Detection of Enteroviruses by Polymerase Chain Reaction in Urine Samples of Patients With Aseptic Meningitis

Enteroviruses (EVs) are the most common cause of aseptic meningitis (AM) in the US, causing as many as 75 000 cases each year; more than 90% of all cases of AM for which an etiologic agent is identified are caused by the EVs.1, 2 PCR analysis of cerebrospinal fluid (CSF) specimens has confirmed the EVs as causative of approximately two-thirds of cases of AM for which an infectious agent has not been identified by other means.3, 4 The ease with which urine samples can be obtained from ill children makes such specimens ideal for rapid diagnosis. Recently urine was shown to be a useful specimen for PCR diagnosis of EV infections in newborns <2 weeks of age.5 We describe the results of EV PCR testing of urine from children with AM. Materials and methods.Patients and specimens. Children were prospectively enrolled between 1986 and 1990 according to criteria previously described.6 Although the method of collection of urine samples was not recorded, the vast majority were by sterile catheterization during sepsis work-ups (JFM, personal observation). Urine samples were aliquoted and frozen at -70°C until testing. CSF specimens were collected by lumbar puncture from all children; most also had a throat swab and a stool specimen or rectal swab for viral culture. Viral cultures. Urine specimens for culture were clarified by centrifugation at 3000 × g at 4°C for 20 min, and the supernatants were inoculated onto each of three continuous cell lines selected for their support of EV replication, as previously reported.6 Throat and stool/rectal cultures were similarly centrifuged and cultured for EVs; CSF specimens were cultured without centrifugation.6 Virus isolates were serotyped by use of the Lim Benyesh-Melnick equine antiserum pools supplied by the World Health Organization Collaborating Center for Virus Reference and Research.7 Polymerase chain reaction. Primers and probe: Anti-sense primer EV1b, sense primer EV2b and probe EV3 were used for the PCR and detection; these oligonucleotides are derived from the conserved sequences of the 5′ end of the EV genome, as previously reported.8 EV1b and EV2b are biotin-labeled; the EV3 oligonucleotide is bound, unlabeled, to the bottom of microwells for the detection step. Reverse transcription-PCR: A 100-μl aliquot of each urine sample was processed as previously described for CSF samples.8 Steps included lysis, RNA precipitation, washes, uracil N-glycosylase treatment to prevent carryover contamination and single tube/single enzyme reverse transcription-PCR.8 From 3 to 8 negative control tubes (containing buffer only) and 1 positive control tube (containing poliovirus RNA) were processed identically with tubes containing samples. All samples were amplified in a thermal cycler (Perkin-Elmer PCR System 9600) on a MicroAmp base (Perkin-Elmer) according to the protocol previously reported.8 Standard techniques to prevent carryover contamination during PCR were used at all times,9 including designated rooms for different stages of the PCR process. Detection of amplification products: Each base-denatured amplicon was combined with hybridization buffer in a well of a 96-microwell plate containing EV3. Hybridization was carried out for 1 h at 37°C, after which the microwells were washed to remove unhybridized material.8 Detection was via avidin-horseradish peroxidase conjugate and substrate.8 The optical density of the wells was read on a plate reader at 450 nm and results were scored as positive if the OD was ≥0.35. Testing for inhibitors. Separate aliquots of each PCR-positive/culture-negative sample were supplemented with two different concentrations of poliovirus type 1, 101 TCID50 or 100 TCID50; a third aliquot was run without supplemented virus. The PCR methods were as described above. Results. Urine samples were prospectively collected from 155 children with AM enrolled during a 5-year period6; 101 patients were confirmed to be infected with an EV by a positive viral culture at one or more body sites (CSF, throat, stool/rectal and/or urine). Forty-three (43%) of the confirmed EV-infected patients had positive PCR assays of urine samples for EVs; this compared with 37 (37%) EV-infected children with urine viral cultures positive for EVs. Applying the same stringent sensitivity criteria to CSF viral culture, that latter test was positive in only 47 of 95 AM patients with proven EV infection (i.e. viral culture from any site positive for EV), for a sensitivity of only 49%. Eight urine samples were PCR-positive for EVs without confirmatory culture results at any body site. Assuming that those samples were all false-positive PCR results, the specificity of urine PCR for detecting EV-infected children was 85%. The sensitivity of urine EV PCR in infants ≤12 weeks of age was 47% with specificity of 85%; by definition the specificity of urine cultures detecting EV-infected babies was 100%, but the sensitivity was 39% in babies ≤12 weeks of age. We separately analyzed the results from 2 time periods within the 5-year study to determine the effect of prolonged storage on PCR sensitivity. Sensitivity of urine PCR for detecting EV infection (as proved by a positive culture at any body site) was 35% for specimens collected during 1986 and 1987, vs. 57% for specimens collected during 1989 through 1990 (Table 1). The sensitivity of urine PCR for detecting EV infection in babies ≤12 weeks of age among specimens collected during the later time period (1989 through 1990) was 65% (Table 1). Comparison between urine EV PCR testing and cultures from various body sites is shown in Table 2. Most patients with culture-positive rectal swabs for EVs were negative by urine EV PCR (34 of 62), and 16 of 44 patients (36%) with positive urine EV PCRs were rectal swab culture-negative (Table 2). These results suggest that the likelihood of fecal contamination explaining the urine results is low, as expected with sterile catheterization technique. Urine EV PCR was most sensitive in detecting positive urine cultures (62%), however, 14 culture-positive urine specimens were negative by PCR (Table 2). The serotypes “missed” by PCR (9 coxsackieviruses B, 4 echoviruses, 1 poliovirus) reflected the overall distribution of serotypes found in the infected patients.6 Adequate volume remained for inhibitor testing of 12 of those 14 urine samples; 5 showed evidence of inhibitory substances, with lower OD readings than anticipated at the dilutions of poliovirus added (data available on request). An aliquot of each of the culture-positive, PCR-negative urine samples was also retested by PCR without supplemented virus; 5 of the 12 were positive on repeat testing, 3 of which were at low OD values suggesting that the urine sample contained a quantity of virus at the threshold of sensitivity for the test (between 100 and 101 TCID50). Only 1 false negative urine PCR sample overlapped the two groups, i.e. was positive on retesting without supplemented virus and had evidence of inhibitory substances (data available on request). Discussion. Urine specimens in this study had been stored for 4 to 8 years before PCR testing; nevertheless PCR proved to be more sensitive than the viral cultures of the same urine samples which had been performed several years previously. The 65% sensitivity observed in our patients ≤12 weeks of age whose specimens had been collected during the later period is comparable to the 62.5% sensitivity observed in a study of stored specimens from ≤2-week-old babies.5 Eight PCR-positive urine samples could not be confirmed by cultures of other body sites in the same children; the sensitivity, specificity and predictive values assume these PCR results to be falsely positive. The assay used in this study is quite resistant to carryover contamination.5, 8 with specificity reported as nearly 100%.5, 8 Hence the 8 “falsepositive” samples in our study may have actually been true infections; those specimens, as is true for all specimens in this study, were from children with AM. False negative results may also occur with any PCR assay, in some cases due to inhibitory substances which interfere with the function of the enzymes used in PCR. Nearly one-half of our retested samples had evidence of inhibitory substances which might explain the false negative results; retesting the other falsenegative samples suggested that the titer of virus was at or near the threshold of sensitivity for the assay. In summary the sensitivity of urine PCR for detecting EV infection in children presenting with AM ranges from 35 to 65% depending on the age of the child and the duration of storage. The specificity of the assay, assuming that all PCR-positive, culture-negative specimens are false-positive PCRs, is 70 to 90%. The sensitivity might be increased by testing freshly collected or recently stored specimens, by testing duplicate aliquots of the same specimen and by taking steps to remove inhibitory substances. Urine PCR for EVs will not replace CSF culture or CSF PCR in making the diagnosis of EV meningitis. However, urine PCR is more sensitive than urine culture for EVs and may well supplement the results obtained by other means and from other specimens. Acknowledgments. The authors gratefully acknowledge Neva Murphy for technical assistance. This study was supported by a grant from Roche Molecular Systems, Branchburg, NJ. Lars P. Nielsen, M.D.; John F. Modlin, M.D.; Harley A. Rotbart, M.D.