Genetic Epidemiology of Smoking Behavior and Nicotine Dependence

Abstract

The genetic epidemiology of smoking behavior and nicotine dependence is reviewed. Twin, family and adoption studies show consistent evidence for genetic effects on many aspects of smoking behavior and nicotine dependence, while molecular genetic analyses have only recently identified genes consistently associated with nicotine dependence and amount smoked. Further studies with more sophisticated phenotypes, larger sample sizes and better measures of the environment are needed.

Keywords:

• Nicotine dependence
• smoking
• genetics
• complex trait

INTRODUCTION

As the readers of this Journal well know, the public health significance of sustained smoking is overwhelming. Persistent smoking is the most preventable cause of disability and death and is associated with Chronic Obstructive Pulmonary Disease (COPD), respiratory cancers as well as many other adverse health outcomes. While the prevention of smoking initiation and the support of effective cessation are the keys to reducing and eventually eliminating the health burden of tobacco use, more knowledge of the determinants of smoking behavior is needed. Thus, understanding why some people smoke while other do not, why some smokers smoke persistently, while others can quit easily and why the degree of dependence to nicotine, a key element in tobacco addiction, varies between subjects within populations can provide essential scientific knowledge to construct a better evidence-based, effective tobacco prevention policy. This knowledge base is also needed for health care providers, in particular clinicians dealing with patients suffering from a tobacco-related condition to deliver the best possible care. For example, in this same issue, Ivan Berlin provides a systematic review on the use of nicotine replacement therapy (NRT) for effective smoking cessation in patients, particularly in COPD patients.

Because of individual differences in many aspects of smoking behavior and nicotine dependence, behavioral and molecular genetic research has been undertaken to understand the contribution of genetic factors in the predisposition to smoking. Twin, family and adoption studies have indicated that smoking is familial, with a major genetic contribution as briefly reviewed below. Molecular genetic research is progressing rapidly, and it is too early to review comprehensively what it will reveal.

Twin and family studies have also been used to better define the role of non-genetic factors (such as peer networks), which is possible through the control of confounding by genetic factors in genetically informative data sets. Thus, among Californian adult twins, achieving a high school education halved the likelihood of persistent smoking, parental smoking almost doubled it, and having an MZ co-twin who ever smoked increased the likelihood nearly 10-fold ([1]).

The first twin studies of smoking behavior were reported during the late 1950s and 1960s as an integral part of ongoing debate on the causal role of smoking in lung cancer. For example, R.A. Fisher argued that studies of smoking-discordant twin pairs would provide a test of causality ([2]). The earliest studies, published by the mid-1980s demonstrated fairly consistently that concordance for smoking was higher in MZ pairs than DZ pairs without providing quantitative estimates of heritability. More sophisticated statistical techniques permitted such estimates to be derived during the past 20 years and to test alternative genetic models under a range of assumptions. Thus multiple twin and family study estimates of the genetic contribution of various aspects of smoking behavior have been published, and these are reviewed by Rose et al. (in press, 3).

Most of the studies on the genetic architecture of smoking behavior have examined smoking initiation and smoking persistence, while there are fewer quantitative genetic studies on other smoking behaviors and outcomes including nicotine dependence. Estimated heritabilities for smoking initiation (with varying definitions) from these studies range from. 22 to. 75 for men and from. 32 to. 63 for women. As heritability is a population and time specific estimate of the relative contribution of genes and environment, such variability can reflect not only sampling variability, but also true variation due to multiple factors affecting smoking prevalence and the likelihood that a genetic predisposition can become manifest. Such factors include birth cohort, religiosity, family structure and status, parental smoking history, socioeconomic conditions and availability of tobacco, to mention just a few.

The review by Rose et al. ([3]) finds that for smoking persistence, again defined in multiple fashions in different studies, there is substantial additive genetic variance, while little or no effect from common environment and no evidence for twin-specific environmental effect. Moreover, quite interestingly, most studies find some degree of overlap in genetic effects for initiation with those for persistence. There is nonetheless quite a discrepancy in the amount of overlap between the two stages, which may be partly due to different measurement instruments and analytical techniques. However, whether or not there is overlap in genetic predisposition of initiation and progression to nicotine dependence has important implications of the design of (linkage) and association studies, especially with respect to the definition of control groups. A better understanding of the etiology of various stages of smoking behavior, both in adults and adolescents, may be important in efforts to finding specific or common genes for these stages.

Twin studies on quantity of smoking indicate heritability estimates ranging from 40 to 56%, that is to say, moderately heritable. Smoking amount is also used as a proxy of nicotine dependence, and is a key element in DSM-IV assessments as well as in the Fagerström Test of Nicotine Dependence (FTND). There are fewer studies on nicotine dependence, but these mainly show fairly high and reasonably consistent heritability estimates in different cultures irrespective of the assessment instrument ([3]).

To progress in identifying genes underlying the genetic component implied by family and twin studies, linkage studies of family co-transmission of genes and traits were conducted starting in the 1990s. Most of the initial studies were from family samples created for studying other diseases, in which information on smoking happened to be available. These samples were small, smoking assessments relatively crude and not unexpectedly, rather inconsistent and unreplicated findings were obtained. Lately, larger, better-designed family studies have been completed, and stronger associations of a number of chromosomal regions with smoking behavior have been reported. For example, in a U.S.-Finnish-Australian collaboration the Nicotine Addiction Genetics (NAG) Project has reported significant linkage in two independent family data sets from Finland and Australia, with a combined LOD score for the combined data set exceeding 5 on chromosome 22 for the maximum number of cigarettes ever smoked in a 24-hour period ([4]). In contrast to many earlier studies, the NAG study samples were ascertained for heavy cigarette use and nicotine dependence with a very large sample size and consistent methods. The finding is now being followed-up in an independent case-control study to identify the underlying gene(s).

While animal models and knowledge of the neurobiology of nicotine have provided many candidate genes to be studied for their role in smoking behavior, such as the nicotinic acetylcholine receptor genes and dopaminergic receptor genes, early studies were quite underpowered and results mostly inconsistent. Over the past two years, large scale whole genome-wide association studies (GWAS) have identified novel genes in many complex diseases. For smoking behavior and nicotine dependence, studies published in 2007 and 2008 have consistently implicated the nicotinic receptor gene complex on chromosome 15, with the alpha 5 gene being the most likely the important one in explaining inter-individual differences in risk of nicotine dependence. These genes are now being intensively investigated to more thoroughly characterize their mode of action and role in different aspects of smoking behavior ([5]). One fascinating emerging finding is that it would appear that they might be associated with increased risk of early onset nicotine dependence ([6]). Nonetheless, this chromosome 15 finding only accounts for a small fraction of the genetic risk in nicotine dependence, as it accounts for only 1% of the variance in amount smoked, with persons having the risk and non-risk homozygote genotypes differing in average cigarette consumption by only two cigarettes per day ([7]).

Another interesting aspect of some of last year's GWA studies was that they were case-control studies of lung cancer, yet implicated a gene cluster of nicotinic receptors that showed association with smoking amount and nicotine dependence. Now, in 2009 a study of COPD cases and controls identified the chromosome 15 nicotinic receptor gene region as being associated with COPD ([8]). However, the study also found gene associations independent of smoking on chromosome 4, which were found in another study of lung function parameters ([8], [9]). The latter may eventually help to understand why some smokers have poorer lung function than others.

It is obvious that both better designed and larger studies with various, clearly defined phenotypes are needed to progress in identifying more of the genes responsible. Larger studies are in fact underway, and we can expect new gene findings over the next year or two. The technologies used to identify genetic variants for GWA studies also will develop to cover more of the genetic variation in the genome that is currently not assessed by the current gene chips. Such variation may include copy number variants, rarer alleles with greater genetic risks associated to them, as well as various epigenetic effects.

The models used to estimate heritabilities from twin and family studies and used to evaluate findings from GWA studies all assume that the effects of genes and environments are independent of each other, yet the estimates of heritability may be biased upwards if gene-environment interactions are present. Likewise, GWA studies will not detect genes expressed only in one environment but not in another. Despite expectations of substantial gene-environment correlations and interactions due to influences on smoking initiation by environmental factors shared within families and extra-familial environmental factors shared with peers and birth cohorts, such effects have not been widely studied or reported. One example comes from our own FinnTwin12 cohort study. We found that genetic influences on adolescent smoking decreased, and common environmental influences increased, at higher levels of parental monitoring ([10], [11]). The analysis suggests that limited parental monitoring may offer an environment that allows greater opportunity for the expression of adolescents' genetic predispositions. Likewise Timberlake and colleagues have reported that religiousness significantly attenuated genetic variance on smoking initiation in the AddHealth study ([12]). Further studies of the importance of gene-environment interactions are needed, which may in part explain why heritability estimates vary greatly for smoking behaviors and why the search for specific genes has not been particularly fruitful to date.

Most of the studies referenced and reviewed above have been conducted on European and U.S. Caucasian populations, in the context of relative affluence. Little is known of how genetic factors affect smoking behavior, if at all, in developing countries subjected to the marketing pressure of the major tobacco companies. Likewise, other tobacco products such as pipe, cigar and smokeless tobacco use have been studied much less than cigarette smoking in genetically informative settings. Modern neuroimaging techniques have led to understanding of nicotine's actions in the brain, but clearly we do not still have a complete picture of how nicotine affects various aspects of our neurobiology and neurophysiology that result in inter-individual differences in smoking behavior.

Such an understanding requires the joint efforts of many disciplines including work by geneticists and epidemiologists. Finally scientific knowledge needs to be translated into clinical practice to help patients and enhance primary prevention of tobacco diseases and disorders. At the moment family history continues to be the best predictor of smoking behavior and the current genetic information cannot be used for predictive purposes in the general population. However, the genetic information does provide more understanding of the basic neurobiology of smoking and nicotine dependence and thus eventually translate to better therapies for smoking cessation.

Declaration of interest

The author reports funding for smoking-related research from NIH and Academy of Finland. Consulted for Pfizer on pharmacogenetics of smoking cessation. The author alone is responsible for the content and writing of the paper.

Support: Academy of Finland Centre of Excellence in Complex Disease Genetics