ACGS: Standardisation of variant interpretation and reports

Sian Ellard

South West NHS Genomic Medicine Centre

Festival of Genomics, 31st January 2018

Genome sequencing for clinical diagnosis

2018

Genome sequencing

as an NHS test

October 1st 2018

2015

100,000 Genomes Project

Whole genome sequencing is (relatively) straightforward

Variant interpretation is not…

Genetic variants

The human genome consists of 3 billion DNA bases that encode ~20,000 genes

Each genome differs by 3-4 million variants

Each individual has ~150 rare variants that are predicted to affect

protein function (Lek et al Nature 2016 ExAC MAF<0.1%)

There are many different types of variants e.g. missense, splicing,

nonsense, small deletions, insertions, gene copy number changes and chromosome rearrangements

Innovation

Standardisation

Quality improvement

The 100,000 Genomes Project is catalysing laboratory service transformation at pace and scale.

5

4

3

2

1

Pathogenic

Likely pathogenic (>90%)

Uncertain significance

Likely benign (<90%)

Benign

New variant interpretation guidelines from USA 2015

Summary of evidence framework

Richards et al 2015 Genetics in Medicine; Jarvik and Browning 2016 Am J Hum Genet

Rules for combining criteria to classify variants

5

4

3

2

1

Pathogenic

Likely pathogenic (>90%)

Uncertain significance

Likely benign (<90%)

Benign

Variant classification evidence

Assess the evidence for a variant across all patients for which information is available

Publications

Public/NHS databases

Laboratory database

Patient referred for testing

Assessing variant pathogenicity

Search databases: HGMD Pro DECIPHER ClinVar Gene specific Google scholar

Protein analysis: In vitro modelling Functional studies Animal models

Clinical data: Muscle biopsy Nerve conduction studies Drug response Metabolic profile Biochemistry Immunology

Fetus 1 TOP 22/40

Rapidly progressing ventriculomegaly No DNA

Fetus 2 TOP 18/40

Rapidly progressing ventriculomegaly No genetic diagnosis

2

Case example – lethal fetal disorder

POMGNT1 p.Arg497Gln/N POMGNT1 p.Arg497Gln/N

POMGNT1 p.Arg497Gln/p.Arg497Gln

Exome sequencing (23,244 genes) of parental DNA samples

Not homozygous

Evidence: PM1, PM2, PP3, PP1 (1/7 need 1/8)

One report of this variant on ClinVar database An e-mail to the Emory lab provides a key piece of

evidence; the variant is homozygous in a patient with the characteristic phenotype

MDT discussion (referring clinical geneticist and lab scientists with input from external experts) concludes that variant is “likely pathogenic”

Case example – lethal fetal disorder

Evidence: PM1, PM2, PP3, PS4_Supporting, PP4, PP1 (1/7 need 1/8)

Putting the ACMG guidelines into practice

Principles

Use the guidelines as a framework to determine whether there is sufficient evidence to classify the variant as likely pathogenic or likely benign

Incorporate clinical data (functional and gene specific phenotype)

Use your professional judgement

Variant classification and interpretation workshop 4th November 2016

Introduction to ACMG guidelines

Early adopters’ experience

Aligning CNV classification

Integrating clinical phenotype and interpretation

Improving the quality of variant interpretation for clinical diagnosis

Train the trainers: February 28th 2017

24 Regional centres represented

Emma Baple

Monthly WebEx for regional trainers

Cases

Two cases per month circulated

Submit classifications before meeting

Review cases at meeting

Produce worked case examples

On-line resource (Jostle)

Share ideas for local implementation

Disease-specific subgroups

Cancer

Cardiac

Familial hypercholesterolaemia

ACGS guidelines published October 2017

PS3 – (Strong) Well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or

gene product.

Functional studies can include in vitro functional assays (eg. reporter gene assays for transcription factors),

measurement of proteins in vivo (eg. biochemical tests on patient samples) mRNA analysis for suspected splicing

variants and other investigations where the results are pathognomonic of a specific single genetic cause of a disorder.

See Table 2 for a list of examples.

Functional domain or mutation hot spot?

PM1 – (Moderate) Located in a mutational hot spot and/or critical and well-established

functional domain (e.g., active site of an enzyme) without benign variation.

Plot functional domains, ExAC variants, HGMDPro/ClinVar/LOVD etc variants and

Consurf/conservation plots to show reported pathogenic variants, proxy population and

benign variants and amino acid conservation for a region of a gene.

PM1 Plot for SOX10 p.(Ala132Val)

NOTCH2 p.Arg1895His variant in patient with ?Alagille syndrome

http://cardiodb.org/allelefrequencyapp/

The ACMG guidelines are evolving…

Aim

To develop an improved, standardised report template for

clinical genomic tests performed in UK laboratories. The

standardised report template will communicate the results of

clinical genomic tests in a format that facilitates

understanding of the results by a specialist or non-specialist

healthcare professional and the patient or their family.

Developing a standardised report template

Working group: Steve Abbs, Ed Blair, Diana Eccles, Sandi Deans, Sian Ellard

(Chair), Helen Firth, David Gokale, Lucy Jenkins, Tracy Lester, Dom

McMullan, Sian Morgan, Bill Newman, Chris Patch and Richard Scott

F-shaped reading pattern

20-30% text actually read

Standard report template

Standard report template

Appendix

Variant classification and MDT meeting summary form

Standard report template

Thank you

ClinGen team (USA) All the UK regional trainers Reporting template working group members