gms | German Medical Science

Background: Common available test systems for children with allergies comprise specific antibody (IgE) testing based on whole extracts. Containing a mix of different proteins, the use of component-resolved-diagnostics is specifically testing for major and minor allergen components. The immunoregulatory profile can be described using cytokine expression data. Therefore, we aimed to identify reproducible cytokine-based clinical phenotypes, correlate them with sensitization profiles, and describe them on a clinical level.

Methods:

Data source: Blood from 49 asthmatic children (median age 9.4 years) was analysed for specific IgE of 24 whole allergens and recombinant allergen components (major/minor allergen). Detailed population characteristics from questionnaires and clinical visits were available including information on allergic rhinoconjunctivitis and related symptoms in the past 12 months. Furthermore, cytokine expression data from these children and 188 distinct children from the same cohort (validation data) were available.

Data analysis: We used an unsupervised approach to identify phenotypes based on cytokine expressions. After data pre-processing (imputation and feature selection) and parameter optimization, we fitted k-means clustering. We validated the clusters on the validation data and compared them on the level of basic clinical characteristics and sensitization profiles. For the latter, a permutation test comparing binary distances of a sample to all samples of other clusters was programmed.

Results: Clustering of children based on cytokine expression yielded three distinct groups. The first group with highest cytokines showed highest sensitization to the major allergen of birch and apple combined with less symptoms of rhinoconjunctivitis and a significantly lower lung function. The second cluster with low IFN-g (a marker cytokine for Th1-cells) was sensitized more often against perennial allergens as compared to the other groups and had more often symptoms of allergic rhinoconjunctivitis. The third cluster with intermediate cytokine secretion was composed of younger children with a higher proportion of food allergy and levels of IgE. Those clusters were reproducible on a distinct larger data set from the same cohort.

Conclusions: We identified novel clinical allergy phenotypes based on cytokine expression and were able to validate them. The results may suggest a more specific clinical supervision of the first cluster (lower lung function) and the third cluster (younger, more food allergy).

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.