Comparing Molecular Methods for Early Detection and Serotyping of Enteroviruses in Throat Swabs of Pediatric Patients

Background Enteroviruses include over 100 serotypes and usually cause self-limited infections with non-specific symptoms in children, with the exceptions of polioviruses and enterovirus 71 which frequently cause neurologic complications. Therefore, early detection and serotyping of enteroviruses are critical in clinical management and disease surveillance. Traditional methods for detection and serotyping of enteroviruses are virus isolation and immunofluorescence assay, which are time-consuming. In this study, we compare virus isolation and two molecular tests for detection and serotyping of enteroviruses in clinical samples. Methods One hundred and ten throat swabs were collected from pediatric outpatients with enterovirus-like illnesses (hand-foot-mouth disease, herpangina, and non-specific febrile illness). Virus isolation was conducted using multiple cell lines and isolated viruses were serotyped using immunofluorescent assay. In the molecular tests, a semi-nested RT-PCR and a novel CODEHOP platform were used to detect the 5′UTR and VP1 genes of enteroviruses, respectively. Amplified nucleotides were sequenced and genotyped. Results Among the 110 cases, 39(35%), 52(47%), and 46(42%) were tested positive with these three tests, respectively. Using the consensus results of these three tests as the gold standard, agreement of the VP1 CODEHOP test was 96%, which is higher than those of the virus isolation (89%) and the 5′-UTR test (88%). The VP1 CODEHOP test also has the best performance on serotyping confirmed with serum neutralization tests. Conclusions The VP1 CODEHOP test performed well for detection and serotyping of enteroviruses in clinical specimens and could reduce unnecessary hospitalization cares during enterovirus seasons.


Introduction
Human enteroviruses include over 100 serotypes and can be divided into four species using molecular typing [1]. With the exceptions of polioviruses and enterovirus 71 (EV71), which frequently cause neurological complications, human enteroviruses usually cause self-limited infections in children [2,3]. Therefore, early detection and serotyping of enteroviruses infections are critical in clinical management and disease surveillance. The traditional standard methods for detection and serotyping of enterovirus infections are virus isolation and immunofluorescence assay (IFA), which are time-consuming and labor-intensive [4,5]. Several clinical studies have documented that molecular diagnosis based on polymerase chain reaction (PCR) is time-saving and more sensitive than virus isolation for detection of enterovirus infections in hospitalized patients [6][7] [8][9][10] but few studies have been conducted in outpatients. Moreover, no study has compared molecular tests and virus isolation/IFA for serotyping of human enteroviruses using clinical specimens. Although these methods had been used to detect enteroviruses in clinical specimens including throat swabs, stool samples and cerebrospinal fluid (CSF) [5][6][7][9][10][11][12], reports elucidating the comparison among the diagnostic approaches are limited.
Molecular tests for the detection of human enteroviruses in clinical specimens usually target highly conserved sites in the 59 untranslated region (59UTR) [13]. Due to low virus titers in clinical specimens, several reverse transcription (RT)-nested or RT-seminested PCR (RT-snPCR) have been developed to further increase its sensitivity and specificity [14][15][16]. However, serotyping of enteroviruses based on 59UTR sequences directly amplified from clinical specimens have not been well-evaluated. In addition, enterovirus VP1 capsid gene has recently been proposed to be an ideal target for detection and serotyping of enteroviruses using the consensus degenerate hybrid oligonucleotide primer (CODEHOP) but this molecular method has not been well-evaluated in clinical specimens [17]. In this study, we compare the traditional methods (virus isolation/IFA) and these two molecular tests for detection and serotyping of human enteroviruses in throat swabs of pediatric outpatients.

Ethics Statement
This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (CGMH) following the Helsinki Declaration; and written informed consents were obtained from all mothers of participating infants.

Clinical Specimens
Previous studies have shown that throat swabs are the most sensitive clinical specimens for isolation of enterovirus 71 [18][19][20]. Clinical specimens were collected from a child cohort study conducted in northern Taiwan [21][22][23]. In the cohort study, sera were obtained from participating children in the following schedule: neonates at birth (cord blood), and children at 6, 12, 24, and 36 months of age. If the participating children developed suspected enterovirus illnesses (herpangina, hand-foot-mouth disease, and non-specific febrile illness), throat swabs and sera were collected from these participating children. In 2008-09, 153 suspected cases were detected in the study cohort and 118 of them provided throat swabs. Among the 118 cases providing throat swabs, 110 had completed virus isolation and molecular tests and 94 of them also provided paired sera samples.

Viral RNA Extraction and Semi-Nested RT-PCR
Viral RNA was extracted from the clinical specimens using a QIAamp Mini Viral RNA Extraction Kit (Qiagen, Germany). A RT-PCR procedure for each amplification was performed in a 50 ml reaction mixture with 5 ml of extracted RNA template and the Access Ouick TM RT-PCR system (Promega, Wisconsin, USA). The highly conserved 59UTR of the enteroviruses was chosen as the target for the synthesis of a 440-bp cDNA with primers EV-F1 and EV-R1 (Table 1). The cycling conditions were as follows: 45 min at 45uC for the RT step, 2 min at 95uC for the initial denature step, 40 cycles for 30 s at 95uC for denature, 30 s at 52uC for annealing, 1 min at 72uC for extension, and a final extension at 72uC for 10 min. Negative controls (sterile water) were included in each amplification series. To improve sensitivity and specificity for identifying enterovirus, a semi-nested PCR system was applied (Table 1) [24]. Semi-nested PCR was performed with 5 ml of each RT-PCR product added to a new PCR tube with a solution containing 10x PCR buffer 5.0 ml (20 mmol/l MgCl 2 ), dNTP mix 2.0 ml (10 mmol/l of each dNTP), EV-F1/EV-R1N 1.0 ml (10 mmol/l for seminested PCR), and VioTaq DNA polymerase 0.5 ml (5 units/ml), (Viogene, Taiwan). Sterile water was added to a final volume of 50 ml. The following temperature program was used: 40 cycles consisting of 95uC for 30 s, 56uC for 30 s, 72uC for 1 min and a final extension at 72uC for 10 min. The semi-nested primers, EV-F1 and EV-R1N, yielded a 400-bp amplicon. Negative controls (sterile water) were included in each amplification series. One EV71 and one CA16 isolate were used as positive controls of RT-PCR for enterovirus detection.

PCR Amplification and Serotyping Using CODEHOP
EV VP1 gene sequences were amplified by a recently described CODEHOP protocol leading to product of 350,400 bp [17,25]. In brief, virus RNA was extracted from clinical specimens using a QIAamp Mini Viral RNA Extraction Kit (Qiagen, Germany) and cDNA was generated using four different primers (AN32-35). The cDNA was then used in the first PCR with primers AN88 and AN89. Product of the first PCR was added to a second PCR with primers 222 and 224 for nested amplification (Table 1).

Sequence Analysis
The amplified DNA was sequenced using the ABI 3730 XL DNA Analyzer (Applied Biosystem Inc., Foster City, CA). Nucleotide sequences of 59UTR or VP1 were checked against the NCBI database by the BLAST search to find the enterovirus serotype with the highest identity. Alignment of the nucleotide sequences of the enterovirus isolates was performed using BioEdit Sequence Alignment Editor software v7.08 (Tom Hall, North Carolina State University, Carolina, USA). A phylogenetic dendrogram was constructed using the neighbor-joining method of the MEGA program v4.0 (Arizona State University, AZ, USA). Nucleotide sequences analyzed in this study have been submitted to GenBank (accession numbers JN896765-JN896862).

Serologic Assay
Serum neutralizing antibody test has been widely used for serotyping of enterovirus isolates. However, it is not suitable for early diagnosis of enterovirus infections due to requirement of collecting paired sera and complexity of preparing virus stocks. Therefore, we only used serum neutralizing antibody test to verify serotypes of enterovirus infections when IFA and molecular tests had disagreed results. Serotypes used for serum neutralization tests included EV71, coxsackievirus A2, A4, A5, A6 and A10. Laboratory methods for measuring serum neutralizing antibody titers followed standard protocols [26]. Twofold serially diluted sera and virus working solution containing 100 TCID 50 of enterovirus were mixed on 96-well microplates and incubated with rhabdomyosarcoma cells. A cytopathic effect was observed in a monitor linked with an inverted microscope after an incubation period of 4 to 5 days. The neutralization titers were read as the highest dilution that could result in a 50% reduction in the cytopathic effect. Each test sample was run simultaneously with cell control, serum control, and virus back titration. The starting dilution was 1:8 and the cutoff level of seropositivity was set at 8.

Statistical Analysis
Since gold standard is not available for detection of human enteroviruses in clinical samples, consensus result of the three tests (virus isolation and two molecular tests) was used as the gold standard to evaluate individual performance of these three tests by calculating sensitivity, specificity, positive prediction value, negative prediction value, and agreement [27,28]. The statistical significance in the agreement rates of different tests were tested by the X 2 test and Fisher's exact test as appropriate. All statistical analyses were performed using Epi-Info (CDC, Atlanta, GA) or SAS (SAS Institute, Cary, NC).

Detection of Enteroviruses in Throat Swabs
Between January 2008 and December 2009, throat swabs were collected from 110 children diagnosed with HFMD (7 cases), herpangina (93 cases), and non-specific febrile illness (10 cases). Among them, 39 (35%), 52 (47%), and 46 (42%) were tested positive by the virus isolation/IFA, the 59UTR RT-PCR, and the VP1 CODEHOP tests, respectively. Overall, results of these 110 samples tested with these three methods have eight combinations, including 30 were tested positive by all of these three tests, 17 were positive by two of these three tests, 13 were positive by only one of these three tests, and 50 were negative with all of these three tests ( Table 2). Since no gold standard method has been established for detection of enteroviruses in throat swabs, it would be reasonable to use consensus result of these three tests as the gold standard for evaluating performance of each individual test. Using the consensus result of these three tests as the gold standard, positive prediction value (PPV), negative prediction value (NPV) and agreement of these three tests were 95%, 86% and 89% for the virus isolation/IFA test, 83%, 93% and 88% for the 59-UTR test, and 96%, 95%, and 96% for the VP1 CODEHOP test, respectively (Table 3). Overall, agreement of the VP1 CODEHOP test was higher than that of the virus isolation and 59UTR tests (P = 0.05, Fisher's exact test).

Serotyping of Enteroviruses in Clinical Specimens
Among the 30 cases tested positive by all of these three tests, 23 had agreed serotypes and 7 had disagreed serotypes. Except one case who did not provide post-infection serum, 6 of the 7 cases with disagreed serotypes were further verified to be consistent with the virus isolation and VP1 CODEHOP tests by serum neutralization test using paired sera collected before and after infections (Table 4). Among the 17 cases tested positive by two of  these three tests, 13 had agreed serotypes and 4 had disagreed serotypes. Among these 4 cases with disagreed serotypes, 2 cases were serologically verified to be consistent with the virus isolation (ID#98050) or the VP1 CODEHOP tests (ID#97009), respectively and the remaining 2 cases (ID#97035 and ID#98055) could not be verified (Table 5). There were 13 cases tested positive by only one of the three tests and they were further serologically verified (Table 5). Among 2 cases (ID#97028 and ID#97065) tested positive only by the virus isolation/IFA, one was serologically confirmed and one did not provide post-infection serum. Among 9 cases tested positive only by the 59-UTR test, seven were confirmed serologically to be false positive and two did not provide post-infection sera. Among two cases (ID#98019 and ID#98044) tested positive only by the VP1 CODEHOP test, both were serologically confirmed . Overall, the serotyping results further confirmed superiority of the VP1 CODEHOP to the 59-UTR test. Table 6 showed the distribution of the top 5 enterovirus serotypes detected in Taiwan, 2008-09. The VP1 CODEHOP test could detect much more EV71 infections than the virus isolation (9 cases vs. 4 cases) and eight of these nine cases developed seroconversion against EV71 except one case who did not provide post-infection serum, which further confirmed accuracy of the VP1 CODEHOP test.

Discussion
Human enteroviruses have more than 100 serotypes which make laboratory diagnosis very challenging. Traditional methods based on virus isolation and IFA are time-consuming and laborintensive and could not detect new serotypes before antisera are available. In this study, we compare virus isolation/IFA and two molecular tests targeting 59UTR and VP1 genes for detection and  serotyping of human enteroviruses in clinical specimens of pediatric outpatients. Overall, the VP1 CODEHOP test has the best performance on detection and serotyping, which confirmed preliminary finding from a previous study which only tested 7 clinical specimens [17]. Molecular tests are more sensitive than virus isolation for detection of enteroviruses but may have more false positive reactions which can be differentiated by determining seroconversion of serotype-specific neutralizing antibody. Most studies did not collect paired sera to measure serotype-specific serum neutralizing antibody in outpatient studies [7,8,29]. In our study, paired sera have been collected to verify the validity of the molecular tests. It is not surprising to found that detection rate of the virus isolation/IFA was lower than those of the 59UTR RT-PCR and VP1 CODEHOP tests (35%, 47%, and 42%, respectively). After verification with serotype-specific serum neutralization test, we found that the 59UTR RT-PCR test was more likely to have false positive reactions (Table 4). In addition, the 59UTR-PCR test was also more likely to have false serotyping.
There are several possible reasons why the 59UTR-PCR test was more likely to have the false positive reaction and mismatched serotyping than the VP1 CODEHOP. First, the 59UTR gene is more conservative than the VP1 gene among viruses of the Enterovirus genus which includes at least 10 species (4 human enteroviruses, 3 human rhinoviruses, and 3 animal enteroviruses) [30] . Second, phylogenetic analysis using the VP1 gene sequences is highly correlated to serotyping using serum neutralization tests in enteroviruses [31]. Third, gene recombinations between enteroviruses have been detected in the 59UTR region but not in the VP1 region [32,33]. At the moment, standard laboratory methods for enterovirus surveillance have not been established, which make international comparison impossible. An international human enterovirus surveillance network would be desirable to establish standard molecular and serological methods for laboratory diagnosis of human enterovirus infections and understand their disease burden. Most of enterovirus infections are self-limited and do not require hospitalization cares. However, EV71 infections in young children frequently cause complications and could progress quickly. Therefore, Taiwanese parents frequently requests hospitalization cares if their children develop enterovirus-like symptoms (HFMD, herpangina and febrile illness) during EV71 epidemics. In a previous study, only 18.9% of enterovirus inpatients were confirmed to have EV71 infections during the 2008 EV71 epidemics in Taiwan [22]. Rapid tests differentiating EV71 infections from other enterovirus infections are urgently needed to reduce unnecessary hospitalization cares. The virus isolation/ IFA test requires 5-14 days to complete detection and serotyping, which is useful to virus surveillance but not clinical management. In contrast, the VP1 CODEHOP test could finish detection within 24 hours and requires another 24 hours for serotyping. Therefore, the VP1 CODEHOP test could be used to reduce unnecessary hospitalization cares during EV71 epidemics. In addition, the VP1 CODEHOP test could more precisely estimate disease burden of EV71 infections than the virus isolation in epidemiological studies. Moreover, about 15,30% of enterovirus isolates could not be serotyped annually using IFA in Taiwan [5]. The VP1 CODEHOP platform could potentially reduce the number of untypable enterovirus isolates and detect new enterovirus serotypes [1,34].