We found that the ADRB2 Arg16Gly genotype contributed to the occurrence of wheeze among children who were exposed to tobacco smoke. Across a spectrum of wheeze outcomes, associations of in utero exposure to maternal smoking and childhood exposure to SHS were larger in children with the homozygous Arg16 genotype. Among the genetically susceptible children, a strong dosage-response relationship was found for wheeze-related outcomes, and the risks increased with the number of household smokers. These associations were consistent in 2 cohorts of children who were recruited 3 years apart. Our findings indicate that there are important long-term effects of in utero and postnatal exposure to tobacco smoke in a genetically susceptible group of children.
, 2 other functional SNPs (Arg
in the 5′ promoter region and Thr
) have been studied in addition to the 2 SNPs that we studied (Arg
). The promoter Arg
SNP is in strong linkage disequilibrium with the coding SNPs Arg
, and 3 common haplotypes (Cys
27, and Cys
Because the MAF for Ile
164 is very low (~1%) and the alleles at the Arg
locus could be predicted from the diplotypes estimated from Arg
genotypes, it was not necessary to genotype Arg
ys and Thr
Previous in vitro and in vivo studies provided conflicting data on the functional significance of these SNPs. These inconsistencies may have resulted from conducting experiments on different cell lines (eg, airway smooth muscle, respiratory epithelium, lymphocyte, mast cell, macrophage) with pleiotropic effects of the SNPs on these cells.13–16
Because these SNPs are tightly linked, some researchers have also argued that assessment of individual SNP effects without considering haplotypes in previous studies may have resulted in inconsistent associations between these SNPs and agonist-promoted β
2AR desensitization. Given the haplotype distributions, the Arg
16 allele is carried in the Cys
27 haplotype (Arg
27 diplotype in this study). Emerging data from in vivo studies suggest that Arg
16 homozygosity is associated with enhanced desensitization of the β
and reduced lung function and lung fluid clearance.16
Our data also suggested a recessive genetic model for the joint effects of smoking and Arg
on wheeze outcomes.
The MAFs of these 2 tightly linked SNPs vary across racial/ethnic groups, and there are geographic differences and temporal trends in childhood smoke exposure. These heterogeneities in exposure prevalence and genotype distributions across populations and differential impact of these SNPs on wheeze outcomes by smoking exposures could explain the inconsistent findings in previous studies. Similar to some of the previous studies, we also did not find statistically significant associations between these 2 functional ADRB2
polymorphisms and wheeze/asthma6,18
; however, tobacco smoke exposures modified the effect of the Arg
16 allele on wheeze outcomes. Smoking is a strong risk factor for respiratory health outcomes (asthma, wheeze, lung function, and AHR).19
The adverse effects of smoking could be mediated directly by an increase in AHR and allergic sensitization and by impairment of lung function.20–22
Tobacco smoke may also reduce β
2AR density and ligand binding.23
Besides these direct adverse effects of smoking on these respiratory outcomes and β
2AR activity, a growing body of evidence also indicates that tobacco smoke exposures could modulate the effects of the Arg
genotype on asthma,7
Additional research is warranted to examine the joint effects of Arg
and tobacco smoke exposures on β
Although the underlying reason for the seeming discrepancy in results observed for wheeze and asthma remains unclear, our results are consistent with previous reports. Holloway et al26
found that ADRB2
variants were associated with asthma severity; however, that study was based on a small sample. That finding was not replicated in a larger study and a meta-analysis by Hall et al,6
which included the data from the study by Holloway et al.26
Given the existing evidence, we can speculate that the effect of Arg
16 allele is mediated by increasing AHR,25
which was further accentuated by smoking's leading to wheezing. It could be that these wheezing episodes are attributable to airway hyperreactivity and results in a clinical pattern that supports an asthma diagnosis. This hypothesis needs to be evaluated in a larger prospective study with direct measures of AHR.
We acknowledge that there were some overlaps among the wheeze phenotypes; however, we presented them to show consistency in our results and to be consistent with our previous articles. In addition, heterogeneity in reporting respiratory symptoms has been a concern in epidemiologic studies; therefore, we decided to present data on a few of the symptoms that were examined previously in other articles on this topic.
We considered the potential effects of limitations in this study. Demographic factors, socioeconomic factors, exposure to maternal smoking during pregnancy, SHS after birth, and prevalence of asthma showed significant differences between participants and nonparticipants (data not shown). To address any potential for selection bias from these modest differences, we adjusted for these factors in the final models and found very little change (<10%) in the risk estimates. Furthermore, restricting our analyses to children with health insurance yielded similar results (data not shown). Several studies found that maternal recall of smoking during pregnancy is reliable even 10 to 30 years later.27–30
Parental report of SHS has been found to be reliable in some31–33
and was found to be underreported in children with asthma34
; therefore, although recall bias is possible for in utero and SHS exposures, this may have attenuated the risk estimates because children with asthma were more likely to be in the unexposed group. This potential bias could account for the lack of significant modification of the association between Arg
and asthma by smoking exposures. Because genotyping for the ADRB2
SNPs is not used in diagnosing asthma, diagnostic bias with respect to the studied SNPs is not likely.
Bias from population stratification is a concern in association studies involving multiethnic populations. To address this potential bias, we excluded children who were black or Asian or belonged to a mixed ethnic background because of insufficient sample sizes to conduct ethnic-specific analyses. We acknowledge that substructure within each ethnic group could lead to population stratification; however, this is unlikely given the homogeneity across non-Hispanic and Hispanic white groups. Furthermore, for examining joint effects of exposures and the genetic variants, such intraethnic substructure would have to be related to both factors, which is unlikely. Because the associations between the SNPs and exposures showed similar results by ethnicity and the joint effects of Arg16Gly and maternal smoking during pregnancy showed similar magnitude of associations by ethnicity, our analytic approach of conducting pooled analyses adjusting for ethnicity is unlikely to introduce a bias that could account for the findings.
The modifying role of smoking on the association between ADRB2 Arg
and asthma/wheeze was reported previously for adults,7
and existing data suggest that both smoking and these functional SNPs modulate β
2AR density and activity13,23
; therefore, our analyses were based on specified a priori hypotheses. In this setting, we do not consider it appropriate to disregard previous information on the biology of β
2AR expression and to adjust the P
values for testing a set of selected a priori hypotheses. Some of the results were based on small sample sizes, and we conducted multiple tests on the basis of a priori hypotheses; therefore, these results should be interpreted with some caution.