| Summary: | 碩士 === 國立臺灣大學 === 醫療器材與醫學影像研究所 === 105 === Background: Antipsychotic drugs are the standard treatment for schizophrenia; however, the treatment outcomes vary. According to the practice guidelines for schizophrenia, at least 3 to 6 weeks are required before treatment response can be determined. Previous studies revealed that different treatment outcomes may be attributed to the genetic and molecular heterogeneity of patients, which may be represented in the white matter structures of the brain. Therefore, a brain correlate that is characteristic of the treatment outcomes is desirable. This study was divided into two parts: Study 1, we aimed to demonstrate that the remission and non-remission groups had considerably distinct alterations in white matter tract integrity. Study 2, we proposed an approach to predict treatment response in each individual drug naïve patient.
Study 1: Ninety-one patients with schizophrenia (remitted, 50; non-remitted, 41) and 50 healthy controls in study, we performed diffusion spectrum imaging and the whole brain tract-based automatic analysis to investigate the relations between white matter microstructures and remission state in patients with schizophrenia. Results showed that 4 association fibers (bilateral fornices and bilateral uncinate fasciculi) and 2 commissure fibers (callosal fibers connecting the temporal poles, and hippocampi) had significantly different GFA values among the 3 groups. Post-hoc analysis showed that the non-remission group had lower GFA values in all 6 tracts than did the control; the remission group had lower GFA values than the control group only in 4 tracts (bilateral fornices and callosal fibers connecting the temporal poles, and hippocampi). Compared with the remission group, the non-remission group had lower GFA values in all 6 tracts. These results suggest that the remission and non-remission groups show considerably distinct severities of white matter tract alterations and it might be a potential prognostic marker for the symptomatic remission in patients with schizophrenia.
Study 2: The study involved a training group (123 patients with schizophrenia; male, 54; age, 31.9±8.7 years) and a testing group (25 drug-naïve patients with schizophrenia; male, 12; age, 26.5±5.9 years). Diffusion images were analyzed by MAP-MRI to produce 7 diffusion indices including GFA, axial diffusivity (AD), mean diffusivity (MD), radial diffusivity (RD), non-Gaussianality (NG), NG in orthogonal direction (NGO), and NG in parallel direction (NGP). The analysis was performed at each step of the tract-specific profiles and across all 7 diffusion indices. Each diffusion index was analyzed by a general linear model with age, sex and treatment response as independent variables and the steps that showed significant correlation with treatment response were selected as the predictors in a quadratic discriminant model. Results showed that the model was successful about 80% in correctly predicting the true response on individual drug-naïve patients with schizophrenia subject basis. The results imply that individualized prediction of treatment response in schizophrenia is feasible based on alterations of the whole brain white matter tracts. Diffusion indices in specific segments of the white matter tracts could serve as potential imaging biomarkers for predicting treatment response before antipsychotic treatment.
Conclusion: This thesis aims to find biomarkers of treatment response in schizophrenia. We approach the aim from group comparison to individualized prediction. In combination with clinical measures, the prediction model could be useful in treatment planning for each individual patient.
|
|---|
