Synthesizing results across multiple GWA studies and integrating them into the existing knowledge base Marta Gwinn, MD, MPH National Office of Public Health.

Synthesizing results across multiple GWA studies and integrating them into the existing knowledge base

Marta Gwinn, MD, MPHNational Office of Public Health Genomics

CDC

Adding Genome-Wide Association to Population StudiesBoston, June 22, 2007

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

DataInformation

Knowledge?

Synthesizing and integrating results of GWA

• Review replication of genetic associations

– Identifying associations vs. measuring effects

– Methodological issues

• Describe network approaches

– Human Genome Epidemiology Network (HuGENet)

– Other examples

• Discuss two important results

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

* HuGE Published Literature

32 39 66 80 109 184

gene-disease association

gene-environment interaction meta-analysis32

Genetic association studies of unrelated personsExtracted from PubMed, 2001-2006*

“We suggest that this rapid, early succession of extreme findings may be called the Proteus phenomenon after the mythologic god who rapidly metamorphosed himself to very different figures.”

• Published literature scan

• Systematic reviews

• Strengthened reporting

• Network collaboration

http://www.hugenet.org.uk

http://www.dhe.med.uoi.gr/hugenet.htm

http://www.hugenet.caHuGENet Canada

Human Genome Epidemiology (HuGE)

www.cdc.gov/genomics/hugenet

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

HuGENet Network

of Networks

Single teamsSingle studies

Published and unpublished data

Systematic reviewsMeta-analyses

Field-widesynopses

Feedback Reporting

SynthesisGrading

Commentary, Nature Genetics 38, 3 - 5 (2006) A road map for efficient and reliable human genome epidemiology

STREGA6/2006

Handbook3/2006

Venice11/2006

Atlanta1/2008

11/2005

Replication of genetic associations

• Heterogeneity – unmeasured factors

• Statistical uncertainty– sampling variability

– low power

• Bias– all the usual epidemiologic biases

– publication bias

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

-- including exposures

-- to detect small effects

-- among many comparisons

Replication of genome-wide associations

• Heterogeneity – unmeasured factors

• Statistical uncertainty– sampling variability

– low power

• Bias– all the usual epidemiologic biases

– publication bias

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

-- genetic background

-- meta-analysis, prior information

-- statistical methods

-- transparency

-- enhanced access to data

HuGE Systematic Reviews and Meta-analysisGTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

• Handbook for systematic reviews / meta-analyses

http://www.cdc.gov/genomics/hugenet/publications/index.htm#guidelines

• Online database of systematic reviews / meta-analyses– >50 reviews published in collaboration with 10 journals– citation database of ~550 meta-analyses

• Proposed guidance for reporting association data

• Proposed criteria for evaluating evidence for association

http://www.cdc.gov/genomics/hugenet

Synthesizing results of GWA studies:different from candidate gene studies?

• Study priorities may differ– GWA: find novel associations

– candidate gene: measure effect size

• Most differences are a matter of degree– Type 1 errors, type 2 errors

– harmonization among studies: phenotyping,

genotyping methods

– population stratification

• All the usual epidemiologic biases prevail

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

Meta-analysis of GWAs?

• Improve power to measure small effects

• Assess heterogeneity among GWAs

• Methodological challenges– use of different genotyping platforms / different SNPs

– harmonization of phenotypic data

– treatment of replication samples within same GWA

• Good for horizontal integration—only one dimension

of evidence

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

Ioannidis JP. Commentary: grading the credibility of molecular evidence for complex diseases. Int J Epidemiol 2006

• Bridge “cottage industry” with “Big Science”

• Domain experts may share: – specific knowledge (e.g., phenotype definitions)– awareness of current research problems– funding sources

• Many networks already exist – NCI Cancer Genetic Markers of Susceptibility (CGEMS )– NINDS Human Genetics Repository – International Collaborative Study on Genetic Susceptibility to Environmental Carcinogens (GSEC) – Preterm Birth International Collaborative (PREBIC)

Why a “Network of Networks?”GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

Hoover RN. The evolution of epidemiologic research: from cottage industry to "big" science. Epidemiology Jan 2007;18:13-7.

Copyright ©2007 by the National Academy of Sciences

Goh, Kwang-Il et al. (2007) Proc. Natl. Acad. Sci. USA 104, 8685-8690

Fig. 2. The HDN and the DGN

The AlzGene database on www.alzforum.org

www.alzforum.org

CARD15 (NOD2) and Crohn’s disease- 2001 positional cloning, candidate gene studies- risk genotype prevalence 0.3, 0.1- relative risk 3, 40- disease risk 0.001

CFH and age-related macular degeneration- 2005 genome-wide association study- risk genotype prevalence 0.2, 0.05- relative risk 5, 7- disease risk 0.05 (by age 60)

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

A Tale of Two Associations

* Note: all estimates are approximate!

0

20

40

60

80

100

120

2000 2001 2002 2003 2004 2005 2006 2007

Year of publication

Pu

bM

ed

art

icle

s

All other genes CARD15 TNF SLC22A4,A5

Genetic Associations with Crohn DiseaseHuGE Published Literature, June 2007

**as of June 2007

***

*indicates number of meta-analyses, including 3 meta-analyses of CARD15

1

1

1

4

• Early success

– “Ideal” combination of genotype prevalence, effect size

and population disease risk?

– Key insights into pathogenesis, phenotype

• Unmet expectations, translation frustrations

– No replication in some populations (Japanese)

– Disappointing clinical trials (infliximab)

• Help from GWA

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

CARD15 (NOD2) and Crohn’s disease

Crohn’s disease: help from GWA

Cardon L. Delivering new disease genes. Science Dec 2006;314:1403-05. Perspective on:Duerr RH, et al. A genome-wide association study identifies IL23R as an inflammatory bowel disease gene. Science Dec 2006:314;1461-63.

0

5

10

15

20

25

30

35

40

45

2000 2001 2002 2003 2004 2005 2006 2007

Year of publication

Pu

bM

ed

art

icle

s

All other genes APOE CFH

Genetic Associations with Macular DegenerationHuGE Published Literature, June 2007

**as of June 2007

**

*

*includes 2 meta-analyses in each group

• Early success of GWA• Insight into pathogenesis and progression

– CFH already related to kidney disease, new focus in cardiovascular disease

– Interaction with smoking and BMI

• Translation? – No interaction with anti-oxidant vitamins/zinc

supplementation (AREDS rx)– No utility for screening

GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

CFH and Age-related macular degeneration

• Why? – increase probability that positive associations are true– increase power to examine small associations, less common variants – zero in on causal variants– start translating data into knowledge

• How?– careful documentation, complete reporting, collaboration– cumulative review of evidence

– integration with other epidemiologic data, as well as knowledge from other fields

Synthesizing and integrating results of GWA GTCGACTGGAGTGTCTGTGAATTGACTTTTGTTGCCAGTTGGCAGCGGCAGAAGCAGCAAAGCCCGGCCAACAGCAACAAGCTCCTGCCAGATCCCAAAAGCAAACACG

Hokusai, Mt. Fuji Off Kanagawa, 1826-1833

Not a tsunami…Not a tsunami--

Hiroshige, A View of Eitai Bridge and Tsukuda Island, 1857

…but a rising tide that lifts all boats.