Moderate-sized studies have suggested that variants of candidate genes can influence laboratory markers of coronary artery disease (CAD), but whether they predict parallel changes in clinical CAD risk is unknown.

We studied a single nucleotide polymorphism (SNP) from each of the 5 candidate genes for intermediate (laboratory) and clinical (angiographic CAD) end points in a large cohort of patients. The 5 gene SNPs were cholesteryl ester transfer protein (CETP) TaqIB (N = 3219), ATP-binding cassette (ABCA1) G596A (N = 3302), lipoprotein lipase (LPL) HindIII (N = 909), plasminogen activator inhibitor, type 1 (PAI1), 4G/5G (N = 1142), and hepatic lipase (HL) C-541T (N = 4704). Intermediate outcomes were high-density lipoprotein cholesterol (HDL-C) and triglycerides (TGs). Cases had 1- to 3-vessel CAD (≥70% stenosis); controls had angiographically normal coronaries.

Cholesteryl ester transfer protein predicted HDL (mean, B1B1 35.0 mg/dL, B2B2 38.6 mg/dL; P < .001) but not CAD (B1B1 74%, B2B2 70%; adjusted P = .35, odds ratio [OR] = 0.89). ABCA1 predicted HDL (mean, GG = 36.4 mg/dL, AA = 39.2 mg/dL; P = .02) but not CAD (GG 74%, AA 75%; adjusted P = .96, OR = 0.99). HL predicted HDL (CC 37.1 mg/dL, TT 40.9 mg/dL; P = .002) but not CAD (CC 71%, TT 68%, adjusted P = .66, OR = 0.94). LPL predicted TG (median: [++] 134, [−−] 98 mg/dL; P < .001) but not CAD ([++] 79%, [−−] 79%; adjusted P = .99, OR = 1.00). PAI1 predicted TG (median, 4G4G 130 mg/dL, 5G5G 148 mg/dL; P = .16), but not CAD (4G4G 77%, 5G5G 76%; adjusted P = .62, OR = 1.11).

Five SNPs predicted differences in risk-related lipids but not angiographic CAD. These discrepancies suggest that genetic determinants of CAD are complex and intermediate phenotypes are poor surrogates. These findings have important implications for future directions in genetic research.