AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Clinical Utility of Combined Rare and Common Variant Polygenic Score for Severe Hypercholesterolemia
By Shim I, Sui Y, Truong B et al.
Background: Common-variant polygenic risk scores (PRS) for low-density lipoprotein cholesterol (LDL-C) underperform at phenotypic extremes such as severe hypercholesterolemia (severe HC; LDL-C ≥190 mg/dL) and in non-European populations. Whether integrating rare-variant effects refines clinical risk identification across diverse ancestries, within monogenic familial hypercholesterolemia (FH) carriers, and among individuals below the diagnostic threshold remains unclear. Hypothesis: Combining rare- and common-variant PRS for LDL-C (PRS-comb) improves severe HC and coronary artery disease (CAD) risk stratification beyond common-variant PRS alone across ancestries and clinical subgroups. Methods: Among 147,534 diverse adults in the All of Us Research Program with whole-genome sequencing, we built a rare-variant PRS (PRS-RV) from per-variant effect estimates across 69 LDL-C-associated genes using UK Biobank-derived RareEffect estimates, and a common-variant PRS (PRS-CV) using weights from GLGC GWAS. PRS-comb was the linear combination of PRS-RV and PRS-CV with weights derived by full-cohort regression of untreated LDL-C on both scores; 5-fold cross-validation confirmed no overfitting. LDL-C underwent time-aware lipid lowering medication correction and was summarized as the person-level median. Monogenic FH was defined by pathogenic or likely pathogenic variants in LDLR , APOB , or PCSK9 . Outcomes were severe HC and prevalent/incident CAD. Models were adjusted for age, sex, principal components, and cardiovascular risk factors. Results: PRS-comb outperformed PRS-CV in incremental LDL-C variance explained across all ancestries (European 9.7%, African 8.1%, Admixed American 6.8%, East Asian 5.4%, South Asian 9.0%, Middle Eastern 5.0%), with larger relative gains in non-European populations reflecting greater rare-variant portability (Figure A). Among monogenic FH carriers, the top 10% of PRS-comb had a mean adjusted LDL-C of 185.9 mg/dL versus 146.7 in the bottom 90%, approaching the severe HC threshold (Figure B). Compared with non-carriers in the bottom 90% of PRS-comb (reference), FH carriers in the top 10% had adjusted ORs of 22.34 (95% CI 18.23-27.33) for severe HC and 2.89 (2.23-3.72) for prevalent CAD (Figure C). Among 12,108 individuals with sub-threshold high-PRS (LDL-C <190 mg/dL, PRS-comb top 10%), cumulative incident CAD was elevated relative to sub-threshold lower-PRS (adjusted HR 1.25, 95% CI 1.14-1.35) and reached an absolute 5-year incidence comparable to severe HC (5.6% vs 5.9%; Figure D) Conclusions: By adding rare-variant effects to a common-variant score, PRS-comb refines severe HC and CAD risk across ancestries, grades severity within monogenic FH carriers, and flags a sub-threshold group whose CAD risk approaches that of overt severe HC—gained at no added genotyping cost from one sequencing test, supporting clinical integration.