Ovomucoid epitope-specific repertoire of IgE, IgG₄, IgG₁, IgA₁, and IgD antibodies in egg-allergic children

Abstract

Background:

Egg-white ovomucoid, that is, Gal d 1, is associated with IgE-mediated allergic reactions in most egg-allergic children. Epitope-specific IgE levels have been correlated with the severity of egg allergy, while emerging evidence suggests that other antibody isotypes (IgG₁, IgG₄, IgA, and IgD) may have a protective function; yet, their epitope-specific repertoires and associations with atopic comorbidities have not been studied.

Methods:

Bead-based epitope assay (BBEA) was used to quantitate the levels of epitope-specific (es)IgA, esIgE, esIgD, esIgG₁, and esIgG₄ antibodies directed at 58 (15-mer) overlapping peptides, covering the entire sequence of ovomucoid, in plasma of 38 egg-allergic and 6 atopic children. Intraclass correlation (ICC) and coefficient of variation (CV) were used for the reliability assessment. The relationships across esIgs were evaluated using network analysis; linear and logistic regressions were used to compare groups based on egg allergy status and comorbidities.

Results:

BBEA had high reliability (ICC >0.75) and low variability (CV <20%) and could detect known IgE-binding epitopes. Egg-allergic children had lower esIgA₁ (P = .010) and esIgG₁ (P = .016) and higher esIgE (P < .001) and esIgD (P = .015) levels compared to the atopic controls. Interestingly, within the allergic group, children with higher esIgD had decreased odds of anaphylactic reactions (OR =0.48, P = .038). Network analysis identified most associations between esIgE with either esIgG₄ or esIgD; indicating that IgE-secreting plasma cells could originate from either sequential isotype switch from antigen-experienced intermediate isotypes or directly from the IgD⁺ B cells.

Conclusions:

Collectively, these data point toward a contribution of epitope-specific antibody repertoires to the pathogenesis of egg allergy.

Graphical Abstract

Egg-allergic children have lower levels of epitope-specific (es)IgA₁ and esIgG₁ but higher esIgE and esIgD compared to the atopic controls. Within the allergic group, children with higher esIgD antibodies have decreased odds of anaphylactic reactions. Network analysis identifies most associations between esIgE with either esIgG₄ or esIgD, indicating potentially different origins of IgE.

1 |. INTRODUC TION

Hen’s egg allergy is one of the most common pediatric food allergies affecting 0.5%−10% of infants and children.^1–9 Allergic reactions occur when antigen-specific immunoglobulin E (IgE) binds to its high-affinity receptors on granulocytes, activating their effector functions and release of inflammatory mediators.^10,11 Several egg-white proteins contribute to IgE sensitization, of which ovomucoid (OVM) is considered a major allergen, designated as Gal d 1 (WHO-IUIS).^12–16 Ovomucoid is resistant to heat and digestive enzymes, and IgE recognition of the OVM protein is indicative of a more persistent allergic phenotype.^15,17–21 Additionally, IgE recognition of a greater number of sequential epitopes is shown to be associated with greater severity of allergic reactions.^22–24

Over the past few years, different groups demonstrated that induction of other antibody isotypes, such as IgA, IgG₁, IgG₄, and IgD, can ameliorate IgE-mediated reactions.^15,25–40 Specifically, several clinical trials of egg oral immunotherapy reported increases of serum antigen-specific IgA, IgG₁, and IgG₄ in patients that responded to treatment.^32,33 Interestingly, the ability of the antibodies to block IgE-mediated responses could be largely dependent on their antigen specificity rather than isotype.³¹ However, recent research in food allergy has been mostly focused on understanding epitope-specific repertoire of IgE and IgG₄, with only a limited number of studies addressing epitope specificity of IgG1, IgA, and IgD.^41,42

We have previously presented a bead-based epitope assay that is highly reliable and sensitive for detecting milk^43,44 and peanut⁴⁵ epitope-specific IgE and IgG4. In this work, we developed an assay to measure levels of five antibody classes, that is, IgE, IgG4, IgG1, IgA1, and IgD, directed at the peptides covering the entire sequence of the OVM protein. We then investigated how epitope-specific antibody repertoires are associated with egg allergy and anaphylaxis, as well as atopic comorbidities, such as asthma, rhinitis, and eczema. To our knowledge, this is the first report investigating the relationships among epitope-specific adaptive humoral responses in pediatric food allergy.

2 |. MATERIALS AND METHODS

2.1 |. Study participants

Plasma from children with specific IgE (sIgE) to the egg-white extract >0.35 kU_A/L (ImmunoCAP) was obtained from the Food Allergy Research Initiative (FARI) biorepository at Mount Sinai. Allergy status was determined by sIgE or skin prick testing and/or a convincing history, with moderate-to-severe, clear-cut allergic reactions following the ingestion of egg, as documented by a patient’s local physician. A negative control pool (a mix of sera from four adults with no history of any food or environmental allergies), a positive control pool (plasma from three children with egg sIgE >100 kU_A/L), and atopic controls with food allergies other than egg were also included. The study was approved by the Institutional Review Board at the Icahn School of Medicine at Mount Sinai. All the study subjects provided informed consent.

2.2 |. Measurement and quantification of epitope-specific (es) Ig levels

A peptide library consisting of 58 15-mer biotinylated peptides (12-mer overlap; CS Bio) was generated covering the entire sequence of hen’s egg white OVM (UniProt ID P01005). The bead-based epitope assay (BBEA) was carried out as described previously,⁴⁵ with a few modifications. A master mix of biotinylated peptides, conjugated to LumAvidin beads (Luminex Corporation), was prepared in PBS-TBN buffer (1xPBS, 0.02% Tween-20, 0.1% BSA), and 100 μL/well of mix was added to 96-well filter plates. Plates were washed twice with PBS-TBN, and 100 μL/well of 1:10 plasma dilution was added in duplicates and incubated on a shaker for 2 hours at room temperature. To account for potential well position and plate effect, samples were randomized to wells using PlateDesigner.⁴⁶ For the nonspecific signal detection, each plate included two wells with only PBS-TBN. Plates were washed and incubated for 30 minutes with 50 μL/well of mouse anti-human PE-labeled secondary antibody (2 μg/mL of IgE, clone BE5, Cat. MA1–10375, Thermo Fisher Scientific; or 0.25 μg/mL of IgG₄, clone HP6025, Cat. 9200–09, Southern Biotech; or 62.5 ng/mL of IgG₁, clone HP6001, Cat. 9054–28, Southern Biotech; or 0.2 μg/mL of IgA₁, clone IS11–8E10, Miltenyi Biotec; or 0.17 μg/mL of IgD, clone IADB6, Cat. OB903009, Southern Biotech). Plates were washed three times, and beads were resuspended in 100 μL of PBS-TBN. The signal was quantified as a median fluorescence intensity (MFI) on the Luminex200 instrument (Luminex Corporation). For each peptide, MFIs were log₂-normalized and converted to nMFI values by subtracting the average of the non-specific binding wells, as previously described.⁴⁵ Since MFI is representative of antibody concentrations,⁴⁷ these epitope-specific immunoglobulin (esIg) nMFI levels represent a relative quantity of an antibody in plasma directed at a particular peptide.

2.3 |. Statistical analyses

The plate effect for the IgG₄ experiment was quantified using the principal variance component analysis (PVCA),⁴⁸ with the principal component threshold set to 0.8. Plate effect was estimated for each peptide by fitting linear models with “plate” as a covariate; then, the coefficients for each plate were subtracted from the original nMFI values.⁴⁵

To assess the reliability of the technical duplicates, a two-way intraclass correlation coefficient (ICC) for agreement was used, with the reliability thresholds defined as excellent (0.75–1.00), good (0.60–0.74), fair (0.40–0.59), or poor (<0.40).⁴⁹ Unsupervised hierarchical clustering of the duplicates was performed using Pearson’s r correlation as a distance metric and an average-linkage algorithm. The coefficient of variation (CV) was calculated for each pair of replicates as (standard deviation [MFI]/average [MFI])*100. After the reliability assessment, the duplicates were averaged for downstream analyses.

An esIg network was constructed from the Pearson’s r using all epitope pairs with the WGCNA algorithm⁵⁰: The epitope-epitope signed correlation matrix was converted to an adjacency matrix using a power function that optimizes a scale-free topology (power = 22). A normalized number of neighbors shared by the nodes on either side of each edge were calculated by transforming the adjacency into a topological overlap matrix (TOM). Then, esIgs were joined using the hierarchical clustering with an average linkage of the TOM and split into modules using a hybrid dynamic tree-cut algorithm (height = 0.99). A network layout was generated using the ForceAtlas2 method in Gephi⁵¹ and visualized using the Cytoscape software.⁵²

The pairwise associations among epitopes of each antibody isotype were measured using Spearman’s ρ correlation and cosine similarity. Antibody-level summary statistics for the pairwise correlation were obtained after using the Fisher Z-transformation of the correlation coefficients and reverse-transformed those summary statistics to Spearman’s ρ.⁵³ The association of esIgE and log₁₀-normalized egg sIgE was calculated using Spearman’s ρ and clustered with mcquitty agglomeration algorithm.

Analysis of the egg-allergic (EA) vs atopic children (AC): Comparisons for each esIg were carried out using linear modeling in the limma framework, where an empirical Bayes approach allows estimation of the variance parameters acquiring information across all esIgs.^54,55 Then, the levels of the significant esIgs (P < .05) were averaged for the downstream Spearman’s ρ estimations; since none of the esIgG₄s were significant, the analyses were performed with the average of all 58 esIgG₄s. Using either the averages of significant epitope-binding antibodies or the mean of all 58, logistic models were fitted to estimate the area under the curve (AUC) for each immunoglobulin. Similarly, analysis of anaphylaxis and the atopic comorbidities, that is, asthma, eczema, and rhinitis, for only EA children was done using limma methodology. Results are summarized as in a Manhattan plot using the -log₁₀ (P-value) for each epitope but including the direction of the association (<0 if negative and >0 if positive). As the resultant plot resembles the skyline of a city by a lake, we have termed “Chicago” plot. Then, for each disease, logistic models were built using a stepwise selection of predictors that included age, atopic comorbidities, and top 5–10 esIgs (based on the smallest P-value). Due to a small sample size, the P-values were not adjusted for multiple comparisons. Statistical analyses were performed in R version 3.2.2.

3 |. RESULTS

3.1 |. Characteristics of the study cohort

Thirty-eight egg-allergic (EA) children, with an average egg-white sIgE of 54 kU_A/L, and 6 atopic controls (AC) were included in the study (Table 1). EA children were on average 6.6 (standard deviation (SD) = 3.6) years old, 22 (57%) were male, and 27 (80%) were White/Caucasian. Twenty-four (63%) children in the EA group had a history of egg-related anaphylaxis, 29 (76%) had eczema, 28 (74%) had asthma, and 22 (58%) had rhinitis. The AC group was slightly older, 10.6 (SD = 3.8) years of age, but comparable in other demographics and medical history.

TABLE 1.

Patients’ characteristics

	AC (n = 6a)	EA (n = 38)	P-valueb
Age	10.6 (3.8)	6.6 (3.6)	.016
Male (%)	3 (50.0)	22 (57.9)	.999
Race (%)
Asian	0 (0.0)	5 (14.7)	.166
Back/African American	1 (20.0)	0 (0.0)
Other	0 (0.0)	2 (5.9)
White/Caucasian	4 (80.0)	27 (79.4)
Not Hispanic or Latino (%)	5 (100.0)	30 (93.8)	.999
Asthma (%)	2 (40.0)	28 (73.7)	.153
Rhinitis (%)	2 (40.0)	22 (57.9)	.64
Eczema (%)	3 (60.0)	29 (76.3)	.589
Anaphylaxis (%)	.	24 (63.2)
Egg sIgE (kUA/L)	0.1 [0.1–0.3]	49.9 [22.8–90.5]	<.001

^aAC group had demographic and medical history data available for five out of six patients.

^bContinuous variables are presented as mean (SD) or median [1st and 3rd quartile]; categorical variables are reported as frequencies and percentages. Groups were compared using Wilcoxon-Mann-Whitney and Fisher’s exact test. AC, atopic controls; and EA, egg allergic.

3.2 |. BBEA to measure antibodies specific to ovomucoid epitopes

The bead-based epitope assay (BBEA) was developed to measure levels of OVM’s esIgs. We previously established in the peanut and milk BBEAs that plate effect (that is, technical experimental artifact common to high-throughput assays) can be detected.⁴⁵ The IgG₄ experiment was performed on multiple plates that included repeats of the same positive and negative pooled samples. The plate effect accounted for 57.2% of the overall variability and samples cluster by experimental date on a PCA plot (Figure S1A). This effect was estimated and eliminated using linear models, and in the adjusted data, the plate effect was contributing to 1.2% of total variability (Figure S1B).

Experimental reliability of the BBEA was assessed using multiple metrics. First, technical duplicates showed excellent agreement based on the intraclass correlation coefficient (ICC) with the average (±SD) of 0.96 ± 0.07 for esIgE, 0.96 ± 0.01 for esIgG₄, 0.99 ± 0.003 for esIgG₁, 0.77 ± 0.05 for esIgA, and 0.97 ± 0.02 for esIgD (Figure 1A), which was consistent across individual peptides (Figure S1C). The average coefficient of variation (CV) was below 20% for all antibodies (Figure 1B, Figure S1D). Additionally, using unsupervised hierarchical clustering, duplicates of the same sample clustered together (Figure 1C).

FIGURE 1.

Characterization of egg-BBEA. A, ICC measured among duplicates and averaged across 58 peptides shows high agreement (ICC >0.75) for all antibody isotypes, presented as mean and SD. B, CV averaged across 58 peptides shows low variability among technical replicates, presented as mean and SD C, Unsupervised clustering presented as a phylogenetic tree of the technical duplicates constructed with the Pearson correlation as distance metric and “average” agglomeration algorithm shows that samples from the same patients cluster together

3.3 |. IgE shows most epitope specificity and is “connected” to esIgG4 and esIgD

Epitope-specific antibody levels, presented as a heatmap (Figure 2A), show heterogeneity among patients. esIgE had greater intra-patient variability with the lowest average epitope-epitope pairwise correlation (ρ = .68) and cosine similarity (0.74, Figure 2B), preferentially recognizing only a few peptides, that is, IgE epitopes. All 58 IgE antibodies were positively associated with the egg white sIgE, with correlations ranging from 0.20 to 0.61 (Figure 2C). Other immunoglobulin isotypes were present at either low or high levels but with similar specificity to all 58 OVM peptides, as demonstrated by an average pairwise correlation of 0.95 for esIgD, 0.87 for esIgG₄, 0.84 for esIgA, and 0.83 for esIgG₁ (Figure 2B).

FIGURE 2.

Epitope-specific immunoglobulin repertoire. A, Heatmap representing the (standardized) levels of five antibody isotypes to 58 peptides (rows, ordered by sequential peptide number) for all 44 patients (columns). B, A median and interquartile range of epitope-epitope pairwise Spearman correlations or cosine similarity calculated separately for each antibody isotype. esIgEs have lowest correlations, potentially indicating epitope specificity C, Barplot showing that all esIgEs have positive Spearman correlation with egg-white sIgE (all P-values < 0.05), ordered by unsupervised hierarchical clustering. D, Left, a dendrogram of the hierarchical clustering of the TOM matrix (colored tips represent module assignment and corresponding classes of the esIgs) shows that aside from esIgE, other antibody isotypes were assigned to the self-containing modules. Right, a multiscale network of esIgs constructed with ForceAtlas2 algorithm and colored by module assignment; larger letters in color (module color for esIgE, blue for esIgA, and pink of the rest) are most connected nodes, that is, module hubs. esIgs in black larger letters highlight epitopes associated with allergy or history of anaphylactic reactions identified in subsequent analysis

The relationship across esIgs was further investigated through the network analysis, which separates distinct modules based on the overall connectivity. Consistent with the previous observations, esIgE antibodies showed the most variability and were split into four modules (green, black, pink, and red), while other esIgs belonged to self-containing groups: brown—esIgA, blue—esIgG₄, turquoise—esIgD, and yellow—esIgG₁ (Figure 2D). There was one exception: 015.esIgG₁ was assigned to the turquoise (IgD) module, while the rest of the 57 esIgG₁ antibodies belonged to the yellow cluster. Interestingly, this esIgG₁ antibody had the most connections to both esIgD (turquoise) and esIgG₁ (yellow) modules, serving as an intermediate link between the two isotypes. The interconnectivity was mostly observed between esIgE nodes and esIgG₄ or esIgD, with no edges connecting esIgA to any other esIg.

3.4 |. Higher esIgE and esIgD and lower esIgA and esIgG1 are associated with egg allergy

To determine which esIgs are associated with egg allergy, we compared repertoires of EA and AC groups. Ten esIgE and 12 esIgD were higher in the EA group (P < .05), while 2 esIgG₁ and 3 esIgA were higher in the AC patients; none of the esIgG₄ reached significance (Figure 3A). The averages of these esIgs were also significantly different among groups (Figure 3B) and provided better group distinction than averaging all 58 epitopes (Figure S2), with AUCs of 0.97 vs 0.92 for esIgE, 0.81 vs 0.69 for esIgD, 0.82 vs 0.62 for esIgA, and 0.80 vs 0.65 for esIgG₁ (Figure 3C).

FIGURE 3.

esIg repertoire in EA vs AC groups. A, Heatmap representing the mean levels of esIgs that were significantly different among EA and AC groups (P < .05): 10 esIgE, 7 esIgD, 3 esIgA, 2 esIgG₁, and none of the esIgG₄. B, Boxplots of the average of the significant esIgs compared using Wilcoxon-Mann-Whitney U test (each point is an individual patient). C, The ROC curves of the logistic models for either the average of only significant esIgs (left) or all esIgs (right) and their respective AUCs in the legend. D, Spearman correlations of the averages of the significant esIgs (and all 58 esIgG₄s) by group; color represents direction and strength of association, and correlation coefficient is a number in bold, followed by a p-value

The associations among esIgs were also different in the EA and AC (Figure 3D, Figure S3A). In the EA children, esIgD was negatively correlated with esIgA (ρ = −0.32, P = .05), while in the AC group, a strong positive trend was observed between esIgE and esIgD (ρ = 0.83, P = .06). Of note, while most patients had some IgD and IgA directed at OVM epitopes, their relative quantities were different: higher levels of one isotype generally meant lower level of the other one, as shown by patients clustering into the opposite groups in Figure S3B.

3.5 |. Lower esIgD and esIgG1 are associated with increased odds of anaphylaxis in egg-allergic patients

Univariate models of esIgs and anaphylaxis, as well as asthma, rhinitis, and eczema, showed different profiles of immunoglobulin isotypes. Eight esIgD and 14 esIgG₁ were lower in patients with a history of anaphylaxis (Figure 4A); and the 017.esIgD antibody was significantly associated with reduced odds of anaphylactic reactions (OR = 0.48, P = .038) in a multivariable model adjusted for asthma, age, and esIgG₁ (AUC = 0.88, Figure 4B). Additionally, nonanaphylactic patients had a positive correlation of esIgD with esIgE (ρ = 0.48, P = .037), which was similar to the AC group, while esIgD and esIgA (ρ = −0.76, P < .001) and esIgE and esIgG₁ (ρ = −0.45, P = .057) were negatively correlated (Figure S4). These associations were in the same direction but with a much lower magnitude in the anaphylactic group.

FIGURE 4.

Associations of atopic comorbidities and anaphylaxis with esIgs in EA patients. A, “Chicago” plot representing –log₁₀(P-value) for univariate associations; values above 0 represent positive association and negative association if below 0. B, Multivariable logistic models selected by a stepwise procedure for anaphylaxis (top) and rhinitis (bottom) and their respective ROCs

Of the atopic comorbidities, EA children with rhinitis had 44 esIgG₄ antibodies that demonstrated significantly greater levels compared to patients without rhinitis; 5 esIgE were higher in children with eczema; and no differences in esIgs were observed in patients with or without asthma (Figure 4A,B). In a multivariable model, only a positive association of rhinitis and esIgG₄ to epitope #054 remained significant, even when adjusted for age and asthma.

4 |. DISCUSSION

Understanding the molecular mechanism of allergic sensitization may help improve diagnostic, prognostic, and treatment approaches in food allergy. Previously, we demonstrated that milk-BBEA and peanut-BBEA have excellent reliability and sensitivity in detecting allergenic epitopes,⁴⁵ providing further insights into the role of IgE and IgG4 repertoires in disease phenotypes⁴⁴ and prognosis of milk oral immunotherapy outcome.⁴³ We have now developed an assay, termed egg-BBEA to measure the levels of five antibody isotypes, that is, IgE, IgG₄, IgG₁, IgA, and IgD to peptides covering the entire sequence of a major hen’s egg white allergen—ovomucoid. We showed that egg-BBEA, similarly to other high-throughput technologies, has plate effects that can be detected and eliminated; and is a reliable assay with high agreement and low variation among technical replicates (Figure 1, Figure S1).

We have identified IgE directed at sequential peptides associated with egg allergy (Figure 3A), constituting immunogenic IgE epitopes previously identified by other groups^15,23–57 and demonstrating sensitivity of egg-BBEA for epitope detection. We also found that EA children had variable but on average higher esIgD, while AC showed greater levels of esIgA and esIgG₁ antibodies (Figure 3A,B, Figure S3). Furthermore, we found that esIgE was highly correlated with esIgD in AC but not EA patients (Figure 3D), which may indicate that while atopic patients have low levels of IgE, it is produced by plasma cells coming from a direct isotype switch from IgD⁺ B cells, resulting in lower-affinity IgE antibodies.⁵⁸ Different antibody isotypes may inhibit IgE-mediated allergic reactions, given they have similar specificity to an antigen,³¹ and in this cohort, EA patients had high egg sensitization with median egg sIgE of 49.9 kU_A/L, which might have rendered antibody repertoire profiles different from patients with lower sIgE levels. However, these results provide basis for further studies that should include broader allergic population and a larger number of control participants.

To identify association patterns across repertoires of antibody isotypes, we applied a network analysis and showed that esIgE had large variability with four distinct modules and low correlations, suggesting higher epitope specificity of this antibody (Figure 2B,D). On the other hand, the rest of the isotypes were assigned into self-containing modules. Interestingly, across antibodies connectivity was observed between esIgE and esIgG₄ or esIgD, while esIgA was disconnected from other modules, and esIgG₁ was only connected to esIgD via one epitope. This may indicate that IgE-secreting plasma cells come from either sequential isotype switch from the intermediate upstream isotypes (ie, IgG₄⁺ B cells) or directly from the IgD⁺ B cells. This concept was shown by several groups that sequenced the heavy-chain loci of the human immunoglobulin gene (IgH) and compared the mutation rates of the variable segment, concluding that high-affinity IgE is a product of sequential isotype switch form “antigen-experienced” B cells of the IgG subclasses; while lower-affinity IgE derived from a direct switch from the IgM/IgD B cells.^58–64

Early food allergy and eczema, especially an extrinsic subtype defined by high IgE levels, are often followed by progression to asthma and allergic rhinitis, commonly referred to as the “atopic march”.^65,66 The development of atopic manifestations progresses with age and having one of the comorbidities is considered a risk factor for the others. Additionally, food-related anaphylaxis was shown to be associated with age as well as the presence of atopic dermatitis.^67,68 We further investigated whether the repertoire of immunoglobulins was associated with atopic diseases and anaphylaxis. While the EA children had higher esIgD than AC (Figure 3B), within the EA group patients with lower esIgD and esIgG₁ had greater odds of anaphylaxis (Figure 4). This finding is consistent with a recent study by Shan et al³⁶ showing that antigen-bound IgD can attenuate IgE-mediated basophils degranulation, potentially increasing immunity against soluble antigens. High levels of esIgG₁ could result in more antigen molecules being cleared through phagocytosis^69,70 rendering them unavailable to IgE. Additionally, while no correlation was observed between esIgE and esIgD in the EA group (Figure 3D, Figure S3A), after stratifying by history of anaphylaxis, this correlation was positive (Figure S4) among children with no history of anaphylaxis to egg, similarly to that of the atopic controls. Zhang et al⁷¹ observed that antigen-specific IgE was correlated with antigen-specific IgD for several aero-allergens, for example, birch pollen, timothy pollen, and cat dander, with higher IgD levels to these antigens in atopic compared to nonatopic patients, which they hypothesized was indicative of similar regulation of the two antibodies.

Of the atopic comorbidities, high levels of almost all esIgG₄ antibodies were associated with rhinitis, and one IgG₄ epitope had significant association in a model adjusted for asthma and age (Figure 4). This association of IgG₄ is not commonly known, but IgG₄ specific to house-dust mite was shown to be on average higher in the rhinitis group than atopic or healthy controls.^72,73 Several esIgE antibodies were associated with eczema; and none of the esIgs reached significance for asthma.

Despite a small sample size and limited clinical information for these biorepository samples, we showed that examining the relationships across several antibody isotypes, including less studied IgG₁, IgA, and IgD antibodies, could distinguish additional endotypes among egg-allergic children at a molecular level. Further investigation of egg epitope-specific humoral immune profiles in patients with egg allergy confirmed by supervised oral food challenges and a control population with serological sensitization to egg in the absence of clinical allergy is likely to provide insight into the potential severity, natural history, and response to immunotherapy in egg-allergic individuals.