Lung function, as measured by spirometry to assess the physiological state of the lungs and airways, is a complex trait that reflects inherent genetic susceptibility and environmental exposures over the lifespan. Cigarette smoking is the most common exposure that increases risk for reduced lung function and chronic obstructive pulmonary disease (COPD), the fourth leading cause of death in the United States. Gene-by-smoking interactions are biologically plausible and repeatedly studied, but overall, statistical evidence for interaction has been tepid. Kim et al1 report on interaction between a polygenic risk score (PRS), a single composite score representing 2.5 million genetic variants, and smoking on reduced lung function in the UK Biobank cohort. Their findings parallel those of Zhang et al,2 showing interaction between the same constructed PRS and smoking on risk of incident COPD in UK Biobank. While smoking posed harm for all genetic risk categories, the reported interactions showed that individuals at high genetic risk were more susceptible than individuals at low genetic risk to the harmful effects of smoking on reduced lung function and risk of developing COPD.
Lung function and COPD are heritable. Common genetic variants at hundreds of loci have been associated with lung function from a series of genome-wide association studies (GWAS) with increasingly large sample sizes (largest N = 400 1023). GWAS of lung function have largely focused on 2 measures used in the diagnosis of COPD—forced expiratory volume in the first second (FEV1) and its ratio to forced vital capacity (FEV1/FVC)—taken at a single time point. These cross-sectional measures provide a snapshot of the lung function trajectory over time: growth from birth to early adulthood, plateau at peak attainment, and decline thereafter. There is overlap, albeit incomplete, in variants identified at genome-wide significance for cross-sectional lung function and COPD. Each GWAS-identified variant exerts a relatively small effect size and explains little phenotypic variance. Genetic variants taken in aggregate to construct a single PRS, weighted by the variants’ effect sizes, explain a much larger portion of the phenotypic variance.4 The PRS has clinical promise in identifying individuals at risk for reduced lung function and COPD and forming the basis for early intervention. In taking that promise one step further, understanding the interplay between genetics and smoking is critical for targeted intervention to attenuate reduced lung function and prevent COPD.
Studying genetic interaction with cigarette smoking is grounded in biological plausibility (eg, genetic variation influencing oxidative stress and inflammatory pathways and altering response to smoking). The new PRS-by-smoking findings1,2 follow a history of interaction studies, beginning with candidate genes and then extending across the genome. In the first genome-wide study that included interaction between single variants and smoking on lung function, my collaborators and I used a method that simultaneously tested main and interactive effects and identified novel genetic loci that were missed in standard GWAS.5 Evidence of single variant interactions per se was not strong in our study or later in UK Biobank,3 possibly because of smoking’s large effect size dominating the small effect sizes of single variants. This lack of strong interactions does not mean that genetic variation and smoking act independently in influencing lung function. Instead, embedding interaction to more fully adjust the model for smoking was valuable in identifying genetic loci underlying lung function.
Aggregating small effects across many variants appears to help overcome smoking’s dominating effect and begins to unravel how genetics and smoking interact in influencing lung function. This advantage of PRS was not immediately obvious in the field. In 2017, Aschard et al6 tested PRS-by-smoking interaction on lung function using the 26 earliest GWAS-identified variants; we detected an interaction with ever vs never smoking on FEV1/FVC in a consortia-based meta-analysis of 50 047 participants, but statistically significant replication was not attained. In 2019, with more genome-wide significant variants identified, Shrine et al3 extended the PRS using 279 variants and found negligible evidence for interaction with smoking in nearly 375 000 individuals in the UK Biobank. Now, by using a PRS that reflects 2.5 million variants and improves the variance explained in lung function,4 Kim et al1 found statistically significant PRS-by-smoking interaction on FEV1/FVC (N = 319 370). In parallel, Zhang et al found a significant PRS-by-smoking interaction on risk of incident COPD (N = 439 255). Both studies used the UK Biobank cohort, the same PRS constructed with 2.5 million variants based on their cumulative evidence for association with FEV1 and FEV1/FVC,4 and similar smoking definitions (ever vs never smoking, current vs former smoking, and pack-years). Interaction was detected when using genome-wide results that was not found when focusing only on genome-wide significant variants, and the interaction patterns emphasize categories of individuals based on genetic risk and smoking behavior. Light smokers with a high PRS showed comparable overall risks for reduced lung function and incident COPD as heavy smokers with a low PRS. These studies suggest that genetic risk amplifies the harmful effects of smoking on reduced lung function and COPD, with smokers of high genetic risk being of gravest concern.
From a public health standpoint, abstaining from smoking is key for all individuals to prevent lung disease and many other adverse health effects. Yet, smoking remains common. Targeted intervention based on genetic risk may offer a more effective strategy for individuals to succeed in quitting smoking. Smokers with high genetic risk for lung function and COPD, based on their PRS, may benefit the most from smoking cessation treatment. While the PRS for lung function and COPD may identify a subset of smokers for targeted intervention, future studies should also consider how smoking behaviors and response to smoking cessation treatment are polygenically influenced, as outlined elsewhere.7
From a biological standpoint, single variant-by-smoking interactions would enable straightforward mechanistic studies of how smoking influences lung function and COPD in disease-relevant tissues and cell types, but such interactions are elusive. In contrast, the PRS reflects a large number of variants involved in a multitude of pathways, some of which likely influence physiological response to cigarette smoke exposure in the lungs. The mechanistic clues offered by the PRS are wide, and integration of multiple layers of biological and environmental exposure data and lung function trajectories over time are needed to better understand the mechanisms underlying the emergent PRS-by-smoking interaction.
In summary, PRS-by-smoking interaction expands upon prior work and provides more conclusive evidence than before, showing that individuals at high genetic risk face the greatest detrimental outcomes of smoking on lung function and COPD. While promising, these interaction findings still merit replication in independent cohorts. This step is limited by other data sets of comparable size as UK Biobank not being readily available. Sample sizes are even further constrained for non-European ancestries and hamper the accuracy of PRSs in diverse populations. PRS-by-smoking studies to date have largely, if not exclusively, focused on individuals of European ancestry. Extension of these studies into other ancestries is critically important to address health disparities in prevention and treatment of smoking behaviors and lung disease. This extension underscores the urgent need for more data from diverse populations.
Published: December 16, 2021. doi:10.1001/jamanetworkopen.2021.40347
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Hancock DB. JAMA Network Open.
Corresponding Author: Dana B. Hancock, PhD, RTI International, 3040 E Cornwallis Rd, PO Box 12194, Research Triangle Park, NC 27709 (firstname.lastname@example.org).
Conflict of Interest Disclosures: None reported.
et al; Understanding Society Scientific Group. New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries. Nat Genet
. 2019;51(3):481-493. doi:10.1038/s41588-018-0321-7PubMedGoogle ScholarCrossref
et al; International COPD Genetics Consortium; SpiroMeta Consortium. Chronic obstructive pulmonary disease and related phenotypes: polygenic risk scores in population-based and case-control cohorts. Lancet Respir Med
. 2020;8(7):696-708. doi:10.1016/S2213-2600(20)30101-6PubMedGoogle ScholarCrossref
DB, Soler Artigas
et al. Genome-wide joint meta-analysis of SNP and SNP-by-smoking interaction identifies novel loci for pulmonary function. PLoS Genet
. 2012;8(12):e1003098. doi:10.1371/journal.pgen.1003098PubMedGoogle Scholar
et al; Understanding Society Scientific Group. Evidence for large-scale gene-by-smoking interaction effects on pulmonary function. Int J Epidemiol
. 2017;46(3):894-904. doi:10.1093/ije/dyw318PubMedGoogle Scholar
et al; Genetics and Treatment Workgroup of the Society for Research on Nicotine and Tobacco (SRNT). Leveraging genomic data in smoking cessation trials in the era of precision medicine: why and how. Nicotine Tob Res
. 2018;20(4):414-424. doi:10.1093/ntr/ntx097PubMedGoogle ScholarCrossref