Polygenic risk scores (PRS) aggregate the effects of common genetic
variants into a single metric of disease predisposition. Many
neurological disorders exhibit a polygenic architecture, thereby
providing a rationale for the application of PRS in risk prediction,
biological subtyping, and stratification of patients to inform clinical
decision-making. Here, we use restless legs syndrome (RLS) as an
informative translational model to discuss both opportunities and
current constraints of PRS use in neurology. RLS has a
well-characterized polygenic component with 164 GWAS risk loci, a PRS
with moderate case-control discrimination (AUC 0.73) when used alone,
but showing potential for higher performance (AUC 0.82–0.91) in
machine-learning models incorporating non-genetic variables. We discuss
how multi-omics integration, PRS-based clinical subgrouping, and rare
variant penetrance modification can advance PRS development and
application in RLS and contextualize these developments within the wider
landscape of PRS in neurological disorders.