Related Experiment Videos
Integrating Mendelian randomization, machine learning and retrospective clinical data: an exploratory analysis of the
Yao Ge1, Hongbin Cai2, Yike Li1
1The First School of Clinical Medicine, Lanzhou University, Lanzhou, China.
Background:
Epidemiological studies on the association between chronic hepatitis B (CHB) and Parkinson's disease (PD) have yielded inconsistent findings, with causality obscured by confounding and limited mechanistic evidence.
Methods:
The study employed an integrated approach encompassing two-sample Mendelian randomization(MR), multi-omics analysis, machine learning-driven gene screening, immune infiltration profiling, and multicenter retrospective clinical validation, with retrospective clinical validation conducted in two independent cohorts.
Results:
MR provided evidence suggesting a genetically predicted inverse association between susceptibility to CHB and the risk of PD (OR = 0.82-0.94, p < 0.05). Through the integration of machine learning and multi-omics data, RTN3 and MAP4K3 were recognized as priority cross-disease genes linking these two conditions. Phenylalanine metabolism emerged as an amino acid pathway showing consistent dysregulated patterns between the two diseases, with peripheral phenylalanine levels exhibiting a divergent trend: elevated in CHB while relatively lower in PD. Immune infiltration analysis and clinical hematological data suggested that eosinophil levels tended to decline in CHB but rise in PD, and such divergent expression patterns may be linked to the observed inverse correlation between CHB and PD susceptibility. (OR = 8.99, p < 0.001).
Conclusion:
Genetic susceptibility to chronic hepatitis B is inversely associated with Parkinson's disease risk. The shared pathophysiological landscape potentially involves MAP4K3, RTN3, phenylalanine metabolism, and eosinophil. These factors represent candidate therapeutic targets and peripheral biomarkers for PD risk reduction.