Defining data-driven subgroups of obsessive–compulsive disorder with different treatment responses based on resting-state functional connectivity

Abstract

Characterization of obsessive–compulsive disorder (OCD), like other psychiatric disorders, suffers from heterogeneities in its symptoms and therapeutic responses, and identification of more homogeneous subgroups may help to resolve the heterogeneity. We aimed to identify the OCD subgroups based on resting-state functional connectivity (rsFC) and to explore their differences in treatment responses via a multivariate approach. From the resting-state functional MRI data of 107 medication-free OCD patients and 110 healthy controls (HCs), we selected rsFC features, which discriminated OCD patients from HCs via support vector machine (SVM) analyses. With the selected brain features, we subdivided OCD patients into subgroups using hierarchical clustering analyses. We identified 35 rsFC features that achieved a high sensitivity (82.74%) and specificity (76.29%) in SVM analyses. The OCD patients were subdivided into two subgroups, which did not show significant differences in their demographic and clinical backgrounds. However, one of the OCD subgroups demonstrated more impaired rsFC that was involved either within the default mode network (DMN) or between DMN brain regions and other network regions. This subgroup also showed both lower improvements in symptom severity in the 16-week follow-up visit and lower responder percentage than the other subgroup. Our results highlight that not only abnormalities within the DMN but also aberrant rsFC between the DMN and other networks may contribute to the treatment response and support the importance of these neurobiological alterations in OCD patients. We suggest that abnormalities in these connectivity may play predictive biomarkers of treatment response, and aid to build more optimal treatment strategies.

Abnormal brain functional network dynamics in obsessive–compulsive disorder patients and their unaffected first‐degree relatives

“…Our results were also consistent with Kwak and colleagues who indicated that not only within‐DMN rs‐FC but also functional connectivity between brain regions involved in the DMN were critical and that rs‐FC features in somatosensory‐motor, visual and auditory, and cingulo‐opercular networks were associated with clinical symptom severity improvement. (Kwak et al, 2020 ). We also found within State I that connectivity within motor and sensory networks (CB‐AUD, CB‐VIS, SMN‐VIS) was greater in OCD compared with HC.…”

Neurocircuit models of obsessive-compulsive disorder: limitations and future directions for research

“…119 Importantly, predictive modelling avoids issues with multiple comparisons and low statistical power that arise when comparing multivariate data between clinical groups. 119 Indeed, a handful of recent large-scale studies have conducted data-driven analyses using machine-learning algorithms to investigate subtypes of OCD based on resting-state functional connectivity patterns and the extent to which those data-driven subtypes differ in treatment response to CBT, 120 and to investigate homogeneous subgroups of children with OCD, autism, or ADHD based on integrated measures of brain structure and clinical symptomatology. 121,122 These studies indicate that it is possible to identify subgroups within OCD and also across traditional diagnostic categories that are characterized by different neural alterations associated with different clinical profiles 121,122 or reductions in OCD symptoms in response to treatment with CBT.…”

“…121,122 These studies indicate that it is possible to identify subgroups within OCD and also across traditional diagnostic categories that are characterized by different neural alterations associated with different clinical profiles 121,122 or reductions in OCD symptoms in response to treatment with CBT. 120 A similar analytical approach is planned for the latest phase of our cross-site global study of OCD; we will apply machine learning methods to ''multi-modal fusion'' data (i.e., combined clinical, neuropsychological, and structural and functional neuroimaging data) to identify brain signatures of OCD. 123 Such data-driven approaches will be important for testing the subgroups of OCD related to underlying neurocircuit dysfunctions proposed in neurocircuit models.…”

Identifying Internet Addiction and Evaluating the Efficacy of Treatment Based on Functional Connectivity Density: A Machine Learning Study

“…Due to the vast amount of data, the selection of informative features from tremendous neuroimaging data is necessary in ML studies that are based on neuroimaging (Chiang et al, 2015) since it decreases the computational burden and improves the performance of SVM models. As in previous study, the twosample t-test (p < 0.005) was used to select features in the current study (Suk et al, 2015;Kwak et al, 2020). After the features with discriminative information were selected, we adopted the Fisher score as the feature weight in the SVC model (Germond et al, 2018), which is defined in the following equation:…”