Hormone Replacement Therapy in Cancer Survivors – Review of the Literature

Hormone replacement therapy (HRT; also known as menopausal hormone therapy, MHT) means substituting estrogen (or compounds exerting estrogenic effects) and progesterone (or compounds exerting progestagenic effects) after the cessation of cyclic ovarian hormone production. In the context of young oncologic patients with premature ovarian insufficiency (POI) the term HRT seems to be more appropriate then MHT. Oncologic risk of MHT requires consideration from two aspects: the potential of MHT to induce tumours in patients who have no oncologic history; and the potential to cause cancer recurrence and progression in cancer survivors. The former question is frequently asked by patients and needs to be explained to those who receive MHT. To date, a lot of data from large randomised controlled trials and relevant guidelines are available in context of the most common malignancies, such as breast or colorectal cancer. Most of these are reviewed in the clinical guidelines of menopausal hormone therapy, such as that of the International Menopause Society, IMS [1], the National Institute for Health and Care Excellence, NICE [2] or the American College of Obstetricians and Gynecologists, ACOG [3].

The situation is far more complex when it comes to cancer survivors. They can experience premature ovarian insufficiency as a consequence of cancer treatment (surgery, chemotherapy or radiotherapy), or as a result of an independent disease (see all the possible causes of POI, e.g. genetic or autoimmune diseases, or the consequence of other benign ovarian pathology), or simply may survive long enough to reach the age of physiological menopause around the age of 50. HRT may be necessary because of menopausal symptoms, but young asymptomatic patients should also receive hormone replacement if not contraindicated. It is well known that refusal of MHT decreases not only the quality of life, but also the life expectancy of young menopausal patients by several years. In a Dutch study, this was found to be 2 years lost over a 17-year follow-up period, mainly due to cardiovascular and osteoporotic morbidity [4]. Thus, not only inadequate initiation of MHT, but also its unsubstantiated denial, ’just to be on the safe side’, harms the patient.

Professional decision-making, however, is hindered by several factors. The major problem is the nearly infinite number of possibilities that biologically characterize the two potentially counteracting participants: the malignant disease and hormone therapy. Our decision depends on

It is obvious from this list that although ’individualized decision-making’ is needed, data on which this should be based can not always be readily available. When trying to find relevant data, depending on the prevalence of the tumour type, papers from preclinical research, case reports, retrospective studies, randomized controlled trials and meta-analyses can be found. Their strength to predict the risks and benefits of MHT in the given clinical setting is varied, yet, these may be the only source of information for the clinician.

In this review we try to summarize the data regarding the advantages and disadvantages of HRT in the most common tumour types.

The effect of estrogen on a tissue or a certain cellular function is not at all straightforward. Classic nuclear estrogen receptors (ERs), ERα and ERβ exert their effect after estrogen binding and dimerization via binding to estrogen response elements (ERE) in the promoter and regulatory regions of their target genes [Fig. 1]. This binding can be direct or can be mediated by transcription factors (TFs). Also, ligand (estrogen) independent activation and ERE-binding of ERs via phosphorilation triggered by for example growth factor receptors (e.g. HER2, IGF-1R), as well as ER independent activation and ERE-binding of TFs is known. This activation can be mediated by endoplasmic reticulum or surface membrane bound classic or G-protein-coupled 7-transmembrane spanning (GPER or GPR30, xER types) receptors, or estrogen independent protein kinase activation. [5, 6]

Estrogen sensitivity and ERE mediated activation of cells or genes can be defined in several ways. When gene upregulation following the addition of estrogen is measured, over 1000 estrogen sensitive genes can be identified, the exact number depending on cell and tissue type [7]. The other approach is identifying EREs in the promoter region of genes. ERE databases of the human genome are available online [8, 9], and include for example the following genes: EBAG9, c-fos, OXT, F12, TFF1, LTF, CTSD, PFDN2, TGF-α, AGT, GREB1, KIAA1243, NRIP1, MADH9, NME3, TPD52L, and ABCG2. [9].

To make the situation even more complex, estrogen-related receptors (ERRα, ERRβ, ERRγ), several estrogen-binding and non-estrogen-binding splice variants (isoforms) of ERα (e.g. ERα-46, ERα-36) and ERβ (ERβ1-5), and interaction of ERα and ERβ via heterodimerization have been known for several years [10, 11]. Taken all these into consideration, it is clear that the presence of estrogen and its receptor is not enough to predict the effect of estrogen in any cell type. Yet, the estrogen (and progesterone) receptor status has long been known to be an important prognostic factor and has been used to determine the mode of oncotherapy in certain cancer types, for example breast cancer. Estrogen receptor overexpression, however, has not only been found in the well-known estrogen dependent tumour types, such as breast, endometrial and ovarian cancer. Altered estrogen and / or progesterone receptor expression has been documented for example in thyroid cancer [12‐15], Hodgkin’s lymphoma [16‐18], B-cell malignancies [19, 20], brain tumours [21‐25], prolactinoma [26, 27], melanoma [28‐31], lung cancer [32, 33], colorectal cancer [34‐36], gastric cancer [37] and liver cancer [38, 39].

Nevertheless, receptor expression pattern is not enough to predict the effect of estrogen or progesterone on a given tumour type. In general, ERα can induce estrogen-dependent proliferation while ERβ can inhibit it, but the overall biologic effect of estrogenic compounds on different cells, tissues and tumour types is determined by the complex interplay of ER subtypes, isoforms, ER-related receptors and heterodimerization, presence of transcriptional coactivators and corepressors, as well as estrogen and non-estrogen mediated phosphorilation of all the above-listed ’players’. As mentioned in the previous section, estrogen effect on the stroma, immune response and participants of tumour metastatization also modifies the hormonal effect in vivo. As the effect is therefore practically impredictable, experience and its proper interpretation in the preclinical and clinical setting is inevitable.

As both the symptoms and the long-term negative effects of menstrual cycle cessation are attributed mainly to estrogen deficiency, HRT needs to substitute estrogen. Yet, to prevent the oncologic risk of unopposed estrogen stimulation of the endometrium, progestagenic compounds are also used in patients with an intact uterus. As the progestagens used do not have a pure progesterone-like effect (except for micronised progesterone) but can also act as activators or inhibitors of different potency on other steroid hormone receptors (androgen, mineralocorticoid and glucocorticoid), prediction of E+P HRT effect on tissues simply based on receptor expression and signalling is nearly impossible. Progesterone signalling is not less complicated than estrogen signalling, and beyond that, estrogen and progesterone interaction, progesterone effect on the tumour stroma, immune cells and the events of metatstatization and the effect on other steroid hormone receptors combine to finally determine the overall impact that progesterone-containing HRT exerts on a certain malignant disease. Detailed discussion of progesterone signalling exceeds the scope of this review. Nuclear progesterone receptor (PR) A and B activation and progesterone response element (PRE) binding leads to ’classical’ progesterone signalling, but rapid progesterone effects are also mediated by membrane-bound PRs, cytoplasmic PRs and receptor-independent signalling via various signal transduction pathways, called ’non-classical’ progesterone signalling, as reviewed in detail by Garg and colleagues [40] and summarized in Fig. 2 [40‐42].

Similarly to what was discussed above in detail with regard to estrogen, progesteron receptor expression pattern is also not enough to exactly predict the effect of progesterone on a given tissue or tumour type. Therefore, each cancer type and its relation to MHT requires an independent search of the literature. Thus, to make the most advantage of this work, we included in our search the most common tumour types, whose survivors are therefore the most probable to show up with the request of HRT. Childhood and young adult (ages 20 to 39) cancer survivors will present with premature ovarian insufficiency, whereas survivors of the most common female cancer types may come for a professional opinion regarding HRT at or after the natural age of menopause. The most common cancer types are listed in Table 1, based on the 2015-2016 data of the American Cancer Society.

In this review, we are going to discuss the relation of MHT and various cancer types according to the following grouping:

One of the most commonly encountered clinical situation in the field of HRT in oncologic patients is the request of breast cancer survivors to relieve their menopausal symptoms. Their menopause can be the result of either chemo- or radiotherapy, or some type of antiestrogenic endocrine therapy. The general attitude is straightforward, as also stated in the guideline of the International Menopause Society: no hormone replacement therapy should be given to these patients [1]. Rather, non-hormonal methods, including lifestyle changes, behavioral therapy, gabapentine, venlafaxine or fluoxetine should be preferred [44]. Although the observational and case control studies before 2002 indicated no increased risk of recurrence [45‐48], HABITS (Hormone Replacement Therapy After Breast Cancer – Is it Safe?), the first large randomized, controlled trial (RCT) of the field changed the attitude and has dominated professional approach ever since. HABITS was stopped in 2003 after 2.1 years as the results showed an increased risk of breast cancer recurrence (n=434, recurrence 26 cases in the HRT group vs. 7 cases in the non-HRT group, HR: 3.3) [49]. The other RCT at the time, the Stockholm Trial was also stopped based on the results of the HABITS trial, although its results showed no increased risk of recurrence, with a RR of 0.82 when the trial was prematurely stopped (n=379). However, there were significant differences between the two trials effecting the lymph node positivity and tamoxifen application. Although the original RCT results seemed to be conflicting, and even now no clear-cut conclusion can be drawn, trials of the last decade, including extended follow-up studies of HABITS and the Stockholm Trial have seemed to indicate an increased recurrance risk of breast cancer after different HRT regimens, with the relative risk (RR) of recurrence varying between 2.0 to 3.6 [50, 51]. In the extended follow-up of HABITS, Holmberg and colleagues found a RR of recurrence of 2.4 (n=442, mean HRT duration 24 months, follow-up 5 years, recurrence 22.2% HRT user vs. 8.0% non-user) [52]. The 10-year follow-up results of the Stockholm Trial also indicate an increased breast cancer recurrence risk. In the study of Fahlen and coworkers, a hazard ratio (HR) of 3.6 was detected for recurrence of the disease (n=378, mean HRT duration 26 months, recurrence HRT users 7,4% vs. non-users 2.1%) [53].

Not only estrogen + progestagen oral HRT regimens have been tested, but also tibolone, a compound that is metabolized to an estrogenic, progestagenic and androgenic isomer and is routinely used for MHT. The LIFT (Long-Term Intervention on Fractures with Tibolone) study proved that not only does tibolone decrease fracture risk in an osteoporotic postmenopausal population, but it also reduces invasive breast cancer risk significantly (odds ratio 0.32) [54]. The LIBERATE (Livial Intervention Following Breast Cancer: Efficacy, Recurrence and Tolerability Endpoints) Trial was conducted to assess the use of tibolone in breast cancer survivors. Although bone mineral density (BMD) and climacteric symptoms significantly improved, the trial was prematurely terminated because of the increased recurrence risk of breast cancer (n=3098, follow-up 3.1 years, breast cancer recurrence 15.2% with tibolone vs. 10.7% with placebo, HR: 1.4) [55‐57]. Interestingly, cancer recurrence was only observed in the normal BMD group, suggesting a local estrogen effect unrelated to the circulating estrogen in the plasma. Unfortunately, due to the disapointing outcome of these prematurely halted trials, we could not get a clear vision on the safety of HRT in breast cancer survivors in the following decade.

Although HRT is generally contraindicated in breast cancer survivors, mainly based on the studies mentioned above, some points are worth consideration during decision-making.

As a result of rapid advance in oncology, both the diagnosis and the treatment of malignant tumours has been constantly improving, leading to the increasing survival of oncologic patients. More and more of them live long enough to reach either the natural age of menopause or, as a side effect of their oncotherapy, experience the cessation of their gonadal function, leading to premature ovarian insufficiency. Thus, an ever increasing number of cancer survivors search endocrinologic help in the form of the replacement of their missing female hormones. Often it is not the endocrinologist who is first asked by the patient, but other members of the professional staff involved in her oncologic care, including the oncologist, gynecologist, general practitioner, or even the radiologist or the oncologic nurse. The general attitude has been defensive: ignorance of the latest results in the field of menopausal hormone therapy and false concepts dating back to the WHI (Women’s Health Initiative) Study results almost two decades ago [173]. The misinterpretation of the results of this study are well known among the specialists of the field of menopausal medicine and will not be discussed here. The net outcome, however, became an irrational fear of female hormone replacement, both by the general population and medical professionals. It might has seemed the logical and safe conclusion to many physicians that it is better to avoid HRT, because this attitude definitely causes no harm, whereas the decision of prescribing estrogen with or without progestins for a patient might bear oncologic and cardiovascular risks, may lead to litigation in case of a potentially related complication, and requires meticulous and continuous self-education in the field not to miss the latest evidence. The major mistake in this reasoning is the baseline theory of not harming by avoiding HRT. It was known even before the WHI results that premature menopause and hypogonadism decreases the life expectancy of women by years through its skeletal and cardiovascular effects, and this negative effect correlates with the length of the hypoestrogenaemic period. 17 years of untreated hypogonadism add up to losing 2 life-years [4], and every year of delay of menopause decreases cardiovascular mortality risk by 2% [174]. The quality of a life lived with severe menopausal symptoms is yet another matter not to forget. Therefore, it is not prescribing HRT for a young patient suffering from premature ovarian insufficiency (POI) that requires professional explanation - as it is clearly stated also by the latest guidelines of the International Menopause Society [1] and the Global Consensus Statement on MHT [175]. It is the denial of HRT, a decision proved to harm the patients health and decrease her life expectancy, that needs to be supported by evidence and should be weighed againts the risks (oncologic or other) of HRT. Denying HRT ’just to be on the safe side’ is unacceptable – we can only make this maleficent decision if the definitive harm caused to our patient by it is smaller than the suspected harm of HRT.

Yet, it is not easy to assess the HRT-related extra risks of HRT in cancer survivors. Practically, every tumour is a new and potentially unprecedented entity, as far as its oncogenic mutations, hormone receptor status and biological beviour are concerned, even if medical science tries to group tumours according to organ, histology or molecular characteristics. The variability of the stage, grade and former oncotherapy received by the patient complicates decesion making even further. It is clear that deciding for or against HRT should always be individualized and decision should be made together with the patient. The risk-benefit assessment can be based on a wide range of scientific evidence of different quality, depending on their availability: from in vitro studies to clinical investigations, and the latter ranging from individual case reports to meta-analysis of several randomized prospective trials, and even indirect evidence (e.g. epidemiologic data of tumour behaviour in the different sexes, during hormonally altered periods like in pregnancy, or HRT effect on tumours not in cancer survivors etc.). In this review we have tried to collect data about the most common types of tumours, in an attempt to help clinical decisions (Table 2). Based on the evidence gathered so far, adequately chosen female hormone substitution can be clearly advantageous in some cancers (e.g. endometrial cancer, haematologic malignancies or colorectal carcinoma) and harmful in hormone dependent others (e.g. breast cancer, meningioma or ER+/PR+ gastric cancer). In certain tumours or in their subgroups, the risk of HRT seems to outweigh the potential benefits and thorough individualized decision-making is necessary (e.g. lung cancer or bladder cancer), whereas in many tumour types MHT is neutral for the malignant disease and thus it should not be denied (e.g. cervical cancer, kidney tumours or pancreatic cancer). In many situations, however, when more rare cancer types are faced, individual search of the literature for the specific tumour will be necessary. Clinical evidence of recurrence risk of the given cancer type, weighed against the negative long-term effects of avoiding HRT, the general state and life-expectancy of the patient, the severity of her hypoestrogenic and other symptoms, and her subjective fear of the recurrence of the disease, which is understandably always dominant in cancer survivors, will help to find the best solution.