- Email: [email protected]
Comparison of procedures for measurement of IgE protein in serum and secretions
Comparison of procedures for measurement of IgE protein in serum and secretions Gerald J. Gleich,
M.D., and Sandra L. Dunnette,
B.A. Rochester, Minn.
Because of conflicting reports in the literature concerning the value of various procedures for measurement of IgE in serum and secretions, we compared four different methods, the radioimmunosorbent test (RIST), the double-antibody radioimmunoassay (RIA), the paper disc immunosorbent test (PRIST), and radial immunodt#usion (RID). The standards used in the assays were tested initially in the double-antibody RIA with the use of the reference from the World Health Organization. The results showed that RID as expected was relatively insensitive and IgE was reliably measured only above approximately 1,000 international units (IU). Moreover, certain sera containing low levels of IgE by the other procedures gave distinct precipitin zones and presumably falsely high levels of IgE protein. Thus RID may yield apparently erroneous results when used as a screening procedure for measurement of IgE levels. Among the other procedures PRIST and the double-antibody RIA showed the best agreement. With serum samples RIST yielded values for IgE in the low level range higher than those given by the PRIST and double-antibody RIA. With breast milk and colostrum, values of IgE between 120 and 690 nglml were found by RIST, whereas IgE was not detected by double-antibody RIA and PRIST. No evidence of an inhibitor of IgE was found in breast milk, so that the apparent elevation of IgE in breast milk by the RIST is likely false. These jndings confirm prior reports of spurious elevations of IgE with the RIST and indicate the usefulness of the PRIST and double-antibody RIA ,for the measurement of IgE in sera and secretions.
Since the discovery of IgE as a distinct protein, a variety of methods have been utilized to measure it in sera and secretions.‘-l2 These include two competitive binding radioimmunoassays, one using a solidphase antibody, the radioimmunosorbent test (RIST),‘~ 2 and the other a double-antibody radioimmunoassay (RIA).4 In addition, IgE has been measured by a noncompetitive radioimmunoassay, the paper disc immunosorbent test (PRIST),5 and by a noncompetitive galactosidase-immunosorbent test (GIST). l* Finally, IgE has been measured by single radial immunodiffusion (RID) lo as well as several varieties of RID with the use of radioactive reagents.3, 6 The RIST procedure is available commercially and has been utilized for measurement of IgE in many reports in the literature.*, i3-15 Polmar and associates9 reported that procedures similar to the RIST yielded -From the Allergic Diseases Research Laboratory, Mayo Clinic and Mayo Medical School, Rochester, Minn. 55901. Supported in part by Grant AI 11483 from the National Institute of Allergy and Infectious Diseases. Received for publication Nov. 29, 1976. Accepted for publication Feb. 8, 1977. Reprint requests to: Dr. G. .l. Gleich, Mayo Clinic, Rochester, Minn. 55901.
spuriously high levels of IgE in serum from patients with agammaglobulinemia, and more recently Underdown and associatesl6 have reported that measurement of IgE levels in breast milk by the RIST may yield erroneous results. Because of these reports as well as the claim by Johansson and associatesi that the double-antibody RIA yields higher values of IgE in the low level range than PRIST, we compared these procedures for the measurement of IgE in serum and secretions. MATERIALS AND METHODS Sera and secretions Sera were obtained from the laboratory of Dr. Nai Jiang of the Department of Laboratory Medicine of the Mayo Clinic and Mayo Foundation. Theseserahad beenanalyzed by the double-antibody RIA4 and were selected so as to provide a range of levels of IgE from elevated to barely detectable. In addition, 13 umbilical cord sera obtained at the time of parturition were acquired. Nasal washes were performed in 10 patients as described by Yunginger and Gleich17and parotid fluid was obtained from 6 subjects as described by Hansen and Gleich.‘* Breast milk was obtained from 6 patients and colostrum from 5 patients. The reference standard of the World Health Organization (WHO), 68/341, was kindly provided by Dr. W. D. Terry and was used to calibrate the standardsfor all of the IgE Vol. 59, No. 5, pp. 377-382
378
Gleich and Dunnette
J. ALLERGY
CLIN. IMMUNOL. MAY 1977
b O s B 2
65-
$
-0.500
-
-1.000
-
-1.500
-
2 3 B \ cl 8 2 ‘;;
.$
@
25-
2
-2.000 0 B
0
0.5
Log,,of
I .o
1.5
IgE, ng
FIG. 1. Comparison of standards used in the various procedures for measurement of IgE to the WHO reference by the double-antibody RIA. Symbols used are as follows: WHO, l -; RIA, o----o; RID, A----A; RIST, O- --o; and PRIST, x- - -x. A, The inhibition curves. 9, The logit-log regression lines. The correlation coefficients of the regression lines in B were all greater than +0.99.
Because this reference serum contains the HbB antigen, precautions were taken to prevent personal contact with it during these initial experiments. In our institution personnel routinely handling human sera are checked for the presence of the HbB antigen in sera. In the past year no one in our laboratory has become ill with symptoms suggestive of hepatitis nor has the HbB antigen been detected in sera by RIA.
assays.
Methods
for the measurement
of IgE
Double-antibody RIA. This was performed as described by Gleich and associates4 with the exception that the second incubation was shortened to 2 hr at 23” C resulting in a 24-hr procedure. Rabbit antibody to IgE was used in this procedure. RIST. This procedure was performed as recommended by the Pharmacia Company, Piscataway, N. J. The RIST is a competitive binding RIA in which rabbit anti-IgE molecules are covalently bound to Sephadex particles. IgE in the sample competes with a fixed amount of radiolabeled IgE for a limited number of binding sites on the anti-IgE-Sephadex. After rotation overnight at 23” C and 3 washings with 0.9% saline, radioactivity bound to the Sephadex is counted. The radioactivity associated with the solid-phase anti-IgE is inversely related to the amount of IgE in the patient’s serum. The procedure takes 24 hr. As directed by Pharmacia, values for sera that were diluted lo-fold were divided by 0.96, while sera diluted 5-fold were divided by 0.91. Sera with
PRIST. This procedure, which is a noncompetitive RIA,5 was performed as recommended by the Pharmacia Company. Briefly, sheep anti-IgE covalently coupled to a paper disc binds IgE in the test serum. After incubation for 3 hr at 23” C the discs are washed thrice with 0.9% saline and radiolabeled anti-IgE is added. After a second incubation for 18 hr during which radiolabeled anti-IgE reacts with IgE, the disc is washed and the radioactivity measured in a gamma counter. The amount of radioactivity is directly related to the amount of IgE present in the sample. The procedure takes approximately 24 hr. We diluted sera having greater than 50 IU/ml so as to fall on the midsection of the standard curve, that is, 10 to 50 III/ml. RID. IgE was also measured by radial immunodiffusion (RID)‘O, ig with the use of kits obtained from the Meloy Company, Springfield, Va., and we followed the detailed procedure described by the company. Briefly, the sample is added to a well in an agarose gel containing goat antiserum to IgE. IgE diffuses out of the well during a 48-hr incubation period and reacts with antibody to form a precipitin ring. The diameter of the ring is proportional to the concentration of the IgE in the sample. The gel plates are washed, dried, and stained to visualize the IgE precipitin rings. The ring diameters were read by 2 individuals and the values averaged. The procedure takes a minimum of 3 days.
lmmunoelectrophoresis Certain sera were tested for the presence of antibodies to goat and sheep sera by immunoelectrophoresis. This procedure was performed as described previously by Macy and associates.20 Undiluted goat or sheep serum was added to
VOLUME NUMBER
Procedures for measurement of IgE 379
59 5
r=O
l * .
I’ 1
'
' """'
1
'
IgE,
FIG. 2. Comparison PMT.
of IgE protein levels by RIST and
methods
The results of analysis of IgE by the double-antibody RIA were analyzed by logit-log transformation of the data as described earlier for measurement of ragweed antigen E by Yunginger and Gleich. 2L The inhibition curves were logit-log transformed and the slopes given by the various inhibitors in the double-antibody RIA were tested by analysis of covariance with the use of a programmable Hewlett-Packard 98 IOA calculator.
' """'
'
"l"Ul
1000
10,000
I.U./ml, double antlbody RIA
10,000
r = 0.995
F
s
I100:
10 =
RESULTS
.
we compared the IgE standards used in the various assays to the IgE reference preparation supplied by the WHO. For this purpose the double-antibody RIA was performed with the use of the WHO reference preparation as a primary calibrating reagent and the quantity of IgE in all of the other standards measured by comparison. In Fig. 1 the inhibition produced by the various standards is shown in both curvilinear and linear form after logitlog transformation. Comparison of the inhibition curves in Fig. 1, A indicates that no major difference exists among the various IgE standards. Analysis of the slopes of the logit-log regression lines in Fig. 1, B indicates a difference among the slopes of the lines (F[4, I I] = 5.48; p < 0.025) and inspection suggests that the difference is likely due to the PRIST standard. Analysis of the slopes for RID, RIST, WHO, and the double-antibody RIA standards did not reveal a statistical difference (F[3,9] = 2.8 1; NS). Although there was a statistical difference among the standards in terms of their ability to inhibit the reaction of anti-IgE (ND) with IgE (PS) in the doubleIn initial
' 100
FIG. 3. Comparison of IgE protein levels by RIST and double-antibody RIA.
the wells, and patient’s serum, as the putative antiserum, was added to the trough after the completion of electrophoresis.
Statistical
' """'
10
experiments
I
/: 1
-
,
"""'
10
' ' """'
IgE,
100
' """"
I.U./ml,
FIG. 4. Comparison of IgE protein antibody RIA and by PRIST.
1000
' ""'Ul
10.000
PRIST
levels by double-
antibody RIA, the slopes of the regression lines were quite similar and therefore all of the standards were expressed in terms of the content of IgE in the WHO reference. The WHO reference contains 10,000 IU or 24,000 ng/ml.17, 22 The IgE levels in the sera are shown in Figs. 2 to 4 and in Table I. First, analysis of the results obtained by RID indicates that, above the manufacturer’s suggested lower limit of 900 IU, this procedure appears to correlate very well with the results of the radioimmunoassay procedures. However, in many instances rings were formed in the gel which evidently were spurious. For example, in the case of sample 34 the various radioimmunoassay procedures showed barely detectable levels of IgE, whereas by RID a level of 4,310 ng was obtained. In this case we assume that
380
Gleich and Dunnette
J. ALLERGY
TABLE I. IgE levels in serum (IUlml) Patient No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14
15 16 17
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39* 40 41 42 43 44 45 46 41 48 49 50 51
RID
RIST
PRIST
5,900 4,580 4,310 4,580 3,810 2,080 2,600 1,460 1,940 1,600 1,280 1,010 1,100 1,120 ~800 <800 <800 <800 4,710 1,150 ~800 <800 <800 <800 <800 2,750 a00 ‘a300 <800 5,430 <800 <800 <800 4,310 <800 <800 <800 <800 ~800 <800 <800 <800
5,260 4,680 4,000 4,650 3,850 2,160 2,330 1,260 1,600 1,340 911 655 846 444 249 170 248 132 219 170 67 59 49 68 45 2,020 47 48 37 18 5 26 2.4 28 <2 8 12 4 <2 <2 70 2 3 5 45 3 4 3 <2 2 2
7,670 4,815 3,960 5,685 4,560 ND 2,225 1,890 1,620 1,425 1,140 710 790 505 162 146 105 115 134 80 51 42 32 40 33 1,590 23 26 15 3 3.4 6.4 1.1 3.8 1.5 2.1
ND: not done. *Samples 39 to 51 were umbilical cord sera.
Table II. Levels in secretions Double-antibody RIA
8,290 6,100 5,200 5,708 5,000 2,421 2,460 1,515 1,820 1,110 925 630 641 433 151 128 70 94 106 65 54 32 28 37 22 1,881 20 17 14 5.8 4.7 5.3 1.7 3.6 ND 2.7 1.6
Secretion
RIST
(W/ml)
PRIST
Nasal wash 1 2 3 4 5 6 7 8 9 10
2.8 Cl.3 2.3 1.7 Cl.25 Cl.25 Cl.25 Cl.25 Cl.25 1.9
Breast milk 11 12 13 14 15 16 17 18 19 20 21
415* 500 335 400 335 690 630 295 120 690 135
CLIN. IMMUNOL. MAY 1977
2.2
Double-antibody RIA
7 2 Cl.4 <1.4 Cl.4 <1.4 Cl.4 <1.4 Cl.4 2 <8 <8 <8 4 <8 -4 <8 <4 <8 <4 <8
*Samples diluted Sfold.
the precipitin ring did not reflect the presence of authentic IgE. Discrepancies were also found with sera 19, 20, 30, and 47. Inspection of these sera (19, 20, 30, 34, and 47) revealed that they were not different in physical appearance from the other sera. Also, analysis of these sera by immunoelectrophoresis did not reveal the presence of antibodies reactive with goat or sheep serum. Thus, the RID procedure yields levels of IgE in certain sera which evidently are fallacious. Second, comparison of the results obtained by RIST with the results obtained by PRIST and double-antibody RIA revealed in most cases quite comparable values. The correlation coefficient for the agreement between the PRIST and the RIST was +0.988 and that between the RIST and the doubleantibody RIA was +0.983. Nonetheless, comparison of the results of RIST with those of PRIST revealed a difference by the paired t test (t = 4.80; p < 0.01). Similarly, there was a significant difference between the results obtained by RIST and by the doubleantibody RIA (t = 4.73; p < 0.01). Inspection of Figs. 2 and 3 reveals that in the lower range of IgE values, from approximately 1.6 to 40 III/ml, the results with RIST are higher than those with the PRIST and the double-antibody RIA. Finally, the best agreement among the procedures was found between the
VOLUME NUMBER
53 5
results with PRIST and the double-antibody RIA. As can be seen by inspection of the data in Table I and Fig. 4, there is extremely good agreement between the results from these two procedures; the correlation coefficient was +0.995. Moreover, no difference was found comparing the results by the paired t test (t = -0.46; NS). The results of analysis of secretions shown in Table II revealed a close agreement between those obtained in the double-antibody RIA and the PRIST. The only exception was the secretion from patient 1 in whom a clearly detectable level of IgE was found by the double-antibody RIA while none was detectable by PRIST. Most of the samples of nasal wash fluid did not contain IgE protein measurable by any of the procedures, and all of the 6 samples of parotid fluid were devoid of IgE protein (data not shown). In contrast, analysis of colostrum and breast milk by RIST revealed significant levels of IgE, whereas none was detectable by double-antibody RIA and PRIST.* The latter result could occur if an inhibitor of IgE were present in t.he breast milk and caused falsely low values for IgE in the double-antibody RIA and the PRIST assays. This possibility was tested by adding a known quantity of IgE to breast milk and to phosphate albumin buffer, respectively, and by determining whether there was a difference between the amounts of Ig,E detected in these mixtures. With both the double-antibody RIA and the PRIST, the quantities of IgE measured in the mixtures were essentially the same. Thus these experiments provided no evidence for the pmsence of an inhibitor of IgE in the breast milk. DISCUSSION In this study we measured IgE by various radioimmunoassay procedures as well as by RID and compared the results in serum and secretions. To determine whether the standards used in the various procedures were comparable, we initially measured the slopes of the transformed inhibition curves. The standards were serially diluted and tested for their ability to inhibit the reaction of specific antibody to IgE (ND) with IgE (PS)-islI. The results of these experiments are shown in Fig. 1 and indicate that all of the IgE standards yielded inhibition curves and inhibition lines after logit-log transformation quite similar to the WHO reference. Therefore the IgE values of each of the standards were expressed in IU. Comparison of the levels of serum IgE as shown in *Analysis of fresh cow’s milk by the double-antibody RIA, the RIST, and the PRIST did not reveal the presence of detectable IgE.
Procedures
for measurement
of IgE
381
Figs. 2 to 4 and Table I revealed that the results obtained by PRIST and by the double-antibody RIA agreed remarkably well. Similarly, both of these procedures gave good agreement with the RIST for values of IgE above 10 IU. However, below 10 IU, the agreement was less good as has been previously reported by Johansson and colleagues. l3 The RIST values had been corrected for serum interference as suggested by the manufacturer, but even after this correction discrepancies between the RIST procedure and the double-antibody RIA and PRIST persisted. Thus, we must conclude that there are certain unknown factors present in undiluted serum which can interfere with the measurement of IgE by RIST. Measurement of IgE by RID in the 12 patients with IgE levels greater than 900 IU yielded quite good correlations with the radioimmunoassay methods (r = 0.98 for RIST, r = 0.98 for double-antibody RIA, and r = 0.95 for PRIST). However, RID was flawed by the appearance of rings with certain sera which appeared to be spurious. These sera were not turbid and they did not contain antibodies reactive with goat or sheep serum when analyzed by immunoelectrophoresis. As seen in Table I these sera which contained relatively little IgE by the other procedures often gave detectable rings by RID. Hence, the use of RID as a screening method for elevated levels of IgE, at least in our hands, is flawed by the occurrence of these false positives. Our results, however, were obtained with a classical RIDiS without use of an intensification step (treatment with dopa)‘O and without pretreatment of sera with dextransulfate, which Wiedermann and colleaguesz3 claim eliminates nonspecific ring formation in the agar gel. Even without these additional steps, if the level of IgE is elevated as a result of screening by one of the other procedures, it would seem reasonable to further analyze the serum by RID to measure the absolute level of IgE. The RID procedure, however, takes approximately 3 days for completion. In contrast, in our laboratory we are now performing the doubleantibody RIA as an overnight procedure so that by testing a series of concentrations one could probably have the same data as one would obtain from RID without any loss of time. Also, both RIST and PRIST require only 24 hr for completion. Finally we analyzed only one lot of the RID reagents and one lot of RIST reagents so that we do not know whether the results found here would be encountered with other lots. Analysis of various secretions including nasal washes, parotid fluid, breast milk, and colostrum showed that low levels of IgE are present in these secretions as measured by PRIST and double-
382
Gleich and Dunnette
antibody RIA in agreement with an earlier report from our laboratory. 24In contrast, the RIST performed on undiluted as well as a 1: 5 dilution of breast secretion showed relatively high levels of IgE. When IgE was added to breast milk and analyzed by double-antibody RIA and by PRIST, the expected amount of IgE was found. This result discourages belief that an inhibitor of IgE was present in breast milk which resulted in falsely low values of IgE by double-antibody RIA and by PRIST. These findings are in agreement with the observation of Underdown and associates,16 who have shown that the RIST procedure yields falsely high levels of IgE in breast milk whey and that an inhibitor of the double-antibody RIA is not present in this secretion. Whether the problem of finding falsely high levels of IgE extends to other secretions is not known, but our findings suggest that the results of measurement of IgE in secretions by RIST should be confirmed by another procedure. Authors’
note added in proof
J. ALLERGY
7.
8.
9.
10.
11.
12.
13.
Additional analysesof sera numbered 15-23, 28-30, 32, 34, 36, 37,41, 44, and 0.1 ml of both undiluted of results with those antibody RIA indicated
45 were performed by RIST using and 1: 5 diluted sera. Comparison obtained by PRIST and doublethat the 1: 5 dilution showed better
14.
15.
agreement in 9 cases, the undiluted sera showed better agreement in 4 cases, while no difference was noted in 6 cases. In all analyses of sera with low levels of IgE, less than 30 IU/ml, RIST still yielded values higher than those given by PRIST and double-antibody RIA.
16.
We thank Mrs. Linda Peterson of the La Leche League of Minnesota, Ms. Delores Czerwinski, and Dr. ThomasB. Tomasi, Jr. for providing the samples of colostrum and breast milk, Dr. Nai Jiang for the sera containing known levels of IgE, and Pharmacia Laboratories, Piscataway, N. J., and Meloy Company, Springfield, Va., for gifts of their test kits.
18.
REFERENCES 1. Wide, L., and Porath, J.: Radioimmunoassay of proteins with the use of Sephadex-coupled antibodies, Biochem. Biophys. Acta 130:257, 1966. 2. Johansson, S. G. 0.: Raised levels of a new immunoglobulin class (IgND) in asthma, Lancet 2:951, 1967. 3. Rowe, D. S.: Radioactive single radial diffusion: A method for increasing the sensitivity of immunochemical quantification of proteins in agar gel, Bull. WHO 40:613, 1969. 4. Gleich, G. J., Averbeck, A. K., and Swedlund, H. A.: Measurement of IgE in normal and allergic serum by radioimmunoassay, J. Lab. Clin. Med. 77:690, 1971. 5. Ceska, M., and Lundkvist, U.: A new and simple radioimmunoassay method for the determination of IgE, Immunochemistry 9:1021, 1972. 6. Arbesman, C. E., Ito, K., Wypych, J. I., and Wither, K.:
17.
19.
20.
21,
22.
23.
24.
CLIN. IMMUNOL. MAY 1977
Measurement of serum IgE by a one-step single radial radiodiffusion method, J. ALLERGY CLIN. IMMUNOL. 49:72, 1972. Smith, H. J., Ozkaragoz, K., and Gokcen, M.: A simplified radioimmunoassay technique for measuring IgE, J. ALLERGY CLIN. IMMWNOL. 50:193, 1972. Hoffman, D. R.: Estimation of serum IgE by an enzymelinked immunosorbent assay (ELISA), J. ALLERGY CLIN. IMMUNOL. 51:303, 1973. Polmar, S., Waldmann, T., and Terry, W. D.: A comparison of three radioimmunoassay techniques for the measurement of serum IgE, J. Immunol. 110~1253, 1973. Sieber, A., and Becker, W.: Quantitative determination of IgE by single radial immunodiffusion. A comparison of three different methods for intensification of precipitates, Clin. Chim. Acta 50:153, 1974. Carson, D., Metzger, H., and Bazin, H.: A simple radioimmunoassay for the measurement of human and rat IgE levels by ammonium sulfate precipitation, J. Immunol. 115561, 1975. Weltman, J. K., Frackelton, A. R., Szaro, R. P., and Rotman, B.: A galactosidase immunosorbent test for human immunoglobulin E, J. ALLERGY CLIN. IMMUNOL. 58:426, 1976. Johansson, S. G. O., Berglund, A., and Kjellman, N. I. M.: Comparison of IgE values as determined with different solid phase radioimmunoassay methods, Clin. Allergy 6:91, 1976. Nerenberg, S. T., and Prasad, R.: Radioimmunoassays for Ig classes G, A, M, D, and E in spinal fluids: Normal values of different age groups, J. Lab. Clin. Med. 86:887, 1975. Nye, L., Merrett, T. G., Landon, J., and White, R. J.: A detailed investigation of circulating IgE levels in a normal population, Clin. Allergy 1:13, 1975. Underdown, B. J., Knight, A., and Papsin, F. R.: The relative paucity of IgE in human milk, J. Immunol. 116:1435, 1976. Yunginger, J. W., andGleich, G. J.: Seasonalchangesinserum and nasal IgE concentrations, J. ALLERGY CLIN. IMMUNOL. 51:174, 1973. Hansen, R. L., and Gleich, G. J.: Immune responses to hemocyanin from limulus polyphemus in allergic and normal individuals, Int. Arch. Allergy 42607, 1972. Mancini, G., Carbonara, A. O., and Heremans, J. F.: Immunochemical quantitation of antigens by single radial immunodiffusion, Immunochemistry 2:235, 1965. Macy, N. E., O’Sullivan, M. B., and Gleich, G. J.: Comparison of sensitivities of immunodiffusion methods, Proc. Sot. Exp. Biol. Med. 128:1098, 1968. Yunginger, J. W., and Gleich, G. J.: Measurement of ragweed antigen E by double-antibody radioimmunoassay, J. ALLERGY CLIN. IMMUNOL. 50~326, 1972. Bazaral, M., and Hamburger, R. N.: Standardization and stability of immunoglobulin E (IgE), J. ALLERGY CLIN. IMMUNOL.49:189, 1972. Wiedermann, G., Stemberger, H., Kraft, D., Ambrosch, F., Schadlbauer, B., and Jarischko, E.: Quantitativer IgEnachweis mittels modifizierter radialer immunodiffusion im vergleich zum radioimmunosorbent-test (RIST), Z. Immunitaetsforsch. 147:366, 1974. Nakajima, S., Gillespie, D. N., and Gleich, G. J.: Differences between IgA and IgE as secretory proteins, Clin. Exp. Immunol. 21:306, 1975.
M.D., and Sandra L. Dunnette,
B.A. Rochester, Minn.
Because of conflicting reports in the literature concerning the value of various procedures for measurement of IgE in serum and secretions, we compared four different methods, the radioimmunosorbent test (RIST), the double-antibody radioimmunoassay (RIA), the paper disc immunosorbent test (PRIST), and radial immunodt#usion (RID). The standards used in the assays were tested initially in the double-antibody RIA with the use of the reference from the World Health Organization. The results showed that RID as expected was relatively insensitive and IgE was reliably measured only above approximately 1,000 international units (IU). Moreover, certain sera containing low levels of IgE by the other procedures gave distinct precipitin zones and presumably falsely high levels of IgE protein. Thus RID may yield apparently erroneous results when used as a screening procedure for measurement of IgE levels. Among the other procedures PRIST and the double-antibody RIA showed the best agreement. With serum samples RIST yielded values for IgE in the low level range higher than those given by the PRIST and double-antibody RIA. With breast milk and colostrum, values of IgE between 120 and 690 nglml were found by RIST, whereas IgE was not detected by double-antibody RIA and PRIST. No evidence of an inhibitor of IgE was found in breast milk, so that the apparent elevation of IgE in breast milk by the RIST is likely false. These jndings confirm prior reports of spurious elevations of IgE with the RIST and indicate the usefulness of the PRIST and double-antibody RIA ,for the measurement of IgE in sera and secretions.
Since the discovery of IgE as a distinct protein, a variety of methods have been utilized to measure it in sera and secretions.‘-l2 These include two competitive binding radioimmunoassays, one using a solidphase antibody, the radioimmunosorbent test (RIST),‘~ 2 and the other a double-antibody radioimmunoassay (RIA).4 In addition, IgE has been measured by a noncompetitive radioimmunoassay, the paper disc immunosorbent test (PRIST),5 and by a noncompetitive galactosidase-immunosorbent test (GIST). l* Finally, IgE has been measured by single radial immunodiffusion (RID) lo as well as several varieties of RID with the use of radioactive reagents.3, 6 The RIST procedure is available commercially and has been utilized for measurement of IgE in many reports in the literature.*, i3-15 Polmar and associates9 reported that procedures similar to the RIST yielded -From the Allergic Diseases Research Laboratory, Mayo Clinic and Mayo Medical School, Rochester, Minn. 55901. Supported in part by Grant AI 11483 from the National Institute of Allergy and Infectious Diseases. Received for publication Nov. 29, 1976. Accepted for publication Feb. 8, 1977. Reprint requests to: Dr. G. .l. Gleich, Mayo Clinic, Rochester, Minn. 55901.
spuriously high levels of IgE in serum from patients with agammaglobulinemia, and more recently Underdown and associatesl6 have reported that measurement of IgE levels in breast milk by the RIST may yield erroneous results. Because of these reports as well as the claim by Johansson and associatesi that the double-antibody RIA yields higher values of IgE in the low level range than PRIST, we compared these procedures for the measurement of IgE in serum and secretions. MATERIALS AND METHODS Sera and secretions Sera were obtained from the laboratory of Dr. Nai Jiang of the Department of Laboratory Medicine of the Mayo Clinic and Mayo Foundation. Theseserahad beenanalyzed by the double-antibody RIA4 and were selected so as to provide a range of levels of IgE from elevated to barely detectable. In addition, 13 umbilical cord sera obtained at the time of parturition were acquired. Nasal washes were performed in 10 patients as described by Yunginger and Gleich17and parotid fluid was obtained from 6 subjects as described by Hansen and Gleich.‘* Breast milk was obtained from 6 patients and colostrum from 5 patients. The reference standard of the World Health Organization (WHO), 68/341, was kindly provided by Dr. W. D. Terry and was used to calibrate the standardsfor all of the IgE Vol. 59, No. 5, pp. 377-382
378
Gleich and Dunnette
J. ALLERGY
CLIN. IMMUNOL. MAY 1977
b O s B 2
65-
$
-0.500
-
-1.000
-
-1.500
-
2 3 B \ cl 8 2 ‘;;
.$
@
25-
2
-2.000 0 B
0
0.5
Log,,of
I .o
1.5
IgE, ng
FIG. 1. Comparison of standards used in the various procedures for measurement of IgE to the WHO reference by the double-antibody RIA. Symbols used are as follows: WHO, l -; RIA, o----o; RID, A----A; RIST, O- --o; and PRIST, x- - -x. A, The inhibition curves. 9, The logit-log regression lines. The correlation coefficients of the regression lines in B were all greater than +0.99.
Because this reference serum contains the HbB antigen, precautions were taken to prevent personal contact with it during these initial experiments. In our institution personnel routinely handling human sera are checked for the presence of the HbB antigen in sera. In the past year no one in our laboratory has become ill with symptoms suggestive of hepatitis nor has the HbB antigen been detected in sera by RIA.
assays.
Methods
for the measurement
of IgE
Double-antibody RIA. This was performed as described by Gleich and associates4 with the exception that the second incubation was shortened to 2 hr at 23” C resulting in a 24-hr procedure. Rabbit antibody to IgE was used in this procedure. RIST. This procedure was performed as recommended by the Pharmacia Company, Piscataway, N. J. The RIST is a competitive binding RIA in which rabbit anti-IgE molecules are covalently bound to Sephadex particles. IgE in the sample competes with a fixed amount of radiolabeled IgE for a limited number of binding sites on the anti-IgE-Sephadex. After rotation overnight at 23” C and 3 washings with 0.9% saline, radioactivity bound to the Sephadex is counted. The radioactivity associated with the solid-phase anti-IgE is inversely related to the amount of IgE in the patient’s serum. The procedure takes 24 hr. As directed by Pharmacia, values for sera that were diluted lo-fold were divided by 0.96, while sera diluted 5-fold were divided by 0.91. Sera with
PRIST. This procedure, which is a noncompetitive RIA,5 was performed as recommended by the Pharmacia Company. Briefly, sheep anti-IgE covalently coupled to a paper disc binds IgE in the test serum. After incubation for 3 hr at 23” C the discs are washed thrice with 0.9% saline and radiolabeled anti-IgE is added. After a second incubation for 18 hr during which radiolabeled anti-IgE reacts with IgE, the disc is washed and the radioactivity measured in a gamma counter. The amount of radioactivity is directly related to the amount of IgE present in the sample. The procedure takes approximately 24 hr. We diluted sera having greater than 50 IU/ml so as to fall on the midsection of the standard curve, that is, 10 to 50 III/ml. RID. IgE was also measured by radial immunodiffusion (RID)‘O, ig with the use of kits obtained from the Meloy Company, Springfield, Va., and we followed the detailed procedure described by the company. Briefly, the sample is added to a well in an agarose gel containing goat antiserum to IgE. IgE diffuses out of the well during a 48-hr incubation period and reacts with antibody to form a precipitin ring. The diameter of the ring is proportional to the concentration of the IgE in the sample. The gel plates are washed, dried, and stained to visualize the IgE precipitin rings. The ring diameters were read by 2 individuals and the values averaged. The procedure takes a minimum of 3 days.
lmmunoelectrophoresis Certain sera were tested for the presence of antibodies to goat and sheep sera by immunoelectrophoresis. This procedure was performed as described previously by Macy and associates.20 Undiluted goat or sheep serum was added to
VOLUME NUMBER
Procedures for measurement of IgE 379
59 5
r=O
l * .
I’ 1
'
' """'
1
'
IgE,
FIG. 2. Comparison PMT.
of IgE protein levels by RIST and
methods
The results of analysis of IgE by the double-antibody RIA were analyzed by logit-log transformation of the data as described earlier for measurement of ragweed antigen E by Yunginger and Gleich. 2L The inhibition curves were logit-log transformed and the slopes given by the various inhibitors in the double-antibody RIA were tested by analysis of covariance with the use of a programmable Hewlett-Packard 98 IOA calculator.
' """'
'
"l"Ul
1000
10,000
I.U./ml, double antlbody RIA
10,000
r = 0.995
F
s
I100:
10 =
RESULTS
.
we compared the IgE standards used in the various assays to the IgE reference preparation supplied by the WHO. For this purpose the double-antibody RIA was performed with the use of the WHO reference preparation as a primary calibrating reagent and the quantity of IgE in all of the other standards measured by comparison. In Fig. 1 the inhibition produced by the various standards is shown in both curvilinear and linear form after logitlog transformation. Comparison of the inhibition curves in Fig. 1, A indicates that no major difference exists among the various IgE standards. Analysis of the slopes of the logit-log regression lines in Fig. 1, B indicates a difference among the slopes of the lines (F[4, I I] = 5.48; p < 0.025) and inspection suggests that the difference is likely due to the PRIST standard. Analysis of the slopes for RID, RIST, WHO, and the double-antibody RIA standards did not reveal a statistical difference (F[3,9] = 2.8 1; NS). Although there was a statistical difference among the standards in terms of their ability to inhibit the reaction of anti-IgE (ND) with IgE (PS) in the doubleIn initial
' 100
FIG. 3. Comparison of IgE protein levels by RIST and double-antibody RIA.
the wells, and patient’s serum, as the putative antiserum, was added to the trough after the completion of electrophoresis.
Statistical
' """'
10
experiments
I
/: 1
-
,
"""'
10
' ' """'
IgE,
100
' """"
I.U./ml,
FIG. 4. Comparison of IgE protein antibody RIA and by PRIST.
1000
' ""'Ul
10.000
PRIST
levels by double-
antibody RIA, the slopes of the regression lines were quite similar and therefore all of the standards were expressed in terms of the content of IgE in the WHO reference. The WHO reference contains 10,000 IU or 24,000 ng/ml.17, 22 The IgE levels in the sera are shown in Figs. 2 to 4 and in Table I. First, analysis of the results obtained by RID indicates that, above the manufacturer’s suggested lower limit of 900 IU, this procedure appears to correlate very well with the results of the radioimmunoassay procedures. However, in many instances rings were formed in the gel which evidently were spurious. For example, in the case of sample 34 the various radioimmunoassay procedures showed barely detectable levels of IgE, whereas by RID a level of 4,310 ng was obtained. In this case we assume that
380
Gleich and Dunnette
J. ALLERGY
TABLE I. IgE levels in serum (IUlml) Patient No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14
15 16 17
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39* 40 41 42 43 44 45 46 41 48 49 50 51
RID
RIST
PRIST
5,900 4,580 4,310 4,580 3,810 2,080 2,600 1,460 1,940 1,600 1,280 1,010 1,100 1,120 ~800 <800 <800 <800 4,710 1,150 ~800 <800 <800 <800 <800 2,750 a00 ‘a300 <800 5,430 <800 <800 <800 4,310 <800 <800 <800 <800 ~800 <800 <800 <800
5,260 4,680 4,000 4,650 3,850 2,160 2,330 1,260 1,600 1,340 911 655 846 444 249 170 248 132 219 170 67 59 49 68 45 2,020 47 48 37 18 5 26 2.4 28 <2 8 12 4 <2 <2 70 2 3 5 45 3 4 3 <2 2 2
7,670 4,815 3,960 5,685 4,560 ND 2,225 1,890 1,620 1,425 1,140 710 790 505 162 146 105 115 134 80 51 42 32 40 33 1,590 23 26 15 3 3.4 6.4 1.1 3.8 1.5 2.1
ND: not done. *Samples 39 to 51 were umbilical cord sera.
Table II. Levels in secretions Double-antibody RIA
8,290 6,100 5,200 5,708 5,000 2,421 2,460 1,515 1,820 1,110 925 630 641 433 151 128 70 94 106 65 54 32 28 37 22 1,881 20 17 14 5.8 4.7 5.3 1.7 3.6 ND 2.7 1.6
Secretion
RIST
(W/ml)
PRIST
Nasal wash 1 2 3 4 5 6 7 8 9 10
2.8 Cl.3 2.3 1.7 Cl.25 Cl.25 Cl.25 Cl.25 Cl.25 1.9
Breast milk 11 12 13 14 15 16 17 18 19 20 21
415* 500 335 400 335 690 630 295 120 690 135
CLIN. IMMUNOL. MAY 1977
2.2
Double-antibody RIA
7 2 Cl.4 <1.4 Cl.4 <1.4 Cl.4 <1.4 Cl.4 2 <8 <8 <8 4 <8 -4 <8 <4 <8 <4 <8
*Samples diluted Sfold.
the precipitin ring did not reflect the presence of authentic IgE. Discrepancies were also found with sera 19, 20, 30, and 47. Inspection of these sera (19, 20, 30, 34, and 47) revealed that they were not different in physical appearance from the other sera. Also, analysis of these sera by immunoelectrophoresis did not reveal the presence of antibodies reactive with goat or sheep serum. Thus, the RID procedure yields levels of IgE in certain sera which evidently are fallacious. Second, comparison of the results obtained by RIST with the results obtained by PRIST and double-antibody RIA revealed in most cases quite comparable values. The correlation coefficient for the agreement between the PRIST and the RIST was +0.988 and that between the RIST and the doubleantibody RIA was +0.983. Nonetheless, comparison of the results of RIST with those of PRIST revealed a difference by the paired t test (t = 4.80; p < 0.01). Similarly, there was a significant difference between the results obtained by RIST and by the doubleantibody RIA (t = 4.73; p < 0.01). Inspection of Figs. 2 and 3 reveals that in the lower range of IgE values, from approximately 1.6 to 40 III/ml, the results with RIST are higher than those with the PRIST and the double-antibody RIA. Finally, the best agreement among the procedures was found between the
VOLUME NUMBER
53 5
results with PRIST and the double-antibody RIA. As can be seen by inspection of the data in Table I and Fig. 4, there is extremely good agreement between the results from these two procedures; the correlation coefficient was +0.995. Moreover, no difference was found comparing the results by the paired t test (t = -0.46; NS). The results of analysis of secretions shown in Table II revealed a close agreement between those obtained in the double-antibody RIA and the PRIST. The only exception was the secretion from patient 1 in whom a clearly detectable level of IgE was found by the double-antibody RIA while none was detectable by PRIST. Most of the samples of nasal wash fluid did not contain IgE protein measurable by any of the procedures, and all of the 6 samples of parotid fluid were devoid of IgE protein (data not shown). In contrast, analysis of colostrum and breast milk by RIST revealed significant levels of IgE, whereas none was detectable by double-antibody RIA and PRIST.* The latter result could occur if an inhibitor of IgE were present in t.he breast milk and caused falsely low values for IgE in the double-antibody RIA and the PRIST assays. This possibility was tested by adding a known quantity of IgE to breast milk and to phosphate albumin buffer, respectively, and by determining whether there was a difference between the amounts of Ig,E detected in these mixtures. With both the double-antibody RIA and the PRIST, the quantities of IgE measured in the mixtures were essentially the same. Thus these experiments provided no evidence for the pmsence of an inhibitor of IgE in the breast milk. DISCUSSION In this study we measured IgE by various radioimmunoassay procedures as well as by RID and compared the results in serum and secretions. To determine whether the standards used in the various procedures were comparable, we initially measured the slopes of the transformed inhibition curves. The standards were serially diluted and tested for their ability to inhibit the reaction of specific antibody to IgE (ND) with IgE (PS)-islI. The results of these experiments are shown in Fig. 1 and indicate that all of the IgE standards yielded inhibition curves and inhibition lines after logit-log transformation quite similar to the WHO reference. Therefore the IgE values of each of the standards were expressed in IU. Comparison of the levels of serum IgE as shown in *Analysis of fresh cow’s milk by the double-antibody RIA, the RIST, and the PRIST did not reveal the presence of detectable IgE.
Procedures
for measurement
of IgE
381
Figs. 2 to 4 and Table I revealed that the results obtained by PRIST and by the double-antibody RIA agreed remarkably well. Similarly, both of these procedures gave good agreement with the RIST for values of IgE above 10 IU. However, below 10 IU, the agreement was less good as has been previously reported by Johansson and colleagues. l3 The RIST values had been corrected for serum interference as suggested by the manufacturer, but even after this correction discrepancies between the RIST procedure and the double-antibody RIA and PRIST persisted. Thus, we must conclude that there are certain unknown factors present in undiluted serum which can interfere with the measurement of IgE by RIST. Measurement of IgE by RID in the 12 patients with IgE levels greater than 900 IU yielded quite good correlations with the radioimmunoassay methods (r = 0.98 for RIST, r = 0.98 for double-antibody RIA, and r = 0.95 for PRIST). However, RID was flawed by the appearance of rings with certain sera which appeared to be spurious. These sera were not turbid and they did not contain antibodies reactive with goat or sheep serum when analyzed by immunoelectrophoresis. As seen in Table I these sera which contained relatively little IgE by the other procedures often gave detectable rings by RID. Hence, the use of RID as a screening method for elevated levels of IgE, at least in our hands, is flawed by the occurrence of these false positives. Our results, however, were obtained with a classical RIDiS without use of an intensification step (treatment with dopa)‘O and without pretreatment of sera with dextransulfate, which Wiedermann and colleaguesz3 claim eliminates nonspecific ring formation in the agar gel. Even without these additional steps, if the level of IgE is elevated as a result of screening by one of the other procedures, it would seem reasonable to further analyze the serum by RID to measure the absolute level of IgE. The RID procedure, however, takes approximately 3 days for completion. In contrast, in our laboratory we are now performing the doubleantibody RIA as an overnight procedure so that by testing a series of concentrations one could probably have the same data as one would obtain from RID without any loss of time. Also, both RIST and PRIST require only 24 hr for completion. Finally we analyzed only one lot of the RID reagents and one lot of RIST reagents so that we do not know whether the results found here would be encountered with other lots. Analysis of various secretions including nasal washes, parotid fluid, breast milk, and colostrum showed that low levels of IgE are present in these secretions as measured by PRIST and double-
382
Gleich and Dunnette
antibody RIA in agreement with an earlier report from our laboratory. 24In contrast, the RIST performed on undiluted as well as a 1: 5 dilution of breast secretion showed relatively high levels of IgE. When IgE was added to breast milk and analyzed by double-antibody RIA and by PRIST, the expected amount of IgE was found. This result discourages belief that an inhibitor of IgE was present in breast milk which resulted in falsely low values of IgE by double-antibody RIA and by PRIST. These findings are in agreement with the observation of Underdown and associates,16 who have shown that the RIST procedure yields falsely high levels of IgE in breast milk whey and that an inhibitor of the double-antibody RIA is not present in this secretion. Whether the problem of finding falsely high levels of IgE extends to other secretions is not known, but our findings suggest that the results of measurement of IgE in secretions by RIST should be confirmed by another procedure. Authors’
note added in proof
J. ALLERGY
7.
8.
9.
10.
11.
12.
13.
Additional analysesof sera numbered 15-23, 28-30, 32, 34, 36, 37,41, 44, and 0.1 ml of both undiluted of results with those antibody RIA indicated
45 were performed by RIST using and 1: 5 diluted sera. Comparison obtained by PRIST and doublethat the 1: 5 dilution showed better
14.
15.
agreement in 9 cases, the undiluted sera showed better agreement in 4 cases, while no difference was noted in 6 cases. In all analyses of sera with low levels of IgE, less than 30 IU/ml, RIST still yielded values higher than those given by PRIST and double-antibody RIA.
16.
We thank Mrs. Linda Peterson of the La Leche League of Minnesota, Ms. Delores Czerwinski, and Dr. ThomasB. Tomasi, Jr. for providing the samples of colostrum and breast milk, Dr. Nai Jiang for the sera containing known levels of IgE, and Pharmacia Laboratories, Piscataway, N. J., and Meloy Company, Springfield, Va., for gifts of their test kits.
18.
REFERENCES 1. Wide, L., and Porath, J.: Radioimmunoassay of proteins with the use of Sephadex-coupled antibodies, Biochem. Biophys. Acta 130:257, 1966. 2. Johansson, S. G. 0.: Raised levels of a new immunoglobulin class (IgND) in asthma, Lancet 2:951, 1967. 3. Rowe, D. S.: Radioactive single radial diffusion: A method for increasing the sensitivity of immunochemical quantification of proteins in agar gel, Bull. WHO 40:613, 1969. 4. Gleich, G. J., Averbeck, A. K., and Swedlund, H. A.: Measurement of IgE in normal and allergic serum by radioimmunoassay, J. Lab. Clin. Med. 77:690, 1971. 5. Ceska, M., and Lundkvist, U.: A new and simple radioimmunoassay method for the determination of IgE, Immunochemistry 9:1021, 1972. 6. Arbesman, C. E., Ito, K., Wypych, J. I., and Wither, K.:
17.
19.
20.
21,
22.
23.
24.
CLIN. IMMUNOL. MAY 1977
Measurement of serum IgE by a one-step single radial radiodiffusion method, J. ALLERGY CLIN. IMMUNOL. 49:72, 1972. Smith, H. J., Ozkaragoz, K., and Gokcen, M.: A simplified radioimmunoassay technique for measuring IgE, J. ALLERGY CLIN. IMMWNOL. 50:193, 1972. Hoffman, D. R.: Estimation of serum IgE by an enzymelinked immunosorbent assay (ELISA), J. ALLERGY CLIN. IMMUNOL. 51:303, 1973. Polmar, S., Waldmann, T., and Terry, W. D.: A comparison of three radioimmunoassay techniques for the measurement of serum IgE, J. Immunol. 110~1253, 1973. Sieber, A., and Becker, W.: Quantitative determination of IgE by single radial immunodiffusion. A comparison of three different methods for intensification of precipitates, Clin. Chim. Acta 50:153, 1974. Carson, D., Metzger, H., and Bazin, H.: A simple radioimmunoassay for the measurement of human and rat IgE levels by ammonium sulfate precipitation, J. Immunol. 115561, 1975. Weltman, J. K., Frackelton, A. R., Szaro, R. P., and Rotman, B.: A galactosidase immunosorbent test for human immunoglobulin E, J. ALLERGY CLIN. IMMUNOL. 58:426, 1976. Johansson, S. G. O., Berglund, A., and Kjellman, N. I. M.: Comparison of IgE values as determined with different solid phase radioimmunoassay methods, Clin. Allergy 6:91, 1976. Nerenberg, S. T., and Prasad, R.: Radioimmunoassays for Ig classes G, A, M, D, and E in spinal fluids: Normal values of different age groups, J. Lab. Clin. Med. 86:887, 1975. Nye, L., Merrett, T. G., Landon, J., and White, R. J.: A detailed investigation of circulating IgE levels in a normal population, Clin. Allergy 1:13, 1975. Underdown, B. J., Knight, A., and Papsin, F. R.: The relative paucity of IgE in human milk, J. Immunol. 116:1435, 1976. Yunginger, J. W., andGleich, G. J.: Seasonalchangesinserum and nasal IgE concentrations, J. ALLERGY CLIN. IMMUNOL. 51:174, 1973. Hansen, R. L., and Gleich, G. J.: Immune responses to hemocyanin from limulus polyphemus in allergic and normal individuals, Int. Arch. Allergy 42607, 1972. Mancini, G., Carbonara, A. O., and Heremans, J. F.: Immunochemical quantitation of antigens by single radial immunodiffusion, Immunochemistry 2:235, 1965. Macy, N. E., O’Sullivan, M. B., and Gleich, G. J.: Comparison of sensitivities of immunodiffusion methods, Proc. Sot. Exp. Biol. Med. 128:1098, 1968. Yunginger, J. W., and Gleich, G. J.: Measurement of ragweed antigen E by double-antibody radioimmunoassay, J. ALLERGY CLIN. IMMUNOL. 50~326, 1972. Bazaral, M., and Hamburger, R. N.: Standardization and stability of immunoglobulin E (IgE), J. ALLERGY CLIN. IMMUNOL.49:189, 1972. Wiedermann, G., Stemberger, H., Kraft, D., Ambrosch, F., Schadlbauer, B., and Jarischko, E.: Quantitativer IgEnachweis mittels modifizierter radialer immunodiffusion im vergleich zum radioimmunosorbent-test (RIST), Z. Immunitaetsforsch. 147:366, 1974. Nakajima, S., Gillespie, D. N., and Gleich, G. J.: Differences between IgA and IgE as secretory proteins, Clin. Exp. Immunol. 21:306, 1975.