Journal of Virological Methods, 9 (1984) 15-26

Elsevier

JVM 00317

15

SPECIFICITY AND SENSITIVITY OF THE IgM CAPTURE IMMUNOASSAY:

STUDIES OF POSSIBLE FACTORS INDUCING FALSE POSITIVE OR

FALSE NEGATIVE RESULTS

MARIE-JOSE BRIANTAIS’, LILIANE

LUnit6 INSERM U 131, Hapita Antoine

Institut Pasteur, F-75015 Paris, France

(Accepted 23 March 1984)

GRANGEOT-KEROS’ and JACQUES PILLOTz

B&Ike, F-92141 Clamart. =UnitP d’lmmunologie Microbienne.

The specificity and sensitivity of the IgM-capture immunoassay (IgM-CI) were evaluated for detection of

rubella specific IgM and hepatitis B core (HBc) specific IgM. For rubella specific IgM, antibodies bound to

the solid phase were detected by haemadsorption and for HBc specific IgM, by using HBc antigen (HBcAg)

and radiolabelled IgG anti-HBc.

Rheumatoid factor (RF) was found to interfere in the test for HBc specific IgM because IgM-RF bound to

the solid phase reacted with aggregated radiolabelled HBc specific IgG. This false positive reaction did not

occur when radiolabelled F(ab’), was used instead of the whole IgG molecule. HBcAg purified from

biological fluids might be coated with host IgG and under these conditions, HBcAg could react with RF. It

was also demonstrated that high levels of IgG antibodies could interfere with IgG anti-u coated-surface by

means of non-immunological protein-protein interactions. In fact, IgG did not interfere in the rubella

assay, whereas it did in the very sensitive anti-HBc test. To prevent this false-positive reaction, different

dilution media were tested. Only the addition of non-specific IgG and fetal calf serum (FCS), to the dilution

medium, seems to improve the specificity of the test. Furthermore, in order to decrease this non-specific

IgG-IgG interaction and an occasional prozoning phenomenon, the dilution of serum to be tested was

taken into account.

Parameters considered to decrease sensitivity were also studied. RF, anti-F(ab’), antibodies and non-spe-

cific IgM did not decrease significantly the sensitivity of the assay.

IgM-capture immunoassay rubella specific IgM HBc specific IgM rheumatoid factor

INTRODUCTION

The detection of specific IgM antibodies is very useful in the serological diagnosis of

primary viral infections. Several methods are available for demonstrating IgM antibo-

dies. Most of these require separation of IgG and IgM either by classical sucrose

density gradient ultracentrifugation and gel filtration or by staphylococcal protein-A

Address correspondence to: J. Pillot, Service de Bacteriologic-Immunologie, Hopital Antoine B&cl&e,

F-92141 Clamart. France

0166-0934/84/$03.00 0 1984 Elsevier Science Publishers B.V.

16

absorption. In recent years, enzyme linked immunosorbent assays and radioimmu-

noassays have been developed and used for detection of specific IgM antibodies.

Sucrose density gradient ultracentrifugation and gel filtration are time consuming and

expensive. IgG absorption by staphylococcal protein A is unreliable for serological

purposes (Field et al., 1980; Grangeot-Keros et al., 1982). As for enzyme linked

immunosorbent assay (ELISA) and radioimmunoassay, different procedures have

been proposed. One of them, the most used, involves sensitizing polystyrene microti-

tre plates, tubes or beads with viral antigen. The patient’s serum presumed to contain

specific IgM antibodies is added, followed by anti-human IgM. The main disadvan-

tages of this method are the occurrence of false positive results with some rheumato’id

factor (RF) positive sera (Vejtorp, 1980; Ziegelmaier et al., 1980) and the lack of

sensitivity due to competition between IgG and IgM for antigenic sites (Knez et al.,

1976). To avoid these problems, another procedure named IgMcapture immunoassay

(IgM-CI) has been developed (Duermeyer and Van den Veen, 1978; Krech and

Wilhem, 1979). In this procedure, anti-u antibody is coated to the solid phase. After

incubation with the patient’s serum, viral antigen is added and the presence of a

reaction will be subsequently revealed by either radioactive or enzyme labels. These

labels may be bound to the added antigen or to a secondary antibody to the antigen.

Our study was undertaken to evaluate the performance of the IgM-CI for detecting

IgM antibodies with emphasis on specificity and sensitivity. Specificity and sensitivity

were studied with two different immunoassays allowing either the detection of rubella

specific IgM by haemadsorption or the detection of hepatitis B core (HBc) specific

IgM by radioimmunoassay. In a first step, the different parameters which may induce

false positive reactions were studied; namely, RF, antigenic preparation and specific

antibodies of the IgG class. In a second step, possible competition between specific

IgM antibodies and different factors (RF, anti-F(ab’), antibodies and non-specific

IgM) for IgG anti-u coated surface was investigated.

MATERIALS AND METHODS

Serum samples and classical serological tests Specimens were collected from patients with acute and chronic hepatitis Band from

patients with recent or remote rubella.

Anti-HBc was detected by radioimmunoassay (Corab, Abbott Laboratories).

RF positive sera were obtained from patients with rheumatoi’d arthritis and anti-

F(ab’), positive sera from patients with endocarditis. RF was detected by direct

latex-agglutination reaction (Roche) and anti-goat F(ab’)* antibodies by passive

haemagglutination (Senet and Pillot, 1971).

IgM and IgG concentrations were determined by radial immunodiffusion (Behring-

werke).

17

Preparation of HBc antigen (HBcAg)

HBcAg was prepared from hepatitis B virus present in plasma. Clarified plasma

from haemodialysed patients infected with hepatitis B virus were pelleted at 48,000 X

g for 6 h. Pellets were washed 5 times with Tris-NaCl buffer (TBS) at pH 7.2. HBcAg

was further purified by isopycnic banding of the pelleted material on a CsCl density

gradient (300,000 X g, 24 h). HBcAg was detected by sandwich radioimmunoassay

using IgG anti-HBc for coating and radiolabelled IgG anti-HBc. HBcAg positive

fractions were pooled and treated with 3 M NaSCN in order to increase their

antigenicity, and dialysed (Pillot and Capel, 1981).

Preparation of immunoglobulins and fragments IgG anti-HBc was isolated from the serum of a hepatitis B surface antigen (HBsAg)

carrier, with a high titre of anti-HBc. Fractionation was carried out on DEAE 52

(Whatman) in 0.01 M phosphate buffer, pH 8.0. The IgG fraction was concentrated to

25 mg/ml.

For detection of rubella specific IgM, IgG anti-human IgM (u-chain specific) was

purchased from Dako.

For detection of HBc specific IgM, F(ab’), anti-human IgM (u-chain specific) were

used. These fragments were prepared from antibodies obtained by affinity chromato-

graphy. Briefly, goat IgG anti-human IgM (p-chain specific) (a gift from Dr. La-

vergne, Institut Pasteur Production) was incubated for 1 h at 37°C and overnight at

4°C with an IgM Sepharose 4B (Pharmacia) immunosorbent prepared as described

by March et al. (1974). Anti-IgM antibodies were eluted with 5 M MgCl,.

F(ab’), fragments were prepared from the pure goat IgG anti-u, from human IgG

anti-HBc and from normal goat IgG (a gift from Dr. Lavergne, Institut Pasteur

Production). The IgG preparations were extensively dialysed against 0.2 M acetate

buffer pH 4.0 and digested for 16 h at 37’C with pepsin (crystallized pepsin, Calbio-

them) using 1 mg of enzyme for 50 mg of IgG at 30 mg/ml in 0.2 M acetate buffer, pH

4.0. The digestion was stopped by adding 1 M KzHPOl solution until neutralization.

The antibody preparations obtained were filtered on Sephadex G 100 (Pharmacia) in

order to separate undigested IgG and digestion products of Fe(y) fragments from the

F(ab’), fragments. The purity of F(ab’), fragments was checked by gel double immu-

nodiffusion and SDS-PAGE electrophoresis.

‘2sI-labeNed anti-HBc IgG and F(ab’), IgG and F(ab’), antibodies were labelled with lz5iodine (Amersham) as described

by Fraker and Speck (1978). Proteins were labelled in the ratio of 1 mg protein to 1

mCi iz51 in tubes coated with 2.5 ug iodogen. Free iodine was separated from labelled

protein by fractionation on Sephadex G 25 (Pharmacia).

18

IgM antibody capture

Detection of anti-HBc IgM Remova wells (Dynatech) were incubated for 2 h at 37’C

and overnight at 4°C with 100 ~1 of a solution containing 4 ug/ml of F(ab’), anti-u in

phosphate buffered saline (PBS). The wells were washed five times in PBS-Tween 20

(0.05%) and each well was filled with 200 ul of 2% bovine serum albumin in PBS

(PBS-BSA). After 4 h incubation at room temperature, the wells were washed again

with PBS-Tween and 100 ul of the diluted test serum was added to each well. Sera were

usually diluted 1 : 1,000 in PBS-BSA for screening purposes. The wells were reincuba-

ted in a moist chamber for 2 h at 37’C and washed live times with PBS-Tween. A 100

ul quantity of an optimal dilution of HBcAg preparation was placed in each well and

the plates were kept overnight at room temperature. After further washing of the wells

with PBS-Tween, 100 ul of 10 ug/ml 1251-IgG or ‘251-F(ab’), anti-HBc in PBS-BSA

was added and incubated for 4 h at 37*C. After washing with PBS-Tween as above,

L25I bound to the wells was measured in a gamma counter. Normal values were

established from the results obtained on 28 different control sera. Under such condi-

tions, the mean normal value was 236 f 70.9 cpm. Each assay on patients’ sera was

performed in parallel with 4 selected control sera. Results were read as positive when

more than 3 SDS above the mean value of the 4 control sera were obtained.

Detection of rubella specific IgM The method used to detect rubella specific IgM is a

modification of the technique described by Denoyel et al. (1981). Briefly, 96 U-well

polystyrene microtitre plates (Greiner) were filled (1OOul per well) with rabbit antibody

to human IgM (u-chain specific, Dako) at a dilution of 1 : 500 in PBS, pH 7.2. Plates

were left for 1 h at 37°C and overnight at 4°C. After three washings with PBS

containing 0.05% Tween 20, non specific binding sites on the solid phase were

quenched by incubation with 200 ~1 of 0.5% BSA in PBS for 1 hat 37°C. Heteroagglu-

tinins were removed from each serum by incubation with chick erythrocytes (100 ul of

a 50% suspension of cells in dextrose-gelatin-Verona1 buffer per 100 ul of serum) for 1

h at 4°C. After removing by aspiration the blocking agent from plates, 100 ul of serial

two-fold dilutions of treated patients’ sera in PBS containing 0.05% Tween 20 and 1%

BSA were added and incubated for 2 h at 37’C. After washing as before, 6 U of rubella

haemagglutinin (Behringwerke) in 100 ~1 0.02 M Hepes-HCl (pH 6.2), 0.15 M NaCl,

0.5% BSA, were added to the wells. After incubation at room temperature for 1 h, 100

pl of 0.5% washed l-day-old chick erythrocytes in BSA were added to each well.

Settling patterns were read after 18 hat 4’C. A haemadsorption-like pattern indicated

the presence of virus specific IgM, and a granular deposit of cells indicated the absence

of rubella specific IgM.

19

RESULTS

Specificity of IgWCI

Interference by RF Ten sera positive for RF and negative for anti-HBc were tested at

1: 80, 1: 400, and 1: 2,OOOdilutions in plates coated with F(ab’), anti-uandexamined by

IgM-CI for anti-HBc. In this experiment, HBcAg was not added. The results shown in

Table 1 indicate that RF gives false positive results at 1 : 2,000 dilution in 8 out of the

10 sera tested, and that the degree of the false positive results is not proportional to the

RF amount detected by latex agglutination. In order to explain the binding of

‘251-labe11ed IgG anti-HBc to the solid-phase-absorbed IgM, further experiments were

performed, without HBcAg, under the following conditions: in the first experiment,

1251-IgG anti-HBc was used either diluted in 2% BSA to reduce non-specific interac-

tions or in PBS after heating for 20 min at 63°C which might increase RF binding. In

the second experiment, 1251-F(ab’)z anti-HBc was used instead of the whole IgG

anti-HBc molecule. One serum without anti-HBc, but containing RF, was tested in

both experiments before and after fractionation on a Sephadex G 200 column (Phar-

macia). The results obtained (Table 2) show that the use of unheated or heated

1251-IgG always leads to false positive results with the native serum, whereas the use of

12SI-F(ab’)2 fragments does not. Furthermore, these false positive results were restrict-

ed to the RF-positive IgM fraction isolated by gel filtration.

RF might give false positive results in another way. Indeed, as HBcAg was isolated

from plasma containing hepatitis B virus and large amounts of anti-HBc, successive

TABLE 1

Interference by rheumatoid factor in the IgM capture immunoassay

Serum no. RF titre IgM (mg/ml) P/N” for sera diluted

1 : 80 1 : 400 I : 2000

1 &O 1.6 2.08 1.13 1.15

2 160 3.2 1.77 1.69 1.21

3 160 1.4 3.84 4.29 3.75

4 160 1 3.07 2.57 2.50

5 320 1.4 2.98 2.61 2.27

6 320 1.2 2.19 2.18 2.01

7 320 0.8 2.30 2.37 2.21

8 640 3.6 2.84 2.98 3.20

9 1280 2.5 6.94 7.17 5.16

10 5120 3.3 4.50 3.31 3.06

P/N value: ratio of test serum/negative mean value; a specimen was considered positive only if the P/N

value was >2.

TA

BL

E

2

Spec

ific

ity

of

igM

ca

ptur

e im

mun

oass

ay;

inte

rfer

ence

by

rh

eum

atoi

d fa

ctor

; ro

le

of

‘2’I

-ant

i-H

Bc

antib

odie

s

RF

titre

Se

rum

or

frac

tion

dilu

tions

P/N

va

luea

iZfI

-IgG

an

ti-H

Bc

dilu

ted

in P

BS-

BSA

iZ51

-IgG

an

ti-H

Bc

dilu

ted

in

PB

S a

nd

heat

ed

20

min

at

63

°C

rZ51

-F(a

b’),

anti-

HB

c

dilu

ted

in P

BS-

BSA

Non

- I

: 10

x.

35

4.39

1.

04

frac

tiona

ted

5120

1

: 10

0 3.

95

2.72

0.

48

seru

m

1 :

1000

3.

50

1.51

0.

96

kM

1 :

10

2.39

2.

90

1.37

frac

tion

640

I :

100

I.72

3.

45

1.24

1 :

1000

2.

06

1.75

N

Db

IgG

I

: 10

1.

73

1.75

0.

84

frac

tion

<20

I :

100

2.04

1.

67

1.15

1 :

1000

1.

04

2 N

Db

a P/

N va

lue:

se

e T

able

1.

b N

D:

not

done

.

21

steps of HBcAg purification cannot always remove completely IgG anti-HBc. Under

these conditions, IgG anti-HBc bound to HBcAg could react with RF. To test this

possibility, IgM and IgG fractions, isolated by gel filtration from a serum containing

RF and devoid of anti-HBc, were tested under the following conditions: wells were

coated with F(ab’), anti-u; IgM or IgG fractions were added to the wells. An HBcAg

preparation was placed in each well and antigen binding was subsequently revealed

by iz51-F(ab’), anti-HBc. ,As shown in Table 3, binding of HBcAg only occurred with

the RF fraction whereas the IgG fraction did not retain HBcAg. The same experiment

performed without adding HBcAg as a control, gave negative results with the IgM

fraction.

Interference by specific ZgG antibodies To investigate whether high levels of IgG

anti-HBc could interfere in the IgM-CI for IgM anti-HBc, IgG anti-HBc fractionated

by gel filtration on Sephadex G 200 was used. IgM-CI was then performed in wells

coated either with goat F(ab’), anti-u or with goat F(ab’), without known antibody

specificity. The results obtained (Table 4) show that IgG anti-HBc can interfere in the

IgM-test at the highest concentrations (1 mg/ml and 0.1 mg/ml). To try to avoid

these non-specific reactions, different dilution media were used. IgG anti-HBc was

diluted in TBS containing either 2% BSA or 2% BSA and 0.05% Tween 20. IgG

anti-HBc was also diluted in 2% BSA and various concentrations (10-5-2.5 mg/ml) of

human IgG (Sigma), then in human IgG with and without 50% fetal calf serum (FCS).

The results shown in Table 5 indicate that except for human IgG (10 mg/ml) with FCS

these different dilution media do not improve the specificity of the IgM-CI.

In the rubella assay, 20 sera from patients who were not infected recently were tested

at a starting dilution of 1 : 4 (instead of the 1 : 50 routine dilution). None of these 20

sera gave false positive results due to their rubella specific IgG content.

Sensitivity of ZgWCZ

By using the whole molecule of IgG anti-u for coating plates or wells, both RF and

TABLE 3

Specificity of IgM capture immunoassay; interference by rheumatoid factor; role of IgG anti-HBc bound to

HBcAg

RF positive serum RF titre Dilutions P/N valuea

With HBcAg Without HBcAg

IgM fraction 640 undiluted 10.8 1.28

1 : 100 6.6 1.04

IgG fraction <20 undiluted 1.65 0.72

1: 100 2.02 1.04

a P/N: see Table 1.

22

TABLE 4

Specificity of IgM capture immunoassay; influence of the IgG anti-HBc concentration

IgG anti-HBca

diluted to:

Goat F(ab’), anti-human Non-immune goat

IgM (u-chain specific) F(ab’), P/Nb PM

1000 ug/ml BSA’ 13.4 10.3

NHSC 7.9 9.5

100 pg/ml BSA 5.3 5.7

NHS 3 1.3

10 ug/ml BSA 0.56 ND*

NHS 1.60 ND

1 ug/ml BSA 0.74 ND

NHS 1.87 ND

a Titre at 20 mg/ml of proteins: 10m6.

b P/N value: see Table 1.

’ IgG isolated on Sephadex G 200 was diluted in PBS-BSA or in undiluted normal human serum (NHS).

d ND: not done.

TABLE 5

Specificity of IgM capture immunoassay; interference by specific IgG antibodies

Dilution media

for IgG anti-HBc

P/N valuea

2% BSA 5.1

2% BSA + 0.05% Tween 20 3

Human lgGb (mg/ml)

10 2.1

5 4.1

2.5 6

2% BSA f Human IgGb (mg/ml)

10 2.5

5 2.1

2.5 2. I Human IgGb (10 mg/ml)

+ 50% fetal calf serum 1.9

Wells were coated with non-immune goat F(ab’), and IgG anti-HBc was used at 350 ug/ml.

a P/N value: see Table 1.

’ Human IgG devoid of anti-HBc antibodies contained about 25% of aggregated IgG (examined by gel

filtration on Sephadex G 200).

23

TABLE 6

Sensitivity of IgM capture immunoassay; lack of interference by rheumatoid factor and anti-F(ab’),

antibodies

Rubella IgM positive Rubella specific

serum mixed with” IgM titre

Normal human serum

Rheumatoid factor positive serum

Latex agglutination titre 640

5120

Anti-F(ab’), positive serum

Haemagglutination titre 40

160

80 000

40 000

40 000

80 000

80 000

a Rubella IgM positive serum diluted 1 : 25 in either normal human serum or rheumatoid factor positive

human serum or anti-F(ab’), positive human serum.

anti-F(ab’), antibodies might react with IgG bound to the solid phase and, in this way,

decrease the sensitivity of the assay. To test this possibility, two sera containing

anti-goat F(ab’), antibodies, obtained from patients with endocarditis and two other

sera containing RF, collected from patients with remote rubella, were mixed as

indicated in Table 6 with a serum known to contain rubella specific IgM antibodies.

As shown in the Table, antibody IgM titres obtained with either mixture were not

significantly different from the antibody titre obtained with the IgM positive serum

tested alone.

The anti-p coated surface of the wells binds viral specific IgM as well as IgM with

other specificities. Therefore competition between specific and non-specific IgM was

studied. For that purpose, one serum containing high levels of polyclonal IgM without

anti-rubella antibodies and one serum without anti-HBc antibodies, were mixed

respectively, as indicated in Tables 7 and 8, with a serum drawn from a patient infected

with rubella virus and another from a patient infected with hepatitis B virus. Under

our working conditions, the titres were not significantly modified by addition of

non-immune IgM.

DISCUSSION

The IgM-CI appears to be the most sensitive of all the tests used for detecting

specific IgM antibodies. Therefore, it must be regarded as somewhat too sensitive, so

that persistence of IgM antibodies as well as their appearance during reinfections with

certain viruses, particularly herpes virus, should be reevaluated with such a method.

To do so, it is essential to undertake a thorough study of the specificity and sensitivity

of such a technique.

Different procedures are used to reveal antibodies bound to the solid phase: specific

24

TABLE 7

Sensitivity of IgM capture immunoassay; lack of competition between rubella specific IgM and non-specific

IgM

IgM concentration (mg/ml)

Rubella IgM Rubella IgM

positive serum negative serum

0.70 0

0.70 1.9

0.70 3.7

0.70 5.6

0.70 7.5

Rubella specific

IgM titre

6400

3200

3200

3200

3200

This experiment was performed by mixing 100 ul of a rubella IgM positive serum with 100 ul of a rubella

IgM negative serum.

TABLE 8

Sensitivity of IgM capture immunoassay; lack of competition between IgM anti-HBc and non-specific IgM

IgM concentration (mg/ml)

IgM anti-HBc IgM anti-HBc

positive serum negative serum

IgM anti-HBc

P/Nd

1.26 0 31.5

1.26 0.68 39.6

1.26 1.37 42.6

1.26 2.11 31.9

1.26 2.75 38.3

a P/N value: see Table I.

This experiment was performed by mixing 100 ul of an IgM anti-HBc positive serum with 100 ul of an IgM

anti-HBc negative serum. The mixture was used diluted at 1 : 2000 in PBS-BSA.

IgM can be detected either with the corresponding antigen present on the microorga-

nism (Desmonts et al., 1981) or with the radiolabelled or enzyme-linked antigen, or

indirectly, by means of a labelled antibody to the respective antigen (Chau et al., 1983)

or with erythrocytes if the antigen has haemagglutinating properties (Denoyel et al.,

198 1). The major drawback of most of these procedures might be the lack of specifici-

ty. It should be stressed that the more sensitive the procedure the higher the risk of

non-specific reactions. For example, non-specific reactions due to non-specific bind-

ing of IgG antibodies onto the solid phase can occur when detecting anti-HBc IgM

by means of radiolabelled antibodies, but not when detecting rubella specific IgM by

haemadsorption, as shown in our experiments.

25

Among the causes of non-specific reactions arising in the detection of antimicrobial

activity in the IgM fraction, RF is the most often cited. However, its interference in

IgM-CI is controversial. Flehmig et al. (1979) reported that RF does not induce false

positive results, while others found the reverse (Duermeyer and Van Der Veen, 1978;

Gerlich and Luer, 1979; Lemon et al., 1980). Therefore, we looked for RF interference

in IgM-CI. In this test, RF did give false positive results when the binding of HBcAg to

the IgM-coated solid phase was revealed with lzSI-IgG anti-HBc. This phenomenon is

probably due to IgG aggregated by fractionation, labelling or storage. This drawback

was overcome by using 1Z51-F(ab’)z fragments. Addition of aggregated IgG to the

samples may partially reduce this phenomenon (Gerlich and Luer, 1979). As RF is

frequently present in infected patients’ sera, the use of 1251-F(ab’)2 antibodies is

essential for the specificity of this procedure.

Specificity is also affected by the origin of the antigen. When the antigen used in the

IgM-CI is purified from biological fluids, it may be bound to antibodies, and the

immune complex is subsequently able to give false positive results with RF positive

sera. This point must be carefully controlled. Good preparations of HBcAg, devoid of

anti-HBc IgG, were obtained from liver (data not shown).

IgG-IgG interaction may also cause false positive reactions: specific IgG from sera

to be tested may physically or immunologically react with the IgG anti-u coated solid

phase. To prevent immune reactions due to anti-IgG antibodies possibly contaminat-

ing the goat IgG anti-u preparation, goat non-immune F(ab’)* was used for coating.

False positive results were only observed with high concentrations of IgG anti-HBc.

Under these conditions, BSA and partially aggregated human IgG were ineffective in

suppressing the false reaction observed for both goat F(ab’), anti-u and non-immune

F(ab’), coated wells. Only high concentrations of non-specific IgG with FCS improv-

ed the specificity of the reaction, Furthermore, with high concentration of IgM

antibodies, a prozoning phenomenon is more likely to occur (Chau et al., 1983). In our

experience, when we used a serum dilution of 1 : 1,000 in the IgM-CI for anti-HBc, we

did not observe significant non-specific IgG binding nor any prozoning phenomenon

(results not shown). No such false positive reactions were found with IgM-CI for

rubella IgM probably because of its lower sensitivity.

Until now, few reports have mentioned competition between specific IgM and other

serum factors. Kato et al. (1979) found that in some immunoassays RF reacted with

the Fc region of IgG bound to the solid phase and hindered specific binding. However,

in our experiments, RF did not inhibit the immune reaction of the anti-u coated solid

phase with the specific IgM of tested sera. In addition, anti-human F(ab’), (idiotypic

antibodies) detected in some patients’ sera may block specific IgM binding to the solid

phase by cross-reacting with goat F(ab’), antibodies. We have investigated and

excluded this possibility. The main inhibition factor might be represented by non-spe-

cific IgM contained in the sera to be tested. In fact, the results obtained indicate that

competition between specific IgM and non-specific IgM is not detectable by the

IgM-CI under physiological conditions (total IgM concentration from 1.2 mg/ml to 4

26

mg/ml) and under our working conditions (sample dilution of 1 : 2,000 in PBS-BSA

and not in NHS). These findings can be explained because in our experiment the

binding capacity of the solid phase is not saturated (data not shown). In this case,

non-specific rubella IgM can be added without decreasing the sensitivity of the test.

When, by contrast, the sample is diluted in NHS, the sensitivity of the assay is in

versely proportional to the concentration of NHS in the dilution medium (Gerlich and

Luer, 1979; Tedder and Wilson-Croome, 1980).

In conclusion, our results show that IgM-CI is a very sensitive and specific reaction.

However, in the indirect procedure, good specificity necessitates the use of labelled

F(ab’), fragments instead of the whole IgG molecule to reveal specific IgM binding.

Furthermore, an optimal dilution of tested sera must be determined to avoid false

reactions and prozoning phenomenon. To date, conflicting results reported in the

literature concerning persistence of IgM antibodies or their appearance in reinfections

seem to be due to variable sensitivity and specificity of IgM-CI. Bearing in mind the

pitfalls of IgM-CI, it is necessary to standardise its use before studying a particular

serological response to an infectious agent.

ACKNOWLEDGEMENTS

The authors are grateful to Francis Cape1 for his help. The secretarial aid of Mrs. 0.

Boudin is gratefully acknowledged.

REFERENCES

Chau, K.H., M.P. Harcie, R.H. Decker, I.K. Mushahwar and L.R. Overby, 1983, Hepatology 3, 142.

Denoyel, G., A. Gaspar and D. Peyramond, 1981, J. Clin. Microbial. 13, 698.

Desmonts, G.A., Y. Naot and J.S. Remington, 1981, J. Clin. Microbial. 14, 486.

Duermeyer, W. and J. van Der Veen, 1978, Lancet ii, 684.

Field, P.R., S. Shanker and A.M. Murphy, 1980, J. Immunol. Methods 32, 59.

Flehmig, B., M. Ranke, H. Berthold and H.J. Gerth, 1979, J. Infect. Dis. 140, 169.

Fraker P.J. and J.C. Speck Jr., 1978, Biochem. Biophys. Res. Commun. 80, 849.

Gerlich, W.H. and W. Luer, 1979, J. Med. Virol. 4, 227.

Grangeot-Keros, L., L. Lebrun, M.J. Briantais and J. Pillot, 1982, J. Immunol. Methods 51, 183.

Kato, K., U. Umeda, F. Suzuki, D. Hayashi and A. Kosaka, 1979, FEBS Lett. 102, 253.

Knez, V., J.A. Stewart and D.W. Ziegler, 1976, J. Immunol. 117, 2003.

Krech, U. and J.A. Wilhelm, 1979, J. Gen. Virol. 44, 281.

Lemon, SM., C.D. Brown, D.S. Brooks, T.E. Simms and W.H. Bancroft, 1980, Infect. Immun. 28,927.

March, SC., I. Parikh and P. Cuatrecasas, 1974, Ann. Biochem. 60, 149.

Pillot, J. and F. Capel, 1981, C.R. Acad. Sci. (Paris) 292, 1133.

Senet, J.M. and J. Pillot, 1971, Rev. Eur. Etud. Clin. Biol. 16, 938.

Tedder, R.S. and R. Wilson-Croome, 1980, J. Med. Virol. 6, 235.

Vejtorp, M., 1980, J. Virol. Methods 1, 1.

Ziegelmaier, R., F. Behrens and G. Enders, 1980, Med. Lab. 8, 16.