Brain Structure Measurements Predict Individualized Treatment Outcome of 12-Week Antipsychotic Monotherapies in First-episode Schizophrenia

Abstract

Background and hypothesis: Early prediction of treatment response to antipsychotics in schizophrenia remains a challenge in clinical practice. This study aimed to investigate if brain morphometries including gray matter volume and cortical thickness could serve as potential predictive biomarkers in first-episode schizophrenia.

Study design: Sixty-eight drug-naïve first-episode patients underwent baseline structural MRI scans and were subsequently randomized to receive a single antipsychotic throughout the first 12 weeks. Assessments for symptoms and social functioning were conducted by eight "core symptoms" selected from the Positive and Negative Syndrome Scale (PANSS-8) and the Personal and Social performance scale (PSP) multiple times during follow-ups. Treatment outcome was evaluated as subject-specific slope coefficients for PANSS-8 and PSP scores using linear mixed model. LASSO regression model were conducted to examine the performance of baseline gray matter volume and cortical thickness in prediction of individualized treatment outcome.

Study results: The study showed that individual brain morphometries at baseline, especially the orbitofrontal, temporal and parietal cortex, pallidum and amygdala, significantly predicted 12-week treatment outcome of PANSS-8 (r[predicted vs observed] = 0.49, P = .001) and PSP (r[predicted vs observed] = 0.40, P = .003) in first-episode schizophrenia. Moreover, the gray matter volume performed better than cortical thickness in the prediction the symptom changes (P = .034), while cortical thickness outperformed gray matter volume in the prediction of outcome of social functioning (P = .029).

Conclusions: These findings provide initial evidence that brain morphometry have potential to be used as prognostic predictors for antipsychotic response in patients, encouraging the future investigation of the translational value of these measures in precision psychiatry.

Keywords: antipsychotic response; predictive biomarker; structural MRI.

Fig. 1.

Clinical and social outcomes after 12 weeks of antipsychotic monotherapy in the studied sample. The black and red lines indicate Individual and group trajectories, respectively. The slope coefficients for symptoms and social functioning were highly significant.

Fig. 2.

Brain structure measurements predict individualized treatment outcome of 12-week antipsychotic monotherapies. The left side of each panel presents the correlations between the predicted and observed slope coefficients across individuals, and the right side of each panel shows the top features (ie, regions consistently selected at all CV iterations). Regions in red were positively correlated with slope coefficients, while regions in blue were negatively correlated with slope coefficients. (A) Prediction of core PANSS symptoms with gray matter volume as predictors. Note that the result remained highly significant even after removing an outlier at the bottom left of the scatter plot. (B) Prediction of core PANSS symptoms with cortical thickness as predictors. (C) Prediction of social functioning with gray matter volume as predictors. (D) Prediction of social functioning with cortical thickness as predictors.