Diagnosis of Enterovirus Infection by Polymerase Chain Reaction of Multiple Specimen Types

Enteroviruses (EVs) are the most common cause of aseptic meningitis1 and nonspecific febrile illness2 among children in the United States. The utility of diagnosis of EV infection by viral culture (VC) is limited by a sensitivity of only 65 to 75%, a turnaround time of 3 to 10 days and the high degree of technical expertise required. The PCR has been shown to be an effective alternative to VC in diagnosing EV meningitis using cerebrospinal fluid (CSF) specimens3,4 as well as in diagnosing neonatal EV infections with serum and urine specimens.5 The development of a 5-h colorimetric microwell detection PCR-based kit (AMPLICOR® EV Test; Roche Diagnostic Systems, Branchburg, NJ) has greatly simplified EV diagnosis by PCR and reduced the chances of carryover contamination.6 In a blinded, prospective, multicenter study of children of all ages, we compared the AMPLICOR® EV Test with VC for the diagnosis of EV infections using CSF, serum, throat and urine samples. Methods.Patients. Children seen in emergency rooms at five university teaching hospitals during a single summer season who required blood culture and/or lumbar puncture to rule out sepsis/meningitis were enrolled by obtaining informed consent from their parent or guardian. Patients were not studied further if a bacterial etiology was identified for their illness. Specimens. Specimens of CSF, serum (or nonheparinized plasma) and urine (many collected by sterile catheterization, others by bag method), obtained at the time of admission, were collected from the routine diagnostic laboratories where they had been stored for ≤48 h at 4°C. Patients' throats were swabbed with two cotton-tipped Dacron® swabs; one swab was then immersed in VC transport medium ((veal infusion broth (0.025 g/ml), bovine serum albumin (0.0025 g/ml), gelatin (0.005 g/ml), NaCl (0.00012 g/ml = 0.2 mM), MgCl2·6H2O (10 mM), penicillin (100 units/ml), streptomycin (100 μg/ml) and amphotericin B (2.5 μg/ml); pH was adjusted to 7.2)) and the second swab was immersed in PCR transport medium ((TRIZMA (10 mM), bovine serum albumin (0.005 g/ml), NaCl (0.0085 g/ml = 145 mM), penicillin (100 units/ml), streptomycin (100 μg/ml) and amphotericin B (2.5 μg/ml); pH was adjusted to 7.2)). If no leftover urine sample was available, a new urine specimen was obtained by bag or clean catch method. Specimens for VC were promptly transported to the diagnostic virology laboratory where they were inoculated onto cell lines as described below. Specimens for PCR were frozen at −70°C until testing. Viral culture. Specimens at each study site were inoculated into tubes of human embryonic lung fibroblast cells (MRC-5), Buffalo green monkey kidney cells and human embryonic rhabdomyosarcoma cells; certain of the study centers additionally inoculated specimens onto primary monkey kidney cells. These cultures were maintained at 36-37°C in 5% CO2 and examined daily for 7 days for the presence of viral cytopathic effect. Specimens exhibiting cytopathic effect suggestive of the EVs were passed once in the same cell line for confirmation. PCR. A 100-μl aliquot of each sample was used for the AMPLICOR® EV Test according to manufacturer's instructions and as previously reported6; specimens from all study sites were tested either in Denver (by HAR) or Dallas (by GHM). Briefly an aliquot of specimen was mixed with lysis solution to free viral RNA from infected cells. EV RNA was then precipitated, resuspended in diluent solution and added to an equal volume of a reverse transcription-PCR master mix containing uracil N-glycosylase, biotinylated primers,3,6 deoxynucleotide triphosphates (with dUTP in place of dTTP) and Thermus thermophilus (rTth) DNA polymerase in a bicine buffer. Samples of CSF, serum (or plasma), urine and PCR transport medium (for throat specimens) were processed identically. With each PCR "run," negative control tubes (containing buffer only) and positive control tubes (containing poliovirus RNA) were processed as for tubes containing patient samples. Standard techniques to prevent carryover contamination during PCR were used at all times,7 including designated rooms for different stages of the PCR process. All samples were amplified through 35 cycles of denaturation, annealing and extension in a thermal cycler (Perkin-Elmer PCR System 9600®) on a MicroAmp® base (Perkin-Elmer). Amplification products were detected as previously reported.6 Each base-denatured amplicon was combined with hybridization buffer in a well of a 96-microwell plate that also contained immobilized oligonucleotide probe specific for the EVs. Hybridization was carried out for 1 h at 37°C, and reaction product was detected by avidin-horseradish peroxidase colorimetric changes. The optical density of the wells was read at 450 nm and results were scored as positive if ≥0.35. Serotyping. Virus isolates from culture-positive, PCR-negative samples 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.8 Interpretation. Sensitivity and specificity calculations were based on VC as the standard; i.e. a PCR-negative/VC-positive sample was considered a false-negative PCR and a PCR-positive/VC-negative sample is considered a false-positive PCR. Because PCR may be more sensitive than culture for certain samples, we attempted to "resolve" PCR-positive/VC-negative samples by considering VC and PCR results of other concurrent specimens from the same patient (e.g. a positive VC or PCR from the CSF would confirm a PCR-positive/VC-negative serum as being a true PCR positive). Results. We enrolled 502 children assessed in Emergency Departments with illnesses requiring blood culture and/or lumbar puncture. Approximately two-thirds of the children were ≤1 year old; there was no difference in the performance of the PCR assay (vs. VC) as a function of the patients' ages. From the 502 patients we obtained 214 paired (paired = one aliquot for VC, one aliquot for PCR) samples of CSF, 395 paired blood samples (373 paired sera, 22 paired nonheparinized plasma), 381 paired throat swabs and 397 paired samples of urine, for a total of 1387 paired samples. Results of PCR vs. VC, by specimen type, are shown in Table 1; the sensitivity of PCR ranged from 77% for urine to 100% for CSF.TABLE 1: Comparison of PCR and viral culture for enterovirus diagnosis in paired samples from multiple body sites There were a total of 16 specimens for which VC was positive and PCR was negative (7 urine, 6 throat, 3 blood, 0 CSF). Twelve of the 16 viral isolates from those cultures were available for serotyping analysis. There were 6 echovirus 9 (all from a single study site), 2 echovirus 30 and one each of coxsackievirus B1, coxsackievirus B2, echovirus 22/23 and poliovirus 3 (in a urine sample of a child vaccinated with live poliovirus vaccine 8 days before study). Most PCR-positive/VC-negative samples could be confirmed as true PCR positives by examining VC and PCR results of other concurrent specimens from the same patient. Twenty-five of 30 PCR-positive/VC-negative CSF samples were confirmed as true PCR-positive samples by this method. Of the remaining 5 samples, all had improvement of CSF pleocytosis (89, 160, 283, 367, 624 white blood cells/mm3) without antibiotic therapy beyond 72 h and no other etiology for their meningitis identified; 2 of the 5 patients had no concurrent samples from other body sites available for testing. Sixty-two of 68 PCR-positive/VC-negative blood specimens were resolved as true PCR positives. Of the remaining 6 samples (all sera, not plasma), 2 were from patients from whom only urine VC and PCR were available as potentially corroborating samples. Twenty-three of 52 PCR-positive/VC-negative throat specimens were resolved as true positive PCR results; of the remaining 29 children, 11 had a history of "recent" live oral poliovirus immunization, (range, 1 day to 4 months before study). Finally, of 43 PCR-positive/VC-negative urine specimens, 35 were resolved as true positive PCRs. Of the remaining 8 children 4 had received recent live poliovirus immunizations (14, 24, 30, 54 days before study); we could not determine whether these 8 samples were collected by bag/clean catch or by catheter. Of 76 children known to have received live poliovirus immunizations within the 2 months preceding study enrollment, 47 (62%) were negative for enteroviruses by culture and PCR at all body sites tested. Discussion. The EVs are estimated to cause between 10 and 30 million infections in the US annually and can be difficult to distinguish from bacterial infections on clinical grounds alone. Diagnosis of EV infection has traditionally depended on VC methods that are time-consuming and of limited sensitivity. In this study a 5-h PCR-based kit for EV diagnosis was evaluated in children undergoing blood culture and/or lumbar puncture to rule out bacterial disease. This is the largest study to date of PCR in EV diagnosis, the first to evaluate specimens from multiple body sites in the same patients and the first to evaluate the AMPLICOR® EV PCR kit in serum and throat samples from children of all ages. The sensitivity of the PCR kit (vs. VC) for EV diagnosis was greatest for CSF (100%) and lowest for urine (77%), consistent with previous, smaller studies.5,6,9 Specimens that were VC-positive/PCR-negative were not the result of infections with EVs not recognized by the PCR assay; the serotypes of these EV isolates are all expected to be detectable with the current assay except echovirus 22/23.6 Serum was demonstrated to be an excellent specimen for PCR diagnosis of EV infection, with high sensitivity and high resolved specificity vs. VC. Throat swab PCR detected 95% of VC-positive EV throat isolates; however, many VC-negative throat specimens were positive by EV PCR and could not be confirmed as true positives with concurrent samples. Increased and prolonged shedding from the throat after systemic EV infection may explain this observation; all of the children in this study were ill enough to require blood culture and/or lumbar puncture and did not have documented bacterial infection. This PCR assay cannot distinguish polioviruses from the non-polio EVs; hence recent oral polio immunization may also explain some of the disparities noted. PCR positivity may outlast VC positivity after immunization (and other, potentially corroborating body sites are unlikely to be positive by either VC or PCR many days to weeks postimmunization). Potential time and cost savings would accrue from a rapid EV diagnostic technique.10,11 The AMPLICOR® EV test kit is a 5-h, accurate and convenient method for diagnosing EV infections and can be applied to specimens from multiple body sites. Acknowledgments. This study was supported by grants from Schering-Plough Research Institute, Kenilworth, NJ, and Roche Diagnostic Systems, Branchburg, NJ. We thank Neva Murphy and Kurt Olsen for excellent technical assistance. Harley A. Rotbart, M.D. Amina Ahmed, M.D. Sheila Hickey, M.D. Ron Dagan, M.D. George H. McCracken Jr., M.D. Richard J. Whitley, M.D. John F. Modlin, M.D. Marianne Cascino, R.N. John F. O'Connell, Ph.D. Marilyn A. Menegus, Ph.D. Deborah Blum, M.D. Departments of Pediatrics and Microbiology; University of Colorado School of Medicine; Denver, CO (HAR) Department of Pediatrics; University of Texas Southwestern Medical Center; Dallas, TX (AA, SH, GHM) Pediatric Infectious Disease Unit; Soroka University Medical Center; Beer-Sheva, Israel (RD) Departments of Pediatrics, Microbiology and Medicine; University of Alabama; Birmingham, AL (RJW) Departments of Pediatrics and Medicine; Dartmouth Medical School; Dartmouth, NH (JFM) Departments of Clinical Immunology and Infectious Disease Research and Antiviral Chemotherapy; Schering-Plough Research Institute; Kenilworth, NJ (MC, JFO, DB) Department of Microbiology and Immunology; University of Rochester Medical Center; Rochester, NY (MAM)