|Home | About | Journals | Submit | Contact Us | Français|
Hepatitis C virus (HCV) RNA viral load (VL) monitoring is a well-established diagnostic tool for the management of chronic hepatitis C patients. HCV RNA VL results are used to make treatment decisions with the goal of therapy to achieve an undetectable VL result. Therefore, a sensitive assay with high specificity in detecting and accurately quantifying HCV RNA across genotypes is critical. Additionally, a lower sample volume requirement is desirable for the laboratory and the patient. This study evaluated the performance characteristics of a second-generation real-time PCR assay, the Cobas AmpliPrep/Cobas TaqMan HCV quantitative test, version 2.0 (CAP/CTM HCV test, v2.0), designed with a novel dual-probe approach and an optimized automated extraction and amplification procedure. The new assay demonstrated a limit of detection and lower limit of quantification of 15 IU/ml across all HCV genotypes and was linear from 15 to 100,000,000 IU/ml with high accuracy (<0.2-log10 difference) and precision (standard deviation of 0.04 to 0.22 log10). A specificity of 100% was demonstrated with 600 HCV-seronegative specimens without cross-reactivity or interference. Correlation to the Cobas AmpliPrep/Cobas TaqMan HCV test (version 1) was good (n = 412 genotype 1 to 6 samples, R2 = 0.88; R2 = 0.94 without 105 genotype 4 samples). Paired plasma and serum samples showed similar performance (n = 25, R2 = 0.99). The sample input volume was reduced from 1 to 0.65 ml in the second version. The CAP/CTM HCV test, v2.0, demonstrated excellent performance and sensitivity across all HCV genotypes with a smaller sample volume. The new HCV RNA VL assay has performance characteristics that make it suitable for use with currently available direct-acting antiviral agents.
The primary goal of therapy for chronic hepatitis C (CHC) is to eradicate the virus by achieving a sustained virologic response (SVR). Hepatitis C (HCV) RNA viral load (VL) results and viral kinetics are used for clinical decisions to shorten, extend, or stop therapy, also referred to as response-guided therapy (1–3).
Achievement of an SVR in patients treated with pegylated interferon plus ribavirin has been defined as the attainment of an HCV RNA level below the lower limit of detection (LOD; <50 IU/ml) 24 weeks after the end of treatment (EOT) using first-generation VL assays (4). In the recent phase III clinical trials of boceprevir and telaprevir, a new treatment response standard was established for genotype 1 patients, requiring HCV RNA “target not detected” for the assessment of early VL kinetics and the EOT response (5–10). Hence, for patients receiving triple-therapy regimens, an early “target not detected” response is an indication of a shorter period of treatment eligibility for treatment-naïve patients (weeks 4 and 12 of therapy with telaprevir and weeks 8 and 24 of therapy with boceprevir). SVR, however, was defined as an HCV RNA level of <25 IU/ml at 24 weeks after the EOT. Given the importance of these VL results at the new clinical decision points, accurate and reproducible results at the low end of an HCV RNA VL assay are critical (11).
Methods for the quantitation of HCV RNA levels have historically been based on target endpoint PCR or signal amplification of conserved regions of the HCV genome and detection with fluorescently labeled probes (12, 13). More recently, real-time PCR-based assays have been introduced, including two different assays from Roche Molecular Systems (Pleasanton, CA), a first version of a fully automated system named the Cobas AmpliPrep/Cobas TaqMan (CAP/CTM) HCV test and the Cobas TaqMan HCV test, v2.0, for use with the High Pure system for manual sample preparation, and the Abbott RealTime HCV test (Abbott Molecular, Des Plaines, IL). Compared to the previous PCR assays, these have greater sensitivity and broader dynamic ranges (14–16). For example, the Abbott RealTime HCV test (LOD of 12 IU/ml, range of 12 to 1 × 108 IU/ml) uses a single probe to detect its target (the 5′ untranslated region [UTR] of the HCV genome) and a noncompetitive internal control from a pumpkin gene (17).
The CAP/CTM HCV test uses magnetic silica bead-based nucleic acid extraction by Cobas AmpliPrep, followed by amplification with primers specific to the 5′ UTR of the HCV genome and detection with a fluorescently labeled hydrolysis probe performed with the Cobas TaqMan thermal cycler to detect the target and a quantitative standard (QS). A different assay using manual extraction, the Cobas TaqMan HCV test for use with the High Pure System, v2.0 (HPS/CTM HCV test, v2.0), was used in all phase III clinical trials for the recently approved direct-acting antiviral (DAA) agents boceprevir and telaprevir (5, 8, 10, 17).
The CAP/CTM HCV test, v2.0, was developed as a second-generation assay by using a novel dual-probe approach and an additional reverse primer to improve genotype 4 inclusivity and performance with known samples found to be difficult to quantify (18–20). In addition, an improved sample preparation method included changes in the elution and lysis buffers, as well as a higher temperature for the reverse transcriptase step and shorter overall PCR cycles. The master mixture composition was optimized by increasing the ZO5 DNA polymerase (Thermus species) concentration used in the first-generation assay, as well as by adding a new polymerase, ZO5D (a D580G mutant form of ZO5; Roche Molecular Systems, Inc., Pleasanton CA). Additionally, as part of the extraction, magnetic glass particles are removed from the eluted nucleic acid. Finally, a reduced sample input volume (650-μl sample requirement with 500 μl processed by the instrument) in comparison to the prior version (1-ml sample input with 850 μl processed) was incorporated into the assay design.
In this study, we evaluated the performance characteristics of the CAP/CTM HCV test, v2.0, and also compared them to those of the CAP/CTM HCV test (version 1) with patient samples across HCV genotypes.
The CAP/CTM HCV test and the CAP/CTM HCV test, v2.0, were performed according to the manufacturer's instructions. Both assays are standardized to the World Health Organization international standard (WHO-IS) for HCV RNA (2nd WHO standard NIBSC code 96/798), and results are reported in international units per milliliter. Platform correlation, matrix correlation, and specificity analysis were performed at the Johann Wolfgang Goethe University Hospital, Frankfurt, Germany. All other studies were performed at Roche Diagnostics, Ltd., in Rotkreuz, Switzerland. The QS invalid rate was calculated (percentage of the total runs) to measure assay reliability.
The LOD was determined by diluting the 3rd WHO-IS (NIBSC code 06/100, genotype 1a) in HCV RNA-negative single-donor EDTA plasma and serum matrices using CLSI guidelines (21). A seven-member panel was tested across three different reagent lots for each matrix with at least 250 replicates per level. The LOD for each genotype (1a, 1b, 2a, 2b, 3, 4, 5, or 6) was confirmed by using dilutions of individual clinical HCV specimens and diluted in either HCV-negative (by HCV antibody and HCV RNA) single-donor EDTA plasma or serum matrices. For each genotype, at least 60 replicates per concentration and matrix were tested. Sensitivity was determined for each panel by either calculating the LOD by using PROBIT analysis (95% confidence interval [CI]) or calculating the detection rate (≥95%, hit rate analysis).
To determine the linear range, two overlapping dilution panels were prepared by using an armored RNA stock solution (Roche, Branchburg, NJ) including high-titer panel members and a panel derived from a diluted HCV RNA-positive clinical sample (genotype 1). Both panels were tested in either an EDTA single-donor plasma matrix or a single-donor serum matrix. Each panel member was tested in 12 or 16 replicates by using two different reagent lots. To verify the linearity of individual HCV genotypes (1a, 1b, 2a, 2b, 3, 4, 5, and 6), genotype-specific linearity panels were prepared with EDTA plasma and subjected to the CAP/CTM HCV test, v2.0. Both studies were performed according to CLSI guidelines (22, 23). The HCV RNA titer was assigned by the calibrator bracketing method, and the original concentration was determined by a secondary standard traceable to the 2nd WHO-IS for HCV using the CAP/CTM HCV test, v2.0. Genotypes were verified by direct DNA sequencing of the 5′UTR HCV genome region by using the ABI BigDye Terminator v3.1 Cycle Sequencing kit on an Applied Biosystems 3730 DNA analyzer and comparison with reference sequences (NCBI and CLC Genomics Workbench).
Accuracy was assessed across the linear dynamic range in both plasma and serum samples by using a panel composed of high-titer samples (armored HCV RNA), the LOD panel for the 3rd WHO HCV RNA international standard, and a genotype 1 clinical sample. The linearity panel results were reported as average log10 (IU/ml) differences from a nominal value (22). The precision of the CAP/CTM HCV test, v2.0, was determined in accordance with CLSI guidelines (24) by analysis of serial dilutions of an HCV-positive clinical specimen (genotype 1) or armored HCV RNA in EDTA plasma or in serum. Six dilution levels were tested in three replicates in 12 runs on 4 days using three different lots of reagents.
For specificity, 300 matched HCV RNA-negative and seronegative EDTA plasma and serum samples from individual blood donors were tested with the CAP/CTM HCV test, v2.0, using two different kit lots. Analytical specificity was assessed by testing 27 different pathogens closely related to HCV or frequently encountered in HCV-infected patients (see Table S1 in the supplemental material). Each sample was tested in triplicate at approximately 1.0E+06 particles/ml, 1.0E+06 PFU/ml, or 1.0E+06 copies/ml independently in either HCV-negative (antibody and RNA) or HCV RNA-positive EDTA plasma specimens (SeraCare Life Sciences, Milford, MA) spiked at an HCV RNA concentration of 45 IU/ml.
To test for PCR interference, elevated levels of triglycerides (3,300 mg/dl), conjugated bilirubin (25 mg/dl), unconjugated bilirubin (20 mg/dl), albumin (6,000 mg/dl), hemoglobin (200 mg/dl), and human DNA (40 mg/dl), as well as autoimmune disease markers (systemic lupus erythematosus, rheumatic factor, and antinuclear antibody), were tested with the CAP/CTM HCV test, v2.0, in HCV RNA-negative individual donors with and without EDTA plasma specimens spiked with HCV RNA (Seracare Life Sciences, Milford, MA) at a concentration of 45 IU/ml. Interference was also investigated in 29 therapeutic agents (see Table S2 in the supplemental material) commonly used for the treatment of viral diseases at physiological concentrations (3 times the peak level in plasma). Specificity studies were performed according to CLSI guidelines (25).
Patient samples were acquired as part of routine testing (using either bDNA [Versant HCV RNA 3.0 assay on a bDNA System 340; Siemens Healthcare Diagnostics, Deerfield, IL] or the CAP/CTM HCV test) and genotyped using the Versant HCV Genotype 1.0 or 2.0 assay (Innogenetics, Ghent, Belgium). The residual sample volume (diluted with negative plasma or serum) from 463 HCV RNA-positive patient samples (244 EDTA plasma and 219 serum samples collected between 2007 and 2011) was tested in parallel with both the CAP/CTM HCV test and the CAP/CTM HCV test, v2.0. Collected samples were tested according to CLSI guidelines (26) in duplicate for each assay in parallel, and results were compared using Deming regression analysis.
Serum and plasma matrices from 25 HCV RNA-negative samples (confirmed by Bayer Versant HCV RNA qualitative assay) and 25 HCV RNA-positive matched samples collected with S-Monovette blood collection tubes (Sarstedt), EDTA Lavender Top blood collection tubes, and BD Vacutainer plasma preparation tubes (PPT).
While the majority of the 25 HCV-positive samples were of genotype 1 (n = 18), genotype 2 (n = 3), genotype 3 (n = 2), and genotype 4 (n = 2) were also represented. To achieve a distribution of HCV concentrations across the linear range of the CAP/CTM HCV test, v2.0, 10 samples with an HCV concentration of approximately 100,000 IU/ml were diluted 1:10, 1:100, 1:1,000, and 1:10,000 with the appropriate matrix (HCV RNA-negative serum or EDTA plasma).
By PROBIT analysis, the LOD determined with the 3rd WHO HCV RNA international standard was determined to be 11 IU/ml (95% CI, 10 to 13 IU/ml) in EDTA plasma and 12 IU/ml (95% CI, 10 to 14 IU/ml) in serum (Table 1). The LOD for both matrices by >95% detection rate using hit rate analysis was 15 IU/ml for all genotypes (Table 2). Hit rate concordance between matrices was observed across genotypes/subtypes. The hit rate for genotype 2b was higher in serum than in plasma at 5 IU/ml (92 versus 67%, respectively), but the hit rates were similar at 15 and 45 IU/ml. Thus, the difference between serum and plasma at 5 IU/ml was most likely due to chance as a systematic error, for example, due to inhibition and would also have been observed at other concentrations.
The CAP/CTM HCV test, v2.0, had a lower limit of quantitation (LLoQ) of at least 15 IU/ml to an upper limit of quantitation of at least 1.0E+08 IU/ml (linear range) by analysis of different HCV genotypes and subtypes and using an acceptable absolute deviation from linearity of ±0.2 log10 (Fig. 1; see Table S3 in the supplemental material). The assay's LLoQ was calculated to be 15 IU/ml based on an absolute deviation from linearity at this level using the polynomial 3rd-order regression line and determination of the allowable maximum differences. The linearity range tested was confirmed for each genotype tested and had an acceptable absolute deviation from linearity within ±0.2 log10 (data not shown; see Table S3 in the supplemental material).
Across the linear range, the accuracy of the test was within ±0.2 log10 for both matrices (titer range, −0.20 to +0.10 log10 IU/ml) (Fig. 2). The CAP/CTM HCV test, v2.0, showed good precision (SD, 0.04 to 0.22 log10 IU/ml) in three different kit lots across an HCV RNA concentration range of 3.0E+02 to 1.0E+08 IU/ml (Table 3).
None of the 300 matched HCV-negative plasma and serum samples gave a false-positive HCV RNA result with the CAP/CTM HCV test, v2.0, achieving a specificity of 100% (one-sided lower 95% confidence limit, ≥99.5%). The 27 microorganisms tested for analytical specificity did not interfere with the performance of the test (see Table S1 in the supplemental material). The mean log10 titers of HCV-positive specimens spiked at a concentration of 45 IU/ml and containing the potentially cross-reacting microorganism were within ±0.3 log10 of the positive spike control.
Clinically relevant substances known to potentially interfere in assay performance were tested. None of the pathogens tested interfered with the accurate quantitation of HCV-positive samples (at 45 IU/ml). Additionally, none of the 29 commonly used therapeutic agents showed any effect on HCV-negative or -positive samples subjected to the CAP/CTM HCV test, v2.0, at physiologically relevant concentrations (for detailed results, see Table S2 in the supplemental material).
Valid results were obtained with 204 EDTA plasma specimens and 208 serum samples with HCV RNA VL titers within the linear range of both tests. Deming regression analysis of the 412 paired results showed good comparability between the CAP/CTM HCV test and the CAP/CTM HCV test, v2.0, with an R2 value of 0.88, including all genotypes (Fig. 3a and andb).b). When genotype 4 specimens (n = 105) were removed from the analysis (data not shown), an R2 value of 0.94 was determined. When genotype 4 was analyzed separately (Fig. 3d), an absolute difference in the mean VL titer of 0.53 log10 (the CAP/CTM HCV test result being lower) in these samples (range, −0.19 to 1.49) was observed.
The 25 serum and EDTA plasma samples that were previously found to contain no detectable HCV RNA were all negative by the CAP/CTM HCV test, v2.0. For the 25 HCV-positive specimens, an R2 value of 0.99 was achieved (Deming regression analysis, data not shown). When the mean titer of the plasma specimens was compared to the titer obtained with the matched serum samples, an absolute difference in the mean VL of −0.1 log10 was obtained by Bland-Altman analysis (Fig. 4).
The QS internal control failure rate was calculated across five reagent lots of the CAP/CTM HCV test, v2.0, during 17 verification studies conducted at two testing sites. The QS failure rate observed was 0.09% (14 of 15,402 samples tested).
In this study, the CAP/CTM HCV test, v2.0, was shown to be sensitive (LOD equal to LLoQ at 15 IU/ml), accurate across a broad linear range (15 to 1.0E+08 IU/ml), and precise at clinically relevant concentrations (±0.20 log10 IU/ml) for HCV genotypes 1 to 6. The previous CE-IVD version, the CAP/CTM HCV test, had an LOD of 15 and an LLoQ of 43 IU/ml. A result between the LOD and the LLoQ was reported as >15 and <43 IU/ml HCV RNA detected. The new assay reports HCV RNA levels of 1 to 14 IU/ml as “HCV RNA detected, below the LLoQ.” This should simplify the results obtained with this new assay and eliminate confusion by providing a single value for both the LOD and LLoQ, making it easier for laboratories and clinicians to interpret the results of DAA treatments (27, 28). It is important to note that, regardless of whether the LOD is equal to or below the LLoQ, HCV RNA can be detected below this level. For example, at 5 IU/ml, HCV RNA is still detectable approximately 70% of the time if the same sample is measured several times (Table 1). Furthermore, any result that is not detected (e.g., “0 IU/ml”) will be reported as “target not detected,” a result that is distinct from “HCV RNA detected, <15 IU/ml” (6, 7, 11).
The CAP/CTM HCV test, v2.0, meets the recommendations published in the latest clinical practice guidelines for monitoring and predicting patients' responses to the most advanced antiviral combination therapies (1, 2, 4, 7). The broad dynamic range of the new assay is appropriate in the context of VL titers usually found during the course of CHC.
HCV RNA VL is a central component in the management of CHC by monitoring viral replication kinetics and guiding antiviral therapy (1, 2, 4, 7). While the CAP/CTM HCV test has been successfully used in the clinical monitoring of antiviral therapy for HCV (28), the CAP/CTM HCV test, v2.0, was developed to provide improved accuracy and specificity with a smaller sample input (650 μl versus 1 ml sample input) with improved sensitivity. The design incorporated two probes and an additional primer, as well as optimized reagents to ensure accurate quantification of HCV RNA across different genotypes, particularly those not quantified well by the previous version of the assay.
All 600 samples expected to be HCV RNA negative were reported as “target not detected.” Additionally, when the test was challenged with 27 different pathogens and 29 therapeutic agents and concomitant conditions, no interference was detected. This feature is particularly important, given that triple therapy regimens require a “target not detected” result for shortened treatment. Therefore, a “false positive” result due to nonspecific amplification would have significant implications for patient management. Additionally, prevention of carryover contamination is another important feature to reduce “false positive” results. The TaqMan-based assays contain AmpErase to reduce the risk of carryover contamination.
The CAP/CTM HCV test, v2.0, also showed a high overall correlation with the CAP/CTM HCV test when assessed using a large collection of HCV-positive clinical specimens. Accurate and precise HCV RNA VL measurements near the low end of the linear range and at clinically relevant titers are imperative for making key medical decisions when treating patients with CHC (30). That the analytical precision was found to be <0.2 log10 IU/ml across the new assays' linear range and at low HCV-RNA concentrations suggests that HCV RNA VL results are highly reproducible. Furthermore, the improvement in more accurate genotype 4 quantification with different PCR master mixtures during the development of the CAP/CTM HCV test, v2.0, has been previously published (29).
In conclusion, the data reported here show that the CAP/CTM HCV test, v2.0, is highly sensitive, accurate, and specific, a useful clinical tool for use in the era of HCV DAA agents.
We thank Thomas Liem, Dagmar Brunner, Sybille Ruettimann, Daniela Eckert, Katja Burchard, Sabine Zielenski, Regine Erismann, Robert Pretsch, Sergio Carrillo-Wilisch, Diana Thamke, Frank Bergmann, Karen Young, Caterina Füller, Dany Perner, Claudia Mergel, and Giuseppe Colucci for their contributions to the assay design; study protocols, design, testing, and analyses; and review of the manuscript.
Ethics approval was obtained from the Johann Wolfgang Goethe University Hospital, Frankfurt, Germany.
These studies were supported and funded by Roche Molecular Diagnostics Ltd., Rotkreuz, Switzerland. Bryan Cobb, Heike Zitzer, Gabrielle Heilek, Karine Truchon, and Dorothea Sizmann are Roche employees. C. Sarrazin has received research grants from Roche Molecular Diagnostics Ltd., Switzerland, and Abbott Diagnostics, Wiesbaden, Germany. We have no competing interests.
Published ahead of print 12 December 2012
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JCM.01784-12.