AHRR (cg05575921) hypomethylation marks smoking behaviour, morbidity and mortality

Abstract

Rationale and objectives: Self-reported smoking underestimates disease risk. Smoking affects DNA methylation, in particular the cg05575921 site in the aryl hydrocarbon receptor repressor (AHRR) gene. We tested the hypothesis that AHRR cg05575921 hypomethylation is associated with risk of smoking-related morbidity and mortality.

Methods: From the Copenhagen City Heart Study representing the Danish general population, we studied 9234 individuals. Using bisulphite treated leucocyte DNA, AHRR (cg05575921) methylation was measured. Rs1051730 (CHRN3A) genotype was used to evaluate smoking heaviness. Participants were followed for up to 22 years for exacerbations of COPD, event of lung cancer and all-cause mortality. Six-year lung cancer risk was calculated according to the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO_M2012).

Measurements and main results: AHRR (cg05575921) hypomethylation was associated with former and current smoking status, high daily and cumulative smoking, short time since smoking cessation (all p values <7×10^-31), and the smoking-related CHRN3A genotype (-0.48% per T-allele, p=0.002). The multifactorially adjusted HRs for the lowest versus highest methylation quintiles were 4.58 (95% CI 2.83 to 7.42) for COPD exacerbations, 4.87 (2.31 to 10.3) for lung cancer and 1.67 (1.48 to 1.88) for all-cause mortality. Finally, among 2576 high-risk smokers eligible for lung cancer screening by CT, observed cumulative incidences of lung cancer after 6 years for individuals in the lowest and highest methylation quintiles were 3.7% and 0.0% (p=2×10^-7), whereas predicted PLCO_M2012 6-year risks were similar (4.3% and 4.4%, p=0.77).

Conclusion: AHRR (cg05575921) hypomethylation, a marker of smoking behaviour, provides potentially clinical relevant predictions of future smoking-related morbidity and mortality.

Keywords: COPD epidemiology; Lung Cancer; Tobacco and the lung.

Figure 1

Smoking behaviour as a function of AHRR (cg05575921) methylation extent, among former and current smokers. The lower number of individuals, for smoking history compared with tobacco consumption, is explained by missing information on smoking history. Bar values are means and SEs. Estimates (95% CI) and p values per lower quintile and per 10% lower methylation extent were calculated with unadjusted linear regression. p Values for sex interaction were from a likelihood ratio test that compared the main effects in a linear regression model to a model also including a two-factor (sex by quintile of methylation extent) interaction term. AHRR, aryl hydrocarbon receptor repressor.

Figure 2

Cumulative incidences of COPD exacerbations, lung cancer and all-cause mortality as a function of ranked AHRR (cg05575921) methylation extent quintiles, among never, former and current smokers. Cumulative incidences were calculated by the Fine and Gray method, taking death and emigration as competing events into consideration, except for all-cause mortality for which only emigration was a competing event. For COPD exacerbations, the trend test is from the Fine and Gray regression instead of the log-rank trend test because the model allowed for multiple events for the same individual. Never smokers in 5th quintile were excluded due to very low numbers. AHRR, aryl hydrocarbon receptor repressor.

Figure 3

HRs of COPD exacerbation, lung cancer and all-cause mortality by AHRR (cg05575921) methylation extent quintiles and per 10% lower extent, among never, former and current smokers using a Cox regression model. Multifactorial adjustment was for age, sex, use of alcohol (continuous), body mass index (continuous), exposure to dust (no/yes), exposure to passive smoking (no/yes), history of cancer prior to attendance (no/yes), history of COPD prior to attendance (no/yes; for lung cancer and all-cause mortality), familial history of lung cancer (no/yes), education level (categorical), smoking status (never, former current), current and cumulative consumption of tobacco. p Values for trend were from the Cox regression model, where quintiles were inserted as a continuous variable. p Values (continuous) for the HR per 10% decrease of methylation extent are also shown. AHRR, aryl hydrocarbon receptor repressor.

Figure 4

Cumulative incidence of lung cancer among 2576 high-risk smokers, and observed and predicted 6-year risk by AHRR (cg05575921) methylation extent quintiles. High risk was defined as predicted PLCO_M2012 6-year risk>1.3455%. AHRR, aryl hydrocarbon receptor repressor; PLCO_M2012, Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial.