Abstract:Key Points Question To what extent do global polygenic risk scores (PRS), molecular pathway-specific PRS, complement component (C4) gene expression, MHC loci, sex, and ancestry jointly contribute to risk for schizophrenia-spectrum disorders (SZ)? Findings Global polygenic risk for schizophrenia, sex, and their interaction most robustly predict risk in a classification and regression tree model, with highest risk groups having 50/50 chance of SZ. Meaning Psychometric risk indicators, such as prodromal symptom assessments, may be enhanced by the examination of genetic risk metrics. Preliminary results suggest that of genetic risk metrics, global polygenic information has the most potential to significantly aide in the prediction of SZ. Abstract Importance Schizophrenia (SZ) has a complex, heterogeneous symptom presentation with limited established associations between biological markers and illness onset. Many (gene) molecular pathways (MPs) are enriched for SZ signal, but it is still unclear how these MPs, global PRS, major histocompatibility complex (MHC) complement component (C4) gene expression, and MHC loci might jointly contribute to SZ and its clinical presentation. It is also unclear whether sex or ancestry interacts with these metrics to increase risk in certain individuals. Objective To examine multiple genetic metrics, sex, and their interactions as possible predictors of SZ risk. Genetic information could aid in the clinical prediction of risk, but it is still unclear which genetic metrics are most promising, and how sex interacts with genetic risk metrics. Design, Setting, and Participants To examine molecular risk in a proof-of-concept study, we used the Wellcome Trust case-control cohort and classified cases as a function of 1) polygenic risk score (PRS) for both whole genome and for 345 implicated molecular pathways, 2) predicted C4 expression, 3) SZ-relevant MHC loci, 4) sex, and 5) ancestry. Main Outcomes and Measures PRSs, C4 expression, SZ-relevant MHC loci, sex, and ancestry as joint risk factors for SZ. Results Recursive partitioning yielded 15 molecular risk classes and retained as significant psychosis classifiers only sex, genome-wide SZ polygenic risk, and one MP PRS. Sex was the most robust classifier in a stepwise regression, and there was a significant interaction of sex with SZ PRS on case status, suggesting males have a lower polygenic risk threshold. By down-sampling case proportion to 1% and 1.4% population base rates in males and females, respectively, high-risk subtypes defined by this model had roughly a 52% odds of developing SZ (individuals with SZ PRS elevated by 2.6 SDs; incidence = 51.8%). Conclusions and Relevance This proof-of-concept suggests that global SZ PRS, sex, and their interaction are robust predictors of risk and that males have a lower PRS threshold for onset. Implications for the integration of these metrics with psychometrically-identified risk are discussed.

Can artificial neural networks supplant the polygene risk score for risk prediction of complex disorders given very large sample sizes?

Prediction of antipsychotics efficacy based on a polygenic risk score: a real-world cohort study

A Non-Parametric Method for Building Predictive Genetic Tests on High-Dimensional Data

Proof of concept: Molecular prediction of schizophrenia risk

Optimization of Multi-Ancestry Polygenic Risk Score Disease Prediction Models

Use of multiple polygenic risk scores for distinguishing schizophrenia-spectrum disorder and affective psychosis categories in a first-episode sample; the EU-GEI study

Improving the Utility of Polygenic Risk Scores as a Biomarker for Alzheimer's Disease

Epistatic Features and Machine Learning Improve Alzheimer's Disease Risk Prediction Over Polygenic Risk Scores

Identifying Genes Associated with Alzheimer's Disease Using Gene-Based Polygenic Risk Score

A comprehensive investigation of statistical and machine learning approaches for predicting complex human diseases on genomic variants

Distinguishing different psychiatric disorders using DDx-PRS

Applications of Polygenic Risk Scores in Psychiatric Genetics.

Feature Selection for Polygenic Risk Scores using Genetic Algorithm and Network Science

Winner's Curse Correction and Variable Thresholding Improve Performance of Polygenic Risk Modeling Based on Genome-Wide Association Study Summary-Level Data

Genomic Prediction of Complex Disease Risk

Interpreting artificial neural networks to detect genome-wide association signals for complex traits

Polygenic Risk Scores in Alzheimer’s Disease Genetics: Methodology, Applications, Inclusion, and Diversity

Systems biology approaches to utilise polygenic risk scores for chronic diseases

Polygenic Risk Scores Provide Strong Evidence for the Biologic Relevance of Molecular Subtypes of Schizophrenia

Effective Genetic Risk Prediction Using Mixed Models

Multi-scale Analysis of Schizophrenia Risk Genes, Brain Structure, and Clinical Symptoms Reveals Integrative Clues for Subtyping Schizophrenia Patients