DNA variant classification-reconsidering "allele rarity" and "phenotype" criteria in ACMG/AMP guidelines

Recent guidance suggested modified DNA variant pathogenicity assignments based on genome-wide allele rarity. Different a priori probabilities of pathogenicity operate where patients already have clinical diagnoses, and are found to have a very rare variant in a gene known to cause their disease, compared to predictive testing of a clinically unaffected individual. We tested new recommendations from the ClinGen Sequence Variant Interpretation Working Group for ClinVar-listed, loss-of-function variants meeting the very strong evidence of pathogenicity criterion [PVS1] in genes for 3 specific diseases where causal gene identification can modify clinical care of an individual- Von Willebrand disease, cystic fibrosis and hereditary haemorrhagic telangiectasia. Across these diseases, current rules leave 20/1,278 (1.6%) of loss-of-function variants as variants of uncertain significance (VUS that may not be reported to clinicians), and 207/1,278 (17.2%) as likely pathogenic. Applying the new ClinGen rule enabling PVS1 and the allele rarity criterion PM2 to delineate likely pathogenicity still left 8/1,278 (0.9%) as VUS (reflecting non-PVS1 calls by the submitters), and the majority of null alleles meeting PVS1 as merely likely pathogenic. We favour an approach whereby, for PVS1 variants in patients who personally meet the phenotypic PP4 criterion for a disease where casual variants are commonly family-specific, that PM2 is upgraded to permit a pathogenic call. Of 1,278 ClinVar-listed frameshift, nonsense and canonical splice site variants that met PVS1 in the 3 conditions, 16.0% (204/1,278) would be newly designated as pathogenic, avoiding misinterpretation outside of clinical genetics communities. We suggest further discussion around variant assessment across different clinical applications, potentially guided by PP4 alerts to distinguish personal versus family phenotypic history.