Blood gene signals reveal Parkinson’s risk years before diagnosis

By Tarun Sai LomteReviewed by Lauren HardakerFeb 2 2026

Blood-based DNA repair and stress-response signals reveal a brief molecular window preceding Parkinson’s diagnosis.

Study: Longitudinal assessment of DNA repair signature trajectory in prodromal versus established Parkinson’s disease. Image credit: Chinnapong/Shutterstock.com

In a recent study published in npj Parkinson’s Disease, researchers examined DNA repair and integrated stress response (ISR) genes in healthy individuals and those with prodromal and established Parkinson’s disease (PD).

DNA damage and repair emerge early in Parkinson’s progression

PD is a progressive neurodegenerative disorder characterized by motor symptoms, such as bradykinesia, postural instability, tremors, and rigidity, which appear only after significant neurodegeneration. Years before symptom onset, people experience a prodromal phase characterized by non-motor features, including olfactory loss, anxiety, depression, constipation, and rapid eye movement sleep behavior disorder.

Understanding the molecular changes and mechanisms in prodromal PD is crucial, as it may enable diagnosis before significant neuronal loss occurs, when interventions are most likely effective. Evidence suggests that oxidative DNA damage and maladaptive or impaired repair responses contribute to PD pathogenesis from the earliest stages. As such, prodromal PD patients may show distinct changes in repair pathways before clinical manifestation.

Longitudinal blood transcriptomics track DNA repair dynamics in PD

In the present study, researchers explored the dynamic regulation of DNA repair gene expression in healthy individuals, prodromal PD individuals, and established PD patients. They included 188 healthy individuals, 58 prodromal PD individuals, and 393 established PD patients from the Parkinson’s Progression Marker Initiative cohort and analyzed their peripheral blood transcriptomic data.

First, global differential expression patterns were examined across groups at baseline. The number of differentially expressed genes (DEGs) progressively increased across comparisons, from no significant differences between healthy and prodromal groups to extensive differences between healthy and established PD groups and between prodromal and established PD groups.

Notably, genes from the ISR, nuclear DNA damage repair (DNArep), and mitochondrial DNA damage repair (mtDNArep) pathways were mostly absent in the top 50 DEGs. Next, the team evaluated whether longitudinal gene expression patterns in the ISR, DNArep, and mtDNArep pathways could differentiate established PD patients from healthy individuals using machine-learning classification models applied across repeated follow-up visits.

Classification accuracy, assessed at baseline, 12, 24, and 36 months, ranged between 50 % and 64 % across time points. This meant that the expression of these pathways in peripheral blood does not provide a sufficiently strong signal to distinguish PD from healthy controls. In contrast, accuracy was high when differentiating between prodromal PD and healthy individuals across the three gene sets after baseline, while baseline classification performance remained comparatively weak.

Accuracy increased steadily over time for the mtDNArep gene set, peaking at 36 months (89 %). The variability in gene expression was highest at baseline and declined over time, suggesting that gene expression becomes more uniform in prodromal individuals as the disease progresses. The authors interpret this pattern as a transient, potentially adaptive transcriptional response that diminishes as PD advances. Further, classification accuracy was high between established and prodromal PD groups.

However, accuracy decreased slightly at later time points, suggesting that differences in gene expression attenuate as the disease progresses. The team also evaluated a core set of known PD-related genes and a broader PD-associated gene set to compare their classification abilities with those of ISR and DNA repair pathways. These PD-specific gene sets failed to reliably distinguish healthy people from established PD patients.

Nevertheless, the PD-specific gene sets distinguished healthy controls from prodromal individuals, with accuracy ranging from 65 % to 87 %. Similar to other gene sets, classification accuracy was lower at baseline and declined at later visits. The PD-specific gene sets also performed well in distinguishing prodromal individuals from those with established PD.

Next, the team assessed how gene expression changed over time in each group. Gene expression was relatively stable in healthy and established PD groups over time. However, the prodromal group exhibited greater variability, with the highest at baseline. Moreover, many genes in the prodromal group displayed dynamic, non-linear trends over time, with approximately half of the DNA repair genes and nearly three-quarters of the ISR genes showing non-linear trajectories.

Finally, feature importance analysis was performed to identify genes contributing the most to group separation. This revealed several predictors of prodromal PD, including excision repair cross-complementation group 6 (ERCC6), primase-polymerase (PRIMPOL), Nth-like DNA glycosylase 1 (NTHL1), and Nei-like DNA glycosylase 2 (NEIL2).

Early Parkinson’s shows fleeting but detectable molecular signatures

In sum, the study provided novel insights into the dynamics of ISR and DNA repair gene expression in prodromal PD. The results illuminate distinct molecular changes that occur before clinical diagnosis of PD, revealing candidate molecular signatures that may help inform earlier-stage disease stratification.

However, the authors caution that peripheral blood expression is an indirect proxy for brain pathology, that immune-related signals may influence observed patterns, that not all prodromal individuals may convert to clinical PD, and that transcript-level changes may not directly reflect protein function. Future studies should validate these results in larger cohorts, examine the underlying biological mechanisms, and assess the diagnostic and therapeutic potential.

Download your PDF copy now!