Archives

  • 2018-07
  • 2018-10
  • 2018-11
  • 2019-04
  • 2019-05
  • 2019-06
  • 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2019-12
  • 2020-01
  • 2020-02
  • 2020-03
  • 2020-04
  • 2020-05
  • 2020-06
  • 2020-07
  • 2020-08
  • 2020-09
  • 2020-10
  • 2020-11
  • 2020-12
  • 2021-01
  • 2021-02
  • 2021-03
  • 2021-04
  • 2021-05
  • 2021-06
  • 2021-07
  • 2021-08
  • 2021-09
  • 2021-10
  • 2021-11
  • 2021-12
  • 2022-01
  • 2022-02
  • 2022-03
  • 2022-04
  • 2022-05
  • 2022-06
  • 2022-07
  • 2022-08
  • 2022-09
  • 2022-10
  • 2022-11
  • 2022-12
  • 2023-01
  • 2023-02
  • 2023-03
  • 2023-04
  • 2023-05
  • 2023-06
  • 2023-07
  • 2023-08
  • 2023-09
  • 2023-10
  • 2023-11
  • 2023-12
  • 2024-01
  • 2024-02
  • 2024-03
  • 2024-04
  • 2024-05
  • 2024-06
  • 2024-07

    • Neuroimaging techniques have shown promise in the identifica

    2018-10-23

    Neuroimaging techniques have shown promise in the identification of neurobiological substrates underlying major depressive disorder and other psychiatric illnesses. The limbic-cortical pathways have been identified as the key brain network that may guide treatment for depression (Fu et al., 2013; Mayberg, 2003; Korgaonkar et al., 2014b). A number of baseline MRI measures of both function and structure of these regions have been associated with treatment remission. Fu et al., 2013 It seems likely, however that more than one tract or brain region may be required for effective treatment prediction. Association analyses are helpful to identify candidate biomarkers; however to employ such a biomarker in a clinical setting would need a demonstrated reliable predictive power along with decision points applicable at an individual level. A major challenge in the field, therefore, is to find ways of combining different brain measures to generate an integrated tool to different clinical treatment decisions on. Data driven classification techniques such as pattern classification are promising in combining whole brain neuroimaging measures to get objective information on diagnosis or prognosis (Fu et al., 2008). This approach has successfully been applied using volumetric data to distinguish between unipolar and bipolar depression (Redlich et al., 2014). The framework of the ROC based signal detection analysis utilized in this paper provides a non-biased and completely data driven method to identify regions and cut-points for categorization of treatment outcome. In our analysis, this approach was robust in identifying non-remitters and was replicated in an independent cohort. However, we do note that the cut-points identified at each partitioning stage using the test cohort were not replicated as significant discriminators using the Fisher\'s exact statistic in the validation cohort. We think this is likely due to the differential non-remission rate for the validation cohort. Our validation data does show robust replication of the main goal of the study which was to identify a sub-group of patients for which a combination of measures predicts non-remission with high classification accuracy. This was also supported by our multiple sampling cross-validation analysis. The decision tree based on the regions and their cut-points identified in our approach provides a measure of classification accuracy which is the probability of either remission or non-remission to ADMs. Defining an overall sensitivity/specificity in the context of our analysis is complex and not clinically meaningful, since we are aiming to isolate groups of subjects that can be classified with a high degree of accuracy, a process that may leave a proportion of the group who are, from a clinical perspective, essentially “unclassified”. For example, based on the DTI decision tree (see Fig. 2), patients for whom left CgC<0.63 & right SFOF>0.54 & right SLF>0.5 cannot be reliably classified. In these cases the decision tree and test will not have helped, and usual clinical care would proceed. An analogous situation of such an approach is the application of an altered management pathway for BRAC1/2+patients in screening for breast cancer. For Orphans reason we have carefully reported the raw numbers of patients (to allow for any transparent post hoc analysis), and have preferred to refer to accuracy of classification as defined above. Major depressive disorder is highly heterogeneous, which provides a significant challenge in the management of this disorder (Goldberg, 2011). This heterogeneity was reflected in our analysis, where only a subset of non-remitters was identified using MRI measures. The selected non-remitters were not demographically or clinically different from the non-selected non-remitters except for the marginally lower proportion of those with melancholic features present in the group. Our data supports the use of neurobiological markers to characterize in meaningful ways the heterogeneity in depression not adequately captured using clinical measures (Insel et al., 2010). It would be of interest to further characterize this group of patients using other biological measures (including other forms of imaging, neurophysiological measures, and genetics). In particular, multimodal imaging, combining functional and structural measures offer the most logical approach to further extend out understanding of the features that govern treatment outcomes in this complex disorder (Teipel et al., 2014). Functional data using resting state fMRI (Li et al., 2013), and metabolic information (McGrath et al., 2013), would be expected to add valuable independent information to a treatment-prediction approach as illustrated by our current analysis.