Genetically determined intelligence and coronary artery disease risk

Abstract

Background: Epidemiological studies have shown inverse association between intelligence and coronary artery disease (CAD) risk, but the underlying mechanisms remain unclear.

Methods: Based on 242 SNPs independently associated with intelligence, we calculated the genetic intelligence score (gIQ) for participants from 10 CAD case-control studies (n = 34,083) and UK Biobank (n = 427,306). From UK Biobank, we extracted phenotypes including body mass index (BMI), type 2 diabetes (T2D), smoking, hypertension, HDL cholesterol, LDL cholesterol, measured intelligence score, and education attainment. To estimate the effects of gIQ on CAD and its related risk factors, regression analyses was applied. Next, we studied the mediatory roles of measured intelligence and educational attainment. Lastly, Mendelian randomization was performed to validate the findings.

Results: In CAD case-control studies, one standard deviation (SD) increase of gIQ was related to a 5% decrease of CAD risk (odds ratio [OR] of 0.95; 95% confidence interval [CI] 0.93 to 0.98; P = 4.93e-5), which was validated in UK Biobank (OR = 0.97; 95% CI 0.96 to 0.99; P = 6.4e-4). In UK Biobank, we also found significant inverse correlations between gIQ and risk factors of CAD including smoking, BMI, T2D, hypertension, and a positive correlation with HDL cholesterol. The association signals between gIQ and CAD as well as its risk factors got largely attenuated after the adjustment of measured intelligence and educational attainment. The causal role of intelligence in mediating CAD risk was confirmed by Mendelian randomization analyses.

Conclusion: Genetic components of intelligence affect measured intelligence and educational attainment, which subsequently affect the prevalence of CAD via a series of unfavorable risk factor profiles.

Keywords: Coronary artery disease; Educational attainment; Genetic association; Genetic risk score; Intelligence; Obesity; Smoking.

Fig. 1

Association of gIQ and CAD risk. The genetic intelligence score was calculated in 10 case-controls studies of CAD and UK Biobank respectively. Logistic regression was performed to evaluate the association between gIQ and CAD risks in each study. Fixed-effect size meta-analysis was performed to combine all studies. Forest plot shows regression result in each study and the overall effect size. The gIQ was inversely associated with CAD risk

Fig. 2

Distribution of cases and controls according to gIQ. Individuals in 10 CAD studies were evenly grouped into a low (score = 1), medium (score = 2) and high (score = 3) group according to their gIQ. The OR is incidence of CAD relative to low group. Risk of CAD decreases along the increases of gIQ

Fig. 3

Associations of gIQ with CAD and it risk factors including BMI, smoking, T2D, HDL cholesterol, LDL cholesterol, and hypertension in UK Biobank. The OR for BMI is shown as logarithm of the linear regression coefficient. ‘Adjusted’ indicates the regression model between gIQ and trait after adjustment for measured intelligence (IQ), or length of school years completed (Edu), or neither of the two (No). The gIQ had inverse effects on BMI, T2D, smoking, and hypertension and a positive effect on HDL cholesterol. The association signals were largely attenuated by measured intelligence and educational attainment

Fig. 4

The result of MR analyses. Error bars indicate 95% confidence intervals around the estimated effects calculated using multivariable two-sample MR. The effects on outcomes for intelligence and educational attainment were generally in consistent directions. But the effects of educational attainment are quantitatively stronger than intelligence

Fig. 5

Pathway from higher gIQ to lower risk of CAD. Our study shows inverse effects of genetic determinants of intelligence on CAD and its risk factors including BMI, smoking, hypertension, and T2D and positive effects on HDL cholesterol. These association signals are mediated by measured intelligence and educational attainment, which two are bidirectionally associated with each other