Clinical Allergy, 1976, Volume 6, pages 91-98
Comparison of IgE values as determined by
different solid phase radioimmunoassay methods
S. G. O. JOHANSSON,* ASTA BERGLUNDf and
N.-I. M. KJELLMANI
* The Blood Centre, University Hospital, and
t Re.search and Development, Pharmacia Diagnostics AB, Uppsala
X Department of Paediatrics, Linkopitig University, Linkoping
Summary
Accurate measurement of low IgE concentrations is technically difficult. In this paper
results obtained by a direct sandwich and three inhibition methods of radioimmuno-
assays are compared. For values above 50 U/ml good correlation was obtained with
all methods. Below 50 U/ml, however, the inhibition methods tended to yield falsely
high values. For very low concentrations, l-lO U/ml the best correlation was obtained
between the direct sandwich test (PRIST*) and the inhibition test using a correction
factor to allow for the non-specific effect of serum.
The four methods were used to quantify IgE in cord serum samples from healthy
individuals. The mean value obtained by PRIST was 0 4 U/ml and by the inhibition
test, using a correction factor, 0-6 U/ml respectively. Because of its greater simplicity
the direct sandwich test is recommended.
IntroductioD
The first report of raised levels of IgE, at that time provisionally designated IgND,
in atopic, extrinsic asthma but not in other types of asthma (Johansson, 1967) stimul-
ated an interest in the measurement of serum IgE in immediate hypersensitivity
disorders as well as other diseases with immunological engagements. To date well over
a hundred papers have confirmed the inilial findings and extended them to other atopic
diseases, parasitic infections and certain immunodeficiency states (Johansson, Bennich
&Berg, 1972).
It was recently found that a raised serum IgE level may be predictive of later
onset of extrinsic asthma in children with wheezy bronchitis (Foucard, 1974) and of
any type of atopic allergy in infancy (Hamburger et aL, 1974). To enable the accurate
measurement ofthe very low IgE levels of early childhood sophisticated technology
is necessary. In this paper two solutions are reported to the problems involved in
* Trade mark applied for by Pharmacia Diagnostics AB, Uppsala.
Correspondence: Dr N.-I. M. Kjellman, Department of Paediatrics, Regionsjukhuset, S-58! 85
Linkoping, Sweden.
91
92 5". G. O. Johansson, Asta Berglund and N-I. M. Kjellman
quantification of low serum IgE concentrations. The results obtained by two different
radioimmunoassays (RIA) in various modifications are compared.
Materials and methods
At the department of paediatrics, Linkoping University, Sweden, 228 serum samples
were collected from children aged 0 weeks (cord serum) to 15 years (for details regard-
ing age and sex distribution see Kjellman, Johansson & Roth, 1976). Venous blood
was collected and serum was separated after 1-2 hr at room temperature and stored at
— 20°C until analysed. The cord blood was stored in a refrigerator for about 1 day
before serum separation and then kept deep frozen until analyses were performed.
IgE determination
The IgE determinations on the serum samples were performed at the Blood Centre,
University Hospital, Uppsala, Sweden. The serum IgE concentrations were measured
in duplicate in two independent experiments by means of two radioimmunoassay
solid phase techniques and the results expressed in U/ml serum (1 U/ml= IkU/ml).
Inhibition technique (RIST). Cord blood serum and sera from children less than 1 year
of age were diluted 1:5 in buffer before assay. The remaining serum samples were
diluted 1:10.
Direct competitive method. The Phadebas® IgE Test kit (Pharmacia, Uppsala)
was used for all serum samples collected. This radioimmunoassay method depends on
the ability of an antibody to bind its antigen which is labelled with a radioactive isotope
and the competitive inhibition of this reaction by the unlabelled antigen. By using
antibodies coupled to a solid phase the antibody 7-bound labelled and unlabelled antigen
can easily be separated from the unbound antigen.
The assays were set up according to the test procedure recommended by the
manufacturer except that two additional IgE standard dilutions were prepared for
the standard curve: 0 1 and 0 5 U/ml. IgE values were calculated according to instruc-
tions for the test.
Direct competitive method with modified calculations. In order to correct for possible
serum interference in the test system where buffer was used as a diluent, the IgE
levels were also calculated by means of a correction factor established in the following
manner. The IgE standard ofthe Phadebas IgE kit was prepared in buffer according to
the instructions in the manual and in IgE-free serum diluted five and ten times in
buffer. The IgE-free serum was prepared by exposing human serum, containing less
than 1 U IgE/ml to a temperature of 56°C for 4 hr. The assay was set up according to
the recommended test procedure. To detect serum interference more easily, standard
curves were constructed by plotting the bound activity as count rates per minute
(c.p.m.) against the concentration ofthe IgE standard. By using IgE standards pre-
pared in IgE-free serum the standard curve was shifted to the left. The results varied
when different serum concentrations were used. The standard curve obtained by using
IgE-free serum ofthe same concentration, both as diluent and zero sample, was found
superimposable on a standard curve obtained by using buffer as a diluent. Both curves
were constructed by plotting IgE concentrations v. the count rate for the standards as a
percentage ofthe zero dose bound activity. Consequently, buffer can be used as diluent
for the standards if the bound activity ofthe unknown sample is expressed as a per-
centage of the bound activity at zero dose in IgE-free serum rather than in buffer, as
Comparison of fgE values 93
used for the standard curve. An investigation was made to find the ratio between the
bound activity of zero samples consisting of IgE-free serum and buffer. Human
serum samples from twenty patients with IgE values less than 10 U IgE/ml were
exposed to 56°Cfor 4hr.The twenty'IgE-free'serum samples were diluted five and ten-
fold with buffer and were tested as zero samples in the IgE test. The ratio between the
c.p.m. of the zero samples consisting of an IgE-free serum and a buff'er respectively
was found to be 0-96 at a serum dilution of I :IO and 0 91 at a serum dilution of 1:5.
The unknown concentrations of IgE in serum samples (diluted 1/5 or 1/10), were
read off from the standard curve using the bound activities of the samples expressed
as percentage of the corrected zero dose bound activity (i.e. 0 91 or 0 96 times the
buffer zero dose bound activity).
Sequential addition method was performed in the following way on ail but one of
the serum samples collected. The Sephadex-coupled anli-lgE from the Phadebas
IgE Test kit was incubated at room temperature overnight with the patient sera.
'^•^I-labelled IgE, diluted according to instructions for the kit, was then added. The
test tubes were subjected to a second incubation for 3-4 hours after which the lubes
were washed five times with saline containing 0-1% Tween 20 before the final
centrifugation and counting of the radioactivity bound to the Sephadex-anti-lgE
complex was carried out. A standard curve was established in the same way as in the
methods described above.
Sandwich technique.
Paper disc RIST (PRIST). The method described by Ceska & Lundkvist (1972)
using anti-lgE antibody immobilized on paper discs as the solid phase was applied for
total lgEdeterminationsin226of the serum samples. The serum samples from children
up to 4 years of age were generally tested undiluted while the remaining sera were
tested diluted 1:5 in buffer. This system is a direct radioimmunoassay where the IgE
to be assayed is first bound to the solid phase coupled antibody and after washing it is
incubated with immunosorbent purified antibodies labelled with a radioactive
isotope. The assay was performed in polystyrene test tubes. One paper disc to which
anti-lgE was coupled was used per test tube. Fifty microlitres of serum or standard
was added and incubated for 3 hr at room temperature, and then the discs were washed
three times in saline with Tween 20. One hundred microlitres '^^l-anti-IgE representing
approximately 100,000 c.p.m. was added and a second incubation overnight followed.
Finally, the tubes were again washed and the radioactivity bound to the solid phase
was determined in a gammacounter. Buffer and IgE standard used in this method were
the same as that in the Phadebas IgE Test kit. The standard curve was prepared by
means of the same standard dilutions as described above.
A certain serum effect was found with PRIST in the determination of IgE in sera
diluted less than 1:5. This has to be compensated for by diluting the standards in an
IgE-free serum.
Results
Radioimmunoassays based on two different principles have been used in this study.
The results of an inhibition test, represented by the radioimmunosorbent test (RIST)
and a direct sandwich test, the paper disc RIST (PRIST) have been compared (Fig. 1).
The correlation is good for IgE concentrations above 50-60 U/ml but for values below
that level there is an increasing tendency for RIST to give higher values than PRIST.
94 S. G. O. Johansson, Asia Berglund and N.-l. M. Kjellman
lOOOcr
100
'0
I I I I 11 m l I I I I I ml I I I I [ ml i i i 11 ii i
0-1 10 too 1000
Phadebos IgE test
IgE U/m;
Fig. 1. Correlation between serum IgE values assayed with PRIST and the Phadebas IgE test.
Thus no RIST value less than 1 U/ml was found, but as many as 22% of the PRIST
values were tbat low.
The results of PRIST and the sequential addition moditication of RIST were
essentially similar in the low level region as were also results of PRIST and the original
RIST (Fig. 2). Tbe former good correlation seems to extend down to somewhat lower
concentrations tban in tbe case of tbe original RIST.
1000 R
01 -
10 100 1000
Sequentiol oddition method
IgE U/ml
Fig. 2. Correlation between serum IgE values assayed with PRIST and the sequential addition melhod.
Comparison of IgE values 95
1000
100
10
JL* / I t * *>8\* •
0-1 -
ml 1 1 1 1 It II
0-1 I 10 100 1000
Phadebos IgE fest modified calculotion
IgE U/ml
Fig. 3. Correlation between serum IgE values assayed with PRIST and the Phadebas IgE test with
modified calculation.
The modified calculation procedure for RIST compensates for the average, non-
specific influence of the serum milieu on the test system. When the results are compared
with those of PRIST, this correlation seems to be very good for values above 30 U/ml
(Fig. 3). Below this level an acceptable correlation is obtained but the scatter is con
siderable. Approximately the same number of values, 22%, were below 1 U/ml for
both methods. However, 5-8% of the samples for which a value below I U/ml was
obtained in one test were found to have a value higher than 1 U/ml in the other test.
Table I. Comparisons by means of paired observa-
tions using Student's /-test of IgE levels in twenty-six
cord blood sera obtained with dilTerent test methods
Test methods X y P
U/ml U/mi
x=Phadebas IgE Test
(modified calculation) 0-64 3-79 <0-00I
y = Phadebas IgE Test
x = Phadebas IgE Test 379 039 < 0-001
y = PRIST
x = Sequential method 311 039 < 0-001
y = PRIST
X = Phadebas IgE Test
(modified calculation) 0-64 039 <005
y = PRIST
96 S. G. O. Johansson, Asta Berglund and N.-f. M. Kjellman
The mean IgE values found in cord serum using the four RIA methods (Table 1)
illustrate the effect of the serum factor on the results obtained.
Discussion
An inhibition RIA is based on the following principle: the non-labelled antigen
(in our case IgE) in the sample or in the reference standard is quantified from its
capacity to inhibit the reaction between the isolated, isotope-labelled antigen (IgE)
and the specific antibody (anti-lgE). Separation of bound and non-bound tracer
(labelled-lgE) can be accomplished by precipitation by a second antibody directed
against the Fc part of the anti-lgE followed by washing of the precipitate, the so-called
double-antibody technique (Gleich, Averbeck & Swedlund, 1971). The separation can
also be obtained by binding the first antibody (anti-lgE) to a solid phase, like the
inner surface of a disposable polystyrene test tube (Catt & Tregear, 1967) or Sephadex
particles (Wide & Porath, 1966). The latter system is usually designed so that the anti-
body can bind only 10-20% of added, labelled antigen, while for double antibody
methods this figure is often considerably higher. This non-saturation of the system
is necessary to ensure high sensitivity of an inhibition test. Any substance that interferes
non-specifically with the reaction of IgE with anti-lgE will decrease the radioactivity
bound to the particles and will thus be interpreted as IgE. The higher protein con-
centration of the serum sample compared to the standard diluted in buffer exerts such
a non-specific effect.
Antibodies to the gamma globulin of the same animal origin as the anti-lgE
used can, when bound to the antiserum, sterically interfere with the reaction between
IgE and anti-lgE. Such antibodies with specificity for bovine gamma globulin occur
in approximately 70% of the population (Foucard et al., 1975). Due to cross reaction
with sheep and goat gamma globulin, anti-lgE from these animals cannot be recom-
mended in inhibition types of RIA.
The direct sandwich RIA (Wide, 1971), PRIST, should theoretically sufTer from
the same non-specific factors as inhibition techniques. The decreased amount of IgE
bound to the solid phase anti-lgE which will lead to a decreased amount of radio-
activity on the disc will in this case be interpreted as a falsely low IgE value. Experi-
mental studies indicate that the influence of such activity on the IgE values is not so
pronounced. However, a certain serum effect cannot be excluded when sera are tested
undiluted and the standard is diluted in buffer.
The reason why PRIST is less influenced by non-specific effects than RIST is
probably two-fold. The washing step performed before the addition of the labelled
reagent will probably eliminate some serum factors. In addition, an excess of anti-lgE
is bound to each disc and therefore much more animal gamma globulin is added to
each test sample. This will make it less likely for antibodies to bovine gamma globulin
to sterically interfere with the binding of IgE to anti-lgE.
The sequential modification of inhibition techniques has been reported to increase
the sensitivity of the assay (Wide, 1971). However, in this study advantage was not
taken of the somewhat higher sensitivity of the sequence test. The serum samples were
tested at the same dilution as in the direct inhibition test leading to comparable serum
effect in both methods. This probably explains why the mean IgE value for cord serum
was found to be as high as 3-1 U/ml compared to 3-8 U/ml for the direct inhibition
test.
It has been suggested that solid phase techniques should be more easily influenced
Comparison of IgE values 97
by non-specific factors than would a double-antibody method. This effect seems to be
especially prominent when the solid phase consists of Sepharose or BAC-cellulose
(Polmar, Waldmann & Terry, 1972) or of carbon-particles (McLaughlan e/o/., 1974).
Under certain conditions also the double-antibody method seems to give falsely high
levels as judged from the mean of 39ng/ml (approximately 16 U/ml) reported for cord
serum by Stevenson ct al. (1971). Our very first investigations into serum IgE levels in
1967 using RIST also seemed, in retrospect, to yield high values. However, the serum
eflect on solid phase inhibition techniques is usually of little clinical importance
unless low levels are to be studied. In the present study the direct inhibition technique
gave a figure for cord serum of 3-8 U/ml which is comparatively low.
When the serum effect was compensated for by the correction factor a significantly
lower mean value was obtained and PRIST gave even lower levels. In fact, with PRIST,
no IgE (less than 0-1 U/ml) was detected in nine out of twenty-six tested cord serum
samples.
The double-antibody method seems sufficiently sensitive to detect IgE at concentra-
tions below 10 U/ml and even below 1 U/ml (Gleich et al.. 1971). To reach such a
high sensitivity with acceptable precision a prolonged incubation time, up to 96 hr,
is necessary and the operation range is narrow (Nye et aL., 1975). Even under optimal
conditions this method tends to give higher values in the low level region than PRIST
(Lundkvist & Merrett, personal communication).
Several approaches have been tried to overcome the difficulties involved when
low levels of IgE are to be assayed by inhibition techniques. Bazaral and co-workers
suggested that the non-specific effect of each serum sample should be measured and
subtracted from the results obtained (Bazaral, Orgel & Hamburger, 1971). However,
this is a rather laborious procedure. It is possible to dilute the standard in an IgE-free
serum, but large quantities of such sera are very difficult to obtain. The compensating
factor suggested here, which takes into account the average serum factor effect of
various dilutions of the tested serum samples is an attractive solution. However,
the best solution would probably be to use a direct sandwich-type of RIA like the
PRIST. Thismethodisalsoa technically very simple method. A test procedure based on
paper discs excludes many steps in laboratory routine. Stoppering of test tubes,
rotation or shaking during incubation and centrifugation are not necessary.
Acknowledgments
This work was supported in part by the Swedish Medical Research Council (Grant No.
I6X-105) and Semper Fund for Nutritional Research.
References
BAZAKAL. M., ORGEI., A . & HAMBURGER. R.N. (1971) IgE levels in normal infants and mothers and
:in inherilance hypothesis. Journal of Immu/iotogy, 107, 794.
CATT, K. & TREGEAR, G.W. (1967) Solid-phase radioimmunoassay in antibody-coaled tubes. Science,
158, 1570.
CESKA, M . & LUNDKVIST, U . (1972) A new and simple radioimmunoassay method for the determina-
tion of IgE. Immmwchemistry, 9, 1021.
FOUCARD, T.. BENNICH, H., JOHANSSON, S.G.O. & LUNDKVIST, U . (1975) Human antibodies to bovine
gamma globulin. Occurrence in immunological disorders and influence on allergy radioimmuno-
assays. tiitirnarUmat Archives of Aiiergy and AppUetl tmmmioiogy. 48, 812.
FoucARp, T. (1974) A follow-up study of children with asthmatoid bronchitis. II. Serum IgE and
eosinophilic counts in relation to clinical course. Actapaediatrka Scandinavica, 63, 129.
98 S. G. O. Johansson, Asta Berglund and N.-I. M. Kjellman
GLEICH, G.J., AVERBECK, A.K., SWEDLUND, H . A . (1971) Measurement of IgE in normal and allergic
serum by radioimmunoassay. Journal of Laboratory and Clinical Medicine 77, 690.
HAMBURGER, R.N., LENOIR, M . , GROSHONG, T.E., MILLER, J.R., WALLACE, W . M . & ORGEL, H.A.
(1974) Development of IgE and allergy during the first year oflife: Preliminary data. Journal of
Allergy and Clinical Immunology, 53, 94.
JOHANSSON, S.G.O. (1967) Raised levels of a new immunoglobulin class (IgND) in asthma. Lancet,
ii,951.
JOHANSSON, S.G.O., BENNICH, H . & BERG, T. (1972) The clinical significance of IgE. In: Progress
in Clinical Imimmology, Volume I (Ed. by R. S. Schwartz), p. 157. Grune & Stratton Inc., New
York and London.
KJELLMAN, N.-I.M., JOHANSSON, S.G.O. & ROTH, A. (1976) Serum IgE levels in healthy children
quantified by a sandwich technique (PRIST). Clinical Allergy, 6, 9.
MCLAUGHLAN, P., STANWORTH, D.R., WEBSTER, A . D . H . & ASHERSON, G . L . (1974) Serum IgE in
immune deficiency disorders. Clinical and E.vpc'riniental Inmiunology, 16, 375.
NYE, L., MERRETT, T.G., LANDON, J. & WHITE, R.J. (1975) A detailed investigation of circulating IgE
levels in a normal population. Clinical Allergy, S, 13.
PoLMAR. S.H., WALDMANN, T.A. & TERRY, W . D . (1972) IgE in immunodeficiency. American Journal
of Pathology, 69:3, 499,
STEVENSON, D.D., ORGEL, H.A., HAMBURGER, R . N . & REID, R . T . (1971) Development of IgE in
newborn human infants. Journal of Allergy and Clinical Immunology, 48, 61.
WIDE, L. & PORATH, J. (1966) Radioimmunoassay of proteins with the use of Sephadex-coupled
antibodies. BiocheniUa el biophysica Ada, 130, 257.
WIDE, L. (1971) Solid phase antigen-antibody systems. In: Radioimmunoassay Methods (Ed. by
K. E. Kirkham and W. M. Hunter), p. 405. European Workshop, E&S Livingstone Ltd.,
Edinburgh and London.
