Endometrial adenocarcinomas are a heterogeneous group of tumors derived from endometrial glandular epithelial cells. Most maintain a resemblance to endometrial glands (‘endometrioid’) but even in these cases mucinous or squamous differentiation occurs frequently. A much less common group of endometrial carcinomas shows non-endometrioid histology and includes clear cell and serous carcinomas. This chapter provides an overview of current diagnostic criteria as well as relevant molecular genetic findings.
Adenocarcinoma of the endometrium is the most common gynecologic cancer in the United States, having a lifetime occurrence risk of 2.5%1 with 44,000 new cases and 7950 deaths annually.2 The median age at presentation is 63 years, of which 90% of cases are found in women past menopause and only 1% are under age 40 years. The absolute prevalence of endometrial cancer is affected by the background hysterectomy rate,3 which varies greatly between populations, and is 40% by age 60 in the United States.4
A two-type view of endometrial adenocarcinoma combines epidemiologic, clinical, histologic, and molecular genetic data, and provides a useful pathogenetic model supported by multiple lines of evidence (Figure 18.1). The two types are: endometrioid carcinomas and their variants (type 1) and the non-endometrioid (type 2) carcinomas.
Figure 18.1 Types of endometrial carcinoma, and their clinically significant subdivisions. Endometrioid (type 1) tumors include many histologic variants, but it is tumor grade that is most important in risk prognostication. Non-endometrioid tumors (type 2) are ungraded, but histotypes should be diagnosed separately as each has a distinctive natural history.
For many years endometrial cancers were subdivided according to histologic grade into well, moderately, and poorly differentiated groups. This was the case until serous carcinoma, the most aggressive type of endometrial cancer, was recognized in 1982,5 and soon found to be unassociated with unopposed estrogens, the risk factor tightly linked to other endometrial carcinomas. Based on this finding, a proposal was made to separate endometrial adenocarcinomas into two distinct groups designated as types 1 and 2 (Table 18.1).6 It was a division into either indolent estrogen-induced or aggressive estrogen-independent tumors, respectively. Type 1 was further designated as ‘endometrioid’ and type 2 ‘non-endometrioid’ to reflect their divergent histologies, thereby suggesting that the noted functional differences might be extrapolated to their histologic classification.7
Molecular genetic analyses of well-differentiated endometrioid and serous carcinomas support this dualistic model of endometrial tumorigenesis (Figure 18.2).8 Studies using cDNA microarrays confirm that these two tumor types have distinctively different gene expression profiles.9 Increasing genetic damage can be seen in precursor lesions within the endometrioid pathway, beginning with PTEN or PAX2 inactivation in normal-appearing glands (latent precancers), followed by positive hormonal selection and clonal outgrowth as endometrial intraepithelial neoplasia (EIN; also known as atypical hyperplasia or AH), and then cancer (Figure 18.3).10 Comparable gradations are less evident in the serous carcinoma pathway, where a noninvasive form of disease, serous endometrial intraepithelial carcinoma (serous EIC), is genetically identical to its invasive counterpart and also capable of metastasizing.
Figure 18.2 Type 1 and 2 endometrial carcinoma differ in the frequency of specific mutations, but there is substantial crossover especially in the grade III endometrioid and mixed categories. p53 inactivation is present in about 90% of serous carcinomas, and much less in clear cell carcinomas and carcinosarcomas.
Although the dualistic model is applicable to a high proportion of endometrial carcinomas, not all tumors fit in. In fact, a gray zone exists between the two broad types, with a significant number of tumors showing overlapping clinical, morphologic, and molecular features (Figures 18.1 and 18.2). Moreover, there is an ongoing debate about whether a histologic subset of endometrioid carcinomas (those that are poorly differentiated or have high nuclear grade) should be assigned to the type 2 group. Furthermore, it is now accepted that a non-endometrioid component may emerge from a pre-existing endometrioid carcinoma. The probable mechanism for this is development of genetic heterogeneity within elements of a type 1 tumor, and progressive expansion of a particularly aggressive tumor subclone with genetic and behavioral features resembling those of type 2 tumors9 (Figure 18.2).
Some unique tumor histotypes, such as clear cell carcinomas and carcinosarcomas, demonstrate clinical and molecular features that do not fit into either of the prototype endometrioid and non-endometrioid categories.133 These other histologic types may achieve even greater significance if and when specific therapeutic agents are developed to their peculiar molecular genetic alterations.
There is one aspect of the type 1/2 divide that has been extended to clinical care. Histologic grade of endometrioid (type 1) carcinomas has independent value in determining prognosis and should be included in the diagnosis, whereas all of the type 2 tumors (serous, clear cell, carcinosarcoma) are sufficiently aggressive that subclassification by grade is unnecessary.
Estrogen-related type 1 tumors frequently demonstrate one or more of the following: microsatellite instability, inactivated PTEN tumor suppressor gene, KRAS mutations, and activation of the β-catenin gene (CTNNB1).11 In contrast, the estrogen-independent type 2 tumors show loss of heterozygosity at different loci, altered p53, and abnormalities in genes regulating mitotic checkpoints.9 However, p53 mutations are found in approximately 10% of endometrioid carcinomas, most frequently in grade 3 and occasionally in grade 2 tumors. Overall, p53 mutations occur in 50% of grade 3 tumors, but not in grade 1 tumors or EIN/AH. This finding suggests that p53 is involved in the progression, but not the initiation, of endometrioid carcinoma.
Our understanding of synergies between multiple independent genetic events that produce endometrioid (type 1) carcinoma centers upon the PI3K pathway, and its ability when perturbed to activate AKT (Figures 18.2–18.4). It is estimated that more than 80% of endometrioid carcinomas have an abnormality in the PI3K pathway,12 which act to increase levels of PIP3, an activator of AKT. Thus PTEN inactivation (60% of cases), or constitutive activation of KRAS (10–30%) or PIK3CA (30%) act in concert to accumulate PIP3, which in turn activates AKT by phosphorylation.13 Once activated, pAKT initiates a cascade of tumorigenic events that includes stimulation of the mTOR pathway, deregulation of cell cycle control, blocking of apoptosis, and prolonged cell survival.13 This model presents new downstream targets for suppression by pharmacologic inhibitors, such as the mTOR pathway.
Figure 18.4 Activation of the AKT pathway in endometrioid (type 1) tumorigenesis. Activation of AKT by PIP-mediated phosphorylation is augmented by primary mutations (*), which activate the KRAS and/or PIK3CA genes, or inactivate PTEN. The combination of increased PIP3 production (red) with diminished PIP3 dephosphorylation (blue) has the net effect of increasing PIP3. pAKT has numerous downstream effects including activation of the mTOR pathway, changes in cell cycle regulation, blockade of apoptosis, and prolonged cell survival.
Recognition of histologic subtypes is an important factor in planning treatment and predicting clinical outcome. A histologic classification of endometrial carcinomas is given in Table 18.2, and this chapter is organized according to these diagnostic categories.
The overriding stimulus behind the development of EIN/AH and the endometrioid type of endometrial carcinoma is the effect of estrogens, both endogenous and exogenous. These are not mirrored in non-endometrioid tumors, with the exception of carcinosarcomas. Beginning in 1969, a notable rise in the number of endometrial cancers occurred, which coincided with a fourfold increase in the use of estrogens for the alleviation of perimenopausal and postmenopausal symptoms in women. The relative risk of developing endometrial carcinoma in women taking unopposed estrogens is elevated 3- to 6-fold,14 rising to 9.5-fold if unopposed estrogen has been used for 10 years or longer.15 The increased risk persists for several years after the estrogen is discontinued.16 The risk is roughly similar whether the estrogens are taken continuously or cyclically. The additional administration of progestins for several days of each month reduces the risk of carcinoma to baseline population levels. Although progestins are usually prescribed for either 7 or 10 days per month in women taking estrogen-replacement therapy, the protection from endometrial cancer is much greater if progestins are used for at least 10 days or given continuously as combined estrogen-progestin therapy.
The overall relative risk for an obese woman to develop endometrial cancer increases proportionally with increasing body mass index, up to sixfold for the morbidly obese (BMI > 40).17–19 This risk can be reduced to near baseline levels with bariatric surgery and successful weight reduction.20 Endometrial cancer is only one of the many comorbidities seen in the obese patient. The association in women past menopause is commonly explained in terms of increased aromatization of androgens to estrogens (estrone and estradiol) in adipose tissue, this being the major source of estrogens in women of this age group. A woman who completes surgical therapy for stage 1 endometrial cancer is more likely to die of cardiovascular disease than any other cause including her cancer.21
The magnitude of endometrial cancer risk in patients with type 2 diabetes has been difficult to measure because of a high frequency of coexisting risk factors such as obesity and polycystic ovarian disease. However, improved quantitative markers have confirmed a positive association of diabetes with endometrial cancer. Insulin resistance, as measured by the (inversely proportional) serum surrogate marker adiponectin, correlates highly with endometrial cancer risk even when corrected for obesity.22,23
Polycystic ovary syndrome (PCOS) is a constellation of endocrine disorders expressing at least two of the following features: anovulation or infrequent ovulation, androgen excess, and polycystic ovaries.24 Ovarian cysts typically are theca–lutein follicles with prominent luteinization of the theca interna and an inconspicuous granulosa cell layer. Primary endocrine defects in PCOS are peripheral insulin resistance and excess ovarian production of androgens.25 The population of affected women is enriched for risk factors such as obesity, with endometrial findings indicating a hyperestrogenic state.26 Endometrial carcinoma occurs at 2.7-fold increased risk in women with PCOS,27 but, as the women are all young, this number comprises a significant proportion of endometrial carcinomas in women under age 45 years.
Granulosa cell tumor is a relatively uncommon tumor that mainly affects women shortly after menopause. Most tumors produce increased estrogens and about half the women affected present with postmenopausal bleeding, one-third having proliferative endometrium. Endometrial carcinomas occur in 9–13% of women with granulosa cell tumors.28
Women with hereditary non-polyposis colorectal carcinoma syndrome (HNPCC), Lynch syndrome, a condition affecting about 1% of the population, have a 70% lifetime risk of endometrial adenocarcinoma.29 They tend to develop disease 15 years earlier than sporadic occurrences and the prognosis is favorable.30 Their tumors show many of the histopathologic and genetic features of endometrioid endometrial adenocarcinomas,31 including transit through a premalignant EIN/AH phase,32 and genetic alterations in mismatch repair genes (mainly MSH-2, MSH-6, and MLH-1)33,34 and PTEN.35 Immunohistochemistry for DNA mismatch repair proteins is often abnormal in endometrial cancers of Lynch syndrome patients, and has been suggested as a screening test for this heritable condition. Immunohistochemistry is nonspecific, however, because it is a somatically acquired non-heritable feature in 18% of all sporadic endometrial carcinomas. One cost–benefit study has recommended restricting immunotesting of primary endometrial tumors to those patients having additional clinical risk factors, such as one first-degree relative with a Lynch-associated carcinoma.36
Germline BRCA mutation, which increases breast and ovarian cancer occurrences, does not significantly alter endometrial cancer risk.37 There may be a small indirect effect on endometrial cancer incidence in those patients managed by prophylactic tamoxifen administration.38
The antiestrogen tamoxifen is widely used as an adjuvant therapy for women with breast cancer. Tamoxifen is a nonsteroidal compound that competes with estrogen for estrogen receptors (ERs). In women of childbearing age it antagonizes endogenous estrogens and induces endometrial inactivity or atrophy, but in postmenopausal women, who are normally hypoestrogenic, it may have a weak estrogenic effect. Tamoxifen administration is associated with an overall slightly increased risk (2–3 times) of endometrial adenocarcinoma.39 Carcinoma occurrences are mainly of early stage and low grade, but a small subset of aggressive high-grade endometrioid carcinomas, clear cell carcinomas, or carcinosarcomas are disproportionately increased.40,41 Placement of a levonorgestrel-impregnated intrauterine device in tamoxifen-treated patients has not yet demonstrated statistically significant protection from tamoxifen-related endometrial cancers.42
Most studies have shown an association between early age at menarche, late age at natural menopause, and total length of ovulation span,43 although these findings are not universal. The use of oral contraceptives reduces the risk of endometrial cancer, in some studies by half.
Nulliparity is a strong independent risk factor for endometrial carcinoma.43 Women with endometrial carcinoma are less likely to have had children than normal controls and, if they are parous, they will have had fewer children. Infertility, particularly that associated with anovulation and progesterone insufficiency, is also associated with the risk of developing endometrial carcinoma.45 Nulliparity is significant only when the endometrial carcinoma develops before menopause and not after.46 This suggests that the hormonal disturbances that prevent conception also encourage malignant change in the endometrium. The protective effect of pregnancy applies only to full-term pregnancy.47
Cigarette smoking reduces the risk of endometrial carcinoma. The effect is limited primarily to women whose disease is detected after menopause and, among these women, current smokers show the greatest reduction in risk, and former smokers are less affected.48 The mechanism whereby cigarette smoking reduces risk is not clear. One study has shown that serum estrogen levels were unaffected but that androstenedione levels were slightly higher in smokers.49 Paradoxically, women with advanced stage endometrial carcinoma (stages II–IV) were more likely to be smokers than women with early stage disease (stages 0–I).50
Endometrioid carcinoma is the most common type of endometrial cancer accounting for approximately 60% of cases. The term ‘endometrioid’ derives from the tumor’s predominant glandular pattern, which resembles proliferative-phase endometrium. Nevertheless, its histologic variants may have a minor, or even predominant, mucinous, secretory, squamous, ciliated, or sex cord-like pattern.
Most tumors develop slowly in the setting of hyperestrogenism against a background of non-atypical endometrial hyperplasia, although some arise in atrophic endometrium.51 Endometrioid carcinoma is predominantly a disease of the sixth and seventh decades and 75% of cases occur after the menopause. Only 5% occur in women less than 40 years old. They are low-grade, non-myoinvasive, associated with a good prognosis and often develop after a long history of anovulatory cycles or estrogen therapy. Endometrial carcinoma rarely occurs in pregnancy.52
Women with endometrioid carcinoma most often present with abnormal vaginal bleeding, which means in the majority of cases postmenopausal bleeding. The fact that they bleed, however, simply means that the tumor is often large or advanced. Smaller carcinomas may be asymptomatic. Surprisingly frequently, asymptomatic tumor is documented in women who have an endometrial biopsy before instituting hormone replacement therapy, or had the tumor discovered initially at autopsy.53 Patients with advanced disease may complain of pelvic pain, which reflects tumor spread.
The diagnosis is made by endometrial biopsy or curettage, but imaging techniques and hysteroscopy are being used more and can play an important role. Outpatient endometrial sampling techniques (Pipelle biopsy) have an excellent diagnostic rate for endometrial carcinoma, similar to that for curettage.54
Endometrioid carcinoma can present variously to the naked eye when the uterus is opened. The uterus may be slightly or grossly enlarged but it may be of normal size or even small and atrophic, particularly in a postmenopausal woman. Most tumors arise in the corpus but some originate in the lower uterine segment. The tumor may present as a single mass or there may be multiple separate masses (Figure 18.5) or a diffuse thickening of the endometrium. Carcinomas are situated more frequently on the posterior than the anterior wall. The most common appearance is of an exophytic, rough, perhaps papillary area of the endometrium with a shaggy surface and ulceration (Figure 18.6). Sometimes the tumor is polypoid, with a fairly narrow base. When this is the case, its surface may be smooth and hemorrhagic and the uterine cavity distended, with concomitant thinning of the uterine wall. When the tumor is polypoid, the remaining endometrium usually appears thin. Myometrial invasion may be obvious to the naked eye (Figure 18.6), with either pushing or infiltrating borders, but frequently it is difficult to appreciate the degree of myometrial invasion grossly. There seems to be no correlation between the degree of exophytic growth of the tumor within the uterine cavity and the presence of myometrial invasion.55 However, a tumor diameter of more than 2 cm generally is associated with poorer prognosis and a higher frequency of distant failure.57
The glandular pattern and cellular features generally resemble the proliferative endometrium. Carcinoma is recognized within the endometrial compartment by the presence of at least one of the following: meandering interconnected lumens formed by folded sheets of neoplastic epithelium (Figure 18.7), irregular angulated and tapering glandular contours (Figure 18.8), a cribriform pattern of the glands (Figures 18.9 and 18.10), or a solid area of glandular epithelium (Figures 18.10 and 18.11). Several features may be present together. These points are summarized in Figure 18.12.
Figure 18.7 Endometrial adenocarcinoma, endometrioid type, grade 1 (well differentiated). The branching glands with maze-like lumens indicate an architecture of folded sheets of neoplastic epithelium, rather than dense packing of tubular structures.
Figure 18.12 (A–E) Intraendometrial adenocarcinoma. Glandular architectural patterns that may be seen in non-myoinvasive areas. (A) Cribriform glands; (B) Maze-like or meandering lumens; (C) Villoglandular; (D) Confluent polygonal molded glands; (E) Solid growth.
Stratification of epithelial cells is almost always seen. Occasionally, cribriform fragments have a microglandular appearance easily confused with a cervical lesion (Figure 18.13). The individual epithelial cells are larger than would be expected in the proliferative phase. Compared with the normal endometrium, the carcinoma cells have a distinctly altered cytology that varies between cases and even within areas of a single tumor, but may include rounded nuclei, clumped chromatin, and prominent nucleoli. Individual tumors frequently show patchy changes in differentiation to mucinous, squamous, tubal, or other cytologies, and in these cases cytoplasmic as well as nuclear features stand out from the normal background. Some endometrioid adenocarcinomas secrete abundant mucin (‘mucin-rich’ variant), and these may or may not have identifiable intracytoplasmic mucin. Mitotic figures are usually present but may be scanty in well-differentiated tumors.
Figure 18.13 Endometrial adenocarcinoma, endometrioid type, with microglandular cribriform architecture. These delicate lesions tend to be exophytic, and may appear highly fragmented in biopsy specimens.
Endometrioid adenocarcinomas, like their precursor EIN lesions, spread within the endometrial compartment by extension of newly formed, well-differentiated neoplastic glands into the adjacent stroma (centripetal growth). In turn, the adjacent endometrial stroma responds by remodeling, rarely showing a desmoplastic change. For this reason, qualitatively assessing the character of the endometrial stroma within the endometrial compartment itself is noncontributory in distinguishing noninvasive from invasive carcinoma.
Foamy histiocytes are commonly seen in the endometrial stroma of patients with a carcinoma. Nearly one-fifth of cases contain stromal cells laden with lipid (Figure 18.14), but there is no correlation between the presence of these cells and the grade of the tumor or the survival of the patient.56 This change is simply a reactive response to tumor cells. The presence of such histiocytic cells in endometrial biopsies showing EIN should always lead to further diagnostic work-up for coexistent carcinoma.
The most commonly used histologic grading system is that of the International Federation of Gynecology and Obstetrics (FIGO) and is recommended by the World Health Organization (WHO; Table 18.3).57 This three-tiered grading system is applied to endometrioid adenocarcinomas which are classified, as well differentiated (grade 1), moderately differentiated (grade 2), or poorly differentiated (grade 3). The grading procedure is based on the amount of solid non-squamous areas within a tumor. Squamous components are excluded. Grade 1 tumors (Figure 18.15) show 5% or less solid growth and often there is no solid tumor at all. Grade 2 tumors (Figure 18.16) show solid growth in 6–50% of the tumor. More than 50% solid growth is considered grade 3 (Figure 18.17). Most endometrioid adenocarcinomas are grade 1 or 2.
|Grade 1||5% or less of non-squamous solid growth|
|Grade 2||6–50% of non-squamous solid growth|
|Grade 3||More than 50% of non-squamous solid growth|
Although the FIGO grading system of endometrioid carcinoma relies first and foremost on the architectural pattern of the glands, the histologic grade should be raised by one level beyond that determined by architecture alone in those cases exhibiting severe nuclear atypia (Table 18.458 and Figure 18.18). Usually the architectural and nuclear grades correspond, but, when at variance, the nuclear grade is often the more reliable indicator of prognosis.57 When the final grade is elevated due to severe nuclear atypia, a note should appear in the report so that the clinician is alerted that the patient falls into this special category.
|Low Grade||High Grade (Severe Atypia)|
|Little variation in shape||Marked variation in shape|
|Little variation in size||Marked variation in size (some markedly enlarged)|
|Hypochromasia||Marked hyperchromasia (may be focal)|
|No variation in staining intensity||Marked variation in staining intensity|
|Evenly distributed chromatin||Coarsely clumped chromatin|
|Nucleoli not prominent||Prominent nucleoli|
|Sparse mitoses||Frequent mitoses with abnormal forms|
Myoinvasion is diagnosed when the malignant glands have transgressed the endomyometrial junction and extend into the underlying myometrium. The normal junction, however, is not a straight line, but a rather vague and irregular border. Thus, in many cases, tumors confined to the endometrium cannot be distinguished microscopically from tumors invading the superficial myometrium. In fact, myoinvasion is overdiagnosed in routine practice in as much as 25% of cases; in contrast, failure to diagnose true myoinvasion is extremely rare.59 This is supported by recent FIGO data, which indicate almost identical 5 year survival rates for women with noninvasive tumors and those with tumors invading less than half of the myometrial thickness.59 In contrast, the distinction of inner half invasion from outer half invasion is usually straightforward, and the probability of recurrence is markedly increased for women with deeply invasive tumors. Thus, in the 2009 FIGO staging classification of endometrial carcinoma, tumors with no myometrial invasion and tumors with less than 50% invasion are combined under stage IA.60
Occasionally, endometrial carcinoma appears confined to foci of adenomyosis. Adenomyosis is really a ‘diverticulum’ of endometrium deep into the myometrium, and carcinoma can extend into these foci without invading the myometrium.59 Tumor involvement of adenomyotic foci occurs in about 25% of cases and is not associated with an adverse prognosis.61,62 When tumor cells extend outward into the myometrium, depth of invasion is measured not from the point of transgression in adenomyosis, but from the superficial endomyometrial junction.
The decision to perform pelvic and para-aortic lymph node dissection is largely based on depth of myometrial invasion (also on cell type and histologic grade). Intraoperatively, the pathologist may be requested to assess these features by frozen section.
Maximum depth of myoinvasion is measured in millimeters from the endomyometrial junction, and expressed as the percentage of the total myometrial thickness. Problems arise, however, when the boundary is distorted by the tumor itself or other lesions such as leiomyomas. Residual areas of normal endometrium (Figure 18.19) or overrun normal glands are informative landmarks, when available. Bulky exophytic tumors can be difficult to orient in a single histologic section and, in such cases, assessment of the endomyometrial boundary at the site of deepest invasion requires a combination of gross and histologic evaluation.
Figure 18.19 This tumor is entirely intraendometrial, but the irregular junction between the endometrium and myometrium imparts a false impression of myoinvasion. Endometrial stroma (arrows) surrounds the tumor.
Myoinvasion is assessed by evaluation of the topographical distribution of glands and the appearance of adjacent stroma.63 Invasion across a broad front (21% of cases) may be difficult to distinguish from an irregular endomyometrial interface unless it can be compared to the adjacent uninvolved endomyometrium. This is sometimes evident by a ‘shoulder’ formed at the juncture of a zone of myoinvasion with an area of surface involvement (Figure 18.20, arrow). Most common (67%) are irregular groups of glands with or without a stromal response (Figure 18.21). ‘Adenomyosis-like’ invasion (7% of cases) (Figure 18.22) can be distinguished from non-myoinvasive tumor extending into foci of adenomyosis (Figures 18.23 and 18.24) by its lack of accompanying endometrial stroma. Unfortunately, CD10 immunoreaction occurs focally in the cells surrounding tumor clusters in the myometrium of women without adenomyosis, and this limits utility. A distinctive pattern of microcystic, elongated or slit-like, and fragmented (MELF) invasive glands is often accompanied by individual invading cells and a loose stromal response with marked inflammation (Figures 18.25 and 18.26). The MELF pattern of invasion has some features of epithelial–mesenchymal transition, including a greatly reduced proliferative activity and acquisition of a cytokeratin (CK)7-positive immunophenotype.64 Least common of all (1%) are widely spaced individual glands lacking any stromal response, akin to adenoma malignum of the endocervix (Figures 18.27 and 18.28).
Figure 18.23 Tumor extension into adenomyosis, non-myoinvasive. Two areas of adenomyosis are seen, one with only benign glands (left), and one with extension of adenocarcinoma from the surface (right).
Figure 18.26 Individual invasive cells in MELF-type invasion are eosinophilic or squamous like, and somewhat obscured by the combined stromal and inflammatory response. (Courtesy of Dr. Marisa R. Nucci.)
Endometrial intraepithelial neoplasia (also known as atypical hyperplasia or AH) is an immediate precursor to endometrioid endometrial adenocarcinoma, which is discussed at length in Chapter 17. It is a clonal expansion of mutated glands, which can be seen as cytologically altered crowded areas of endometrial glands lacking the architectural (solid, cribriform, maze-like, myoinvasive) characteristics of adenocarcinoma (Figure 18.12). Often residual EIN is present within the endometrium at the time of presentation with carcinoma (Figures 18.29 and 18.30).
There are no established guidelines for routine endometrial cytologic screening, but exfoliated neoplastic endometrial cells are sometimes encountered in cervical Pap smears. Up to two-thirds of women with endometrial adenocarcinoma have malignant cells in their cervical cytology specimen and high-grade as well as high-stage tumors seem to be detected more frequently.65 In cervical smears, malignant endometrial cells characteristically appear as small clusters with darkly stained nuclei (Figure 18.31) or perhaps as single discrete cells that are easily overlooked. On rare occasions following the recognition of malignant glandular cells in a cervical Pap smear, the endocervical curettage and endometrial biopsy may be negative. The possibility of an ovarian lesion should be considered and excluded.
Transcervical cytology sampling devices such as the Tao brush have been designed to directly access the endometrium.66 With this device malignant cells in an endometrial brushing are freshly removed and rapidly fixed so the features are well preserved. Appearance is widely variable, from large single cells to small clusters as well as sheets (Figure 18.32). Although superior to a cervical Pap smear specimen, transcervical instrumentation with a brush introduces many of the potential morbidities of the Pipelle biopsy while yielding less tissue.
Shed tumor DNA incidentally collected by liquid cervical Pap smear may be sequenced to demonstrate mutations known to be present in a coincident endometrial carcinoma.67 Because the DNA may be shed from anywhere above it is not possible to localize a lesion, nor is the sensitivity and specificity of such a screen known when large numbers of mutations are screened in the asymptomatic patient.
Endometrioid carcinoma with squamous differentiation (one-fourth of endometrial adenocarcinomas) display focal squamous differentiation (Figures 18.33 and 18.34). Formerly, the distinction was made between tumors where the squamous component was well or poorly differentiated. The former tumors were called ‘adenoacanthomas’ and the latter ‘adenosquamous carcinomas.’ Several studies have confirmed that prognosis relates largely to the grade of the glandular component. In fact, the glandular component is much easier to grade in a reproducible fashion, and superior in predicting lymph node metastasis and 5 year survival. Therefore, it is recommended that endometrioid carcinomas with squamous epithelium are classified as endometrioid carcinoma with squamous differentiation and graded, on the basis of the glandular component, as well, moderately, or poorly differentiated (grade 1, 2, or 3, respectively). Besides, the clinical features of adenocarcinoma containing squamous epithelium and those of endometrioid adenocarcinoma are essentially the same.
While the squamous component is often reported as ‘benign’ or even as ‘squamous metaplasia,’ it is in reality both neoplastic and metaplastic, as shown by conservation of β-catenin mutations in both squamous and glandular elements. The squamous elements generally have low mitotic activity and lack ERs and thus may be described as being ‘terminally differentiated’ and ‘hormonally incompetent.’
Well-differentiated tumors (grade 1) are composed of glands and squamous nests but usually the glandular component predominates. Squamous epithelium can appear in strips or sheets (Figure 18.34), and when present as oval nests within gland lumens it is referred to as morules. Intercellular bridges and keratin deposits may be seen. The nuclei of the squamous cells are uniform, and lack prominent nucleoli (Figure 18.35). Mitotic figures are rare. In poorly differentiated tumors, the squamous cells show grade 2 or 3 nuclear atypia and are not confined to the lumens of glands. Occasionally, the squamous cells have a spindle morphology, simulating a sarcoma, and may invade the myometrium or vascular spaces. Keratin pearls are often found. Care must be taken in distinguishing between squamous differentiation and solid foci of adenocarcinoma. The undifferentiated epithelial component should be considered glandular unless intercellular bridges are seen or the cells show ample eosinophilic cytoplasm with well-defined borders (Table 18.5).68
Foreign-body-type granulomas may form in the peritoneum in response to the keratin component of endometrioid carcinomas with squamous differentiation. These lesions do not seem to affect the prognosis adversely in the absence of viable-appearing tumor cells. The granulomas probably result from transtubal spread of exfoliated necrotic tumor cells.