JOURNAL OF CLINICAL MICROBIOLOGY, Nov. 1984, p. 951-9580095-1137/84/110951-08$02.00/0Copyright C 1984, American Society for Microbiology

Vol. 20, No. 5

Diagnosis of Postnatally Acquired Rubella by Use of Three Enzyme-Linked Immunosorbent Assays for Specific Immunoglobulins G andM and Single Radial Hemolysis for Specific Immunoglobulin G

PETER R. FIELD* AND CHUN MEI GONGtVirology Department, Institute of Clinical Pathology and Medical Research, Westmead, New South Wales, Australia

Received 29 May 1984/Accepted 14 August 1984

Three commercial indirect enzyme-linked immunosorbent assays (ELISAs) (RUBELISA, Enzygnost-Rubella, and Rubazyme) and a commercial single radial hemolysis (SRH) test (Rubazone) were evaluated forthe diagnosis of acute rubella by testing 41 acute- (<7 days postonset) and convalescent-phase (8 to 82 dayspostonset) serum pairs from cases of rubella previously confirmed by signfficant change in the hemagglutina-tion inhibition test titer. Specificity was tested by using 10 acute- and convalescent-phase serum samples frompatients with rash not confirmed as rubella (control group). In testing for rubella-specific immunoglobulin G(IgG) antibody, Enzygnost-Rubella and Rubazyme confirmed infection in 40 and RUBELISA in 39 of the 41proven rubella patients. For one patient all three ELISAs failed to show a significant titer rise. No false-positivediagnoses occurred in the control group, although a suspected infection was shown by Rubazyme in one patient.No specific IgM could be detected in this case. Single radial hemolysis confirmed infection in 39 of the 41 provenrubella patients, and one false-positive diagnosis occurred in the control group patients. Of the 43 convalescent-phase serum samples, rubella-specific IgM was detected in 42 by Enzygnost-Rubella, in 41 by RUBELISA M,and in 39 by Rubazyme-M. For a rapid diagnosis with acute-phase sera, specific IgM detection by ELISA wasmost reliable in hemagglutination inhibition test-positive sera; of 18 such serum samples IgM was shown in 15by Enzygnost-Rubella, in 13 by RUBELISA M, and in 11 by Rubazyme-M. Conversely, of 24 acute-phaseserum samples with a hemagglutination inhibition test titer of <10, specific IgM was detected in only 5 byEnzygnost-Rubella, in 4 by RUBELISA M, and in 1 by Rubazyme-M. False-positive specific IgM results wereshown by Rubazyme-M in two serum samples from one patient in the control group. These serum samples werenegative with the other two ELISA methods.

Since 1967 diagnosis of rubella infection has been carriedout mainly by serological means by the hemagglutinationinhibition test (HAI). This has been done by demonstratingfourfold or greater rises in HAI titers between paired serumspecimens; however, in cases where an acute-phase bloodspecimen is not available, and it is not possible to demon-strate a rise in antibody titer, specific immunoglobulin M(IgM) antibody testing has been done by separating the IgMfraction of serum from IgG and testing this fraction forrubella HAI antibody activity (24). The isolation of rubellavirus is seldom used, as a diagnosis of rubella is obtainedmore readily by serological testing (12).During recent years several other techniques have been

described for the measurement of rubella IgG or IgM anti-body, or both. Antibody measured by passive hemagglutina-tion (3) is not of the IgM class and shows a considerablydelayed immune response and, owing to the urgency of rapiddiagnosis of rubella infection in pregnant women, demon-stration of seroconversion by this technique is impractical.Single radial haemolysis (SRH) (21) provides a simple meansfor confirmation of recent rubella by showing an increase inantibody level in paired serum specimens, but it cannot beused to detect IgM antibody (13).On the other hand, enzyme-linked immunosorbent assay

(ELISA) (23) and solid-phase radioimmunoassay (16) havebeen shown to detect significant IgG antibody rises in pairedserum specimens from cases of rubella. They can alsomeasure specific IgM antibodies in nonfractioned sera.

* Corresponding author.t Present address: institute of Medical Biology, Chinese Acade-

my of Medical Sciences, Kunming, People's Republic of China.

These immunoassays have advantages over HAI, passivehemagglutination, and SRH in that they have the potentialfor automation, and precise numerical read-out replacesvisual interpretation. However, solid-phase radioimmuno-assay suffers from the disadvantages of short shelf life, highcost of reagents and counters, and possible safety hazards.For these reasons the ELISA technique is the one which isbeing increasingly developed, evaluated, and used in viralserology. Other studies have shown ELISA (4, 10, 20, 23,25) to have precision similar to that of HAI in determinationof immune status and diagnosis of recent infection. Howev-er, to our knowledge no fully documented comparative studyevaluating commercial ELISA systems for the diagnosis ofrecent acquired rubella infection has been published. Simi-larly, other studies using "in-house" laboratory-preparedreagents have shown the suitability of SRH for diagnosingrecent infection (4, 11, 13, 19, 21), but only brief mention hasbeen made of the use of commercial reagents for suchpurpose (22).

This communication describes the evaluation of threecommercial indirect ELISA systems for IgG, three commer-cial indirect ELISA systems for IgM, and one commercialSRH system and compares them with an in-house HAI forthe diagnosis of postnatally acquired rubella. CommercialSRH tests are not widely available, due to their short shelflife. All the ELISA systems and the SRH used a singledilution of serum. Besides evaluating the measurement ofrubella IgG by ELISA and SRH for demonstrating signifi-cant antibody rises in cases of recent rubella, rubella IgMELISA testing was carried out on each serum pair for thefollowing reasons: (i) to determine whether acute-phaseserum specimens could be used to obtain an early diagnosis,

951

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

952 FIELD AND GONG

(ii) to obtain an indication of the persistence of rubella IgMas measured by ELISA so that its role in retrospectivediagnosis could be more accurately defined, (iii) to confirmthat the rising titers are evidence of primary rubella infec-tion, and (iv) to resolve any discrepancies occurring in serumpairs between HAI tests and ELISA rubella IgG results.

MATERIALS AND METHODS

Clinical specimens. The study material consisted of a totalof 105 serum specimens. The age of the patients ranged from5 to 40 years, with the great majority being adults. Theseserum specimnens were divided into two groups.

(i) Confirmed rubella infection group. Paired serum speci-mens were collected from 41 patients during the acute andconvalescent stages of rubella infection (patients 1 through41). These cases were confirmed as rubella by demonstrationof seroconversion or significant rise in antibody titer as

determined by HAI. In two cases three additional serumspecimens were collected, so that a total of 85 specimenswas tested. It was not possible to determine the exact date ofonset of symptoms for some patients, but in all cases thecollection of the acute-phase serum specimens was within 1week of onset. Convalescent-phase serum specimens were

collected at periods ranging from 8 to 82 days after onset, butmainly in the second and third weeks after onset.

(ii) Control group. Paired serum specimens were collectedfrom 10 patients during the acute and convalescent stages ofan illness resembling rubella (patients 42 through 51). Noneof these cases could be shown to be due to rubella infec,tion,as all patients had stationary rubella HAI titers. Two caseswere confirmed by seroconversion as measles, two as ade-novirus infections, and one as dengue fever. The remainingfive cases with stationary rubella titers could not be associat-ed with any other etiological agent, and tests for rubella-specific IgM in the three cases with HAI antibody were

negative.HAI. A microtiter technique was used (9) with trypsinized

human 0 cells at a concentration of 0.3% with HEPES (N-2-

hydroxyethylpiperazine-N'-2-ethanesulfonic acid) buffer(pH 6.2) containing 0.5% bovine albumin fraction V asdiluent and 4 U of commercially available rubella antigen.Sera were pretreated with heparin-manganous chloride be-fore testing.ELISA for rubella IgG. Enzyme immunoassays were con-

ducted on all serum specimens by three indirect ELISAsystems for IgG with commercial reagents. The reagentsused were RUBELISA (M. A. Bioproducts, Walkersville,Md.), Enzygnost-Rubella (Calbiochem-Behring, La Jolla,Calif.), Rubazyme (Abbott Laboratories, North Chicago,Ill.).

All the test systems used the principle of indirect ELISA,and the procedure followed was that recommended by themanufacturer, with the exception of one modification in theEnzygnost-Rubella system. This involved the stopping of theenzyme reaction after a shorter time (usually 10 to 15 mins)than that recommended by the manufacturer (45 min). Whenthe reaction was stopped after 45 min, background (controlantigen) absorbance values rose considerably, resulting inmany false-negative results. A comparison of the principlecharacteristics of the three ELISA test systems and SRH isshown in Table 1. The solid-phase component of eachsystem consisted of antigen attached to either the well in a

plastic microplate (RUBELISA, Enzygnost-Rubella) or a

bead (Rubazyme). With the exception of Rubazyme, allsystems employed a control antigen to test for nonspecificreactivity.

All systems used an anti-human IgG conjugate. The com-

position of serum and conjugate diluents, as well as washingsolutions, could not be obtained from the manufacturer, withthe exception of RUBELISA, which has a washing solutionof phosphate-buffered saline plus 0.05% Tween and Ruba-zyme, which uses water for washing. For RUBELISA andEnzygnost-Rubella, washing was performed with a Dyna-tech Miniwash, and absorbances of the solutions in each wellwere measured directly in the plate with a vertically measur-

ing photometer (Dynatech MR580 MicroELISA Auto Read-

TABLE 1. Comparison of test procedure for three indirect ELISA systems and SRH for rubella IgG for diagnosis of recent rubellainfection

ELISA

Characteristic Enzygnost- SRHRUBELISA Rubella Rubazyme (Rubazone)

(Calbiochem-Behring)

Solid-phase antigen Microplate in strips Microplate Bead NAaControl antigen Yes Yes No YesSerum dilution 1/25 1/200 1/20 UndilutedSerum incubation 0.75 h at 20 to 25°C 2 h at 20 to 25°C 1 h at 37°C 20 hEnzyme/substrate Alkaline phosphate/pNPPb Alkaline phos- Peroxidase/OPDC NA

phate/pNPPConjugate incubation 0.75 h, RTd 1 h, 370C 1 h, 370C NASubstrate incubation 0.75 h, RT 0.75 h, RTe 0.5 h, RT NAAbsorbance reading 405 nm 405 nm 492 nm NATime 3.25 h per 48 tests 3.75 h per 48 tests 5 h per 100 tests 20 h per 60 testsApprox. costf of reagents per $4.30Y $2.10 $3.309 $1.60 to $2.00

test serumStability of reagents >1 yr >1 yr >1 yr 3 weeksEvaluation of paired sera Absorbance ratio (.1.65) HAI ratio (.4x) Absorbance ratio IUh (HAI) ratio

(.1.65) (24x)a NA, Not applicable.b pNPP, p-Nitrophenyl phosphate.c OPD, o-Phenylenediamine-2-hydrochloride.d RT, Room temperature (20 to 25°C).eIn practice, 10 to 15 min was found to be the optimal time.f Costs are shown in Australian dollars.8 Based on cost of duplicate tests as recommended by manufacturer.h IU, International units.

J. CLIN. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

DIAGNOSIS OF POSTNATALLY ACQUIRED RUBELLA 953

TABLE 2. Comparison of test procedure for three indirect ELISA systems for rubella IgM

Characteristic RUBELISA M Enzygnost-Rubella Rubazyme-M(Behringwerke)Solid-phase antigen Microplate in strips Microplate BeadControl antigen Yes Yes NoSerum pretreatment" Yes No YesSerum dilution 1/45 1/40 1/200Serum incubation 2 h at 37°C 1 h at RTh 2.5 h at 45°CEnzyme/substrate Alkaline phosphate/pNPP' Alkaline phosphatase/pNPP Peroxidase/OPD"Conjugate incubation 2 h at RT 1 h at 37°C 1.5 h at 45°CSubstrate incubation 0.75 h at RT 0.75 h at RT 0.5 h at RTAbsorbance reading 405 nm 405 nm 492 nmRF false-positive results No Yes NoTime 6 h per 48 tests 3.5 h per 48 tests 6 h per 50 testsApprox. cost of reagents per $9.50 $2.30 $6.90'

test serumStability of reagents >1 yr >1 yr >1 yrInterpretation of results Absorbance values of -0.28 as positive Absorbance values of .0.2 as posi- Rubazyme-M index

tive >1.09 as positive

To remove the interfering effect of IgG or RF.RT. Room temperature (20 to 25'C).

C pNPP. p-Nitrophenyl phosphate.d OPD. o-Phenylenediamine-2-hydrochloride.' Based on cost of duplicate tests as recommended by the manufacturer. Costs are shown in Australian dollars.

er; Dynatech Laboratories, Inc., Alexandria, Va.). ForRubazyme, washing was done with a Pentawash 1I (AbbottLaboratories) and absorbances were measured in a QuantumII spectrophotometer (Abbott Laboratories).The methods for evaluating the results for paired serum

samples for the presence of a significant increase in rubellaIgG appear in detail as footnotes to Table 4.SRH test. All sera were tested with commercial reagents

(Rubazone; Scientific Measuring Instruments, Sydney, Aus-tralia). Serum samples (5 jl), inactivated at 56°C, were

added to wells in a system that was similar to that describedby Vaananen and Vaheri (21), except that sheep erythro-cytes were used and complement was added to the platesafter overnight radial diffusion of IgG antibody. To minimizeinterlaboratory variability, a standard curve was producedby using calibrators standardized against the World HealthOrganization 1,000-IU standard for anti-rubella serum, andall SRH titers were expressed in international units. Four-fold increases in antibody level indicated suspected currentinfection.ELISA for rubella IgM. All serum specimens were tested

for rubella IgM antibody by three indirect ELISA systemswith commercial reagents. The reagents used were RU-BELISA M, Enzygnost-Rubella (Behringwerke, Marburg,West Germany), and Rubazyme-M.

All the test systems used the principal of indirect ELISA.and the procedure used was that recommended by themanufacturer. The shortened enzyme action time, whichwas found necessary in the Enzygnost-Rubella IgG ELISA,was not required in the IgM ELISA. The principal differ-

ences in methods among the three tests systems are shown inTable 2.

The solid-phase component of each system, the method ofwashing, and the absorbance reading were similar to thatdescribed for the IgG ELISA. All systems used an anti-human IgM conjugate. The composition of serum pretreat-ment mixtures and serum and conjugate diluents could notbe obtained from the manufacturers.To overcome the interfering problem of IgG and rheuma-

toid factor (RF), an initial serum pretreatment step was usedin RUBELISA M to remove IgG. In Rubazyme-M the serum

incubation buffer is said to neutralize high concentrations ofserum RF (Martin Armstrong, Abbott Laboratories, person-al communication). With Enzygnost-Rubella there was noserum pretreatment to remove IgG or RF, and in cases ofpositively reacting sera it was necessary to repeat the IgMdetermination after removal of possible RF by a methodsuch as absorption with IgG-coated latex particles (15).(Behringwerke now recommends pretreatment of sera with a

new reagent, RF-Absorbent, for removing RF. At the timethis study was conducted this new reagent was unavailablein Australia.)The World Health Organization recommends that all viral-

specific IgM determinations should be made on a purifiedIgM fraction (1).The calculation and interpretation of the results varied for

the three methods and are shown as footnotes to Table 4.Discrepant and equivocal results. All patient serum speci-

mens showing discrepant results were repeat tested to showthat consistent results were obtained. In addition, equivocal

TABLE 3. Comparison of ELISA and SRH for 41 acute and convalescent-phase serum pairs from cases of rubella confirmed bysignificant change in HAI titer and 10 acute- and convalescent-phase serum pairs from cases with rash not confirmed as rubella (control

group) for diagnosis of acute rubella by demonstration of significant rise in antibody level"

ELISANo. of SRH (Rubazone)

Group sr p RUBELISA Enzygnost-Rubella Rubazymeserum pairs

Positive Negative Positive Negative Positive Negative Positive Negative

Confirmed rubella 41 39 (95.1) 2 (4.9) 39 (95.1) 2b (4.9) 40 (97.6) 1" (2.4) 40 (97.6) 16 (2.4)Control group 10 1 (10) 9 (90) 0 10 (100) 0 10 (100) 0 10' (100)

"Values represent the number of serum pairs (%).Patient 12 gave results indicating suspected infection.Patient 47 gave results indicating suspected infection.

VOL. 20, 1984

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

954 FIELD AND GONG

results were retested in duplicate. The SRH results frompatient 42 were confirmed in an independent laboratory.

RESULTSDetection of a significant rise in rubella IgG antibody levels

in paired serum specimens determined by the three ELISAmethods and SRH. The ability of the three ELISA methodsand SRH to detect significant rises in rubella IgG antibodybetween acute- and convalescent-phase sera from 41 cases

of clinical rubella confirmed by significant HAI titer rises issummarized in Table 3. Also included are the results ob-tained by testing 10 serum pairs from patients with rash notconfirmed as rubella (control group). Table 4 shows theactual test results and time of collection of specimens forsera in test and control groups which gave discrepant results.(The data on patient serum specimens not shown are avail-able from the authors upon request.) SRH failed to confirminfections in two patients (patients 3 and 12). RUBELISAdetected significant changes in 39 serum pairs. It failed toshow any change in antibody level in patient 3 but gave

results indicating a suspected infection in patient 12. Enzyg-nost-Rubella and Rubazyme detected significant antibodyrises in all cases, except for patient 12 (Table 4), in whom theresults from these two methods indicated a suspected but notconfirmed infection. Patient 12 was not hypogammaglobulin-emic.With the control group, SRH showed a seroconversion in

one patient (patient 42). This patient also showed serocon-

version by HAI to dengue fever virus, and the convalescentserum contained dengue fever virus IgM. However no

rubella IgM antibody was demonstrated in either the acute-or convalescent-phase serum (Table 4). None of the threeELISA techniques gave positive results in the control group,

except for Rubazyme, which in one case (patient 47) indicat-ed a suspected infection. Although the Diagnostic Ratio was.1.65, the Rubazyme index was <1.00 for both serum

specimens.Comparison of the three ELISA methods for detection of

rubella IgM antibody in acute- and convalescent-phase serum

specimens. To determine whether acute-phase serum speci-mens could be used to obtain an early and hence rapiddiagnosis, the 42 acute-phase serum specimens (collected <7days after onset) were compared with the 43 convalescent-phase serum specimens (collected .7 days after onset)obtained from the 41 cases of confirmed rubella infection(Table 5).Of the 42 acute-phase serum specimens, 17 (40.5%) were

positive by RUBELISA M, 20 (47.6%) were positive byEnzygnost-Rubella, and 12 (28.6%) were positive by Ruba-zyme-M. Of the 43 convalescent-phase serum specimens(collected between 7 and 82 days after the onset of infec-tion), 41 (95.4%) were positive by RUBELISA M, 42(97.7%) were positive by Enzygnost-Rubella, and 39 (90.7%)were positive by Rubazyme-M. Details about specimenswhich gave discrepant results are shown in Table 4.

Rubella-specific IgM could be demonstrated in only 6 of

TABLE 4. Patients with discrepant results in a comparison of three ELISA systems for rubella IgG and IgM with SRH for the diagnosisof acquired rubella in 85 serum specimens from 41 cases of rubella confirmed by significant change in HAI titer (patients 1 through 41)

and 10 acute- and convalescent-phase serum pairs from 10 cases with rash other than rubella (patients 42 through 51).IgG IgM

Days RUBELISA Enzyg- RubazymedPatient no. after HAI SRHa Gb nost' RUBELISA Rubazyme-

onset Ab Critical Estimated Rubazyme Diag. Me Enzygnostr ml

s ratio HAI index Ratio

Proven rubellainfection group

1 1 <10 <16 0.12 <2 0.84 0.43 0.52 1.318 640 100 0.18 1.5 24 2.47 3.0 '-0.80 0.75 3.77

3 2 20 <8 0.11 <8 0.40 0.53 0.31 1.209 640 <8 0.11 1.0 256 1.18 2.98 '0.80 _1.05 3.61

4 3 10 8 0.10 <8 0.49 0.42 0.27 1.0710 320 250 0.21 2.1 128 1.70 3.48 _0.80 1.07 3.93

5 <7 20 16 0.05 10 1.51 0.37 0.16 0.5610 160 100 0.25 4.0 _128 3.31 2.19 '0.80 0.86 2.88

6 2 <10 <8 0.09 <8 0.29 0.29 0.30 0.5311 320 160 0.23 2.5 64 1.06 3.6 _0.80 0.84 3.93

10 2 10 <8 0.07 <8 0.31 0.17 0.18 0.4113 640 160 0.18 2.57 105 1.89 6.09 '0.80 _1.25 3.93

12 1 <10 <8 0.05 <8 0.26 0.13 0.19 0.5414 80 <8 0.11 2.20 14 0.84 3.22 '-0.80 0.38 1.30

14 4 10 26 0.08 <8 0.45 0.19 0.35 0.6414 320 160 0.30 3.7 128 2.12 4.69 0.49 0.63 2.01

17 1 <10 8 0.07 <8 0.47 0.08 0.01 0.6415 40 250 0.43 6.1 64 2.58 5.42 0.06 0.09 0.33

18 2 20 <16 0.05 <8 0.57 0.29 0.31 0.5415 320 250 0.43 8.6 115 4.72 8.22 '-0.80 0.54 1.50

19 1 <10 8 0.08 <4 0.33 0.29 0.66 0.6015 640 250 0.35 4.3 '56 2.71 8.10 0.47 0.51 1.75

20 1 <10 <8 0.06 <4 0.44 0.25 0.08 0.4415 80 200 0.38 6.3 160 5.22 10.09 0.78 0.85 1.55

22 2 <10 <8 0.09 <8 0.27 0.17 0.23 0.5618 320 160 0.32 3.32 100 2.6 9.60 _0.80 _1.26 3.49

29 2 <10 <8 0.06 <8 0.32 0.19 0.25 0.7222 160 160 0.33 5.5 90 2.28 7.02 '-0.80 '0.93 3.72

Continued on following page

J. CLIN. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

DIAGNOSIS OF POSTNATALLY ACQUIRED RUBELLA 955

TABLE 4-ContiniuedIgG IgM

Days RUBELISA Enzyg- Rubazyme"Patient no. after HAl SRH" Gb nost RUBELISA Rubazyme-

onset Ab Critical Estimated Rubazyme Diag. Me Mq

Abs. ratio HAI index Ratio

30 1 <10 <8 0.05 <4 1.12 0.06 0.02 0.1422 640 250 0.22 4.4 '-32 3.48 3.12 _0.80 '1.17 3.93

31 <7 <10 <8 0.07 <8 0.44 0.55 0.14 0.9623 320 160 0.30 4.28 50 2.38 5.41 _0.80 0.51 3.65

32 1 <10 <8 0.07 <8 0.26 0.22 0.08 0.6024 160 160 0.33 4.7 48 2.90 11.5 '-0.80 0.90 1.33

33 2 20 <8 0.12 12 0.59 0.45 0.31 1.1726 320 250 0.43 3.54 _64 2.89 4.88 0.58 _1.01 1.87

34 1 <10 <8 0.05 <4 0.46 0.14 0.11 0.3228 320 250 0.53 10.6 '-80 5.15 >10.00 0.54 0.64 0.76

35 2 10 <8 0.04 <4 0.76 0.16 0.26 0.6028 160 250 0.47 11.8 _100 5.85 7.70 0.50 0.73 2.02

36 1 40 <8 0.06 <2 1.06 '0.80 0.30 1.199 640 65 0.33 5.50 64 2.27 2.14 -0.80 1.21 2.3019 320 NTh NT NT NT _.0.80 0.98 1.5229 160 NT NT NT NT _0.80 0.40 0.65

37 <7 20 65 0.17 16 1.44 0.16 0.18 0.4830 2,560 '-400 0.70 4.10 64 3.15 2.19 0.29 0.50 1.19

38 1 <10 <8 0.06 <4 1.54 0.05 0.06 0.3035 640 .400 0.66 11.00 140 5.72 3.71 0.70 0.40 1.17

39 <7 20 <8 0.18 4 0.96 _0.80 0.72 1.5936 320 350 0.32 1.77 _128 5.18 5.39 _0.80 0.72 1.71

40 1 20 <8 0.04 <4 0.79 0.25 0.23 1.1963 160 160 0.44 11.00 80 5.38 6.83 0.28 0.39 0.56

41 1 <10 <8 0.09 <8 0.28 0.13 0.14 0.7082 80 65 0.20 2.22 32 1.85 6.58 0.27 0.29 1.27

Patients with rashother than ru-bella (con-trol group)

42 2 160 <8 0.52 80 4.45 0.13 0.19 0.508 160 160 0.35 0.67 64 4.71 1.06 0.11 0.10 0.89

44 4 640 .400 0.68 '-140 5.64 0.13 0.16 1.1613 640 .400 0.47 0.68 '140 4.80 0.85 0.05 0.12 1.13

47 2 10 <8 0.06 <8 0.34 0.05 0.01 0.1615 10 <8 0.08 1.33 <8 0.69 2.03 0.05 0.02 0.16

a SRH results are expressed in international units. Values of .8 indicate evidence of previous infection.RUBELISA G absorbance values of .0.17 (obtained from a calibration curve) are equivalent to an HAI titer of .1:8. The critical ratio is the ratio of the mean

RUBELISA value of the convalescent-phase serum to that of the acute-phase serum. A critical ratio of '1.65 indicates a highly significant increase in antibodytiter, and a ratio of 1.47 to 1.64 is comparable to a four-fold increase in HAI titer. Abs., Absorbance values.

Enzygnost-Rubella also involves the construction of a calibration curve, but the test ELISA values are converted back to an estimated HAI titer.d Rubazyme absorbance values are expressed as the Rubazyme index. This is the ratio of the mean specimen absorbance to the mean of duplicate low-positive

control absorbance values. An index of _1.00 is positive for rubella IgG antibody. The Diagnostic Ratio is the ratio of the Rubazyme index of the convalescent-phase serum to that of the acute-phase serum. A ratio of '1.65 is confirmation of recent rubella infection.

e RUBELISA M values of .0.28 (from a calibration curve) are positive. Values of '0.19 are negative, and intermediate values are equivocal.f Enzygnost-Rubella values are expressed as the net absorbance. Values of .0.2 are positive.8 Rubazyme-M values are expressed as an index by dividing the mean duplicate test specimen absorbances by the mean of the triplicate low-positive control

values. A Rubazyme-M index of .1.09 is positive and an index of '0.91 is negative. Intermediate values are equivocal.h NT, Not tested.

the 24 (26.1%) acute-phase serum specimens with an HAItiter of <10 by any of the three ELISA methods. Of these sixacute-phase specimens, rubella IgM was detected by allthree ELISA methods in one case (patient 1). RUBELISA Mand Enzygnost-Rubella were positive in two cases (patients6 and 19). Enzygnost-Rubella was the only test positive intwo cases (patient 22 and 29), and RUBELISA M was theonly test positive in one case (patient 31). In summary, of the24 HAI-negative acute-phase serum specimens, specific IgMwas detected in only 5 (20.7%) by Enzygnost-Rubella, 4(16.7%) by RUBELISA M, and 1 (4.2%) by Rubazyme-M.On the other hand, specific IgM was detected in 10

(55.6%) of 18 acute-phase serum specimens with HAI titersof .10 by all three ELISA methods. Of the remaining eightacute-phase serum specimens, RUBELISA M and Enzy-gnost-Rubella were positive in two (patients 4 and 18).Enzygnost-Rubella and Rubazyme-M were positive in onealone (patient 40), RUBELISA M was positive in one alone(patient 5), and Enzygnost-Rubella was positive in two alone(patients 14 and 35). Two serum specimens were negative byall three methods (patients 10 and 37). In summary, of the 18HAI-positive acute-phase serum specimens, Enzygnost-Ru-bella was positive in 15 (83%), RUBELISA M was positivein 13 (72.2%), and Rubazyme-M was positive in 11 (61.1%).

VOL. 20, 1984

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

956 FIELD AND GONG

TABLE 5. Comparison of the three rubella-specific IgM ELISA methods for the 85 serum samples from the confirmed rubella infectiongroup (42 acute-phase serum specimens collected <7 days after onset and 43 convalescent-phase serum specimens) and the 20 serum

specimens from the control group (rash not confirmed as rubella)'GroupNo. of serum RUBELISA M Enzygnost-Rubella Rubazyme-M

specimens Positive Negative Positive Negative Positive Negative

Confirmed rubella 42 (acute) 17 (40.5) 25b (59.5) 20 (47.6) 22 (52.4) 12 (28.6) 30' (71.4)43 (convalescent) 41 (95.4) 2d (4.6) 42 (97.7) 1 (2.3) 39 (90.7) 4 (9.3)

Control 20 0 20 (100) 0 20 (100) 2 (10) 18 (90)a Values represent the number of serum specimens (%).b Three specimens gave equivocal results.c Two specimens gave equivocal results.d One specimen gave equivocal results.

The overall detection rate of rubella IgM antibody byusing the three ELISA methods with the 85 serum specimensfrom the confirmed rubella infection group (acute- andconvalescent-phase specimens combined) was 58 (68.2%) forRUBELISA M, 62 (72.9%) for Enzygnost-Rubella, and 51(60%) for Rubazyme-M. Equivocal results were obtainedwith four serum specimens tested by RUBELISA M (pa-tients 20, 32, 40, and 41) and two serum specimens tested byRubazyme-M (patients 4 and 31).

In the group with rash not confirmed as rubella (controlgroup) 2 of 20 serum specimens (10%) were positive byRubazyme-M (Table 5). These two specimens were acute-and convalescent-phase sera from patient 44. No false-positive results were obtained with either RUBELISA M orEnzygnost-Rubella.

Persistence of rubella-specific IgM. Enzygnost-Rubella de-tected rubella-specific IgM in all sera collected between 7and 82 days. RUBELISA M detected specific IgM in all suchsera up to 63 days after onset (patient 40). At 82 days (patient41) this test gave an equivocal result which was just belowthe positive cut-off point (Table 4). Rubazyme-M was reli-able in detecting rubella-specific IgM up to 26 days afteronset (patient 33) but detected this antibody irregularlythereafter. For example, Rubazyme-M was negative at 28days (patient 34), 29 days (patient 36), and 63 days (patient40), but on the other hand, it detected rubella IgM in adifferent patient (patient 35) at 28 days. It also was positiveat 30 days (patient 37), 35 days (patient 38), 36 days (patient39), and 82 days (patient 41).

It should be pointed out that the serum collected frompatient 17 at 15 days after onset was an exception to theabove in that none of the three ELISA systems detectedrubella IgM, despite the fact that significant rises in antibodylevel were shown in this patient by HAI, SRH, RUBELISA,Enzygnost-Rubella, and Rubazyme. There was insufficientserum remaining to test for rubella IgM by sucrose densitygradient ultracentrifugation. Thus, all of the 41 patients withHAI rising titers were confirmed as having rubella byspecific IgM testing, with the sole exception of patient 17.

DISCUSSIONBoth the ELISA and SRH have the advantages over HAI

in that nonspecific inhibitors of hemagglutination which maycause false-positive results do not need to be removed beforetesting (14), results can be read objectively, and quantitationof antibody levels can be estimated on single dilutions ofserum. In comparison to conventional truncated HAI serumdilution titers, where reproducibility is acceptable so long asreplicate titers remain within a twofold range, ELISA testresults are more accurate because they are free to take onany of a continuum of values, usually with a reproducibilityof ±10% (coefficient of variation) (5). RUBELISA and

Rubazyme use the continuous nature of ELISA values,whereas Enzygnost-Rubella IgG results have to be convert-ed back to an HAI titer, with the inherent disadvantage thatonly antibody increases of fourfold or greater are significant.Enzygnost-Rubella IgM uses a predetermined fixed cut-offabsorbance level, which does not deal effectively with theproblem of interassay variation. There is no general agree-ment as to what method of reporting of test results ispreferable in ELISA (5).Other investigators (4, 20, 23, 25) have found that enzyme

immunoassay detects rising levels of rubella IgG antibodylater in the course of infection than the HAI test, and thismay in fact be an advantage in the diagnosis of someinfections where the first serum specimen is drawn late and arise in HAI antibodies cannot be shown. However, this wasnot seen in the present study, as acute-phase serum speci-mens were collected early in the first week after onset ofinfection. As SRH detects only IgG antibody, the advan-tages in relation to the timing of collection of specimens arethe same (21) as for IgG ELISA.

Rubella-specific IgM was detected in 29 to 48% of acute-phase serum specimens from confirmed rubella cases, de-pending on the ELISA method used, and thus there appearsto be a limit to the use of these tests for rapid diagnosis onsingle acute-phase serum specimens. However, the acute-phase serum specimens tested in this study were collectedwithin 2 to 3 days of the onset of a rash, and this may explainwhy higher detection rates of specific IgM were not ob-tained. Other reports have also indicated that rapid diagnosisby estimating specific IgM in acute-phase serum specimensis limited. For example, Coxiella burnetii-specific IgM wasdetected by ELISA in only 4% of the acute-phase serumspecimens collected during the first week after onset of Qfever infection and in 38% of the serum specimens collectedduring the second week (7). In another study (C. M. Gong,D. W. T. Ho, P. R. Field, and A. M. Murphy, J. Virol.Methods, in press) to evaluate the use of specific IgM byELISA for the diagnosis of echovirus type 11 infections,echovirus type 11-specific IgM was detected in only 58% ofthe acute-phase serum specimens. Using an ELISA formumps-specific IgM, Meurman et al. (17) detected mumps-specific IgM in 96% of the first available serum specimensfrom clinically typical mumps infection. However, in theirreport, the time of collection after onset was not stated, andit appears that many of the first serum specimens werecollected during convalescence, as rising titers by comple-ment fixation tests were obtained in only 71% of these cases.However, in some viral infections, such as hepatitis A,which have long incubation periods, specific IgM is detect-able at the time of onset.

It is difficult to explain the failure to detect rubella-specificIgM by all three methods in one case (patient 17), and there

J. CLIN. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

DIAGNOSIS OF POSTNATALLY ACQUIRED RUBELLA 957

was no other evidence that this patient was undergoing asecondary infection. The possibility of hypogammaglobulin-emia was ruled out, as total immunoglobulin levels wereshown to be normal. Alternatively, the IgM response mayhave been of very short duration.

In the control group, the two false-positive results detect-ed by Rubazyme-M are of concern. It has been shown thatRubazyme-M can produce false-positive results due to thepresence of heterophil antibodies (18), but the manufacturerclaims that false-positive results only occurred in earlyRubazyme-M kits and not in subsequently modified kits. Thekits used in the present study were of the latter type. Despitecontinuous improvements being made, a later study (2) hasshown that false-positive results can still occur in Ruba-zyme-M with sera without heterophil antibodies. A Ruba-zyme-M confirmatory neutralization test was not available inAustralia at the time of our studies. In the absence of areliable confirmatory test or the incorporation into the test ofa control antigen bead to test for nonspecific reactivity, allpositive results obtained by Rubazyme-M should be con-firmed by another method.Commercial kits and reagent sets are now being increas-

ingly promoted for use in microbiology, especially for sero-logical diagnosis of viral infections, but unlike in America,where the U.S. Department of Health, Education and Wel-fare has granted authority to the Food and Drug Administra-tion to regulate this expanding field, there is no control of thesale and use of such products in Australia and certain othercountries. Thus, laboratory workers in these countries haveno idea as to the accuracy and precision of a given kit apartfrom what is provided by unsolicited, independent studies,such as the present report, or the performance characteris-tics described as a result of the evaluation of the manufactur-er. Although no bias can be directly ascribed to the in-housetesting of the manufacturer, it would seem to be desirable forcountries lacking regulatory approval to have some authori-tative control over the use of commercial diagnostic reagentsin their countries. This could perhaps be accomplished bythe testing of new commercial diagnostic reagents andsystems in large government laboratories where the neces-sary clinical material and scientific and technical expertiseare available for such evaluations. This would guard againstproducts or batches of products not approved by the Foodand Drug Administration which are either unsuitable orlacking in diagnostic precision being used in other countries.

In many kits the identity of some of the reagents is notrevealed, and this applies to the test systems evaluated inthis communication. It has been suggested that no clinicalchemist should ever agree to employ secret reagents (6). It isappreciated that companies expend much effort in producingreagent systems in competition with rival manufacturers andthat, accordingly, they may not wish to make known somedetails of such systems. However, until more detailed infor-mation on the ingredients used is available from manufactur-ers, many reputable microbiologists will be reluctant to usesuch commercial reagent sets.For the diagnosis of rubella when acute- and convalescent-

phase serum pairs are available, two commercial rubella IgGELISA systems, Enzygnost-Rubella and Rubazyme, ap-peared to be suitable alternatives to the more laborious, yetwell-proven and understood, HAI test. RUBELISA G didnot detect infection in one patient. The commercial SRH,Rubazone, although the simplest test to perform, was theleast satisfactory, as it failed to detect seroconversion in twopatients and gave a false-positive result in another patient.Of the three commercial ELISA systems tested for rubel-

la-specific IgM, Enzygnost-Rubella and RUBELISA M ap-peared to be less costly yet comparable alternatives to thecumbersome, time-consuming, and relatively expensive ref-erence technique, HAI testing on IgM fractions obtainedafter sucrose density gradient ultracentrifugation (8). Ruba-zyme-M was less satisfactory in our hands. It was lesssensitive than the other two systems, and false-positiveresults occurred. For a rapid diagnosis, acute-phase serawith low levels of HAI antibodies can be tested by Enzy-gnost-Rubella or RUBELISA M provided it is recognizedthat this is only a screening test. If equivocal or negativeresults are obtained with an acute-phase serum specimen, asecond early-convalescent specimen should be tested.

It is essential that all available evidence (all serologicalresults, previous test results, and as much accurate clinicaldata as possible) be taken into consideration before making adiagnosis of rubella when this occurs in pregnancy. Whenthe result is difficult to interpret or results are not consistentwith the history, we would recommend that the tests berepeated and that sera giving equivocal results should betested in a different assay.

ACKNOWLEDGMENTS

We are indebted to all our colleagues in the Virology Departmentfor their whole-hearted assistance and A. M. Murphy for his advicein preparation of the manuscript.The Australian Development Assistance Bureau provided finan-

cial assistance to C.M.G.

LITERATURE CITED

1. Almeida, J., P. Atanasiu, A. J. Beale, K. Chernesky, P. S.Gardner, E. H. Houwink, J. E. Maynard, P. K. Nakane, L.-A.Nilsson, 0. Sobeslavsky, and A. Voller. 1979. Detection ofantigens and IgM antibodies for rapid diagnosis of viral infec-tions: a WHO memorandum. Bull. W.H.O. 57:925-930.

2. Best, J. M., S. J. Palmer, P. Morgan-Capner, and J. Hodgson.1984. A comparison of Rubazyme-M and MACRIA for thedetection of rubella-specific IgM. J. Virol. Methods 8:99-109.

3. Birch, C. J., B. P. Glaun, V. Hunt, L. G. Irving, and I. D. Gust.1979. Comparison of passive haemagglutination and haemagglu-tination-inhibition techniques for detection of antibodies torubella virus. J. Clin. Pathol. 32:128-131.

4. Champsaur, H., E. Dussaix, and P. Tournier. 1980. Haemagglu-tination inhibition, single radial haemolysis, and ELISA testsfor the detection of IgG and IgM to rubella virus. J. Med. Virol.5:273-286.

5. de Savigny, D., and A. Voller. 1980. The communication ofELISA data from laboratory to clinician. J. Immunoassay1:105-128.

6. Di Giorgio, J. 1974. Commercial kits and reagent sets, p. 283-286. In R. J. Henry, D. C. Cannon, and J. W. Winkelman (ed.),Clinical chemistry-principles and technics, 2nd ed. Harper &Row, Publishers, New York.

7. Field, P. R., J. G. Hunt, and A. M. Murphy. 1983. Detection andpersistence of specific IgM antibody to Coxiella biurnetii byenzyme-linked immunosorbent assay: a comparison with immu-nofluorescence and complement fixation tests. J. Infect. Dis.148:477-487.

8. Field, P. R., and A. M. Murphy. 1972. The role of specific IgMglobulin estimations in the diagnosis of acquired rubella. Med. J.Aust. 2:1244-1248.

9. Field, P. R., A. M. Murphy, S. F. Cape, and M. B. Albrey. 1977.Experience with the use of typsinized human group 0 erythro-cytes in the rubella haemagglutination inhibition test. Aust. J.Med. Technol. 8:48-54.

10. Forghani, B., and N. J. Schmidt. 1979. Antigen requirements,sensitivity, and specificity of enzyme immunoassays for mea-sles and rubella viral antibodies. J. Clin. Microbiol. 9:657-664.

11. Grillner, L., and 0. Strannegard. 1976. Evaluation of thehemolysis-in-gel test for the screening of rubella immunity and

VOL. 20, 1984

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.

958 FIELD AND GONG

the demonstration of recent infection. J. Clin. Microbiol. 3:86-90.

12. Grist, N. R., E. J. Bell, E. A. C. Follett, and G. E. D. Urquhart.1979. Diagnostic methods in clinical virology, 3rd ed. p. 164-171. Blackwell Scientific Publications, Oxford.

13. Harnett, G. B., C. A. Palmer, and E. M. Mackay-Scollay. 1979.Single-radial-hemolysis test for the assay of rubella antibody inantenatal, vaccinated, and rubella virus-infected patients. J.Infect. Dis. 140:937-944.

14. Haukenes, G., and H. Blom. 1975. False positive rubella virushaemagglutination inhibition reactions: occurrence and disclo-sure. Med. Microbiol. Immunol. 161:99-106.

15. Henle, G., E. T. Lennette, M. A. Alspaugh, and W. Henle. 1979.Rheumatoid factor as a cause of positive reactions in tests forEpstein-Barr virus-specific IgM antibodies. Clin. Exp. Im-munol. 36:415-422.

16. Meurman, 0. H. 1978. Antibody responses in patients withrubella infection determined by passive haemagglutination,haemagglutination inhibition, complement fixation, and solid-phase radioimmunoassay tests. Infect. Immun. 19:369-372.

17. Meurman, O., P. Hanninen, R. V. Krishna, and T. Ziegler. 1982.Determination of IgG- and IgM-class antibodies to mumps virusby solid-phase enzyme immunoassay. J. Virol. Methods 4:249-257.

18. Morgan-Capner, P., R. S. Tedder, and J. E. Mace. 1983.

Rubella-specific IgM reactivity in sera from cases of infectiousmononucleosis. J. Hyg. 90:407-413.

19. Neumann, P. W., and J. M. Weber. 1983. Single radial hemoly-sis test for rubella immunity and recent infection. J. Clin.Microbiol. 17:28-34.

20. Schekarchi, I. C., J. L. Sever, N. Tzan, A. Ley, L. C. Ward, andD. Madden. 1981. Comparison of haemagglutination inhibitiontest and enzyme-linked immunosorbent assay for determiningantibody to rubella virus. J. Clin. Microbiol. 13:850-854.

21. Vaanninen, P., and A. Vaheri. 1979. Hemolysis-in-gel test inimmunity surveys and diagnosis of rubella. J. Med. Virol.3:245-252.

22. Vaheri, A., 0. Saksela, and P. Vaananen. 1979. Rubella screen-ing by haemolysis-in-gel test. Lancet ii:644.

23. Vejtorp, M., E. Fanoe, and J. Leerhoy. 1979. Diagnosis ofpostnatal rubella by the enzyme-linked immunosorbent assayfor rubella IgM and IgG antibodies. Acta. Pathol. Microbiol.Scand. Sect. B 87:155-160.

24. Vejtorp, M. 1983. Serodiagnosis of postnatal rubella. A surveyof methods with special reference to the enzyme-linked immu-nosorbent assay. Dan. Med. Bull. 30:53-66.

25. Vejtorp, M., and J. Leerhoy. 1980. Comparison of the sensitivityof ELISA and the haemagglutination inhibition test for routinediagnosis of rubella. Acta. Pathol. Microbiol. Scand. Sect. B88:349-350.

J. CLIN. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

cm o

n 01

Jan

uary

202

2 by

173

.32.

128.

74.