How Well Do Whole Exome Sequencing Results Correlate with Clinical Findings? A Study of 89 Mayo Clinic Biobank samples. S. Middha1, N. M. Lindor2, S. K. McDonnell1, K. J. Johnson3, J. E. Olson1, E. D. Wieben4, G. Farrugia3, J. R. Cerhan1, S. N. Thibodeau1,5 1) Department of Health Sciences Research, Mayo Clinic, Rochester, MN; 2) Department of Health Sciences Research, Mayo Clinic, Scottsdale, AZ; 3) Center for Individualized Medicine, Mayo Clinic, Rochester, MN; 4) Department of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, MN; 5) Department of Laboratory Medicine and Pathology, Mayo Clinic College of Medicine, Rochester, MN.

   Whole Exome Sequencing (WES) is increasingly used for diagnosis of known Mendelian disorders and for gene discovery. Yet there is inadequate information on the number and type of clinically reportable variants typically found from WES on any given individual and their correlation with clinical phenotype. WES was performed on 89 deceased individuals (mean age at death 74 years) from the Mayo Clinic Biobank. Significant clinical diagnoses were abstracted from each individuals electronic medical record (EMR) via chart review. Variants (SNV, INDEL) were filtered based on quality (accuracy>99%, read-depth>20, alternate-allele read-depth>5, allele-freq<0.1) and available HGMD/OMIM phenotype information. Variants were defined as Tier 1 (stop-gain, splice or frame-shifting) and Tier 2 (missense, predicted damaging by SIFT or PolyPhen). The number and type of variants reported were evaluated in a list of 56 ACMG-reportable genes and 58 cancer predisposition genes, as well as examining overall genotype-phenotype correlations. Following the filtering, we found a total of 6992 variants (79 per person, 623 Tier 1 and 6369 Tier 2). Of these, 161 variants (1.8 per person, 13 Tier 1 and 148 Tier 2) were found among the 56 ACMG-reportable genes, and 115 variants (1.3 per person, 3 Tier 1 and 112 Tier 2) were found among 58 cancer predisposition genes. The number and type of variants in the 58 cancer predisposition genes did not significantly differ between individuals with a history of invasive cancer and those without. For the broader genotype-phenotype correlation, only nine Tier 1 variants were found in autosomal-dominant (AD) inherited genes known to cause a phenotype that correlated with an observation in the individuals EMR. However potential phenotype from 115 additional Tier 1 variants identified in AD genes did not correlate with data in the individuals EMR. We used WES to evaluate potential phenotypes from identified genetic variants and their correlation with the individuals medical records. The list of genes with no evident phenotype correlation was more than 10 times longer than the list of genes with phenotype correlation to EMR. These data highlight challenges that need to be addressed including both phenotype issues (e.g., disease penetrance, uncertainty about what is clinically reportable) and sequencing issues (e.g., incomplete sequencing coverage, thresholds for data filtering, lack of high quality databases to determine functional annotation).

You may contact the first author (during and after the meeting) at