Trajectory analysis of BMI increase induced by second-generation antipsychotics in first-episode schizophrenia: a secondary analysis based on CNFEST****

Antipsychotic-induced weight gain (AIWG) exhibits marked heterogeneity. We conducted a secondary analysis of the Chinese First-Episode Schizophrenia Trial, leveraging frequent body mass index (BMI) measurements over 12 months. Our aims were to identify latent BMI trajectories in first-episode schizophrenia (FES) patients treated with second-generation antipsychotics (SGAs) and to explore predictors of trajectory membership. Subjects in this study were from the Chinese First-Episode Schizophrenia Trial (CNFEST). After quality control, a total of 361 drug-naïve FES patients treated with olanzapine, risperidone, or aripiprazole were included. BMI was measured at 7 timepoints over 12 months. Latent class trajectory modeling (LCTM) was used to identify distinct BMI trajectories. Multinomial logistic regression was applied to detect predictors of trajectory membership. Four BMI trajectories were emerged, including Low Baseline BMI with Rapid Increase (LBRI) (6.1%, +3.5kg/m² within the first 3 months), Moderate Baseline BMI with Gradual Increase (MBGI) (33.8%, steady rising during 9 months), and Low/High Baseline BMI with Slight Increase (LBSI/HBSI) (46%/14.1%, Minimal change (<1.5 kg/m²)). Baseline BMI (χ² = 144.5, p < 0.001) was the strongest predictor of the LBRI trajectory. A numerically higher, though not statistically stable, odds were observed for olanzapine vs. aripiprazole (OR = 20.4, 95% CI = 2.48-166.67). Shorter duration of untreated psychosis (DUP < 1 year) (OR = 4.12, 95% CI = 1.31-12.93) and lower education (OR = 5.40, 95% CI = 1.19-24.52) further increased LBRI risk. A high-risk subgroup (LBRI) with rapid early weight gain was identified, driven by olanzapine use, shorter DUP, and lower educational attainment. These findings advocate for dynamic risk stratification and early preventive interventions in vulnerable FES patients (Trial Registration: This trial was registered at ClinicalTrials.gov (Identifier: NCT01057849) on January 26, 2010).