Menopausal Hormone Therapy and Risk of Endometrial Cancer: A Systematic Review

Background: Menopausal hormone therapy (MHT) is an appropriate treatment for women with the climacteric syndrome. The estrogen component of MHT effectively alleviates climacteric symptoms but also stimulates the endometrium and thus may increase the risk of endometrial cancer (EC). Materials and Methods: We performed a systematic literature search of the databases PubMed and Cochrane Central Register of Controlled Trials to identify controlled and uncontrolled clinical trials reporting on the prevalence and/or incidence of EC among women using MHT. Results: 31 publications reporting on 21,306 women with EC diagnosed during or after MHT were identified. A significantly reduced risk of EC among continuous-combined (cc)MHT users with synthetic progestins (SPs) was demonstrated in 10/19 studies with odds ratios (ORs)/hazard ratios (HRs) between 0.24 and 0.71. Only one study documented an increased risk of EC among long-term users (≥10 years), not confirmed in three other sub-group analyses of women with ≥6, ≥5, and >10 years of ccMHT use. A significantly increased risk of EC among users of sequential-combined (sc)MHT with SPs was demonstrated in 6/12 studies with ORs/HRs between 1.38 and 4.35. Number of days of progestin per month was a significant modulator of EC risk. A decreased risk of EC was seen in obese women. Two studies documented an increased risk of EC among users of cc/scMHT with micronized progesterone. A significantly increased risk of EC among estrogen-only MHT users was demonstrated in 9/12 studies with ORs/HRs between 1.45 and 4.46. The adverse effect of estrogen-only MHT was greatest among obese women. Conclusion: ccMHT with SPs reduces the risk of EC, whereas estrogen-only MHT increases the risk. scMHT with SPs and cc/scMHT with micronized progesterone increase the risk of EC depending on type of progestin, progestin dosage, and duration of MHT use.


Introduction
Endometrial cancer (EC) is a malignancy derived from the epithelial lining of the uterine cavity. It is the most common female pelvic malignancy with an estimated life-time incidence of 4% [1]. Typically, older women are affected with a mean age at the time of diagnosis of 69 years [2]. EC generally has a good prognosis and therefore accounts for only 2% of cancer-related deaths in females despite its high incidence and prevalence. The reason for the favorable outcome of most

Materials and Methods
We performed a systematic literature search of the databases PubMed and Cochrane Central Register of Controlled Trials using the following search terms: "hormone replacement therapy" MeSH Terms OR "hormone" All Fields AND "replacement" All Fields AND "therapy" All Fields OR "hormone replacement therapy" All Fields AND "endometrial neoplasms" MeSH Terms OR "endometrial" All Fields AND "neoplasms" All Fields OR "endometrial neoplasms" All Fields OR "endometrial" All Fields AND "cancer" All Fields OR "endometrial cancer" All Fields (search date: 2020-06-20). The methodology followed the PRISMA criteria. The PICO question was as follows: Does a hormone therapy with estrogens and/or estrogens combined with progestins increase the risk of endometrial cancer in peri-or postmenopausal women? Screening, eligibility, and data analysis were performed by two authors independently (CBT and GAR). Discrepancies were solved by consensus. Study investigators were not contacted to obtain further information. The main outcome of interest was the risk of EC in relation to different forms of peri-and postmenopausal MHT. Bias of individual studies and grading of the strength of the evidence were not assessed. With the above described search strategy, 1359 citations were identified. Therefore, the search was restricted to the last 20 years, i.e., January 2000 until May 2020 yielding 883 citations. After screening all abstracts, 29 appropriate citations were selected reporting on the incidence and prevalence of EC among women undergoing MHT. MHT was defined for the purpose of this systematic review as systemic (oral, transdermal or vaginal) use of estradiol or conjugated equine estrogens or estradiol or conjugated equine estrogens combined with a synthetic progestin (norethisterone acetate, medroxyprogesterone acetate, megestrol acetate, chlomadinone acetate, medrogestone, levonorgestrel, cyproterone acetate, drospirenone, dienogest) or combined with progesterone (natural progesterone or dydrogesterone) in a continuous combined therapy scheme (ccMHT) or in a sequentially-combined therapy scheme (scMHT). Studies not reporting individual patient data and studies containing no extractable clinical data were excluded. We also excluded studies reporting on MHT with dehydroepiandrosterone, tibolone, or androgens with or without estrogens (n = 854). All 29 appropriate citations were then retrieved in full and cross reference searching was performed identifying two further citations for a total of 31 appropriate citations reporting on the incidence and prevalence of EC among women using MHT published between January 2000 and May 2020. Figure 1 shows a flow diagram of the literature search algorithm. Data were extracted and analyzed in a descriptive manner. Meta-analysis was not performed due to the heterogeneity of studies.

Materials and Methods
We performed a systematic literature search of the databases PubMed and Cochrane Central Register of Controlled Trials using the following search terms: "hormone replacement therapy" MeSH Terms OR "hormone" All Fields AND "replacement" All Fields AND "therapy" All Fields OR "hormone replacement therapy" All Fields AND "endometrial neoplasms" MeSH Terms OR "endometrial" All Fields AND "neoplasms" All Fields OR "endometrial neoplasms" All Fields OR "endometrial" All Fields AND "cancer" All Fields OR "endometrial cancer" All Fields (search date: 2020-06-20). The methodology followed the PRISMA criteria. The PICO question was as follows: Does a hormone therapy with estrogens and/or estrogens combined with progestins increase the risk of endometrial cancer in peri-or postmenopausal women? Screening, eligibility, and data analysis were performed by two authors independently (CBT and GAR). Discrepancies were solved by consensus. Study investigators were not contacted to obtain further information. The main outcome of interest was the risk of EC in relation to different forms of peri-and postmenopausal MHT. Bias of individual studies and grading of the strength of the evidence were not assessed. With the above described search strategy, 1359 citations were identified. Therefore, the search was restricted to the last 20 years, i.e., January 2000 until May 2020 yielding 883 citations. After screening all abstracts, 29 appropriate citations were selected reporting on the incidence and prevalence of EC among women undergoing MHT. MHT was defined for the purpose of this systematic review as systemic (oral, transdermal or vaginal) use of estradiol or conjugated equine estrogens or estradiol or conjugated equine estrogens combined with a synthetic progestin (norethisterone acetate, medroxyprogesterone acetate, megestrol acetate, chlomadinone acetate, medrogestone, levonorgestrel, cyproterone acetate, drospirenone, dienogest) or combined with progesterone (natural progesterone or dydrogesterone) in a continuous combined therapy scheme (ccMHT) or in a sequentially-combined therapy scheme (scMHT). Studies not reporting individual patient data and studies containing no extractable clinical data were excluded. We also excluded studies reporting on MHT with dehydroepiandrosterone, tibolone, or androgens with or without estrogens (n = 854). All 29 appropriate citations were then retrieved in full and cross reference searching was performed identifying two further citations for a total of 31 appropriate citations reporting on the incidence and prevalence of EC among women using MHT published between January 2000 and May 2020. Figure 1 shows a flow diagram of the literature search algorithm. Data were extracted and analyzed in a descriptive manner. Meta-analysis was not performed due to the heterogeneity of studies.
The clinical characteristics of the subset of 20 studies reporting individual patient data on EC among women using various forms of MHT are shown in Tables 1-4. Specifically, the clinical characteristics of individual studies reporting on ccMHT with synthetic progestins and EC risk are shown in Table 1. The clinical characteristics of individual studies reporting on scMHT with synthetic progestins and EC risk are shown in Table 2. Table 3 shows the clinical characteristics of individual studies reporting on ccMHT and scMHT with progesterone or dydrogesterone and EC risk, and Table 4 shows the clinical characteristics of individual studies reporting on MHT with estrogens in women with an intact uterus and EC risk.

ccMHT with Synthetic Progestins Reduces The Risk of EC
The clinical characteristics of individual studies reporting on ccMHT with synthetic progestins and EC risk are shown in Table 1. In 8 case-control studies (CCS), 9 cohort studies (CS), and 2 randomized controlled trials (RCTs), 10,265 women with EC and 1,969,172 controls were analyzed. A significantly reduced risk of EC among ccMHT users with synthetic progestins was demonstrated in 9/19 studies with ORs/HRs between 0.24 [21] and 0.71 [30]. In the remaining studies, neither a risk reduction nor a risk increase was observed. Only one study documented an increased risk of EC in the sub-group of long-term users (≥10 years) of ccMHT with synthetic progestins [7], which was not confirmed in three other sub-group analyses of long-term users with ≥6 years of use [26], ≥5 years of use [27], and >10 years of use [19].

scMHT with Synthetic Progestins May Increase The Risk of EC
The clinical characteristics of individual studies reporting on scMHT with synthetic progestins and EC risk are shown in Table 2. In 6 CCS and 6 CS, 9663 women with EC and 1,843,377 controls were analyzed. A significantly increased risk of EC among scMHT users with synthetic progestins was demonstrated in 6/12 studies with ORs/HRs between 1.38 [19] and 4.35 [7]. In the remaining six studies, no effect on EC risk was observed. Only one study documented a decreased risk of EC in the sub-group of short-term users (<5 years) of ccMHT with synthetic progestins [19]. A beneficial effect of scMHT with synthetic progestins was also seen in the subgroup of obese women [30]. Number of days per month of progestin use was also a modulator of EC risk increase with a higher risk in case of <10 days/month [7,26,33].

ccMHT and scMHT with Progesterone or Dydrogesterone Increases The Risk of EC
The clinical characteristics of individual studies reporting on ccMHT and scMHT with progesterone or dydrogesterone and EC risk are shown in Table 3. Only two CS were identified [8,21] with 902 cases and 180,202 controls. Both studies documented an increased risk of EC among users of micronized progesterone (MP). Short-term use (≤5 years) of MP and ever use of dydrogesterone and progesterone derivatives were not associated with an increased risk of EC.

Estrogen-only MHT in Women with an Intact Uterus Increases the Risk of EC
The clinical characteristics of individual studies reporting on estrogen-only MHT in women with an intact uterus and EC risk are shown in Table 4. In 4 CCS and 8 CS, 10,241 women with EC and 1,952,004 controls were analyzed. A significantly increased risk of EC among estrogen-only MHT users was demonstrated in 9/12 studies with ORs/HRs between 1.45 [30] and 4.46 [7]. In the remaining three studies, no effect on EC risk was observed. No study documented a decreased risk of EC. The adverse effect of estrogen-only MHT was greatest in the subgroup of obese women [30]. Both past and current use and type of estradiol (estradiol, conjugated estrogens, non-conjugated estrogens) increased EC risk.

Narrative Reviews on MHT and EC Risk
In a narrative review of the literature, Brinton and Felix stated that in most studies including women who had ever used ccMHT (>25 days/months) EC risk was reduced relative to non-users (meta-analysis relative risk (RR) 0.78; 95% confidence interval (CI) 0.72-0.86 based on observational studies) [15]. The reduced risk was greatest among obese women. In contrast, women who had ever used scMHT with <10 days of progestins per month were at a significantly increased risk of EC with meta-analysis results showing an overall RR of 1.76 (95% CI 1.51-2.05). However, when progestins were given for 10-24 days/month, scMHT appeared unrelated to the risk of EC with an RR of 1.07 (95% CI 0.92-1.24). Sjögren et al. performed a systematic review of the literature and summarized 28 studies reporting on menopausal women with intact uteri treated with estrogen only or with estrogen plus progestin for a minimum of one year. [13]. They found that observational studies confirmed an increased risk among users of estrogen alone. Continuous combined therapy showed a lower risk than sequential combined therapy. The newer marketed micronized progesterone increased the risk notably, also when administered continuously.
A stepwise model of MHT and endometrial carcinogenesis with abnormal endometrial proliferation, atypical hyperplasia, and endometrial cancer was summarized by Horn et al. [31]. They report that about 15% of endometrial biopsies taken from women on scMHT show abnormal proliferative activity with atypical endometrial hyperplasia identified in 1% of cases. In contrast, endometrial biopsies from women using ccMHT typically demonstrate endometrial atrophy with only 2-3% of biopsies showing abnormal proliferative activity but without atypical hyperplasia. The risk of atypical hyperplasia or carcinoma under unopposed estrogen-mono MHT varies from 2 to 10%.

Body Mass Index, MHT, and EC Risk
In a meta-analysis of nine studies analyzing EC risk by body mass index (BMI) and MHT use, Crosbie et al. found that ever use of MHT reduced the risk of EC conferred by being overweight [22]. Specifically, the overall risk ratio of EC per 5 kg/m 2 increase in BMI was 1.60 (95% CI 1.52-1.68). Among never users of MHT, the overall risk ratio of EC per 5 kg/m 2 increase in BMI was 1.90 (95% CI 1.57-2.31) compared to only 1.18 (95% CI 1.06-1.31) among ever users of MHT. This analysis did not differentiate between ccMHT and scMHT. Similarly, in a cohort study of 103,882 postmenopausal women with 677 cases of EC, Chang et al. also found that MHT significantly modified the relations of BMI and EC risk [25]. The RR of EC for a base-line BMI ≥30 versus < 25 kg/m 2 was 5.41 (95% CI 4.01-7.29) among non-users of MHT and was markedly reduced to 2.53 (95% CI 1.21-5.30) among former users of MHT. In line with these findings, another large cohort study with 33,436 postmenopausal women and 318 EC cases confirmed that BMI was a strong predictor of EC risk (RR 4.41; 95% CI 2.7-7.2 for BMI ≥ 35 kg/m 2 ), but this association was no longer evident among ever users of cc/scMHT [24].

Influence of MHT on EC-Specific Mortality
In a sub-analysis of the National Institutes of Health-American Association of Retired Persons (NIH-AARP) Diet and Health Study, the effect of MHT on the 10-year all-cause EC-specific mortality was assessed based on 890 EC cases [14]. Interestingly, pre-diagnosis use of MHT had a significant and variable effect on mortality depending on MHT type. For example, pre-diagnosis use of ccMHT or scMHT was associated with a significantly lower 10-year all-cause mortality (HR 0.65, 95% CI 0.43-0.99) and EC-specific mortality (HR 0.51, 95% CI 0.26-0.98) compared with never users. In contrast, former estrogen-only MHT users had a significantly higher all-cause mortality (HR 1.71, 95% CI 1.02-2.88) and EC-specific mortality (HR 2.17, 95% CI 0.96-4.90). A case-control study from Sweden also investigated the influence of MHT on EC-specific mortality and found significantly improved survival rates for ever users of any form of MHT (relative survival ratio [RER] 0.40; 95% CI = 0.16-0.97), in particular ever users of any form of estrogens (RER 0.38; 95% CI = 0.15-0.99) [23].

Genetic Modulation of the Association between MHT and EC
EC risk conferred by MHT may be modulated by the carriage of polymorphisms in genes encoding enzymes involved in sex steroid metabolism. McKean-Cowdin et al. examined the association between EC risk and estrogen-mono MHT by CYP17 genotype using 51 incident cases and 391 randomly selected controls from a multiethnic cohort in Hawaii and Los Angeles, California [34]. The risk of EC was significantly higher for women homozygous for the CYP17 T allele (OR 4.10; 95% CI 1.64-10.3), but not for women with the C allele (OR 1.31; 95% CI 0.53-3.21), suggesting that CYP17 variants affecting biosynthesis and metabolism of estrogen are potential individual modulators of EC susceptibility due to MHT.
An additional genetic variant modulating the risk of EC among MHT users has been described by Razawi et al. [18]. They found that a variation in the CYP11A1 gene may modify the risk of postmenopausal EC conferred by estrogen-only MHT. Specifically, in a nested case-control study within the California Teachers Study, 286 EC cases and 488 controls were genotyped for polymorphisms in six sex steroid metabolism genes. The strongest interaction was observed between duration of estrogen-only MHT and haplotype 1A of CYP11A1 (p = 0.010). The OR for EC per copy of haplotype 1A was 2.00 (95% CI: 1.05-3.96) for long-term estrogen-only MHT users. All other genetic variants were not associated with any form of MHT after correction for multiple testing.

Circulating Estrogen Levels during MHT and EC Risk
In a nested case-control study among postmenopausal women using MHT at baseline in the Women's Health Initiative Observational Study (230 endometrial cancers, 253 controls), EC risk was unrelated to estrogen or estrogen metabolite levels among women who took ccMHT [10]. This is an interesting finding because it is counter-intuitive that circulating estrogens do not influence EC risk among women with MHT-induced high-estrogen levels.

MHT and Epidemiology of EC
In an interesting analysis of the Surveillance, Epidemiology, and End Results (SEER) data base, Constantine et al. noted an increase in EC incidence after the publication of the WHI Study coinciding with a decrease in ccMHT prescriptions in the US [9]. In detail, EC rates were constant from 1992 to 2002 but increased by 2.5% annually with a 10% increase from 2006 to 2012. Use of approved prescription ccMHT products decreased after the publication of the WHI data, whereas other risk factors either remained constant or decreased during the same time. Based on these observations, the authors concluded that the EC rate increase after the first publication of WHI data in 2002 may be associated with the decreased use of approved ccMHT. In accordance with these data, Wartko et al. confirmed that EC incidence rates increased in the US after 2002 [17]. In contrast to the constant EC rate pattern observed from 1992 to 2002 (annual percentage change: 0.0%), rates increased after 2002 in women 50-74 years old (annual percentage change: 2.5%) and in women 20-49 years old (annual percentage change: 2.1%). Post-2002 increases in incidence among women ages 50-74 were specific to Type I endometrial tumors.
ccMHT or scMHT is the standard of care for women with an intact uterus in order to avoid endometrial stimulation with subsequent endometrial hyperplasia and EC [4]. Our systematic review confirms that women with an intact uterus should not be treated with estrogen-only MHT. Independent of the woman's age, the endometrial epithelium will start to proliferate when estrogens without adequate progestin-opposition are being used and the risk of developing EC will be significantly increased as demonstrated in >10,000 women with EC. The risk of EC while using estrogen-only MHT is greatest in obese women, who therefore represent an especially vulnerable population. Our review also confirms that ccMHT with synthetic progestins is safe regarding endometrial proliferation and carcinogenesis. Substitution of synthetic progestins with natural progesterone such as MP, however, is problematic based on the data in the literature. Our analysis documents an increased risk of EC among users of MP. If MP is being used as part of ccMHT or scMHT, short-term use for ≤5 years is recommended. In addition, care should be taken that ccMHT is used instead of scMHT with an adequate dosage of at least 200 mg/day based on the results of the Postmenopausal Estrogen/Progestin Interventions (PEPI) trial [37].
The most controversial issue identified in our review is the potentially elevated risk of EC in women using scMHT. Our analysis based on >9000 EC cases shows that half of all studies found a significantly elevated risk of EC with ORs/HRs between 1.3 and 4.3. Number of days per month of progestin use was an important modulator of EC risk increase with a higher risk in case of <10 days/month. Therefore, women with climacteric symptoms seeking medical care should be counselled that scMHT is probably best avoided completely. There is no physiological or medical advantage of scMHT compared to ccMHT [3]. Thus, ccMHT, which clearly does not increase the risk of EC, is a reasonable alternative to scMHT avoiding the uncertainty of a potentially increased risk of EC.
The strongest effect size regarding an increase in EC risk among users of MHT with estrogens only was documented by Doherty et al. with an OR of 11 for long-term users with ≥6 years of therapy [26]. The strongest effect size regarding an increase in EC risk among scMHT users was observed by Razawi for long-term users with an OR of 4.35 [7], whereas the strongest effect regarding a decrease in EC risk among ccMHT users was found by Allen et al. with a HR of 0.24 among ever users [21].
Of note, epidemiological data strongly suggest that MHT has a primary preventive effect regarding EC. Two independent analyses of US population-based data found a significant increase in EC incidence coinciding with a decrease in MHT prescriptions. Specifically, Constantine et al. noted that EC rates were constant from 1992 to 2002 but increased by 2.5% annually with a 10% increase from 2006 to 2012 [9]. In accordance with these data, Wartko et al. also confirmed that EC incidence rates increased in the US by 2.5% annually after 2002 [17]. Since ccMHT is especially effective for EC risk reduction among obese women, ccMHt should be recognized as an efficient preventive measure in light of the ongoing obesity epidemic.
The present review adds to our knowledge of EC as an unwanted side effect of MHT providing an up-to-date analysis of the evidence. The difference of the present review compared to other studies is its comprehensive design including narrative reviews and work related to genetic analyses, interaction with BMI, and the effects of MHT prescription practice on EC incidence. In addition, the latest review on this topic dates back to 2016 and thus, an update of the evidence related to this clinically relevant topic is appropriate.
The available data do not show that MHT-associated ECs have a specific spectrum of histological subtypes. In this regard, most MHT-associated ECs are thus endometrioid adenocarcinomas [7,8,19,21,25,30]. A preference for aggressive histological subtypes such as clear cell carcinomas or malignant mixed müllerian tumors, which has been noted for tamoxifen-associated ECs [1,2], has not been found when women using MHT are diagnosed with EC.
Limitations of this review include the lack of meta-analysis, heterogeneity of data, and restriction to the last 20 years omitting older studies. However, based on the broad spectrum of studies identified, we believe that this review is a comprehensive and representative assessment of the current state of knowledge regarding this topic. In the future, the association between genetic markers such as variants of CYP17 and CYP11A1 and EC risk among MHT users should be tested in further studies. Individualization of risk profiles by genetic assessment carries a high potential for improvement and personalization of medical care. However, at this time, genetic testing remains experimental, as demonstrated by the available literature identified in this review. Thus, there is no reason to recommend any form of genetic testing in MHT users except for thrombophilia testing (prothrombin G20210A mutation or factor V Leiden mutation) in women with a personal and/or family history of thrombosis [3].

Conclusions
In conclusion, MHT affects the risk of EC in different ways. MHT with estrogens only increases the risk of EC. We also found that type and duration of progestin use in the context of MHT is an important factor influencing a woman's risk of developing EC. Specifically, ccMHT with synthetic progestins reduces the risk of EC, whereas scMHT with synthetic progestins and ccMHT and scMHT with natural progestins increases the risk of EC depending on progestin dosage and duration of use.
Author Contributions: C.B.T. and G.A.R. collected data and wrote the manuscript. G.A.R. and C.B.T. analyzed the data. I.J.-B., Z.H. and P.K. co-wrote the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.

Conflicts of Interest:
The authors declare that they have no competing interests.