Of the eleven SNPs that we analyzed, we observed a statistically significant interaction by occupational radiation dose with genotype for rs2107425 in the H19 gene. In the BCAC study (1
), carrying one or two rs2107425 variant alleles was associated with a decreased risk of breast cancer, which we also observed in the low-dose group (OR=0.8, p = 0.05). In contrast, breast cancer risk in our study was increased in the high-dose group among carriers of the rs2107425 variant (OR= 1.6, p = 0.009). SNP rs2107425 of the H19
gene was found to be only weakly statistically significant in stage 3 of the BCAC study after adjustment for rs3817198 in the LSP1
). We did not observe significant interaction with rs3817198 or any of the other four SNPs found to be most significant in stage 3 of the BCAC study (1
). We did not find any evidence of interaction between the H19
SNP and personal diagnostic radiation breast dose score. This may be explained by the attenuated effect of personal diagnostic radiation breast dose score on breast cancer risk as compared to occupational radiation dose (see footnotes for ).
gene is located on 11p15, a region linked to Beckwith-Weidemann syndrome and is known to be associated with breast cancer and other cancer types by the so called “multiple tumor-associated chromosome region 1”. H19
is a maternally expressed imprinted non-coding mRNA whose specific function is unknown but is closely involved in regulating the insulin growth factor gene, IGF2
). A polymorphism in H19
that increases IGF2
expression may promote carcinogenesis by allowing cells with radiation induced DNA damage to survive, proliferate, and maintain the malignant phenotype (15
). This suggests that rs2107425 in H19
may be important in a radiation-related pathway associated with breast cancer risk (15
). Whole-body radiation exposure in BALB/c mice was associated with an altered H19
methylation pattern (16
) and methylation status of H19
in rats was related to hepatic neoplasms (17
), suggesting that epigenetic phenomena might also play a role in radiation associated carcinogenesis as hypothesized by others (18
). Further testing of the H19
gene and the rs2107425 variant in biologically-based radiation assays may illuminate possible functional relevance for this gene.
Without replication of our finding and laboratory based studies of the 11p15 locus, there are few, if any clinical implications for our finding presently. Based on the apparent relationship with other genes and variants near the 11p15 locus, there could be complex polygenic factors underlying the interaction with occupational radiation exposure. As the number of convincing disease-SNP associations grow, further epidemiologic study of their potential interaction with other established risk factors and association with disease sub-types, ideally in prospective cohort settings where biases may be reduced, will be important to conduct. Such studies may give clues to the function of the variants/genes, potentially guiding laboratory analyses that can more definitively evaluate them and eventually lead to clinical applications.
Strengths of the present study are that the occupational breast doses are derived from a comprehensive dose reconstruction system and have been corroborated by biodosimetry in a separate effort (7
). Limitations include the use of prevalent rather than incident breast cancer cases; however, the prevalence of genotype frequencies by survival time between breast cancer diagnosis and blood collection showed no significant differences (results not shown). A similar analysis considering occupational and personal diagnostic ionizing radiation exposures was not possible because increased survival time was associated with greater age, which is associated with greater cumulative exposure among our study subjects. However, an analysis considering all types of cancers among atomic bomb survivors demonstrated no association between survival time and radiation dose (19
). Furthermore, this was an exploratory analysis with no prior hypothesis regarding radiation interaction with the 11 variants, so chance may explain our finding, which needs to be replicated in other groups.
This case-control study nested within the USRT cohort presented a unique opportunity to evaluate effect modification of SNPs conferring susceptibility to breast cancer by ionizing radiation, an established breast cancer carcinogen (3
). We believe the H19
gene may be a good candidate for functional studies because: the risk estimates for the H19
SNP in the low dose group were consistent with the BCAC study, carefully reconstructed dose estimates were used, the H19
SNP is unlikely to be a correlate of survival, and H19
appears to be related to IGF2
regulation and has some indirect relationships with ionizing radiation in animal models.