Pharmacogenomics of CRP response to statins: a GIST consortium study

Abstract Introduction Statins are first line treatments in the primary and secondary prevention of cardiovascular disease. Prior clinical studies have shown that statins act independently of lipid lowering mechanisms to decrease C-reactive protein (CRP), a marker of inflammation. Purpose To elucidate genetic loci associated with CRP response to statins. Methods CRP response was specified as the change from baseline in log CRP after at least 4 weeks of statin therapy. Cohort level Genome-wide Association Studies (GWAS) of this CRP response was performed by linear regression analysis adjusted for baseline CRP, age, sex and BMI covariates using genetic data imputed to 1000 Genomes, testing ~10 million single nucleotide polymorphisms (SNPs) of minor allele frequency (MAF) >2%. Following quality control with EasyQC, a meta-analysis was conducted with METAL to combine GWAS results from six different cohorts of European ancestry (CARDS, FHS, JUPITER, MESA, PARC, PROSPER) within the GIST consortium. 1Mb loci regions were defined, centred +/-500kb around the lead SNP. Conditional analysis was performed using GCTA. Results There were 11,075 statin-treated individuals. The meta-analysis results revealed two loci achieving genome wide significance (P<5e-8): APOE (chr 19) and HNF1A (chr 12) (Figure 1). A signal at the CRP locus was highly suggestive (P=2.3e-7). All three loci are known to be associated with CRP levels in the absence of statin. Conditional analysis did not reveal secondary signals. The most associated variant at the APOE locus was the missense SNP rs429358, which contributes to the APOE E4 haplotype and is a risk locus for both dyslipidaemia and Alzheimer’s dementia. It is a C>T missense variant of MAF 0.12 leading to a R/C amino acid change, more common in African ancestry populations (MAF 0.27), and least common in Asian populations (MAF 0.09). A nominally significant association (P=0.09) for interaction with randomized allocation to statin v. placebo within samples derived from clinical trials supports this pharmacogenetic effect at APOE beyond genetic effects on baseline CRP levels. The most associated variant at the HNF1A locus was the intronic SNP rs11065384, which is in strong LD with the HNF1A missense SNP (rs1169288). The HNF1A locus is associated with diabetes, cholesterol levels, and coronary artery disease. There was no between-study heterogeneity at the top three loci (P>0.1), and forest plots show contribution to these signals from all 6 studies within the meta-analysis (Figure 2 A,B,C). Conclusions APOE and HNF1A are associated with CRP response to statins at the level of genome-wide significance. TheAPOE E4 signal is also known to be associated with LDL-Cholesterol response to statins, whereas the HNF1A locus is identified here for the first time as a pharmacogenetic determinant of statin response. The effect of this interindividual variability in CRP response on non-cardiovascular clinical outcomes should be elucidated.Figure 1Figure 2 A,B,C