1. www.creal.cat EXPERIENCES IN POOLING OF DATA: OPPORTUNITIES FOR HARMONIZATION OF I4C COHORTS Manolis Kogevinas, MD, PhD CREAL, Barcelona and National School of Public Health, Athens

2. Outline Ad hoc and wider collaborations New cohorts, need to harmonise without restricting new ideas Pooling or meta-analysis

3. ad hoc collaborations between mother-child cohorts Most collaborations are ad hoc around a specific project with specific hypotheses and normally short term (max 5 years) –NewGeneris: genotoxicity-biomarkers –Obelix: obesity –Earnest; nutrition and programming –Mobi-Kids: mobile phones, late childhood, young adults –HiWate: water contaminants –ESCAPE: air-pollution –EnviroGenoMarkers: environment and omics

4. Mother-Child Birth Cohorts and Biobanks in NewGeneris

5. Micronuclei, a biomarker predictive of future cancer risk

6. NewGeneris. Adjusted mean birth weight (grams) and 95% confidence intervals per acrylamide Hb adduct quartiles in cord blood (M Pedersen, ISEE2011) Adjusted for gestational age and country; n=903

7. Two initiatives on coordination of European birth cohorts funded by the EU (FP7) Enrieco: environmental health Chicos: all exposures and policy Main task coordination, but also several case studies

8. Inventory of European cohorts with data on environmental exposures. ENRIECO project (www.enrieco.org)

9. Expand ENRIECO/CHICOS worldwide Harmonization and websearch achieved in Europe Know-how developed Expand worldwide (using I4C as platform for doing this?) Funding: EU, RO1, other?

10. New cohorts/1 Necessary for I4C objectives (numbers) Increases exposure variation (and protocol variation) Extremely positive commitment for long term funding Extremely positive commitment to evaluate environmental exposures

11. New cohorts: questionnaires Development of new protocols and questionnaires coordinated with older cohorts, but allowing innovation –Basic exposure and outcome variables and biological material should be combinable –e.g. asthma/allergies: inclusion of main ISAAC questions absolutely necessary but some cohorts may also include new more detailed questions

12. During your entire lifetime, have you ever smoked a total of 100 cigarettes or more (which is 5 or more packs)? YES1 NO(SKIP TO B19)2 DK(SKIP TO B19)8 Did you ever smoke cigarettes regularly, that is, at least one per day for six months or longer? YES1 NO(SKIP TO B19)2 DK(SKIP TO B19)8 New cohorts: questionnaires- cultural adaptation

13. New cohorts: mother–child cohorts are complex projects Absolute need for multidisciplinary group leading each cohort including a very strong group in epidemiology

14. Pooling or meta-analysis?

15. Some analyses can be done through meta-analysis, i.e. without tranfer of raw data Depending on hypotheses and adjustment factors individual cohorts may need to run hundreds of prespecified models that should then be meta- analysed Some analyses impossible to do through meta- analysis, e.g. Gene-environment interactions or pathway analyses Ethical considerations for data transfer

16. Genomewide Association Study of Asthma - GABRIEL Consortium (Moffatt, NEJM 2010)

17. One or two stages analytical approach to detect GxE interactions 1.a global genome-wide analysis 2.a two-stage approach-first G x exposure (Murcray AJE 2009): - evaluate association of SNPs with exposure; select SNPs with p-value<10E-05 - then GxE

18. Association tests Logistic Regression-GE When information on the exposure is available both in cases and controls Test of GxE interaction effect LRT (1 df-χ²) : H1 : Logit P(M|G,E) = β 0 + β E E + β G G + β GE GE H0 : Logit P(M|G,E) = β 0 ’ + β E ’ E + β G ’ G

19. 2 Step-Analysis to identify genes involved in GxE Murcray et al. Am J Epidemiol 2008 Step 1: Screening test: to find SNPs most likely to be involved in a GxE interaction by testing for G-E association Case only analysis (combined case/control sample ) For each of N SNPs: LR Test for association between G and E → Select m SNPs with P <  1 Step 2: Case-Control analysis LR Test for GxE applied to m SNPs selected at step 1 → Significance based on P <  /m Comparison with classical one-step approach applied to case-controls → Significance based on P <  /N

20. Power for one-step and two-step analyses to detect GxE for varying levels of interaction effect size 10,000 markers and 500 cases/500 controls Murcray et al. Am J Epidemiol 2009;169:219-26

21. Metabolism of benzo[a]pyrene by CYPs and other drug-metabolizing enzymes (IARC 2010)

22. Pathway analysis: Gene set enrichment analysis (GSEA) -example European adult asthma cohorts 488,058 variants included mapping 16,656 genes 258 gene sets evaluated

23. Pathway analysis, example European adult asthma cohorts Gene set nameSourceGene set p- value Gene set FDR q value Sign. genes Selected gene # All gene # Cell adhesion molecules KEGG≤0.0010.01871122134 Tight junctionKEGG≤0.0010.02860118136 Type I diabetes mellitus KEGG≤0.0010.168284245 Wnt signalling pathway KEGG≤0.0010.17571136149 Antigen processing and presentation KEGG0.0020.228276483

24. Expand ENRIECO/CHICOS worldwide Harmonization and websearch achieved in Europe Know-how developed Expand worldwide (using I4C as platform for doing this?) Funding: EU, RO1, other?