Clinical Chemistry 46:2
273–278 (2000)
Clinical Immunology
False Positivity in a Cyto-ELISA for
Anti-Endothelial Cell Antibodies Caused by
Heterophile Antibodies to Bovine Serum Proteins
Ronan Revelen, Anne Bordron, Maryvonne Dueymes, Pierre Youinou, and
Josiane Arvieux*
Background: ELISAs with fixed endothelial cells or cell
lines are widely used screening tests for anti-endothelial cell antibodies (AECAs), but spurious increases
occur. We examined interferences by heteroantibodies
and means to eliminate them.
Methods: AECAs were measured by ELISA on fixed
layers of the human endothelial cell line, EA.hy 926, in
a panel of 60 patient serum samples diluted in bovine
serum albumin. Heteroantibodies against fetal calf serum (FCS) proteins were demonstrated and characterized in an ELISA—the interference assay—that used
FCS-coated plates and Tween 20-containing buffer as
blocking agent and sample diluent, as well as by immunoblotting.
Results: In 12 of 60 patient serum samples, spurious
increases of AECA titers were produced by endogenous
antibodies reacting with FCS proteins from culture
medium that were coated onto the solid-phase at the
time of cell plating. This mechanism of interference was
supported experimentally by exposing extracellular matrix, varying cell density, and incubating wells with FCS
alone. The heterophile antibodies were mainly IgG and
IgA, and in inhibition experiments, they recognized
serum proteins from goat, sheep, and horse. Washing
cells free of FCS before plating, or adding FCS (100
mL/L) to the patient sample diluent eliminated spurious
signals from all 30 tested sera, but the latter method had
practical advantages.
Conclusions: Antibodies against animal serum proteins
are a frequent cause of erroneous results in cyto-ELISAs.
The interference can be eliminated by simple antibody
absorption in FCS-containing dilution buffer.
© 2000 American Association for Clinical Chemistry
Anti-endothelial cell antibodies (AECAs)1 represent a
heterogeneous group of antibodies against poorly characterized targets that have been reported in a variety of
inflammatory disorders, and they may be valuable as
markers of disease activity (1 ). Often considered as an
epiphenomenon of vascular injury, AECAs may also play
a role in the pathophysiology of associated diseases,
especially by inducing endothelial cell activation and/or
apoptosis (1, 2 ). Numerous methods have been developed to assay AECAs, including immunofluorescence,
immunoblotting, functional assays, and more commonly,
solid-phase immunoassays such as ELISAs and RIAs, but
parallel studies of the same samples in different tests have
been infrequent (3–5 ). One of the main problems faced in
this field is the lack of agreement on a standardized
method to detect AECAs, which renders the interlaboratory comparison of the results somewhat hazardous (6, 7 ).
The large discrepancies observed in clinical correlations of
AECA concentrations may be ascribed to several variables
that affect their measurement. For example, the choice of
cells (with possible interdonor variability) or the use of
cell membrane extracts, cell culture and fixation conditions, heating of serum samples (8 ), and the expression of
results may affect the outcome.
While attempting to optimize cell fixation in such an
ELISA, we identified another possible confounding factor:
a substantial proportion of patient samples containing
IgG antibodies against bovine serum proteins gave falsepositive results in the AECA test. Human antibodies
reactive with animal proteins, including immunoglobu-
Laboratory of Immunology, Institut de Synergie des Sciences et de la Santé,
Brest University Medical School, F29609 Brest Cedex, France.
*Address correspondence to this author at: Laboratory of Immunology,
Brest University Medical School Hospital, BP 824, F 29609, Brest Cedex, France.
Fax 33-298-80-10-76; e-mail [email protected].
Received October 20, 1999; accepted November 10, 1999.
1
Nonstandard abbreviations: AECA, anti-endothelial cell antibody; BSA,
bovine serum albumin; ␤2GPI, ␤2-glycoprotein I; FCS, fetal calf serum; PBS,
phosphate-buffered saline; MAb, monoclonal antibody; and ECM, extracellular matrix.
273
274
Revelen et al.: Heterophile Antibody Interference in a Cyto-ELISA
lins and common blocking agents such as cow’s milk,
bovine serum albumin (BSA), and non-immune animal
sera, are also designated as heterophile antibodies when
they seemingly do not arise against a well-defined immunogen (9 ). These antibodies represent an often unrecognized source of interference in all immunoassays, potentially giving rise to false-positive results (10 –14 ). The
present study was therefore designed (a) to delineate
heterophile antibody interference in our ELISA for
AECAs, (b) to propose strategies for resolving the problem, and (c) to better characterize the antibodies involved.
A heightened awareness of this type of interference on the
part of investigators performing the AECA ELISA should
improve its reliability, a prerequisite to better evaluate the
role of AECAs in clinical practice and unravel their
putative pathogenicity.
Materials and Methods
patients and controls
Serum samples, obtained from 60 patients with various
non-organ-specific autoimmune disorders, were selected
to cover a wide range of IgG AECA concentrations, from
weakly to strongly positive. Of these, eight were part of a
previous study, and their apoptosis-inducing effect on
endothelial cells has been described (2 ). In addition, 30
patient sera that previously had been shown to react in
ELISAs to bovine, but not human, ␤2-glycoprotein I
(␤2GPI) (15 ) were used to determine the prevalence of
heterophile antibodies under study (see below) in this
population. Sera from 45 healthy blood donors served as
controls. All sera were stored at ⫺20 °C until use.
blocked with PBS containing 10 g/L BSA, and then were
successively exposed to patient sera (100 ␮L of a 1:100
dilution in the same buffer) and to peroxidase-conjugated
rabbit F(ab⬘)2 anti-human IgG, IgM (both diluted 1:4000),
or IgA (diluted 1:2000) antibodies (Dako), followed by
incubation with o-phenylenediamine (0.2 g/L in 0.05
mol/L phosphate buffer, pH 5, containing 0.5 mL/L
H2O2). Incubations were for 1 h at 37 °C, separated by
three PBS washes. For each serum, the absorbance at 492
nm of control wells (blocked with PBS-BSA but without
cells) was subtracted from the absorbance in the wells
with EA.hy 926 cells to account for nonspecific binding.
The mean ⫹ 3 SD of 45 control sera was taken as the
threshold for positivity.
When we became aware of heterophile antibody interference in the above ELISA, we modified its format by
diluting serum samples and conjugates in PBS containing
100 mL/L FCS and by blocking control wells with PBSFCS.
inhibition with animal sera
The ability of fluid-phase serum proteins from various
animal species to inhibit the binding of patient heterophile antibodies to solid-phase-bound FCS proteins was
measured by ELISA. In this FCS-ELISA—the interference
assay—plates were coated with PBS containing 100 mL/L
FCS for 4 h at 37 °C, and PBS containing 1 mL/L Tween 20
served as the blocking and wash buffer. Patient serum
samples were diluted 1:100 in PBS-Tween alone or in
PBS-Tween containing the indicated concentrations of
competing animal serum. The development was conducted as described above in the AECA test.
endothelial cells
Fixed monolayers of the human endothelial hybrid cell
line EA.hy 926 (a kind gift from C.S. Edgell, University of
North Carolina, Chapel Hill, NC), obtained by fusing
human umbilical vein endothelial cells with a human
lung carcinoma, have been shown to be appropriate to
detect AECAs by ELISA (3, 4 ). Cells were cultured at
37 °C and 5% CO2 in DMEM (Eurobio) supplemented
with 100 mL/L heat-inactivated fetal calf serum (FCS;
BioWhittaker), 2 mmol/L glutamine, and 50 mL/L hypoxanthine aminopterin thymidine (FCS medium). After
cultures had reached confluency, cells were detached
using a mixture of 1.25 g/L trypsin and 0.2 g/L EDTA
(1:4, by volume) in Tris buffer (Eurobio) for 3 min at 37 °C,
and then were washed once in FCS medium.
ELISAs for AECAs
EA.hy 926 cells were plated (1 ⫻ 104 cells/well in FCS
medium) in flat-bottomed 96-well microtiter plates
(Nunc). Confluent cell layers from 2- to 3-day-old cultures
were washed with phosphate-buffered saline (PBS) and
fixed with 100 ␮L of 1 g/L glutaraldehyde (or lower
concentrations, as indicated) for 10 min at 4 °C. Alternatively, cells were fixed with 100 ␮L of absolute ethanol for
5 min at 4 °C. After three washings in PBS, plates were
western blot analysis
FCS proteins (7 ␮L of crude FCS per lane) were resolved
by sodium dodecyl sulfate-polyacrylamide gel electrophoresis in 10% acrylamide gels under nonreducing conditions, followed by electrotransfer to nitrocellulose.
Membranes were blocked for 2 h with PBS containing 1
mL/L Tween and were subjected to successive 1-h incubations with serum samples (1:100 in PBS-Tween) and the
same anti-IgG conjugate as for ELISAs (1:4000 in PBSTween). Peroxidase activity was detected with diaminobenzidine.
statistical analysis
The Student t-test for paired data and the ␹2 test were
used when appropriate. P ⬍0.05 was considered statistically significant.
Results
cause of spurious increases in aeca titers
In a study designed to optimize cell fixation in ELISA
procedures for AECA detection, we noticed that apparent
AECA activity did not decrease with cell loss in some
patients’ samples (Fig. 1). This group, exhibiting spurious
increases in antibody titers, included 12 of 60 (20%)
Clinical Chemistry 46, No. 2, 2000
patient serum samples and was designated group I. Cell
losses during the fixation step were estimated by phasecontrast microscopy, and reactivity with a monoclonal
antibody (MAb) against a cell surface molecule of endothelial cells, thrombomodulin, that served as a probe for
residual cell density. As shown in Fig. 1, the most important cell loss occurred in the absence of fixative and led to
the strongest ELISA responses with group I sera. As a
whole, all 12 group I sera showed increased relative
absorbances that were negatively correlated with residual
cell density, suggesting steric hindrance by cells and
antibody recognition of noncellular targets. The remaining 48 patients (group II) possessed exclusively genuine
AECAs, with a reactivity pattern resembling the one
observed with the anti-thrombomodulin MAb, except for
consistently increased binding after ethanol (vs glutaraldehyde) fixation.
To gain further insight into the understanding of the
mechanism of the false positivity that characterized group
I sera, we compared different treatments of the ELISA
plate (Table 1). To avoid interference from added proteins, blocking and washing buffers contained Tween 20
alone in these experiments. Adherent EA.hy 926 cells were
either detached with trypsin-EDTA or by exposure of their
extracellular matrix (ECM) with detergent because antibodies to ECM components such as collagen and laminin
275
have been reported in AECA-positive patients (1 ). The
coating by EA.hy 926 cells in FCS-containing culture
medium was sensitive to digestion with trypsin, which
suggests the involvement of proteic structures in antigenicity. These proteins originated entirely from FCS and
probably included BSA, as demonstrated when cells were
omitted and replaced by crude FCS or purified BSA.
redesign of aeca-elisa to prevent heterophile
antibody interference
We next investigated different strategies to solve the
problem of interference by heteroantibodies against FCS
proteins in our ELISA for AECAs. Typical results are
shown in Fig. 2. We first washed EA.hy 926 cells twice
and switched to serum-free DMEM before plating for
ELISA. Although this approach was efficient for decreasing to baseline the binding of 9 of 12 group I sera when
BSA served as sample diluent, it suffered from the requirement for approximately four times as many cells per
well because the culture time was reduced to the few
hours necessary to get firm cell adherence to the solid
phase. We next evaluated the substitution of 100 mL/L
FCS for the 10 g/L BSA used to dilute the test samples
(Fig. 2B), based on the rationale that reactivity against
bound FCS proteins should be preabsorbed. This clearly
was the case, and spurious signals were totally eliminated
in the nine group I sera that were negative when incubated with FCS-depleted washed cells. This observation
was extended to a set of 30 patient samples possessing
IgG specific for bovine (but not human) ␤2GPI, associated
in 18 cases with IgG to FCS proteins (see below). Finally,
the simple subtraction of the binding to FCS-coated wells
(sample blank) from the binding to EA.hy 926-coated
wells was inappropriate (at least in the absence of FCS in
the sample dilution buffer) in view of the higher values of
the former. Consequently, this practice would overcorrect
specific antibody estimates in samples in which heteroantibodies coexist with genuine AECAs. Three such cases
were identified in group I (group I subset), on the basis of
reactivity profiles, which were identical to group II sera
when FCS was used as sample diluent. Thus, the dilution
of serum samples in PBS containing 100 mL/L FCS, along
with the use of control wells coated with 100 mL/L FCS
medium alone, were eventually chosen as the most convenient means of measuring AECAs without interference
by heterophile antibodies.
properties of heterophile antibodies
Fig. 1. Influence of the mode of fixation of the monolayer of EA.hy 926
cells on the ELISA responses.
Assayed samples were diluted 1:5000 for the MAb against thrombomodulin and
1:100 for sera from a healthy control and from one representative patient of
group I (apparent AECA activity) and group II (genuine AECA activity). The
means ⫾ SD of absorbance values from two independent experiments are
shown. Based on the reactivity with the anti-thrombomodulin MAb, the residual
cell densities were in the rank order: no fixative (;) ⬍⬍ 0.25 g/L glutaraldehyde
(䡺) ⬍ 1 g/L glutaraldehyde (u) or ethanol ( ).
Using a FCS-ELISA (the interference assay), designed to
detect involved heteroantibodies, we sought to further
characterize the heterophile antibodies in our samples. To
assess the species specificity of IgG antibodies against FCS
proteins, decreasing concentrations of a variety of competing animal sera were added to PBS-Tween at the
antibody-binding step (Fig. 3). As expected, the heteroantibody interference was quantitatively inhibited by the
addition of FCS or adult bovine serum, with 50% inhibi-
276
Revelen et al.: Heterophile Antibody Interference in a Cyto-ELISA
Table 1. FCS proteins are responsible for IgG antibody binding from group I sera in the ELISA for AECAs.
Solid-phase treatments
Step 1:
EA.hy 926 in
FCS mediuma
Step 2:
buffer (purpose)
Yes
Yes
Yes
No
No
Ethanol (fixation)
Trypsin/EDTA (cell detachment)
5 mL/L Triton X-100 (ECM exposure)
100 mL/L FCS (protein coating)b
5 g/L BSA (protein coating)b
a
b
Absorbance values of assayed sera
Healthy control
Group I (n ⴝ 12),
mean ⴞ SD
(no. of positives)a
Group II (n ⴝ 6),
mean ⴞ SD a
(no. of positives)
0.030
0.072
0.068
0.056
0.043
1.120 ⫾ 0.362 (12)
0.565 ⫾ 0.124 (11)
1.713 ⫾ 0.450 (12)
1.644 ⫾ 0.503 (12)
0.590 ⫾ 0.538 (8)
0.636 ⫾ 0.224 (6)
0.047 ⫾ 0.009 (0)
0.089 ⫾ 0.031 (0)
0.065 ⫾ 0.017 (0)
0.025 ⫾ 0.022 (0)
To calculate the number of positive sera, the cutoff was set at 0.200 in these experiments.
In PBS buffer.
tion of binding reached at serum concentrations as low as
0.002– 0.1%. These antibodies exhibited a broad specificity, recognizing serum proteins from goat, sheep, and to a
lesser extent horse, although they bound less efficiently
than their bovine counterparts. Only slight inhibition was
achieved in most cases when concentrations of serum of
human, pig, rabbit, and rat origin up to 50% were used.
IgG antibodies against bovine serum proteins, including BSA and ␤2GPI, appear to be a common finding in
healthy blood donors, the prevalence being ⬃13–18%
when direct ELISAs are used (Table 2). In addition, these
types of antibodies often coexist in patient populations,
and the frequencies observed in 30 patients selected for
the presence of species-specific IgG to bovine ␤2GPI, as
well as in group I patients, largely exceeded the ones in
controls. However, anti-BSA and anti-bovine ␤2GPI antibodies contributed only marginally to the whole anti-FCS
reactivity in double- or triple-positive sera, because fluidphase BSA or ␤2GPI caused poor inhibition in the FCSELISA, contrasting with total blockade in the homologous
assays. Among the patients with anti-FCS antibodies
Fig. 2. IgG binding in various ELISA formats
for AECAs, comparing sample dilution in PBS
containing 10 g/L BSA (A) or PBS containing
100 ml/L FCS (B) and three types of coating.
Wells were incubated for 6 h with medium containing 100 mL/L FCS (sample blank; ;), or with EA.hy
926 cells (5 ⫻ 104 cells/well) resuspended, after
being washed with PBS, in FCS-depleted culture
medium (b) or medium containing 100 mL/L FCS
( ). The plates were then washed twice in PBS and
fixed with ethanol before probing with control and
patient serum samples. A representative experiment is shown (mean ⫾ SD of absorbance values)
that included three sera in each category, i.e., group
I (heteroantibodies only; I), group II (genuine AECAs
only; II), and group I subset (genuine AECAs associated with heteroantibodies; I subset).
shown in Table 2, the most prevalent isotype was IgG,
alone in 14 cases, or associated with IgA, IgM, and IgA ⫹
IgM in 12, 2, and 2 cases, respectively. Two patients
possessed only IgM.
We also performed immunoblotting of FCS, using the
12 group I sera as the overlaying IgG antibodies. Seven of
the sera were reactive to a limited number of proteic
bands often shared by the sera, in particular at ⬎250, 220,
160, 140 and 95 kDa. No attempt was made to identify the
bands. There was no band detectable in the 66- and
50-kDa regions, which correspond to BSA and ␤2GPI,
respectively.
Discussion
The present study confirms that antibodies (mostly IgG
and IgA) against animal serum proteins are widely distributed in human serum (10 –12, 16 ), and it establishes
their potential to give falsely increased AECA results, as
detected by a cyto-ELISA. During unsuccessful attempts
to set up an ELISA with unfixed endothelial cells, we
incidentally noticed strong ELISA responses with some
277
Clinical Chemistry 46, No. 2, 2000
Fig. 3. Comparative efficiency of FCS and
serum from various species to compete with
patient IgG binding to FCS-coated ELISA
plates when used as sample diluent.
ELISA plates were coated with 100 mL/L FCS. The
concentrations of animal sera responsible for 50%
inhibition are shown for the 12 patients from group
I (each symbol corresponds to an individual patient). Animal serum concentrations ⬎⬎50% indicate marginal inhibition.
sera despite very low residual cell density, suggesting that
the antibodies were not cell specific. This is reminiscent of
the report by Westphal et al. (5 ) of sera that reacted
strongly with the gelatin used to encourage endothelial
cell adhesion, both in the healthy control and in the
patient sera. These investigators noted that the signal in
control wells sometimes overcame the test values in the
presence of cells, in keeping with the behavior of samples
from 12 of our patients (referred to as group I), in which
IgG reacted to FCS from culture medium (Figs. 1 and 2).
These observations stress the usefulness of running appropriate sample blank wells, the absorbance values of
which are usually subtracted from those of the samples.
However, the fact that the binding of group I sera was
much higher in wells coated with FCS than in cell-seeded
wells (Fig. 2A) casts doubt on the value of sample blank
subtraction in these cases, with a risk of missing genuine
AECAs that happen to coexist with heteroantibodies (e.g.,
group I subset). The approach that we selected for both
circumventing the problem of false positives and properly
detecting associated genuine AECAs is to include 100
mL/L FCS in the sample dilution buffer (Fig. 2B). This
ability of FCS or adult bovine serum to inhibit, in a
concentration-dependent manner, spurious signals from
all assayed samples confirms that antibodies against bovine serum proteins were responsible for the interference.
In this respect, it is intriguing that Craig et al. (17 )
observed blocking-specific background in immunoblot
assays that was attributable to a reaction between endogenous IgG and the membrane blocking agent (i.e., milk or
BSA), although sera were diluted in blocking buffer;
reactivity against bound blocking proteins should have
been absorbed. Interestingly, Mauracher et al. (18 ) had to
use heat-denatured blocking proteins as the sample diluent to best reduce blocking-specific background in an
ELISA. Alternatively, washing endothelial cells free of
FCS before plating for ELISA also blocks heteroantibody
interference, although it is more tedious than the above
method. It thus is likely that none of the FCS proteins
targeted by our series of heteroantibodies interacted with
Table 2. Prevalence of IgG antibodies to crude FCS, BSA, and bovine ␤2GPI, as determined by direct ELISAs in control and
patient groups.
Number (%) of positive results on solid-phasea
Population (n)
100 mL/L FCS
5 g/L BSA
10 mg/L bovine ␤2GPI
Group I (12)
Patients with anti-bovine ␤2GPI IgG (30)
Healthy controls (45)
12 (100%)
18 (60%)b
8 (18%)
8 (67%)b
7 (23%)c
6 (13%)
9 (75%)b
30 (100%)
7 (15%)
a
b
c
The threshold for positivity was set at 2 SD above the mean of 45 control sera in the three ELISAs.
P ⬍0.001 vs the corresponding frequencies in controls.
Difference in frequency vs controls not significant.
278
Revelen et al.: Heterophile Antibody Interference in a Cyto-ELISA
cell membranes strongly enough to resist in vitro washes
in PBS.
There has been recent emphasis on human anti-animal
immunoglobulins, particularly human anti-mouse antibodies, as a cause of both positive and negative interferences in two-site immunoassays based on murine MAbs,
in view of the increasing clinical application of the latter
reagents for targeted imaging and immunotherapy. In
most cases, so-called heterophile antibodies are weak,
polyreactive antibodies produced in the absence of obvious exposure to animal proteins in a social or iatrogenic
setting (9, 10 ). This raises the possibility that they arise
from the natural process of antibody diversity (9 ). Of note
is that heterophile antibodies, identified because of interference in interleukin-4 measurements by a two-site
ELISA, have been reported to be under genetic control
associated with self-tolerance and resistance to progression of type I diabetes (19 ). Alternatively, a major route by
which animal protein antigens may be presented to the
immune system and trigger antibody formation is the
transfer of dietary antigens across the gut wall (10, 13 ). In
keeping with this explanation is the increased incidence of
antibodies to a variety of food antigens, including BSA
and proteins from cow’s milk, which is higher in children
than among older individuals or in association with
pathological conditions such as celiac disease and selective IgA deficiency (16, 20 ). In our group I patients,
favorable differential inhibition by FCS or adult bovine
serum (Fig. 3) points to bovine proteins as the main
immunogen, although clearly there was marked crossreactivity with serum proteins from other animal species
such as goat, sheep, and horse. Also in line with an
antigen-driven response is the observation that antibodies
to several bovine serum proteins frequently coexist in
human sera (Table 2).
One puzzling fact from the literature dealing with
AECAs is the variability of their prevalence in various
diseases, and the basis for most of these discrepancies is
presumably methodologic (1, 6, 7 ). We demonstrated
here that endogenous antibodies to animal serum proteins
frequently can interfere in any system that uses cultured
cells as the solid-phase binder but that these antibodies
are easily neutralized through assay redesign. This
knowledge is of practical importance to achieve better
interlaboratory agreement in the measurement of AECAs
by ELISA.
Ronan Révélen is a recipient of a scholarship from the
Communauté Urbaine de Brest, which is gratefully acknowledged. We thank C.S. Edgell, University of North
Carolina, Chapel Hill, NC for providing us with the
EA.hy 926 cell line.
References
1. Belizna C, Cohen Tervaert JW. Specificity, pathogenicity, and
clinical value of antiendothelial cell antibodies. Semin Arthritis
Rheum 1997;27:98 –109.
2. Bordron A, Dueymes M, Levy Y, Jamin J, Leroy JP, Piette JC, et al.
The binding of some human antiendothelial cell antibodies induces endothelial cell apoptosis. J Clin Investig 1998;101:2029 –
35.
3. Heurkens AHM, Gorter A, de Vreede TM, Edgell CS, Breedveld FC,
Daha MR. Methods for the detection of anti-endothelial antibodies
by enzyme-linked immunosorbent assay. J Immunol Methods
1991;141:33–9.
4. Meyer O, Kaiser P, Haim T, Edgell CJ, Pasquier C, de Bandt M, et
al. Anti-vascular endothelial cell antibodies: comparison of two
assay methods and clinical applications. Rev Rheum 1995;62:
737– 47.
5. Westphal JR, Boerboms AMTH, Schalkwijk CJM, Kwast H, de
Weijert M, Jacobs C, et al. Anti-endothelial cell antibodies in sera
of patients with autoimmune diseases: comparison between
ELISA and FACS analysis. Clin Exp Immunol 1994;96:444 –9.
6. Meroni PL, Khamashta MA, Youinou P, Shoenfeld Y. Mosaic of
anti-endothelial antibodies. Review of the first international workshop on anti-endothelial antibodies: clinical and pathological
significance. Lupus 1995;4:95–9.
7. Youinou P, Meroni PL, Khamashta MA, Shoenfeld Y. A need for
standardization of the anti-endothelial-cell antibody test. Immunol
Today 1995;16:363– 4.
8. D’Cruz DP, Keser G, Direskeneli H, Khamashta MA, Hughes GRV.
Anti-endothelial cell antibodies in systemic vasculitis and systemic lupus erythematous: effects of heat inactivation on binding
and specificity. Clin Exp Immunol 1999;115:567–70.
9. Kaplan IR, Levinson SS. When is a heterophile antibody not a
heterophile antibody? When it is an antibody against a specific
immunogen. Clin Chem 1999;45:616 – 8.
10. Kricka LJ. Human anti-animal antibody interferences in immunological assays. Clin Chem 1999;45:942–56.
11. Boscato LM, Stuart MC. Heterophilic antibodies: a problem for all
immunoassays. Clin Chem 1988;34:27–33.
12. Howanitz PJ, Howanitz JH, Lamberson HV, Ennis KM. Incidence
and mechanism of spurious increases in serum thyrotropin. Clin
Chem 1982;28:427–31.
13. Hunter WM, Budd PS. Circulating antibodies to ovine and bovine
immunoglobulin in healthy subjects: a hazard for immunoassays.
Lancet 1980;ii:1136.
14. Willman JH, Martins TB, Jaskowski TD, Hill HR, Litwin CM.
Heterophile antibodies to bovine and caprine proteins causing
false-positive human immunodeficiency virus type 1 and other
enzyme-linked immunosorbent assay results. Clin Diagn Lab
Immunol 1999;6:615– 6.
15. Arvieux J, Regnault V, Hachulla E, Darnige, L, Roussel B, Bensa
JC. Heterogeneity and immunochemical properties of anti-␤2glycoprotein I autoantibodies. Thromb Haemost 1998;80:393– 8.
16. Ammann AJ, Hong R. Anti-antiserum antibody as a cause of double
precipitin rings in immunoglobulin quantitation and its relation to
milk precipitins. J Immunol 1971;106:567–9.
17. Craig WY, Poulin SE, Collins MF, Ledue TB, Ritchie RF. Background
staining in immunoblot assays. Reduction of signal caused by
cross-reactivity with blocking agents. J Immunol Methods 1993;
158:67–76.
18. Mauracher CA, Mitchell LA, Tingle AJ. Reduction of rubella ELISA
background using heat-denatured sample buffer. J Immunol Methods 1991;145:251– 4.
19. She JX, Ellis TM, Wilson SB, Wasserfall CH, Marron M, Reimsneider S, et al. Heterophile antibodies segregate in families and are
associated with protection from type 1 diabetes. Proc Natl Acad
Sci U S A 1999;96:8116 –9.
20. Rothberg RM, Farr RS. Anti-bovine serum albumin and anti-alpha
lactalbumin in the serum of children and adults. Pediatrics
1965;35:571– 88.