To the best of our knowledge, the present study is the first to explore the potential role of endogenous estrogen on risk of CRC in men. We found serum estradiol level to be positively associated with increased risk of CRC in men. Our study is also the first to evaluate the relationship between ESR2 rs1256049 and rs4986938 and risk of CRC in Han Chinese and found both SNPs were associated with susceptibility to CRC in men. We observed a significant association between the risk genotypes of these two SNPs in combination and CRC risk, which was statistically significant in a dose–response manner. In addition, men possessing the risk genotypes of rs4986938 and high level of serum estradiol had significant increased risk for developing CRC.
The role of estrogens in CRC risk remains controversial. In the WHI trial, postmenopausal women who randomly assigned to combination of estrogen plus progestin therapy group were at reduced risk for CRC compared to those assigned to placebo group, the reduction (37%) was nominally significant
]. An early meta-analysis also reported a 20% reduction in risk of colon cancer and 19% reduction in risk of rectal cancer associated with postmenopausal hormone therapy, and particularly current hormone use (34% reduction of CRC risk)
]. These observations strongly supported an inverse association between exogenous postmenopausal estrogens and CRC risk. However, endogenous estrogens may have an opposite effect on CRC risk. In the nested CRC case–control study from the WHI cohort, circulating estradiol concentration was positively associated with increased risk of CRC
]. Similarly, the New York University Women’s Health study reported a 1.8-fold increase of CRC risk in the postmenopausal women having the highest quartile of circulating estrone level compared with those having lowest quartile
]. Although the exact mechanisms are unclear, several observations may help explain the disparate results. First, in the WHI clinical trial, estrogen plus progestin exerted a protective role, whereas oral estrogen alone achieved no such effect
]. Second, the use of oral estrogens usually couples with increased expression of sex-hormone-binding globulin, which leads to reduced concentrations of bioavailable estrogen
]. Finally, oral estrogens showed a negative correlation with the synthesis and activity of insulin-like growth factor I axis components, which have a protective role against CRC tumorigenesis
]. In our study, in agreement with these previous results, we found that high circulating estradiol concentrations were associated with increased risk of CRC in men, supporting the risk effect of endogenous estrogens on CRC development. Further studies to evaluate the sex-difference and dose-difference effect of endogenous estrogens on CRC tumorigensis are warranted.
One of the most intriguing findings in our study was the joint effect of ESR2
rs4986938 genotype and estradiol levels on CRC risk. Although there was no significant interaction between rs4986938 genotypes and serum estradiol levels for CRC risk in our study (data not shown), the joint effect was likely close to the expected additive scale, that is, individuals with both high estradiol level and rs4986938 risk genotypes had a significant OR of 2.3, compared with individuals with low estradiol level and wild-type genotype, which was higher than the sum of ORs in those with high estradiol level only (OR, 1.1) and in those with wild-type genotype alone (OR, 1.0). This result supports the possible influence of interaction between ESR2
gene and endogenous estrogens on the risk of CRC. It is noteworthy that the biological action of estrogens has to be through its receptors, and colon tissue is characterized by a predominance of ERβ, the protein encoded by ESR2
]. Although the mechanisms remain largely unknown, frequent loss of ESR2
expression in CRC tissue has been observed, suggesting involvement of ERβ in development and progression of CRC
]. Moreover, some ESR2
variants have been shown to alter the function of the receptor, affecting the tissue’s response to estrogens
]. There is also evidence that rs4986938 could affect RNA stability of the ESR2
]. Our findings of the joint effect of rs4986938 and estradiol levels on CRC risk and the significant associations between ESR2
SNPs and CRC risk provides supporting evidence of functional potential of ESR2
gene in estrogen-related CRC tumorigenesis, while in vivo
and in vitro
research as well as evaluation in larger population are required for further elucidation.
rs1256049 CT/TT genotypes conferred a reduced risk of CRC in this Han Chinese case–control population. This is in agreement with the study of Slattery et al. which evaluated ESR2
rs1256049 in a case–control population of vast majority of Caucasian and found that the C allele of rs1256049 conferred an increased risk of rectal cancer among the total population if diagnosed before 60
years of age, and an increased risk in colon cancer if estrogen positive
]. A study in Japanese population, however, found an increased risk of CRC linked with rs1256049 TT genotype
]. The ethnic difference may lead to the discrepancy among these studies. More large-scale studies in different ethnic background population are warrant to validate the potential role of ESR2
rs1256049 in susceptibility to CRC.
There are certain limitations to our findings. First, because all patients enrolled in this study were ethnic Han Chinese, our ability to adequately assess risk in other ethnic populations is limited. Further studies using sufficient numbers of subjects from other races and ethnic groups are needed to extend and confirm our findings. Second, only a limited number of candidate SNPs were selected in this study, which is not fully representative of the complexity of genes in the estrogen receptor pathway. Finally, due to the nature of hospital-based case–control study design, a potential selection bias should be taken into consideration when interpreting the results. One of the reasons for the difference between cases and controls in smoking status could be the selection bias. Additionally, using hospital-based controls could generate Berkson bias which might influence the frequencies of ESR2 genotypes and the susceptibility to CRC risk. However, the genotype frequencies in the controls did not deviate significantly from the Hardy-Weinberg equilibrium, thus, the selection bias may have been minimized in this study.