INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Cytomegalovirus (CMV) is an important cause of morbidity and mortality in human immunodeficiency virus (HIV)-infected subjects with low CD4 T-cell counts (8, 11). In addition to the level of cellular immunosuppression, the presence of CMV viremia appears to be an important risk factor for the development of CMV disease in that population (12, 20).

Recent studies have shown a good correlation between high viral loads in the leukocytes of AIDS patients (as determined by sensitive methods such as an antigenemia assay [15] and quantitative PCR [Q-PCR] assays [4, 19]) and the development of symptomatic CMV infections. Alternatively, the detection of CMV DNA in plasma has been shown to be a good diagnostic and possibly predictive marker for the occurrence of CMV disease in some studies (16, 22, 23). The assay to detect CMV DNA in plasma offers many technical advantages over cell-based assays, including a convenient specimen type and simpler qualitative PCR testing.

However, most PCR protocols in use to detect CMV are still time-consuming and require a fair amount of expertise. In that context, we evaluated a new commercially available assay, the AMPLICOR CMV test (Roche Diagnostic Systems, Inc., Branchburg, N.J.), which allows simple batch testing of a large number of plasma samples. Results obtained by this test were compared to results of other conventional CMV assays as well as to those of in-house Q-PCR-based methods by using plasma and leukocytes from HIV-infected subjects with and without CMV disease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Subjects and samples. HIV- and CMV-seropositive subjects with CD4 T-cell counts below 250 per mm³ were enrolled in this study. The vast majority of these individuals were not receiving HIV protease inhibitors at that time. Subjects were divided into two groups based on the presence or absence of CMV disease as previously defined (21). Briefly, the diagnosis of CMV retinitis required compatible ophthalmologic lesions whereas the diagnosis of visceral CMV disease required the presence of typical intranuclear inclusions in tissues or bronchoalveolar lavage cells. EDTA-treated blood samples were collected before antiviral therapy was initiated, and they were processed within 6 h. The blood was first centrifuged for the recovery of plasma and cells. Plasma was filtered through a 0.45-µm-pore-size filter (Corning Costar, Cambridge, Mass.) and kept at 70°C until extraction. The cells were separated on a 6% dextran gradient, and polymorphonuclear leukocytes (PMNL) were counted and used in nonmolecular and molecular assays.

Nonmolecular CMV assays. An aliquot of 10⁶ PMNL was inoculated onto human foreskin fibroblasts in a 24-well microplate for conventional blood culture. The cells were observed for 28 days for the appearance of typical CMV-induced cytopathic effect. A second aliquot of 10⁶ PMNL was centrifuged in a shell vial and stained after 48 h with a monoclonal antibody (CMV early nuclear protein; Dupont, NEN, Boston, Mass.) directed against the CMV p72 antigen. A third aliquot of PMNL was dedicated to the CMV pp65 antigenemia assay (1C3 clone; Argene, Parc Technologique Delta Sud, France). The exact number of cells spotted on each well was counted with a grid, and results were reported per 10⁵ PMNL. The remainder of the PMNL aliquot was frozen at 70°C until cell lysis.

Q-PCR-PMNL assay. PMNL pellets were thawed and resuspended in a lysing solution (1× PCR buffer; Promega Corporation, Madison, Wis.) containing proteinase K (final concentration, 120 µg/ml; Sigma, Mississauga, Ontario, Canada) to obtain a concentration of 10⁴ cells per µl. The mixture was incubated for 1 h at 56°C, and proteinase K was inactivated by heating the solution for 10 min at 95°C. Ten microliters (10⁵ PMNL) of the mixture was added to the PCR master mix. The CMV DNA load in PMNL was evaluated by a recently described Q-PCR protocol that uses PMNL (Q-PCR-PMNL) and nonisotopic hybridization (4). Hybrids were detected with a commercial kit (SHARP Signal System; Digene Corporation, Silver Spring, Md.). The lower limit of detection of this assay is 25 copies per 10⁵ PMNL. All samples negative for CMV DNA by PCR were screened for the presence of PCR inhibitors by testing for the -actin gene (7).

QC-PCR-plasma assay. A 100-µl aliquot of filtered plasma was added to a 100-µl solution containing (final concentrations) 100 mM KCl, 20 mM Tris-HCl (pH 8.3), 5 mM MgCl₂, and 0.9% Tween 20 solution. After proteinase K digestion (final concentration, 120 µg/ml) at 55°C for 1 h, the reaction was inactivated at 95°C for 10 min and the reaction mixture was microcentrifuged (12,000 × g for 5 min), and then 10 µl of supernatant (corresponding to 5 µl of initial plasma) was used for PCR (1). A quantitative-competitive PCR assay used with plasma (QC-PCR-plasma) has recently been described and is similar to that used with PMNL except that 50 copies of an internal control (IC) are added to each sample in order to rule out the presence of PCR inhibitors in cell-free specimens (3). This competitor consists of a plasmid containing the human papillomavirus type 31 genome with CMV major immediate-early primer sequences at its ends so that both the IC and patient DNAs containing CMV are amplified by the same CMV primers. After PCR amplification, half of the reaction mixture is hybridized to a specific human papillomavirus type 31 probe, the other half is hybridized to a specific CMV major immediate-early probe, and a colorimetric signal is obtained by the SHARP Signal System described above. The lower limit of detection of this assay is about 12 copies per 5 µl of plasma.

AMPLICOR CMV test. The AMPLICOR CMV test is a qualitative PCR test for the detection of CMV DNA in plasma and is based on four major processes: specimen preparation, PCR amplification of target DNA, hybridization of amplified products to specific probes, and detection of probe-bound amplified products by colorimetric determination. In addition, the test permits the simultaneous amplification of an IC plasmid that has primer binding regions identical to those of the CMV target sequence but that has a different internal probe binding region. The AMPLICOR CMV test was performed according to the manufacturer's recommendations except that plasma samples were filtered (0.45-µm-pore-size filter) before use. Briefly, CMV DNA is first isolated from 50 µl of plasma by lysis of virus particles at 100°C for 30 min with a detergent solution containing proteinase K. Processed specimens (equivalent to 5 µl of initial plasma) are then added to the PCR master mix containing the IC. Specific biotinylated CMV primers amplify a sequence of approximately 365 bp within the human CMV DNA polymerase gene (UL 54). Selective amplification of target nucleic acids is achieved by the use of AmpErase (Roche Diagnostic Systems, Inc.) and dUTP. Amplification is performed in a GeneAmp PCR System 9600 (Perkin-Elmer, Norwalk, Conn.) under the following conditions: 10 min at 50°C and then 40 cycles of 30 s at 94°C followed by 30 s at 65°C and finally 10 min at 72°C.

Statistical analyses. CMV assays were compared by both observation of concordance and analysis with the kappa coefficient, which is a chance-corrected measure of agreement. For most purposes, kappa values greater than 0.75 may be taken to represent excellent agreement beyond chance, values below 0.40 may be taken to represent poor agreement, and values between 0.40 and 0.75 may be taken to represent fair-to-good agreement (9). Student's t test was used to compare the viral loads from concordant (Q-PCR- and AMPLICOR CMV test-positive) and discordant (Q-PCR-positive and AMPLICOR CMV test-negative) assay results.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Concordance of results and kappa coefficient agreement among CMV assays. Blood samples were obtained from 168 HIV- and CMV-seropositive subjects with CD4 T-cell counts of 250 per mm³, including 26 patients with newly diagnosed CMV disease (20 with retinitis, 4 with gastrointestinal disease, and 2 with pneumonitis). Additionally, blood samples from 11 HIV-infected subjects with high CD4 T-cell counts and/or CMV-seronegative status were included as controls. The observed concordance (number of concordant results divided by the number of concordant plus discordant results) and the agreement (kappa coefficients) among the results of the AMPLICOR CMV test and other CMV assays are reported in Table 1. Excellent agreement (Kappa value of >0.75) was found among the results of the AMPLICOR CMV test, the shell vial assay, and the QC-PCR-plasma assay. The agreement between the results of the AMPLICOR CMV test and those of the Q-PCR-PMNL assay was low (kappa value = 0.35) but it increased considerably (kappa value = 0.81) when a diagnostic cutoff value of 690 copies per 10⁵ PMNL was selected for the cell-based assay (see below).

Comparison between results of the AMPLICOR CMV test and those of the QC-PCR-plasma assay. The AMPLICOR CMV test was compared to the in-house QC-PCR-plasma assay by using 179 plasma samples. Results of both tests were concordant for 169 specimens (147 negative and 22 positive). The two assays showed discordant results for 10 samples (seven AMPLICOR-positive and three QC-PCR-plasma-positive tests). Based on results obtained with the other CMV assays and on clinical information, all seven AMPLICOR CMV test results were considered true positives. Four of the seven false-negative test results obtained with the QC-PCR-plasma assay were due to the presence of PCR inhibitors. Such inhibitors were encountered in none of the samples tested by the AMPLICOR CMV test but were encountered in 3.4% (6 of 179) of those tested by the QC-PCR-plasma assay. The three samples that tested positive only by the QC-PCR-plasma assay were all considered true positives. The viral DNA loads in these three samples were very low (mean, 18 copies per 5 µl; median, 24 copies per 5 µl [all contained 25 copies]) compared to the viral loads in the 22 samples that were positive by both assays (mean, 1,548 copies per 5 µl; median, 216 copies per 5 µl; range, <25 to 10,091 copies).

Comparison between results of the AMPLICOR CMV test and those of the Q-PCR-PMNL assay. The AMPLICOR CMV test was also compared to the Q-PCR-PMNL assay by using 177 blood samples. Results were concordant for 122 samples (94 negative and 28 positive samples). On the other hand, discordant results were found for 55 specimens (all positive by the Q-PCR-PMNL assay only). Specimens that tested positive by the Q-PCR-PMNL assay only were from two different groups of subjects. Forty-six of the 55 (83.6%) samples were from asymptomatic HIV-infected subjects (mean, 242 CMV DNA copies per 10⁵ PMNL; range, <25 to 4,376 copies), whereas 9 of the 55 (16.4%) samples were from subjects with established CMV disease (mean, 3,563 CMV DNA copies per 10⁵ PMNL; range, <25 to 16,259 copies). Overall, the mean viral load for the 55 samples positive by the Q-PCR-PMNL assay but negative by the AMPLICOR CMV test was 785 copies per 10⁵ PMNL (median, 96 copies; range, <25 to 16,259 copies). In comparison, the mean CMV DNA load for the 28 samples that tested positive by the two assays was 21,452 copies per 10⁵ PMNL (median, 9,784 copies; range, 292 to 140,430 copies) (P = 0.003). The AMPLICOR CMV test had sensitivity and specificity values of 33.7 and 100.0%, respectively, when results were compared to results of the Q-PCR-PMNL assay with no preselected cutoff value. The best agreement between the two tests was found when a threshold of 690 copies per 10⁵ cells was selected for the Q-PCR-PMNL assay. In that context, the AMPLICOR CMV test had a sensitivity of 96.4% and a specificity of 95.3% when results were compared to results of the cell-based PCR assay.

Diagnostic values of the assays for CMV disease. The sensitivities, specificities, and positive and negative predictive values of the conventional and molecular assays obtained with data from the 168 HIV- and CMV-seropositive subjects with 250 CD4 T-cells per mm³ are reported in Table 2. The incidence of CMV disease for these individuals during the study period was 15.5%. The highest sensitivity and negative predictive values were obtained by using the leukocyte-based assays (Q-PCR-PMNL and pp65 antigenemia assays), whereas the highest specificity and positive predictive values were obtained by using the plasma-based assays (in-house QC-PCR-plasma and AMPLICOR CMV test). Twenty-four of the 29 (82.8%) samples that were positive by the AMPLICOR CMV test were from subjects with documented CMV disease (n = 17) or from patients who developed disease within 4.3 months of testing (n = 7).

Longitudinal analysis of the CMV viral load. Sequential blood samples were analyzed from six subjects before they developed CMV disease (Table 3). Plasma samples from five of six patients were positive by the AMPLICOR CMV test at the time CMV disease was diagnosed, although all six patients had positive samples prior to development of disease. The AMPLICOR test was positive for at least 48 days before the onset of CMV disease (up to 350 days for one subject). There was generally a good concordance between a positive AMPLICOR CMV test and a high viral load in PMNL as determined by the Q-PCR-PMNL and the antigenemia assays.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Until recently, PCR testing of blood samples has been used mainly to confirm the diagnosis of CMV disease especially when retinal lesions are atypical or to complement histopathological examination in the case of visceral disease (4, 19, 23). Recently, CMV PCR assays have also been evaluated for their potential to identify high-risk individuals in the context of preemptive therapeutic strategies (16, 22). Finally, molecular CMV assays can also be used for monitoring antiviral therapy and assessing the emergence of drug-resistant mutants (2, 5, 6, 13). However, many basic problems, such as the need for a Q-PCR instead of a qualitative PCR assay in some of the clinical situations described above, remain unsolved. Similarly, there exists a controversy over the type of specimen to use for PCR testing, i.e., PMNL or plasma. More importantly, there is a lack of standardization for CMV PCR testing, in part because simple and reproducible commercial kits have not been developed. The purpose of our study was twofold. First, we compared various nonmolecular and molecular CMV assays (using both PMNL and plasma specimens) for their potential to confirm CMV disease and also, to a lesser extent, to predict the development of disease in subjects with AIDS. At the same time, we performed an evaluation of a new commercially available PCR kit, the AMPLICOR CMV test.

The results of this study demonstrate an excellent agreement (kappa value > 0.75) between the results of the AMPLICOR CMV test and those of another plasma-based PCR assay in use at our institution (3). The agreement between the results of the AMPLICOR CMV test and those of other cell-based assays was not as good in general. In particular, a very low correlation (kappa value = 0.35) was found with the results of the Q-PCR-PMNL assay that we recently described (4). However, a better correlation was found between the results of the two assays when a cutoff of 690 copies per 10⁵ cells was selected for the PCR-PMNL test (kappa value = 0.81). This cutoff value is close to the one (1,000 copies per 10⁵ PMNL) associated with the best sensitivity (87.5%) and specificity (96.2%) for the diagnosis of CMV disease in a recently published study (4). A positive result by the AMPLICOR CMV test was also clearly associated with a high viral DNA load, and conversely, a negative test was associated with a low viral DNA load. This finding can be illustrated by comparing the mean CMV DNA load in samples positive by both PCR assays (AMPLICOR test and Q-PCR assays) with the one found in samples positive by the Q-PCR assays only. Indeed, we found an 84.1- and 27.3-fold difference in the mean viral loads of concordant and discordant samples using, respectively, QC-PCR-plasma and Q-PCR-PMNL results as a reference. Similarly, Freymuth et al. showed that the presence of CMV DNA in plasma as detected by nested PCR was in correlation with high levels of antigenemia in AIDS patients and organ transplant recipients (10).

As a diagnostic test for CMV disease in HIV-infected subjects, the AMPLICOR test was highly specific (92%) but not as sensitive (65%) as other cell-based assays. The lower sensitivity of the AMPLICOR test than those of the Q-PCR-PMNL and antigenemia assays is in agreement with our own data obtained by another PCR-plasma assay (sensitivity, 65%) as well as results from other groups who reported a higher viral load in PMNL than in plasma for an equivalent blood volume (14, 24). However, some investigators have found much higher sensitivity values (ranging from 74 to 83%) using other plasma- or serum-based PCR assays (16, 23). The reasons for such discrepancies need to be further examined, but it might be related to sample preparation, the volume of sample tested, or the PCR conditions. Also, the use of filtered plasma in our study may have resulted in a lower than expected sensitivity value for the AMPLICOR CMV test, although preliminary data indicate that filtration does not seem to alter test results significantly (data not shown). Recently, the AMPLICOR CMV test was found to be more sensitive than the antigenemia assay in a small study realized with samples from bone marrow and solid organ transplant recipients (17). Although those results differ somewhat from ours, the population studied and the definition of CMV disease were not the same. In particular, both cases of active CMV infection and CMV disease were considered in the analysis of the latter study whereas only proven cases of CMV disease were included in our analysis. The discrepant results emphasize the fact that CMV diagnostic tests should be evaluated in each of the different clinical settings.

Interestingly, the AMPLICOR CMV test readily identified the risk of progression to CMV disease in a small subset of patients who had longitudinal PCR evaluations (Table 3). For these subjects, the AMPLICOR test was positive at least 48 days before the development of disease. A recent larger study showed that the median time between a first positive result by an in-house PCR-plasma assay and diagnosis of CMV disease was approximately 6.0 months (22). In that context, PCR-plasma assays may be a better laboratory marker than PCR-PMNL assays for enacting preemptive therapy for AIDS patients, since cell-based assays may be positive for an unacceptably long period of time before development of disease (3, 14). Clearly, a larger prospective longitudinal study is needed to verify this aspect.

Overall, the AMPLICOR CMV test is a very convenient assay that combines rapidity (<6 h), simplicity, and the possibility of batch testing. Unlike with the antigenemia assay, specimens do not need to be processed immediately and require minimal preparation before testing (18). Also, compared to most PCR protocols in use, the AMPLICOR CMV test minimizes contamination risks by taking advantage of the AmpErase- dUTP system and facilitates detection of PCR products with the use of microwell plates and a colorimetric detection system. On the other hand, this test does not provide quantitative results and still may need to be used in conjunction with additional methodologies for virus recovery and antiviral susceptibility testing.

In conclusion, the place of the AMPLICOR CMV test is not completely defined at present and is likely to depend on the type of clinical situations encountered and the workload of the virology laboratory. For large reference laboratories, the AMPLICOR CMV test appears particularly suitable, since blood samples may be kept for up to 24 h at room temperature before being processed and batch testing is possible. Clinicians need to keep in mind that a negative test does not rule out CMV disease but that a positive test almost always identifies an individual with an established CMV disease or one at high risk for subsequent disease. Indeed, our results show that 82.8% (24 of 29) of the positive AMPLICOR CMV tests were found with samples from patients with active CMV disease or subjects who developed disease within 4.3 months. For smaller laboratories, the CMV antigenemia assay is still a very good alternative, but it has the technical limitations described previously. Although, we did not address this aspect of testing in our study, it is probable that Q-PCR assays will be necessary for monitoring anti-CMV therapy. In that context, a Q-PCR assay, which can be used with either plasma or PMNL specimens, from Roche Diagnostic Systems (COBAS AMPLICOR CMV MONITOR test) is under evaluation.