Abstract
Background: The differential DNA methylation (DNAm) of the CpG site cg05575921, within the gene body of AHRR, has been shown to be a robust indicator of tobacco smoking exposure. This study aimed to determine if cg05575921 at AHRR DNAm had a clinically relevant association with prevalent cardiovascular disease (CVD) and compare its predictive performance with self-reported smoking assessments.
Methods: Whole blood cg05575921 (AHRR) DNAm was assessed using a simple droplet digital PCR assay in a cohort of 2784 elderly individuals. DNAm values were used to impute smoking status and its independent association with prevalent CVD.
Results: Consistent with previous reports, cg05575921 (AHRR) had significant, sensitive, and specific associations with self-reported smoking status (SRSS: never, ex-, current. <0.0001 between groups). This DNAm marker was also associated with the length of smoking cessation in ex-smokers (rho 0.55, p < .0001, Spearman rank correlation). A key finding of this study was that cg05575921 DNAm had a distinct non-linear relationship with a history of prevalent CVD. DNAm-imputed smoking status (ex-smoker quit >21 years, ex-smoker quit <21 years, and current smoker), remained significantly associated with CVD after covariant adjustment, which included SRSS (OR 2.5, 3.1, and 6.0, p < .0001, respectively). Even among never smokers, cg05575921 values were also significantly associated with CVD (Mann-Whitney U test p < .0001).
Conclusions: Imputed (cg05575921 AHRR) DNAm smoking status appears to have a significant and independent association with the prevalence of CVD. This association is stronger than that of conventional SRSS and may suggest potential utility to improve existing CVD risk prediction scores.
Implications: The results of this study were consistent with previous reports showing that the DNA methylation status of cg05575921 is highly accurate in identifying smoking status (current, ex-smokers, or never smokers), including the length of time an ex-smoker had quit. The marker's levels also had a strong relationship with prevalent CVD, including after adjusting for other risk factors, such as age, diabetes, and even self-reported smoking status. This study makes a strong contribution to a growing body of work suggesting that this DNAm marker may have substantial clinical utility to aid "precision-medicine" chronic disease risk prediction.