Studies have found that brain function connectivity predicts the efficacy of antipsychotics in the treatment of first-episode schizophrenia
August 28, 2018 Source: Institute of Psychology, Chinese Academy of Sciences
Window._bd_share_config={ "common":{ "bdSnsKey":{ },"bdText":"","bdMini":"2","bdMiniList":false,"bdPic":"","bdStyle":" 0","bdSize":"16"},"share":{ }};with(document)0[(getElementsByTagName('head')[0]||body).appendChild(createElement('script')) .src='http://bdimg.share.baidu.com/static/api/js/share.js?v=89860593.js?cdnversion='+~(-new Date()/36e5)];Recent studies have found that severe mental illness, including schizophrenia and affective disorder, gradually worsens after the first episode, and repeated episodes can cause irreversible changes in the brain. Therefore, it is of great clinical significance to confirm the diagnosis and effective treatment of the patient as soon as possible, which will reduce the onset and alleviate the tendency of the disease to worsen.
The rapid development of magnetic resonance imaging (MRI) technology in the past two decades has provided a unique opportunity to find biological indicators of schizophrenia. At the same time, using advanced machine learning techniques, researchers have been able to make a more objective diagnosis of schizophrenia. However, compared with chronic patients, a clear diagnosis of patients in the early stages of the disease is still facing many challenges. The study also found that antipsychotic treatment can change the brain of patients with schizophrenia. Therefore, in the case of schizophrenia patients with first episodes and no drug treatment, non-invasive brain imaging technology is used as a biological indicator to diagnose patients, especially the efficacy of antipsychotic drugs can be predicted early in the course of the disease. It has important clinical significance, but research on this aspect is still rare. At present, the precise medicine for first-episode schizophrenia mainly includes two aspects: one is to make a clear diagnosis for each patient, and the other is to predict the efficacy of each patient.
Zhang Xiangyang, Key Laboratory of Mental Health, Institute of Psychology, Chinese Academy of Sciences, analyzed the functional connections between the epithelial and other cortical areas, and used machine learning algorithms to successfully diagnose individualized schizophrenia and predict antipsychotic disease. The efficacy of each patient. In the study, the researchers used a new method—mutual information and correlation methods—to calculate the functional connection between the epithelial and other cortical regions, while nucleus through the resting state function (rs-fMRI) The blood oxygen level-dependent (BOLD) signal between the epithelial layer and the other cortical regions was examined and the correlation between them was analyzed.
The study included 38 patients with first-episode and non-medicated schizophrenia, with an average age of 28.3±10.1 years, and 29 healthy controls, with an average age of 27.7±7.8 years. A structured clinical assessment (SCID) was used to determine the diagnosis of schizophrenia. All patients were treated with a single dose of risperidone for 10 weeks. The dose was increased to 3-6 g/day in the first week, and then the fixed dose was maintained until the end of the study. . Clinical symptoms were assessed before and after treatment using the PANSS scale.
The study used GE's 3T MRI machine to perform structural scanning and resting-state functional nuclear magnetic scanning. FreeSurfer software was used to analyze cortical reconstruction and functional imaging. Desrieux Atlas was used to extract the BOLD average time series of each brain region. There are a total of 68 brain regions, mutual information (MI) and zero-lag correlation of the regional time series BOLD signal between each pair of cortex, using the MATLAB toolkit to calculate functional connections. In addition, MATLAB was used for statistical analysis. For each pair of cortical intervals, the t-test was used to compare the difference between the patient group and the control group, and the false discovery rate (FDR) was used to correct multiple comparisons. Pearson correlation analysis was further used to detect the correlation between mutual information and cortical correlation and PANSS score.
By analyzing the mutual information of the signals paired with different cortical regions, a total of 2278 connections were detected, and the mutual information of the patients was generally reduced. However, only 8 connections were verified by FDR and remained significant (Fig. 1). Interestingly, the mutual information between the temporal epithelium and the dorsal prefrontal, cingulate, temporal and parietal cortex was associated with PANSS positive symptoms and hallucinations, but these significance were not multiplied by FDR.
Because the study found that 7 of the 8 low-MI connections were associated with the sacral epithelium, the investigators further tested whether this cross-information could be used to determine the diagnosis of a single patient. The study uses the mutual information between the epithelial layer and other cortical regions as input parameters, enters a machine learning algorithm called "Support Vector Machine" (SVM), performs SVM classification, and adopts a leave-one-out Mutual verification methods predict whether a person who has never met has schizophrenia. The study model can accurately determine whether a patient is schizophrenia with an accuracy of 77.8%, a sensitivity of 74.4%, and a specificity of 82.8% (Table 1A).
After 10 weeks of risperidone treatment, the investigators further followed the patient. Both the PANSS total score and the three subscale scores were significantly lower. The PANSS total score was reduced by 30% as the improvement criterion, and the regression model was constructed by correlation of cortical functional connections. The results showed that the regression model can predict the percentage reduction of PANSS total score (r=0.69; p<0.0001). Based on this predictive percentage reduction in PANSS, 88.0% of patients with improved treatment were predicted, and 76.9% of patients with no response (balance accuracy of 82.5%) (Table 1B).
The study found that patients with primary and non-medicated schizophrenia had abnormalities in the functional connection of the superior temporal cortex. The support vector machine method is used to input the functional connection from the STC to the machine learning algorithm, which successfully determines the diagnosis of each schizophrenia patient and predicts the therapeutic effect against psychosis. This is the first study in the world to use functional connectivity data derived from resting-state functional MRI to determine the diagnosis of schizophrenia and to predict the efficacy of antipsychotic patients.
The study was funded by the National Natural Science Foundation of China. The paper has been published online at Molecular Psychiatry.
Cardio-Cerebrovascular,Felodipine Tablets,Telmisartan Capsules,Nicergoline Tablets
Shandong Qidu Pharmaceutical Co., Ltd. , https://www.qdyypharma.com