Oral immunotherapy has garnered increasing attention as a strategy to induce tolerance in individuals with food allergies. Adjunct therapies such as omalizumab, an anti-IgE monoclonal antibody, can enhance both safety and efficacy by mitigating early hypersensitivity events and elevating the threshold for reactivity. Nevertheless, the immunologic mechanisms underlying successful desensitization to specific components like cashew remain incompletely understood. We investigated system-wide and targeted proteomic changes associated with cashew tolerance within an omalizumab-facilitated OIT cohort. We analyzed 22 successfully desensitized participants from a phase 2 trial of omalizumab-facilitated oral immunotherapy. The protocol combined omalizumab administration (weeks 0-16) with dose-escalation OIT (weeks 8-30). Matched plasma samples were collected at baseline (week 0) and post-OIT follow-up (week 52) and analyzed using the targeted 250-plex NULISAseq™ Inflammation Panel and untargeted LC-MS/MS global profiling. Data analysis utilized variance partitioning, paired statistics, and XGBoost machine learning with SHapley Additive exPlanations (SHAP) to evaluate predictive performance and identify distinct biomarker signatures. Initial exploratory analysis utilizing principal component analysis revealed that unsupervised clustering cleanly separated participants into high- and low-inflammation subgroups. Furthermore, variance partitioning and linear regression models with covariates identified biological sex as a primary, pronounced driver of proteomic variance, thereby necessitating a strictly sex-stratified analytical approach. In female participants, targeted analysis revealed a profile consistent with attenuated Th2-driven inflammatory responses and a shift toward immune regulation. This was characterized by significant decreases in IL-4 and CXCL8, alongside decreased TGF-β1—a key immune suppressor—and increased IL-36A, collectively suggesting a profile that supports improved clinical outcomes through dampened allergic inflammation. Machine learning on untargeted proteomics data (AUC: 0.678) identified alpha-enolase (ENO1) and FRMD4B as top downregulated predictors of the post-treatment state. Conversely, male participants exhibited a highly distinct response centered on systemic metabolic regulation and extracellular matrix remodeling. Targeted analysis showed significant upregulation of FGF21 and FGF19, suggesting enhanced metabolic regulation, alongside elevated MMP3, which specifically indicates active extracellular matrix remodeling. Males also displayed altered inflammatory and chemotactic signaling through the notable downregulation of CXCL13 and MERTK. Furthermore, despite their well-known roles in bone remodeling, a decrease in TNFRSF11B coupled with upregulated TNFSF11 (RANKL) indicated important alternative implications for B-cell regulation and overall immune activation. The XGBoost model for males demonstrated high predictive performance (AUC: 0.876), identifying the kynurenine pathway enzyme AADAT (upregulated) and the vascular remodeling protein VASN (downregulated) as the top predictors following the 52-week treatment. Omalizumab-facilitated oral immunotherapy induces sexually dimorphic immunomodulation during cashew desensitization. Females display classical inflammation suppression, whereas males undergo profound systemic metabolic and structural remodeling alongside altered B-cell regulation. Integrating dual-proteomics successfully identified these novel sex-specific biomarkers and immunologic mechanisms, highlighting the absolute necessity for personalized, sex-stratified monitoring strategies in advanced allergy treatments.
