Extracting and calibrating evidence of variant pathogenicity from population biobank data

Area of Science:

• Genomic Medicine
• Human Genetics
• Population Health

Background:

• Genomic medicine relies on understanding variant phenotypic impacts, but evidence is often incomplete for monogenic disease genes.
• Assessing the clinical significance of genetic variants, especially rare variants of unknown significance (VUS), is a major challenge.

Purpose of the Study:

• To evaluate the utility of population cohort data for assessing genetic variant pathogenicity.
• To establish a framework for using population-based evidence to support variant interpretation in clinical practice.

Main Methods:

• Analyzed variant-level odds ratios of disease enrichment in 41 genes across 18 phenotypes using UK Biobank data (469,803 participants).
• Calibrated odds ratio evidence strength against American College of Medical Genetics and Genomics (ACMG/AMP) guidelines (PS4 criterion).
• Integrated computational, functional, and population data to reclassify variants.

Main Results:

• Significant differences in odds ratios were found between pathogenic and benign variants for 11 phenotypes.
• Population-based odds ratios can provide 'moderate' to 'very strong' evidence for variant pathogenicity.
• 2.6% of participants harbor rare VUS with at least 'moderate' evidence of pathogenicity.
• 12.4% of rare VUS in the LDLR gene met criteria for likely pathogenic classification.

Conclusions:

• Population cohort data offer a powerful resource for building the evidence base required for genomic medicine.
• This approach can systematically reclassify variants of unknown significance, potentially identifying individuals at unrecognized risk.
• The findings demonstrate a scalable method to improve genetic variant interpretation and clinical decision-making.