Common Disease Findings (case study on diabetes)

GWAS Workshop

Francis S. Collins, M.D., Ph.D.National Human Genome Research Institute

May 1, 2007

No Data <4% 4-4.9% 5-5.9% ≥6%

19941994 20042004

Estimates of Diagnosed Diabetes Among Adults in the U.S.

Diabetes Pathophysiology

Genetic Predisposition Environmentgenetic defects

Beta-cell Dysfunction Insulin Resistance

impaired insulin secretion

poor glucose utilization

obesity

TYPE 2 DIABETES

BETA-CELL EXHAUSTION

HYPERGLYCEMIA

Type 2 Diabetes:“The geneticist’s nightmare”

• Family history as a substantial risk factor– Relative risk to a sibling ~3

• Environment as a major contributor• Family linkage studies relatively disappointing• Validated genes prior to 2007:

– PPARG (candidate gene)

– KCNJ11 (candidate gene)

– TCF7L2 (linkage study)

Applying Genome Wide Association to Type 2 Diabetes

• Three groups, each with 1000 – 1500 cases and 1000 – 3000 controls in Stage 1:– FUSION (Boehnke, Bergman, Collins, Mohlke, Tuomilehto)– Diabetes Genetics Initiative of Broad, Novartis, and Lund

(Altshuler, Groop)– Wellcome Trust Case Control Consortium/UK Type 2

Diabetes Consortium (McCarthy, Hattersley, Donnelly)

• Genotyped with Illumina 317K or Affy 500K panel• Compared results across all three studies• Followed up promising signals in Stage 2 validation set

FUSION Two-Stage GWA Study of T2DStage 1: GWA

Stage 2: Validation of best hits

Cases

NGT

Controls Total

Familial /

population-based cases (%)

Stage 1 1161 1174 2335 68/32

Stage 2 1215 1258 2489 0/100

Total 2392 2432 4824 27/73

80% power to detect:• Stage 1: OR = 1.4 - 1.5 • Stage 1 + 2: OR = 1.3 - 1.4

• 317,503 SNPs genotyped on Illumina HumanHap300 BeadChip at the Center for Inherited Disease Research (CIDR)

• Dropped 1,868 SNPs from analysisFailed to meet expectations for Hardy-Weinberg< 90% successful genotypes> 3 Mendelian or duplicate sample errors< 10 detections of the rare allele

GWA Genotyping of Stage 1 Samples

FUSION Stage 1 Association Results-l

og

10(

p-v

alu

e)

1161 Finnish T2D cases + 1174 Finnish normal glucose tolerant controls

No results meet genome-wide significance (p < 1.7 x 10-7)

Known associations can serve as positive controls-l

og

10(

p-v

alu

e) TCF7L2

KCNJ11PPARG

1161 Finnish T2D cases + 1174 Finnish normal glucose tolerant controls

Three Groups Working Together

Sweden

Poland

United States

(off map)

# cases + controls

FUSIONS1: 1161 + 1174S2: 1215 + 1258

DGIS1: 1464 + 1467S2: 5065 + 5785

WTCCC/UKT2DS1: 1924 + 2938S2: 3757 + 5346

Totals S1 = 4549 + 5579S2 = 10053 + 12389 (n=32,554)

Combined Results Now Highly Significant-l

og

10(

p-v

alu

e)

Chromosome

15

10

5

0

49 TCF7L2

KCNJ11

PPARG

//

Top 10 Results From Combined Analysis

FUSION DGI WTCCC/UKT2D All Samples

Gene OR p-value OR p-value OR p-value OR p-value

TCF7L2 1.34 1.3 x 10-8 1.38 2.3 x 10-31 1.37 6.7 x 10-13 1.37 1.0 x 10-48

IGF2BP2 1.18 2.1 x 10-4 1.17 1.7 x 10-9 1.11 1.6 x 10-4 1.14 8.9 x 10-16

CDKN2A/B 1.20 .0022 1.20 5.4 x 10-8 1.19 4.9 x 10-7 1.20 7.8 x 10-15

FTO 1.11 0.016 1.03 0.25 1.23 7.3 x 10-14 1.17 1.3 x 10-12

CDKAL1 1.12 0.0095 1.08 0.0024 1.16 1.3 x 10-8 1.12 4.1 x 10-11

KCNJ11 1.11 0.013 1.15 1.0 x 10-7 1.15 0.0013 1.14 6.7 x 10-11

HHEX 1.10 0.026 1.14 1.7 x 10-4 1.13 4.6 x 10-6 1.13 5.7 x 10-10

SLC30A8 1.18 7.0 x 10-5 1.07 0.047 1.12 7.0 x 10-5 1.12 5.3 x 10-8

Chr 11 1.48 5.7 x 10-8 1.16 0.12 1.13 0.068 1.23 4.3 x 10-7

PPARG 1.20 0.0014 1.09 0.019 1.23 0.0013 1.14 1.7 x 10-6

Top 10 Results From Combined Analysis

FUSION DGI WTCCC/UKT2D All Samples

Gene OR p-value OR p-value OR p-value OR p-value

TCF7L2 1.34 1.3 x 10-8 1.38 2.3 x 10-31 1.37 6.7 x 10-13 1.37 1.0 x 10-48

IGF2BP2 1.18 2.1 x 10-4 1.17 1.7 x 10-9 1.11 1.6 x 10-4 1.14 8.9 x 10-16

CDKN2A/B 1.20 .0022 1.20 5.4 x 10-8 1.19 4.9 x 10-7 1.20 7.8 x 10-15

FTO 1.11 0.016 1.03 0.25 1.23 7.3 x 10-14 1.17 1.3 x 10-12

CDKAL1 1.12 0.0095 1.08 0.0024 1.16 1.3 x 10-8 1.12 4.1 x 10-11

KCNJ11 1.11 0.013 1.15 1.0 x 10-7 1.15 0.0013 1.14 6.7 x 10-11

HHEX 1.10 0.026 1.14 1.7 x 10-4 1.13 4.6 x 10-6 1.13 5.7 x 10-10

SLC30A8 1.18 7.0 x 10-5 1.07 0.047 1.12 7.0 x 10-5 1.12 5.3 x 10-8

Chr 11 1.48 5.7 x 10-8 1.16 0.12 1.13 0.068 1.23 4.3 x 10-7

PPARG 1.20 0.0014 1.09 0.019 1.23 0.0013 1.14 1.7 x 10-6

Top 10 Results From Combined Analysis

FUSION DGI WTCCC/UKT2D All Samples

Gene OR p-value OR p-value OR p-value OR p-value

TCF7L2 1.34 1.3 x 10-8 1.38 2.3 x 10-31 1.37 6.7 x 10-13 1.37 1.0 x 10-48

IGF2BP2 1.18 2.1 x 10-4 1.17 1.7 x 10-9 1.11 1.6 x 10-4 1.14 8.9 x 10-16

CDKN2A/B 1.20 .0022 1.20 5.4 x 10-8 1.19 4.9 x 10-7 1.20 7.8 x 10-15

FTO 1.11 0.016 1.03 0.25 1.23 7.3 x 10-14 1.17 1.3 x 10-12

CDKAL1 1.12 0.0095 1.08 0.0024 1.16 1.3 x 10-8 1.12 4.1 x 10-11

KCNJ11 1.11 0.013 1.15 1.0 x 10-7 1.15 0.0013 1.14 6.7 x 10-11

HHEX 1.10 0.026 1.14 1.7 x 10-4 1.13 4.6 x 10-6 1.13 5.7 x 10-10

SLC30A8 1.18 7.0 x 10-5 1.07 0.047 1.12 7.0 x 10-5 1.12 5.3 x 10-8

Chr 11 1.48 5.7 x 10-8 1.16 0.12 1.13 0.068 1.23 4.3 x 10-7

PPARG 1.20 0.0014 1.09 0.019 1.23 0.0013 1.14 1.7 x 10-6

How could IGF2BP2 be related to diabetes?

• IGF2BP2 codes for the insulin-like growth factor 2 mRNA binding protein 2

• But we know very little about what this does

• A related gene, IGF2BP1, codes for a protein that binds to the upstream leader sequence of the insulin-like growth factor 2 (IGF2) mRNA and regulates IGF2 protein production

• IGF2 is involved in development, growth, and stimulation of insulin action

rs13266634is in the codingregion of SLC30A8, andchanges a highlyconservedarginine to atryptophan

SLC30A8 - Beta Cell Zinc Transporter

Chimienti (2005) BioMetals 18:313

SNPs Upstream of CDKN2A/B

• Cyclin-dependent kinase inhibitors CDKN2A and CDKN2B, also known as p16INK4a and p15INK4b

• Inhibit the activity of cyclin-dependent protein kinases CDK4 and CDK6, have been implicated repeatedly in cancer.

• Cdk4 activity also influences beta cell proliferation and mass in mice; loss of Cdk4 leads to diabetes

Embargoed until 5 PM Thursday M

ay 3!

SNPs Within Introns of CDKAL1• CDK5 regulatory subunit associated protein 1-like 1

• Shares protein domain similarity with CDK5RAP1, which inhibits CDK5

• CDK5 activity is influenced by glucose, stimulates insulin gene transcription, and may influence other beta cell processes

• Over activity of CDK5 in pancreas may lead to beta cell degeneration