Purpose: Utilization of sequencing to screen the general population for preventable monogenic conditions is receiving substantial attention because of its potential to decrease morbidity and mortality. However, the selection of which variants to return is a serious implementation challenge. Procedures must be investigated to ensure optimal test characteristics and avoidance of harm from false-positive test results.
Methods: We scanned exome sequences from 478 well-phenotyped individuals for potentially pathogenic variants in 17 genes representing 11 conditions that are among the most medically actionable Mendelian disorders in adults. We developed five variant selection algorithms with increasing sensitivity and measured their specificity in these 17 genes.
Results: Variant selection algorithms with increasing sensitivity exhibited decreased specificity, and performance was highly dependent on the genes analyzed. The most sensitive algorithm ranged from 88.8 to 99.6% specificity among the 17 genes.
Conclusion: For conditions with very low prevalence, small reductions in specificity greatly increase false positives. This inescapable test characteristic governs the predictive value of genomic sequencing in the general population. To address this issue, test performance must be evaluated systematically for each condition so that the false negatives and false positives can be tailored for optimal outcomes, depending on the downstream clinical consequences.Genet Med 18 6, 593-599.