New clinicopathological concept of endometrial carcinoma with integration of histological features and molecular profiles.

Mesonephric adenocarcinoma, recognized as being human papillomavirus (HPV)‐independent and of presumable normal or hyperplastic mesonephric remnant origin, is exceedingly rare and occurs most commonly in the cervix and rarely in the vagina (Figures  8 and  9 ). 127 , 128 , 129 , 130 , 131 Occasional cases have been reported in the uterine corpus, 132 , 133 , 134 , 135 , 136 but primary ovarian mesonephric adenocarcinoma has not been reported. A novel histological entity, MLA, which closely mimics mesonephric adenocarcinoma, was first introduced as a subtype of both endometrial and ovarian tumors in the 2020 WHO classification. MacFarland et al. described the MLA characteristics of the uterine corpus and ovary in detail in 2016. 31 They argued that “on the balance of probabilities, we believe these neoplasms to represent mesonephric adenocarcinomas, but recommend that they be termed mesonephric‐like adenocarcinomas until their histogenesis is firmly established.” 31 A variable histological pattern of MLA. (a) Fused tubules composed of atypical columnar cells, containing eosinophilic secretes in the lumina; (b) dilated and fused glandular architecture; (c) ambiguous tubular manner and solid growth; (d) complex papillary protrusions in dilated glands; (e) glomeruloid or labyrinth‐like architecture; (f) chondroid matrix as heterotopic component. (a) Immunohistochemical expressions in MLA. (A) Estrogen receptor, completely negative; (B) TTF‐1, positive for tubular components; (C) GATA3, focally positive; (D) CD10, extensively positive. (b) Molecular analysis of MLA. Sanger sequencing demonstrating KRAS mutation [p.G12A (c.35 G > C)] but PIK3CA wild pattern (same case as (a)). MLA shows various characteristic histological patterns, such as an admixture of architectural patterns with small glands or tubules, some containing luminal eosinophilic colloid‐like material, typically predominating. Solid and papillary architectures are also often present, 31 , 32 , 137 and a component of spindle cells may also be present. 130 Marked histological heterogeneity is often observed with an admixture of varying architectural patterns, which may be useful for diagnosis. 32 According to IHC, MLAs are “flat” negative with ER and PgR and exhibit nuclear staining with TTF‐1. p53 expression is a wild‐type. CD10, calretinin, and GATA‐3 are positive in variable proportions. 31 , 32 , 137 , 138 MLA is categorized as NSMP subgroup in TCGA classification. 94 , 95 Although the morphological, immunohistochemical, and molecular profiles of MLA have recently been defined, little is known about its clinical behavior because of the small number of cases analyzed. MLA is rare, with an estimated prevalence of 1% of ECs, 139 but this seems uncertain because MLAs can be easily mistaken for various tumors, including EMA, CCC, SEC, and CS. 139 , 140 , 141 Some studies have concluded that MLA has a worse prognosis, presenting at higher stages and with frequent recurrences, especially showing a predilection for pulmonary recurrence. 32 , 139 , 140 In the new FIGO staging system for ECs, MLA is included in the aggressive tumor group. Distinct molecular aberrations have been defined, with both mesonephric adenocarcinoma and MLA classically demonstrating mutations in KRAS and NRAS 32 , 139 , 142 , 143 , 144 and a subset of MLA additionally demonstrating mutations in PIK3CA and ARID1A. 139 , 140 , 143 , 144 These findings suggest a biological overlap between MLA and EAC. 144 The variable clinicopathological findings support the prevailing notion that MLA is of a Müllerian ductal lineage exhibiting mesonephric differentiation. One case showed the coexistence of MLA and typical EAC with a transition/admixture between these two tumors. 137 In this case, EAC was evaluated to have been cured completely using hormone therapy (administration of medroxyprogesterone acetate [MPA]), but only MLA recurred, suggesting that MLA is refractory to MPA therapy. In ovarian MLA, the morphologic patterns of MLA overlap with those of endometrioid ovarian carcinoma, where endometriosis‐associated tumorigenesis has been suggested. 145

Masanori Yasuda drafted and completed this manuscript with figures.

TCGA and ProMisE have contributed immensely to establishing histomolecular classifications to improve the management of ECs. Grouping ECs based on molecular TCGA and ProMisE signatures are believed to help reduce inter‐ and intra‐observer variability in assigning histological types and grades of ECs. Further research is needed to resolve the remaining issues, including the substratification of NSMP ECs, the prognostic value of clinicopathological variables in MMRd ECs, treatment of advanced POLE mut ECs, and therapeutic strategies for SEC. 34 The new FIGO staging system for ECs may contribute to risk stratification with the assistance of molecular‐based techniques and immunohistochemical approaches. However, controversial and unresolved issues are present in the new staging system. ESGO/ESTRO/ESP designates the new staging system as “too much, too soon?.” The authors propose that the system should be first appraised at a multi‐institutional and global level. 146 According to the survey of the pathologists and clinicians for the new staging system, a slight preference for incorporating tumor histological type and molecular classification into staging is conveyed by both groups but the support was short of the 75% threshold formal consensus. 147 Amid the growing trend of molecular‐centered histopathological classification, the emergence of MLA as a new entity has emphasized the significance of morphological assessment from clinical and pathological aspects. Histological diagnoses, not limited to the gynecological tumor field, are consistently committed by pathologists. However, the diagnostic accuracy and risk stratification cannot be guaranteed without a close collaboration between pathologists and clinicians.

Endometrial carcinoma (EC) is the most commonly diagnosed gynecological cancer in industrialized countries and the second most common worldwide after cervical cancer. 1 EC is the 6th most common female cancer in Japan, 2 4th in the United States, 3 and 6th worldwide. 1 Approximately 417 367 women were diagnosed with uterine corpus cancer globally in 2020 (288 000 in 2008, 4 and the estimated age‐standardized incidence rate is 8.7 per 100 000, 1 and the estimated age‐standardized mortality rate is 1.8 per 100 000. 1 EC is the 11th leading cause of cancer‐related death in women worldwide, 1 and in Japan it ranks 13th between cervical cancer (12th) and ovarian cancer (9th) (Figures  1 and  2 ). 2 Conventional EC categorization. The endometrial environment drastically changes, causing differences in the tumor characteristics and propensities between pre‐ and post‐menopause. EAC is a representative tumor throughout any age and has been conventionally thought to develop from endometrial hyperplasia as a precursor lesion through hyperplasia‐carcinoma sequences. EACs G1 and G2, “so‐called type I tumors,” are categorized into low‐grade (non‐aggressive) EC. However, some EACs are proposed to arise de novo from the normal or atrophic endometrium in post‐menopause. High‐grade (aggressive) ECs, that is, “so‐called type II tumors,” which include EAC G3, SEC, CCC, MC, UDC/DDC, and CS, tend to increase in the post‐menopause. Histomolecular classification/subgroups of ECs. The 2020 WHO classification is a “hybrid” of TCGA and ProMisE, characterized by the four subgroups defined by distinctive prognoses as follows: POLE mut, excellent; MMRd, intermediate; NSMP, intermediate; and p53mut, poor. Using the surrogate markers provided by ProMisE for applying TCGA contributes to the risk stratification of ECs. The diagnostic algorithm of 2020 WHO classification is different from that of ProMisE in the initial analysis, whether selecting POLE mut or MMRd. 5 , 6 , 7 Around menopause, the endometrial landscape drastically changes from a dynamic to a static environment, reflecting the differences in tumor characteristics and propensities between pre‐ and post‐menopausal women. Post‐menopausal women with higher estrogen concentrations are at an increased risk of ECs, as are women with polycystic ovary syndrome or estrogen‐producing ovarian tumors, earlier age at menarche, later age at menopause, nulliparity, and obesity. 8 , 9 , 10 A family history of Lynch 11 , 12 or Cowden syndrome increases EC risk. 13 , 14 Protective factors against EC include later age at first and last birth, 15 , 16 continuous combined hormone replacement therapy, oral contraceptives (high progestin), intrauterine devices, smoking, and tubal ligation. 17 , 18 Endometrioid adenocarcinoma (EAC), a major histological type of ECs throughout any age, is defined as a malignant tumor displaying varying proportions of glandular, papillary, and solid architectures. EACs are heterogeneous in prognoses, because they range from being well to poorly differentiated. Based on the Federation of International Gynecology and Obstetrics (FIGO) grading system, EACs are divided into grade 1–3 (G1–G3) tumors and exhibit solid, non‐glandular, and non‐squamous growth in G1: 50%, and diagnosed as 54.1% for G1, 18.9% for G2, and 8.7% for G3. 19 The presence of severe cytological atypia in the majority of cells (>50%) increases the grade by one level, but serous carcinoma (SEC) should be excluded in cases with nuclear atypia that is out of proportion to the architecture. 20 Binary grading is recommended, whereby G1 and G2 EACs are classified as low‐grade and G3 EACs as high‐grade. 5 In terms of frequency, EAC is followed by SEC (6.1%), including serous endometrial intraepithelial carcinoma (SEIC), carcinosarcoma (CS; 4.6%), clear cell carcinoma (CCC; 2.3%), mixed carcinoma (MC; 1.7%), and undifferentiated carcinoma (UDC), which includes dedifferentiated carcinoma (DDC) (1.1%). 19 It has been four decades since Bokhman proposed the dualistic model, wherein ECs are pathologically, clinically, and epidemiologically divisible into type I and II tumors. 21 Type I tumors, predominantly endometrioid type, which tend to occur in young and obese patients, associated with excess estrogen, and show a favorable prognosis. Type II tumors, predominantly non‐endometrioid type, which tend to arise in older and non‐obese patients and are typically of serous histology without endometrial hyperplasia, and show a poor prognosis. 21 , 22 EAC is conventionally thought to develop from endometrial hyperplasia through the hyperplasia–carcinoma sequence in a hyperestrogenic background. However, a handful of previous studies have described a non‐negligible fraction of patients with EAC who lack concurrent hyperplasia. 23 A subset of EACs is proposed to arise de novo from a background of normal, but often atrophic, endometrium in post‐menopausal women. 23 The assessment of pathological factors that define high risk, especially in high‐grade ECs, is constrained by substantial inter‐observer variability. 24 , 25 Identifying patients who will benefit from chemotherapy remains challenging, and molecular classification may help determine appropriate adjuvant treatments. 24 , 26 For intermediate‐ or high‐risk patients, radiotherapy might also be considered and another merit of the molecular classification is that it can contribute to determining indication of minimally invasive surgery, that is, avoiding this procedure for potentially high‐grade endometrial carcinoma. 27 Key parameters such as histological type and grade in classical histopathological diagnosis, have been shown to have poor reproducibility, 24 , 28 , 29 reflected by pathologists often classifying similar tumors differently. 6 Thus, integrating histological features with molecular profiles may resolve these issues and may be the best approach for stratifying patient risk with EC in terms of prognosis and therapeutic strategy. The molecular classification of ECs introduced by The Cancer Genome Atlas (TCGA) has promoted a transition toward molecular‐based classification with a clear prognostic value for ECs. 30 Histological diagnosis of low‐grade EAC is consistently reproducible in most cases. 5 Therefore, in routine practice, no molecular analysis is required for most cases of low‐grade EAC. More than two‐thirds of ECs are well‐differentiated or low‐grade EACs at a low stage, suggesting an excellent prognosis. 5 However, high‐grade EACs are subject to considerable inter‐observer variability. Most EACs have a few copy number alterations or TP53 mutations. 30 SECs and 25% of high‐grade EACs show extensive copy number alterations and frequent TP53 mutations, indicating that high‐grade EACs have a molecular phenotype similar to that of SECs. A novel polymerase‐ε ( POLE )‐ultramutated ( POLE mut) subtype has previously been found in 10% of ECs, and a subset of EACs has been shown to harbor hotspot mutations in POLE. 30 Microsatellite instability (MSI) testing using seven repeat loci reveal MSI in 40% of EACs and 2% of SECs, which share genomic features with ovarian serous and basal‐like triple‐negative breast carcinomas. 30 EACs share many characteristics with colorectal carcinomas, including a high frequency of MSI (40% and 11%, respectively) and POLE mutations (7% and 3%, respectively), leading to ultrahigh mutation rates. 30 The compelling similarities between this subset of EACs and SECs suggest that genomic‐based classification may lead to the improved management of patients with these diseases. 30 Mesonephric‐like adenocarcinoma (MLA) and gastric‐type mucinous carcinoma (GMC) represent the novel histological types as new entities in The 2020 WHO Classification of Female Genital Tumors, 5th edition (2020 WHO classification). MLA histologically shows an admixture of varying architectural patterns mimicking mesonephric adenocarcinoma 31 and is immunohistochemically positive for thyroid transcription factor‐1 (TTF‐1) and GATA‐binding protein 3 (GATA‐3), but negative for estrogen receptor (ER) and progesterone receptor (PgR). 31 Although a rare entity, MLA is of clinicopathological significance in differentiation from other histological types. 32 This review focuses on exploring the newly published essential evidences in ECs, which have been brought consistently by using histomolecular‐based analyses, and incorporating them into the practical pathological diagnosis to help refine the risk stratification in ECs.

The author, Masanori Yasuda, is an Editorial Board member of Pathology International and was excluded from all editorial decision‐making related to the acceptance of this article for publication to minimize bias.

To date, EC risk has been stratified based on histological type, tumor grade, myometrial invasion depth, cervical and adnexal involvement, lymphovascular invasion, and lymph node metastasis. 34 Nevertheless, we are currently moving beyond treating disease according to anatomic site and dependence on histomorphology alone and toward tailoring therapy according to molecular profiles within an individual's tumor. 35 TCGA integrated genomic, transcriptomic, and proteomic analysis using array‐ and sequencing‐based technologies, 30 resulting in the classification of ECs into four subgroups: POLE mut; MSI hypermutated, copy number low (CNL), and copy number high (CNH). The integration of molecular classification with clinicopathological features has improved the prognostic accuracy of EC, highlighting the potential of molecular classification to refine and individualize patient risk stratification. 7 , 36 Although patients with EC generally have a good prognosis, 15%–20% of ECs have high‐risk disease with an increased incidence of distant metastases and tumor‐related death (Figures  2 , 3 , 4 , 5 , 6 ). 26 Frequency of TCGA subgroups across different histological types. POLE mutations are more frequent in EAC G3 than in EAC G1. A relatively high frequency of POLE mutations is also found in UDC/DDC, while the lowest frequency in CCC. p53mut is the second most frequently observed in CS, following SEC. 32 Prognoses demonstrated by disease‐free survival (DFS) or progression‐free survival (PFS). In these analyses, POLE mut subgroup is prognostically excellent in any EC (a–c), especially in a subset of EAC G3 (c), while p53mut/abn ECs are unexceptionally poor. MMRd ECs and p53wt/NSMP ECs are intermediate and close in prognosis. 3 , 26 , 33 (a) POLE ‐mutated EC. (A) Aggressive EC: EAC G3 focally characterized by tumor cells having large and/or multiple hyperchromatic nuclei; (B) p53 expression showing subclonal expression pattern. (b) POLE mutation analysis. Sanger sequencing demonstrating [p.P286R (c.857 C＞G)] in exon 9, (same case as (a)). (a) Immunohistochemical MMR evaluation. PMS2 and MSH6 can be initially selected in the routine practice. When PMS2 loss and/or MSH2 loss are recognized, MLH1 and/or MSH2 analyses should be performed as the next step, respectively. (b) Representative MMRd pattern. PMS2 loss and MSH6 loss may be due to loss of MLH1 and MSH2, respectively. Many cases with MLH1 loss can be caused by hypermethylation. The significant prognostic differences among the four TCGA molecular subgroups have been replicated by Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) using surrogate markers in formalin‐fixed, paraffin‐embedded tissues and molecular‐analytical techniques. 36 ProMisE is characterized as being analogous, but not identical, to the genomic classification defined in TCGA 7 , 36 : Mismatch repair deficiency (MMRd), showing loss of one or more mismatch repair proteins and corresponding to the hypermutated subtype; POLE mutations in the exonuclease domain in exons 9–14, corresponding to the ultramutated subtype; p53mut, demonstrating aberrant p53 immunohistochemical staining, corresponding to the CNH subtype; and p53 wild type (p53wt), corresponding to the CNL subtype. In contrast to TCGA methods, which depend on freshly frozen tissue and require costly and complex methodologies, ProMisE can be easily adopted in usual pathology laboratories at most cancer centers. In that, TCGA can be applied in pathological practice using simplified surrogates that include three immunohistochemical markers (p53, MMR‐related proteins) and one molecular analysis ( POLE mutations). The applicability of molecular classification to diagnostic specimens, that is, biopsy and curettage, is highly concordant with that of the hysterectomy specimens. 37 , 38 To validate these tools, ProMisE was applied to a large population‐based cohort of consecutive ECs from an independent institution. It has been mentioned that “integration of microscopic features with molecular characteristics is the best approach to stratify patients to predict prognosis in regions with the resources available to incorporate such techniques.” This classification presents the “diagnostic algorithm for integrated histomolecular EC classification ( POLE mut, MMRd, no specific molecular profile (NSMP)/p53wt, p53mut/abn)” by integration of TCGA and ProMisE. POLE mut ECs with POLE mutations are characterized by an overwhelmingly favorable prognosis, showing progression‐free survival of 92%–100%. 26 , 30 , 33 , 36 , 39 , 40 , 41 This is particularly useful for assessing prognosis in G3 EAC 26 , 42 and also potentially in other EACs. However, the molecular characterization of ECs is recommended to be restricted to high‐grade tumors. Unlike in the MMRd and p53mut subgroups, no immunohistochemical surrogate markers are available to identify the POLE mut subgroup. 7 , 43 The POLE mut subgroup is the least common molecular group in the TCGA series (7.3% of all ECs) and is labeled as an “ultramutated” based on its exceptionally high mutational burden (232 × 10 6 mutations per Mb). 30 Ultramutated ECs show pathogenetic mutations in the exonuclease domain of POLE. 30 Over 80% of the pathogenetic POLE mutations fall into one of five hotspots (P286R, V411L, S297F, A456P, and S459F). Sequencing methods that limit the analysis to these hotspots can identify 67%–92% of all POLE mut ECs. 7 , 44 Sanger sequencing is an acceptable option, as is panel‐based next‐generation sequencing (NGS), which incorporates POLE exons 9–14 or 9, 13, and 14. However, for clinical implementation, to avoid missed detection in this prognostic subgroup, it may be preferable to cover the complete exonuclease domain using NGS. POLE mutations have been shown to be significantly more frequent in high‐grade EC (12.1%) than in low‐grade EC (6.2%). 45 Morphological heterogeneity and marked atypia are common features of POLE mut ECs, which may also present as giant anaplastic cells. 46 , 47 , 48 A relatively high frequency of POLE mutations has been found in UDC/DDC (12.4%), 49 whereas low frequencies were found in CCC (3.8%) 50 and CS (5.3%). 51 Remarkably, POLE mutations have been found in a significant proportion of mixed endometrioid‐serous carcinomas in young women (16%), and these tumors are thought to arise as EAC that secondarily develop a serous morphology with or without p53 mutations. 52 Both high‐grade features and p53 mutations result from a high mutational load and have no clinical significance. A small subset of tumors (approximately 5%) combines more than one molecular feature (e.g., POLE mut and p53mut or MMRd and p53mut), and these are referred to as “multiple classifiers.” In the case of multiple classifiers with POLE mut or MMRd and secondary p53mut, the available scientific evidence indicates that these cases should not be classified as p53mut because they retain a favorable prognosis as those with POLE mut or MMRd. Therefore, patients with both POLE mut and p53mut should be classified as POLE mut, and those with both MMRd and p53mut should be considered MMRd. 33 , 53 , 54 POLE mutations prognostically supersede both MMRd and p53 mutations, 33 , 53 , 54 and POLE mut non‐endometrioid ECs still show good prognosis. 43 , 55 , 56 , 57 The mutational load itself causes a strong immune response due to exposure to several neoantigens, which is reflected in the lymphocytic infiltration that accompanies most (~79%) POLE mut ECs and could be responsible for their good prognosis. 46 , 58 However, Talhouk et al. found that the immune response is not independently associated with prognosis, suggesting that other factors drive prognosis in POLE mut ECs. 59 Given the prominent lymphocytic infiltrates found in most POLE mut ECs, it is reasonable to hypothesize that these tumors may benefit from immunotherapy. 59 The European Society of Gynecological Oncology (ESGO), the European Society for Radiotherapy and Oncology (ESTRO), and the European Society of Pathology (ESP) guidelines consider POLE mut ECs up to FIGO stage II as low‐risk tumors that do not need adjuvant treatment. 27 Irrespective of much data regarding the usefulness of identifying POLE mutations in ECs, this sequencing technique is not widely available, and its relatively high cost may constrain routine pathology. For the disadvantage of POLE mutation analysis, it is suggested that morphological screening of particularly ambiguous histological features and rare bizarre nuclei can be useful for the selective enrichment of ECs for POLE mutation analysis. 60 A simple combination of morphological and immunohistochemical characteristics (tumor type, grade, peritumoral lymphocytes, MMR‐related proteins and p53 expression) reportedly increases the probability of POLE mutations in EC from 7% to 33%. 61 MMRd The MMRd subgroup was first defined by TCGA as a “hypermutated” (18 × 10 6 mutations per Mb), with fewer mutations than those of the ultramutated POLE mut subgroup. The MMRd‐hypermutated subgroup was the second most common, accounting for 28% of ECs. 30 All hypermutated ECs exhibit high MSI, which is associated with the rapid accumulation of genomic mutations. 30 , 62 Because MSI is typically caused by a deficiency in the MMR system, immunohistochemistry (IHC) for MMR‐related proteins has been used as a surrogate test to identify the hypermutated subgroup. 35 Historically, MMR IHC and MSI assay have been optimized and designed to triage patients with colorectal cancer to identify Lynch syndrome. 63 , 64 The MMR IHC approach is reliable, with concordance rates for MSI assay ranging from 86% to 99%. 65 , 66 MMR‐related proteins heterodimerize into MutSa (MSH2–MSH6) and MutLa (MLH1–PMS2), 67 , 68 where MSH2 and MLH1 stabilize MSH6 and PMS2, respectively. 69 Therefore, performing MMR IHC for MSH6 and PMS2, which has the same accuracy as testing the full MMR panels, 70 has been demonstrated to be both sensitive and specific and saves cost. 42 , 71 As a positive internal control, the endometrial stroma and lymphocytes were clearly evaluated. Normal/retained expression of MMR‐related proteins consists of positivity in tumor cell nuclei, which should be stronger than stromal positivity. 72 The routine MMR IHC of all ECs is increasingly being advocated because ECs are the most common tumors in women with Lynch syndrome. 73 , 74 Subclonal or regional loss of MMR‐related protein expression has recently been shown to indicate that MMRd can occur during disease progression. 65 Notably, the concordance between MMR IHC and MSI assay was greater when subclonal/regional loss of expression was accounted for. 35 Compared with the MSI assay, MMR IHC is a more reliable method for identifying ECs with MSH6 mutations. 65 The triage cascade includes an assessment of the methylation status of the MLH1 promoter. 73 The MMRd subgroup shows several similarities with the POLE mut group and mainly comprises EACs (85.8%), which are more frequent in high‐grade than in low‐grade EACs (39.7 vs. 24.7%). 75 The MMRd frequencies are as follows: 44% for UDC/DDC, 49 16 to 66% for MC, 52 , 76 9.8% for CCC, 50 and 7.3% for CS. 51 Histological types have been suggested to have no prognostic value in MMRd ECs. 53 The overall prognosis of MMRd ECs is intermediate; MMRd prognostically supersedes p53mut but is superseded by POLE mut. 33 , 53 , 54 SWI/SNF‐deficient carcinomas have shown an exceedingly poor prognosis even in the presence of an MMRd signature. 43 Notably, MMRd ECs associated with MLH1 promoter methylation have been suggested to have a poorer prognosis than MMRd EC associated with MMR gene mutation. 43 MMRd ECs have shown higher susceptibility to radiotherapy than MMRp ECs 77 and are candidates for immunotherapy. 78 p53mut/CNH After excluding ultra‐ and hype‐mutated ECs, TCGA subdivided ECs with a low mutational load into two subgroups: CNH and CNL. The CNH subgroup is characterized by a high frequency of TP53 mutation (85%) and serous morphology (73.3%), designated the “serous group”, 30 representing prototypical type II tumors. 30 , 79 , 80 The CNH/serous group has subsequently been termed the p53mut group. 27 , 30 , 36 , 53 p53 IHC is a widely used surrogate marker that reflects the mutational status of TP53 , and recent reports have shown approximately 95% concordance. 49 The systematic assessment of p53 immunohistochemical expression in TP53 ‐mutant ECs has shown three possible aberrant patterns of p53: overexpression (strong diffuse expression in >70%–80% of tumor cells, accounting for 85.6% of cases), complete loss/null pattern (11.5% of cases), and cytoplasmic expression (1.9% of cases). 81 A cytoplasmic pattern of p53 immunohistochemical expression was recently reported in a rare subset of pelvic and endometrial cancers, with TP53 mutations involving domains that affect nuclear localization. 82 The p53 cytoplasmic pattern is observed in 100% of tumors with TP53 mutation in the nuclear localization domain and in 33%–44% of tumors with a mutation in the adjacent tetramerization domain or nuclear exclusion sequence. 82 Alternative mechanisms for cytoplasmic localization (other than mutations involving the nuclear localization domain) may exist, possibly due to conformational changes or post‐translational modifications of the aberrant p53 protein. 83 The presence of a subclonal p53mut pattern is often associated with MMRd or POLE mutation. 81 In the presence of these signatures, p53mut has no prognostic value. 54 A small subset of TP53 ‐mutant ECs (~5%) was shown not to display abnormalities in p53 expression and thus could not be identified by IHC. 84 In addition, a subset of CNH ECs do not show TP53 mutations and can only be classified by molecular analysis of copy number variations. 30 The p53mut signature is far more common in high‐grade than in low‐grade EACs (21.3% vs. 4.7%). 45 p53mut low‐grade EACs can be observed in elderly patients. 85 , 86 , 87 A p53mut signature is present in virtually all serous ECs, 53 including the vast majority of CSs (73.9%) 51 and almost half of CCCs (42.5%). 50 The biological behavior of p53mut ECs is consistently aggressive across different histological types. 27 , 36 , 53 , 56 , 57 The ESGO/ESTRO/ESP guidelines include all p53mut ECs in the high‐risk group (except for non‐myoinvasive cases). 27 However, SEC and p53mut EAC can be difficult to strictly differentiate because their morphological features do not distinctly differ. Therefore, combining all p53mut ECs into the same risk group seems appropriate. 48 , 53 Regarding treatment, p53mut ECs with myoinvasion always need adjuvant treatment. 27 Human epidermal growth factor receptor type 2 (HER2) amplification has been identified as a therapeutic target in a subset of p53mut carcinomas, regardless of histological type. 88 , 89 , 90 , 91 High poly ADP‐ribose polymerase‐1 (PARP‐1) expression has also been observed in p53mut ECs, suggesting the possibility of using PARP inhibitors. 92 NSMP/CNL The remaining TCGA group showed neither a high mutational load nor significant copy number variations and was designated the “CNL/endometrioid group,” considered a prototypical type I tumor. 30 Therefore, the group has been classified as NSMP, 27 , 53 which is the most frequent TCGA subgroup (~40% of cases), showing an intermediate prognosis, and is similar to the MMRd subgroup. 27 , 30 , 36 , 53 The vast majority (84.4%) of NSMP ECs are low‐grade EACs, 80 and the NSMP subgroup can be found in almost any EC histological type. 45 , 49 , 51 , 76 Minor SECs with a CNH signature may lack abnormal immunohistochemical expression of p53 and TP53 mutations, resulting in NSMP classification. 34 These cases are considered biologically and prognostically analogous to p53mut serous ECs. 30 , 81 , 84 The critical features, that is, low‐grade and ER expression, characterize the low‐risk NSMP subgroup for which de‐escalation and/or endocrine therapy strategies can be applied. G3 EAC and/or ER‐negative status indicates a high‐risk NSMP subgroup, including high‐grade ECs such as CCC, DDC, and MLA, which are responsible for most NSMP‐related deaths. 93 The ESGO/ESTRO/ESP guidelines recommend substratifying the NSMP group based on the same criteria as the MMRd subgroup. 27 However, it is suggests that NSMP is more prognostically heterogeneous and more heavily affected by other clinicopathological factors than the MMRd subgroup. 53 Non‐endometrioid NSMP ECs have a poor prognosis, similar to that of p53mut ECs, 56 , 57 , 94 , 95 whereas NSMP EACs have a heterogeneous prognosis, ranging from as good as that of POLE mut ECs to as poor as that of p53mut ECs. 42 , 47 , 96 Therefore, several authors have proposed possible substratification of NSMP ECs based on histological, immunohistochemical, and molecular markers, some of which may constitute to therapeutic targets. 92 , 97 , 98 The subclassification of NSMP ECs is currently under evaluation in the PORTEC‐4a study. 97 POLE mut ECs with POLE mutations are characterized by an overwhelmingly favorable prognosis, showing progression‐free survival of 92%–100%. 26 , 30 , 33 , 36 , 39 , 40 , 41 This is particularly useful for assessing prognosis in G3 EAC 26 , 42 and also potentially in other EACs. However, the molecular characterization of ECs is recommended to be restricted to high‐grade tumors. Unlike in the MMRd and p53mut subgroups, no immunohistochemical surrogate markers are available to identify the POLE mut subgroup. 7 , 43 The POLE mut subgroup is the least common molecular group in the TCGA series (7.3% of all ECs) and is labeled as an “ultramutated” based on its exceptionally high mutational burden (232 × 10 6 mutations per Mb). 30 Ultramutated ECs show pathogenetic mutations in the exonuclease domain of POLE. 30 Over 80% of the pathogenetic POLE mutations fall into one of five hotspots (P286R, V411L, S297F, A456P, and S459F). Sequencing methods that limit the analysis to these hotspots can identify 67%–92% of all POLE mut ECs. 7 , 44 Sanger sequencing is an acceptable option, as is panel‐based next‐generation sequencing (NGS), which incorporates POLE exons 9–14 or 9, 13, and 14. However, for clinical implementation, to avoid missed detection in this prognostic subgroup, it may be preferable to cover the complete exonuclease domain using NGS. POLE mutations have been shown to be significantly more frequent in high‐grade EC (12.1%) than in low‐grade EC (6.2%). 45 Morphological heterogeneity and marked atypia are common features of POLE mut ECs, which may also present as giant anaplastic cells. 46 , 47 , 48 A relatively high frequency of POLE mutations has been found in UDC/DDC (12.4%), 49 whereas low frequencies were found in CCC (3.8%) 50 and CS (5.3%). 51 Remarkably, POLE mutations have been found in a significant proportion of mixed endometrioid‐serous carcinomas in young women (16%), and these tumors are thought to arise as EAC that secondarily develop a serous morphology with or without p53 mutations. 52 Both high‐grade features and p53 mutations result from a high mutational load and have no clinical significance. A small subset of tumors (approximately 5%) combines more than one molecular feature (e.g., POLE mut and p53mut or MMRd and p53mut), and these are referred to as “multiple classifiers.” In the case of multiple classifiers with POLE mut or MMRd and secondary p53mut, the available scientific evidence indicates that these cases should not be classified as p53mut because they retain a favorable prognosis as those with POLE mut or MMRd. Therefore, patients with both POLE mut and p53mut should be classified as POLE mut, and those with both MMRd and p53mut should be considered MMRd. 33 , 53 , 54 POLE mutations prognostically supersede both MMRd and p53 mutations, 33 , 53 , 54 and POLE mut non‐endometrioid ECs still show good prognosis. 43 , 55 , 56 , 57 The mutational load itself causes a strong immune response due to exposure to several neoantigens, which is reflected in the lymphocytic infiltration that accompanies most (~79%) POLE mut ECs and could be responsible for their good prognosis. 46 , 58 However, Talhouk et al. found that the immune response is not independently associated with prognosis, suggesting that other factors drive prognosis in POLE mut ECs. 59 Given the prominent lymphocytic infiltrates found in most POLE mut ECs, it is reasonable to hypothesize that these tumors may benefit from immunotherapy. 59 The European Society of Gynecological Oncology (ESGO), the European Society for Radiotherapy and Oncology (ESTRO), and the European Society of Pathology (ESP) guidelines consider POLE mut ECs up to FIGO stage II as low‐risk tumors that do not need adjuvant treatment. 27 Irrespective of much data regarding the usefulness of identifying POLE mutations in ECs, this sequencing technique is not widely available, and its relatively high cost may constrain routine pathology. For the disadvantage of POLE mutation analysis, it is suggested that morphological screening of particularly ambiguous histological features and rare bizarre nuclei can be useful for the selective enrichment of ECs for POLE mutation analysis. 60 A simple combination of morphological and immunohistochemical characteristics (tumor type, grade, peritumoral lymphocytes, MMR‐related proteins and p53 expression) reportedly increases the probability of POLE mutations in EC from 7% to 33%. 61 MMRd The MMRd subgroup was first defined by TCGA as a “hypermutated” (18 × 10 6 mutations per Mb), with fewer mutations than those of the ultramutated POLE mut subgroup. The MMRd‐hypermutated subgroup was the second most common, accounting for 28% of ECs. 30 All hypermutated ECs exhibit high MSI, which is associated with the rapid accumulation of genomic mutations. 30 , 62 Because MSI is typically caused by a deficiency in the MMR system, immunohistochemistry (IHC) for MMR‐related proteins has been used as a surrogate test to identify the hypermutated subgroup. 35 Historically, MMR IHC and MSI assay have been optimized and designed to triage patients with colorectal cancer to identify Lynch syndrome. 63 , 64 The MMR IHC approach is reliable, with concordance rates for MSI assay ranging from 86% to 99%. 65 , 66 MMR‐related proteins heterodimerize into MutSa (MSH2–MSH6) and MutLa (MLH1–PMS2), 67 , 68 where MSH2 and MLH1 stabilize MSH6 and PMS2, respectively. 69 Therefore, performing MMR IHC for MSH6 and PMS2, which has the same accuracy as testing the full MMR panels, 70 has been demonstrated to be both sensitive and specific and saves cost. 42 , 71 As a positive internal control, the endometrial stroma and lymphocytes were clearly evaluated. Normal/retained expression of MMR‐related proteins consists of positivity in tumor cell nuclei, which should be stronger than stromal positivity. 72 The routine MMR IHC of all ECs is increasingly being advocated because ECs are the most common tumors in women with Lynch syndrome. 73 , 74 Subclonal or regional loss of MMR‐related protein expression has recently been shown to indicate that MMRd can occur during disease progression. 65 Notably, the concordance between MMR IHC and MSI assay was greater when subclonal/regional loss of expression was accounted for. 35 Compared with the MSI assay, MMR IHC is a more reliable method for identifying ECs with MSH6 mutations. 65 The triage cascade includes an assessment of the methylation status of the MLH1 promoter. 73 The MMRd subgroup shows several similarities with the POLE mut group and mainly comprises EACs (85.8%), which are more frequent in high‐grade than in low‐grade EACs (39.7 vs. 24.7%). 75 The MMRd frequencies are as follows: 44% for UDC/DDC, 49 16 to 66% for MC, 52 , 76 9.8% for CCC, 50 and 7.3% for CS. 51 Histological types have been suggested to have no prognostic value in MMRd ECs. 53 The overall prognosis of MMRd ECs is intermediate; MMRd prognostically supersedes p53mut but is superseded by POLE mut. 33 , 53 , 54 SWI/SNF‐deficient carcinomas have shown an exceedingly poor prognosis even in the presence of an MMRd signature. 43 Notably, MMRd ECs associated with MLH1 promoter methylation have been suggested to have a poorer prognosis than MMRd EC associated with MMR gene mutation. 43 MMRd ECs have shown higher susceptibility to radiotherapy than MMRp ECs 77 and are candidates for immunotherapy. 78 p53mut/CNH After excluding ultra‐ and hype‐mutated ECs, TCGA subdivided ECs with a low mutational load into two subgroups: CNH and CNL. The CNH subgroup is characterized by a high frequency of TP53 mutation (85%) and serous morphology (73.3%), designated the “serous group”, 30 representing prototypical type II tumors. 30 , 79 , 80 The CNH/serous group has subsequently been termed the p53mut group. 27 , 30 , 36 , 53 p53 IHC is a widely used surrogate marker that reflects the mutational status of TP53 , and recent reports have shown approximately 95% concordance. 49 The systematic assessment of p53 immunohistochemical expression in TP53 ‐mutant ECs has shown three possible aberrant patterns of p53: overexpression (strong diffuse expression in >70%–80% of tumor cells, accounting for 85.6% of cases), complete loss/null pattern (11.5% of cases), and cytoplasmic expression (1.9% of cases). 81 A cytoplasmic pattern of p53 immunohistochemical expression was recently reported in a rare subset of pelvic and endometrial cancers, with TP53 mutations involving domains that affect nuclear localization. 82 The p53 cytoplasmic pattern is observed in 100% of tumors with TP53 mutation in the nuclear localization domain and in 33%–44% of tumors with a mutation in the adjacent tetramerization domain or nuclear exclusion sequence. 82 Alternative mechanisms for cytoplasmic localization (other than mutations involving the nuclear localization domain) may exist, possibly due to conformational changes or post‐translational modifications of the aberrant p53 protein. 83 The presence of a subclonal p53mut pattern is often associated with MMRd or POLE mutation. 81 In the presence of these signatures, p53mut has no prognostic value. 54 A small subset of TP53 ‐mutant ECs (~5%) was shown not to display abnormalities in p53 expression and thus could not be identified by IHC. 84 In addition, a subset of CNH ECs do not show TP53 mutations and can only be classified by molecular analysis of copy number variations. 30 The p53mut signature is far more common in high‐grade than in low‐grade EACs (21.3% vs. 4.7%). 45 p53mut low‐grade EACs can be observed in elderly patients. 85 , 86 , 87 A p53mut signature is present in virtually all serous ECs, 53 including the vast majority of CSs (73.9%) 51 and almost half of CCCs (42.5%). 50 The biological behavior of p53mut ECs is consistently aggressive across different histological types. 27 , 36 , 53 , 56 , 57 The ESGO/ESTRO/ESP guidelines include all p53mut ECs in the high‐risk group (except for non‐myoinvasive cases). 27 However, SEC and p53mut EAC can be difficult to strictly differentiate because their morphological features do not distinctly differ. Therefore, combining all p53mut ECs into the same risk group seems appropriate. 48 , 53 Regarding treatment, p53mut ECs with myoinvasion always need adjuvant treatment. 27 Human epidermal growth factor receptor type 2 (HER2) amplification has been identified as a therapeutic target in a subset of p53mut carcinomas, regardless of histological type. 88 , 89 , 90 , 91 High poly ADP‐ribose polymerase‐1 (PARP‐1) expression has also been observed in p53mut ECs, suggesting the possibility of using PARP inhibitors. 92 NSMP/CNL The remaining TCGA group showed neither a high mutational load nor significant copy number variations and was designated the “CNL/endometrioid group,” considered a prototypical type I tumor. 30 Therefore, the group has been classified as NSMP, 27 , 53 which is the most frequent TCGA subgroup (~40% of cases), showing an intermediate prognosis, and is similar to the MMRd subgroup. 27 , 30 , 36 , 53 The vast majority (84.4%) of NSMP ECs are low‐grade EACs, 80 and the NSMP subgroup can be found in almost any EC histological type. 45 , 49 , 51 , 76 Minor SECs with a CNH signature may lack abnormal immunohistochemical expression of p53 and TP53 mutations, resulting in NSMP classification. 34 These cases are considered biologically and prognostically analogous to p53mut serous ECs. 30 , 81 , 84 The critical features, that is, low‐grade and ER expression, characterize the low‐risk NSMP subgroup for which de‐escalation and/or endocrine therapy strategies can be applied. G3 EAC and/or ER‐negative status indicates a high‐risk NSMP subgroup, including high‐grade ECs such as CCC, DDC, and MLA, which are responsible for most NSMP‐related deaths. 93 The ESGO/ESTRO/ESP guidelines recommend substratifying the NSMP group based on the same criteria as the MMRd subgroup. 27 However, it is suggests that NSMP is more prognostically heterogeneous and more heavily affected by other clinicopathological factors than the MMRd subgroup. 53 Non‐endometrioid NSMP ECs have a poor prognosis, similar to that of p53mut ECs, 56 , 57 , 94 , 95 whereas NSMP EACs have a heterogeneous prognosis, ranging from as good as that of POLE mut ECs to as poor as that of p53mut ECs. 42 , 47 , 96 Therefore, several authors have proposed possible substratification of NSMP ECs based on histological, immunohistochemical, and molecular markers, some of which may constitute to therapeutic targets. 92 , 97 , 98 The subclassification of NSMP ECs is currently under evaluation in the PORTEC‐4a study. 97