Evaluating Copositivity Patterns in Fragrance/Botanical Patch Testing through Hierarchical Clustering and Network...

Background: Fragrances/botanicals are ubiquitous allergens. Patients allergic to one fragrance/botanical are frequently sensitive to other fragrances/botanicals and are typically counseled to avoid all fragrances/botanicals. However, broad avoidance of all fragrances/botanicals may not be clinically necessary.Objectives: We examined copositivity patterns in fragrance/botanical patch testing.Methods: The Mayo Clinic patch test database was queried for pairwise copositivity rates for fragrances/botanicals between 1997 and 2022, representing a total of 43 allergens. Data analyzed included 4706 positive reactions out of 252,485 total patches applied to 15,864 patients. After background correction for general positivity, copositivity rates were organized through unsupervised hierarchical clustering to determine copositivity subgroups and then evaluated through network analysis.Results: After background correction, clustering revealed distinct copositivity subgroups: Fragrance Mix I-Myroxylon pereirae-limonene hydroperoxides-linalool hydroperoxides; Compositae Mix-sesquiterpene lactone-parthenolide; Fragrance Mix II-Lyral; lichen acid mix-treemoss extract; menthol-Mentha piperita; narcissus-dandelion; and Santalum album-trans-anethole-tea tree-lemongrass-clove-turpentine-Rosa damascena-Lavandula-Geranium-Cananga odorata-neroli-bergamot. In addition, there were further isolated intergroup copositivity reactions seen in network analysis.Conclusions: Background correction followed by hierarchical clustering demonstrated the fragrance/botanical group can be divided into multiple copositivity subgroups. Combined with network analysis, patients with a positive patch test to one fragrance/botanical allergen may be preferentially guided to use specific other fragrances/botanicals.