Articles••

Molecular typing guiding treatment and prognosis of endometrial cancer

•,,

...

Abstract

Genetic abnormalities, such as PTEN, PIK3CA, CTNNB1, ARID1A, and ERBB2, which frequently occur in endometrial cancer (EC), are potential therapeutic targets. In 2013, integrated genomic analysis conducted by The Cancer Genome Atlas identified four molecular subtypes, including POLE ultra-mutated, microsatellite instability hypermutated, copy-number low, and copy-number high, which strongly correlate with prognosis. Surrogate markers-based molecular classification methods have been developed to make these molecular classifications accessible and affordable, achieving classification into POLEmut, mismatch repair deficient (MMRd), p53abn, and no specific molecular profile (NSMP) with normal p53 expression. Although POLEmut EC has aggressive pathologic features, there are few cases of advanced and/or recurrence. Therefore, the possibility of de-escalating adjuvant therapy can be considered. Additionally, immune checkpoint inhibitors (ICI) may be a candidate for treating advanced and recurrent POLEmut EC because of their high immunogenicity. MMRd EC shows an intermediate prognosis between those of POLEmut and p53abn EC. MMRd EC is generally characterized by high immunogenicity similar to POLEmut EC, suggesting that ICI can also be a potential therapeutic agent. Among the four molecular subtypes, p53abn EC has the worst prognosis. However, some p53abn tumors have the molecular hallmark of homologous recombination deficiency and could be treated with poly (ADP-ribose) polymerase inhibitors. In addition, some p53abn tumors overexpress the human epidermal growth factor receptor 2, which can also be a potential therapeutic target. NSMP EC are a heterogeneous population because they lack characteristic molecular biological features. Approximately half of the NSMP EC show high expression of estrogen receptor/progesterone receptor, suggesting the possibility of hormonal therapy. In addition, the PI3K/AKT/mTOR pathway frequently altered in EC may be a therapeutic target. This review summarizes the molecular biological characteristics and potential therapeutic agents in molecularly featured EC. Several clinical trials are in progress to stratify EC into molecular classifications and demonstrate the efficacy and safety of molecularly matched treatment and management strategies.

1 Introduction

Endometrial cancer (EC) is the sixth most common cancer in women worldwide, accounting for 417,000 new cases and 97,000 deaths in 2020.¹ EC is geographically more prevalent in North America, Europe, Micronesia/Polynesia, and Australia/New Zealand, whereas it is less prevalent in Africa and South-Central Asia. Generally, risk factors for EC include no history of pregnancy or delivery, menstrual irregularities, infertility, estrogen exposure, use of hormonal agents, higher body mass index (BMI), diabetes, genetic predisposition (such as Lynch syndrome).² For decades, we have determined the recurrent risk of EC based on pathological findings, such as histology, stage, grade, myometrial invasion (MI), lymph vascular space invasion (LVSI), and cervical stromal invasion (CSI). However, risk stratification based on pathologic findings may not be reproducible among expert gynecologic pathologists.³^,⁴ Central pathology review in previous clinical trials revealed interobserver differences in histology, grade, stage, and risk classification criteria, such as CSI, MI, and LVSI.⁵ Thus, risk stratification based on inconsistent pathologic criteria may result in some patients not receiving adequate treatment.

Recent breakthroughs in genome analysis technology have revealed genomic abnormalities in EC. Furthermore, integrated genomic analysis has identified molecular subgroups that correlate with prognosis and will be widely used in the future to inform treatment and management decisions in EC.⁶^,⁷ In addition, various mechanisms of biological abnormalities occurring in EC cells have been discovered, and the development of new therapeutic agents and biomarkers targeting these abnormalities has been proceeding actively. This review aimed to summarize the genomic and molecular alterations in EC and further explore the possibility of potentially beneficial therapeutic agents based on molecular characteristics.

2 Review of the genomic and molecular alterations in EC

Molecular testing is becoming more common in EC, as in many other types of cancer, with the goal of personalizing therapeutic approaches. Several targeted agents are being evaluated for EC treatment. The future potential of targeted approaches to EC treatment is highly promising. Bokhman proposed the EC subtype classification for the first time in 1983.⁸ This classification is widely accepted, dividing EC into two types: type 1 (endometrioid), affecting 80% of patients, and type 2 (non-endometrioid), affecting 20% of patients.

Type 1 EC is histologically intermediate to highly differentiated endometrioid endometrial carcinoma (EEC) associated with long-term estrogen stimulation and is frequently preceded by endometrial hyperplasia.^9–11 Type 1 EC is significantly associated with phosphatase and tensin homolog (PTEN) and AT-rich interaction domain 1A (ARID1A) tumor suppressor gene mutations, catenin beta 1 (CTNNB1), phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA), and KRAS oncogene mutations (Table 1).¹²^,¹³ In addition, microsatellite instability (MSI) is present in approximately one-third of the tumors.¹⁴^,¹⁵ EEC is the most common type of EC, characterized by a high frequency of PTEN mutations (52%–78%).⁶^,¹⁶^,¹⁷ Loss of PTEN function, a tumor suppressor gene, is implicated in the pathogenesis of complex atypical hyperplasia (CAH), a precursor lesion of EEC.¹⁸ Furthermore, PTEN is a negative regulator of the phosphoinositide 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) pathway; thus, PTEN mutations activate the PI3K/AKT/mTOR pathway, leading to the progression of CAH to EEC.¹⁹^,²⁰ CTNNB1 gain-of-function mutations are present in approximately 25% of EEC.²¹ Mutational inactivation of the ARID1A tumor suppressor gene is also frequently observed in approximately 40%–50% of EEC cases, which is implicated in the progression of CAH to EEC.²²

Table 1

•

Somatic aberration frequencies of essential driver genes in EC and corresponding molecular targeted drugs.

Type 2 EC typically includes serous endometrial carcinoma (SEC), clear cell endometrial carcinoma (CCEC), and uterine carcinosarcomas (UCS), which are characterized by histologically poorly differentiated tumors.²³ They are significantly associated with TP53 mutations and amplification of the Erb-B2 receptor tyrosine kinase 2 (ERBB2) oncogene (Table 1).²⁴ SEC is a high-grade tumor that typically occurs in an estrogen-independent manner and accounts for approximately 10% of EC.²⁵ SEC shows TP53 mutations in most cases and frequent copy number alterations.²⁶ Human epidermal growth factor receptor 2 (HER2) is a transmembrane receptor encoded by the proto-oncogene ERBB2. Amplification or overexpression of HER2 is found in 26%–45% of SEC.²⁷^,²⁸ Overall mutation rate is low; MSI is rare.⁶ Serous endometrial intraepithelial carcinoma (SEIC) often precedes SEC, and SEIC may occur sequentially with an endometrial p53 signature.²⁹^,³⁰ CCEC is a rare histological type that comprises 1%–6% of newly diagnosed cases of EC.^31–33 In contrast to SEC, the TP53 mutation is relatively uncommon in CCEC, at approximately 40%.³⁴^,³⁵ Mutations in PIK3CA and ARID1A are also frequent in CCEC, with approximately 14%–36% and 16%–26% of cases, respectively.⁶^,^34–36 MSI was present in approximately 11%–19% of cases.³⁴^,³⁵ UCS is a biphasic tumor consisting of high-grade epithelial and mesenchymal elements.¹⁶ Most of the mutations that are significantly recurring in the UCS are also recurring mutations in the EEC.¹⁷ The most frequent was the TP53 mutation, occurring in 91% of UCS cases.¹⁶ Almost half of the UCS had mutations in more than one PI3K pathway gene: 22%–41%, 11%–41%, and 6%–17% were mutated in PIK3CA, PTEN, and PIK3R1 genes, respectively. In contrast to most other EC subtypes, in which PTEN and TP53 mutations are mutually exclusive, the majority of UCS with PTEN mutations also have TP53 mutations.³⁷^,³⁸

Approximately 5% of EC cases are caused by inherited genetic mutations associated with diseases such as Lynch syndrome (LS), polymerase proofreading-associated polyposis (PPAP), and Cowden syndrome. LS is an autosomal dominant inherited disease caused by germline mutations in mismatch repair (MMR) genes, specifically the MLH1, MSH2, MSH6, and PMS2. LS-associated EC is the most common extraintestinal sentinel cancer, classified as type 2 of LS, along with ovarian, breast, urinary tract, and other cancers.³⁹^,⁴⁰ Approximately 2–6% of EC are attributed to LS.⁴¹ PPAP is an autosomal dominant inherited disease caused by germline missense mutations in the exonuclease domain of polymerase epsilon (POLE) and delta (POLD1), predisposing to EC, adenomatous polyps, colorectal cancer, and other malignancies.⁴² They exhibit specific mutation signatures.⁴³ Cowden syndrome is an autosomal dominant inherited syndrome characterized by PTEN mutations and multiple hamartomas in various tissues.⁴⁴ The risk of malignancy is increased in many tissues, particularly in the endometrium, breast, thyroid gland, colorectum, and kidney.⁴⁵

3 Molecular classifications guiding prognosis

3.1 Molecular classifications identified by integrated genomic analysis

For decades, the EC risk stratification has been determined based on pathological findings, such as histological subtype, grade, MI, LVSI, and CSI, for surgically removed specimens. However, while conventional pathological risk stratification remains an important tool, interobserver variability is one of the problems. The Cancer Genome Atlas (TCGA) conducted an integrated genomic study in 2013 that included whole-genome or exome sequencing, deoxyribonucleic acid (DNA) copy number analysis, ribonucleic acid (RNA) sequencing, micro RNA sequencing, DNA methylation analysis, and reverse phase protein arrays were performed for 373 ECs, consisting of EEC (G1/G2/G3), SEC, and mixed EC (MEC) cases, revealed that EC was divided into four molecular subgroups based on molecular biological characteristics: POLE ultra-mutated, microsatellite instability hypermutated, copy-number high, and copy-number low (Table 2).⁶ In addition, it revealed that this molecular classification strongly associates with the prognostic outcome. Other studies have proven the validity of this classification and its association with prognosis.⁴⁶^,⁴⁷ Of these four classifications, the POLE ultra-mutated has the best prognosis, the copy-number high type has the worst prognosis, and the microsatellite instability hypermutated and copy-number low types have an intermediate prognosis. Because molecular classification and prognosis are closely related, this molecular classification was intended to be utilized to decide EC management strategies, including surgical methods and the necessity for postoperative adjuvant therapy. Subsequently, two research groups developed an algorithm using surrogate markers to make the classification more practical and cost-effective in clinical settings.⁴⁶^,^48–51 These groups developed the algorithms using immunohistochemistry of p53, MMR-related proteins, and mutational analysis by sequencing the exonuclease domain of the POLE gene. Further, the algorithm divides EC into four groups constructed from POLEmut, mismatch repair deficiency (MMRd), p53abn, and no specific molecular profile (NSMP p53 wt). With the development of the algorithm with surrogate markers, molecular analysis became more accessible because it can be performed using standard formalin-fixed paraffin-embedded specimens. The fifth edition of the 2020 World Health Organization Female Genital Tumors recommends incorporating molecular parameters into standard pathology reports. Moreover, in the 2020 ESGO/ESTRO/ESP guideline, the molecular classifications were incorporated into the risk stratification algorithm for determining the requirement and methods of postoperative adjuvant therapy.⁷ In addition, these molecular classifications correlate with prognosis and suggest that certain molecularly targeted therapeutic agents might be effective. With these backgrounds, molecular classification should be an effective tool in determining EC treatment and management strategies.

Table 2

•

Clinicopathological and molecular features, and potential therapeutic agents in four molecular subtypes identified by TCGA.

3.2 POLE ultra-mutated/POLEmut

POLEmut EC is detected in approximately 10% of all EC cases and is commonly associated with clinicopathologic features in younger age, lower BMI, endometrioid histology, and early-stage (Table 2).⁶^,⁴⁶^,⁴⁷^,⁵¹ The POLE gene encodes the catalytic subunit of DNA polymerase epsilon and is associated with DNA replication and repair. Therefore, a mutation in the exonuclease domain of POLE results in an ultra-mutated tumor with increased CD8 lymphocytic infiltration and up-regulation of cytotoxic T-cells.⁵² POLE mutations are called ultra-mutated because they can accumulate several mutations, with more than 100 genetic variants per megabase (MB).⁶ Histologically, EEC grade 3 accounted for 56.9% of POLEmut cases, followed by SEC and CCEC at 11.8% each.⁴⁷ In contrast, EEC grades 1 and 2 account for only 7.8%. A meta-analysis of 3185 EEC cases from nine studies showed that grade 3 EEC cases were significantly more likely to be assigned to POLEmut or MMRd ECs, while grade 1 and 2 cases were less likely to be assigned to NSMP.⁵³ Although most POLEmut EC cases exhibit pathologically aggressive features, such as high grade and LVSI, it has the best prognosis among the four molecular subtypes.⁶^,⁴⁷^,⁵⁴ All 17 POLEmut EC identified in the TCGA analysis had no recurrence, and there was only one case with recurrent among the 93 POLEmut EC identified in the PORTEC-3 analysis. Thus, omitting postoperative adjuvant therapy has been suggested for this molecular subtype because POLEmut EC has a favorable prognosis.⁷ Moreover, targeted therapy using immune checkpoint inhibitors (ICI) may be candidates for recurrent and advanced POLEmut EC treatment because of their high immunogenicity. POLEmut EC with concurrent TP53 mutations has been reported as a multiple-classifier, which occurs when two or more surrogate markers are positive.⁵⁵ In a study of 3518 ECs, all POLEmut-p53abn cases showed similar molecular characteristics and prognosis to those of POLEmut EC because 107 EC cases (3%) were assigned as multiple-classifiers. Therefore, these results suggest that p53abn may be a secondary alteration in POLEmut cases, implying that POLEmut-p53 cases should be treated the same as POLEmut cases.

3.3 Microsatellite instability hypermutated/MMRd

MMRd EC is characterized by the presence of an MMR function deficiency. MMR function deficiency can be determined using microsatellite instability testing or immunohistochemical staining of DNA mismatch repair-related genes, such as MLH1, MSH2, MSH6, PMS2, and approximately 30% of EC lack the mechanism that leads to mutation accumulation (Tables 1 and 2).⁶^,⁴⁶^,⁴⁷^,⁵¹ Therefore, MMRd EC shows a comparatively high tumor mutational burden (TMB), with more than 10 mutations detected per MB.⁶ Genetically, PTEN and PIK3CA mutations are more frequent in MMRd EC, occurring in 88% and 54% of cases, respectively. ARID1A mutations have also been reported in 37% of cases.⁶^,⁵⁶^,⁵⁷ MMRd EC with germline abnormalities related to the mismatch repair mechanism is called LS, which increases the lifetime incidence of uterine, colon, stomach, ovarian, renal/ureter, biliary tract, and pancreatic cancers. MMRd EC, as with POLEmut EC, is a highly immunogenetic tumor in which highly increased production of tumor mutant antigen (neoantigen) is associated with prolonged survival.⁵⁸ Increased neoantigen production upregulates tumor-infiltrating lymphocytes, especially CD8⁺ cytotoxic T cells, and increases T cell-mediated anti-tumor responses.^58–60 Tumor cells develop immunological escape mechanisms, primarily by upregulating inhibitory immune checkpoint receptors, such as programmed death ligand 1 (PD-L1), which subsequently blocks tumor cell death by activated T cells, suggesting that ICI can be an effective therapeutic agent.⁶⁰ Furthermore, similar to POLEmut EC, MMRd EC with concurrent TP53 mutations have been discovered, which is termed as multiple classifiers. However, the prognosis of these cases is similar to that of MMRd EC, suggesting that the TP53 mutation is a secondary genomic event.⁵⁵

3.4 Copy-number high/p53abn

p53abn EC represents approximately 12.2%–25.9% of EC and has a prognosis similar to SEC. Furthermore, it has the worst prognosis among the four molecular classifications identified by TCGA, accounting for half of all EC deaths.⁶^,³⁴^,^48–51^,^61–63 Of the 60 p53abn EC identified by TCGA, 44 (73.3%) had serous or mixed histologies.⁶ P53abn EC accounts for 93% of SEC, 85% of CS, 38% of CCEC, 22% of grade 3 EEC, and 5% of grades 1 and 2 EEC.⁶^,³⁴^,⁴⁶^,^48–51^,^61–65 Identifying p53abn tumors among them is crucial in determining the need and method of postoperative adjuvant therapy because grades 1 and 2 EEC have a good prognosis. P53abn EC is genetically characterized by high copy number and low mutational burden (Table 2).⁶ Genetic mutations in TP53 are found in more than 90% of cases, subsequently followed by mutations in PIK3CA (47%), FBXW7 (22%), PPP2R1A (22%) and focal amplification of ERBB2 leading to HER2 overexpression (25%).⁶ Conversely, PTEN mutations, frequently detected in other molecular types, are present in approximately 10% of cases.⁶^,⁵⁶^,⁵⁷ Comprehensive genomic analysis by TCGA has demonstrated that the molecular profile of p53abn EC is similar to that of high-grade serous ovarian cancer (HGSOC) and basal-like breast cancer.⁶ These HGSOC and basal cell-like breast cancers frequently have homologous recombination deficiency (HRD), poly (ADP-ribose) polymerase (PARP) inhibitors (PARPi); hence, targeting tumors with HRD are generally used in the standard therapy of these cancers, implying that PARPi could be a potential molecular agent for p53abn EC with HRD. TP53 mutations and abnormal p53 expression occasionally occur in POLEmut and MMRd EC, but these abnormalities are commonly considered secondary changes and should not be classified as p53abn EC because the prognosis is similar to that of POLEmut and MMRd types.⁵⁵

3.5 Copy-number low/NSMP p53 wt

NSMP EC lack specific molecular features, such as POLE mutation, TP53 mutation, and MMRd, genetically have low copy number and TMB, and account for approximately half of all EC (Table 2).⁶ NSMP EC has a moderate prognosis similar to that of MMRd, and is suspected to be a heterogeneous population because it lacks a unique molecular profile in NSMP p53 wt EC. Clinicopathologic features of NSMP p53 wt EC include a high incidence of grades 1 and 2 EEC, BMI, and prevalence of diabetes mellitus. Approximately half of the patients with NSMP exhibit high expression of estrogen receptor (ER) and progesterone receptor (PR), suggesting their potential for effective hormonal treatment. Genetically, POLE and TP53 gene abnormalities are not detected, whereas somatic mutations, such as PTEN, PIK3CA, CTNNB1, and ARID1A genes, are frequently identified (Table 2). Mutations in exon3 of the CTNNB1 gene are detected in approximately half of the cases, and patients with mutations have more recurrent and poorer prognoses than those without mutations.⁶^,⁴⁸^,⁶⁶^,⁶⁷ In addition, overexpression of the L1 cell adhesion molecule (L1CAM) is associated with distant metastasis and poorer prognosis.⁴⁸^,⁶⁸ Thus, other potentially useful biomarkers for the heterogeneous population of NSMPs, such as CTNNB1 mutations and L1CAM expression, may exist.

4 Molecular typing-guided treatment

4.1 Conventional therapy (radiotherapy, chemoradiotherapy, and chemotherapy)

The results of the PORTEC-3 trial can be considered as a reference regarding molecular classification and chemotherapy efficacy.⁴⁷ In this study, the 5-year recurrence-free survival (RFS) of the concurrent chemoradiotherapy and radiotherapy groups in p53abn EC was 58.6% and 36.2% (p = 0.021), respectively. These results suggested that chemotherapy may be effective in p53abn EC, and a similar trend was observed in early-stage cases in the subgroup analysis. In contrast, there were no significant prognostic differences in 5-year RFS between concurrent chemoradiotherapy and radiotherapy in the POLEmut EC (100% vs. 97%, p = 0.637), MMRd EC (68% vs. 76%, p = 0.428), or NSMP EC (80% vs. 68%, p = 0.243). However, radiation therapy is beneficial in MMRd EC in other studies.^69–71 A recent retrospective multi-center study compared the combination of radiotherapy and chemotherapy with chemotherapy alone in MSI-high (MSI-H) advanced EC and discovered that adding radiation prolonged disease-free survival (DFS). In the PORTEC-4a study, postoperative high- and intermediate-risk patients with EC were divided into favorable, intermediate, and unfavorable groups according to a molecular-integrated risk profile, and each group received observation, vaginal brachytherapy (VBT), and external beam radiotherapy (ERBT), respectively (Table 3).⁴⁷ The molecular-integrated risk profile for this study used the parameters of POLE mutation, MMRd, TP53 mutation, CTNNB1 mutation, L1CAM expression, and LVSI. Vaginal recurrence rates will be investigated in each group and compared with that of control patients who received VBT, the standard adjuvant therapy in high- and intermediate-risk EC. Another clinical study (NCT05524389) will compare 3-year loco-regional recurrence between patients who received molecular classification-based treatments, including observation (for the favorable group), VBT(for the intermediate group), and ERBT(for the unfavorable group), and those who received conventional risk stratification-based treatments, including VBT(for the intermediate group) and ERBT(for the unfavorable group), in early-stage EC (Table 3). Additionally, the CAN-STAMP trial (NCT04159155) is ongoing, and 3-year progression-free survival (PFS) between chemotherapy alone and chemoradiotherapy plus chemotherapy groups will be compared in the early-stage SEC/p53abn EC population (Table 3). Another study (NCT05489848) will compare 2-year PFS between the chemotherapy group and the radiotherapy plus chemotherapy group in newly diagnosed p53mut EC.

Table 3

•

Remarkable ongoing phase III clinical trials targeting molecularly specific ECs.

4.2 Immune checkpoint blockade therapy alone or in combination

Given their high immunogenicity, ICI may be effective for advanced or recurrent POLEmut and MMRd EC. The efficacy of ICI in MMRd/MSI-High (MSI-H) solid tumors has been evaluated, and single-agent responses to nivolumab (anti-PD-1), avelumab (anti-PD-L1), durvalumab (anti-PD-L1), and dostarlimab (anti-PD-1) have been reported as approximately 25%–47%.^72–75 A clinical trial, KEYNOTE-158, of pembrolizumab monotherapy in MMRd/MSI-H non-colorectal cancer patients who had failed prior therapy included 49 EC cases with an overall response rate (ORR) of 57.1%.⁷⁶ In the recent GARNET study of dostarlimab efficacy, the ORR in the MMRd arm was 43.5% (95% CI 34.0–53.4%), with complete response in 11 patients and partial response in 36, while the ORR in the MMR proficient (MMRp) arm was 14.1% (95% CI 9.1–20.6%), with complete response in 3 patients and partial response in 19.⁷⁷ Although immune checkpoint blockade therapy is effective in advanced and recurrent MMRd ECs, whether there is an add-on effect of ICI to standard therapy in newly diagnosed EC remain unclear because of limited data on benefit in first-line therapy. The TransPORTEC refining adjuvant treatment in EC based on molecular profile, RAINBO program, plans to solve this issue (Table 3).⁷⁸ The RAINBO program comprises four multinational clinical trials and a comprehensive research program in EC. Eligible patients are classified into four molecular subgroups, including POLEmut, MMRd, NSMP, and p53abn, and divided into standard and interventional therapies, except for patients with POLEmut. In the randomized phase III trial named MMRd-GREEN in the RAINBO program, stage II with LVSI and stage III patients with MMRd will be assigned to external beam radiotherapy (EBRT) and EBRT plus duruvalumab groups to evaluate the add-on effect of duruvalumab on EBRT. Other two phase III trials, KEYNOTE-C93/GOG-3064/ENGOT-en15 (NCT05173987) and DOMENICA (NCT05201547), compare the efficacy of the PD-1 antibody alone, pembrolizumab and dostarlimab, versus paclitaxel and carboplatin chemotherapy (TC) as a first-line treatment for MMRd EC in advanced and recurrent settings, respectively (Table 3). One more phase III trial aims to demonstrate the additional efficacy of pembrolizumab on EBRT plus VBT in stage I (with the combination of defined age and risk factors) and stage II EECs with MMRd/MSI-H (NCT04214067) (Table 3). More recently, the Australia New Zealand Gynaecological Oncology Group (ANZGOG) is planning a phase II trial, ADELE trial (ACTRN12621000273886), to evaluate the add-on effect of Tislelizumab (PD-1) on postoperative chemoradiation in high-risk EC. Thus, these several trials will provide important results on the add-on effect of ICI on postoperative adjuvant therapy in EC patients with MMRd.

Lenvatinib, a multi-kinase inhibitor targeting vascular endothelial growth factors 1–3, has shown efficacy for thyroid cancer, hepatocellular carcinoma, and renal cell carcinoma. KEYNOTE-146 is a phase IB/II trial investigating the efficacy of the combination of pembrolizumab and lenvatinib in advanced patients with solid tumors. Results from a cohort of 108 EC cases were published in 2020, with an overall population response rate of 38.0%, consisting of 63.6% in the MMRd group and 36.2% in the MMRp group.⁷⁹ These results led to a randomized phase III trial (KEYNOTE-775) comparing pembrolizumab plus lenvatinib to the physician's choice of single-agent chemotherapy in patients who had received at least one platinum-based chemotherapy regimen, and the prolonged PFS and overall survival were observed in the overall and MMRp cohorts.⁸⁰ Recently, a phase III trial, ENGOT-en9/LEAP-001 trial (NCT03884101), comparing pembrolizumab plus lenvatinib versus TC in advanced and recurrent ECs is ongoing.⁸¹ Another phase II trial (NCT03367741) is comparing the efficacy of nivolumab in combination with or without cabozantinib in advanced, recurrent, and metastatic EC. Similarly, other ongoing phase II trials, including NRG-GY025 (NCT05112601), PODIUM-204 (NCT04463771) and EndoMAP (NCT04486352), evaluate the safety and efficacy of ICIs in combination with other molecular targeted agents, including other ICIs, in advanced and/or recurrent EC. In terms of the add-on effect of ICI to platinum-based conventional chemotherapy, the following four phase III trials are ongoing. In the NRG-GY108 (NCT03914612) and KEYNOTE-B21/ENGOT-en11/GOG- 3053 studies (NCT04634877), the add-on effect of pembrolizumab to TC will be evaluated in advanced/recurrent EC and newly diagnosed advanced EC, respectively. Moreover, AtTEnd trial (NCT03603184) is assessing the add-on effect of atezolizumab to TC in patients with advanced and recurrent EC. Another phase III RUBY trial (NCT03981796）examines the add-on effect of dostarlimab to conventional chemotherapy in primary stage III/IV or first recurrent EC.

4.3 Targeting homologous recombination deficiency

The homologous recombination repair mechanism is a major mechanism in DNA double-strand break repair, along with non-homologous end joining.⁸² PARP is an accumulation of DNA damage via multiple mechanisms, including synthetic lethality and PARP trapping, leading to cell death with HRD, such as cancer cells with pathogenic variants of breast cancer susceptibility (BRCA) 1 and 2 genes.⁸² PARPi are effective for tumors with HRD and are commonly used in treating advanced and platinum-sensitive recurrent ovarian cancer settings.^83–89 Therefore, HRD may be a potential target for treating p53abn EC with HRD. According to the TCGA analysis, the molecular profile of SEC is similar to that of HGSOC and basal-like breast cancer, and approximately half of HGSOC cases and basal-like breast cancer show deficiencies in homologous recombination repair (HRR).⁶^,⁹⁰ As previous mall-scale study reported that 46% of EC with TP53 gene aberrations had HRD.⁹⁰ Lin Dong et al. studied 60 SEC cases and discovered 22 (36.7%) patients with HRR-related gene abnormalities, such as ATM, BRCA1, and BRCA2. Furthermore, they reported that patients with HRD had longer PFS and DSS than patients with HRP in the p53abn population.⁹¹ Conversely, a study that investigated HRD scores for EC using Myriad myChoice reported a poorer prognosis in EC with higher HRD scores.⁹² Siedel et al. reported that cell lines with high HRD were more sensitive to olaparib than cell lines with low HRD scores.⁹² The transPORTEC RAINBO program is planning a Red-p53abn trial to evaluate the effect of adding a DNA damage response targeting agent to chemotherapy in stage I–III p53abn EC because clinical data on the efficacy of PARPi in EC are limited.⁷⁸ The CAN-STAMP trial (NCT04159155), an ongoing phase III clinical trial targeting SEC and p53abn, will compare the 3-year DFS of TC alone with that of TC plus the PARPi niraparib in an advanced EC cohort (Table 3). Additionally, a phase II NRG-GY012 trial (NCT03660826) is currently underway to evaluate the efficacy of olaparib alone and in combination with cediranib, capivasertib (AKT inhibitor), and durvalumab in patients with recurrent, advanced, and refractory EC. Another ongoing phase II DOMEC trial (NCT03951415) is evaluating the efficacy of olaparib in combination with durvalumab in patients with recurrent and advanced EC.

4.4 Anti-angiogenic therapy

Anti-angiogenesis therapy has been employed in treating several cancers, including gynecological cancers, and its utility in EC has been studied.⁹³ The MITO-END-2 trial was a randomized phase 2 study evaluating the effect of adding bevacizumab to TC therapy, and the bevacizumab plus TC group had better PFS than the TC group, but it was insignificant.⁹⁴ Although the GOG-86P trial assigned patients with advanced and recurrent EC to three groups: TC plus bevacizumab, TC plus temsirolimus, and ixabepilone and carboplatin plus bevacizumab, and compared their PFS with that of TC cohort in the GOG209 trial, no significant difference of PFS was observed in any experimental arm compared to historical controls.⁹⁵ However, an additional study analysis suggested that bevacizumab may be more effective in patients with overexpressed p53 protein.⁹⁶ A meta-analysis involving the above two randomized controlled trials and five single-arm trials concluded that chemotherapy with bevacizumab might improve PFS and overall survival in advanced and recurrent EC compared with chemotherapy alone.⁹⁷ Therefore, further validation of the utility of anti-angiogenic therapy in EC is required, given molecular classification and the development of new biomarkers for anti-angiogenesis therapy.

4.5 Targeting PI3K/AKT/mTOR pathway

The PI3K and mTOR signaling pathways are commonly altered in EC and regulate cell growth, survival, protein synthesis, and angiogenesis.⁹⁸ TCGA and other molecular studies have identified that mutations in components of these pathways, including PTEN, PIK3CA, PIK3R1, and KRAS, are frequently found in EC.⁶^,¹⁷^,²⁰ Inhibitors of mTOR, such as ridaforolimus, everolimus, and temsirolimus, have shown some clinical efficacy for EC and are being investigated.⁹⁹^,¹⁰⁰ One study demonstrated that the anti-angiogenic drug bevacizumab and temsirolimus combination was effective, although it caused adverse events such as bowel perforation.¹⁰¹ Inhibiting mTOR is advocated to overcome endocrine resistance. Phase II trial of everolimus and letrozole in women with recurrent EC who received less than two cytotoxic regimens, the clinical benefit rate (CBR), which was defined as complete response (CR), partial response, or stable disease (16 weeks) by RECIST 1.0 criteria was 40% (14 of 35 patients). The confirmed objective response rate (RR) was 32% (11 of 35 patients; nine CR and two partial responses).¹⁰² Everolimus and letrozole therapy also reported a high median PFS (28 months) for recurrent EC without prior chemotherapy. Metformin, a drug commonly used to treat diabetes, downregulates the AKT/mTOR pathway and may improve response to treatment with mTOR inhibitors in EC. A phase II trial with everolimus, letrozole, and metformin therapy in advanced or recurrent EEC showed 50% CBR and 28% overall response in women with recurrent EEC.¹⁰³^,¹⁰⁴ PR positive tumors can respond better to treatment. Recently, a phase I/II clinical trial of vistusertib in combination with the aromatase inhibitor anastrozole for hormone receptor-positive recurrent or metastatic endometrial cancer was reported.¹⁰⁵ With a median follow-up of 27.7 months, the median PFS was 5.2 in the vistusertib and anastrozole combination therapy arm and 1.9 months in the anastrozole monotherapy arm.

4.6 Targeting HER2/Neu

Amplification or overexpression of HER2 is frequently observed in SEC and is associated with a worse prognosis (Table 2).¹⁰⁶ Clinical trials of trastuzumab, a humanized monoclonal antibody against HER2, have been associated with improved prognosis in recurrent, metastatic, and advanced SEC. Recent reports indicated that HER2 and ERBB2 abnormalities were observed in more than 20% of p53abn cases, with no association between them and histology.²⁸^,¹⁰⁷ Thus, HER2/Neu can be an effective therapeutic target in a subset of p53abn EC. A phase II trial on a new treatment modality involving trastuzumab added to chemotherapy for SEC compared TC with TC and trastuzumab (TC + Tr) in recurrent and advanced SEC.¹⁰⁸^,¹⁰⁹ The latest analysis confirmed a benefit from trastuzumab maintenance therapy in addition to standard therapy TC. The median PFS benefit was 12.9 months in the TC + Tr arm and 8.0 months in the TC arm. In patients with stage III and IV disease, the benefit was 17.7 months in the TC + Tr arm and 9.3 months in the TC arm. The overall survival benefit was 29.6 months in the TC + Tr arm and 24.4 months in the TC arm. In patients with stage III and IV disease, median survival was not reached in the TC + Tr arm, while it was 24.4 months in the TC arm. Toxicity did not differ between the two arms. Trastuzumab's characteristic inhibits HER2 homodimerization, while pertuzumab inhibits heterodimerization. Thus, adding pertuzumab to TC + Tr therapy should increase its anti-tumor activities, which has been demonstrated in breast cancer.¹¹⁰ Based on these results, a randomized phase II/III trial, NRG-GY026, has been initiated to evaluate trastuzumab efficacy in combination with TC + Tr therapy in patients with HER2-positive SEC or UCS. Recently, trastuzumab deruxtecan (T-DXd), a novel antibody-drug conjugate that covalently conjugates trastuzumab with the topoisomerase I inhibitor deruxtecan, was shown to be effective in HER2-positive recurrent gastric cancer.¹¹¹ Therefore, a phase II trial is currently being designed to evaluate the efficacy and safety of T-DXd in HER2-expressing tumors, including EC.¹¹²

4.7 Targeting ARID1A

ARID1A is a gene that encodes for a protein called BAF250a, a component of the SWI/SNF chromatin remodeling complex. This complex plays a role in gene transcription and is involved in various cellular processes such as tissue differentiation, proliferation, and DNA repair.¹¹³ Approximately 40% of low-grade and high-grade EEC have ARID1A gene mutations.¹¹⁴^,¹¹⁵ Furthermore, ARID1A protein loss was reported in 29% and 39% of low-grade and high-grade EEC, respectively.¹¹⁶ Inactivating ARID1A mutations are frequently detected in EEC and are associated with poor prognosis; however, ARID1A expression loss in grade 3 EEC has recently been associated with significantly longer relapse-free survival (RFS).¹¹⁷^,¹¹⁸ Loss of ARID1A expression is also observed in focal areas of atypical endometrial hyperplasia, indicating a tumor suppressive role for ARID1A in endometrial tumorigenesis.¹¹⁴^,¹¹⁹ Thus, ARID1A mutations are a promising biomarker for predicting EC and may also be used to assess new therapies. For instance, in preclinical studies, ARID1A mutated cancers are sensitive to ataxia telangiectasia and Rad3-related (ATR) inhibitors. Recently, the ATARI trial is ongoing to evaluate the clinical activity of the ATR inhibitor seracertib as a single agent and in combination with olaparib in ARID1A-stratified gynecologic cancers.¹²⁰ EZH2, together with ARID1A, targets the PI3K-interacting protein one gene (PIK3IP1) and regulates cell proliferation and antiapoptotic effects through the PI3K/AKT pathway.¹²¹ Loss of ARID1A expression results in an imbalance in EZH2 activity and promotes tumorigenesis.¹²² In addition, previous studies indicated that EZH2 inhibition induces apoptosis in ARID1A mutant cells and suppresses cell proliferation by enhancing PIK3IP1 expression.¹²³ Based on these findings, in anticipation of a synthetic lethal interaction between ARID1A mutations and EZH2-targeted inhibitors are being developed.

4.8 De-escalation of adjuvant therapy

The omission of postoperative therapy for the POLEmut EC may be considered because of their extremely good prognosis. The PORTEC-1 trial, which compared postoperative radiation therapy to observation in high intermediate-risk patients, discovered 10-year survival rates of 100% and 80% for patients with and without pathologic mutations in the observation cohort, respectively. Furthermore, experiments using pole-mutant mouse-derived embryonic stem (mES) showed that POLE mutation did not enhance the sensitivity of mES cells to radiotherapy and chemotherapy.¹²⁴ Similarly, in the PORTEC-3 trial, which compared the prognostic benefit of postoperative chemoradiation versus radiation therapy in patients with high-risk EC, the 5-year survival rates for patients with POLE mutations were 100% and 97% (all cohort 98%), respectively.⁴⁷ Moreover, a recent meta-analysis reported that postoperative adjuvant therapy was not associated with outcomes in patients with pathological POLE mutations.⁵⁴ Of the 294 patients with pathological POLE mutations analyzed in the study, 11 (3.7%) had recurrences and 3 (1%) had disease-related deaths. In the ESGO guidelines, POLEmut EC without residual tumor was assigned a low-risk status, not requiring adjuvant therapy.⁷ Active clinical studies constructed from PORTEC-4a, Tailored Adjuvant Therapy in POLE-mutated and p53-wildtype Early-Stage Endometrial Cancer (TAPER) trials to investigate the safety of de-escalation of adjuvant therapy in POLEmut EC. In the PORTEC-4a trial, patients who should have favorable EC based on a molecular-integrated risk profile will be observed and compared to patients with high intermediate-risk EC who did not undergo a molecular-integrated risk profile and received standard postoperative VBT.⁶⁶ POLEmut-BLUE trial, a part of RAINBO program, will demonstrate 3-year pelvic RFS in POLEmut EC without adjuvant therapy.⁷⁸ Moreover, the TAPER trial is a multicenter, single-arm study evaluating the omission of postoperative adjuvant therapy in POLEmut and early-stage NSMP EC (NCT04705649). However, because the evidence of management strategy for advanced and recurrent POLEmut EC is limited now, therapy omission may not be easily selected for these cases.

4.9 Hormonal therapy

Although the efficacy of the hormonal therapy for ER/PR-positive EC has been examined in clinical trials, the results have not been favorable enough to merit incorporation into standard therapy, and its utilization has not been established yet. In contrast, hormonal therapy is important in standard care for breast cancer. Studies examining the relationship between molecular characteristics and effects of hormonal therapy have been intensively conducted in breast cancer, and such studies are required in EC.¹²⁵ Given that NSMP p53abn EC is associated with mutations in the PI3K/Akt/mTOR pathway, the results of clinical trials utilizing letrozole in combination with the mTOR inhibitor everolimus and metformin, a treatment for diabetes mellitus, have been reported.^102–104 However, prospective clinical trials with appropriate molecular biological stratification are required because these studies have not focused on molecular types. The RAINBW program is planning the NSMP-ORANGE trial to compare adjuvant chemotherapy with radiation plus hormonal therapy in patients with ER-positive stage II (with LVSI) or stage III with NSMP p53 wt EC.⁷⁸ Additionally, the NCT05454358 trial is ongoing to compare 3-year PFS between NSMP patients with or without postoperative letrozole maintenance therapy (Table 3).

5 Conclusion

As described in this review, the molecular abnormalities and classifications in EC that are relevant to prognosis and therapeutic drug decisions are gradually being clarified through the development of innovative genomic analysis techniques. In addition, a convenient classification method using surrogate markers has been developed to enable molecular classification, and some guidelines have already incorporated molecular classification into risk stratification.⁷ However, most previous EC clinical trials did not consider molecular classification, making it difficult to directly integrate the results of previous clinical trials into a management and treatment strategy that considers molecular classification. Several clinical trials are in progress to stratify EC into molecular classifications and select appropriate treatment and management strategies to address this issue.⁶⁶^,⁷⁸ Hopefully, these trials will establish more appropriate treatment and management strategies for EC using molecular classification. However, molecular characterization has not yet been achieved; thus, there remains a population that cannot avoid conventional treatment and management. In such populations, further molecular biological analysis and development of therapeutic agents will be required.

Authors contribution

JT: writing, review, and/or revision of the manuscript, and preparation of tables; MT: writing, review, and/or revision of the manuscript, preparation of tables and article supervision; AO: article supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.