sequence (P<10−60). However, these SNPs were not associated with CAD (this result was obtained in a yet unpublished GWAS comparing GHS individuals to a cohort of CAD patients), whereas proxies of the CAD-associated SNPs were unrelated with CDKN2B expression (see legend of Figure 4 for more details). The SNPs associated with CDKN2B expression are located within the sequence of the non-coding alternatively spliced gene ANRIL (also named CDKN2BAS) whose implication in the association with CAD has been hypothesized [31]. Although our results are limited by the fact that neither CDKN2A nor ANRIL expressions could be evaluated, they reveal that in humans, SNPs that affect CDKN2B expression are different from those that are known to affect CAD risk ( Figure 4 ).
