Evaluation of two commercially available ELISAs for the diagnosis of Japanese encephalitis applied to field samples

Authors


Corresponding Author Penny Lewthwaite, Brain Infections Group, 8th Floor Duncan Building, University of Liverpool, Liverpool L69 3GA, UK. E-mail: pennylewthwaite@doctors.org.uk

Summary

Objective  To compare two commercially available kits, Japanese Encephalitis-Dengue IgM Combo ELISA (Panbio Diagnostics) and JEV-CheX IgM capture ELISA (XCyton Diagnostics Limited), to a reference standard (Universiti Malaysia Sarawak – Venture Technologies VT ELISA).

Methods  Samples were obtained from 172/192 children presenting to a site in rural India with acute encephalitis syndrome.

Results  Using the reference VT ELISA, infection with Japanese encephalitis virus (JEV) was confirmed in 44 (26%) patients, with central nervous system infection confirmed in 27 of these; seven patients were dengue seropositive. Of the 121 remaining patients, 37 (31%) were JEV negative and 84 (69%) were JEV unknown because timing of the last sample tested was <10 day of illness or unknown. For patient classification with XCyton, using cerebrospinal fluid alone (the recommended sample), sensitivity was 77.8% (59.2–89.4) with specificity of 97.3% (90.6–99.2). For Panbio ELISA, using serum alone (the recommended sample), sensitivity was 72.5% (57.2–83.9) with specificity of 97.5% (92.8–99.1). Using all available samples for patient classification, sensitivity and specificity were 63.6% (95% CI: 48.9–76.2) and 98.4% (94.5–99.6), respectively, for XCyton ELISA and 75.0% (59.3–85.4) and 97.7% (93.3–99.2) for Panbio ELISA.

Conclusion  The two commercially available ELISAs had reasonable sensitivities and excellent specificities for diagnosing JEV.

Abstract

Evaluation de deux tests ELISA disponibles dans le commerce appliqués à des échantillons de terrain pour le diagnostic de l’encéphalite japonaise

Objectif:  Comparer deux kits disponibles dans le commerce: ELISA encéphalite japonaise/dengue IgM combo (Diagnostics Panbio) et ELISA capture JEV-Chex IgM (Diagnostics XCyton Limited) à un étalon de référence (Universiti Malaysia Sarawak-Venture Technologies VT ELISA).

Méthodes:  Des échantillons ont été obtenus de 172/192 enfants présentant le syndrome encéphalitique aigu (SEA) dans un site rural en Inde.

Résultats:  En utilisation la référence VT ELISA, l’infection par le virus de l’EJ (JEV) a été confirmée chez 44 (26%) patients avec infection confirmée du SNC pour 27 d’entre eux. 7 patients étaient séropositifs pour la dengue. Sur les 121 patients restants, 37 (31%) étaient négatifs pour JEV et 84 (69%) n’ont pas eu de résultat JEV car la période de test du dernier échantillon était < 10 jours de maladie ou inconnue. Pour la classification des patients avec XCyton, en utilisant le LCR (échantillon recommandé) seul, la sensibilitéétait de 77,8% (59,2-89,4) et la spécificité de 97,3 (90,6-99,2). Pour ELISA Panbio, en utilisant le sérum seul (échantillon recommandé), la sensibilitéétait de 72,5 (57,2-83,9) et la spécificité de 97,5% (92,8-99,1). En utilisant tous les échantillons disponibles pour la classification des patients, la sensibilité et la spécificitéétaient de 63,6% (IC95%: 48,9-76,2) et 98,4% (94,5-99,6) respectivement, pour XCyton ELISA et 75,0% (59,3-85,4) et 97,7% (93,3 -99,2) pour ELISA Panbio.

Conclusion:  Les deux tests ELISA disponibles dans le commerce ont des sensibilités et des spécificités raisonnablement excellentes pour le diagnostic de JEV.

Abstract

Evaluación de dos ELISAs disponibles comercialmente para el diagnóstico de la encefalitis japonesa en muestras de campo

Objetivo:  Comparar dos kits comerciales, el ELISA IgM Combo Encefalitis japonesa-Dengue (comercializada por Panbio Diagnostics) y el ELISA de captura ELISA IgM (JEV-CheX), comercializada por XCyton Diagnostics Limited, frente a un estándar de referencia (Universiti Malaysia Sarawak -Venture Technologies VT ELISA).

Métodos:  Se obtuvieron muestras de 172/192 niños que se presentaron con un síndrome de encefalitis aguda (SEA) en un emplazamiento de una zona rural de la India.

Resultados:  Utilizando el VT ELISA como referencia, se confirmó la infección por el virus de la Encefalitis Japonesa (VEJ) en 44 (26%) pacientes, con confirmación de infección en el SNC en 27 de ellos; 7 pacientes eran seropositivos para dengue. De los 121 pacientes restantes, 37 (31%) eran negativos para el VEJ, y 84 (69%) tenían un estatus desconocido para VEJ puesto que el momento de toma de la última muestra examinada era a <10 días de caer enfermo o desconocida. En cuanto a la clasificación de los pacientes con XCyton, utilizando solo líquido cefalorraquídeo (tipo de muestra recomendado) la sensibilidad era del 77.8% (59.2-89.4) con una especificidad de 97.3 (90.6-99.2). Para el Panbio ELISA, utilizando solo suero (tipo de muestra recomendado), la sensibilidad era de 72.5 (57.2-83.9) con una especificidad de 97.5 (92.8-99.1). Utilizando todas las muestras disponibles para la clasificación del paciente, eran 63.6% (95% IC: 48.9-76.2) y 98.4% (94.5-99.6), respectivamente, para XCyton ELISA y 75.0% (59.3-85.4) y 97.7% (93.3-99.2) for Panbio ELISA.

Conclusión:  Los dos ELISAs disponibles en el comercio tenían sensibilidades razonables y una excelente especificidad para el diagnóstico del VEJ.

Introduction

Japanese encephalitis virus (JEV) is a mosquito-borne flavivirus; it is the leading cause of encephalitis in South-east Asia. Over 3 billion people live in areas where Japanese encephalitis (JE) is endemic; it causes an estimated 20 000–175 000 cases annually with 6000–15 000 deaths (Tsai 2000; Halstead & Tsai 2004; Ghosh & Basu 2009). The virus continues to spread into new areas and often co-circulates with the related flavivirus, dengue.

Affordable vaccines are now becoming available, but implementation programmes are hampered because the epidemiological data and surveillance for JE are poor; this is largely because of lack of standardised diagnostics. JE is endemic in resource-poor areas and in these settings; patients often have a clinical diagnosis only. Serological tests are the gold standard for diagnosis of JEV as the period of viraemia is short, making methods to detect genomic material or viral antigen unreliable (WHO 2006). The timing of sample collection and type of sample collected impact on the ability to confirm the diagnosis of JEV. Cerebrospinal fluid (CSF) is the sample of choice, as detection of antibodies in the CSF confirms viral infection of the central nervous system (CNS). Serum samples demonstrate recent peripheral infection only. Even demonstrating a rising serum IgM is insufficient for confirmation of JEV as the causative agent for an acute encephalitis syndrome (AES), as this does not distinguish peripheral from CNS infection. Studies from Thailand suggest that high titres of JEV IgM may be found in sera of asymptomatic individuals (Grossman et al. 1973). Burke et al. (1985) found that by day 7 of illness, 100% of patients with acute JE had IgM to JEV detectable in CSF and serum. Other studies have since shown it can take up to 10 days to mount an antibody response to JEV (Solomon et al. 1998).

Thus, patients with early negative samples only should be classified as having a ‘JEV unknown’ status as detectable levels of antibody may not have had time to develop. Current WHO surveillance standards recommend a sample from day 10 or more of illness to confirm a true negative (TN) result (WHO 2006). There has also been concern that lower CSF IgM antibody responses may be associated with a poorer outcome, and if the CSF IgM rise is only transient, then it may be difficult to exclude JE, even with a negative CSF IgM (Libraty et al. 2002). Because of cross-reactivity between JEV and dengue virus, serological testing for both viruses is recommended in areas where they co-circulate, which is the case in most of Asia. It is, however, important not to become complacent about diagnosis and overlook other aetiological agents. As in the field, in the absence of robust alternative diagnostics, it is all too easy to attribute disease to one infection when it may in fact be attributable to another, as has occurred in Uttar Pradesh in India where recurrent epidemics of JE have occurred for many years and have apparently continued despite a recent immunisation campaign. It is only with further investigations that EV71 has recently been identified as a causative agent of encephalitis, which otherwise would have been attributed to JEV (Parida et al. 2006; Sapkal et al. 2009).

There are only a limited number of enzyme-linked immunosorbent assay (ELISA)–based diagnostic kits commercially available for diagnosing JEV, and not all are designed to test CSF. The two most widely available kits being used across Asia are the JEV-CheX IgM capture ELISA (XCyton) and the JE-Dengue combo test (Panbio). They differ in terms of which samples they test and ability to test for dengue. For the XCyton test, the preferred sample, according to manufacturer’s recommendations, is CSF, a positive result confirming JE, while a positive serum result is described as being only suggestive of recent JEV infection. In contrast, the Panbio test is designed for serum only; although the manufacturers do advise, upon being contacted directly, that if CSF is to be tested, it should be at a dilution of 1:10. The Panbio ELISA tests for dengue antibodies in parallel to JEV and the XCyton ELISA tests for JEV antibodies and recent flavivirus infection such as dengue or West Nile.

There are limited data on the sensitivity and specificity of these kits, although in practice, they are being used to test both serum and CSF samples in many laboratories across Asia (Cuzzubbo et al. 1999; Jacobson et al. 2007; Shrivastva et al. 2008; Ravi et al. 2009). Recent studies have been based on samples collected without benefit of additional clinical data using well-characterised collections of samples with a complete set for each patient. No published studies have looked at a prospective cohort of patients with AES over several JEV seasons, which include those with indeterminate and borderline status and incomplete sample sets, a situation very common in the field.

We decided to determine sensitivity and specificity of these two commercially available kits, against a reference standard, which has been used widely for many years in South-east Asia (Sarawak, Malaysia, Vietnam and Indonesia) and that we have used previously (Solomon et al. 1998; Cuzzubbo et al. 1999; Cardosa et al. 2002; Ravi et al. 2006; Ooi et al. 2008). The kits were used according to the manufacturer’s recommendation and also for the non-preferred specimens, to reflect what is actually happening in the field.

Methods

Participants

Samples were obtained from children presenting with suspected AES to the paediatric department of the Vijayanagar Institute of Medical Sciences (VIMS), Bellary, Karnataka, India, between October 2005 and March 2007. AES was suspected in those with a febrile illness (<2 weeks’ duration), with at least one of: meningism, photophobia, severe headache, altered mental status, seizures or focal neurological signs (WHO 2006). Those with pre-existing neurological conditions or with Plasmodium falciparum parasitaemia were excluded. The study protocol was approved by the hospital’s scientific and ethical committee and the University of Liverpool, UK. Informed consent was obtained from the child’s parent or guardian.

A detailed history was taken, and full neurological examination was performed by a member of the study team. Routine blood samples were taken, and lumbar puncture was performed on admission to the study and, where possible, on days 2–5 and day 7 of their in-patient stay. Samples were immediately processed and frozen at −70 °C. Treatment for JE is supportive, and the children were given fluids, steroids and symptomatic treatment for seizures, at the discretion of the treating clinician.

Test methods

Laboratory personnel performing the diagnostic assays were aware that the patient had presented with AES but were blinded to the severity of the clinical presentation, outcome and to other test results. Because of the small sample volumes available from children, it was only possible to perform each ELISA once, and there was insufficient sample for further confirmatory testing.

The JEV-CheX IgM capture ELISA (XCyton, Bangalore, India), known as the XCyton test, was prospectively applied to samples in accordance with local testing guidelines by a technician trained in the use of the kit in either Bellary or the National Institute of Mental Health and Neurological Science (NIMHANS), Bangalore, where the tests have been in use for some time (Ooi et al. 2008).

The XCyton test recommends CSF as the diagnostic specimen and provides results as JEV positive, JEV negative or flavivirus positive; positive CSF samples are confirmatory for encephalitis, whereas positive serum samples are suggestive of recent peripheral infection. The XCyton ELISA was performed according to the manufacturer’s instructions; serum samples were diluted 1:20 and CSF 1:10 using the diluent supplied. The XCyton JEV ELISA unit was calculated by the formula [(test sample optical density (OD) value − mean of negative control performed in triplicate OD) × 100/(Mean OD of weak positive control performed in duplicate − mean OD of negative control performed in triplicate)]. Samples with ≥100 ELISA units are considered JEV positive, samples with 30–99 ELISA units are suspected recent flavivirus infection and those with <29 ELISA units, JEV negative. For the purposes of this analysis, serum samples positive by XCyton ELISA were designated to be JEV peripheral infection positive and all samples that were negative or flavivirus positive were classed as ‘JEV unknown’ if sample was <day 10 of illness or JEV negative if from ≥day 10 of illness.

Samples were also tested by Japanese Encephalitis-Dengue IgM COMBO ELISA (Panbio, Australia) (Cuzzubbo et al. 1999). The Panbio ELISA was performed according to the manufacturer’s instructions on serum samples. After discussion with the manufacturer for advice on testing CSF, CSF samples were diluted 1:10 in serum diluent provided (serum samples were diluted 1:100). The Panbio JE-Dengue combo tests gave positive, negative or equivocal results for JEV and dengue. Panbio Units (PBU) were calculated using the formula [sample absorbance OD value multiplied by 10, divided by the cut-off value (the cut-off value is calculated by averaging OD of a calibrator measured in triplicate, multiplied by a batch dependent calibration factor)]. The same cut-off values were used for both serum and CSF samples. A PBU of >11 was diagnostic, 9–11 equivocal and <9 negative. For the purpose of this analysis, the JEV/dengue ratio was calculated (JEV PBU/Dengue PBU) for samples with a PBU >11 for both JEV and dengue; for ratios ≥1, a presumptive diagnosis of JEV was made, and for those ≤1, a presumptive diagnosis of dengue was made. For the purposes of analysis, Panbio equivocal results were considered to be negative.

The two ELISA test kit results were compared with the results of the JEV-dengue reference VT ELISA (Solomon et al. 1998; Cuzzubbo et al. 1999; Ravi et al. 2006; Ooi et al. 2008). Using the reference VT ELISA, a patient was considered to have had JEV encephalitis if their CSF OD against JEV had a Sample/Negative control ratio >5 and the OD was higher against JEV than against dengue. Patients with antibody detected only in serum but not in CSF were classed as having had a recent peripheral JEV infection. If CSF was not available but the serum was positive, the patient was also classed as having had a recent peripheral JEV infection with the caveat that the status of intrathecal IgM was unknown.

For each ELISA kit, results were obtained for each sample tested (CSF or serum). Each patient was classified according to the overall results from all samples for each kit, as having a CNS or peripheral infection, which was positive or negative for JEV, dengue or unspecified flavivirus (see below). Patients with ELISA negative results were classed as ‘JEV unknown’ if their samples had been taken earlier than on day 10 of illness or unknown date of onset and JEV negative if any of their samples had been taken on day 10 of illness or later. These results were compared to those obtained using the reference VT ELISA results for each sample.

Results were examined using the preferred specimen only for each test, i.e. CSF for XCyton and serum for Panbio; the overall classification for each patient was determined using all available samples. For simplicity of analysis, patients with any CSF antibody positive results were classified as having CNS infection; patients with no CSF or negative CSF JEV antibody results but with antibody in serum only were defined as having peripheral infection only (although their actual CNS infection status was indeterminate). This differentiation between CSF infection status is however clinically important, as data from Southern Thailand suggest that lower IgM levels in CSF could indicate poor prognosis (Libraty et al. 2002).

Analysis

Test specificity and sensitivity and positive and negative predictive values were calculated for both new commercial kits using the reference VT ELISA. Results were classified as either JEV positive or JEV negative; Panbio equivocal results were classified as JEV negative as were XCyton’s non-JEV (i.e. flavivirus or dengue positive) results i.e. ‘JEV negative’. Results of the test assays were classified as true positive (TP) or TN if in agreement with the reference standard. If results differed from the standard, they were classified as false positive (FP) or false negative (FN). Diagnostic sensitivity was calculated as TP/(TP + FN) and diagnostic specificity as TN/(TN + FP). Sensitivities and specificities were calculated for each individual sample and the patients’ overall classification. Confidence intervals for sensitivity and specificity were calculated using the Wilson score method without continuity correction (Newcombe 1998). Positive predictive value was calculated as TP/(TP + FP), and negative predictive values were calculated as TN/(TN + FN).

Results

Of the 232 children screened, 192 were recruited to the study, and of these, 172 had sufficient sample volume for this comparative work. In total, 266 serum samples were obtained from 159 patients and 105 CSF samples from 102 patients. For 90 patients, both serum and CSF samples were available; of these, 87 had a single CSF and serial serum samples available and only three patients had both serial CSF and serum samples. The 172 patients had a median age of 7 years (range 8 months to 16 years), and 83 (48%) were girls. The median age of the 20 children not included in the analysis was 4 years (range 1–15), and 6 (30%) were female.

Using the reference VT ELISA, 44 (26%) of the children were confirmed to have had JEV infection (Table 1). CNS infection was confirmed in 27 (25 of whom were also JEV ELISA positive in peripheral blood); 17 were JEV positive in peripheral blood only; of these, five had negative CSF samples at ≥10 days of illness and so had peripheral infection only, and the remaining 12 had no CSF sample tested and so their CNS infection status was unknown. Of the JEV ELISA negative patients, seven were dengue IgM positive in serum. Of the 121 remaining patients, 37 (31%) were JEV negative, and 84 (69%) were JEV unknown as they only had samples available from <10 day of illness or an unknown date of onset of illness. Only three patients had both serial CSF and serum samples, and one of these was JEV positive and two were JEV negative.

Table 1.   Overall ELISA results
 SamplesPatients
CSFSerum
  1. Table 1 shows results for each sample by ELISA and overall patient result for each ELISA test. Patients with ELISA negative samples taken prior to 10th day of illness are not true JEV negatives and so are termed ‘JEV unknown’.

  2. CSF, cerebrospinal fluid; JEV, Japanese encephalitis virus.

  3. *Three patients with no information about date of onset of illness are included as ‘JEV unknown’. A sample from day 10 or more of illness is regarded as a true negative result (WHO 2006).

Reference Venture Technologies ELISA
 JEV positive287544
 Dengue positive0137
 ‘JEV unknown’ <day 10 sample*  84
 Negative7717837
XCyton IgM capture ELISA
 JEV positive234630
 Flavivirus positive92415
 ‘JEV unknown’ <day 10 sample*  87
 Negative7319640
Panbio JE-dengue IgM COMBO ELISA
 JEV positive215736
 Dengue positive03118
 ‘JEV unknown’ <day 10 sample*  83
 Negative8417835
Total105266172

Fourteen of the children (8%) died: four had JEV CNS infection confirmed by positive JEV ELISA on CSF, and one had peripheral dengue infection, she was a 10-year-old girl who presented with reduced consciousness and died of dengue shock syndrome. All nine deceased children without diagnosis had negative samples from day 2–8 of illness and so were ‘JEV unknown’.

Table 2 shows agreement between the reference VT ELISA and test ELISAs for CSF samples and serum samples. In Table 3, the patient classifications using CSF samples alone or serum samples alone are compared. In Table 4, the sensitivity, specificity positive and negative predictive values are given for the two test kits, compared to the reference standard. Using the manufacturer’s recommended sample (CSF for XCyton, serum for Panbio), the specificity of both kits was very good (>97%); the sensitivity was better for the XCyton kit (77.8% assessing patients using their CSF samples) than for Panbio (72.5% assessing patients by their serum samples). However, when assessing patients using both CSF and serum samples, the specificity was marginally better for the XCyton kit (98.4%vs. 97.7%), whereas the sensitivity was better for the Panbio kit (75.0%vs. 63.6%).

Table 2.   Comparison of ELISA results for either CSF samples alone or serum samples alone
 ResultReference VT ELISA
CSF samplesSerum samples
JEV positiveDengue positiveJEV/Dengue negativeJEV positiveDengue positiveJEV/Dengue negative
  1. Table 2 shows the comparison of ELISA results for either CSF samples alone or serum samples alone using XCyton and Panbio kits compared to the reference VT ELISA. Numbers in bold are those where kits under test and the reference standard are in agreement.

  2. CSF, cerebrospinal fluid; VT, Venture Technologies; JEV, Japanese encephalitis virus.

XCytonJEV positive21024303
ELISAFlavivirus positive6031851
Negative1072148174
Panbio JEJEV positive19025205
DengueDengue positive00031117
ELISANegative9075202156
Totals 280777513178
Table 3.   Overall patient classifications using either CSF samples alone or serum samples alone
Reference VT ELISA
(a) Patient classification based on CSF samples only(b) Patient results from serum samples only
XCytonJEV positiveDengue positive‘JE unknown’NegativeXCytonJEV positiveDengue positive‘JEV unknown’Negative
  1. Overall Patient classifications using either CSF samples alone or serum samples alone, results from XCyton and Panbio compared to the reference VT ELISA. Patients with negative ELISA results from samples from <10 day of onset of illness are classified as ‘JEV unknown’.

  2. VT, Venture Technologies; JEV, Japanese encephalitis virus; CSF, cerebrospinal fluid; CNS, central nervous system; Periph, peripheral infection.

JEV positive21020JEV positive25010
Flavivirus positive4020Flavivirus positive6311
‘JEV unknown’10610‘JEV unknown’51740
Negative10010Negative42036
Panbio    Panbio    
 JEV positive18020JEV positive29030
 Dengue positive0000Dengue positive2547
 ‘JE unknown’70620‘JEV unknown’61690
 Negative20110Negative30030
Total2706510Total4067637
Table 4.   Sensitivities, specificities, positive and negative predictive values for XCyton and Panbio assays compared with Reference VT ELISA
 Sensitivity(95% Confidence interval)Specificity(95% Confidence interval)Pos pred valueNeg pred value
  1. Table 4 shows sensitivities and specificities using each test kit compared to the reference VT ELISA for CSF samples alone, serum samples alone, patient classification using only CSF and only serum and classification using all available samples. Bold values show results for preferred samples used according to manufacturer’s recommendations: CSF for XCyton and serum for Panbio ELISA and combined results using all available samples.

  2. VT, Venture Technologies; CSF, cerebrospinal fluid; CNS, central nervous system; Periph, peripheral infection; Pos pred value, positive predictive value; neg pred values, negative predictive value.

XCyton compared to Reference VT ELISA
 CSF samples75.0(56.6–87.3)97.4(91.1–99.3)0.910.91
 Serum samples57.3(46.067.9)98.4(95.599.5)0.930.85
 Patients assessed by CSF sample only77.8(59.2–89.4)97.3(90.6–99.2)0.910.92
 Patients assessed by serum sample only62.5(47.075.8)99.2(95.399.8)0.960.89
Patients assessed by both CSF and serum samples63.6(48.9–76.2)98.4(94.5–99.6)0.930.89
Panbio ELISA compared to Reference VT ELISA
 CSF samples67.9(49.382.1)97.4(91.099.3)0.900.89
 Serum samples69.3(58.2–78.6)97.4(94.0–98.9)0.910.89
 Patients assessed by CSF sample only66.7(47.881.4)97.3(90.699.3)0.900.89
 Patients assessed by serum sample only72.5(57.2–83.9)97.5(92.8–99.1)0.910.91
Patients assessed by both CSF and serum samples75.0(59.3–85.4)97.7(93.3–99.2)0.920.92

Discussion

With the global spread of JE, there is a need for standardisation of diagnostic testing. WHO, in collaboration with JE endemic countries, academic institutions and the Program for Appropriate Technology and Health (PATH), has established a JE laboratory network across Asia. Expensive vaccine programmes are based around disease detection, so it is important to have accurate epidemiological information to inform governments and funding bodies as to where best to introduce immunisation and disease prevention strategies.

Although there are several commercial kits available for diagnosing JE, the two being most widely used are manufactured by XCyton and Panbio. Although the kits were designed for use with CSF or serum, respectively, we chose to test both serum and CSF samples, even if this was outside manufacturer’s recommendations, because this reflects how the kits are actually being used. The detection of anti-JEV IgM in CSF is diagnostic of JEV encephalitis, whereas IgM in serum may reflect recent asymptomatic infection or vaccination. We used the Venture Technologies (VT) ELISA as our reference standard because this is a well-established diagnostic test, which was developed at Universiti Malaysia, Sarawak, and has been in use in Southeast Asia for over 10 years (Cardosa et al. 2002; WHO 2003; Wong et al. 2008). It has compared favourably with another widely used ELISA test Armed Forces Research Institute of Medical Sciences (AFRIMS), Thailand (Cardosa et al. 2002; Ooi et al. 2008; Wong et al. 2008). Neither PCR nor virus isolation were used because it is rarely possible to detect JEV by these methods (Desai et al. 1994; WHO 2003).

When all samples available for a patient were used to give an overall classification for each patient, the specificity for both kits was excellent (98.4% for XCyton and 97.7% for Panbio). However, sensitivity was improved by using the sample type recommended by the manufacturer for XCyton (77.8% (95% CI: 59.2–89.4)) CSF and using both serum and CSF for Panbio 75.0% (95% CI 59.3–85.4). The availability of multiple samples improved the sensitivity for serum but not for CSF diagnoses; this may reflect the small number of serial CSF samples. Both assays require a trained technician to perform them. XCyton and Panbio provide all solutions required either as concentrates or in solutions. The preferred diagnostic sample for the XCyton kit is CSF, as only by confirming JEV infection in the CSF can a diagnosis of JE be made. Presence of JEV antibody in blood, the preferred specimen for the Panbio ELISA, may reflect a recent peripheral JEV infection only and does not confirm that the neurological presentation was attributable to JEV. This is of importance, as vaccines are being used more widely, and the possibility of FPs owing to recent vaccination, or FNs owing to muted IgM responses after vaccination, may occur. The reference standard and Panbio ELISAs have the advantage that they also diagnose dengue; this is important in clinical practice as JEV and dengue viruses often co-circulate. XCyton only identifies the presence of other non-JE flavivirus.

The main shortcoming of our study was the lack of sufficient sample volume to repeat equivocal results and carry out further tests including neutralisation tests to confirm the ELISA diagnosis. But when these tests are used in the field, often only one assay is performed and sample number and volumes are limited. Our study therefore reflects the reality of how these kits will perform in the field. Diagnosis in the most severe cases (i.e. children who died before samples were taken or shortly after samples were taken) remains problematic. Those patients who died but were JEV negative may have had JE, but there had been insufficient time to mount a detectable immune response (Libraty et al. 2002).

Using the current WHO definition, the importance of obtaining at least one sample on or after day 10 illness was highlighted; only 37 (22%) patients with negative JEV ELISA results were TNs with samples from ≥day 10 of illness, the remainder being ‘JEV unknown’, which may lead to an underestimate of disease burden (Solomon et al. 1998; WHO 2003, 2006). The majority of our samples, 87 (83%) of all CSF and 177 (76%) of all serum samples, were taken prior to day 10 of illness (Figure 1), illustrating that obtaining later samples is a difficult undertaking in a study setting and likely to be even more difficult in routine clinical practice. If instead the cut-off of ≤7 days of illness is used, as described by Burke et al. (1985), then 80 (76%) of CSF samples and 145 (54%) of serum samples were within this time frame, although there are concerns that this earlier cut-off may lead to under-diagnosis (Solomon et al. 1998).

Figure 1.

 (a) Distribution of serum samples by day of illness and reference standard result JEV VT ELISA. (b) Distribution of cerebrospinal fluid (CSF) samples by day of illness and reference standard result JEV VT ELISA. Serial serum and CSF results were obtained where possible from patients. The sample distribution by day of illness is shown in these graphs together with the reference standard VT ELISA result for JEV. JEV, Japanese encephalitis virus.

Our study suggests that both the ELISA kit tests are reasonably good, but both could be improved. The choice of which test to use will depend on the setting. If it is possible to obtain CSF samples, then XCyton is the preferred test. Even in areas where dengue is circulating, this kit appears to accurately diagnose CNS infection with JEV, based on a CSF sample only. This is especially important in areas where vaccination has occurred because of the likelihood that vaccinees will be JEV antibody sero-positive. In our study setting, we were able to perform lumbar punctures on 102 (59%) of AES cases. In areas where it is not possible to obtain CSF samples, the Panbio test may be preferable because it is more sensitive when applied to serum and is able to distinguish JEV from dengue. However, in general, lumbar punctures should be encouraged, not only for accurate diagnosis of JE, but also to rule out bacterial infections that require specific life-saving treatment (WHO 2006). Our results will help to guide the WHO JEV laboratory network and partner countries as to which kit to use in which circumstances.

In summary, this study has shown that both the commercially available XCyton and Panbio ELISA tests have reasonable sensitivities and excellent specificities for diagnosing JEV. Their wider usage should facilitate the JEV control programme.

Acknowledgements

We are grateful to all the patients and their carers for assisting with this work. Thanks also are due to the Hospital Director at Vijayanagar Institute of Medical Sciences, Bellary, the ward staff and nurses and doctors who assisted with the study and in particular to Mr Iqbal Sayed and also the staff of the Neurovirology Department, NIMHANS. We also thank Dr Susan Hills and Dr Julie Jacobson, PATH JE Project, for their support and encouragement. We thank PATH JE Project and MRC UK for their financial support and Panbio Australia for donating the ELISA test kits and for technical advice.

Ancillary