Unsupervised Clustering of children with positive c-sIgE
Among preschool children with at least one positive c-sIgE (n=61), three clusters (clusters 1-3) of different size were generated: Cluster 1 (C1, n=4, 6.6%), with “multiple sensitizations, mainly to grass pollens and pathogenesis–related protein family 10 (PR-10)”, Cluster 2 (C2, n=4, 6.6%), with “multiple sensitizations, mainly to food, grass pollens, animal dander, and nsLTP”, and Cluster 3 (C3, n=53, 86.9%), with ”few sensitizations, mainly to HDM” (Sup Fig 5). The distribution of SRW within the three clusters did not differ, but three of the four patients of Cluster 2 had SRW. Overall, lung function parameters were similar between the three clusters (Table 3).
Among school-age children with positive c-sIgE (n=128), four clusters (clusters 4-7) were generated: Cluster 4 (n=4, 3.1%), with “multiple sensitizations, mainly to grass pollens, HDM, PR-10, and nsLTP”, Cluster 5 (n=6, 4.7%) with “multiple sensitizations, mainly to airborne allergens, including grass pollens and HDM”, Cluster 6 (n=24, 18.8%), with “multiple sensitizations, mainly to grass pollens, HDM, and PR-10”, and Cluster 7 (n=94, 73.4%) with “few sensitizations, mainly to HDM” (Sup Fig 6). All four patients from Cluster 4 had SA, vs 33% in Cluster 5, 25% in Cluster 6, and 34% in Cluster 7 (p=0.036). Lung function parameters were comparable between the four clusters (Table 4, Sup Table 5).