INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Cytomegalovirus (CMV) infections pose a serious threat to AIDS, bone marrow, solid organ transplant, and other immunocompromised patient populations (14, 20). A rapid and accurate laboratory diagnosis of CMV is important for the appropriate administration of antiviral therapy.

CMV PCR technology has been used in clinical diagnostics and research since the mid-to-late 1980's. Such testing has been shown to serve as a marker of current or impending disease, and the assay has application in the monitoring of anti-CMV drug efficacy (4, 5, 8, 10, 13, 14).

Conventional PCR requires detection and confirmatory testing of the amplicon by gel electrophoresis, incorporating an isotopically labeled or nonradioactive hybridization assay or a nested PCR to detect targets present in very low copy numbers (8, 17, 20). In an effort to simplify diagnostic PCR technology for the clinical microbiology or virology laboratory, a PCR-solid-phase enzyme immunoassay (EIA) plate technology system was developed. The advantage of the EIA plate detection system over that of conventional PCR rests with the test's improved turnaround time and its simplicity of use. The purpose of this study, accordingly, was to evaluate the relatively new PrimeCapture CMV DNA Detection Plate System (PC-PCR) for incorporation into the typical clinical setting.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Clinical specimens. A total of 51 blood specimens (34 patients) and 50 cerebrospinal fluid (CSF) specimens (49 patients) were tested in this study. All experimental testing was performed on specimen volumes remaining after use in routine diagnostic testing (i.e., pathologic discard). Blood specimens were consecutively collected from AIDS patients and, to a lesser extent (i.e., <5%), bone marrow transplant recipients. All specimens were collected from patients with clinical syndromes suspected to be caused by a CMV infection. Patients were not on anti-CMV therapy at the time of testing. Retrospective testing of CSF was performed on specimens which had been frozen at 70°C. CSF was collected from pediatric patients, and to a much lesser extent, from AIDS patients suspected of having central nervous system infections. Blood specimens were processed within several hours of collection (i.e., by dextran sedimentation, fixation, DNA extraction; see below). Extracted DNA was temporarily maintained at 4°C if assayed within 1 week, or stored at 20°C for longer periods prior to the performance of PC-PCR and nested PCR. Extraction of DNA from CSF was performed in batches from frozen specimens, and either tested immediately in parallel by PC-PCR and nested PCR, or temporarily (<1 week) maintained at 4°C.

PC-PCR. PC-PCR was performed in accordance with the manufacturer's specifications, incorporating slight modifications suggested by Stephen Day (ViroMed Laboratories, Minneapolis, Minn.) (product currently available from Synthetic Genetics, San Diego, Calif.). Supplemental reagents not supplied with the PC-PCR kit were obtained from the GeneAmp PCR Core kit (PE Applied Biosystems, Foster City, Calif.).

(i) Isolation of PMNL from peripheral blood. Five milliliters of peripheral blood was collected in EDTA-containing vacutainer tubes from AIDS patients and other patient groups seen in clinics and by physicians in private practice. Isolation of polymorphonuclear leukocytes (PMNL) was performed as described previously (14). Briefly, a PMNL-dextran solution was incubated at 37°C for 20 min, followed by collection, washing, and counting of the leukocyte-dextran mixture fraction. The isolated cellular compartment was adjusted to aliquot concentrations of 2 × 10⁶ PMNL/ml.

(ii) DNA extraction. The QIAamp DNA extraction kit (Qiagen, Santa Clarita, Calif.) was used throughout the study. Briefly, 2 × 10⁶ PMNLs or 200 µl of CSF was added to 1.5-ml microfuge tubes. Twenty-five microliters of the Qiagen protease and 200 µl of buffer AL were added to each sample. Mixtures were vortexed for 15 s (i.e., until the pellets were dissolved) and then heated to 70°C for 10 min. Absolute alcohol was added to each sample, which was then mixed by vortexing and applied to the spin column. After centrifugation at 6,000 × g, the columns were washed with buffer AW, followed by elution of the DNA with buffer AE. The extracted DNA was either amplified immediately or temporarily stored at 20°C. PMNL and CSF negative controls were obtained from seronegative donors.

(iii) Preparation of positive control. MRC-5 tube cultures were inoculated with the laboratory-adapted CMV strain AD-169 (ATCC VR-538). Upon the development of a cytopathic effect (CPE) infecting 70 to 80% of the tube culture monolayer, the cells were scraped and washed with phosphate-buffered saline, followed by DNA extraction as described above. AD-169 DNA extracts were tested in parallel with all experimental runs.

(iv) Amplification. A reaction mix of 50 µl was prepared for amplification and consisted of 30.25 µl of Ultra Pure water, 5 µl of 10× PCR buffer II, 1 µl of each deoxynucleoside triphosphate dNTP/(dATP, dUTP, dGTP, and dCTP), 0.25 µl of Taq polymerase, 1 µl of CMV primer mix (Primer 1 and Primer 2), 4 µl of 25 mM MgCl₂, and 0.5 µl of uracil-N-glycosylase (UNG) to prevent PCR product carry-over. Forty-five microliters of this master mixture and 5 µl of each DNA template were added to the appropriate PCR tubes. This complete reaction mixture was heated to 50°C for 2 min (to destroy any dUTP-containing amplicons from previous amplifications) followed by five cycles at 95°C for 10 s (denaturation), 58°C for 30 s (annealing), 72°C for 10 s (primer extension). An additional 35 cycles consisted of 94°C for 30 s (denaturation), 58°C for 30 s (annealing), and primer extension for 30 s at 72°C. After the last cycle, primer extension was continued for 5 min to allow all templates to be completed. Cycling was performed using a GeneAmp PCR system thermal cycler model 2400 (PE Applied Biosystems).

(v) Postamplification processing. Hybridization solution and conjugate diluent were preheated to 37°C and, before use, mixed until the solutions were clear. Ten microliters of each amplicon were added to detection plate wells, followed by the addition of the supplied denaturation and hybridization solutions. The plate was incubated at 45°C for 55 min, washed (using the supplied wash solution), and followed by the addition of the streptavidin-horseradish peroxidase conjugate mixture. Following the tetramethylbenzidine substrate additive and stop solution, absorbance was determined at 450 nm (enzyme-linked immunosorbent assay [ELISA] plate reader, SYVA MicroTrak Autoreader, part no. 3801153, version 1.0; Wampole Laboratories [Cranbury, N.J.]). An optical density (OD) of 3 was considered an indication of a PC-PCR positive specimen and signal. Extracted DNA obtained from uninoculated MRC-5 cells served as the negative control. Control OD readings were equivalent to background measurements observed on a standard curve (see Results and Table 1). Plate sealers (Dynex Technologies, Inc., Chantilly, Va.) were used during intermediary plate-shaking steps to prevent aerosol contamination of adjacent wells.

(vi) Construction of a standard curve. Using the laboratory-adapted CMV strain AD-169, a standard curve was constructed to determine the minimal detection limit of PC-PCR, for comparison to both quantitative (plaque) and quantal (endpoint) infectivity assays. The curve was constructed using a CMV stock aliquot previously determined to have a titer of 3 × 10⁵ PFU/ml (9.6 × 10⁴ 50% tissue culture infective doses [TCID₅₀/ml]). DNA extraction and amplification were performed by using log₁₀ viral serial dilutions ranging from 10³ to 10¹¹. Amplification, plate inoculation, and OD readings were performed as described above.

Nested PCR. Nested PCR was performed only for the purpose of assisting in the resolution of discordant results. Due to the potential occurrence of a primer-to-template binding error in any given PCR, or to the failure of the initial PCR with primer pairs CB-1 and CB-2 to help resolve discordant results, a second or third in-house PCR assay, encompassing different CMV gene segments, was used in subsequent confirmatory tests.

(i) Amplification of the CMV glycoprotein B (gB) gene segment flanking the glycoprotein 55 cleavage site. Five microliters of the extracted DNA was added to a PCR mixture (400 µM each dNTP, 4 mM MgCl₂, 0.8 µM each primer, 2.5 U of Taq polymerase) to achieve a 50-µl final volume. The two oligonucleotide primer pairs used to perform nested PCR consisted of outer primers CB-1 (5'-CTG GGA AGC CTC GGA ACG-3') and CB-2 (5'-ACC CAT GAA ACG CGC GGC-3') and inner primers CB-3 (5'-ACG TAC TAT CCG TTC CGA-3') and CB-4 (5'-GGC AAT CGG TTT GTT GTA-3') (550 to 556 bp). The sequences of the primers correspond to nucleotides 1200 to 1217, 1765 to 1782, 1215 to 1232, and 1750 to 1767, respectively. The mixtures were run for 45 cycles. Cycle 1 included 5 min at 94°C (denaturation), 1 min 50 s at 50°C (annealing), and 1 min at 72°C (extension). The remaining 44 cycles were identical to the first except that denaturation was performed for 1 min. The nested reaction was performed by adding 0.1 µl of the previously amplified product to a second reaction mixture as described above, except that the final MgCl₂ concentration was 3 mM/100 µl. This second amplification was run for 35 cycles at 94°C for 1 min (denaturation), 52°C for 1 min 50 s (annealing), and 72°C for 1 min (extension). Cycle 1 of the nested PCR was performed at 94°C for 2 min. An additional cycle of 72°C for 10 min was performed on both the PCR and the nested reaction. The gels (1.5% agarose) were stained with ethidium bromide and examined by UV transillumination (19).

(ii) Amplification of the CMV UL97 gene. PCR conditions were identical to that described for the amplification of the CMV gB gene segment as described in section i above, except for the following. The outer primer oligonucleotides VS976 (5'-ATT CGT GCA GCA TGG TCT-3') and VS977 (5'-TAC GGC GTT ATT GCA TGT-3') (nucleotides 1992 and 1635, respectively) were used to amplify a 357-bp fragment (16, 22). The inner primers consisted of C97E (5'-ATG TCG GAG CTG TCG GCG-3') and C97F (5'-GTC TGC GAG CAT TCG TGG-3') (nucleotides 1649 and 1978, respectively) to amplify a 329-bp fragment (17a). The Mg concentration used in the performance of the outer segment amplification was 2.25 mM. An annealing temperature of 52°C was used in both amplification components within this nested (UL97) PCR.

(iii) The CMV variable region at the 5' end of the gB ORF (primers CB-5 through CB-8). Additional testing was performed by utilizing primers to a variable region of the 5' end of the gB open reading frame (ORF) to further assess and resolve discordant data. Both PCR mixtures and conditions were identical to that described in section ii above, except for the following oligonucleotide primers: outer primers CB-5 (5'-CCT CAT CGC TGC TGG ATT-3') and CB-6 (5'-TGA CTC CCA CCA CAT CTC-3') and inner primers CB-7 (5'-ATT TGG CCC GCG ACG AAC AT-3') and CB-8 (5'-CTC CGT ACT TGA GGG TAG TG-3') (250 to 256 bp). These primer sequences correspond to nucleotides 124 to 141, 392 to 409, 139 to 158, and 375 to 394, respectively, of the gB gene. A magnesium concentration of 4 mM was used for the amplification of the genome segment corresponding to the first round of the nested PCR. An annealing temperature of 55°C and a magnesium concentration of 2 mM were used in the (inner primer) second amplification round (3, 20).

The CMV antigenemia (CMV-Ag) assay. The CMV-Ag assay was performed according to the methodology of Lipson et al. (14). From the purified PMNL preparation, cell spots were prepared in duplicate by cytocentrifugation, each consisting of 4 × 10⁵ PMNLs (Cytospin 2 cytocentrifuge; Shandon, Pittsburgh, Pa.). Cells were fixed in formalin-sucrose followed by immunostaining with antibody 1C3 (pp65-67, catalog no. 11-001; Argene Biosoft Department, North Massapequa, N.Y.). The second immunostaining reagent consisted of a fluorescein isothiocyanate-conjugated F(ab')₂ GAM immunoglobulin G fragment gamma chain-specific antibody (ICN Pharmaceuticals, Costa Mesa, Calif.).

Virus isolation (TC-CPE). From the dextran-purified PMNL preparation, 0.2 ml was added to two MRC-5 tube cultures. The cultures were read three times per week for a period of 30 days. Tubes suspected of a CMV-induced CPE were subcultured. Confirmatory testing of the CMV-induced CPE was performed by indirect immunofluorescence, using the monoclonal antibody 1C3.

Plaque assay. The plaque assay was performed as previously described (18). Briefly, MRC-5 tube cultures were trypsinized, followed by seeding of 24-well, flat-bottom cell culture plates. Plates were incubated at 36.5°C for 3 to 5 days in a 5% CO₂ atmosphere. Upon monolayer confluency, the growth medium was replaced with Eagle minimal essential medium supplemented with 2% fetal bovine serum, antibiotics, and L-glutamine. CMV strain AD-169 was serially diluted (log₁₀) from neat to 10⁹. Viral dilutions were inoculated in quadruplicate. The positive control consisted of CMV strain AD-169 diluted to yield 50 to 100 PFUs per well. The negative control consisted of four uninoculated wells. Medium was changed after 7 days. Final plate readings were performed after 14 days.

TCID₅₀ and endpoint titration. Viral endpoints were determined as described previously (11). Briefly, 0.2 ml of each viral dilution (10³ to 10⁹) was added in pentuplicate to MRC-5 tube cultures. Cultures were read three times per week for 30 days. Endpoints were calculated after the last tube culture reading.

Interpretation of data and resolution of discordant results. A specimen was considered a true positive upon viral (CMV) isolation. The criterion of consensus testing was utilized to further differentiate the true-positive from the true-negative specimen. The consensus tests were as follows: (i) TC-CPE negative but nested PCR (CB-1/2) positive upon retesting, with a previous and subsequent viral isolation; (ii) TC-CPE negative but PC-PCR positive and currently and previously (within 30 days) antigenemia positive; (iii) TC-CPE and PC-PCR negative but antigenemia positive and with the development of a CMV retinitis within 30 days; (iv) PC-PCR positive only but previously (<10 days) isolation positive or with the development of a laboratory-confirmed CMV enteritis within 30 days; (v) PC-PCR positive only but antigenemia and TC-CPE positive within 45 days and PC-PCR positive upon retesting; and (vi) PC-PCR, antigenemia, and TC-CPE negative but with development of a laboratory-confirmed CMV enteritis within 30 days. Additional testing to resolve discordant data was performed by nested PCR, using one or a combination of primer pairs to gene segments CMV gB, UL97 or gB ORF, designated CB-1/2 (i.e., CB-1 and CB-2), VS976-7 (VS976 and VS977), and CB-5/6 (CB-5 and CB-6), respectively.

Statistics. The calculation of sensitivity, specificity, negative and positive predictive values were performed in accordance with the method described by Strongin (21). McNemar's tests were used to determine whether any pairs of assays were significantly different. Significance was identified by determining the kappa coefficient ±95% confidence interval. A kappa value of <0.4 was considered poor agreement. A value of >0.75 was considered good agreement. Kappa values between 0.4 and 0.75 were considered fair agreement (6). A P value of <0.05 was considered statistically significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Prior to the performance of PC-PCR, an OD cutoff value was established to differentiate the viral-induced reactive (PC-PCR) signal from that of nonspecific background. To achieve this goal, a standard curve was constructed utilizing serial 10-fold input viral (CMV strain AD-169) dilutions ranging from 10³ through 10¹¹ (input virus titer, 3 × 10⁵ PFU [9.6 × 10⁴ TCID₅₀]). OD readings attained unity between infected and uninfected cultures, as viral titers approached the theoretical infectivity level of 1 PFU or the quantitative value of 1 TCID₅₀. These findings suggested our use of an OD cutoff ratio of 3 (Table 1).

Among 51 consecutively collected blood specimens, 23 (45%) were deemed positive by isolation and by the resolution of discordant results by using data obtained from PC-PCR, antigenemia, in-house nested PCR, and/or patient clinical evaluation and follow-up (Table 2). Among 28 negative blood specimens, 23 (82%) were negative by PC-PCR, antigenemia, and isolation. Four of 23 positive blood specimens (no. 1191, 1273, 1430, and 1482) were isolation positive and reactive by PC-PCR and antigenemia. Results for twenty-seven blood specimens tested by PC-PCR, the CMV-Ag assay, and isolation were in accord.

PC-PCR identified 18 true-positive blood specimens. Five false negatives and seven false positives were identified by PC-PCR (Table 2). Following the resolution of discordant results, the sensitivities, specificities, positive predictive values (PPV), and negative predictive values (NPV) for PC-PCR, CMV-Ag, and TC-CPE were 78, 75, 72, and 81%; 46, 100, 100, and 70%; and 39, 100, 100, and 67%, respectively. There was no significant difference in the rate of CMV detection and isolation by CMV-Ag assay or conventional tube culture (P = 0.480). CMV-Ag was significantly less sensitive than PC-PCR, detecting 11 positive blood specimens (P = 0.001). No false-positive CMV-Ag assay signals, or nonspecific cellular degenerations mimicking a CPE, were identified. In contrast to that observed upon parallel testing utilizing the CMV-Ag assay (P = 0.001) and TC-CPE (P = 0.001), no significant difference was identified between PC-PCR (P = 0.564) and the total CMV detection rate in blood. These differences might be explained in part by the ability of the PC-PCR plate capture assay to detect latent virus (viz., PC-PCR true-positive specimen no. 1283, 1374, 1375, 1422, 1437, 1504, and 1505). All 50 CSF specimens were negative by both PC-PCR and isolation; CMV disease did not occur in any patient whose CSF was determined negative by PC-PCR or isolation. The combined sensitivity, specificity, PPV, and NPV for blood and CSF by PC-PCR were 77, 90, 68, and 93%, respectively.

Reproducibility testing by PC-PCR was performed on three confirmed positive PMNL and three negative (two CSF, one blood) specimens over a 3- to 4-month period. PC-PCR signals obtained from two to three subsequent runs per specimen were unchanged in comparison to initial results (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

With the introduction of new and modified molecular diagnostic technologies into the clinical setting, a continual need exists to evaluate the efficacy of such tests and to compare these new assays to those which are widely and routinely performed. Accordingly, we compared the new PC-PCR assay to the well-established quantitative CMV-Ag assay and the TC-CPE "gold standard." Among specimens for which results were deemed discordant following testing by PC-PCR, antigenemia, and TC-CPE, and in consideration of our patients' clinical histories and courses, supplementary testing was performed utilizing one or a combination of three in-house nested PCR assays.

In the current study, with the utilization of assays differing in sensitivity and specificity (most noticeably, PC-PCR versus isolation; Table 3), interpretation of data obtained from consensus testing was necessary to effect discordancy resolution. PC-PCR readings equal to or greater than our determined OD cutoff value of 3 (specimen and patient no. 1289, 1323, 1324, 1330, 1441, 1443, and 1493), for example, were judged false positives due to our inability to isolate, detect, or clinically identify the virus by TC-CPE, the CMV-Ag assay, nested PCR, or patient evaluation and follow-up, respectively. Conversely, several peripheral blood specimens displaying PC-PCR OD values of <3 (specimen and patient no. 1208, 1246, 1267, 1332, and 1465) were deemed false negatives, due to either the appearance of CMV organ-specific disease upon patient follow-up within 30 days of testing or the detection of a CMV DNAemia by nested PCR, positive antigenemia, and/or viremia. These data point out, with the increasing array of laboratory tests available to the technologist, the importance of performing multiple assays to effect discordancy resolution.

The failure to identify CMV in peripheral blood by gene amplification utilizing PC-PCR, in-house nested PCR assays, antigenemia, or viremia does not unequivocally preclude impending CMV disease. Biopsy-confirmed CMV enteritis was evidenced in patient and specimen no. 1465 despite repeated (negative) testing of the patient in question 1 month prior to colonoscopy (Table 2). The development of CMV organ-specific disease among PCR-, antigenemia-, and isolation-negative AIDS patients has been reported by others but is uncommon (2, 9).

It would not be unreasonable to suggest that one or more technically defined PC-PCR false-positive specimens (i.e., no. 1289, 1323, 1324, 1330, and 1493), might actually have been indicative of latent virus within the blood leukocyte compartment of patients with asymptomatic infections (7, 15, 24). Our inability to identify CMV among the aforementioned specimens using different in-house primers synthesized to one or more late gene segments, as well as to any indication of clinical CMV disease, mandated the decision to finalize these PC-PCR results as reflective of false-positive PC-PCR signals.

The use of PCR technology in clinical diagnostics has been questioned. Amplicon build-up in the laboratory as a source of PCR contamination for example, has been recognized (1, 23). The potential for contamination by PCR product carry-over was addressed in our laboratory by the incorporation of UNG into the PCR cocktail.

It is interesting to note a failure of nested PCR (CB-1/2 primers) to detect CMV in three consecutively collected blood specimens from the same patient (specimen and patient no. 1422, 1465, and 1505) in whom CMV enteritis was confirmed upon biopsy. Specimen no. 1422, however, was PC-PCR, VS976-7, and CB-5/6 PCR positive. Specimen no. 1505 was PC-PCR positive only, while specimen and patient no. 1465 was nonreactive by PC-PCR, negative by nested PCR (CB-1/2, VS976-7, and CB-5/6 PCR), antigenemia, and TC-CPE. The failure to detect CMV in three consecutively collected specimens from the same patient might be ascribed either to a CB-1/2 primer template (12) or to an amplicon-EIA capture mismatch. The mechanism(s) responsible for our failure to detect specimen no. 1465 or 1505 by using primer pairs CB-1/2, VS976-7, or CB-5/6 might similarly be ascribed to a primer-template mismatch.

Among 50 CSF specimens tested, no positive signals were identified by PC-PCR, nor was CMV identified by TC-CPE. No patients in this segment of our study presented with or went on to develop CMV disease. Although CMV was not detected by PC-PCR in the CSF specimens in question, the data importantly point out the assay's excellent specificity with this specimen source.

PC-PCR was significantly more sensitive than antigenemia (P = 0.001), TC-CPE (P = 0.001), or combined CMV-Ag and TC-CPE assays (P = 0.003). There was no statistical difference in the sensitivity between PC-PCR and the total CMV-positivity rate (P = 0.564). The improved sensitivity of PC-PCR must be considered, however, with respect to the test's reduced specificity. Some refinements in the PC-PCR detection system protocol might improve specificity. Alternate inoculation of ELISA capture plate wells might reduce the potential of aerosol contamination between adjacent wells. Although plate sealers were used during the ELISA plate-shaking step, amplicon aerosol contamination, perhaps during sealer removal, although unlikely, may not have been unequivocally prevented.

PC-PCR offers a simple-to-perform gene amplification and complementary EIA detection system, readily amenable to the needs of the general microbiology or virology laboratory using molecular diagnostics. Importantly, PC-PCR obviates the need for gel electrophoresis and a confirmatory hybridization test (viz., Southern blotting) or a nested PCR. The EIA component of the PC-PCR test requires less than 90 min to complete, effecting simplicity in amplicon identification and an improved assay turnaround time compared to the conventional PCR gel-hybridization detection system.