Estrogen receptor-β gene polymorphism and colorectal cancer risk: Effect modified by body mass index and isoflavone intake

Authors


Abstract

Estrogen receptor (ER)-β signaling has generally been implicated in protection against colorectal cancer. The ER-β gene cytosine-adenine (ESR2 CA) repeat polymorphism was reported to be associated with colorectal cancer, although showing contradicting results probably caused by ethnicity or age distribution of the subjects. We investigated the association between this polymorphism and the colorectal cancer risk in a community-based case-control study in Japan (685 cases/778 controls), including only subjects younger than 75. The effect modifications of the body mass index (BMI) and isoflavone intake were also examined. ESR2 CA repeat polymorphism was determined by polymerase chain reaction using fluorescein-labeled primers. CA repeat alleles were classified into short (S) allele (<22 repeats) and long (L) allele (≥22 repeats). Subjects were divided into three genotype groups (SS/SL/LL). The risk of colon cancer, but not of rectal cancer, was increased with an increasing number of L alleles among postmenopausal women; age-adjusted odds ratio (OR) for SL and LL genotypes compared with the SS genotype were 1.78 and 2.91, respectively (trend p = 0.002). Increased risks of colon cancer associated with the L allele were more evident among postmenopausal women with low BMI (<25 kg m−2) or with high isoflavone intake. Such associations were not observed among men or premenopausal women. Having longer ESR2 CA repeat increases colon cancer risk among postmenopausal women younger than 75, possibly with modification of BMI and isoflavone intake. Aging and estrogenic condition may be important in the colon cancer pathogenesis associated with ESR2 CA repeat polymorphism.

A large number of in vitro and in vivo studies have suggested that estrogen may confer a decreased risk of colorectal cancer.1, 2 Estrogens suppress the proliferation of colorectal cancer cells,1 and hormone replacement therapy (HRT) has consistently been related to a decreased risk of colorectal cancer3, 4; however, the evidence is not necessarily consistent. Endogenous estrogen levels have been shown to be positively associated with colorectal cancer risk among postmenopausal women,5, 6 and it was recently reported that reproductive history putatively leading to greater exposure to endogeneous estrogens was related to an increased risk of colorectal cancer in postmenopausal women.7 Moreover, estrogen concentrations were higher in colon cancerous tissues than in nonneoplastic tissues in colon cancer patients, and their prognosis was poorer when intratumoral estrogen concentrations were higher.8

Colorectal epithelial cells frequently express a second estrogen receptor, ER-β, but not the classic receptor, ER-α, and ER-β signaling has generally been implicated as protective in colorectal carcinogenesis.1, 2 Dinucleotide (cytosine-adenine, CA) repeat polymorphism at intron 5 of the ER-β gene (ESR2), which was first reported for a Japanese population,9 has been reported to be associated with bone mineral density,10–12 menopausal symptoms13, 14 and Alzheimer's disease.15 In a previous study on an autopsy series of elderly Japanese (mean age, men, 79; women, 82), colon cancers occurred more frequently among women having a shorter ESR2 CA repeat, although this was not true for men.16 In contrast, Slattery et al. reported an increased risk of colon cancer among women with long CA repeats in a community-based case-control study in the United States (mostly composed of Caucasians).17 Among the differences in the study settings between the two studies, ethnic difference (Caucasians versus Japanese) and difference of age distribution seemed to be most essential. To elucidate which influenced the discrepancy, we examined the association between the ESR2 CA repeat polymorphism and the risk of colorectal cancer in a community-based case-control study in Japan, which was composed of subjects younger than 75. We also addressed effect modifications of the body mass index (BMI) and isoflavone intake. High BMI is known to be associated with elevated production of nonovarian estrogens,18 and isoflavones have an estrogenic property, showing higher binding affinity for ER-β than ER-α.2

Material and Methods

The Fukuoka Colorectal Cancer Study was a case-control study of Japanese residents living in Fukuoka City and three adjacent areas. The study protocol was approved by the ethics committees of the Kyushu University Faculty of Medical Sciences and the participating hospitals. Informed consent was obtained from each subject. Details of the methods have been reported previously.19

Subjects

The cases were a consecutive series of patients with histologically confirmed incident colorectal adenocarcinomas who were admitted to two university hospitals or six affiliated hospitals for surgical treatment from September 2000 to December 2003. Eligible cases were aged 20–74 years at the time of diagnosis; lived in the study area; had no prior history of partial or total removal of the colorectum, familial adenomatous polyposis or inflammatory bowel disease; and were mentally competent to give informed consent and to complete the interview. Of the 1,053 eligible cases, 840 (80%) participated in the interview, and 685 (65%) gave informed consent for genotyping. Colon and rectal cancer cases numbered 384 and 290, respectively; 11 cases had cancers at both sites.

Eligibility criteria for controls were the same as those of the cases except for having no diagnosis of colorectal cancer at the time of selection. A total of 1,500 control candidates were selected by two-stage random sampling using residential registry, and were invited to participate in the study by mail. Among them, 1,382 were found to be eligible. Of these, 833 persons (60%) participated in the survey and 778 (56%) gave informed consent for genotyping.

In the analysis of the interaction between genotype and isoflavone intake, we excluded those who were in the top 1% or bottom 1% of total energy intake within each stratum of sex and age (<55, 55–64 and ≥65 years) in all subjects.

Interview

Research nurses interviewed cases and controls in person regarding dietary and nondietary lifestyle factors.19, 20 The index dates were the date of the onset of symptoms or screening for cases, and the time of interview for controls. Anthropometric questions inquired about height (cm) and weight (kg), current and 10 years earlier. BMI, weight per height2 (kg m−2), 10 years earlier was used in the analysis because current BMI was unrelated to risk.21

The dietary interview ascertained consumption frequencies and portion sizes of 148 food/dish items on average over the past year by using a computer-assisted interview.20 Isoflavone intake was estimated from the average daily intake of nine soy food items, as described in detail previously.22 Energy-adjusted intake of isoflavone estimated by the dietary interview was found to be fairly valid in comparison with the intake based on the 28-day diet record over 1 year among 28 control subjects.22

Other lifestyle factors considered as potential confounders were habitual alcohol consumption 5 years prior to the index date, cumulative exposure to cigarette smoking until the beginning of the previous decade of age, type of job (sedentary or non-sedentary), and leisure-time physical activity as expressed by metabolic equivalents (MET)-hours per week. Women who had been menopausal for 3 years or longer were classified as postmenopausal. Those with a history of hysterectomy were regarded as postmenopausal if they were aged 55 years or older.

Genotyping

Deoxyribonucleic acid (DNA) was extracted from buffy coat using a commercial kit (QIAGEN, Hilden, Germany), and DNA concentrations were adjusted to 50 ng μL−1. Microsatellite polymorphisms were determined by polymerase chain reaction (PCR) using fluorescein-labeled oligonucleotide primers designed to amplify the polymorphic (CA)n repeat in the human ER-β gene (ESR2 gene accession number, ID 2100; ESR2 CA repeat polymorphism, rs3223460). The forward primer was labeled with NED fluorescent dye and used together with the tailed reverse primer (5′-NED-TCCCTGCTACCTTTGTGGAC-3′, 5′-CAGCATGGGACACCACTG-3′). The size of the fluorescence-labeled PCR product was determined by electrophoresis using an ABI PRISM 3730 Genetic Analyzer (Applied Biosystems, Foster City, CA), and alleles were assigned with Gene Mapper software (Applied Biosystems). Using the same cutoff as in previous Japanese studies,13, 16 we designated the allele with CA repeats < 22 as S allele and ≥ 22 as L allele. The subjects were divided into three genotypes, i.e., SS, SL and LL.

Statistical methods

Statistical analyses were performed in men and women separately. Logistic regression analysis was used to estimate the odds ratio (OR) and 95% confidence interval (CI) for the association between the ESR2 CA repeat genotype and colorectal cancer risk. The age at the index date was used in the analysis, and was regarded as age at onset in the cases. Statistical adjustment was made for age, residence area (Fukuoka City or adjacent areas), smoking (0, 1–399, 400–799 or ≥800 cigarettes per year), alcohol intake (0, 0.1–0.9, 1.0–1.9 or ≥2.0 U day−1), BMI 10 years before (<22.5, 22.5–24.9, 25.0–27.4 or ≥27.5 kg m−2), job (sedentary or nonsedentary), leisure-time physical activity (0, 1–15.9 or ≥16 MET-hours per week) and parental colorectal cancer. Results from the multivariate adjustment did not differ from those with adjustment for age alone, and the latter are presented. The trend of increasing (or decreasing) OR with the number of L alleles was tested by the Wald statistic for an ordinal variable representing the number of L alleles. The interaction between the genotype and a covariate was evaluated using the product term of an ordinal variable for the L alleles and a dichotomous variable for a factor of interest. Statistical significance was declared if the two-sided p value was <0.05. Statistical analyses were carried out using SAS version 9.2 (SAS Institute, Cary, NC).

Abbreviations

BMI: body mass index; CA: cytosine-adenine; CI: confidence interval; DNA: deoxyribonucleic acid; ER: estrogen receptor; HRT: hormone replacement therapy; IGF: insulin-like growth factor; MET: metabolic equivalents; MSI: microsatellite instability; OR: odds ratio; PCR: polymerase chain reaction; SHBG: sex hormone binding globulin

Results

The number of ESR2 CA repeats was distributed from 14 to 31 with two large peaks at 18 and 23 among the controls (Fig. 1). Age-adjusted OR of colorectal cancer increased modestly with an increasing number of the L allele in women (trend p = 0.09), but not in men. In women, a more evident increase in the OR associated with the L allele was observed for colon cancer (trend p = 0.002), while no such association was observed for rectal cancer. Neither colon nor rectal cancer showed a measurable association with ESR2 CA repeats genotype in men (Table 1). There were 42 patients with onset at < 45 years old (15 females and 27 males), who may have had hereditary and/or HNPCC-related colorectal cancers. The analysis excluding such cases showed almost the same results. For instance, the age-adjusted OR of colon cancer for the SL and LL genotypes compared with the SS genotype were 1.77 (95% CI 1.07–2.95) and 2.53 (95% CI 1.40–4.59), respectively (trend p = 0.002) in women, and the corresponding OR for men were 0.81 (95% CI 0.55–1.20) and 1.03 (95% CI 0.65–1.65), respectively (trend p = 0.96).

Figure 1.

Distribution pattern of the number of CA repeats of ESR2 in controls. Two major peaks are observed at 18 and 23.

Table 1. ESR2 CA repeat polymorphism and colorectal cancer risk by sex
inline image

When stratified by menopausal status (Table 2), an increased risk of colorectal cancer associated with the L allele was observed among postmenopausal women (trend p = 0.03), but not among premenopausal women (trend p = 0.88). An increased risk among postmenopausal women was evident only for colon cancer (colon cancer, trend p = 0.002; rectal cancer, trend p = 0.95). In premenopausal women, the exclusion of the early-onset cases did not change the results for either colorectal (trend p = 0.46) or colon cancer (trend p = 0.39), but slightly accentuated the decreased OR of rectal cancer for the SL and LL genotypes (trend p = 0.05). In men, the overall and site-specific risks of colorectal cancer showed no measurable difference in the association with the polymorphism between those aged <55 and ≥55 years (data not shown).

Table 2. ESR2 CA repeat polymorphism and colorectal cancer risk by menopausal status among women
inline image

Effect modifications of BMI and isoflavone intake on the association with ESR2 CA repeat polymorphism are summarized in Tables 3 and 4, respectively. Increased risks of colon cancer associated with the L allele were more pronounced among women, especially of postmenopausal status, with low BMI (Table 3) and with high isoflavone intake (Table 4). The interaction among postmenopausal women was nearly statistically significant with respect to BMI (p = 0.07) and isoflavone intake (p = 0.07). Neither BMI nor isoflavone intake modified the association between ESR2 CA polymorphism and colon cancer risk among premenopausal women and among men. The analyses based on continuous variables also suggested effect modifications of BMI (p = 0.09) and isoflavones (p = 0.08) for colon cancer in postmenopausal women.

Table 3. ESR2 CA repeat polymorphism and colon cancer risk with stratification by body mass index
inline image
Table 4. ESR2 CA repeat polymorphism and colon cancer risk with stratification by isoflavone intake
inline image

Discussion

The L allele of the ESR2 CA repeat was associated with an increased risk of colon cancer, but not of rectal cancer, among postmenopausal women. An increased risk of colon cancer with the L allele among postmenopausal women was influenced by BMI and isoflavone intake.

The distribution pattern of the number of ESR2 CA repeats with two major peaks and four intermediate CA repeats was almost identical to that reported for Japanese in other studies,10, 14, 16 but was different from the distribution observed in a study population in the United States consisting predominantly of Caucasians, in which a large peak and a small peak with three intermediate CA repeats were observed.17

The present study provides evidence that longer ESR2 CA repeat is associated with the pathogenesis of colon cancer in postmenopausal women younger than 75. In women, longer ESR2 CA repeat lengths have been associated with lower bone mineral density,11, 12 higher osteoporosis risk,23 and increased sex hormone binding globulin (SHBG).24 Considering that estrogens prevent osteoporosis, and SHBG reduces free sex steroid concentration, those observations suggest that the estrogen-ER-β signaling pathway is less functional in women with longer ESR2 CA repeat alleles. The ESR2 CA repeat polymorphism has been postulated to affect the transcription or alternative splicing of ESR2 Exon 8, resulting in the determination of ER-β isotype.15 Of the several isotypes,25 only ER-β1 (the wild type of ER-β) has ligand binding ability for estrogens, and has estrogen signaling function. Epithelia of colonic cancer and adjacent mucosa in women with the SS genotype expressed ER-β1 more frequently than those with the SL genotype, which suggests that the L allele is associated with less ER-β1 transcriptional activity.16 ER-β signaling has generally been implicated as protective against colorectal carcinogenesis1, 2 and tumor development.26 The increased risk of colon cancer associated with the L allele among postmenopausal women seems to reflect impaired ER-β signaling caused by lower ER-β1 expression.

The present results that a longer ESR2 CA repeat was associated with colon cancer risk among postmenopausal women are consistent with the observation in a case-control study in the United States,17 and are opposite of the previous results for Japanese elderly autopsy series.16 Our findings demonstrate that the discrepancy between the two previous studies is not ascribed to the ethnic difference (Caucasians versus Japanese). In the Japanese autopsy series, the mean age of women was 82, and the subjects with latent colorectal cancer, first discovered at autopsy, were also considered as colorectal cancer cases. The opposite results between the two Japanese studies may reflect a complex role of ER-β and estrogens in colon carcinogenesis, which differs according to age. For example, microsatellite instability (MSI) is an important colon carcinogenic mechanism, and MSI-positive colon cancer is frequently observed among elderly women.27 Although ER-β expression has been generally considered to be protective against colorectal carcinogenesis,1, 2 high expression of ER-β1 was reported in MSI-positive colon cancer.28 It may be useful to examine whether the ESR2 repeat genotype affects the age at the onset of colon cancer among women; however, the mean age at onset did not differ according to the ESR2 repeat genotypes in either women aged 45 years or more as a whole (SS genotype 61.4; SL genotype 61.4; and LL genotype 61.8 years) or postmenopausal women (SS 65.1; SL 65.4 and LL 64.3 years) at least in the present study. Although colon cancer was frequently found among elderly women with SS genotype in the autopsy series, the tumors in those women were less invasive compared with SL/LL genotypes.16 Difference of age at onset according to the ESR2 repeat genotypes may appear at a more advanced age.

Another novel finding was that an increased risk of colon cancer associated with the L allele of ESR2 CA repeat polymorphism was confined to postmenopausal women with low BMI. BMI affects circulating concentrations of estrogens in postmenopausal women, in whom the main source of estrogens is peripheral fatty tissue where aromatase converts adrenal and/or ovarian androgens into estrogens.18 High BMI has been consistently associated with an increased risk of colon or colorectal cancer in men, but the association between BMI and colon cancer has been controversial among women.29–32 It was shown that high BMI was associated with an increased risk of colorectal cancer in premenopausal women, but not in postmenopausal women.33, 34 Hyperinsulinemia and elevated levels of insulin-like growth factor (IGF)-1 are postulated to be a biological mediator for an increased risk of colon cancer associated with overweight and obesity.6, 35 Among postmenopausal obese women, elevated levels of estrogens due to high BMI may offset the risk related to elevated levels of insulin or IGF-1. In other words, the effect of the ESR2 CA repeat polymorphism may have been less discernible among postmenopausal women with a higher expression of mediators other than estrogen-related signaling molecules.

It has been suggested that isoflavone intake may protect against the risk of colorectal cancer2; however, epidemiologic evidence remains sparse and inconsistent.22 The present study suggested an effect modification of isoflavone intake on the association between ESR2 CA repeat polymorphism and colon cancer in postmenopausal women. A strong positive association with the L allele observed for postmenopausal women with high isoflavone intake was largely ascribed to a substantial decrease in the risk among those with high isoflavone intake who had the SS genotype. The binding affinity of a representative isoflavone, genistein, for ER-β1 is more than 10 times higher than that for ER-α, and almost the same as the affinity of estradiol for ER-β1 and about 40 times higher than that of estrone.36, 37 As described above, ER-β1 was expressed more frequently in epithelia of colon cancer and adjacent mucosa of women with the SS genotype.16 Colonic mucosa of women with the SS genotype may be more sensitive to the protective effect of isoflavones. In premenopausal women, who are exposed to abundant ovarian estrogens, the effect of the differential expression of ER-β according to ESR2 CA repeat genotypes may have been masked irrespective of isoflavone intake or BMI levels.

The present results seem to have an important implication regarding studies on estrogens and colorectal cancer. There have been many such studies; however, ESR2 CA repeat polymorphism has not been examined simultaneously. Inconsistent results regarding estrogen effects on colorectal cancer previously reported among women may be resolved considering ESR2 CA repeat genotype, which seemingly influences estrogenic action on postmenopausal colonic epithelia. Further, age should be also considered, because aging may be important in the colon carcinogenic mechanism associated with ESR2 CA repeat polymorphism.

It is notable that none of the three studies indicated an association between the ESR2 CA repeat polymorphism and colon cancer in men. Furthermore, the ESR2 CA repeat polymorphism was unrelated to rectal cancer even among postmenopausal women, suggesting that estrogens are less important in the pathogenesis of rectal cancer.

In conclusion, the L allele of the ESR2 CA repeat was associated with an increased risk of colon cancer in Japanese postmenopausal women younger than 75, especially with low BMI or high isoflavone intake. The opposite results from the previous results for Japanese elderly autopsy series may explain the differences in female colon carcinogenic mechanisms according to age. The study substantiated a role for ESR2 CA repeat polymorphism and estrogens in colon cancer pathogenesis among postmenopausal women, although the mechanisms in relation with aging remain to be clarified.

Acknowledgements

The authors acknowledge support from Emeritus Professors Keizo Sugimachi, Seiyo Ikeda, Sumitaka Arima and Takayuki Shirakusa and from Drs. Motonori Saku, Yoichi Ikeda, Soichiro Maekawa, Kazuo Tanoue, Kinjiro Sumiyoshi and Shoichiro Saito in conducting the survey of cases. The authors also acknowledge support from Drs. Motoji Sawabe, Kaiyo Takubo, Shinobu Ikeda, Kenji Shimizu and Hironobu Sasano for their useful advice. The following physicians kindly supervised the survey of controls at their clinics: Drs. Hideaki Baba, Tomonori Endo, Hiroshi Hara, Yoichiro Hirokata, Motohisa Ikeda, Masayoshi Ishibashi, Fumiaki Itoh, Yasuhiro Iwanaga, Hideki Kaku, Shoshi Kaku, Minoru Kanazawa, Akira Kobayashi, Ryunosuke Kumashiro, Shinichi Matsumoto, Soukei Mioka, Umeji Miyakoda, Osamu Nakagaki, Nobuyoshi Nogawa (deceased), Nobuyuki Ogami, Toyoaki Okabayashi, Hironao Okabe, Nishiki Saku, Masafumi Tanaka, Masahiro Ueda, Bunichi Ushio and Koheisho Yasunaga. The authors have nothing to declare.

Ancillary