English Longitudinal Study of Ageing (ELSA): An example of data use from the ELSA DNA Repository Meena Kumari 17 November 2008

ELSA DNA Repository (EDNAR) HSE ’98, ’99, 2001 Wave 1 (2002-3) 11391 Wave 2 (2004-5) Wave 2 nurse 7666 Wave 3 (2006-7) DNA consent 6551 Refresher 1892 issued Wave 4 (2008-9) Wave 4 nurse

AIMS: EDNAR Using genetics to understand social and psychosocial processes impact health Gene-environment interactions Using genetics to understand biological pathways Mendelian randomisation approach contribution of data to the wider academic community Within UCL as part of London based consortium (n=35,000) Wider academic community in the form of the repository Genetic repository (EDNAR) m.kumari@ucl.ac.uk

Progress in the EDNAR Funded in November 2005 by NIA 16 applications (as of October 2008) 1,450 SNPs measured 2 papers published Caulfield et al., Plos Med. 2008 Rice et al., J. Allergy. Exp. Immunol. 2008 4 papers under review

Randomisation to test causality the Mendelian randomisation approach Drug intervention Genetics RCT Mendelian randomisation Sample Population Randomisation Random allocation of alleles Intervention Control Genotype aa Genotype AA Biomarker lower Biomarker higher Biomarker lower Biomarker higher CV event rate lower CV event rate higher CV event rate lower CV event rate higher

Application of mendelian randomisation Separating the mechanism-based and off-target actions of torcetrapib using cholesteryl ester transfer protein gene polymorphisms (Sofat et al., under review) Torcetrapib is a drug that raises ‘good’ cholesterol by its action on cholesteryl ester transfer protein (CETP) Recently the ‘Illuminate’ trial was stopped because the drug was found to have increased adverse cardiovascular outcomes in the treatment arm compared to the control arm Torcetrapib associated with increases in systolic and diastolic blood pressure

Adverse effects of torcetrapib on CVD risk factors Is the adverse effect due to the mechanism of the drug or something ‘off target’ about this particular drug? Can we use genetic variation in the CETP gene to understand whether the effect of Torcetrapib is due to its mechanism of action (‘on target’) or to an idiosyncracy of the drug itself (‘off target’)? Is genetic variation in the CETP gene associated with changes in systolic and diastolic blood pressure?

Association of CETP TaqIB and CVD risk factors Examination of the association of CETP TaqIB (B1B1, B1B2 vs B2B2) and -629C>A variants (rs708272 and rs1800775 respectively on CETP HDL-cholesterol (‘good’ cholesterol) Diastolic and systolic blood pressure A total of 31 studies and 67,687 individuals of mean age 55.8 (SD 9.6) years

Association of genetic variation in CETP and CETP protein Genotype stratified by ethnicity Caucasian Japanese Mean Difference (95% CI) B1B2* B2B2* Individuals (No of studies) 2,763 (6) 1,149 (5) 4,086 (6) 750 (5) p value for 2 test of heterogeneity CETP Concentration mg/ ml <0.001 a. -0.23 (-0.32, -0.14) -0.24 (-0.32, -0.16) -0.47 (-0.67, -0.26) -0.52 (-0.74, -0.31) -0.25 -0.5 -0.75

Association of genetic variation in CETP and HDL-cholesterol 0.06 0.06 (0.05, 0.07) 0.06 (0.05, 0.06) 0.13 (0.11, 0.14) 0.16 (0.10, 0.22) 0.12 (0.08, 0.15) 0.10 (0.08, 0.13) 0.14 (0.12, 0.17) 0.11 (0.09, 0.13) 0.16 (0.14, 0.18) 0.12 (0.10, 0.14) 0.12 (0.11, 0.14) 0.15 (0.10, 0.20) 0.13 B1B2 v B1B1 Caucasian B1B2 v B1B1 Japanese B2B2 v B1B1 Caucasian B2B2 v B1B1 Japanese >1000 <1000 Affected Unaffected Mixed-affected & unaffected Male only Female only Males and Females rs708272 rs1800775 Stratified analyses: Caucasians only, B2B2 v B1B1 Study Size Baseline coronary disease status Gender SNP 54,971 (30) 1,876 (6) 34,432 (30) 1,179 (6) 31,772 (20) 3,664 (10) 4,385 (9) 6,538 (10) 23,083 (13) 12,822 (16) 6,343 (9) 16,749 (12) 33,208 (27) 2,706 (3) Group Comparisons Mean Difference (95% CI) Individuals (No of studies) HDL-Cholesterol p value for 2 test of heterogeneity 0.002 <0.001 0.008 0.494 0.198 0.003 mmol/L c.

Is CETP variation associated with blood pressure? -0.27 (-0.64, 0.10) 0.16 (-0.28, 0.60) 0.23 (-0.02, 0.69) -0.47 (-1.90, 0.95) -0.16 (-1.64, 1.33) -0.10 (-0.10, 0.90) 0.28 (-0.24, 0.80) 0.15 (-0.55, 0.85) -0.35 (-1.59, 0.89) 0.31 (-0.38, 0.99) 0.23 (-0.22, 0.69) -0.80 (-2.49, 0.90) -0.74 (-1.86, 0.38) 0.15 (-1.93, 2.23) -4 -2 2 4 B1B2 v B1B1 27,877 (20) Low LDL High LDL B2B2 v B1B1 >1000 <1000 Affected Unaffected Mixed-affected & unaffected Male only Female only Males and Females rs708272 rs1800775 Stratified analyses B2B2 v B1B1 Study Size Baseline coronary disease status Gender SNP 46,412 (21) 29,050 (21) 28,047(16) 1,711(6) 2,551(3) 4,312(5) 23,184(14) 9,489 (11) 4,793 (6) 15,270 (11) 2,070 (2) SBP by LDL level 6,596 (6) 6,587 (6) Group Comparisons Individuals (No of studies) Mean Difference (95% CI) p value for 2 test of heterogeneity 0.72 0.65 0.26 0.46 0.36 Systolic Blood Pressure 0.15 mmHg a

Is CETP variation associated with blood pressure? -0.23 (-0.43, -0.04) -0.04 (-0.35, 0.28) -0.00 (-0.27, 0.26) 0.02 (-0.15, 1.62) -0.60 (-1.42, 0.23) -0.09 (-1.03, 0.86) 0.08 (-0.28, 0.44) -0.12 (-0.60, 0.35) -0.02 (-0.07, 0.67) -0.02 (-0.58, 0.53) 0.05 (-0.26, 0.36) -1.13 (-2.08, -0.17) -0.58 (-1.69, 0.54) 0.24 (-0.37, 0.84) -4 -2 2 4 B1B2 v B1B1 27,877 (20) Low LDL High LDL B2B2 v B1B1 >1000 <1000 Affected Unaffected Mixed-affected & unaffected Male only Female only Males and Females rs708272 rs1800775 Stratified analyses B2B2 v B1B1 Study Size Baseline coronary disease status Gender SNP 46,412 (21) 29,050 (21) 28,047(16) 1,711(6) 2,551(3) 4,312(5) 23,184(14) 9,489 (11) 4,793 (6) 15,270 (11) 2,070 (2) SBP by LDL level 6,596 (6) 6,587 (6) Group Comparisons Mean Difference (95% CI) Individuals (No of studies) p value for 2 test of heterogeneity Diastolic Blood Pressure 0.33 0.86 0.02 0.21 0.10 0.29 0.55 mmHg b

Comparing the effect of gene and drug DBP 5mg B1B2 allele 10mg B2B2 allele -0.50 -0.25 0.00 0.25 0.50 Diastolic Blood Pressure (mmHg) 1.0 SBP 0.5 Systolic Blood Pressure (mmHg) 0.0 -0.5 -1.0 5mg B1B2 allele 10 mg B2B2 allele HDL-C 5 mg B1B2 allele 10 mg B2B2 allele 0.025 0.075 0.125 0.175 HDL (mmol/L) Observed from genetic studies Expected, as calculated from trials

Conclusions 1. Discordance in the effect of CETP SNPs and torcetrapib treatment on blood pressure, despite the concordant effects of gene variants and drug on eight blood lipid and lipoprotein traits indicates that the hypertensive effect of torcetrapib is unlikely to be due to CETP-inhibition. 2. The findings are important for regulators and manufacturers considering randomised trials of other CETP inhibitor molecules in development. 3. Using genetic studies as a type of natural trial could have wider application in drug development, helping to validate targets, model drug effects, and distinguish on and off-target effects in man.