Our study suggests that paternal smoking increases the risk of childhood leukemia, especially ALL, and the effect may be modified by polymorphisms in CYP1A1, an enzyme which is responsible for activation of PAHs.
Many previous epidemiological studies found that prenatal paternal smoking is associated with increased risk of childhood leukemia [7
]. Our meta-analysis of previous reports [7
], showed that lifetime paternal smoking and paternal smoking before pregnancy of the child was significantly associated with increased risk of childhood leukemia. A plausible biological mechanism for the association with paternal smoking before the pregnancy could be cigarette smoke-induced oxidative DNA damage in human semen, which causes chromosome breaks that ultimately lead to translocations in utero
and childhood leukemia development [33
]. Support for this comes from studies that found that the level of DNA damage [34
], or benzo[a]pyrene diol epoxide (BPDE)-DNA adducts [35
], was higher in the sperm of smokers compared with nonsmoking men.
The results of our study suggest that postnatal paternal smoking may also play a role in the development of childhood leukemia and that paternal smoking at home, rather than paternal smoking itself, significantly increases risk of childhood leukemia. Pack-years smoked after birth and the number of smokers in the home during the child's life showed a moderate association with childhood leukemia risk. According to the “two hit” hypothesis proposed by Greaves [36
], for ALL, the initiation step includes chromosome translocations originating in utero
whereas, postnatal exposures provide the second hit. Thus, postnatal exposure to cigarette smoke may increase risk of ALL by causing a secondary genetic event. The interactive effect between CYP1A1
genotype of the child and paternal smoking that we observed in our study also supports a role for children's postnatal exposure to cigarette smoke in the etiology of leukemia risk. We note that our meta-analysis for paternal smoking after birth of the child included only four studies with relatively small number of subjects, and that more studies need to be conducted to evaluate the association between smoking exposure after birth and childhood leukemia.
Environmental exposure to cigarette smoking among children has been related to the level of biomarkers of genetic damage such as sister chromatid exchange [37
], and hemoglobin adducts with BPDE or ethylene oxide which can serve as surrogates for DNA adducts induced by the same chemicals [37
]. Cigarette smoke has been causally related to adduct formation between BPDE and DNA or protein [39
], and increases aromatic hydrocarbon hydroxylase (AHH) activity [40
], and CYP1A1 expression [41
]. AHH activity level has been associated with the CYP1A1
*2A allele [28
], and 462Val [42
], whereas the mRNA level of CYP1A1 was associated only with 462Val [42
]. Given this experimental evidence, it has been hypothesized that a CYP1A1
haplotype including these two SNPs may play a role as an effect modifier for the association between postnatal smoking exposure and childhood leukemia. However, individual SNP analysis does not support a role of these SNPs in childhood leukemia risk and further investigation is needed.
The primary limitation of our study is the small sample size and, as a consequence, low statistical power to detect associations. In addition, the small sample size may have resulted in false positive result by chance due to the increased likelihood of a false positive finding, particularly if the risk factor has a low prior probability [43
]. The suggestive interactive effect found between CYP1A1
diplotype and paternal smoking while plausible also needs cautious interpretation. Thus, it is important to attempt to replicate the findings in other studies of childhood leukemia.
We evaluated possible selection bias since controls were recruited from only one hospital (Seoul National University Hospital), whereas cases were recruited from three hospitals in Seoul including SNUH. Although all three hospitals are considered to be among the top ranking hospitals in Seoul, the socio-economic status of the patients of SNUH tended to be somewhat lower compared to those of the other two hospitals. Smoking rates are higher among lower socio-economic groups in Korea [44
]. However, restricting the analyses to cases and controls from SNUH, we found similar associations for paternal smoking and smoking at home to those for the overall study population; thus, there seems little evidence for selection bias.
Potential recall or reporting bias is common in case–control studies. However, this bias would likely be relatively small because both case and control subjects were admitted to the hospitals and their mothers would be expected to have similar recall [45
]. We observed non-significant results for paternal smoking before pregnancy and significant results for paternal smoking after birth. These results may be partly explained by the fact that the information about smoking history of the father was collected by the mother of the index child, thus starting or quitting age might be relatively inaccurate compared to smoking information provided by the father after birth.
This study is one of several that have evaluated the interactive effects between postnatal smoking exposure and genetic polymorphisms in relation to risk of childhood leukemia and we found some suggestive results in support of effect modification. Additionally, the results of our meta-analysis support an association between prenatal smoking of the father and childhood cancer risk.
In conclusion, paternal smoking may increase the risk of childhood leukemia and the effect may be modified by the CYP1A1 genotype of the child. However, these findings need to be replicated in other populations using accurate smoking exposure data, molecular biomarkers, and complete coverage of tagging SNPs in CYP1A1 and other genes that play a role in the metabolism of carcinogens in cigarette smoke.