Abstract
Worldwide, colorectal cancer has a higher incidence rate in men than in women, suggesting a protective role for sex hormones in the development of the disease. Preclinical data support a role for estrogen and its receptors in the initiation and progression of colorectal cancer and establishes that protective effects of estrogen are exerted through ERβ. Hormone replacement therapy (HRT) in postmenopausal women as well as consumption of soy reduces the incidence of colorectal cancer. In the Women's Health Initiative trial, use of HRT in postmenopausal women reduced the risk of colon cancer by 56% [95% confidence interval (CI), 0.38–0.81; P = 0.003]. A recent meta-analysis showed that in women, consumption of soy reduced the risk of colon cancer by 21% (95% CI, 0.03–0.35; P = 0.026). In this review, using the preclinical data, we translate the findings in the clinical trials and observational studies to define the role of estrogen in the prevention of colorectal cancer. We hypothesize that sometime during the tumorigenesis process ERβ expression in colonocytes is lost and the estrogen ligand, HRT, or soy products, exerts its effects through preventing this loss. Thus, in the adenoma-to-carcinoma continuum, timing of HRT is a significant determinant of the observed benefit from this intervention. We further argue that the protective effects of estrogen are limited to certain molecular subtypes. Successful development of estrogen modulators for prevention of colorectal cancer depends on identification of susceptible colorectal cancer population(s). Thus, research to better understand the estrogen pathway is fundamental for clinical delivery of these agents. Clin Cancer Res; 19(21); 5842–8. ©2013 AACR.
Background
Age is the strongest risk factor for colorectal cancer, with 90% of cases and 94% of deaths reported in individuals older than 50 years. Yet the incidence is lower in women across all age groups (1). This epidemiologic observation can be explained by differences in the colorectal cancer–related risk factors between men and women such as obesity, smoking, dietary exposures, and physical activity. However, persistence of the gender differences across several decades, in different racial groups and across the world, suggest that sex hormones (estrogen and progesterone) play a role in the pathogenic pathways of colorectal cancer and this role is most likely protective.
In addition, studies from relationship between gender and mortality of colorectal cancer consistently show lower mortality for women, especially premenopausal women. Epidemiologic studies over several decades suggest a decline in mortality of women attributed to use of hormone therapy (2, 3).
The protective role of the estrogen pathway in development of colon cancer has been studied and established in animal models. In several studies, exposure of ovariectomized rats to estrogen reduced the rate of colon tumors by 71% even in the presence of an ERα knockout (4, 5). In addition, in the ApcMin/+ mice ovariectomy resulted in an increased number of polyps, whereas replacement of estrogen in these mice reduced the number of polyps to the baseline values (6).
Over the past two decades use of hormone replacement therapy (HRT) provided the opportunity to discover the role of this intervention in prevention and outcome of colorectal cancer. Data from multiple prospective and retrospective cohorts support a protective role for HRT in the development and outcome of colon cancer.
Experience with soy products complements the hypothesis that estrogens have a protective effect against colorectal cancer. Lower incidence of colorectal cancer in Asians is attributed to high soy content. In animal models genistein, estrogen in soy, inhibits colon cancer cell proliferation and enhances apoptosis through interaction with several pathways (7, 8).
The estrogen pathway has the potential to be targeted for preventive and therapeutic strategies in colorectal cancer; therefore, its role and relationship with other pathways warrant further investigation. In this review, we provide comprehensive evidence on the role of estrogen and progesterone and how their signaling pathways play a critical role in colorectal cancer development and outcome.
Molecular effects of estrogen and progesterone in colorectal cancer
Estrogens are members of the steroid hormone family and are traditionally associated with female reproductive development. The most abundant and most potent estrogen in humans is 17β-estradiol (E2). Other natural estrogens include estrone (E1) and estriol (E3), metabolites of E2. The estrogen receptor (ER) recognizes a molecule as estrogen based on its three-dimensional configuration and charge, but preferentially binds E2 over E3 (2×) and E1 [3× (ref. 9)]. Estrogen production and metabolism are tissue specific and are different in colon compared with breast tissue. In the colon, 17βHSD2 and 4 convert E2 to E1, which is antiproliferative in human colon cancer cell lines (10, 11).
The discovery of different ERs has revealed their significant role in tissue types other than the female reproductive tract, including the gastrointestinal system (12). Estrogens regulate these cellular effects through their intracellular receptors, ERα and ERβ. The two receptors are coded for by separate genes, ESR1 (ERα) and ESR2 (ERβ), with each gene producing different receptor isoforms from alternative splicing, resulting in three ERα and five ERβ variants. The two receptors share DNA homology, coreceptors, and downstream signaling, however; their dissimilarities dictate the differences observed in their effects (13). Estrogen interacts differently with ERα and ERβ downstream pathways (14). The differences in the downstream signaling of these two receptors may explain their distinctive intracellular effects. Furthermore, it has been shown that the biologic response of activated ERα and ERβ is dependent on the ratio of ERα to ERβ in the cell and that ERβ inhibits transcriptional activity of ERα (15).
The intracellular effects of estrogen are exerted through two main pathways: genomic and nongenomic.
Genomic effects of estrogen in colorectal cancer.
An activated ER activates gene transcription through either direct interaction with specific DNA sequences known as estrogen response elements (ERE) or other transcription factors such as c-Jun and/or c-Fos, resulting in transcription (Fig. 1). While interaction of both ERs with ERE is similar, ERα interacts with other transcription factors such as c-Jun and c-Fos of the activating protein-1 complex (AP1) and SP1, whereas ERβ's interaction with these transcription factors is less significant (13). In the case of AP1, it has been shown that binding of E2 to ERα results in transcription activation and that binding to ERβ inhibits transcription through diversion of estrogen away from the ERα pathway (16, 17).
Genomic effects can also be seen in a ligand-independent manner, where ERs are phosphorylated through an activated kinase pathway in the absence of the estrogen ligand (Fig. 1). As an example, the EGF receptor (EGFR) can activate the Ras/Raf/MAPK pathway, which in turn will phosphorylate the ER, resulting in dimerization and ligand-independent activation of gene expression (18, 19).
The genomic-mediated effects of estrogen result in activated transcription of a wide array of genes including genes involved in angiogenesis, such as VEGF, cellular adhesion, such as cadherins and laminins, and proliferation and apoptosis, including the TGFβ family.
Nongenomic effects of estrogen in colorectal cancer.
In addition to genomic effects, estrogen and its receptors can activate several signaling pathways without direct interaction with DNA, resulting in modulation of additional cellular processes. Transmembrane ERs have been shown to activate diverse intracellular pathways, including protein kinase C (PKC; ref. 20), intracellular Ca2+ (21, 22), cytosolic cAMP (23), nitric oxide (24), and MAPK (ref. 25; Fig. 1). For example, transmembrane ERα signaling through PI3K can result in cell proliferation and survival, whereas transmembrane ERβ signaling causes the influx of Ca2+ in the cell, resulting in suppression of PKC signaling (Fig. 1). An additional role of ERβ is through modulation of cell-cycle control through interactions with c-Myc, cyclin D1 (26), and cyclin A (27) resulting in inhibition of cell-cycle progression.
In both normal and cancerous colonocytes, ERα expression levels remain low. In contrast, ERβ is the predominant ER in the normal colon (28, 29) with higher expression level in the ascending colon (30). The expression level of ERβ in tumor tissue compared with normal colon mucosa is decreased and correlates with stage of the disease (31, 32).
Hartman and colleagues showed that transfection of SW480 cell lines with ERβ resulted in inhibition of proliferation and cell-cycle arrest. SW480 xenografts with ERβ expression had 70% reduction in the tumor weight (33). Furthermore, knockout of ERβ in ApcMin/+ mice resulted in a dramatic increase in the polyps, and addition of E2 treatment did not prevent development of polyps in these mice (34).
Edvardsson and colleagues had shown that transfection of colon cancer cell lines with ERβ affects the MAPK signaling pathway (35). Furthermore, ERβ results in downregulation of interleukin-6 and thus in reducing inflammation (35). Giroux and colleagues postulate that the effects of ERβ in ApcMin/+ mice are through modulation of the TGFβ signaling pathway (34).
Several studies support the cross-talk between Wnt/β-catenin signaling and ERα. In one experiment, ERα transfection of SW480 and HCT116 resulted in activation of Wnt signaling and addition of an ER antagonist resulted in deactivation of the pathway. Moreover, in this experiment adding an antibody against β-catenin resulted in activation of ERE in an estrogen-dependent manner (36).
Despite the established cellular role for the ERs in colorectal cancer, the exact role of estrogen in colorectal cancer remains unclear. Effects of estrogen on colonocytes may be related to the direct effects of the estrogen ligand and its interaction with the receptors (directly related to ligand) or changes in the ER ratio and function independent of the ligand (indirectly related to ligand). For example, the rise in the incidence of colon cancer after menopause in women, when the circulating amount of estrogen ligand is decreased, could be related to the loss of cell-cycle regulatory effects of ERβ (directly related to ligand), or that the decrease in the endogenous estrogen level at the time of menopause results in a change in the ratio of ER (indirectly related to ligand). In fact, in animal models the transcriptional activity of ERs changes over time and is influenced by estrogen level (37). In APC mouse models, E2 treatment results in an increase in the ratio of ERβ and protection against colon cancer (6). Therefore, it is likely that HRT in women protects against colon cancer through an increased ratio of ERβ. As ERβ expression decreases throughout the process of tumorigenesis the timing of HRT to prevent loss of ERβ is critical (38).
As for estrogen metabolism, polymorphism in the metabolic pathway of estrogen may result in a higher level of E2 and a lower level of E1 in colonocytes, thereby stimulating growth. The expression level of 17β-HSD2 and 17β-HSD4 is lower in colon cancer than in normal colon tissue, resulting in lower levels of E1 in the tumor. The imbalance in the level of the estrogen metabolites can result in an increase in proliferation or a decrease in apoptosis (10, 11, 39). In a recent study, the level of E2 was higher in a cohort of Chinese men with colon cancer compared with controls. In this population combination of genetic variations in the ESR2 and E2 level was predictive of the risk of colon cancer, whereas in the group with high E2 level and CT/TT genotypes of ESR2 the risk of colon cancer was 2.3 (95% CI, 1.4–3.9) compared with the risk in those with a low level of E2 and CC genotype (40).
The literature on soy supports that use of isoflavones, a strong ERβ agonist, in the DLD-1 human colon cancer cell line, resulted in G2 cell-cycle arrest. Use of soy isoflavones after silencing of ERβ in this cell line failed to suppress G2 arrest (41). Pintova and colleagues had shown that genistein reduces proliferation in HT29, DLD1, and RKO cell lines through inhibition of the Wnt/β-catenin pathway (42). While these finding are exciting, results of a meta-analysis of the relationship between use of soy and risk of colorectal cancer in humans suggests that the protective effects of soy are limited to women (43), thus suggesting that effects of soy depend on the endogenous estrogen level through altering estrogen metabolism (44). Furthermore, use of soy isoflavones in rats resulted in increased expression of ERβ and decreased expression of ERα in osteocytes; exposure of colonocytes to soy may also result in increased expression of ERβ.
In addition to estrogen, the steroid hormone progesterone has been shown to play a distinct role in the regulation of the normal female reproductive function and regulating the bone matrix, but further has antiestrogen effects in other tissue types (45). Cellular effects of progesterone are mediated through a similar pathway to that of the ER, in which progestins bind either progesterone receptor (PR) A or B and mediate transcription through interaction with progesterone response elements. The only prominent nongenomic effect of progesterone that is reported is a rapid and transient activation of the MAPK pathway (46). The observed cellular effects of progesterone are largely attributed to its ability to oppose the action of estrogen, causing an inhibition of ER expression and through abrogating the induction of estrogen-related genes (47). Cellular effects of progesterone in colonocytes have been minimally studied. In mammary cells, interception of the PR by progesterone results in upregulation of EGFR and Wnt-1 (48–50). In animal breast cancer models, combination of estrogen and progesterone results in increased amphirgulin and augmented proliferation compared with estrogen alone. Assuming that EGFR and Wnt-1 can be upregulated in colonocytes, endogenous progesterone may have a proliferative effect on the colonocytes. These findings are, presumably, fitting the interactions of progesterone with ERα, and interactions between ERβ and progesterone are unexplored. Medroxyprogesterone, the progesterone formulation in HRT, has a direct inhibitory effect on colon cancer cell lines, supporting the observation that HRT is protective against colon cancer (50).
Epidemiologic data
The results of the Women's Health Initiative (WHI) study raised the awareness of the scientific community about the interactions between sex steroids and incidence and outcome of colon cancer. One limitation of the WHI study is that colon cancer was not the primary endpoint of the study and therefore the number of colorectal cancers in the study was small (122 during the trial and 263 during and after follow-up), making the data hypothesis generating. Hartz and colleagues published an analysis inclusive of all WHI enrollees and its observational cohort and found that use of any form of HRT reduces the risk of colon cancer by 30% [95% confidence interval (CI), 0.62–0.80; P < 0.01] and the risk of rectal cancer by 43% (P < 0.0001; ref. 51).
As the publication of WHI results, several observational studies have examined the role of HRT in the incidence and outcome of colon cancer. Table 1 summarizes the data from a few selected observational studies. The majority of these studies suggest that use of HRT reduces the incidence of colon cancer (52–55), but, this finding is not confirmed in other studies (52, 53, 56, 57). It is important to point out that these studies are planned and analyzed over several decades, with different formulations of hormones being used by participants. There is a large variation in the timing of start of these hormones after menopause and duration of hormonal use. It is unknown whether they are used to alleviate symptoms of menopause or for other reasons. More importantly, due to the observational nature of these studies, compliance with the treatment and indication for treatment is unknown. Furthermore, in all these studies, prior use of hormonal therapy (postmenopausal or contraceptives) may have resulted in a difference in the observed outcome. Although, statistical methods are used to adjust for these variations, one should not ignore the potential role of bias in the results given the multitude of underlying factors.
Study (pub year; ref.) . | Risk . | Product . | Mean age . | Colorectal cancer . | HR (95% CI; P) . |
---|---|---|---|---|---|
French E3N prospective cohort (2012; ref. 52) | I | E | 51.1 | 525 | 0.72 (0.56–0.94) [S] |
French E3N prospective cohort (2012; ref. 52) | I | E + P | 56 | 1 (0.83–1.21) | |
Cancer Prevention Study II Nutrition Cohort (2009; ref. 53) | I | E | 61&66 | 776 | 0.76 (0.59–0.97; 0.01) |
Cancer Prevention Study II Nutrition Cohort (2009; ref. 53) | I | E + P | 57&59 | 0.84 (0.54–1.30; 0.72) | |
UK General Practice Research Database (GPRD; 2007; ref. 56) | I | E | 60.5 | 1.18 (0.72–1.92; 0.77) | |
UK General Practice Research Database (GPRD; 2007; ref. 54) | I | E + P | 60.5 | 0.56 (0.35–0.87) | |
European Prospective Investigation into Cancer and Nutrition (57) | I | HRT | 49.7 | 1,186 | 1 (0.86–1.16; 0.06) |
Nested case control (2012; ref. 55) | I | HRT | 73.8 | 4,708 | 0.81 (0.73–0.91; 0.001) |
Cancer Prevention Study II (1995; ref. 58) | M | HRT | 65.8 | 897 | 0.71 (0.63–0.81; 0.001) |
Utah, California, and Minnesota Cancer Registry Data (59) | M | HRT | 0.4 | ||
Seattle HMO database (60) | M | HRT | 699 | 0.41 | |
Nurses Health Study (61) | M | HRT | 63 | 834 | 0.64 |
Study (pub year; ref.) . | Risk . | Product . | Mean age . | Colorectal cancer . | HR (95% CI; P) . |
---|---|---|---|---|---|
French E3N prospective cohort (2012; ref. 52) | I | E | 51.1 | 525 | 0.72 (0.56–0.94) [S] |
French E3N prospective cohort (2012; ref. 52) | I | E + P | 56 | 1 (0.83–1.21) | |
Cancer Prevention Study II Nutrition Cohort (2009; ref. 53) | I | E | 61&66 | 776 | 0.76 (0.59–0.97; 0.01) |
Cancer Prevention Study II Nutrition Cohort (2009; ref. 53) | I | E + P | 57&59 | 0.84 (0.54–1.30; 0.72) | |
UK General Practice Research Database (GPRD; 2007; ref. 56) | I | E | 60.5 | 1.18 (0.72–1.92; 0.77) | |
UK General Practice Research Database (GPRD; 2007; ref. 54) | I | E + P | 60.5 | 0.56 (0.35–0.87) | |
European Prospective Investigation into Cancer and Nutrition (57) | I | HRT | 49.7 | 1,186 | 1 (0.86–1.16; 0.06) |
Nested case control (2012; ref. 55) | I | HRT | 73.8 | 4,708 | 0.81 (0.73–0.91; 0.001) |
Cancer Prevention Study II (1995; ref. 58) | M | HRT | 65.8 | 897 | 0.71 (0.63–0.81; 0.001) |
Utah, California, and Minnesota Cancer Registry Data (59) | M | HRT | 0.4 | ||
Seattle HMO database (60) | M | HRT | 699 | 0.41 | |
Nurses Health Study (61) | M | HRT | 63 | 834 | 0.64 |
Abbreviations: E, estrogen; I, incidence; M, mortality; P, progesterone; [S], significant.
In addition, four studies evaluated the role of hormones in mortality from colon cancer in patients who were diagnosed with colon cancer. All of these studies reported that HRT reduces the risk of death from colon cancer (Table 1; refs. 58–61). However, Newcomb and colleagues evaluated the risk of death in patients with colon cancer who were users of HRT and found that HRT does not reduce the risk of death from colon cancer (HR, 1.09; 95% CI, 0.81–1.47; ref. 62). These results further support a role for estrogen in reducing the mortality of colon cancer.
Clinical–Translational Advances
Although the estrogen pathway is not the central pathway in colorectal cancer, it plays an important role in the initiation and progression of the disease. Despite expression of ERβ in normal colonocytes, none of the existing colon cancer cell lines express a sufficient amount of ERβ, thus suggesting that loss of expression of ERβ is part of the tumorigenesis process. This review establishes that the ERs play a major role in the fate of the colonocytes and ligands maintain the receptors' balance.
Data from the WHI and observational cohorts collectively support that HRT, estrogen surrogate, reduces the risk of developing colon cancer. Nevertheless, the degree of risk reduction differs among individuals, likely based on the activated pathways leading to colorectal cancer. In the WHI study colon cancer in the hormone replacement group seemed to be more advanced and had a higher number of positive lymph nodes, suggesting that tumors developing while on HRT seem to be biologically more aggressive (63). Interestingly, in the observational studies, obesity and smoking were associated with lesser degree of benefit from HRT. These phenotypic features are linked to aberrant activation of the pathways that are independent from the estrogen pathway and thus HRT may be less effective in preventing the tumors driven by these ER-independent pathways. For example, obesity results in hyperinsulinemia, which in turn results in PI3K activation and increased risk of colorectal cancer (64). As shown in Fig. 1, there is no relationship between PI3K and ERβ (Fig. 1), and therefore estrogen may not protect against PI3K-driven tumors. In addition, smoking is linked to increased risk of MSI-high tumors and HRT does not decrease the risk of MSI-high tumors, which emphasizes that HRT may be effective in molecularly defined populations (65, 66).
In the WHI study, women older than 60 years on HRT had a lesser degree of colorectal cancer risk reduction, suggesting that timing of start of HRT (shorter interval from menopause) may play a crucial role in its protective effects on colon cancer. On the basis of the existing preclinical evidence, early start of HRT prevents changes in the ratio of the receptors and results in a lower rate of colorectal cancer through maintaining ERβ receptors. Furthermore, it is likely that start of estrogen before activation of the aberrant pathways will prevent colorectal cancer, while if the precancerous process is already initiated and ERβ is lost, estrogens will not be able to reverse the process.
Besides HRT, high estrogen content of soy is implicated to be protective against colorectal cancer. However, recent evidence suggests that higher soy consumption is protective against colon cancer only in women (43).
To maximize the benefit of estrogen for prevention of colorectal cancer, exploring the phenotypic and molecular subgroups of the individuals who achieve maximum benefit from this intervention is essential. Until then, use of estrogens for prevention of colon cancer should remain investigational.
Translational Relevance
Use of estrogens for prevention of colon cancer is an attractive concept in women; however, the increased rates of cardiovascular events with HRT limit use of these agents in clinical practice.
While there are no clinical data to prove that ERβ modulation reduces the incidence of colon cancer, preclinical models of ERβ modulation are promising. In the study by Schleipen and colleagues, treatment of ovariectomized rats with ERβ agonists or genistein increased the apoptosis and reduced proliferation in the ileum and colon, further supporting the role of these agents in the prevention of colon cancer (67). Also, Giroux and colleagues showed that ERβ agonist treatment reduces the number of small-intestinal polyps in ApcMin/+ male and female mice (68). Although development of selective ERβ modulators has proven to be very challenging, two agents are currently in phase III and IV trials. MF101 and raloxifen are being studied for their effects on menopausal symptoms and osteoporosis, respectively (69, 70). Long-term follow-up will show if any of the agents in clinical development have an effect on the incidence of colon cancer in women. The challenge is that benefit from these agents happens primarily in the presence of ERβ; thus, secondary prevention in patients who express ERβ is the most logical first step to test the clinical efficacy of these agents.
We conclude by stating that although estrogen modulation has a role in colorectal cancer prevention, the burden is on investigators to identify the appropriate population and product to move the field forward.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: A. Barzi, A.M. Lenz, M.J. Labonte, H.-J. Lenz
Development of methodology: A.M. Lenz, H.-J. Lenz
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Barzi, A.M. Lenz, H.-J. Lenz
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.M. Lenz, H.-J. Lenz
Writing, review, and/or revision of the manuscript: A. Barzi, A.M. Lenz, M.J. Labonte, H.-J. Lenz
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.M. Lenz, M.J. Labonte, H.-J. Lenz
Study supervision: H.-J. Lenz
Grant Support
The work was supported by NIH support, P30, and Wunderglo Foundation.