Review

Classification of endometrial carcinoma: more than two types Rajmohan Murali, Robert A Soslow, Britta Weigelt

Endometrial cancer is the most common gynaecological malignancy in Europe and North America. Traditional classification of endometrial carcinoma is based either on clinical and endocrine features (eg, types I and II) or on histopathological characteristics (eg, endometrioid, serous, or clear-cell adenocarcinoma). Subtypes defined by the different classification systems correlate to some extent, but there is substantial heterogeneity in biological, pathological, and molecular features within tumour types from both classification systems. In this Review we provide an overview of traditional and newer genomic classifications of endometrial cancer. We discuss how a classification system that incorporates genomic and histopathological features to define biologically and clinically relevant subsets of the disease would be useful. Such integrated classification might facilitate development of treatments tailored to specific disease subgroups and could potentially enable delivery of precision medicine to patients with endometrial cancer.

Lancet Oncol 2014; 15: e268–78

Introduction

Correspondence to: Dr Rajmohan Murali, Department of Pathology, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, New York, NY 10065, USA [email protected]

Endometrial cancer is the most common gynaecological malignant disease, and the fourth most common cancer in European and North American women, accounting for about 6% of new cancer cases and 3% of cancer deaths per year.1,2 Incidence is steadily increasing;3 age-adjusted annual incidence was 24·3 per 100 000 women in the USA in 2006–10, and 19·4 per 100 000 in the UK in 2008.1,3 Around 75% of patients with endometrial cancer are diagnosed in the early stages (International Federation of Gynecology and Obstetrics [FIGO] stages I or II), and 5-year overall survival is 74–91%.1,4 For women with advanced stage III or IV disease, 5-year overall survival of 57–66% and 20–26%, respectively, has been reported.4 Traditionally, endometrial carcinomas have been classified as type I or type II, as defined by Bokhman,5 on the basis of clinical, endocrine, and epidemiological observations. Type I tumours were oestrogen dependent, and associated with endometrial hyperplasia, whereas type II tumours were oestrogen independent and associated with endometrial atrophy.5 Endometrial carcinoma is also classified according to histopathological characteristics,6 with the most common subtypes being endometrioid carcinoma, serous carcinoma, carcinosarcoma, and clear-cell carcinoma. Correlations have been noted between the subtypes in these two classification systems—type I cancers generally have endometrioid histology and most type II cancers are serous carcinomas—but these correlations are imperfect. In the past decade it has become increasingly clear that endometrial cancer comprises a biologically, clinically, morphologically, and genetically heterogeneous group of tumours. Traditional classifications do not entirely take into account this heterogeneity and, being prognostic in nature, are limited in predicting response to therapy. A genomic classification of endometrial carcinoma has been proposed7 in an attempt to identify potential targets for treatment in different subgroups of the disease. In this Review, we provide an overview of traditional and genomic classifications of endometrial carcinoma, www.thelancet.com/oncology Vol 15 June 2014

and discuss their potential and their limitations. In view of the substantial morphological and molecular heterogeneity in endometrial cancer, we suggest that classification systems based on limited sets of parameters are insufficient for the development of effective individualised treatments. We propose a rationale for establishing an integrated classification system that incorporates molecular and histopathological features to define biologically and clinically relevant subsets of endometrial cancer.

Department of Pathology (R Murali MD, Prof R A Soslow MD, B Weigelt PhD) and Center for Molecular Oncology (R Murali), Memorial Sloan Kettering Cancer Center, New York, NY, USA; and Department of Pathology and Laboratory Medicine, Weill Cornell Medical College, New York, NY, USA (Prof R A Soslow)

Traditional classification Dualistic and histological classification Bokhman5 proposed that endometrial cancers can be categorised into two pathogenetic types that are primarily based on clinical, metabolic, and endocrine characteristics (table 1).7–17 Type I tumours were associated with oestrogen excess, obesity, hormone-receptor positivity, and endometrial hyperplasia, were moderately or highly differentiated, and had favourable outcomes; type II tumours were more common in non-obese women, arose in the absence of endocrine and metabolic disturbances, were associated with an atrophic endometrium, were poorly differentiated, and had less favourable outcomes.5 Subsequent studies aimed to elucidate the clinicopathological, histological, and molecular correlates of type I and type II cancers. However, although these studies advanced understanding of endometrial cancers, some misconceptions emerged, as discussed in this Review. Tumours of the uterine corpus comprise several distinct histological types that WHO classifies as epithelial carcinomas (endometrioid, serous, clear cell, mucinous, squamous cell, transitional cell, small cell, and undifferentiated), mixed epithelial and mesenchymal tumours (eg, carcinosarcomas), or mesenchymal tumours (eg, endometrial stromal and smooth-muscle tumours), gestational trophoblastic diseases, and other malignant tumours.6 In this Review we focus on epithelial tumours—of which endometrioid, serous, and e268

Review

Type I

Type II

Clinical, endocrinological, and morphological components (Bokhman classification5) Distribution

60–70%

30–40%

Reproductive function

Decreased

No disturbances

Onset of menopause

After age 50 years

Younger than age 50 years

Background endometrium

Hyperplasia

Atrophy No

example, endometrioid (type I) carcinomas are preferentially associated with mutations in PTEN, KRAS, CTNNB1, and PIK3CA, and microsatellite instability, whereas serous (non-endometrioid, type II) carcinomas show HER2 amplification and recurrent TP53 mutations (table 1).7,9,10 Endometrioid and serous carcinomas are also generally distinct at the transcriptional level and in gene copy numbers.23,24 These findings were perceived as being consistent with the type I and type II division of endometrial cancers; subsequently, histological type and molecular features became integral components of the dualistic Bokhman classification.

Oestrogen associated

Yes

Associated obesity, hyperlipidaemia, and diabetes mellitus

Yes

No

Tumour grade

Low (grades 1–2)

High (grade 3)

Myometrial invasion

Superficial

Deep

Potential for lymphogenic metastatic spread

Low

High

Prognosis

Favourable

Unfavourable

Sensitivity to progestagens

High

Low

Limitations of traditional classification schemes

Outcome (5-year survival)

86%

59%

Endometrioid

Serous

The Bokhman and histological classification systems are undoubtedly conceptually useful. Their implementation has advanced understanding of endometrial cancers, and provided a framework for studies, particularly those of molecular features. Yet, the correlations between the subtypes defined by the traditional taxonomies are imperfect (table 2).5,6,11 Furthermore, there has been widespread misconception that the Bokhman types define diseases that are homogeneous with respect to their biological, genetic, and pathological features. Several lines of evidence, however, have shown that there is not only overlap between type I and type II tumours, but that there is also heterogeneity within each of these types. Bokhman’s dualistic model was based on clinical and epidemiological characteristics of women with endometrial cancer in the former Soviet Union more than 30 years ago.5 Characteristics in current patients might, therefore, differ—eg, because of increased use of hormone-replacement therapy and increased numbers of overweight or obese patients.25–27 Bokhman’s model also does not account for endometrioid cancers occurring in patients with Lynch syndrome, who generally are thin and whose tumours are not often associated with hyperplasia.28 Furthermore, epidemiological data suggest that obesity is also associated with type II cancers, although to a lesser extent than with type I cancers.25,29 Type I and II tumours also share multiple risk factors with a history of diabetes being associated with increasing parity, age at menarche, use of oral contraceptives, and pack-years of smoking being associated with reduced risk.25,29 While low-grade endometrioid and serous carcinomas integrate well into Bokhman’s model (being, respectively, prototypical type I and II tumours), many in the range of endometrioid cancers fall outside a simple dichotomous classification. Between 10% and 19% of endometrioid carcinomas are high grade4,29 and have clinical, histopathological, and molecular features that are either intermediate between those of types I and II or are more akin to those of type II cancers, including lack of association with endometrial hyperplasia and poor outcomes.29,30 By contrast, not all serous carcinomas behave as prototypical type II cancers. For example, 2%

Clinicopathological and molecular correlates7–10 Prototypical histological type

Oestrogen-receptor or progesterone-receptor expression High

Low

Stage at diagnosis

Early (FIGO stage I–II)

Advanced (FIGO stage III–IV)

PTEN mutation

52–78%

1–11%

PIK3CA mutation

36–52%

24–42%

PIK3R1 mutation

21–43%

0–12%

KRAS mutation

15–43%

2–8%

ARID1A mutation

25–48%

6–11%

CTNNB1 mutation

23–24%

0–3%

TP53 mutation

9–12%

60–91%

PPP2R1A mutation

5–7%

15–43%

HER2 amplification

0

27–44%

Microsatellite instability

28–40%

0–2%

Common genetic alterations10–17

FIGO=International Federation of Gynaecology and Obstetrics.

Table 1: Dualistic classification of epithelial endometrial cancer, including clinical, pathological, and common molecular genetic correlates

clear-cell carcinomas account for 75%, 5–10%, and 1–5%, respectively6—as they are the most extensively studied (figure 1).18–21 Endometrioid adenocarcinomas represent a range of neoplasms, from well to poorly differentiated tumours (ie, low to high grade), whereas serous and clear-cell carcinomas are high grade by definition. Lowgrade endometrioid carcinomas are often seen in premenopausal women, are associated with endometrial hyperplasia, and generally exhibit indolent clinical behaviour. By contrast, serous carcinomas frequently develop in postmenopausal women in association with atrophic endometrium, and generally show an aggressive clinical course (figure 1).4,6,9,22 Bokhman’s model5 formed the basis of the tenet that type I tumours comprise low-grade endometrioid carcinomas associated with unopposed oestrogen exposure and excellent prognosis, and that type II tumours are largely non-endometrioid tumours (ie, serous and clear-cell carcinomas) with poor outcomes (tables 1 and 2).7,8 Molecular data to support this dichotomous classification were also available. For e269

www.thelancet.com/oncology Vol 15 June 2014

Review

of serous carcinomas arise in association with endometrial hyperplasia, and at least 20% lack deep myometrial invasion.31 Furthermore, some common high-grade endometrial cancers, such as carcinosarcomas, or recognised histological subtypes, such as clear-cell or undifferentiated carcinomas, were either not taken into account by Bokhman in his original classification or do not readily fit into the Bokhman model.30,32 At the molecular level, the Bokhman types are frequently perceived to define homogeneous entities that are characterised by unique sets of genetic alterations. Although specific mutations are more common in one type than in the other, there is substantial overlap in genetic alterations. For example, PIK3CA mutations, which are found in about 52% of endometrioid tumours, are also present in up to 42% of serous carcinomas.11,16 TP53 mutations, which are present in about 75% of serous cancers, are also found in up to 12% of endometrioid tumours.11,15 Furthermore, within each of the Bokhman and histological types, a great deal of genetic heterogeneity exists. For instance, only a subset of endometrioid (type I) cancers harbour PTEN, KRAS, CTNNB1, or PIK3CA mutations, and the same is true for TP53 or PPP2R1A mutations in serous (type II) cancers.7,15 Histologically, some tumours are difficult to classify accurately because of overlapping morphological features. In particular, distinguishing between highgrade endometrial cancers (including FIGO grade 3 endometrioid and serous carcinomas) can be challenging (figure 2), despite the use of ancillary immunohistochemical markers.20 Findings are subject to interobserver variability, even between expert gynaecological pathologists.33

Genetic alterations and molecular classification Many studies have assessed genetic alterations in endometrial cancer, almost exclusively in endometrioid and serous carcinomas. Early research focused mainly on single candidate genes or pathways,9 but nextgeneration sequencing studies have provided insights into genome-wide genetic alterations in these tumours.

Candidate gene studies Genes altered in other cancer types have been studied for mutations or copy-number aberrations in endometrial cancers.7,8,10 Endometrioid carcinoma is the most extensively studied type of endometrial cancer, probably because of its high prevalence and the availability of representative mouse models and cell lines.4,34–36 Endometrioid cancers generally have a high mutational load, particularly in the PI3K/AKT/mTOR signalling pathway, which regulates cell growth and survival, and the Wnt/β-catenin signalling pathway, which regulates gene transcription and development. The tumour suppressor gene PTEN, a negative regulator of the PI3K/AKT/mTOR pathway, is mutated and lost in up to 80% of endometrioid www.thelancet.com/oncology Vol 15 June 2014

A

B

C

D

Histological type

Endometrioid

Endometrioid

Serous

Clear cell

Histological grade

Low

High

High

High

Metastasis

Uncommon

Lymph nodes Distant organs

Lymph nodes Peritoneal Distant organs

Lymph nodes Peritoneal –/+

Prognosis

Favourable

Poor

Poor*

Poor*†

Molecular markers18–21 ER/PR expression PTEN expression DNA MMR loss Aberrant P53 Ki-67/MIB-1

+ –/+ –/+ – Low

+/– –/+ –/+ –/+ High

–/+ + – + High

– + –/+ –/+ Low or high

Figure 1: Clinicopathological and molecular characteristics of the common types of epithelial endometrial carcinoma (A) FIGO grade 1 endometrioid carcinoma composed of well formed glands. (B) FIGO grade 3 endometrioid carcinoma displaying solid growth pattern. (C) Serous carcinoma composed of atypical cells with pleomorphic nuclei. (D) Clear-cell carcinoma composed of cuboidal cells with clear cytoplasm. ER=oestrogen receptor. PR=progesterone receptor. MMR=mismatch repair; +=present/high. –=absent/low. –/+=occasional. +/−=frequent. FIGO=International Federation of Gynaecology and Obstetrics. *Frequently diagnosed at advanced stages. †When diagnosed at an early stage, prognosis is better than that of serous carcinoma of the same stage.

tumours.15,37–41 PTEN mutations have also been detected in about 55% of patients with endometrial hyperplasia,37,38 and a subset of heterozygous Pten mice develop endometrioid tumours.35,36 PTEN loss of function has, therefore, been proposed as an early event in the pathogenesis of endometrioid cancer. Furthermore, several other components of the PI3K/AKT/mTOR pathway are frequently targeted by mutations in endometrioid cancers—eg, PIK3CA, which encodes the catalytic subunit of PI3K, p110α, is mutated in about 52%,37,39–41 PIK3R1, which encodes the regulatory subunit of PI3K, p85α, is mutated in about 43%,16,41 and KRAS is mutated in about 43% of endometrial cancers.13,16,41 These mutations frequently coexist within a given endometrioid tumour in different combinations.16,39,41 PIK3CA mutations and amplifications are seen, respectively, in 35% and 46% of serous endometrial carcinomas, whereas PTEN, PIK3R1, and KRAS mutations are less frequently observed (about 11%, 12%, and 8%, respectively).13,16,41 In contrast to breast and colorectal carcinomas, in which most PIK3CA mutations occur in two hotspots in the helical and kinase domains,42 mutations in endometrial cancer are distributed throughout the gene.40,41 In up to 45% of serous cancers, the PI3K/AKT/mTOR pathway might also be activated by HER2 gene amplifications.7,43 Alterations in the Wnt/β-catenin signalling pathway, particularly loss of E-cadherin expression, are seen in up to 50% of endometrioid, and up to 80% of serous endometrial cancers.7,10,44 Nuclear accumulation of β-catenin, which is a surrogate marker of activated canonical Wnt/β-catenin signalling, however, is almost entirely restricted to endometrioid cancers (up to 47% of e270

Review

Basis

Bokhman5

WHO6*

The Cancer Genome Atlas7

Clinical and epidemiological features

Histological features

Genome-wide genomic characterisation

Categories Type I Type II

Endometrioid POLE (ultramutated), MSI (hypermutated), copy-number low (endometrioid), copy-number high (serous-like) Serous Copy-number high (serous-like) Clear cell NA

MSI=microsatellite instability. *WHO mucinous, squamous-cell, transitional-cell, small-cell, and undifferentiated carcinoma subtypes were not considered in Bokhman’s classification.

Table 2: Comparison of classification systems of endometrial cancer

A

B

C

D

E

F

Figure 2: Histologically ambiguous high-grade endometrial carcinomas (A) Suggestive of both endometrioid and serous carcinoma. (B) Suggestive of serous and clear-cell carcinoma. (C) P53 immunohistochemical analysis (clone Do7) may be used to assign an ambiguous endometrioid-serous tumour to the serous category, as shown in (A). (D) Serous and (E) endometrioid carcinoma components from one tumour, which showed loss of PTEN expression (clone 6H2·1) in both components (F).

endometrioid vs less than 3% of serous carcinomas).7,8,10,45 CTNNB1 (β-catenin) gain-of-function mutations are present in around 25% of endometrioid cancers, but are seen very rarely in serous carcinomas (table 1).8,15 Microsatellite instability, the alteration of repetitive nucleotide sequence lengths mainly due to MLH1 promoter hypermethylation,46 is a recurrent event in up to a third of sporadic endometrioid carcinomas,13,47 and may explain the high mutational load in these cancers. By contrast, in serous carcinomas, microsatellite instability is rare (table 1). Mutations in the tumour suppressor gene TP53 are present in up to 90% of serous carcinomas and in up to 10% of low-grade and 30% of high-grade endometrioid cancers.13,15,16,41,48,49 Because endometrial serous intraepithelial carcinoma (the putative precursor of invasive serous carcinoma) harbours TP53 mutations in 75% of cases, this genetic alteration is thought to be an early event in endometrial serous carcinogenesis.48,49 Mutations in ARID1A (a component of the switch/ sucrose non-fermentable chromatin-remodelling complex), e271

or loss of expression of its protein product BAF250a, were identified in 29% of grade 1 and 2 and in 39% of grade 3 primary endometrioid endometrial cancers, in 26% of clear-cell carcinomas, and 18% of serous carcinomas.50 Loss of BAF250a expression has also been documented in 16% of complex and atypical endometrial hyperplasias, and in 28% of metastatic endometrioid carcinomas. Thus, ARID1A mutations and loss of BAF250a expression might contribute to the progression of endometrioid carcinomas.51,52 Mutations in PPP2R1A encoding the α-isoform of the scaffolding subunit of the PP2A holoenzyme, a putative tumour suppressor complex, were identified in 40% and 5% of serous and endometrioid carcinomas, respectively.53 Whether PPP2R1A mutations have a role in endometrial cancer tumorigenesis and tumour progression is yet to be determined.54 Data on the genetic aberrations characteristic of other non-endometrioid endometrial cancers is scarce. Carcinosarcomas harbour recurrent TP53 mutations (44–64%) and mutations in PTEN (11–33%), PIK3CA www.thelancet.com/oncology Vol 15 June 2014

Review

(22–29%), PIK3R1 (6%), ARID1A (24%), KRAS (17%), PPP2R1A (21%), and CTNNB1 (up to 5%).15,41 McConechy and colleagues,15 however, noted two groups of carcinosarcomas: an endometrioid type harbouring mutations in PTEN and ARID1A, and a group with TP53 and PPP2R1A mutations, similar to serous cancers. The molecular characterisation of uterine carcinosarcomas and other non-endometrioid subtypes is so far insufficient to draw definitive conclusions. The mutational landscape of endometrial cancers is complex and heterogeneous. Although the prevalence of individual somatic genetic aberrations varies between endometrioid and serous carcinomas, no mutations in any of the genes studied so far have been exclusively present in either tumour type.

Genome-wide studies using next-generation sequencing The development and refinement over the past decade of technologies for next-generation sequencing and bioinformatic methods for analysis of large-scale genomic datasets has enabled unbiased characterisation of genome-wide genetic alterations. A study that used whole-exome sequencing (ie, the protein-coding genomic regions) for the analysis of 11 endometrioid and two mixed endometrioid and serous cancers revealed mutations in ten tumour-suppressor candidates, including ARID1A, IGFBP3, and WNT11, and two oncogene candidates (HER3 and RPS6KC1).55 The biological effects and clinical relevance of these mutations remain to be elucidated. Targeted sequencing studies have led to the identification of novel classes of genes mutated in human cancers, including chromatin remodelling genes, such as KDM6A (formerly known as UTX) in transitional-cell carcinoma of the bladder,56 and genes that have roles in the RNA processing machinery, such as DICER1 mutations in non-epithelial ovarian tumours.57 Targeted exome sequencing of endometrial serous carcinomas by three independent groups revealed frequent mutations in chromatin-remodelling genes, such as CHD4 in 17–19% of tumours, mutations or deletions in ubiquitin ligase complex genes, such as FBXW7, in 20–29%, and amplification of CCNE1, a target of FBXW7-mediated ubiquitination, in 44%.12,14,17 Mutations in TAF1 (13%), a component of the core RNA polymerase II machinery, have also been identified,17 as have recurrent TP53 and PIK3CA mutations in 60–81% and 23–31% of serous carcinomas, respectively.12,14,17 TP53, PIK3CA, and PPP2R1A mutations were also suggested to be early events in serous tumorigenesis, as they were present in invasive carcinoma and associated serous intraepithelial carcinoma.12 Genome-wide studies have validated genetic alterations shown in previous candidate-gene investigations and identified several novel mutations, such as in CHD4, FBXW7, and SPOP,14 in a subset of serous carcinomas. www.thelancet.com/oncology Vol 15 June 2014

Resequencing of these genes showed that the mutations are present in 15% of endometrioid cancers, which provides further evidence that type I and type II cancers are heterogeneous with respect to their mutational repertoires. Taken together, the focused and genomewide studies suggest some degree of overlap between the molecular characteristics of high-grade endometrioid carcinomas and serous carcinomas, with the greatest differences between the two types being the higher prevalence of PTEN mutations in the former and the higher prevalence of TP53 mutations in the latter.

Genomic classification of endometrial carcinoma The Cancer Genome Atlas Research Network (TCGA) has reported a comprehensive genomic and transcriptomic analysis of endometrial cancers.11 A large set of endometrial cancers was assessed with next-generation sequencing technologies, in combination with analysis of DNA methylation, reverse-phase protein array, and microsatellite instability. The study focused on common histological types, namely endometrioid (n=307), serous (n=53), and mixed endometrioid and serous (n=13) carcinomas. On the basis of integration of mutation spectra, copy-number aberrations, and microsatellite instability status, endometrial cancers were categorised into four genomic classes: 1) POLE (ultramutated) tumours characterised by very high mutation rates and hotspot mutations in the exonuclease domain of POLE58 (a subunit of DNA polymerase ε that has a role in DNA replication), few copynumber aberrations, increased frequency of C→A transversions, mutations in PTEN, PIK3R1, PIK3CA, FBXW7, and KRAS, and favourable outcome; 2) a microsatellite-instable (MSI [hypermutated]) group of endometrioid tumours characterised by microsatellite instability due to MLH1 promoter methylation, high mutation rates, few copy-number aberrations, recurrent RPL22 frameshift deletions, and KRAS and PTEN mutations; 3) copy-number low (endometrioid) tumours, comprising microsatellite-stable grade 1 and 2 endometrioid cancers with low mutation rates, characterised by frequent CTNNB1 mutations; and 4) copy-number high (serous-like) tumours, characterised by extensive copynumber aberrations and low mutation rates, recurrent TP53, FBXW7, and PPP2R1A mutations, infrequent PTEN and KRAS mutations, and poor outcome (table 3).11 The last genomic class included all but one serous and one mixed epithelial carcinoma and a quarter of grade 3 endometrioid carcinomas. Independent review by two experienced gynaecological pathologists of digitised images of 82 tumours diagnosed as FIGO grade 3 endometrioid carcinoma in the TCGA study led to reclassification of 20–25% of the tumours as serous carcinoma (interobserver agreement 88%).11,20 The TCGA study further revealed several notable genomic patterns: a subset of tumours diagnosed as high-grade endometrioid carcinomas harboured copynumber and mutational profiles similar to those of e272

Review

POLE (ultramutated)

MSI (hypermutated)

Copy-number aberrations

Low

Low

Copy-number low (endometrioid) Copy-number high (serous-like) Low

MSI/MLH1 methylation

Mixed MSI high, low, stable

MSI high

MSI stable

MSI stable

Mutation rate

Very high (232 × 10⁶ mutations/Mb)

High (18 × 10⁶ mutations/Mb)

Low (2·9 × 10⁶ mutations/Mb)

Low (2·3 × 10⁶ mutations/Mb)

Genes commonly mutated (prevalence)

POLE (100%) PTEN (94%) PIK3CA (71%) PIK3R1 (65%) FBXW7 (82%) ARID1A (76%) KRAS (53%) ARID5B (47%)

PTEN (88%) RPL22 (37%) KRAS (35%) PIK3CA (54%) PIK3R1 (40%) ARID1A (37%)

PTEN (77%) CTNNB1 (52%) PIK3CA (53%) PIK3R1 (33%) ARID1A (42%)

TP53 (92%) PPP2R1A (22%) PIK3CA (47%)

Histological type

Endometrioid

Endometrioid

Endometrioid

Serous, endometrioid, and mixed serous and endometrioid

High

Tumour grade

Mixed (grades 1–3)

Mixed (grades 1–3)

Grades 1 and 2

Grade 3

Progression-free survival

Good

Intermediate

Intermediate

Poor

The four genomic classes were identified by The Cancer Genome Atlas Network11 by combining information on mutations, copy-number aberrations, and microsatellite instability. General characteristics of the genomic classes are shown. Mb=megabase. MSI=microsatellite instability.

Table 3: Characteristics of four genomic classes of endometrioid and serous carcinomas

serous carcinomas; apart from POLE hotspot mutations, no mutations were unique to any of the four genomic classes; more than 92% of MSI (hypermutated) and copynumber low (endometrioid) cancers and 60% of copynumber high (serous-like) endometrial cancers showed aberrations in the PI3K pathway, with mutations in PIK3CA and PIK3R1 being mutually exclusive in most cases; and the receptor tyrosine kinase/RAS/β-catenin pathway was altered in copy-number low, hypermutated, and copy-number high tumours (82%, 71%, and 50%, respectively), with mutations in CTNNB1, KRAS, and in SOX17, FBXW7, FGFR2, and HER2 being mutually exclusive.11 Finally, the TCGA genomic characterisation of endometrial cancer might permit reclassification of endometrioid and serous carcinomas, which could directly affect treatment decisions and guide clinical trials of targeted therapies. The study, however, has important limitations. First, as it focused solely on endometrioid, serous, and mixed carcinomas of the uterus,11 the genomic diversity of other non-endometrioid carcinomas, including carcinosarcomas and clear-cell carcinomas, remains to be established. Whether additional molecular subtypes will emerge from these analyses is yet to be defined. Second, despite the high standards adopted by the pathology group of TCGA, the histological typing of high-grade endometrial cancers is challenging even for experienced gynaecological pathologists.20,33 Third, highgrade endometrioid carcinomas classified as being of copy-number high (serous-like) subtype might have included mixed tumours in which only the endometrioid component was sampled.

Towards an integrated classification system In view of the substantial genetic and morphological heterogeneity in endometrial carcinomas, the current approach of histopathology-based classification requires e273

refinement. We propose that an integrated classification system incorporating molecular and clinicopathological features would define biologically and clinically relevant subsets of endometrial cancer. Most cancer taxonomies are prognostic, although some, such as lung cancer classifications, provide predictive information.59 In endometrial carcinoma, clinical and surgical features, including FIGO stage, and pathological features, such as histological type and grade, are well established prognostic factors.4,9,22,60 The integration of molecular markers with established clinicopathological parameters has the potential to improve prognostic accuracy and to provide predictive information. Because many women with endometrioid cancers are likely to be cured by surgical treatment alone, overtreatment of these patients should be avoided. The identification of reliable prognostic and predictive markers is, therefore, of great clinical importance. We posit that standardised evaluation of molecular biomarkers, such as P53 (figure 2) and POLE, might refine prognostic assessment of endometrioid cancers. For instance, the TCGA study showed that most P53dysfunctional endometrioid cancers have copy-number and mutational profiles similar to those found in serous carcinomas and, therefore, might benefit from adjuvant treatment approaches used for the latter because of their aggressive clinical behaviour.11 By contrast, assessment of POLE mutations in endometrioid cancers by sequencing or surrogate immunohistochemical analysis of tumours might identify a subset of patients likely to have favourable outcomes,11 who could be spared chemotherapy or radiation irrespective of histological tumour grade. These associations between genotypes and clinical behaviour require further investigation and independent validation, as the TCGA study was retrospective in nature and included heterogeneously treated patients. www.thelancet.com/oncology Vol 15 June 2014

Review

The TCGA data, however, do have immediate implications for pathology practice. The integration of genomic markers may aid histological categorisation of tumours that are intrinsically challenging, particularly high-grade neoplasms (figure 2).19,20 The assessment of genes or gene products found to be differentially altered between subtypes may provide important ancillary information for accurate diagnosis. When faced with a differential diagnosis of endometrioid versus serous carcinoma, sequencing (or immunohistochemical) analysis for TP53 (P53), PTEN (PTEN), PPP2R1A (PPP2R1A), and ARID1A (BAF250a) or DNA mismatch repair genes (proteins) might provide important ancillary information for accurate diagnosis. For instance, tumours that have mutated (altered) TP53 (P53) or PPP2R1A (PPP2R1A) but have wild-type (retained) PTEN (PTEN), ARID1A (BAF250a), and DNA mismatch repair genes (proteins) are likely to be diagnosed as serous carcinoma, whereas the converse pattern would favour endometrioid carcinoma. Immunohistochemical analysis of PTEN, however, has proven technically challenging in terms of preanalytical variables, antigen retrieval, antibody, and clones, and interpretation of the immunohistochemical expression patterns.18 The genomic classification of endometrioid and serous carcinomas highlights the importance of stratification of patients according to genomic alterations. Such biomarker-driven trials will help to identify subsets of patients with poor prognosis who are most likely to benefit from specific drugs and to ascertain genetic determinants of response. Predictive markers with proven clinical utility61 defined in such trials should then be integrated into the diagnostic workup for endometrial carcinomas.

Future challenges and opportunities Next-generation sequencing studies have confirmed and expanded knowledge of recurrently altered signalling pathways in endometrial cancer, and have laid the foundations for rational design of genomics-based clinical trials. In view of the high mutation rates, patterns of epistatic interactions between known cancer genes in a subset of endometrioid carcinomas might differ from those in other epithelial malignant disease. Additionally, whether different mutations (eg, in PTEN, PIK3CA, PIK3R1, or KRAS) and combinations of mutations that affect the same pathway have different effects on its activation and response to targeted agents is unclear. The wealth of data generated provides opportunities for refinement of disease classification, prognostic assessment, and prediction of responses to specific therapies and disease monitoring.

a heterogeneous group—tumours with MSI (hypermutated), POLE (ultramutated), and copy-number low phenotypes. Type II (high-grade) cancers, contrary to the prevailing view, might include serous carcinomas and a subset of high-grade endometrial carcinomas. The development of a comprehensive classification system for endometrial carcinomas will require detailed study of rare histological types, including clear-cell, squamous-cell, and small-cell carcinomas, and carcinosarcomas (believed to be metaplastic carcinomas62). The latter are being characterised for their genomic aberrations by the TCGA.11 Genomic analyses of rare endometrial cancers might result in the identification of novel driver genes or novel combinations of somatic genetic aberrations that define specific subtypes. They might also permit reclassification of cancers defined as undifferentiated carcinomas and their underlying mechanisms. Risk factor profiles for endometrioid and nonendometrioid (type II) cancers are broadly similar.25,29 When stratified by grade, however, the relationship to risk factors, especially obesity, differs between type II and high-grade endometrioid cancers and low-grade endometrioid cancers.25,29 Whether tumours of different genomic subtypes have similar or distinct risk factor profiles and causal pathways remains to be determined. Next-generation sequencing has revealed the existence of spatial intratumour genetic heterogeneity in other cancer types.63 The degree of intratumour genetic heterogeneity in endometrial cancers is not yet known and requires detailed analysis of clonal structures. On the basis of the high mutation rates in endometrioid cancers of POLE (ultramutated) and MSI (hypermutated) subtypes, it could be speculated that not all cells in a tumour harbour all mutations identified. Biomarkers assessed in biopsy or curettage specimens might differ based on the sampled areas of tumours, or between primary tumour and metastatic or recurrent disease.63 Devising optimum tissue sampling approaches for genetic characterisation of endometrial cancers and their precursors will be essential for successful translation of genomic findings to clinical management. With use of available genetic information, it is potentially possible to define a target gene panel that would suffice to recapitulate the current histological classification for endometrioid and serous carcinomas and to identify the subtypes defined by the TCGA analysis. The challenge is to incorporate the information into clinically useful prognostic and predictive models that combine clinicopathological features and genetic aberrations.

Prognostication and prediction of response Disease classification Although genetic data can confirm the validity of Bokhman’s type I and type II model to some extent, it is clearly insufficient to encompass the diversity of endometrial cancers. Type I (low-grade) cancers comprise www.thelancet.com/oncology Vol 15 June 2014

Surgical stage (ie, extent of myometrial invasion and local or distant spread) is the most important prognostic factor in endometrial cancer.60,64 Other characteristics, such as the patient’s age and tumour histological grade, type, and size, also provide important prognostic e274

Review

information,9,64 and have been incorporated into nomograms for risk assessment.22 The development of assays or algorithms for accurate identification of patients with high-grade or advanced endometrial cancers who might not need systemic therapy is a priority. Lessons learned from other malignant diseases ought to be embraced, particularly the finding that a combination of clinicopathological and molecular features almost invariably outperforms the use of anatomical or biological markers alone.15 Tools to assess the prognosis of defined molecular subtypes are already available, but the information required for predictive testing is yet to be accrued. Preclinical models of tumours that recapitulate POLE (ultramutated), MSI (hypermutated), copy-number low, and copy-number high genomic subtypes should be developed. Several molecular alterations detected in the genomic subtypes, most of which were previously reported to be recurrently altered in endometrial cancer, might be associated with response to specific targeted therapies.7,9,34,65–73 For example, the presence of mutations in PTEN, PIK3CA, or both, and of FGFR2 mutations could be predictive of response to inhibitors of their respective pathways (table 4).7,9,65,67,68 Identification of novel mutations by next-generation sequencing, including FBXW7 mutations in serous-like cancers and HER2 mutations in cancers with microsatellite instability, might predict response to histone deacetylase inhibitors

Aberration

Prevalence in genomic subtypes7,73

Potential drugs

Method of detection

PTEN

Mutation, homozygous deletion

POLE (ultramutated) 94%; MSI (hypermutated) 89%; copy-number low 77%; copy-number high 15%

PI3K, AKT, and mTOR inhibitors,7,9,65 olaparib, and veliparib66,72

DNA sequencing, immunohistochemistry

PIK3CA

Mutation, amplification

POLE (ultramutated) 71%; MSI (hypermutated) 54%; copy-number low 53%; copy-number high 60%

PI3K, AKT, and mTOR inhibitors7,9,65

DNA sequencing

FGFR2

Mutation, amplification

POLE (ultramutated), 29%, MSI (hypermutated) 14%; copy-number low 13%; copy-number high 10%

Dovitinib, NVP-BGJ398, PD173074, AZD454767,68

DNA sequencing

HER2

Amplification

POLE (ultramutated) 0; MSI (hypermutated) 2%, copynumber low 0; copynumber high 25%

Trastuzumab, pertuzumab, lapatinib69

Immunohistochemistry, FISH, CISH

HER2

Mutation

Neratinib70 POLE (ultramutated) 12%; MSI (hypermutated) 8%; copy-number low 1%; copynumber high 0

DNA sequencing

FBXW7

Mutation

HDAC inhibitors POLE (ultramutated) 82%; (eg, vorinostat)34,71 MSI (hypermutated) 9%; copy-number low 6%; copynumber high 22%

DNA sequencing

FISH=fluorescence in-situ hybridisation. CISH=chromogenic in-situ hybridisation. HDAC=histone deacetylase. MSI=microsatellite instability.

Table 4: Genomic alterations in genomic subtypes of endometrial cancer potentially predictive of response to targeted theray

e275

and the irreversible HER2 inhibitor, neratinib, respectively (table 4).11,12,14,17,34,70,71 Although alterations of the PI3K/AKT/mTOR pathway are known in endometrial cancers, clinical trials of single agents targeting this pathway have shown no substantial therapeutic benefits, or identified any robust biomarkers of therapeutic response.74,75 These studies, however, did not take histological and molecular subtypes into account and the biomarkers tested were analysed retrospectively. These trials might, however, provide opportunities to define the biological basis of the cancer-cell dependency on the PI3K/AKT/mTOR pathway and to identify the mechanisms of sensitivity to specific agents. Other approaches, such as the analysis of tumour genomes in outlier patients,76 might provide clues about the genetic determinants of sensitivity or the role of epistatic interactions between mutations affecting this pathway. Crucial to these endeavours are optimum tissue procurement coupled with detailed histopathological characterisation of the tumours to be studied. Clinical trials testing novel agents generally involve previously treated patients who have advanced or recurrent disease. Whether the repertoire of genetic aberrations differs between early, advanced, or recurrent endometrial carcinoma (as it does in primary and metastatic breast cancer77,78) is unknown. Low-grade endometrioid carcinomas occasionally recur as highgrade or histologically ambiguous tumours, and frequently the progesterone receptor profile differs from that of the primary tumour.21 These characteristics suggest that clonal selection occurs during tumour progression. Additionally, a clinical trial of the allosteric mTORC1 inhibitor temsirolimus in women with recurrent or metastatic endometrial cancer revealed lower response rates in patients who had previously received chemotherapy than in chemotherapy-naive patients.74 These factors need to be taken into account in the design of clinical trials. TCGA analyses confirmed that a substantial proportion (up to 45%) of serous carcinomas harbour PIK3CA mutations or amplifications.7,11 Furthermore, in the above mentioned clinical trial, response to allosteric mTORC1 inhibition was seen in a subset of endometrioid and serous carcinomas.74 Whether the serous carcinomas sensitive to mTORC1 inhibitors have PI3KCA alterations is unclear. Nevertheless, the findings suggest that treatment options for serous carcinoma patients are not limited to chemotherapy. Genomic analyses are likely to identify additional subsets of endometrial cancers, even within individual TCGA genomic classes, that are driven by distinct genetic aberrations and might be targeted by specific therapeutic agents. Finer-scale stratification of endometrial carcinomas on the basis of hormone receptor expression could also be possible. Oestrogen-receptor and progesterone-receptor status are prognostic in endometrial cancers9,79,80 and might help to define a subset of patients who will benefit from endocrine www.thelancet.com/oncology Vol 15 June 2014

Review

Search strategy and selection criteria Data for this Review were identified by searches of PubMed, references from relevant articles, and internet searches, with the search terms “endometrium”, “endometrial”, “uterus”, “uterine corpus”, and “cancer”, “carcinoma”, “adenocarcinoma”, “endometrial carcinoma”, or “endometrial adenocarcinoma”. Only articles published in English between January, 1983, and September, 2013, were included. The final reference list was generated on the basis of originality and relevance to this paper.

therapies.81,82 Before the incorporation of these markers into the diagnostic and management algorithms for endometrial cancer, preanalytical and analytical methods and cutoff values for immunohistochemistry-based assays for oestrogen and progesterone receptors must be standardised,9 akin to the American Society of Clinical Oncology/College of American Pathology guidelines for breast cancer.83 Furthermore, studies to investigate the prognostic importance of biomarkers, and whether they offer additional value to that provided by the genomic taxonomy, are warranted.

Disease screening and detection Genomic characterisation of endometrial cancers will undoubtedly have applications beyond tumour taxonomy and therapeutic prediction. Such information would probably provide means for the development of highly accurate screening tests and methods for early detection and monitoring of disease. In a proof-of-principle study endometrial cancer cells were detected with massively parallel sequencing analysis of DNA extracted from routine liquid-based cervical cytological specimens.84

Conclusions The results of next-generation sequencing studies have substantially broadened understanding of endometrial cancers. Nevertheless, detailed analyses of rare subtypes and clarification of the biological importance of novel mutations are required for translation of these insights into clinical benefit for patients. All information needed to define the optimum therapy for each patient is unlikely to be obtained from the analysis of somatic genetic aberrations, particularly because of the complexity of genetic aberrations affecting components of the same pathway in endometrial cancers. Understanding the relevance of specific mutations might require carefully designed functional experiments and optimised models of the disease, coupled with innovative clinical trial designs.85 Finally, as technologies for epigenomic and proteomic profiling evolve, global, integrative analyses will undoubtedly provide the bases for development of a definitive classification system combining the anatomical and molecular characteristics of endometrial cancers. www.thelancet.com/oncology Vol 15 June 2014

Contributors All authors contributed to the literature search, preparation of figures and tables, and writing of the paper. Declaration of interests We declare that we have no competing interests. References 1 Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin 2013; 63: 11–30. 2 Cancer Research UK. Uterine cancer statistics. http://info. cancerresearchuk.org/cancerstats/types/uterus/ (accessed Jan 23, 2014). 3 Wartko P, Sherman ME, Yang HP, Felix AS, Brinton LA, Trabert B. Recent changes in endometrial cancer trends among menopausalage U.S. women. Cancer Epidemiol 2013; 37: 374–77. 4 Creasman WT, Odicino F, Maisonneuve P, et al. Carcinoma of the corpus uteri. FIGO 26th annual report on the results of treatment in gynecological cancer. Int J Gynaecol Obstet 2006; 95 (suppl 1): S105–43. 5 Bokhman JV. Two pathogenetic types of endometrial carcinoma. Gynecol Oncol 1983; 15: 10–17. 6 Silverberg SG, Kurman RJ, Nogales F, Mutter GL, Kubik-Huch RA, Tavassoli FA. Epithelial tumours and related lesions. In: Tavassoli FA, Devilee P, eds. World Health Organization classification of tumours: pathology and genetics— tumours of the breast and female genital organs. Lyon: IARC Press, 2003: 217–32. 7 Dedes KJ, Wetterskog D, Ashworth A, Kaye SB, Reis-Filho JS. Emerging therapeutic targets in endometrial cancer. Nat Rev Clin Oncol 2011; 8: 261–71. 8 Matias-Guiu X, Prat J. Molecular pathology of endometrial carcinoma. Histopathology 2013; 62: 111–23. 9 Salvesen HB, Haldorsen IS, Trovik J. Markers for individualised therapy in endometrial carcinoma. Lancet Oncol 2012; 13: e353–61. 10 Weigelt B, Banerjee S. Molecular targets and targeted therapeutics in endometrial cancer. Curr Opin Oncol 2012; 24: 554–63. 11 Kandoth C, Schultz N, Cherniack AD, et al, for the Cancer Genome Atlas Research Network. Integrated genomic characterization of endometrial carcinoma. Nature 2013; 497: 67–73. 12 Kuhn E, Wu RC, Guan B, et al. Identification of molecular pathway aberrations in uterine serous carcinoma by genome-wide analyses. J Natl Cancer Inst 2012; 104: 1503–13. 13 Lax SF, Kendall B, Tashiro H, Slebos RJ, Hedrick L. The frequency of p53, K-ras mutations, and microsatellite instability differs in uterine endometrioid and serous carcinoma: evidence of distinct molecular genetic pathways. Cancer 2000; 88: 814–24. 14 Le Gallo M, O’Hara AJ, Rudd ML, et al, for the NIH Intramural Sequencing Center (NISC) Comparative Sequencing Program. Exome sequencing of serous endometrial tumors identifies recurrent somatic mutations in chromatin-remodeling and ubiquitin ligase complex genes. Nat Genet 2012; 44: 1310–15. 15 McConechy MK, Ding J, Cheang MC, et al. Use of mutation profiles to refine the classification of endometrial carcinomas. J Pathol 2012; 228: 20–30. 16 Urick ME, Rudd ML, Godwin AK, Sgroi D, Merino M, Bell DW. PIK3R1 (p85α) is somatically mutated at high frequency in primary endometrial cancer. Cancer Res 2011; 71: 4061–67. 17 Zhao S, Choi M, Overton JD, et al. Landscape of somatic single-nucleotide and copy-number mutations in uterine serous carcinoma. Proc Natl Acad Sci USA 2013; 110: 2916–21. 18 Garg K, Broaddus RR, Soslow RA, Urbauer DL, Levine DA, Djordjevic B. Pathologic scoring of PTEN immunohistochemistry in endometrial carcinoma is highly reproducible. Int J Gynecol Pathol 2012; 31: 48–56. 19 Hoang LN, McConechy MK, Köbel M, et al. Histotype-genotype correlation in 36 high-grade endometrial carcinomas. Am J Surg Pathol 2013; 37: 1421–32. 20 Soslow RA. High-grade endometrial carcinomas—strategies for typing. Histopathology 2013; 62: 89–110. 21 Soslow RA, Wethington SL, Cesari M, et al. Clinicopathologic analysis of matched primary and recurrent endometrial carcinoma. Am J Surg Pathol 2012; 36: 1771–81.

e276

Review

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42 43

44

e277

Abu-Rustum NR, Zhou Q, Gomez JD, et al. A nomogram for predicting overall survival of women with endometrial cancer following primary therapy: toward improving individualized cancer care. Gynecol Oncol 2010; 116: 399–403. Cao Q J, Belbin T, Socci N, et al. Distinctive gene expression profiles by cDNA microarrays in endometrioid and serous carcinomas of the endometrium. Int J Gynecol Pathol 2004; 23: 321–29. Fles R, Hoogendoorn WE, Platteel I, et al. Genomic profile of endometrial tumors depends on morphological subtype, not on tamoxifen exposure. Genes Chromosomes Cancer 2010; 49: 699–710. Setiawan VW, Yang HP, Pike MC, et al, and the Australian National Endometrial Cancer Study Group. Type I and II endometrial cancers: have they different risk factors? J Clin Oncol 2013; 31: 2607–18. Stefanick ML. Estrogens and progestins: background and history, trends in use, and guidelines and regimens approved by the US Food and Drug Administration. Am J Med 2005; 118 (suppl 12B): 64–73. Kelly T, Yang W, Chen CS, Reynolds K, He J. Global burden of obesity in 2005 and projections to 2030. Int J Obes (Lond) 2008; 32: 1431–37. Garg K, Soslow RA. Lynch syndrome (hereditary non-polyposis colorectal cancer) and endometrial carcinoma. J Clin Pathol 2009; 62: 679–84. Brinton LA, Felix AS, McMeekin DS, et al. Etiologic heterogeneity in endometrial cancer: evidence from a Gynecologic Oncology Group trial. Gynecol Oncol 2013; 129: 277–84. Voss MA, Ganesan R, Ludeman L, et al. Should grade 3 endometrioid endometrial carcinoma be considered a type 2 cancer-a clinical and pathological evaluation. Gynecol Oncol 2012; 124: 15–20. Soslow RA, Bissonnette JP, Wilton A, et al. Clinicopathologic analysis of 187 high-grade endometrial carcinomas of different histologic subtypes: similar outcomes belie distinctive biologic differences. Am J Surg Pathol 2007; 31: 979–87. Zannoni GF, Scambia G, Gallo D. The dualistic model of endometrial cancer: the challenge of classifying grade 3 endometrioid carcinoma. Gynecol Oncol 2012; 127: 262–63. Gilks CB, Oliva E, Soslow RA. Poor interobserver reproducibility in the diagnosis of high-grade endometrial carcinoma. Am J Surg Pathol 2013; 37: 874–81. Garnett MJ, Edelman EJ, Heidorn SJ, et al. Systematic identification of genomic markers of drug sensitivity in cancer cells. Nature 2012; 483: 570–75. Podsypanina K, Ellenson LH, Nemes A, et al. Mutation of Pten/ Mmac1 in mice causes neoplasia in multiple organ systems. Proc Natl Acad Sci USA 1999; 96: 1563–68. Stambolic V, Tsao MS, Macpherson D, Suzuki A, Chapman WB, Mak TW. High incidence of breast and endometrial neoplasia resembling human Cowden syndrome in pten+/- mice. Cancer Res 2000; 60: 3605–11. Hayes MP, Wang H, Espinal-Witter R, et al. PIK3CA and PTEN mutations in uterine endometrioid carcinoma and complex atypical hyperplasia. Clin Cancer Res 2006; 12: 5932–35. Mutter GL, Lin MC, Fitzgerald JT, et al. Altered PTEN expression as a diagnostic marker for the earliest endometrial precancers. J Natl Cancer Inst 2000; 92: 924–30. Oda K, Stokoe D, Taketani Y, McCormick F. High frequency of coexistent mutations of PIK3CA and PTEN genes in endometrial carcinoma. Cancer Res 2005; 65: 10669–73. Rudd ML, Price JC, Fogoros S, et al. A unique spectrum of somatic PIK3CA (p110α) mutations within primary endometrial carcinomas. Clin Cancer Res 2011; 17: 1331–40. Cheung LW, Hennessy BT, Li J, et al. High frequency of PIK3R1 and PIK3R2 mutations in endometrial cancer elucidates a novel mechanism for regulation of PTEN protein stability. Cancer Discov 2011; 1: 170–85. Samuels Y, Wang Z, Bardelli A, et al. High frequency of mutations of the PIK3CA gene in human cancers. Science 2004; 304: 554. Santin AD, Bellone S, Van Stedum S, et al. Amplification of c-erbB2 oncogene: a major prognostic indicator in uterine serous papillary carcinoma. Cancer 2005; 104: 1391–97. Moreno-Bueno G, Hardisson D, Sarrió D, et al. Abnormalities of E- and P-cadherin and catenin (β-, γ-catenin, and p120ctn) expression in endometrial cancer and endometrial atypical hyperplasia. J Pathol 2003; 199: 471–78.

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60 61

62

63

64

65

66

67

Schlosshauer PW, Ellenson LH, Soslow RA. β-catenin and E-cadherin expression patterns in high-grade endometrial carcinoma are associated with histological subtype. Mod Pathol 2002; 15: 1032–37. Simpkins SB, Bocker T, Swisher EM, et al. MLH1 promoter methylation and gene silencing is the primary cause of microsatellite instability in sporadic endometrial cancers. Hum Mol Genet 1999; 8: 661–66. Salvesen HB, MacDonald N, Ryan A, et al. Methylation of hMLH1 in a population-based series of endometrial carcinomas. Clin Cancer Res 2000; 6: 3607–13. Tashiro H, Isacson C, Levine R, Kurman RJ, Cho KR, Hedrick L. p53 gene mutations are common in uterine serous carcinoma and occur early in their pathogenesis. Am J Pathol 1997; 150: 177–85. Jia L, Liu Y, Yi X, et al. Endometrial glandular dysplasia with frequent p53 gene mutation: a genetic evidence supporting its precancer nature for endometrial serous carcinoma. Clin Cancer Res 2008; 14: 2263–69. Wiegand KC, Lee AF, Al-Agha OM, et al. Loss of BAF250a (ARID1A) is frequent in high-grade endometrial carcinomas. J Pathol 2011; 224: 328–33. Mao TL, Ardighieri L, Ayhan A, et al. Loss of ARID1A expression correlates with stages of tumor progression in uterine endometrioid carcinoma. Am J Surg Pathol 2013; 37: 1342–48. Werner HM, Berg A, Wik E, et al. ARID1A loss is prevalent in endometrial hyperplasia with atypia and low-grade endometrioid carcinomas. Mod Pathol 2013; 26: 428–34. McConechy MK, Anglesio MS, Kalloger SE, et al, for the Australian Ovarian Cancer Study Group. Subtype-specific mutation of PPP2R1A in endometrial and ovarian carcinomas. J Pathol 2011; 223: 567–73. Shih IM, Wang TL. Mutation of PPP2R1A: a new clue in unveiling the pathogenesis of uterine serous carcinoma. J Pathol 2011; 224: 1–4. Liang H, Cheung LW, Li J, et al. Whole-exome sequencing combined with functional genomics reveals novel candidate driver cancer genes in endometrial cancer. Genome Res 2012; 22: 2120–29. Gui Y, Guo G, Huang Y, et al. Frequent mutations of chromatin remodeling genes in transitional cell carcinoma of the bladder. Nat Genet 2011; 43: 875–78. Heravi-Moussavi A, Anglesio MS, Cheng SW, et al. Recurrent somatic DICER1 mutations in nonepithelial ovarian cancers. N Engl J Med 2012; 366: 234–42. Church DN, Briggs SE, Palles C, et al, and the NSECG Collaborators. DNA polymerase ε and δ exonuclease domain mutations in endometrial cancer. Hum Mol Genet 2013; 22: 2820–28. Travis WD, Brambilla E, Riely GJ. New pathologic classification of lung cancer: relevance for clinical practice and clinical trials. J Clin Oncol 2013; 31: 992–1001. Pecorelli S. Revised FIGO staging for carcinoma of the vulva, cervix, and endometrium. Int J Gynaecol Obstet 2009; 105: 103–04. Hayes DF, Bast RC, Desch CE, et al. Tumor marker utility grading system: a framework to evaluate clinical utility of tumor markers. J Natl Cancer Inst 1996; 88: 1456–66. McCluggage WG. Uterine carcinosarcomas (malignant mixed Mullerian tumors) are metaplastic carcinomas. Int J Gynecol Cancer 2002; 12: 687–90. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med 2012; 366: 883–92. Barlin JN, Zhou Q, St Clair CM, et al. Classification and regression tree (CART) analysis of endometrial carcinoma: seeing the forest for the trees. Gynecol Oncol 2013; 130: 452–56. Slomovitz BM, Coleman RL. The PI3K/AKT/mTOR pathway as a therapeutic target in endometrial cancer. Clin Cancer Res 2012; 18: 5856–64. Forster MD, Dedes KJ, Sandhu S, et al. Treatment with olaparib in a patient with PTEN-deficient endometrioid endometrial cancer. Nat Rev Clin Oncol 2011; 8: 302–06. Konecny GE, Kolarova T, O’Brien NA, et al. Activity of the fibroblast growth factor receptor inhibitors dovitinib (TKI258) and NVP-BGJ398 in human endometrial cancer cells. Mol Cancer Ther 2013; 12: 632–42.

www.thelancet.com/oncology Vol 15 June 2014

Review

68 69

70

71

72

73

74

75

76 77 78

Byron SA, Pollock PM. FGFR2 as a molecular target in endometrial cancer. Future Oncol 2009; 5: 27–32. English DP, Roque DM, Santin AD. HER2 expression beyond breast cancer: therapeutic implications for gynecologic malignancies. Mol Diagn Ther 2013; 17: 85–99. Bose R, Kavuri SM, Searleman AC, et al. Activating HER2 mutations in HER2 gene amplification negative breast cancer. Cancer Discov 2013; 3: 224–37. He L, Torres-Lockhart K, Forster N, et al. Mcl-1 and FBW7 control a dominant survival pathway underlying HDAC and Bcl-2 inhibitor synergy in squamous cell carcinoma. Cancer Discov 2013; 3: 324–37. Dedes KJ, Wetterskog D, Mendes-Pereira AM, et al. PTEN deficiency in endometrioid endometrial adenocarcinomas predicts sensitivity to PARP inhibitors. Sci Transl Med 2010; 2: 53ra75. Gao J, Aksoy BA, Dogrusoz U, et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 2013; 6: pl1. Oza AM, Elit L, Tsao MS, et al. Phase II study of temsirolimus in women with recurrent or metastatic endometrial cancer: a trial of the NCIC Clinical Trials Group. J Clin Oncol 2011; 29: 3278–85. Trédan O, Treilleux I, Wang Q, et al. Predicting everolimus treatment efficacy in patients with advanced endometrial carcinoma: a GINECO group study. Target Oncol 2013; 8: 243–51. Iyer G, Hanrahan AJ, Milowsky MI, et al. Genome sequencing identifies a basis for everolimus sensitivity. Science 2012; 338: 221. Ding L, Ellis MJ, Li S, et al. Genome remodelling in a basal-like breast cancer metastasis and xenograft. Nature 2010; 464: 999–1005. Shah SP, Morin RD, Khattra J, et al. Mutational evolution in a lobular breast tumour profiled at single nucleotide resolution. Nature 2009; 461: 809–13.

www.thelancet.com/oncology Vol 15 June 2014

79

80

81

82

83

84

85

Zannoni GF, Monterossi G, De Stefano I, et al. The expression ratios of estrogen receptor α (ERα) to estrogen receptor β1 (ERβ1) and ERα to ERβ2 identify poor clinical outcome in endometrioid endometrial cancer. Hum Pathol 2013; 44: 1047–54. Huvila J, Talve L, Carpén O, et al. Progesterone receptor negativity is an independent risk factor for relapse in patients with early stage endometrioid endometrial adenocarcinoma. Gynecol Oncol 2013; 130: 463–69. Singh M, Zaino RJ, Filiaci VJ, Leslie KK. Relationship of estrogen and progesterone receptors to clinical outcome in metastatic endometrial carcinoma: a Gynecologic Oncology Group Study. Gynecol Oncol 2007; 106: 325–33. Thigpen JT, Brady MF, Alvarez RD, et al. Oral medroxyprogesterone acetate in the treatment of advanced or recurrent endometrial carcinoma: a dose-response study by the Gynecologic Oncology Group. J Clin Oncol 1999; 17: 1736–44. Hammond ME, Hayes DF, Dowsett M, et al. American Society of Clinical Oncology/College of American Pathologists guideline recommendations for immunohistochemical testing of estrogen and progesterone receptors in breast cancer. J Clin Oncol 2010; 28: 2784–95. Kinde I, Bettegowda C, Wang Y, et al. Evaluation of DNA from the Papanicolaou test to detect ovarian and endometrial cancers. Sci Transl Med 2013; 5: 167ra4. Baselga J. Bringing precision medicine to the clinic: from genomic profiling to the power of clinical observation. Ann Oncol 2013; 24: 1956–57.

e278