We determined the association of quantitative indices of severity of coronary atherosclerosis, as determined by serial QCA, with 4 SNPs at the 9p21 locus; previously shown to be associated with the risk of CAD in case-control GWAS [1
]. We found no significant association between the quantitative indices of severity of coronary atherosclerosis and the SNPs at the 9p21 in the well-characterized LCAS population. The availability of serial QCA also provided the opportunity to examine the potential association of development of new coronary lesions and progression of the existing coronary lesions. We found no significant SNP and haplotype effect on the development of new coronary lesions and the progression of coronary atherosclerosis over a period of 2 1/2 years neither in the placebo group nor in the entire population when treatment with fluvastatin was included as a covariate. In contrast, plasma HDL-C levels and treatment with fluvastatin were the only independent predictors of progression of coronary atherosclerosis (inverse relationship). There were also no genotype-by-treatment interactions between the 4 SNPs at the 9p21 locus and the response of coronary atherosclerosis to treatment with fluvastatin. Moreover, we detected no significant association between the 9p21 locus SNPs and plasma levels of total cholesterol, HDL-C, LDL-C, triglycerides or apolipoproteins. The findings suggest absence of a strong association between the severity, development or progression of coronary atherosclerosis, as determined by QCA, and the 4 SNPs at the 9p21 locus.
The results of the present study is in accord with the results of a recent study showing no significant association between the severity of carotid atherosclerosis as determined by the measurement of carotid intima-media thickness by ultrasound and the 4 SNPs at the 9p21 locus [23
]. The null results of these two studies should be weighed against the results of the case-control GWAS showing a strong association between the 9p21 locus and risk of CAD. The previous null study had a very large sample size [23
] and low chance of type II statistical error. It is also noteworthy that the Ottawa Heart Study-1 and Ottawa Heart Study-2 each had a sample size of approximately 320 cases and controls [2
]. Despite the modest sample sizes there were significant differences in the distribution of genotypes of the 9p21 locus SNPs between the cases and controls in both datasets [2
]. Nonetheless, we considered the possibility of type II statistical error in the present study. Therefore, we calculated the power of the present study, post hoc, to detect differences in MLD, the primary quantitative phenotype of severity of coronary atherosclerosis. Accordingly, the number of observed genotypes in the study population provided 95.5%, 92.9% and 73.9% power to detect a 20% difference in the baseline MLD between subjects with the GG vs. GA, GA vs. AA and GG vs. AA genotypes, respectively, and 98.1% between GG vs. GA and AA genotypes, when α is set at 0.05. Likewise, the power to detect a 15% difference in mean MLD between GG vs. GA and AA genotypes was 86%. For a 10% or smaller differences in the mean MLD among the genotypes, the sample size of the present study did not offer sufficient power. Therefore, we can not exclude the possibility that the 9p21 locus imparts an effect on MLD that is less than 10% (0.165 mm). Hence, smaller differences among the genotypes are beyond the level of detection of the present study. The presence of undetected small differences in the MLD among the genotype may impart clinical significance and affect the clinical outcome. As for the progression of coronary atherosclerosis during the follow up period, the time interval between the two QCA was 2.5 years, which was sufficient to detect the effects of treatment and plasma HDL-C levels. Accordingly, one would have expected to detect the effects of SNPs at 9p21 locus on progression of coronary atherosclerosis if they were greater than the effects of treatment or plasma HDL-C levels. Smaller effects, however, could remain undetected. We also note that while QCA is a robust tool for the measurement of lumen diameter and area, but it may not be sufficiently sensitive to changes in plaque burden. As for the plasma LDL-C and HDL-C levels, the sample size of the study provided 80% power to detect approximately 8 mg/dl and 4 mg/dl differences between the two common genotypes of the SNPs, respectively. Overall, we noted no trend toward an association for any of the quantitative indices of severity of coronary atherosclerosis or plasma levels of lipid to suggest the possibility of type II statistical error. Finally, we note the prospective design of the present study wherein each individual was his/her own control as regards the progression of coronary atherosclerosis and the detailed phenotypic definition of atherosclerosis as a continuous phenotype as potential strengths of the present study.
Various phenotypes including ischemic stroke, abdominal aortic aneurysm, intra-cranial aneurysm, coronary artery calcification, osteoporosis, type 2 diabetes mellitus and possibly Alzheimer's disease have been associated with the 9p21 locus [5
]. Many but not all of the associated phenotypes may share a common vascular etiology with atherosclerosis as an underpinning milieu. The 9p21 locus contains the tumor suppressor proteins p16INK4a and P16INK4b, which are frequently lost or mutated in various human cancers, such as in leukemia, glioma, head and neck cancers and bladder cancers [13
]. The naturally occurring mutations at this locus are not known to predispose to premature atherosclerosis. Similarly, genetic deletion of the locus in the mouse is not known to predispose to or protect from atherosclerosis [16
]. Accordingly, given that LD in the human genome can span large chromosomal segments [29
], one may speculate that the 9p21 locus is in LD with a remote locus that serves as a true susceptibility locus for coronary atherosclerosis. Alternatively, the locus also contains a gene that transcribes a non-coding RNA, with yet to-be-defined function. Whether one of the two known genes or the non-coding RNA at the 9p21 locus or a distant locus/gene in LD with the 9p21 locus is responsible for the observed and consistent association of the 9p21 locus with CAD remains to be established.
The most significant impact of the results of the GWAS is likely to be in elucidation of the novel mechanistic pathways involved in the pathogenesis of atherosclerosis and its thrombotic complications. Such mechanistic studies could lead to identification of new therapeutic targets and the development of new drugs. In general, the clinical utility of the results of GWAS for risk stratification and prognostication is expected to be relatively modest; partly because of the low pre-test likelihood of the clinical events and relatively modest impact of the vast majority of the risk alleles for a complex trait. Therefore, delineation of molecular mechanisms by which the 9p21 locus imparts susceptibility to broadly defined CAD could have considerable clinical implications. An integral part of and likely a pre-requisite for the mechanistic studies pursuant to the results of the GWAS is identification of the precise phenotype of coronary heart disease that is associated with the 9p21 locus.