We wish to use the ALSPAC cohort to investigate the effects of novel and previously-confirmed type 2 diabetes loci on fetal growth, growth in childhood and intermediate traits related to type 2 diabetes.

Following our analyses of the first UK type 2 diabetes genome-wide association (GWA) study, which identified 6 new type 2 diabetes genes (FTO, CDKAL1, CDKN2A/2B, HHEX, IGF2BP2, SLC30A8) [1, 2], we have performed a meta-analysis of GWA data, combining our UK study with the DGI [3] and FUSION [4] GWA scans (N=10,128 individuals and ~2.2 million SNPs, directly genotyped and imputed). We followed-up promising signals by performing replication studies of up to 63,532 independent samples [5]. Four loci showed robust evidence for association at Pless than 5x10-8. We are therefore confident that the results exceed the stringent levels of statistical support needed for genetic association studies.

The associated SNPs are as follows:

JAZF1 rs864745

CDC123/CAMK1D rs12779790

ADAMTS9 rs4607103

THADA rs7578597

The precise nature of the biological mechanism by which each of these genes predisposes to type 2 diabetes is largely unknown. We have used replication and robust statistical data, rather than biology, to identify these genes. Further analyses of their roles in intermediate traits, using ALSPAC, are likely to help to uncover further biological pathways.

In addition to the above, we would like to complete the list of currently-confirmed type 2 diabetes loci genotyped in ALSPAC by including the following:

WFS1 rs10010131

PPARG rs1801282

KCNJ11 rs5219

We therefore propose to analyse the polymorphisms in ALSPAC to test the following hypotheses:

1. Fetal genotype and maternal genotype alter fetal growth.

2. Fetal genotype and maternal genotype alter growth velocity in childhood

3. Genotype alters intermediate traits related to type 2 diabetes including fasting insulin, fasting glucose and insulin secretion (in the subset of offspring with OGTT data), triglycerides, HDL, LDL and total cholesterol, anthropometric measures including BMI, lean/fat body mass, WHR, waist circumference, skin folds where available.

4. In the event of associations with intermediate traits in mothers we will propose to test the role of maternal genotype on fetal and childhood growth and body composition and biochemistry. For example, if the SNP alters BMI the SNP can be used in a Mendelian randomisation framework to test the role of maternal BMI in fetal growth and body composition free from confounding factors such as socio-economic status.

Whether the results are negative or positive they will help in our understanding of how the novel genes function and, if positive, provide important insights into growth and other diabetes intermediate phenotypes.

* Specific ALSPAC phenotypes being considered:

To do this we would like to genotype (at Kbioscience) all ~20,000 ALSPAC samples. We will need the following phenotypes to test our hypotheses:

1. Birth weight, length and head circumference

2. Growth measures in childhood (height, weight and BMI aged 7-11)

3. Covariates of birth weight to check if genotype is acting through them: gestational age, maternal age, maternal BMI, smoking , parity, twin status to exclude non-singletons, ethnicity as genotype frequency may alter with ethnic origin and confound analyses.

4. Type 2 diabetes-related intermediate traits including fasting insulin, fasting glucose and insulin secretion (in the subset of offspring with OGTT data), triglycerides, HDL, LDL and total cholesterol, anthropometric measures including BMI, lean/fat body mass, WHR, waist circumference, skin folds where available.

In addition, we are keen to examine the KCNJ11 variant in relation to physical activity in the ALSPAC children. This is to follow up a nominal association observed in the RISC cohort (Mark Walker, personal communication).

Plans for replication:

ALSPAC will provide the largest dataset for association studies with fetal growth. However, we have access, through our own studies and extensive collaborations to samples from the Exeter family study (950 population based parent-newborn trios), Plymouth Earlybird study (300 parent-5 yr old trios - T Wilkin) the North Cumbria community genetics project (3000 mother child pairs - C Relton), the Northern Finnish 1966 Birth Cohort (4600 individuals with own birth measures) and the 1958 cohort (7000 individuals with own birth measures). We hypothesize that real genetic associations will be consistent across all these studies - i.e. even if individually studies show only nominal significance, a meta-analysis of all studies will provide highly significant results.