Urinary exosomal microRNAs as predictive biomarkers for persistent psychotic-like experiences

Psychotic-like experiences (PLEs) are relatively common during adolescence (estimated prevalence 5–17%) and often transient, but a subset persists and is linked to increased risk for schizophrenia, depression, and suicidality. Persistent PLEs are thus a meaningful early-intervention target. However, biological predictors of persistence are scarce. MicroRNAs (miRNAs), especially those contained in exosomes detectable in biofluids such as urine, have emerged as promising, minimally invasive biomarkers in psychiatric disorders. Given urine’s non-invasive collection and evidence that some urinary exosomes may originate from brain tissue, the study aimed to test whether urinary exosomal miRNA profiles at age 13 predict persistence of PLEs over one year in adolescents.

Prior work indicates PLEs in youth predict later psychopathology, with persistence conferring higher risk for psychotic disorders and poor outcomes. Few biological markers have been explored for predicting persistent PLEs, with fingertip advanced glycation end products being one example. Exosomal miRNAs have been implicated in psychiatric conditions, with altered miRNA expression reported in first-episode psychosis and in serum/plasma of schizophrenia and major depressive disorder cases. Exosomal miRNAs are stably measurable across biofluids and urine offers advantages for adolescent populations. No prior studies directly examined miRNAs in relation to PLEs, motivating this investigation.

Design: Population-based biomarker subsample of the Tokyo Teen Cohort (TTC) study. Participants: 345 adolescents (mean age 13.5 years) and caregivers at baseline; 282 (82%) completed one-year follow-up. PLE assessment: At ages 13 (baseline) and 14 (follow-up), participants first completed the Adolescent Psychotic-Like Symptom Screener (APSS, 7 items). Adolescents endorsing “Maybe” or “Yes, definitely” on any item underwent semi-structured interviews by experienced psychiatrists using Scales for the Assessment of Positive Symptoms to validate timing and nature of experiences, exclude hypnagogic/hypnopompic states, fever, substances, and reach consensus ratings. PLE groups were defined across two time points: no-experience, remitted (baseline PLEs only), incident (follow-up only), persistent (both time points). For miRNA analysis, 15 adolescents with persistent PLEs were age- and sex-matched to 15 with remitted PLEs. Sample collection and processing: Morning urine at baseline (age 13) was collected, initial centrifugation at 3000×g for 10 min, supernatant stored at −80 °C. Exosomes isolated from ~5–10 mL urine using miRCURY Exosome Cell/Urine/CSF Kit (QIAGEN). RNA extracted with miRNeasy Micro Kit (QIAGEN). RNA quality assessed with Agilent Bioanalyzer 2100 (clear small RNA peak <200 nt). Libraries prepared using QIA-seq miRNA Library Kit (QIAGEN) from 2 ng RNA. Sequencing and quantification: NovaSeq 6000 (Illumina). Total 1,409,409,273 sequences (~29,362,693 per donor). Reads aligned to miRBase v22.1 using CLC Genomics Workbench v12.0.3 (Biomedical Genomics Analysis Plugin v1.2.1) after adapter trimming. Unique molecular identifier (UMI) counts used for expression. Normalization and differential expression: UMI counts normalized to TPM using edgeR (v3.30.3). Differential expression tested via likelihood ratio test in edgeR comparing persistent vs remitted groups; full model included group and sex, reduced model included sex; p-values adjusted by Benjamini–Hochberg FDR. Prediction modeling: Logistic regression (R 4.0.4 glm) to estimate probability of persistent PLEs using normalized expression of significantly different miRNAs as predictors. Performance evaluated by ROC AUC with 95% CI via DeLong method (PROC R package). Cross-validation: five-fold CV using multivariate model; each fold trained on 80% and tested on 20%, average AUC and performance metrics reported. Pathway analysis: DIANA-mirPath v3 using microT-CDS v5.0 to identify miRNA–mRNA interactions and KEGG pathway enrichment (significance threshold p<0.05). Ethical approval and consent were obtained from the Tokyo Metropolitan Institute of Medical Sciences ethics committee.

• Cohort classification (n=282 at follow-up): no-experience 200 (70.9%), remitted 62 (22.0%), incident 5 (1.77%), persistent 15 (5.32%). Baseline PLEs in 92/345 (26.7%); follow-up PLEs in 20/282 (7.09%). Downstream analysis included 15 persistent and 15 age- and sex-matched remitted participants; no significant differences in age or sex between groups.
• miRNA detection: 2631 miRNAs detected overall; 427 expressed >10 TPM in at least 3 subjects used for analysis.
• Differential expression: Six miRNAs were significantly altered (FDR<0.05; |log2FC|>1): hsa-miR-486-5p, hsa-miR-199a-3p, hsa-miR-144-5p, hsa-miR-451a, hsa-miR-143-3p, hsa-miR-142-3p. All six were downregulated in the persistent group vs remitted.
• Prediction performance: Multivariate logistic model using the six miRNAs achieved AUC 0.853. Univariate AUCs ranged 0.48–0.747 (best: miR-144-5p at 0.747). Five-fold cross-validation of the six-miRNA model: average AUC 0.847 (95% CI 0.690–0.994); sensitivity 89.5±10.7%; specificity 74.1±13.3%; accuracy 82.5±3.12%; Youden index 0.451±0.0974.
• Pathway enrichment: Target genes of the six miRNAs enriched in multiple KEGG pathways, including dopaminergic synapse, ECM-receptor interaction, PI3K-Akt, mTOR, ErbB, estrogen signaling, Ras, TGF-beta, sphingolipid signaling, cholinergic synapse, among others.

The study demonstrates that urinary exosomal miRNA profiles at age 13 can distinguish adolescents whose PLEs will persist over the following year from those whose PLEs remit. The six-miRNA panel performed substantially better than individual miRNAs, supporting a multivariate signature as a predictive biomarker. These findings extend prior evidence that exosomal miRNAs are informative in psychiatric conditions and provide the first direct link between miRNAs and PLE persistence, a marker of elevated psychiatric risk. Enrichment of pathways related to dopaminergic synapse and ECM-receptor interaction aligns with hypothesized neurobiological mechanisms of PLEs and psychosis risk. Experimental links between identified miRNAs and targets such as KIF5B (implicated in dopaminergic synaptic function and neuronal transport) and ITGB8 (involved in ECM-mediated synaptic processes) provide plausible mechanistic routes by which altered miRNA levels could contribute to PLE pathophysiology. Together, results suggest the six-miRNA urinary exosomal signature could aid early identification and stratification for preventive interventions.

Six urinary exosomal miRNAs (hsa-miR-486-5p, hsa-miR-199a-3p, hsa-miR-144-5p, hsa-miR-451a, hsa-miR-143-3p, hsa-miR-142-3p) were significantly downregulated in adolescents with persistent PLEs versus remitted PLEs and, combined in a logistic regression model, predicted persistence over one year with high accuracy (AUC ~0.85). The work supports urinary exosomal miRNAs as minimally invasive biomarkers for early risk assessment of psychiatric disorders linked to persistent PLEs. Future research should validate these findings in larger, independent cohorts, extend prediction horizons beyond one year, and investigate functional mechanisms linking urinary exosomal miRNAs to brain pathophysiology.

• Small sample size (n=30 for modeling), raising concerns about statistical power and generalizability; although FDR control and five-fold cross-validation were applied, external validation is needed.
• Unclear physiological/functional linkage between urinary exosomal miRNAs and PLEs, as urine-based biomarkers are underexplored in psychiatry.
• Predictive evaluation limited to a one-year follow-up; performance over longer periods remains to be determined in ongoing cohort follow-up.