during the routine clinical diagnostic test using the same real-time CMV or EBV PCR preceded by manual extraction. Twenty blood samples were negative (<2.70 log10 copies/ml) with both extraction methods, and 33 samples yielded a positive result with both extraction methods. The correlation for these 33 samples was good (r = 0.95) (Fig. 1C). The mean log10 copies/ml difference for the samples positive by both methods was 0.33. The 95% CI for the mean difference was –0.23 and +0.89 log10 copies/ml. Six samples with a viral load between 2.70 and 2.93 log10 copies/ml were negative after automated extraction. For one of these samples, the IC1 was invalid (inhibition of PCR) after automated extraction. One negative sample was positive (3.41 log10) after automated extraction. Both extraction methods were repeated on the same sample aliquot for these seven samples. For the sample with inhibition of the PCR after extraction with the m1000 system, inhibition of the PCR was confirmed after automated extraction, while the positive result with manual extraction was also confirmed. The discordant results on the sample positive after automated extraction and negative after manual extraction were confirmed with both extraction methods. In the five remaining samples, a low viral load was detected with manual and automated extraction, with most manual extractions just above (between 2.71 and 3.36 log10 copies/ml) and most automated extractions just below (between 2.17 and 2.80 log10 copies/ml) the limit of detection. With the same automated m1000 (blood) protocol, the samples of the CMV 2004 QCMD and EBV 2004 QCMD proficiency program panel were tested with manual and automated extraction. The 12 CMV panel samples consisted of 12 lyophilized samples in a plasma matrix. The eight EBV panel members contained different concentrations of EBV diluted in Dulbecco's modified Eagle's medium with fetal calf serum. An overview of the panel result range, the routine diagnostic result (manual extraction), and the result after automated extraction is given in Table 3. CMV and EBV viral load results after manual and automated extraction with the samples of the proficiency program panels were within the reported result range for six out of seven CMV-positive samples and six out of seven EBV-positive samples. The single CMV-positive sample out of range (CMV04-12) after automated extraction was detected but outside the quantification range of our assay (<2.70 log10 copies/ml) both with manual and automated extraction. One weakly EBV-positive sample (EBV04-01) was outside the quantification range of our assay (<2.7 log10 copies/ml) after automated extraction but had an EBV load of 3.0 log10 copies/ml after manual extraction.
Cross-contamination. Twenty-four strongly positive and 24 negative samples were loaded on the m1000 instrument in such a way that a checkerboard pattern was obtained in the m1000 "1 ml subsystem carrier." The "1 ml subsystem carrier" contains the reaction vessels, and therefore it can be regarded as the main processing unit of m1000, where all processes of sample and reagent mixing, heating, incubation, magnetic capture, washing, and elution take place. After real-time polyomavirus PCR, none of the 24 negative samples gave a positive signal, while all 24 positive samples gave a nice amplification plot resulting in a mean BK viral load of 8.10 log copies/ml with a standard deviation of 0.07 log copies/ml.
Time estimates. The estimated time required to perform the manual and m1000 extractions is presented in Table 4. The estimated hands-on setup time for the m1000 system was ca. 30 min. Batches for manual extractions were limited to 24 samples due to the capacity of our microcentrifuge. Therefore, for batches of 48 samples, two consecutive batches of 24 samples had to be performed. Compared with the manual sample preparation for the HCV and HBV assay on 48 plasma or serum samples, a 114-min decrease in total time and 210-min decrease of hands-on time was achieved. The automated extraction of 48 EDTA-blood samples had no influence on total time, but there was a 140-min decrease of hands-on time compared to the manual DNA extraction protocol. When only 24 samples were extracted, the total assay time for automated extraction was identical to or even higher than the total assay time for manual extraction, but there still was a significant decrease in hands-on time, with a 90-min decrease in hands-on time for the plasma/serum DNA or RNA extraction and a 55-min decrease for the blood DNA extraction. In addition to the time required for sample preparation, the reaction setup and the amplification on ABI 7900 for 48 samples (96 reactions) required ca. 3.5 h total time with 1 h hands-on time.
DISCUSSION
Conventional manual extraction protocols for the extraction of pathogen DNA or RNA from clinical samples are the most labor-intensive and critical part in current NA diagnostics assays for pathogens. Recently, Abbott developed m1000, an automated extraction system for nucleic extraction in routine diagnostic laboratories. The m1000 system not only provides NA protocols as a front end for the Abbott molecular assays ("closed-mode") but also allows the protocols to be adapted to the specific needs of the user ("open-mode"). This makes the m1000 system a truly open platform that allows the user to define protocols for a variety of different samples and applications. In this study, m1000 was evaluated as a NA extraction system for samples of five in-house real-time PCR assays. The in-house real-time HCV and HBV PCR assays with the automated sample preparation protocol showed sufficient linearity and sensitivity on dilution series of, respectively, a HCV and HBV standard. The interassay and intra-assay percent coefficient of variation for the logarithmic viral load value ranged from 2.78% to 6.27% for HCV and from 1.85% to 8.11% for HBV. These results are in concordance with those reported for manual or automated extraction systems and a real-time HCV-PCR (4, 6) or HBV-PCR (1, 15, 19). Samples of HCV and HBV proficiency program panels were tested by the respective in-house PCR assays after automated extraction. All eight HCV panel members had a result within a 0.5-log10 HCV IU/ml difference with the expected result. Six out of eight HBV panel members had a result within a 0.5-log10 HBV IU/ml difference with the expected result. The two remaining HBV panel members had differences of 1.22 and 1.03 log10 HBV IU/ml with the expected results. Compared to the mean result of 82 data sets generated for this panel, these differences were, respectively, reduced to 0.92 and 0.68 log10 HBV IU/ml and had a +2.3 SD and +0.9 SD difference from this mean result. Both samples contained HBV genotype A, as did three other samples from this panel. A former (2003) HBV proficiency panel evaluated with this "in-house" HBV PCR assay and manual extraction contained six samples of HBV genotype A that were correctly quantified (data not shown), excluding the possibility of genotype-dependent quantification. All samples from the HBV proficiency panel had an HBV load result higher than the expected value, with a mean difference of +0.55 log10 HBV IU/ml. This may be due to the factor used to convert HBV copies to HBV IU or to the standards used to calibrate the assay. This is confirmed by the mean +0.33-log10 HBV IU/ml difference obtained with this assay after manual extraction on the former (2003) HBV proficiency panel (data not shown), where the same conversion factor and standards were used. The in-house HBV, HCV, CMV, and EBV PCR assays all yielded a good concordance between the manual and automated extraction performed on patient samples. Correlation for HCV was lower than for HBV and CMV/EBV. This is in agreement with our observation that the variation of the in-house HCV assay is higher than, and for the HBV assay is equal to, the variation inherent in these in-house assays on RNA or DNA dilutions (data not shown). So, the lower correlation for HCV could be caused mainly by a higher variability of the automated and manual viral RNA extraction methods.
The DNA extraction from a large range of sample types, using different m1000 extraction protocols, has been shown to be successful (8, 17). Using the same m1000 blood protocol as was used for the blood samples, the plasma samples of a CMV proficiency panel and the Dulbecco's modified Eagle's medium with fetal calf serum samples of an EBV proficiency panel were extracted. The results for these samples suggest that the blood protocol can be used for other sample types as well, allowing the possibility of extracting different sample types in a single extraction run. Being able to extract the majority of our samples automatically with such a "generic extraction protocol" would lead to a large reduction of the turnaround time of most of our assays. Nevertheless, the advantage of a generic extraction protocol has to be weighed against the required sample equivalent to be used in the downstream assay. Other (bacterial) pathogens may require a higher sample starting volume and a lower elution volume to obtain maximum sensitivity.
A major concern in the use of automation to extract nucleic acids for use in amplification reactions is the risk of cross-contamination of negative specimens by strongly positive specimens as a consequence of aerosols, pipette leaks, or faulty robotics. The absence of cross-contamination during automated extraction on m1000 was demonstrated with high-positive BKV urine samples. These samples are a known risk in our laboratory for cross-contamination during the numerous pipetting and handling steps inherent in manual extraction. The manual extraction therefore requires rigorous handling and expertise from the analyst.
The m1000 instrument contains an eight-nozzle pipette head that can process various numbers of samples (1 to 48) in one run. Reagents are delivered in kits for 96 samples, divided into 4 reagent sets for 24 samples. Depending on the protocol used, the total assay time for the extraction of 48 samples was reduced from or identical to the total assay time needed for manual extraction. When only 24 samples were extracted, the total extraction time needed for the m1000 DNA-blood protocol was higher than the total assay time of manual extraction. For the moment, the m1000 software allows only the extraction of a maximum of 48 samples, but the hardware of the m1000 system is designed for the extraction of up to 96 samples in 1 run. A future m1000 software upgrade will result in a higher reduction of total assay time for the high-throughput assays. With all protocols, an important reduction of hands-on time was achieved, even when only 24 samples were extracted. Only the setup of the m1000 system requires hands-on time, allowing the technologists to perform other laboratory duties without interruption for 2 to 2.5 h, once the extraction has started. The reduction of hands-on time, compared to manual extraction, increases with an increasing number of samples. These findings correlate well with the evaluations of other automated extraction systems, such as MagNA Pure LC (Roche Molecular Biochemicals, Basel, Switzerland) and BioRobot M48 (QIAGEN, Hilden, Germany), where for 24 or 30 samples a small increase in total assay time but a large decrease in hands-on time were observed (7, 11). Another important factor to consider in the evaluation of an automated extraction system is the cost per test. The list prices of the manual QIAamp DNA mini-kit (Blood) and the QIAamp Ultrasens virus kit (serum/plasma) are, respectively, 2.9 and 5.7 euro/specimen. Both (DNA/RNA) m1000 sample preparation packs have a list price of 8.7 euro/specimen. Clearly, the m1000 reagents have a higher list price than the reagents for the manual extraction or even the reagents of other automated extraction systems (11).
In summary, our findings demonstrate that the m1000 instrument provides a user-friendly and versatile platform that is able to replace different manual extraction protocols as a front-end extraction system for high-throughput quantitative real-time PCR assays, resulting in a strong reduction of the required hands-on time.
ACKNOWLEDGMENTS
We thank Abbott Molecular Diagnostics (Abbott Laboratories, Abbott Park, Illinois) for providing the m1000 reagents used in this study. We are grateful to Sven Thamm for his helpful comments.
REFERENCES
Abe, A., K. Inoue, T. Tanaka, J. Kato, N. Kajiyama, R. Kawaguchi, S. Tanaka, M. Yoshiba, and M. Kohara. 1999. Quantitation of hepatitis B virus genomic DNA by real-time detection PCR. J. Clin. Microbiol. 37:2899-2903.
Candotti, D., J. Temple, S. Owusu-Ofori, and J.-P. Allain. 2004. Multiplex real-time quantitative RT-PCR assay for hepatitis B virus, hepatitis C virus, and human immunodeficiency virus type 1. J. Virol. Methods 118:39-47.
Clewly, J. P. 1989. The polymerase chain reaction, a review of the practical limitations for human immunodeficiency virus diagnosis. J. Virol. Methods 25:179-188.
Cook, L., K.-W. Ng, A. Bagabag, L. Corey, and K. R. Jerome. 2004. Use of the MagNA Pure LC automated nucleic acid extraction system followed by real-time reverse transcription-PCR for the ultrasensitive quantitation of hepatitis C virus RNA. J. Clin. Microbiol. 42:4130-4136.
Elnifro, E. M., A. M. Ashshi, R. J. Cooper, and P. E. Klapper. 2000. Multiplex PCR: optimisation and application in diagnostic virology. Clin. Microbiol. Rev. 13:559-570.
Forman, M. S., and A. Valsamakis. 2004. Verification of an assay for quantification of hepatitis C virus RNA by use of an analyte-specific reagent and two different extraction methods. J. Clin. Microbiol. 42:3581-3588.
Germer, J. J., M. M. Lins, M. E. Jensen, W. S. Harmsen, D. M. Ilstrup, P. S. Mitchell, F. R. Cockerill III, and R. Patel. 2003. Evaluation of the MagNA Pure LC instrument for extraction of hepatitis C virus RNA for the COBAS Amplicor hepatitis C virus test, version 2.0. J. Clin. Microbiol. 41:3503-3508.
Grattard, F., S. Thamm, H. Tuzet, A. Ros, E. Gaillard, and B. Pozetto. 2005. Optimisation and evaluation of the detection of Chlamydia trachomatis from sperm and urine specimens by combining an automated extraction protocol and a commercial real-time PCR assay. Clin. Microbiol. Infect. 11(Suppl. 2):493.
Jungkind, D. 2001. Automation of laboratory testing for infectious diseases using the polymerase chain reaction—our past, our present, our future. J. Clin. Virol. 20:1-6.
Kkanna, M., J. Fan, K. Pehler-Harrington, C. Waters, P. Douglass, J. Stallock, S. Kehl, and K. J. Henrickson. 2005. The Pneumoplex assays, a multiplex PCR-enzyme hybridization assay that allows simultaneous detection of five organisms, Mycoplasma pneumoniae, Legionella pneumophila, Legionella micdadei, and Bordetella pertussis, and its real-time counterpart. J. Clin. Microbiol. 43:565-571.
Knepp, J. H., M. A. Geahr, M. S. Forman, and A. Valsamakis. 2003. Comparison of automated and manual nucleic acid extraction methods for detection of enterovirus RNA. J. Clin. Microbiol. 41:3532-3536.
Kwok, S., and R. Higuchi. 1989. Avoiding false positives with PCR. Nature (London) 339:237-238.
Loeb, K. R., K. R. Jerome, J. Goddard, M.-L. Huang, A. Cent, and L. Corey. 2000. High throughput quantitative analysis of hepatitis B virus DNA in serum using the TaqMan fluoregenic detection system. Hepatology 37:626-629.
Shyamala, V., P. Arcangel, J. Cotrell, D. Coit, A. Medina-Selby, C. McCoin, D. Madriaga, D. Chien, and B. Phelps. 2004. Assessment of the target-capture PCR hepatitis B virus (HBV) DNA quantitative assay and comparison with commercial HBV DNA quantitative assays. J. Clin. Microbiol. 42:5199-5204.
Siu-Man Sum, S., D. Ka-Ho Wong, M.-F. Yuen, H.-J. Yuan, J. Yu, C.-L. Lai, D. Ho, and L. Zhang. 2004. Real-time PCR assay using molecular beacon for quantitation of hepatitis B virus DNA. J. Clin. Microbiol. 42:3438-3440.
Stuyver, L., A. Wyseur, W. V. Arnhem, F. Hernandez, and G. Maertens. 1996. Second-generation line probe assay for hepatitis C genotyping. J. Clin. Microbiol. 34:2259-2266.
Thamm, S., K. Beuselinck, T. Bourlet, S. Kares, R. Karhu, H. Mairhofer, H. Nitschko, B. Pozetto, J. Van Eldere, and G. Michel. 2005. Automated extraction of RNA and DNA from different specimen types feeding the detection of several markers in the field of molecular diagnostics. Clin. Microbiol. Infect. 11(Suppl. 2):527.
Thamm, S., J. Smith, and G. Michel. 2004. European evaluation of the m1000 nucleic extraction system, an automated solution for generic RNA and DNA extraction. J. Mol. Diagnostics 6:155.
Zhao, J.-R., Y.-J. Bai, Q.-H. Zhang, Y. Wan, D. Li, and X.-J. Yan. 2005. Detection of hepatitis B virus DNA by real-time PCR using Taqman-MGB probe technology. World J. Gastroenterol. 11:508-510.(Kurt Beuselinck, Marc van)
ABSTRACT
The performance of the m1000 system (Abbott Laboratories, Illinois) as a front-end extraction system for high-throughput "in-house" quantitative real-time PCR assays was analyzed and compared to that of manual extraction of plasma and serum samples (hepatitis C virus [HCV] and hepatitis B virus [HBV]) and EDTA-blood samples (cytomegalovirus [CMV] and Epstein-Barr virus [EBV]). Linearity of extraction was tested on dilution series of HCV and HBV reference materials. The correlation coefficient for standard curves based on repeated extraction runs was 0.97 ± 0.06 for HCV and 0.97 ± 0.03 for HBV, indicating a linear extraction from 100 to 1.0 x 105 HCV IU/ml and from 100 to 1.0 x 106 HBV IU/ml. Intra- and interrun variability was below 0.23 log10 IU/ml for 2.98 to 5.28 log10 HCV IU/ml and 2.70 to 5.20 log10 HBV IU/ml. Correlation between automated and manual extraction was very good. For HCV, the correlation coefficient was 0.91 and the mean difference in viral load was 0.13 log10 HCV IU/ml. For HBV, the correlation coefficient was 0.98 and the mean difference in viral load 0.61 log10 HBV IU/ml. For CMV and EBV, the correlation coefficient was 0.98 and the mean difference in viral load 0.33 log10 copies/ml. Accuracy was confirmed with a reference panel (QCMD, Glasgow, Scotland) for all four assays. No cross-contamination was observed when extracting strongly positive polyomavirus samples (8.10 log10 copies/ml) interspersed with polyomavirus-negative samples. Automated extraction via the m1000 system offers a high reliability of extraction and resulted in a strong reduction of the required extraction hands-on time for high-throughput PCR compared to manual extraction protocols.
INTRODUCTION
Real-time thermocyclers have greatly decreased the amount of hands-on time in nucleic acid (NA)-based diagnostics for pathogens. The use of commercially available universal master mixes that contain all reagents except the pathogen-specific primer-probe combination has further decreased the number of pipetting steps and thus reduced labor and the possibility of errors.
NA extraction has now become the most critical and labor-intensive step in NA-based diagnostic assays. The overall sensitivity of the assay is determined by the NA yield, its purity, and the amount of sample equivalents that can be transferred into the amplification reaction. Conventional manual sample preparation methods are labor intensive and susceptible to contamination, handling variations, or errors (3, 9, 12).
Since both the pathogen range and the number of different sample types are expanding and since multiplex downstream testing is becoming standard practice, there is a need for a generic extraction method (2, 5, 10). Ideally, the NA extraction procedure should yield pure NA from different pathogens and from a broad range of sample types.
Recently, Abbott introduced m1000, an automated generic RNA and DNA extraction system using magnetic microparticle processing (17, 18). The m1000 system provides the necessary features for full automation of NA extraction of up to 48 samples in 2 h or less, depending on the protocol. It uses ready-to-use reagents and hands-off operation from the moment of loading of the patient specimen in primary tubes onto the m1000 instrument up to completed NA extract.
Both the RNA and DNA extraction chemistries of the m1000 system are based on lysis and nuclease inactivation with guanidinium isothiocyanate. The RNA extraction chemistry further uses uncoated iron particles as the capture medium. The DNA extraction chemistry uses silica-coated magnetic particles as the solid phase.
The processes of sample and reagent pipetting, heating and incubation, magnetic capture and washing, and the final elution are optimized in ready-to-use m1000 protocols. These protocols are provided by Abbott and can be altered in accordance with the specifications and needs of the user.
In this study, the performance of m1000 as a front-end extraction system for high-throughput "in-house" real-time virological PCR assays (m1000 "open mode") was evaluated. We verified m1000 protocols for use with the following quantitative tests: hepatitis C virus (HCV) RNA, hepatitis B virus (HBV) DNA, cytomegalovirus (CMV) DNA, and Epstein-Barr virus (EBV) DNA.
The linearity for automated extraction preceding the HCV RNA and HBV DNA PCR was determined using dilution series of reference materials, while intra- and interrun variations were determined using three dilutions of an HCV- or HBV-positive sample. Accuracy was tested by proficiency panels for these two assays, and the HCV RNA and HBV DNA viral load results for automated and manual extraction were compared for clinical plasma and serum samples. For the CMV DNA and EBV DNA quantitative PCR, the correlation of viral load results for EDTA blood samples and two proficiency panels by the automated and manual methods were assessed.
The possible risk of sample cross-contamination on m1000 was evaluated through an "in-house" real-time polyomavirus (JC virus [JCV]/BK virus [BKV]) PCR assay with urine samples.
MATERIALS AND METHODS
Manual extraction protocols. Manual extraction of HCV RNA and HBV DNA starting from 1 ml serum or plasma was done according to the protocol of the QIAamp Ultrasens virus kit (QIAGEN, Hilden, Germany). Internal control (6 x 109 copies; IC1, see "Viral load determinations") were added together with the carrier RNA to the sample. Elution was performed in a 50-μl volume.
All other manual extractions were performed according to the protocol of the QIAamp DNA minikit. IC1 (6 x 109 copies) were added to the sample during the first step of the protocol together with proteinase K. The elution volume was 100 μl.
Automated extraction protocols. For fully automated sample preparation, all extraction protocols were provided by Abbott. RNA extractions were done using the Sample Preparation System kit (Abbott) after addition of 1.8 x 1011 copies of IC1 to a bottle of lysis buffer. The Sample Preparation System-kit DNA (Abbott) was used for DNA extractions after addition of 1.8 x 1012 copies of IC1 to a bottle of lysis buffer. The RNA-plasma-BA-500-110-v1 extraction protocol was used to extract HCV RNA from 0.5 ml serum or plasma, resulting in a final elution volume of 110 μl. HBV DNA was extracted from 0.5 ml serum or plasma with the DNA-plasma-500-110-v1 extraction protocol, resulting in an elution volume of 110 μl. Automated extractions for CMV and EBV DNA were performed with the DNA-Blood-BA-300-150-v1 protocol, starting from 0.3 ml of sample and eluting in a 150-μl volume. For the cross-contamination study with the BKV positive and negative urine samples, the DNA-urine-LL-400-70-v2 protocol was used to extract DNA from 0.4 ml in a end volume of 70 μl.
Viral load determinations. All "in-house" real-time virological PCR assays were performed on an ABI 7900 real-time thermocycler (Applied Biosystems, Foster City, California) in duplicate: one reaction well for a single PCR for the viral target of interest and the second reaction well for a multiplex reaction for the viral target and IC1. IC1 is a heterologous 66-bp oligonucleotide added to the sample during the extraction procedures. In the absence of inhibition of the reaction, IC1 gives a constant result. Reported results are based on the results of the single reactions. The multiplex reaction is used as a control for inhibition (result IC1) and as a confirmation of the result of the single reaction (viral target).
The real-time one-step reverse transcriptase (RT)-PCR for the amplification and detection of HCV RNA was carried out using the QIAGEN OneStep RT-PCR kit with addition of 0.15 μM of in-house carboxy-X-rhodamine dye PRTM (6-carboxyfluorescein [FAM]-5'-TATCTAGT-3' noncoding region) as a passive reference. A 262-bp gene fragment in the HCV 5' carboxy-X-rhodamine was amplified with 0.5 μM of each biotinylated primer: HCV1FB (5'-Biotin-CTAGCCATGGCGTTAGTATGAGTGT-3') and HCV2RB (Biotin-5'-GGTGCACGGTCTACGAGACCT-3'; Biotin). Primers were biotinylated so that the amplification product could be used directly in the InnoLiPA HCV II assay (Innogenetics, Belgium) (16). The single one-step RT-PCR further contained a final concentration of 1x OneStep RT-PCR buffer, 400 μM of each deoxynucleoside triphosphate, and 0.4 μM probe HCPr96FTM (FAM-5'-CACTCGCAAGCACCCTATCAGGCAGT-3'-TAMRA) (16). RT-PCR after manual RNA extraction was carried out in a reaction volume of 40 μl containing 10 μl of the manually extracted RNA and 1.6 μl OneStep RT-PCR enzyme mix. RT-PCR after automated extraction was carried out in a reaction volume of 50 μl containing 25 μl of the m1000-extracted RNA and 2.0 μl OneStep RT-PCR enzyme mix. The multiplex one-step RT-PCR was performed in the same way as the single reaction with the addition of 0.2 μM (each) primers IC1P1 (5'-TCATGGTCAAATGCTGGATGA-3') and IC1P2 (5'-CTCTGCAATCAGCTCACGAAAC-3') and 0.1 μM ICPTM (VIC-5'-TCCATGGAGACAGCCGCACCA-3'-TAMRA), i.e., the primers and probe to detect IC1. The reactions were carried out with an initial reverse transcription step at 48°C for 30 min, followed by PCR activation at 95°C for 10 min and 50 cycles of amplification (95°C for 15 s, 60°C for 1 min). Real-time data collection was performed at 60°C. Quantitative results are reported from 2.0 log10 HCV IU/ml to 6.7 log10 HCV IU/ml. All real-time DNA PCRs were carried out in a reaction volume of 40 μl containing 10 μl of the DNA extract, 20 μl 2x Universal Mastermix (Applied Biosystems, Foster City, California), and the primers/probe specific for the viral target: 0.15 μM HBV3F (5'-TCCCCGTCTGTKCCTTCTC-3'), 0.15 μM HBV4R (5'-GCGTTCAYGGTGGTYKCCAT-3'), and 0.10 μM HBV1T (FAM-5'-CCGTGTGCACTTCGCTTCACCTCTGC-3'-TAMRA) (13) for HBV; 0.25 μM CMCP11 (5'-CGTAACGTGGACCTGACGTTT-3'), 0.25 μM CMCP12 (5'-CACGGTCCCGGTTTAGCA-3'), and 0.20 μM CMCP3TM (FAM-5'-TATCTGCCCGAGGATCGCGGTTACA-3') for CMV; 0.25 μM EBBKT-1 (5'-TTTGGACCCGAAATCTGACACT-3'), 0.25 μM EBBKT2 (5'-GCCAACCATAGACCCGCTTC-3'), and 0.20 μM EBBKT-F (FAM-5'-CCATTTTGTCCCCACGCGCG-3'-TAMRA) for EBV; 0.5 μM PEP-3 (5'-GGAAAGTCTTTAGGGTCTTCTACCTTT-3'), 0.25 μM PEP-5 (5'-GATGAAGATTTATTYTGCCATGARGA-3'), 0.25 μM PEP-6 (5'-GAAGACCTGTTTTGCCATGAAGA-3'), and 0.20 μM HuPo-1 (FAM-5'-ATCACTGGCAAACAT-MGB-3') for polyomaviruses JCV and BKV. The multiplex DNA PCRs were performed in the same way as the single reactions with the addition of 0.075 μM (each) primers IC1P1 and IC1P2 and 0.15 μM ICPTM. The reactions were carried out with an initial uracyl-N-glycosylase incubation step at 50°C for 2 min, followed by PCR activation at 95°C for 10 min and 45 cycles of amplification (95°C for 15 s, 60°C for 1 min). Real-time data collection was performed at 60°C. Quantitative results for HBV are reported from 2.0 log10 HBV IU/ml to 6.7 log10 HBV IU/ml, for CMV and EBV from 2.0 log10 copies/ml to 6.7 log10 copies/ml, and for polyomavirus from 2.0 log10 copies/ml to 7.70 log10 copies/ml.
All "in-house" real-time virological PCR assays with the manual sample preparation protocol have been validated using commercial standards. For CMV and EBV, a quantitated CMV and EBV DNA control (Advanced Biotechnologies, Inc., Columbia, Maryland) was used, and for HCV and HBV, a quantitated HCV and HBV serum control (Acrometrix, Benica, California) was used. The performance of each assay with the manual sample preparation protocol has been verified with a virus-specific proficiency panel (QCMD, Glasgow, Scotland). Each of the four proficiency panels contained eight samples, and all results were quantified within the range of acceptation (geometric mean ± 1 standard deviation [SD] of all reported quantitative data sets) for this panel.
Study design. In the first part of the study, the linearity and accuracy of the in-house real-time HCV and HBV PCR with the automated sample preparation protocol was determined with dilutions of commercially available HCV and HBV controls (NAP-HCV007 and NAP-HBV006; Acrometrix) and the samples of the HCV and HBV 2004 QCMD proficiency program panels.
In the second part, the intrarun and interrun variations for the HCV and HBV assays with automated sample preparation were determined using three different dilutions of a positive sample for both assays. All dilutions were tested five times in each of four runs. At the same time, the performance of both assays with automated sample preparation was compared to the performance of the assays with the manual extraction method. A total of 40 clinical plasma/serum samples were investigated for HCV and 40 clinical plasma/serum samples for HBV.
In the third part, the correlation and agreement of CMV and EBV viral load results for 60 EDTA blood samples and the samples of the CMV and EBV 2004 QCMD proficiency program panels, extracted via the automated and the manual extraction method, were assessed.
In a last and fourth part, the possible risk of cross-contamination with m1000 was tested with the DNA-urine-LL-400-70-v2 protocol as a front-end for the "in-house" real-time polyomavirus (JCV/BKV) PCR assay. Two milliliters of a highly positive urine sample for BKV (1.8 x 109 copies/ml) was diluted to 25 ml with a viral transport medium. This diluted urine sample was loaded onto m1000 to obtain a checkerboard pattern with the pure viral transport medium.
RESULTS
Linearity. Four dilutions of the commercially available HCV control (NAP-HCV007) (2 x 106 HCV IU/ml), which contained 1.0 x 105, 1.0 x 104, 1.0 x 103, and 100 HCV IU/ml, were each extracted five to seven times in as many separate extraction runs together with the HCV-negative diluent. Five dilutions of the commercial HBV control (NAP-HBV006) (2 x 107 HBV IU/ml), which contained 1.0 x 106, 1.0 x 105, 1.0 x 104, 1.0 x 103, and 100 HBV IU/ml, were extracted two to eight times, in as many different extraction runs, together with the HBV-negative diluent. None of the negative specimens yielded a positive result in either reaction, while 100 HCV IU/ml and 100 HBV IU/ml were detected in all replicates.
The correlation coefficient for the standard curves based on the different extraction runs of these dilutions was 0.970 ± 0.060 for HCV and 0.971 ± 0.025 for HBV, indicating a linear response for 100 to 1.0 x 105 HCV IU/ml and for 100 to 1.0 x 106 HBV IU/ml. The mean PCR efficiencies calculated from these standard curves was 83% for HCV and 101% for HBV.
Accuracy. Samples of the HCV 2004 QCMD proficiency program panel were tested with automated extraction and the in-house real-time HCV-PCR. All results were found to be within 0.5 log10 HCV IU/ml of the expected result (Table 1). One weakly positive sample with an expected result of 2.23 log10 HCV IU/ml had a result (1.90 log10 HCV IU/ml) below the quantification range of our assay and was reported as "< 2.0 log10 HCV IU/ml." Automated extraction on m1000 followed by in-house real-time HBV-PCR performed on the samples of the HBV 2004 QCMD proficiency program panel resulted in an HBV viral load difference with the expected results, varying from +0.12 to +1.22 log10 HBV IU/ml for all panel members (Table 1). Our in-house HBV PCR results are expressed in HBV IU/ml because we used a commercially available reference material calibrated against the international standard to generate a standard curve. The expected HBV results were expressed in HBV copies/ml and were divided by 5 (– 0.70 log HBV IU/ml) (14) to obtain the reference results expressed in HBV IU/ml.
Interassay and intra-assay variations. Determination of interassay and intra-assay variations with automated sample preparation was performed with three dilutions of a sample positive for either HCV or HBV. These dilutions had viral loads, determined with manual extraction, of 5.28, 3.38, and 2.98 log10 HCV IU/ml for HCV and 5.20, 3.20 and 2.70 log10 HBV IU/ml for HBV. All dilutions were tested five times in each of four runs. The mean results and the variations for each of the dilutions are given in Table 2. For HBV, the two weakest samples both had one negative result (<2.0 log10 HBV IU/ml). Because the IC1 signal for these two samples was stronger than the IC1 result for all other samples in this study, these two results were removed as outliers. PCR inhibition normally results in a weakened or negative IC1 signal. The cause of this stronger IC1 signal remains unclear to us.
Correlation with manual extraction. Extraction with the m1000 system followed by real-time HCV PCR for 40 clinical samples and comparison with the results obtained with the routine manual extraction resulted in a positive result for all 34 samples which were also positive with the manual extraction. The total correlation for these 34 samples between the two methods was good (r = 0.91), as shown in Fig. 1A. The mean difference in viral load was 0.13 log10 HCV IU/ml. The 95% confidence interval (CI) (± 2 standard deviations) for the mean difference was –0.75 and +1.01 log10 HCV IU/ml. The three samples with a log10 difference of more than 0.71 between the two extractions were repeated by both extraction methods on the same sample aliquot and analyzed again. When these three reanalysis results were taken into account, a final mean difference in viral load of 0.04 log10 HCV IU/ml and a 95% CI for the mean difference of –0.70 to +0.79 log10 HCV IU/ml were obtained.
Extraction with the m1000 system followed by real-time HBV PCR for 40 clinical samples and comparison with the results obtained during the routine clinical diagnosis resulted in a positive result for 25 samples which were also positive with the manual extraction. The total correlation for the 25 samples positive by both methods was very good (r = 0.98), as shown in Fig. 1B. The mean difference in viral load is 0.61 log10 HBV IU/ml. The 95% CI for the mean difference was –0.44 and+1.66 log10 HBV IU/ml. The three samples with a log10 difference of more than 1.0 between the two extractions were repeated by both extraction methods on the same sample aliquot and analyzed again. When these three reanalysis results were taken into account, a final mean difference in viral load of 0.128 log10 HBV IU/ml and a 95% CI for the mean difference of –0.75 to +1.01 log10 HBV IU/ml were obtained. Nine samples gave a different result after automated extraction compared to the routine manual extraction: six samples that had been negative with manual extraction gave a positive result with m1000, and three weakly positive samples with manual extraction (2.20, 2.09 and 2.24 log10 HBV IU/ml) were negative with m1000. Only four out of these nine samples had sufficient fluid left to repeat both extraction methods on the same sample aliquot. These four samples that were positive with automated extraction and negative with manual extraction were now positive with both extraction methods, with a mean viral load ranging from 2.40 to 3.38 log10 HBV IU/ml and a mean log difference of 0.25 log10 HBV IU/ml between the two methods.
Sixty EDTA-blood samples were analyzed using automated extraction on m1000 prior to real-time CMV or EBV PCR. Results were compared with the results obtained during the routine clinical diagnostic test using the same real-time CMV or EBV PCR preceded by manual extraction. Twenty blood samples were negative (<2.70 log10 copies/ml) with both extraction methods, and 33 samples yielded a positive result with both extraction methods. The correlation for these 33 samples was good (r = 0.95) (Fig. 1C). The mean log10 copies/ml difference for the samples positive by both methods was 0.33. The 95% CI for the mean difference was –0.23 and +0.89 log10 copies/ml. Six samples with a viral load between 2.70 and 2.93 log10 copies/ml were negative after automated extraction. For one of these samples, the IC1 was invalid (inhibition of PCR) after automated extraction. One negative sample was positive (3.41 log10) after automated extraction. Both extraction methods were repeated on the same sample aliquot for these seven samples. For the sample with inhibition of the PCR after extraction with the m1000 system, inhibition of the PCR was confirmed after automated extraction, while the positive result with manual extraction was also confirmed. The discordant results on the sample positive after automated extraction and negative after manual extraction were confirmed with both extraction methods. In the five remaining samples, a low viral load was detected with manual and automated extraction, with most manual extractions just above (between 2.71 and 3.36 log10 copies/ml) and most automated extractions just below (between 2.17 and 2.80 log10 copies/ml) the limit of detection. With the same automated m1000 (blood) protocol, the samples of the CMV 2004 QCMD and EBV 2004 QCMD proficiency program panel were tested with manual and automated extraction. The 12 CMV panel samples consisted of 12 lyophilized samples in a plasma matrix. The eight EBV panel members contained different concentrations of EBV diluted in Dulbecco's modified Eagle's medium with fetal calf serum. An overview of the panel result range, the routine diagnostic result (manual extraction), and the result after automated extraction is given in Table 3. CMV and EBV viral load results after manual and automated extraction with the samples of the proficiency program panels were within the reported result range for six out of seven CMV-positive samples and six out of seven EBV-positive samples. The single CMV-positive sample out of range (CMV04-12) after automated extraction was detected but outside the quantification range of our assay (<2.70 log10 copies/ml) both with manual and automated extraction. One weakly EBV-positive sample (EBV04-01) was outside the quantification range of our assay (<2.7 log10 copies/ml) after automated extraction but had an EBV load of 3.0 log10 copies/ml after manual extraction.
Cross-contamination. Twenty-four strongly positive and 24 negative samples were loaded on the m1000 instrument in such a way that a checkerboard pattern was obtained in the m1000 "1 ml subsystem carrier." The "1 ml subsystem carrier" contains the reaction vessels, and therefore it can be regarded as the main processing unit of m1000, where all processes of sample and reagent mixing, heating, incubation, magnetic capture, washing, and elution take place. After real-time polyomavirus PCR, none of the 24 negative samples gave a positive signal, while all 24 positive samples gave a nice amplification plot resulting in a mean BK viral load of 8.10 log copies/ml with a standard deviation of 0.07 log copies/ml.
Time estimates. The estimated time required to perform the manual and m1000 extractions is presented in Table 4. The estimated hands-on setup time for the m1000 system was ca. 30 min. Batches for manual extractions were limited to 24 samples due to the capacity of our microcentrifuge. Therefore, for batches of 48 samples, two consecutive batches of 24 samples had to be performed. Compared with the manual sample preparation for the HCV and HBV assay on 48 plasma or serum samples, a 114-min decrease in total time and 210-min decrease of hands-on time was achieved. The automated extraction of 48 EDTA-blood samples had no influence on total time, but there was a 140-min decrease of hands-on time compared to the manual DNA extraction protocol. When only 24 samples were extracted, the total assay time for automated extraction was identical to or even higher than the total assay time for manual extraction, but there still was a significant decrease in hands-on time, with a 90-min decrease in hands-on time for the plasma/serum DNA or RNA extraction and a 55-min decrease for the blood DNA extraction. In addition to the time required for sample preparation, the reaction setup and the amplification on ABI 7900 for 48 samples (96 reactions) required ca. 3.5 h total time with 1 h hands-on time.
DISCUSSION
Conventional manual extraction protocols for the extraction of pathogen DNA or RNA from clinical samples are the most labor-intensive and critical part in current NA diagnostics assays for pathogens. Recently, Abbott developed m1000, an automated extraction system for nucleic extraction in routine diagnostic laboratories. The m1000 system not only provides NA protocols as a front end for the Abbott molecular assays ("closed-mode") but also allows the protocols to be adapted to the specific needs of the user ("open-mode"). This makes the m1000 system a truly open platform that allows the user to define protocols for a variety of different samples and applications. In this study, m1000 was evaluated as a NA extraction system for samples of five in-house real-time PCR assays. The in-house real-time HCV and HBV PCR assays with the automated sample preparation protocol showed sufficient linearity and sensitivity on dilution series of, respectively, a HCV and HBV standard. The interassay and intra-assay percent coefficient of variation for the logarithmic viral load value ranged from 2.78% to 6.27% for HCV and from 1.85% to 8.11% for HBV. These results are in concordance with those reported for manual or automated extraction systems and a real-time HCV-PCR (4, 6) or HBV-PCR (1, 15, 19). Samples of HCV and HBV proficiency program panels were tested by the respective in-house PCR assays after automated extraction. All eight HCV panel members had a result within a 0.5-log10 HCV IU/ml difference with the expected result. Six out of eight HBV panel members had a result within a 0.5-log10 HBV IU/ml difference with the expected result. The two remaining HBV panel members had differences of 1.22 and 1.03 log10 HBV IU/ml with the expected results. Compared to the mean result of 82 data sets generated for this panel, these differences were, respectively, reduced to 0.92 and 0.68 log10 HBV IU/ml and had a +2.3 SD and +0.9 SD difference from this mean result. Both samples contained HBV genotype A, as did three other samples from this panel. A former (2003) HBV proficiency panel evaluated with this "in-house" HBV PCR assay and manual extraction contained six samples of HBV genotype A that were correctly quantified (data not shown), excluding the possibility of genotype-dependent quantification. All samples from the HBV proficiency panel had an HBV load result higher than the expected value, with a mean difference of +0.55 log10 HBV IU/ml. This may be due to the factor used to convert HBV copies to HBV IU or to the standards used to calibrate the assay. This is confirmed by the mean +0.33-log10 HBV IU/ml difference obtained with this assay after manual extraction on the former (2003) HBV proficiency panel (data not shown), where the same conversion factor and standards were used. The in-house HBV, HCV, CMV, and EBV PCR assays all yielded a good concordance between the manual and automated extraction performed on patient samples. Correlation for HCV was lower than for HBV and CMV/EBV. This is in agreement with our observation that the variation of the in-house HCV assay is higher than, and for the HBV assay is equal to, the variation inherent in these in-house assays on RNA or DNA dilutions (data not shown). So, the lower correlation for HCV could be caused mainly by a higher variability of the automated and manual viral RNA extraction methods.
The DNA extraction from a large range of sample types, using different m1000 extraction protocols, has been shown to be successful (8, 17). Using the same m1000 blood protocol as was used for the blood samples, the plasma samples of a CMV proficiency panel and the Dulbecco's modified Eagle's medium with fetal calf serum samples of an EBV proficiency panel were extracted. The results for these samples suggest that the blood protocol can be used for other sample types as well, allowing the possibility of extracting different sample types in a single extraction run. Being able to extract the majority of our samples automatically with such a "generic extraction protocol" would lead to a large reduction of the turnaround time of most of our assays. Nevertheless, the advantage of a generic extraction protocol has to be weighed against the required sample equivalent to be used in the downstream assay. Other (bacterial) pathogens may require a higher sample starting volume and a lower elution volume to obtain maximum sensitivity.
A major concern in the use of automation to extract nucleic acids for use in amplification reactions is the risk of cross-contamination of negative specimens by strongly positive specimens as a consequence of aerosols, pipette leaks, or faulty robotics. The absence of cross-contamination during automated extraction on m1000 was demonstrated with high-positive BKV urine samples. These samples are a known risk in our laboratory for cross-contamination during the numerous pipetting and handling steps inherent in manual extraction. The manual extraction therefore requires rigorous handling and expertise from the analyst.
The m1000 instrument contains an eight-nozzle pipette head that can process various numbers of samples (1 to 48) in one run. Reagents are delivered in kits for 96 samples, divided into 4 reagent sets for 24 samples. Depending on the protocol used, the total assay time for the extraction of 48 samples was reduced from or identical to the total assay time needed for manual extraction. When only 24 samples were extracted, the total extraction time needed for the m1000 DNA-blood protocol was higher than the total assay time of manual extraction. For the moment, the m1000 software allows only the extraction of a maximum of 48 samples, but the hardware of the m1000 system is designed for the extraction of up to 96 samples in 1 run. A future m1000 software upgrade will result in a higher reduction of total assay time for the high-throughput assays. With all protocols, an important reduction of hands-on time was achieved, even when only 24 samples were extracted. Only the setup of the m1000 system requires hands-on time, allowing the technologists to perform other laboratory duties without interruption for 2 to 2.5 h, once the extraction has started. The reduction of hands-on time, compared to manual extraction, increases with an increasing number of samples. These findings correlate well with the evaluations of other automated extraction systems, such as MagNA Pure LC (Roche Molecular Biochemicals, Basel, Switzerland) and BioRobot M48 (QIAGEN, Hilden, Germany), where for 24 or 30 samples a small increase in total assay time but a large decrease in hands-on time were observed (7, 11). Another important factor to consider in the evaluation of an automated extraction system is the cost per test. The list prices of the manual QIAamp DNA mini-kit (Blood) and the QIAamp Ultrasens virus kit (serum/plasma) are, respectively, 2.9 and 5.7 euro/specimen. Both (DNA/RNA) m1000 sample preparation packs have a list price of 8.7 euro/specimen. Clearly, the m1000 reagents have a higher list price than the reagents for the manual extraction or even the reagents of other automated extraction systems (11).
In summary, our findings demonstrate that the m1000 instrument provides a user-friendly and versatile platform that is able to replace different manual extraction protocols as a front-end extraction system for high-throughput quantitative real-time PCR assays, resulting in a strong reduction of the required hands-on time.
ACKNOWLEDGMENTS
We thank Abbott Molecular Diagnostics (Abbott Laboratories, Abbott Park, Illinois) for providing the m1000 reagents used in this study. We are grateful to Sven Thamm for his helpful comments.
REFERENCES
Abe, A., K. Inoue, T. Tanaka, J. Kato, N. Kajiyama, R. Kawaguchi, S. Tanaka, M. Yoshiba, and M. Kohara. 1999. Quantitation of hepatitis B virus genomic DNA by real-time detection PCR. J. Clin. Microbiol. 37:2899-2903.
Candotti, D., J. Temple, S. Owusu-Ofori, and J.-P. Allain. 2004. Multiplex real-time quantitative RT-PCR assay for hepatitis B virus, hepatitis C virus, and human immunodeficiency virus type 1. J. Virol. Methods 118:39-47.
Clewly, J. P. 1989. The polymerase chain reaction, a review of the practical limitations for human immunodeficiency virus diagnosis. J. Virol. Methods 25:179-188.
Cook, L., K.-W. Ng, A. Bagabag, L. Corey, and K. R. Jerome. 2004. Use of the MagNA Pure LC automated nucleic acid extraction system followed by real-time reverse transcription-PCR for the ultrasensitive quantitation of hepatitis C virus RNA. J. Clin. Microbiol. 42:4130-4136.
Elnifro, E. M., A. M. Ashshi, R. J. Cooper, and P. E. Klapper. 2000. Multiplex PCR: optimisation and application in diagnostic virology. Clin. Microbiol. Rev. 13:559-570.
Forman, M. S., and A. Valsamakis. 2004. Verification of an assay for quantification of hepatitis C virus RNA by use of an analyte-specific reagent and two different extraction methods. J. Clin. Microbiol. 42:3581-3588.
Germer, J. J., M. M. Lins, M. E. Jensen, W. S. Harmsen, D. M. Ilstrup, P. S. Mitchell, F. R. Cockerill III, and R. Patel. 2003. Evaluation of the MagNA Pure LC instrument for extraction of hepatitis C virus RNA for the COBAS Amplicor hepatitis C virus test, version 2.0. J. Clin. Microbiol. 41:3503-3508.
Grattard, F., S. Thamm, H. Tuzet, A. Ros, E. Gaillard, and B. Pozetto. 2005. Optimisation and evaluation of the detection of Chlamydia trachomatis from sperm and urine specimens by combining an automated extraction protocol and a commercial real-time PCR assay. Clin. Microbiol. Infect. 11(Suppl. 2):493.
Jungkind, D. 2001. Automation of laboratory testing for infectious diseases using the polymerase chain reaction—our past, our present, our future. J. Clin. Virol. 20:1-6.
Kkanna, M., J. Fan, K. Pehler-Harrington, C. Waters, P. Douglass, J. Stallock, S. Kehl, and K. J. Henrickson. 2005. The Pneumoplex assays, a multiplex PCR-enzyme hybridization assay that allows simultaneous detection of five organisms, Mycoplasma pneumoniae, Legionella pneumophila, Legionella micdadei, and Bordetella pertussis, and its real-time counterpart. J. Clin. Microbiol. 43:565-571.
Knepp, J. H., M. A. Geahr, M. S. Forman, and A. Valsamakis. 2003. Comparison of automated and manual nucleic acid extraction methods for detection of enterovirus RNA. J. Clin. Microbiol. 41:3532-3536.
Kwok, S., and R. Higuchi. 1989. Avoiding false positives with PCR. Nature (London) 339:237-238.
Loeb, K. R., K. R. Jerome, J. Goddard, M.-L. Huang, A. Cent, and L. Corey. 2000. High throughput quantitative analysis of hepatitis B virus DNA in serum using the TaqMan fluoregenic detection system. Hepatology 37:626-629.
Shyamala, V., P. Arcangel, J. Cotrell, D. Coit, A. Medina-Selby, C. McCoin, D. Madriaga, D. Chien, and B. Phelps. 2004. Assessment of the target-capture PCR hepatitis B virus (HBV) DNA quantitative assay and comparison with commercial HBV DNA quantitative assays. J. Clin. Microbiol. 42:5199-5204.
Siu-Man Sum, S., D. Ka-Ho Wong, M.-F. Yuen, H.-J. Yuan, J. Yu, C.-L. Lai, D. Ho, and L. Zhang. 2004. Real-time PCR assay using molecular beacon for quantitation of hepatitis B virus DNA. J. Clin. Microbiol. 42:3438-3440.
Stuyver, L., A. Wyseur, W. V. Arnhem, F. Hernandez, and G. Maertens. 1996. Second-generation line probe assay for hepatitis C genotyping. J. Clin. Microbiol. 34:2259-2266.
Thamm, S., K. Beuselinck, T. Bourlet, S. Kares, R. Karhu, H. Mairhofer, H. Nitschko, B. Pozetto, J. Van Eldere, and G. Michel. 2005. Automated extraction of RNA and DNA from different specimen types feeding the detection of several markers in the field of molecular diagnostics. Clin. Microbiol. Infect. 11(Suppl. 2):527.
Thamm, S., J. Smith, and G. Michel. 2004. European evaluation of the m1000 nucleic extraction system, an automated solution for generic RNA and DNA extraction. J. Mol. Diagnostics 6:155.
Zhao, J.-R., Y.-J. Bai, Q.-H. Zhang, Y. Wan, D. Li, and X.-J. Yan. 2005. Detection of hepatitis B virus DNA by real-time PCR using Taqman-MGB probe technology. World J. Gastroenterol. 11:508-510.(Kurt Beuselinck, Marc van)