Citation: Melén E, Bottai M (2012) On Lung Function and Interactions Using Genome-Wide Data. PLoS Genet 8(12): e1003174. https://doi.org/10.1371/journal.pgen.1003174
Editor: Greg Gibson, Georgia Institute of Technology, United States of America
Published: December 20, 2012
Copyright: © 2012 Melén, Bottai. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: EM has received research funding from the Swedish Heart and Lung Foundation. The funder had no role in the preparation of the article.
Competing interests: The authors have declared that no competing interests exist.
Lung function reflects the physiological state of the lungs and airways, and abnormalities are present in asthma and chronic obstructive pulmonary disease (COPD), for example. Function measurements (e.g., forced expiratory volume in one second, FEV1) are often used in clinical practice to detect obstructive or restrictive conditions. Smoking is well-known to affect lung function negatively, presumably through induction of oxidative stress, inflammation, and lung damage. Studies also report negative effects on lung function and later asthma risk in children whose mothers smoked during pregnancy , which supports the “Barker hypothesis” that cardiovascular and respiratory diseases in adulthood may have their origin during fetal life . Interestingly, transgenerational pulmonary effects of nicotine exposure during pregnancy, possibly mediated by epigenetic mechanisms, have recently been observed in animal studies . Twin studies suggest a substantial genetic contribution to the variability of lung function, and several important loci have been identified in recent genome-wide association studies (GWASs) . In addition, genetic factors of relevance for respiratory diseases are also proposed to influence lung growth in utero .
Although smoking is the most well-studied and established lifestyle risk factor for respiratory diseases, not all smokers develop diseases such as asthma or COPD . Large individual variability in responses to environmental factors exists, and genetic susceptibility may partly account for this. For example, IL13 single nucleotide polymorphisms (SNPs) have been shown to modulate the adverse effects of long-term cigarette smoking on pulmonary function . MMP12, a protease involved in tissue degradation, has been associated with lung function and risk of COPD, but only in high-risk populations such as smokers and asthmatics . The lungs develop during fetal life and throughout childhood , which is likely why children have been reported to be more susceptible to hazardous airborne substances compared to adults . Whether this has any relevance for the identification of gene–environment interactions in adults or children remains to be investigated.
As the effects of gene-smoking interactions on lung function have not been extensively studied in large data sets so far, the paper by Hancock et al. in this issue of PLOS Genetics adds value to the current literature in many ways . The authors present a large scale gene–environment interaction study based on GWAS data, smoking status, and lung function outcomes in over 50,000 adults from 19 studies primarily from the CHARGE and SpiroMeta consortia (Figure 1). Few similar studies on a complex trait where interaction effects have been thoroughly explored have been published to date.
This Perspective highlights the main findings in Hancock et al.  and discusses why interaction effects are so difficult to identify even in large, well-characterized data sets.
Hancock et al. used a recently developed joint meta-analysis method (JMA)  primarily designed to detect genetic effects while taking environmental factors into account as the main method for their analyses. The method jointly tests main genetic (SNP) and interaction effects and is attractive for gene–environment interaction analyses thanks to its robustness and superior power over standard interaction models under certain conditions. Three novel regions of potential importance for lung function were identified in the present study: DNER (2q36), HLA-DQ (6p21), and KCNJ2/SOX9 (17q24). Further, using the publicly available GEO repository, the authors were able to show differential expression of DNER and SOX9 in epithelial cells from smokers versus non-smokers, which supports the involvement of these genes in smoking-induced lung function deterioration. In addition, recent experimental data from studies on mice show that tracheal and bronchial cartilage formation is regulated by Tbx4 and Tbx5 through Sox9 expression . Interestingly, HLA-DQ has been associated with asthma in independent GWASs of asthma , .
The most significant association was for rs7594321 in DNER with a p-value for the joint test equal to 5.0×10−11 in the pack-years model and 2.6×10−9 in the ever-smoking model. The corresponding interaction p-values were non-significant. In the ever-smoking model, stratified genetic effects of per allele change 0.049 in never-smokers (comparable to approximately a year and a half of FEV1 decline) and 0.035 in ever-smokers were observed. Thus, the effect of this particular variant seemed to only marginally differ between never-smokers and ever-smokers. In addition, standard interaction analyses identified no SNP by smoking interaction at the genome-wide significant level (p<5×10−8).
Using the JMA method, Hancock et al. successfully identified three novel loci not previously associated with lung function in GWAS or candidate gene studies. Yet, it is likely that important gene–environment interaction effects for lung function remain to be identified. Alternatively, one must accept that there are no gene-smoking interaction effects of importance for lung function after all. The latter alternative, albeit possible, seems to contrast with clinical and epidemiological evidence. Besides, clear and abundant evidence of interactions between gene and exposures has been found in animal studies and epigenetic projects , .
So why are interaction effects so difficult to identify even in large, well-characterized data sets? This question has not found a definite answer yet, and excellent reviews of methodological challenges and the current status of the research have recently been published , .
It has long been recognized that combinations of multiple variants of different genes may together raise the risk of complex diseases more than any single variant alone. Combinations of variants may in turn interact with one or more environmental factors. It is also sensible to allow for the possibility that any such combinations, if they exist, may not be identical for all individuals. The identification of possible combinations of variants, interacting environmental factors, and inter-individual heterogeneity presents major challenges that are best met by interdisciplinary efforts.
So far, most attempts to find interactions between individual or multiple gene variants and environmental factors at the genome-wide level have been undermined by insuperable limitations of statistical power and sample size. The recent advances in analytical methodology may have alleviated these limitations but surely not overcome them. The sheer number of gene variants in GWAS or genome sequencing studies, and the even greater number of their possible combinations, make it statistically unreasonable to pursue such quests with standard statistical hypothesis testing.
The wealth of information contained in the human genome, however, cannot be left untapped, and much can be done with the resources that are already available or expected to become available in the near future. Promising trails point to (1) development of exploratory analytical approaches that may help tackle high-order interactions between multiple variables; (2) acknowledgment and evaluation of the inter-individual heterogeneity observed clinically and experimentally through genomic, transcriptomic, and proteomic profiling; (3) investigation of the negative findings through power analyses that might help restrict the potential effect of some gene variants and rule out biologically or clinically relevant effects of many of them; and (4) critical assessment of the applicability to this context of the frequentist and Bayesian inferential paradigms and exploration of possible alternatives. Whatever trails are to be followed, research can only progress through close collaboration between disciplines like biostatistics, epidemiology, biology, and medicine.
- 1. Neuman A, Hohmann C, Orsini N, Pershagen G, Eller E, et al. (2012) Maternal smoking in pregnancy and asthma in preschool children: a pooled analysis of 8 birth cohorts. Am J Respir Crit Care Med 186 (10) 1037–1043.
- 2. Barker DJ, Martyn CN (1992) The maternal and fetal origins of cardiovascular disease. J Epidemiol Community Health 46: 8–11.
- 3. Rehan VK, Liu J, Naeem E, Tian J, Sakurai R, et al. (2012) Perinatal nicotine exposure induces asthma in second generation offspring. BMC Med 10: 129.
- 4. Soler Artigas M, Loth DW, Wain LV, Gharib SA, Obeidat M, et al. (2011) Genome-wide association and large-scale follow up identifies 16 new loci influencing lung function. Nat Genet 43: 1082–1090.
- 5. Melén E, Kho AT, Sharma S, Gaedigk R, Leeder JS, et al. (2011) Expression analysis of asthma candidate genes during human and murine lung development. Respir Res 12: 86–86.
- 6. Lokke A, Lange P, Scharling H, Fabricius P, Vestbo J (2006) Developing COPD: a 25 year follow up study of the general population. Thorax 61: 935–939.
- 7. Sadeghnejad A, Meyers DA, Bottai M, Sterling DA, Bleecker ER, et al. (2007) IL13 promoter polymorphism 1112C/T modulates the adverse effect of tobacco smoking on lung function. Am J Respir Crit Care Med 176: 748–752.
- 8. Hunninghake GM, Cho MH, Tesfaigzi Y, Soto-Quiros ME, Avila L, et al. (2009) MMP12, lung function, and COPD in high-risk populations. N Engl J Med 361: 2599–2608.
- 9. Narayanan M, Owers-Bradley J, Beardsmore CS, Mada M, Ball I, et al. (2012) Alveolarization continues during childhood and adolescence: new evidence from helium-3 magnetic resonance. Am J Respir Crit Care Med 185: 186–191.
- 10. van Leeuwen DM, Pedersen M, Hendriksen PJ, Boorsma A, van Herwijnen MH, et al. (2008) Genomic analysis suggests higher susceptibility of children to air pollution. Carcinogenesis 29: 977–983.
- 11. Hancock DB, Artigas MS, Gharib SA, Henry A, Manichaikul A (2012) Genome-wide joint meta-analysis of SNP and SNP-by-smoking interaction identifies novel loci for pulmonary function. PLoS Genet 8: e1003098 .
- 12. Kraft P, Yen YC, Stram DO, Morrison J, Gauderman WJ (2007) Exploiting gene-environment interaction to detect genetic associations. Hum Hered 63: 111–119.
- 13. Arora R, Metzger RJ, Papaioannou VE (2012) Multiple roles and interactions of Tbx4 and Tbx5 in development of the respiratory system. PLoS Genet 8: e1002866 .
- 14. Moffatt MF, Gut IG, Demenais F, Strachan DP, Bouzigon E, et al. (2010) A large-scale, consortium-based genomewide association study of asthma. N Engl J Med 363: 1211–1221.
- 15. Lasky-Su J, Himes BE, Raby BA, Klanderman BJ, Sylvia JS, et al. (2012) HLA-DQ strikes again: Genome-wide association study further confirms HLA-DQ in the diagnosis of asthma among adults. Clin Exp Allergy 42(12): 1724–1733.
- 16. Willis-Owen SA, Valdar W (2009) Deciphering gene-environment interactions through mouse models of allergic asthma. J Allergy Clin Immunol 123: 14–23; quiz 24-15.
- 17. Holloway JW, Savarimuthu Francis S, Fong KM, Yang IA (2012) Genomics and the respiratory effects of air pollution exposure. Respirology 17: 590–600.
- 18. Aschard H, Lutz S, Maus B, Duell EJ, Fingerlin TE, et al. (2012) Challenges and opportunities in genome-wide environmental interaction (GWEI) studies. Hum Genet 131: 1591–1613.
- 19. Thomas DC, Lewinger JP, Murcray CE, Gauderman WJ (2012) Invited commentary: GE-Whiz! Ratcheting gene-environment studies up to the whole genome and the whole exposome. Am J Epidemiol 175: 203–207; discussion 208–209.