Shared genetic architecture of Parkinson's disease and cutaneous melanoma: Heterogeneous pleiotropy and Colocalization at the SOX6 locus.

Parkinson's disease (PD) and cutaneous melanoma show epidemiological co-occurrence, but their shared genetic architecture remains unclear

METHODS

We combined bidirectional time-dependent cox regression in the UK biobank (N ≈ 500,000) with a multi-layered genomic analysis using large-scale PD and melanoma GWAS summary statistics (N = 745,746). Genetic analyses included linkage disequilibrium score regression (LDSC), bidirectional Mendelian randomization (MR), PLACO, LAVA, and Bayesian colocalization

RESULTS

Melanoma was associated with increased subsequent PD risk (HR = 1.62, 95% CI: 1.18-2.23), whereas the reverse PD-to-melanoma analysis was not statistically significant and was limited by low event counts. LDSC showed a modest positive genetic correlation (rg = 0.181, P = 0.04), while MR provided no evidence of bidirectional causality. PLACO identified 134 pleiotropic SNPs across 25 gene loci, including SOX6, the GDF5-GSS-EDEM2 cluster, and immune-related loci at 17q21.32. LAVA identified 17 locally correlated loci, with antagonistic effects predominating (11/17, 64.7%). Bayesian colocalization highlighted SOX6 as the strongest shared-signal locus (PP·H4 = 0.904; PP·H3 = 0.054)

CONCLUSIONS

PD-melanoma co-occurrence is supported by shared but heterogeneous genetic architecture rather than bidirectional causality. Antagonistic local effects may partially offset genome-wide correlation, while SOX6 represents a prioritized colocalized candidate locus linking dopaminergic neuron vulnerability and melanocytic biology.