ABSTRACT
The ReSSQ CMV assay is a novel commercially available kit for quantification of cytomegalovirus (CMV), based on real-time PCR with a peptide nucleic acid probe coupled with a single dye. In combination with the LightCycler, the ReSSQ CMV assay was evaluated with respect to specificity, PCR inhibition, linearity, reproducibility, and sensitivity. All nontested CMV materials were negative, and the assay was not inhibited by the use of different anticoagulants or other factors that may influence blood samples. The dynamic range was between 10 and 5 × 108 copies/PCR, and intra- and interassay variabilities were below 0.10 and 0.12 log10 standard deviations, respectively. Assay sensitivity was validated by analysis of 24 samples from a proficiency panel and by comparison to a nested in-house CMV PCR and the COBAS AMPLICOR CMV MONITOR test, using 159 clinical samples. Results from the proficiency panel were well in accordance with input values over the entire range of viral concentrations tested (50 to 31,250 copies/ml). The association between the ReSSQ CMV assay and the in-house PCR was in agreement in 90% of the clinical samples, and discordant results were found for all types of sample materials analyzed. The ReSSQ CMV and COBAS AMPLICOR assays showed no significant differences for samples containing >1,000 CMV copies/ml, but results differed to a greater extent at lower viral concentrations. The results demonstrate that the ReSSQ CMV assay is a CMV-specific, robust, and reproducible method and hence is well suited for routine use in clinical virology laboratories.
Cytomegalovirus (CMV) is one of the major pathogens in immunocompromised patients, such as organ or bone marrow transplant recipients and persons infected with human immunodeficiency virus type 1 (HIV-1). Early detection and close monitoring of high-risk patients is important for prevention of CMV disease. Today, qualitative methods for CMV detection are mostly used, however, quantitative methods are becoming the methods of choice, and their clinical utility in managing immunocompromised patients has been well documented (1, 2, 10, 11, 12, 13, 19). In transplant recipients, CMV load is used to diagnose active CMV disease, screen patients for the use of preemptive therapy, and monitor the response to antiviral therapy (1, 2, 11). For HIV-1 patients, the risk of developing CMV disease has been reported to be directly related to the quantity of CMV in the plasma (12, 13).
During recent years, a number of molecular assays for detecting and quantifying CMV DNA from clinical samples have become commercially available. The COBAS AMPLICOR CMV MONITOR test (Roche Diagnostics, Indianapolis, IN) is an automated system for PCR amplification, detection, and quantitation of CMV DNA from plasma. The linear range of the test is reported to be between 400 and 100,000 copies/ml of plasma by using a plasmid DNA standard. The Hybrid Capture System CMV DNA test (Digene Corp., Gaithersburg, Md.) and the NucliSens CMV pp67 assay (bioMérieux, Durham, N.C.) have also been shown to be useful for the diagnosis of CMV infection. In addition to these commercially available tests, in-house-developed DNA PCR methods are still widely used in clinical laboratories (8, 17). However, the results from such assays are often not reproducible between laboratories due to insufficient standardization.
The ReSSQ CMV assay (LightUp Technologies, Huddinge, Sweden) is a new commercial kit for qualitative and quantitative detection of CMV DNA in clinical samples. The assay is based on real-time PCR in combination with a LightUp probe, i.e., a peptide nucleic acid (PNA) oligomer coupled to a single reporter group (14). This approach offers some important advantages. PNA is more sequence specific than DNA and is not degraded by nucleases, peptidases, or proteases (3, 4). In addition, it binds more quickly and more strongly to complementary nucleic acids than traditional DNA probes (4, 6, 15). The single-dye system ensures that only one reaction needs to occur, making the reaction less prone to artifacts.
The assay can be used on several real-time PCR instrument platforms, e.g., LightCycler (Roche Diagnostics), RotorGene (Corbett Research), and ABI Prism 7000 SDS (Applied Biosystems)
The Department of Clinical Microbiology, Karolinska Hospital, carries out approximately 1,500 CMV DNA tests yearly. The majority of samples are whole-blood clinical specimens from either untreated blood (serum) or EDTA-treated blood (plasma), but samples of other origins, such as urine, bronchoalveolar lavage (BAL), fluid organ biopsies, bone marrow, breast milk, and cerebrospinal fluid (CSF), also constitute significant amounts. For the qualitative assessment of CMV DNA, a nested in-house CMV PCR is used (18), and positive plasma samples are quantified using the COBAS AMPLICOR assay. In general there is an increasing demand for quantitative CMV DNA analysis, particularly for specimens other than plasma. Because the AMPLICOR CMV test is limited to plasma samples, we have in the present study evaluated the ReSSQ CMV assay in combination with the LightCycler to determine whether it can replace both our in-house CMV PCR and the AMPLICOR CMV test for the qualitative and quantitative testing of CMV DNA in a broader range of clinical specimens.
MATERIALS AND METHODS
Clinical specimens and samples.The clinical specimens were received by the Department of Clinical Microbiology, Karolinska Hospital, for routine diagnosis of CMV infection. The total number of clinical specimens analyzed was 159: 88 plasma, 30 serum, 15 BAL, 16 urine, 5 breast milk, 2 CSF, 2 amniotic fluid, and 1 biopsy sample. Blood, BAL, CSF, and biopsy specimens were obtained in most cases from solid-organ transplant patients or patients with hematological disorders for monitoring of CMV or when CMV infection was suspected. Urine, breast milk, and amniotic fluid were collected from women and children to determine a congenital or neonatal CMV infection.
The samples were analyzed and evaluated blindly. CMV-negative serum and plasma were drawn from healthy volunteers and purified using a QIAamp DNA Blood Mini kit (QIAGEN) prior to utilization as the background matrix.
Artificial samples were prepared by spiking the purified CMV-negative plasma with the standardized commercially available control material CMV AD169 (Advanced Biotechnologies, Columbia, Md.). The accuracy and variability (coefficient of variation [CV]) of the ReSSQ assay were determined by repeatedly (three consecutive runs) by measuring four artificial samples containing 2,000 CMV copies/ml (10 replicates) or 4,000, 20,000, or 100,000 copies/ml (7 replicates each). The artificial samples were within the standard-curve range of 10 to 1 × 105 copies/PCR. To determine the variability below the standard-curve range, samples containing 200 copies/ml were analyzed (28 replicates). Interference test samples were whole-blood samples from healthy donors, collected in tubes containing EDTA, citrate, or heparin. Plasma was isolated and stored at −80°C. Interference was tested by spiking plasma with CMV genomic DNA extracted from a CMV AD169 supernatant (Swedish Institute of Infectious Diseases Control). The effects of bilirubin (Sigma-Aldrich, Stockholm, Sweden), intralipid (Fresenius-Kabi, Uppsala, Sweden), and hemoglobin on the ReSSQ CMV assay were evaluated using EDTA-plasma samples. Nonspiked samples served as negative controls. The quality control (QC) panel samples used were from the Quality Control for Molecular Diagnostics (QCMD) CMV DNA 2002 and 2003 panels, obtained from QCMD (Glasgow, United Kingdom). High concentrations of virions purified from human herpesvirus 6 (HHV-6) and HHV-8, varicella-zoster virus (VZV), herpes simplex virus type 1 (HSV-1) and HSV-2, and Epstein-Barr virus were obtained from the Swedish Institute for Infectious Diseases Control and were used to test the specificity of the ReSSQ CMV assay. CMV plasmid DNA used for determination of the linear range of the ReSSQ CMV assay was obtained from LightUp Technologies.
DNA extraction.DNAs from all different sample materials (plasma, serum, bronchoalveolar lavage fluid, fluid organ biopsy specimens, bone marrow, and breast milk) were extracted manually (200 μl original sample to 50 μl eluate) using the QIAamp DNA Blood Mini kit (QIAGEN, Germany) according to the manufacturer's recommendations. Cerebrospinal fluid was heated at 95°C for 10 min, and no pretreatment was used for urine samples. Isolation of CMV DNA from plasma for the COBAS AMPLICOR CMV MONITOR test was performed according to the manufacturer's instructions.
CMV DNA testing.Qualitative CMV PCR testing was performed by an in-house PCR in a nested protocol with primers from the CMV DNA polymerase gene (18). Viral load was assayed using the ReSSQ CMV assay and the COBAS AMPLICOR CMV MONITOR test. The COBAS AMPLICOR CMV MONITOR test was used according to the manufacturer's instructions for quantitative detection of CMV DNA in plasma (detection limit, 400 copies/ml).
The ReSSQ CMV assay was used according to the manufacturer's instructions. Briefly, a CMV master mix for each 20-μl reaction mixture was prepared by mixing 10.1 μl PCR-grade water, 4 μl 5× CMV master mix, 0.5 μl solution A, and 0.4 μl JumpStart Taq (Sigma; 2.5 U/μl). Standard curves were prepared by serial 10-fold dilutions of the supplied CMV control DNA in dilution buffer, yielding a standard curve spanning 10 to 1 × 105 copies/PCR. The sample volume added to the master mix was 5 μl. The real-time PCR experiments were carried out on a LightCycler (Roche Diagnostics). ViroQalc software, provided with the ReSSQ CMV assay, was used for viral load determinations and experimental-performance monitoring. The ReSSQ CMV assay is linear over at least an 8-log range. For clinical samples the lower limit of quantification is 500 copies/ml when the extraction is performed from 200 μl original sample to 50 μl eluate, and the lower limit of detection under the same conditions is 150 copies/ml.
Statistical analysis.Accuracy describes the closeness of the mean test results obtained by the method to the true value and is calculated as (mean measured value/theoretical value) × 100. Precision, or CV, describes the closeness of individual measurements of the analyte when the method is used repeatedly and is calculated as (measured standard deviation/measured average) × 100. The t test was used, and P values of <0.05 were considered significant. The CMV load values determined by the ReSSQ assay were correlated with those of the AMPLICOR assay by use of a scatter diagram incorporating data from samples with measurable CMV loads by both assays. Correlation analysis was calculated using the nonparametric Spearman rank order correlation test.
RESULTS
Evaluation of the ReSSQ CMV assay.The specificity of the ReSSQ CMV assay was evaluated using the CMV-related human herpesvirus species HHV-6, HHV-8, varicella-zoster virus, HSV-1, HSV-2, and Epstein-Barr virus or human genomic DNA as the template material. A CMV standard curve was generated in the same run, serving as a positive control. All non-CMV materials tested were negative (data not shown), demonstrating high specificity of the assay.
The interference of different anticoagulants in blood tubes and of increased levels of bilirubin, hemoglobin, and lipids in patient samples was evaluated. Plasma samples, prepared from EDTA-, citrate-, and heparin-blood from three donors each, were either analyzed directly or spiked with CMV DNA (approximately 4,000 copies/ml for donor 1 and 14 × 104 copies/ml for donors 2 and 3) before analysis in duplicate. All spiked samples (n = 9) were positive for CMV DNA, and there was no significant difference between matrices, as shown by identical threshold cycle (Ct) values for EDTA (27.9 ± 2.7) (n = 3), citrate (28.2 ± 2.8) (n = 3), and heparin (28.2 ± 2.8) (n = 3). Nonspiked samples were negative. To investigate the interference of bilirubin and intralipid, EDTA-plasma samples from three donors were pooled and either analyzed directly or spiked with CMV DNA (approximately 14 × 104 copies/ml) before analysis. Bilirubin (end concentrations, 0, 12, 24, and 38 μg/ml) or intralipid (end concentrations, 0, 6, 12, and 24 mg/ml) was added, and Ct values were compared to examine substance interference in the ReSSQ CMV assay. In addition, the interference of hemoglobin was tested by removing one aliquot of EDTA-blood before hemolysis, after which three additional aliquots were withdrawn at increasing degrees of hemolysis. Hemolyzed plasma samples were spiked and analyzed as described above. No significant difference in the Ct values of spiked samples with or without bilirubin or intralipid was found at any of the concentrations tested, and no significant difference between untreated controls and hemolyzed samples was found. The estimated Ct values for control, bilirubin-treated, intralipid-treated, and hemolyzed samples were 26.4 ± 0.3 (n = 3), 26.4 ± 0.3 (n = 3), 26.8 ± 0.2 (n = 3), and 26.1 ± 0.2 (n = 3), respectively. Nonspiked samples were negative.
The dynamic range of the assay was determined by establishing standard curves from serial dilutions of a plasmid containing the CMV target sequence. The samples were analyzed in quadruplicate in two separate experiments (Fig. 1). The ReSSQ CMV assay was found linear over at least 8 orders of magnitude (between 10 and 5 × 108 copies/PCR), since within this range a standard curve with a correlation coefficient (R2) of >0.99 was obtained.
Dynamic range of the ReSSQ CMV assay. Serial dilutions of CMV DNA were analyzed in quadruplicate, and the Ct was plotted against the log CMV DNA copy number. Values are from two separate experiments, represented by gray and black boxes and lines, respectively. The regression line was obtained by the least-squares method.
Intra-assay variability was evaluated using four artificial CMV-positive samples ranging from 2,000 to 100,000 CMV copies/ml. Each of these samples was tested in 7 to 10 replicates in the same PCR run. The intra-assay variability (CV) was less than 3% (minimum and maximum values, 0.3 and 2.7%), and the accuracy was estimated at 100 to 101% (Table 1). Interassay variability was determined by repeating the above experiments in three different PCR runs. Here, the variability was less than 4% (minimum and maximum values, 1.9 and 3.4%), with the highest variability at the lowest template concentration (Table 1). Imprecision below the lowest point of the standard curve was 22% as determined by limiting dilution assay.
Intra- and interassay variability of viral DNA load determined by the ReSSQ CMV assay
Proficiency panel results.The quality assessment program QCMD provides a means for standardization and quality control by yearly releasing a proficiency panel. The panel composition is designed to test for assay performance with respect to sample matrix, sample type, and accuracy of viral load determination. The samples from the 2002 and 2003 panels were extracted and quantified by the ReSSQ CMV assay. The results are shown in Table 2. Viral load estimations by the ReSSQ CMV assay are in good agreement with the input reference (QCMD) values, with log difference values from −0.5 to 0.6.
CMV DNA load in QCMD CMV 2002 and 2003 panels measured by the ReSSQ CMV assay
Comparison of the ReSSQ CMV assay with an in-house nested PCR.A total of 159 patient samples of different origins were analyzed in parallel by both the qualitative in-house nested PCR and the ReSSQ CMV assay (Table 3). For 93% of the plasma samples, 90% of the serum samples, 80% of the BAL samples, and 81% of the urine samples, there was agreement between the results of the in-house PCR and those of the ReSSQ CMV assay. A very limited number of samples of other origins were tested. In total, CMV DNA was concordantly detected in 88/159 (55%) of the samples, and 55/159 (36%) showed concordant negative results. The agreement between the two assays was thus 90%. Discrepant results (10%) included nine samples that were CMV DNA positive by the ReSSQ assay but repeatedly negative by the nested in-house PCR and seven samples positive by the in-house PCR but repeatedly negative by the ReSSQ CMV assay. Of the 16 samples showing discrepant results, 12 nonserum samples were further tested by CMV isolation. True-positive results by the ReSSQ CMV assay were confirmed by virus isolation in five out of nine samples (three urine, one plasma, and one BAL sample). Results from virus isolation were available only for three of the seven samples that were positive by in-house PCR and negative by the ReSSQ CMV assay. All three were negative by virus isolation. The discrepancies seen might be explained by samples with low copy numbers resulting in an uneven distribution of DNA copies in the clinical specimen.
Comparison between the ReSSQ CMV assay and an in-house CMV PCR with samples from different clinical specimens
Comparison of the ReSSQ CMV assay with the COBAS AMPLICOR CMV MONITOR test.The ReSSQ CMV assay was further compared with the quantitative COBAS AMPLICOR assay. Thus, 45 randomly chosen CMV DNA-positive plasma samples and 4 negative samples, as determined by the qualitative in-house PCR method, were selected for comparison. The viral load estimations from the two assays are shown in Fig. 2A. CMV DNA was quantified in 44 and 35 samples by the ReSSQ CMV and COBAS AMPLICOR assays, respectively. Thirty-four samples were positive in both assays, and 11 samples showed discordant results. The COBAS AMPLICOR assay cannot discriminate between positive and negative samples below a viral load of 400 copies/ml. This is also seen in our study (Fig. 2A). As expected, 9 of the 11 samples with fewer than 400 copies/ml by the COBAS AMPLICOR assay were determined to be positive and quantified as <400 copies/ml by the ReSSQ CMV assay, and the remaining 2 samples were quantified as <600 copies/ml. Sample 46 was positive by COBAS AMPLICOR but negative by the ReSSQ CMV assay.
(A) CMV DNA loads obtained by the COBAS AMPLICOR CMV assay (black) and the ReSSQ CMV assay (gray). Plasma samples (n = 49) were analyzed by the two methods. Results are presented in descending order according to the ReSSQ CMV assay. (B) Comparison of samples positive in both assays. Results were analyzed by least-squares regression (black line), and the linear range was examined by comparison to a 45° line (gray line). Correlation was calculated by the Spearman rank order correlation test (r).
Figure 2B shows the scatter diagram of all CMV loads measured by the COBAS AMPLICOR assay plotted against those measured by the ReSSQ CMV assay. The regression indicates good agreement between the two assays, most notably for CMV loads higher than 1,000 copies/ml. Below this level, values obtained by the COBAS AMPLICOR assay were higher than the corresponding results obtained by the ReSSQ CMV assay. The results obtained may equally well represent an overestimation by COBAS AMPLICOR as an underestimation by ReSSQ. It is not possible to draw a conclusion about a difference in sensitivity between the two assays from these data.
To further compare the two assays, the CMV DNA load in one allogeneic bone marrow transplant patient was monitored over a period of 175 days (Fig. 3). After 2 months of prophylactic Valtrex therapy with very low or undetectable CMV DNA levels, the viral load increased significantly at day 65 and continued to rise, as determined by both assays. The patient was therefore treated with Cymvene from day 74 but did not respond, and because resistance to ganciclovir was suspected, the patient was switched to Foscavir therapy on day 87. Ganciclovir resistance was indeed verified subsequently. The high CMV loads were successfully suppressed by the Foscavir therapy, but because the drug was not well tolerated by the patient, therapy was interrupted. Increasing levels of CMV were again detected at day 153 by both assays and suppressed after a second Foscavir treatment.
CMV DNA load change determined by the COBAS AMPLICOR CMV assay (black) and the ReSSQ CMV assay (gray) in an allogeneic bone marrow-transplant patient who had received Valtrex, Cymvene, and Foscavir therapy. Boxes indicate the time periods of antiviral therapy. For clarity, the time axis is not drawn to scale.
DISCUSSION
Quantitative CMV DNA tests are being increasingly used to diagnose or predict CMV disease in transplant recipients and in HIV-1-infected patients. For this purpose, many laboratories use various assays developed in-house, for instance, a qualitative PCR in combination with a commercial quantitative test (e.g., the AMPLICOR CMV MONITOR test or Hybrid Capture test) or a stand-alone quantitative PCR assay with poorly standardized calibrators and variable denominators for quantification. As a consequence, the use of such different tests complicates the interpretation of results and leads to studies that are not always comparable.
The ReSSQ CMV assay was carefully validated by testing specificity, PCR inhibition, dynamic range, and reproducibility. All non-CMV materials tested were negative in the assay, and the assay was not inhibited by the use of different anticoagulants (EDTA, citrate, and heparin) or other factors that may influence the whole-blood sample (bilirubin, intralipid, and increasing degrees of hemolysis). The dynamic range of the assay was linear over at least 8 orders of magnitude. Both the intra-assay and interassay variabilities were below 6%.
For quantitative analysis of HIV-1 and human hepatitis B virus, there are established standards against which all kits are calibrated, but there is no such “gold standard” for quantification of CMV. However, the QCMD panel is well accepted and is one of the most widely used CMV standards, with more than 100 participating laboratories in recent years (9). Quantification by the ReSSQ CMV assay of viral loads in the 2002 and 2003 QCMD panels was well in accordance with input values over the entire range of viral concentrations tested (Table 2). It is noteworthy that the ReSSQ CMV assay demonstrated good performance for the five samples with low CMV concentrations, in a range (250 to 400 copies/ml) well below the limit of the AMPLICOR CMV MONITOR assay. Moreover, for reference samples at concentrations below 100 copies/ml, the ReSSQ CMV assay still performed well and scored no false positives. It is important to bear in mind that at such very low concentrations, stochastic effects should be taken into consideration.
During recent years, a variety of real-time PCR methods for CMV quantification and/or detection have been published, as well as a number of studies comparing the performance of these methods against commercially available assays (5, 8, 16, 17). These comparative studies have mostly been limited in the range of starting materials, since manufacturers of the most widely used assays recommend the use of whole-blood or plasma specimens. The present study, however, used a variety of clinical specimens (i.e., plasma, serum, BAL fluid, urine, and breast milk) representing the composition of sample materials that are routinely analyzed at our laboratory. The association between the qualitative in-house PCR and the ReSSQ CMV assay was in agreement for 90% of the samples. Discordant results (n = 16 [10%]) were found for all types of sample materials analyzed, although with a higher relative frequency in the urine, BAL, and breast milk samples (Table 3). To further analyze the discordant results, CMV was isolated from 12 of the 16 samples. This revealed a better concordance between virus isolation and the ReSSQ CMV assay (67%) than between virus isolation and the in-house PCR (33%), most notably with respect to the urine samples. It is well known that urine contains factors inhibitory in PCRs (7), and the results indicate that the ReSSQ CMV assay is less sensitive to inhibitory factors. Although plasma and serum are the sample materials recommended for use in the ReSSQ CMV assay, this study shows that other sample materials may well be used for CMV DNA quantification with this assay.
The ReSSQ CMV assay and the COBAS AMPLICOR assay are both technically straightforward quantitative methods based on PCR. The ReSSQ CMV assay is a homogeneous real-time PCR method performed in a closed-tube format, while the COBAS AMPLICOR assay is a heterogeneous PCR method requiring separation of bound and unbound probes prior to signal detection. Another difference lies in the sample volume requirement, which is 50 μl for COBAS AMPLICOR and 5 μl for the ReSSQ CMV assay. It should also be noted that at the time of this study, the lower quantification limit of the COBAS AMPLICOR assay was 400 copies/ml, but recently the manufacturer (Roche) has increased that level to 600 copies/ml. Below this limit the method does not discriminate between positive and negative samples. The ReSSQ CMV assay, on the other hand, has a lower quantification limit of 500 copies/ml if the extraction is performed from 200 μl original sample to 50 μl eluate. Below this limit, the viral load is still reported, although with a notification of increased error (approximately 20%) in the reported result. In addition, the lower limit of detection (where 95% of the reactions will be determined to be positive) for the ReSSQ CMV assay is 150 copies/ml under the same conditions. In this study, the ReSSQ CMV and COBAS AMPLICOR assays gave comparable results, and the measured values showed small differences for samples containing >1,000 CMV copies/ml. When the assays were compared at lower viral concentrations, the measured values differed substantially, with the COBAS AMPLICOR assay consistently estimating higher values (Fig. 2). This discrepancy, however, is only apparent and is due to the higher detection limit of the COBAS AMPLICOR. Taken together, the results suggest that the ReSSQ assay has a higher analytical sensitivity.
The clinical importance of monitoring CMV infection during therapy has been demonstrated for a number of major patient groups. In the present study the ReSSQ and COBAS AMPLICOR CMV assays were utilized in parallel to monitor a symptomatic CMV-infected allogeneic bone marrow transplant patient (Fig. 3). In this case also, the two methods showed good agreement. This is the first example of use of the ReSSQ CMV assay in a clinical setting, and more studies will be needed to define the clinical cutoff value for the assay.
From a routine diagnostic laboratory perspective, it is advantageous to use a method that functions with a wide variety of sample materials and provides both qualitative and quantitative CMV DNA determination in one assay. The present study demonstrates that the ReSSQ CMV assay is a CMV-specific, robust, and reproducible method and therefore well suited for routine use in clinical virology laboratories and that the ReSSQ CMV assay, in combination with the LightCycler, can replace in-house PCR and the COBAS AMPLICOR CMV MONITOR test for this purpose.
FOOTNOTES
- Received 23 June 2004.
- Returned for modification 7 November 2004.
- Accepted 25 April 2005.
- Copyright © 2005 American Society for Microbiology
