Psychosis MRI Shared Data Resource (Psy-ShareD): Partnership Grant

Schizophrenia and psychotic illness affects approximately 1% of individuals; those diagnosed are often unable to function normally within society, as current treatment options are limited in their effectiveness. Schizophrenia causes hallucinations and delusions which profoundly impact the individual's wellbeing, causing huge suffering for patients and families. Although some progress has been made in terms of understanding risk factors and how the brain is affected in schizophrenia, we are still far from certain about what causes it. Modern brain scanning techniques such as Magnetic Resonance Imaging (MRI) have been used to study how the structure of the brain differs in schizophrenia, and much research time and money has been invested in this. However, most of this research has been conducted by individual research groups on relatively small numbers of patients. Schizophrenia symptoms vary a lot between individuals, so we need to run investigations across large numbers of brain scans in order to draw reliable conclusions. Also, we need to consider all stages of the disease, and also consider brain structural differences in individuals at high risk, such as those with a family history of schizophrenia. This will allow us to understand what changes occur in the brain, to cause the highly debilitating symptoms of psychosis. To achieve this, we need to combine data from a large number of previous brain scanning studies, so all researchers can access and conduct analyses on a large collection of brain scans drawn from a wide range of different people.

To address these challenges we propose to combine existing MRI scans collected by researchers worldwide into one freely available database resource (Psy-ShareD). The scans will be anonymised meaning that patients won't be able to be identified. The brain scans will have other information linked to them such as symptom information and also results of memory/attention tests, which will help researchers understand more about how the brain is affected in psychosis. The database will provide an essential resource for many researchers who study schizophrenia and will allow others to undertake new research. It is sorely needed, as currently, no database of this type exists in the UK or internationally. Psy-ShareD will be accessible to all researchers around the world, guidance and instructional materials will also be made available, to help novice users in particular. We will involve people who have experienced psychosis when planning this research. We will seek their advice in the best way to ensure the privacy of patients whose data is included in the database, the best way to use the database and also how to promote the database to other researchers and patients. Psy-ShareD has large potential to impact research, by allowing researchers at all institutions, and at all career stages, to analyse high-quality data across disease stages. This will lead to important insights and progress in understanding what occurs in the brain in patients with psychosis. In addition to the initial 2,500 MRI scans used to establish Psy-ShareD, we will actively seek out other researchers willing to donate their MRI data in the UK and overseas. We will specifically seek MRI data from international contacts and collaborators that can provide relevant MRI datasets in non-European populations, to ensure Psy-ShareD is more representative. Also, we will undertake a range of promotional activities, and make use of social media channels, to maximise visibility and uptake. We want to make sure that all interested researchers know about Psy-ShareD, ensuring this new and valuable resource gets fully utilised.

The Psy-ShareD database will host pre-existing structural MRI data collected at all sites (N=2,481) that will be suitable for region of interest, voxel based morphometry, cortical thickness and surface area analytical approaches. All MRI T1 data included in Psy-ShareD will be linked to standardised clinical, demographic, IQ/cognitive data in patient, at risk and control populations. MRI datasets will be harmonised using software that can remove unwanted variation induced by scanner differences, while preserving biological variability between individuals using an empirical Bayes framework.
For clinical data, founding datasets in patient and at risk cohorts have linked PANSS and SAPS/SANS (21) ratings, which will allow classification of symptoms into 'very mild, mild, moderate, severe, very severe, positive/negative/other symptoms', regardless of the scale used. Cognitive data will be harmonised by creating standardised scores, using data from respective control cohorts for specific cognitive domains, such as working memory, executive function, and social cognition.

The Psy-ShareD database will be hosted using Figshare at King's College London (https://kcl.figshare.com/). The data from all sites will be curated into BIDS format (Brain Imaging Data Structure, https://bids.neuroimaging.io) an international standard for organising neuroimaging data. Data dictionaries will be developed for clinical data ensuring the information is readily transferable between datasets. Curated data will be hosted on the Figshare server and allocated a DOI number ensuring a persistent and durable link to the files. Psy-ShareD data will be used to conduct (and publish) a proof-of-principle study addressing a current knowledge gap in the schizophrenia/psychosis MRI literature. We will test the hypothesis that positive symptom severity, and illness stage, affect grey matter volume in temporal lobe regions (lateral and medial).