- Email: [email protected]
Development and comparative evaluation of a multiple-antigen RAST as a screening test for inhalant allergy
Development and comparative evaluation of a multiple-antigen RAST as a screening test for inhalant allergy Dennis R. Ownby, M.D., John A. Anderson, M.D., Gregory and Henry A. Hornburger, M.D. Detroit, Mich., and Rochester,
L. Jacobs,
B.S.,
Minn.
We have developed a mod$ed in vitro test for IgE antibodies, the multi&UT, to detect antibodies of dtzerent allergen spectficities simultaneously in a single tube. The multi-RAST is as sensitive for detecting low concentrations of individual IgE antibodies as the discrete RAST. We also evaluated the multi-RAST as a screening test for respiratory allergy to inhalant allergens in children by comparing the results of the multi&UT performed by use of a mixture of SRI’, TGP, and DF allergen-immunosorbents with the results of skin tests, discrete RAST tests for the same allergens, serum total IgE concentrations, and nasal smears for eosinophils in 100 children referred for allergic-disease evaluation. The results of the multi-RAST were more sensitive, spectj2, and eflcient than the results of tests for serum IgE concentration and nasal eosinophils in establishing the diagnosis of inhalant allergy; the multi-RAST was the only diagnostic test that yielded results that were significantly associated with the clinician’s impression of allergy. We conclude that the multi-RAST is a useful and cost-efSective screening test for inhalant allergy in children. (J ALLERGY CLJN IMMUNOL 73:4&G472, 1984.)
Screening tests for allergic disease are frequently ordered from clinical laboratories. There are several reasons for the popularity of these tests. Many physicians feel the need for some laboratory justification before referring a patient to an allergist for evaluation. Furthermore, some types of patients may be extremely difficult to skin test reliably, and it is important to be sure that skin testing is justified in such individuals. Screening laboratory tests may be useful also in clinical research. For example, a recent study of antihistamine-decongestant medications in children with otitis media and effusion was criticized because no screening tests for allergy were performed.’
From the Pediatric Allergy/Immunology Laboratory and Department of Pediatrics, Henry Ford Hospital, Detroit, Mich. and Department of Laboratory Medicine, Mayo Clinic, Rochester, Minn. Presented in part at the Thirty-ninth Annual Meeting of the American Academy of Allergy and Immunology, March 21, 1983, Hollywood, Pla. Received for publication April 27, 1983. Accepted for publication Oct. 18, 1983. Reprint requests: D. R. Ownby, M.D., Division of Allergy and Clinical Immunology, Henry Ford Hospital, 2799 W. Grand Blvd., Detroit, MI 48202.
466
Several different laboratory tests have been proposed to screen for respiratory allergy. Two tests that are often used are measurement of the concentration of total IgE in serum and the examination of nasal secretions for eosinophils. Although these tests are easy to perform and positive results of either test are relatively specific for allergic disease, neither is a sensitive screening test.2, 3 Another laboratory test used widely in the evaluation of respiratory allergy is the IgE-antibody test RAST. Several clinical studies have shown that RAST results agree well with the results of skin tests and allergen challenges in patients suspected of having allergies to inhalant allergens,4-7 but there is uncertainty regarding the number of RAST tests and the allergen specificities required to comprise an adequate screen. As more allergens are tested, fewer allergic individuals have false negative test results, but the cost of testing quickly becomes prohibitive. Compared with skin tests, the discrete RAST performed to many allergens as a screening test is not costeffective. A possible solution to this problem is to test for IgE antibodies of several different allergen specificities simultaneously in a single tube by a multiple-antigen RAST.8 This modified version of the
VOLUME 73 NUMBER 4
Multiple-antigen
Abbreviations used RAST: Radioallergosorbent test SRP: Short ragweed pollen TGP: Timothy grass pollen DF: Dermatophagoides farinae SRW: Short ragweed PRIST: Paper radioimmunosorbent
RAST, hereinafter
called
test
the multi-RAST,
may be
with disk type or microparticulate alllergen-immunosorbents. In this article we describe the analytical characteristics of the multi-RAST performed by adding aliquots of different microcrystalline cellulose allergen-immunosorbents to the same tube to test for the presence of IgE antibodies of different specificities simultaneously.’ We have also compared the results of standard skin tests, tests for eosinophils in nasal secretions, serum IgE concentrations, and tests for IgE antibodies to SRP, TGP, and DF performed by the discrete RAST and the multi&UT techniques in a group of 100 children evaluated clinically for respiratory allergy to inhalant allergens.
performed
MATERIAL Laboratory
either
AND METHODS methods
Allergen-immunosorbents used in the discrete RAST and the multi-RAST were prepared by coupling dialyzed allergen extracts to CNBr activated, microcrystalline cellulose particles. lo Commercial allergen extracts (Hollister Stier Laboratories, Inc.; Downers Grove, Ill.; and Greer Laboratories, Inc.; Lenoir, N.C.) were dialyzed versus 0.15 mol/L NaCl and coupled to activated microcrystalline cellulose particles. The cellulose allergen-immunosorbents were suspended in 0.1 mol/L phosphate buffer, pH 7.5, with 0.1% NaNa, 1% Tween 20, and 0.2% bovine serum albumin (phosphate buffer) at 1 mg of cellulose per milliliter . The discrete RAST was performed by reacting 500 ~1 of a single allergen-immunosorbent with 50 ~1 of serum, an allergic serum pool, or a diluted allergic serum pool in polystyrene tubes after adding phosphate buffer to a final volume of 1 .O ml. After incubating overnight at room temperature, the particles were pelleted by centrifugation and were washed twice. Radiolabeled anti-IgE antibody (Pharmacia Diagnostics, Inc.; Piscataway, N.J.), 8 ng, was added to each tube, and the volumes were adjusted to 1 .O ml with phosphate buffer with 4% (v/v) fetal bovine serum (without bovine serum albumin). After a second overnight incubation at room temperature, the tubes were centrifuged, and the pellets were washed as before. Radioactivity in the pellets was measured by a Tracer model (Tracer Analytic, Inc., Elk Grove Village, Ill.) 1190 gamma counter. The
RAST
for
inhalant
allergy
467
multi-RAST was performed by the same method as the discrete RAST except that 150 ~1 each of SRP, TGP, and DF allergen-immunosorbents were added to the same tube in the first stage of the assay. These three allergens were chosen for use in the multi-RAST based on the results of a preliminary study that demonstrated that children with positive skin tests to any inhalant allergen had a positive skin test to at least one of these three allergens in 96% of cases (D R. Ownby, unpublished observations). The analytical sensitivity of the multi-RAST was compared with the discrete RAST by examining the dose response, i.e., percent of total counts bound of iz51-anti-IgE versus ~1 of allergic serum added, to determine the minimum dose of allergic serum required to produce specific binding of ‘251-anti-IgE more than 400% of the negative control level .I’ The results of the discrete RAST were used to identify sera from children in the clinical study with antibodies to a single allergen in the inhalant allergen mixture, e.g., antibodies to SRW. These sera were then used to compare the analytical sensitivities of two RAST techniques by performing dose-response analyses as described above. Nasal secretions, obtained by having the patient blow into wax paper or with a swab were air-dried, and the slides were stained with Wright’s stain and examined without knowledge of the patient’s allergy status. The smears were scored according to the percentage of eosinophils and neutrophils per 100 nucleated cells as follows: number of eosinophils (or neutrophils) per total nucleated cells (eosinophils + neutrophils + epithelial cells) x 100 = percent eosinophils (neutrophils). Serum IgE concentrations were measured by PRIST (Pharmacia Diagnostics; Piscataway, N.J.). The results were classified in five categories on the basis of ageadjusted concentrations as follows: (1) less than the mean for age, (2) mean to 1 SD above the mean, (3) 1 to 2 SD above the mean, (4) 2 to 4 SD above the mean, and (5) more than 4 SD above the mean.
Comparative
evaluation
of screening
tests
Patients referred for allergy evaluation were included in the comparative study if they were less than 21 yr of age and if a complete allergy evaluation was indicated by the presenting clinical history. Most patients were referred by their personal physicians, but a few were self-referred. A standardized questionnaire was completed for each patient. On the basis of the patient’s history of symptoms and clinical signs only, the allergist recorded the clinical impression of allergy as “doubtful,” “possible,” “probable,” or “highly probable. ” Skin tests were performed initially by the prick method in all patients. The standard battery included four trees, three grasses, five weeds, five molds, and six household allergens. The household allergens tested were: house dust, HMRU dust (Greer Laboratories), DF, cockroach, dog, and cat. Intradermal tests were performed if the tests were indicated by a suggestive clinical history and negative prick test results. In a few children less than 5 yr of age, only selected tests were performed. Prick tests were performed with 1: 20
468
J. ALLERGY CUN. IMMUNOL. APRIL 1994
Ownby et al.
TABLE I. Comparison of the results of IgE-antibody tests performed with an aflergen-immunosorbent mixture and single allergen-immunosorbents on individual sera with IgE antibodies to one allergen Percent bound
Serum D894 D983 D1066 D1087 D189 D524 D868 D947 D1265 D1027 Negative control D154 E29 D538 D338 D775 D789 Negative control no. 1 Negative control no. 2 Negative control no. 3 D955 D747 D1121 D749 D974 D872 D48 D541 D774 Negative control no. I Negative control no. 2 Negative control no. 3
SRW 11.15 8.91 8.29 1.38 2.41 12.85 9.25 10.73 0.77 1.11 0.15 TGP 17.79 7.99 19.44 8.76 8.21 4.16 0.14 0.11 0.15 DF 4.72 13.95 6.36 3.33 25.84 24.20 6.45 1.33 11.17 0.17 0.20 0.23
total counts of 1251-anti-lgE
Mixture 8.05 4.45 4.47 0.79 1.04 9.17 5.96 5.44 0.63 0.62 0.14 Mixture 13.61 2.83 16.55 5.72 6.09 3.36 0.25 0.15 0.14 Mixture 3.95 9.70 4.38 1.79 17.97 15.98 4.41 0.68 8.07 0.25 0.15 0.14
w/v extracts (Hollister Stier Laboratories, Inc. and Greer Laboratories, Inc.). Both prick and intradermal skin test reactions were graded on a scaleof negative to 4+ by use of standard criteria.12 The overall skin test score for each patient was determined from the single strongest reaction to any allergen as follows: (1) no significant reactions, (2) no ieaction more than l+ by the in@adermal method, (3) a positive reaction of 1 to 2+ by the prick method or more than 1+ by the intradermal method, (4) a positive reaction of more than 2+ by the prick method. The sera from all children in the clinical study were tested for IgE concentration and for IgE antibodies to SRW, TGP, and DF by the discrete RAST and by the multi-RAST techniques; nasal secretionsfrom all children were examined for eosinophils.
The results of discrete RAST and multi-RAST tests on sera from children in the clinical study were classifiedas negative, borderline, or positive by calculating the ratio in percent of total counts bound by each serum divided by the percent of total counts bound by a negative control serum pool tested simultaneously as follows: (1) negative, less than 150% of control; (2) borderline, 150% to 400% of control; and (3) positive, more than 400% of control.” The results of all laboratory tests in children were analyzed by use of the Michigan Interactive Data Analysis System through the Wayne State University Computer Center.13 The values of x2 and significancegiven for each section of the tables presented were computed for the entire sectionof the table without further grouping of the data. The sensitivity, specificity, and predictive valuesof positive and negative results for each test and the efficiency of each test were calculated by standard methods.“’ For the purpose of calculating these parameters, only class 3 and 4 skin reactions were considered significant positive reactions. Similarly, only positive RAST scores were considered significant positive reactions, and negative and borderline reactions were considered negative.
RESULTS Analytical characteristics of the multi-RAST The results of the multi-RAST to SRW, TGP, and DF in several sera from children in the comparative study with IgE antibodies to only one allergen are summarized in Table I. The specific binding of radiolabeled anti-IgE antibody mediated by 50 ~1 of serum in the multi&UT was less than in the discrete RAST, but the multi-RAST results were definitely positive in each case. The dose-response curves of three of these sera with IgE antibodies of single specificities to SRW, TGP, or DF are compared in Fig. 1, A, B, and C. The minimum amount of allergic serum required for a definite positive response was the same in each analytical system, approximately 1 to 5 ~1 of serum, but the maximum binding detected in the multi&UT in the presence of 100 ~1 of allergic serum was less than in the discrete RAST (Fig., 1). Comparative evaluation of the multi-RAST as a screening test for inhalant allergy The mean age of the children in the comparative study was 9.8 + 4.6 yr with a range of 2 to 20 yr. There were 41 female and 59 male children. The presenting clinical symptoms and signs were rhinitis in 72 children and asthma in 28. Before evaluating the results of the laboratory tests, we compared two methods for scoring the results of skin tests. One method was to classify skin test results on the basis of the greatest reaction to any single allergen, and the other method was to classify on the basis of the summation of the scores for each group of allergens tested. The two methods produced
VOLUME 73 NUMBER 4
Multiple-antigen
RAST
for
inhalant
allergy
469
SERUM D - 886
SERUM D - 154
SERUM D - 774
/LL serum
/AL serum
FIG. 1. Comparative analysis of dose response of an allergen-immunosorbent TGP, and DF; multi-RAST and discrete SRW, TGP, or DF allergen-immunosorbents individual allergic sera with single allergen specificities.
nearly identical results (x2 = 106.3, df = 6, p < O.O00005), and only those results classified on the basis of the single greatest skin reaction are presented. Table II displays the relationships observed between the results of screening laboratory tests and the skin test scores. The first section of Table II shows the relationship between the highest discrete RAST score to any of the three individual allergens teste’d and the skin test reaction class. As expected, a highly significant relationship was found (p < 0.00005). Not shown is the relationship between the discrete RAST tests and the skin test for the corresponding group of allergens. The RAST to TGP was highly related to the skin test score for any grass pollen h” = 58.7, df = 6, p < 0.00005), and the RAST to SRW was highly related to the skin test score for any weed pollen k2 = 63.1, df = 6,p < 0.00005). By comparison, the association between the discrete RAST to DF and the skin reactions to any of the allergens in the household group was not as strong (x2 = 19.4, df = 6, p = 0.004). This lower association between the discrete RAST to DF and the skin tests to the household allergen group was expected since some patients with positive skin tests reacted only to cat, dog, or cockroach. The relationship of the results of the multi-RAST to skin tests is displayed in the second section of Table
mixture
of SRW, by use of
II. The association between the multi-RAST results and skin test results was slightly greater ($- = 36.5 versus 34.7 with equal degrees of freedom) than for the discrete RAST results and was much stronger than the association of skin test results and tests for nasal eosinophils or serum IgE concentration. The results of the nasal smear tests were divided into three categories on the basis of eosinophil percentages of less than 5%, 5% to 25%, and more than 25%. The relationship of these results to the skin test reactivity scores in all the patients is displayed in the third section of Table II. The association of these data is of minimal statistical significance. Thirty-five of the 75 patients with clinically significant skin test reactions had fewer than 5% eosinophils in nasal secretions, and most of these had no eosinophils. The relationship between the concentrations of total serum IgE and the skin test reactions is shown in the fourth section of Table II. There was a highly statistically significant association. Not shown is the significant association (p < 0.0004) between serum IgE concentrations and the results of the discrete RAST and multi-RAST tests. The results of skin tests, discrete RAST tests, multi-RAST tests, total serum IgE concentrations, and nasal smears are compared to the clinician’s impression of the likelihood of allergy in Table III. Only
470
TABLE
J. ALLERGY CLIN. IMMUNOL. APRIL 1984
et al.
Ownby
II. Comparative
analysis
of the results
of laboratory
tests and skin tests
Class Test
1. Discrete RAST to SRW, TGP, and DF
Negative Borderline Positive Negative Borderline Positive <5% 5 to 25% >25%
2. Multi-RAST to SRW, TGP, DF mixture 3. Nasal eosinophils
4. Serum IgE concentration
TABLE
III.
impression
Comparative of allergy
.
analysis
2 SD >4 SD
of the results
Test
1. Skin test reactions
2. Combination of discrete RASTs to SRW, TGP, and DF 3. Multi-RAST to SRW, TGP, and DF mixture 4. Nasal eosinophils
5. Serum IgE levels
1 2 3 4 Negative Borderline Positive Negative Borderline Positive <5% 5 to 25% >25% 5 mean mean + 1 SD 1 to 2 SD >2 SD >4 SD
of skin
test
reaction
1
2
3
4
3 10 0 \9 4 0 11 1 1
4 6 2 6 5 1 10 2 0
2 11 9 8 6 8 13 4 5
2 12 39 7 7 39 22 17 14
5 6 1 0 1
1 4 2 2 3
1 5 7 6 3
1 6 18 10 18
of skin tests, laboratory
x2 = 34.7, df = 6, p < o.oooo5 x2 = 36.5, df = 6, p -c o.oooo5 x2 = 13.5, df = 6, p < 0.05 x2 = 36.1 I df = 12, p < 0.0003
tests, and the clinicians’
Doubtful
Possible
Probable
Highly probable
0 I 1 1 0 3 0 0 3 0 2 0 1 1 1 1 0 0
7 4 4 14 3 14 12 10 7 12 18 7 4 5 5 7 5 7
4 6 10 19 4 16 19 13 8 18 19 11 9 1 12 14 5 7
2 1 7 19 4 6 19 7 4 18 17 6 6 1 3 6 8 11
the multi-RAST results were significantly associated with the clinician’s impression of the likelihood of allergy. Serum IgE concentrations and the combination of discrete RAST results were weakly associated with the clinician’s impression, but these associations did not reach statistical significance. Table IV displays the predictive value parameters commonly used to evaluate a laboratory test: sensitivity, specificity, the predictive values of positive and negative results, and the diagnostic efficiency of the test. FOF the calculation of these values, each labora-
Significance
x2 = 10.3, df = 9, p = 0.33
x2 = 9.9, df = 6, p = 0.13 x2 = 13.8, df = 6, p = 0.03 x2 = 2.8, df = 6 p = 0.83 ’ x2 = 19.1,df p = 0.09
= 12,
tory test was compared to the skin test score. Skin test scores of 1 and 2 were considered negative, and scores of 3 and 4 were considered positive. Smears for nasal eosinophils were considered positive if there were more than 4% eosinophils present. The usefulness of serum IgE as a screening test was calculated both for concentrations more than 1 SD and for concentrations more than 2 SD above the mean for age. The combination of discrete RAST tests was considered positive if any one of the tests was positive. The most sensitive screening test was measurement
VOLUME 73 NUMBER 4
TABLE
IV. Predictive
Multiple-antigen
values
of screening
laboratory
for
value
of
inhalant
allergy
471
tests Predictive
1. Multi-RAST to SRW, TGP, and DF mixture 2. Combination of discrete RASTs to 3. Serum IgE > 1 SD for age 4. Serum IgE > 2 SD for age 5. Nasal eosinophils 5 5%
RAST
Sensitivity (%I
Specificity (%I
63
96
98
46
71
64
92
96
50
71
83
64
87
55
78
49
76
86
33
56
53
84
91
60
61
of serum IgE with a positive result defined as more than 1 SD above the mean, but the specificity of an IgE concentration above this level was only 64%. When the cutoff concentration was increased to more than 2 SD above the mean, diagnostic specificity was improved to 76%, but the sensitivity fell to 49%. On the basis of high specificity (96%) and moderate sensitivity (63%) of test results, we considered the multi-RAST to be a better screening test than measurement of serum IgE. The results of the three discrete RAST tests had comparable predictive values and efficiency compared with the multi-RAST. DISCUSSION The multi&UT as performed with microcrystalline cellulose allergen-immunosorbents is a sensitive analytical procedure comparable to the discrete RAST for detecting low concentrations of IgE antibodies in vitro. The multi-RAST is also a relatively sensitive, specific, and efficient screening test for respiratory allergy to inhalant allergens. In the comparative study described above, the results of the multi-RAST and discrete RAST tests agreed in all sera, confirming the findings of Merrett and MerretP who used a multi-RAST performed with disk-type immunosorbents .8 We, also compared the multi-RAST results to the results of other screening tests, skin tests, and to the clinician’s impression of the likelihood of allergy. On the basis of criteria used in this evaluation, the multi-RAST is a better screening test than the others for respiratory allergy to inhalant allergens in children. The results of the multi-RAST had greater diagnostic specificity and positive predictive value for inhalant allergy than the results of the IgE test. The multi-RAST results were as useful as the results of three discrete RAST tests and cost only about one third as much to perform. An assumption underlying the comparison of the predictive values of laboratory tests for allergy in this
Positive test (%I
Negative test (%)
Efficiency (%I
study is that positive skin tests are absolute indicators of allergic disease. It is generally recognized, however, that skin test results may be falsely positive or falsely negative in some cases. In this context, it is noteworthy that the clinical impression of the examining allergist as to the likelihood of allergy was significantly related to the results of the multi&UT. This was not surprising since others have observed significant correlations between RAST scores and clinical symptoms during specific pollen seasons.4’ 5 Our findings suggest that the clinician’s impression of the likelihood of allergy is related to the quantity of IgE antibodies to three of the allergens most commonly associated with inhalant allergy. It should be stressed that the children recruited as subjects into this comparative evaluation were not randomly selected from a pediatric age population. The high percentages of positive skin tests and elevated serum IgE concentrations are consistent with the fact that the patients had been referred for allergy evaluation. We feel that this is a valid population for study for two reasons. If a screening test does not perform well in a population with a high prevalence of disease, it will be of little value in a population with a lower prevalence of disease. In clinical practice, screening tests are more likely to be ordered in patients suspected of having allergic disease such as the patients included in this study. On the basis of the high predictive value of a positive multi-RAST (98%), one can be quite certain that a patient with a positive multi-RAST will have positive allergen skin test(s). The multi-RAST, as well as nasal eosinophiba and elevated serum IgE, is relatively insensitive as demonstrated by the low predictive values for a negative test ranging from 33% to 60%. Thus a negative screening test does not rule out inhalant allergy. None of the screening tests evaluated can replace good clinical judgement or evaluation by a trained allergist. An area of discrepancy between RAST results and
472
Ownby
et al.
the allergen skin tests was the lack of a cutaneous reaction to DF allergen by many patients sensitive to other allergens in the household group. Haahtela and JaakonmakP have reported similar results. This discrepancy was especially important in some inner city children in whom cockroach is a major recognized allergen. I5 The choice of the best allergen to represent the household group in a multi-RAST screen deserves further study and may need to be individualized depending on the patient population. Although the three allergens chosen for use in this multi-RAST are probably adequate for much of the eastern United States, the best allergens for other geographic areas need to be determined. In summary, we have found the multi-RAST to be a highly specific screening test for inhalant allergy in . . a pe&atnc population. The multi-RAST, like most screening tests for allergy, was more specific than sensitive. The results of the multi-RAST agreed better with the clinician’s impression than the results of examinations of nasal smears for eosinophils or serum IgE tests. REFERENCES 1. Johnstone DE: Decongestant-antihistamine combination for otitis media with effusion in children. N Engl J Med 308: 1598, 1983 2. Saarinen UM, Juntunen K, Kajosarri M, Bjorksten F: Serum immunoglobulin E in atopic and non-atopic children aged 6 months to 5 years. Acta Paediatr Stand 71489, 1982 3. Kajosarri M, Saarinen UM: Evaluation of laboratory tests in childhood allergy. Allergy 36329, 1981
J.ALLERGYCLIN.
IMMUNOL. APRIL 1984
4. Norman PS, Lichtenstein LM, Ishizaka K: Diagnostic tests in ragweed hay fever. 3 ALLERGY CLIN IM~WNOL 52:210, 1973 5. Lichtenstein LM, Ishizaka K, Norman PS, Sobotka AK, Hill BM: IgE antibody measurements in ragweed hay fever. Relationship to clinical severity and the results of immunotherapy. J Clin Invest 52472, 1973. 6. Haahtela T, Jaakonmaki I: Relationship of allergen-specific IgE antibodies, skin prick tests, and allergic disorders in unselected adolescents. Allergy 36:251, 1981 7. Grtolani C, Miadonna A, Adami R, Restuccia M, Zanussi C: Correlation of specific IgE in serum and nasal secretiotis with clinical symptoms in atopics. Clin Allergy 11:249, 1981. 8. Merrett J, Merrett TG: RAST atopy screen. Clin Allergy 8~235, 1978 9. Jacob GL, Ownby DR, Homburger HA: The simultaneous detection of IgE antibodies to different allergens by a modified radioallergosorbent test. J ALLERGY CLIN IMMUNOL 71:122, 1983 10. Gleich GJ, Jones RT: Measurement of IgE antibodies by the radioallergosorbent test. I. Technical considerations in the performance of the test. J ALLERGY CLIN IMMUNOL55:334,19X 11. Homburger HA, Jacob GL: Analytic accuracy of specific immunoglobulin E antibody results determined by a blind proficiency survey. J ALLERGY CLIN IMMUNOL70:474, 1982 12. Vanselow NA: Skin testing and other diagnostic procedures. In Sheldon JM, Lovell RG, Mathews KP, editors: A Manual of Clinical Allergy. Philadelphia, 1967, WB Saunders Co, pp 59-62 13. Fox DJ, Guire KE: Documentation for MIDAS, ed 3. Ann Arbor, 1978, University of Michigan Press 14. Vecchio TJ: Predictive value of a single diagnostic test in unselected populations. N Engl J Med 274: 117 1, 1966 15. Fromer JM, Anderson JA, Yanari S, Bailey JA: Cockroach sensitivity among children: exposure history, skin test, and IgE-radioallergosorbent (RAST) reactivity. J ALLERGY Ct.m IWMUNOL65:203, 1980
L. Jacobs,
B.S.,
Minn.
We have developed a mod$ed in vitro test for IgE antibodies, the multi&UT, to detect antibodies of dtzerent allergen spectficities simultaneously in a single tube. The multi-RAST is as sensitive for detecting low concentrations of individual IgE antibodies as the discrete RAST. We also evaluated the multi-RAST as a screening test for respiratory allergy to inhalant allergens in children by comparing the results of the multi&UT performed by use of a mixture of SRI’, TGP, and DF allergen-immunosorbents with the results of skin tests, discrete RAST tests for the same allergens, serum total IgE concentrations, and nasal smears for eosinophils in 100 children referred for allergic-disease evaluation. The results of the multi-RAST were more sensitive, spectj2, and eflcient than the results of tests for serum IgE concentration and nasal eosinophils in establishing the diagnosis of inhalant allergy; the multi-RAST was the only diagnostic test that yielded results that were significantly associated with the clinician’s impression of allergy. We conclude that the multi-RAST is a useful and cost-efSective screening test for inhalant allergy in children. (J ALLERGY CLJN IMMUNOL 73:4&G472, 1984.)
Screening tests for allergic disease are frequently ordered from clinical laboratories. There are several reasons for the popularity of these tests. Many physicians feel the need for some laboratory justification before referring a patient to an allergist for evaluation. Furthermore, some types of patients may be extremely difficult to skin test reliably, and it is important to be sure that skin testing is justified in such individuals. Screening laboratory tests may be useful also in clinical research. For example, a recent study of antihistamine-decongestant medications in children with otitis media and effusion was criticized because no screening tests for allergy were performed.’
From the Pediatric Allergy/Immunology Laboratory and Department of Pediatrics, Henry Ford Hospital, Detroit, Mich. and Department of Laboratory Medicine, Mayo Clinic, Rochester, Minn. Presented in part at the Thirty-ninth Annual Meeting of the American Academy of Allergy and Immunology, March 21, 1983, Hollywood, Pla. Received for publication April 27, 1983. Accepted for publication Oct. 18, 1983. Reprint requests: D. R. Ownby, M.D., Division of Allergy and Clinical Immunology, Henry Ford Hospital, 2799 W. Grand Blvd., Detroit, MI 48202.
466
Several different laboratory tests have been proposed to screen for respiratory allergy. Two tests that are often used are measurement of the concentration of total IgE in serum and the examination of nasal secretions for eosinophils. Although these tests are easy to perform and positive results of either test are relatively specific for allergic disease, neither is a sensitive screening test.2, 3 Another laboratory test used widely in the evaluation of respiratory allergy is the IgE-antibody test RAST. Several clinical studies have shown that RAST results agree well with the results of skin tests and allergen challenges in patients suspected of having allergies to inhalant allergens,4-7 but there is uncertainty regarding the number of RAST tests and the allergen specificities required to comprise an adequate screen. As more allergens are tested, fewer allergic individuals have false negative test results, but the cost of testing quickly becomes prohibitive. Compared with skin tests, the discrete RAST performed to many allergens as a screening test is not costeffective. A possible solution to this problem is to test for IgE antibodies of several different allergen specificities simultaneously in a single tube by a multiple-antigen RAST.8 This modified version of the
VOLUME 73 NUMBER 4
Multiple-antigen
Abbreviations used RAST: Radioallergosorbent test SRP: Short ragweed pollen TGP: Timothy grass pollen DF: Dermatophagoides farinae SRW: Short ragweed PRIST: Paper radioimmunosorbent
RAST, hereinafter
called
test
the multi-RAST,
may be
with disk type or microparticulate alllergen-immunosorbents. In this article we describe the analytical characteristics of the multi-RAST performed by adding aliquots of different microcrystalline cellulose allergen-immunosorbents to the same tube to test for the presence of IgE antibodies of different specificities simultaneously.’ We have also compared the results of standard skin tests, tests for eosinophils in nasal secretions, serum IgE concentrations, and tests for IgE antibodies to SRP, TGP, and DF performed by the discrete RAST and the multi&UT techniques in a group of 100 children evaluated clinically for respiratory allergy to inhalant allergens.
performed
MATERIAL Laboratory
either
AND METHODS methods
Allergen-immunosorbents used in the discrete RAST and the multi-RAST were prepared by coupling dialyzed allergen extracts to CNBr activated, microcrystalline cellulose particles. lo Commercial allergen extracts (Hollister Stier Laboratories, Inc.; Downers Grove, Ill.; and Greer Laboratories, Inc.; Lenoir, N.C.) were dialyzed versus 0.15 mol/L NaCl and coupled to activated microcrystalline cellulose particles. The cellulose allergen-immunosorbents were suspended in 0.1 mol/L phosphate buffer, pH 7.5, with 0.1% NaNa, 1% Tween 20, and 0.2% bovine serum albumin (phosphate buffer) at 1 mg of cellulose per milliliter . The discrete RAST was performed by reacting 500 ~1 of a single allergen-immunosorbent with 50 ~1 of serum, an allergic serum pool, or a diluted allergic serum pool in polystyrene tubes after adding phosphate buffer to a final volume of 1 .O ml. After incubating overnight at room temperature, the particles were pelleted by centrifugation and were washed twice. Radiolabeled anti-IgE antibody (Pharmacia Diagnostics, Inc.; Piscataway, N.J.), 8 ng, was added to each tube, and the volumes were adjusted to 1 .O ml with phosphate buffer with 4% (v/v) fetal bovine serum (without bovine serum albumin). After a second overnight incubation at room temperature, the tubes were centrifuged, and the pellets were washed as before. Radioactivity in the pellets was measured by a Tracer model (Tracer Analytic, Inc., Elk Grove Village, Ill.) 1190 gamma counter. The
RAST
for
inhalant
allergy
467
multi-RAST was performed by the same method as the discrete RAST except that 150 ~1 each of SRP, TGP, and DF allergen-immunosorbents were added to the same tube in the first stage of the assay. These three allergens were chosen for use in the multi-RAST based on the results of a preliminary study that demonstrated that children with positive skin tests to any inhalant allergen had a positive skin test to at least one of these three allergens in 96% of cases (D R. Ownby, unpublished observations). The analytical sensitivity of the multi-RAST was compared with the discrete RAST by examining the dose response, i.e., percent of total counts bound of iz51-anti-IgE versus ~1 of allergic serum added, to determine the minimum dose of allergic serum required to produce specific binding of ‘251-anti-IgE more than 400% of the negative control level .I’ The results of the discrete RAST were used to identify sera from children in the clinical study with antibodies to a single allergen in the inhalant allergen mixture, e.g., antibodies to SRW. These sera were then used to compare the analytical sensitivities of two RAST techniques by performing dose-response analyses as described above. Nasal secretions, obtained by having the patient blow into wax paper or with a swab were air-dried, and the slides were stained with Wright’s stain and examined without knowledge of the patient’s allergy status. The smears were scored according to the percentage of eosinophils and neutrophils per 100 nucleated cells as follows: number of eosinophils (or neutrophils) per total nucleated cells (eosinophils + neutrophils + epithelial cells) x 100 = percent eosinophils (neutrophils). Serum IgE concentrations were measured by PRIST (Pharmacia Diagnostics; Piscataway, N.J.). The results were classified in five categories on the basis of ageadjusted concentrations as follows: (1) less than the mean for age, (2) mean to 1 SD above the mean, (3) 1 to 2 SD above the mean, (4) 2 to 4 SD above the mean, and (5) more than 4 SD above the mean.
Comparative
evaluation
of screening
tests
Patients referred for allergy evaluation were included in the comparative study if they were less than 21 yr of age and if a complete allergy evaluation was indicated by the presenting clinical history. Most patients were referred by their personal physicians, but a few were self-referred. A standardized questionnaire was completed for each patient. On the basis of the patient’s history of symptoms and clinical signs only, the allergist recorded the clinical impression of allergy as “doubtful,” “possible,” “probable,” or “highly probable. ” Skin tests were performed initially by the prick method in all patients. The standard battery included four trees, three grasses, five weeds, five molds, and six household allergens. The household allergens tested were: house dust, HMRU dust (Greer Laboratories), DF, cockroach, dog, and cat. Intradermal tests were performed if the tests were indicated by a suggestive clinical history and negative prick test results. In a few children less than 5 yr of age, only selected tests were performed. Prick tests were performed with 1: 20
468
J. ALLERGY CUN. IMMUNOL. APRIL 1994
Ownby et al.
TABLE I. Comparison of the results of IgE-antibody tests performed with an aflergen-immunosorbent mixture and single allergen-immunosorbents on individual sera with IgE antibodies to one allergen Percent bound
Serum D894 D983 D1066 D1087 D189 D524 D868 D947 D1265 D1027 Negative control D154 E29 D538 D338 D775 D789 Negative control no. 1 Negative control no. 2 Negative control no. 3 D955 D747 D1121 D749 D974 D872 D48 D541 D774 Negative control no. I Negative control no. 2 Negative control no. 3
SRW 11.15 8.91 8.29 1.38 2.41 12.85 9.25 10.73 0.77 1.11 0.15 TGP 17.79 7.99 19.44 8.76 8.21 4.16 0.14 0.11 0.15 DF 4.72 13.95 6.36 3.33 25.84 24.20 6.45 1.33 11.17 0.17 0.20 0.23
total counts of 1251-anti-lgE
Mixture 8.05 4.45 4.47 0.79 1.04 9.17 5.96 5.44 0.63 0.62 0.14 Mixture 13.61 2.83 16.55 5.72 6.09 3.36 0.25 0.15 0.14 Mixture 3.95 9.70 4.38 1.79 17.97 15.98 4.41 0.68 8.07 0.25 0.15 0.14
w/v extracts (Hollister Stier Laboratories, Inc. and Greer Laboratories, Inc.). Both prick and intradermal skin test reactions were graded on a scaleof negative to 4+ by use of standard criteria.12 The overall skin test score for each patient was determined from the single strongest reaction to any allergen as follows: (1) no significant reactions, (2) no ieaction more than l+ by the in@adermal method, (3) a positive reaction of 1 to 2+ by the prick method or more than 1+ by the intradermal method, (4) a positive reaction of more than 2+ by the prick method. The sera from all children in the clinical study were tested for IgE concentration and for IgE antibodies to SRW, TGP, and DF by the discrete RAST and by the multi-RAST techniques; nasal secretionsfrom all children were examined for eosinophils.
The results of discrete RAST and multi-RAST tests on sera from children in the clinical study were classifiedas negative, borderline, or positive by calculating the ratio in percent of total counts bound by each serum divided by the percent of total counts bound by a negative control serum pool tested simultaneously as follows: (1) negative, less than 150% of control; (2) borderline, 150% to 400% of control; and (3) positive, more than 400% of control.” The results of all laboratory tests in children were analyzed by use of the Michigan Interactive Data Analysis System through the Wayne State University Computer Center.13 The values of x2 and significancegiven for each section of the tables presented were computed for the entire sectionof the table without further grouping of the data. The sensitivity, specificity, and predictive valuesof positive and negative results for each test and the efficiency of each test were calculated by standard methods.“’ For the purpose of calculating these parameters, only class 3 and 4 skin reactions were considered significant positive reactions. Similarly, only positive RAST scores were considered significant positive reactions, and negative and borderline reactions were considered negative.
RESULTS Analytical characteristics of the multi-RAST The results of the multi-RAST to SRW, TGP, and DF in several sera from children in the comparative study with IgE antibodies to only one allergen are summarized in Table I. The specific binding of radiolabeled anti-IgE antibody mediated by 50 ~1 of serum in the multi&UT was less than in the discrete RAST, but the multi-RAST results were definitely positive in each case. The dose-response curves of three of these sera with IgE antibodies of single specificities to SRW, TGP, or DF are compared in Fig. 1, A, B, and C. The minimum amount of allergic serum required for a definite positive response was the same in each analytical system, approximately 1 to 5 ~1 of serum, but the maximum binding detected in the multi&UT in the presence of 100 ~1 of allergic serum was less than in the discrete RAST (Fig., 1). Comparative evaluation of the multi-RAST as a screening test for inhalant allergy The mean age of the children in the comparative study was 9.8 + 4.6 yr with a range of 2 to 20 yr. There were 41 female and 59 male children. The presenting clinical symptoms and signs were rhinitis in 72 children and asthma in 28. Before evaluating the results of the laboratory tests, we compared two methods for scoring the results of skin tests. One method was to classify skin test results on the basis of the greatest reaction to any single allergen, and the other method was to classify on the basis of the summation of the scores for each group of allergens tested. The two methods produced
VOLUME 73 NUMBER 4
Multiple-antigen
RAST
for
inhalant
allergy
469
SERUM D - 886
SERUM D - 154
SERUM D - 774
/LL serum
/AL serum
FIG. 1. Comparative analysis of dose response of an allergen-immunosorbent TGP, and DF; multi-RAST and discrete SRW, TGP, or DF allergen-immunosorbents individual allergic sera with single allergen specificities.
nearly identical results (x2 = 106.3, df = 6, p < O.O00005), and only those results classified on the basis of the single greatest skin reaction are presented. Table II displays the relationships observed between the results of screening laboratory tests and the skin test scores. The first section of Table II shows the relationship between the highest discrete RAST score to any of the three individual allergens teste’d and the skin test reaction class. As expected, a highly significant relationship was found (p < 0.00005). Not shown is the relationship between the discrete RAST tests and the skin test for the corresponding group of allergens. The RAST to TGP was highly related to the skin test score for any grass pollen h” = 58.7, df = 6, p < 0.00005), and the RAST to SRW was highly related to the skin test score for any weed pollen k2 = 63.1, df = 6,p < 0.00005). By comparison, the association between the discrete RAST to DF and the skin reactions to any of the allergens in the household group was not as strong (x2 = 19.4, df = 6, p = 0.004). This lower association between the discrete RAST to DF and the skin tests to the household allergen group was expected since some patients with positive skin tests reacted only to cat, dog, or cockroach. The relationship of the results of the multi-RAST to skin tests is displayed in the second section of Table
mixture
of SRW, by use of
II. The association between the multi-RAST results and skin test results was slightly greater ($- = 36.5 versus 34.7 with equal degrees of freedom) than for the discrete RAST results and was much stronger than the association of skin test results and tests for nasal eosinophils or serum IgE concentration. The results of the nasal smear tests were divided into three categories on the basis of eosinophil percentages of less than 5%, 5% to 25%, and more than 25%. The relationship of these results to the skin test reactivity scores in all the patients is displayed in the third section of Table II. The association of these data is of minimal statistical significance. Thirty-five of the 75 patients with clinically significant skin test reactions had fewer than 5% eosinophils in nasal secretions, and most of these had no eosinophils. The relationship between the concentrations of total serum IgE and the skin test reactions is shown in the fourth section of Table II. There was a highly statistically significant association. Not shown is the significant association (p < 0.0004) between serum IgE concentrations and the results of the discrete RAST and multi-RAST tests. The results of skin tests, discrete RAST tests, multi-RAST tests, total serum IgE concentrations, and nasal smears are compared to the clinician’s impression of the likelihood of allergy in Table III. Only
470
TABLE
J. ALLERGY CLIN. IMMUNOL. APRIL 1984
et al.
Ownby
II. Comparative
analysis
of the results
of laboratory
tests and skin tests
Class Test
1. Discrete RAST to SRW, TGP, and DF
Negative Borderline Positive Negative Borderline Positive <5% 5 to 25% >25%
2. Multi-RAST to SRW, TGP, DF mixture 3. Nasal eosinophils
4. Serum IgE concentration
TABLE
III.
impression
Comparative of allergy
.
analysis
of the results
Test
1. Skin test reactions
2. Combination of discrete RASTs to SRW, TGP, and DF 3. Multi-RAST to SRW, TGP, and DF mixture 4. Nasal eosinophils
5. Serum IgE levels
1 2 3 4 Negative Borderline Positive Negative Borderline Positive <5% 5 to 25% >25% 5 mean mean + 1 SD 1 to 2 SD >2 SD >4 SD
of skin
test
reaction
1
2
3
4
3 10 0 \9 4 0 11 1 1
4 6 2 6 5 1 10 2 0
2 11 9 8 6 8 13 4 5
2 12 39 7 7 39 22 17 14
5 6 1 0 1
1 4 2 2 3
1 5 7 6 3
1 6 18 10 18
of skin tests, laboratory
x2 = 34.7, df = 6, p < o.oooo5 x2 = 36.5, df = 6, p -c o.oooo5 x2 = 13.5, df = 6, p < 0.05 x2 = 36.1 I df = 12, p < 0.0003
tests, and the clinicians’
Doubtful
Possible
Probable
Highly probable
0 I 1 1 0 3 0 0 3 0 2 0 1 1 1 1 0 0
7 4 4 14 3 14 12 10 7 12 18 7 4 5 5 7 5 7
4 6 10 19 4 16 19 13 8 18 19 11 9 1 12 14 5 7
2 1 7 19 4 6 19 7 4 18 17 6 6 1 3 6 8 11
the multi-RAST results were significantly associated with the clinician’s impression of the likelihood of allergy. Serum IgE concentrations and the combination of discrete RAST results were weakly associated with the clinician’s impression, but these associations did not reach statistical significance. Table IV displays the predictive value parameters commonly used to evaluate a laboratory test: sensitivity, specificity, the predictive values of positive and negative results, and the diagnostic efficiency of the test. FOF the calculation of these values, each labora-
Significance
x2 = 10.3, df = 9, p = 0.33
x2 = 9.9, df = 6, p = 0.13 x2 = 13.8, df = 6, p = 0.03 x2 = 2.8, df = 6 p = 0.83 ’ x2 = 19.1,df p = 0.09
= 12,
tory test was compared to the skin test score. Skin test scores of 1 and 2 were considered negative, and scores of 3 and 4 were considered positive. Smears for nasal eosinophils were considered positive if there were more than 4% eosinophils present. The usefulness of serum IgE as a screening test was calculated both for concentrations more than 1 SD and for concentrations more than 2 SD above the mean for age. The combination of discrete RAST tests was considered positive if any one of the tests was positive. The most sensitive screening test was measurement
VOLUME 73 NUMBER 4
TABLE
IV. Predictive
Multiple-antigen
values
of screening
laboratory
for
value
of
inhalant
allergy
471
tests Predictive
1. Multi-RAST to SRW, TGP, and DF mixture 2. Combination of discrete RASTs to 3. Serum IgE > 1 SD for age 4. Serum IgE > 2 SD for age 5. Nasal eosinophils 5 5%
RAST
Sensitivity (%I
Specificity (%I
63
96
98
46
71
64
92
96
50
71
83
64
87
55
78
49
76
86
33
56
53
84
91
60
61
of serum IgE with a positive result defined as more than 1 SD above the mean, but the specificity of an IgE concentration above this level was only 64%. When the cutoff concentration was increased to more than 2 SD above the mean, diagnostic specificity was improved to 76%, but the sensitivity fell to 49%. On the basis of high specificity (96%) and moderate sensitivity (63%) of test results, we considered the multi-RAST to be a better screening test than measurement of serum IgE. The results of the three discrete RAST tests had comparable predictive values and efficiency compared with the multi-RAST. DISCUSSION The multi&UT as performed with microcrystalline cellulose allergen-immunosorbents is a sensitive analytical procedure comparable to the discrete RAST for detecting low concentrations of IgE antibodies in vitro. The multi-RAST is also a relatively sensitive, specific, and efficient screening test for respiratory allergy to inhalant allergens. In the comparative study described above, the results of the multi-RAST and discrete RAST tests agreed in all sera, confirming the findings of Merrett and MerretP who used a multi-RAST performed with disk-type immunosorbents .8 We, also compared the multi-RAST results to the results of other screening tests, skin tests, and to the clinician’s impression of the likelihood of allergy. On the basis of criteria used in this evaluation, the multi-RAST is a better screening test than the others for respiratory allergy to inhalant allergens in children. The results of the multi-RAST had greater diagnostic specificity and positive predictive value for inhalant allergy than the results of the IgE test. The multi-RAST results were as useful as the results of three discrete RAST tests and cost only about one third as much to perform. An assumption underlying the comparison of the predictive values of laboratory tests for allergy in this
Positive test (%I
Negative test (%)
Efficiency (%I
study is that positive skin tests are absolute indicators of allergic disease. It is generally recognized, however, that skin test results may be falsely positive or falsely negative in some cases. In this context, it is noteworthy that the clinical impression of the examining allergist as to the likelihood of allergy was significantly related to the results of the multi&UT. This was not surprising since others have observed significant correlations between RAST scores and clinical symptoms during specific pollen seasons.4’ 5 Our findings suggest that the clinician’s impression of the likelihood of allergy is related to the quantity of IgE antibodies to three of the allergens most commonly associated with inhalant allergy. It should be stressed that the children recruited as subjects into this comparative evaluation were not randomly selected from a pediatric age population. The high percentages of positive skin tests and elevated serum IgE concentrations are consistent with the fact that the patients had been referred for allergy evaluation. We feel that this is a valid population for study for two reasons. If a screening test does not perform well in a population with a high prevalence of disease, it will be of little value in a population with a lower prevalence of disease. In clinical practice, screening tests are more likely to be ordered in patients suspected of having allergic disease such as the patients included in this study. On the basis of the high predictive value of a positive multi-RAST (98%), one can be quite certain that a patient with a positive multi-RAST will have positive allergen skin test(s). The multi-RAST, as well as nasal eosinophiba and elevated serum IgE, is relatively insensitive as demonstrated by the low predictive values for a negative test ranging from 33% to 60%. Thus a negative screening test does not rule out inhalant allergy. None of the screening tests evaluated can replace good clinical judgement or evaluation by a trained allergist. An area of discrepancy between RAST results and
472
Ownby
et al.
the allergen skin tests was the lack of a cutaneous reaction to DF allergen by many patients sensitive to other allergens in the household group. Haahtela and JaakonmakP have reported similar results. This discrepancy was especially important in some inner city children in whom cockroach is a major recognized allergen. I5 The choice of the best allergen to represent the household group in a multi-RAST screen deserves further study and may need to be individualized depending on the patient population. Although the three allergens chosen for use in this multi-RAST are probably adequate for much of the eastern United States, the best allergens for other geographic areas need to be determined. In summary, we have found the multi-RAST to be a highly specific screening test for inhalant allergy in . . a pe&atnc population. The multi-RAST, like most screening tests for allergy, was more specific than sensitive. The results of the multi-RAST agreed better with the clinician’s impression than the results of examinations of nasal smears for eosinophils or serum IgE tests. REFERENCES 1. Johnstone DE: Decongestant-antihistamine combination for otitis media with effusion in children. N Engl J Med 308: 1598, 1983 2. Saarinen UM, Juntunen K, Kajosarri M, Bjorksten F: Serum immunoglobulin E in atopic and non-atopic children aged 6 months to 5 years. Acta Paediatr Stand 71489, 1982 3. Kajosarri M, Saarinen UM: Evaluation of laboratory tests in childhood allergy. Allergy 36329, 1981
J.ALLERGYCLIN.
IMMUNOL. APRIL 1984
4. Norman PS, Lichtenstein LM, Ishizaka K: Diagnostic tests in ragweed hay fever. 3 ALLERGY CLIN IM~WNOL 52:210, 1973 5. Lichtenstein LM, Ishizaka K, Norman PS, Sobotka AK, Hill BM: IgE antibody measurements in ragweed hay fever. Relationship to clinical severity and the results of immunotherapy. J Clin Invest 52472, 1973. 6. Haahtela T, Jaakonmaki I: Relationship of allergen-specific IgE antibodies, skin prick tests, and allergic disorders in unselected adolescents. Allergy 36:251, 1981 7. Grtolani C, Miadonna A, Adami R, Restuccia M, Zanussi C: Correlation of specific IgE in serum and nasal secretiotis with clinical symptoms in atopics. Clin Allergy 11:249, 1981. 8. Merrett J, Merrett TG: RAST atopy screen. Clin Allergy 8~235, 1978 9. Jacob GL, Ownby DR, Homburger HA: The simultaneous detection of IgE antibodies to different allergens by a modified radioallergosorbent test. J ALLERGY CLIN IMMUNOL 71:122, 1983 10. Gleich GJ, Jones RT: Measurement of IgE antibodies by the radioallergosorbent test. I. Technical considerations in the performance of the test. J ALLERGY CLIN IMMUNOL55:334,19X 11. Homburger HA, Jacob GL: Analytic accuracy of specific immunoglobulin E antibody results determined by a blind proficiency survey. J ALLERGY CLIN IMMUNOL70:474, 1982 12. Vanselow NA: Skin testing and other diagnostic procedures. In Sheldon JM, Lovell RG, Mathews KP, editors: A Manual of Clinical Allergy. Philadelphia, 1967, WB Saunders Co, pp 59-62 13. Fox DJ, Guire KE: Documentation for MIDAS, ed 3. Ann Arbor, 1978, University of Michigan Press 14. Vecchio TJ: Predictive value of a single diagnostic test in unselected populations. N Engl J Med 274: 117 1, 1966 15. Fromer JM, Anderson JA, Yanari S, Bailey JA: Cockroach sensitivity among children: exposure history, skin test, and IgE-radioallergosorbent (RAST) reactivity. J ALLERGY Ct.m IWMUNOL65:203, 1980