Comparative evaluation of the QIAsymphony RGQ system with the easyMAG/R-gene combination for the quantitation of cytomegalovirus DNA load in whole blood

One hundred whole blood specimens were included in the study. Samples were obtained from immunocompromised patients hospitalised at the University Hospital of Saint-Etienne, France, for the following clinical pictures scenarios: bone-marrow graft, kidney transplantation or chronic HIV infection. For all of them, the CMV DNA load was routinely evaluated as part of standard patient care. The study was conducted on the residual clinical specimens stored at −20°C. The research was approved by the Ethics Committee of the University Hospital of Saint-Etienne on the 30^th of May 2012. For the study, the samples were thawed, divided into 2 aliquots and tested by both systems on the same day. Specimens were then stored at 4°C until validated results of quantitation with both amplification methods (<24h).

The 2010 QCMD (Quality Control for Molecular Diagnostics) program (Glasgow, UK) included 1 negative and 9 positive (AD169 strain) specimens with viral loads ranging from 230 to 2,552,701 copies/ml, diluted either in CMV-negative human plasma or in virus transport medium. The specimens were reconstituted, separated in aliquot fractions and stored at −80°C as recommended by the QCMD instructions. After thawing, they were tested in the same way as clinical samples with both extraction and amplification systems.

All the experiments were conducted with in vitro diagnosis (IVD) certified kits according to the manufacturers’ instructions.

Just before extraction, 300 μl volumes of the whole blood samples were manually transferred into 2-ml vials. The tubes were placed into 24-tube capacity carrier racks and loaded into the SP module. The extraction step was performed using the QIAsymphony DNA Mini Kit and the Virus Blood 200 protocol with automated eluate dispensing. The internal control of the artus CMV QS-RGQ kit was added into ATE buffer and automatically distributed during the lysis of samples. With this protocol, 200 μl volumes of whole blood were extracted and eluted into a total volume of 90 μl, allowing a minimum accessible volume of 60 μl. The rack containing the elution tubes was then automatically transferred to the AS module. A 30 μl volume of the PCR mix prepared by the AS module using the artus CMV QS-RGQ kit was automatically distributed into RG 72 strip tubes (all reagent and eluate positions are continually cooled) and after distribution of 20 μl of template by the AS module, the RG 72 strip tubes were closed and placed manually into the 72-well rotor of the Rotor-Gene Q. The amplification step was performed according to the manufacturer’s instructions. Each PCR run included a set of quantitative calibrators corresponding to 10, 100, 1000 and 10000 copies/μl of template; the CMV DNA load was calculated from the standard curve and expressed as the number of CMV DNA copies/ml of whole blood (after calculating the input and output volume). The presence of PCR inhibition was detected when the CT value for the internal control of the sample was more than 3 cycles higher than the CT value for the internal control of a negative whole blood control.

Just before extraction, 200-μl volumes of the whole blood samples were manually transferred into 4-ml cryotubes containing 2 ml of lysis buffer. The whole extraction process was performed on the NUCLISENS easyMAG instrument according to the previously validated protocol [9]. With this protocol, 200 μl of whole blood was extracted and eluted into a total volume of 50 μl. The eluates were manually transferred into 2ml-tubes. Fifteen micro liters of ready-to-use PCR mix of the CMV R-gene kit and 10 μl of template were manually distributed into a plate stored in a pre-cooled rack. The amplification step was performed using an ABI7500 instrument (Applied Biosystems) according to Argene’s instructions. Each PCR run included a set of quantitative calibrators corresponding to 5, 50, 500 and 5000 copies/μl of template; the CMV DNA load was calculated from the standard curve and expressed as the number of CMV DNA copies/ml of whole blood. The presence of PCR inhibition was detected when the CT value for the internal control of the sample was more than 3 cycles higher than the CT value for the internal control of a negative whole blood control.

The analytical performances of the QIAGEN workflow were studied with frozen pooled blood samples from patients of our hospital with known CMV DNA load values. The limit of detection of the test was assessed using seven 0.5-log₁₀ dilution series of CMV DNA positive whole blood in CMV DNA negative blood to yield concentrations ranging from 14 to 10000 CMV DNA copies/ml. Each measure was tested on 5 replicate aliquots.

The reproducibility of the QIAGEN system was determined in comparison with the easyMAG/Argene technique using 4 pools prepared from positive specimens and exhibiting different DNA loads: one low (< 3 log₁₀ copies/ml), two intermediate (between 3 and 4.5 log₁₀ copies/ml) and one high (> 4.5 log₁₀ copies/ml). For each concentration, 8 frozen aliquots were tested, 4 for intra-run reproducibility assay and 4 for inter-run reproducibility assay (performed each on a different day).

The correlation analysis between the two systems was performed using a set of paired aliquots of whole blood specimens. The extraction and amplification steps were performed the same day for the same sample with both techniques.

The lower limit of detection of the QIAGEN assay was determined by probit analysis [10]. The agreement between methods’ results was tested by the kappa coefficient. Bivariate correlation analysis was performed with the Pearson r coefficient and a two-tailed test of significance. Differences in mean values were analysed by the paired t test. P values of 0.05 were considered as the threshold of significance.

The limit of detection of the QIAGEN workflow, calculated by probit analysis, was 72 copies/ml, with a confidence of 95% (Figure 1). By comparison to the easyMAG/Argene system, the inter-run and intra-run variability was consistently lower with the QIAGEN platform, notably for the low and intermediate viral loads (Table 1).

Ten samples belonging to the 2010 QCMD panel were tested in duplicate with the two systems. The negative sample was correctly assigned by both techniques. The results for the 9 positive samples are illustrated in Figure 2: the two systems exhibited the same range of quantitation, despite the fact that the viral load was consistently higher with the easyMAG/Argene method.

From the 100 clinical samples tested comparatively with both systems, 3 needed to be tested again for an invalid internal control value, 1 for easyMAG/Argene and 2 for QIAGEN and could be validated after retest.

Of the 100 pairs of clinical samples, 62 were positive with both methods and 24 were negative, which corresponds to a substantial agreement of 86% (kappa coefficient of 0.67). Fourteen samples exhibited discrepant results: 11 were tested positive with the QIAGEN platform and negative with the easyMAG/Argene method and 3 samples exhibited the opposite result; all of these discrepant results corresponded to viral loads < 2.5 log copies/ml.

The 62 samples found positive with both tests were used to draw the correlation curve (Figure 3A) and the Bland-Altman plot (Figure 3B). The Pearson r coefficient was 0.70 (P < 0.01) and the mean difference in viral loads was 0.25 log₁₀ copies/ml to the benefit of the easyMAG/Argene strategy (P < 0.001). Samples corresponding to low CMV DNA loads exhibited a high variability. Among those with a CMV DNA load above 1000 copies/ml, 6 exhibited discrepancy of at least 0.5 log copies/ml: 5 with CMV load higher by the easyMAG/Argene method, one of them being below the mean of the difference – 2 SD in Figure 3B, and one with CMV load higher by the QIAGEN platform.

The time necessary to complete a run was evaluated to 3.5 and 3.7 hours for the easyMAG/Argene strategy and the QIAGEN system, respectively. However, the hands-on time was approximately of 0.3 hour for the QIAGEN system versus 1 hour for the easyMAG/Argene strategy.