Optimising diagnostic yield in highly penetrant genomic disease

BackgroundPediatric disorders include a range of highly genetically heterogeneous conditions that are amenable to genome-wide diagnostic approaches. Finding a molecular diagnosis is challenging but can have profound lifelong benefits. MethodsThe Deciphering Developmental Disorders (DDD) study recruited >33,500 individuals from families with severe, likely monogenic developmental disorders from 24 regional genetics services around the UK and Ireland. We collected detailed standardised phenotype data and performed whole-exome sequencing and microarray analysis to investigate novel genetic causes. We developed an augmented variant analysis and re-analysis pipeline to maximise sensitivity and specificity, and communicated candidate variants to clinical teams for validation and diagnostic interpretation. We performed multiple regression analyses to evaluate factors affecting the probability of being diagnosed. ResultsWe reported approximately one candidate variant per parent-offspring trio and 2.5 variants per singleton proband, including both sequence and structural variants. Using clinical and computational approaches to variant classification, we have achieved a diagnosis in at least 34% (4507 probands), of whom 67% have a pathogenic de novo mutation. Being recruited as a parent-offspring trio had the largest impact on the chance of being diagnosed (OR=4.70). Probands who were extremely premature (OR=0.39), had in utero exposure to antiepileptic medications (OR=0.44), or whose mothers had diabetes (OR=0.52) were less likely to be diagnosed, as were those of African ancestry (OR=0.51). ConclusionsOptimising diagnosis and discovery in highly penetrant genomic disease depends upon ongoing and novel scientific analyses, ethical recruitment and feedback policies, and collaborative clinical-research partnerships.