Contemporary Polygenic Scores of Low-Density Lipoprotein Cholesterol and Coronary Artery Disease Predict Coronary Atherosclerosis in Adolescents and Young Adults

The development of coronary atherosclerosis is a lifelong process that begins as early as childhood.¹ We assessed whether contemporary polygenic scores (PGSs) for coronary artery disease (CAD) or for risk factors of CAD can predict the presence of raised coronary plaque among adolescents and young adults examined in the PDAY study (Pathobiological Determinants of Atherosclerosis in Youth) involving 15 laboratories that conducted autopsies on 2876 people aged 15 to 34 years dying of external causes between 1987 and 1994.¹

Our study included a subset of 1068 subjects genotyped as part of the SNPs and Extent of Atherosclerosis study (Single Nucleotide Polymorphisms and Extent of Atherosclerosis).² Genotype and phenotype data were downloaded from the publicly available National Center of Biotechnology Information database of Genotypes and Phenotypes (https://www.ncbi.nlm.nih.gov/gap/), and the former was used to impute genotypes on the Trans Omics for Precision Medicine reference panel and to calculate genetic principal components within White and Black subjects separately. Institutional Review Board approval was not required to conduct this study.

Data from dbGaP were accessed through approved research project 2638, study accession: phs000349.v1.p1. The source code used to perform all analysis is available at https://github.com/rodrigoguarischi/sea.

Our main outcome was the presence of raised lesions in the right coronary artery as such lesions represent the earliest stage of coronary atherosclerosis associated with traditional risk factors and community rates of clinical CAD.² Cases were defined as subjects having raised lesions, while controls had no raised lesions based on the evaluation of 3 independent pathologists.¹

The main exposure was genetic predisposition for a given trait or outcome defined by the relevant contemporary PGS. Scoring files were downloaded from the PGS catalog³ and high-risk allele scores were calculated using PGS-Calc. We assessed 1 PGS for CAD and 8 for risk factors of CAD from recent large-scale discovery and validation studies of PGSs.³ Genetic variant coverage was uniformly high for all PGSs (95.4%–99.6%). Odds ratios per SD increase in PGS were estimated using logistic regression stratified by race and adjusted for age at death, sex, and the first 10 race-specific genetic principal components. Pooled analyses adjusted for race were conducted for PGSs showing no statistical interaction with race (P<0.05). We corrected for multiple testing of PGSs within a cohort using the Benjamini-Hochberg method and a false discovery rate of 0.05. In a sensitivity analysis, we further adjusted models with a significant PGS for body mass index.

Among 564 White subjects, the mean age and body mass index was 26.7 years and 25.3; 436 (77%) were male and 181 (32%) were cases. The odds ratios for low-density lipoprotein (LDL) cholesterol level in blood and for CAD PGSs were significantly >1 (Figure [A]) even after further adjustment for body mass index (1.44, 1.17–1.77 for LDL; 1.38, 1.12–1.69 for CAD) and when modeling both PGSs concurrently (1.35, 1.10–1.66 for LDL; 1.38, 1.12–1.69 for CAD). Using the case definition implemented in this study, the odds ratio for CAD was also notably higher than a less complex PGS we examined in PDAY 8 years ago (1.21, 1.00–1.47).² Among 504 Black subjects, the mean age and body mass index was 27.5 and 25.1; 412 (82%) were male, and 165 (33%) were cases. No PGS was significant among Black subjects, but the LDL PGS showed the highest magnitude odds ratio (Figure [B]). In a pooled analysis, LDL was the only significant PGS (1.34, 1.13–1.60) showing no effect modification by race.

Figure.

Association between contemporary polygenic scores for traits related to coronary artery disease and the presence of right coronary artery atherosclerosis in adolescents and young adults. Odds ratio and 95% CIs per 1 SD increase in polygenic risk score for the presence of raised lesions among White (A) and Black (B) subjects in the Pathobiological Determinants of Atherosclerosis in Youth study. Polygenic scores (PGSs) are named as per the PGS catalog, but associations are classified into those related to adiposity, coronary artery disease (CAD), dysglycemia, dyslipidemia, and hypertension. P values shown are unadjusted with an asterisk indicating significance at a false discovery rate of 0.05. C, Prevalence of raised lesions among all White subjects (left) and burden of raised lesion among White cases only (right) by PGS quintiles for the CAD and LDL cholesterol PGSs. BMI indicates body mass index; fat %, body fat percentage; GWS, genome-wide significant variants; Ins, insulin; LDL, low-density lipoprotein cholesterol; Lp(a), lipoprotein a; P. Reg., penalized regression; P+T, pruning and thresholding; SBP, systolic blood pressure; Trig, triglycerides; and T2D, type 2 diabetes.

Next, we plotted the prevalence of raised lesions among all White subjects and the burden of raised lesions among White cases only by quintile of LDL and CAD PGS after collapsing the middle 3 quintiles into 1 group (Figure [C]). As expected, the prevalence of cases increased with successive quintiles of both scores given our per SD results. The percent surface area involvement among cases successively increased for LDL but not for CAD. The trend for LDL was not statistically significant (Kruskal–Wallis, P=0.76) but also not well powered to detect a difference (β=0.41).

Our analyses provide evidence that contemporary genome-wide polygenic risk scores track with the earliest phases of coronary atherogenesis in adolescents and young adults. PGSs for CAD and for LDL stand out as the strongest correlates with early raised lesions among White subjects, with LDL also being significant and consistent in the White and Black cohorts combined. We note the LDL PGS is the only score that was not only derived from a multiancestry genome-wide association study but also shown to perform equally well across all populations including those of African origin.⁴ Our observations support the hypothesis that genetic predisposition to elevated LDL cholesterol exposure early in life is a particularly important risk factor for premature atherosclerosis. Although not possible to demonstrate with PDAY, other studies have shown that markers for genetic susceptibility to high LDL improve prediction of CAD even in the presence of measures of LDL.⁵ Thus, genetic risk assessment may be one strategy for targeting primordial prevention efforts, particularly relating to LDL cholesterol. Such targeting is expected to become even more efficient as PGSs continue to improve.

ARTICLE INFORMATION

Sources of Funding

Genotype data was generated by an National Heart Lung Blood Institute grant to Dr Herrington (1U01HL080443). Dr Clarke is supported by a National Institutes of Health-Clinical and Translational Science Awards grant (KL2TR003143). Dr Assimes is supported by a grant from the National Institute of Diabetes and Digestive and Kidney Diseases (R01DK114183).

Disclosures

None.