Blog @ Illumina
Real scientists. Real commentary.

ESHG Satellite Meeting: Clinical & Quality Issues When Introducing New Technologies in Genomics

Scott Brouilette, Ph.D.
| May 30, 2014

the future of NGS in clinical careAhead of the official start of ESHG 2014, a number of satellite meetings have already taken place. One that was particularly relevant in the current climate was "Clinical & Quality Issues When Introducing New Technologies in Genomics". Here, I review some of the key presentations from the session.

The meeting kicked off with Hans Scheffer (Human Genetics, Nijmegen) discussing how improvements in diagnosis using NGS has resulted in changes to clinical practice. Dr Scheffer highlighted some of his own experiences with exome sequencing and the transition from Sanger sequencing; in particular, the need for “more data” to minimize mistakes in calling rare variants from NGS output. Dr. Scheffer also called out deficiencies in existing gene panels and the need for wider collaboration to enable the creation of “better panels” for future studies of particular clinical phenotypes. This is very topical, but does rely on researchers taking a more altruistic approach in the future… a move which the audience seemed broadly supportive of, based on the ensuing Q&A session.

Of course, the issue of incidental findings was raised, but Hans quickly showed that his exome-seq pipeline filters much of this data early on in the 2-step workflow: the first step focuses on known genes to confirm any initial diagnosis, but the exome is then “opened up” if nothing reveals itself. And this clearly highlights the double-edged sword issue with whole-exome versus panels: if the panel misses the causative mutation more sequencing is required (another panel or a move to whole-exome), but whole-exome increases the potential to reveal those incidental findings. As Hans had moved from Sanger to NGS, he discussed the relative diagnostic yields of the two approaches, showing (for his blindness panel) an impressive 58% for exome sequencing vs 25% for Sanger sequencing.

Following Hans was Wendy Jones from the Sanger Institute to discuss clinical reporting based on the Deciphering Developmental Disorders (DDD) project. After a brief overview of the project, she announced that they had recruited an impressive 9,000 families to date. In the early stages of this project, aimed initially at CNV analysis, they were using  array comparative genomic hybridization (aCGH), but Wendy quickly went on to detail how they are now moving over to exome-sequencing of their trios using HiSeq. In doing so, they find ~20,000 variants per exome, then filter this down using various approaches, including comparison to parental genotypes. Finally a clinical review is performed to further filter, typically resulting in between 0-2 variants/exome. For those with access to trios and the resources to sequence the exome, the schematic of this pipeline is certainly a useful resource. The move from arrays to WES has also allowed the team to report not just CNVs, but SNVs too.

Again incidental findings were discussed, but in the DDD project they are not reported back. However, Wendy stated that this is due in part to the inconsistent guidelines relating to incidental findings. As a result, Dr. Anna Middleton, a registered genetic counselor and ethics researcher at Sanger, has initiated a survey to assess attitudes to reporting of incidental findings.

We then moved on to discuss potential quality issues with “new technologies”, jointly presented by Gert Matthijs & Els Dequeker. They began by reiterating the need for quality control at multiple steps of an NGS workflow, including run quality, sample quality, variant calling and interpretation. The overarching issue is how to take existing quality rules and apply them to novel technology (such as NGS) in the clinic? Doing so will require a focus on international quality standards. While there remians an absolute need for accreditation/certification, it was clear that there are already many resources. For example, the Molecular Pathology Checklist of the CAP Document discusses many elements that relate to the introduction of NGS into the clinic.

Standardization to enable cross-lab comparisons was also discussed with the . suggestion of quality metrics at three levels: the technical target (what genomic regions are targeted by the method, which is platform/kit dependent), the clinical target (what regions are clinically relevant), and a transcript list (which genes, based on a specific transcript database, are included). With this information, it should become much easier to make direct comparisons between studies. Bioinformatics pipelines also require standardization for the same reason – and this is covered by a new chapter in ISO15189.

In a later session, Jan Traeger-Synodinos echoed many of these points, highlighting the near complete absence of best-practice guidelines in bioinformatics and stated the need for a “diagnostic-grade .vcf” using standardised variant-calling pipelines. Despite all of the perceived issues, however, the final statement was clear: NGS is currently used or will be used in clinical practice - "there is no way back. Overall, this was a very interesting and informative session with a mixture of clinicians, laboratory scientists, core facility heads and bioethicists. For some, this session may have provided the first awareness of the issues surrounding the utility of NGS in a clinical setting, which only serves to highlight the pressing need to educate and continue the open discussions as we all move forward.