| Objective: Major depression disorder(MDD)is a mental disease with high prevalence,high recurrence rate and high suicide rate.However,the diagnosis of mental illness still depends on the description of subjective symptoms and lacks objective biological markers.Functional magnetic resonance imaging(f MRI)is widely used in the exploration of brain dysfunction in the field of psychiatry,but there is no consistent conclusion about the core imaging damage of MDD.Previous studies suggested that the frontal-posterior functional imbalance of emotion processing system based on amplitude of low frequency(ALFF)may be the potential neurobiological mechanism of MDD.By combining machine learning approach,the heterogeneous MDD population can be divided into homogeneous neuroimaging subtypes.The subtyping strategy based on reproducible neuroimaging biomarker is helpful to guide the diagnosis,differential diagnosis and treatment of MDD.At present,the first-line clinical treatment strategy of MDD is still limited to traditional drug treatment,well its precise neural mechanism remains unclear.Transcranial magnetic stimulation(TMS)is widely used in the treatment of MDD as a green and non-invasive neuromodulation approach.At present,the most commonly used treatment strategy is to deliver high-frequency TMS targeting the left dorsolateral prefrontal cortex(dlPFC).Current studies mainly focus on the accurate localization of dlPFC based on brain imaging.However,the core damage of different MDD subtypes may be located in seperate brain regions.By exploring the core neuroimaging deficits of different MDD subtypes and combining machine learning methods to determine precise treatment targets in the whole brain,it is helpful to achieve precise TMS of MDD.Although depression may occur at any stage of life,up to 40% of MDD patients will experience the first onset from adolescence to early adulthood.However,the clinical diagnoses of depressed youths are confounding: depressive symptoms and anxiety symptoms are inseparable;moreover,depressed youths usually show more complex clinical manifestations such as suicidal tendencies.Based on the consideration of clinical efficacy and other factors,the current first-line treatment strategy for depressed youths is still unknown.TMS has been applied to a certain extent in the clinical treatment of depressed youths.However,the existing research has limitations,and innovative treatment strategies are urgently needed to improve its clinical efficacy.To sum up,combining brain imaging and machine learning,this study aims to explore the objective neurobiomarkers of MDD,based on which the study will build a subtyping framework for MDD,develop precise TMS targets of different MDD subtypes,and validate the subtyping and precise TMS strategy in depressed youths.Methods:Study 1: This study recruited a total of 304 subjects aged from 13 to 55 years from the People’s Hospital of Wuhan University,including 238 MDD outpatients and 66 healthy controls(HC).The whole-brain voxel-wise ALFF was calculated based on resting-state f MRI data,and the automatic anatomical labeling(AAL-90)template was used for ALFF filtering.The filtered ALFF of MDD patients is dimensionally reduced and clustered by Gaussian mixture model algorithm.The obtained MDD neuroimaging subtypes and HC were subjected to two-sample t-test of the whole-brain voxel-wise ALFF,and multiple comparison correction was performed using Gaussian random field.The neuroimaging characteristics,stability,reliability and diagnostic performance of MDD subtypes were used for validation.Study 2: Based on the subjects included in study 1,this study further collected structural MRI data.In the process of precise TMS target selection,the AAL-90 template was used to extract the ALFF of 90 brain regions of all subjects.For the MDD subtype obtained in study 1,the support vector machine classifiers of each subtype and HC are established by taking 90 brain regions ALFF as features respectively;for each subtype,90 classifiers can be obtained.Based on the performance of the classifiers,the top 10 brain regions are selected as candidate targets,and the dominant features of each subtype are further selected as their precise TMS targets based on the neurobiological basis.Further combine the electroencephalogram international 10-10 system and structural MRI,a precise TMS target localization strategy can be developed in a personalized way.Study 3: In this study,72 depressed youths aged from 13 to 24 years were recruited from the Early Intervention Unit of Nanjing Brain Hospital affiliated Nanjing Medical University for a two-week precise TMS(Chi CTR210045391).The patient needs to be diagnosed with mood disorder and is in the stage of a major depressive episode.The study can be divided into three time points: baseline T0,midpoint T1 and endpoint T2;during each timepoint the patients underwent MRI scanning and clinical evaluation.After the baseline MRI scan,based on the results of study 1 and study 2,the subtyping and precise TMS strategy was carried out.For each subtype,the neuroimaging outcome was evaluated by paired t-tests of ALFF post-pre treatment;the symptomatic outcome was evaluated by the 17-item Hamilton Rating Scale for Depression and the 14-item Hamilton Rating Scale for Anxiety.Results:Study 1: This study identified and validated two neuroimaging subtypes of MDD based on frontal-posterior functional imbalance.Specifically,143 patients identified as archetypal depression had significantly increased ALFF in frontal regions and significantly decreased ALFF in posterior regions,whereas 95 patients identified as atypical depression exhibited significantly decreased ALFF in frontal regions and significantly increased ALFF in posterior regions.The stability of archetypal depression is 0.93,and that of atypical depression is 0.73.The reliability of the two subtypes was statistically significant(p<0.002).The classification performance of the two subtypes is better than the total sample of MDD.Study 2: The dominant features of the top 10 brain regions of the archetypal subtype were located in the prefrontal lobe,and that of the atypical subtype was located in the occipital lobe.The precise TMS target of the archetypal subtype is dorsomedial PFC(dm PFC),and that of the atypical subtype is occipital cortex(OCC).AFz and Oz in the international 10-10 system are used as the reference points of the two subtypes,and the anatomical distances based on structural MRI are further calculated to achieve precise TMS positioning of different subtypes.Study 3: For depressed youths included in this study,16 patients were identified as archetypal depression and 56 patients were identified as atypical depression.After 2-week precise TMS,the frontal-posterior functional imbalance of both subtypes was significantly improved.In terms of clinical symptoms,for archetypal depression,the response rate reached 90.0%,and the remission rate reached 30.0%;for atypical depression,the response rate reached 70.7%,and the remission rate reached 46.3%;both subtypes showed a decrease in suicidal tendencies.Conclusion: By combining brain imaging and machine learning methods,this study identified and validated the frontal-posterior functional imbalance as an objective neurobiomarker of MDD,based on which this study identified archetypal and atypical neuroimaging subtypes of MDD,developed precise TMS targets of both subtypes,and validated the subtyping and precise TMS strategy in depressed youths through neuroimaging and symptomatic effectivenesses,hence building a precision medicine framework for depression. |
PDF Full Text Request