Aging is associated with an increase in the development of cancers. Among them some can be influenced by the gonadal hormones. Cancers which are the most hormone dependent are breast and endometrial cancer. Among breast cancer (BC) most of those occurring in postmenopausal women are estradiol (E2) receptor (ER) positive. Endometrial cancer is extremely estrogen dependent. Ovarian cancer (OC) can also mildly be influenced by hormones. Cervical adenocarcinoma (but not squamous cell carcinoma) is possibly a target of estrogen. But in addition to cancer arising from hormone dependent organs, gonadal hormones can also impact some other cancers arising in organs which are not classically hormone dependent.
Aging is associated with an increase in the development of cancers. Among them some can be influenced by the gonadal hormones. Cancers which are the most hormone dependent are breast and endometrial cancer. Among breast cancer (BC) most of those occurring in postmenopausal women are estradiol (E2) receptor (ER) positive. Endometrial cancer is extremely estrogen dependent. Ovarian cancer (OC) can also mildly be influenced by hormones. Cervical adenocarcinoma (but not squamous cell carcinoma) is possibly a target of estrogen. But in addition to cancer arising from hormone-dependent organs, gonadal hormones can also impact some other cancers arising in organs which are not classically hormone dependent.
First we will address the question of the level of risk associated with MHT and cancer, and in the second part which treatment can be used in cancer survivors.
MHT and the Risk of Cancer
BC is the paradigm of hormone dependent cancer. A total 2 088 849 new cases were diagnosed in 2018 worldwide and 626 679 deaths . The mortality has decreased in most of the western countries but not in eastern Europe. There are different types of BC. Most of them express ER but with different levels. They are defined by histological types, mostly ductal type (85 per cent), lobular type (15 per cent) and their content in E2, progesterone (P) receptors (PR) and HER-2 amplification. Tumours expressing high levels of ER and expressing PR correspond to Luminal A BC, have a good prognosis and are extremely hormone dependent. Luminal B contain ER but usually not PR. The prognosis of luminal B BC is less favourable. About 15 per cent of BC express also an amplification of HER-2. They are treated with antibodies against HER-2, which has significantly improved their prognosis. The last type is called triple negative (TN) BC, as they do not express ER, PR or HER-2 amplification and have the worst prognosis. The relative proportion of ER+/TN BC increases with age. After menopause, 70–80 per cent of BC contain ER and 40 per cent PR. It can thus be easily understood that MHT can feed a preexisting cancer or promote the growth of precancerous lesions or ductal carcinoma in situ (DCIS).
Evidence from most of the studies is that
– an estrogen-only treatment (ET) is associated with a lower risk than a combined (estrogen+ progestogen) therapy (CT)
– risk increases with duration and decreases after cessation of the treatment arguing for a promoter effect on preexisting lesions since the lag time to develop BC is usually 20–25 years.
– ER-positive breast cancers are associated with MHT. Luminal A type is the predominant type but luminal B is also potentially associated with MHT [2, 3]. HER-2+ and TN BC are not modified by MHT [2, 3].
In most observational studies, the use of ET is associated with a mild risk at long term. In the French cohort study E3 N, a small increase was observed using E2 after long-term use (>5 years) but on a small number of patients . In the Nurses’ Health study, a large prospective cohort study conducted in the US, the increase in the risk is observed after 10–20 years of treatment [2, 5].
The Women’s Health Initiative (WHI), which has included 11 000 postmenopausal hysterectomized women randomized between conjugated estrogens (CEE) and placebo, observed no increase in the risk and even a decrease in adherent women . After a mean treatment duration of 6.8 years, the relative risk (RR) of invasive breast cancer was 0.77 (0.59–1.01) in the whole population and 0.67 (0.47–0.97) in adherent women. Only women naive of treatment had a decrease in the RR . Longer follow-up of the WHI population confirmed the significant decrease of BC in the whole population .
The WHI population was different from the one included in the other studies by the fact that a majority of women were far from menopause: 45 per cent were 60–69 years old, 24 per cent were 70 or above. Observational studies include women at menopause. Clinical characteristics were different: in the WHI, mean BMI was 30.1 (6.1), and 45 per cent were obese whereas in the Nurses’ and the E3 N cohort, women were much leaner and healthier. The prevalence of obesity can influence the effects of MHT on breast cancer risk. In several studies it was shown the leaner women were at higher risk than obese women using MHT. In the Nurses’ the lower risk of BC was seen with 5 to 9.9 current years of ET use and to be limited to women with a BMI of 25 or higher (RR: 0.74; 0.55–1.00) . Insulin resistance is decreased by estrogen. It is likely that the women in the WHI ET trial could have benefitted from the treatment by decreasing insulin resistance. Another possible explanation is that CEE which contain mixed compounds having both agonist and antagonist estrogen potencies, thereby have different impacts on the breast. The influence of type of estrogen is, however, not confirmed. In the WHI observational cohort, the dose of CEE (usual and low dose <0.625 mg) and oral and transdermal E2 were not associated with a different level of risk. In that cohort, the RR with CEE was not increased nor decreased, with an average of follow-up of 8.2 years . Similarly, in a study from Finland, there was no difference in the risk of oral E2 and transdermal E2 combined with norethisterone acetate (NETA) .
There is no study which has compared head to head an ET or CT treatment, and the characteristics of women in both groups may vary. Nevertheless, in most observational studies, CT is associated with a higher risk of breast cancer than ET. The level of this risk varies with the studies and the progestogen used. The RR decreases progressively after cessation of treatment, suggesting a promoter effect on preexisting lesions. The WHI, conducted another randomized control trial (RCT) consisting of 16 000 women randomized between CEE+ medroxyprogesterone acetate (MPA) and placebo. In that trial, there was a limited increase in the RR =1.24 (1.01–1.53) of BC . The characteristics of the patients were different from the observational studies: mean age at inclusion was 63.3 years, 34 per cent were obese. Observational studies in most of cases show an increase in the RR of a larger magnitude around 1.7/2 fold with CT and synthetic progestogens (SP) (Table 21.1). Different levels of risk are reported with different progestogen (for review, see ). Observational studies reported a lower risk of BC in women treated with progesterone/dydrogesterone compared to synthetic progestins. Micronized progesterone (P) is practically only used in France. Dydrogesterone (DYD) has some properties in common with P. There are now four studies, two from France, one from UK and one from Finland, reporting no or lower risk of BC with these two compounds, P and DYD (Table 21.1) .
|Study||Population||Follow-up (mean)||Risk of BC|
|Fournier ||8.1 years|
|Espie ||2.5 years|
|Schneider ||6.0 years|
|Lyytinen ||E2 + NET:|
Note. ET = estrogen only. CT = combined MHT. P = progesterone. DYD = dydrogesterone.
Duration is also a factor influencing the risk, the longer the duration the more important the risk. In the French E3 N study, a small increase in the risk was observed with P/DYD after more than 6 years of use in current users (mean: 8.7 years) RR = 1.31 (1.15–1.48) compared to SP (mean use: 8.4 years) RR = 2.02 (1.81–2.26). After 3 months to 5 years of cessation, the risk disappeared in women treated by P/DYD, RR = 1.15 0.93–1.42) and for SP, the RR reached 1 after >10 years since last use . In a meta-analysis the RR of BC comparing P and SP, was RR = 0.67 (0.55–0.81) .
The type of regimen, continuous or sequential, has also been associated with a different level of risk, the sequential treatment being at lower risk. There is no head-to-head comparison between both regimens, which precludes firm recommendations since most women prefer a continuous regimen.
Tibolone is a normethyltestosterone derivative and thus a progestogen. It is metabolized into an androgenic and two mild estrogenic metabolites. The risk associated with tibolone is likely to be in the same range as that with CT. One RCT in BC patients reported an increased risk of recurrence whereas a RCT in osteoporotic women of a mean age of 68.3 ± 5.2 years reported a decrease in the RR of BC (0.32 [0.13–0.80]), but based on six cases in the tibolone group and 19 in the placebo group and after only 3 years of use. This is a short time and those women were at lower risk of BC since osteoporotic [13, 14]. In the Million Women Study (MWS), a large observational study conducted in UK, the RR with tibolone was intermediary between ET and CT risks .
Levonorgestrel IUD (LnorgIUD): there are controversial data on the RR of BC with LnorgIUD. Most of the results are from its use as contraceptive. Some studies based on registers observed an increase in the risk of BC when used as contraceptive or in the menopausal transition (see ). At the opposite in a cohort study, in contraception use, there was no increase in the RR of BC . Another observational study reported an increase in women who used LnorgIUD at the age of 40–45 years but not in women of 46–50 years old. It is thus difficult to conclude but these studies suggest that a small risk cannot be excluded.
A combination of CEE and a SERM, Bazedoxifene is available in some countries. Breast density and mastalgia are not increased by this treatment as well as the endometrial thickness, whereas breast density and mastalgia are both increased by CT. There are no data on breast cancer in humans and it may increase the risk of thrombosis. There may be a place for this regimen in women who develop mastalgia using standard MHT or benign breast disease but no study available.
The concept of combining an estrogen with a SERM is elegant. A randomized study (Hot study) included more than 1800 women using MHT and 5 mg of tamoxifen versus placebo demonstrated a decrease in luminal A BC . The potential increase in venous thrombosis should be discussed in the risk/benefit assessment.
How to Decrease the RR of BC in Women Using MHT?
– Selecting women with a lower risk?
Women at very high risk for BC are those belonging to a family with strong history of BC/ovarian cancer at a young age, who have an history of atypical hyperplasia at biopsies, women who have been treated by thoracic radiotherapy at a young age, and in cases of birad-d category of breast density. Those women are not the best candidates for MHT.
Other risk factors are: late first full-term pregnancy, alcohol, sedentarity, obesity, diabetes, late age at menopause, young age at menarche, biopsy with proliferative disease, mastalgia.
– Treatment with lower risk: P or DYD are associated with a lower risk. Sequential therapy seems more favourable for the breast but not for the endometrium (see below).
– Education on lifestyle.
Physical exercise, losing weight, low breast density, low bone density are factors associated with lower risk. Several studies using models to predict the risk in their cohort showed that women who had the best score for lifestyle factors were at lower risk of BC even using MHT . In the Nurses’ cohort, evaluating the attributable part of BC to modifiable factors at menopause, they concluded that risk factors that are modifiable account for more than one-third of postmenopausal BC . This is crucial information, and education on lifestyle may help to decrease the mild RR of BC observed with MHT.
– In practice, before prescribing a MHT, to evaluate the risk of BC is mandatory. This can be done by gathering the risk factors and, if necessary, using scores which are available online. The Gail score (bcrisktool.cancer.gov/) and the IBIS score (www.ems-trials.org/riskevaluator/). The last one contains more clinical items including breast density category and a detailed family history as part of the analysis to predict the probability of being a BRCA carrier. Its accuracy remains, however, about 0.70, whereas the Gail score has a lower accuracy (0.6).
Prevalence of EC is increased in postmenopausal women with a peak of incidence between 60 and 75 years of age. It is the sixth commonest cancer in women worldwide with 382 069 new cases and 89 929 deaths in 2018 . Eighty-five per cent of endometrial cancers belong to type I: hormone dependent, well differentiated and with a good prognosis. Type II includes serous carcinoma, mixed cell and clear cell carcinoma. These are more invasive and their prognosis much less favourable. Type I is extremely sensitive to estrogen, type II can also be estrogen sensitive. The main risk factors are endogenous or exogenous estrogen, obesity, nulliparity, early menarche, late menopause, diabetes, hypertension, whereas smoking is associated with a reduced risk. It can rarely occur in the context of hereditary risk with colon cancers (Lynch syndrome). A major factor of protection is the pill with a long-term residual effect on its prevention. ET alone in women with a uterus is a high risk factor increasing with treatment duration. It progresses from hyperplasia initially, followed by hyperplasia with atypia, and then carcinoma. Sequential CT is at higher risk than continuous CT. The MWS demonstrated that obese women and overweight women were protected from EC by CT, due to the benefits of the progestogen . In long-term studies, the increase in EC risk observed with ET and CT was remnant lasting up to 5–10 years after cessation of treatment [21, 22]. The duration of progestogen administration combined with estrogen is crucial . In sequential administration it has to be at least 12–14 days according to the length of estrogen administration. P was suspected to be at higher risk than synthetic progestogen in the E3 N cohort but this might be due to a lack of compliance since two RCT using endometrial biopsies did not show an increase in hyperplasia in women using P nor a recent oral combination of E2+P ( Lobo et al Menopause 2018) (for review, see ). LnorgIUD (Mirena, no data on the other IUDs) is protective of the endometrium to the same extent as an oral progestogen; they equally can reverse hyperplasia and well-differentiated adenocarcinoma [24, 25].
An uncertainty was raised about tibolone and endometrial cancer but was not confirmed in RCT .
OC occurs predominantly in postmenopausal women, with a peak in women aged 55–64 years and a median age of 63 . It is the seventh most commonly diagnosed cancer among women in the world with 295 414 new cases in 2018 and 184 799 deaths . The percentage surviving at 5 years is 30–40 per cent . The most frequent belong to the epithelial group and among them high-grade and low-grade serous tumours represent 70 per cent. Clear cell carcinoma, endometrioid and mucinous types are the other differentiations. Other types are germ cell, stroma cell and small cell carcinomas. Borderline tumours can be of the same type, but with a lower potential of evolution.
Risk factors are a family history of OC (and BC) or colon cancer (CC) (Lynch syndrome), infertility, endometriosis (for the clear cell and endometrioid carcinoma), obesity. A major factor for protection is the pill (with a long-term remnant effect on prevention) and tubal ligation. There are only data from observational studies. No data were obtained in the WHI on OC, due to its relative low frequency. In the MWS, an increased risk was reported as well as in a meta-analysis which included 52 studies but with a high weight for the MWS itself . There was a RR = 1·43 (1·31–1·56). Not all the risk or protective factors were available in part of the population, in particular hormonal contraception, which is a strong factor suggesting that bias could be present in that study. In most observational studies, ET appeared to be associated with a mildly higher risk than CT. In this meta-analysis ET and CT were equivalent. The mucinous type is not influenced either by MHT or by hormonal contraception.
Cervical cancers belong to squamous small cell carcinoma (SSCC) (over 90 per cent) and adenocarcinoma. Worldwide, 569 847 new cases were diagnosed in 2018 and 311 365 deaths occurred. It is the fourth commonest cancer in women, predominant in countries with low income (no screening). The main risk factor is HPV infection. The SSCC is not dependent on hormones and thus not influenced by MHT. Adenocarcinoma is considered as estrogen dependent. However, there are few data on adenocarcinoma and its risk factors. A publication suggested that the pill could increase the risk of adenocarcinoma in a larger instance than for SSCC  but not MPA when used as a contraceptive . In postmenopausal women, ET was associated with an increase of adenocarcinoma but based on a very small number of cases (14).
CC represents 520 812 new cases in women worldwide and 260 760 deaths . Colon is not considered a hormone-dependent organ but the normal epithelium contains ERβ. The main risk factors are genetic, obesity, processed meat, sedentarity and alcohol. Exercise and healthy diet help to decrease the risk. Both the oral contraceptive pill (OCP) and MHT are reported to be associated with a decrease in CC in most but not all of the studies. In the WHI, a significant decrease in CC was observed in the CT trial, but not in the ET trial (more obesity?, role of MPA?) . Several case control and observational studies also reported less CC in women using MHT [32, 33]. One possible bias is the prevalence of screening and colonoscopy in different populations, which may vary and is an effective means of CC prevention. In the E3 N cohort ET use was associated with a decrease risk of CC but an increase in adenomas, which could indicate colonoscopy and thus prevention of cancer .
Lung cancer has an increasing incidence in women and will be in most countries the first cause of death by cancer in women . New cases in women represent 725 352 and 576 060 deaths, just after BC . The main risk factor is smoking: 80 per cent of lung cancers occur in smokers. Lung cancer may contain ERα, ERβ and PR and their contents have been associated with prognosis [35, 36]. The prognosis in women is more favourable than in men. In the WHI CT trial, there were less lung cancers but a higher mortality , whereas there was no effect in the ET trial. The effect in the CT trial decreased after stopping CT . The results from other studies are contradictory. Smoking is a confounding factor since it may modulate ER and PR status in lung cancer . There are different kinds of lung cancer and the effect of hormones may vary according to their biology. A recent meta-analysis did not observe increase in mortality in women using MHT .
It represents 137 025 new cases and 25 881 deaths . Melanomas are more prevalent in women before menopause than in men of same age. The main factor is sun exposure. The effect of MHT is controversial showing no effect in a meta-analysis nor increase by ET and decrease by CT in a more recent cohort study performed in Norway .