JOURNAL OF CLINICAL MICROBIOLOGY, Dec. 1993, p. 3240-3246 Vol. 31, No. 120095-1137/93/123240-07$02.00/0Copyright © 1993, American Society for Microbiology

Coxsackievirus B1-Based Antibody-Capture Enzyme-LinkedImmunosorbent Assay for Detection of Immunoglobulin G

(IgG), IgM, and IgA with Broad Specificity for EnterovirusesC. M. A. SWANINK,l* L. VEENSTRA,l Y. A. G. M. POORT,1 J. A. KAAN,2 AND J. M. D. GALAMA'Department ofMedical Microbiology, University ofNijmegen, P.O. Box 9101, 6500 HB Nijmegen, 1 and

Microbiology Department, Public Health Laboratory, 7512 AD Enschede,2 The Netherlands

Received 12 May 1993/Returned for modification 22 July 1993/Accepted 23 September 1993

An antibody-capture enzyme-linked immunosorbent assay (ELISA) with coxsackievirus Bi as the antigenwas evaluated for detection of immunoglobulin G (IgG), IgM, and IgA antibodies and showed broad specificityfor enteroviruses. In total, 116 serum or cerebrospinal fluid samples from 62 patients were tested by ELISA andthe complement fixation test (CF17). Additionally, 15 serum samples that contained poliovirus-specific IgMantibody were tested. Serum samples from 200 healthy blood donors were used for standardization of theassays. The sensitivity of the ELISA varied with time of serum sampling, with a relatively low sensitivity whenserum was collected within 3 days after the onset of symptoms (23%; 5 of 22) but good sensitivity when serumwas collected later (83%; 20 of 24). The sensitivity was better than that of the CFT. The ELISAs were broadlyreactive as concluded from typing of virus isolates that were simultaneously obtained. The assay did,furthermore, detect antibody against poliovirus type 3. Sera that contained rheumatoid factor, antinuclearantibody, or cardiolipin antibody (by the Venereal Disease Research Laboratory test) did not react in thisELISA. Nonspecific reactivity did occur, however, in cases of infectious mononucleosis and in Mycoplasmapneumoniae infection. The enterovirus-specific ELISA is found to be simple to perform, more sensitive than theCFT, and far less laborious than the neutralization test.

Enteroviruses are common pathogens that can cause avariety of symptoms ranging from mild respiratory infectionto severe central nervous system disease. Enterovirusesmay furthermore play a role in chronic diseases such aspostviral fatigue syndrome (1, 12, 14, 37), chronic myocar-ditis/dilated cardiomyopathy (8), polymyositis/dermatomyo-sitis (7, 38), postpolio syndrome (30), and insulin-dependentdiabetes mellitus (2, 20, 34). The diagnosis of the acuteinfection is mainly based on virus isolation and serologicaltests such as virus neutralization or complement fixation(CF). Virus isolation is not successful in cases of the chronicdiseases mentioned above, and a relationship with enterovi-ruses can be investigated only by molecular techniques (39)or serology. Enteroviruses include 69 serotypes. Hence, aneutralization test is laborious as long as the virus type is notknown. CF tests (CFTs) are useful but of limited diagnosticvalue, particularly in a presumed persistent phase of infec-tion. Enzyme-linked immunosorbent assays (ELISAs) forthe diagnosis of enteroviral infections based on either anti-body capture (3, 22, 29) or an indirect technique (6, 19) havebeen described previously. These tests made use of either asingle serotype (6) or multiple serotypes (3, 19) as theantigen. Tests for immunoglobulin G (IgG) antibody wereshown to be of limited value because of the presence ofanamnestic antibody from earlier infections (6).We describe an antibody-capture ELISA based on a

heat-inactivated coxsackievirus Bi antigen for the detectionof enterovirus-specific IgM, IgA, and IgG antibodies. Wehave deliberately chosen the antibody-capture ELISA be-cause false reactions due to the interference with rheumatoidfactors seldom occur and competition between antibodies ofvarious isotypes is avoided. The antibody-capture ELISAs

* Corresponding author.

are highly sensitive, with exception of the IgG-captureassay. A high level of nonspecific IgG in serum reduces thesensitivity of this assay to a relatively low level. Theadvantage of this low sensitivity is that the assay will notdetect low concentrations of IgG class antibodies from pastinfections but only increased levels from recent infections orfrom persistent infections. The capture principle leads to anincreased sensitivity (also for IgG antibody) when cerebro-spinal fluid (CSF) is tested (32). In this report, we describethe performance of this ELISA for the diagnosis of acuteenterovirus infection with routine samples.

MATERIALS AND METHODS

Patients. Routine serum samples were collected during an8-month period (May to December 1992) from patients withsymptoms of respiratory infection, gastroenteritis, meningi-tis, pleuritis, and myocarditis. One hundred sixteen samplesfrom 62 patients were tested for enteroviral antibodies byCFT and ELISA. A stool specimen, a throat swab, or CSFwas obtained from 35 of 62 patients for viral culture. For 14patients, cultures remained negative for enterovirus; entero-virus was isolated from 21 patients. Enteroviruses weretyped by the use of pools of neutralizing sera (17). The typesof enteroviruses that were isolated were coxsackievirus A9(once), coxsackievirus B5 (four times), echovirus 6 (once),echovirus 9 (twice), echovirus 11 (three times), echovirus 19(once), echovirus 30 (eight times), and poliovirus type 3(once). Additionally, 15 serum samples that were obtainedduring a recent Dutch epidemic of poliomyelitis type 3 werekindly provided by the National Institute of Public Healthand Environmental Protection. The latter sera were all takenfrom asymptomatic contacts of patients with poliomyelitis.All sera contained poliovirus type 3-specific IgM antibodies

3240

COXSACKIEVIRUS Bl-BASED ELISA 3241

that were detected by a poliovirus type-specific ELISA (23,27, 36).

Control sera. Sera collected in March 1991 from 200healthy blood donors were used as controls. To investigatethe specificity of the assay, rheumatoid factor IgM-positiveserum samples (n = 20), serum samples with antinuclearantibodies (n = 21), syphilitic serum samples (positive by theVenereal Disease Research Laboratory test and the Trepo-nema pallidum microhemagglutination test) (n = 20), serumsamples with heterophile antibodies (Monospot; Ortho Di-agnostic Systems, Raritan, N.J.) (n = 20), and Mycoplasmapneumoniae IgM- and IgA-positive serum samples (n = 15)were tested as well. Serum samples were kept at -20°C untilanalysis.

Preparation of antigen. Coxsackievirus Bi was grown inbuffalo green monkey cells. Buffalo green monkey cells weregrown in 850-cm2 roller bottles (Corning Glass Works,Corning, N.Y.). Confluent monolayers were infected withcoxsackievirus Bi (Tucson strain [28]) at a multiplicity ofinfection of 0.5. The cultures were maintained in 150 ml ofminimal essential medium with 3% fetal bovine serum, 100 Uof penicillin per ml, and 50 ,ug of gentamicin per ml. Theroller bottles were incubated at 36°C. Virus was harvestedwhen complete cytopathic effect was observed, usually after2 to 3 days. Cells were freeze-thawed three times andtogether with the culture medium firmly shaken with 10%chloroform for 15 min and centrifuged at 2,000 x g for 30 min(18). The supernatant was centrifuged through a 30% (wt/wt)sucrose cushion at 150,000 x g for 3 h. The pellets wereresuspended in a final volume of 6 ml of phosphate-bufferedsaline (PBS) and heated at 56°C for 60 min in order todenature the virions and convert antigenicity from typespecific (N antigen) to group specific (H antigen) (19, 24).Control antigens were prepared in the same way by usingnoninfected cell cultures. The protein contents of the viralantigen and the control antigen preparations were deter-mined as described by Lowry et al. (21). The antigens wereconjugated with horseradish peroxidase as described byWilson and Nakane (35). The conjugated viral and controlantigens were suspended in PBS containing 0.005% mercu-rothiolate and 2% fetal bovine serum. The labelled antigenswere stored in small aliquots at 4°C.

Capture ELISA technique. The antibody-capture ELISAsfor detection of enterovirus IgG, IgM, and IgA were carriedout essentially as described previously (32, 33). Polyethyleneterephthalate glycol microtiter plates (96-well PETG assayplates, no. 6595; Costar Europe, Badhoevedorp, The Neth-erlands) were coated with 125 ,ul of goat antihuman IgG(Cappel Laboratories, Cochranville, Pa.), goat antihumanIgA (Cappel), or monoclonal antihuman IgM (kindly pro-vided by J. P. Coutelier, Universite Catholique de Louvainand Institute of Cellular and Molecular Pathology, Brussels,Belgium) diluted in Tris buffer (pH 9.0). The optimal dilu-tions were determined by checkerboard titration and were1:500 for anti-IgG and 1:1,000 for anti-IgM and anti-IgAantibodies. The plates were incubated overnight at 4°C,washed four times with wash buffer (0.01 M PBS [pH 7.2]with 0.05% Tween 20), and shaken dry. Next, 100 ,ul ofpatient serum diluted at 1:100 in PBS-Tween with 2% fetalbovine serum and 0.005% mercurothiolate (PFT-M) wasadded and the plates were incubated at 37°C in a humidifiedatmosphere for 2 h. All sera were tested in duplicate. Plateswere washed again four times and shaken dry, and 100p,l ofhorseradish peroxidase conjugated antigen diluted in PFT-Mwas added. The optimal dilution of conjugated antigen wasdetermined by block titration and was 1:1,000 for the IgG

Frequency (%) cut-off36-

30 7

25250 ....... ...... ........................................................................................ .......................... ................................................... ..........

1 5 ............................................................................................. .............................................................

10105 - - - .------- ----- ----

5-0.-.Z;

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4Extinction

0.45 0.5 0.55 0.6 p0.6

FIG. 1. Results of the IgG-capture assay with serum samplesfrom 200 healthy blood donors.

assay and 1:1,500 for the IgM and IgA assay. After overnightincubation at 4°C, the plates were washed again and 100 ,ul ofsubstrate solution was added. The substrate solution wasprepared immediately before use by dissolving 4 mg oforthophenylenediamine per ml in 0.05 M citrate buffer (pH5.2) and then adding 0.045% hydrogen peroxide (final con-centration). The reaction was stopped after exactly 10 minby the addition of 100 p.l of 3 M H2SO4. The A492 was read(Titertek Multiskan; Flow Laboratories, Irvine, UnitedKingdom). The buffer control was used as a blank. Astrongly positive serum sample, a cutoff serum sample, anda negative control serum sample were included in each test.The cutoff was chosen on the assumption that at most 5% ofthe serum samples from healthy blood donors were reactive.A serum sample was considered positive when the absor-bance was at least twice the absorbance of the cutoff serumfor IgM and IgA or at least the same as cutoff value for IgG.The ELISA with the control antigen was performed in asimilar way.CFT. The CFT was performed according to the microtiter

technique described by Casey (9). The enterovirus antigen inthe CFT consisted of a mixture of antigens including cox-sackieviruses Bi to B5 and A9 and echoviruses 4, 6, 9, 14,24, and 30 (Behringwerke AG, Marburg, Germany). The titeris expressed as the reciprocal of the highest dilution showing50% hemolysis. The CFT was considered positive when afourfold increase of antibodies between acute and convales-cent sera was found or when the titer was >128.

Neutralization test. Sera from some of the patients weretested for neutralizing antibodies as described by Melnick etal. (25). IgM was prepared by a fractionation-concentrationtechnique after gel chromatography as described by Inouyeand Kono (15). Because of the limited amount of serum, theIgM fraction was tested only against the virus strain againstwhich the highest neutralizing antibody titer was found inunfractionated serum.

RESULTS

Standardization of enterovirus-specific ELISA. Two hun-dred serum samples from healthy blood donors were testedfor enterovirus IgG, IgM, and IgA antibodies. From theoutcome, cutoff values were determined on the assumptionthat at most 5% of the serum samples were reactive. This isillustrated for the IgG assay in Fig. 1. Serum with reactivity

VOL. 31, 1993

J. CLIN. MICROBIOL.3242 SWANINK ET AL.

TABLE 1. Patients with culture-positive enterovirus infectiona

Patient Age Clinical diagnosis, Virus isolate Source of isolate Material No. of days CFtr IgG gM IgAno. Ae symptoms, and signs after onset titer

1 4 mo Near SIDS Echovirus 11 Thr Serum +13 <4 - +2 1 yr Fever, aplasia Echovirus 11 F Serum -15 8 - -

Serum +4 16 - ++ +3 1 yr Respiratory infection Echovirus 19 Thr Serum 0? 32 - - ++4 5 yr Fever e.c.i. Poliovirus type 3 F (CSF, negative) Serum +8 >256 + +5 2 1/2 yr Pyelonephritis (E. coli) Coxsackievirus B5 F Serum 0 >256 - -

6 5 wk Staphylococcal sepsis Echovirus 11 F (CSF, negative) CSF 0 <4 - -

7 4 yr Astasia, encephalitis/ Echovirus 6 F (CSF, negative) CSF +1 <4 - -

cerebellitis?Serum +1 NS - +

8 1 wk Neonatal meningitis Coxsackievirus A9 CSF CSF +1 <4 - -

CSF +8 NS - -

Serum +7 8 - -

9 24 yr Meningitis Echovirus 9 CSF CSF 0 <4 - -

Serum +1 16 - -

10 4 mo Meningism? Echovirus 30 F (CSF, negative) CSF +2 <4 - -

Serum +2 <4 - -

11 25 yr Meningitis Coxsackie virus B5 CSF CSF 0 <4 - -

CSF +9 <4 + - +Serum +1 16 - - -

Serum +9 >256 - - -

12 12 yr Meningitis Coxsackievirus B5 CSF CSF +3 <4 + + + +Serum +3 >256 - ++ -

13 11 yr Viral meningitis Echovirus 9 F Serum +15 128 - + -

14 40 yr Meningitis Coxsackievirus B5 CSF CSF 0 <4 - - -

Serum +7 >256 - - -

15 10 yr Aseptic meningitis Echovirus 30 F Serum +7 16 - + + +Serum +14 32 - + +

16 5 yr Meningitis Echovirus 30 Thr (CSF, negative) Serum 0 <4 - - -

17 5 yr Viral meningitis Echovirus 30 CSF Serum +10 128 + + + +18 22 yr Viral meningitis Echovirus 30 CSF Serum 0 8

Serum +14 128 +19 33 yr Viral meningitis Echovirus 30 CSF Serum 0 NT

Serum +19 >256 ++20 7 yr Viral meningitis Echovirus 30 CSF Serum +10 NT - + -

21 6 yr Viral meningitis Echovirus 30 Thr Serum +10 4 - + + -

a Abbreviations: SIDS, sudden infant death syndrome; F, fecal sample; Thr, throat swab; NS, nonspecific; NT, not tested. Symbols: -, negative (IgG < thecutoff; IgM and IgA < 2x the cutoff); +, positive (IgG, 1x to 2x the cutoff; IgM and IgA, 2x to 5x the cutoff); + +, strongly positive (IgG > 2x the cutoff; IgMand IgA > 5x the cutoff).

at the cutoff level as well as a strongly reactive serum sampleand a negative serum sample were selected for standardiza-tion of the assays. The strongly positive serum sample thatwas used was derived from a patient with culture-provenenteroviral infection who also had a high titer in the CFT(titer, >256). TheA492 of this serum sample was greater thanfive times the absorbance of the cutoff serum sample for theIgM and IgA ELISA and greater than three times theabsorbance of the cutoff serum sample for the IgG ELISA.To reveal any nonspecific reactions, all 200 donor serumsamples were tested against the horseradish peroxidase-labelled control antigen made from uninfected cells. None ofthe sera reacted with the control antigen.

Patients. One hundred sixteen serum samples from 62patients were tested for enteroviral antibodies by CFT andELISA. A stool specimen, a throat swab, or CSF wasobtained from 35 patients for viral culture. For 14 patients,culture remained negative for enterovirus, and enteroviruswas isolated from 21 patients. Data for these 21 patients arerepresented in Table 1. A recent enterovirus infection wasdiagnosed in 7 patients by CFT (33%) and in 14 patients byELISA (67%; IgG 6 times, IgM 10 times, and IgA 6 times).In seven patients, no enteroviral antibodies were detected byELISA. In two of them (patients 5 and 14), such antibodies

were detected by CFT. Two patients were eventually diag-nosed as also having a bacterial infection at the time of bloodsampling (patients 5 and 6). Fifteen of the 21 patientspresented with meningeal signs (Table 1, patients 7 to 21).Enterovirus was isolated either from CSF (n = 9), from athroat swab (n = 2), or from a fecal sample (n = 4). In 5 of15 patients, recent enteroviral infection was diagnosed byCFT (33%), and in 10 patients, it was diagnosed by ELISA(67%; IgG five times, IgM seven times, and IgA four times).Five patients presented with meningeal signs during a smallepidemic of echovirus 30 meningitis (patients 17 to 21). Serafrom these patients were also tested for neutralizing antibod-ies against echovirus 30. All patients had a neutralizingantibody titer of 2128. After fractionation of these serumsamples, neutralizing antibodies of the IgM class could bedetected in all samples as well. In 2 of the 14 patients whosecultures were negative for enterovirus, a recent enterovirusinfection was diagnosed by serology. The clinical diagnosisfor these two patients was Bornholm disease and myoperi-carditis (Table 2, patients 26 and 28). Among the remaining12 of 14 culture-negative patients, other infections weredocumented for 6: parainfluenza virus (n = 2), herpessimplex virus (n = 1), varicella-zoster virus (n = 1), urea-plasma infection (n = 1), and streptococcus infection (n = 1);

COXSACKIEVIRUS Bl-BASED ELISA 3243

TABLE 2. Patients with cardiomyopathy, myocarditis/pericarditis, or pleuritisa

Patient Clinical diagnosis Virus No. of days CFT Neutralizingno. Age symptoms and signs isolate Material after onset titer IgG IgM IgA antibodyafteronsettiter ~~~~~~~~~~~~titer22 1 yr Cardiomyopathy NT Serum +4 NT ++ + + NT

Serum +44 64 ++ - -23 1 day Cardiomyopathy NT Serum 0 >256 - - - NT24 20 yr Pleuritis NT Serum +4 128 + + - - NT

Serum +14 128 + +25 25 yr Pleuritis NT Serum 0 32 - - - NT

Serum +12 >256 + + + + +26 37 yr Myopericarditis Negative Serum +9 NT - - + CVB1, 256

Serum +34 >256 + - - CVB1, 25627 26 yr Pleuritis NT Serum +3 >256 - + - CVB4, 256

Serum +11 >256 - ++ + CVB4, 51228 28 yr Bornholm disease Negative Serum +6 128 - + - CVB4, 128

Serum +16 128 - + - CVB4, 64

a Abbreviations: NT, not tested; CVB1 and CVB4, coxsackieviruses B1 and B4, respectively. Symbols: -, negative (IgG < the cutoff; IgM and IgA < 2x thecutoff); +, positive (IgG, 1x to 2x the cutoff; IgM and IgA, 2x to 5x the cutoff); + +, strongly positive (IgG > 2x the cutoff; IgM and IgA > 5x the cutoff).

in three cases no infection was documented, and in threecases a noninfectious cause was found. All serum samplesfrom these patients were negative by ELISA and CFT.For 27 patients, no material for culture was available;

thus, only serologic tests were performed. For 21 of thesepatients, enterovirus-specific serology was negative. In fiveof these patients there was serologic evidence for other viralinfections, namely, by herpes simplex virus (n = 1), parain-fluenza virus (n = 3), and respiratory syncytial virus (n = 1).For six patients, serology indicated a past or recent entero-virus infection. Data for four of these patients are presentedin Table 2 (patients 22, 24, 25, and 27). One patient presentedwith convulsions, and serologic results indicated a recententerovirus infection: high titers were determined by CFT,and the patient was positive for both IgM and IgA antibod-ies. In one patient, IgM antibodies against enterovirus weredetected. However, in this patient there were also IgMantibodies against cytomegalovirus, Epstein-Barr virus, her-pes simplex virus, and varicella-zoster virus, which means

that these antibody responses were probably nonspecific.Seven patients with cardiomyopathy, myocarditis/pericar-

ditis, or pleuritis (Bornholm disease) as a clinical diagnosisare represented in Table 2. One patient died of cardiomyop-athy at birth (patient 23). The high titer in the CFT probablyrepresents maternal antibodies. Among the remaining sixpatients, recent enteroviral infection was diagnosed in fourby CFT (67%) and in all by ELISA (100%; IgG four times,IgM four times, and IgA four times). For three patients,serum was also tested for neutralizing antibodies againstcoxsackieviruses Bi, B2, B3, B4, and B5. Two patients hadneutralizing antibodies against coxsackievirus B4, and one

patient had neutralizing antibodies against coxsackievirusBi. In these samples neutralizing antibodies of the IgM classcould be detected, which indicates recent infection.

Fifteen serum samples that were drawn from asymptom-atic contacts of patients with poliomyelitis were tested bythe coxsackievirus Bl-based antibody-capture ELISA andby CFT. All sera contained poliovirus type 3-specific IgMantibodies as determined at the National Institute of PublicHealth and Environmental Protection. The results are sum-

marized in Table 3. All sera were positive by the IgMELISA, 12 serum samples were positive by the IgA ELISA,and 9 serum samples were positive by the IgG-captureELISA. Seven serum samples were positive by all three

ELISAs. Nine serum samples were positive by CFT (titer,>256).

Specificity of the assays. The results for the control serawith known antibody specificity are summarized in Table 4.Of the 20 serum samples with rheumatoid factors of the IgMtype, only 1 was reactive in the IgM and IgA ELISA. Threeof 21 serum samples with antinuclear antibodies reacted inthe IgM ELISA, but none of them reacted in the IgA ELISA.One of 20 Venereal Disease Research Laboratory test-positive serum samples reacted with enteroviral antigen inthe IgM ELISA as well as in the IgA ELISA. Unfortunately,reactivity to the enteroviral antigen was encountered in half(10 of 20) of the Monospot-positive sera and in the majority(10 of 15) of sera containing IgM and IgA antibodies againstM. pneumoniae. All sera that reacted with the enteroviralantigen in the ELISA were also tested with the controlantigen and by CFT. None of the sera reacted with thecontrol antigen in the IgG and IgA ELISAs. In the IgM

TABLE 3. Sera taken during recent Dutch epidemic ofpoliomyelitis type 3a

Patient Poliovirus GETno. Age type-specific titer IgG IgM IgA

29 9 yr 3 >256 + +30 17 yr 3 4 ++ ++ +31 10yr 3 64 ++ + +32 15yr 3 >256 ++ ++ -

33 11yr 3 8 - + +34 9 yr 3 16 - ++ ++35 18 yr 1,2,3 >256 ++ ++ ++36 23 yr 1,2,3b 64 - +37 5 yr 3 64 ++ + ++38 9 yr 3 >256 ++ ++ ++39 17 yr 3 >256 ++ ++ +40 3 yr 3 >256 ++ ++ ++41 31 yr 1,2,3 >256 - + ++42 36yr 3 >256 - ++ +

43 31 yr 1,2,3 >256 - + ++

a Symbols: -, negative (IgG < the cutoff; IgM and IgA < 2x the cutoff); +,positive (IgG, 1x to 2 x the cutoff; IgM and IgA, 2x to 5 x the cutoff); ++,strongly positive (IgG > 2x the cutoff; IgM and IgA > 5 x the cutoff).

b Vaccinated.

VOL. 31, 1993

3244 SWANINK ET AL.

TABLE 4. Control sera with known antibodies

No. of isolates positive/Known antibodies no. tested for:

IgG IgM IgA

Rheumatoid factor IgM 0/20 1/20 1/20Antinuclear antibodies 1/21 3/21 0/21VDRLU and TP-MHAb 0/20 1/20 1/20Heterophile antibodies 1/20 10/20 6/20Mycoplasma IgM and IgA 0/15 10/15 2/15

a VDRL, antibodies detected by the Veneral Disease Research Laboratorytest.

b TP-MHA, antibodies detected by the Treponema pallidum microhemag-glutination test.

ELISA with the control antigen, some reactivity was foundin the Monospot-positive sera and sera with antibodiesagainst M. pneumoniae. However, reactivity with the con-trol antigen was less prominent than reactivity with theenteroviral antigen. In 2 of 10 Monospot-positive serumsamples and in 2 of 10 serum samples containing mycoplas-mal antibodies, high titers were found by CFT, suggestingrecent enterovirus infection or nonspecific reactivity occur-ring in CFT as well. All patient sera that were positive byELISA with the enteroviral antigen were tested with thecontrol antigen as well. None were reactive.

DISCUSSION

The aim of this study was to evaluate the sensitivity andspecificity of a coxsackievirus Bl-based antibody-captureELISA as a simple test with broad reactivity for enterovi-ruses. It is known that cross-reactive epitopes are located onthe empty capsids of enteroviruses. Heating to 56°C con-verts virions to empty capsids, thereby changing the antige-nicity to group-specific reactivity (19, 24, 29). The presentstudy confirms this broad specificity. During the develop-ment of these ELISAs we have made conjugated antigens ofthe 10 enterovirus types with the highest incidence in TheNetherlands. These conjugates have been tested either sep-arately or as antigen mixtures. So far, infections by thefollowing known serotypes have been recognized: echovi-ruses 6, 7, 9, 11, 14, 19, 25, and 30; coxsackieviruses Bi, B2,B4, and B5; and poliovirus type 3. A combination of variousserotypes pooled with an antigen mixture did not furtherbroaden reactivity for enteroviruses, nor did the use of otherserotypes as the antigen (data not shown).The cutoff values were arbitrarily chosen at the 95%

confidence level of reactivity in a population of 200 healthyblood donors. In March (the time of collection of the donorsera), the incidence of acute enterovirus infections is low,but the exact prevalence was not known. This might haveinfluenced the sensitivity and specificity of the tests.Among the patients with culture-positive enterovirus in-

fections, seven patients were missed by ELISA. In two ofthese patients (patients 5 and 6), there was also evidence ofbacterial infection (Eschenchia coli and Staphylococcusaureus). Therefore, the onset of enteroviral infection is notaccurately known for these cases. It may be that the sampleswere drawn too early to find specific antibodies (patients 9and 16). Unfortunately, there was no second sample. It mayalso be that the ELISA was falsely negative. Consistent withthis possibility, enteroviral antibodies were detected in pa-tients 5 and 14 by CFT but not by ELISA. Patient 8 sufferedfrom neonatal meningitis and could have had a delayed

antibody response, as seen more often with neonatal infec-tions. Enteroviral antibodies were detected by ELISA inCSF from only two of eight patients. This is probably due tothe time of sampling. CSF is usually taken at the onset ofsymptoms, when antibody production is just starting. Anti-bodies were indeed detected in two of three CSF samplesthat were taken more than 2 days after onset. Thus, ELISAwith CSF may be useful when CSF is taken later than 2 daysafter onset, comparable to what is found with serum (seebelow). This has also been described by van Loon et al. (32),who did not detect antibodies against herpes simplex virus inCSF within 6 days after the onset of illness. In contrast, theyfound herpes simplex virus IgG and IgA in CSF from allpatients with herpes simplex virus encephalitis more than 10days after onset.

In patients with cardiomyopathy, myocarditis/pericarditis,or pleuritis, cultures are usually not performed or are per-formed with a considerable delay. When clinical symptomsbecome clear, cultures are usually negative. In such pa-tients, only serology may reveal a relationship with entero-viral infection. We have tested serum samples from seven ofsuch patients by the coxsackievirus group Bl-based ELISAand by CFT with the antigen mixture (Table 2). For onepatient, serology was negative by ELISA. This patient(patient 23) was born with cardiomyopathy and died at birth.Polymerase chain reactions with heart biopsies from thispatient, taken at autopsy, were negative for enterovirus.There is a discrepancy in IgM and IgA responses: nine

patients had an IgM response but no IgA response, and threepatients had an IgA response but no IgM response. This hasbeen described before (26), but the reason for this discrep-ancy is not known.The sensitivity of the antibody-capture ELISA is good

when the outcome of the IgG, IgM, and IgA assays arecombined. The sensitivity ranges from 67% in cases ofculture-proven enteroviral infections and meningitis (14 of 21and 10 of 15, respectively) to 86% (6 of 7) in cases ofmyocarditis, pericarditis, or pleuritis. The sensitivity in thecase of poliovirus type 3-specific IgM-positive sera was100% (15 of 15). The sensitivity is clearly related to the timewhen serum samples are taken. When serum samples aretaken within 3 days after the onset of symptoms, thesensitivity is low (5 of 22 [23%]). When serum is drawn laterthan 3 days after onset, the sensitivity is good (20 of 24[83%]). The CFT with the antigen mixture is less sensitivethan the ELISA (44 versus 81%), especially when only oneserum sample is available. When the ELISA and CFT werecombined, the sensitivity was 88%. This ELISA was foundto be as sensitive as an antibody-capture ELISA usingdifferent antigens (3, 22). The sensitivity of the IgM ELISAalone is 58%. When IgM and IgA are combined, the sensi-tivity increases to 67%. The IgG-capture assay adds another16% to the sensitivity. The relatively low sensitivity of theIgM and IgA ELISA may be due to the recurrent characterof enterovirus infections. Recurrent enteroviral infectionwith different serotypes causes boosts in responses to com-mon epitopes. This may result in diminished antibody re-sponses of both IgM and IgA. For cytomegalovirus, as anexample, it is known that IgM and IgA responses are lessprominent or even absent in recurrent infection comparedwith the primary infection (31, 33).

Antibody-capture ELISAs have proven to perform excel-lently with CSF (32). This is because of the low backgroundlevels of nonspecific Igs in CSF as long as the blood-brainbarrier is not damaged. This is particularly so for theIgG-capture assay as is illustrated by patients 11 and 12 in

J. CLIN. MICROBIOL.

COXSACKIEVIRUS Bl-BASED ELISA 3245

our study. It can be envisaged that the IgG-capture ELISAwill also perform well in case of high antibody levels(hyperimmunization), as has been reported to occur inpersistent infection (2, 7, 37). This has already been inves-tigated by using poliovirus type-specific ELISAs for postpo-lio syndrome, with which we did not find evidence for viralpersistence (23). Similar studies are underway for postviralfatigue syndrome, dilated cardiomyopathy, and polymyositis(16).As expected, there was no interference by rheumatoid

factor. There was little cross-reactivity in sera with antinu-clear antibodies and none in sera with positive VenerealDisease Research Laboratory test reactions. Unfortunately,there was significant cross-reactivity with sera that con-tained heterophile antibodies or M. pneumoniae IgM andIgA antibodies. This is probably due to polyclonal B-cellactivation, which is common in Epstein-Barr virus infectionand mycoplasmal infection (4, 5, 11). Cross-reactivity oc-curred mainly in the IgM ELISA and less in the IgA ELISA.

Pitfalls in our assay seem to be similar to those reported byothers (13). The major disadvantage of this test is its cross-reactivity with Epstein-Barr virus and M. pneumoniae.However, the clinical symptoms of these infections are quitedifferent from those of enteroviral infections, and thereforewe do not think this will be a major problem in clinicalpractice. Perhaps the use of synthetic peptides, which hasrecently been described by Cello et al. (10), may reducecross-reactivity. The great advantage of the test we de-scribed is that it is simple to perform, more sensitive than theCFT, and far less laborious than the neutralization test.

ACKNOWLEDGMENTS

This work was in part supported by grant 28-1969 from thePraeventiefonds and by a grant from the University Hospital Nij-megen.We thank A. M. van Loon from the National Institute of Public

Health and Environmental Protection for providing sera with polio-virus type 3-specific IgM.

REFERENCES1. Archard, L. C., N. E. Bowles, P. 0. Behan, E. J. Bell, and D.

Doyle. 1988. Postviral fatigue syndrome: persistence of entero-virus RNA in muscle and elevated creatine kinase. J. R. Soc.Med. 81:326-329.

2. Banatvala, J. E., G. Schernthaner, E. Schober, et al. 1985.Coxsackie B, mumps, rubella and cytomegalovirus specific IgMresponses in patients with juvenile-onset insulin-dependent di-abetes mellitus in Britain, Austria and Australia. Lancet i:1409-1412.

3. Bell, E. J., R. A. McCartney, D. Basquill, and A. K. R.Chaudhuri. 1986. ,u-Antibody capture ELISA for the rapiddiagnosis of enterovirus infections in patients with asepticmeningitis. J. Med. Virol. 19:213-217.

4. Biberfeld, G. 1977. Activation of human lymphocyte subpopu-lations by mycoplasma pneumoniae. Scand. J. Immunol.6:1145-1150.

5. Biberfeld, G., and E. Gronowicz. 1976. Mycoplasma is a poly-clonal B-cell activator. Nature (London) 261:238-239.

6. Boman, J., B. Nilsson, and P. Juto. 1992. Serum IgA, IgG, andIgM responses to different enteroviruses as measured by acoxsackie B5-based indirect ELISA. J. Med. Virol. 38:32-35.

7. Bowles, N. E., V. Dubowitz, C. A. Sewry, and L. C. Archard.1987. Dermatomyositis, polymyositis, and coxsackie-B-virusinfection. Lancet i:1004-1007.

8. Bowles, N. E., E. J. G. Olsen, P. J. Richardson, and L. C.Archard. 1986. Detection of coxsackie B virus specific RNAsequences in myocardial biopsy samples from patients withmyocarditis and dilated cardiomyopathy. Lancet i:1120-1122.

9. Casey, H. L. 1965. Adaptation of the LBCF microtechnique, p.31-34. In Standardized diagnostic complement fixation andadaptation to microtest, part II. U.S. Public Health Servicemonograph no. 74. U.S. Public Health Service, Washington,D.C.

10. Cello, J., A. Samuelson, P. Stalhandske, B. Svennerholm, S.Jeansson, and M. Forsgren. 1993. Identification of group-com-mon linear epitopes in structural and nonstructural proteins ofenteroviruses by using synthetic peptides. J. Clin. Microbiol.31:911-916.

11. Cremer, N. E., V. L. Devlin, J. L. Riggs, and S. J. Hagens. 1984.Anomalous antibody response in viral infection: specific stimu-lation or polyclonal activation? J. Clin. Microbiol. 20:468-472.

12. Cunningham, L., N. E. Bowles, and L. C. Archard. 1991.Persistent virus infection of muscle in postviral fatigue syn-drome. Br. Med. Bull. 47:852-871.

13. Glimaker, M., A. Samuelson, L. Magnius, A. Ehrnst, P. Olcen,and M. Forsgren. 1992. Early diagnosis of enteroviral meningitisby detection of specific IgM antibodies with a solid-phasereverse immunosorbent test (SPRIST) and ,u-capture ELISA. J.Med. Virol. 36:193-201.

14. Gow, J. W., W. M. H. Behan, G. B. Clements, C. Woodall, M.Riding, and P. 0. Behan. 1991. Enteroviral RNA sequencesdetected by polymerase chain reaction in muscle of patientswith postviral fatigue syndrome. Br. Med. J. 302:692-698.

15. Inouye, S., and R. Kono. 1981. Simple technique for separationof concentrated immunoglobulin M fractions by multi-samplegel chromatography and its application to rubella serodiagnosis.J. Clin. Microbiol. 13:248-251.

16. Jongen, P. J. H., G. J. Zoll, M. Beaumont, W. J. G. Melchers,L. B. A. van de Putte, and J. M. D. Galama. 1993. Polymyositisand dermatomyositis: no persistence of enterovirus and enceph-alomyocarditis-virus RNA in muscle. Ann. Rheum. Dis. 52:575-578.

17. Kapsenberg, J. G. 1988. Picornaviridae: the enteroviruses (po-lioviruses, coxsackieviruses, echoviruses), p. 692-722. In E. H.Lennette, P. Halonen, and F. A. Murphy (ed.), Laboratorydiagnosis of infectious diseases: principles and practice, vol. II.Viral, rickettsial and chlamydial diseases. Springer-Verlag,Inc., New York.

18. Kapsenberg, J. G., A. Ras, and J. Korte. 1979. Improvement ofenterovirus neutralization by treatment with sodium deoxy-cholate or chloroform. Intervirology 12:329-334.

19. Katze, M. G., and R. L. Crowell. 1980. Indirect enzyme-linkedimmunosorbent assay (ELISA) for the detection of coxsackie-virus group B antibodies. J. Gen. Virol. 48:225-229.

20. King, M. L., A. Shaikh, D. Bidwell, A. Voller, and J. E.Banatvala. 1983. Coxsackie B-virus specific IgM responses inchildren with insulin-dependent diabetes mellitus. J. Pediatr.100:15-20.

21. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall.1951. Protein measurement with the Folin phenol reagent. J.Biol. Chem. 193:265-275.

22. McCartney, R. A., J. E. Banatvala, and E. J. Bell. 1986. Routineuse of ,u-antibody-capture ELISA for the serological diagnosisof coxsackie B virus infections. J. Virol. Methods 19:205-212.

23. Melchers, W., M. de Visser, P. Jongen, A. van Loon, R.Nibbeling, P. Oostvogel, D. Willemse, and J. Galama. 1992. Thepostpolio syndrome: no evidence for poliovirus persistence.Ann. Neurol. 32:728-732.

24. Melnick, J. L. 1990. Enteroviruses: polioviruses, coxsackievi-ruses, echoviruses, and newer enteroviruses; antigenic charac-teristics, p. 551-553. In B. N. Fields and D. M. Knipe (ed.),Virology, 2nd ed. Raven Press, Ltd., New York.

25. Melnick, J. L., H. A. Wenner, and C. A. Philips. 1979. Entero-viruses: serologic diagnosis, p. 521-526. In E. H. Lennette andN. J. Schmidt (ed.), Diagnostic procedures for viral rickettsialand chlamydial infections, 5th ed. American Public HealthAssociation, Inc., Washington, D.C.

26. Muir, P., N. B. Singh, and J. E. Banatvala. 1990. Enterovirus-specific serum IgA antibody responses in patients with acuteinfections, chronic cardiac disease, and recently diagnosedinsulin-dependent diabetes mellitus. J. Med. Virol. 32:236-242.

VOL. 31, 1993

3246 SWANINK ET AL.

27. Nibbeling, R., M. Agboatwala, T. Naquib, et al. A poliovirustype-specific IgM antibody capture enzyme linked immunosor-bent assay for the rapid diagnosis of poliomyelitis. Submittedfor publication.

28. Ray, C. G., L. L. Minnich, and P. C. Johnson. 1979. Selectivepolymyositis induced by coxsackievirus Bi in mice. J. Infect.Dis. 140:239-243.

29. Samuelson, A., E. Skoog, and M. Forsgren. 1990. Aspects onserodiagnosis of enterovirus infections by ELISA. Serodiagn.Immunother. Infect. Dis. 4:395-406.

30. Sharief, M. K., R. Hentges, and M. Ciardi. 1991. Intrathecalimmune response in patients with the post-polio syndrome. N.Engl. J. Med. 325:749-755.

31. Van Loon, A. M., F. W. A. Heessen, and J. T. M. van der Logt.1987. Antibody isotype response after human cytomegalovirusinfection. J. Virol. Methods 15:101-107.

32. van Loon, A. M., J. T. M. van der Logt, F. W. A. Heessen, B.Postma, and M. F. Peeters. 1989. Diagnosis of herpes simplexvirus encephalitis by detection of virus-specific immunoglobu-lins A and G in serum and cerebrospinal fluid by using anantibody-capture enzyme-linked immunosorbent assay. J. Clin.Microbiol. 27:1983-1987.

33. van Loon, A. M., J. T. M. van der Logt, F. W. A. Heessen, andJ. van der Veen. 1985. Quantitation of immunoglobulin Eantibody to cytomegalovirus by antibody capture enzyme-

linked immunosorbent assay. J. Clin. Microbiol. 21:558-561.34. Vella, C., C. L. Brown, and D. A. McCarthy. 1992. Coxsackie-

virus B4 infection of the mouse pancreas: acute and persistentinfection. J. Gen. Virol. 73:1387-1394.

35. Wilson, M. B., and P. K. Nakane. 1978. Recent developments ofthe periodate method of conjugating horseradish peroxidase(HRPO) to antibodies, p. 215-224. In W. Knap, K. Holobar, andG. Wick (ed.), Immunofluorescence and related staining tech-niques. Elsevier/North-Holland Biomedical Press, Amsterdam.

36. World Health Organization. 1992. New approaches to poliovirusdiagnosis using laboratory techniques: memorandum from aWHO meeting. Bull. W.H.O. 70:27-33.

37. Yousef, G. E., E. J. Bell, G. F. Mann, V. Murugesan, D. G.Smith, R. A. McCartney, and J. F. Mowbray. 1988. Chronicenterovirus infection in patients with postviral fatigue syn-drome. Lancet i:146-149.

38. Yousef, G. E., D. A. Isenberg, and J. F. Mowbray. 1990.Detection of enterovirus specific RNA sequences in musclebiopsy specimens from patients with adult onset myositis. Ann.Rheum. Dis. 49:310-315.

39. Zoll, G. J., W. J. G. Melchers, H. Kopecka, G. Jambroes,H. J. A. van der Poel, and J. M. D. Galama. 1992. Generalprimer-mediated polymerase chain reaction for detection ofenteroviruses: application for diagnostic routine and persistentinfections. J. Clin. Microbiol. 30:160-165.

J. CLIN. MICROBIOL.