Two-sided role of estrogen on endometrial carcinogenesis: stimulator or suppressor?

Abstract

Endometrial carcinoma (EC) often expresses estrogen receptors (ER), and the growth of EC is stimulated by estrogen. Therefore, EC is considered to be an estrogen-dependent tumor. However, the role of estrogen in endometrial carcinogenesis is somewhat unclear because the majority of EC occurs at peri- or post menopause when serum estrogen levels are generally decreased. In this article, we describe the double-edged role of estrogen in the genesis of EC, especially in terms of mismatch repair functions in vitro and in vivo, i.e. when serum estradiol (E2) levels are relatively low (approximately less than 90 pg/ml), and E2 enhance the carcinogenesis, whereas high E2 levels may suppress the carcinogenesis. This will deepen mechanistic insight into unopposed estrogen.

摘要

子宫内膜癌(EC)常表达雌激素受体(ER), 其生长受雌激素刺激。因此, EC被认为是一种雌激素依赖性肿瘤。然而, 雌激素在正常子宫内膜癌变过程中的作用尚不清楚, 因为大多数EC发生在绝经前后, 此时血清雌激素水平普遍下降。在本文中,我们描述了雌激素在EC发生过程中的双刃剑作用, 特别是在体外和体内的错配修复功能方面,即当血清雌二醇(E2)水平相对较低(大约不到90 pg / ml)时, E2有致癌作用, 而高E2水平可能抑制其致癌作用。这将加深对雌激素致癌机制的认识。

The Chinese abstracts are translated by Prof. Dr. Xiangyan Ruan and her team: Beijing Obstetrics and Gynecology Hospital, Capital Medical University, Beijing 100026, China.

Keywords:

• Endometrial carcinoma
• estradiol
• mismatch repair
• next-generation sequencing
• carcinogenesis

Introduction

The number of endometrial carcinoma (EC) patients is steadily increasing worldwide [1]. Among diverse tumors, steroid hormone dependency, i.e. estrogen-induced proliferation and progesterone-mediated growth suppression, is the hallmark of EC [2]. Therefore, the pathogenetic process of this particular malignancy is thought to be affected by sex steroid hormones. In this article, we summarize reports on circulating estrogen levels and genetic changes in EC patients, including those detected using recent next-generation sequencing. We then describe the estrogen-induced growth mechanisms of endometrial cells and the specific role of estrogen in the pathogenesis of EC in terms of mismatch repair (MMR) activity. Accordingly, we propose a double-edged role for estrogen in endometrial carcinogenesis, and clarify the significance of unopposed estrogen, i.e. low and chronic estrogen, as a well-known causative factor of EC.

Serum estrogen levels in a woman’s life

Because estradiol (E2) is the most potent estrogen among the three estrogens (estrone, estradiol, and estriol) in terms of the human endometrium, we focus on the roles of E2 in this article. Serum levels of E2 fluctuate drastically during the menstrual cycle, as well as during a woman’s entire life [3]. In reproductive age women with a normal menstrual cycle, E2 levels range from ∼50 to 300 pg/ml during the menstrual cycle. Over 200 pg/ml of E2 is needed to induce an LH surge and ovulation. During peri- and early post-menopause, E2 levels decrease to less than 80 pg/ml and become lower than 10 pg/ml in late postmenopause. In this article, a serum E2 level of more than ∼90 pg/ml is tentatively classed as ‘normal-high E2’, which is often seen in reproductive age women with regular menstruation (Figure 1). The level which is lower than 10 pg/ml is classed as ‘low E2’, which is mostly seen in late postmenopausal women or those older than 70 years of age. The intermediate area, i.e. E2 levels ranging from ∼10 pg/ml to 90 pg/ml is classed as ‘relatively-low E2’. This value is often seen in peri- or early postmenopausal women, and in reproductive age women with irregular menstruation. Actually, the majority of EC patients are included in this ‘relatively-low E2’ classification [4–9], and only a limited number of EC patients had serum levels E2 higher than 80 pg/ml [4], possibly due to follicle persistence.

Figure 1. Terminology of estradiol (E2) concentrations used in this article, and respective populations, menstrual status, and proliferative activity of the normal human endometrium. The majority of endometrial carcinoma (EC) patients are in the ‘Relatively-low’ E2 range. OHSS: ovarian hyperstimulation syndrome, LH: luteinizing hormone.

The EC ‘paradox’

Malignant transformation tends to occur in actively proliferating cells, since DNA replication errors accumulate along with DNA duplication and mitosis [10]. Normal endometrial glandular (NEG) cells and a subset of EC cells express estrogen receptor (ER) and their proliferation is stimulated by estrogen [11]. In this article, carcinogenesis of E2 is defined as the capability of producing new cancer cells from (i) actively replicating cells lacking in DNA repair system, and (ii) E2 metabolites which can lead mutation. Thus, physiologically high levels of serum estrogen, i.e. normal-high E2 levels, as often seen in women with normal menstrual cycles, has been considered to stimulate endometrial carcinogenesis. However, ∼80% of ECs arise in peri- or post-menopausal women, whose E2 levels are in the relatively-low range [12]. Serum E2 levels are also often relatively-low in women receiving estrogen replacement therapy (ERT), a well-known risk factor for EC [12]. But, some women with ERT showed high E2 status, suggesting the E2 levels to be drug dose-dependent [13,14]. Moreover, polycystic ovary syndrome (PCOS) is a representative risk factor in young women [15], and PCOS patients often show slightly lower serum E2 levels than healthy women [16], and their E2 concentrations are at relatively-low E2 levels. But a study showed slightly higher E2 levels in PCOS patients than controls [17]. It also should be noted the risk of EC may be dependent not only on E2 but also hyperandrogenemia in PCOS [18]. In this regard, we consider these clinical observations as a ‘paradox’ of EC. Thus, the actual role of estrogen on endometrial carcinogenesis has not been fully elucidated. The present review focuses on explaining this contradiction.

Serum estrogen levels in EC patients

Several studies have reported the serum levels of E2 in women with EC (Supplementary Figure 1) [4–9]. In postmenopausal women, although most patients showed relatively-low E2 concentrations, the average serum E2 levels in EC patients were slightly higher than those of normal control women in most studies. These results indicate that cumulative exposure of E2 for long period has a critical role in endometrial carcinogenesis. One study reported that serum E2 levels in premenopausal EC patients were lower than in healthy young women [6]. These reports suggest that low and chronic E2 levels are advantageous for endometrial carcinogenesis.

Clinicopathological and genetic characteristics of EC: two types of ECs

This type-1/2 classification is not always applicable to all EC cases; however, it has been widely accepted in recent decades [19]. Briefly, ∼80% of EC belongs to type-1 carcinoma, arising with a background of unopposed estrogen. It is frequently associated with atypical endometrial hyperplasia (AEH) and expresses ER and progesterone receptors (PR). The major cause is persistent exposure to estrogen due to perimenopause, obesity, PCOS, or ERT. Histologically, most type-1 tumors are endometrioid carcinoma with low histological grade characterized by favorable biological behavior.

In contrast, ∼20% of ECs are classified as type-2, which develop via an estrogen-unrelated pathway with from atrophic endometrium [19]. These tumors usually occur in older women, about 5–10 years older than those with type-1 carcinoma. Type-2 carcinomas are typically high-grade tumors with non-endometrioid histology, such as serous and clear cell tumors. The expression of ER and PR is usually negative, and serum levels of estrogen are very low. Both serous and clear cell carcinomas have a poor patient prognosis.

Genetic alteration of EC

Typical genetic changes are summarized in Supplementary Table. Notably, several important mutations in the early stage carcinogenesis, such as PTEN [20], KRAS [21], and CTNNB1 [22], are frequently found in microsatellite instability (MSI)-positive EC. MSI is defined by the accumulation of insertions and/or deletions at short DNA repeats (microsatellites), leading to different repeat lengths, which is caused by impaired MMR protein expressions such as for MLH1 and MSH2 due to the mutation or methylation of MMR-related genes [23]. Moreover, even in the normal endometrium, a lack of MLH1expression caused by promoter methylation has been reported [24]. These findings suggest that reduced MMR activity is closely involved in early endometrial carcinogenesis.

Lynch syndrome (LS)

LS, a hereditary non-polyposis colorectal cancer (HNPCC), is characterized by the development of colorectal cancer, EC, and various other cancers and is caused by a germline mutation in one of the MMR genes MSH2, MLH1, MSH6 or PMS2 [25]. The cumulative lifetime risk of EC in LS is estimated to be between 20 and 70%, depending on the type of MMR mutation [25]. The LS indicates the MMR abnormality is a pivotal event in endometrial carcinogenesis.

The cancer genome atlas (TCGA)

TCGA study newly revealed four genomic subtypes; POLE-ultramutated (favorable outcome), MSI, copy number low (intermediate risk/outcome), and copy number high (the worst outcome) [26]. These categories roughly corresponded to the classical dualistic model, because 97% of POLE-ultramutated, MSI and copy number low were type-1, and 95% of copy number high were serous tumors. Importantly, the TCGA data also revealed that MSI was found in the POLE-ultramutated and MSI groups which account for ∼40% of all cases, supporting the importance of MSI in endometrial carcinogenesis.

Genetic alteration of AEH

Since AEH is considered to be a precursor of type-1 EC [19], Supplementary Table summarizes the mutations in AEH such as PTEN, PIK3CA, KRAS, and MSI which were reported with a lower percentage than those in EC. These alterations seem to be important in the early stage of endometrial carcinogenesis. However, the EC and AEH tissues samples examined in these analyses were obtained from different patients. Therefore, the pathogenic significance of each gene alteration remains unclear. To address this issue, we performed whole exome sequencing in the AEH and EC tissues obtained from one patient, and directly compared the mutations in AEH and EC. The result indicated that our case already had about 4000 mutations in AEH and 5700 mutations in EC, with 2200 common mutations in both tissues [27]. Of note, this case had a POLE mutation as early as in AEH, so this was regarded as an ultramutated-type. In addition, genetic alterations like MMR genes, PTEN, PIK3CA were observed in both AEH and EC, whereas that of ARID1A was noted only in EC [27]. These results indicate that the mutation-prone environment evoked by mutations in the POLE and MMR genes associated with the activated PI3K pathway played a pivotal role in this case, suggesting that the functional significance of each genetic alteration may be different in the carcinogenetic process.

Estrogen and endometrial carcinogenesis in vitro

Carcinogenetic effect of estrogen

Malignant transformation tends to occur in actively proliferating cells because DNA replication errors accumulate during DNA duplication and mitosis [10]. Therefore, E2 is potentially carcinogenic to NEG cells. Although E2 itself is not carcinogenetic, catechol estrogen such as 4-hydroxyestradiol, a metabolite of E2, is reportedly genotoxic and carcinogenic in animal models [28]. We reported that 4-hydroxyestradiol induces DNA damage on codon 130/131 of PTEN in EC cells [29]. However, actual tissue concentrations of these metabolites have not been measured in normal and carcinomatous endometria, and the levels of estrogen metabolites are shown to be generally low when the parent E2 level is low [30]. It should be noted that the dependency of carcinogenicity on E2 metabolism is complex, with follow up reactions within the cells. Therefore, E2 levels are sometimes less representative for this risk.

Anti-carcinogenetic effect of estrogen

Because MMR deficiency is an important factor in the early stages of type-1 endometrial carcinogenesis, we examined the effect of estrogen on MMR protein expression and functions [31]. Consequently, we observed the following findings (Figure 2):

Figure 2. Schematic demonstration of the E2 concentration, proliferation of normal endometrial glandular cells, mismatch repair protein expression, and cancer risk. Cells under a low E2 status show the weak MMR protein expression and no proliferating activity. Cells under a high E2 status show the strong MMR protein expression and the high proliferating activity, indicated by the darkest gray nuclei. Cells under a relatively-low E2 concentration show a few proliferating cells with the weak MMR protein expression, indicated by darker gray nuclei.

1. The expression of MMR proteins (MLH1 and MSH2), and actual in vitro MMR function were increased by E2 in NEG cells.

2. Immunohisto/cytochemical studies revealed that

   1. When the E2 concentration was high, the NEG cells had strong proliferative activity, but had strong MMR functions, suggesting that active proliferation-related carcinogenesis is protected.

   2. When E2 was very low, although the expression of MMR proteins was negative, the NEG cells did not proliferate, indicating low cancer risk.

   3. When E2 was relatively-low, the expression of MMR was almost negative but NEG cells still had considerable growth activity, implying that the potential cancer risk of NEG cells is not counteracted by MMR proteins, and increases cancer risk.

Collectively, our data indicated an increased cancer risk at relatively-low E2 levels, and we tentatively termed this E2 concentration range the ‘Cancer window’ (Figure 3). When E2 concentrations are within this range, it is an important risk factor, and this explains the risk of unopposed, i.e. low-chronic estrogen, in endometrial carcinogenesis. This also suggests that short-term exposure of E2 may potentially protect the endometrial carcinogenesis.

Figure 3. Graphic summary of E2 concentration-dependent proliferation, MMR activity and risk of EC.

Estrogen and endometrial carcinogenesis in animal models

Estrogen as a stimulator

Several studies demonstrated the carcinogenic effects of estrogen. The frequency of a potent carcinogen, N-methyl-N-nitrosourea (MNU),-induced EC was increased by the oral intake of E2 [32]. The number of EC lesions in Pten^−/+ mice was reduced by the administration of a pure ERα antagonist, suggesting a positive role of estrogen in tumorigenesis [33]. The knockout of mitogen-inducible gene-6 (Mig6), a pivotal downstream molecule in the progestin-induced growth suppression of EC cells, resulted in the development of EC following the addition of E2 [34]. These studies indicated the positive role of estrogen in endometrial carcinogenesis.

Estrogen as a potential suppressor

There are also several studies suggesting the protective role of estrogen on endometrial carcinogenesis. A study showed that Pten^−/+ mice developed EC, whereas the simultaneous knockout of ERα resulted in a higher incidence of EC, suggesting that endometrial carcinogenesis is independent of estrogen and ERα status [35]. Moreover, another report demonstrated that in the endometria Pten conditional knockout mice using the Cre-loxP system developed EC; however, the tumor formation was suppressed by the addition of E2 [36]. Interestingly, high dosage of E2 metabolite, 2-methoxyestradiol, suppress the growth of breast cancer cells [37], but such reports are lacking in EC.

This discrepancy may be due in part to the methodological differences. However, there were two unresolved problems in these studies. Firstly, the genetically engineered mice used in the above-mentioned studies may have influenced innate MMR systems, potentially disturbing the possible relationship between estrogen and MMR. Secondly, although the physiological serum levels of E2 in mice reportedly range between 1 and 10 pg/ml [38], none of those animal studies examined the concentration of E2 in each mouse used. Therefore, we do not know whether the E2 concentration of these model mice precisely reflects the human serum E2 levels. Based on these considerations, we examined the effect of various concentrations of E2 on endometrial carcinogenesis using mice with an intra-uterine injection of MNU. The result indicated that the mean E2 concentrations of mice with atrophic, normal, hyperplastic, and carcinomatous endometria were 0.2, 3.8, 190.0, and 6.7 pg/ml, respectively, with significant differences (Supplementary Figure 2) [39]. Our study indicated that EC often occurs at intermediate E2 levels, and elevated E2 somewhat suppressed the development of EC, but led to endometrial hyperplasia. We consider this result reflects the E2 concentration-dependent vulnerability of human EC.

EC in young women and estrogen

Approximately 20% of EC occurs at a relatively younger (less than 50 years old) generation. Among these young EC patients, about 50% of premenopausal EC patients had irregular menstruation, and irregular menstruation or amenorrhea increases the risk of EC in young women [40]. These anovulatory women often show lower serum E2 levels than that with regular menstruation because the continuous elevation of E2, i.e. more than 200 pg/ml, is needed to evoke LH surge and ovulation [3]. Therefore, young women with irregular menstruation suffer from unopposed estrogen status, i.e. low and chronic estrogen status such as seen in PCOS, are thus regarded to be at high risk of EC.

However, about 50% of young EC patients have normal ovulation and normal menstrual cycle, with normal-high serum E2 levels. The pathogenic mechanism of EC patients with normal menstruation is still unclear. The increased silencing promoter methylation of the hMLH1 gene associated with MSI was reported in EC [41], and a study showed frequent loss of the DNA MMR protein in non-obese young EC patients with undifferentiated histology, comprising a distinct clinico-pathological subset [42]. We have observed the reduced expression of MIG6, which is an important progestin-induced tumor suppressor [34] in young EC patients with normal menstruation (unpublished data).

The significance of estrogen in human reproduction and EC prevention

In order to become pregnant, young women must have elevated estrogen levels to thicken the endometrium and induce ovulation. However, if the elevated circulating estrogen levels are left unattended, the risk of EC is likely to increase because of the accumulation of proliferation-related DNA errors. However, young women are protected from potential EC risk of estrogen by the ‘double block’ system; one is estrogen-induced activation of DNA MMR function, and another is estrogen-induced ovulation and subsequent progesterone secretion. In the peri- or postmenopausal women, this double block system does not work because of the unopposed estrogen status by relatively-low E2 level and anovulation.

Prevention of EC in endocrinological aspect

For postmenopausal women, as long as E2 is within a relatively low level, E2 is an important risk factor. Therefore, drug or factors which increase circulating E2 should be avoided. Although the potential risk of exogenous E2 for EC may be different from endogenous E2, ERT apparently increases serum E2 levels, so, concurrent use of progestin is recommended [12]. Obesity is an important risk factor for EC [12]. Although the mechanisms are not fully understood, obese women tend to have elevated serum E2 levels because the production of E2 from androgen through aromatase in fat tissue may be increased [43]. Although the effect is still not conclusive, the benefits of physical exercises and body weight reduction in the prevention of EC are suggested [44]. In reproductive-age women, anovulatory status leading to unopposed estrogen status should be avoided using appropriate drugs or ovulation induction.

Conclusion

A clear understanding of the hormonal characteristics of EC in the pathogenesis and growth mechanisms is necessary for the adequate management of this tumor type. Unopposed estrogen has been considered to be involved in endometrial carcinogenesis, however, our studies showed that elevated E2 levels may rather suppress the carcinogenesis possibly by the up-regulation of MMR activity, suggesting a two-sided role of estrogen in endometrial carcinogenesis. Further studies are needed to clarify the molecular mechanism of prolonged exposure of low E2 in endometrial carcinogenesis, and the genetic background in young EC patients with normal menstruation. In addition, the possibility of EC as a preventable neoplasm should be explored.

Disclosure statement

No potential conflict of interest was reported by the authors.