Polygenic Risk, Rare Variants, and Family History: Independent and Additive Effects on Coronary Heart Disease

Abstract

Background: Genetic factors are not included in prediction models for coronary heart disease (CHD).

Objectives: The authors assessed the predictive utility of a polygenic risk score (PRS) for CHD (defined as myocardial infarction, coronary revascularization, or cardiovascular death) and whether the risks due to monogenic familial hypercholesterolemia (FH) and family history (FamHx) are independent of and additive to the PRS.

Methods: In UK-biobank participants, PRS_CHD was calculated using metaGRS, and 10-year risk for incident CHD was estimated using the pooled cohort equations (PCE). The area under the curve (AUC) of the receiver operator curve and net reclassification improvement (NRI) were assessed. FH was defined as the presence of a pathogenic or likely pathogenic variant in LDLR, APOB, or PCSK9. FamHx was defined as a diagnosis of CHD in first-degree relatives. Independent and additive effects of PRS_CHD, FH, and FamHx were evaluated in stratified analyses.

Results: In 323,373 participants with genotype data, the addition of PRS_CHD to PCE increased the AUC from 0.759 (95% CI: 0.755-0.763) to 0.773 (95% CI: 0.769-0.777). The AUC and NRI_Event for PRS_CHD were higher before the age of 55 years. Of 199,997 participants with exome sequence data, 10,000 had a PRS_CHD ≥95th percentile (PRS_P95), 673 had FH, and 46,163 had FamHx. The CHD risk associated with PRS_P95 was independent of FH and FamHx. The risks associated with combinations of PRS_CHD, FH, and FamHx were additive and comprehensive estimates could be obtained by multiplying the risk from each genetic factor.

Conclusions: Incorporating PRS_CHD into the PCE improves risk prediction for CHD, especially at younger ages. The associations of PRS_CHD, FH, and FamHx with CHD were independent and additive.

Keywords: cardiovascular diseases; coronary disease; genetic predisposition to disease; genome-wide association study; risk assessment.

Graphical abstract

Figure 1

UKBB Cohort Analyzed for This Study Creating a UKBB validation cohort to compare the IRS vs PCE (right arm), and a UKBB cohort with exome sequencing data to asses the effect of genetic factors on CHD, along with their independent, and additive effects on CHD (left arm). ∗The numbers for each exclusion item may overlap. CHD = coronary heart disease; ES = exome sequencing; FamHx = family history of CHD; FH = familial hypercholesterolemia; IRS = integrated risk score; LLT = lipid lowering treatment; PCE = pooled cohort equation; PRS_P# = percentile of PRS; UKBB = UK-biobank.

Figure 2

Kaplan-Meier Curves for Survival Free of CHD Events The numbers of at-risk in each group are displayed. The green line represents low PCE/low IRS, the blue line represents high PCE/low IRS, the purple line represents low PCE/high IRS, and the red line represents high PCE/high IRS. Abbreviations as in Figure 1.

Figure 3

The Association of Genetic Factors With CHD Across Age Using a regression model adjusted for sex and the first 4 principal components of ancestry, we estimated the ORs (95% CI) for CHD associated with PRS_P95, monogenic FH, and FamHx in different age categories. Abbreviations as in Figure 1.

Central Illustration

The Incremental Predictive Utility of PRS_CHD, as Well as the Independent and Additive Effects of Genetic Factors for the Risk of CHD (Top panel) The UKBB data set was used to calculate the PRS, identify FH variants, and ascertain FamHx of CHD. By combining PCE and PRS, an IRS was developed to predict the incidence of the primary outcome, CHD. (Middle panel) The IRS outperformed the PCE in predicting the incidence of CHD, particularly in younger and lower-risk individuals. (Bottom panel, left) Three genetic susceptibility factors independently increased the risk of CHD, with each factor exerting its effect on CHD risk regardless of the presence or absence of the other genetic risk factors. (Bottom panel, right) The effect of 3 genetic susceptibility factors was also found to be additive. When more than one genetic factor was present, the risk of CHD could be estimated by multiplying the individual risks contributed by each genetic factor. APOB = apolipoprotein B; AUC = area under curve; CHD = coronary heart disease; CV = cardiovascular; FamHx = family history of CHD; FH = familial hypercholesterolemia; IRS = integrated risk score; LDLR = low density lipoprotein receptor gene; MI = myocardial infarction; NRI = net reclassification improvement; P diff = P value for difference; PCE = pooled cohort equations; PCS9 = proprotein convertase subtilisin/kexin type 9 gene; PRS = polygenic risk score; PRS_P# = percentile of PRS; UK = United Kingdom.

Figure 4

Additive Effect of PRS_CHD, FH, and FamHx on CHD The observed ORs (95% CI) for CHD associated with different combinations of PRS_CHD, FH, and FamHx at age 65 years were estimated using logistic regression models adjusted for sex and the first 4 principal components of ancestry. The expected ORs were estimated by multiplying the crude independent risk, based on Supplemental Table 10. The FH^-FamHx^- group was considered as the reference for studying the additive effect of FH and FamHx (P interaction between FH and FamHx: 0.986) (top panel). PRS_P20-80FH^- was considered as the reference group for studying the additive effect of PRS_CHD and FH (P interaction between PRS_P20 and FH: 0.522, P interaction between PRS_P80 and FH: 0.771) (middle panel). PRS_P20-80FamHx^- was considered as the reference group for studying the additive effect of PRS_CHD and FamHx (P interaction between PRS_P20 and FamHx: 0.916, P interaction between PRS_P80-95 and FamHx: 0.607, P interaction between PRS_P95 and FamHx: 0.559) (lower panel). CI = confidence interval; PRS_P# = Percentile of PRS_CHD.