Everolimus, Letrozole, and Metformin in Women with Advanced or Recurrent Endometrioid Endometrial Cancer: A Multi-Center, Single Arm, Phase II Study

Endometrial cancer is the most common gynecologic malignancy in the United States with an estimated 61,880 new diagnoses and 12,160 expected deaths this year. Both the incidence and mortality associated with endometrial cancer have continued to increase; in part because of the limited treatment options for women with advanced or recurrent disease. A number of studies done through the Gynecologic Oncology Group in women with recurrent endometrial cancer have shown poor responses to salvage chemotherapy. Paclitaxel was the only active single-agent chemotherapeutic with a response rate (RR) of 27.3% in women with advanced or recurrent endometrial cancer. These data however, were in women without prior treatment with taxane-based chemotherapy (1). Currently, the combination of paclitaxel and carboplatin is often used at first-line treatment in advanced or recurrent endometrial cancer (2). As a result, the focus of second-line therapy has been on non-taxane treatment. Other agents studied including etoposide, liposomal doxorubicin, and topotecan have shown less than promising results with RR between 4% and 13% (3, 4). Bevacizumab was approved in the recurrent setting based on 6-month progression-free survival (PFS) of 40% and overall RR of 13.5% (5). As a result, the major focus of clinical research studies has included alternative treatment strategies.

The Cancer Genome Atlas and other molecular studies have identified that aberrations in the PI3K/AKT/mTOR pathway are common in endometrioid endometrial cancer (EEC), with loss of PTEN found in up to 80% of tumors (6). In addition, mutations in PTEN and PIK3CA have been identified in up to 50% of tumors (7). A number of studies have evaluated the role of single-agent mTOR inhibition in recurrent endometrial cancer as both primary and second-line therapy (8–10). While objective RR ranged from 0% and 24%, stable disease (SD) rates have been as high as 90% resulting in further study.

A phase II study evaluating the combination of everolimus and letrozole in women with advanced and recurrent endometrial cancer showed a clinical benefit (CB) rate of 40% with an objective RR of 32% including nine complete and two partial responses (11). In this trial, 9 patients were on metformin either prior to entry or started on metformin because of elevated glucose, a common side effect of everolimus. Among the small number of patients on metformin, the CB rate was 78% with an objective RR of 56% compared with CB rate of 38% and an objective RR of 23% in patients not taking metformin (11). Previously, we demonstrated that a short course of oral metformin prior to endometrial cancer surgery resulted in downregulation of the AKT/mTOR pathway at the tissue level (12). In addition, preclinical studies using a xenograft mouse model of endometrial cancer showed that metformin inhibited cell proliferation, induced apoptosis, and decreased tumor growth with the greatest response seen in cells harboring activating KRAS mutations. Metformin displaced constitutively active KRAS from the cell membrane causing uncoupling of the MAPK signaling pathway. These findings provided rationale for combining metformin and a PI3K-targeted agent, particularly in KRAS mutation–positive endometrial cancer (13).

The objective of this study was to estimate the CB rate of the combination of everolimus, letrozole, and metformin in women with advanced or recurrent EEC. In addition, we performed an exploratory analysis to determine whether specific molecular features, including presence of a KRAS mutation were associated with response to therapy.

This was a phase II, open-label trial performed within the MD Anderson Cancer Research Network. Patients enrolled at MD Anderson Cancer Center (main campus and Houston area locations), Lyndon B. Johnson Hospital (Houston, TX), and Cooper Health System (Camden, NJ). The primary objective of the study was to determine the CB of everolimus, letrozole, and metformin in patients with recurrent or advanced, noncurable EEC. Secondary objectives included treatment toxicity, PFS, time to disease progression, and overall survival (OS) in this patient population. Exploratory objective included molecular analysis on tumor tissue obtained at study entry to determine whether these findings informed response to therapy. The study was registered on clinicaltrials.gov as study number NCT01797523. The institutional review board at each site approved the study in compliance with the Declaration of Helsinki.

Eligible patients had advanced or recurrent histologically confirmed endometrioid or mixed-endometrioid endometrial adenocarcinoma, which was refractory to curative therapy. At least 18 patients had to have a KRAS mutation to be able to determine whether presence of this mutation was associated with response. Patients were excluded if they had a carcinosarcoma, sarcoma, pure clear cell carcinoma, or any component of serous carcinoma. Patients were allowed up to two prior chemotherapeutic regimens for recurrent disease. Chemotherapy administered in conjunction with primary radiation as a radio-sensitizer was not counted as prior therapy. Prior hormonal therapy, including letrozole was allowed. Patients with known diabetes, including those already on metformin were also eligible. All patients were required to have RECIST version 1.1 measurable disease and Eastern Cooperative Oncology Group performance status of 0–2. (14). Pretreatment hematologic, renal, and hepatic function tests were required to be grade 0 or 1 according to Common Terminology Criteria for Adverse Events (CTCAE, version 4.0).

The study schema is shown in Fig. 1. After informed consent was obtained, patients were enrolled on the trial and underwent a pretreatment biopsy for molecular analysis. If a patient had a biopsy for recurrence within 3 months of enrollment, this tissue was requested and used for molecular testing. If inadequate tissue was obtained during the research biopsy, archival tissue from either the primary tumor or a recurrence biopsy were requested for molecular testing. After completion of the biopsy, patients underwent a 7–10 day lead-in with metformin only. This included 500 mg oral daily for at least 3 days, then increase to 500 mg twice a day until day 1, cycle 1. If patients were already on metformin, their baseline dose was titrated up to the study dose of 500 mg orally twice a day. If their dose was ≥1,000 mg daily, they continued their usual dose.

The starting dose was everolimus 10 mg orally daily, letrozole 2.5 mg orally daily, and metformin 500 mg twice a day (unless they were on a higher dose of metformin prior to enrollment). Each cycle consisted of 4 weeks of therapy. Patients continued treatment until disease progression, dose-limiting toxicity, or withdrawal of consent. The primary endpoint was CB rate defined as prolonged SD (≥16 weeks), partial response (PR) or complete response by (CR) RECIST 1.1 criteria. Imaging was performed every 2 cycles until cycle 6 and then every 3 cycles, unless there was a new symptom suggestive of progression or obvious progression on physical exam. Toxicity was assessed using the NCI CTCAE (v4.0).

Paraffin-embedded sections of endometrial tumor tissue were cut at 4 μm thickness and used for IHC analysis following antigen retrieval with citrate buffer (pH 6.0). Sections were incubated in primary antibody against Phospho-S6 Ribosomal Protein (Ser235/236; pS6p, 1:75), Progesterone receptor (PgR, 1:500; Cell Signaling Technology), Phosphatase and tensin homolog (PTEN, 1:100), and Estrogen receptor (ER, 1:35; DAKO Incorporation), followed by the incubation with biotinylated anti-rabbit or anti-mouse IgG and streptavidin-HRP (Dako Incorporation). Diaminobenzidine solution was applied to visualize the complex. The sections were counterstained with Mayer's hematoxylin. To evaluate differential expression levels of the markers, the following scoring systems were used: PTEN was scored as positive (>90% of tumor cells show positive cytoplasmic and nuclear staining), negative (<1% of tumor cells show positive staining in cytoplasmic and nuclear), and heterogeneous (tumors show positive and negative staining foci; ref. 15). For PgR and ER, the proportion and intensity of staining was used to determine H score—percent of tumor cells stained with certain intensity multiplied by intensity score (0, none; 1, weak; 2, intermediate; and 3, strong) was calculated, with the score ranging from 0 to 300. Tumor tissue staining with H score ≥ 1 was reported as positive, and negative staining was reported as H score < 1 (16). For pS6P: H score ranging from 0 to 300 was used to report tumor cells staining.

Molecular evaluation of mutational status was performed using a next-generation sequencing panel of hotspots from 50 genes (see Supplementary Data) in a clinical molecular diagnostics laboratory (17).

A Simon two-stage design was employed with 25 patients evaluable for efficacy in the first phase (18). If there were at least 12 patients with CB measured after 2 cycles of therapy (at 8 weeks), the additional 29 patients enrolled in the second stage for 54 evaluable patients. This study design had a 91% power with a 10% significance level to reject a CB of 40% in favor of a CB rate of 60% (the hypothesized rate). This design had a 0.73 probability of stopping after the first stage if the CB rate was actually 40% and a 0.08 probability of stopping after the first stage if the CB rate was actually 60%.

Descriptive statistics were used to summarize the demographic and clinical characteristics of patients, as well as tumor response. The CB rate (CR + PR + SD) measured after 8 weeks (2 cycles) of therapy and confirmed after 16 weeks (4 cycles) of therapy, was estimated with an exact 95% binomial confidence interval. Duration of CB for those patients with at least SD after 8 weeks (2 cycles) and confirmed after 16 weeks (4 cycles), was calculated from the end of cycle 2 to the date of disease progression or date of last contact. PFS was calculated from day 1, cycle 1 until date of progression or last contact. OS was calculated from day 1, cycle 1 until date of death or last contact. Patients, who came off trial due to toxicity or in the absence of progression, were censored at last known follow-up. Patients on active treatment were censored at last tumor assessment if they were alive and without progression. Kaplan–Meier method was used to analyze PFS and OS, and Fisher exact test was used to compare CB rate in patients with and without molecular alterations. Study data were collected and managed using REDCap (19). All statistical analyses were performed using SAS 9.4 for Windows (Copyright 2002–2012 by SAS Institute Inc).

Sixty-two patients enrolled in the study between October 2013 and May 2016. Eight patients were consented but were not evaluable for response due to withdrawal of consent (2), bowel obstruction/rapid progression prior to completion of cycle 1 (2), diagnosis of thyroid cancer prior to initiation of therapy (1), change in treatment plan to radiation (1), no measurable disease (1), and death due to a car accident during cycle 1 (1). The demographic characteristics of the 54 evaluable patients are shown in Table 1.

Median age was 62 years. Median body mass index (BMI) was 33.3 kg/m². Ninety-three percent had pure endometrioid histology; 18% grade 1, 41% grade 2, 37% grade 3, and 4% unknown grade. A majority of patients (85%) had received at least one prior chemotherapy agent for recurrence and 65% had received prior radiotherapy. Five patients (9%) were on metformin prior to enrollment in the study.

Among the 54 evaluable patients, median number of cycles was 6 (range 1–31) with a total of 410 cycles completed. Clinical outcomes are shown in Table 2. The overall CB rate was 50% [95% confidence interval (CI), 36.1–63.9]; with 28% PR and 22% SD confirmed after 16 weeks of therapy. Figure 2 is a swimmer plot depicting both clinical outcome and duration of response. The median duration of response for those with a confirmed CB after 16 weeks was 7.1 months (range 2.9–26.6). The 6-month PFS was 0.41 (95% CI, 0.27–0.54) and 12-month PFS 0.25 (95% CI, 0.14–0.37). A majority of patients (90%) came off study due to progression of disease. One patient remains on active treatment. She has received 31 cycles and had a PR. The median follow-up time was 17.9 months (range 2–47 months). The median PFS was 5.7 months (95% CI, 3.6–8.2) and median OS was 19.6 months (95% CI, 14.2–26.3). At the time of this analysis, 67% of patients were deceased.

Toxicity was evaluated in the 59 patients who had at least one dose of treatment and had a follow-up evaluation for side effects. The majority of adverse events were grade 1 and 2 and manageable with supportive care. The grade2, 3, and 4 adverse events are listed in Table 3. There were no grade 5 events. Anemia was the most common adverse event with 85% of patients having at least grade 2. Grade 3 anemia was found in 24%, followed by hypertriglyceridemia (15%), hyperglycemia (9%), hyponatremia (7%), fatigue (6%), and thrombocytopenia (6%). Four patients required a dose reduction to everolimus 5 mg daily due to pneumonitis after 4 cycles (2), elevated creatinine after 9 cycles (1), and persistent grade 3 thrombocytopenia after 4 cycles. All patients remained on study at the reduced dose. One patient had to discontinue metformin due to persistent grade 2 diarrhea that was not tolerable for the patient. Only 1 patient was taken off study due to toxicity; grade 3 liver function tests after cycle 1, which did not resolve within 28 days. Of note, she had a PR on follow-up imaging after only receiving 1 cycle of therapy.

Forty-seven patients had tissue available for molecular analysis (Table 4). A majority were biopsies done at the time of recurrence (51% on study and 24% archived tissue). For 19%, the primary tumor tissue was evaluated.

The majority of tumors were ER and PgR positive, 91% and 63%, respectively. Of the 15 samples that had data on ER status at both primary diagnosis and recurrence, 100% matched (95% CI, 78%–100%). Of the 12 samples that had data on PgR status at both primary and recurrence, 75% matched (95% CI, 43%–95%). There was no difference in CB rate based on ER status [ER negative 33% (95% CI, 0.8%–91%) vs. ER positive 69% (95% CI, 49%–85%), P = 0.27]. However, there was a significant difference in CB rate based on PgR status [PgR negative 27% (95% CI,6%–61%) vs. PgR positive 90% (95% CI, 67%–99%), P = 0.001]. The difference in overall response [PgR negative 9% (95% CI, 0.2%–41%) vs. positive 45% (95% CI, 23%–68%), P = 0.06] was present but not statistically significant.

Next-generation sequencing was performed on 47 (87%) tissue samples. The median number of mutations detected was 2 (range 1–4). The most common mutations were PTEN (60%), PIK3CA (47%), KRAS (38%), TP53 (26%), and CTNNB1 (23%). There was no difference in CB rate based on PTEN (P = 0.78) or KRAS mutation status (P = 0.99). There were 9 patients who had mutational analysis on their primary tumor in addition to the research biopsy performed as part of the study. Among these, the presence of a mutation in both samples matched 100% for PTEN (7 patients, 95% CI, 59%–100%), 100% AKT (2 patients, 95% CI, 16%–100%), 100% for PIK3CA (5 patients, 95% CI, 48%–100%), 100% for KRAS (5 patients, 95% CI, 48%–100%), 60% for CTNNB1 (5 patients, 95% CI, 15%–95%), and 50% for p53 (2 patients, 95% CI, 1%–99%).

The combination of everolimus, letrozole, and metformin resulted in CB for 50% of women with advanced or recurrent endometrial cancer, with an overall RR of 28%. The median duration of response was 7 months, with a 6-month PFS of 41%. These results compare favorably with currently approved therapies for recurrent endometrial cancer. The combination was well tolerated, with a manageable toxicity profile. PgR status was associated with response, with a 90% CB rate and 45% overall RR among women who had PgR expression in their tumor. This is among the first phase II studies in recurrent endometrial cancer to find a candidate biomarker associated with CB. Further validation studies are needed.

A number of single-agent studies targeting mTOR have shown RR between 0% and 24% with SD rates as high as 90% (8–10, 20). The high CB rate has led to additional studies combining targets, including hormonal therapy. The concept of combining hormonal therapy with mTOR inhibition stems from preclinical data that suggest that blocking the PI3K/AKT/mTOR pathway may overcome resistance to hormonal therapy (21). Cotargeting the mTOR pathway and ER has shown benefit for patients with hormone receptor–positive (HR+) breast cancer in clinical trials (22). In a phase III randomized trial BOLERO-2, the addition of everolimus to endocrine therapy had significant improvement in progression free but not OS (23, 24).

The combination of mTOR inhibition and hormonal therapy in endometrial cancer has also been reported, however, with less consistent results. Fleming and colleagues conducted a randomized phase II trial of intravenous temsirolimus ± megesterol alternating with tamoxifen (considered the most effective hormonal regimen) in women with recurrent endometrial cancer. There was no added benefit seen with the addition of hormones to temsirolimus. The study was closed early, however, due to toxicity with an increase in venous thrombosis seen in women on the hormonal arm (20). At our institution, a single arm phase II study of everolimus and letrozole by Slomovitz and colleagues reported a CB rate of 40% and an overall RR of 32% (11). The promising results from this study laid the foundation for this study.

Elevated serum glucose is a common side effect of many mTOR inhibitors. As a result, oral hypoglycemic agents will be initiated to manage this on-target event. During the phase II trial by Slomovitz and colleagues evaluating everolimus and letrozole, patients appeared to benefit from the addition of metformin (11). In a post hoc analysis, the RR among patients on metformin was higher than those not receiving metformin. Specifically, there were 9 patients on metformin either prior to study enrollment or started on metformin due to elevated glucose. Among this subgroup, the CB rate was 78% with an objective RR of 56% compared with CB rate of 38% and an objective RR of 23% in patients not taking metformin.

Metformin is thought to have both a direct and indirect effect on cell growth and metabolism (25, 26). In the direct model, metformin activates AMPK, which results in phosphorylation of tuberous sclerosis 2 protein. This inhibits mTOR signaling which ultimately inhibits cell growth. Metformin also acts indirectly by increasing insulin sensitivity, increasing uptake of glucose in the cell, and ultimately decreasing circulating levels of insulin. Both insulin and IGF-1 are known growth factors that promote cell growth, so decreasing insulin would have a negative effect on cell proliferation.

On the basis of preclinical data using metformin in endometrial cancer cell lines, we anticipated that tumors with a KRAS mutation, would be more likely to respond and this study was powered to detect this difference (13). In this study, patients were required to have a baseline biopsy for molecular analysis to determine whether molecular features at the time of recurrence could predict response to therapy. Interestingly, 100% of patients had ≥1 mutation detected by next-generation sequencing. PTEN was the most common, followed by PIK3CA, KRAS, TP53, and CTNNB1. Although available for comparison in only 9 patients, there was a high concordance between mutational analysis in the primary tumor compared with the research biopsy done at recurrence. Both ER and PgR positivity were high, 91% and 63%. Like many other studies in endometrial cancer, presence of PTEN or PIK3CA mutation was not associated with response to mTOR inhibition. Although we hypothesized that KRAS mutation would be associated with response to therapy, there was no association between this biomarker or any of the detected mutations and response to therapy. PgR status measured by IHC, however, correlated with response to therapy. Interestingly, there was a high correlation between ER and PgR status in the primary tumor and the recurrent tumor among those who had testing at both time points. This is important, as it is a readily available marker when determining the best treatment for patients with EEC. Prospective integrated trials are warranted to validate this association.

Although mechanism was not evaluated in this study, we suggest that PgR status in advanced/recurrent endometrial cancer could provide a readout of coupled, functional ER/PgR signaling. As a prototypical estrogen-regulated gene, expression of PgR suggests that downstream estrogen-regulated signaling is active, and thus, potentially more sensitive to hormonal agents. The response to antiestrogen activity of letrozole is expected to be highest in this setting. ER status alone may not indicate the functional status of downstream ER signaling. Furthermore, metformin has been shown to influence both ER/PgR signaling and PI3K/AKT/mTOR signaling providing further support to the relevance of PgR in predicting response to this regimen (27–29). Previous studies have shown cross-talk between ER/PgR and PI3K/AKT/mTOR signaling (30). Alternatively, the differential response in PgR+ cases could reflect increased benefit in cases that are ER+ and PgR+. A larger cohort of ER− and PgR− cases are required to evaluate this relationship. Further studies will be necessary both for validation and to understand the mechanism underlying PgR status and response to letrozole, metformin, and everolimus.

Future studies focus on novel combination therapies, which may add benefit to mTOR inhibition and/or hormonal therapy, as well as studies evaluating molecular diagnostics to identify subgroups most likely to respond to these combinations. CDK4/6 inhibition has shown benefit in combination with endocrine therapy in several HR+ breast cancer populations (31–33). The combination of everolimus, exemestane, and ribociclib (CDK4/6 inhibitor) is ongoing in HR+ breast cancer (clinicaltrials.gov, NCT01857193). Similarly, we are enrolling in a phase II randomized study combining everolimus, letrozole ± ribociclib to determine whether the addition of a CDK4/6 inhibitor will add benefit for women with recurrent EEC (clinicaltrials.gov, NCT03008408). These trials should continue to explore the relationship between molecular features in the tumor and benefit from targeted therapies, to identify the best treatment for individuals with EEC.

Everolimus, letrozole, and metformin resulted in a CB rate of 50% and an overall RR of 28% in all patients. Response may be enhanced in women with PgR-positive tumors. These findings, as well as the preliminary report from GOG-3007, suggest that the backbone of everolimus and letrozole are effective in treating women with recurrent EEC and may be a reasonable choice in second-line therapy (34). This is the second trial to suggest there is benefit from the addition of metformin, without an increase in toxicity for patients. This study confirmed that the combination of everolimus, letrozole, and metformin showed activity in the treatment of recurrent EEC.

S.N. Westin is an employee/paid consultant for AstraZeneca, Clovis Oncology, Tesaro, Roche/Genentech, Merck, Pfizer, Takeda, Circulogene, and Novartis, and reports receiving commercial research grants from AstraZeneca, ArQule, Clovis Oncology, Tesaro, Bayer, Roche/Genentech, and Cotinga Pharmaceuticals. Y. Yuan is a paid consultant for Boehringer Ingelheim Pharmaceuticals, Servier Pharmaceuticals, Amgen Inc., Deciphera Pharmaceuticals, Citius Pharmaceuticals, and Midas Medical Technologies. B.M. Slomovitz is a paid consultant for GlaxoSmithKline, Clovis, Genentech, and AstraZeneca. R.L. Coleman reports receiving other commercial research support from Merck, Abbvie, AstraZeneca, Janssen, Clovis, Roche/Genentech, and Novartis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P.T. Soliman, S.N. Westin, D.A. Iglesias, B.M. Fellman, Y. Yuan, M.S. Yates, B.M. Slomovitz, K.H. Lu, R.L. Coleman

Development of methodology: P.T. Soliman, S.N. Westin, D.A. Iglesias, B.M. Fellman, Y. Yuan, Q. Zhang, M.S. Yates, B.M. Slomovitz, R.L. Coleman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.T. Soliman, S.N. Westin, Q. Zhang, R.R. Broaddus, B.M. Slomovitz, R.L. Coleman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.T. Soliman, S.N. Westin, B.M. Fellman, Y. Yuan, M.S. Yates, R.R. Broaddus, B.M. Slomovitz, R.L. Coleman

Writing, review, and/or revision of the manuscript: P.T. Soliman, S.N. Westin, D.A. Iglesias, B.M. Fellman, Y. Yuan, Q. Zhang, M.S. Yates, R.R. Broaddus, B.M. Slomovitz, K.H. Lu, R.L. Coleman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.T. Soliman, B.M. Slomovitz, R.L. Coleman

Study supervision: P.T. Soliman, B.M. Slomovitz, K.H. Lu, R.L. Coleman

This work was supported in part by Cancer Center Support Grant (NCI Grant P30 CA016672), Andrew Sabin Family Fellowship, NCI SPORE in Uterine Cancer (2P50 CA098258-06), and Novartis.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.