Tobacco Genomics: Complexity and Translational Challenges

This issue of Nicotine & Tobacco Research includes four studies that address critical research gaps in the genetics of tobacco use behaviors. There is currently limited research using polygenic risk scores (PRSs) of tobacco use to understand the use of other tobacco products such as electronic cigarettes (e-cigarettes) and water pipes, with comorbid disorders, or that adequately represents different ancestral groups using stratified analyses or trans-ethnic analyses. These studies present examples of current methods used to address questions of association, correlation, variance explained, and causal inference, in tobacco use behaviors and related disorders, from population, disease, and twin-based cohorts of multiple ancestries.

Allegrini et al.¹ generated PRSs from summary statistics of the International Cannabis Consortium and the Tobacco And Genetics Consortium and analyzed associations of these with e-cigarette and water pipe use phenotypes in the Netherlands Twin Register. PRSs for cigarettes per day (CPD) were positively associated with lifetime e-cigarette use in the entire sample (N = 4050). After stratification by smoking history, PRSs for CPD were associated with lifetime e-cigarette use in former smokers (n = 951), and with age at initiation of water pipe use in never cigarette smokers (n = 425). These results suggest that genetic vulnerability to smoking heaviness may increase risk for other tobacco or nicotine product use behaviors and that this may differ by smoking history.

Gibson et al.² calculated genetic correlations and causal effects (using Mendelian randomization) between smoking and sleep behaviors using summary and individual level data in more than 300 000 UK Biobank participants. They found novel genetic correlations between sleep behaviors and smoking behaviors. For example, positive associations were found between undersleeping and smoking initiation, as well as between undersleeping and CPD. Negative associations were found between sleep duration and smoking initiation, and between the chronotype of “morningness” and smoking cessation. The authors confirmed positive genetic correlations of insomnia and smoking behaviors (initiation and CPD). Causal analyses with individual-level data provided evidence that smoking heaviness (CPD) decreases the odds of the “morningness” chronotype, and analyses with summary-level data provided evidence that insomnia causally increases smoking heaviness and reduces cessation. These correlation and causal analyses suggest shared genetic vulnerability underlying both sleep (reduced sleep or reduced “morningness” chronotype) and smoking behaviors (initiation, CPD, or reduced cessation).

Jiang et al.³ conducted an exome chip-based exome-wide association scan (EWAS) of common single nucleotide polymorphisms (SNPs), a gene-based association analysis, and a heritability analysis of age of smoking initiation in heavy smokers of African American (N = 1695 individuals, N = 48 095 SNPs) and European American (N = 856 individuals, N = 34 992 SNPs) ancestry. In two ancestry groups, this study identified four novel susceptibility loci and significant heritability for age of smoking initiation. Three SNPs (two in the African American sample) and one gene in the African American sample remained exome-wide significant after Bonferroni correction, and exome chip-based SNPs accounted for 13% and 7% of age of smoking initiation phenotypic variance in the African American and European American samples.

Lutz et al.⁴ conducted GWAS using common genome-wide SNPs, and exome chip-based EWAS using uncommon SNPs (minor allele <5%) with average CPD in Non-Hispanic white (N = 6658) and African American (N = 3260) individuals from three chronic obstructive pulmonary disorder (COPD) consortia (COPDGene, GenKOLS, and ECLIPSE). Analyses included replications of SNPs identified in the study of smoking behavior, lung function, and COPD from the UK BiLEVE study,⁵ and stratified analyses by COPD case status and sex. These investigators identified: genome-wide significant associations on chromosome 15q25 and replications of BiLEVE SNPs in all individuals and in Non-Hispanic whites on chromosomes 11, 19, and 20; a replication of a BiLEVE SNP in African Americans on chromosome 8 (CHRNB3); one novel gene (MYLIP) identified in African Americans; and, replication of one BiLEVE candidate gene (CYP2A7) in Non-Hispanic whites. This study’s novel findings included evidence of similar effect sizes of common variants among COPD cases and controls at SNPs on chromosomes 11, 15, and 19, and a novel candidate gene in African Americans.

In the space of a few decades, the research community has moved from initial findings and confirmation of a handful of variants and genes unequivocally associated with alcohol and tobacco consumption,⁶ to the discovery of hundreds of loci that are associated with smoking and drinking behaviors. The GWAS and Sequencing Consortium of Alcohol and Nicotine Use (GSCAN, https://genome.psych.umn.edu/index.php/GSCAN; last accessed April 3, 2019) was developed as a collaborative endeavor to identify genes associated with nicotine and alcohol use measures across over 30 studies and 1 million research participants. A joint webinar hosted by the Genetics and Omics & Basic Science Networks of the Society for Research on Nicotine & Tobacco held on September 11, 2018 provided an overview of GSCAN’s discoveries.^7–9 In addition to replicating findings from previous GWAS, GSCAN revealed hundreds of novel loci for further research and translation. Specifically, 564 independent genetic variants assigned to 405 genes were found to be associated with smoking and drinking behaviors. Top-ranked results included: pharmacokinetic (CYP2A6) and pharmacodynamic (all CHRN excluding CHRNA7) loci associated with cigarette consumption; and, pharmacokinetic (ADH1B), metabolic (GCKR), and endocrine (KLB) loci associated with alcohol consumption. Of the 405 genes with SNPs associated with the five behaviors, 171 exhibited association with one behavior only, whereas 150 were found to affect two to five smoking and drinking behaviors. Although only three genes were associated with all five of the measured behaviors and only one gene with all four smoking behaviors, there were 88 genes associated with alcohol consumption and one or more smoking behaviors.

The identification of hundreds of loci provides challenges for researchers regarding the integration of diverse data for analyses, biological interpretation, and translation toward further research and clinical applications.¹⁰ Research and clinical care communities and research sponsors are continually organizing to address the complex scientific and social challenges of addiction research. How these communities organize existing resources and leverage translational models will influence public health prevention, interventions, and clinical care for tobacco and alcohol dependence and related outcomes in the decades to come.

Funding

This work was supported by the National Institutes of Health (9R44AA027675-02 to A.W.B., 3U54MD010724 to S.P.D., and National Institute on Drug Abuse R01 DA038076 and National Cancer Institute P30 CA091842-16S2 to L.C.). The funders/sponsors had no role in the study design, data collection, analysis, preparation of the manuscript, or decision to submit the manuscript for publication.

Declaration of Interests

A.W.B. is a consultant to BioRealm, LLC, which intends to commercialize an analysis platform for substance use disorder studies and applications.