Cerebrospinal fluid proteomics reveals synaptic and immune dysregulation in treatment-resistant schizophrenia.

IMPORTANCETreatment-resistant schizophrenia (TRS) affects around one-third of individuals with schizophrenia and is associated with substantial morbidity, mortality, and healthcare burden. Despite its impact, the pathophysiology of TRS remains poorly understood, and no validated biomarkers exist to predict or stratify treatment resistance and identify patients who may benefit from earlier second-line interventions. OBJECTIVETo identify cerebrospinal fluid protein signatures associated with treatment-resistant schizophrenia using high-throughput immune-based proteomic profiling, and to characterize their biological relevance and evaluate their predictive potential. DESIGN, SETTING AND PARTICIPANTSThis cross-sectional study analyzed cerebrospinal fluid samples, collected between 2001 and 2024 during routine clinical practice in two centers, of 64 patients with treatment-resistant schizophrenia, and 84 sex-, age-, and center-matched non-treatment-resistant patients. Proteomic profiling was performed using Olink Explore HT, a highly sensitive, multiplexed proximity extension assay capable of quantifying up to 5,416 proteins. Univariate differential expression analysis, Gene Set Enrichment Analysis, and Support Vector Machine modelling were conducted to identify treatment-resistant schizophrenia-associated proteins, explore underlying biological mechanisms, and uncover predictive biomarker signatures. MAIN OUTCOMES AND MEASURESDifferentially expressed proteins, associated functional pathway enrichments, and classification performance of predictive models based on cerebrospinal fluid proteomics. RESULTS1,839 protein assays passed detection threshold in cerebrospinal fluid proteomic analysis using Olink Explore HT. 98 proteins were differentially expressed in individuals with treatment-resistant schizophrenia and 95 displayed consistent fold changes across both study centers. Gene Set Enrichment Analysis revealed that upregulated proteins were linked to immune-related pathways, while downregulated proteins were predominantly enriched in synapse-related processes, including glutamatergic signaling. Multivariate support vector machine modelling identified 64 proteins predictive of treatment-resistant schizophrenia, of which 63 overlapped with the differentially expressed proteins. CONCLUSIONS AND RELEVANCECerebrospinal fluid proteomic profiling indicates activation of immune-related processes and synaptic dysfunction in treatment-resistant schizophrenia, supporting the hypothesis that it represents a neurobiologically distinct subtype of the disorder. The identified candidate biomarker signature may support the development of liquid biopsy-based diagnostic tools and guide targeted treatment strategies. Key Points QuestionAre there proteomic alterations in the cerebrospinal fluid (CSF) of patients with treatment-resistant schizophrenia (TRS)? FindingsUnbiased CSF proteomics identified 98 proteins dysregulated in TRS, implicating pathways of neuroinflammation and synaptic dysfunction. Machine-Learning Validation pinpointed protein features with the strongest predictive value for TRS, of which 63 overlapped with the differentially expressed proteins. MeaningThe intersection of univariate and multivariate results yields a prioritized panel of protein biomarkers with potential to distinguish TRS patients from those responsive to first-line antipsychotic therapy. The identified CSF-based proteomic signature reveals immune and synaptic alterations in TRS and may pave the way for early diagnostic tools and targeted interventions to alter its disease-course.