Molecular Heterogeneity in Clinically Ambiguous Atopic Dermatitis-Psoriasis Overlap: Compartment-Resolved Transcriptomic Profiling

Atopic dermatitis (AD) and psoriasis (PsO) are chronic inflammatory skin diseases with distinct features, but clinical and molecular overlap exists. These ambiguous phenotypes complicate diagnosis and treatment, and their immunologic basis remains poorly defined. This study aimed to delineate compartment-specific immune profiles in AD-PsO overlap lesions and explore candidate markers associated with molecular stratification.

Spatial transcriptomic profiling (GeoMx) was performed on 14 biopsies (AD = 4, PsO = 4, clinically ambiguous AD-PsO overlap = 6), followed by immunohistochemical (IHC) analysis of candidate markers in an expanded cohort (n = 35; AD = 9, PsO = 9, overlap = 17) to support the transcriptomic findings.

AD-PsO overlap lesions demonstrated intermediate Th2 and Th17 signatures relative to AD and PsO and showed relative alignment toward AD-like or PsO-like molecular patterns rather than forming a single uniform state. Patient-level analyses showed distinct pathway profiles, with AD-like alignment associated with epidermal barrier-related pathways and dermal neural/sensory signaling, whereas the PsO-like alignment was associated with Th1, Th17, interferon, IL-15, and neutrophil-related pathways. Candidate genes associated with AD-like alignment included CCL13, CCL19, P2RY4, CXCR4, IL4I1, and EBI3, whereas MAVS and C1orf43 showed relative association with PsO-like alignment. IL36G, which was not used for alignment assignment, showed higher staining in PsO and PsO-like overlap cases than in AD and AD-like overlap cases, providing additional biological support for the observed molecular alignment.

AD-PsO overlap lesions are molecularly heterogeneous and align toward AD-like or PsO-like inflammatory patterns. Compartment-resolved profiling provides a hypothesis-generating framework for molecular characterization of clinically ambiguous cases that merits validation in larger independent cohorts.