Relevance of the determination of serum-specific IgE antibodies in the diagnosis of immediate β-lactam allergy
Pascal Demoly MD, PhD
Exploration des allergies – Maladies Respiratoires and INSERM U454 – IFR3
Hôpital Arnaud de Villeneuve
University Hospital of Montpellier
34295 Montpellier Cedex 5
Background: Allergic reactions to β-lactams are the most frequent cause of adverse drug reactions mediated by specific immunologic mechanisms. They can be explored by in vivo and/or in vitro tests. The measurement of serum-specific immunoglobulin E (IgE) presents several advantages: safety, simplicity, and availability to nonallergologist physicians.
Objectives: To establish the diagnostic value of specific IgE determination in the diagnosis procedure of immediate β-lactam allergy.
Methods: The in vitro determination of β-lactam-specific IgE antibodies was compared in three well-defined groups of patients (n = 45): one with negative skin tests and a positive drug provocation test, another with positive skin tests, and a third control exposed population with good tolerance. Two techniques were used: the CAP-FEIA system (Phadia®) commercially available and a homemade radioallergosorbent test (RAST).
Results: The specificity of CAP-FEIA ranged from 83.3% to 100% and sensitivity from 0% to 25% depending on initial clinical manifestations. The specificity of RAST was between 66.7% and 83.3% and sensitivity 42.9% and 75%. In the subgroup of patients with an anaphylactic shock and negative skin tests, the sensitivity and specificity of RAST were 75%. Positive and negative predictive values were 45.5% and 77.1% with CAP-FEIA and 38.5% and 81.5% with RAST, respectively.
Conclusion: These results indicate that, although the specificity of β-lactam-specific IgE measurement is good, sensitivity is low. Immunoglobulin E measurement should be limited to patients with a clinical history of anaphylactic shock and negative skin tests in order to avoid a drug provocation test. More sensitive assays should be developed.
Allergic reactions to β-lactams are the most common cause of adverse drug reactions mediated by specific immunologic mechanisms. They could be life-threatening and require the identification of the molecule in question. Several clinical entities have been described and those occurring immediately after drug exposure are immunoglobulin E (IgE)-mediated and explored by skin testing and by the in vitro measurement of serum-specific IgE. The sensitivity of these tests is not 100% and even for patients with a clear positive history, a drug provocation test may be required in order to confirm the diagnosis (1).
The advantages of the in vitro determination of specific IgE antibodies when compared with in vivo testing are that the former poses no direct risk to the patient and does not require personnel with expertise. Even though in vitro tests are recommended in immediate hypersensitivity reactions (2), their exact place in the diagnostic procedure is not clear and certain authors do not use this method in daily practice (3). In one study, in terms of sensitivity, 11 of 26 patients (42%) with negative skin tests and a positive drug provocation challenge (or repeated clinical history) had specific IgE to benzylpenicilloyl or amoxicilloyl (4). The specificity of the test was 95–100%. Therefore, IgE measurements can avoid a potentially harmful drug provocation test.
This study included patients who were already diagnosed as well as negative controls. They were all classified into three groups: one with negative skin tests and a positive drug provocation test, another with positive skin tests, and a third control exposed population with good tolerance. The aim was to establish the relevance of both commercially available and homemade-specific IgE antibody measurements, by determining the respective sensitivities, specificities and, when possible, positive (PPV) and negative predictive values (NPV).
β-Lactam allergic cases with a positive history, negative skin tests and a positive drug provocation test were compared with two groups: one composed of patients with a positive history and positive skin tests (allergic controls) and another with negative skin tests and good tolerance after β-lactam administration (nonallergic exposed controls). These groups were chosen in order to address two different questions: Could β-lactam-specific IgE measurement avoid a potentially harmful drug provocation test for patients with negative skin tests? Could IgE measurement avoid the β-lactam skin tests performed by allergologists?
Material and methods
Subjects who developed an immediate reaction after the administration of a β-lactam derivative documented by the referring physician was included in the study. All manifestations occurred <6 h after the drug intake. Three clinical entities were defined: urticaria, anaphylaxis without shock, and anaphylactic shock. When urticaria was associated with any other noncutaneous symptom, the patient was classified as ‘anaphylaxis’ or ‘anaphylactic shock’ if there was a drop in blood pressure
Between September 1996 and August 2004, subjects were explored for drug allergy in our clinic (University Hospital of Montpellier, Montpellier, France). All patients gave their written consent. If skin tests proved negative (2), they underwent skin tests with different β-lactams and a drug provocation test, according to the ENDA/EAACI recommendations.
The determination of specific IgE antibodies was not undertaken at this stage in order to avoid any interference with the diagnostic procedure. Three groups of patients were considered, based on the results of the β-lactam skin tests and provocation tests. The first group included patients with negative skin tests and positive oral provocation. The second group included those with positive skin tests, and the third served as a control group with negative skin tests and good tolerance. Patients from the two latter groups were matched as much as possible with patients from the first group in terms of age, clinical history, and delay between clinical manifestations and testing. Our Drug Allergy and Hypersensitivity Database (3) was used for this procedure. The patients’ characteristics are mentioned in Table 1. Subjects presenting one or more positive skin prick tests to common aeroallergens were considered as atopic patients.
Table 1. Patient characteristics and β-lactam-specific immunoglobulin E results
Samples were obtained from the patients whilst they were being evaluated in the clinical outpatient department. These samples were taken before the skin test procedure and were frozen at −20°C until use. All samples were run in parallel for the in vitro tests.
Specific IgE antibody determination by the commercially available Phadia CAP System®
The Phadia CAP System FEIA® (Uppsala, Sweden) is an in vitro test system based on ImmunoCap technology. It is used for the determination of circulating specific IgE antibodies. The β-lactam c1 (penicilloyl G), c6 (amoxicillin), c5 (ampicillin), and c7 (cefaclor), covalently coupled to ImmunoCap, interact with the specific IgE in the serum samples taken from the patient. After washing away nonspecific IgE, enzyme-labeled antibodies against IgE were added to form a complex. After incubation, unbound enzyme anti-IgE were washed away and the bound complex was then incubated with a developing agent. After stopping the reaction, the florescence of elute was measured in FluoroCount 96 (Phadia, Uppsala, Sweden). To classify test results, the fluorescence in patient samples was compared directly with that for standards run in parallel. Values above 0.35 kUA/l were considered positive and those below 0.35 kUA/l were negative. This assay was performed in the INSERM U 454 Unit, University Hospital of Montpellier, France.
Homemade-specific IgE antibody determination
This determination was performed in the Research Unit for Allergic Diseases, Carlos Haya Hospital, Malaga, Spain. It was produced by radioallergosorbent test (RAST) using three penicillins: penicillin G (BP; Normon SA, Madrid, Spain; 105 IU/ml), amoxicillin (AX; SB Smithkline Beecham, Madrid, Spain; 20 mg/ml), ampicillin (AMP; Normon SA; 20 mg/ml) and eight cephalosporins (2 mg/ml): cefaclor (Lilly, Madrid, Spain), cefuroxime (GlaxoSmithKline S.A, Madrid, Spain), ceftazidime (GlaxoSmithKline S.A), ceftriaxone (Roche, Basel, Switzerland), cefotaxime (Roussel Iberica Laboratorios, Madrid, Spain), cefazolin, cefixim and cefatrizin (GlaxoSmithKline S.A) conjugated to poly-l-lysine (PLL; Sigma, St Louis, MO, USA) in the solid phase as described previously (5). About 30 μl of the patients’ sera was incubated in the solid phase with different β-lactam conjugates. After washing, radiolabeled anti-IgE antibody was added and incubated overnight. The disks were then washed and their radioactivity was measured in a gamma counter, Cobra II auto-gamma (Packard BioScience Company, Frankfurt, Germany). Samples were considered positive if they were higher than 2.5% of the label uptake, which were the mean ± 2 SD of the negative control group. All the determinations were blind measurements of specific IgE. An IgE measurement technique is considered positive when one (or more) determination (using this technique) is positive.
Age, clinical entity, and the time between clinical manifestations and testing had been paired one by one as much as possible using our Drug Allergy and Hypersensitivity Database. The Kruskal–Wallis test and Mann–Whitney U-test with the Bonferroni's correction were used for statistical comparisons between the three groups. Determinations of sensitivity, specificity, and PPV and NPVs were performed according to the prevalence of allergic patients in our patient series (3; Table 2).
Table 2. Prevalences of demonstrated allergy in our Drug Allergy and Hypersensitivity Database (3) for patients with a suspicion of immediate drug allergy to β-lactams and depending on the clinical presentation
|anaphylaxis without shock||57||29.8|
The 45 patients included in our study are described in Table 1. For each patient, the following factors are mentioned: age, sex, clinical manifestation, the incriminated β-lactam, the intervals separating clinical manifestation and testing, allergy in vivo and in vitro explorations.
The sex ratio was 2 with 30 women (66.7%) and 15 men (33.3%). The mean age was 38.5 years with a range of 7–67. Twenty-six (80%) patients were atopic.
No significant differences existed between the groups in terms of sex, atopy, time separating the clinical manifestation, and allergy explorations. However, there was a difference in age between groups 1 and 3, as well as between groups 2 and 3 (P = 0.018; Table 3) in spite of the matching.
Table 3. Patient group comparisons (median, 25–75 percentile)
|1||Patients with negative skin tests and positive provocation challenge||15||34 [22–39]*||4/11||12/3||5 [3–18]|
|2||Patients with positive skin tests||15||37 [26–49]*||3/12||8/7||3 [3–10]|
|3||Patients with both negative skin and provocation tests||15||46 [39–56]*||8/7||7/8||7 [3–13]|
After taking cephalosporins, 11 patients presented an immediate reaction in the first group and four in the second and third groups. Penicillins were involved in the other cases. According to the previously defined clinical entities, each group was composed of seven urticaria, four anaphylaxis, and four anaphylactic shock.
Determination of sensitivity and specificity of IgE measurements in the negative skin test population
All results are presented in Table 4. The specificity of CAP-FEIA ranged between 85.7% and 100% and sensitivity between 0% and 25% depending on initial clinical manifestations. The specificity of RAST ranged from 71.4% to 75.0% and sensitivity from 25.0% to 75.0%. It was best for both techniques in the subgroup of patients with an anaphylactic shock and negative skin tests. When we withdrew the five cases and five controls assessed more than 1 year after the allergy episode, sensitivity were 10.0% for CAP-FEIA and 60.0% for RAST; specificity were 90.0% and 70.0%, respectively.
Table 4. Determination of sensitivity, specificity, and positive and negative predictive values in two clinical situations: patients with negative skin tests and all patients before skin testing
|Negative skin test|
| Whole population||15||15||CAP FEIA||6.7||93.3||/||/|
| Urticaria||7||7||CAP FEIA||0||85.7||/||/|
| Anaphylaxis||4||4||CAP FEIA||0||100||/||/|
| Anaphylaxis and anaphylactic shock||8||8||CAP FEIA||12.5||100||/||/|
| Anaphylactic shock||4||4||CAP FEIA||25.0||100||/||/|
|Before skin testing|
| Whole population||30||15||CAP FEIA||16.7||93.3||45.5||77.1|
| Urticaria||14||7||CAP FEIA||14.3||85.7||17.8||82.2|
| Anaphylaxis||8||4||CAP FEIA||25.0||100||100||75.9|
| Anaphylaxis and anaphylactic shock||16||8||CAP FEIA||18.8||100||/||/|
|RAST||56.3||75.0|| || |
| Anaphylactic shock||8||4||CAP FEIA||12.5||100||100||70.1|
Positive and negative predictive values could not be calculated as there were no available data on the prevalence of allergic patients in a negative skin test population.
Determination of sensitivity, specificity, and positive and negative predictive values of IgE measurements before skin testing in all patients
The specificity of CAP-FEIA varied from 85.7% to 100% and sensitivity from 12.5% to 25.0% depending on the initial clinical manifestations (Table 4). The sensitivity of RAST was higher with 42.9–62.5% and specificity lower (ranging from 71.4% to 75.0%). Predictive values were superior in the population with anaphylaxis or anaphylactic shock when using the CAP-FEIA measurement when compared with the homemade RAST. The PPV reaches 100% in this situation and the NPV reaches 75.9% for the population presenting an anaphylaxis and 70.1% in situations of an anaphylactic shock. In an urticarian presentation, the PPV and NPVs of RAST were superior with 24.5% and 85.2%, respectively, vs 17.8% and 82.2% with the CAP-FEIA technique. When we withdrew the eight cases and five controls assessed more than 1 year after the allergy episode, sensitivity were 18.0% for CAP-FEIA and 50.0% for RAST; specificity were 90.0% and 70.0%, respectively.
Determination of sensitivity and specificity of IgE measurements in patients with cephalosporin- or aminopenicillin-induced manifestations
In patients with cephalosporin-induced manifestations, sensitivity and specificity displayed by RAST were 40.0% and 75.0%, respectively, vs 6.7% and 75.0% by the CAP-FEIA for the whole population before skin testing (lower when compared with the total patient group; Table 5). In patients with aminopenicillin-induced manifestations, sensitivity and specificity for RAST were 64.3% and 72.7%, respectively vs 28.6% and 100.0% by the CAP-FEIA for the whole population before skin testing (better when compared with the total patient group).
Table 5. Determination of sensitivity and specificity in the cephalosporin- and aminopenicillin-induced hypersensitivity populations
| Whole population||15||4||CAP FEIA||6.7||75.0|
| Negative skin test population||11||4||CAP FEIA||9.1||75.0|
| Whole population||14||11||CAP FEIA||28.6||100|
| Negative skin test population||4||11||CAP FEIA||0||100|
Several in vitro methods (5–8) are reported to be useful in the diagnosis of patients with suspected immediate allergic reactions to β-lactams, but not all have been standardized or evaluated. For safety reasons, IgE measurements are of particular interest.
In a first approach, usually undertaken by general practitioners who do not have access to drug skin tests, one is interested in the determination of the sensitivity and specificity of the IgE immunoassays in the whole population. Our results demonstrate that specificity is good for both assays and better for CAP-FEIA (85.7–100% depending on initial clinical presentations) when compared with homemade RAST (71.4–75.0%). These results have to be confronted with the data available in the literature. However, to our knowledge, no study has ever shown (as we have) the diagnostic accuracy of the specific IgE CAP-FEIA in immediate allergy to β-lactams. In a study performed by Blanca et al. (5), where data concerned only amoxicillin and penicillin G, sensitivity was the same (42%) and specificity better (97%) using RAST technology in a similar patient population. There was no information on predictive values.
Although the two IgE assays were not (significantly) concordant, the homemade RAST did not bring any additional benefit in terms of predictive values. However, its significantly higher sensitivity is an interesting element (with a higher NPV), certainly due to the increased number of β-lactams available and particularly to the number of cephalosporins (three penicillins and eight cephalosporins, when compared with three and one for the CAP-FEIA).
It should be noted that the β-lactams incriminated in our study are not all available in the Phadia CAP-FEIA (19 of 45 cases and especially 19 cephalosporins). Results were best for aminopenicillin allergic patients. Nonpenicillin β-lactams are given particular attention in our study, especially in skin test-negative patients (11 of 15). The exploration of cephalosporin allergies is not standardized as well (as much or as thoroughly) as penicillins. Even if most immediate reactions to cephalosporins appear to be IgE-mediated, the cephalosporin allergenic determinants have not been properly identified, and cross-reactivity between penicillins and cephalosporins, as well as among cephalosporins, is low if they do not share the same side chain (9). Some experts have mentioned that skin testing with antigenic determinants of penicillin G, amoxicillin, and ampicillin is no longer sufficient for evaluating patients for β-lactam-induced allergy. The use of an extended panel of cephalosporin is necessary. Techniques of specific IgE detection-like sepharose-radioimmunoassay seem efficient (10), but the commercially available biologic assay CAP-FEIA lacks all but cefaclor IgE assays. The identification of β-lactam-specific IgE in patients’ sera requires the development of a large panel of molecules (in order to increase sensitivity), epitopes, and β-lactam carriers.
We could consider the PPV and NPVs depending on the relative frequencies of demonstrated drug allergy in our Drug Allergy and Hypersensitivity Database (Table 2). For serious clinical situations, PPVs displayed by the Phadia CAP-FEIA are 100%. For anaphylactic manifestations with or without anaphylactic shock, the estimated PPV and NPVs are high. Although this is a small case series, it suggests that IgE measurement could therefore have a place in the diagnosis procedure of anaphylaxis patients seen by nonallergologist physicians. Skin tests do, however, have a higher sensitivity and specificity. Immunoglobulin E assays might therefore be indicated in situations with no access to β-lactam skin testing when using nonsoluble cephalosporins and in high-risk patients.
In a second situation corresponding to allergologists who can carry out drug skin tests, one is interested in the determination of the sensitivity and specificity of IgE assays in the population with a suspicion of β-lactam allergy and negative skin tests. Does the practice of IgE assays make it possible to avoid an oral provocation test? Our results demonstrate that although specificity is good (93.3%), sensitivity is low (6.7%) for the CAP-FEIA. Sensitivity is improved by the homemade RAST (46.7%) but specificity is deteriorated (73.3%). In a published population of patients with negative skin tests, only 40 of 89 patients (44.9%) with negative β-lactam RAST had a negative provocation test (11).
It was impossible to estimate PPV and NPVs because of the lack of information on the prevalence of the positive diagnosis of β-lactam allergy in the negative skin test population. Only 15 patients with sera were identified in our database. So this situation seems to be rare and skin testing powerful in the diagnosis strategy of β-lactam allergy. The low sensitivity of the assays could be explained by a decrease over time of specific IgE levels, as it has been demonstrated for β-lactam immediate positive skin tests (2). However, when we withdrew the eight cases and five controls assessed more than 1 year after the allergy episode, sensitivity and specificity for the whole group did not significantly improve.
Moreover, immediate clinical manifestations because of β-lactams with positive oral challenge do not necessarily indicate an IgE-induced manifestation, even if it happened <6 h after drug intake. Other immunologic models proposed recent theories about non-IgE drug hypersensitivity (12). This could partly explain the lack of sensitivity of IgE assays in the population with negative skin tests especially with cephalosporin derivatives. For cephalosporins, the most reasonable explanation for the lack of sensitivity is the inadequate understanding and knowledge of the IgE determinants with this group.
However, this technique has a sensitivity of 75% in the population of patients having a history of anaphylactic shock. This population corresponds clinically to the patients who present a vital risk if the drug is re-administered. Thus, if these results are duplicated, a positive measurement of β-lactam-specific IgE in patients having had an anaphylactic shock but negative skin tests could be regarded as sufficient to avoid a life-threatening oral provocation test.
These results indicate that, although the specificity of β-lactam-specific IgE measurement is good, sensitivity is low. The place for IgE measurement should be limited to situations of severe manifestation in order to avoid drug provocations. Sensitive assays, in particular for cephalosporin allergies, should be developed. Increasing the number of β-lactam derivatives available in vitro is as important as for the in vivo diagnosis (13). A prospective multicenter study with systematic IgE measurement in situations of negative skin tests could provide us with more information about the performance of these assays.