Commentary on Chen et al. (2009): Gene-environment interactions in nicotine dependence


Smoking patterns aggregate in families, but only few specific genes are known to consistently associate with smoking and nicotine dependence. Twin and family studies have found that additive genetic effects significantly influence smoking initiation, amount and persistence [1]. However, these estimates of heritability vary substantially between studies. With respect to nicotine dependence, heritability estimates are more consistently moderate to high (56–75%). Studies on smoking and nicotine dependence based on linkage scans and candidate gene associations have yielded fairly inconsistent and poorly replicated results, in part due to at most moderate size samples and variability in the definition and assessment of smoking behaviours and different measures of nicotine dependence [1] but are nonetheless consistent in indicating the presence of genetic effects on liability to nicotine dependence.

Since the advent of large genome-wide association studies, a number of such studies in the past two years, together with further confirmatory studies, have consistently implicated the role of genetic variation in the nicotinic receptor gene (α5/α3/β4) complex on chromosome 15 on smoking behaviour and nicotine dependence [2–5], as well as lung cancer and chronic obstructive pulmonary disease risk [3, 6]. However, the proportion of variance accounted for by genetic variants at this locus on chromosome 15 is small, with approximately 1% of the variability in cigarettes per day consumption accounted for the best associating genetic variant, the single nucleotide polymorphism (SNP) rs1051730 [3]. This is in line with findings from many other genome-wide association studies of common traits and diseases, where the identified genes account typically for at most 5–10% of the variance of the trait and substantially less than the heritability estimated from twin and family studies. While the ‘missing heritability’ might be accounted for by other types of genetic variation (rare variants not assessed in current genotyping chips, gene-gene interactions, epigenetic effects, copy number variants) [7], gene-environment interactions are often invoked to account for this apparent discrepancy between results from family studies and genome-wide association studies.

The new article by Chen et al.[8] takes the study of gene-environment interactions in smoking behaviors clearly forward because it links a specific gene variant known to consistently relate to disease risk with a documented moderator of genetic effects seen in twin studies. Statistical models of twin data permit assessing whether the relative contribution of genes and environment is dependent on the context for that gene expression [9], i.e. is there lower or higher heritability at different levels of putative environmental factors. Prior evidence for moderation of genetic and environmental effects on smoking comes from a longitudinal study of Finnish twins, the FinnTwin12 study [10,11]. Analyses indicated that aspects of parenting may moderate the expression of genetic predispositions on adolescent smoking and alcohol use. For example, significant moderating effects were associated with parent monitoring (the measure also used by Chen et al.[8]), such that genetic influences on adolescent smoking decreased, and common environmental influences increased, at higher levels of parent monitoring [10,11]. An interpretation of this result is that limited parent monitoring (i.e. parents are not aware or not interested in their adolescent children's whereabouts or doings) may offer an environment that allows greater opportunity for the expression of adolescents' genetic predispositions. One might speculate that these adolescents at both genetic and environmental risk might then be more strongly influenced by peers to initiate smoking. Chen et al.[8] now show that a genetic variant (the SNP rs16969968, which is highly correlated with rs1051730), which increases the risk of nicotine dependence and amount smoked, has a greater effect on risk of nicotine dependence when parental monitoring (assessed retrospectively for grades 7–8) was reported to be the least present. Interestingly, in the FinnTwin12 study, the effects associated with parenting on adolescent smoking remained significant from age 14 to age 17, consistent with other findings from Finnish longitudinal analyses that common environmental influences on adolescent smoking vary little across adolescence [11]. Nevertheless, replication of Chen et al.'s finding in prospective studies of adolescents is needed, with concurrent assessment of multiple aspects of parenting. The variance explained by the combined effects of genes at chromosome 15 and parental monitoring was over 4%, which is a first step towards better accounting for the large heritabilities reported by family and twin studies, and the variability between studies. Additional work to determine the extent to which gene-environment interactions may exist on adolescent and adult nicotine addiction phenotypes is needed.

As more specific genes related to nicotine addiction and other addictions are identified, we need to study their role and impact in different environments. On the other hand the existence of such genetic variation helps explain why some individuals are more susceptible than others to harmful environmental influences. The impact of individual genes will mostly be very small, thus in the population at large there will exist hundreds to thousands of combinations of environmental and genes. Exact prediction of future outcomes, such as development of substance dependence or abuse, will not become any easier.

Declaration of interest

The author has undertaken consultancy for Pfizer.