Volume 253, Issue 4 p. 355-365
Invited Review
Free Access

Endometrial carcinoma: molecular subtypes, precursors and the role of pathology in early diagnosis

Jutta Huvila

Jutta Huvila

Department of Pathology, University of Turku, Turku, Finland

Department of Pathology and Laboratory Medicine, Vancouver General Hospital and University of British Columbia, Vancouver, Canada

Search for more papers by this author
Jennifer Pors

Jennifer Pors

Department of Pathology and Laboratory Medicine, Vancouver General Hospital and University of British Columbia, Vancouver, Canada

Search for more papers by this author
Emily F Thompson

Emily F Thompson

Department of Pathology and Laboratory Medicine, Vancouver General Hospital and University of British Columbia, Vancouver, Canada

Search for more papers by this author
C Blake Gilks

Corresponding Author

C Blake Gilks

Department of Pathology and Laboratory Medicine, Vancouver General Hospital and University of British Columbia, Vancouver, Canada

Correspondence to: C Blake Gilks, MD, FRCPC, Department of Pathology, Vancouver General Hospital, Rm 1435, 1st Floor JPPN, 855 West 12th Avenue, Vancouver, BC, V5Z 1M9, Canada.

E-mail: [email protected]

Search for more papers by this author
First published: 25 December 2020
Citations: 45

No conflicts of interest were declared.


Endometrial carcinoma (EC) is classified into a wide range of morphological variants; this list has expanded over the past decade with the inclusion of mesonephric-like and dedifferentiated carcinoma as EC variants in the fifth edition of the WHO Classification of Female Genital Tumours, and recognition that carcinosarcoma is a biphasic carcinoma rather than a sarcoma. Each EC variant has distinct molecular abnormalities, including TCGA-based molecular subtypes, allowing further subclassification and adding complexity. In contrast to this rapid progress in understanding EC, there are only two recognized EC precursor lesions: endometrial atypical hyperplasia/endometrioid intraepithelial neoplasia (EAH/EIN) and serous intraepithelial carcinoma, a situation that has not changed for many years. Diagnosis of EC precursors is a cornerstone of surgical pathology practice, with early diagnosis contributing to the relatively favorable prognosis of EC. In this review we relate the precursor lesions to each of the EC morphological variants and molecular subtypes, discuss how successful early diagnosis is for each variant/molecular subtype and how it might be improved, and identify knowledge gaps where there is insufficient understanding of EC histogenesis. © 2020 The Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Endometrial carcinoma (EC) is the most common cancer of the female reproductive organs in the developed world, and the fourth most common cancer in women in Canada, the UK, and the USA, after breast, lung, and colorectal [1-3]. Although many EC patients are cured with surgery alone, there are significant numbers of women with more aggressive variants of EC for whom the prognosis remains poor. The main clinical challenges include risk stratification based on the diagnostic biopsy to guide the extent of surgery, and risk stratification/predictive biomarker assessment post-hysterectomy to determine the need for and type of adjuvant treatment [4].

Carcinoma of the uterus was first subclassified based on site of origin into those tumors involving the uterine corpus and arising from the endometrium, and those arising in the cervix [5]. The former were subclassified in 1983 by Bokhman, who recognized two types of EC based on clinical and pathological features; type I ECs were more common, related to excess unopposed estrogenic stimulation of the endometrium, and were mostly endometrioid histotype and of lower grade. The prognosis of patients with type I EC was favorable, with a greater than 85% 5-year disease-free survival rate. In contrast, type II ECs were high-grade tumors of clinically aggressive histotypes (e.g. serous, clear cell; although it should be noted that within this ‘non-endometrioid’ category not all clear cell carcinomas behave in an aggressive fashion). Type II ECs were associated with a poor response to hormonal therapy and relatively poor outcomes [6, 7]. This categorization system was a major conceptual advance, but type I/type II classification identified loose constellations of cases illustrating biological truths, rather than specific diagnoses; although there are prototypical examples of type I and type II EC, too many ECs have features that are not readily classified as one or the other with certainty. The category of grade 3 endometrioid carcinoma, for example, has been considered type I or type II [8, 9].

In 2013, The Cancer Genome Atlas (TCGA) identified four molecular subtypes of EC based on genomic features [10]. These molecular subtypes differ from each other with respect to molecular abnormalities, hereditary risk factors, environmental risk factors, prognosis, and response to treatment [11, 12]. Furthermore, clinically applicable surrogate markers for the four molecular subtypes have been independently developed by two different groups of investigators [13-16]; as a result, molecular subtype can be diagnosed in practice with a high degree of inter-observer reproducibility [17], and a diagnosis based on a biopsy specimen shows excellent correlation with the molecular classification based on the subsequent hysterectomy specimen [18-21].

The relationships between TCGA-based molecular subtypes and histotypes are shown in Figure 1 and Table 1 [10, 13, 15, 16, 19, 22-34], and the four molecular subtypes are described below. It should be noted that there may be publication bias in the cases included in Table 1, such that they may not be representative of a population-based case series. We have chosen not to include mixed EC in the figure as most so-called mixed carcinomas are examples of morphological mimicry rather than a true admixture of different tumor types [35].

Details are in the caption following the image
The relationship between histotype (top row) and molecular subtype (bottom row) in endometrial carcinoma. The thickness of the lines correlates with percent of tumors of each histotype that are of the corresponding molecular subtype. MMRd, mismatch repair deficient; NSMP, no specific molecular profile; abn, abnormal; mut, mutated; EEC, endometrioid endometrial carcinoma; CCC, clear cell carcinoma; SIEC, serous intraepithelial endometrial carcinoma; SEC, serous endometrial carcinoma; DDEC, dedifferentiated endometrial carcinoma; UDEC, undifferentiated endometrial carcinoma; CS, carcinosarcoma; Gr, grade.
Table 1. Relationship between endometrial carcinoma histotype and molecular subtype.
POLEmut MMRd NSMP p53abn
Total n % n % n % n %
EEC grade 1–2 2515 150 6.0 720 28.6 1515 60.2 130 5.2
EEC grade 3 900 107 11.9 342 38.0 252 28.0 199 22.1
CCC 61 1 1.6 6 9.8 31 50.8 23 37.7
SEC 122 0 0.0 3 2.5 6 4.9 113 92.6
CS 244 13 5.3 30 12.3 23 9.4 178 73.0
UDEC 60 4 6.7 22 36.7 17 28.3 17 28.3
DDEC 28 8 28.6 12 42.9 6 21.4 2 7.1
Mixed/NEEC 187 11 5.9 31 16.6 18 9.6 127 67.9
Total 4117 294 7.1 1166 28.3 1868 45.4 789 19.2
  • References: [10, 13, 15, 16, 19, 22-34].
  • POLEmut, pathogenic mutation in the exonuclease domain of POLE; MMRd, mismatch repair deficient; NSMP, no specific molecular profile; abn, abnormal; EEC, endometrioid endometrial carcinoma; CCC, clear cell carcinoma; SEC, serous endometrial carcinoma; DDEC, dedifferentiated endometrial carcinoma; UDEC, undifferentiated endometrial carcinoma; CS, carcinosarcoma; NEEC, non-endometrioid endometrial carcinoma.

POLEmut, ultramutated group

These copy-number-stable ECs feature pathogenic mutations in the exonuclease domain of DNA polymerase epsilon (POLE), a gene involved in DNA replication and repair [36-39]. These ECs have one of the highest somatic mutation frequencies of any solid tumor, frequently exceeding 100 mutations per megabase (Mb). POLEmut ECs are most often, but not exclusively, of endometrioid histotype. They may show intratumoral morphological heterogeneity and ambiguous morphology, with features of both endometrioid and serous histotypes. Women with POLEmut ECs tend to be younger, with a normal BMI, and are associated with a favorable prognosis despite often having high-risk pathologic features such as high tumor grade and extensive lymphovascular space invasion (LVSI) (>96% 5-year survival, confirmed across multiple studies) [40-43].

MMRd, hypermutated/microsatellite unstable group

These tumors have low levels of somatic copy-number alterations but a very high mutational burden secondary to dysfunctional mismatch repair (MMRd) proteins, i.e. MLH1, PMS2, MSH2, or MSH6 [44]. Epigenetic silencing of MLH1 is responsible for the majority of this subgroup but both somatic and germline mutations in any of the mismatch repair genes will lead to a high mutational frequency (>10 mut/Mb). These ECs occur across a wide age range (younger in patients with Lynch syndrome than in sporadic cases) and are not associated with an increased BMI. They are predominantly of endometrioid histotype and are often higher grade. There is FDA approval of PD-1/PD-L1 inhibitors for this molecular subtype of EC, in the setting of recurrent or advanced disease, when there are no other treatment options [45-47].

NSMP (no specific molecular profile) group

These ECs are generally genomically stable, with low levels of somatic copy-number alterations [48], and are MMR-proficient, POLE wild-type and show wild-type-pattern p53 immunoreactivity/wild-type TP53 on sequencing. This group encompasses mostly endometrioid tumors with high levels of estrogen and progesterone receptor (ER, PR) expression. Mutations in exon 3 of CTNNB1 are associated with a worse prognosis in NSMP EC [49]. This molecular subtype is associated with increased BMI.

p53abn, serous-like group

ECs in this molecular subgroup have high somatic copy-number (CN) alterations, similar to high-grade serous tubo-ovarian and basal-like breast carcinomas. Abnormal p53 immunostaining/TP53 mutations are the hallmarks of this molecular subtype, and also high-grade serous tubo-ovarian and basal-like breast carcinomas. In contrast to serous tubo-ovarian and basal-like breast carcinomas, however, p53abn ECs have recurrent mutations in PIK3CA, FBXW7, and PPP2R1A [10]; they are less likely to have mutations in BRCA1 or BRCA2 compared with tubo-ovarian high-grade serous carcinoma [10, 50, 51]. HER2 amplification occurs in ~20% of CN-high ECs and a proportion of CN-high ECs have homologous recombination deficiency [51, 52]. A large majority of serous carcinomas and carcinosarcomas are of this molecular subtype, but it also includes most ECs of mixed histology, and a minority of high-grade endometrioid ECs (Table 1). Although only accounting for 15% of ECs, this molecular subtype accounts for 50–70% of EC-associated mortality [10, 13-15, 19]. Patients with this molecular subtype are older, and it is not associated with increased BMI. The PORTEC-3 clinical trial results suggest that patients with p53abn ECs have superior outcomes when treated with chemotherapy in addition to radiation, compared with radiotherapy alone [12]. It has become clear that EC is too heterogeneous a disease to be classified using a simple binary system, as more than 30% of ECs are of MMRd or POLEmut molecular subtypes, which do not correspond exclusively to either type I or type II [25, 26, 53].

Endometrial biopsy performance in diagnosis of endometrial carcinoma

Endometrial biopsy and curetting specimens are one of the most common gynecological pathology specimen types. For the purposes of this discussion, we will consider biopsy and curettage specimens to be equivalent, given their equivalent diagnostic performance in the diagnosis of EC, and will hereafter refer to them as ‘biopsy/biopsies’. With a markedly reduced role of endometrial biopsy in the assessment of patients with infertility [54], most endometrial biopsies are taken for investigation of abnormal bleeding in peri- or post-menopausal women, in order to rule out EC or a precursor lesion. In a review of more than 22,000 endometrial biopsies from our region, there were 172 that were negative for carcinoma or a precursor from patients who subsequently developed EC within 5 years of the negative biopsy [55]. The sensitivity of endometrial biopsy, based on this study, was 89%, a figure unchanged from a study performed more than 15 years previously [56]. This is probably a conservative estimate of the sensitivity as some lesions may have truly been absent at the time of biopsy and developed in the interval between biopsy and hysterectomy, but as this was less than a year in a majority of the cases, it is unlikely to explain most of the lack of sensitivity. Thus, endometrial biopsy is a good but not perfect test for the diagnosis of EC, with false-negative results in approximately 10% of patients with EC. There remain patients who present with advanced-stage EC and these disproportionately consist of patients with p53abn EC, as noted previously. This suggests that the use of endometrial biopsy as a tool for early diagnosis of EC, i.e. at low stage, based on biopsy of women with abnormal bleeding, is more effective in some subtypes of EC than in others; in particular, early diagnosis, while the norm in low-grade endometrioid NSMP carcinomas, is less likely in the other molecular subtypes (Table 2). As for the case series included in Table 1, there may be publication bias in the cases included in Table 2, such that they are not representative of a population-based series of EC. The goal of reducing mortality due to any carcinoma is based on one or more elements of the triad of prevention, early detection, and more effective treatment. In the case of EC, early detection has been possible for many patients, but it is worth considering why it has not been more successful and if there are opportunities for improvement, in order to mitigate the impact of the increasing incidence of EC.

Table 2. Relationship between molecular subtype and endometrial carcinoma stage at presentation.
POLEmut MMRd NSMP p53abn
Stage n % n % n % n %
I 687 78 94.0 170 73.6 358 83.6 81 48.8
II–IV 221 5 6.0 61 26.4 70 16.4 85 51.2
Total 908 83 9.1 231 25.4 428 47.1 166 18.3
  • References: [13, 15, 19].
  • POLEmut, pathogenic mutation in the exonuclease domain of POLE; MMRd, mismatch repair deficient; NSMP, no specific molecular profile; abn, abnormal; mut, mutated.

Precursors of endometrial carcinoma

Recognition of EC precursors and their accurate diagnosis on biopsy or curetting specimens are important because of the opportunity for early diagnosis and cure before progression to EC. Two morphological precursor lesions of EC are recognized:
  1. Endometrial atypical hyperplasia/endometrioid intraepithelial neoplasia (EAH/EIN) is characterized by a clonal proliferation of endometrial glands lined by atypical cells, with a predominance of glands over stroma (Figure 2A) [57]. Atypical hyperplasia is associated with unopposed estrogenic stimulation of the endometrium and acquisition of mutations in PTEN, KRAS, PIK3CA, CTNNB1, and/or ARID1A [57]. EAH/EIN is the precursor lesion of almost all endometrioid carcinomas and a subset of serous carcinomas [58]. The mutational profile of EAH/EIN and the concurrent EC is highly concordant [18, 59-62] in the majority of cases; however, Li et al showed that, in 5 out of 30 EAH/EIC cases with concurrent EEC, the EAH/EIC and EC shared less than 5% of the mutations identified, indicating clonality but with a high degree of divergence [62].
  2. Serous intraepithelial carcinoma (SEIC) is characterized by pre-existing endometrial glands lined by markedly atypical glandular epithelial cells in the absence of invasive disease. These epithelial cells show cytological features typical of serous carcinoma, i.e. nucleomegaly and pleomorphism, and are associated with a high proliferative index (Figure 2B,C) [63-65]. As the name implies, this lesion is the precursor of some invasive serous carcinomas, but the exact percentage of cases that arise from it is not known as serous EC can also arise through progression of a morphologically low-grade lesion, i.e. EAH/EIN, with genetic features of endometrioid EC, e.g. PTEN loss. One hypothesis is that the SEIC is preceded by a sequence similar to that of tubo-ovarian high-grade serous carcinoma, i.e. endometrial p53 signature lesion and glandular dysplasia followed by SEIC [66-68], analogous to the p53 signature, serous tubal intraepithelial lesion (STIL), and serous tubal intraepithelial carcinoma (STIC). The term SEIC is not used by many pathologists because such lesions, which are apparently confined to the endometrium without endometrial or myometrial invasion, can be associated with extra-uterine spread; thus, the alternative diagnosis of ‘early serous carcinoma’ has been proposed [63, 65].
Details are in the caption following the image
(A) Endometrial atypical hyperplasia/endometrioid intraepithelial neoplasia, with glandular crowding and cytological atypia. Note the abrupt change in cytological features from the benign glands on the left (arrow) (H&E). (B, C) Serous intraepithelial carcinoma (SEIC) involving an endometrial polyp, with pre-existing glands lined by markedly atypical cells (B) (H&E), that show mutant pattern p53 immunostaining, with nuclear overexpression (C). (D, E) p53abn endometrial carcinoma with a low-grade component on the right and a high-grade component on the left (D) (H&E). Mutant pattern p53 (overexpression) is seen in the high-grade but not the low-grade component (E). Note that in some such carcinomas both the low-grade and the high-grade components show mutant pattern p53 staining. (F) MMRd endometrial carcinoma showing MSH6 loss of expression in both the carcinoma (bottom) and EAH/EIN (top). (G–J) Clear cell carcinoma (G) with adjacent atypical glandular proliferation (glandular crowding and mild cytological atypia (H). Another clear cell carcinoma showing transition to endometrioid glands with mild cytological atypia (arrow) (I); there is loss of ER expression in the atypical endometrioid glands with patchy expression in the associated clear cell carcinoma (J). (K, L) Dedifferentiated endometrial carcinoma with loss of ARID1A expression in both the low-grade (K) and the high-grade (L) components.

Although only two EC precursors are well characterized, it is entirely possible that additional precursor lesions, especially precursors of some of the less studied molecular subtypes/histotypes, such as POLEmut, MMRd, clear cell carcinoma, or mesonephric-like carcinoma, may be described with further study. Just as a dualistic model of EC was an oversimplification, so too the description of only two precursor lesions may prove to be insufficient. We also devote little attention in this review to the phenomenon of tumor progression, whereby loss of MMR expression or abnormal (mutant pattern) p53 expression occurs after tumor initiation; this phenomenon is seen in practice as subclonal p53abn or MMRd expression, with part of the tumor showing wild-type p53 expression or intact MMR expression, adjacent to a contiguous area of p53abn or MMRd, respectively. Subclonal MMRd staining is predominantly observed in the context of MLH1 hypermethylation [69], and subclonal p53 staining is frequently encountered in ECs known to carry high mutational burdens (MMRd, POLEmut), where TP53 mutation is a secondary event and not associated with an adverse prognosis (so-called ‘multiple classifier’ ECs) [70].

Morphological features of the endometrial carcinoma precursors associated with the four molecular subtypes

Very little data on the TCGA-based molecular subtype classification of EC precursor lesions is available. In recent work comparing resolving and progressing EAH/EIN and comparing EAH/EIN and subsequent carcinoma, Russo et al identified NSMP, MMRd, and p53abn EAH/EIN lesions [61, 71]. Evidence of clonal evolution has been demonstrated in whole-exome sequencing studies of paired EAH/EIN and concurrent EC (mostly of NSMP and MMRd molecular subtypes [62].
  1. p53abn: This is the prototypical type II EC and accounts for a large majority of serous ECs. The precursor lesion can be SEIC, but some p53abn ECs progress from low-grade EC and there may be co-existing EAH/EIN, especially in younger patients [59]. In these latter tumors, the acquisition of TP53 mutations can be an early event, i.e. present in the low-grade areas, or it can occur later and only be seen in the high-grade serous component (Figure 2D,E).
  2. MMRd: These carcinomas are mostly of endometrioid histotype and there can be co-existing EAH/EIN as the precursor lesion [60]. There is identical loss of MMR protein(s) in the EAH/EIN as in the associated carcinoma in most but not all cases (Figure 2F). In one study of more than 100 EAH/EINs, fewer than 5% were MMRd; thus, MMRd is much more commonly identified in EC than in the precursor lesion, suggesting that there is rapid transit through the in situ phase in these genomically unstable tumors [60]. This is similar to the situation in vulvar squamous cell carcinoma, where purely in situ lesions are predominantly HPV-associated while invasive vulvar squamous cell carcinoma is predominantly HPV-independent, an observation attributable, at least in part, to the more rapid progression of HPV-independent VIN [72].
  3. NSMP: These are prototypical type I ECs, arising from EAH/EIN, which may persist for years before progressing to invasive carcinoma, and may be treated with hormonal therapy, i.e. progestins [73]. Progression from AEH/EIN to EC is associated with the acquisition of driver mutations [62].
  4. POLEmut: The histogenesis of this molecular subtype is the least understood. EAH/EIN has been described in association with POLEmut EC (6/43 cases in one study) [74] and there is a single study demonstrating the presence of POLE mutations in EAH/EIN adjacent to EC [75], but there are no studies on the presence of POLE mutations in isolated EAH/EIN. The low frequency of associated precursor lesions identified with POLEmut EC suggests that this molecular subtype may be associated with a rapid transit through the phase of EAH/EIN, due to the high mutation rate.

Rare endometrial carcinoma histotypes: molecular subtypes and precursor lesions

Most of the work on molecular characterization of EC and its precursors has focused on the most common histotypes, i.e. endometrioid and serous carcinomas. In this section we discuss rare EC histotypes, including mesonephric-like and dedifferentiated, which are now included in the fifth edition of the WHO Classification of the Female Genital Tract [57], focusing on their molecular abnormalities and (purported) precursor lesions.

Clear cell carcinoma

Clear cell EC is an uncommon histotype, comprising less than 5% of ECs, and patients are usually older, postmenopausal women [33, 76]. Clear cell carcinomas can be of any of the four molecular subtypes. POLEmut clear cell carcinomas have the most favorable prognosis, while p53abn clear cell carcinomas are associated with aggressive behavior, i.e. the behavior of clear cell carcinoma is similar to that of non-clear cell EC of these molecular subtypes [26, 77, 78]. NSMP clear cell ECs, however, differ from NSMP endometrioid carcinomas, as they are associated with a worse prognosis [26]. Tumors that show mixed clear cell and endometrioid components are MMRd in most cases,[35] whereas MMRd pure clear cell carcinomas are uncommon [26, 32].

To date, data on precursor lesions of clear cell carcinoma remain scarce. Fadare et al identified a putative precursor lesion, described as isolated glands or surface epithelium displaying cytoplasmic clarity and/or eosinophilia with varying degrees of nuclear atypia, in approximately half of pure clear cell carcinomas [79]. In later work, an intraepithelial carcinoma-like growth pattern was identified in 20% of clear cell ECs adjacent to the invasive tumor [76] (Figure 2G,H). Clear cell carcinomas have been described to be associated with an endometrial polyp [76, 80]. As well as an intraepithelial carcinoma-like precursor, we have seen NSMP clear cell EC associated with atypical glands with low-grade cytological features. Although the morphological appearance is consistent with EAH/EIN, the cells are ER-negative, and this may also be a precursor of clear cell EC (Figure 2I,J).

Undifferentiated and dedifferentiated endometrial carcinoma

Undifferentiated carcinoma is a highly aggressive EC subtype that lacks all patterns of differentiation and consists of monotonous cancer cells growing in a sheet-like pattern. Most express epithelial antigens (e.g. cytokeratin) but often focally/weakly. Dedifferentiated carcinomas are composed of FIGO grade 1 or 2 endometrioid EC adjacent to areas of undifferentiated carcinoma. This category of tumors is the least well understood of the major EC histotypes, as it was only recently described [81-83].

The most common molecular subtype of undifferentiated and dedifferentiated carcinomas is MMRd, accounting for 37% and 43%, respectively [22, 27-30]. POLEmut undifferentiated carcinomas are rare, whereas nearly one-third of the dedifferentiated carcinomas harbor a POLE mutation. Most of the POLEmut dedifferentiated carcinomas came from one study, which could also reflect the challenges of histomorphological classification of POLEmut tumors [16]. p53abn has been reported to account for 7% and 28% of dedifferentiated and undifferentiated carcinomas, respectively. This difference is concerning and it is possible that some p53abn ‘undifferentiated carcinomas’ are better classified as solid pattern serous or high-grade endometrioid carcinoma of p53abn molecular subtype. Undifferentiated and dedifferentiated carcinomas often have mutations in genes encoding proteins of the SWI/SNF complex [29, 84]. There is not a molecular marker or panel that is considered diagnostic of dedifferentiated/undifferentiated EC, which hampers research on this uncommon tumor type, as the reported case series may not be comparable.

The precursor of undifferentiated and dedifferentiated carcinomas is thought to be low-grade carcinoma that, in the case of undifferentiated carcinoma, has been overgrown by the undifferentiated component (Figure 2K,L).


Carcinosarcoma is an uncommon, aggressive, biphasic neoplasm that accounts for less than 5% of ECs. Carcinosarcomas contain both epithelial and mesenchymal cell types that share mutation profiles, indicating a carcinoma origin with metaplastic conversion due to EMT [85]. While a large majority of carcinosarcomas are of the p53abn molecular subtype, they can also be of the other three molecular subtypes of EC, including MMRd [85-87]. In TCGA, over 90% of carcinosarcomas harbor a TP53 mutation [85]. Some carcinosarcomas share mutational profiles with the endometrioid lineage, e.g. PTEN mutation, indicating that, like serous carcinoma, an alternative route of carcinogenesis is via a low-grade endometrioid carcinoma and its precursors [85, 88].

Mesonephric-like endometrial carcinoma

Mesonephric-like carcinoma is a rare and aggressive EC histotype [89-91]. Mesonephric-like carcinomas characteristically show a variety of histologic patterns, including tubular, ductal, papillary, solid, spindled, retiform, glomeruloid, and sex cord-like [90, 92].

Although few studies have examined the molecular profile of these rare neoplasms, all available data suggest that they are of NSMP molecular subtype. To our knowledge, all cases of mesonephric-like carcinoma of the endometrium reported in the literature are TP53 wild-type (by molecular or immunohistochemical analysis) [89, 90, 92-94], MMR-proficient (by immunohistochemical analysis) [95], and POLE wild-type [89]. Like NSMP clear cell carcinoma, mesonephric-like ECs (all of which are NSMP) have a worse prognosis than NSMP endometrioid carcinoma [89, 96, 97].

Unlike mesonephric carcinoma of the uterine cervix, mesonephric neoplasms of the uterine corpus do not appear to be associated with mesonephric remnants/hyperplasia [90, 92, 98]. Furthermore, they appear to arise in the endometrium rather than the myometrium [92], and have considerable molecular overlap with endometrioid adenocarcinoma (including PIK3CA and ARID1A mutations) [94, 99]. Mixed carcinoma, consisting of both mesonephric-like and endometrioid components, has been reported, suggesting the possibility that mesonephric-like carcinomas may arise from Müllerian origins, with subsequent mesonephric transdifferentiation [85].

Knowledge gaps and opportunities

Progression to carcinoma can be prevented by early detection of precursor lesions of EC. The relationships between the two EC precursor lesions, EAH/EIN and SEIC, and the molecular subtypes and histotypes of EC are depicted in Figure 3. How can diagnosis of EC precursor lesions be optimized to effect earlier diagnosis and improve outcomes? The current approach of endometrial biopsy based on symptoms (irregular perimenopausal bleeding or postmenopausal bleeding) works relatively well only in the detection of EAH/EIN as a precursor of type I/low-grade endometrioid/NSMP EC. Patients with EAH/EIN have a uniformly favorable prognosis and young patients may be treated conservatively with hormonal therapy and preservation of fertility. Even when there has been progression to carcinoma, most low-grade endometrioid/NSMP ECs are low stage and cured by surgery alone. As noted previously, EAH/EIN associated with MMR loss is rarely diagnosed except when adjacent to carcinoma, and the same appears to be true for EAH/EIN associated with POLEmut EC, suggesting that EAH/EINs associated with these molecular subtypes, with their resulting high mutation rate, progress rapidly. In addition, there are no data on whether POLEmut or MMRd EAH/EINs are as responsive to the more common NSMP EAH/EIN treatments, and thus these patients might not be candidates for conservative treatments. Similarly, early diagnosis of SEIC is rare; isolated SEIC, without associated invasive carcinoma, is reported in only a few small case series, suggesting that SEIC is not consistently associated with postmenopausal bleeding that would trigger a biopsy, and thus it may routinely go undetected until invasive carcinoma develops. The precursor lesions of clear cell carcinoma, carcinosarcoma, dedifferentiated/undifferentiated carcinoma, and mesonephric-like carcinoma are not well characterized, but given their aggressive behavior and the paucity of descriptions of precursor lesions adjacent to these carcinomas, the study of precursors will be challenging and the prospect for early detection, at a pre-invasive point in their natural history, is unlikely.

Details are in the caption following the image
Proposed histogenesis of endometrial carcinoma from benign endometrium to carcinoma. A proposed model for endometrial cancer histogenesis is shown, with likely progression between lesions depicted (solid line). The dashed line depicts possible progression from NSMP to p53abn by an acquired TP53 mutation. The tumor histology and molecular subtypes (POLEmut, MMRd, NSMP, p53abn) are indicated within each box. EAH/EIN, endometrial atypical hyperplasia/endometrioid intraepithelial neoplasia; mut, mutation; MMRd, mismatch repair deficient; NSMP, no specific molecular profile; abn, abnormal; EC, endometrial carcinoma; EEC, endometrioid endometrial carcinoma; CCC, clear cell carcinoma; SIEC, serous intraepithelial endometrial carcinoma; SEC, serous endometrial carcinoma; DDEC, dedifferentiated endometrial carcinoma; UDEC, undifferentiated endometrial carcinoma; CS, carcinosarcoma.

In a review of a large series of endometrial biopsies reported as negative, from patients who subsequently were diagnosed with EC, we found that there were no cases of SEIC that had been missed on these earlier biopsies, and only a handful of cases of EAH/EIN, indicating that the lack of sensitivity of endometrial biopsy (only 90% sensitivity) [55] is attributable to sampling rather than misinterpretation errors. Of the carcinomas with negative earlier biopsy in this series, 7.7% were serous carcinomas and the remaining endometrioid EC; thus, it appears that focusing on improved pathologist diagnosis of EAH/EIN or SEIC for early diagnosis is unlikely to impact on mortality due to EC; as noted previously, EC mortality is mostly attributable to p53abn and MMRd EC, and uncommon histotypes that tend to only become symptomatic when there has been development of more advanced disease.


Only two EC precursor lesions have been well characterized: EAH/EIN and SEIC. These correspond to the classic precursor lesions of type I and type II EC, respectively, but can also give rise to a wide range of EC histotypes and molecular subtypes. Only a small number of studies have examined the molecular subtype characteristics of EAH/EIN and SEIC, especially when diagnosed as an early lesion, without associated carcinoma. The precursor lesions associated with rare EC histotypes are not well characterized and warrant further study.

The apparently rapid transit through the precursor lesion to invasive (and often advanced stage) carcinoma means that diagnosis of MMRd or p53abn EC at the precursor stage is rare, and our current approach of endometrial biopsy based on symptoms is unlikely to impact on the outcomes of patients with these EC molecular subtypes. Similarly, uncommon EC histotypes such as mesonephric-like, clear cell, dedifferentiated/undifferentiated, and carcinosarcoma are not consistently detected as low-stage disease based on the current approach. As these less common molecular subtypes and histotypes collectively account for most EC mortality, alternative approaches to prevention or early detection will be needed to significantly reduce EC mortality. One possible approach is to look for key EC driver gene mutations in cervical cytology or intrauterine brush samples [100]. Such an approach, while promising, will have to take into account the presence of such mutations in normal endometrium [101], and the low disease prevalence, which will make screening difficult.

Ethical approval

No specific ethical approval was received for this review, but it was conducted in accord with the ethical guidelines for research of the University of British Columbia.

Author contributions statement

JH, ET and JP generated the first draft of this manuscript and provided all photomicrographs. JH generated the figures that are not photomicrographs and the tables. All the authors participated in revising the manuscript into its final state.