Allergen sensitization in a birth cohort at mid-childhood: focus on food component IgE and IgG4 responses

To the Editor

Standard IgE assays which measure a single allergen extract or component attached to a solid phase (such as ImmunoCAP) and microchip component assays (such as ISAC) represent two strikingly different approaches for measuring IgE antibodies to allergens (1). While the latter approach is commonly used to assess sensitization profiles in cohort studies, the microchip assay may be less sensitive than traditional allergen testing (2). This could have important consequences when trying to characterize the true prevalence of IgE responses in a population, or in situations where low-level IgE to allergen may be important, such as eosinophilic esophagitis (3, 4). We embarked on the present study to characterize the levels and prevalence of sensitization to common allergens in an unselected birth cohort at mid-childhood, to compare findings using a total extract assay (ImmunoCAP) and a microchip component assay (ISAC), and to relate differences in results from the two assays to specific IgE and IgG4 levels.

Project Viva is a longitudinal cohort of mother-child pairs enrolled at initial prenatal visits in urban and suburban areas of Eastern Massachusetts from 1999-2002. Study details and demographics have been previously described, but importantly there were no selection criteria relating to asthma or atopic disease at the time of enrollment of the mothers (5). The study design is available in the supplemental data (Fig E1). Sera from 616 children at the mid-childhood visit (median age 7.7 years) were available for assay. Allergen extract–specific IgE antibodies were measured by ImmunoCAP (Thermo Fisher Scientific/Phadia, Kalamazoo, MI) to 8 inhalants and 5 foods where values > 0.35 kU/L were considered positive. Our findings on sensitization prevalence were largely consistent with results from 6-9 year olds reported in NHANES 2005-06. In particular, sensitization to food, while largely low titer, was common (Fig 1A and Table E1) (6).

Fig 1.

A, IgE sensitization to inhalant and food allergen extracts in 6-10 year olds (n = 616) as measured by ImmunoCAP. B, Comparison of IgE results from microchip assay with results from the extract assay. C, Comparison of food-specific components with ImmunoCAP singleplex and ISAC. Class sensitization (on x-axis) was determined by extract testing and statistics reflect the χ² test with Yates correction.

Samples that were positive for one or more of the 13 allergens were assayed by microchip component assay (ISAC, Thermo Fisher Scientific/Phadia) (3). Results of this analysis are semi-quantitative and expressed in standardized units, with 0.3 units representing the lower limit of detection. For dust mite, cat and peanut there was a convincing correlation between the results of the two testing techniques in terms of the prevalence of positive titers (Fig 1B). This was true even at low levels of sensitization (i.e – class 1 and 2). As a clinical correlation, we evaluated the relationship of the presence of IgE to peanut extract and IgE to component allergens with the answer to the question “Has your child had an allergic reaction to peanuts?” Out of the cohort there were 27 individuals who reported a history of such a reaction and univariate analysis revealed a significant odds ratio with the different testing modalities (Table E2). However, there were some allergens where the agreement between extract and microchip assays was not as strong (Table E3). Focusing on food our analysis showed that the microchip assay detected positive results in less than 20% of the sera that were positive to milk, egg or wheat by extract testing (Fig 1B).

To address the possibility that the relevant components were not included on the microchip we assayed IgE to specific components derived from milk (Bos d 4, Bos d 5 and Bos d 8), peanut (Ara h 1, Ara h 2 and Ara h 8) and hen's egg (Gal d 1 and Gal d 2) with the ‘singleplex’ ImmunoCAP. These results showed that the specific component assays correlated more closely with extract testing for milk and egg as compared to microchip component testing, and this was true even at low-level sensitization (Fig 1C). As an example, testing for specific IgE to Bos d 4 (α-lactalbumin), Bos d 5 (β-lactoglobulin) and Bos d 8 (caseins) was positive for at least one of these components in 68 of the 81 cases that were positive by milk extract testing, whereas the microchip approach detected only a single positive out of these 81 samples despite incorporating the same components. This indicates that the relevant components are present on the microchip, but that the sensitivity of the microchip for IgE to cow's milk or hen's egg is much less than the specific component assays. Of note, our analysis of wheat was limited by the fact that not all of the components used on the microchip were available for the singleplex ImmunoCAP for wheat.

To address the hypothesis that the discrepancy in results between specific and microchip component assays could be explained by the presence of ‘blocking’ antibodies, we assayed specific IgG4 to milk, egg and peanut components in a subset of the sensitized samples. For milk and egg there was abundant IgG4 to the same components that were recognized by IgE (Fig 2A and B). The analysis revealed that milk (& egg) components have a much higher IgG4/IgE ratio (Fig E2), as well as a distinct IgG4/IgE ‘signature’ as compared to peanut components (Fig 2C).

Fig 2.

A, Representative results of samples that were tested for IgE and IgG4 to cow's milk and peanut components by ImmunoCAP, and for IgE to components by ISAC. B, Specific IgG4 levels from samples that were positive to the component-relevant food specific IgE by extract testing where threshold of detection of specific IgG4 was 0.07 μg/ml. Data represents geometric means where error bars are the 95% CI. Values less than the threshold of detection are shown below the dotted line and are not included in the geometric mean. C, Milk (n=20) and peanut (n=23)-specific components were plotted for IgE and IgG4. Where IgE and IgG4 were less than the detection limit they were depicted with the values 0.25 kU/L and 0.04 μg/mL, respectively.

Taken together these results suggest that microchip assays are most likely to be discordant with extract or specific component assays for allergens with high concomitant IgG4 levels, such as milk and egg. Allergen-specific differences in the ratio of specific IgG4/IgE are also consistent with the hypothesis that sensitization to milk and egg occur via a different route than peanut. This latter observation is limited, however, by the fact we do not have detailed dietary history on these subjects to account for consumption differences.

The results described here highlight potential problems with component assays that should be considered when using component-based assays. These include:

1. Component assays are only possible if the relevant allergen epitopes have been identified, purified and included on the microchip or the specific component assay (see www.allergen.org).

2. Quantities of allergen used in assays vary markedly, with estimates ranging from less than 1 picogram on a microchip to 1 microgram or more on a solid-phase singleplex assay.

3. With lower quantities of allergen, i.e.-microchip techniques, the potential effects of IgG4 or IgG1 ‘blocking’ antibodies become much greater (for example, in the case of abundant IgG4 to milk as seen in Fig 2a).

4. With higher quantities of purified allergen, i.e.-singleplex component assays, minor impurities can give erroneous positive signals.

5. When IgE is directed against a minor component of the extract, the component assay can give a result that is much higher than the result with the extract (for example, Ara h 8 as seen in Fig 2a).

This work builds on a number of earlier reports describing allergen-specific IgG4 and performance characteristics of allergen multiplex assays (7-9). We also emphasize that this work does not address the suitability of multiplex component testing for evaluating immediate hypersensitivity reactions (which typically involve high titer specific IgE). The results demonstrate a striking difference between the response to peanut and milk proteins within the cohort. It also demonstrates limitations of microchip assays for determining IgE sensitization to some allergens, a finding that could be particularly important in disease states such as eosinophilic esophagitis (EoE) where skin testing has had little utility but it has been shown that low titer IgE to food allergens may be relevant to diagnosis and management (4).

Supplementary Material

Table E1. IgE sensitization to 8 aeroallergens and 5 foods by ImmunoCAP extract, stratified by sensitization class. In cases with insufficient sera not all allergens were tested.

Table E2: Analysis of relationship between history of peanut reactions and serum testing for peanut extract and specific components to Ara h 1, Ara h 2 and Ara h 8 with microchip (from n = 300). * Reactions were assessed based on the answer to the question ‘has your child ever had an allergic reaction to peanut?’

Table E3. Comparison of IgE to common allergens using extract and ISAC. Classes shown are for extract where class 0 = < 0.36 kU/L, class 1-2 = 0.36-3.5 kU/L, class 3-4 = 3.51-50 kU/L and class 5-6 = > 50 kU/L.

Fig E1. Overview of study design. The number of individuals or sera that were tested are depicted. Sera that were tested are represented in bold.

Fig E2. Component-specific IgG4 to IgE ratio in sera that were positive to the component-relevant food by extract testing. In cases where IgG4 or IgE was not detected a ratio could not be calculated. IgE was normalized with the formula 1 kU/L = 2.4 ng/mL.

Abbreviations

EoE

Eosinophilic esophagitis

ISAC

Immuno Solid-phase Allergen Chip

IgE

Immunoglobulin E

IgG4

Immunoglobulin G4

IL-10

Interleukin-10