Mendelian randomization in inflammatory conditions- the exception rather than the rule?

Objectives: To assess causality of inflammatory RNA molecules for myocardial infarction (MI) using a Mendelian randomization approach. Methods: Samples were 524 men with a history of MI and 628 controls from the Study of Myocardial Infarction Leiden (SMILE). Macrophage migration inhibitory factor (MIF) G-173C, protein tyrosine phosphatase, non-receptor type, 1 (PTPN1) 1484InsG and chemokine (C-C motif) ligand 3 (CCL3 or MIP1-alpha) C1685T were selected since they are in genes that exhibited elevated mRNA expression measured using multiplex ligation probe-dependent amplification in cases compared to controls. Results: The age-adjusted odds ratio (OR) of MI for MIF G-173C (GG vs CC) was 1.0 (95%CI 0.6-1.8), CCL3 C1685T (TT vs CC) 0.7 (0.4-1.1), and PTPN1 1484InsG (heterozygotes vs wildtype) 0.9 (0.6-1.2). None of the polymorphisms had clearcut effects on circulating levels of its respective mRNA. However, in controls, the MIF -173 CC genotype was associated with higher levels of expression of both inflammatory and anti-inflammatory molecules (PDE4B, TNFRSF1A, IL1RA, TLR4). CCL3 C1685T CT and TT genotypes were associated with higher levels of functionally related CCL4-like transcripts, which are expressed by genes tightly linked to CCL3. Heterozygotes for the PTPN1 1484InsG polymorphism were associated with lower HDL cholesterol (1.18 vs 1.31 mmol/l) and higher triglyceride levels (1.44 vs 1.21 mmol/l). Furthermore smoking and age had effects on some mRNA expression and on the effect of genotypes. Conclusion: MIF G-173C, PTPN1 1484InsG and CCL3 C1685T polymorphisms have no obvious influence on the risk of myocardial infarction. However each of them is associated with more than one intermediate phenotype related to MI -due to linkage disequilibrium (CCL3 with CCL4-like genes), through effects on the expression of genes in related biological pathways (MIF), or via effects on biochemical intermediate phenotypes (PTPN1, HDL and triglycerides). Polymorphisms in such genes cannot be used in a simple Mendelian Randomisation analysis.