Development of a Conceptual Disease Model of Atopic Dermatitis Using Data from the Phase 3 ARCADIA 1 Trial.

INTRODUCTIONPatients with atopic dermatitis (AD) often experience severe skin disease, intense itching, sleep disturbance, and impaired quality of life (QoL). This study aimed to generate a clinically credible, data-driven conceptual disease model (CDM) of AD using baseline data from the phase 3 ARCADIA 1 trial of nemolizumab in patients with moderate-to-severe AD (NCT03985943).

METHODSLatent constructs were defined for skin severity, itch severity, sleep disturbance, and QoL/health status using validated clinical outcome assessments (COAs). Baseline COA data were obtained for the adult intent-to-treat population in ARCADIA 1. Structural equation modelling was conducted to assess direct and indirect relationships among AD domains. Factor loadings were calculated to evaluate COA-to-latent construct relationships, and regression coefficients were calculated to ensure that constructs had clinically interpretable directionality. To achieve consensus on the final model structure, model development and refinement were complemented by a modified Delphi process. Dermatologists with extensive experience managing moderate-to-severe AD and conducting AD clinical trials iteratively reviewed COA selection and latent construct definitions and proposed directional pathways.

RESULTSCOA data from 807 adults with moderate-to-severe AD were used to generate the CDM. In the model, itch severity has a substantial impact on QoL, exerting both a direct effect (β = 0.427) and an indirect effect via sleep disturbance (itch severity → sleep disturbance: β = 0.702; sleep disturbance → QoL: β = 0.431). A significant bidirectional relationship between skin severity and itch severity is also observed (β = 0.367). COAs show strong loadings on their respective latent constructs.

CONCLUSIONSThe CDM identifies itch severity as a central driver of the patient burden in moderate-to-severe AD and supports prioritization of itch and sleep disturbance in clinical assessment, therapeutic planning, and outcome evaluation in AD. Our model, whose validity is strengthened by expert consensus, can potentially guide clinical decision-making and evaluation of treatment responses across key domains of AD burden.

Abstract from PubMed