Clinical Relevance of Specific IgE in Penicillin Allergy Investigation

Approximately 10% of hospitalized patients report to have a penicillin allergy [1, 2], but only 10% of these are confirmed on testing [1, 3]. In a large Danish study, immediate reactions (IRs) accounted for approximately 20% of confirmed drug allergies [1]. Anaphylaxis is rare and as few as 0.00048% of the population is reported to have experienced true anaphylaxis following oral administration of a penicillin [2]. Although rare, it is important that patients at high risk of a severe reaction on reexposure are identified and investigated, without exposing the patient to undue risk.

Most penicillin allergy labels are based on nonallergic side effects or viral exanthemas and can be removed after allergy evaluation and investigation [4‒11]. In the literature, there is increased focus on optimizing investigations by using risk stratification based on the history and tailoring the investigation to the risk profile. Risk stratification has been recommended in recent European and Danish guidelines but suggested investigation protocols differ [12, 13]. For IR, the primary focus in the European guidelines is on skin testing as the first-line investigation [12]. In Denmark, testing for specific IgE to penicillins has been routinely used for many years and previous guidelines from 2006 recommended measuring specific IgE in all patients with a suspected penicillin allergy before referral. Updated Danish guidelines from 2019 only recommend specific IgE as the first-line investigation in patients with a history of IR [11, 13, 14]. A recent study has suggested that measuring specific IgE to penicillin minor determinants (MDs) identifies a further 29% of patients not identified on the other four available penicillin IgE assays and that this analysis should be included in investigations [15]. The objectives of this study were to evaluate whether clinical characteristics of patients with a suspected allergic reaction to a penicillin, with elevated specific IgE to one or more penicillins, differ from patients without specific IgE, to identify predictors for elevated specific IgE to penicillins.

Specific IgE data from patients referred to the Allergy Clinic at Gentofte Hospital with a suspected penicillin allergy between 2010 and 2020 were retrieved from a laboratory database. A total of 9,100 patients had 15,000 specific IgE analyses performed. There were 5 penicillin assays available using the ImmunoCAP technology (Thermo Fisher Scientific, Uppsala, Sweden): Specific IgE to penicillin V (PV), penicillin G (PG), amoxicillin (AMX), ampicillin (AMP), and penicillin MDs. Specific IgE values above or equal to 0.35 kUA/L were defined as elevated (IgE positive) and values below 0.35 kUA/L were defined as not elevated (IgE negative) as per the manufacturer’s recommendation at the time of the study.

A total of 477 patients had elevated specific IgE to one or more of the five penicillins and were classified as IgE positive. Medical notes were retrieved from the electronic patient journal, and 430 of these had a relevant clinical history with suspected penicillin allergy and were included in further analysis (laboratory cohort). The remaining 47 patients were excluded due to missing information on clinical history.

In our clinic, IgE negative patients undergo drug provocation (DPT), in most cases with the suspected culprit penicillin. In the study period, skin testing was rarely performed. Clinical data including investigation results are collected prospectively in a drug provocation database. In total 4,508 IgE negative patients with a penicillin as culprit drug, who had undergone DPT were identified from the database. Of these 414 patients were excluded due to missing information about index reaction and a total of 4,094 IgE negative patients were included in the analyses (DPT cohort).

Statistical analyses were performed using a χ² test and a multiple logistic regression adjusting for age, sex, reaction type, symptoms, and treatment to compare the IgE positive laboratory cohort to the IgE negative patients in the DPT cohort. All analyses were performed in R version 4.1.0 and p values <0.05 were considered statistically significant. Some types of reactions were pooled in the analysis. Bearing in mind the severity of respiratory and circulatory symptoms, it was considered unlikely that patients would not recall those. Therefore, patients answering “do not know” to these symptoms were categorized as not having had respiratory and circulatory symptoms.

Of the 9,100 patients where specific IgE results were available, 6,668 (73%) were female, 6,771 (75%) were between 18 and 65 years of age, 398 (4%) were below the age of 18, and 1,931 (21%) were above the age of 65. A total of 477 patients (5.2%) had at least one positive specific IgE analysis, more frequently among males 154/2,432 (6.3%) than females 323/6,668 (4.8%) (p = 0.004).

More patients above the age of 65, 129/1,931 (6.7%) (p = 0.04), were IgE positive and fewer patients under the age of 18, 13/398 (3.3%) (p = 0.003), had a positive specific IgE test compared to the reference group of patients between 18 and 65 years of age 335/6,771 (4.9%). Most IgE positive patients had specific IgE to PV 326/477 (68.3%) and to MD 301/477 (63.1%), whilst IgE to the other three penicillins occurred less often: AMX (19.1%), ampicillin (23.3%), and PG (24.3%), shown in Figure 1. Figure 2 shows the highest specific IgE value for either of the five penicillins in relation to time between reaction and blood sampling (days). The 24 highest values of specific IgE ranged between 14.7 and 276 kUA/L (data not shown). 10 were IgE positive to MD, 12 to PV, 1 to AMP, and 1 to PG. In all these cases, at least one of the following had been reported: urticaria, angioedema, circulatory symptoms, respiratory symptoms, or treatment with adrenaline. Fifteen of the 24 had either circulatory or respiratory symptoms suggesting a severe reaction.

PV, dicloxacillin, and AMX were the culprit drugs most frequently identified from the medical charts (data not shown). Unless the culprit drug was specified clearly in the medical notes, the drug was registered as unknown. Some patients had more than one suspected culprit drug.

Of the 430 patients in the laboratory cohort with specific IgE and a history with suspected penicillin allergy, 139 (32.3%) had a history with an IR. In the univariate analysis significantly fewer IgE negative patients (263/4,094) (6.4%) in the DPT cohort had a medical history with an IR (p < 0.001), shown in Table 1. In the multiple logistic regression analysis, a medical history with an IR was a significant predictor of a positive specific IgE analysis with the highest OR = 4.34 (3.10–5.58, p < 0.001), as shown in Figure 3. Adrenaline treatment was the second strongest predictor with OR = 2.21 (1.32–4.63, p = 0.005). Other significant predictors of a positive specific IgE test were circulatory symptoms, angioedema, male sex and age above 65 years. Urticaria was not a significant predictor: OR = 0.98 (0.75–1.28, p = 0.88).

In this large study of specific IgE to five penicillins, 5.2% of patients referred to our clinic for penicillin allergy had elevated specific IgE to at least one penicillin. The most significant predictor for elevated specific IgE to one or more penicillins was having a history of an IR index reaction. This is consistent with findings in other studies where a history of anaphylaxis and IR were associated with greater likelihood of confirmed allergy to penicillins [16, 17]. In our study, immediate type symptoms such as circulatory symptoms and angioedema were significant predictors for elevated specific IgE to one or more penicillins. Treatment with adrenaline was a very strong predictor and treatment with antihistamines or steroids were also significant predictors. Interestingly, male gender was a significant predictor despite 73% of referred patients being women. Age above 65 was also a significant predictor of elevated specific IgE. This could be explained by more exposures to penicillins over a lifetime, increasing the risk of developing an allergy. Patients with the 24 highest values of specific IgE in our cohort (above 14.7 kUA/L) had at least one of the following: urticaria, angioedema, circulatory symptoms, respiratory symptoms, or treatment with adrenaline. This further supports that higher values of specific IgE are likely to be associated with more severe immediate type symptoms.

The distribution of suspected culprit drugs in this study corresponds well with prescription rates of primarily narrow spectrum penicillins in Denmark [18], so there is no suggestion of single drugs having an increased risk of causing IgE mediated allergy. Different prescription patterns for penicillins in other countries may change the distribution of the positive specific IgE’s as shown in a study from Switzerland, where the IgE to aminopenicillins show a higher positive rate, and IgE to MD a lower positive rate, compared to our study [19].

The literature about specific IgE as a diagnostic tool is scarce. Specific IgE has been used routinely in penicillin allergy investigation in Denmark for many years but is infrequently used in most other countries. Studies of the sensitivity and specificity of specific IgE for penicillins have mainly been performed in mixed cohorts of patients with both IR and NIR and specific IgE is commonly concluded to have a low sensitivity, but a high specificity [17, 20‒23]. One study examined the clinical relevance of sIgE to penicillins in 389 patients with confirmed penicillin allergy based on skin testing and/or clinical history suggesting a sensitivity of 39.6% and a specificity of 68.8% with a cut-off value of 0.35 kUA/L [23]. The result of specific IgE testing may be influenced by the total IgE level, and false positive results can be seen in patients with elevated total IgE levels such as in atopic dermatitis [17, 24, 25]. Lowering the cut-off value may increase sensitivity, but at the same time reduce specificity causing more patients to be incorrectly labelled with a penicillin allergy. One study hypothesizes that applying the specific IgE/total IgE ratio may improve diagnostic performance compared to using the specific IgE level alone [25]. Another study suggested the presence of a non-clinically relevant epitope, phenylethylamine, which might lead to falsely elevated specific IgE to PV and PG [26]. Due to the risk of false positive results, it is recommended that specific IgE should not stand alone as a diagnostic method but should always be interpreted in the light of a clinical history of an IR [16, 22, 27].

In theory, a true IgE-mediated reaction should result in positive specific IgE, but false negatives can occur [20, 28]. One reason for this can be a long delay between index reaction and time of blood test as IgE levels decline over time [29]. Thus, testing within 12 months from index reaction is recommended, as the half-life of serum IgE is less than a year in most patients [13, 29]. Our study shows, however, that specific IgE can remain elevated for longer in some patients and could be sampled later and to avoid false negative values earlier sampling is advised. In addition, a recent study from our group shows that 29.4% more patients have a positive specific IgE if specific IgE to penicillin MDs is added to the investigations, thereby improving sensitivity [15]. A recent study investigating the role of specific IgE in patients with IR to penicillins, suggested that sensitivity of specific IgE was low in a Spanish and Italian population, but this study only analyzed specific IgE for AMX and PG [23]. The pattern of penicillin use in Denmark differs from Southern Europe [18] and in our cohort, the AMX, PG, and AMP assays only test positive in a small subgroup of patients (19.1%, 24.3%, and 23.3%, respectively), whilst specific IgE PV and MD identify a much higher proportion of patients (68.3% and 63.1%, respectively). Our experience from Denmark in using specific IgE for all penicillins is that specific IgE does not always test positive on the assay corresponding to the culprit drug, and therefore we recommend that all five specific IgE assays should be performed in patients with a suspected IR to penicillins.

In the rest of Europe, the USA and Australia skin testing is the first-line investigation for penicillin allergy and specific IgE is not usually recommended due to low sensitivity [12, 30, 31]. The findings of this study suggest that specific IgE can be used as a first-line investigation in patients with a history of IR, especially when severe, and presenting with symptoms from the circulatory system or angioedema requiring treatment.

In Denmark, recent guidelines recommend that general practitioners measure specific IgE for penicillins in patients with a clear history of an IR, ideally 4–6 weeks after the reaction. If one or more specific IgE are elevated, patients should be considered allergic and referral for allergy investigation is not needed, saving time for patients and resources for the specialist allergy service. In our center, skin testing is not used routinely, except for IgE negative patients with a severe IR. Other IgE negative patients with milder IR will go on to provocation typically titrated in three steps. We have previously published results of this protocol showing that out of 1,590 IgE negative patients undergoing provocation with a penicillin only one patient developed anaphylaxis, which responded well to treatment with adrenaline [1]. It is therefore likely that the most severe IgE mediated reactions are identified by the specific IgE analysis when taken at the appropriate time. The proportion of false positive IgE is unknown since very few patients with specific IgE have undergone a DPT, the gold standard in allergy investigation. This problem has also been addressed by other groups [20, 23].

In a recent study on IR to beta-lactam antibiotics, the authors speculate that urticaria occurring within 1 h after first dose of penicillin and disappearing in less than 1 day has a very high diagnostic value [32]. The authors suggest that patients fulfilling this 1-1-1-criterion should be considered high-risk and referred for an allergy workup. This supports the use of urticaria as a high-risk criterion in Danish guidelines [14]. In our study, however, urticaria was not a predictor for specific IgE in the multiple logistic regression analysis. This could be explained by diagnostic uncertainty in distinguishing between urticaria and other rashes, and the fact that we did not have detailed data on timing of skin symptoms.

In conclusion, this is the largest study examining the role of specific IgE to penicillins. The study findings are based on performing all five commercially available IgE assays for penicillins and diagnosing penicillin allergy based on one or more elevated specific IgEs combined with a history of an IR to a penicillin. Our study shows that a history of an IR and treatment with adrenaline are the strongest predictors for a positive specific IgE to penicillins. Using specific IgE for penicillins in the risk stratification of patients with IR to penicillins is likely to improve sensitivity and safety of allergy investigation, as these patients have a higher risk of developing systemic reactions if they undergo drug provocation or future treatment with penicillins. Furthermore, if specific IgE is measured in the primary sector, selectively in patients with a clinical history of a severe IR, patients with positive IgE could be diagnosed in primary care preventing allergy referral and thereby allowing resources to be used more efficiently.

The Danish Patient Safety Authority approved data retrieval from relevant patients on July 14, 2020, journal No. 31-1521-457, which included exemption from written informed consent from patients. Ethical approval was not required under Danish law.

None of the authors have any conflicts of interest relevant to the content of this publication.

This study was not supported by any sponsor or funder. This study was carried out with departmental funding only.

Conception and design: V.L., L.H.G., H.M., and L.K.P. Acquisition of data: V.L., N.K., S.F., and L.K.P. Data analysis and interpretation of data: V.L., L.H.G., H.M., L.K.P., N.K., S.F., J.B.B., and L.B. Drafting the manuscript: V.L., L.H.G., and H.M. Critical revision for important intellectual content and final approval: V.L., L.H.G., H.M., L.K.P., N.K., S.F., J.B.B., and L.B.

Edited by: H.-U. Simon, Bern.

The data that support the findings of this study are not publicly available due to data containing information that could compromise the privacy of research participants but are available from the corresponding author (V.L.) upon reasonable request.