A high-resolution plasma proteomic framework for stratifying treatment resistance and ultra-resistance in...

Biological mechanisms underlying heterogeneous antipsychotic response in schizophrenia remain incompletely understood. In this exploratory cross-sectional study, we applied deep data-independent acquisition LC-MS/MS plasma proteomics to 56 adults initially enrolled (14 per group), including healthy controls and patients meeting TRRIP criteria for treatment-sensitive (TSS), treatment-resistant (TRS), and ultra-treatment-resistant schizophrenia (UTRS). Following proteomic quality control and exclusion of outliers, 52 individuals comprised the final analytical cohort. Proteomic data were analyzed using a layered strategy integrating covariate adjustment, variance partitioning, mediation analysis, monotonicity filtering, LASSO stability assessment, and redundancy reduction. More than 1400 plasma proteins were quantified; 450 differed across groups before adjustment, and 310 reviewed proteins remained significant after covariate modeling. A sensitivity-associated profile distinguishing TSS from controls (CRP, CCDC62, FBXW7, GULP1, CALD1, COPS6) was consistent with lower inflammatory tone and relative preservation of proteostatic and cytoskeletal regulation. In contrast, a resistance-associated profile separating TRS/UTRS from TSS (SHROOM1, MYH7, ABR, EZR, SERPINF2) converged on cytoskeletal organization, actin-membrane dynamics, and extracellular regulatory processes. Directionally concordant but quantitatively amplified changes were observed in UTRS relative to TRS, although multivariate separation between resistant subgroups was limited after full covariate adjustment. Several proteins enriched in resistant groups corresponded to intracellular or nuclear factors rarely detected in plasma and require cautious interpretation. Overall, these findings are compatible with a progression-like molecular pattern in which treatment sensitivity and resistance may reflect shifts in cellular adaptability and structural regulation. Replication in larger and longitudinal cohorts is required.