National Cancer Advisory Board

Genome-Wide Association Studies and the Road Ahead

Stephen Chanock, M.D.

Chief, Laboratory of Translational GenomicsDirector, Core Genotyping Facility

Division of Cancer Epidemiology and Genetics

December 7, 2010

Basic principle of genetic association studies in unrelated individuals

Genome wide SNP chips

Selection of surrogate SNPs to capture common genetic variation

Millions of common SNPs across the genomeUse linkage disequilibrium across the genome

Association with disease risk in discovery followed by replication studiesMapping of susceptibility loci

“What’s past is prologue”William Shakespeare

• In the mid 60’s, DCEG began collection of biospecimens from high risk families.

• This approach was extended in the 80’s to high-quality population studies.

• Scientific progress results from application of new technology.

• Remarkable opportunity for discovery of germ-line contributions to cancer.

PSCA

BNC2

ADH1B

C20orf54

ALDH2

FOXE1

NKX2-1

6p21.33

CTBP2

11p15THADA

EHBP1

ITGA6

3p12.1

EEFSECPDLIM5

TET2

SLC22A3

JAZF1

LMTK2

8p21

8q24.21(x5)

MSMB/NCOA4

11q13 (x3)

HNF1B (x2)

17q24.3

19q13.2

KLK2/KLK3

BIK

NUDT10/NUDT11

38 Prostate

C2orf43

5p15

FOXP4GPRC6A

13q22

10p14

11q23.1

BMP4

GREM1CDH1

SMAD7

RHPN2

BMP2

EIF3H

8q24.21

13 Colorectal

PRKD215q21.315q23 16q24.1

11q24.1

2q13

FARP2

IRF4

8q24.21

CDKN2A/CDKN2B

RTEL1

PHLDB1

TERT

CDKN2A/CDKN2B

CCDC26

5 Glioma

1p36

1q42

KIF1B

TERT/CLPTM1L

7q32

7 Basal Cell Carcinoma

TYR

KRT5

ASIP

MC1R

CDKN2A/CDKN2B

TYRPLCE1

ASIP

4 Melanoma

1q21.1

BARD1

3 Neuroblastoma

6p22

KLF5/KLF12

ABO1q32.1

4 Pancreas

CLPTM1L

ITGA9

7 Nasopharyngeal

GABBR1HLA-FHLA-A

IKZF1

ARIDB5

CEBPE

CHRNA3/CHRNA5

6p21.32

TERT/CLPTM1L

3 Lung

SPRY4

BAK1 KITLG

6 Testicular

GSTM1deletion

6** Bladder

TACC3

NAT2

PSCA

MYCFGFR2

LSP1

TOX3 COX11/STXBP4

RAD51L1

1p11.2

2q35

SLC4A7/NEK10

5p125q11.2

ECHDC1/RNF146C6orf97/

ESR1

19* Breast

4 Pediatric Acute Lymphoblastic Leukemia 4 Esophageal Squamous

2 Thyroid 2 Non-Hodgkin4 Ovary 3 Gastric

CASP8

10 CLL

Published GWAS Etiology Hits: 12.1.10~140 Loci marked by SNPs1 Locus marked by a CNV

Chung & Chanock 2010

ZNF365

10p15.1

10q22.3

DMRT1

ATF7IP

8 Multiple

1q22

1 Liver

2q31

9p13

12q13.13

3q26

1q41TP63

GATA3REL

3 Hodgkins

Profile of Cancer GWAS Regions

• Nearly all map to non-coding regions of the genome

• ~20% map to ‘gene’ deserts• Estimated Effect Sizes (Odds Ratios) are

small (< 1.5)Exception is KITLG in Testicular Cancer

• Multi-Cancer Susceptibility RegionsExamples: 8q24, 5p33.15 (TERT), 11q13

Gene-Environment Interaction for Bladder Cancer Risk: Large International Study Shows NAT2 Slow Acetylation

Increases Risk Only for Smokers

P-interaction = 2.8x10-4 Rothman et al., Nat Genet, 2010

Prostate Cancer Risk Factors2006

• Age• Ethnic Background• Family History

CTBP2

11p15THADA

EHBP1

ITGA6

3p12.1

EEFSECPDLIM5

TET2

SLC22A3

JAZF1

LMTK2

8p21

8q24.21

MSMB/NCOA4

11q13

HNF1B

17q24.3

19q13.2

KLK2/KLK3

BIK

NUDT10/NUDT11

Prostate Cancer (38 as of Nov 2010)

PSA or Prostate Cancer or both??Type 2 Diabetes

No distinction between advanced and early disease

2

3

5

10q11.23

5p15.33

13q22.1

FOXP4

2p24.1

RFX6

3p11

2q27.3

12q13

Prostate Cancer Risk Factors2010

• Age• Ethnic Background• Family History• Multiple common alleles

• Each common variant provides a small contribution

• Inverse Relationship of Type 2 Diabetes & Prostate Cancer

1. Association Finding

2. Mapping Region

3. Assessment of Variants

4. Secondary Association

ATGGAAAGGGAGCATATTACGCACTATGAAGCGGGGGAACGAATGGAAAGGGAGCATATTATGCACTATGAAGCGGGGGAACGAATGGAAAGGGAGCATATTACGCACTATGAAGCGGAGGAACGAATGGAAAGGGAGCATATTATGCACTATGAAGCGGAGGAACGAATGGAAAGGGAGCATATTACGCACTATGAAGCGGGGGAACGAATGGAAAGGGAGCATATTACGCACTATGAAGCGGAGGAACGAATGGAAAGGGAGCATATTACGCACTATGAAGCGGGGGAACGAATGGAAAGGGAGCATATTACGCACTATGAAGCGGGGGAACGAATGGAAAGGGAGCATATTATGCACTATGAAGCGGAGGAACGAATGGAAAGGGAGCATATTATGCACTATGAAGCGGGGGAACGAATGGAAAGGGAGCATATTACGCACTATGAAGCGGGGGAACGA

GWAS or Linkage

Regional Re-Sequence HapMap1000 Genomes Data

Choose Variants for Testing

Follow-up Genotyping in Large Studies

Nominating Optimal Variants for Functional Studies

Follow-up of GWAS Hits

Non-ProteinCoding

BioinformaticAnalysis

ExperimentalStrategy

UnannotatedTranscript

RegulatoryElement

EpigeneticAlteration ofGene Levels Effect on Genes

ElsewhereNovel

Transcripts

Test FunctionalElements

In vitro/vivo

miRNARNASeq

ExpressionQuantitative

Trait LociAnalysis

Analyze HistoneMethylation

Elements

The Majority of GWAS Map to Non-Coding Regions

Functional Elements

10q11.2 & Prostate CancerTop Hit: rs10993994Promoter of MSMB

MSMB= β-microseminoproteinProstate specificSerum marker in studies

20,000 subjects

Risk Allele “T”Lower expression levels

Reporter assaysElectromobility Shift AssaysLevels in Prostate TissueTumor Tissue

TT Decrease Expression in NCI-60 Cell lines

Functional Analysis

10q11.2 Could Be More Complex……MSMB and NCOA4

RNA Expression MSMB and NCOA4 Normal Tumor Tissue

Anchorage Independent Growth is Specific to ProstateMSMB- SuppressionNCOA4- Over-expression

Cancer Susceptibility Loci Across 8q24

MYC209 Kb126 Kb231 Kb 58 Kb

Pros

tate

Reg

ion

2

Pros

tate

Reg

ion

5

Pros

tate

Reg

ion

4

Pros

tate

Reg

ion

3

Pros

tate

Reg

ion

1

Bre

ast R

egio

n

Col

orec

tal R

egio

n*

CLL

Reg

ion

Bla

dder

Reg

ion

Ova

rian

Reg

ion

* Also Colorectal Adenoma

Chung and Chanock

Genome-wide association studies

Association testing

Biometrics Nutrient levelsBehavioral traits

Tobacco Height, Weight, BMI, Vitamins D,B12Caffeine Menarche/Menopause CaroteneAlcohol

International Consortia (e.g., GIANT, SUNLIGHT)

Genome-wide association studies

Relationship testingSex verificationPopulation genetics

and ancestry

Quality Control

Pursuit of X-Chromosome Aneuploidy and Cancer Risk

Genome-wide association studies

Privacy & Confidentiality GWAS membership

Large chromosomal abnormalities, structural variation, aneuploidy in

Germ-line DNA

Rodriguez-Santiago AJHG 2010 Jacobs Nature Genetics 2009

Deeper Understanding

Genetic Predisposition to Breast CancerEuropean Population

1

1.1

10 0.1 0.9

3

10 BRCA1BRCA2

TP53PTENATM

CHEK2

PALB2BRIP1RAD51C

1.2

1.3

1.4

1.5

0.2 0.3 0.4 0.5 0.6 0.7 0.8

Popu

latio

n ge

noty

pe r

elat

ive

risk

Population risk-allele frequency

BCACCGEMS/BCACWTCCCOther

??

E

Risk prediction utility Crohn’s Disease

Sibling relative-risk=20-35Common (BPC) cancersSibling relative risk=2-3

RandomUsing known lociUsing all estimated lociIdeal (if we could explain all heritability) Park et al Nature Genetics 2010

Predicting Prostate Cancer Risk

35 common SNPs from GWASMen between 60 and 65 yearsNCI Breast and Prostate Cohort

Consortium Area Under the CurveFH Alone= 0.52FH + 35 SNPs= 0.66

ROC=Receiver Operator Curve

Status of Genome-Wide Association Studies

Discovery of New Regions in the Genome Associated with Diseases/Traits

• New “Candidate Genes” Clues for Mechanistic Insights Using Common Variants

• Etiology• Gene-Environment/Lifestyle Interactions• Outcomes & Pharmacogenomics

Challenge of Genetic Markers for Risk Prediction for Individual or Public Health Decisions Common Variants Represent a Fraction of the Genetic

Contribution to Risk Integration of Lifestyle/Environment

Acknowledgements

NCI-DCEGJoseph FraumeniGilles ThomasRobert HooverMeredith YeagerKevin JacobsNilanjan ChatterjeeJuHyun ParkSholom WacholderMila Prokunina-OlssonLaufey AmundadottirPeggy TuckerCharles Chung

NCI- OCGDaniela GerhardNCI-CCRMike DeanHong Lou

HSPHDavid HunterPete Kraft

CGEMS StudiesACS (M Thun)ATBC (D Albanes)CAPS (H Gronberg/J Xu)CeRePP (O Cussenot)CONOR (L Vatten)EPIC (E Riboli)JHU (W Issacs/J Xu)MEC (B Henderson)PLCO (R Hayes)WHI (R Prentiss)

DFCIMatt FreedmanMark Pomerantz

Bladder Cancer GWASNat RothmanDebra SilvermanMontse Garcia-Closas

Renal Cancer GWASM PurdueP BrennanM LathropM Johansson

Cohort Consortium Strategic PlanB

iom

edic

al R

esea

rch

Com

mun

ity Cohort Consortium

Large, High-Quality Cohorts

Biospecimen Collection

Open (Participation and Access)

International

Interdisciplinary

Collaborative

Integrated Databases and Biomarkers

New PopulationCohortsSpecific Cancer ProposalsOther DiseaseProposals

Genetic SusceptibilityMarkers (Replication)

Genetic Pathways and Gene-Environment Interactions (Pooling)

Early Detection Markers

Mechanistic Insights(Integrative Biology)

Mapping Complex Diseases:Identifying Genetic Markers

Comprehensive Catalog of Genetic Variation Enables ‘Agnostic’ Search

Frequency of Study Design forSNP Markers DiscoveryCommon >10% GWASUncommon 1-10% GWAS or SequencingRare < 1% Linkage or Sequencing

SNP= Single Nucleotide Polymorphism

Next Generation Sequencing of

Germline

Regional GWAS and linkage follow-up

Candidate exon

Whole genome

Whole exome Functional Work to Define:Direct Association

11q13: Multi-Cancer Susceptibility Region

660 kB across the region

Recombination hotspots and LD structure in 11q13:GWAS Region with no candidate genes

Sampled 900 controls5 times drawn from 10 studies

2.5 M SNP Chip• Beta tested at CGF• High performance at CGF (~750 samples)

– Completion rates of 99% with concordance > 99%• 2.5 M Array

– Average MAF=10.4% (range 3-15%)– Based on 3 algorithms (UofM, Broad and WTSI)– New common regions not covered on prior chips

• 2.5 M S (2nd half of 5 M)– Design begins in November 2010– Projected beta test in Spring 2011

IMPUTATION STUDY @ CGF750 Caucasians (250 of ACS, ATBC & PLCO)

2.5 M SNP array1.0M Omni array1.0M Duo or 660w arrayAffymetrix 6.0

100 AA & 75 Chinese with 2.5 M and 1 M/660wDeposit in dbGaP in Winter of 2010Approximately 3.4 M genotyped SNPs in CEU for

analysis with 1000 CEU data (new download)

Validation for 42 events: 100% Validation Allelic imbalance &- normal dosage UPD- abnormal dosage CNV

5 / 6 events: also found in bladder tissue; early developmental eventsRodríguez-Santiago et al. Am J Hum Genet. 2010;87:129-38

Large Structural Events (>5 Mb) by Chromosome

50,520 Elderly Subjects>550,000 SNPsIllumina Infinium

MDS Region

Germline Mosaic EventsCopy Neutral LOHDuplication/trisomyDeletionUPD