Abstract
Approximately one-third of patients with first-episode schizophrenia-spectrum disorders develop treatment-resistant schizophrenia (TRS) within five years, yet reliable early predictors remain lacking. Routine cerebrospinal fluid (CSF) biomarkers may offer insights into TRS pathophysiology and enable early prognostic stratification. In this longitudinal study, we examined whether baseline CSF parameters-total protein, glucose, and lactate dehydrogenase (LDH)-predicted TRS, defined according to TRRIP consensus criteria and clozapine use. Forty-four patients with first-episode schizophrenia spectrum disorders underwent lumbar puncture during index hospitalization and were followed clinically for five years at the Mataró Mental Health Care Centre. TRS status was confirmed through detailed electronic health record review. Thirteen patients (29.5%) met TRS criteria. At baseline, these individuals had significantly lower CSF LDH concentrations compared to non-TRS patients (p = 0.014), while glucose and protein levels showed no significant differences. In adjusted logistic regression models, lower LDH remained independently associated with TRS (OR = 0.043, p = 0.031). A combined model incorporating LDH and early antipsychotic response achieved an AUC of 0.86, outperforming LDH alone (AUC = 0.73), and demonstrating good discriminative accuracy. Lower baseline CSF LDH concentrations predicted treatment resistance at five years, especially when combined with poor early antipsychotic response. This two-variable prognostic model-BIOERES (Bioenergetic and Early Response Stratification)-may facilitate early identification of high-risk patients and support personalized treatment strategies. Validation in larger, independent cohorts is needed.