Complexity of genomic diagnosis: Lessons learnt from the UK Biobank and Generation study newborn genome sequencing...

Pathogenic variants in COL1A1 and COL1A2 account for 85-90% of osteogenesis imperfecta (OI). Glycine substitutions within the triple-helical domain are traditionally considered highly pathogenic. The COL1A2 c.2827G>A p.(Gly943Arg) variant was previously classified as pathogenic under ACGS 2019/2020 criteria.

One proband with this heterozygous variant was identified through national newborn genomic screening prompting further evaluation from the UK National OI Reference Laboratory. Four additional patients with the same variant were identified through referral to tertiary metabolic bone services/ Genetics clinic and phenotypic data were gathered to understand better clinical presentations and inheritance patterns. The variant was reassessed following the release of gnomAD v4.1, incorporating updated ACGS 2024 criteria.

The expanded gnomAD v4.1 dataset incorporating UK Biobank (UKBB) demonstrated the presence of p.(Gly943Arg) in an unselected population at a frequency inconsistent with the pathogenicity of COL1A2, resulting in reclassification from pathogenic to variant of uncertain significance (VUS). Clinical presentations of 5 patients ranged from no fracture history to moderately severe OI. Cascade testing revealed multiple asymptomatic adult heterozygotes. UKBB data demonstrated no recorded OI diagnoses among 44 adult heterozygotes with evidence of a shared common haplotype in 40/45 enriched in Northwest England.

The COL1A2 p.Gly943Arg variant demonstrates marked clinical heterogeneity and population frequency inconsistent with a fully penetrant dominant OI allele. This study highlights the dynamic nature of variant classification, the importance of large population datasets, and the ethical and clinical challenges of variant reclassification in the era of genomic screening.