Endometrial cells are practically never seen in normal pregnancy. The decidual cells and particularly the large Arias-Stella cells with dark, polyploid nuclei, either derived from the endometrium or the endocervix, both discussed and illustrated in Chapter 8, may be mistaken for endometrial cancer cells in cervicovaginal material. Pregnancy does not rule out endometrial cancer. On the rarest occasion, we have observed normal pregnancy occurring in women with endometrial carcinoma documented by prior biopsy and confirmed postpartum. A similar case was described by Kowalczyk et al (1999) who also summarized the very scanty literature on this topic. Apparently, normal implantation of the ovum may occur under these circumstances. Also on record are several cases of normal pregnancies occurring in women with documented endometrial hyperplasia (Kurman et al, 1985).

As has been described in Chapters 8 and 10, the wearers of intrauterine contraceptive devices (IUDs) may shed endometrial cells at midcycle. Occasionally, such cells have a vacuolated cytoplasm and poorly preserved nuclei that may appear to be somewhat enlarged and slightly hyperchromatic and that may be mistaken for cells of an adenocarcinoma (Fig. 13-1). Sometimes, the cervicovaginal smears may also contain inflammatory cells and macrophages, creating a cytologic background, not unlike that seen in endometrial carcinoma (see below). The young age of most wearers of IUDs is usually against this latter diagnosis. Another potential source of error is the presence of endocervical “repair” caused by IUD, in which the reactive endocervical cells may be mistaken for abnormal endometrial cells (see Chapter 10; and comments below).

An important histologic finding in wearers of the IUD
is the presence of small, round foci (morulae) of squamous cells in the superficial layers of the endometrium, presumably a form of squamous metaplasia, induced by the mechanical effect of the devices. Lane et al (1974) suggested that this abnormality is transient, although evidence of reversal of this process is poor. These abnormalities are very rarely seen and should not be mistaken for an endometrioid carcinoma with squamous component or an adenoacanthoma (see below).

Iezzoni and Mills (2001) described 5 symptomatic patients in whom routine endometrial tissue samples contained aggregates of benign signet ring cells with small nuclei. The authors traced these cells to decidualized stromal cells. There is no record of such cells in cytologic samples.

Johnson and Kini (1996) described the presence of atypical endometrial cells in the presence of eosinophilic, papillary, squamous and tubal metaplasia of the endometrium. Five of seven patients were postmenopausal and three had abnormal bleeding. The nature of this observation is questionable and it cannot be excluded that some of the patients had a poorly defined neoplastic process.

Women receiving this medication occasionally bleed or spot and shed endometrium at mid-cycle (breakthrough bleeding) until the dosage is adjusted. Long-term usage of these agents may result in decidua-like changes in endometrial stroma, followed by atrophy; neither of these conditions is known to cause endometrial shedding. Abnormalities of nuclei of endocervical cells may occur in women receiving progesterone-rich contraceptive agents (see Chapter 10). Accurate clinical history is helpful in preventing errors but, in some cases, may require biopsies for clarification.

In patients receiving steroid hormones, particularly estrogens, two important cytologic changes may be observed.

For further comments on effects of steroid hormones, see Chapter 9.

A resurgence of tuberculosis in patients with immune deficiency caused by AIDS has revived interest in this disease in the developed countries. The disease is common in the developing world.

This granulomatous disease of unknown etiology may affect the endometrium (Chalvardijian, 1978; Skehan and McKenna, 1986; Elstein et al, 1994). Noncaseating granulomas, characteristic of this disorder, are observed in histologic material but, so far, have not been observed in cytologic material. For a description of cytologic presentation of pulmonary sarcoidosis, see Chapter 19.

Astin and Askin (1975) and Wenckebach and Curry (1976) described endometritis due to cytomegalovirus. The tissue showed evidence of chronic inflammation and formation of lymphocytic deposits, in addition to large cells containing the characteristic viral inclusions. Wenckebach and Curry confirmed the diagnosis by electron microscopy. Duncan et al (1989) described a case of necrotizing endometritis
associated with herpesvirus infection. Neither of these viral infections of the endometrium have been reported in cytologic writing.

A case of malacoplakia was described by Thomas et al (1978). For further comments on histologic and cytologic presentation of malacoplakia, see Chapter 22.

In the late 1960s and in the 1970s, a statistically significant increase in the rate of endometrial carcinoma has been observed in many institutions throughout the United States. Smith et al, Ziel and Finkle simultaneously pointed out in 1975 that widespread administration of conjugated and nonconjugated exogenous estrogens to alleviate menopausal symptoms and prevent osteoporosis was statistically associated with this increase. Mack et al (1976) calculated the risk ratio for endometrial carcinoma in estrogen users when compared with nonusers at 8.0 times, and for conjugated estrogens at 5.6 times; these investigators also demonstrated a dose-related effect on endometrial carcinoma. In a study by a writers group for the PEPI Trial (1996), the administration of unopposed estrogens was shown to cause endometrial hyperplasia and occasional adenocarcinoma. The effect could be prevented by the administration of progesterone. Exogenous estrogens have been shown to be associated with endometrial carcinoma, even in the absence of ovarian function, for example in ovarian agenesis (Gray et al, 1970;
Cutler et al, 1972) or in Sheehan’s syndrome (Reid and Shirley, 1974).

Although the evidence is substantial that estrogens may cause endometrial carcinoma, it has been shown that such lesions observed in estrogen-treated patients are usually fully curable, low-grade and low-stage cancers (Robboy et al, 1982). Horwitz and Feinstein (1978) addressed this issue and reported on the status of peripheral endometrium in a case control study of 233 postmenopausal women, 112 of whom had endometrial carcinoma. Peripheral, simple endometrial hyperplasia was more commonly observed with grade 1 cancer among estrogen users than in cancer of higher grades among nonusers of estrogen. The authors concluded that “it was likely that many otherwise asymptomatic tumors might have remained undetected except for the manifestations of the estrogen-related comorbid condition” (hyperplasia). The observation was repeated by Horwitz et al (1981) who proposed that the effect of estrogens on endometrium is indirect: the drugs cause endometrial hyperplasia and, hence, uterine bleeding that leads to curettages and results in incidental discovery of small foci of early endometrial cancer. In fact, in our own study of occult endometrial carcinomas, estrogen treatment has not been shown to be a risk factor except for women with lower than average weight. It was hypothesized that this observation may perhaps be explained by the inability of this group of women to store the estrogens in their subcutaneous fat, resulting in more direct action on the endometrium (Koss et al, 1984; see below). The use of either estrogen therapy or estrogens combined with progesterone, also increases the risk of breast cancer (Colditz et al, 1995; Schairer et al, 2000) (see Chap. 29).

Tamoxifen is a steroid agent best characterized as an estrogen agonist or estrogen-receptor modulator, which blocks estrogen receptors in a variety of tissues and is now extensively used for prevention and treatment of breast cancer (summary in Osborne, 1998). The drug has several side effects affecting the female genital tract and, specifically, the endometrium.

Measuring the thickness of the endometrium by ultrasound is a favored method of follow-up of patients receiving tamoxifen and other hormones (Achiron et al, 1995; Levine et al, 1995; Hann et al, 1997). It has been suggested that endometrial thickness of 8 mm or more should trigger an endometrial investigation by biopsy or curettage. Langer et al (1997), using the thickness of 5 mm as a trigger for endometrial biopsies in women receiving estrogen replacement therapy, noted that at this level of endometrial thickness, the technique has a very poor positive predictive value but a high negative predictive value for important endometrial disorders.

Endometrial carcinoma has been observed in approximately 0.05% of women treated with a variety of hormones for carcinoma of the breast (Hoover et al, 1976). Hormonal contraceptive agents usually cause endometrial atrophy. It is not known, at this time, whether these agents may also contribute to the genesis of endometrial cancer, although a few such cases have been recorded (Silverberg and Makowski, 1975).

Malignant tumors of the endometrium (carcinomas and occasionally mesodermal mixed tumors) have been observed in patients who received a curative dose of radiation for invasive carcinoma of the uterine cervix (Fehr and Prem, 1974).

Carcinoma of the endometrium has been traditionally thought to be associated with diabetes, obesity, hypertension, a past history of abnormal menses, and late menopause (Wynder et al, 1966; Elwood et al, 1977). Our own epidemiologic studies of asymptomatic women with occult carcinoma failed to confirm these observations (Koss et al, 1984) but this cohort may have differed from symptomatic women who have been the common target of such studies. The only statistically significant factor in the Koss study was delayed onset of menopause (see Table 13-8). The full extent of the clinical epidemiology of the disease is deserving of further studies comparing symptomatic with asymptomatic patients.

As the name indicates, this malignant tumor is characterized by a disorderly proliferation of the endometrial glands resulting in a grotesque image of the endometrium. These tumors are usually primary in the endometrium but may also develop in endometrial polyps and in foci of endometriosis that may be located in a variety of primary sites, including the ovary and even the regional lymph nodes (Koss, 1963). The cancerous glands vary in size and configuration, are often crowded, and adjacent to each other without intervening endometrial stroma. Papillary projections into the lumen of the glands is not uncommon (Fig. 13-6A). The cancerous glands are lined by cells that are larger than normal, usually cuboidal but sometimes columnar (tall-cell carcinoma) in configuration. The nuclei of these cells vary from simple enlargement and slight hyperchromasia in low grade tumors to markedly enlarged, sometimes hyperchromatic nuclei in high grade tumors. A characteristic feature of cells of endometrioid carcinoma is the presence of clearly visible nucleoli. The number and size of the nucleoli also vary with tumor type, with one or two small nucleoli present in well differentiated tumors, when compared with up to four larger nucleoli in
high grade tumors (Long et al, 1958). The degree of nuclear abnormalities is the basis for nuclear grading that is thought to be of prognostic value. The frequency of mitotic figures varies.

The stroma separating the cancerous glands may occasionally show rather remarkable changes in the form of clusters of very large macrophages, first described by Dubs in 1923 (Fig. 13-6B,C). Rarely, concentric, often calcified protein secretions (psammoma bodies) may be formed by some of these tumors (Parkash and Carcangiu, 1997).

The degree of architectural differentiation of endometrial cancer may vary considerably and is of prognostic significance. Some tumors present only a slight deviation from the normal endometrial pattern (grade I carcinomas, sometimes referred to as adenoma malignum); at the other extreme, there is a grade III carcinoma, presenting as a nearly solid growth of cancer cells in sheets with only an occasional attempt at gland formation. Most of the endometrial cancers fall somewhere between the two extremes and are graded II.

Villoglandular carcinoma is an uncommon variant of endometrioid carcinoma, characterized by formation of slender papillary fronds on the surface of the tumor (see Fig. 13-17B). The tumor cells are similar to those of a well-differentiated endometrioid cancer.

Hyperplasia, which occurs mainly in premenopausal women, is caused by a hormonal imbalance in favor of estrogens and may result from disturbances of ovulation, such as the Stein-Leventhal syndrome, in which the estrogenic phase is not followed by a progesterone phase. Hormone-producing ovarian tumors, such as theca or granulosa cell tumors, may also produce endometrial hyperplasia. Simple hyperplasia may also be associated with leiomyomas (Deligdisch and Loewenthal, 1970). In postmenopausal women, administration of unopposed exogenous estrogens is a known cause of hyperplasia (the Writing Group for the PEPI Trial, 1996).

The essential clinical feature of endometrial hyperplasia, regardless of type, is a period of amenorrhea followed by uterine bleeding that may be excessive in amount (menorrhagia) or irregular (metrorrhagia). In some patients, the bleeding may be fairly cyclic in character, whereas in others it is very irregularly spaced.

Although current textbooks and atlases of gynecologic pathology (e.g., Silverberg and Kurman, 1992) offer a variety of terms to describe various forms of endometrial hyperplasia, according to the configuration of the glands and the level of abnormalities in the epithelial lining, a simple classification is used here. Three forms of endometrial hyperplasia can be distinguished:

This classification disregards the configuration of the glands, but experience has shown that in most hyperplasias with nuclear abnormalities, the endometrial glands are abnormally configured.

Evidence for progression of atypical hyperplasia to carcinoma of the endometrium is relatively poor because most of these lesions cause symptoms and are treated, at least by curettage and hormonal manipulation, but not infrequently by hysterectomy. At the time of this writing (2004), few patients with these abnormalities are left untreated. The evidence of progression is based on older studies. A frequently cited study is that by Gusberg and Kaplan (1963) in which a group of patients with “adenomatous hyperplasia” were prospectively followed; several of them (about 10%) developed endometrial cancer. Anecdotal evidence of progression of endometrial hyperplasia to carcinoma was also provided by Foster and Montgomery (1965). In a retrospective study of 170 patients, Kurman et al (1985) classified hyperplasias according to the degree of nuclear abnormality. Carcinoma developed in only 2 of 122 patients without significant cytologic atypia and in 11 of 48 women (23%) with “atypical” glands. The “progression” also depended on the complexity of the glandular pattern with “simple” lesions less likely to progress than “complex” lesions. Many of the lesions illustrated in the Kurman paper as “atypical complex hyperplasia” could be classified by other observers as a well-differentiated endometrioid carcinoma. Further, even though none of these patients were initially treated by hysterectomy, most received some form of treatment such as hormonal manipulation, curettage, or both. Hence, the rate of development of invasive cancer in untreated patients could be much higher.

However, there is substantial evidence suggesting that endometrial hyperplasia is not a mandatory stage in the development of endometrioid carcinoma (or other types of endometrial cancer) that may also develop de novo, particularly in postmenopausal women. The search for occult endometrial cancer (Koss et al, 1981, 1984) strongly suggested this possibility (see below). In an older contribution, Greene et al (1959) observed peripheral hyperplasia in only 10 of 120 cases of endometrial carcinomas. These authors expressed the view that, “some (and probably the minority) of endometrial carcinomas are preceded by or possibly induced in or developed from areas of endometrial hyperplasia.” These observations are particularly valuable because they were published in 1959, before widespread use of hormones obscured endometrial pathology.

Based on a case control study, cited above, Horwitz and Feinstein (1978) proposed that “endometrial hyperplasia and carcinoma may represent separate expressions of endometrial pathology, which may occur side by side, but do not necessarily follow each other. It is further suggested that the so-called atypical hyperplasia, a lesion most likely to ‘progress’ to invasive carcinoma, does in fact represent a low-grade endometrial carcinoma. The two lesions can only be separated from each other by a series of intricate and generally nonreproducible morphologic criteria.” Still, endometrial hyperplasia of whatever type must be construed as a warning sign that an endometrium is not cycling or not cycling properly and, therefore, is susceptible to neoplastic events. With luck and skill, the cytologic diagnosis of occult endometrial hyperplasia is sometimes possible either in cervicovaginal smears or in direct endometrial samples.

It has been reported that hormonal manipulation of atypical hyperplasia with progesterone and related drugs may occasionally restore the cycling endometrial pattern (the Writing Group for the PEPI Trial, 1996). Yet, in our experience, these drugs are rarely, if ever, curative of the disease. There is little doubt, however, that the presence of these abnormalities puts the untreated patient at risk for the development of endometrial carcinoma, although the degree of risk cannot be estimated in any individual patient.

Recent studies of this group of tumors traced their origin to malignant changes in the surface endometrium and adjacent glands that has been labeled endometrial intraepithelial carcinoma (Fig.13-8A,B), and which is characterized by expression of mutated protein p53 (Sherman et al, 1992, 1995, 2000). On the surface, the lesion is composed of a single or double layer of large cancer cells with large nuclei and nucleoli, sometimes in a palisade arrangement. Adjacent glands show similar changes. Mitotic activity is abundant. The proponents of EIC avoided the use of the term endometrial carcinoma in situ, an abnormality of endometrial glands, described by Hertig et al (1949) as a precursor lesion of endometrioid carcinoma, discussed above. It has been proposed that the genesis of serous endometrial carcinoma follows a different pathway from endometrioid carcinoma and is unrelated to endometrial hyperplasia (Sherman et al, 1992, 1995, 2000).

DNA ploidy measurements have been shown to be of prognostic value in endometrial carcinoma (Atkin, 1984; Iverson and Laerum, 1985; Iverson and Utaaker, 1988; and others). It has been documented that tumors with approximately diploid DNA content have a better prognosis than aneuploid tumors. In general, well-differentiated endometrioid carcinomas have a diploid DNA content but occasionally higher grade tumors are also in the diploid range of measurements.

Baak et al (1988) reported that combined architectural and nuclear morphometric features in tissue sections were a more accurate predictor of behavior of endometrial hyperplasia than nuclear features alone. This elaborate study requiring costly instrumentation and dedicated personnel is not likely to be of practical value in the laboratory.

These studies have documented the presence of estrogen and progesterone receptors in most endometrial carcinomas and in some metastases (Ehrlich et al, 1981

Stay updated, free articles. Join our Telegram channel

Join