INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Human cytomegalovirus (HCMV) is an important opportunistic pathogen in human immunodeficiency virus type 1 (HIV-1)-infected patients with low CD4 counts and is a significant cause of morbidity and mortality (12, 18). Central nervous system (CNS) involvement can occur, resulting in several neurologic manifestations including myelitis/polyradiculopathy, encephalitis with dementia, ventriculoencephalitis, and mononeuritis multiplex (18, 24). The clinical presentations of HCMV CNS disease can be nonspecific, and the diagnosis of HCMV CNS disease is often one of exclusion, primarily based upon clinical presentation, neuroradiologic studies, cerebrospinal fluid (CSF) chemistries, serologic testing, CSF viral culture (18), and DNA PCR (1, 2, 5, 8, 9, 14, 29). However, CSF viral culture can be insensitive (18) and qualitative DNA PCR can detect both latent and replicating virus, indicating that alternative methods, such as quantitative PCR, may be necessary to differentiate between latent infection and active viral replication (3, 6, 20, 21, 22, 25, 28).

The NucliSens (CMV) pp67 assay (4; F. Roeles, P. Sillekens, N. Tacken, F. Cromme, J. Middeldorp, T. Oosterlaken, and R. Schoones, Program Abstr. 6th Eur. Conf. Clin. Aspects Treatment HIV Infect., abstr. 313, 1997) uses nucleic acid sequence-based amplification (NASBA) to detect the presence of mRNA encoding pp67, a phosphorylated matrix tegument protein encoded on the L-unique (U_L) 65 gene of HCMV (10, 11). pp67 mRNA is one of the most abundant late gene transcripts that is detectable when viral replication is occurring (11). Consequently, the presence of late gene transcripts is indicative of active HCMV infection but will not be detected during viral latency (10, 11, 19). Therefore, we examined the clinical utility of using the NucliSens CMV pp67 assay for the diagnosis of active HCMV CNS disease. The results obtained with the NucliSens CMV pp67 assay were compared to those obtained by CSF DNA nested PCR for the detection of HCMV glycoprotein B DNA (23) and by CSF viral culture. Overall results were also correlated with clinical diagnosis determined by patient clinical history and presentation, standard laboratory tests, and neuroradiology studies.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Patient population and clinical samples. All patients enrolled in the study were HIV-1 seropositive and were recruited from the North Shore University Hospital Center for AIDS Research and Treatment, Manhasset, New York, from September 1994 through April 1998. After informed consent was obtained, a total of 76 CSF specimens were collected by lumbar puncture from 65 patients undergoing spinal tap for standard medical care. Whole bloods were collected by venipuncture in EDTA-anticoagulated and serum separator VACUTAINER tubes (Becton Dickinson, Franklin Lakes, N.J.) and centrifuged for 20 min at 1,000 × g in a swinging bucket rotor (Spinchron R; Beckman Instruments, Inc., Palo Alto, Calif.). CSF, plasma, and serum were stored in cryovials at 80°C until tested.

Clinical laboratory testing. CSF protein and glucose levels and cell counts were determined according to standard practices. When clinically indicated, HCMV blood cultures and pp65 antigenemia assays (27) were performed according to standard practices. CMV immunoglobulin G (IgG) and CMV IgM assays were performed by using the Vidas immunoassays (bioMerieux, St. Louis, Mo.) according to the manufacturer's protocol. CSF and blood were routinely cultured for bacterial, fungal, viral, and mycobacterial pathogens. When indicated, additional serologic tests for cryptococcal antigen, toxoplasma IgG and IgM, syphilis, herpes simplex virus types I and II, hepatitis C, and Lyme disease were performed.

Subject neurologic disease classification. Patients were initially stratified to specific disease groups by clinical presentation, history, laboratory data, neuroradiologic examination (computerized tomography [CT] and magnetic resonance imaging when indicated) and neurologic dysfunction, without knowledge of NucliSens pp67 NASBA or DNA PCR results. Correlation of NucliSens pp67 assay and DNA PCR results to disease classification was retrospective at completion of study. In 10 cases a definitive diagnosis could not be made until DNA PCR and NucliSens pp67 assay results were included in the evaluation criteria. Review of long-term patient clinical course and/or autopsy results determined the clinical relevance of the PCR and NucliSens pp67 assay results.

NucliSens CMV pp67 qualitative (QL) assay. Testing for the presence of HCMV pp67 mRNA transcripts was performed using the NucliSens CMV pp67 assay (Organon Teknika, Durham, N.C.) according to the manufacturer's instructions. Briefly, the in vitro-generated system control RNA present in the kit and 200 µl of patient CSF were added to 0.9-ml NucliSens lysis buffer tubes. Total nucleic acids were isolated based on the guanidine isothiocyanate-acidified silica procedure (7). After isolation, nucleic acid extract (5 µl) was used in the amplification reaction. Amplified products were hybridized with pp67 wild-type and system control probes labeled with ruthenium (Ru²⁺). Results were obtained by analysis of the hybridization reaction products in the NucliSens Reader System (Organon Teknika, Durham, N.C.), an electrochemiluminescence reader.

CMV DNA detection by PCR amplification. Amplification and detection of a conserved region of HCMV glycoprotein B DNA were performed according to the method described by Shepp et al., with the following modifications (23). Nested PCR was done using external oligonucleotide primers CB1 (5' CTG GGA AGC CTC GGA ACG 3') and CB2 (5' ACC CAT GAA ACG CGC GGC 3') and internal oligonucleotide primers CB3 (5' ACG TAC TAT CCG TTC CGA 3') and CB4 (5' GGC AAT CGG TTT GTT GTA 3'). Sequences of the four primers correspond to nucleotides 1200 to 1217, 1765 to 1782, 1215 to 1232, and 1757 to 1767, respectively. The initial PCR mixture contained 10 µl of sample DNA from the NASBA isolation, primers CB1 and CB2 (each at a concentration of 0.5 µM), a 200 µM concentration of each deoxynucleoside triphosphate, 1.5 mM MgCl₂, and 2.5 U of Taq polymerase in PCR buffer (Applied Biosystems, Foster City, Calif.). PCR analysis was done using an automated thermal cycler (PCR 9600 System; Perkin-Elmer Corporation, Foster City, Calif.). PCR conditions were as follows: 94°C for 5 min followed by 35 cycles of 94°C for 60 s, 56°C for 90 s, and 72°C for 60 s and an extension at 72°C for 10 min. The conditions of the nested PCR with CB3 and CB4 were identical, except that only 1 µl of the initial PCR product was used and the concentration for primers CB3 and CB4 was 0.6 µM. Following the PCR, the HCMV-amplified DNAs were electrophoretically separated on a 1.6% agarose gel stained with ethidium bromide. DNA fragments were visualized by UV scanning with an Ultra Violet Products (UVP) gel documentation system (Integral Technologies, Inc., Indianapolis, Ind.). PCR segments varied in size between 550 and 556 bp depending upon the HCMV isolate subgroup.

Amplification control materials. Previously characterized HCMV clinical isolates (23) and stock culture samples (courtesy of D. Shepp) representing HCMV subtypes 1 to 4, including the AD-169 and Towne strains, were tested for amplification efficiency and detection by both the NucliSens pp67 and glycoprotein B DNA PCR assays. The sensitivity of the nested DNA PCR assay for glycoprotein B was determined by amplifying log₁₀ viral serial dilutions (range, 10¹ to 10¹⁰) of stock cultures of the AD169 and Towne strains and detection as described above.

Assessment of albumin BBB leakage. Assessment of blood-brain barrier (BBB) integrity was performed on 56 of the 65 study patients, including 13 patients with HCMV CNS disease, 9 patients with nonneural HCMV disease, 12 patients with AIDS dementia complex (ADC) and 22 patients with other CNS disease. CSF and serum albumin and IgG levels were measured with the Behring nephelometric analyzer (Dade Behring, San Jose, Calif.) according to manufacturer's instructions. The rate of albumin BBB leakage was calculated with the following formula of Tourtellotte et al. (26): (CSF albumin level [in milligrams per deciliter] minus serum albumin [in milligrams per deciliter]) divided by 230 and multiplied by 5. In normal individuals the albumin BBB leakage range is 5 to 75 mg/day.

Statistical analysis. For each test, the clinical sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively) were calculated according to standard formulas by comparing test results for patients with HCMV CNS disease with test results from patients who did not have HCMV CNS disease. Nonparametric analysis of variance for CD4 T-cell counts and CSF white blood cell (WBC), protein, and glucose levels, classified by variable group, and Wilcoxon two-sample test (rank sums) for each variable classified by group were performed.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sensitivity and specificity of DNA PCR and NucliSens pp67 assays. Prior to testing the CSF samples by DNA nested PCR and the NucliSens pp67 assay, studies were conducted to determine the analytical sensitivity and specificity of the PCR assay and the analytical specificity of the NucliSens pp67 assay. Nested PCR performed on serial dilutions of AD-169 and Towne strains demonstrated that the assay was capable of detecting between 1 and 5 copies of HCMV and all 36 previously characterized HCMV isolates (23), including 8 from group 1, 23 from group 2, 10 from group 3, and 4 from group 4 (data not shown). The NucliSens pp67 assay was efficiently able to detect the same 36 HCMV isolates (data not shown).

Stratification of patients by clinical diagnosis. In total, 76 CSF samples were collected from 65 randomly selected patients. The study population consisted of 13 females (mean age ± standard deviation [SD], 36 ± 7 years) and 52 males (mean age ± SD, 39 ± 7 years). The CSF samples were initially grouped according to patient diagnosis determined by clinical presentation and history, laboratory data, and neuroimaging studies without knowledge of either the NucliSens pp67 assay or DNA PCR results (Table 1). Group A comprised seven CSF samples from seven control patients who did not exhibit any evidence of neurologic disease. Group B contained 39 CSF specimens obtained from 31 patients diagnosed with a non-HCMV-related neurologic syndrome. Group C included 21 CSF samples from 19 patients. A total of 11 CSF samples were from patients with a definitive diagnosis of HCMV CNS disease, including encephalitis, neuropathy, or radiculopathy and retinitis concurrent with HCMV encephalitis. The additional 10 CSF samples were from patients with HCMV included in their admission differential diagnosis. Group D comprised nine CSF samples from nine patients with either HCMV retinitis and no other HCMV CNS disease manifestations or other nonneural HCMV disease.

CD4 T-cell counts and CSF parameters. Shown in Table 2 are the mean CD4 T-cell counts, CSF WBC counts, and CSF protein and glucose concentrations for groups A to D and an additional group, E, which comprises a subset of group C CSF samples associated with a definitive diagnosis of HCMV CNS disease based upon a final review of all data. As expected, the mean CD4 T-cell count for patients with nonneural and neural HCMV disease was <100 cells/mm³. Interestingly, one patient with confirmed HCMV retinitis and encephalitis had a CD4 T-cell count of 355 cells/mm³. Group E patients had in general an elevated CSF protein level (mean, 106 mg/dl) compared to patients in group A (mean, 53 mg/dl), group B (mean, 61 mg/dl), and group D (mean, 57 mg/dl). CSF WBC counts for group E (mean, 40 cells/mm³) and group B (mean, 44 cells/mm³) were significantly higher (P = 0.012 and 0.0348 for groups E and B, respectively) than for control group A. CSF glucose levels for group E were within the normal range (mean, 45 mg/dl), with the exception of one patient with a slightly decreased level.

BBB assessment. Using the formula of Tourtellotte et al. (26) we measured the rate of albumin BBB leakage in patients with and without HCMV disease to determine if the BBB is altered and if BBB leakage is a marker of HCMV CNS disease. This formula is based upon the fact that the CSF albumin concentration in a normal individual depends on both the normal escape of albumin almost exclusively through junctions of endothelial cells of the capillaries of the choroid plexus into the brain extracellular space and the blood concentration. A significant abnormal BBB-to-albumin ratio exists or a significant number of capillary endothelial tight junctions are leaking when an individual has an albumin BBB leakage greater than 75 mg/day. The rate of albumin BBB leakage in patients with HCMV CNS disease (n = 13) was significantly elevated (mean, 285.2 mg/day; range, 32 to 1,017 mg/day), with 10 of 13 patients exhibiting albumin leakage greater than the normal range. Similarly, patients with nonneural HCMV disease (n = 9) also demonstrated an increased rate of albumin BBB leakage (mean, 107 mg/day; range, 39 to 124 mg/day). In comparison, patients with ADC (n = 12; mean leakage rate, 64.9 mg/day; range, 14 to 215 mg/day) and patients with non-HCMV related CNS disease (n = 20; mean leakage rate, 33.2 mg/day; range, 3.95 to 261.7 mg/day) generally demonstrated normal albumin BBB leakage rates. Specific clinical exceptions in the non-HCMV related groups which resulted in above-normal albumin BBB leakage rates included Mycobacterium tuberculosis meningitis, other viral meningitis, and lymphoma.

Performance of DNA PCR, NucliSens pp67 assay, and viral culture. HCMV DNA PCR, NucliSens pp67 assay, and viral culture results for CSF groups A to D are shown in Table 1, and the results are summarized in Table 3. The CSF DNA PCR and/or NucliSens pp67 assay results confirmed the original diagnosis for 11 of the 13 positive CSF samples. For the 10 remaining patients in group C the HCMV CNS disease status could not be definitively determined based upon the initial diagnostic criteria and final diagnosis was made after inclusion of the PCR and NucliSens pp67 assay results. HCMV CNS disease was confirmed for one patient based upon positive PCR and NucliSens pp67 assay results and subsequently proven during the autopsy. In an additional patient the final diagnosis of HCMV CNS disease was based on a positive PCR result and confirmed by clinical course and response to anti-HCMV therapy. HCMV blood cultures and/or pp65 antigenemia assays were positive in 11 of 13 patients with CNS disease, and HCMV was detected in the blood by PCR in the remaining two patients. Negative HCMV CSF PCR, CSF NucliSens pp67 assay, and CSF and blood culture results for the remaining eight patients ruled out HCMV CNS disease, and in all cases the neurologic symptoms were attributed to ADC. Inclusion of the PCR and NucliSens pp67 assay results did not bias the final patient classifications and test predictive values since subsequent long-term patient follow-up confirmed the diagnoses of all patients included in or excluded from group E. 

Resolution of discordant results. Resolution of the 10 discordant CSF CMV pp67 assay, DNA PCR, and culture results was based upon clinical, laboratory, and radiographic studies and a review of antiviral therapy. A total of 6 patients with positive CSF DNA PCR results but negative NucliSens pp67 assay and CSF viral culture results did not demonstrate clinical symptoms indicative of active HCMV disease, nor did they progress to active HCMV disease for up to 3 years posttesting (mean, 26.4 months). Blood cultures for HCMV were negative for all six patients. HCMV DNA PCR on blood was positive for three patients, two of whom had lymphoma and decreased BBB integrity as demonstrated by elevated albumin leakage (129.7 mg/dl and 110 mg/dl). Based upon these results, the six positive CSF DNA PCR results were attributed to the detection of latent or defective genomes. Two patients with positive CSF DNA PCR results but negative NucliSens pp67 and culture results were diagnosed with active HCMV CNS disease. The remaining two patients with negative NucliSens pp67 and HCMV culture results but positive DNA PCR results were on anti-HCMV therapy (one on foscarnet for 3 days and one on ganciclovir for 7 days) prior to and at the time of CSF collection.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In our present study we examined the clinical correlation between the presence of HCMV pp67 late gene transcripts in CSF and active HCMV CNS disease. Currently, the NucliSens pp67 assay has been validated and cleared for use with whole-blood specimens and has not been previously described for use with CSF. We felt that both the target and format of this assay had several advantages over the other currently used methods for detecting active HCMV CNS disease, including increased sensitivity over viral culture and increased specificity over DNA PCR. In addition, other studies have assessed the clinical utility of monitoring HCMV immediate early and late transcripts in whole blood for predicting the onset of HCMV infection and for monitoring response to anti-HCMV therapy (4, 13, 19). Studies by Blok et al. demonstrated that the detection of pp67 transcripts in whole blood by the NucliSens assay was very specific (100%) and highly predictive (100%) of the onset of HCMV infection in renal allograft recipients (4). Further studies by Gerna et al. (13) in heart, lung, and bone marrow transplant patients found a similarly high specificity and PPV for the NucliSens pp67 assay (90 to 100%).

Both DNA PCR and the NucliSens pp67 assay were able to detect subtypes 1 to 4 of HCMV, including mixed isolates, with equal efficiency. However, the specificity of detection as related to clinical disease appeared to be directly related to the analytical sensitivity of each assay. As demonstrated by our DNA PCR assay, very sensitive assays will result in an excellent NPV (100%) but a significantly lower PPV (68.4%) for HCMV CNS disease. This is presumably due to the detection of latently infected cells or extracellular DNA fragments in patients without active HCMV disease at the time of testing who did not progress to disease for upward of 3 years posttesting. Our results were consistent with other studies which demonstrated that quantitative PCR (5, 20, 21, 22) or antigenemia assays (25, 28) are preferable to qualitative PCR to improve correlation with clinical disease. Therefore, the PPV value for DNA PCR may have been more correlative with active HCMV CNS disease in our patient population if the assay was quantitative rather than qualitative. Although previous studies clearly indicate that higher levels of HCMV in CSF are correlative with active neural disease, the number of HCMV copies in CSF which define active versus latent CNS infection can be dependent upon several factors including neurologic syndrome, BBB integrity, patient immune status, assay technology, quantitation method, and laboratory variability. One significant benefit of the NucliSens pp67 assay is that positivity correlates with active disease without the need for accurate quantitation of the mRNA copies present.

The diagnostic performance of a molecular amplification assay is determined by two factors: the biological relevance of the presence or levels of the target molecule that is detected and the sensitivity and specificity of the underlying technology to detect that level in a reproducible fashion. The NucliSens pp67 assay has been shown to turn positive only at an infection level that is considered clinically relevant in solid organ and bone marrow transplant patients (13). Low DNA or antigenemia levels in peripheral blood leukocytes were often not associated with detectable levels of pp67 mRNA and did not coincide with clinical symptoms that could be attributed to HCMV. This is the result of the set sensitivity of the assay to detect pp67 mRNA, and the ability of the amplification technology NASBA to detect single-stranded RNA in a double-stranded DNA background (16). Our data suggest that this detection sensitivity and specificity are well suited to diagnose only clinically relevant HCMV, yielding a PPV of 100% and an NPV of 97% for HCMV CNS disease.

The NucliSens pp67 assay confirmed the diagnosis of HCMV CNS disease for nine patients, indicated response to antiviral therapy in two patients, and was instrumental in ruling out HCMV CNS disease in an additional eight patients. The NucliSens pp67 assay result was negative for four CSF samples from three patients with HCMV CNS disease. One patient was not treated for HCMV CNS disease at the time of initial CSF collection (sample 56) because a definitive diagnosis of HCMV disease had not been made (PCR results were not available to the clinician). At that time the CSF DNA PCR was positive and both NucliSens pp67 and CSF culture were negative. However, the patient's symptoms progressed and the patient was readmitted 3 months later with polyradiculopathy and HCMV was again detected in the CSF by DNA PCR (sample 57). However, both CSF culture and NucliSens pp67 assay results were again negative. A review of the patient's chart determined that ganciclovir therapy had been initiated 1 week prior to CSF collection, when HCMV blood cultures became positive. Similarly, an additional patient (PCR positive, culture negative, and NucliSens pp67 assay negative) was diagnosed with HCMV retinitis and encephalitis and was on foscarnet therapy for 3 days prior to and at the time of CSF collection (sample 51). In both cases negative NucliSens pp67 assay results would be expected since HCMV late mRNAs are only transcribed during HCMV replication (10, 11) and transcription should cease upon inactivation of viral polymerase by anti-HCMV agents (17). These data suggest that the NucliSens pp67 assay could be used as a tool to monitor both disease progression and response to antiviral therapy in patients with HCMV CNS disease. Our observation is supported by data from Blok et al. (4) and Gerna et al. (13), who found that the NucliSens pp67 assay, when performed on whole blood, was an effective tool for monitoring initiation and termination of antiviral therapy in renal allograft recipients and in heart, lung, and bone marrow recipients. In both studies the detection of pp67 transcripts was very specific and was highly predictive of the onset of HCMV infection.

The remaining discordant PCR-positive, culture-negative, and pp67 assay-negative CSF results (sample 59) were from a patient admitted to rule out HCMV encephalopathy, ADC, or progressive multifocal leukoencephalopathy (PML). A clinical diagnosis of HCMV encephalopathy was made based upon clinical symptoms, neuroimaging studies, and HCMV-positive blood cultures. The inability to detect pp67 transcripts in this sample and in sample 56 may be directly related to the stage of the disease, the level of pp67 transcripts present, and/or the detection limit of the assay. Additional studies are needed to determine the optimal CSF input volume needed to detect early infection, while still limiting the incidence of detecting latent infection as seen with qualitative DNA PCR. Unfortunately, repeat testing with increased volumes of CSF was not performed on this patient due to lack of sufficient sample. Finally, instability of the pp67 mRNA in CSF samples stored at 80°C could have contributed to the false-negative results for the two samples from the patients with active HCMV CNS disease and positive DNA PCR. Although pp67 mRNA degradation cannot be definitively ruled out, this appears to be unlikely, since we have efficiently detected pp67 mRNA in numerous plasma and CSF samples frozen for up to 5 years at 80°C (data not shown).

In addition to both DNA PCR and NucliSens pp67 assay results, various clinical parameters can also aid in the differential diagnosis of HCMV-related CNS disease. Our study determined that patients with neural HCMV disease had elevated CSF protein and WBC counts and normal glucose levels, all of which were consistent with the results of previously published studies that examined the CSF characteristics from patients diagnosed with HCMV CNS disease (15, 18).

The infection of brain endothelial cells by HCMV may cause a loss of BBB integrity that allows infected cells to travel across the damaged BBB and introduce virus and/or viral proteins into the CNS. As demonstrated in this study patients with both HCMV CNS disease and nonneural HCMV disease generally have increased albumin BBB leakage, indicating that a significant number of capillary endothelial tight junctions are leaking. In addition, altered BBB integrity due to other disease processes could account for the detection of HCMV in latently infected cells in the CSF of patients without active HCMV CNS disease. In this study, two patients, one with M. tuberculosis meningitis and one with lymphoma, had positive CSF HCMV DNA PCR results, negative NucliSens pp67 assay results, negative HCMV blood culture results and no progression to HCMV CNS disease. Both patients had decreased BBB integrity, as indicated by an increase in albumin leakage (261 mg/dl and 130 mg/dl).

In summary, we found the NucliSens CMV pp67 assay to be a highly specific and sensitive method to detect HCMV CNS disease. This test will be quite useful when the diagnosis of HCMV CNS disease in AIDS patients is difficult due to the variety of neurologic manifestations HIV-1 itself causes, coinfections with opportunistic pathogens, and/or malignancies which affect the central nervous system (18). The assay format is simple, results are available in 1 day, and only 200 µl of CSF is sufficient for analysis. In addition, the Boom nucleic acid extraction procedure provides an eluate containing both RNA and DNA (7). This eluate is available for additional nucleic acid testing for the identification of other CSF pathogens. Finally, this method provides an accurate alternative to quantitative PCR testing for the diagnosis of HCMV CNS disease.

Fortunately, since the initiation of this study the incidence of HCMV disease has decreased dramatically as a result of highly active antiretroviral therapy (HAART) and immune reconstitution in patients with successful courses of therapy. However, as patients are challenged with the complications of long-term HAART, including the emergence of multidrug-resistant quasispecies, drug toxicity, compliance failure, and limited therapeutic options, clinical failures and their complications will remain a major concern. Rising HIV-1 viral loads and immune depletion in patients who have experienced failed HAART can lead to the same clinical and diagnostic challenges faced prior to the HAART era. In addition, the availability of HAART is still critically limited on a worldwide level. Therefore, advances in the diagnosis of HCMV disease are still important not only for their application to HIV disease but also for other immunosuppressive conditions due to chemotherapy and transplantation.