Endometritis, Metaplasias, Polyps, and Miscellaneous Changes
Infections of the gynecologic tract are relatively common, and cause an enormous health impact. It is estimated that there are 19 million cases of sexually transmitted diseases (STDs) in the United States, with a total healthcare cost of roughly $16.4 billion.1 These infections can be both symptomatic and asymptomatic, and it is believed that up to 24,000 cases of infertility in the United States are related to STDs. Symptomatic infections are most commonly diagnosed based on the clinical examination and basic laboratory studies, and, as such, a tissue biopsy is generally not needed. Approximately 20% of cases of STD can result in pelvic inflammatory disease (PID), which is inflammation of the uterus, fallopian tubes, or ovaries. An unknown fraction of STDs lead to asymptomatic infections, or to signs that are not recognized to be related to infection, such as dysfunctional uterine bleeding. Chronic endometritis has been shown to interfere with local expression of estrogen and progesterone receptors in the endometrium, which may also contribute to infertility.2 It is especially in the evaluation of women with bleeding that a diagnosis of infection is critical. STDs can be readily treated, and proper treatment can prevent later complications. In a minority of cases the pattern of inflammation in the endometrium is indicative of a specific organism (Table 16.1). In the majority of cases the pattern of inflammation is nonspecific.
Endometrial inflammation is recognized, as in other organs, primarily by the presence of inflammatory cells. However, as the endometrium may normally contain inflammatory cells, the diagnosis of endometritis is often not straightforward. Lymphocytes, even when focally collected around blood vessels, are not necessarily indicative of endometritis. As plasma cells are seldom found in a normal endometrium, their presence is indicative of chronic endometritis. Neutrophil polymorphonuclear leukocytes may also be found in the normal cycle, especially in appreciable numbers once menstruation is well established. Granulated lymphocytes, or CD56+ uterine natural killer cells, present in the stroma during the second half of the cycle, and especially in predecidua, are easily mistaken for neutrophils.3,4 Macrophages are always present in the normal endometrium, but in their usual appearance resemble endometrial stromal cells, rendering separation difficult.
Patients with endometritis often present with the clinical features of PID. They will have a history of fever and lower abdominal pain, with leukocytosis and bilateral adnexal or cervical motion tenderness on bimanual pelvic examination. If the infecting organism is less virulent or if the disease is early in its natural history, the symptoms, such as menstrual abnormalities, pelvic pain, or infertility, may not initially indicate the clinical diagnosis of endometritis. Although most cases of acute salpingitis are accompanied by endometritis, it is not known how often endometritis occurs in isolation, without involving the fallopian tubes. The diagnosis of endometritis is therefore often a histologic one that, particularly in the case of nonspecific endometritis, is not associated with a specific clinical presentation or cause.
The histologic diagnosis of acute endometritis is made when significant numbers of neutrophils are present in the endometrial stroma at a time other than menstruation (Figure 16.1). Isolated or rare neutrophils are insufficient for a diagnosis of acute endometritis. The usual pattern is that of patchy distribution, with or without microabscesses. As with all forms of endometritis, it is not only the presence of a particular inflammatory cell that is important, but also the reaction pattern of the endometrium. A common appearance in the patient of reproductive age is glandular dyssynchrony in response to circulating sex hormones. In the case of acute endometritis it should be accompanied by other features of inflammation and repair, including edema, hemorrhage, venule ectasia, microabscesses, or pus-filled glands. In the case of menstrual endometrium other features of menstruation are typically seen, including secretory vacuoles in the glandular epithelium and predecidual change of the stroma with diffuse stromal breakdown. It may not always be possible to distinguish between acute endometritis and menstrual endometrium.
The acute inflammation seen in acute endometritis is often confined to the upper layers of the functionalis, leading to shedding of the inflamed tissue during menstruation. Assuming that the organism that precipitated the endometritis is no longer present, a new functionalis can develop from a healthy basalis layer. If the infection is not treated then the inflammation will proceed to chronic endometritis.
If plasma cells are present, often seen together with aggregates of lymphocytes and/or germinal centers, then a diagnosis of chronic endometritis is made (Figure 16.2). A not uncommon problem occurs when only a rare plasma cell is identified or it is a minor component of the infiltrate. If on a low-power magnification the endometrium looks ‘dirty’ with numerous lymphocytes (Figure 16.3), a careful search under higher magnification often shows scattered plasma cells that justify a diagnosis of chronic endometritis. Areas worthy of particular attention include glands engorged with inflammatory cells. As plasma cells may be a normal component within the stroma of the endocervix and lower uterine segment, care must be taken to ensure that only the functionalis is assessed.
Glands may be involved in the inflammatory process in two ways. There may be an infiltrate of inflammatory cells, predominantly neutrophils, affecting the epithelium of the glands and the lumina may contain pus. In addition, as the process becomes more advanced, the glandular response to hormonal stimulation becomes impaired, the glands appearing inactive, with neither proliferation nor secretion. Once an endometrium shows signs of endometritis, precise dating is not possible. It is probably not even reliable to categorize the endometrium as either proliferative or secretory. Dating is unnecessary in a patient who has significant endometritis. The histologic features of endometritis are summarized in Table 16.2.
Several studies have investigated the use of biomarkers (both histochemical and immunohistochemical methods) to aid in the identification of plasma cells. Because rare plasma cells can be identified in the background of women without the other diagnostic features of chronic endometritis described earlier, the routine use of these methods is not currently recommended. Instead, a careful histologic examination by routine H&E stain using the previous criteria is recommended.
The most difficult aspect in the diagnosis of endometritis is simply recognizing it and, when necessary, initiating a search for plasma cells and other histologic features consistent with the diagnosis. The differential diagnosis includes normal menstrual endometrium, where (nonplasma cell) inflammatory cells are widespread, and plasma-cell-bearing fragments of cervix or lower uterine segment, which are mistaken for endometrial tissue.
Many upper genital tract infections result from the sexually transmitted organisms Chlamydia trachomatis and Neisseria gonorrhoeae.5,6 Infection by C. trachomatis exhibits supranuclear, intracytoplasmic vacuoles containing inclusions, but these are not readily apparent by microscopic examination. Immunohistologic or molecular methods are usually required for confirmation of chlamydial infection. Polymerase chain reaction identifies C. trachomatis in one-fourth of plasma cell endometritis specimens, and these can also be cultured.7,8 Endometritis due to either Chlamydia or Neisseria has large numbers of plasma cells in the endometrial stroma, neutrophils in the endometrial surface epithelium, and lymphoid aggregates with transformed lymphocytes,9 thus providing clues to the diagnosis. Intraluminal neutrophils and subepithelial hemorrhages are also usually present, but these findings are less specific. Histopathology alone can suggest, but not reliably distinguish between, these organisms. Prominent lymphoid follicles with transformed lymphocytes points more to chlamydial infection while extensive stromal fragmentation favors gonococcal infection.5
Infection with the mycoplasma Ureaplasma urealyticum produces subtle but distinctive changes that may be easily missed. Mycoplasma is the second most common etiologic agent after common bacteria.10 The changes are focal, composed of lymphocytic clusters with histiocytes and occasional neutrophils. Plasma cells are not prominent. These focal collections often lie beneath the surface epithelium, around glands, and surrounding spiral arterioles. In many of these patients, U. urealyticum also has been cultured from the uterine cervix, suggesting an ascending infection. Mycoplasma genitalium is a common pathogen associated with acute endometritis.11
Cytomegalovirus (CMV) infection of the endometrium has been reported rarely and may be associated with pregnancy and immunosuppression as well as being found in women with no known underlying disorder. The affected epithelial and sometimes endothelial cells are grossly enlarged and each contains a usually single large, basophilic intranuclear inclusion that is surrounded by a narrow clear space separating it from the nuclear membrane (Figure 16.4).
Like CMV infection, herpes simplex virus (HSV) infection of the endometrium is rare. It may occur as an ascending process associated with cervical infection or result from disseminated virus in immunosuppressed patients. It may cause postpartum endometritis in cases where infants have died from disseminated HSV infection.5 Infection of the endometrium produces changes similar to those associated with HSV elsewhere, namely extensive acute inflammation with necrosis. Cowdry type A inclusions and, occasionally, multinucleated giant cells with ground-glass nuclei may be identified in glandular epithelium or in the stroma.
Although tuberculosis has become infrequent in the United States, it remains an endemic disease in some parts of the world12 and more may be seen in the future as a result of the emergence of drug-resistant strains of Mycobacterium tuberculosis and the immunosuppression of AIDS. Tuberculosis of the female genital tract is virtually always secondary to disease of the lungs or gastrointestinal tract, which often occurred many years previously. About three-fourths of women with tuberculosis of the genital tract have endometrial involvement. The fallopian tubes are almost always involved and the endometrium is affected secondarily by passage of the bacteria down the tube into the uterine cavity.
The disease may present without gynecologic symptoms or it may be associated with infertility, irregular bleeding, or amenorrhea. The principal histologic feature is the epithelioid cell granuloma. These may be sparse, so that a thorough search of the material is required if tuberculosis is suspected clinically. The granulomas are usually small and central caseation is unusual, especially in women who have normal, or near normal, menstrual cycles. Caseation is found only when there is amenorrhea or after menopause, for it is only when the endometrium no longer periodically sheds that there is sufficient time for the granuloma to develop de novo and then caseate. The disease process affects all layers of the endometrium.
The granuloma of tuberculous endometritis contains a central collection of epithelioid cells with both Langhans and foreign body-type giant cells (Figure 16.5). There is usually a peripheral collar of lymphocytes. Because caseation is absent, the granulomas more closely resemble those of sarcoidosis, a feature reinforced because they are well circumscribed. Uterine sarcoidosis rarely presents without proven sarcoidosis elsewhere in the body.
Culture of the endometrial tissue best confirms the diagnosis. In a classic study that has not been repeated, acid-fast bacilli were demonstrated in less than 2% of 1134 women with culture-proven organisms.13 More recently molecular methods have been developed that allow the specific detection, and strain classification, of M. tuberculosis in formalin-fixed paraffin-embedded materials at a level of sensitivity exceeding that of special stains.14–16
Infection by Actinomyces israeilii has long been associated with intrauterine contraceptive device (IUD) usage.17 This organism is a slow-growing, filamentous, Gram-positive anaerobe that is not normally found in the lower genital tract (Figure 16.6). When found in cervical smears or curettings, half the women lack symptoms. Microscopic examination of the curettings shows densely matted filaments of the Gram-positive organisms (Figure 16.7). Gomori methenamine silver stain is positive in Actinomyces, which is a useful means of identifying noninfectious pseudoactinomycotic radiate granules (negatively staining) that are even more common than actual organisms in women with an IUD.18,19 There is usually some degree of associated endometritis, with a dense infiltrate of neutrophils. The infection may progress to pelvic actinomycosis. The risk of infection by A. israeli relates less to the type of IUD in use than the length of time it has been in place in the uterine cavity; 85% of cases occur when the IUD has been worn for 3 or more years.
The finding of granulomas in an endometrial biopsy can be caused by a variety of diseases. In a recent review granulomas were identified in 0.15% of endometrial biopsies.20 Focal granulomas, that is the finding of a single granulomatous focus, were significantly associated with a prior history of cesarean section or dilatation and curettage, and none of the women had any signs of a systemic disorder or infection upon follow-up. In terms of infection, granulomas can be caused by tuberculosis (as described previously), atypical mycobacteria, endemic mycosis, actinomycosis, and parasites.20,21 Noninfectious causes can include a reaction to foreign material, Crohn disease, sarcoidosis, or lymphoma. Granulomas can also be seen postsurgically. Sarcoidosis is associated with well-formed, clearly demarcated epithelioid cell granulomas lacking central necrosis. It occurs usually, but not always, in women with known systemic disease. However, as necrosis is unusual in tuberculous endometritis, endometrial sarcoid should only be diagnosed if the disease is present elsewhere in the body. An effort should be made to identify a treatable, infective cause if it is present. Fungi may be identified using periodic acid–Schiff (PAS) or Grocott stains, but it is questionable whether a time-consuming search for acid-fast bacilli is worth undertaking because of the very low chance of identifying the organism, even in culture-proven cases.13 The identification of pigment or birefringence under polarized light points to a foreign body reaction. A clinical history may also be useful in coming to a conclusion.
On rare occasions, the endometrium discloses a massive infiltrate of histiocytes that contain abundant lipid, together with multinucleated giant cells, hemosiderin, cholesterol clefts, plasma cells, lymphocytes, and occasional neutrophils (Figure 16.8). This may occur in association with pyometra, due to postmenopausal cervical stenosis, or in association with an endometrial glandular neoplasm. The terms ‘pseudoxanthoma’ and ‘xanthogranuloma’ have also been used but the term ‘histiocytic endometritis’ is preferred. The condition bears no relation to the focal clusters of foamy macrophages sometimes found in the stroma of endometrial carcinomas. Appearances similar to histiocytic endometritis may be seen in the fallopian tube. Malakoplakia is a rare variant of histiocytic endometritis. The appearances are generally similar to those already described but in addition Michaelis–Gutmann bodies are present. These are small, round, laminated calcospherules that may be found both in the cytoplasm of the histiocytes and also extracellularly.
Endometritis occurs in 2–5% of women following vaginal delivery, 20–55% of women following cesarean section, and 1–8% of women having termination of pregnancy in a hospital setting.22 The risk of endometritis is even higher in women with retained products of conception. Plasma cell infiltrates, which are not a normal postpartum response, are characteristic of postpartum chronic endometritis (Figure 16.9). Microbiologic examination shows that the infection is caused by two or more microorganisms, the most frequently cultured being U. urealyticum and Gardnerella vaginalis. The latter is most commonly associated with bacterial vaginosis and clinical evidence of bacterial vaginosis is associated with a sixfold increase in the risk of post-cesarean endometritis.23 Furthermore, recent studies have shown that bacterial vaginosis is associated with a higher risk of endometritis9 and upper genital tract infection in the nonpregnant state.24
The histopathologic features of the endometrium in these forms of endometritis are usually those of an acute infection, with prominence of neutrophils and tissue necrosis. Organisms are often identified by Gram stain.
Inflammation of the endometrium is most dramatic when pyometra, pus filling the uterine cavity, develops (Figure 16.10). This is often due to a blocked cervical canal in conditions such as cervical carcinoma or, more frequently, postmenopausal stenosis. Introduction of foreign material through fistulous tracts is another common cause. As the term implies, the cavity of the distended uterus is filled with pus and the endometrium usually shows the features of severe acute and chronic inflammation (Figure 16.11). Squamous metaplasia may develop in the surface epithelium and glands, and at its most extreme is known as ichthyosis uteri.25 Squamous cell carcinoma, rarely, may also occur.
Endometrial metaplasia is a change in cellular differentiation to a type that is not present in the normal endometrium. Terms such as metaplasia, differentiation, and ‘change’ are used, often interchangeably, to reflect the wide variety of cell types that can be seen in the endometrium. These can include alterations in cytoplasm, nuclei, and even architecture. Metaplasia most commonly involves the epithelial compartment but can also involve the stromal compartment. Because altered differentiation may be seen in benign, premalignant, or malignant endometria, a simple diagnosis of endometrial metaplasia is rarely sufficient to fully convey the lesion’s clinical significance. Endometrial metaplasias are best considered a spectrum of altered differentiation states in which the pathologist should attempt to determine the underlying mechanism, and use terminology that clearly communicates clinical significance to the referring physician.
The epithelium of the upper female genital tract, whether endocervical, endometrial, or tubal, is all of paramesonephric (müllerian) origin, but each area shows its own characteristic cell type: mucus secreting in the cervix, ciliated in the fallopian tube, and columnar with secretory potential in the endometrium. The most common forms of endometrial metaplasia are those in which endometrial glands assume a morphology resembling that seen elsewhere along the müllerian duct, such as serous, mucinous, or squamous. Müllerian-derived structures thus maintain an element of plasticity in which any of a number of pathogenetic mechanisms may incite differentiation inappropriate for the particular anatomic location. The resultant metaplasia is accompanied by the appearance of differentiated structures and protein expression patterns usually associated with a specific nonendometrial cell type, such as intracellular mucin in mucinous metaplasia, cilia in tubal metaplasia, or keratin in squamous metaplasia. In practice, use of the term ‘metaplasia’ is not constrained to a change in differentiation, having been extended as a descriptive label for some nonspecific degenerative changes.
The term metaplasia applies to a broad range of cytologic appearances caused by degenerative, hormonal, or neoplastic processes (Table 16.3). The overall low-magnification topography of metaplastic changes within the endometrial compartment, combined with the local histologic context, are extremely useful in ascertaining a likely cause.26,27 Degenerative changes can be localized to areas of inflammation, stromal breakdown, or recent surface repair. Hormonally induced changes tend to be diffusely and randomly scattered throughout the upper endometrial compartment (functionalis), whereas premalignant and malignant processes are monoclonal expansions that begin as localizing lesions offset by the normal background. The genetic underpinnings of müllerian metaplasia are just now emerging. Genetic mediators of müllerian differentiation include members of the homeobox gene superfamily, HOX, where coordinate expression of HOX genes determines differentiation state.28 Altered HOX gene expression, through hormonal29,30 or epigenetic mechanisms,31,32 is capable of re-determining differentiation state. For example, HOXA11 expression can be induced by progestins and in müllerian epithelium is associated with acquisition of mucinous differentiation.33
Many instances of degenerative metaplasia will show an underlying cause, such as physical irritation due to an intrauterine mass lesion (submucous fibroid, polyp), inflammatory conditions (endometritis), or breakdown of endometrial stroma. Most are not true metaplasias demonstrating features of specialized altered differentiation, but rather histologic mimics where use of the term descriptively augments a more specific diagnosis of endometritis or breakdown. An example is ‘papillary syncytial metaplasia,’ which exhibits papillary fronds of epithelium that have lost nuclear polarity and can be seen in association with breakdown of supporting stroma, irritation of the uterine lining epithelium, or as part of a reparative process.
Hormonally induced metaplasias are common and not always abnormal, particularly in the case of estrogen-induced tubal change. Estrogens may induce tubal metaplasia of endometrial glands in a random interspersed pattern throughout the entire endometrial compartment at a density depending on dose and duration of exposure. Estrogen-induced tubal metaplasia is most frequently seen in women around the time of menopause and just afterward, and is associated with abnormal uterine bleeding.
A particular pitfall in dealing with endometrial neoplasms with altered (metaplastic) differentiation is the incorrect assumption that metaplasia and neoplasia are mutually exclusive processes. Squamous, mucinous, tubal, or secretory differentiation may be acquired features intrinsic to a premalignant (endometrial intraepithelial neoplasia; EIN) or malignant (adenocarcinoma) neoplastic clone.34–37 In these instances, altered differentiation is best noted descriptively appended to a primary diagnosis of EIN or adenocarcinoma (Table 16.4). Individual examples of EIN or adenocarcinoma can demonstrate quite unstable differentiation states in which admixed cytologies line individual glands or populate different regions of the neoplastic field.34,38,39 In the case of metaplastic premalignant lesions, all EIN criteria must be met (Chapter 17),40 with the observed change in cellular differentiation sometimes contributing to the altered cytology requirement. Overtly malignant processes, especially mucinous and secretory adenocarcinomas, may have unusually bland nuclei as part of their altered differentiation state.
Tubal differentiation is the most commonly encountered form of endometrial metaplasia, accounting for 60% of examples of metaplasia in the endometrium. Cilia are the most distinctive manifestation, but changes evident even at low magnification are increased cytoplasmic volume and eosinophilic staining. Interspersed between cuboidal ciliated cells are occasional round secretory cells in a pattern resembling that of the normal fallopian tube. The density of cilia varies greatly, and even a cilia-poor epithelium that demonstrates characteristic cytoplasmic properties of admixed round and columnar cells may be considered serous like (‘seroid’) or tubal in character (Figure 16.12). Some cilia-depleted tubal metaplasias display such prominent micropapillary or eosinophilic features that they may be considered mixed-type metaplasias.