On 2016 May 17, Marcus Munafò commented:

Price and colleagues rightly celebrate the success of genomewide association studies (GWAS) in identifying genetic loci associated with a range of physical and mental health outcomes [1]. We agree that the GWAS approach continues to prove extraordinarily successful, in stark contrast to the candidate gene approach that preceded it [2]. However, what they do not emphasise is that in many cases these genetic loci can tell us as much about modifiable behavioural risk factors as they can about underlying biology.

Perhaps the clearest example of this comes from GWAS of lung cancer, which identified an association with a nicotinic receptor gene cluster CHRNA5-A3-B4 on chromosome 15 (at 15q25) [3]. This cluster encodes three nicotinic acetylcholine receptor subunit proteins: alpha-5, alpha-3 and beta-4. The same locus was shown, in the same GWAS, to be associated with peripheral arterial disease, and has also been shown to be associated with chronic obstructive pulmonary disease [4]. What is critical to interpreting these results is the knowledge that the same CHRNA5-A3-B4 locus has been consistently associated with heaviness of cigarette smoking [3,5]. Functional studies have demonstrated that the minor allele at rs16969968 (i.e., the risk variant for heavier smoking) is associated with a decreased maximal response to a nicotine agonist in vitro [6], while animal studies using alpha-5 knock-out mice have clarified the behavioural effects on nicotine self-administration: knock-out mice respond far more vigorously for nicotine infusions at high doses and do not self-titrate nicotine delivery [7].

There was initially some debate as to whether there is an independent effect of this locus on lung cancer risk (based on evidence of residual association after adjustment for self-reported smoking quantity). However, it is likely this residual association is due to the imprecision of self-report measures of heaviness of smoking, and misclassification of smoking status. For example, the locus accounts for a far greater proportion of variance in nicotine metabolite levels relative to self-report measures of daily tobacco consumption, and this is sufficient to fully account for the association with lung cancer risk [8]. In other words, smoking causes lung cancer, peripheral arterial disease and chronic obstructive pulmonary disease (as well as many other diseases), and this is confirmed by GWAS.

The implication is clear – GWAS can tell us about modifiable behavioural risk factors that contribute to disease [9], and the results of GWAS should therefore be interpreted with this in mind. For example, a recent GWAS of schizophrenia identified the same CHRNA5-A3-B4 locus [10], raising the intriguing possibility that smoking may be a risk factor for schizophrenia [11]. With this in mind, Figure 3 in the commentary by Price and colleagues is striking – one gene associated with a range of different outcomes is ALDH2. The genes in this figure are described as pleiotropic, but here the distinction between biological (or horizontal) and mediated (or vertical) pleiotropy is critical. The former refers to a genetic variant influencing multiple separate biological pathways, while the latter refers to the effects of a genetic variant on multiple outcomes via a single biological pathway. In the case of ALDH2, which encodes the aldehyde dehydrogenase enzyme involved in the metabolism of alcohol and acetalydehyde, it is well established that a variant in this gene influences alcohol consumption [12]. Individuals with one or two copies of the inactive variant experience unpleasant symptoms following alcohol consumption, due to slow metabolism of acetaldehyde and its subsequent transient accumulation [12]. For example, ALDH2 was not identified in GWAS of blood pressure that recruited predominantly European samples [13], where the variant is rare, but was identified in studies that recruited East Asian samples [14,15], where the variant is common. A parsimonious explanation for the associations shown in Figure 3 of Price and colleagues, therefore, is that alcohol consumption causes these outcomes – a results of mediated (rather than biological) pleiotropy.

Behavioural traits such as tobacco and alcohol use can be regarded as intermediate traits, which are under a degree of genetic influence, but which are themselves direct causal agents influencing various health outcomes. This logic also applies to intermediate phenotypes that may lie on the causal pathway, and may be amenable to behavioural or pharmacological intervention, such as LDL cholesterol. While GWAS have been extraordinarily successful in identifying genetic loci associated with disease outcomes, making full use of this knowledge will require an appreciation that GWAS can tell us as much about modifiable – including behavioural – risk factors as it can about underlying biology.

Marcus Munafò and George Davey Smith

1. Price, A.L., et al. Progress and promise in understanding the genetic basis of common diseases. Proc Biol Sci, 2015. 282: 20151684.
2. Flint, J. and, Munafò, M.R. Candidate and non-candidate genes in behavior genetics. Curr Opin Neurobiol, 2013. 23: p. 57-61.
3. Thorgeirsson, T.E., et al. A variant associated with nicotine dependence, lung cancer and peripheral arterial disease. Nature, 2008. 452: p. 638-642.
4. Pillai, S.G., et al. A genome-wide association study in chronic obstructive pulmonary disease (COPD): identification of two major susceptibility loci. PLoS Genet, 2009. 5: e1000421.
5. Ware, J.J., et al. Association of the CHRNA5-A3-B4 gene cluster with heaviness of smoking: a meta-analysis. Nicotine Tob Res, 2011. 13: 1167-1175.
6. Bierut, L.J., et al. Variants in nicotinic receptors and risk for nicotine dependence. Am J Psychiatry, 2008. 165: p. 1163-1171.
7. Fowler, C.D., et al. Habenular alpha5 nicotinic receptor subunit signalling controls nicotine intake. Nature, 2011. 471: p. 597-601.
8. Munafò, M.R., et al. Association between genetic variants on chromosome 15q25 locus and objective measures of tobacco exposure. J Natl Cancer Inst, 2012. 104: p. 740-748.
9. Gage, S.H., et al. G = E: What GWAS Can Tell Us about the Environment. PLoS Genet, 2016. 12: e1005765.
10. Schizophrenia Working Group of the Psychiatric Genomics Consortium. Biological insights from 108 schizophrenia-associated genetic loci. Nature, 2014. 511: p. 421-427.
11. Gage, S.H. and Munafò M.R. Smoking as a causal risk factor for schizophrenia. Lancet Psychiatry, 2015. 2: p. 778-779.
12. Quertemont E. Genetic polymorphism in ethanol metabolism: acetaldehyde contribution to alcohol abuse and alcoholism. Mol Psychiatry, 2004. 9: p. 570-581.
13. International Consortium for Blood Pressure Genome Wide Association Studies, et al. Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature, 2011. 478: p. 103-109.
14. Kato, N., et al. Meta-analysis of genome-wide association studies identifies common variants associated with blood pressure variation in east Asians. Nat Genet, 2011. 43: p. 531-538.
15. Lu, X., et al. Genome-wide association study in Chinese identifies novel loci for blood pressure and hypertension. Hum Mol Genet, 2015. 24: p. 865-874.

On 2016 May 17, Marcus Munafò commented:

Price and colleagues rightly celebrate the success of genomewide association studies (GWAS) in identifying genetic loci associated with a range of physical and mental health outcomes [1]. We agree that the GWAS approach continues to prove extraordinarily successful, in stark contrast to the candidate gene approach that preceded it [2]. However, what they do not emphasise is that in many cases these genetic loci can tell us as much about modifiable behavioural risk factors as they can about underlying biology.

Perhaps the clearest example of this comes from GWAS of lung cancer, which identified an association with a nicotinic receptor gene cluster CHRNA5-A3-B4 on chromosome 15 (at 15q25) [3]. This cluster encodes three nicotinic acetylcholine receptor subunit proteins: alpha-5, alpha-3 and beta-4. The same locus was shown, in the same GWAS, to be associated with peripheral arterial disease, and has also been shown to be associated with chronic obstructive pulmonary disease [4]. What is critical to interpreting these results is the knowledge that the same CHRNA5-A3-B4 locus has been consistently associated with heaviness of cigarette smoking [3,5]. Functional studies have demonstrated that the minor allele at rs16969968 (i.e., the risk variant for heavier smoking) is associated with a decreased maximal response to a nicotine agonist in vitro [6], while animal studies using alpha-5 knock-out mice have clarified the behavioural effects on nicotine self-administration: knock-out mice respond far more vigorously for nicotine infusions at high doses and do not self-titrate nicotine delivery [7].

There was initially some debate as to whether there is an independent effect of this locus on lung cancer risk (based on evidence of residual association after adjustment for self-reported smoking quantity). However, it is likely this residual association is due to the imprecision of self-report measures of heaviness of smoking, and misclassification of smoking status. For example, the locus accounts for a far greater proportion of variance in nicotine metabolite levels relative to self-report measures of daily tobacco consumption, and this is sufficient to fully account for the association with lung cancer risk [8]. In other words, smoking causes lung cancer, peripheral arterial disease and chronic obstructive pulmonary disease (as well as many other diseases), and this is confirmed by GWAS.

The implication is clear – GWAS can tell us about modifiable behavioural risk factors that contribute to disease [9], and the results of GWAS should therefore be interpreted with this in mind. For example, a recent GWAS of schizophrenia identified the same CHRNA5-A3-B4 locus [10], raising the intriguing possibility that smoking may be a risk factor for schizophrenia [11]. With this in mind, Figure 3 in the commentary by Price and colleagues is striking – one gene associated with a range of different outcomes is ALDH2. The genes in this figure are described as pleiotropic, but here the distinction between biological (or horizontal) and mediated (or vertical) pleiotropy is critical. The former refers to a genetic variant influencing multiple separate biological pathways, while the latter refers to the effects of a genetic variant on multiple outcomes via a single biological pathway. In the case of ALDH2, which encodes the aldehyde dehydrogenase enzyme involved in the metabolism of alcohol and acetalydehyde, it is well established that a variant in this gene influences alcohol consumption [12]. Individuals with one or two copies of the inactive variant experience unpleasant symptoms following alcohol consumption, due to slow metabolism of acetaldehyde and its subsequent transient accumulation [12]. For example, ALDH2 was not identified in GWAS of blood pressure that recruited predominantly European samples [13], where the variant is rare, but was identified in studies that recruited East Asian samples [14,15], where the variant is common. A parsimonious explanation for the associations shown in Figure 3 of Price and colleagues, therefore, is that alcohol consumption causes these outcomes – a results of mediated (rather than biological) pleiotropy.

Behavioural traits such as tobacco and alcohol use can be regarded as intermediate traits, which are under a degree of genetic influence, but which are themselves direct causal agents influencing various health outcomes. This logic also applies to intermediate phenotypes that may lie on the causal pathway, and may be amenable to behavioural or pharmacological intervention, such as LDL cholesterol. While GWAS have been extraordinarily successful in identifying genetic loci associated with disease outcomes, making full use of this knowledge will require an appreciation that GWAS can tell us as much about modifiable – including behavioural – risk factors as it can about underlying biology.

Marcus Munafò and George Davey Smith

1. Price, A.L., et al. Progress and promise in understanding the genetic basis of common diseases. Proc Biol Sci, 2015. 282: 20151684.
2. Flint, J. and, Munafò, M.R. Candidate and non-candidate genes in behavior genetics. Curr Opin Neurobiol, 2013. 23: p. 57-61.
3. Thorgeirsson, T.E., et al. A variant associated with nicotine dependence, lung cancer and peripheral arterial disease. Nature, 2008. 452: p. 638-642.
4. Pillai, S.G., et al. A genome-wide association study in chronic obstructive pulmonary disease (COPD): identification of two major susceptibility loci. PLoS Genet, 2009. 5: e1000421.
5. Ware, J.J., et al. Association of the CHRNA5-A3-B4 gene cluster with heaviness of smoking: a meta-analysis. Nicotine Tob Res, 2011. 13: 1167-1175.
6. Bierut, L.J., et al. Variants in nicotinic receptors and risk for nicotine dependence. Am J Psychiatry, 2008. 165: p. 1163-1171.
7. Fowler, C.D., et al. Habenular alpha5 nicotinic receptor subunit signalling controls nicotine intake. Nature, 2011. 471: p. 597-601.
8. Munafò, M.R., et al. Association between genetic variants on chromosome 15q25 locus and objective measures of tobacco exposure. J Natl Cancer Inst, 2012. 104: p. 740-748.
9. Gage, S.H., et al. G = E: What GWAS Can Tell Us about the Environment. PLoS Genet, 2016. 12: e1005765.
10. Schizophrenia Working Group of the Psychiatric Genomics Consortium. Biological insights from 108 schizophrenia-associated genetic loci. Nature, 2014. 511: p. 421-427.
11. Gage, S.H. and Munafò M.R. Smoking as a causal risk factor for schizophrenia. Lancet Psychiatry, 2015. 2: p. 778-779.
12. Quertemont E. Genetic polymorphism in ethanol metabolism: acetaldehyde contribution to alcohol abuse and alcoholism. Mol Psychiatry, 2004. 9: p. 570-581.
13. International Consortium for Blood Pressure Genome Wide Association Studies, et al. Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature, 2011. 478: p. 103-109.
14. Kato, N., et al. Meta-analysis of genome-wide association studies identifies common variants associated with blood pressure variation in east Asians. Nat Genet, 2011. 43: p. 531-538.
15. Lu, X., et al. Genome-wide association study in Chinese identifies novel loci for blood pressure and hypertension. Hum Mol Genet, 2015. 24: p. 865-874.