INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Enteroviruses (EVs) are the most frequent etiologic agents of aseptic meningitis and are estimated to be the cause of 70 to 90% of cases of viral meningitis (4, 23). The clinical features associated with EV infections of the central nervous system (CNS) are often indistinguishable from those of other infections. For optimal patient management (avoidance of unnecessary hospitalization and presumptive treatment of the patient), a rapid and specific method for the diagnosis of acute EV infection is required (24).

The current method of choice for the diagnosis of EV infections is still isolation of the virus by cell culture with several cell lines which are examined for the development of a cytopathic effect during a 10- to 14-day incubation period. However, viral culture does have a limited sensitivity and some serotypes do not grow in cell culture (8). The laboratory diagnosis may also be based on serology, i.e., the detection of an antibody titer increase between acute- and convalescent-phase serum specimens or by the detection of specific immunoglobulin M antibody (9). However, the serological diagnosis of EV infection is complicated due to the large number of EV serotypes, and therefore, serological diagnosis has only a limited role in diagnostic investigations. Furthermore, serology is not suitable for the early, rapid diagnosis of enteroviral infections with the possible exception of poliomyelitis, in which an immunoglobulin M response is detectable in the acute phase (18).

Recent developments in molecular biology have enabled the detection of EV genomes in various clinical samples by molecular amplification methods such as PCR (1, 2, 6, 10, 11, 17, 19, 22-25, 28). However, the application of an in-house-developed PCR assay for routine diagnostic investigations is often limited by time-consuming procedures for sample preparation and by the lack of standardization. Commercial amplification test systems may present attractive alternatives to circumventing these problems.

Within the framework of the European Union Concerted Action on Virus Meningitis and Encephalitis, we carried out a multicenter study to evaluate the diagnostic performance of the Amplicor EV PCR test (Roche Diagnostics, Branchburg, N.J.) for the detection of EVs in cerebrospinal fluid (CSF) specimens from patients with aseptic meningitis. The results obtained by PCR analysis and culture were compared to assess the sensitivity and specificity of the Amplicor EV PCR test. The study included 476 CSF specimens collected from nine different laboratories in five European countries. To our knowledge this is the first time that a clinical evaluation of an EV PCR assay has been performed with such a large number of CSF specimens obtained from different countries. The present evaluation enabled the detection of a large variety of EV serotypes and epidemiologically unrelated isolates.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Study design. The samples tested in the study consisted of 476 CSF specimens which were collected at nine different centers in Europe. The CSF samples were examined by viral culture at the participating laboratories within 3 days of the time of collection and were subsequently stored for PCR analysis. Viral culture was performed by the standard methods used by the various laboratories. Detection of enteroviral RNA by the Amplicor EV PCR test was performed at the coordination site, the Department of Virology, University Hospital Utrecht. The CSF specimens were shipped on dry ice to the coordinating laboratory over a period of 10 months (1995 to 1996). Upon arrival the CSF specimens were aliquoted and PCR analysis of the 476 samples was performed in a total of 30 independent runs.

Sample and patient selection. CSF samples were collected from patients with symptoms suggestive of aseptic meningitis. After viral culture the CSF samples were stored at 20 to 70°C. Almost 47% of the CSF specimens had been stored for less than 1 year; of these, 27.3% had been stored for less than half a year. The remainder of the samples had been stored for from 1 to 4 years at 70°C. A collection of CSF specimens (n = 29) from patients with confirmed poliomyelitis collected during the 1992-1993 poliomyelitis outbreak in The Netherlands was also included in the study. Throat and/or rectal swabs or stool specimens collected for routine diagnostic investigations were stored at some centers for discrepancy analysis. The patients' records were reviewed, and clinical data including the patient's history, signs, and symptoms and the results of laboratory examination of CSF (such as specimen appearance, bacterial, fungal, and non-EV viral culture; total leukocyte count; and total protein and glucose concentrations) were provided by the participating centers. CSF pleocytosis was defined as a leukocyte count of 10/mm³.

Detection of EV by culture. At each center, the CSF specimens were inoculated onto three cell lines (including human diploid fibroblasts and primary or tertiary monkey kidney cells) which were observed for the development of a cytopathic effect characteristic of EVs. Cultures were held for at least 2 weeks. Cultures became positive for the majority of specimens after 2 to 6 days. Typing of the virus isolates was carried out by neutralization or complement fixation with intersecting antiserum pools by standard procedures. To assess the ability of the participating laboratories with regard to EV culture and typing, a proficiency panel of EVs was prepared. The proficiency test included 10 specimens, of which 8 contained at least one type of EV. Correct virus isolation results were obtained by the participating centers for more than 98% of the samples (26).

Detection of EV by the Amplicor EV PCR test. The Amplicor EV PCR test (Roche Diagnostics) was performed in accordance with the instructions of the manufacturer. In the assay the reverse transcription and molecular amplification steps are combined through the use of the thermostable enzyme rTth (recombinant Thermus thermophilus) polymerase. Biotinylated EV-specific primers are located at the 5' noncoding region of the EV genome and are used for the amplification and detection of the amplified products. Detection is performed colorimetrically on a microwell plate with an immobilized oligonucleotide probe specific for enteroviruses (22). According to the manufacturer the Amplicor EV PCR test detects most of the 66 different serotypes of EV at a sensitivity of 1 50% tissue culture infective dose (TCID₅₀) (14). Three serotypes, coxsackievirus A2 (CA2), CA4, and CA8, have not been tested. As expected, echovirus 22 (EC22) and EC23, whose sequences are not similar to those of other EVs, cannot be detected (13).

Discrepancy analysis. The remainder of the previously amplified sample was retested for samples that initially had discordant results by PCR in the detection part of the assay. However, if discordant PCR results for the duplicate samples persisted, PCR testing was repeated with another aliquot of the original, unprocessed CSF sample. Discrepancy analysis for samples with discrepant results by PCR and culture was performed by Roche Molecular Systems. Samples were retested by Amplicor PCR, followed by PCR testing with an alternate primer pair-probe set also located at the 5' noncoding region. PCR results were finally clarified on the basis of the number of nondiscordant OD values for the duplicates in the repeated PCR assays. When at least three of the four duplicates or at least four of the six duplicates showed a positive or a negative PCR result, the sample was classified PCR positive or PCR negative, respectively. Statistical comparisons of the PCR and culture results were performed by chi-square analysis.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Patient and specimen characteristics. Samples from a total of 476 patients living in five different European countries were included in the study. All patient groups (neonates, infants, children, adolescents, and adults) were represented. The patients' ages ranged from 2 days to 89 years. The age distribution was as follows: 0 to 5 months, 66 patients; 6 to 11 months; 17 patients; 1 to 4 years, 64 patients; 5 to 14 years, 100 patients; 15 to 24 years, 55 patients; 25 to 59 years, 110 patients; older than 60 years, 33 patients. The age was not known for 31 patients. The median age of the population was between 5 and 14 years. There were no major differences in the age and sex distributions of the patient populations between the different centers and countries.

Comparison of Amplicor PCR and culture. Table 2 provides a comparison of viral culture and initial Amplicor EV PCR test results. Among the 50 CSF specimens that were culture positive, 38 (76%) had an initial PCR-positive result by the initial PCR; 12 samples (24%), however, were PCR negative. In contrast, no EV could be isolated from 28 (42%) of the 66 specimens that were PCR positive. The remaining 398 specimens were negative by the Amplicor EV PCR test and culture.

Amplicor EV PCR test detection of EV serotypes. Forty-five of the 50 EV isolates were typed and yielded 15 different EV serotypes. Table 4 presents the final PCR results for the different EV serotypes. The most common serotype was EC30, which was isolated from 30% of the specimens and which originated from two different sites: Sweden (16 patients) and Oxford, United Kingdom (2 patients). The samples from 13 of the Swedish patients were obtained over a period of 3 months, indicating that these isolates are probably epidemiologically related. The 14 other EV serotypes isolated were equally distributed over the nine different centers, indicating that none of the other EV serotypes could be placed in the context of a single outbreak. All CSF specimens from which an EC30 strain had been isolated were PCR positive. At least one sample containing each of the other EV serotypes except EC6 and CB4 was positive by PCR. A poliovirus type 3-positive CSF sample which was PCR positive was also included in the study. This was an archival CSF sample that had been collected during the Dutch poliovirus epidemic of 1992 and 1993 (18). No poliovirus could be detected by either viral culture or PCR in the CSF specimens from the other 28 poliomyelitis patients involved in the same epidemic.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our study demonstrates the value of the Amplicor EV PCR test for the diagnosis of EV infection of the CNS. The multicenter character of the study made it possible to evaluate the test on a large-scale basis; 476 patients with suspected enteroviral meningitis were examined. Most other studies with large numbers of samples were derived from community outbreaks involving one or a small number of EV serotypes (12, 23, 27). The aim of the present study was to evaluate the diagnostic performance of the Amplicor EV PCR test with a heterologous population of EV serotypes. Because detection of EVs by PCR for diagnostic purposes relies on the conservation of parts of the nucleic acid sequences between strains, it is important to demonstrate that the test broadly reacts with different serotypes circulating in the community and with epidemiologically unrelated isolates of a single serotype.

Viral cultures of CSF specimens from patients with CNS disease, mostly aseptic meningitis, were performed at the various participating laboratories, whereas all PCR analyses were performed at one site, the coordination site. Our experience with the Amplicor EV PCR test with regard to implementation in the diagnostic laboratory is similar to that of others; it is easy to perform and yields rapid results (15, 22, 27). There is a clear-cut separation between the OD readings for positive and negative samples. The discordance rate between the duplicate samples in the amplification and detection phases of the assay was low (0.6%). Duplicate tests therefore seem to be unnecessary. However, because of the design of the study we can make no statement about the discordance rate for independently processed sample aliquots. We encountered problems with the positive controls included in the test kit during the study period. Five of the 35 runs (comprising four batches) had to be repeated because positive controls did not meet acceptable criteria. The manufacturer of the test kit has been made aware of this problem and has informed us that the positive controls have been modified and are now more stable during storage.

The sensitivity of the Amplicor EV PCR test compared to the results of culture was initially 76% when viral culture was used as the gold standard. A total of 12 specimens were initially classified as negative by PCR analysis, whereas they were culture positive. Five samples, however, had PCR-positive results upon repeat testing. Since these samples with discrepant results had low ODs (OD values ranged between 1.2 and 3.0 times the cutoff value), we believe that these CSF specimens had low amounts of viral RNA and that these amounts were at the threshold of the sensitivity of the assay, which may have resulted in the difference in results between initial and repeat tests. The low amount of viral RNA is probably due to the degradation of RNA during storage after viral culture. The positive result by repeat testing by the Amplicor EV PCR test with another aliquot of the original, unprocessed CSF specimens, in contrast to the result obtained in the initial test, underlines the impact of sample extraction on the detection of EV in samples with low viral loads. In a recent evaluation of the Amplicor EV PCR test by different laboratories, it was shown that the variability in sensitivity is markedly influenced by the viral load of the sample (15), which is also partly related to the Poisson distribution of the viral particle within different aliquots of the same specimen. Seven of the 49 culture-positive samples were classified as PCR negative after discrepancy analysis. The failure of PCR to detect the culture-positive samples may be explained by storage conditions which resulted in degradation of enteroviral RNA and/or the presence of amplification inhibitors. Since these specimens were archival samples, we cannot ensure that handling and storage had always been adequate. No internal control is used in the Amplicor EV PCR test, so the possibility that amplification inhibitors are present during analysis cannot be excluded. Unfortunately, the amount of sample available was not sufficient for testing for inhibitors. The final sensitivity of the Amplicor EV PCR test was 86% compared to that of culture. This result is similar to those of other studies, in which the degree of sensitivity ranged from 60 to 98%, depending on the viral load in the sample (15, 22). A major difference with other evaluations is that the earlier studies were prospective studies, whereas most of the samples in the present evaluation were examined retrospectively. A recent study has shown that the sensitivity of the PCR analysis increased significantly when freshly collected or recently stored specimens were tested (20).

The specificity of the Amplicor EV PCR test obtained in this study was 94%, which is similar to the specificities of 95 and 98% reported by others (15, 22). Twenty-eight samples were initially PCR positive and culture negative. The initial PCR-positive result for 26 samples could be confirmed by Amplicor EV PCR test analysis or by the alternate PCR. Four samples appeared to have discordant results between the initial PCR analysis and repeat Amplicor EV PCR testing, with clearly distinct OD values between the independent runs. We have no explanation for this observation other than interlaboratory variability in the performance of the Amplicor EV PCR test, possibly during sample extraction.

As with many evaluations of PCR assays, viral culture is not the ideal gold standard for use in the evaluation of the specificity of a PCR test. It cannot be proven absolutely that samples with culture-negative and Amplicor EV PCR test-positive results are false positive. Since the PCR-positive results were confirmed once or twice at a different laboratory, we believe that these specimens were from patients with true enteroviral infections. The increased sensitivity of PCR over that of virus culture may be explained in several ways. First, the technical sensitivity of PCR has been shown to be below that of viral culture and has been reported to be as low as 0.01 TCID₅₀ for EVs (15, 16). Second, it is well known that some EVs do not grow in standard cell cultures (7). Unfortunately, no additional specimens from other body sites had been collected from these patients. Those specimens could have provided additional evidence of enteroviral infection. During the discrepancy analysis for the 40 samples, the results for 3 samples had to be classified as inconclusive, and the samples were excluded from the comparison of culture and PCR since no aliquots were available for final classification, eventually resulting in a total of 473 samples for final analysis.

The EVs isolated in this study are the most common EV serotypes which have been reported consistently in association with aseptic meningitis and encephalitis (21). EV serotypes other than those isolated in our study are reported to be less common causes of CNS disease. All 15 EV serotypes which were found in our study, including the EV serotypes isolated from the PCR-negative CSF specimens, have been reported to be detectable by the Amplicor EV PCR test (14). With our own collection of EV serotypes isolated from fecal specimens, we could demonstrate that a positive signal in the Amplicor EV PCR test was obtained with serotypes EC6 and CB4, which initially could not be detected in the CSF specimens. Examination of other EV serotypes from our own collection, including CA4, CA16, CB1, CB4, EC3, EC6, EC7, EC18, EC19, EC21, and EC31, demonstrated that the Amplicor EV PCR test is able to detect a wide range of the most common serotypes (data not shown). The performance characteristics of the Amplicor EV PCR test were further evaluated by participation in the quality assurance program organized by the European Concerted Action on Virus Meningitis and Encephalitis (16). All EV serotypes included in this panel except EC16 and EC22 could be detected. An EC16 isolate from our own collection did, however, give a positive result in the Amplicor EV PCR test. Studies have shown that EC16 fails to react with some primers, indicating some possible sequence divergence of EC16 compared to the sequences of the other EV serotypes (3). According to the manufacturer, the Amplicor EV PCR test does not detect serotypes EC22 and EC23. It is well known that the sequences of EC22 and EC23 have very low levels of similarity to those of other EV serotypes, and it is therefore disputed whether they belong to the genus of the human EV types (13). In addition, the enterovirus PCR quality control panel also included various different clinical samples. For all non-CSF specimens nucleic acid extraction was performed by the method of Boom et al. (5), as described in Materials and Methods. By combining the nucleic acid extraction procedure and the Amplicor EV PCR test, EV serotypes could be detected in stool filtrates, tissue culture medium, and throat specimens at a dilution of up to 1.0 TCID₅₀ (data not shown) (16).

CSF specimens collected from 29 patients confirmed to have poliomyelitis during the poliovirus type 3 outbreak in The Netherlands in 1992 and 1993 were included in the study (18). With only one exception, all CSF specimens remained negative by PCR as well as by culture. In contrast, poliovirus type 3 had been isolated from the fecal samples of all of these patients (18). Thus, fecal samples from poliomyelitis patients appear to have a much higher titer of poliovirus than CSF specimens, demonstrating the need for the collection of fecal samples for a reliable laboratory diagnosis of poliomyelitis. It is remarkable that the patient whose CSF was positive for poliotype type 3 by culture and PCR was one of the few patients with meningitis.

Our data do not support the diagnostic relevance of feces examination for the laboratory diagnosis of aseptic meningitis. An EV was detected by culture or PCR in only 6 and 4 patients, respectively, of a total of 26 patients with an EV-positive fecal specimen.

EVs have been estimated to account for up to 80% of aseptic meningitis cases. However, as has been shown in previous studies, the proportion of cases without an etiology remains high (29 to 54%) (4). Evaluation of the medical records in the present study also revealed a high rate of etiologically undiagnosed cases of infection. In our study an etiology was not identified from the records of 270 of the 355 patients (76%) examined. In contrast to our expectations, the average rate of EV detection in our study was low: 10.5% by culture for EV and 14.6% by PCR analysis after final testing. This is in contrast to the results of an earlier study, in which the rates of detection of EV by the Amplicor EV PCR and viral culture were 66 and 34%, respectively (27). That study, however, was of a geographically concentrated outbreak among children, and samples were collected prospectively. It is remarkable that in the present study the rate of EV detection between the laboratories varied from 3.8 to 31.6%, which might be explained by the different patients seen at the various laboratories participating in the study. There were no major differences in the age distributions of the patients from the various laboratories.

In summary, we conclude that the Amplicor EV PCR test is a rapid and reliable assay that represents an important development in the diagnosis of aseptic meningitis, particularly since it has a higher sensitivity than viral culture. The PCR test therefore offers an attractive alternative to virus isolation for routine diagnostic investigations. We must keep in mind that, so far, the EVs detected by PCR cannot be serotyped, and although typing of EV strains causing meningitis is usually of less importance for clinical diagnostic considerations, it is essential for understanding the epidemiology of enteroviral infections. The value of the Amplicor EV PCR test lies mainly in the fact that it has a higher sensitivity and it provides a result in 1 day, which is considerably sooner than viral isolation by culture (4 to 5 days) and therefore a benefit for patient care and patient management.