In the current investigation, we first explored the effect of genetic variants in the dopamine pathway on smoking initiation in a large, cross-sectional sample of Chinese Han adolescents. After an extensive pathway-based candidate gene selection, a comprehensive SNP selection, genotyping, and imputation for untyped SNPs, 2,723 SNPs in 57 genes were examined for association with smoking initiation. Thereafter, SNPs surviving stringent multiple-testing correction in the Chinese population were examined in an independent Caucasian adolescent cohort sample in Los Angeles, California, a dramatically different sociocultural setting. When screening for the effects of the candidate genes in the WSPT, 1 SNP in CALY (rs2298122) was associated with smoking initiation only in females. In the CHS, highly consistent patterns were observed for rs2298122. This replication in a completely independent adolescent sample from a different population with a distinct design appears promising in implicating CALY in smoking initiation.
Previous investigations have demonstrated that CALY
might be important in brain function and psychiatric disorders. CALY
codes for calcyon neuron-specific vesicular protein (65
), a novel brain-specific protein. Recent investigations revealed that the calcyon C terminal can stimulate the self-assembly of clathrin in a dose-dependent fashion and is therefore involved in clathrin-mediated endocytosis, which is crucial for efficient synaptic transmission and optimizing the levels of releasable pools of neurotransmitters (66
) and is thereby involved in dopamine-related signaling and dopamine activity. Both dopamine-related signaling and dopamine activity have been implicated in various brain functions, such as motor control and cognitive processing (67
). Specifically, it has been shown that dopaminergic neurotransmission plays an important role in impulsive decision-making (45
), sensation-seeking (70
), novelty-seeking (71
), and extraversion (72
), all of which have been identified as risk factors for reaction to smoking cues and smoking initiation (46
The top-hit SNP, rs2298122, is located in intron 5 of CALY
, and the role of this variant in human cognition and behavior is not well understood. Within HapMap 3, 2 SNPs in the CEU population (rs7085530 and rs2105341) and 4 SNPs in the CHB population (rs11101694, rs2275723, rs2105341, and rs7085530) are in notable linkage disequilibrium with rs2298122 (r2
> 0.5). However, none of these SNPs are nonsynonymous SNPs. Among the few association studies on CALY
, Laurin et al. (75
) performed a nuclear-family-based association study of attention deficit/hyperactivity disorder in Caucasian adolescents. Although their results are only suggestive, the association between rs2298122 and the impulsive dimension of attention deficit/hyperactivity disorder was marginally significant (P
= 0.08) (75
). The molecular mechanism for the association between CALY
and smoking initiation might reflect a potential effect via impulsivity, which has been shown to play a complex role in smoking behavior and the reaction to smoking cues (45
The association between CALY
and smoking initiation was observed only in females. It is well known that smoking behavior differs between the genders, especially in some populations, such as the Chinese (76
). Furthermore, a recent investigation in 32,359 pairs of native California twins indicated a gender difference in the pattern of genetic and environmental determinants of smoking initiation (77
). Moreover, a gender difference in the association between some psychological characteristics and smoking-related behaviors has been repeatedly reported (78
). These studies indicate that there may be a subtle difference in the mechanisms or underlying genetic factors for smoking initiation between the genders. Our observation of an association only in females may partly reflect this.
In the WSPT, we also observed a strong effect of CRHR1
, although this association was not replicated in the CHS. CRHR1
codes for the type I receptor of corticotrophin-releasing hormone (80
), and it has been associated with depression (81
), suicide (83
), and alcohol dependence (84
) in previous investigations. However, the role of CRHR1
in smoking-related behaviors has not been demonstrated. There are 2 possible explanations for the lack of replication of the positive findings in the CHS. First, in the WSPT, this association was only marginally significant for a pathway-wide significance level. Considering the fact that the best-fitting heredity model for this SNP was recessive, sparse data bias (86
) might have arisen and led to unstable estimates or false positives. Furthermore, even if the findings for CRHR1
in WSPT are true, statistical power to detect the recessive genetic effects would have been low in the CHS because of the limited sample size. Second, none of the CRHR1
SNPs associated with smoking initiation in the WSPT are located in the coding region, indicating that those SNPs might be markers for another rare SNP in linkage disequilibrium with them in a Chinese population. Therefore, the different linkage disequilibrium structure between the Chinese and Caucasian populations might have led to inconsistent association results in the 2 populations. In either scenario, further functional or population-based data may help to clarify the role of CRHR1
in smoking initiation.
The primary phenotype of the current investigation, smoking initiation in young adolescents, is different from lifetime smoking initiation. Specifically, given the high prevalence of smoking in adult Chinese males (>60%) (75
), some persons who will initiate smoking later in adolescence or adulthood might have been treated as controls in the WSPT. This may have led to decreased statistical power if we believe that lifetime smoking initiation is a homogenous phenotype. However, since smoking is integral to social interactions among adults in the Chinese population, there may exist different mechanisms or competing risk factors for smoking initiation at later ages, as compared with early smoking initiation, in which personality factors such as impulsivity, behavior control, decision-making, and novelty-seeking might play more important roles—especially in adolescent females. In addition, we limited our analysis in the WSPT to smoking status at the time of enrollment and did not include longitudinal follow-up data. This was primarily done to avoid any heterogeneity in inference that may have existed postintervention, since the WSPT was specifically designed to test the effectiveness of a school-based intervention aimed at altering adolescent smoking trajectories. In contrast, because the CHS is a cohort representing more natural smoking trajectories, our replication analysis utilized the entire longitudinal history. If we restrict our analysis to comparable ages (12 years and 13 years), we see that the pattern of association remains (odds ratio = 2.25, 95% CI: 0.9, 5.61), although it is not statistically significant because of decreased information in the small sample.
Subtle differences between early and lifetime initiation highlight the difficulty of research in this area, especially when considering the numerous nongenetic (host, social, and environmental) factors that are known to influence smoking initiation. The impact and temporal dynamics of these factors may lead to many gene-environment interactions, making discovery and replication across cohorts more difficult. As an example, in 2 recent genome-wide association studies on smoking behavior (Cancer Genetic Markers of Susceptibility and Genetic Association Information Network), no significant association (defined as P
< 0.00001) between adult smoking initiation and variants in the CALY
) was identified in an analysis combining both males and females. These differences in findings may be partly attributed to differences in the definition of smoking initiation, gender interaction, or differences in exposure to nongenetic factors. In any case, further replication and characterization via stratified analysis is required in this area of genetic research.
While a long-term goal of research in this area is aimed at elucidating potential interactions, we focused the current analyses on main effects of genes and heterogeneity due to gender. Although the positive findings of our exploratory investigation in the WSPT are supported by an independent sample, there are no functional data to further illustrate the putative association. Furthermore, when the WSPT cohort study was initiated in 1998, information on related psychological traits associated with smoking initiation, such as impulsivity and novelty-seeking, was not collected, and measures of depression and hostility were not derived from standardized, established items. This has greatly limited our ability to explore the role of personality characteristics as mediators of or intermediates in the association between the dopamine pathway and smoking initiation. Despite these caveats, we believe the evidence is sufficient to encourage further investigation to verify the potential role of CALY variants in smoking initiation.
In summary, we performed a pathway-based association study in a homogenous Chinese Han adolescent population to explore the role of the dopamine pathway in smoking initiation. We identified 1 SNP (rs2298122) in the CALY gene that was positively associated with smoking initiation only in females. Supportive evidence for this association was subsequently observed in an independent sample of Caucasian adolescents. While these findings will ultimately need to be replicated in additional population-based samples and explored in functional studies, the results of our investigation provide a foundation for future research in understanding the genetic mechanism of smoking initiation in adolescents.