Blog @ Illumina
Real scientists. Real commentary.

AGBT 2013, Day 2: WGS Trends

Abizar Lakdawalla, Ph.D.
| Feb 23, 2013

whole genomeMany of the talks today focused on the positive impact and challenges of NGS in the clinic. Jon Siedman (Harvard Medical) showed that NGS based sequencing of 150 exons in 10 genes plus the massive titin gene provides accurate, pre-clinical diagnosis for inherited cardiomyopathies in high-risk families. Christine Eng (Baylor College of Medicine), showed the clinical utility of exome sequencing, achieving  a 25% diagnostic rate, in spite of high level of stringencies used in filtering the variants. She mentioned that even if diagnosis was not always clear, just stopping the continuous diagnostic testing odyssey is a big win for her patients.

The biggest challenge in implementing clinical sequencing was having a standardized method for filtering SNVs for clinical relevance. Quite often, published associations may be erroneous. Liz Worthey (Medical College of Wisconsin) specifically stated that genotype-phenotype associations in extant databases need to be urgently re-curated. Zivana Tezak (FDA) expected clinical performance to be straightforward if there are well-known associations in literature. If no association exists in the literature, then a clinical trial would be required.

Out of the many variants discovered by whole genome, exome or targeted sequencing, the vast majority are not clinically reportable. For example, Liz Worthey's group selected about 100 variants from whole genome sequencing for in depth review, about 1/4 had errenous associations in the literature (mutation present, but the associated phenotype is not seen) and often there is insufficient evidence in the literature. As an example, whole-genome sequencing from a 4 year-old with chronic and debilitating diarrhoea and episodes of sepsis showed that the SKIV2L gene ( involved in 3' mRNA degradation) may be the causative mutation, but there was no evidence to support the gene-disease association, until a serendipitously timed publication reporting mutations in the same gene in 6 individuals with trichohepatoenteric syndrome supported this diagnosis. Unfortunately, no guidance for treatment was available. In a 180 to the above example, Liz presented the analysis on a 4 month-old with a large spinal hemorrage causing paralysis, multiple myofibrotomas, soft tissue laxity, and complex arteriovenous malformations who had been tested extensively until whole genome sequencing identified a de novo mutation in a gene previously identified with hereditary HT. This variant was NOT reported, as the evidence was considered to be insufficient.

In spite of the large number of variants produced by gene panel or exome sequencing requiring interpretive filtering Liz Worthey was crystal clear on the utility of whole-genome sequencing as compared to exome or targeted assays due to the risk of missing a phenotype causing variation. The dearth of family or medical history are also compensated by the inclusiveness of whole genome sequencing though the long process time to generate and interpret WGS data is a big hindrance. For newborns, treatment intervention often starts in the first 48 hrs, and quite often data is delivered after the death of the patient. Sadly, a more rapid pipeline would have saved lives. At the Medical College of Wisconsin, they are transitioning to a 39 hr timeline from a 2 week pipeline in 2012. Their new rapid pipeline will be based data from the HiSeq 2500 combined with software from Genelogics, Bina for secondary analysis, and CHILIAD for interpretation.