Objective: To improve the prediction of Parkinson’s disease (PD) by examining aggregate profiles of gene expression in the form of transcriptional risk scores (TRS).

Background: Although genome-wide association studies (GWAS) capture substantial variant-based heritability for PD, polygenic score (PGS) prediction remains limited. PGS prediction has been successfully improved for outcomes such as ADHD by integrating multiple PGS in the same model (multi-PGS). TRS – calculated as the weighted sum of an individual’s observed gene expression – offers a complementary approach to understanding the biological contributions to PD risk and improving predictive models. Our previous work (forthcoming) has demonstrated for the first time that TRS can identify novel within- and cross-trait associations between gene expression profiles, PD and its phenotypes not identified using PGS alone. However, the multi-score modelling strategy has not yet been applied to TRS nor within PD.

Method: We conduct transcriptome-wide association analysis (TWAS) across multiple phenotypes, including neurodegenerative diseases, psychiatric disorders, cardiovascular outcomes, metabolic traits, and cognitive performance measures. TWAS-derived expression weights were applied to RNA-seq data from AMP-PD (https://www.amp-pd.org) (Ncases = 2,071; Ncontrols = 1,465). After univariate analysis, a multi-TRS model for PD case/control status combining significant TRS from phenotypes will be trained using logistic regression with elastic net regularisation.

Results: In preliminary univariate analyses we find several novel associations, including that depression, schizophrenia and Alzheimer’s disease TRS predict between 1.5-2% of the variance in PD status (p_FDR ≤ 0.05). We expect combined multi-TRS models to significantly improve the prediction of PD status.

Conclusion: This is the biggest TRS study on PD and represents the first application of a multi-TRS approach to PD prediction. By integrating TRS from multiple phenotypes, this work aims to provide novel biological insights into PD and improve disease prediction.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/a-multi-transcriptional-risk-score-approach-to-parkinsons-disease-prediction/