Biomarkers for predicting transition from at-risk mental state to psychosis: A systematic review

Identifying biomarkers to predict psychosis transition in clinical high-risk (CHR) individuals represents a critical challenge in early intervention psychiatry. While 20-30 % of CHR individuals develop psychosis within two years, most do not transition, necessitating improved prognostic tools. This systematic review evaluates neurobiological, neuroimaging, genetic, and peripheral biomarkers for predicting psychosis onset.

Following PRISMA guidelines, we searched five databases (PubMed, Scopus, Web of Science, PsycINFO, Embase) through November 2024 for longitudinal studies assessing baseline biomarkers in CHR populations. Study quality was evaluated using the Newcastle-Ottawa Scale and PROBAST. Due to substantial heterogeneity, findings were synthesized narratively.

From 7894 citations, 46 studies encompassing ∼15,500 CHR unique individuals were included. Neurophysiological measures showed the most consistent replication, with mismatch negativity and P300 event-related potentials demonstrating moderate effect sizes in multiple independent cohorts. Structural neuroimaging identified prefrontal and anterior cingulate volume reductions as predictive markers. Polygenic risk scores explained ∼10 % of transition variance in European ancestry populations but performed poorly in non-European samples. Multimodal prediction models achieved 70-86 % accuracy in development samples, though external validation remains absent. Clinical prediction models using symptoms and functioning often matched or exceeded biomarker-based approaches.

Despite two decades of research, no biomarker achieves clinical implementation standards. Methodological limitations include small samples, validation deficits, and population biases. While biomarkers illuminate psychosis pathophysiology and identify potential intervention targets, their current value lies primarily in research rather than clinical practice. Future priorities include adequately powered multi-site studies with rigorous external validation across diverse populations.