A novel approach to predict cetuximab‐induced hypersensitivity reaction: detection of drug‐specific IgE on basophils

Abstract Cetuximab is remarkable for the relatively high rate and severity of hypersensitivity reactions (HR) being reported in the literature. Screening for cetuximab‐specific IgE in serum via immunoassay has been found to be useful in preventing HR; however, these tests are known to have a low positive predictive rate. In an attempt to remedy this, we evaluated the interaction between cetuximab and IgE on basophils for predicting severe cetuximab‐induced HR. Twelve head and neck cancer patients were enrolled in this single‐institution study: four with a history of cetuximab‐induced HR and eight with no such history. Cetuximab‐specific and galactose‐α‐1,3‐galactose (α‐gal) specific IgEs in serum were measured in vitro using an enzyme‐linked immunosorbent assay (ELISA). IgE‐cetuximab binding on basophils was also analyzed to evaluate the decrease in cetuximab molecules on basophils after dissociation of IgE from FcεRI. The positive predictive value associated with the presence of cetuximab‐ or α‐gal‐specific IgE in serum was found to be only 0.67, whereas the negative predictive value was 1.00. On the other hand, in all four patients who developed HR, the cetuximab molecules on basophils were decreased significantly due to the dissociation of IgE from basophils (P < 0.05). However, this was not the case in patients who did not develop HR. In conclusion, our results strongly imply that the IgE‐cetuximab interaction on basophils may be key to developing improved methods for predicting severe cetuximab‐induced HR.


Introduction
Indeed, previous studies have demonstrated a clear link between cetuximab-induced severe HR and the presence of cross-reactive IgE antibodies in the serum [12]. The IgE antibody has also been shown to be crossreactive with the oligosaccharide, galactose-α-1,3galactose (α-gal), present in the Fab portion of the cetuximab heavy chain [12,13]. And in patients with severe cetuximab-induced HR, the cross reaction between IgE antibodies and both αgal and cetuximab have already been observed [12]. In this study, anti-cetuximab IgE antibodies were detected in 17 out of 25 patients who displayed severe HR, whereas in patients without HR, this was only found in one out of 51. This finding suggests that the cetuximab-specific IgE in serum may be a possible biomarker for predicting severe HR in clinical settings. Furthermore, there have been two largescale studies evaluating the presence of cetuximab-specific IgE in serum. These studies demonstrated that the absence of cetuximab-specific IgE was definitively associated with HR negative status in almost all cases. However, positive test results were not found to be strongly associated with severe HR (positive predictive value 0.333-0.577) [12,14]. Therefore, a screening test with much higher positive predictive value is clearly needed in order to predict and prevent severe HR due to cetuximab infusion.
Due to their role in the release of inflammatory mediators, basophils have become a useful target for understanding and diagnosing IgE-mediated allergic reactions [15]. CD203c has been identified as a specific surface marker for basophils and mast cells of the hematopoietic lineage [16]. It is expressed on resting cells at low levels with a rapid up-regulation of expression following activation [17]. CD203c was also considered to be a more specific marker than CD63 for IgE-mediated basophil activation [18]. Recently, ex vivo exposure to cetuximab was found to induce basophil activation in two out of two patients with an allergy to red meat, in whom αgalspecific IgE were detected [19]. Considering the relationship between αgal-and cetuximab-specific IgE, these findings lead us to conclude that a more precise understanding of the pharmacodynamic action of cetuximab on basophils could be useful for predicting cetuximabinduced severe HR.
The aim of this study is to construct a more effective way of predicting cetuximab-induced HR in the clinical setting. To accomplish this, we examine differences in the interaction between cetuximab and IgE on basophils in patients with and without cetuximab-induced HR. This research was designed to contribute to the development and discovery of improved pharmacodynamics-based methods for the early detection of potentially life-threatening allergic reactions.

Study subjects
This study included patients who were admitted to the Mie University Hospital between April 2014 and June 2015. The entry criteria were as follows: diagnosis of head and neck cancer, history of cetuximab-induced HR within the past 3 months or initiation of cetuximab therapy within the next week. Twelve patients with head and neck cancer were enrolled in this study; characteristics of the patients are shown in Table 1. Two patients had a history of grade 3 allergy during the first cetuximab infusion. Ten patients were started on cetuximab therapy after enrollment; of these, two patients developed grade 1 allergy during the first cetuximab infusion, whereas eight had no allergic reactions. The degree of HR was evaluated according to the Common Terminology Criteria for Adverse Events Version 4.0. All 12 patients received dexamethasone and an H 1 -receptor antagonist just prior to cetuximab administration. Upon study entry, blood samples were obtained from the patients on two separate days.
This study was conducted in accordance with the Declaration of Helsinki and its amendments. This study's protocol was reviewed and approved by the Ethics Committee of Mie University (certification No. 2717), and written informed consent was obtained from each subject.

Serum assays for specific IgEs and total IgE detection
Cetuximab-specific IgE in serum was measured in vitro using an enzyme-linked immunosorbent assay (ELISA) according to the method reported by Mariotte et al. [14], with minimal modifications. High protein-binding capacity polystyrene 96-well plates (Maxisorp Nunc, Thermo Scientific, Yokohama, Japan) were coated with 100 μL of a 0.5 μg/ mL cetuximab solution (Erbitux ® , Merck Serono, Tokyo, Japan) overnight at 4°C. Next, the plates were saturated with 1% human albumin solution for 2 h at 37°C. Duplicate serum samples (diluted 1/10) were added and incubated overnight at 4°C. The binding of IgE antibodies to cetuximab was detected using 500 ng/mL biotinylated goat monoclonal anti-human-IgE (Vector laboratories, Burlingame, CA) and allowed to react for 1.5 h at 37°C. Streptavidin-alkaline phosphatase (Sigma-Aldrich, Osaka, Japan), 1/2000 dilution, was added, followed by 1 mg/mL paranitrophenyl phosphate solution (PNPP, Interchim, Montluçon, France). Optical density was measured at 415 nm by microplate reader (iMark ™ , Bio-Rad, Hercules, CA).
The results were expressed as international units (IU) per milliliter (with 1 IU equivalent to approximately 2.4 ng IgE). Human myeloma IgE (Calibiochem, Billerica, MA) used as standard samples were prepared from 0.2 to 170 IU/ mL. Total IgE concentration was also measured according to the rate of patients' serum samples directly binding to the solid phase of Maxisorp Nunc microplate. The limit of detection for cetuximab-and αgal-specific IgEs and total IgE were 0.2 IU/mL. Calibration curves for cetuximab-and αgal-specific IgE and total IgE had good linear correlations between respective IgE concentration and absorbance; these correlation coefficients (r 2 ) were 0.999 (n = 2), 0.999 (n = 2), and 0.998-0.999 (n = 2), respectively.

Measurement of CD203c expression on basophils
Measurement of CD203c expression on basophils was carried out within 6 h of blood sampling. Patients' peripheral blood mononuclear cells (PBMCs) were isolated from whole blood anticoagulated with EDTA. The Allergenicity Kit (Beckman Coulter, Fullerton, CA) was used for the quantification of basophil CD203c expression according to the instructions supplied by the manufacturer [20]. We used 10 μg/mL cetuximab, anti-IgE antibody (4 μg/mL) as a positive control, and 0.9% NaCl solution as a negative control. Leukocytes in each sample were then analyzed using a flow cytometer (FACSCanto II, Becton Dickinson Japan, Tokyo, Japan). Basophils were identified by their characteristic forward and side scatter, by the expression of CRTH2, and by the absence of CD3.

Statistical analysis
The non-parametric Wilcoxon signed-rank test was used to assess intergroup differences between two groups in cases where two samples were correlated. We also performed the Spearman rank-order correlation coefficient to evaluate the linear relationship between two variables. All statistical analyses were carried out using GraphPad Prism 5 (version 5.01; GraphPad Software, CA). The P values were two-sided and P < 0.05 was considered statistically significant.

Serum assays for IgE antibody
Serum samples found to be positive for cetuximab-or αgal-specific IgE had antibody titers ranging from 0.2 to 30.0 IU/mL. As for total IgE, antibody titers ranged from Prediction for Cetuximab-Induced Allergy T. Iwamoto et al. 34.2 to 208.4 IU/mL. Statistically significant correlation was observed between the serum levels of total IgE and cetuximab-specific IgE (P < 0.05) (Fig. 1A) as well as cetuximab-specific IgE and αgal-specific IgE (P < 0.001) (Fig. 1C). Regarding the predictive value, 67% of patients who tested positive for IgE cross reactive with cetuximab or αgal were found to have developed HR (positive predictive value: 0.67), whereas 100% of those who did not test positive never experienced HR (negative predictive value: 1.00) ( Table 2). The sensitivity and specificity of the assay were 1.00 and 0.75, respectively. (Table 2).

Basophil activation in response to ex-vivo cetuximab exposure
Basophil activation tests with cetuximab were performed in all 12 patients (Table S1). Positive CD203c basophils were identified by the relatively higher fluorescence intensity of anti-CD203c antibodies in comparison with that of the top 2-3% of unstimulated control cells (negative control). Ex vivo cetuximab exposure was found to increase activated basophil count in only one patient (Patient 1) who had a history of grade 3 allergy; CD203c-positive cells (%) and ΔMFI (increase in median fluorescence intensity against negative control exposure) were 18.0% and 135, respectively (Fig. 2). No other patients, including another patient with grade 3 allergy, demonstrated any such increase in basophil activation upon ex vivo exposure to cetuximab.

Evaluation of IgE-cetuximab binding on basophils
Flow cytometry analysis was subsequently used to measure the IgE-cetuximab binding on basophils. This binding was assessed by observing the decrease in fluorescence intensity of Alexa Fluor 405-conjugated cetuximab affixed to basophils upon IgE dissociation. This dissociation of IgE from basophils was confirmed by staining either with the anti-FcεRI antibodies, CRA1 and CRA2, or with an anti-IgE antibody. As already discussed, CRA1 does not bind directly with the IgE-binding site, whereas CRA2 does. [24] We were thus able to confirm the successful dissociation of IgE by confirming that lactic acid treatment increased the staining levels of CRA2 (P < 0.001) (Fig. 3C and D) and decreased the staining levels of anti-IgE antibody (P < 0.001) (Fig. 3E and F). CRA1 staining levels, on the other hand, remained consistent irrespective of whether lactic acid was applied or not. (Fig. 3A and B). Figure 4A shows the respective levels of binding of Alexa Fluor 405-conjugated cetuximab measured via basophil staining with or without lactic acid treatment in 4 HR patients (patients 1, 2, 3, and 4). The staining levels of cetuximab on basophils in HR patients were significantly decreased by lactic acid treatment (P < 0.05), especially in patients with grade 3 allergy (Fig. 4A and B), whereas this was not observed in patients with no HR (Fig. 4C).

Discussion
Several studies have been conducted in order to predict cetuximab-induced severe HR based on the IgE antibody's recognition of cetuximab's Fab glycosylation site. These studies have demonstrated that the absence of the specific IgE in serum consistently results in negative HR in almost all cases. The same, however, cannot be said for positive test results; hence, the relatively low positive predictive value [12,14]. In our study, we obtained similar results: the positive predictive value associated with the presence of cetuximab-or αgal-specific IgE was found to be only 0.67, whereas the negative predictive value was 1.00, with a good linear relationship shown between the expression levels of the cetuximab-and αgal-specific IgEs in all study patients (Fig. 1C).
Interestingly, we found extraordinarily elevated total IgE levels in the serum of the two patients with false positive results for both cetuximab-and αgal-specific IgEs. This may help to further explain the significant linear relationship observed between the levels of total IgE and cetuximabspecific IgE observed in this study. One may interpret this result to be a probabilistic binding of serum IgE with the structure of either cetuximab or αgal in vitro, due to overall higher total IgE levels in these patients. Indeed, other studies have shown that individuals with higher overall IgE levels are likely to have antibodies specific to common allergens, such as, in this case, bee / wasp venom and rapeseed [25]. However, these dissociations were not considered to be clinically relevant at the time.
To improve the positive predictive value for cetuximabinduced severe HR, we focused our efforts on the precise characterization of the interaction between cetuximab and the IgE present at the site of action most closely associated with IgE-mediated HR, basophils. In one other chimeric monoclonal antibody, rituximab, ex vivo exposure was reported to have induced basophil activation in five patients with HR [26]. Although there have been no reports directly examining the relationship between the occurrence of cetuximab-induced HR and basophil activation tests, cetuximab-induced basophil activation was observed in two out of two patients with αgal-specific IgE and allergies to red meat [19]. In another study investigating the relation between mammalian meatsinduced allergy and basophil activation, 10 out of 12 subjects with IgE specific to αgal had demonstrated clinical evidence of allergy during the mammalian meat challenge, and basophil activation correlated with the appearance of clinical symptoms [27].
In our study, only one of our two grade 3 allergy patients demonstrated increased basophil activation based on CD203c expression due to ex vivo cetuximab exposure. Our other grade 3 allergy patient, however, exhibited no such activation. This latter patient's basophils also exhibited a weak response against the positive control (anti-IgE antibody), which makes it likely possible that they may have simply been a weak responder to this particular basophil activation test. This contention is supported by previous studies, which demonstrated that nonresponders to the basophil activation test using cetuximab as an antigen may indeed exist [19]. Furthermore, in both patients with grade 1 allergy, no basophil activation was detected following ex vivo exposure to cetuximab. We previously demonstrated that one patient with repeated Prediction for Cetuximab-Induced Allergy T. Iwamoto et al. carboplatin administration showed no basophil activation just prior to and after the occurrence of grade 1 allergy [28]. Thus, we find it reasonable to conclude that basophil activation tests may not always be applicable for predicting or diagnosing HR, particularly at lower grades.
Next, we tried to evaluate the difference between patients with and without cetuximab-induced HR in terms of IgEcetuximab binding on basophils. In all four patients who developed HR, the staining levels of Alexa Fluor 405-conjugated cetuximab bound to basophils were found to have been significantly decreased by the dissociation of IgE from basophils ( Fig. 4A and B). However, this was not the case in patients who did not develop HR (Fig. 4C). In the two patients who produced false-positive results via ELISA, we also obtained negative results for cetuximab-IgE binding on basophils. This is the first evidence of its kind to demonstrate that evaluation of cetuximab-IgE binding on basophils might be a more reliable predictor of HR in terms of reducing false-positive results.
The timing of blood sampling may have influenced the ex vivo basophil tests in the two patients with grade 3 allergy. In these patients, the tests were performed more than 44 days after final cetuximab administration. The elimination half-life of cetuximab is approximately 5 days. Cetuximab was therefore presumed to have been almost completely eliminated from circulation in such cases. Consequently, we expect that prior administration of cetuximab had minimal bearing on these results. In addition, IgE-cetuximab binding on basophils was found to have decreased due to dissociation of IgE from basophils one day prior to (12.4-8.5%) and 7 days after the occurrence of hypersensitivity reaction (12.2-8.1%). This result supports the idea that pre-exposure to cetuximab had negligible influence on the results reported in this study.
Recently, it has been reported that the basophil activation test may be much more sensitive than the respective IgE detection methods offered by ELISA or ImmunoCAP, at least in the case of recombinant allergens such as those found in wasp venom [29]. This supports our results that interaction between cetuximab and specific IgE on basophil rather than detection of specific IgE in serum correctly reflect the clinical onset of allergic reaction. Furthermore, evaluation of specific IgE on basophils via flow cytometer has an advantage for clinical application in that the measurement times are usually less than 3 h, whereas detection of antigen-specific IgE via conventional ELISA usually takes 2-3 days [14]. It has been known that, in addition to high-affinity FcεRI, human basophils also express low-affinity IgG receptors, FcγRIIA and FcγRIIB as well as minute amounts of the low-affinity IgG receptor, FcγRIIIB [30]. Although cetuximab, an IgG1 monoclonal antibody, has a possibility to bind to these Fcγ receptors on basophils, the binding rate seems relatively low because of its low affinity and/or low expression on basophil.
Although virtually no premedication can be shown to definitively prevent HR due to cetuximab, the administration of corticosteroids before infusion of cetuximab has been shown to be effective in limiting the occurrence of severe HR [31]. Accordingly, an improved understanding of the pharmacodynamic property of cetuximab on basophils may be helpful for identifying and developing appropriate treatment plans for high-risk patients requiring pretreatment with corticosteroids.
In conclusion, our results strongly imply that quantification of the IgE-cetuximab interaction on basophils is an attractive method for identification of high-risk patients