Flavonoid-rich dietary patterns and incident hospital-recorded Parkinson's disease: a prospective UK Biobank study

Evidence linking flavonoid-rich dietary patterns to the risk of Parkinson's disease (PD) remains limited and inconsistent, and molecular clues to this association remain scarce. This study examined the association between a flavonoid diet score (FDS) and incident hospital-recorded PD and explored related plasma proteomic and cellular expression patterns.

In this prospective analysis of 123,655 UK Biobank participants, we used multivariable Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between FDS and incident hospital-recorded PD. In an exploratory analysis, we integrated a plasma proteomic screen with publicly available single-cell and single-nucleus transcriptomic data to provide preliminary molecular and cellular context.

During follow-up, 796 incident hospital-recorded PD cases were identified. Higher FDS was associated with a lower risk of incident hospital-recorded PD per 1-standard deviation (SD) increase (HR, 0.93; 95% CI, 0.86-1.00), and the risk was lower in the highest than in the lowest FDS quartile (HR, 0.75; 95% CI, 0.61-0.93). In decomposition analyses, no individual food source or flavonoid subclass remained associated with incident hospital-recorded PD after multiple-testing correction. The inverse association was directionally consistent across most sensitivity analyses and in 1-5-year lag analyses, but the evidence became borderline after adjustment for coffee intake (HR, 0.93; 95% CI, 0.86-1.00) and was no longer statistically supported in 6-10-year lag analyses. In exploratory omics analyses, the proteomic screen prioritized a direction-compatible signature comprising 24 proteins, including proteins related to immune and adhesion processes, while single-cell and single-nucleus transcriptomic data provided peripheral immune and glial-vascular cellular context.

In this prospective cohort, higher FDS was modestly associated with a lower incidence of hospital-recorded PD. Residual confounding and reverse causation remain possible, and the proteomic and transcriptomic findings should be considered hypothesis-generating.