Mendelian randomisation and cardiovascular disease

Background and aims: Homocysteine (Hcy), C-reactive protein (CRP), and Lipoprotein- associated phospholipase A2 (Lp-PLA2) have been associated with a high risk of cardiovascular disease. If casual, they are expected to provide additional tools for prevention. Utilising the unique properties of genetic variants (randomly allocated and unmodifiable), they could be used as unconfounded proxies of environmental exposures to investigate disease aetiology, known as Mendelian randomisation. Herein I conduct a series of Mendelian randomisation experiments to: (i) investigate the role of Hyc in stroke; (ii) judge causality of CRP in cardiovascular disease; (iii) investigate the validity of Lp-PLA2 as a therapeutic target in coronary heart disease (CHD) and; (iv) describe how the integration of cis-acting variants and their cognate proteins can be used to dissect causal pathways. Methods: For aim (i), I conducted synthesis research of published and unpublished studies investigating the MTHFR/C677T variant, Hcy and stroke. For aims (ii) to (iv), a series of prospective collaborations conducting de novo genotyping for CRP and PLA2G7 genes, were established using European-based cardiovascular genetic studies of adults. Results: The meta-analyses of studies on MTHFR/C677T-Hcy-Stroke to 2003, showed that subjects with the TT genotype had on average 1.93 µmol/L higher levels of Hcy compared with subjects with CC genotype, and an odds ratio (OR) of stroke of 1.26 (95%Cl: 1.14,1.40). An update analyses to 2008 showed that in both MTHFR-Hcy and MTHFR-stroke associations, studies in Asia had the largest effect, followed by Europe with intermediate effect, and lower or negligible effect in the Americas and Australasia. Analysis on CRP, indicated that subjects homozygous for the T-allele of CRP/+1444C>T variant despite having 0.68 mg/L higher levels of CRP, had no increase in risk of myocardial infarction (OR of 1.01 [95CI: 0.74,1.38]). A tagging- haplotype approach showed a gradual increase on CRP levels by haplotype, but no effect on CHID, diabetes or stroke. Analysis of the seven PLA2G7 tagging-SNPs showed that the best variant (rs1051931) had small to moderate effects on the Lp-PLA2 activity (up to 7% relative differences). No genetic signal with CHD was observed for any PLA2G7 variant, despite some comparisons including up to 8412 CHID cases. A description of the proof of principle, illustrating how to utilise cis-acting variants and their cognate proteins to distinguish causal from non-causal effects among correlated blood proteins was presented, using as an example 6 proteins that have been associated with CHD risk. Conclusions: Genetic evidence reported in this document suggested that the MTHFR effect is more pronounced in geographic regions associated with low folate intake. Genetic studies on CRP presented in this document indicated that CRP is unlikely to be causal in cardiovascular disease. A genetic approach using common variants in PLA2G7 had a reduced ability to confirm or reject Lp-PLA2 as a valid drug target. Finally, integration of cis-acting variants with their cognate proteins seems to be an effective way of dissecting biological pathways.