Overall, our results suggest that genetic variants that can influence nicotine metabolization (mainly, CYP2A6 polymorphisms) are those showing the strongest association with smoking status and smoking-related phenotypes. No significant association was observed for those genetic polymorphisms that are involved in the brain effects of nicotine through nicotinic acetylcholine (CHRNA3, CHRNA5), serotoninergic (HTR2A), opioid (OPRM1) or cannabinoid receptors (CNR1) and serotonin transporters (5HTT). The only candidate polymorphism involved in the brain effects of nicotine that was associated with smoking status and with tobacco consumption (expressed as PYS) was DRD2-ANKK1 2137G>A (Taq1A).

The strongest genetic association we found in our study with smoking status and smoking-related phenotypes was for polymorphisms in CYP2A6, the gene encoding the principle nicotine C-oxidase [24]. Smokers who are partially or totally deficient in this enzyme owing to carriage of the variant allele of some CYP2A6 polymorphisms are ‘poor’ (or ‘slow’) nicotine metabolizers; as such, they are theoretically expected to have a reduced need for cigarette consumption compared with the wild-type genotype [25], [26] However, it is also possible that prolonged high levels of brain nicotine owing to reduced metabolization might increase the risk for nicotine dependence, leading to a certain ‘nicotine tolerance’ phenomenon [27], [28]. With regards to these considerations, it must be kept in mind that the decrease in enzyme activity is considerably more marked with carriage of the CYP2A6*2 allele than with the CYP2A6*12 variant. Thus, *2 allele-carriers, who are ‘null-slow’ rather than ‘intermediate metabolizers’ could experience a phenomenon of ‘nicotine tolerance’ with high cigarette consumption (>20 CPD) early in their smoking lifetime [27]. In other words, smokers with the CYP2A6*2 allele might experiment more negative effects when they start to become smokers; but when they continue smoking they may experience prolonged nicotine levels in the brain, thereby becoming more rapidly tolerant and thus needing to smoke more [29]. Our findings are consistent with the aforementioned biological implications of CYP2A6*12 and CYP2A6*2 variants. First, both CYP2A6*2 and CYP2A6*12 polymorphisms were strongly associated with smoking status, yet the variant *2 and *12 alleles were underrepresented and overrepresented respectively in smokers. Second, the CYP2A6*2 variation, but not the CYP2A6*12 polymorphism was associated with smoking phenotypes within the smokers' group, with those individuals homozygous for the *2 allele showing the highest levels of long-term cigarette consumption (PYS). On the other hand, carriage of the CYP2A6*1×2 duplication allele, leading to faster nicotine metabolization was also associated with higher PYS. Our results are in overall agreement with those reported by Rao et al, who showed that individuals with the duplication allele CYP2A6*1×2 had higher nicotine consumption [23].

Regarding those genes involved in the central effects of nicotine, we only found a significant association for DRD2-ANKK1 (Taq1A). The variant A1 allele: (i) was associated with an increased chance of being a non-smoker, (ii) tended to be overrepresented in non-smokers compared with smokers (yet the between-group comparison did not withstand statistical correction for multiple comparisons), and (iii) positively associated with PYS in smokers. There is controversy in the literature: previous studies suggested and association of the A1 allele with susceptibility to smoking [30] but more recent studied failed to replicate such association. The DRD2-ANKK1 gene is involved in the nicotine effects through dopaminergic pathways, with the variant A1 allele being associated with lower density of dopaminergic receptors (DRD2) in the striatum [30]–[32]. Our findings might indeed suggest that people with a functional deficit in the dopamine reward pathway do not experience a reward with smoking initiation, which might confer a protective role to the A1 allele against smoking initiation. However, once they have become smokers, A1-carriers might need to consume more nicotine to enhance the dopaminergic system [30], [33]. This might explain why the A1 allele was positively associated with PYS in our smokers' group.

A novelty of our study stems from the fact that we analyzed the association of the HTR2A-1438A>G polymorphism with smoking status and all smoking-related phenotypes, including nicotine dependence. The serotoninergic system could theoretically be implicated in habitual smoking because nicotine increases brain serotonin secretion and nicotine withdrawal has the opposite effect [34], [35]. Polina et al found a higher frequency of the variant A allele in European-derived Brazilian smokers than in their non-smoking controls [35]. However, our results do not provide evidence for an association between HTR2A -1438A>G and smoking status. Reasons for disparity between the findings reported by Polina et al and the present ones might lie, at least partly, in the different ethnic background of the two study cohorts. Notably, the frequency of the A allele was considerably lower in their non-smoking controls (40%) compared with ours (53.6%).

On the other hand, we found no association for those genetic polymorphisms that are involved in the brain effects of nicotine through nicotinic acetylcholine (CHRNA3, CHRNA5), serotoninergic (HTR2A), opioid (OPRM1), cannabinoid receptors (CNR1) or serotonin transporters (5HTT). Some studies reported a significant association between the aforementioned variants and nicotine dependence [35]–[39] while others failed to corroborate such association [40]–[43]. A marked racial/ethnic diversity exists in smoking behavior and smoking-related phenotypes (such as age of smoking initiation, smoking rate or level of dependence), as well as in the genotype frequencies of the functional polymorphisms we studied here [44], [45], which could explain, at least partly, differences between studies.

We believe there are several novelties and strengths in our design. This the first association study in the field that takes into account the most important polymorphisms that are strong candidates to influence smoking behavior, i.e. those involved in nicotine metabolization, as well as in the brain effects of nicotine. The results of our study are overall valid, as all the following criteria were met [46]: the studied phenotypes (smoking status and smoking-related phenotypes) were properly defined and accurately recorded by a researcher who was blind to the genetic information; both groups (smokers and non-smokers) were ethnically matched; genotype assessment was unbiased and accurate; we adjusted all statistical inferences for multiple comparisons; and the results are overall consistent with previous research in the field [45], [47]. A weakness of our study was the low sample size of both cohorts, yet we believe this can be partly overcome by the fact that both cohorts were homogeneous and well defined in terms of phenotype assessment.

In conclusion, our results suggest that genetic variants potentially involved in nicotine metabolization (mainly, CYP2A6 polymorphisms) are those showing the strongest association with smoking status and smoking-related phenotypes, as opposed to most genetic variants that can influence the brain effects of nicotine, except for the DRD2-ANKK1 2137G>A polymorphism. We believe studies as the present ones might help understanding the role of genetics in smoking behavior and on potential smoking cessation, and to better focus therapeutic approaches based on the knowledge of each individual's genetic predisposition to smoking.