Symptom Network Dynamics during Antipsychotic Treatment in First-Episode Psychosis

Abstract

BACKGROUND AND HYPOTHESIS: Treatment response in first-episode psychosis varies substantially, yet underlying factors remain poorly understood. Symptom network theory suggests that inter-symptom relationships may influence treatment response. We hypothesized that symptom networks at baseline, as well as dynamic interactions over time, would differ between remitters and non-remitters, and that specific antipsychotics would show differential symptom-targeting effects. STUDY DESIGN: We compared baseline and temporal symptom networks between remitters (n = 250) and non-remitters (n = 196) from the OPTiMiSE trial using 21-item Positive and Negative Syndrome Scale (PANSS) data. Baseline networks were estimated using Gaussian graphical models and compared with the Network Comparison Test. Temporal networks across baseline, week 2, and week 4 were modeled using Cross-Lagged Panel Network analysis. Key symptoms were identified by in- and out-prediction values. Group differences were assessed via non-zero edge weight correlations and Jaccard Index (JI). Network Intervention Analysis was used to examine differential effects of continuing amisulpride versus switching to olanzapine in non-responders (n = 85). STUDY RESULTS: Baseline networks did not differ between outcome groups. However, temporal networks showed substantial differences: remitters and non-remitters had minimal overlap in symptom connections (baseline→week 2: JI = 0.014; week 2 → week 4: JI = 0.055) and virtually no correlation in connection strengths (baseline→week 2: r = -0.089, P = .447; week 2 → week 4: r = 0.005, P = .968). Key nodes (highest in/out-prediction) differed between groups. No robust symptom-specific medication effects emerged. CONCLUSIONS: Temporal symptom dynamics, rather than static baseline relationships, differentiate response trajectories and could inform future research on early markers of non-remission. Absence of antipsychotic-specific effects suggests generic treatment mechanisms.