Pilot Study of Daily Exemestane in Women with Endometrial Intraepithelial Neoplasia or Low-Grade Endometrial Cancer.

Forty participants accrued to the study between January of 2018 and March of 2021 ( Figure 1 ). The mean age was 63 (SD 6.4) Thirty-nine participants (97.5%) self-identified as White, one participant (2.5%) identified as Asian. One participant (2.5%) identified as Hispanic/Latino; the remaining 39 participants identified as Not Hispanic/Latino (97.5%). The mean BMI of study participants was 40.3 (SD 9.8, range 22.8-60.5) ( Table 1 , Supplementary Table 1 ). At baseline, 65% of participants had cardiovascular disease or risk factors (most commonly, hypertension, coronary artery disease, hyperlipidemia), 72.5% of participants had endocrine/metabolic disorders (most commonly pre-diabetes, diabetes mellitus, hypothyroidism), and 85% of participants had musculoskeletal disorders (most commonly osteoarthritis, carpel tunnel syndrome, fibromyalgia). Preoperative diagnoses included CAH/EIN (n=11, 27.5%), or grade 1 (n=26, 65.0%) and grade 2 (n=3, 7.5%) endometrial adenocarcinoma. Thirty-eight participants had evaluable specimens for both pre- and post-treatment Ki-67 analysis ( Figure 1 ). There was no tissue available for analysis for two participants. Median Ki-67 score decreased from 40.7% [IQR (33.9, 50.3)] at baseline to 18.1% [IQR (8.8, 31.8)] after exemestane treatment ( Figure 2A / B ). This represents a median absolute change from baseline of −20.4% [IQR (−29.9, −6.7), p<0.001). In total, percentage of Ki-67 in tumor cells declined in 31 participants and increased in seven participants. There were no outstanding features identified for the participants in whom Ki-67 increased. Specifically, compliance with dosing, tumor grade, BMI, and ER/PR positivity in those with increased Ki-67 was similar to that of the entire cohort ( Supplementary Table 2 ). Age and BMI were not associated with change in Ki-67. Pre-operative diagnosis was correlated with Ki-67 decrease, with greater decreases seen in participants with CAH/EIN and grade 1 tumors compared to grade 2 tumors (r=0.38, p=0.019, Figure 2C ). The historical control group also demonstrated a decrease in Ki-67 score between preoperative biopsy and surgical specimen with a median absolute change from baseline of −6.7% [IQR (−12.7, −1.3), p<0.001]. However, the decrease in Ki-67 was greater in the study participants than the matched historic controls, with a median difference between the groups of −13.4% [IQR (−23.3, 6.9), p≤0.01]. The majority of participants (n=27, 71%) had the same histologic diagnosis at the time of enrollment compared to the final histologic diagnosis at the time of surgery. Nine participants (22.5%) had an upgrade in pathology (for example, CAH/EIN to grade 1) including four participants with grade 3 on final histologic diagnosis. Two participants (5%) had a downgrade in pathology (for example, grade 1 to CAH/EIN). In the historical control cohort of 40 patients, there were seven patients (17.5%) who had an upgrade in pathology, and one patient (2.5%) who had a downgrade in pathology ( Supplementary Table 3 ). Median serum estradiol level prior to treatment was 26 pg/mL (range 5-200). Serum estradiol levels did not change significantly between baseline and post-treatment ( Figure 3 A / B ). BMI positively correlated with pre-treatment (r=0.56, p=0.001) and post-treatment (r=0.451, p=0.004) estradiol levels, but there was no association between BMI and change in estradiol levels ( Supplementary Figure 2 ). Baseline estradiol levels did not show a significant correlation with changes in Ki-67 (r=−0.265, p=0.113) ( Supplementary Figure 3 ). Because estradiol measurement was performed locally, we considered that there may be variability between assays that may contribute to our findings. We analyzed baseline, post-treatment, and change in estradiol at each site and there was no significant decline in estradiol at any of the 3 sites ( Supplementary Figure 4 ). There was a small but statistically significant increase in median serum progesterone from baseline to post-treatment (median change of 0.15 (IQR 0.10 – 0.25) , p<0.001) ( Figure 3C ). Serum exemestane was detected at the time of surgery in 38/39 participants, with a median level of 2.8 ng/mL (IQR 1.8-3.1). The one participant in whom exemestane was not detected was not compliant with dosing. One participant did not have a post-treatment blood sample collected for analysis. All tumors were estrogen receptor (ER) and progesterone receptor (PR) positive. After exemestane treatment both ER and PR levels decreased significantly ( Figure 4 ). Specifically, median ER H-score dropped from 175.6 [IQR (−159.5,− 235.9)] to 128.0 [IQR (−74.9, −169.5)] representing a median absolute change of −72.7 [IQR (−110.3, −21.4)] (p<0.001). Median PR H-score dropped from 225.9 [IQR (−181.4, −268.0)] to 76.9 [IQR (−30.7, −134.5)]) representing a median absolute change of −115.0 [IQR (−179.4, −80.3)] (p<0.001). Baseline ER and PR H-score as well as change in ER and PR H-score did not correspond to change in Ki-67 ( Supplementary Figures5 and 6 ). Thirty-nine participants (97.5%) were compliant with exemestane dosing, defined by completing at least 80% of study medication. The one participant who was non-compliant did not receive enough study drug. All participants completed the study and no participant discontinued or interrupted therapy due to treatment-related adverse events. Adverse events were experienced by 38 participants (95%). Nine participants (22.5%) had grade 3 events: eight participants (20%) with grade 3 hypertension and one participant (2.5%) with grade 3 hyperhidrosis ( Table 2 ). Of the eight participants who experienced grade 3 hypertension, all eight had baseline hypertension (six participants with grade 2 baseline hypertension and two participants with grade 1 baseline hypertension). No participants experienced grade 4 or 5 adverse events and the side effects in this study were consistent with the known adverse event profile of exemestane. The most common adverse events included hot flashes (53%), hypertension (40%), diarrhea (25%), fatigue (23%), abdominal pain (20%) and headache (20%).

This was a multi-center, phase IIA, single-arm ‘window of opportunity’ pilot study of 25 mg per day of exemestane for 21 to 42 days in individuals undergoing planned hysterectomy for EIN or low grade endometrial cancer ( NCT03300557 ). The study was approved by the Central IRB of the National Cancer Institute (UWI2016-08-01). Participants were recruited from 3 sites (University of Wisconsin-Madison, University of Alabama at Birmingham, and University of Minnesota). Individuals were eligible if they had biopsy proven CAH/EIN, or low grade (grade 1 or 2) endometrioid endometrial adenocarcinoma. Other eligibility criteria included Eastern Cooperative Oncology Group (ECOG) performance status ≤1, body mass index (BMI) >20, no prior treatment for EIN or endometrial cancer, no systemic hormonal therapy, and no suspicion for advanced disease. Participants were age 45 or older and postmenopausal. To ensure postmenopausal status, participants who were 45-55 years old or had less than 2 years of amenorrhea underwent measurement of follicle stimulating hormone (FSH), which had to be within the institutional post-menopausal range. The study period was limited to the time between the initial pre-surgical visit and definitive surgery (hysterectomy). At the pre-surgical visit, an endometrial pipelle biopsy was collected during the standard pelvic exam. Participants then began oral exemestane daily, 25 mg/day, for at least 21 days (but not longer than 42 days). Participants were contacted on day 15 of therapy as well as the day before surgery for determination of adverse events and other symptoms. The study utilized the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) v4.0 to assess adverse events. At the time of surgery, a portion of endometrial tumor was collected for post-treatment analysis ( Supplementary Figure 1 ). The study was non-randomized and not blinded. The primary objective was to determine if there is a decrease in proliferation index, measured by percent Ki-67 expression, in CAH/EIN or grade 1/2 endometrial cancer cells from baseline to post-treatment. Tissue collected from the endometrial pipelle biopsy at baseline and the surgical specimen were fixed in 10% neutral buffered formalin and processed into paraffin blocks. Immunohistochemical (IHC) staining and analysis of pre- and post-treatment tissue samples was performed by the University of Minnesota BioNet Histology and IHC Laboratory, and the Digital Imaging Facility. Unstained slides were de-paraffinized and rehydrated using standard methods. For antigen retrieval, slides were incubated in pH 6.0 buffer (Reveal Decloaking reagent, Biocare Medical, Concord, CA) in a steamer for 30 min at 95-98°C, followed by a 20 min cool down period. Slides were incubated with Proteinease K for 5 minutes at room temperature, rinsed well and place into Tris Buffered Saline, 0.05% Tween 20 (TBST). Subsequent steps were automated using an immunohistochemical staining platform (Nemesis 7200, Biocare). Endogenous peroxidase activity was quenched by slide immersion in 3% hydrogen peroxide solution (Peroxidazed, Biocare) for 10 min followed by TBST rinse. A serum-free blocking solution (Background Punisher, Biocare Medical, Concord, CA) was placed on sections for 10 min. Blocking solution was removed and slides were incubated in primary antibody to Ki-67 (ThermoFisher; diluted 1:300 in 10% blocking solution/90% TBST followed by TBST rinse), ER (Cell Marque, diluted 1:50), and PR (Cell Marque, diluted 1:500). This study used the Visiopharm Integrator System (Horsholm, Denmark, version 2022.09.03) for image analysis. Initially, an application was built to distinguish cells based on nuclear coloration. Then, 3 mm sections of endometrial tumor/atypia were manually selected. The application was used on these sections to calculate the total number of tumor/atypical cells, quantity and percentage of Ki-67 positive and negative cells, and the Ki-67 index. ER and PR were manually quantified using the H-score which was calculated as follows: ( 1 x percentage of weak staining ) + ( 2 x percentage of moderate staining ) + ( 3 x percentage of strong staining ) . Results of Ki-67 were compared with a historic control cohort of tumors from patients obtained through the Department of Laboratory Medicine and Pathology at the University of Minnesota. The historic samples (no more than 5 years old) were matched by preoperative histologic diagnosis, age (within 5 years) and BMI category (20-24, 25-30, >30). Specimens from the historic control cohort included a preoperative endometrial biopsy as well as a surgical specimen, separated in time by less than 42 days. This analysis was performed to determine if Ki-67 changes could be explained by variation in sampling technique (pipelle biopsy versus surgical specimen) or time between sample collection. Exemestane was quantified in plasma by LC-MS/MS using a Waters Acquity ultra-performance liquid chromatography (UPLC) system coupled to an AB Sciex Q-Trap 5500 triple quadrupole mass spectrometer. An eight-point standard curve (1 to 80 ng/ml) was prepared in plasma. Samples (200 μl) of participant plasma, standards, and quality controls were processed on a Waters’ Ostro 96-well plate, using a positive-pressure manifold after addition of three volumes acetonitrile with 1% formic acid containing the internal standard norgestral. Following processing, samples were dried under nitrogen and resuspended in 150 ml 50% acetonitrile/50% water/0.1% formic acid. Samples were separated on a Phenomenex XB C18 column (100 x 2.1 mm, 1.7 μm particle size) using an increasing gradient of solvent B (0.1% formic acid in acetonitrile) to solvent A (0.1% formic acid in MilliQ water). The gradient started at 65% B and increased to 80% B by 1.6-min, then to 98% B by 1.8 min, with a hold to 2.4 min, then back to 65% B by 3 min with a hold for 0.5 min. The column temperature was 28°C and the flow was 0.35 ml/min. Samples were analyzed with triplicate injections on the LC-MS/MS in positive ion ESI mode. Gas flow, gas pressure, temperatures and voltages were optimized for each compound. All dwell times were 50 ms. Transitions for exemestane were 297 to 93, 121, and 149. Transitions for the internal standard norgestral were 313 to 91, 109, and 245. Quantitation was based on sum of transitions for exemestane. Data analysis was performed using MultiQuant version 3.0.3 software (AB Sciex), fitting the data to a quadratic model with 1/x 2 weighting. The lower level of quantification (LOQ) was 0.5 ng/ml for exemestane. Pretreatment and posttreatment serum estradiol and progesterone were measured locally at each institution. Post-treatment levels were measured the day of surgery (within 24 hours of last exemestane dose). Estradiol assays and platforms were as follows: UAB used a paramagnetic particle chemiluminescent immunoassay (Beckman DXI; Brea, CA) with a post-menopausal reference range of <30 pg/mL, UMN used a competitive electrochemiluminescence immunoassay (Roche; Basil, Switzerland) with a post-menopausal reference range of <5-138, and UWI used a chemiluminescent microparticle immunoassay (Abbott Alinity; Lake Forest, IL) with a post-menopausal reference range of <10-28. Progesterone assays and platforms were as follows: UAB used liquid chromatography/mass spectrometry (Quest Diagnostics, Secaucus, NJ) with a post-menopausal reference range of ≤ 0.2 ng/mL, UMN used a electrochemiluminescence immunoassay (Roche; Basil, Switzerland) with a post-menopausal reference range of ≤ 0.1 ng/mL, and UWI used a chemiluminescent immunoassay (Abbott Alinity; Lake Forest, IL) with a post-menopausal reference range of <0.1-0.2. The change in absolute Ki-67 and difference in change between study participants and matched historic controls were evaluated using the Wilcoxon signed-rank test. Tissue ER, PR and serum were progesterone and estrogen levels were also compared using the Wilcoxon signed-rank test. We estimated Ki-67 changes based on prior research of aromatase inhibition in endometrial adenocarcinoma. Additionally, we estimated that half of participants would have CAH/EIN and may have a lesser treatment effect compared to participants with adenocarcinoma. With a sample size of 30, there will be power of 0.8-0.9 to detect a mean decrease ranging from 3.5%−4.1% according to a one-sided level 0.05 test. We estimated 25% of participants would either not complete treatment or not have sufficient tissue for analysis and therefore planned recruitment was 40 participants. The data generated in this study are publicly available via the Cancer Data Access System (CDAS) site ( https://cdas.cancer.gov/learn/eppt/browse/uwi2016-08-01/?key=nWFhLQ6-bFggWpA8IfW4z-SOj7v2golAhB3a3qRCzY8 ) Additional queries, including for the raw data behind the figures, can be directed to the corresponding author.

Endometrial cancer is a major global public health concern, given the rising incidence and death from disease. With no screening test available, early symptom recognition is paramount to detecting this disease at an early stage. After decades of no new drug approvals, there are finally new FDA-approved therapies and the number of clinical trials exploring targeted therapy are at an all-time high. However, given the epidemiologic trends and challenges associated with surgery and systemic therapy, prevention is an appealing alternative to the current focus on invasive disease. In this single arm, pre-surgical window of opportunity pilot study, oral exemestane demonstrated anti-proliferative effects on the endometrium as measured by decrease in Ki-67 expression. Additionally, tumor ER and PR levels declined significantly. Exemestane was well tolerated with all participants completing therapy and only one participant demonstrating non-compliance with study drug. Aromatase inhibitors have a biologic rationale in treating endometrial cancer. They can be used in the palliative setting to treat metastatic disease ( 20 , 21 ) and small studies have demonstrated efficacy in pre-operative windows similar to this study ( 20 , 21 ). Specifically, in a study of 12 postmenopausal patients with low grade endometrial cancer who received letrozole pre-operatively, 41% had significant reduction in Ki-67 ( 22 ). The proposed mechanism of anti-tumor activity is through decreased circulating estradiol, which is thought to be a primary driver of endometrial carcinoma and EIN. In our study, BMI levels did correlate with baseline and post-treatment estradiol levels, which is consistent with the existing literature regarding the correlation between obesity and circulating estrogen. Surprisingly, we did not demonstrate reduction in estradiol in our participant population. In this study, we used the standard clinical assay at each of the three participating sites to measure baseline and post-treatment estradiol and progesterone levels. Although centralized testing was not performed, each participant had both their pretreatment and posttreatment serum hormone levels tested at the same local laboratory. When we evaluated estradiol levels at each site, no site’s patient cohort demonstrated a significant decrease in estradiol. Additionally, none of the three sites utilized ultrasensitive assays (that is, assays able to detect and distinguish very low levels of serum estradiol) which was a limitation of this study. However, there were very few participants whose estradiol were below 10 pg/mL (2 participants pre-treatment and 4 participants post-treatment). Therefore, it is unlikely that an ultrasensitive assay would have detected a decrease in estradiol in our study cohort. The lack of estradiol decrease contrasts with the results of a pre-surgical exemestane study in participants with breast cancer ( 23 ). This study demonstrated the non-inferiority of three times weekly exemestane dosing compared to daily exemestane in maintaining estradiol suppression while still decreasing breast cancer proliferation (measured with a Ki-67 index). In this trial, investigators used a ultrasensitive liquid chromatography tandem mass spectrometry (LC-MS/MS) based platform to measure serum estradiol, and they demonstrated decrease in serum estradiol in both obese (BMI >30) and non-obese participants. However, mean baseline estradiol levels were markedly lower in this population compared to our study. Additionally, in our study, mean BMI was 40.3 (IQR 34-49) and 85% of subjects were obese. In the breast cancer study, mean BMI was 28 (IQR 23.9-32.6) and only 33% of participants were obese. One possible explanation for lack of serum estradiol modulation in our study is that the dose of exemestane in this study (25 mg PO daily) could be insufficient to block estradiol production in the setting of such high BMIs, which translate to higher baseline estradiol levels. Available data on the pharmacokinetics of exemestane in postmenopausal women is limited to patients with breast cancer, where the BMI was significantly lower than in the current study ( 24 - 26 ). Further studies to optimize the dose required for sustained estrogen blockade in this patient population are needed. Nevertheless, a systemic effect on lowering estradiol cannot be invoked as the mechanism of action for decreased proliferation by exemestane in the current study. It is therefore possible that the anti-proliferative effects observed may be due to local tissue effects of aromatase inhibition. Aromatase is not expressed in normal endometrial tissue, although it may be overexpressed in endometriosis as well as in endometrial cancer ( 27 - 30 ). Thus, there is a possibility for blocking paracrine signaling as a mechanism of action, although we were not able to evaluate this specifically in this study. Notably, we did find a significant decrease in tissue expression of PR and ER. Typically, the mechanism of decreased ER and PR expression after aromatase inhibition is thought to be due to decrease in circulating estradiol, which blocks ER and subsequent PR transcription ( 31 , 32 ). However, given we did not see a decline in circulating estradiol, we hypothesize that potentially local aromatase inhibition could affect steroid receptor expression. Endometrial cancer cannot always be treated with surgery. Individuals with obesity (who are among the most likely to develop this condition) often have medical co-morbidities including heart disease, diabetes mellitus, and obstructive sleep apnea, that make the safe administration of anesthesia and completion of surgery challenging and high-risk. Non-surgical approaches for treating cancer can be invaluable for these patients. While progestins can be effective for some patients, they can have significant side effects and do not always provide a durable response. Given the positive results of this study, aromatase inhibition should be studied further as a prevention strategy (potentially in combination with progestins) for post-menopausal individuals with EIN or endometrial cancer who are not surgical candidates. One major limitation to this study is the lack of diverse participant enrollment. Black individuals have been markedly under-represented in clinical trials ( 33 ) including the present study. This has implications regarding generalizability and could potentiate existing disparities. Lack of diverse enrollment was identified mid-study and we made efforts to improve non-white participation including opening the trial at a site with more Black patients, meeting with research coordinators to discuss communication methods, and ensuring adequate pre-screening efforts. Notably, this study was intensive for the participants. It required a repeat baseline biopsy in clinic and taking a medication without known direct benefit. It is possible that given this risk versus benefit ratio, individuals with previously negative health care experiences due to structural racism may have been less likely to participate. Future studies will need to incorporate diverse patient advocacy earlier on in study development, and efforts will be made to have a more diverse clinical research staff. Other limitations of this study include a relatively small cohort with historic controls rather than contemporaneous controls. Given the global burden of endometrial cancer, the development of prevention strategies is imperative. Exemestane demonstrated efficacy in this pre-surgical window of opportunity trial, and therefore warrants study in a larger population as a cancer prevention agent.

Despite improvement in the understanding of risk factors, molecular subtypes, and pathogenesis, endometrial cancer incidence is increasing in the U.S. and worldwide ( 1 - 3 ). In 2024, it is estimated that there were over 67,000 new cases in the U.S. and over 13,000 deaths due to this disease ( 4 ). This makes endometrial cancer the most prevalent and deadly gynecologic cancer and the fourth most common cancer in women (behind breast, lung, and colorectal cancer). In people over 50 years old with an intact uterus, it is the second most common malignancy ( 5 ). Worldwide, it is estimated that there are over 400,000 new cases per year with the highest rates seen in North America, Europe, Australia/New Zealand and Micronesia/Polynesia ( 6 ). More than 80% of endometrial cancers are estrogen receptor (ER) positive ( 7 , 8 ). Excess estrogen exposure is strongly linked to the development of low-grade (grade 1 and 2) endometrial cancer and its precursor, endometrial intraepithelial neoplasia (EIN). Conditions of high estrogen exposure (often coupled with inadequate cyclical progesterone) such as obesity, nulliparity, polycystic ovarian syndrome, and unopposed estrogen replacement therapy are well established risk factors for the development of endometrial cancer ( 9 - 13 ). Given the rising incidence of endometrial cancer, there is an urgent need to identify prevention strategies that can be targeted to individuals with a high risk of developing endometrial cancer. EIN, formerly known as complex atypical hyperplasia (CAH), is the well-established precursor lesion of invasive endometrial adenocarcinoma, with at least 40% of individuals with EIN demonstrating invasive adenocarcinoma on their final hysterectomy specimen ( 14 ) and thus making this an important therapeutic target in the prevention of endometrial cancer ( 15 ). The standard of care for the treatment of both endometrial cancer and EIN is hysterectomy, which, given the comorbidities of this patient population (including diabetes mellitus, obesity, and cardiovascular disease), is often associated with significant perioperative morbidity and cost ( 16 ). Some patients can be treated with systemic or local progesterone therapy. However, this treatment is not always effective and has limited long-term efficacy. In a large systematic review of progestin therapy in fertility preservation, durable response was noted in only 53% of women ( 17 ). Therefore, a non-surgical, safe and tolerable therapy to prevent endometrial cancer and its precursors is needed. In pre-menopausal women, most circulating estrogen comes from the ovary. However, in post-menopausal women, estrogen levels are driven by the conversion of adrenal androgens into aromatic estrogens via the aromatase enzyme, which has high concentrations in adipose tissue. Aromatase is a cytochrome p450 enzyme and is responsible for the conversion of androgens (specifically testosterone and androstenedione) into estrogens (specifically estradiol and progesterone, respectively). Blocking aromatase therefore lowers circulating estrogen levels, which may have an anti-proliferative effect on the endometrium. Breast cancer patients treated with aromatase inhibitors have a lower risk of developing endometrial cancer when compared to patients treated with tamoxifen, a selective estrogen receptor modulator ( 18 , 19 ). The strong link between estrogen levels and endometrial carcinoma, and the role of aromatase inhibitors in reducing circulating estrogen levels suggest that aromatase inhibitors could potentially be used as a prevention strategy in individuals with EIN or low-grade endometrial cancer. Establishing biologic activity and tolerability is the first step in developing aromatase inhibition as a preventive strategy.