Dose-Response Associations of Lipid Traits With Coronary Artery Disease and Mortality

Abstract

Importance: Apolipoprotein B (apoB), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG) are associated with coronary artery disease (CAD). However, trial evidence for the association of intensive LDL-C lowering and TG lowering with mortality is less definitive.

Objectives: To investigate the associations of apoB, LDL-C, and TG with CAD and mortality, both overall and by sex and age, and to characterize the shapes of these associations.

Design, setting, and participants: This genetic association study used linear and nonlinear mendelian randomization (MR) to analyze a population-based cohort of individuals of European ancestry from the UK Biobank, which recruited participants from 2006 to 2010 with follow-up information updated until September 2021. Data analysis occurred from December 2022 to November 2023.

Exposures: Genetically predicted apoB, LDL-C, and TG.

Main outcomes and measures: The primary outcomes were CAD, all-cause mortality, and cause-specific mortality. Genetic associations with CAD were calculated using logistic regression, associations with all-cause mortality using Cox proportional hazards regression, and associations with cause-specific mortality using cause-specific Cox proportional hazards regression with censoring for other causes of mortality.

Results: This study included 347 797 participants (mean [SD] age, 57.2 [8.0] years; 188 330 female [54.1%]). There were 23 818 people who developed CAD and 23 848 people who died. Genetically predicted apoB was positively associated with risk of CAD (odds ratio [OR], 1.65 per SD increase; 95% CI 1.57-1.73), all-cause mortality (hazard ratio [HR], 1.11; 95% CI, 1.06-1.16), and cardiovascular mortality (HR, 1.36; 95% CI, 1.24-1.50), with some evidence for larger associations in male participants than female participants. Findings were similar for LDL-C. Genetically predicted TG was positively associated with CAD (OR, 1.60; 95% CI 1.52-1.69), all-cause mortality (HR, 1.08; 95% CI, 1.03-1.13), and cardiovascular mortality (HR, 1.21; 95% CI, 1.09-1.34); however, sensitivity analyses suggested evidence of pleiotropy. The association of genetically predicted TG with CAD persisted but it was no longer associated with mortality outcomes after controlling for apoB. Nonlinear MR suggested that all these associations were monotonically increasing across the whole observed distribution of each lipid trait, with no diminution at low lipid levels. Such patterns were observed irrespective of sex or age.

Conclusions and relevance: In this genetic association study, apoB (or, equivalently, LDL-C) was associated with increased CAD risk, all-cause mortality, and cardiovascular mortality, all in a dose-dependent way. TG may increase CAD risk independent of apoB, although the possible presence of pleiotropy is a limitation. These insights highlight the importance of apoB (or, equivalently, LDL-C) lowering for reducing cardiovascular morbidity and mortality across its whole distribution.

Figure 1.. Univariable Estimates for Genetically Predicted Lipid Traits and Risk of Coronary Artery Disease (CAD), All-Cause Mortality, and Cause-Specific Mortality

The figure shows univariable estimates for apolipoprotein B (apoB), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG). Estimates are expressed as odds ratios (ORs) and hazard ratios (HRs) with 95% CIs per SD increase in genetically predicted level of each lipid trait (approximately 23.8 mg/dL for apoB [to convert to grams per liter, multiply by .01], 33.2 mg/dL [to convert to millimoles per liter, multiply by 0.0259] for LDL-C, and 87.5 mg/dL [to convert to millimoles per liter, multiply by .0113] for TG). CVD indicates cardiovascular disease. Squares denote stratum estimates, and diamonds denote overall estimates. Error bars denote 95% CIs. ^aP < .001. ^bP < .05.

Figure 2.. Multivariable Estimates for Genetically Predicted Lipid Traits and Coronary Artery Disease (CAD), All-Cause Mortality, and Cause-Specific Mortality

The figure shows multivariable estimates for apolipoprotein B (apoB), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG). Multivariable estimates for apoB and LDL-C were adjusted for TG, and multivariable estimates for TG were adjusted for apoB. Estimates are expressed as odds ratios (ORs) and hazard ratios (HRs) with 95% CIs per standard deviation increase in genetically predicted level of each lipid trait (approximately 23.8 mg/dL for apoB [to convert to grams per liter, multiply by .01], 33.2 mg/dL [to convert to millimoles per liter, multiply by 0.0259] for LDL-C, and 87.5 mg/dL [to convert to millimoles per liter, multiply by .0113] for TG). CVD indicates cardiovascular disease. Squares denote stratum estimates, and diamonds denote overall estimates. Error bars denote 95% CIs. ^aP < 001. ^bP < .05.

Figure 3.. Shapes of the Associations of Genetically Predicted Apolipoprotein B (apoB) and Triglycerides (TG) With Coronary Artery Disease (CAD) and All-Cause Mortality

The x-axis depicts apoB level in g/L or TG level in mmol/L. The y-axis depicts the odds ratio (OR) for CAD or hazard ratio (HR) for all-cause mortality with respect to the reference, plotted on a log scale. Reference is set to the median of each lipid trait (1.0 g/L for apoB and 1.5 mmol/L for TG). The black lines represent the dose-response association, and the blue shading represent the 95% CIs. The trend test assesses whether a linear trend in the stratum-specific estimates exists, and the fractional polynomial test examines whether a nonlinear model fits the exposure-outcome association better than a linear model.