Diabetes mellitus in breast cancer survivors: metabolic effects of endocrine therapy

Abstract

Breast cancer is the most common invasive malignancy in the world, with millions of survivors living today. Type 2 diabetes mellitus (T2DM) is also a globally prevalent disease that is a widely studied risk factor for breast cancer. Most breast tumours express the oestrogen receptor and are treated with systemic therapies designed to disrupt oestrogen-dependent signalling. Since the advent of targeted endocrine therapy six decades ago, the mortality from breast cancer has steadily declined; however, during the past decade, an elevated risk of T2DM after breast cancer treatment has been reported, particularly for those who received endocrine therapy. In this Review, we highlight key events in the history of endocrine therapies, beginning with the development of tamoxifen. We also summarize the sequence of reported adverse metabolic effects, which include dyslipidaemia, hepatic steatosis and impaired glucose tolerance. We discuss the limitations of determining a causal role for breast cancer treatments in T2DM development from epidemiological data and describe informative preclinical studies that suggest complex mechanisms through which endocrine therapy might drive T2DM risk and progression. We also reinforce the life-saving benefits of endocrine therapy and highlight the need for better predictive biomarkers of T2DM risk and preventive strategies for the growing population of breast cancer survivors.

Introduction

Breast cancer is the most prevalent invasive cancer in the world¹. At the end of 2020, 7.8 million individuals were alive who had been diagnosed with breast cancer in the previous 5 years. In many instances, the 5-year survival rates for early-stage disease are greater than 90%. Long-term prognosis is directly related to the tumour stage at diagnosis, but breast cancer remains a highly treatable disease with most patients living decades after therapy ends². More than 70% of breast cancers express the receptors for oestrogen or progesterone hormones. Men can also develop breast cancer and, although it is less common than in women, most tumours are hormone receptor-positive in men. Patients with oestrogen receptor (ER)-positive tumours receive treatments designed to interfere with ER signalling, either by blocking the receptor itself, with selective ER modulators (SERMs) or selective ER degraders (SERDs), or by preventing the production of the oestrogens with aromatase inhibitors. Through emphasis on early detection and diagnosis and comprehensive breast cancer treatment, mortality from the disease is declining, due in part to the efficacy of anti-oestrogen or ER-targeted therapies, collectively referred to as endocrine therapy.

Over the past decade, evidence has emerged to suggest that patients with breast cancer, particularly those treated with endocrine therapy, might have an elevated risk of type 2 diabetes mellitus (T2DM). Both obesity and T2DM are known risk factors for ER-positive breast cancer, and the observation of that T2DM is more prevalent in breast cancer survivors than in the general population suggests a vicious cycle involving both diseases (Fig. 1).

Fig. 1 |. The vicious cycle surrounding T2DM and breast cancer.

Type 2 diabetes mellitus (T2DM) increases the risk of oestrogen receptor (ER)-positive breast cancer. This disease can be effectively treated with endocrine therapies, including selective ER modulators (for example, tamoxifen) or aromatase inhibitors, but these treatments in turn might increase the risk of T2DM in survivors with breast cancer. This effect could create a vicious feedback loop.

This Review highlights the relationship between endocrine therapy and T2DM risk and poses key unanswered questions. Some patients with ER-positive breast cancer receive non-hormonal therapies, such as inhibitors of CDK4/6 or PI3K; however, we do not review those in detail here. We present a short history of endocrine therapy, focusing on drug development, key trials and early and later reported adverse metabolic effects of treatment that are linked to T2DM. These include effects on serum lipids, hepatic steatosis, body composition and glucose tolerance. We also discuss paradigms to explain T2DM development and suggest potential mechanisms that might contribute to T2DM risk after endocrine therapy, based in part on preclinical studies. Finally, we briefly revisit the role of ER signalling in metabolic tissues, an understanding of which will be important in the prevention of endocrine therapy-driven T2DM.

Endocrine therapy for breast cancer

Tamoxifen is an SERM that was synthesized in 1963 (ref. 3), becoming one of the first targeted therapies for cancer treatment⁴ (Fig. 2). By 1978, 19 clinical studies of tamoxifen as a single agent had been conducted in patients with advanced breast cancer³, and patients with ER-positive tumours were recognized to have a better response to tamoxifen than those with ER-negative tumours. In addition, older women (who are more likely to have ER-positive tumours) were reported to respond favourably to tamoxifen³. In 1981, the National Surgical Adjuvant Breast and Bowel Project (NSABP) began a randomized clinical trial testing the effect of tamoxifen in patients with ER-positive breast cancer (Fig. 2). By the late 1980s, the available evidence clearly showed that tamoxifen prolonged survival for women with breast cancer, especially after menopause^5,6. Tamoxifen treatment notably reduced both local and distant disease recurrence, as well as new tumour development in the opposite breast, which was an important observation. Similar results were seen with long-term follow-up in the tamoxifen trial STO-3 from the Stockholm Breast Cancer Study Group⁷, including the reduction in contralateral breast cancer diagnosis.

Fig. 2 |. A brief history of endocrine therapy for breast cancer.

Notable events surrounding the development of endocrine therapy and the reports of adverse effects are shown on the top of the timeline in red. The FDA approval years for endocrine therapies are shown on the bottom of the timeline in blue. ER, oestrogen receptor; NAFLD, nonalcoholic fatty liver disease; NSABP, National Surgical Adjuvant Breast and Bowel Project; SERM, selective oestrogen receptor modulator.

On the basis of this research, attention turned to the preventive potential of tamoxifen for women at high risk of breast cancer, and the NSABP P-1 trial began in 1992 (refs. 8,9). By 1998, the ability of tamoxifen to reduce the risk of breast cancer recurrence was obvious, and tamoxifen received approval for breast cancer prevention, but some adverse effects such as endometrial cancer and thromboembolic events were concerning. Nonetheless, tamoxifen offered the potential to save millions of lives and that is exactly what it did⁴. In addition to tamoxifen, several other SERMs, such as raloxifene, toremifene and bazedoxifene, are now FDA-approved (Fig. 2) and are potentially effective breast cancer preventive or therapeutic agents¹⁰.

Throughout the mid-1900s, scientists were studying the synthesis of oestrogens from androgens, catalysed by the aromatase enzyme (CYP19A1)¹¹. This knowledge led to the development of aromatase inhibitors in the early 1970s (Fig. 2). The steroidal aromatase inhibitor formestane (also called α-hydroxy-androstenedione) showed efficacy for metastatic breast cancer treatment in postmenopausal women in the early 1980s (ref. 12). Concurrently, other investigators were exploring nonsteroidal aromatase inhibitors, leading to the development of letrozole and anastrazole (Fig. 2). By the late 1990s, evidence clearly demonstrated that aromatase inhibitors prolonged disease-free survival as a first-line therapy for breast cancer, especially in postmenopausal women^13,14. Thus, the FDA approved letrozole, anastrazole and the steroidal aromatase inhibitor exemestane by the end of the decade (Fig. 2).

Rather than inhibiting ER directly, aromatase inhibitors prevent oestrogen production, depriving cancer cells of critical ER ligands. Aromatase inhibition is especially important after menopause when adipose tissue, including breast adipose¹⁵, becomes a primary source of oestrogens in humans. Early trials had established the benefit of extending tamoxifen treatment to 5 years in women with ER-positive breast cancer, so switching to the newer aromatase inhibitor drugs after 2–3 years of tamoxifen was proposed to offer the same protective effect without potential adverse effects. The use of either steroidal or nonsteroidal aromatase inhibitors alone or after a short duration of tamoxifen provided a superior survival benefit compared with tamoxifen alone in postmenopausal women with breast cancer^16–20. In the past decade, studies have reported effects of extended treatment (up to 10 years with tamoxifen or an aromatase inhibitor in women with ER-positive breast cancer), showing some evidence that the long treatment duration might provide additional protection against recurrence^21,22. Given the prevalence of breast cancer and the effectiveness of these drugs, millions of women might be living for years in a state of severe oestrogen deprivation.

Other approaches to treating breast cancer now include ovarian function suppression and the use of SERDs, such as fulvestrant²³ and the recently approved (January 2023) elacestrant²⁴. Elacestrant is prescribed to patients with breast tumours that have mutations in ESR1; a genomic aberration that often makes ER insensitive to tamoxifen or oestrogen deprivation. In some high-risk premenopausal patients with breast cancer, aromatase inhibition combined with ovarian function suppression is now the standard of care^25,26, potentially further prolonging the time that women live without oestrogens. Any long-term metabolic effects from the newly approved drugs will take some time to emerge, but on the basis of the observed T2DM risk after endocrine therapy (described in depth later), closely monitoring endocrine function in patients with breast cancer should be an important routine part of care.

Dyslipidaemia and hepatic steatosis: acute and persistent effects

Potential effects on lipid metabolism

Some of the first reported adverse effects of endocrine therapy were on lipid profiles, with a surge of data appearing through the late 1980s into the early 1990s (Fig. 2). These investigations were inspired by the concern that cardiovascular disease risk might increase with tamoxifen treatment, owing to the known cardioprotective effects of oestrogen. Early studies were reassuring, however, showing a consistent link between tamoxifen use and low circulating levels of LDL cholesterol^27–31. High total cholesterol, and especially elevated LDL cholesterol, is linked to exacerbated cardiovascular disease; the emerging benefit of tamoxifen on circulating levels of LDL cholesterol was seen after as little as 2 months and up to 3 years of use, but tamoxifen was also associated with elevated triglycerides (TAGs)^27,29–31. Switching from tamoxifen to aromatase inhibitors lowered TAGs and raised LDL cholesterol relative to baseline levels, compared with staying on tamoxifen^32,33. This change in lipids could indicate a rebound effect from tamoxifen-mediated suppression of LDL cholesterol and augmentation of TAGs, or it could indicate a specific benefit of aromatase inhibitors. The changes in lipid profiles that occur once tamoxifen is discontinued suggest mechanistic differences between blocking ER and oestrogen deprivation in terms of lipid metabolism. Specifically, lower LDL cholesterol after tamoxifen use similarly occurs with oestrogen treatment, which suggests that tamoxifen might have ER agonist effects in the liver³⁰.

The effects of aromatase inhibitors alone on lipid metabolism are varied and depend on the study design and control groups. For example, no adverse effects of exemestane were found on cholesterol or TAGs compared with placebo treatment³⁴. By contrast, others reported that letrozole treatment after 5 years of tamoxifen use was associated with greater circulating levels of all lipids relative to placebo-treated patients³⁵. A third study found no clinically important effects of exemestane or anastrazole on circulating levels of lipids compared with tamoxifen-treated patients³⁶. In 2020, the cumulative incidence of lipid events (for example, TAG or cholesterol measurements falling outside the normal range) was found to be greater in patients taking nonsteroidal aromatase inhibitors letrozole and anastrazole compared with those taking the steroidal aromatase inhibitor exemestane^37,38. Each of these studies was conducted in postmenopausal women. The differences in outcomes could be explained by distinct mechanisms of aromatase suppression for steroidal versus nonsteroidal aromatase inhibitors or potentially by relative differences in suppression of plasma levels of oestrogen between drug classes; however, further studies are needed to confirm these links.

A 2022 meta-analysis reported no link between endocrine therapy use and dyslipidaemia³⁹. Here, dyslipidaemia was defined as hyperlipidaemia, hypertriglyceridaemia or hypercholesterolaemia, so any specific changes could potentially be obscured by this analysis. Furthermore, comparisons were made between aromatase inhibitor use and tamoxifen as a monotherapy, or between aromatase inhibitor use and tamoxifen and aromatase inhibitor sequential therapy, but not between control individuals who were untreated or who did not receive endocrine therapy³⁹. Altogether, multiple studies suggest that in some situations, endocrine therapy can disrupt whole-body lipid metabolism, but the mechanisms might depend on the class of the drug, the age of the patient and the duration of treatment. Whether this disruption in lipid metabolism relates to later development of T2DM in individuals treated with endocrine therapy is not known, but warrants further investigation considering dyslipidaemia is linked to T2DM in people without cancer.

Associations with hepatic steatosis

Around the same time as the described lipid effects emerged in the mid-1990s, case reports began to surface that hepatic steatosis, now referred to as metabolic-associated fatty liver disease (MAFLD)⁴⁰, was surprisingly prevalent in individuals taking tamoxifen⁴¹ (Fig. 2). This phenomenon was confirmed in several studies, particularly for women with overweight or obesity^41–43. In 2001, visceral adipose tissue was reported to be greater and liver density was lower (indicating MAFLD) in tamoxifen-treated postmenopausal women versus control women. Notably, the authors concluded that prospective longitudinal studies would determine whether tamoxifen causes metabolic dysfunction, especially in patients with obesity, and/or whether patients with underlying metabolic disease are at increased risk of adverse effects of tamoxifen⁴⁴. This conclusion remains valid today.

Observations of tamoxifen-induced MAFLD have continued, including when individuals taking tamoxifen are compared with those taking aromatase inhibitors⁴⁵. Furthermore, MAFLD seems to occur within the first 2 years of tamoxifen treatment^46,47. A retrospective study, comparing postmenopausal women with early breast cancer treated with nonsteroidal aromatase inhibitors (letrozole or anastrazole) with age-matched women without cancer, found a greater prevalence of MAFLD with aromatase inhibitor use regardless of BMI or diabetes mellitus status⁴⁸, which is consistent with a protective role for oestrogens against hepatic steatosis. MAFLD is a concerning adverse effect of tamoxifen; however, it often resolves once tamoxifen is discontinued. As we describe subsequently, hepatic steatosis is also seen with tamoxifen use in mice. Although MAFLD is often associated with T2DM in people, it is not clear from clinical or preclinical data if the link between tamoxifen use and MAFLD is related to the later development of T2DM in breast cancer survivors.

Changes in body mass and composition

Body composition and weight changes are anecdotally reported by patients with breast cancer and survivors; however, rigorous data to support these claims are difficult to capture. Importantly, although weight gain is easy to measure, it does not reflect a gain in adipose tissue mass or a change in body composition, so using BMI is not the most precise way to infer adipose tissue mass. Weight (and specifically adipose tissue) gain during cancer therapy is a relevant concern, not only from the perspective of the patient but also given the link between excess adiposity and the risk of breast cancer recurrence⁴⁹. Monitoring total body weight change during treatment does not adequately reveal adiposity, as lean and adipose tissue mass changes with age, physical activity and cancer progression or treatment, although BMI and total body mass are fairly easy measures to obtain.

Millions of breast cancer survivors exist who could theoretically provide a wealth of data on changes on body mass and composition related to endocrine therapies; however, patients are prescribed various agents including cytotoxic chemotherapy plus corticosteroids, or endocrine therapies including SERMs, SERDs or aromatase inhibitors, and for different durations. Corticosteroid use alone is linked to weight gain and diabetes mellitus development, which limits appropriate controls and confounds analyses of body mass changes in patients with breast cancer taking endocrine therapies. For example, a 1990 study reported greater weight gain in patients with breast cancer taking chemotherapy plus prednisone, regardless of whether this regimen was combined with tamoxifen, compared with observation groups of premenopausal and postmenopausal women with breast cancer⁵⁰.

Some premenopausal patients with breast cancer receive ovarian function suppression combined with aromatase inhibitors, which further complicates the interpretation of metabolic changes. The loss of oestrogens associated with menopause or with ovarian function suppression, such as that achieved with gonadotrophin-releasing hormone agonists, is associated with body adipose tissue gain and with alterations in regional adipose tissue distribution⁵¹. During natural menopause, oestradiol levels drop, whereas follicle-stimulating hormone levels gradually rise. The increase in follicle-stimulating hormone could account for some adverse changes in body composition, but gonadotrophin-releasing hormone agonists that suppress both hormones are also associated with adipose tissue gain in premenopausal women⁵¹, suggesting a prominent role for oestrogens in maintaining a healthy body composition.

Several large breast cancer prevention trials have reported no differences in body mass after 12 months of treatment comparing tamoxifen with placebo, anastrozole with placebo, or anastrozole with tamoxifen, although, in each group, body mass was greater at 12 months relative to baseline⁵². Studies on SERMs and weight gain in the settings of treatment and prevention initially emerged in the 1990s^53–56, with some reporting greater overall body adipose tissue content and central adiposity after treatment, but inconsistent changes in total body mass^57,58. Aromatase inhibitor use is associated with a lower body adipose tissue percentage over time and with favourable changes in body adiposity when women switch from tamoxifen to aromatase inhibitors compared with those who stay on tamoxifen^32,33. Specifically, in women with overweight or obesity, lean mass was greater and adipose tissue mass was lower in those who switched from tamoxifen to exemestane compared with those who stayed on tamoxifen³³. In a small study evaluating body composition changes over 2 years in postmenopausal women with breast cancer, aromatase inhibitor use was associated with a gain in lean mass and stable adipose tissue mass compared with non-aromatase inhibitor use⁵⁹. However, the non-aromatase inhibitor group included women who received tamoxifen, toremifene (an SERM) or fulvestrant (an SERD), and these patients gained adipose tissue mass with no change in lean mass, which impacted body composition. A small case–control study in postmenopausal women who received aromatase inhibitors reported greater insulin resistance and body adipose tissue percentage, driven by statistically significantly lower lean body mass rather than higher adipose tissue mass, compared with age-matched and BMI-matched women who did not have breast cancer⁶⁰.

The effect of aromatase inhibitors on lean mass might result from elevated circulating levels of androgen that occur with aromatase inhibition, which attenuate muscle loss. In general, the studies on biologically important weight changes during endocrine therapy are limited by a lack of body composition data and participants matched by age, race/ethnicity, menopausal status, cancer stage, comorbidities and drug regimen. A 2012 review assembled data from 15 observational and 21 interventional trials that evaluated body composition during or after breast cancer treatment⁵⁷. The investigators concluded that body mass does not accurately depict potentially meaningful changes in lean or adipose tissue mass as a result of treatment, highlighting an area that continues to warrant investigation. Changes in body composition, especially gains in adipose tissue mass, can not only indicate a potential risk of T2DM but might also drive T2DM development.

T2DM risk in breast cancer survivors

Observations of excess diabetes mellitus diagnoses

In 2006, Lipscombe et al. aimed to determine whether hyperinsulinaemia and insulin resistance, which precede T2DM diagnosis by several years⁶¹, were associated with breast cancer risk⁶². They found that previous breast cancer was a substantially over-represented past event in people with newly diagnosed T2DM, initiating what has become a major area of scientific investigation. In a similar study of women with breast cancer risk variants in BRCA1 or BRCA2, T2DM diagnosis was statistically significantly more likely at or after the time of breast cancer diagnosis compared with women without breast cancer, particularly in women with an elevated BMI⁶³. When the treatment type was considered (that is, comparing those with T2DM with those without T2DM within the group of women who had breast cancer), only chemotherapy, but not tamoxifen, significantly predicted T2DM risk.

A 2022 report from The Pathways Heart Study, a National Cancer Institute supported cohort study in the US Kaiser Permanente healthcare system, substantiated and added to these earlier observations⁶⁴. In fully adjusted models, T2DM incidence was greater in people previously diagnosed with invasive breast cancer (n = 14,942) compared with control individuals (n = 74,702), matched for birth year and race/ethnicity, at all time points analysed (2, 5 and 10 years). In this cohort, however, the association with T2DM remained in subgroups (any chemotherapy, any left-sided radiation therapy or any endocrine therapy) compared with their control groups and was particularly evident in patients with normal weight versus those with overweight or obesity⁶⁴.

In 2012, a population-based study showed that in patients with breast cancer, current or recent past use, but not remote past use, of tamoxifen was associated with a greater risk of T2DM diagnosis, compared with patients with breast cancer not exposed to tamoxifen. This greater risk of T2DM was not seen with aromatase inhibitor use, but this patient group had less than 150 individuals⁶⁵. The investigators concluded that tamoxifen was unlikely to have a persistent or delayed effect on T2DM risk, but we now know that T2DM risk does persist, up to 10 years after treatment⁶⁴. Another short-term study found no association between aromatase inhibitor or tamoxifen use and T2DM diagnosis within 2 years of starting treatment⁶⁶. Each of these analyses was limited to postmenopausal women, but the control groups and data sources were different. The greater T2DM risk seen with current or recent (≤180 days) tamoxifen use was expressed relative to patients with breast cancer who did not use tamoxifen and was centred on population-based data in Ontario, Canada⁶⁵. By contrast, the null study compared individuals on endocrine therapy with matched cancer-free control individuals using SEER-Medicare data from the USA⁶⁶. In each setting, both the case and control study populations probably have different demographics that contribute to the discrepancies.

Together, the findings of more prevalent T2DM after breast cancer diagnosis and treatment could be simply explained by the more frequent medical monitoring experienced by patients with breast cancer or survivors, which would certainly increase the likelihood of T2DM diagnosis⁶³. However, another possibility is that something about breast cancer and its underlying risk factors, or its treatments, promotes T2DM development.

T2DM risk and specific endocrine therapies

In the past decade, substantial evidence has accumulated that endocrine therapy use is a risk factor for T2DM, especially when compared with women without cancer^67–70. In most studies, the relationship is strongest with tamoxifen, although toremifene was a notable risk factor in one study comparing individuals with breast cancer who used endocrine therapy with those who did not⁷¹. Raloxifene is approved for breast cancer prevention in high-risk women⁷², but compared with tamoxifen, less is known about the long-term metabolic complications^73,74. One study on healthy postmenopausal women reported impaired insulin sensitivity with raloxifene use, which is consistent with a risk of T2DM⁷⁵. The combination of bazedoxifene and conjugated oestrogens is being explored in the prevention setting⁷⁶, but long-term studies on T2DM risk are lacking^77–79. Although the addition of oestrogens to an SERM might sustain metabolic health, this strategy is unlikely to be adopted for women at high risk of or diagnosed with breast cancer without strong evidence of a benefit over current approaches.

The link between T2DM risk and aromatase inhibitor use is less robust than with SERM use, potentially because aromatase inhibitors are newer than tamoxifen and are typically prescribed to older women, with less long-term follow-up data to date. Several current reports show elevated T2DM risk with aromatase inhibitor use, but some studies did not reach statistical significance^68,69, whereas others did^70,80. In a 2022 meta-analysis, endocrine therapy exposure in patients with breast cancer was associated with a 30% excess T2DM risk compared with patients with breast cancer who did not use endocrine therapy and 19% excess compared with matched cancer-free control participants⁶⁹. The analysis included tamoxifen and aromatase inhibitors together and when evaluated separately tamoxifen showed the strongest association with T2DM risk⁶⁹.

One challenge with collectively interpreting epidemiological studies is the lack of standardized control groups. In the studies comparing individuals with breast cancer either using or not using endocrine therapy, interpreting a snapshot of T2DM risk is difficult. Patients with breast cancer who do not use endocrine therapy probably will have been diagnosed with ER-negative tumours and would therefore receive chemotherapy with corticosteroids, a known driver of glucose intolerance and weight gain, or therapy targeting the human epidermal growth factor receptor 2. Comparing patients with cancer on endocrine therapy with cancer-free control individuals does not take into consideration the complex link between T2DM risk and breast cancer risk in some people. Despite the differences in study design, 48% of T2DM occurring in breast cancer survivors was estimated to be attributable to endocrine therapy use⁸¹, indicating an urgent need for strategies to predict and prevent metabolic disease risk in this population.

T2DM development: potential mechanisms

Dysfunctional adipose tissue as a driver of T2DM

Diabetes mellitus is a complex, progressive disease that affects over 500 million people globally⁸². The majority (90–95%) of people with diabetes mellitus have T2DM, with a 2022 study reporting phenotypic and genetic subclasses⁸³. To better predict who is at risk of T2DM after breast cancer therapy, we need to understand how T2DM develops in the first place. Several paradigms can explain its aetiology in adults^84–86. One model, the portal hypothesis, is based on the observation that visceral or central adiposity is associated with obesity complications that lead to T2DM. Lipolysis in visceral adipose tissue is accompanied by elevated levels of free fatty acids in the portal vein that impair glucose oxidation, promote hepatic gluconeogenesis and inhibit insulin action, according to the Randle hypothesis⁸⁷. This model would predict that hyperglycaemia occurs before insulin resistance and hyperinsulinaemia, but this scenario is not always the case⁶¹. Patients on endocrine therapy have excess visceral adiposity and elevated circulating levels of TAGs or free fatty acids in some studies, exacerbated by overweight or obesity^30,31,44. Besides visceral adipose tissue, overall and subcutaneous adipose tissue mass are independently associated with T2DM risk. The subcutaneous depot does not drain into portal circulation, suggesting that excess adipose tissue in these regions might influence T2DM progression through other mechanisms. The quality of adipose tissue has not been thoroughly analysed after endocrine therapy and could also contribute to T2DM development.

Another paradigm is the endocrine hypothesis, which centres on the role of adipose tissue as an endocrine organ and not simply a storage site for excess energy^84,85. This model postulates that hypertrophic adipocytes secrete distinct signalling hormones and cytokines compared with small, healthy adipocytes. Large adipocytes might also recruit macrophages and promote a condition of sustained chronic inflammation that can contribute to the altered secretome of adipose tissue^88,89. A shift in the production of leptin and adiponectin or in pro-inflammatory cytokines can affect other peripheral organs and promote insulin resistance. Oestrogens and endocrine therapies have immune-modulating effects, but these have primarily been studied in the context of tumour immunity⁹⁰. In a short-term study of premenopausal and postmenopausal patients with breast cancer, both tamoxifen and aromatase inhibitor use were associated with elevated circulating levels of leptin and adiponectin, but a lower leptin–adiponectin ratio compared with baseline measurements⁹¹, which is a favourable change. Further research will show if breast cancer treatments affect the endocrine function of adipose tissue, but to date the existing literature does not support a role for harmful adipose-derived cytokines in T2DM development after breast cancer treatment.

A third paradigm is the ectopic lipid storage syndrome, based on excess lipid accumulation in muscle, liver and pancreas that occurs with lipodystrophy, or the failure of adipose tissue to adequately store excess nutrients^84,85. When this scenario occurs, adipose tissue exceeds its capacity and excess lipids are deposited ectopically in peripheral tissues. Ectopic lipid accumulation is observed in individuals with severe adipose tissue loss or during a chronic positive energy balance when adipose tissue expansion is limited to hypertrophy as the only mechanism⁹². Thiazolidinediones, a class of drugs used to treat diabetes mellitus, activate peroxisome proliferator-activated receptors to promote adipogenesis, thus improving glucose control despite driving weight gain in some people and providing evidence that adipose tissue expansion can be metabolically healthy⁹³. Healthy adipose tissue expansion can fail for several reasons. A lack of proliferation or differentiation of adipocyte precursor cells would eventually exhaust the potential for new adipocyte formation during a chronic positive energy balance⁹². Failure to oxidize fat in peripheral tissues could also contribute to lipid accumulation. Indeed, one of the earliest reported adverse effects of tamoxifen was MAFLD, but we speculate that the timing of the development of hepatic steatosis in people taking tamoxifen might indicate acute hepatotoxicity from the drug rather than ectopic lipid deposition from chronic adipose dysfunction.

Preclinical studies on endocrine therapy: insights and challenges for clinical translation

Similar to epidemiological data, the reports on body weight and metabolic changes in rodents treated with endocrine therapies are somewhat inconsistent. Tamoxifen is given to rats or mice either as an injection in oil vehicle or as an implanted pellet, whereas it is given orally to people. Modelling aromatase inhibitor therapy also presents challenges. Aromatase inhibitors are given to women whose ovaries are not functioning, owing either to natural menopause or to active ovarian function suppression. Although many preclinical studies use aromatase inhibitors in rodents, research is lacking that evaluates the peripheral metabolic effects of aromatase inhibition in ovariectomized mice or rats, which would reveal roles for extragonadal oestrogen synthesis. The regulatory regions of CYP19A1 are distinct in mice and humans, resulting in tissue-specific expression profiles. Whether aromatase is expressed or functional in mouse adipose tissue after ovarian oestrogen production ceases remains unclear^11,94. Regardless, oestrogen deprivation is one tool to mimic the severe loss of hormone ligand that women experience with aromatase inhibitor treatment⁹⁴.

A single injection of tamoxifen in neonatal mice was sufficient to support weight gain induced by a high-fat diet (HFD) in adult female mice many weeks later, but this effect was not seen in males⁹⁵. Glucose tolerance was better in tamoxifen-treated male mice, but worse in females compared with their respective controls, and liver lipid metabolism was unaffected⁹⁵, which potentially indicates a role for the brain in mediating the effect of early tamoxifen exposure on obesity predisposition. However, food intake was not altered in adult mice regardless of sex. In some studies, acute tamoxifen treatment in mice leads to rapid but transient lipoatrophy, followed by adipose tissue regain^96–100. Mechanistically, weight regain after tamoxifen-induced lipoatrophy can result from de novo adipogenesis and not just mature adipocyte hypertrophy that could lead to MAFLD⁹⁹. However, in isolated human subcutaneous adipose stromal cells, tamoxifen treatment inhibited proliferation and impaired adipogenic differentiation¹⁰¹. The doses of tamoxifen used in rodents substantially exceed what patients with breast cancer receive¹⁰² and whether people experience transient adipose lipolysis shortly after beginning endocrine therapy is unknown. Tamoxifen can promote browning of adipose tissue in rodents, which should increase fat oxidation^96,100; however, adipose tissue browning has not been reported in people taking SERMs or aromatase inhibitors.

One study in HFD-fed ovariectomized female mice showed that tamoxifen alone prevented adipose tissue gain, improved glucose tolerance and prevented hepatic steatosis; effects that were not explained by decreased food intake but that were mediated by ERα¹⁰³. Separately, we evaluated ovariectomized HFD-fed and low-fat diet-fed female mice treated with tamoxifen plus oestradiol or with oestrogen deprivation, compared with oestradiol-supplemented controls¹⁰⁴. The rationale was based on observations that T2DM and associated sequelae might be more common in women with overweight or obesity^{41–43,80,81,105}. No effects of tamoxifen were seen in lean females outside the uterus, and findings in mice that are obese contrasted with the study described at the beginning of this paragraph¹⁰³. That is, chronic exposure (7 weeks) of female mice to tamoxifen augmented weight gain and adipose accumulation and impaired glucose tolerance¹⁰⁴. In this study, tamoxifen was given at ~3 mg/kg per day and resulted in serum levels of ~75 ng/ml in mice, which is lower than what is measured in women (150–200 ng/ml)¹⁰⁴. Short-term effects (2 weeks) of treatment on adipose tissue stromal cells obtained from the mice included more preadipocytes but fewer adipocyte progenitor cells present compared with oestrogen-treated mice, which is consistent with the initiation of adipogenesis, but treated animals did not show hepatic steatosis at 2 weeks. By the end of the study, hepatic steatosis and adipocyte hypertrophy were greater in tamoxifen and oestrogen deprivation-treated (a model of aromatase inhibition) mice compared with oestrogen-treated controls. The effects of tamoxifen or oestrogen deprivation on adipose tissue were only seen in mice that were obese. The development of metabolic phenotypes in this model is consistent with the ectopic lipid deposition paradigm that suggests impaired adipose tissue expansion as an early effect of endocrine therapy. They also align with reports of weight gain, glucose intolerance and MAFLD in patients. These studies were both done in ovariectomized females but differed in the presence of opposing supplemental oestradiol^103,104. As oestrogens might modulate the agonist properties of tamoxifen in adipose or any other tissues, the experimental design for mouse studies has implications for how women of different ages respond metabolically to tamoxifen and could help inform who is at risk of T2DM.

The different reported effects of tamoxifen on mouse metabolism could be explained by housing temperature. At room temperature, tamoxifen without supplemental oestrogen (that is, in ovariectomized mice) acted as an ER agonist, promoting an overall healthy metabolic phenotype compared with ovariectomized controls¹⁰³. At thermoneutral temperatures (28–32 °C), tamoxifen in the presence of oestradiol in ovariectomized mice promoted dysmetabolism compared with oestradiol-only-treated ovariectomized controls; however, this effect was not as severe as that seen in ovariectomized mice without tamoxifen or oestradiol (that is, in oestrogen-withdrawal-treated mice)¹⁰⁴. When given alone to ovariectomized mice at thermoneutrality, tamoxifen had no effect on body mass, composition or fasting glucose and insulin¹⁰⁴. In some studies, mice with tamoxifen-induced adipose tissue loss and regain have more brown adipose tissue and elevated levels of UCP1 (refs. 96,97,100,102). As this effect has not been reported in humans, it could be an artefact because the weight regain occurs in experimental settings of relative hypothermia that enhances brown adipose tissue activity. Optimal housing temperature for rodents continues to be debated but could considerably influence the metabolic outcomes in preclinical studies, making the translation of potential mechanisms to humans difficult.

ER regulation of metabolic homeostasis

Oestrogens regulate whole-body metabolism through central and peripheral mechanisms. In rodents and humans, circulating levels of oestrogen inversely associate with food intake and body mass and directly associate with glucose tolerance¹⁰⁶, so it is no surprise that endocrine therapy has long-term metabolic consequences in breast cancer survivors. The studies covered in this Review include SERMs and aromatase inhibitors that have been used for decades, but even this year new drugs have been approved for clinical use for the management of breast cancer. Targeting ERs to prevent breast cancer growth has been highly successful and will remain a strategy for the future. As we develop new therapies, we might be able to predict adverse metabolic outcomes as we learn more about the biological roles of ER in metabolism.

The ER isoforms, including ERα, ERβ and GPER1, are each expressed in peripheral tissues, such as the liver, skeletal muscle, pancreas and adipose tissues, with different relative levels of gene expression in each¹⁰⁷ (Fig. 3). Esr1-knockout mice, Cyp19a1-knockout mice and ovariectomized female mice all have similar phenotypes characterized by excess adiposity, hepatic steatosis and impaired glucose tolerance that worsens with age^108–111. Oestrogen replacement can improve most of these disruptions, and many effects depend on ERα¹¹². Mice lacking GPER1 have greater adiposity and impaired insulin secretion compared with wild-type mice, similar to those lacking ERα^113,114. Sexually dimorphic effects on glucose homeostasis and hepatic steatosis occur with GPER1 loss, with females showing a more impaired phenotype¹¹³. In female mice, ERα signalling in skeletal muscle maintains glucose tolerance and insulin sensitivity^115–117, suppresses triglyceride synthesis¹¹⁸ and might participate in the maintenance or differentiation of progenitor cells^119–121. Both GPER1 and ERβ can also regulate whole-body insulin sensitivity through effects on skeletal muscle that might depend on the relative abundance of each protein, which can be influenced by age and sex^122–125. In the pancreas, oestrogen protects β-cells from death, associates with increased insulin content and release and suppresses lipid synthesis; effects that can be mediated by each ER^126–132. In mice, tamoxifen counteracts the effects of oestradiol in the pancreas, potentially through ERα antagonism¹²⁹.

Fig. 3 |. ER expression in peripheral metabolic tissues.

The expression levels of genes encoding oestrogen receptors (ERs) in humans vary across peripheral metabolic tissues. Gene expression levels of ESR1 (encoding ERα), ESR2 (encoding ERβ) and GPER1 (encoding GPER1) are shown in the liver, pancreas, adipose tissue and skeletal muscle in humans. Processed data of ER expression across multiple human tissues, reported as fragments per kilobase of transcript per million mapped reads (FPKM), were obtained from the ArrayExpress (accession E-MTAB-1733)¹⁰⁷ and used to make heatmaps in GraphPad Prism 9.

ERs also regulate metabolism through central mechanisms. ERα signalling in the mouse hypothalamus regulates food intake, energy balance and glucose tolerance¹³³, and hypothalamic ERα influences locomotion, bone density and thermoregulation^134,135. In the hypothalamus, tamoxifen acts as an ER agonist^136,137. The effect of endocrine therapy in the brain could contribute substantially to metabolic changes experienced by patients with breast cancer during treatment, but almost no data are available on how SERMs or aromatase inhibitors influence eating behaviour in people, especially with short-term exposure. The long-term effects in patients with breast cancer, including the elevated risk of T2DM, are not consistent with a role for tamoxifen as an ERα agonist in all tissues. Importantly, tamoxifen, raloxifene and fulvestrant can act as GPER1 agonists¹¹⁴. In this way, studying endocrine therapy as one entity, and not considering the individual types of therapy, might result in specific mechanistic information being overlooked.

ER signalling is well known to affect liver and adipose tissue function; tissues that can influence each other. Oestrogen loss in mice promotes MAFLD through direct effects on hepatic lipid synthesis¹³⁸. In ovariectomized female mice lacking liver ERα expression, oestrogen reduces adiposity but does not improve hepatic steatosis or insulin resistance¹³⁹, indicating a cell-autonomous, protective role for oestrogen and ERα. However, liver lipid accumulation in HFD-fed female mice is also mediated by ERα transcriptional activity in non-hepatic tissues¹⁴⁰. Acute tamoxifen treatment is reported to increase MAFLD in male mice, through direct effects on hepatocytes^141,142.

As discussed earlier, tamoxifen and oestrogen deprivation have profound effects on adipose tissue biology, which can influence hepatic lipid deposition. A 2023 study showed that adipose precursor cell hyperplasia in female mice was affected by oestrus stage and the timing of HFD initiation¹⁴³. Adipose expansion by precursor cell hyperplasia helps prevent adipocyte hypertrophy, accompanying lipolysis and ectopic lipid deposition. Proliferation of adipocyte precursor cells in HFD-fed mice was highest when the diet was introduced in proestrus, when oestrogen levels peak. Effects were cell autonomous and mediated by ERα, which is consistent with previous studies showing a link between oestrogen and depot-specific precursor proliferation in male and female mice¹⁴⁴. Importantly, the timing of diet and oestrogen exposure influenced weight gain in mice later in life, with greater adipose tissue accumulation seen in females who began HFD feeding in proestrus compared with metestrus or dioestrus¹⁴³. This finding raises intriguing questions about whether the timing of endocrine therapy administration, for example, in the luteal versus follicular phase of the menstrual cycle, affects the long-term risks for adverse metabolic effects in women with breast cancer.

Conclusions

For the past several decades, targeted therapies for breast cancer have markedly improved patient prognosis and they remain life-saving treatments to this day. As with most drugs, potential adverse effects exist. For endocrine therapies, the metabolic adverse effects include dyslipidaemia, hepatic steatosis, body composition changes and T2DM development, but we still do not know who is at risk. For example, do elevated levels of TAGs and hepatic steatosis, which occur relatively soon after starting tamoxifen, represent canaries in the coal mine for later T2DM risk in breast cancer survivors? Supporting evidence for this hypothesis would require longitudinal studies that follow women for the duration of their treatment and potentially for decades afterwards, analysing blood and body composition along the way. Can we identify early circulating biomarkers of later T2DM risk, such as hyperinsulinaemia? The ideal scenario for preventing T2DM in breast cancer survivors involves an integrated care team with oncology and endocrinology expertise that can identify potential problems early during treatment, helping patients with breast cancer gain the most benefit from their therapies.

Mechanistically, endocrine therapies probably promote T2DM development through complex effects on multiple tissues (Fig. 4). Preclinical data suggest that adipose tissue is an early target of endocrine therapy, particularly given the role of ER in mature adipocytes and precursor cells. Disrupting ER function or oestrogen production would be expected to promote adipocyte hypertrophy and impair hyperplastic adipose expansion. This effect could elevate circulating levels of TAGs and free fatty acids that might be stored in the liver, muscle or pancreas; these effects would attenuate insulin sensitivity and potentially insulin secretion on their own. However, ERs also have cell-intrinsic roles in each of these tissues that would compound the detrimental effects of endocrine therapy on whole-body metabolism (Fig. 4). Future challenges include defining the precise effects of breast cancer treatments throughout the body and identifying interventions that can be offered to patients with breast cancer to prevent later risk of T2DM.

Fig. 4 |. Proposed model of adipose tissue as an early target of endocrine therapy.

Oestrogens, through oestrogen receptors (ERs), maintain metabolic heath by acting directly on adipose, liver, muscle and pancreas. One acute effect of endocrine therapies, such as aromatase inhibitors, selective ER modulators (SERMs) or selective ER degraders (SERDs), is impaired adipose tissue expansion, which might contribute to elevated circulating levels of fatty acids and dyslipidaemia, hepatic steatosis, insulin resistance and hyperinsulinaemia. Together, these effects might culminate in type 2 diabetes mellitus for some women after breast cancer treatment.

Key points.

• Endocrine therapies for breast cancer might increase the risk of type 2 diabetes mellitus (T2DM) development in some patients.

• Oestrogens and oestrogen receptor activation protect against metabolic disease and are disrupted with breast cancer treatment.

• Tamoxifen treatment promotes dyslipidaemia and hepatic steatosis in some people and also has adipose-specific effects in preclinical and clinical studies.

• One paradigm of T2DM development centres on dysfunctional adipose tissue expansion.

• Preclinical studies indicate that adipose tissue might be an early target of endocrine therapies for breast cancer.

• Endocrine therapies save lives, so an urgent need exists to understand any associated T2DM risk and offer interventions for patients with breast cancer.