Accelerated aging in schizophrenia: a systematic review and meta-analysis of multi-organ aging clocks

Introduction

Schizophrenia-spectrum disorders (SSD) are associated with a 15–20 year mortality gap that cannot be fully explained by modifiable risk factors. An emerging neuroscience framework proposes that psychosis itself is accompanied by accelerated biological aging, detectable through structural brain changes and peripheral epigenomic signatures. Two complementary biomarker strategies now allow quantification of this process: machine-learning algorithms that estimate brain age from structural magnetic resonance imaging (MRI) data, providing a proxy for cumulative grey matter loss, synaptic pruning, and white matter degradation; and DNA methylation (DNAm)-based epigenetic clocks that index biological ageing via inflammation-related and senescence-related CpG signatures. Whether these two measurement domains reflect a shared biological aging process, and the extent to which the aging signal is driven by illness biology versus iatrogenic or lifestyle confounders, remains unresolved.

Aims

We aimed to (1) quantify brain-predicted age difference (brain-PAD) across clinical stages of psychosis; (2) determine whether epigenetic age acceleration varies by clock generation; (3) evaluate the extent to which included studies controlled for key confounders; and (4) compare the magnitude of central nervous system (brain) versus peripheral (blood-derived DNAm) aging signals.

Methods

Following PRISMA 2020 guidelines (PROSPERO: CRD420251163851), we searched PubMed, Embase, PsycINFO, Web of Science, and Scopus from inception to March 2026. Eligible studies reported the discrepancy between biologically predicted and chronological age in SSD versus healthy controls. Effect sizes were synthesised using three-level random-effects meta-analysis. Pre-specified moderators included illness stage (clinical high-risk [CHR], first-episode psychosis [FEP], established schizophrenia), clock generation (first-generation versus mortality-trained second-generation clocks), and neuroimaging model architecture. Publication bias was evaluated using Precision-Effect Test-Precision-Effect Estimate with Standard Errors (PET-PEESE) regression.

Results

Forty-seven studies met inclusion criteria (30 neuroimaging; 17 epigenetic), encompassing 13,275 patients and 14,869 controls. The pooled brain-PAD was +3.14 years (95% CI [2.39, 3.89]; p < 0.001; I² = 78%). A stage-dependent gradient emerged: CHR +0.00 years (95% CI [−0.92, 0.93]), FEP +2.35 years (95% CI [1.08, 3.62]), and established schizophrenia +3.56 years (95% CI [2.70, 4.42]; Q-test p = 0.009). Among epigenetic clocks, only second-generation measures trained on morbidity and mortality outcomes detected significant acceleration (GrimAge: +2.64 years, p < 0.001; PhenoAge: +1.15 years, p = 0.03). After PET-PEESE correction, the adjusted brain-PAD (1.43 years) converged with second-generation epigenetic estimates. Critically, the majority of included studies did not adequately control for key confounders — including antipsychotic dosage (/chlorpromazine equivalent), BMI, type 2 diabetes, and smoking status — precluding definitive conclusions about whether the observed aging acceleration is attributable to illness neurobiology, cumulative pharmacological exposure, or cardiometabolic comorbidity.

Conclusions

These findings provide meta-analytic evidence that psychosis is accompanied by a progressive, stage-dependent acceleration of biological aging measurable across both central and peripheral tissues. The convergence of neuroimaging and epigenomic signals supports a translational model potentially linking systemic immunosenescence with premature brain aging. However, the field’s systematic failure to account for antipsychotic exposure and metabolic confounders represents a critical gap. Disentangling iatrogenic from disease-intrinsic aging mechanisms is an urgent priority for designing neuroprotective intervention strategies — including anti-inflammatory agents, senolytic compounds, and metabolic modulators — in early psychosis.

References

No references