Abstract
Peanut allergy (PA) remains a major diagnostic challenge in pediatric allergy, largely due to the frequent discrepancy between immunological sensitization and clinically relevant disease. This study aimed to develop a real-life diagnostic prediction model to distinguish true peanut allergy from asymptomatic peanut sensitization in children referred for evaluation of suspected PA. In this cross-sectional study, 80 children aged 1-18 years were assessed in a tertiary allergy center in Poland. Sixty-five children with peanut sensitization underwent detailed clinical history assessment, skin prick testing, measurement of serum specific IgE including component-resolved diagnostics, basophil activation testing, and oral food challenges where clinically indicated. Clinically confirmed peanut allergy was diagnosed in 42 sensitized children. In univariate analyses, several clinical and immunological factors were associated with PA, including atopic comorbidities, peanut component sensitization, and basophil activation. Multivariate analysis identified food-induced anaphylaxis and walnut sensitization as independent factors associated with PA. In addition, a penalized diagnostic prediction model was developed to support clinical risk stratification. A multivariable diagnostic prediction model integrating clinical history and laboratory parameters demonstrated good discriminative performance in internal validation (area under the ROC curve 0.83). In conclusion, peanut allergy in sensitized children is determined by a combination of clinical and immunological factors rather than a single biomarker. Integrative diagnostic models may support risk stratification and help optimize the use of oral food challenges in specialized clinical settings, although external validation is required before broader implementation.