Epigenetic clocks capture disease burden in early psychosis

Epigenetic clocks provide a powerful framework to quantify deviations from normative aging trajectories, which may yield important insights into the underlying pathophysiology of psychiatric disorders. Across psychiatric conditions - and particularly in psychotic disorders - there is converging evidence for early functional decline and disability, alongside findings of accelerated brain aging. Previous work has indicated that epigenetic clocks trained on age-related biological and health-related features are more sensitive to psychiatric disease burden than clocks trained on chronological age alone [1]. A key open question is whether these signals are already detectable at early stages and could thus support early detection and preventive strategies.

Here, we investigated epigenetic age dynamics in the Personalised Prognostic Tools for Early Psychosis Management (PRONIA) cohort, using high-resolution DNA methylation data generated with Illumina MethylationEPIC v2.0 arrays (N = 963). The sample comprised four diagnostic groups: healthy controls (HC, n = 324), clinical high risk for psychosis (CHR-P, n = 209), recent onset depression (ROD, n = 204), and recent onset psychosis (ROP, n = 226). All participants were assessed at baseline and characterized by multimodal data, including structural and functional neuroimaging, polygenic risk scores, and comprehensive clinical and neurocognitive profiles.

Data preprocessing followed established pipelines, including stringent quality control (detection p-values, beta density inspection, and sex concordance checks), probe filtering (removal of low-quality, SNP-affected, cross-hybridizing, polymorphic, and sex chromosome probes), normalization (stratified quantile normalization with BMIQ), outlier detection, and batch effect correction using ComBat. Epigenetic age estimates were derived using multiple clocks capturing distinct aging dimensions: chronological clocks (Horvath, Hannum), biological clocks (PhenoAge, GrimAge), and pace of aging (DunedinPACE). Age acceleration metrics were calculated relative to the HC subsample and subsequently applied to patient groups, with adjustment for estimated cell-type composition.

No significant group differences were observed for chronological clocks (Horvath: p = 0.21; Hannum: p = 0.15) or PhenoAge (p = 0.98), suggesting that traditional age-predictive models may lack sensitivity to psychiatric disease-related aging processes. In contrast, GrimAge showed robust group differences (p = 9.65 × 10⁻⁸), with all patient groups exhibiting significant acceleration relative to HC. Effect sizes indicated a graded pattern across diagnostic groups (HC-ROD: Cohen’s d = 0.25; HC-CHR-P: d = 0.37; HC-ROP: d = 0.49), consistent with differences across clinical groups. Additionally, DunedinPACE revealed a modest but significant effect (p = 0.014), driven primarily by increased pace of aging in the ROP group compared to HC. These findings align with prior evidence indicating that biologically informed epigenetic clocks, particularly those incorporating health-related and functional aging dimensions, are more sensitive to disease-associated aging alterations [1].

Overall, the results support the view that epigenetic aging measures may index aspects of psychiatric disease burden, with potential relevance even at early stages of psychosis. Future directions include integrating epigenetic aging metrics with clinical, neurocognitive, genetic, and neuroimaging data to identify multimodal predictors of disease trajectories. Additionally, comparisons with complementary aging biomarkers derived from neuroimaging and proteomic data may clarify whether these measures capture shared or distinct dimensions of biological aging in psychiatric populations.

References

1.Yusupov, N., Dieckmann, L., Erhart, M., Sauer, S., Rex-Haffner, M., Kopf-Beck, J., … & Binder, E. B. (2023). Transdiagnostic evaluation of epigenetic age acceleration and burden of psychiatric disorders. Neuropsychopharmacology, 48(9), 1409-1417.