Accelerating Medicines Partnership ^® Schizophrenia (AMP ^® SCZ): developing tools to enable early intervention in the psychosis high risk state

Schizophrenia is a severe mental illness that presents with pos­itive, negative and cognitive symptoms and ranks among the top 15 leading causes of disability worldwide1. Signs of risk for developing this illness can occur months to years before diagnosis. This early period, referred to as the clinical high risk (CHR) for psychosis state, reflects a time during which attenuated psychotic symptoms, marked declines in social and role functioning, help-seeking behavior, and non-psychotic comorbidity are noted. Intervention in the CHR state can prevent future illness-related dis­ability2. Longitudinal studies of CHR individuals show that, at two-year follow-up, approximately 20% transition to psychosis3, 41% undergo remission4, but many of the remainder experience significant symptoms and problems in functioning4. Studies are underway to establish risk calculators and biomarkers that can help identify CHR individuals who are most likely to convert to psychosis, but more work is needed to develop tools that use mechanistic input to stratify CHR populations by predicted clinical outcomes beyond psychosis5. The CHR stage represents a unique opportunity to develop interventions guided by such tools, focused on reducing conversion to psychosis and improving long-term functional outcomes. Aimed at capitalizing on this opportunity, the Accelerating Medicines Partnership® Schizophrenia (AMP® SCZ) is a large international collaboration to develop algorithms using a set of clinical and cognitive assessments, multi-modal biomarkers, and clinical endpoints that can be used to predict the trajectories and outcomes of CHR individuals and advance the testing of pharmacological interventions for CHR individuals in need. The goal is to accurately predict which individuals are likely to remit, experience an acute psychotic episode, or have intermediate outcomes that feature persistent attenuated psychotic and/or mood symptoms along with functional impairment. The algorithms will have the potential to serve as early indicators of treatment efficacy in CHR persons. The AMP SCZ partnership, managed by the Foundation for the National Institutes of Health (FNIH), brings together a breadth of scientific and regulatory expertise and lived experience from the partners: the US National Institute of Mental Health (NIMH), the US Food and Drug Administration (FDA), and the European Medicines Agency (EMA); private industry (Boehringer Ingelheim; Janssen Research & Development; Otsuka Pharmaceutical Development & Commercialization); non-profit and patient advocacy organizations (American Psychiatric Association Foundation; National Alliance on Mental Illness; One Mind; Schizophrenia & Psychosis Action Alliance); and a charitable foundation (Wellcome). The partnership will contribute $117.7 million over 5 years ($99.4 million from NIMH, $7.5 million from industry, and $10.8 million from non-profit organizations) to support implementation of the program. The AMP SCZ program is composed of two Research Networks – the Psychosis-Risk Outcomes Network (ProNET) at Yale University, and the Trajectories and Predictors in the CHR for Psychosis Population: Prediction Scientific Global Consortium (PRESCIENT) at the University of Melbourne/Orygen – and a Data Processing, Analysis and Coordination Center (DPACC) at Harvard Medical School6. ProNET and PRESCIENT form a harmonized research network focused on CHR individuals: identifying biological markers, clinical endpoints, and other measures that predict disease trajectory and outcomes for this group. The DPACC is responsible for managing, processing, disseminating, archiving and analyzing AMP SCZ data, which will be rapidly disseminated and made accessible to all qualified researchers and the public within the NIMH Data Archive7. The AMP SCZ research network will recruit a large cohort (N=1,977) of individuals between the ages of 12 and 30 years who meet CHR criteria – based on the Positive SYmptoms for CAARMS Harmonized with SIPS (PSYCHS) interview, a new psychometric instrument for defining CHR and associated outcomes – and healthy controls (N=640) across 42 sites from 14 countries (US, Canada, UK, Spain, Italy, Switzerland, The Netherlands, Germany, Denmark, Australia, Singapore, South Korea, Chile and China). CHR participants will complete screening, baseline assessments, and a battery of follow-up assessments across 24 months. Healthy controls will complete screening and baseline assessments, and a subset (approximately 5 per site) will complete month 2, 12 and 24 visits. The CHR cohort and healthy controls will be assessed with a core set of measures at baseline and 2 months post-baseline, with additional assessments completed at other timepoints. CHR subjects will be assessed longitudinally for up to 2 years. Subjects who develop their first episode of psychosis (“converted” cases) over the course of study participation will continue to be followed and assessed as scheduled. Measures will include clinical and cognitive assessments; neurophysiology, neuroimaging, genetics and fluid biomarkers; speech and facial expression (audio/video recording); and digital assessments8. The digital assessments will collect active (e.g., daily survey on social interactions and feelings of connectedness) and passive (e.g., time spent sleeping, number of texts and phone calls received or made; time participants spend in green space, home, school, exercising, therapy visits, and social relationships) data, along with an automated assessment of social and community functioning from global positioning system (GPS) data. Through the digital measures, AMP SCZ will be able to assess bio-psycho-social data in CHR individuals and elucidate their role in affecting trajectories which could be targeted by psychosocial interventions. The primary endpoint of interest is conversion to psychosis by 24-month follow-up as defined by psychosis threshold criteria on the PSYCHS. Secondary clinical endpoints of interest include remission or recovery of CHR state, and non-conversion/non-remission. Clinical outcomes of interest cover multiple domains such as attenuated positive symptoms, mood and anxiety, psychosocial functioning, and persistent negative symptoms8. The biomarker data collected by ProNET and PRESCIENT will be analyzed by the DPACC to develop multi-modal prediction models and risk calculators by drawing on recent theoretical and methodological advances (e.g., dynamic prediction, probabilistic multimodal modeling). These models will leverage existing prediction models in the field9 and guide selection and stratification of CHR participants for future clinical trials based on the primary endpoint of interest. For example, stratification can identify a subset of CHR participants who are at higher risk of developing psychosis relative to the rest. The developed tools may have clinical utility in decision making about stepping interventions up or down as risk is assessed over time (clinical trajectory, treatment response) and in response to incoming biomarker information. Some tools, such as the risk calculators, will prioritize the less invasive and more readily available biomarkers for prediction, to enable clinical tools that could be used in community-based settings and are more tolerable by subjects. The novel prediction models generated for the AMP SCZ dataset will be tested using cross-validation approaches designed to improve generalizability of the derived algorithms to other CHR cohorts. By integrating the strengths of multiple international stakeholders, sharing discoveries openly, and priming future research, the AMP SCZ program aims to catalyze advances in knowledge about the CHR population to enable intervention at the earliest stages of schizophrenia, with the goal of maximizing functional outcomes for CHR patients.