Improved immunoglobulin M serodiagnosis in Lyme borreliosis by using a mu-capture enzyme-linked immunosorbent assay with biotinylated Borrelia burgdorferi flagella.

Abstract

A mu-capture enzyme-linked immunosorbent assay (ELISA) for detection of serum immunoglobulin M (IgM) antibodies to Borrelia burgdorferi by using biotinylated purified B. burgdorferi flagella was developed. The diagnostic performance of the mu-capture ELISA was compared with that of a conventional indirect ELISA. Sera from untreated patients with erythema migrans (n = 50), neuroborreliosis (n = 100), and acrodermatitis chronica atrophicans (ACA; n = 48) were investigated. The cutoff of the ELISAs was adjusted to a diagnostic specificity of 98% on the basis of examination of 200 serum specimens from healthy controls. The mu-capture ELISA increased the diagnostic sensitivity in patients with erythema migrans from 32 to 48% (P less than 0.01) and in patients with neuroborreliosis from 37 to 57% (P less than 0.001). Because of an increased signal/noise ratio, the mu-capture ELISA yielded a significantly better quantitative discrimination of individual positive measurements from the cutoff (P less than 0.001). The increased signal/noise ratio was most likely a consequence of the elimination of IgG competition for the test antigen. This may also explain why 12% of patients with ACA showed significantly increased specific IgM levels only by the mu-capture ELISA. Of patients with ACA, 27% had IgM rheumatoid factor. The mu-capture principle with a directly labeled antigen showed no interference with IgM rheumatoid factor, in contrast to the indirect ELISA. The high diagnostic performance and ease of this three-step mu-capture ELISA make it suitable for routine anti-B. burgdorferi IgM serodiagnosis.

Full text

Page 1

JOURNAL OF CLINICAL MICROBIOLOGY, Jan. 1991, p. 166-173
0095-1137/91/010166-08$02.00/0
Copyright © 1991, American Society for Microbiology
Improved Immunoglobulin M Serodiagnosis in Lyme Borreliosis by
Using a ps-Capture Enzyme-Linked Immunosorbent Assay with
Biotinylated Borrelia burgdorferi Flagella
KLAUS HANSEN,'* KURT P11,2 AND ANNE-METTE LEBECH1
Borrelia Laboratory, Department ofInfection Immunology, Statens Seruminstitut,l
and Dakopatts,2 Copenhagen, Denmark
Received 4 June 1990/Accepted 18 October 1990
A ,u-capture enzyme-linked immunosorbent assay (ELISA) for detection of serum immunoglobulin M (IgM)
antibodies to Borrelia burgdorferi by using biotinylated purified B. burgdorferi flagella was developed. The
diagnostic performance of theFL-captureELISA was compared with that of a conventional indirect ELISA.
Sera from untreated patients with erythema migrans (n = 50), neuroborreliosis (n = 100), and acrodermatitis
chronica atrophicans (ACA; n = 48) were investigated. The cutoff of the ELISAs was adjusted to a diagnostic
specificity of 98% on the basis of examination of 200 serum specimens from healthy controls. The ,u-capture
ELISA increased the diagnostic sensitivity in patients with erythema migrans from 32 to 48% (P < 0.01) and
in patients with neuroborreliosis from 37 to 57% (P < 0.001). Because of an increased signal/noise ratio, the
,u-captureELISAyieldedasignificantlybetterquantitativediscrimination of individualpositivemeasurements
from the cutoff (P < 0.001). The increased signal/noise ratio was most likely a consequence of the elimination
of IgG competition for the test antigen. This may also explain why 12% of patients with ACA showed
significantly increased specific IgM levels only by the ,-capture ELISA. Of patients with ACA, 27% had IgM
rheumatoid factor. The ,u-capture principle with a directly labeled antigen showed no interference with IgM
rheumatoid factor, in contrast to the indirect ELISA. The high diagnostic performance and ease of this
three-stepFL-captureELISA make it suitable for routine anti-B. burgdorferi IgM serodiagnosis.
Lyme borreliosis is an increasingly recognized multisys-
temic infection caused by the recently discovered tick-borne
spirochete Borrelia burgdorferi (4, 27). Since the direct
demonstration of the organism in clinical specimens by
cultivation is a low-yield procedure, confirmation of the
clinical diagnosis
specific antibody response. The diagnostic sensitivity of
currently used serological assays, however, has been lim-
ited, especially in early disease (13, 14, 18, 25, 32), probably
because of a low antigen load, a slow and late immune
response, and also the different and often low specificities of
the test antigens used. Recently, significantly improved
diagnostic performance ofimmunoglobulin M (IgM) and IgG
serodiagnosis has been reported by using either a flagellum-
enriched protein fraction (7) or purified native B. burgdorferi
flagella (13, 14, 18) instead of a sonic extract as a test antigen
for an enzyme-linked immunosorbent assay (ELISA). Since
the detection of specific IgM is of special interest in early
disease and may be also an indicator of active or recent
infection, further improvements of methods for specific IgM
detection are essential. Two types of ELISAs are available
for detecting antigen specific IgM: indirect and the ,u-capture
ELISAs (10, 31). In the indirect ELISA, the test antigen
adsorbed to the solid phase binds specific antibodies of all
isotypes in amounts proportional to their concentrations in
the test sample. Bound specific IgM is subsequently detected
with a second antibody. The main limitations of the indirect
ELISA are false-positive results due to interference of IgM
rheumatoid factor (RF) (6, 30) and false-negative or false-low
is mainly achieved by detection of a
*Corresponding author.
IgM results due to competition with specific IgG for anti-
genic sites (6). Both ofthese limitations may be overcome by
the >-capture principle, as the selective binding of patient
IgM eliminates IgG competition and furthermore, in combi-
nation with a directly labeled test antigen, avoids interfer-
ence of IgM RF.
Here we describe a simple, sensitive, and specific three-
step ,u-capture ELISA for B. burgdorferi-specific IgM using
biotinylated purified B. burgdorferi flagella. The diagnostic
performance was compared with an indirect IgM flagellum
ELISA by using 198 serum specimens from patients with
erythema migrans (EM), neuroborreliosis (NB), and acro-
dermatitis chronica atrophicans (ACA).
MATERIALS AND METHODS
Serum specimens from patients. A total of 198 serum
specimens from patients with active untreated Lyme borrel-
iosis were used in this study. They were divided into three
groups according to clinical manifestations.
(i) Sera from 50 Swedish patients with EM. The diagnoses
of 50 Swedish patients with EM were based on clinical
evidence and made by Eva Asbrink (Department of Derma-
tology, Sodersjukhuset, Stockholm, Sweden). These sera
were collected from 1984 to 1988 from 9 males and 41
females between 6 and 71 years of age (median age, 45
years). The disease duration ranged from 1 to 12 weeks
(median duration, 3 weeks). Furthermore, consecutive se-
rum samples from six Danish patients with EM were in-
cluded. These sera were taken up to 6 months after treat-
ment.
(ii) Sera from 100 Danish patients with NB. One hundred
166
Vol. 29, No. 1

Page 2

i-CAPTURE ELISA WITH B. BURGDORFERI FLAGELLA
Danish patients with NB were all hospitalized between 1984
and 1989 (61 males and 39 females between 5 and 74 years of
age; median age, 47 years). For 91 patients the diagnosis was
lymphocytic meningoradiculitis with typical painful radicu-
litis, mononeuritis multiplex (Bannwarth's syndrome), or
both. Nine patients had chronic progressive encephalomy-
elitis with definite intrathecal antibody synthesis against B.
burgdorferi (15). Before treatment all patients had lympho-
cytic pleocytosis in cerebrospinal fluid, with counts of 13 x
106 to 830 x 106 cells per liter (median cell count, 211 x 106
cells per liter). Of the 100 patients, 56 recalled a previous
EM-like skin lesion.
The disease duration, defined as the time after onset of
neurological symptoms, ranged from 4 days to 6 years
(median duration, 26 days). Consecutive serum samples
from six patients with lymphocytic meningoradiculitis fol-
lowed up -4 months after treatment were also included in
this study.
(iii) Sera from 48 Danish patients with ACA. Sera from 48
Danish patients with ACA were collected from 1985 to 1988
from 15 males and 33 females between 17 and 80 years of age
(median age, 54 years). The diagnosis was in every case
made by a dermatologist on the basis of the typical clinical
appearance of ACA and a high IgG titer to B. burgdorferi in
serum. The disease duration ranged from 0.5 to 20 years,
with a median duration of 3 years.
Control serum specimens. Sera from 200 healthy controls
(age range, 16 to 71 years; median age, 40 years) were used
for determination of the 98%-specific cutoff level in the
ELISAs. None of the 200 healthy control serum specimens
contained anticardiolipin antibodies by the Wassermann
reaction and rapid plasma reagin test. Additionally, we
investigated sera from patients with active primary syphilis
(n = 25) and secondary syphilis (n = 15). All sera from
patients with syphilis showed a strong positive Wassermann
reaction, a positive rapid plasma reagin test, and a reactivity
23+ in the fluorescent treponemal antibody absorption test.
Furthermore, we included sera from patients with active and
serologically proven leptospirosis (n = 22) and infectious
mononucleosis (n = 15). Leptospira serological tests were
performed by using a microscopic agglutination test (cutoff
titer, 1:300). Of the 22 patients, 16 were seropositive for
Leptospira icterohaemorrhagiae and 6 were seropositive for
Leptospira sejroe (titer range, 1:1,000 to 1:10,000; median
titer, 1:3,000). We also tested 15 serum samples from pa-
tients with anIgM RF level of .100 IU/ml of serum and sera
from 259 forestry workers which were used in a recent
serosurvey for anti-B. burgdorferi IgG (22).
Biotinylation ofB.burgdorferiflagella. We used the 41-kDa
native B. burgdorferi flagellar protein as the test antigen,
which was purified from strain DK1 as described previously
(14). The protein concentration of the antigen preparation
was determined by the modified Bradford technique (11).
Biotinyl N-hydroxysuccinimide (BNHS) with a spacer con-
sisting of six aminocaproic acid residues (B2643; Sigma
Chemical Co., St. Louis, Mo.) was used to covalently bind
biotin to the flagellar protein (19). A ratio of 0.14 mg of
BNHS to 1 mg of flagellar protein was the optimal binding
ratio. BNHS (20 mM) dissolved in dimethylformamide (Sig-
ma) was added to the flagellar protein in 0.125 M carbonate
(NaHCO3) buffer (pH 9.0). After incubation for 8 h at 20°C,
the reaction mixture was dialyzed for at least 24 h against
several changes of 0.05 M phosphate (KH,PO4) buffer-0.1 M
NaCl (pH 7.3). Biotinylation of a 1-ml suspension of purified
flagella (protein concentration, 1 mg/ml) yielded a sample of
biotin-labelled flagella which could be diluted 1:500 for
ELISA use.
FL-CaptureELISA. Flat-bottom microdilution plates (Max-
isorb; Nunc, Roskilde, Denmark) were coated with 100 ,u1 of
,u-chain-specific rabbit anti-human IgM (code A425; Dako-
patts, Copenhagen, Denmark) diluted 1:1,000 in phosphate-
buffered saline (PBS) (pH 7.2) and incubated for 2 h at 20°C
on a rocker platform and overnight at 4°C. After the plates
were washed, 100 ,ul of serum diluted 1:200 in PBS contain-
ing 0.5 M NaCl, 0.1% (vol/vol) Tween 20, and 1% (wt/vol)
bovine serum albumin (pH 7.2) was added to the wells and
incubated for 2 h at 20°C on a rocker platform. Biotinylated
B. burgdorferi flagella and avidin peroxidase (code P347;
Dakopatts) were mixed at dilutions of 1:500 and 1:5,000,
respectively, in the same buffer that was used for dilution of
sera. After the plates were washed, 100 ,ul of the solution
containing the biotinylated B. burgdorferi flagellum-avidin
peroxidase complex was added to the wells and incubated
for 3 h at 20°C on a rocker platform. Unbound biotinylated
flagella were removed by washing. Next, 200
substrate o-phenylenediamine (0.41 mg/ml; Sigma) in citrate
buffer (pH 5.0) with 0.04% (vol/vol) H,02 was added to each
well. After 15 min protected from light, the enzymatic
reaction was stopped by the addition of 50 p.l of 3 M H2SO4.
The optical density (OD) at 492 nm was read spectrophoto-
metrically (Immuno Reader NJ2000; Nippon Inter Med,
Tokyo, Japan). All washings were done three times with
PBS containing 0.5 M NaCl and 0.1% (vol/vol) Tween 20 (pH
7.2). Samples were tested in duplicate and retested if the two
OD values differed more than 10% from the mean. To limit
interassay variation, three calibrator serum samples, one
each with a low, median, and high OD value, were included
on every plate for construction of a standard curve. The
average OD obtained from every sample was adjusted to this
standard curve.
Indirect ELISA for IgM and IgG antibodies to B. burgdor-
feri flagella. The indirect ELISA for IgM and IgG antibodies
to B. burgdorferi flagella was performed as previously de-
scribed (14). Briefly, microdilution plates (Immunoplates;
code 2-69620; Nunc) were coated with purified B. burgdor-
feri DK1 flagellar protein. After being blocked with PBS
containing 1% (wt/vol) bovine serum albumin, sera diluted
1:200 in PBS containing 0.1% (vol/vol) Tween 20 and 0.5%
(wt/vol) bovine serum albumin (pH 7.2) were incubated for 2
h at 20°C. Bound antibody specific to B. burgdorferi was
then detected with a peroxidase conjugate, either rabbit
anti-human IgM (code P215; Dakopatts) or rabbit anti-
human IgG (code P214; Dakopatts). The substrate reaction
and reading of the test were performed as described above.
The diagnostic cutoff OD was adjusted to be 98% specific on
the basis of the examination of sera from 200 healthy
controls and was 0.160 for the IgG assay and 0.375 for the
IgM assay.
IgM RF detection and absorption. The 48 serum specimens
from patients with ACA were investigated for the presence
ofIgM RF by an ELISA using human IgG, and the results
were evaluated according to the 95%-specific upper limit of
the assay, 8 IU/ml (16). IgM RF was absorbed by immuno-
precipitation of IgG with a commercial kit (RF-Absorbans;
Behringwerke, Marburg, Federal Republic of Germany).
Statistical analysis. The diagnostic
,u-capture and indirect ELISAs were compared by the non-
parametric McNemar's test (1) by considering only the
number of samples which disagreed in positivity or negativ-
ity in the two assays. These observations are assumed to
follow a binomial distribution of paired data. Furthermore,
p.l of the
sensitivities of the
VOL. 29, 1991
167

Page 3

168
HANSEN ET AL.
a
10
10
10
10
10
10
B
TABLE 1. Diagnostic sensitivity of the B. burgdorferi flagellar
>-capture, indirect IgM, and indirect IgG ELISAs
in EM and NB
Clinical
% of patients positive by indicated
ELISA
Indirect
Capture
IgM
manifestation and
no. of wk after
onset
No. of
patients
Indirect
IgG
EM
<2
2-5
.6
10
27
13
50
44
54
20
41
23
10
26
46
Total
50
48
32
28
NB
<2
2-5
.6
23
47
30
65
60
47
48
45
17
70
81
100
Total
100
57
37
84
b
0
1:2
1:8
1:32
1:128
1:512
NS
FIG. 1. Measurement of anti-B. burgdorferi IgM by the ,u-cap-
ture ELISA in a 10-fold dilution series of a strongly IgM-positive
serum in sample dilution buffer (B) only (a) and in a 2-fold dilution
series of the same IgM-positive serum into negative serum (NS)
before the final dilution of 1:200 in sample dilution buffer (b). The
very different effects of dilution with serum and sample dilution
buffer on the IgM signal demonstrate that the ,u-capture ELISA
measures not an absolute concentration of specific antibody but the
relative amount of specific antibody versus total IgM. Addition of
negative serum displaces the specific IgM fraction, thus reducing the
signal.
the quantitative discrimination of the two assays between
controls and seropositive patients was estimated. For each
individual sample positive in both ELISAs, the difference
between the achieved OD value and the cutoff level was
calculated for both assays. These differences were then
compared by the Wilcoxon rank sum test for paired data.
RESULTS
Figure 1 shows the IgM reactivity in the u-capture ELISA
of a serum sample strongly positive for anti-B. burgdorferi
IgM diluted in either sample dilution buffer (Fig. la) or
negative serum (Fig. lb). According to Fig. la, the ,u-capture
layer was saturated with patient IgM, yielding an almost
maximal specific IgM signal, until the sample was diluted
1:104. On the other hand, if the positive serum was diluted in
a negative serum, the specific IgM signal decreased rapidly
(Fig. lb). These curves demonstrate that the ,u-capture
ELISA measures the relative amount of specific IgM anti-
body versus total IgM and not the concentration of specific
IgM.
The interassay variation of the ,u-capture ELISA was
determined by testing three serum specimens on 11 indepen-
dent days. Serum specimen 1 had a mean OD of 0.535
(standard deviation, 0.0446) and a coefficient of variation of
8.33%; serum specimen 2 had a mean OD of 0.880 (standard
deviation, 0.0769) and a coefficient of variation of8.73%, and
serum specimen 3 had
deviation, 0.180) and a coefficient of variation of 5.86%. The
98%-specific diagnostic cutoff of the ,u-capture assay was
fixed at an OD of 0.300 (Fig.
examination of the same 200 healthy control serum samples
used for cutoff determination in the indirect IgM and IgG
ELISAs.
The overall diagnostic sensitivity of anti-B. burgdorferi
IgM detection increased significantly from 32 to 48% (P <
0.01) when the ,u-capture ELISA was used for patients with
EM and from 37 to 57% (P < 0.001) when this assay was
used for patients with NB (Fig. 2; Table 1). Furthermore, the
,u-capture ELISA generally yielded much higher OD signals
than the indirect ELISA, resulting in a significantly im-
proved quantitative discrimination of individual seropositive
samples from the cutoff (Fig. 3). Except for serum from one
patient with NB which showed a slightly elevated 1gM
antibody level (OD, 0.470) in the indirect ELISA (Fig. 3b),
all sera from patients with EM and NB that were reactive in
the indirect assay were also reactive in the ,u-capture ELISA
(Fig. 3). Regarding the IgG and ,u-capture ELISA results of
the sera from 50 patients with EM, 6 were only IgG positive,
16 were only IgM positive, 30 were either IgG or IgM
positive, and 8 were IgG and IgM positive. The correspond-
ing figures for the 100 patients with NB were as follows: 38
were only IgG positive, 11 were only IgM positive, 95 were
IgG or IgM positive, and 46 were IgG and IgM positive.
None of the 200 serum specimens from healthy controls
contained significant IgG and IgM anti-B. burgdorferi anti-
a mean OD of 1.714 (standard
2) on the basis of the
OD
2.6
22
1.8
1.4
1.0
0.6
0.2
OD
2.6
2.2
1.8
1.4
1.0
0.6
Q2
J. CLIN. MICROBIOL.

Page 4

V
9-CAPTURE ELISA WITH B. BURGDORFERI FLAGELLA
VOL. 29, 1991
OD
1
4
I
ills.
r
I
!
In
ai
S.3
I
CONTROLS
EM
NB
I
2.0
1.5
1.0
0.5
0.1
ACA
I
J-
CONTROLS
EM
NB
ACA
OD
2.0
1.5
1.0
0.5
-
0.1
I-
_1-
I
.1.
..
I
t
*
II
0 127< 0050
CONTROLS
EM
NB
ACA
FIG. 2. Anti-B. burgdorferi IgM measured by ,u-capture ELISA (left) and indirect ELISA (center) and anti-B. buirgdorferi IgG measured
by indirect ELISA (right) in sera of 200 healthy controls, 50 patients with EM, 100 patients with NB, and 48 patients with ACA. The horizontal
lines mark the 98%-specific diagnostic cutoff levels in the indicated test. Open circles indicate patients with ACA who had an IgM RF level
of >20 IU/ml.
body levels. Dividing the patients with EM and patients with
NB into three groups according to disease duration (Table 1)
demonstrates the relatively higher frequency of IgM-positive
tests in early disease and the increasing frequency of detect-
able specific IgG with increasing disease duration. Only one
of the nine patients with chronic NB was low-grade IgM
positive in the ,t-capture ELISA (OD, 0.410).
We also tested 48 serum specimens from patients with
ACA. All had high IgG antibody levels to B. burgdorferi,
whereas only six and five serum specimens were IgM
positive in the ,u-capture and indirect IgM ELISAs, respec-
tively (Fig. 2). To rule out the possible influence of IgM RF
on the outcome of the IgM detection, the 48 serum speci-
mens were investigated for IgM RF. An IgM RF level of .8
IU/ml was found in 13 patients (27%), and in all but 6
patients this level was <20 IU/ml. The five IgM-positive
samples in the indirect IgM ELISA all had a significantly
increased concentration of IgM RF (range, 38 to 222 IU/ml;
median, 55 IU/ml). Absorption ofthe IgM RF was performed
on the six serum samples with IgM RF levels of >20 IU/ml
and on eight serum samples with IgM RF levels of <8 IU/ml.
Measurements of IgM antibodies by the indirect assay
before and after RF absorption showed that the absorption
also lowered the OD values in sera without a detectable RF
(Fig. 4). The previously high IgM reactivity of five serum
specimens in the indirect ELISA was reduced significantly in
four and to a lesser degree in one serum specimen (Fig. 4).
All four samples with significant reductions after RF absorp-
tion were negative in the >-capture ELISA (Fig. 4). In the
>.-capture assay, on the other hand, a total of six ACA serum
specimens (12%) were clearly positive (Fig. 2). Only one of
these contained IgM RF, and it was this serum that showed
a decreased but not abolished IgM reactivity in the indirect
assay after RF absorption. This may indicate that the IgM
reactivity in this patient partly was due to IgM RF and partly
was specific IgM reactivity. A further 15 serum samples from
patients selected for IgM RF levels of .100 IU/ml were
studied. All 15 serum samples were negative for IgG anti-
bodies to B. burgdorferi. Only one ofthem had a low level of
IgM reactivity in the indirect ELISA (OD, 0.410) and none
was positive in the p,-capture ELISA.
The diagnostic specificity of the >.-capture ELISA was
further evaluated with sera from patients with active primary
and secondary syphilis, leptospirosis, and infectious mono-
nucleosis. The results are summarized in Table 2. Only
patients with infectious mononucleosis showed considerable
anti-B. burgdorferi IgM reactivity in the ,-capture ELISA as
well as in the indirect IgM ELISA (Table 2).
Figure 5 shows the investigation of consecutive serum
samples from six patients with EM and six patients with NB.
Sharp titer rises were detectable within 1 to 7 weeks in
patients with EM. The specific IgM response in EM and NB,
although declining, was still above the cutoff for at least 4 to
6 months.
OD
2.0-
1.5
1.0
-
Q5
0.1
-
.L
L
169
I
I
t
lqp----

Page 5

170
HANSEN ET AL.
00
2.0
1.5
U)
usww
E
U
1.0
0.5
0.1
0D
2.0
1.5
U)
:3
,,
1.0
U
0.5
0.1
A
I
a
OD
1.2
1.0
0.8
0.6
0.4
0.2
01
0.5
1.0
1.5
2.0
INDIRECT 1gM EUISA
asa
00
a
s
...............b
---
a1
._ .1
. --
0.5
._ ._
I
.
I
.
.
I
I I.
.
1.0
1.5
2.0
INDIRECT IgM ELISA
00
FIG. 3. Correlation of specific anti-B. burgdorferi IgM measure-
ment in sera of 50 patients with EM (a) and 100 patients with NB (b)
by the >t-capture and indirect IgM ELISAs. The vertical lines
represent the cutoff of the indirect ELISA, and the horizontal lines
mark the cutoff of the ,u-capture ELISA. The ,u-capture ELISA
demonstrated a significantly improved signal/noise ratio, as esti-
mated by comparing the differences between the achieved OD
values and the cutoff level in each test by the Wilcoxon rank sum
test for paired data. (a) P < 0.001; (b) P < 0.001.
Sera from 259 forestry workers which in a previous
serosurvey for anti-B. burgdorferi IgG (22) had shown a high
prevalence (35%) of specific IgG antibodies were tested by
the ,u-capture ELISA. Only 4.6% of the samples were IgM
positive (OD range, 0.310 to 1.520; median, 0.400).
DISCUSSION
Since the introduction of the ,u-capture ELISA in 1978
(10), this technique has been shown to be very useful for the
a
b
c
RF:20 lU/mi
-n
a
b
c
RF < 8 lU/mi
FIG. 4. Anti-B. burgdorferi 1gM levels in six serum specimens
from patients with ACA with high levels of1gM RF (38 to 222 lU/ml)
and in eight serum specimens from patients with ACA with IgM RF
levels of <8 IU/ml. The specific IgM was measured before (a) and
after (b) absorption of IgM RF in the indirect IgM ELISA and
without absorption in the ,u-capture ELISA (c). Open circles indi-
cate the only RF-positive patient who tested positive in the ,u-cap-
ture ELISA (see text). The horizontal lines mark the cutoff levels in
the indicated tests.
detection of antigen-specific IgM, especially in viral diseases
but also in syphilis (23) and toxoplasmosis (26). Recently the
pu-capture principle was applied for specific IgM detection in
Lyme borreliosis by Berardi et al. (3) and Karisson and
Granstrom (17). Although both studies used a very similar
assay design with five incubation steps and a sonic extract of
B. burgdorferi as a test antigen, only Berardi et al. (3)
reported significantly improved IgM detection. The concept
of our ,u-capture assay was different, with only three steps
and a single, directly labeled purified B. burgdorferi antigen.
This assay improved specific IgM detection significantly
compared with an indirect ELISA (Table 1). The improve-
ment consists of increased sensitivity due to a much higher
signal/noise ratio and increased specificity due to elimination
ofIgM RF interference. There are two main explanations for
the increased sensitivity. (i) The selective binding of IgM
avoids competition with specific IgG for the test antigen,
which may give false-low or false-negative IgM results in the
indirect assay. This may explain the significantly increased
signal/noise ratio (Fig. 3) and why only the ,u-capture ELISA
detected specific IgM in 12% of the patients with ACA, all
with high specific IgG levels (Fig. 2). (ii) The binding of a
representative part of the total IgM and subsequent detec-
tion of anti-B. burgdorferi specific IgM reflect the relative
amount of specific versus total IgM (Fig. 1) and not the
absolute concentration of specific IgM as measured by the
indirect ELISA. An increase in the relative amount of
specific IgM seemed to be a more sensitive parameter than
an increase in the absolute concentration of specific IgM in
serum measured by an indirect ELISA (26).
Comparing our assay with the recent attempts to measure
B. burgdorferi-specific IgM with a ,i-capture ELISA (3, 17),
J. CLIN. MICROBIOL.