Abstract
Whole-exome/genome sequencing analyses lead to detect disease-causing variants that are unrelated to the initial clinical question. Irrespective of any actionable gene list, only pathogenic variants should be considered. The pathogenicity of 55 cystic fibrosis transmembrane conductance regulator (CFTR) variants of known various impacts was assessed by a group of experts by comparing data from specialized databases CFTR-France and CFTR2 with those of general clinical databases ClinVar and Human Gene Mutation Database (HGMD®) Professional and data aggregators VarSome and InterVar. The assessment of cystic fibrosis (CF) variants was correct with ClinVar and HGMD® Professional while less reliable with VarSome and InterVar. Conversely, the risk of overclassifying variants as CF-causing was up to 82% with HGMD® Professional. The concordance between data aggregators was only 50%. The use of general databases and aggregators is thus associated with a substantial risk of misclassifying variants. This evaluation may be extrapolated to other disease conditions and incites to remain cautious in interpreting and disclosing incidental findings.