Enter Note Done


Keywords: Genomics; Microarrays; Clinical testing; Genetic testing; Bioinformatics

K.M. Bowling 1; M.L. Thompson 1; D.E. Gray 1; K.T. Williams 1; K.M. East 1; A. Cannon 2; W.V. Kelley 1; I.P. Moss 2; J.C. Edberg 2; A.C.E. Hurst 2; E.C. Partridge 1; D.M. Absher 1; G.S. Barsh 1; B.R. Korf 2; G.M. Cooper 1

1) HudsonAlpha Institute for Biotechnology, Huntsville, Alabama; 2) University of Alabama at Birmingham, Birmingham, Alabama

The Alabama Genomic Health Initiative (AGHI) is a state-wide effort to conduct genetic testing for ~10,000 individuals not ascertained for any phenotype. We are using an Illumina Global Screening Array (GSA) to detect pathogenic/likely pathogenic (P/LP) variation in medically actionable disease genes; predominantly associated with cancer/cardiac risk, that overlap largely the 59 actionable genes defined by the American College of Medical Genetics and Genomics.

To date, 4,355 participants have been analyzed. 63 (1.4%) harbor a P/LP variant deemed suitable for return, comprising 48 unique variants across 17 genes. P/LP variants were validated by CAP/CLIA-certified Sanger before return to participants.

Array-based genotyping is more efficient and cost-effective than sequencing, but does have limitations in terms of sensitivity and specificity. We found that ~60% of secondary P/LP variants found from sequencing ~1,000 healthy individuals are present on the array, indicative of a sensitivity upper limit. To assess sensitivity of P/LP variants on the array, we tested 24 individuals known to harbor a medically actionable variant targeted by the GSA; all expected variants were identified, suggesting high sensitivity to array-targeted variants.

We also assessed specificity. Hundreds of potential P/LP variants on the array are seen more frequently in AGHI than reported in databases, or is plausible given associated disease prevalence, suggesting false positives (FPs). After filtration to remove obvious FPs, the remaining P/LP variants in actionable genes were manually curated and Sanger tested. We found that 41 unique P/LP variants, in 39 individuals, detected by GSA were FPs; added to the 48 validated P/LP variants (above), these data indicate a false discovery rate among array-identified P/LP variants of ~46%, even after extensive automated filtration. In 28 of 39 individuals, Sanger detected benign missense/synonymous variation at, or nearby, the tested position, indicating that the array detected a non-targeted variant. For 11 individuals, no nearby alteration was identified by Sanger; reason for array detection is unknown.

Through AGHI, we have implemented an array-based process to screen a population of Alabamians to cost-effectively identify highly penetrant genetic variants in actionable disease genes. We demonstrate the need for clinical follow-up of array-identified variants, especially when considered in the context of direct-to-consumer genetic testing.