Assessing dose-response relations of lipid traits with coronary artery disease, all-cause mortality, and cause-specific mortality: a linear and non-linear Mendelian randomization study

ImportanceApolipoprotein B (apoB), low-density lipoprotein cholesterol (LDL-C), and possibly triglycerides (TG) play causal roles in the aetiology of coronary artery disease (CAD). However, trial evidence for effects of intensive LDL-C lowering and TG lowering on mortality is less definitive. ObjectivesTo investigate dose-response relations of apoB, LDL-C, and TG with CAD and mortality overall, by sex, and by age. DesignWe performed linear Mendelian randomization (MR) analyses to assess the associations of genetically-predicted apoB, LDL-C, and TG with CAD, all-cause mortality, and cause-specific mortality. We also performed non-linear MR analyses, dividing the population into 10 strata, to assess stratum-specific estimates and characterise the shape of these associations. SettingUK Biobank. Participants347,797 European ancestry participants (23,818 CAD cases and 23,848 deaths). ExposuresGenetically-predicted apoB, LDL-C, and TG. Main outcomes and measuresCAD, all-cause mortality, cardiovascular mortality, cancer mortality, and non-cardiovascular/cancer mortality. ResultsGenetically-predicted apoB was positively associated with CAD (odds ratio (OR) 1.65 per standard deviation increase [95% confidence interval 1.57, 1.73]), all-cause mortality (hazard ratio (HR) 1.11 [1.06, 1.16]), and cardiovascular mortality (HR 1.36 [1.24, 1.50]), with some evidence for stronger associations in men than women. Findings were similar for LDL-C. Genetically-predicted TG was positively associated with CAD (OR 1.60 [1.52, 1.69]), all-cause mortality (HR 1.08 [1.03, 1.13]), and cardiovascular mortality (HR 1.21 [1.09, 1.34]); however, sensitivity analyses suggested evidence of pleiotropy. The association of genetically-predicted TG with CAD persisted but its associations with mortality outcomes were attenuated towards the null after controlling for LDL-C. Non-linear MR suggested the shapes of 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 relevanceOur findings suggest that apoB (or equivalently LDL-C) increases CAD risk, all-cause mortality, and cardiovascular mortality all in a dose-dependent way. TG likely increases CAD risk, although the possible presence of pleiotropy is a limitation. These insights highlight the importance of LDL-C lowering for reducing cardiovascular morbidity and mortality across its whole distribution. Key points QuestionDo apolipoprotein B (apoB), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG) increase risk of coronary artery disease (CAD), all-cause mortality, or cause- specific mortality, and if so, what are the shapes of these relations? FindingsIn this Mendelian randomization study including 347,797 European ancestry participants from UK Biobank, genetically-predicted apoB and LDL-C were positively associated with CAD, all- cause mortality, and cardiovascular mortality all in a dose-dependent way. Genetically-predicted TG was positively associated with CAD, although the presence of pleiotropy was suggested. MeaningApoB (or equivalently LDL-C) lowering reduces cardiovascular morbidity and mortality across its whole observed distribution.