Comparative Evaluation of the Automated Roche TaqMan Real-Time Quantitative Human Immunodeficiency Virus Type 1 RNA PCR Assay and the Roche AMPLICOR Version 1.5 Conventional PCR Assay

The need to evaluate antiviral treatment response and the emergenceof resistance have made the human immunodeficiency virus (HIV)viral load assay a major feature of the diagnostic monitoringof HIV-infected individuals. The objective of this study wasto evaluate the utility of the recently In Vitro DiagnosticMedical Devices Directive-approved Roche COBAS AmpliPrep/TaqMan96real-time PCR assay by comparison with the existing Roche COBASAmpliPrep/AMPLICOR MONITOR conventional PCR assay. EDTA-treatedplasma samples from 191 HIV-1-infected individuals were testedfor HIV-1 RNA by the AMPLICOR assay and the TaqMan assay. Thiswas a prospective study using 191 pairs of samples from thesame bleed per patient. The correlation coefficient of the assayswas 98.08%. The mean difference between the assays was 0.05log₁₀ copy/ml plasma, with a standard deviation (SD) of 0.27log₁₀ copy/ml plasma. Thirteen samples gave results with variancesgreater than 0.5 log₁₀ copy/ml plasma, which is our clinicalcutoff. Two samples were more than 3 SD different (0.81 log₁₀copy/ml plasma). The TaqMan assay appeared to be slightly moresensitive at the lower end of the dynamic range. The assayscorrelated significantly (P > 0.95) with each other, andthe regression analysis was also highly significant (R² >0.95).

INTRODUCTION

Top

INTRODUCTION

The utility of monitoring human immunodeficiency virus type1 (HIV-1) RNA viral load in plasma is well documented (1). Whilethe HIV-1 RNA viral load and CD4 lymphocyte count are relatedto prognosis, the issue now seems more complex than first thought(4, 9). However, with the advent of combination HIV drug therapyin 1996, the need to evaluate antiviral treatment response andthe emergence of resistance mean this assay has remained a majorfeature in monitoring HIV infections (8).

Many patented methodologies have been made commercially availableto diagnostic laboratories, and the most widely used in largeUnited Kingdom diagnostic laboratories have been quantitativePCR and quantitative branched-chain DNA assays. The early PCRassays were difficult to perform as high-volume diagnostic tests,and it was not until the introduction of semiautomated systems,such as the Roche COBAS AMPLICOR analyzer, that PCR became amore robust tool for use in this clinical setting. The automationof the nucleic extraction step of the assay using the AmpliPrepextractor further facilitated reliable high-volume testing.

The Roche COBAS TaqMan quantitative HIV 1 RNA assay is moreautomated than the AMPLICOR assay, requiring no manual interventionbetween the initial addition of sample to an assay tube andthe generation of the quantitative result. Additionally theassay, being a real-time PCR assay, has a wider dynamic rangethan its predecessor when trying to establish results with sampleswith high viral loads (>4.87 log₁₀ copies/ml plasma) (COBASAmpliPrep/COBAS TaqMan HIV-1 test kit insert; Roche MolecularDiagnostics, Basel, Switzerland). The assay uses the TaqManreal-time PCR technology (3). The primer and probes of bothof the Roche assays in comparison target conserved regions ofthe gag region of HIV-1.

Here we present a prospective study, carried out to evaluatethe utility of the COBAS AmpliPrep/TaqMan96 (referred to hereafteras TaqMan) assay for HIV-1 RNA in comparison with the existingRoche COBAS AmpliPrep/AMPLICOR Monitor (referred to hereafteras AMPLICOR) assay in use in our laboratory. The quantitativeresults of the assays are compared.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

EDTA plasma samples from 191 HIV-1-infected individuals weretested for HIV-1 RNA by the AMPLICOR assay and the TaqMan assay.The HIV clinics at the Royal London and St. Bartholomew's Hospitalswere asked to provide duplicate samples from the same bleedfor a 2-week period to enable testing by both methods.

All samples were centrifuged at 3,000 rpm for 20 min, and theplasma was frozen at –80°C before testing as per thelaboratory's standard operating procedure. From each pair, onesample was tested by the AMPLICOR assay and the second by theTaqMan assay. The assays were performed strictly according tothe manufacturer's instructions.

When performing the AMPLICOR assay, a decision was made, basedupon the available patient information, as to whether the "ultrasensitive"assay (dynamic range, 50 to 75,000 copies/ml) or the "standard"assay (dynamic range, 400 to 750,000 copies/ml) would be performed.Of the 191 samples, 47 were tested with the "standard" assayand 144 with the "ultrasensitive" assay. For the purposes ofthe analysis, the cutoff for the limit of detection was deemedto be 50 copies/ml for both the TaqMan and AMPLICOR assays.

Three samples initially giving results outside the dynamic rangeof the AMPLICOR ultrasensitive assay required dilution and retesting.These were diluted using the kit's negative control.

In routine clinical use, a difference in viral load of morethan 0.5 log₁₀ copy/ml is deemed clinically significant andthis cutoff has been used in the analysis for guidance of significance.

Results were transformed into log₁₀ copies/ml plasma beforethe statistical analyses, which involved the paired z test ofall results and a Student's t test of the clade results. Linearregression and correlation tests were also performed to judgethe relationship of the assays to one another.

RESULTS

Top

RESULTS

A total of 189 (98.95%) pairs of results gave differences fallingwithin 3 standard deviations (SD), and those within our clinicalcutoff of ±0.5 log₁₀ copy/ml numbered 178 (93.19%). Figure1 shows the correlation data for the two assays plotted as ascattergram. The correlation coefficient of the assays was 98.08%.The difference between the data sets was not significant (P= 0.73), as determined by the paired z test.

View larger version (21K):
[in this window]
[in a new window]

FIG. 1. HIV-1 RNA by TaqMan and AMPLICOR scatter plot.

Figure 2 demonstrates the log difference of the TaqMan assayfrom the AMPLICOR assay plotted against the AMPLICOR log result.The mean log difference was 0.05 log₁₀ copy/ml, and the SD was0.27 log₁₀ copy/ml, giving a 99% confidence interval at 3 SDof 0.81 log copy/ml. Fifteen samples were $≥$ 0.5 log differentfrom the mean, and 2 of these were $≥$ 3 SD different. Seven (3.67%)samples gave TaqMan results $≥$ 0.5 log copy/ml higher than theAMPLICOR, and 6 (3.14%) samples gave results $≥$ 0.5 log₁₀ copy/mllower. The mean AMPLICOR viral load of the seven samples $≥$ 0.5log₁₀ copy/ml higher than the AMPLICOR result was 2.44, andthat for six samples $≥$ 0.5 log₁₀ copy/ml lower than the AMPLICORresult was 3.70. One (0.52%) sample was >0.81 log₁₀ copy/mlhigher than the AMPLICOR result, having an AMPLICOR value of1.70 log₁₀ copies/ml, and one sample (0.52%) was >0.81 log₁₀copy/ml lower, with an AMPLICOR value of 3.39. The data showeda cluster of 12 samples which were detectable by TaqMan at 1.7to 2.5 log₁₀ copies/ml but were <1.7 log₁₀ copies/ml by AMPLICORassay. Four of the 13 samples differing by >0.5 log₁₀ copy/mlwere tested with the standard AMPLICOR assay and 9 with theultrasensitive assay.

View larger version (21K):
[in this window]
[in a new window]

FIG. 2. Difference plot of HIV-1 RNA by TaqMan and AMPLICOR assays.

Figure 3 demonstrates the frequency of difference between logvalues of the two assays.

View larger version (17K):
[in this window]
[in a new window]

FIG. 3. Difference frequencies in HIV-1 RNA by TaqMan and AMPLICOR assays.

Table 1 shows the previous AMPLICOR results and the number ofdays the sample was taken prior to the TaqMan-AMPLICOR pairs.Some inference may be drawn from previous values of three samples(05M008218, 05M008560, and 05M008665) due to the temporal proximityof the previous sample to the evaluation pair. In all threecases, the previous sample was closer to the TaqMan value andin two cases would make the difference fall within the clinicalcutoff.

View this table:
[in this window]
[in a new window]

TABLE 1. AMPLICOR results from previous samples

Twenty-four patients tested were of known genotype: 11 B cladesand 13 non-B clades. Four patients of known genotype variedby more than 0.5, and of these three were non-B clades. Themean of the clade B differences was 0.22 log₁₀ copy/ml withan SD of 0.23. The mean of the difference of non-B clades of0.37 log₁₀ copy/ml was less than the clinical cutoff; it wasstill higher than the clade B mean difference and had a higherSD of 0.46 log₁₀ copy/ml. The difference, however, was not statisticallysignificant as determined by one-tailed Student's t test (P= 0.15).

DISCUSSION

Top

DISCUSSION

The aim of this study was to evaluate the utility of the TaqManassay by comparing it with the AMPLICOR assay. Our results demonstratethat the assays compare favorably. The AMPLICOR assay is oneof the most widely used assays both for clinical monitoringand for clinical trial purposes, and it has been evaluated inmany different studies (2). While the technology used to producethe signal has changed from conventional PCR to real-time PCR,the principle of comparing the signal of the unknown viral loadin a sample to that of the introduced control of known copynumber remains the same. This minimizes interassay variationand interlot variation. Such an approach is essential for high-throughputlaboratories requiring tight quality control (6). It is possiblethat the results having the greatest variance may be due tothe manual manipulation stages of the AMPLICOR assay. Full automationshould further improve the reproducibility of these results.A recent study (7) compared the AMPLICOR assay with the semiautomatedTaqMan 48 assay; however, the extraction system used (High Purekit) is different from the automated AmpliPrep extraction systemused on the fully automated TaqMan96 analyzer.

Previous studies (5, 10, 11) have shown that clade is an importantfactor in the quantitation of HIV plasma viral load. Our claderesults, although limited in number, have shown the variationis within the clinical cutoff and well within statistical significancelimits, although non-clade B genotype virus strains gave morevariable results than those of clade B.

In conclusion, the assays correlate significantly (P > 0.95),and the regression analysis is also highly significant (R² >0.95). With respect to the practical considerations of a diagnosticvirology laboratory, we considered that the TaqMan assay wassuitable for direct substitution for the AMPLICOR assay. Dueto the increased sensitivity at the lower end of the dynamicrange, the advice to our requesting clinicians following theintroduction of the assay was that patients previously with<1.7 log₁₀ copies/ml may now have detectable low-level viralload. In addition to this, the lower limit of the linear rangefor the TaqMan assay is 1.6 log₁₀ copies/ml, which will addto the numbers of samples with quantifiable RNA being detectedif the manufacturer's reporting guidance is used. This willobviously entail close liaison with the requesting cliniciansto ensure that the significance of these results is fully explainedin the light of the change of assay.

ACKNOWLEDGMENTS

We gratefully acknowledge the staff in Specimen Reception andthose of the Molecular Virology Section at the Royal LondonHospital for their assistance with the management of the samples.We also thank Roche Diagnostics for supplying the TaqMan reagentsfor this comparison. Finally, we thank the staff in Infectionand Immunity at Barts and the London NHS Trust for assistingwith the procurement of the samples.

FOOTNOTES

* Corresponding author. Mailing address: Diagnostic Virology (Division of Infection), Barts and the London NHS Trust, Pathology & Pharmacy Building, 80 Newark St., London E1 2ES, United Kingdom. Phone: 44 203 246 0329. Fax: 44 203 246 0325. E-mail: tony.oliver{at}bartsandthelondon.nhs.uk

Published ahead of print on 5 September 2007.

REFERENCES

Top