Analysis of large populations: The Bridge study

Suthesh Sivapalaratnam

On behalf of the ThromboGenomicsand BRIDGE-BPD consortia

University of CambridgeBarts and the London Trust

Disclosures for In compliance with COI policy, ISTH requires the following disclosures to the session audience:

Research Support/P.I. No relevant conflicts of interest to declare

Employee No relevant conflicts of interest to declare

Consultant No relevant conflicts of interest to declare

Major Stockholder No relevant conflicts of interest to declare

Speakers Bureau No relevant conflicts of interest to declare

Honoraria No relevant conflicts of interest to declare

Scientific Advisory Board No relevant conflicts of interest to declare

Presentation includes discussion of the following off-label use of a drug or medical device:NA

- 2 -

Overview

• Bridge - 10K Pilot of 100K study• The Bridge-BPD Study• Two recent findings

Rare diseases – Unmet medical need

• Affect 1 in 20 individuals in the UK.• There are more than 7000 rare diseases.• The genetic basis of only half of these has been resolved.• It takes currently on average 2.5 years to reach a diagnosis.

Objectives:

• To identify the cause of disease

• To improve the rate of diagnosis

(High-Throughput Sequencing Techniques method development)

The NIHR BioResource - Rare Diseases

• Funded by the NIHR

• Aims to establish a comprehensive BioResource of participants with rare diseases, and their relatives

• Works in partnership with Genomics England Ltd to deliver the 100,000 Genomes Project

10K The Study

• Started in 2013• 8,000 patients• 15 rare disorder categories• 50 NHS Trusts• Whole Genome Sequencing

in a clinically accredited lab• Clinical feedback + research

• Plus RD pilot for the 100 000 Genome Project

The process

Recruitment QC, WGS, Data QC

Analysis

Validation, Co-segregation,

Papers

Pertinent Findings

Phenotyping

Research Reports

Breakdown of recruitment sites

68 unique sites recruiting

16 non UK sites

Project Sites recruiting Non UK sitesBPD 37 12CSVD 9 0EDS 6 0HCM 4 0ICP 15 0LHON 1 0MPMT 9 0NPD 8 0PAH 11 2PID 28 4PMG 2 0SMD 10 1SPEED 8 0SRNS 2 0

Breakdown of genetically determined ethnicity

0 20 40 60 80 100

BPD

CNTRL

CSVD

GEL

HCM

ICP

LHON

MPMT

NPD

PAH

PID

PMG

SMD

SPEED

SRNS

Total

Overall Ethnicity of subjects recruited

European 88.6%African 2.6%East Asian 1.1%South Asian 7.7%

UK 2011 Census data Ethnic groups

White 87.1%Asian 6.9%Black 3.0%Mixed 2.0%Other 0.9%

Predicted ethnicity by principal component analysis compared to 1000 genomes ethnic control groups

0% 20% 40% 60% 80% 100% 120%

SPEED

PID

BPD

LHON

PAH

CSVD

PMG

HCM

MPMT

SRNS

EDS

ICP

NPD

SMD

Progress for each project

WGS quota

% of quota sent to WGS (end June 2016)

100%

600

250

270

400

250

700

300

213

250

1250

70

1250

1250

1250

NIHR BioResource - Rare Diseases

WGS data quality and

statistics

Typical WGS10k genome coverage profile

Coverage is the average number of reads representing a given nucleotide in the reconstructed sequence

For detection of SNPs and rearrangements, publications recommend from 10× to 30× depth of coverage

Example WGS10k project coverage vs WES

Minimum percentage genome coverage%

gen

ome

cove

red

More than 90% of the genome has 25x coverage

Batch data and average variation counts per participant

Average 3.9 Million SNVs per personAverage 3.6 Million post QC filtering

55,030,983 unique SNVs detected in project so far

100bp 125bp 150bp

Total number of SNVs per subject : breakdown by ethnicity

Total number of SNVs per genome depends on the population the genome originated from :Populations are based on closest HapMap population using PCA analysis

Batch data and average variation counts per participant

7,733,760 unique Indels detected in project so far

100bp 125bp 150bp 100bp 125bp 150bp

Maximum SNV location/functions in 7204 genomes

Genic SNVs26,118,493

Intergenic SNVs27,092,648

Regulatory SNVs1,819,469

Intronic25,488,758

Splice site39,567

UTR297,383

Synonymous

102,759Non-synonymous

184,946

Exonic590,168

Stop Gain4,443

Start Lost475

Stop Lost162

55,030,983 unique SNVs detected in project so far

Comparison to Imputation panel

MAFs in WGS10K Europeans vs UKBB Imputed panel (150,000)

Almost all the variants common in WGS10K have good quality imputed allele frequencies

Only ~ 7% of the variants in WGS10K with MAFs between 0.1% and 1% in Europeans do not appear in the UKBB imputation panel.

WGS10K MAF

Impu

ted

UKB

MAF

Rare variants do not impute well

Achievements

• More than 50 sites opened and recruiting

• 10968 samples sent to WGS = ¾ BRIDGE finished

• 7204 WGS data

• Great data quality

• Many facilities in place to assist analysis

• Pertinent finding feedback has started

Rare diseases of blood and immune system

Known Genes involved: 466

Bleeding, Thrombotic & Platelet Disorders (BPD)

(Michael A. Laffan/ Willem H. Ouwehand)

Stem Cell & Myeloid Disorders (SMD)

(Irene Roberts)

Primary Immune Disorders (PID)

(Ken Smith/ Adrian Thrasher)

Chen

et a

l, Sc

ienc

e 20

14

Bleeding and platelet disorders (BPD)

• Frequency in general population of – 1:100 (e.g. Von Willebrand’s disease) to 1:1,000,000

(e.g. Prothrombin deficiency) • Symptoms: easy bruising -> life threatening post

procedural bleeds• Laboratory investigations

– Coagulation screen– Platelet count – Platelet function test

• Treatment consist of replacement of factor(s) and platelets

Example - Royal London Haemophilia clinic

- Catchment area of 2,500,000- Clinic of 1000 bleeding and platelet disorder patients- Molecular basis not always known

85 % 15 %

– Diagnosis – Screen family members– Tailoring treatment based on mechanism– Improve understanding of BPD– Develop novel therapeutic options for BPD

Benefit of identifying molecular mechanism

Bridge - BPD

Patient with inherited bleeding and platelet disorder

Known molecular aetiology

Yes

EnrolThromboGenomics

EnrolBRIDGE-BPD

Mutation found?

No

Yes NoReport

Enhance Clinical Phenotype

Pertinent findings/Gene Tiers

Pertinent Findings:Reports can only be generated for variants that are clearly pathogenic (Class 5) or likely pathogenic (Class 4) in a Tier 1 gene.

Gene

Tier 1 Genes that have been shown to be causal in at least 4 independent pedigrees in the literature

Tier 2 Genes where an association with a disorder has been shown but in < 4 pedigrees

Tier 3 Genes suspected of being associated with a disorder

ISTH Tier 1 geneswww.thrombogenomics.org.uk

PLATELET

100 % known genetic disorders(n=159)

92 % suspected etiology(n=61)

10% for uncertain cause(n=76)

Thrombogenomics

Simeoni et al. Blood 2016

Clinical and laboratory phenotypes are registered using Human Phenotype Ontology (HPO) terms

NIHR BioResource - Rare Diseases

Patient with inherited bleeding and platelet disorder

Known molecular aetiology

Yes

EnrolThromboGenomics

EnrolBRIDGE-BPD

Mutation found?

No

Yes NoReport

Enhance Clinical Phenotype

In- and exclusion criteriaPlatelet count <100 x 109/L or >400 x 109/L

Pathological bleeding of unknown aetiology

Abnormal platelet morphology

Highly likely genetic

Mean platelet volume < 6 fl or > 12 fl

Reproducible abnormal platelet function

Acquired BPD including

Bone Marrow Aplasia

Malignancy

TTP/HUS

Medication

HIV and other acute viral infections

Uremia/Hepatic Failure

Splenomegaly

Phenotypic data entry : Paper CRF

Covers BRIDGE fields & Pedigree, No HPO terms

NIHR BioResource - Rare DiseasesPhenotype data requirements

Clinical and Laboratory phenotype to be captured, including HPO terms

Central database for data entry

Pedigree information is essential

The Human Phenotype Ontology (HPO)

General terms

Specific terms

StandardisedControlled terminologyPhenotype not diagnosis

COMPUTABLEWestbury et al, Genome Medicine 2015

General terms

Specific terms

Westbury et al, Genome Medicine 2015

The Human Phenotype Ontology (HPO)

Phenotypic complexity of BPDs

Median HPO terms per case = 7

Number of HPO terms per index case

Freq

uenc

y

Westbury et al. Genome Medicine 2015

Phenotypic complexity of BPDs

Median HPO terms per case = 7

Number of HPO terms per index case

Freq

uenc

y 60% have pathologies outside the blood

Westbury et al. Genome Medicine 2015

HPO coding enabled us to capture the phenotypic complexity of BPD casesneurological, immunological and skeletal disorders are highly over-represented

Westbury et al. Genome Medicine 2015

HPO enabled novel clustering algorithms

50 pedigrees

Westbury et al. Genome Medicine 2015

Roifman n = 4

HPS n = 8

Gorham Stout n = 7

WAS n = 3

50 pedigrees

ACTN1 n=24p = 3.4 x 10-8

PPTHsm n = 4

Westbury et al. Genome Medicine 2015

Blue = BRIDGE-BPD; Green= Other groups

24 New Genes since 2011and another 42 genes to follow

Albers et al, Nat Genetics, 2011; Albers et al, Nat Genetics, 2012; Cvejic et al, Nat Genetics, 2013;Chen et al, Science, 2014; Westbury et al, Genome Medicine, 2015; Green et al, AJHG, 2016; Stritt et al, Blood, 2016; Stritt et al, Nat Communications,2016; Turro et al, Science Translational Medicine, 2016; Simeoni et al, Blood; Lentaigne et al, Blood Accepted

GP1BB

Two examples

• Pedigree – SRC

• Clustering on phenotype similarity– GP1BB

Pedigree with BPD

Thrombocytopenia, large gray platelets, bleeding, abnormal alpha granules, myelofibrosis, skeletal abnormalities

Turro et al. Science Translational Med. 2016

Dysmorphic platelets of variable size and reduced numbers of alpha granules

Control

Patient

Patients

****

HPO in variant prioritisationThrombocytopenia, large gray platelets, bleeding, abnormal alpha granules, myelofibrosis, skeletal abnormalities

67 shared rare variants Turro et al. Science Translational Med. 2016

Thrombocytopenia, large gray platelets, bleeding, abnormal alpha granules, myelofibrosis, skeletal abnormalities

67 shared rare variants

HPO in variant prioritisation

Turro et al. Science Translational Med. 2016

Terms also present in OMIM/MPO for SRC

HPO in variant prioritisation

Turro et al. Science Translational Med. 2016

Gene Prioritization

SRC was the only gene in the top 5 in all 3 methods

HPO phenotype similarity MK-specific up-regulation Pathogenicity score

Turro et al. Science Translational Med. 2016

Src-E527K megakaryocytes have proplatelet formation defect

Increased numbersof CFU-MK and -GEMMcolonies for patients

** **

****

Lentiviral E527K in normal stem cellsZebrafish model that has TP and reduced bone density that is rescued by TKI

Two examples

• Pedigree – SRC

• Clustering on phenotype similarity– GP1BB

Bernard Soulier

• Giant platelets• Thrombocytopenia• Severe bleeding• No ristocetin response

and reduced to thrombin• Autosomal Recessive

Phenotype – Bridge Cases

Variation in GP1BB

Summary

• Largest BPD collection to date• International collaboration• High quality WGS• HPO phenotyping• Combining novel algorithms with available databases• Improving clinical diagnostics• Discovery new players BPD

FundingMembers of BRIDGE Bleeding and Platelet Disorders Consortium