Developing inclusive, neurobiologically-informed stratification of clinical dimensions in Obsessive Compulsive Disorder (OCD) as a first step to effective targeted treatment (INISOCD)

Background: Obsessive-compulsive disorder (OCD) is a chronic and debilitating mental health disorder which significantly impacts the quality of life of the diagnosed individual as well as their relatives and caregivers [1,2]. Although evidence-based treatments exist, a substantial proportion of people do not experience significant benefit from these, with an estimated 20-50% of patients achieving long-term symptom remission [3,4]. Poor outcomes are compounded by sociocultural factors, as underrepresented racial groups experience increased misdiagnosis and barriers to treatment [5]. Additionally, there are societal and economic costs; in the UK, the estimated Cost of Illness of OCD, both through direct costs and lost productivity, is over five billion pounds a year [6]. One factor which limits the success of current OCD treatments is the significant variation in symptom presentation between patients. A more personalized treatment approach, recognising subtypes defined by symptom clusters, can substantially improve outcomes [7,8]. However, current stratification methods for OCD are unsuccessful in creating consistent subtypes and fail to demonstrate links between symptoms and treatment success [9,10]. A new model by Shephard et al., [7] has proposed that heterogeneity in OCD can be encapsulated by the occurrence of five clinical dimensions, namely, dysregulated fear, intolerance of uncertainty, sensory phenomena, altered habit formation/ impaired inhibition, and altered reward responsiveness, each linked to an imbalance in underlying neurocircuitry. This model accounts for a greater variation of symptoms between patients as it treats profiles as dimensions rather than categories, and allows for overlapping symptom dimensions within patients. It also accounts for less well-defined symptoms and motivations such as sensory phenomena and incompleteness. Practically, it has the crucial advantage of allowing treatments to be personalized to individuals, chosen on the basis of their clinical dimensions’ associated neurocircuitry. Objectives: In the current project we aim to: i) Validate Shephard et al.’s [7] model in a sample of OCD patients representing a diverse range of symptoms and sociocultural populations. ii) Examine the relationships between the clinical dimensions and activity of associated neurocircuitry. iii) Validate a novel paradigm for assessing sensory phenomena, which we have already successfully piloted in an independent non-patient sample. Hypotheses: 1. Thematic analysis of qualitative interviews will uncover themes corresponding to Shephard et al.’s five clinical dimensions. 2. A structural equation model using five latent variables, representing the five clinical dimensions, will adequately fit a body of self-report data on OCD-related symptoms. 3. Patients’ functional activation and connectivity (measured by source-localised current source density and synchronisation at rest and during symptom provocation) will show evidence of alterations in neurocircuitry associated with their presenting clinical profiles. For example, based on the model [7], we predict that the dysregulated fear dimension will be linked to hyperactivity of the fronto-limbic circuit, and reduced coherence between fronto-limbic and dorsal cognitive circuit hubs. Participants and Sample Participants will be 50 OCD patients and 50 non-clinical control participants, aged between 18-55. OCD patients will have a primary diagnosis of OCD, but no diagnosis of autism spectrum disorders, psychosis, bipolar disorder, or neurological disorders. Excluding these conditions is necessary as they are likely to introduce symptoms and alterations in neurocircuitry unrelated to OCD which could reduce the accuracy of our models. Depression and anxiety will be measured and included as statistical covariates. The sample size strikes an adequate balance between economic feasibility and statistical power. In support of the latter, our pilot data (N = 50) shows strong correlations (r > .50) between our clinical profile provocations and sub-clinical obsessive-compulsive symptoms, providing promising basis for testing associations between clinical profiles and neurocircuitry. Recruitment: i) NHS recruitment will take place through Southern Health NHS Trust, where a band 4 employee will contact patients who fit the inclusion criteria and have previously expressed an interest in participating in research. ii) An advert will be posted on the website of the charity OCD-UK. iii) Recruitment materials will be shared with local (Wessex) community groups, representing underserved populations. By recruiting through community groups, as well as through the NHS and OCD-UK, we expect to increase the ethnic diversity in our sample. The importance of including patients from underrepresented groups cannot be overstated, particularly with regards of race, as there is concern that the low level of diversity in neuroscience research leads to biased findings [11] and research on the influence of culture on OCD symptoms is lacking [12]. Methods and Materials: The study comprises three sessions for OCD patients and two for non-clinical controls: Session 1. The validity of the Shephard model’s [7] clinical dimensions will be assessed through questionnaires and a full qualitative interview. Questionnaires will be: The Yale-Brown Obsessive-Compulsive Scale [13]; the Obsessive-compulsive Inventory – Revised [14]; the Intolerance of Uncertainty Scale-12 [15]; the Not Just Right Experiences Questionnaire – Revised [16]; Obsessive-compulsive Core Traits Questionnaire [17]; Depression Anxiety and Stress Scale-21 [18]. Session 2. We record participants’ EEG during resting state and during a novel paradigm designed to provoke the sensory phenomena dimension (see pre-registration for details: https://doi.org/10.17605/OSF.IO/KC4FQ), and a behavioral inhibition task. Initial pilot data (N = 50, sample independent of current proposal) for our novel paradigm shows strong significant correlations between induced sensory phenomena and trait obsessive-compulsive symptoms. Session 3. We record participants’ EEG during a resting state and provocation of dysregulated fear and intolerance of uncertainty dimensions using a modification of Mataix-Cols et al.’s [19] symptom provocation task. Data Analysis We will assess the validity of clinical profiles by applying thematic analysis to the qualitative interview data. We will assess whether the five clinical dimensions are consistent with themes arising from a clinical sample. Comparisons will be made between themes occurring in each patient and scores on self-report questionnaires from session 1 to confirm the validity of these questionnaires in assessing clinical dimension phenomenology. If these relationships are confirmed, the questionnaire scores will be used as measures of each clinical profile. We will fit a structural equation model, with five latent variables representing the five dimensions. Responses to the questionnaire items will function as indicators for these five latent variables. We will use item factor loadings and model fit indices to determine model fit. Subsequently, we will calculate factor scores of the five dimensions for each participant. Next, we will test associations between clinical profiles and neurocircuitry activity. Our EEG data will be nested in resting and provocation state conditions and subsequently in participants. We will utilise linear multilevel models which can appropriately model such nested data. The previously obtained factor scores of the five dimensions will be predictors for neurocircuitry activity, defined in the first instance as the source-localised EEG power in each neurocircuit (fronto-limbic; sensorimotor; ventral cognitive; ventral affective; dorsal cognitive), and in the second model as lagged phase synchronization between the hub regions of each neurocircuit. We hypothesize that clinical dimension scores predict congruent neurocircuit activity (e.g., the dysregulated fear profile will predict fronto-limbic circuit activity, and reduced synchronization between fronto-limbic and dorsal cognitive hubs). The inclusion of control participants is imperative in this analysis. We hypothesize we can distinguish OCD patients from control participants based on neurocircuit activity. We expect increased neuro activity in OCD patients (vs. controls) in a neurocircuit when that OCD patient scores high on a congruent OCD profile. We expect no difference in neuroactivity in OCD patients (vs. controls) in non-congruent neurocircuits. Hence, a three-way interaction between 1) OCD profile, 2) participant (OCD patient vs. control) and 3) neurocircuit (congruent vs. non-congruent) on neurocircuit activity should emerge. Finally, we will exploratory run models which include demographics (e.g., ethnicity), measures of associated mental health issues (e.g., depression), and other clinically relevant predictors (medication status, age of onset), to test the robustness of our findings. Environment Qualitative data collection will take place either at the University of Winchester, online (e.g. via Teams), or at a public space chosen by the participant. EEG data collection will take place in the EEG Lab at the University of Winchester. Ethics and Research Governance All protocols and procedures will be approved by the NHS ethics committee. Exploitation and Dissemination In addition to peer reviewed outputs and conference presentations, the project findings will be disseminated through the NIHR ARC Wessex and at a Celebration Event for all contributors. Lay summaries of findings will also be hosted on the project website, which will be publicly accessible. The findings will form the basis of a future grant proposal to develop a digital tool for stratification which will be rolled out in several larger settings to test its ability to predict clinical outcomes. DOI List 1. https://doi.org/10.1