Evaluation of rubella immune status by three commercial enzyme linked immunosorbent assays

0095-1137/88/050990-05$02.00/0

CopyrightC 1988, AmericanSocietyforMicrobiology

Evaluation of Rubella

Immune

Status

by

Three

Commercial

Enzyme-Linked Immunosorbent Assays

PETERR. FIELD,* DAVID W. T. HO, ANDANTHONY L. CUNNINGHAM

Virology Department, Instituteof Clinical Pathology and Medical Research, Westmead, NewSouth Wales

2145,

Australia

Received29 September 1987/Accepted 16 February 1988

Three commercial indirect enzyme-linked immunosorbent assays (ELISAs) (Enzygnost-Rubella, RUBE-LISA, and ORTHO Rubella)wereevaluatedfor thedeterminationof immune statusby testing 1,090serum

specimens, 410 of whichwerefromnonimmunepatients. In comparison with the standard reference technique,

thehemagglutination inhibition (HAI)test,thesensitivities ofORTHORubella(100%)andEnzygnost-Rubella (99.26%)wereexcellent, whereasthesensitivityof RUBELISA(95.60%)wasmarginallylower becauseof the

inabilityof thisassaytodetectantibody in 22% of theserumspecimenswith HAI titers of 10 and11% ofsera

withHAI titers of 20. Thespecificity of all threesystems was>97%. Therewas alinearcorrelationbetween meanELISA values and increasing HAI titers (r 2 0.94). Both ORTHORubella andEnzygnost-Rubellawere

shown tobe suitable replacements for the HAItest, providedthatan equivocal zoneis incorporated in the ORTHOsystem andonlyunheated sera areused in theEnzygnost system.

Rubella is usuallyarelatively mild exanthematous illness inchildren oradults.Infection ofwomeninearlypregnancy may result in congenital abnormalities of the fetus. The

control of rubella infectionin Australia isbeingattempted by routine vaccination of schoolgirls aged 10 to 14 years and selected immunization ofnonpregnant seronegative women

of childbearing age (13). Accurate determination of the immune status ofwomenofchildbearing age anddiagnosis of recent infection in pregnant women are of paramount importance.

The hemagglutination inhibition (HAI)testhas been

con-sidered the standard procedure for diagnosis of rubella infection and evaluation of immune status (17). This test shows excellent correlationfor protection tothefetus with themorecumbersome andtechnically difficult neutralization test (12), which some regard as the ultimate standard.

However, the HAI test is time-consuming, the incomplete removal of nonspecific inhibitors and agglutinins (7) can

diminish its accuracy, and standardization between labora-tories is difficult. For these reasons, most laboratories

rec-ommend furthertesting foranypregnantwomanwithatiter of 10or20 whocomesintocontactwitha caseof rubellaor

whodevelopsarubella-like rash. On the other hand,asthe

HAI test has been in use for such a long time, many

clinicians arefamiliarwith the significance of the results of thistest.

In recent years, solid-phase enzyme-linked

immunosor-bent assay (ELISA) methods have been advocated as

re-placements for the HAI test. ELISA methods have the advantage of the potential for automation by whichaprecise

numerical readout which replaces visual interpretation and quantitation can be achieved on a single dilution ofserum.

Further, given the required reagents and equipment, these testsaresimple and rapid toperformand have been shown to have precisions similar to that of HAI (8). In-house ELISAs are being progressively replaced by a number of

commercial kits. Prestandardized reagents are available so that results can be meaningfully compared between

labora-tories. By linking microplate readers to microcomputers

* Corresponding author.

through comprehensive butflexible softwarepackages, it is now possible to take advantage of the prodigious data-handling capacity of desk-topcomputers,thereby allowinga

single operator to process more than 2,000 immunoassays

per day. The computers save time in the analysis and reporting of results. Hence,anincreasing number of

labora-tories withoutexperience and expertise in rubella serology

are using the relatively simple commercial kits for rubella

antibody assay. All kits available in Australia are of the

indirect type, employing rubella antigen attachedtoa solid

phase. However, few in-depth independent evaluations of such kits have been carried out. This communication de-scribes the evaluation of three commercial indirect ELISA systems and compares them with an in-house HAI testfor the determination ofimmunestatus.

MATERIALSAND METHODS

Clinicalspecimens.Thestudy materials consisted ofserum

specimens from 1,090pregnant women (680serawere from

immunepatients and 410sera werefrom susceptible patients as determined by HAI). The sera tested were submitted

mainly by hospitals, pathologists, and medical practitioners in NewSouth Wales forroutine antenatal testing. The sera wereinitially tested by HAI and then storedat-20°C forno

longer than 10 months until subsequent testing by the three ELISAsystems.

The range of serum samples studied was purposely

skewed toward seronegative and low-positive levels,

be-cause the aim of an immunity screening test is to detect

susceptible individuals. Furthermore, previous works inthis laboratory (unpublished) and elsewhere (3, 18, 22) have shownthatdiscrepancies tendtooccurwith seracontaining

lowlevels of HAI antibodies.

HAI. Anin-house microtiter techniquewasused (5),with trypsinized human O cells ata concentration of 0.3% with

HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) buffer (pH 6.2) containing 0.5% bovine albumin frac-tion V as the diluent and 4 U of commercially available

rubella antigen (Behringwerke AG, Marburg, Federal Re-public of Germany). Sera were pretreated with heparin-manganouschloride to remove nonspecificinhibitors.

Con-990

on April 11, 2020 by guest

http://jcm.asm.org/

trol serawith titers of <10, 20, 80, and 320were included in

each run.

An antibody titer of <10 was considered nonimmune,

whereas atiter of -10wasconsidered immune.

ELISA. Enzyme immunoassays were conducted on all

serum specimens by three indirect ELISA systems for

rubella immunoglobulin G (IgG) with commercial reagents.

The systems used were RUBELISA (M.A. Bioproducts,

Walkersville, Md.), Enzygnost-Rubella (Behringwerke) and ORTHO Rubella IgG ELISA Test Systemn (Ortho Diagnostic Systems, Don Mills, Ontario, Canada).

Alltest systems usedthe principle of indirect ELISA and the procedures followed were those recommended by the

manufacturers, except for one modification in the

Enzy-gnost-Rubella system. This modification involved the

stop-ping of theenzymereactionat20to25°C afterashortertime

(usually 30 min) than that recommended by the manufacturer (45 min). When the reaction was stopped after 45 min,

background (control antigen) absorbance valuesrose

consid-erably, resulting in somefalse-negative results. In compari-son, it was found in an earlier study (4) that the enzyme

reaction in theEnzygnost-Rubellatestneededtobe stopped after10 to 15 min.

Therearesomedifferencesintheprincipalcharacteristics

of thethree ELISA systems. Thesolid-phasecomponentin the RUBELISA and Enzygnost-Rubella systems consisted ofantigen attached to microwells in removable strips in a

plastic tray. TheORTHO Rubellasystemuseda

microdilu-tion plate, but only the central 48 wells were coated with antigen, presumably to remove edge effects. (The latter systemisnowavailable inaremovablestrip format). Except

for the ORTHO Rubella system, all systems employed a

control antigento testfornonspecific reactivity.

Serumdilutions were heatinactivated at56°Cfor 30 min prior to testing by the ORTHO Rubella system. Unheated

sera were used in the RUBELISA and Enzygnost-Rubella

systems.

All systems usedan anti-human IgG conjugate; this

con-jugate waspolyclonal with the RUBELISA and

Enzygnost-Rubella systems, but the ORTHO Rubella system used a

monoclonal anti-human IgG(murine) highly specific for the Fc portion of the heavy chain of human IgG (10). The compositions of serum and conjugate diluents were not available from the manufacturer. Washing was performed

with a processor (Behring ELISA Processor II;

Behring-werke). Theabsorbances ofthe solutions in each wellwere

measured

directly

in the plate with a vertically measuring photometer. For the Enzygnost-Rubella system, Behring

ELISAProcessorIIwas used, and for theRUBELISAand ORTHO Rubella systems, an automatic reader (MR 580

MicroELISA Auto Reader; Dynatech Laboratories, Inc.,

Alexandria,

Va.)was employed.

Cost. The costofreagentsand materials forourin-house

HAI test was approximately $0.40 per test. However, the

use of commercial HAI kit sets can increase this cost to

approximately $2.00, which is comparable to ELISA costs

($1.90, $2.30,

and

$2.95

for theORTHO Rubella, Enzygnost-Rubella, and RUBELISA systems, respectively). (Allcosts areexpressedin Australiandollars.)

The methodsfor determiningthe presenceof rubellaIgG appearin detailbelow (seeTable 3,footnotes).

Discrepantandequivocal results. All patientserum

speci-mensshowing discrepantresultsweretestedrepeatedlyuntil

consistent results were obtained. Equivocal results were

retestedin duplicate.

Measurementof intraassay and interassay variations. The

precision (coefficientofvariation)wasdetermined bytesting

commercial controlserawithhigh- and low-positive ELISA

values in each of the three ELISA systems. Theintraassay

precision was calculated by testing seven replicates of the twocontrolsoneach ofthree consecutive days by the three

ELISA systems. For thedetermination ofinterassay preci-sion, the two controls had 34, 28, and 30 runs in the RUBELISA,Enzygnost-Rubella, andORTHO Rubella sys-tems, respectively.

RESULTS

Evaluation of immunestatus.Thecomparison of thethree ELISA methods with the HAItestfor thedetermination of immunestatusin 1,090 antenatal patients isshown in Table 1. Performance characteristics were calculated with refer-ence to HAI. The RUBELISA, Enzygnost-Rubella, and ORTHO Rubellasystemsshowedsensitivities of 95.6,99.26,

and 100% respectively, while their respective specificities

were99.76, 100, and 97.32%.

The Enzygnost-Rubella and ORTHO Rubella systems showed excellent overallagreementwith the HAItest(99.54

and 98.99%, respectively), and similarly, the

Enzygnost-Rubella and ORTHO Rubella systems demonstrated the TABLE 1. Comparisonof threeindirectELISA methods with the HAItestforimmunestatus in1,090antenatalpatients'

No.ofspecimensthatgive Predictive value (%) for

theindicatedresult with

Method andresult the HAI test Sensitivity Specificity Agreement the

indicated

result

Positive Negative Positive Negative

(n=680) (n =410)

RUBELISA 95.60 99.76 95.60 99.84 89.69

Positive 633 1

Negative 47 409

Enzygnost-Rubella 99.26 100 99.54 100 98.79

Positive 675 0

Negative 5 410

ORTHO Rubella 100 97.32 98.99 98.41 100

Positive 680 il

Negative 0 399

aPerformance characteristics(sensitivity,specificity,agreement, andpredictivevalue)arein relationtothe HAItest.

bPercentsensitivity = 100 x ([totalnumber ofspecimenspositive totheHAI test] - [numberofspecimensnegativetoELISA but positivetotheHAI

test])/totalnumber ofspecimenspositivetothe HAItest.

CPercentspecificity = 100x ([totalnumber ofspecimens negativetothe HAItest] - [numberofspecimens positivetoELISA but negativetothe HAI

test])/total number ofspecimensnegativetothe HAItest.

on April 11, 2020 by guest

http://jcm.asm.org/

TABLE 2. Discrepancies in threeELISA systemsaccordingtoHAItiter

No.of serumspecimens(%)giving the indicated result in the ELISAsystem

HAItiter No.ofsera RUBELISA Enzygnost-Rubella ORTHO Rubella

Positive Negative Positive Negative Positive Negative

<10 410 1(0.24) 409(99.76) 0(0) 410(0) 11(2.68) 399 (97.32)

10 101 79(78.22) 22(21.78) 98(97.03) 3(2.97) 101(100) 0(0)

20 218 193(88.53) 25(11.47) 216(99.08) 2(0.92) 218(100) 0(0)

highest combined predictive values for positivity (100 and 98.41%, respectively) and negativity (98.79 and 100%,

re-spectively).

Comparison of discordant ELISA resultsinrelationtoHAI titer. Itis noteworthy that the only discrepancies occurred with patients having HAI titers of <20. Table 2 shows a

breakdown of the discrepancies according to the HAI titer.

Of319serawithHAI titers of 10 and 20, 47 (14.7%) showed anabsence of priorexposure torubella when testedby the RUBELISA system, whereas only 5 (1.6%) in this group

showednoantibody when tested by the Enzygnost-Rubella system. Itwas significant that the disagreementwasgreater withserahaving HAI titers of 10 than with those having HAI

titers of 20. Of101 sera with HAI titers of10, 22 (21.8%) were nonreactive when tested by the RUBELISA system, whereas of 218serawith HAItiters of 20, 25 (11.5%)were

negative. In comparing the results obtained by the Enzy-gnost-Rubella system with HAI titers, 3 of 101 (3%) sera

withHAI titers of 10werenegative while 2 of 218(0.9%)sera

with HAI titers of 20were negative. The ORTHO Rubella system exhibited paramount sensitivity by detecting anti-body in all the HAI low-titer positive sera. These results

clearly demonstrate the differences in sensitivityof the three ELISA systems in the HAI low-titer range. In this range

(Table2), only the RUBELISAsystemshowedasensitivity

significantly different from those ofthe Enzygnost-Rubella and ORTHO Rubella systems and the HAI test by the chi-squaretest (P< 0.001), butoverall (Table 1),the three ELISAs andHAI showed no significant differences in

sen-sitivity (P > 0.10).

The difference in specificity waslesspronounced. Of 410

HAI seronegative samples, only 1 was repeatedly positive

bytheRUBELISAsystem(absorbancecutoff,0.17; RUBE-LISA value, 0.18). Of 11 (2.68%) sera reactive in the

ORTHO Rubellasystem, 10werejust positive in therange

from 0.20 to 0.25 (the actual standardized ELISA values

obtainedwere0.20[5 samples],0.21[2samples], 0.23,0.24,

and0.25) and 1 hadavalue of 0.30.

Comparisonof HAI titers withELISA values. The

relation-shipbetween HAI titers andmean absorbance values(with

standarderror) obtainedfor each titergroupwhen testedby

each of the three ELISAsystemsis shown in Table 3. This

comparison showed good correlation (r 2 0.94) between

HAI titers, after transformation to natural logs, and the corresponding ELISA mean absorbance values. All three ELISA absorbance values were found to increase propor-tionately with increasing HAI titers; thus, the antibody

levels determined by the ELISAs reflect HAI titers. Precision. The results showeda high degree of

reproduc-ibility for each ELISAsystem. For the RUBELISAsystem, the coefficients of variation were 8.1 and 9.1 for the

high-positive and low-high-positivesera,respectively, for the

intraas-sayand10.0 and9.6, respectively, for theinterassay;forthe

Enzygnostsystem,the coefficients of variation for the

high-positive andlow-positiveserawere5.6 and5.9, respectively,

for the intraassay and 8.4 and 9.8, respectively, for the

interassay.FortheORTHORubellasystemthe

correspond-ing coefficients of variationwere 6.2 and 6.5, respectively,

for the intraassay and 9.8 and 9.2, respectively, for the

interassay.

DISCUSSION

Although ELISA methods are potentially more sensitive

than HAItests,little is known about the clinicalsignificance (i.e., the protective immunity) of low levels of antibody whichmaybe detected inHAI-negativeseraby ELISA (11). Therefore, rubella ELISA tests are calibrated to correlate with the currently accepted standard, the HAI test. This correlation has been the aim of the manufacturers of the three ELISAsystems evaluated in thecurrentstudy.

Whilenotabsolutelyaccuratein somehands (2), the HAI testisthe bestapproximationforimmunityatpresent,and in

TABLE 3. Correlation betweenmeanabsorbance values determined by the three ELISA systems for rubella IgG and HAI titers Meanabsorbance value(SE) determinedbya:

HAItiter No.of sera

RUBELIS.Ab Enzygnost- ORTHO

RUBELISAb

~~~~~Rubellac

Rubellad

<10 410 0.08(0.004) 0.05(0.002) 0.12 (0.002)

10 101 0.29 (0.029) 0.38(0.038) 0.47 (0.017)

20 218 0.32 (0.022) 0.45(0.030) 0.52 (0.035)

40 100 0.35(0.035) 0.55(0.055) 0.62 (0.062)

80 105 0.39(0.038) 0.80(0.078) 0.75 (0.073)

160 89 0.42 (0.045) 0.81(0.086) 0.91 (0.097)

320 47 0.48 (0.070) 1.34(0.195) 1.15 (0.168)

640 20 0.55 (0.123) 1.38(0.308) 1.23 (0.275)

a The

correlation

coefficients for the

correlation

of absorbance values determined by ELISA and HAI titers are 0.94, 0.97, and 0.98 for the RUBELISA,

Enzygnost-Rubella,and ORTHO Rubella systems,respectively.

bAbsorbancevalues of

.0.17

(obtainedfromacalibrationcurve) are

equivalent

to an HAI titer of -1:8. çValues are expressed as net absorbances. Values of -0.2 are positive.

dStandardizedvaluesof -0.20 (obtained by multiplying absorbance values with a high-positive control ratio) indicate immune status.

on April 11, 2020 by guest

http://jcm.asm.org/

our experience (unpublished), low HAI titers of 10 and 20 have shown complete agreement with single radial hemolysis (Rubazone; ScientificMeasuring Instruments, Sydney, Aus-tralia) and the passive latex agglutination test (Virogen; Wampole Laboratories, N.J.), two tests which are

unaf-fected by nonspecific inhibitors. It is well established that the presenceof rubella antibodyas detected by the HAI test

accurately correlates with clinicalprotection of the fetus (8).

The three ELISA systems showed good agreement with the HAI test and werefoundto be simple, rapid, and economical

alternatives.

The only disagreement in sensitivity was observed with sera having HAI titers of10 and 20. Whereas the ORTHO

Rubella and Enzygnost-Rubella systems showed excellent

sensitivities compared with the HAI test, the RUBELISA system was marginally less satisfactory. In some cases, it was apparent that the lack of sensitivity of the RUBELISA system maybe attributedto anelevatedabsorbance value of

the control antigen well relative to the absorbance value of the antigen well. The correlation curve of ELISA absorb-ance versus log ofHAItiter (datanot shown) had the most

gradual slope for the RUBELISA system, reflecting the

lower sensitivity of this system in comparison with the

Enzygnost-Rubella and ORTHO Rubella systems, whose

correlation curveshad steeperslopes. However, becauseof individual serum absorbances scattering around the mean

andoverlapping multiple HAIvalues, theELISA valueofa

singleserumshould not be used to report anequivalentHAI

titer.Hence,ELISAresultsfrom clinicallaboratoriesshould be accompanied by an

interpretative

comment.

All three ELISA systems achieved excellent specificity.

An explanation forthe discrepant sera, which all had posi-tive

reactions

veryclosetothe absorbance cutoff point,may

be that boththe ORTHO Rubellaand RUBELISA systems

have hadtheircut-off points between immunityand

suscep-tibility calibratedtocorrelatewithan HAItestcutoff titer of

8, whereas our in-house HAI test has a marginally higher cutoff titer of 10. To minimize the occurrence of false low-positive results in the ORTHO Rubella system, one of

two modifications could be incorporated. First, those sera

with standardized ELISA values in the range from 0.20 to 0.25could beregardedasequivocal,andsecond,the results mustbeconfirmed byadifferent

procedure.

Theinstructions for theORTHO Rubellasystem(OrthoDiagnostics Systems, Instructional booklet, 1986)drawattentiontothedifficulty of

categorizing

results close to the 0.20 criterion for immune status.

Raising

thecutoff

point,

thus

increasing

the

propor-tion of

false-negative

results, is

probably

aless

satisfactory

alternative.

Anotherreason for the

discrepancies

in both

sensitivity

and specificity could be attributed to the

difficulty

in

stan-dardizing

theHAItestbetween laboratories

(2).

Surveys by

theCenters for Disease Controland the

College

of American

Pathologists,

aswellasthe

Royal College

of

Pathologists

of

Australia,

have confirmed these differences in HAI tests

betweenlaboratories

(1).

Therefore, only

approximate

eval-uationscanbe madeofdifferentELISAmethods for rubella

IgG by

comparison

withHAI tests

(20).

Itmay be that the HAI test used by M.A.

Bioproducts

for calibration of the

RUBELISAsystem was less sensitivethanourHAI test. There have been few

independent

reports ofthe

evalua-tion ofcommercial reagent sets for rubella

IgG by

micro-ELISA methods. The

study

reported

herein is an

in-depth

and

comprehensive

comparative

evaluation of three such reagent sets. Previous individual studies

(3,

8,

18,

22)

have shown thattheRUBELISAsystem hasa

specificity

of99to

100%, but sensitivityrangedfrom91.4 to 67.7%, which was

attributed, inthemain,to the lack ofdetectionof some sera

with HAI titers of 10 and 20. All these published studies show that whilespecificityis excellent sensitivity is lacking. Anew rapid ELISAthat overcomes this lack of sensitivity

with low-HAI-titer sera has been reported (1).

These three assays use markedly contrasting

configura-tions (Table 3). Cutoff absorbance levels were fixed in the

Enzygnost-Rubella system, compared with the floating

thresholdin others,which ismore tolerant of interlaboratory

variation. This assay also does not have a negative control serum, which may lead to a failure to detect inadequate washing. Control antigenshave been used to detect nonspe-cific binding in the Enzygnost-Rubella and RUBELISA

systems(6), whereas highly purified antigen, asused in the ORTHO Rubella system, bypasses the need for such con-trols (19, 21).

Heat-treated sera,usedin concurrentcomplement fixation

tests or to inactivate infectious human immunodeficiency virus (16), weresatisfactory substrates forthe RUBELISA

and ORTHO systems but not for the Enzygnost-Rubella

system, asfound with some other ELISAsystems (14, 15).

The heated sera show increased

reactivity

with control

antigen

wells, oftenabove thecutoff, butthe reasonfor this reactivity is not known (9; P. W. Robertson, personal communication).

Inconclusion,onthe basis ofthe presentstudy,wewould prefer to use eitherthe Enzygnost-Rubella orthe ORTHO Rubella system rather than the RUBELISA system instead of the

time-consuming

HAI test for the determination of immune status,

provided

that (i) sera for testing in the

Enzygnost-Rubella system must not have been heat

inacti-vated and(ii) forthe ORTHO system, standardized ELISA

values in the rangefrom0.20 to0.25 should be regarded as

equivocal

unlessconfirmedby adifferent

procedure.

LITERATURE CITED

1. Boteler, W. L., K. J. Barnes, E. Buimovici-Klein, and A. J. O'Beirne. 1984. Multicenter evaluation ofa1-henzyme-linked immunosorbent assay for rubellaserology. J. Clin. Microbiol. 20:1140-1144.

2. Bradstreet, C. M. P., B. Kirkwood, J. R. Pattison, and J. O. Tobin. 1978. The derivation of a minimum immune titre of rubella haemagglutination-inhibition (HI) antibody. A Public Health Laboratory Service collaborative survey. J. Hyg. 81: 383-388.

3. Castellano,G.A.,D. L.Madden,G. T.Hazzard,C. S.Cleghorn, D. V. Vails, A. C. Ley, N. R. Tzan, and J. L. Sever. 1981. Evaluation ofcommercially available diagnostic test kits for rubella. J. Infect. Dis. 143:578-584.

4. Field, P. R., andC. M. Gong. 1984. Diagnosis ofpostnatally acquiredrubellabyuseofthreeenzyme-linkedimmunosorbent assaysforspecificimmunoglobulin G andMandsingleradial hemolysisforspecific immunoglobulinG.J.Clin. Microbiol.20: 951-958.

5. Field,P.R.,A. M.Murphy,S. F.Cape,and M. B.Albrey.1977. Experience withthe useoftrypsinizedhuman group O

eryth-rocytesin the rubellahaemagglutinationinhibitiontest.Aust. J. Med.Technol. 8:48-54.

6. Forghani, B.,and N.J. Schmidt. 1979. Antigen requirements,

sensitivity,andspecificityofenzymeimmunoassaysfor measles andrubellaviralantibodies.J. Clin. Microbiol. 9:657-664. 7. Haukenes, G., and H.Blom. 1975. Falsepositive rubella virus

haemagglutination inhibitionreactions: occurrence and

disclo-sure.Med. Microbiol. Immunol. 161:99-106.

8. Herrman, K. L. 1985. Available rubella serologic tests. Rev. Infect. Dis.7(Suppl. 1):S108-S112.

9. Husby S., U. Holmskov-Neilsen, J. C. Jensenius, and K. Erb. 1985. Increasednon-specificbindingof heat-treatedproteinsto

on April 11, 2020 by guest

http://jcm.asm.org/

plasticsurfaces analyzed by ELISA andHPLC-fractionation. J. Immunoassay 6:95-110.

10. Jankowski, M. A., W. Gut, E. Nawrocka, and M.Kantoch. 1980. Detection and differentiation ofcytomegalovirus antibodies by radioimmunoassay. J. Virol. Methods 1:133-138.

11. Kleeman, K.T., D. J. Kiefer, and S.P. Halbert.1983. Rubella antibodies detected by several commercial immunoassays in hemagglutination inhibition-negative sera. J. Clin. Microbiol. 18:1131-1137.

12. Lennette, E. H., N.J. Schmidt, and R. L.Magoffin. 1967. The hemagglutination inhibition test for rubella:acomparison of its sensitivity to that of neutralization, complement fixation and fluorescent antibody tests for diagnosis ofinfection and deter-mination of immunity status. J. Immunol. 99:785-793. 13. Menser, M. A., J. R. Hudson, A. M. Murphy, and Y. E. Cossart.

1984. Impact of rubella vaccination in Australia. Lancet i:1059-1062.

14. Mortimer, P. P., J. V. Parry, and J. Y. Mortimer. 1985. Which anti-HTLV III/LAV assays for screening and confirmatory testing? Lancet ii:873-877.

15. Ronalds, C. J., A. E. Hardiman, P. C. A. Grint, and H. D. Kangro. 1985. Rubella-specific IgM determination on heat-treatedsera. Lancetii:1071-1072.

16. Spire, B., D. Dormont, F. Barré-Sinoussi, L. Montagnier, and J. C. Chermann. 1985. Inactivation of

lymphadenopathy-asso-ciated virusby heat,gamma rays, andultravioletlight.Lancet i: 188-189.

17. Stewart,G.I.,P. D.Parkman,H.E.Hopps, R.D.Douglas, J.P. Hamilton,andH. M.Meyer,Jr. 1967.Rubella-virus hemagglu-tination-inhibitiontest.N. Engl. J. Med. 276:554-557. 18. Truant,A.L.,B. L.Barksdale, T. W.Huber,and L. B. Elliott.

1983. Comparison ofan enzyme-linked immunosorbent assay withindirecthemagglutination and hemagglutination inhibition fordetermination of rubella virus antibody: evaluation of

im-mune statuswith commercial reagents inaclinicallaboratory. J. Clin. Microbiol. 17:106-108.

19. Vaheri, A.,and E. M.Salonen. 1980. Evaluation ofsolid-phase enzyme-immunoassay procedure in immunity surveys and diag-nosis ofrubella. J. Med. Virol.5:171-181.

20. Vejtorp, M.1983. Serodiagnosis of postnatal rubella. Asurvey of methods with special reference to the enzyme-linked im-munosorbent assay. Dan. Med. Bull.30:53-66.

21. Vejtorp, M., and J. Leerhoy. 1980. Comparison of thesensitivity of ELISA and thehaemagglutination inhibitiontestfor routine diagnosis of rubella. Acta Pathol. Microbiol. Scand. Sect.B88: 349-350.

22. Wittenburg,R.A.,M. A.Roberts,L.B. Elliott, and L. M.Little. 1985.Comparative evaluation ofcommercial rubella virus anti-bodykits. J. Clin.Microbiol. 21:161-163.

on April 11, 2020 by guest

http://jcm.asm.org/