The Predictive Value Of Short-term Changes In Hippocampal Metabolites For The Efficacy Of Drug Therapy In Patients With Major Depressive Disorder

Objective1.We have developed and evaluated a localization voxel placement technology that maximizes the overlap between MRS voxels and hippocampus to reduce the signal pollution of surrounding non-hippocampus regions and achieve reliable voxel registration for multiple scans.2.We dynamically detected the metabolism of hippocampus in patients with severe depression by combining quantitative MRS and new voxel placement technology,including N-acetylaspartic acid（NAA）,creatine（Cr）,choline（Cho）,glutamate and glutamine（Glx）,inositol（Ins）.To explore the correlation between the change characteristics of short-term metabolites and the follow-up treatment response,and establish a relatively more specific and stable prediction model to provide reference for breaking the 6-week drug treatment response window.Methods1.Twenty volunteers were recruited in the community to participate in the study,and their time was compensated.All 20 subjects completed baseline MRI scans.Ten subjects used manual voxel placement,and the other 10 subjects used the Maximum Voxel Overlap Location Algorithm（MVOLA）for voxel placement.MVOLA is implemented on the MATLAB platform based on iterative algorithm.In addition,all participants completed follow-up MRI scans one week later to evaluate the reliability of the two voxel placement methods in subsequent scans.Both manual voxel placement and MVOLA are operated by an experienced MR technical expert.The accuracy of voxel placement is defined as the percentage of 3D geometric overlap between each subject’s voxel and the template space voxel（acquired by algorithm）.The reliability of voxel placement is defined as the percentage of 3D geometric overlap between subject spatial voxels at baseline and follow-up.The data obtained from the scan were used to analyze the accuracy of the subject’s voxel placement during the two methods of scanning,and the reliability of the subject’s follow-up scan.2.A retrospective analysis of 74 patients with major depressive disorder（MDD）and 20 healthy controls was conducted.Subjects received magnetic resonance spectroscopy（MRS）combined with MVOLA once a week during the 6-week treatment period to screen for the refractory depression group and the non-refractory depression group’s short-term differences in hippocampal metabolites within and between groups.Then,the relationship between changes in hippocampal metabolites and clinical treatment responses was analyzed,and a predictive model based on logistic regression was constructed.In addition,a validation set（n=60）was collected from another medical center to verify the predictive ability of the model.Results1.MVOLA was able to obtain the maximum overlapping position with the hippocampus.The average overlapping rate of voxel and hippocampus was about 83.05±4.42%（baseline and follow-up）,which was significantly higher than the traditional manual voxel placement（baseline:t=7.43,P<0.001;follow-up:t=7.12,P<0.001）;And it also showed a relatively stable placement effect in multiple scans（the average overlap rate of the two scans was about 95.3±1.8%,t=10.2,P<0.001）and consistency in voxel tissue composition（average gray and white matter composition ratio,both P>0.05）.2.After 2-3 weeks of antidepressant treatment,the differential indicators(d（t Cho）_week0-2、d（t Cho）_week0-3、d NAA_week0-3)were successfully screened.Then,the prediction ability of these three indicators was revealed in the logistic regression model,and it was found that the combination of Cho change and NAA change in 0-3 weeks had the best prediction ability（AUC=0.841,95%CI=0.736-0.946）.In addition,their prediction ability was also confirmed in the validation set（accuracy=85.96%,AUC=0.837）.Conclusion1.Our research shows that MVOLA is a feasible,accurate and reliable method for the placement of representative MRS single voxel based on the hippocampus.2.The prediction model established in this study has high prediction accuracy and great validation effect.It can provide early guidance for the adjustment of the treatment plan for depression,and be used as a checkpoint for predicting the final drug treatment results.