Uterine mesenchymal tumors are neoplasms derived from or differentiating toward mesodermally derived tissues. Differentiation is typically toward normal constituents of the uterine corpus—endometrial stroma and myometrial smooth muscle. However, differentiation toward heterologous tissues such as skeletal muscle, cartilage, or bone may also be seen. Recent updates on diagnosis and classification in uterine mesenchymal tumors include reclassification of endometrial stromal tumors to include a high-grade endometrial stromal sarcoma and the improved recognition of hereditary leiomyomatosis renal cell carcinoma syndrome–associated smooth muscle tumors.
Keywordsadenomyoma, adenosarcoma, endometrial stromal neoplasia, leiomyoma, leiomyosarcoma, smooth muscle tumor
Uterine mesenchymal tumors are neoplasms derived from or differentiating toward mesodermally derived tissues. Differentiation is typically toward normal constituents of the uterine corpus—endometrial stromal and myometrial smooth muscle cells. However, differentiation toward heterologous tissues—that is, mesenchymal tissue not normally present in the uterus (e.g., striated muscle, cartilage, or bone)—also may be seen.
Mixed epithelial and mesenchymal uterine tumors are composed of benign or malignant mesenchymal elements combined with benign or malignant epithelium. The classification of such mixed tumors depends on the morphologic assessment of both components.
Identifying Patients at Risk for Mesenchymal Neoplasia
The degree of risk for developing mesenchymal tumors is not uniform for all types of uterine mesenchymal neoplasia, and assessment of risk is further hindered by the rarity of most of these tumors. In fact, uterine sarcomas account for only 4% to 9% of uterine malignancies. Moreover, it is estimated that there are only 0.01 to 0.02 cases of uterine sarcoma/1000 women.
Despite their rarity, there have been recent insights into potential risk factors for certain subtypes of uterine mesenchymal neoplasia. Tamoxifen, an antiestrogenic drug widely prescribed for women at risk for breast cancer or used to treat women with breast cancer, is known to increase the risk for uterine carcinomas, presumably because of its agonistic effect at this site. The risk of uterine and endometrial cancer, of which most represent stage I adenocarcinoma, is almost doubled with tamoxifen use; however, this elevated risk appears to be only in women older than 50 years. Use of this drug may also increase a woman’s risk for certain subtypes of uterine mesenchymal neoplasia, particularly adenosarcoma and carcinosarcoma. In some cases, there is a long latency period, so that long-term follow-up of all patients on tamoxifen is warranted. Follow-up ultrasound evaluation in patients with clinical symptoms may be useful in stratifying those women who need to undergo sampling of their endometrium. An abnormal ultrasound finding (i.e., thickened endometrial stripe, polyp, or uterine mass) in this group should prompt sampling. The risk for a uterine malignancy (both epithelial and mesenchymal types) associated with raloxifene, a drug currently prescribed to prevent osteoporosis and being tested in women with and at risk for breast cancer, is not clear.
In contrast to malignant mesenchymal tumors, the frequency of benign uterine leiomyomata is extremely high, with estimates of up to 85% of women of reproductive age and older. Consequently, leiomyomata are the most common uterine tumor, and this type is one of the most common tumors to affect women. Although their high frequency may complicate their epidemiologic analysis, more is known about risk factors for the development of uterine leiomyomata than for any other uterine mesenchymal tumor. As will be discussed in greater detail in subsequent sections, there is substantial evidence that genetic factors contribute to the development of leiomyomata. These factors are recognized at the familial and population levels. In particular, the risk of benign and possibly malignant uterine smooth muscle tumors are two- to threefold higher in black women.
Nongenetic (i.e., environmental and genetic by environmental) factors also have been implicated in the development or growth of uterine leiomyomata. The most frequently identified factor, smoking, appears to reduce the clinical frequency of leiomyomata in some but not all studies. Increased parity also is associated with a reduced risk of leiomyomata. Oral contraceptive use has been linked to leiomyomata in most studies. In particular, older age at menarche and early oral contraceptive use may be protective. In contrast, obesity is associated with larger or more symptomatic tumors. Environmental (perineal) exposure to talc increases the risk of leiomyomata by a factor of 2. Finally, increased dietary consumption of soy products, alcohol, and red meats and decreased consumption of green vegetables or foods rich in beta-carotenes have been implicated. Individually, these apparent risk factors for leiomyoma are a confusing hodge-podge. A potentially unifying explanation emphasizes the unopposed estrogen hypothesis invoked in endometrial carcinogenesis. Diet would tie into this hypothesis as a source of environmental estrogens. Another potential explanation is that gravidity alters the propensity for leiomyomata by structural, biochemical, and epigenetic remodeling of the uterus.
General Clinical Features
In general, most uterine mesenchymal tumors are intramural or intracavitary lesions. Clinical symptoms are related to their size; disruption of the uterine lining, with pelvic pain, pressure, and dysfunctional uterine bleeding, is the most common clinical manifestation. Although size is not a guarantee of a tumor being malignant, in general, most mesenchymal malignancies of the uterine corpus tend to be large (>10 cm) and may have already spread to involve contiguous structures at the time of presentation. Radiographic studies, particularly ultrasound and magnetic resonance imaging (MRI), may be useful in clinically assessing whether a mass is more likely to be benign or malignant. In addition to spread beyond the uterus, radiographic features suggestive of tumor necrosis may be present. Caveats include central infarction and degeneration of benign tumors, such as those that can be seen in large leiomyomata.
Endometrial Stromal Tumors
Definition and Classification
Currently, the World Health Organization (WHO) recognizes four main categories of endometrial stromal tumors: (1) endometrial stromal nodule; (2) endometrial stromal sarcoma (ESS), low grade; (3) ESS, high grade; and (4) undifferentiated uterine sarcoma. This categorization represents a significant change from the prior WHO categorization (2002), which had removed high-grade ESS from the lexicon and merits some discussion here to bring clarity to the current scheme.
The 2002 classification reflected terminology first proposed in 1966 by Norris and Taylor in their seminal paper on endometrial stromal neoplasms, although the definitions of these categories have changed. The main difference in the prior classification centered on the definition of low-grade ESS and its separation from what was originally termed high-grade endometrial stromal sarcoma. As originally proposed in 1966, separation into low- and high-grade categories was solely based on mitotic activity—if a tumor that morphologically resembled endometrial stroma had fewer than 10 mitoses/10 high-power fields (hpf), it was considered low grade, whereas a tumor with more than 10 mitoses/hpf was considered high grade. However, since this original publication, the clinical relevance of separating those examples of ESS with morphology similar to normal proliferative phase stroma into low- and high-grade categories based on mitotic activity has not been confirmed in subsequent studies, most notably by Evans and Chang et al. Moreover, the category of high-grade ESS had represented a heterogenous group of tumors, including those that resembled endometrial stroma and those that were poorly differentiated, being composed of larger cells with a greater degree of nuclear anaplasia akin to the mesenchymal component of carcinosarcoma. Therefore, in the 2002 WHO classification scheme, there remained two categories of endometrial sarcoma—low grade and undifferentiated—based on differences in tumor morphology rather than mitotic activity. A low-grade ESS is a tumor composed of cells that morphologically resemble non-neoplastic proliferative phase endometrial stroma that infiltrates the surrounding myometrium in a characteristic finger-like permeative fashion and typically invades lymphatic or vascular spaces. In contrast, an undifferentiated endometrial sarcoma is a poorly differentiated sarcoma composed of cells that do not resemble proliferative phase endometrial stroma, which usually shows destructive infiltration of the myometrium. Despite this attempt at a neat separation into diagnostic categories, it is now clear that there is a subset of ESSs that are histologically higher grade, with distinctive uniform morphology, immunophenotype, and genetic alterations. This is now recognized as high-grade ESS, as will be discussed below.
Endometrial Stromal Nodule
Endometrial stromal nodules may occur at any age but are typically encountered in women in their fifth and sixth decades. They may be located intramurally, with no connection to the endometrial surface (raising interesting possibilities concerning histogenesis and a shared developmental origin of uterine stroma and smooth muscle), may be located submucosally at the endomyometrial junction, or may project into the endometrial cavity as a polypoid mass. On gross examination, they are characteristically well circumscribed and may be mistaken for a leiomyoma; however, endometrial stromal nodules are usually softer in consistency, tend more commonly to be yellow, and lack the characteristic whorled bulging appearance of a typical leiomyoma ( Fig. 20.1 ). Hemorrhage and cystic degeneration may be seen (see Fig. 20.1B ). Endometrial stromal nodules can vary in size but are usually less than 10 cm, although larger tumors have been described.
Histologically, an endometrial stromal nodule is characterized by sharp circumscription between the neoplastic cells and surrounding endometrium or myometrium, with a pushing interface ( Fig. 20.2 ). Sometimes, however, there may be some irregularities in the border in the form of small lobulated or finger-like extensions into surrounding myometrium. By definition, no lymphatic or vascular invasion is present, and these foci should be less than three in number and less than 3 mm from the main mass. Tumors with more than three foci or with extension more than 3 mm from the main mass are considered as low-grade ESS with limited infiltration (see below). The neoplastic cells of endometrial stromal nodules resemble proliferative phase endometrial stroma, being composed of cells with uniform round to ovoid nuclei that have scant to moderate amounts of eosinophilic to amphophilic cytoplasm ( Fig. 20.3 ). These cells appear to whorl around the prominent vascular component, which resembles the spiral arterioles of non-neoplastic endometrium; foamy stromal cells may be conspicuous ( Fig. 20.4 ). The vessels are typically evenly spaced and uniform in caliber throughout the neoplasm; however, on occasion, larger, thick-walled vessels may be present, although this is usually only a focal finding in a minority of cases.
Low-Grade Endometrial Stromal Sarcoma
Women with low-grade ESS are typically perimenopausal but tend to be younger than those who present with other types of uterine sarcoma. They usually present with abnormal uterine bleeding or symptoms related to a uterine mass or extrauterine spread. The gross appearance can be variable, although a common feature is its tan to yellow color and soft consistency. Tumors may present as the following: (1) intramyometrial nodular masses; (2) an intracavitary polypoid mass; (3) diffuse myometrial infiltration, with expansion of the uterine wall; or (4) any combination of these patterns ( Fig. 20.5 ). Involvement of parametrial vessels may impart a wormlike appearance on gross examination.
Histologically, the neoplastic cells of low-grade ESS are morphologically indistinguishable from those of an endometrial stromal nodule; they are composed of small, bland-appearing cells resembling proliferative phase endometrial stroma. The mitotic rate is typically low (usually <5 mitoses/10 hpf), although it can be variable and sometimes brisk; necrosis is uncommon. A diagnosis of sarcoma is based on the presence of extensive myometrial infiltration ( Fig. 20.6 ) or invasion of lymphatic or vascular spaces ( Fig. 20.7 ). Characteristically, the neoplastic cells invade as rounded to irregularly shaped nests of varying size, which have been described as a finger-like permeation of the surrounding myometrium.
Endometrial Stromal Nodule and Low-Grade Endometrial Stromal Sarcoma
Endometrial stromal nodules and low-grade ESS, can occasionally exhibit a diverse range of differentiation including sex cord–like, epithelial (e.g., endometrioid type glands), smooth muscle, and, rarely, skeletal muscle differentiation. A fibroblastic appearance due to extensive stromal hyalinization secondary to increased collagenous matrix production may sometimes occur, making recognition as a stromal neoplasm difficult ( Fig. 20.8 ). However, in general, other areas of morphologically typical stromal neoplasia are usually present. In addition, the characteristic vascular component and arrangement of the tumor cells around it are maintained. Other less common variants include features such as epithelioid differentiation, tumor cells with clear or granular cytoplasm, bizarre cells, pseudopapillary growth, and adipose tissue.
Mixed Endometrial Stromal–Smooth Muscle Tumors
Endometrial stromal neoplasms with smooth muscle differentiation have been the focus of a limited number of studies, partly because of underrecognition and their uncommon occurrence. If more than 30% of the tumor is comprised of smooth muscle, they are designated as mixed endometrial stromal–smooth muscle tumors. Histologic features of smooth muscle differentiation include the following: (1) typical smooth muscle morphology reminiscent of that seen in leiomyomata; (2) nodules with central prominent hyalinization (the so-called starburst pattern) ( Fig. 20.9 ); and (3) irregular islands that can be discrete or merge imperceptibly with the areas characteristic of stromal differentiation ( Figs. 20.10 and 20.11 ). These tumors are of endometrial stromal derivation (see later, “ Molecular Genetics ”); therefore, determination of benign versus malignant should be made using the criteria for endometrial stromal tumors (e.g., presence or absence of myometrial invasion, lymphatic or vascular invasion).
Endometrial Stromal Tumor Resembling Ovarian Sex Cord Tumor
Endometrial stromal nodules and low-grade ESS may exhibit morphologic features of ovarian sex cord stromal tumors. These elements may be present focally, most often in the form of interanastomosing trabeculae ( Fig. 20.12 ), cords and, less commonly, tubules. They may also be the predominant pattern in which the stromal element is less conspicuous or absent; in this scenario, the term uterine tumor resembling ovarian sex cord tumor (UTROSCT) has been used. UTROSCTs show striking morphologic overlap with ovarian sex cord tumors mimicking granulosa cell, Sertoli cell, and even retiform Sertoli cell tumors ( Fig. 20.13 ). Whether these tumors represent endometrial stromal neoplasms has been a subject of debate, although molecular evidence has suggested that they may not be of endometrial stromal derivation (see later, “ Molecular Genetics ”). In addition to the morphologic resemblance, the sex cord–like elements are often positive for markers of sex cord differentiation, such as inhibin and O-13/CD99, suggesting that these foci represent true sex cord differentiation akin to that seen in the ovary.
The presence of sex cord–like differentiation in an endometrial stromal nodule and low-grade ESS does not appear to have an impact on biologic behavior. Histopathologic criteria for the distinction between a stromal nodule and stromal sarcoma, as outlined previously, are applied, regardless of the presence of sex cord–like elements. Less is known concerning the natural biologic behavior of tumors that show exclusive sex cord–like differentiation (UTROSCTs). Although evidence has suggested that they are benign—they are generally well-circumscribed, yellow, gray, or tan masses, and there have been no reported incidences of metastasis —the outcome in patients treated conservatively is not known, because all patients typically have undergone hysterectomy. We have also seen rare examples of UTROSCT with worrisome histologic features, including necrosis and brisk mitotic rate; in such cases, the biologic behavior should not be presumed to be benign but rather best considered uncertain.
Endometrial Stromal Tumor With Endometrioid Glands
Divergent differentiation down epithelial lines also may be occasionally seen in endometrial stromal nodule or low-grade ESS, usually as well-formed endometrioid glands ( Fig. 20.14 ). Although divergent differentiation is the most likely explanation for their presence, some examples may represent entrapped non-neoplastic endometrial glands. In general, this type of differentiation is focal, and the main differential diagnosis is the distinction from adenomyosis, particularly adenomyosis with sparse glands. The latter is usually encountered as an incidental finding in postmenopausal women and can be distinguished from low-grade ESS primarily based on the atrophic appearance of the stroma and the absence of a grossly identifiable mass.
High-Grade Endometrial Stromal Sarcoma
High-grade ESS is a recently described subtype of endometrial stromal neoplasia that is recognized by its distinctive morphologic appearance, immunohistochemical profile, and molecular genetic abnormality. These uncommon tumors occur in patients over a wide age range (mean, 50 years). Most patients present with abnormal uterine bleeding and/or symptoms related to extrauterine extension because these are often high-stage tumors. On gross examination, they typically form intracavitary polypoid masses, often with hemorrhage and necrosis ( Fig. 20.15A ), are poorly circumscribed, and show extensive permeation of the uterine wall, often with extension to the serosa (see Fig. 20.15B ). Histologically, the tumor shows extensive finger-like permeation of the myometrium and lymphovascular invasion ( Fig. 20.16 ); however, destructive infiltration with splaying of individual muscle fibers is also commonly seen ( Fig. 20.17 ). The tumor typically contains morphologically low- and high-grade areas, which can sometimes be appreciated on low-power examination as hypercellular and hypocellular areas ( Fig. 20.18A ); these areas can be juxtaposed with transition from a low- to high-grade area or intermingled (see Fig. 20.18B ). The morphologically low-grade areas can have a fibroblastic or myxoid appearance, similar to the fibrous and myxoid variants of low-grade ESS, and are composed of cells with round to oval nuclei and scant cytoplasm ( Fig. 20.19 ). The morphologically high-grade areas typically show nested and corded growth and are composed of a uniform population of histologically higher grade epithelioid tumor cells, with scant to moderate amounts of eosinophilic cytoplasm and round to oval vesicular nuclei with irregular nuclear contours and nucleoli, but without nuclear pleomorphism notable at low-power magnification (4–10× objective; Fig. 20.20 ). As a point of reference, the nuclei are typically four to six times larger than a lymphocyte. There is a prominent delicate arborizing capillary network in which small concentric arterioles are only present occasionally. Mitotic activity is brisk, typically more than 10 mitoses/10 hpf.
Undifferentiated Uterine Sarcoma (Undifferentiated Endometrial Sarcoma)
Usually, an undifferentiated endometrial sarcoma presents as one or more tan-yellow to gray, fleshy, intracavitary polypoid masses ( Fig. 20.21 ). Hemorrhage and necrosis are common. Histologically, the neoplastic cells show marked cellular atypia and numerous mitoses, including atypical forms, without evidence of differentiation toward endometrial stroma both cytomorphologically ( Fig. 20.22 ) and by pattern of growth within the myometrium. In general, an undifferentiated endometrial sarcoma has a diffuse and destructive infiltrative pattern as opposed to the characteristic wormlike pattern and propensity for intravascular extension characteristic of low-grade ESS. Because the tumor cell morphology of undifferentiated endometrial sarcoma is often likened to the mesenchymal component of a carcinosarcoma, exclusion of this latter entity should always be considered. In fact, the diagnosis of undifferentiated endometrial sarcoma should only be considered following exclusion of a poorly differentiated carcinoma, leiomyosarcoma, and carcinosarcoma, all of which may be morphologically similar in appearance ( Fig. 20.23 ). Extensive sampling and immunohistochemical staining (e.g., epithelial and smooth muscle markers) may be required. Undifferentiated uterine sarcomas have been subdivided into pleomorphic and uniform types. Many tumors in the latter category are now known to represent a high-grade ESS, as described above.
Low-Grade Endometrial Stromal Sarcoma
Chromosomal rearrangements involving chromosomes 6, 7, and 17 are the most frequent cytogenetic abnormalities that have been reported. The most common rearrangement is a reciprocal, balanced translocation between chromosomes 7 and 17, the t(7;17)(p15;q21). Two previously unknown genes, termed JAZF-1 and SUZ12 ( JJAZ-1) , were identified at the chromosomal sites of breakage in 7p15 and 17q21, respectively. In both stromal nodules and low-grade ESS, chimeric JAZF-1 – SUZ12 ( JJAZ-1) mRNA transcripts can be detected by reverse transcription polymerase chain reaction (PCR) in most cases. Moreover, fluorescence in situ hybridization (FISH) analysis can be performed on formalin-fixed, paraffin-embedded tissue. In this technique, two fluorescently labeled human genomic probes that flank the chromosome 7 breakpoint in the t(7;17)(p15;q21) are used to test for the presence of the translocation. In a tumor that contains the translocation, the fluorescent-labeled probes are split apart, yielding separate red and green signals ( Fig. 20.24 ). A yellow signal is produced by the close apposition of the two probes, marking the normal copy of chromosome 7.
The second most frequent translocation is the t(6;7)(p21;p15), in which there is fusion of JAZF-1 with PHF-1 . The PHF1 gene at 6p21 can also fuse with EPC1 at 10p11 or MEAF6 at 1p34. Interestingly, ESSs showing sex cord–like differentiation often have a PHF1 genetic rearrangement. Two additional translocations have been described in ESSs, a t(X;22) (p11;q13) and t(X;17) (p11.2;q21.33) associated with a ZC3H7B- BCOR fusion and MBTD1-CXorf67 fusion, respectively. Although endometrial stromal tumors are genetically heterogeneous, the different genes involved in low-grade ESSs are functionally related ( PHF1, SUZ12, EPC1, MBTD1 ), being members of the polycomb gene family. Of interest, ZC3H7B-BCOR , MEAF6-PHF1 , and EPC1-PHF1 fusions were also identified in ossifying fibromyxoid tumors and JAZ1-PFH1 in an ossifying sarcoma of the heart. Rarely, ESSs can show MDM2 amplification by FISH as well as MDM2 expression by immunohistochemistry, a potential pitfall, particularly in tumors occurring in locations more common to liposarcoma (e.g., peritoneum, retroperitoneum).
Endometrial stromal nodules and low-grade ESSs may show evidence of the t(7;17) translocation by FISH, suggesting that the stromal nodule may be a precursor lesion for low-grade ESS and that the t(7;17) is an early genetic abnormality in the development of low-grade ESS. In addition, FISH of mixed endometrial stromal–smooth muscle tumors has shown evidence for the translocation in the endometrial stromal and smooth muscle components, supporting the concept that these tumors are of endometrial stromal derivation.
High-Grade Endometrial Stromal Sarcoma
This tumor has a characteristic genetic abnormality—t(10;17)(q22;p13)—associated with a YWHAE-FAM22 (NUTM2AB) fusion. FISH analysis on formalin-fixed, paraffin-embedded tissue for the YWHAE-FAM22 rearrangement has become more widely available; however, in most cases, confirmatory FISH does not need to be performed if the morphology and immunophenotype are characteristic (see below).
Uterine Tumor Resembling Ovarian Sex Cord Tumors
These tumors have been thought to represent endometrial stromal neoplasms that show entirely sex cord differentiation. However, they do not have the JAZF1-SUZ12(JJAZ1) fusion, which is the most common aberration seen in low-grade endometrial stromal neoplasms. Thus, it is now believed that these may not represent an endometrial stromal neoplasm.
Interpretation of Curettings
Recognition of a morphologically low-grade endometrial stromal neoplasm in a curettage can sometimes be difficult because it must be distinguished from potential mimics, including endometrial basalis, aglandular functionalis, endometrial polyp, and adenosarcoma. Multiple fragments of aglandular cellular stroma containing spiral arteriole-like vessels with an expansile sheetlike growth are characteristic of a stromal neoplasm. In one study, curettage of a stromal neoplasm typically produces fragments of aglandular stroma measuring more than 5 cm ( Fig. 20.25 ). Distinction from fragments of basalis is made by the presence of an orderly component of glands and lack of the rich vasculature so characteristic of stromal neoplasia ( Fig. 20.26A ). Strips of aglandular functionalis, usually associated with submucosal leiomyomata, tend to be less cellular and show features of compression or reactive surface changes (see Fig. 20.26B ). Fragments of stroma-rich endometrial polyps usually exhibit other features of polyps, including large, thick-walled vessels and abnormal glandular architecture (see Fig. 20.26C ). An adenosarcoma also may exhibit a cellular stroma but typically has glandular cuffing, albeit sometimes subtle (see Fig. 20.26D ). Appreciation of a more spindled atypical stroma without a rich vascular network facilitates this distinction.
Once a stromal neoplasm is suspected based on histopathologic findings in the endometrial sampling, the next step in interpretation is determination of whether the tumor is benign or malignant. Unfortunately, this distinction is based on whether or not there is myometrial or lymphovascular invasion, two criteria that are difficult to interpret in biopsy or curettage material. One could raise the possibility of sarcoma if some fragments of tissue contain myometrium infiltrated by stromal tumor; however, prudent clinicopathologic correlation is critical. In general, the diagnosis of an endometrial stromal neoplasm, with a comment on one’s inability to distinguish between a stromal nodule and low-grade endometrial sarcoma in the submitted material, will be the most likely course of action.
Management and Prognosis
Endometrial Stromal Nodule
Endometrial stromal nodules are considered to be benign, nonrecurring neoplasms. Most women with stromal nodules have been treated by hysterectomy, due not only to their incidental discovery in hysterectomy specimens but also to the difficulty in distinguishing between stromal nodules and stromal sarcoma in biopsy or curettage material (discussed previously). In some cases, depending on the location and size of the tumor mass, preservation of fertility may be possible with partial uterine resection, which would include the mass and a rim of myometrium to assess for invasion. Preoperative imaging as to the feasibility of such an approach would be mandatory to assess for circumscription of the mass.
Only a few women with stromal nodules have been treated conservatively (local excision), and none of the tumors have recurred. In a case of an endometrial stromal nodule treated by local excision, an 8-year follow-up has been benign. Occasionally, some endometrial stromal nodules exhibit more than a 3-mm extension into the surrounding myometrium but lack the typical overt myometrial permeation seen in an ESS. The term endometrial stromal tumors with limited infiltration has been proposed for such tumors. Difficulty in predicting biologic behavior for these tumors is compounded by the fact that few cases have been described, all have been treated by hysterectomy, and long-term follow-up is not known. In practice, we recommend diagnosing these as ESS with limited infiltration or as endometrial stromal neoplasm with limited infiltration, with a comment suggesting that the tumor may pursue a benign clinical course but clinical follow-up is nevertheless recommended.
Low-Grade Endometrial Stromal Sarcoma
Patients with a low-grade ESS are treated by hysterectomy and bilateral salpingo-oophorectomy. Patients with tumors confined to the uterus (stage I) have an excellent prognosis, with a 5-year survival rate over 90%. Recurrence is not uncommon, ranging up to 25% in patients with stage I disease. Unfortunately, there are no histopathologic parameters to predict which patients with tumors confined to the uterus are at risk for recurrence. Distant metastases, principally involving the lung, may occasionally occur, often nearly a decade following the initial presentation. Following surgery, treatment options include local radiation therapy, which may reduce local failure. The impact of local radiotherapy on long-term survival, however, is not known. A low-grade ESS typically is positive for progesterone receptor. Therefore, hormonal therapy, particularly progestin therapy, is an option often considered in patients who present with advanced-stage disease or have recurrences.
High-Grade Endometrial Stromal Sarcoma
Patients typically present with advanced-stage disease (stage III or IV ≫ stage I) and are treated with hysterectomy, bilateral salpingo-oophorectomy, and tumor debulking, if indicated. Patients frequently have recurrences, usually within a few years of initial surgery. Although experience is limited, adjuvant therapy may provide survival benefit. Hormonal therapy is likely ineffective because the morphologically high-grade component is typically negative for estrogen and progesterone receptor (see below).
Undifferentiated Uterine Sarcoma
Most of the studies of clinical outcomes with different treatment regimens do not clearly distinguish between undifferentiated uterine sarcoma and high-grade ESS. The latter category might encompass tumors that would be classified according to current WHO criteria as a low-grade ESS as well as undifferentiated uterine sarcoma. Undifferentiated uterine sarcoma is an aggressive neoplasm and treatment options, in addition to surgery, include consideration of radiation therapy for local control, as well as chemotherapy for systemic control.
Biomarkers and Differential Diagnosis
Endometrial Stromal Nodule and Low-Grade Endometrial Stromal Sarcoma
Endometrial stromal tumors with smooth muscle differentiation or those that have a more fibrous or myxoid appearance may occasionally be confused with uterine smooth muscle tumors. Conversely, uterine smooth muscle tumors may mimic endometrial stromal tumors, particularly when the former is markedly cellular (e.g., highly cellular leiomyoma; see later, Figs. 20.75 and 20.76 ) or has prominent vascular invasion (e.g., intravenous [IV] leiomyomatosis; see later, Fig. 20.90 ) and IV leiomyosarcomatosis. In difficult cases, application of a panel of biomarkers may be helpful, provided that one is aware of potential pitfalls.
Endometrial stromal nodules and low-grade ESSs are usually only focally positive for smooth muscle actin and desmin; however, a subset of morphologically typical cases may show more extensive expression of these markers. In contrast, h-caldesmon is a more specific marker of smooth muscle differentiation than desmin and may be useful in this differential. Non-neoplastic and neoplastic endometrial stromal cells are typically negative for this marker, whereas non-neoplastic and neoplastic uterine smooth muscle is positive ( Fig. 20.27 ). In some cases of highly cellular leiomyoma, h-caldesmon may only show patchy or focal positivity (see later, Fig. 20.78 ).
The role of CD-10 as a potential marker of endometrial stromal differentiation was proposed based on its unexpected expression in ESS. Subsequent analyses have shown that CD-10 is expressed in endometrial stromal cells of the cycling endometrium (less so in decidua), adenomyosis, endometriosis, and endometrial stromal tumors, both nodules and low-grade ESS ( Fig. 20.28 ). CD-10 is typically strongly and diffusely positive in non-neoplastic and neoplastic endometrial stroma; however, some endometrial stromal tumors may be negative for this marker. As a further caveat, smooth muscle tumors, particularly highly cellular leiomyomata and leiomyosarcoma, may be positive for CD-10 ( Fig. 20.29 ; see later, Fig. 20.79 ), but usually not to the degree seen in endometrial stromal tumors.
In general, the morphologic appearance of tumor cells and the growth pattern within myometrium can distinguish a low-grade ESS from leiomyosarcoma. In cases in which there is prominent lymphatic or vascular permeation by leiomyosarcoma, ESS may be considered. Although morphologic features such as the presence of a fascicular architecture, even in the intravascular component, can help facilitate its recognition as a malignant smooth muscle tumor, a panel of antibodies, including h-caldesmon, desmin, and CD-10, may be performed in difficult cases. Most low-grade ESSs will be CD-10–positive (diffusely), h-caldesmon–negative, and desmin variable (but usually focal); most leiomyosarcoma will be h-caldesmon–positive, desmin–positive, and CD-10 variable (often positive). The main pitfalls to consider are that nearly 50% of leiomyosarcomas can be CD-10–positive, sometimes diffusely (see Fig. 20.29 ), some low-grade ESSs can be diffusely positive for desmin, and some ESSs can be CD-10–negative. Low-grade ESSs are typically not h-caldesmon–positive unless there are areas of smooth muscle differentiation. In the latter situation, interpretation of areas with classic morphology will facilitate the correct diagnosis. In the distinction from leiomyosarcoma, areas of smooth muscle differentiation in a low-grade ESS tend to be bland and will not exhibit the degree of cellularity and nuclear pleomorphism that can be present in a leiomyosarcoma.
High-Grade Endometrial Stromal Sarcoma
A high-grade ESS is characterized by different patterns of staining in the morphologically low-grade versus high-grade component. The low-grade component is typically positive for CD-10, estrogen receptor (ER), and progesterone receptor (PR) and will show variable positivity for cyclin D1, whereas CD117 is negative. In contrast, the high-grade component is typically negative for CD-10, ER, and PR and shows strong and diffuse positivity (>70% tumor cell nuclei) for cyclin D1 ( Fig. 20.30 ); CD1-17 is often positive.
Low- and high-grade ESSs both show nodular permeative growth within the myometrium; however, the latter is more likely to present at an advanced stage, show destructive myometrial infiltration, and, in most cases, show areas of higher grade morphology, as described above. Moreover, the vascular pattern is different, with a high-grade ESS having numerous delicate and arborizing vessels as opposed to the spiral, arteriolar-like vascular network of a low-grade stromal sarcoma. Immunoperoxidase stains are helpful in this distinction, but correlation with morphology is key. The high-grade epithelioid areas are typically negative for CD-10, ER, and PR but strongly and diffusely positive for cyclin D1. In some cases, correlation with molecular testing may be useful in this distinction.
The morphologically high-grade component of a high-grade ESS has an epithelioid appearance and may be confused with an undifferentiated carcinoma, particularly in limited material, such as a biopsy or curettage. One pitfall to keep in mind is that undifferentiated carcinomas of the endometrium can also show strong and diffuse positivity for cyclin D1; however, they will usually show focal positivity for epithelial membrane antigen (EMA) and broad-spectrum cytokeratin or be positive for CD-10. In cases lacking positivity for these markers, molecular studies might be necessary for the diagnosis.
Tumors of the Myometrium
Definition and Classification
Benign and malignant uterine smooth muscle tumors occur throughout the female genital tract, from the vulva to the broad ligament and ovaries. The vast majority of these tumors, however, are located in the uterine corpus, where they are presumed to arise from benign myometrial cells. Their location in the uterus, in combination with their size, largely determines the resulting symptoms. For example, submucosal tumors often present with abnormal bleeding and may be associated with a poor reproductive outcome. In contrast, subserosal and intramural tumors typically present with pain or impingement on nearby pelvic organs. Subserosal leiomyomas are even thought by some to have the capacity to detach from the uterus after the development of a new blood supply, earning them the colorful term of parasitizing leiomyoma.
In most cases, classification of a smooth muscle neoplasm as benign or malignant is straightforward. Such determination rests entirely on histopathologic features, particularly the presence or absence of atypia, proliferative activity, and particular pattern of necrosis. As shall be discussed, uterine smooth muscle tumors are well known for their benign variants, which have one of the features of malignancy in isolation. Tumors with several features of malignancy—but that do not meet the criteria for the diagnosis of leiomyosarcoma—add further complexity to classification schemes. The prediction of clinical behavior of such morphologic intermediates is difficult at best. Superimposed on this spectrum of smooth muscle neoplasia is a number of morphologically benign smooth muscle proliferations with the biologic features of malignancy—namely, dissemination or distant metastasis, vascular invasion, or local infiltration. This morphologic and biologic diversity makes smooth muscle neoplasia a fascinating area of study.
Benign tumors of the muscular uterine wall are known variously as leiomyomata, myomas, fibromyomas, fibromas, or fibroids. Leiomyomata are observed in nearly 77% of hysterectomy specimens, regardless of the indication for surgery, and the average number of independent tumors per uterus has been estimated to be more than six. Consequently, they are the most common human tumor. Fortunately, only about 25% of women of reproductive age are symptomatic.
Women with symptomatic leiomyomata generally present after the age of 35 years. Symptoms may include abnormal uterine bleeding, pelvic pressure or pain, and reproductive dysfunction. Abnormal uterine bleeding as a result of leiomyomata may be characterized as menorrhagia or hypermenorrhea. Leiomyoma-induced menorrhagia may be so severe as to require changing sanitary napkins hourly for more than 5 days; such profound bleeding can result in significant anemia. Leiomyoma-associated bleeding also may become a source of social embarrassment and result in significant lost productivity. Pelvic pain or pressure is typically associated with tumors large enough to distort the uterine corpus. In a testament to the size that a uterus with leiomyomata may reach, these so-called fibroid uteri are clinically assessed by comparing them to a gravid uterus. Thus, a symptomatic uterus may be as large as a gestational uterus at 16 to 20 weeks. In addition to pain or pressure, large tumors may compress nearby structures and may occasionally cause constipation, urinary frequency, or ureteral obstruction. Although uncommon, the range of reproductive dysfunction disorders associated with leiomyomata includes infertility, spontaneous abortion, premature labor, and fetal malpresentation. Pseudo-Meigs syndrome (ascites and hydrothorax) rarely may be attributed to uterine leiomyoma.
The severity of symptoms associated with leiomyomata is broadly related to tumor size and location. The myometrium can conceptually be divided into three zones—submucosal, intramural, and subserosal. Submucosal leiomyomata, as well as larger intramural tumors that distort the endometrial cavity, may cause abnormal uterine bleeding ( Fig. 20.31 ). Although all the pathophysiologic details have yet to be elucidated, attenuation of the endometrium overlying leiomyomata is frequently found. Sampling of these attenuated areas produces strips of aglandular endometrium in curettings, which can be used to suggest the diagnosis of leiomyomata (see Fig. 20.25 ) and a cause for bleeding, provided that no other endometrial pathologies capable of causing bleeding are present. Subserosal or deeper leiomyomata are less likely to cause uterine bleeding but are more likely to be associated with pelvic pain or pressure.
A fibroid uterus is the most common indication for hysterectomy, and 2.1 hysterectomies/1000 women are performed annually for this diagnosis in the United States. Despite the benign nature of leiomyomata, the impact of more than 200,000 major surgical procedures each year on public health and medical economics is considerable. A hysterectomy performed for leiomyomata is one of the most common procedures in the practice of surgical pathology.
Fibroids also may be managed expectantly if associated with minimal or no symptoms. For some time, gynecologists and patients have expressed a growing interest in avoiding hysterectomy by developing or selecting less invasive alternatives to hysterectomy. Myomectomy, resection with uterine conservation, is the most widely used alternative for women who wish to preserve fertility. Myomectomy may be performed using an open abdominal approach or by various closed techniques involving laparoscopy, hysteroscopy, or myolysis, which involves in situ coagulation using a laparoscopic probe. Tumor size and location play an important role in determining which technique is most appropriate. When compared with hysterectomy, these alternatives are associated with several unique risks—namely, the risks of tumor recurrence and uterine rupture during pregnancy following myomectomy. The risk of symptomatic recurrence is more likely due to the growth of a second crop of tumors and requires a second operation in 10% to 26% of cases. Morcellation with a powered device may result in peritoneal dissemination of benign and malignant smooth muscle tumors. Studies have shown that the rate of unexpected sarcoma after a laparoscopic morcellation procedure may be far higher than once thought, and that dissemination of leiomyosarcoma by morcellation occurs in nearly two-thirds of cases, with some resulting in mortality. Uterine artery embolization with particles of polyvinyl alcohol has been used to treat leiomyoma noninvasively. Although uterine artery embolization results in symptomatic improvement for most patients, this technique has been associated with adverse outcomes, ranging from postprocedure fevers to amenorrhea, uterine rupture, endomyometritis, and fatal sepsis. Uterine artery embolization also has been implicated as a factor delaying the diagnosis of uterine sarcomas. The latest noninvasive technique for the treatment of leiomyomata is MRI-guided focused ultrasound, which causes thermolysis of targeted smooth muscle cells (see later, Figs. 20.71 and 20.72 ).
Although surgery has been the mainstay of treatment for benign uterine smooth muscle tumors, medical therapy may be helpful in particular cases. Androgenic steroids, such as danazol and gestrinone, cause amenorrhea, which may be helpful in treating leiomyoma-associated anemia. Gestrinone also may reduce leiomyoma volume. Treatment with gonadotropin-releasing hormone (GnRH) agonists, the most widely used medical therapy for leiomyoma, also results in reduction of leiomyoma volume by 35% to 65% by producing a pseudomenopausal hypoestrogenic state. This therapy places the patient at risk for osteoporosis and other significant complications; consequently, GnRH agonists must be used only for a short time (e.g., until definitive surgery can be performed or while waiting for natural menopause to occur). Unlike gestrinone, the reduction in tumor volume associated with GnRH agonists is rapidly reversible and, in part, may reflect volume shifts in the extracellular matrix of leiomyomata rather than changes in the neoplastic cells. Volume changes in treated leiomyomata may be due to changes in apoptosis and insulin-like growth factor 1 (IGF-1) receptor activity as well. Mitigation of the risks associated with hypoestrogenism has been attempted by adding exogenous hormones after an initial period of complete suppression, but these protocols have not been widely adopted. Of note to pathologists, there are few or no histologic changes in leiomyomata after treatment with GnRH agonists. Finally, selective estrogen receptor modulators such as raloxifene may inhibit the growth of leiomyomatous smooth muscle cells and reduce leiomyoma volume. Similar to other noninvasive therapies, the medical treatment of leiomyomata may delay the diagnosis of leiomyosarcoma.
Pathobiologic Features of Typical Leiomyoma
As their response to GnRH agonists illustrates, leiomyomata are hormonally responsive tumors. They are rare before menarche, may grow rapidly during pregnancy or in response to clomiphene administration, and often decrease in size after menopause. Moreover, the smooth muscle cells in myometrium and leiomyomata have receptors for estrogen and progesterone, with some studies suggesting that the abundance of steroid receptors is greater in leiomyoma compared with myometrium. Mitotic activity in leiomyoma also varies over the course of the menstrual cycle, with the greatest activity in the periphery of the leiomyoma during the secretory (luteal) phase. Hormone replacement therapy after menopause stimulates the growth of leiomyomata. Finally, in some but not all studies, the presence of a polymorphism in the estrogen receptor has been associated with an increased risk of leiomyomata.
Leiomyomata are independent clonal neoplasms. The unicellular origin of leiomyomata was first established by detecting nonrandom inactivation of the X chromosome by Linder et al. and Townsend et al. In these early studies, the pattern of X chromosome inactivation was established by measuring glucose-6-phosphate dehydrogenase isoforms. All tumor cells expressed only one of the allelic isoforms, indicating that they arose from a single cell in which the other allele had been inactivated by lyonization, of which X chromosome inactivation by methylation is an important feature. This strategy has been replicated using a size polymorphism within the androgen receptor locus at Xp12 as a marker of X chromosome inactivation and, by inference, clonality. Within a single uterus, the pattern of allelic inactivation is random as well, indicating that each clonal leiomyoma arises from an independent transformation event. The mechanisms accounting for this high rate of transformation in myometrial smooth muscle cell are unknown.
There is substantial evidence of a genetic basis for uterine leiomyomata. One clue to the genetic contribution to leiomyoma tumorigenesis comes from twin studies. It has been observed that the risk for hysterectomy, a surrogate for the diagnosis of leiomyoma, doubles with monozygous twins (i.e., genetically identical) relative to dizygous (i.e., genetically nonidentical) twins. In addition, the genetic background contributes to the risk of having symptomatic leiomyomata; premenopausal black women have a two- to threefold increase in symptomatic fibroids compared with women in other ethnic groups. Finally, studies of families have also suggested a heritable predisposition to the development of leiomyomata.
Leiomyomata of the female genital tract are featured in several inherited syndromes. The most prominent of these genetic tumor syndromes is Reed syndrome, also known as multiple cutaneous and uterine leiomyomata or hereditary leiomyomatosis and renal cell carcinoma (HLRCC) syndrome (Online Mendelian Inheritance in Man [OMIM], 150800). This syndrome is characterized by autosomal dominant inheritance of multiple cutaneous leiomyomas (cutaneous leiomyomatosis), which arise from the smooth muscle of erector pili muscles beginning in late adolescence or early adulthood. In addition to cutaneous tumors, affected women in these families frequently have symptomatic uterine leiomyomata at an early age and often undergo hysterectomy before the age of 30 years. Recognition of this syndrome is important because it is also associated with renal cell carcinoma. Of note, the syndromic form of renal cell carcinoma, which often has a papillary architecture but is distinct from the papillary subtype of renal cell carcinoma, behaves aggressively, frequently presenting at a high stage and leading to mortality. It also has been suggested that this syndrome is associated with uterine leiomyosarcoma; however, when the uterine pathology was carefully evaluated in one case initially diagnosed as leiomyosarcoma, the tumor was subsequently determined to be an atypical leiomyoma, a histologic variant of leiomyoma with characteristic nucleolar and chromatin alterations (described in detail later). HLRCC has been mapped to a locus on chromosome 1, band q42.1. Surprisingly, this locus turns out to be fumarate hydratase (FH; fumarase), an enzyme that converts fumarate to malate in the tricarboxylic acid (citric acid or Krebs) cycle. The mechanism is not yet understood, but the loss of heterozygosity for the wild-type allele and near-complete loss of enzymatic activity suggests that FH behaves like a tumor suppressor gene. Of note, mutations in a gene encoding a succinate dehydrogenase subunit, also in the tricarboxylic acid cycle, is associated with familial paraganglioma, and this parallel system tends to dispel any residual skepticism that enzymes of intermediary metabolism can act as tumor suppressors. Abnormalities of FH, however, are not frequently found in nonsyndromic, histologically typical leiomyomata, suggesting that the pathogenetic mechanism causing presumably sporadic leiomyoma is distinct from that causing this familial syndrome. Finally, loss of FH function is found only rarely in nonsyndromic leiomyosarcoma.
Another genetic syndrome causing smooth muscle proliferation is a variant of Alport syndrome (OMIM 308940 ). In addition to the well-known kidney manifestations, some affected individuals have esophageal and vulvar leiomyomatosis. The molecular defect in this syndrome is the deletion of the genes for the alpha-5 and alpha-6 chains of type IV (basement membrane) collagen ( COL-4A5 and COL-4A6 ), which are arranged head to head on Xp22.3.
It has been found that about 40% of typical uterine leiomyomata have at least one clonal cytogenetic aberration. In contrast to leiomyosarcoma, the karyotypes of benign smooth muscle tumors are simple. At least six cytogenetic subgroups have been recognized in garden variety leiomyomata, as follows: (1) deletion of 7q; (2) translocation between 12q15 and 14q24; and (3) trisomy 12 and (4 to 6) rearrangements of 6p, 10q22, and 13q. The large number and variety of cytogenetic subtypes suggests that there are multiple pathways to tumorigenesis in uterine smooth muscle. In cases with mosaicism, the cytogenetic abnormalities represent a secondary change acquired after transformation and clonal expansion. Increased tumor size and the presence of clonal cytogenetic abnormalities are related, suggesting that chromosomal abnormalities enhance the growth of leiomyomata.
The best-studied chromosomal aberration in leiomyoma is t(12;14)(q15;q24; Fig. 20.32 ). This balanced translocation involves the HMGA-2 and RAD-51B loci on chromosomes 12 and 14, respectively. Interestingly, HMGA-2 is associated with a number of other benign mesenchymal tumors, including lipoma, breast fibroadenomas, endometrial polyps, pulmonary chondroid hamartomas, pleomorphic adenomas of the salivary gland, and vulvar aggressive angiomyxoma. HMGA-2 , also termed high-mobility group protein I-C or HMGI-C , is an A-T hook, DNA-binding protein that contributes to transcriptional regulation. RAD-51B , or RAD-51-like 1 , is the human homologue of the bacterial DNA repair gene RecA . It appears that the critical effect of this rearrangement is inappropriate expression of HMGA-2 , either as a full-length or truncated transcript containing the DNA-binding domains. This specific rearrangement has been implicated in the development of pseudo-Meigs syndrome. Interestingly, the rearrangement involving 6p in leiomyoma involves the family member of HMGA-2 —namely, HMGA-1 —suggesting that the two chromosomal aberrations share a common pathogenetic mechanism. An allelic variant of HMGA-2 also may contribute to leiomyoma predisposition and reduce stature.
Deletion of 7q22 defines another distinct subgroup of leiomyomata. An array comparative genomic hybridization analysis has determined that 9.6 Mb from 7q22-7q31.1 is the minimal region consistently deleted. A number of genes located in this region, including the proliferation inhibitor HPB-1 as well as the mitosis integrity maintenance tumor suppressor RINT-1 , have reduced expression.
Macroscopic and Microscopic Features of Typical Leiomyoma
Although the pathologic features of leiomyomata of the usual type are straightforward, a clear understanding of them is required to recognize tumors requiring more scrutiny. In addition, such an understanding is required if we are to forgo sampling every fibroid mass in routine surgical specimens. Fortunately, the features of typical leiomyomata are quite consistent.
The typical leiomyoma is grossly well circumscribed (see Fig. 20.31 ) and has a firm rubbery texture. On incision, this tumor often bulges out because of increased intratumoral pressure. The cut surfaces are white or slightly pink, and the bands of neoplastic smooth muscle are often whorled, giving the impression that the smooth muscle bundles are wrapped around a central core. Finally, there should be minimal variation in the appearance of the cut surface. Any significant variation or deviation from this appearance must receive additional consideration by the pathologist.
A histologic examination of a typical leiomyoma shows fascicles of bland smooth cells ( Fig. 20.33 ), which often have larger diameters and tend to be arranged with tighter packing as compared with myometrium. Consequently, the small venules are less conspicuous and less random in leiomyoma when compared with myometrium. The neoplastic smooth muscle cells themselves are virtually indistinguishable from their normal counterparts. Specifically, the cells are long and tapered, have abundant pink cytoplasm, and contain spindle-shaped nuclei, which have a relatively uniform shape and size ( Fig. 20.34 ). The chromatin is lightly stained, finely textured, and uniformly dispersed. Nucleoli may be noted but should be small and inconspicuous. As we shall see, mitotic figures may be present to varying extents, particularly in the luteal phase. Atypical mitotic figures, however, are worrisome and should prompt further examination and exclusion of malignancy. Interspersed among the benign smooth muscle cells of a leiomyoma, one may see varying numbers of mast cells and, occasionally, even prominent infiltrates of other chronic inflammatory cells. Occasionally, but rarely, dense lymphocytic infiltrates may simulate lymphoma. Although the pathobiologic basis for the recruitment of inflammatory cells is yet to be elucidated, no specific or worrisome clinical significance has been attached to their presence. One potential exception might be the observation that fewer mast cells are found in leiomyosarcomata, whereas more are found in cellular and atypical leiomyomata. One also may see a wide variation in the amount of collagenous extracellular matrix within benign leiomyomata. All but the most extremes in the spectrum of cellularity and hyalinization may be readily dismissed without specific comment in routine practice.
Most clinicians will find sampling every fibroid impractical. This raises the question as to how many fibroids should be sampled. As will be detailed below, leiomyosarcomata are frequently the largest mass and usually have distinctive gross features. Selective sampling of fibroids, therefore, requires a thorough understanding of these distinctive features of malignancy. Minimal deviations from the typical appearance of a benign fibroid merit additional tissue sampling. In typical fibroids, we prefer to sample up to three of the largest tumors in hysterectomies and every fragment in myomectomy specimens.
Alternative Patterns of Differentiation
Deviations from the typical spindle cell morphology will sometimes be observed. Usually, these alternate patterns of differentiation represent a minor component, but occasionally these unusual patterns may predominate. In most cases, the diagnosis of a benign tumor can be made when the alternative pattern is appreciated as such.
Plexiform and Epithelioid Leiomyoma
In our practice, the most frequent alternative pattern of differentiation has an epithelioid appearance, which was previously referred to as being plexiform. In plexiform leiomyoma, small ribbons or islands of rounded smooth muscle cells are present ( Fig. 20.35 ), and it is this nonfascicular component for which the tumors are named. Although most attention is focused on the epithelioid appearance of the cells in this pattern, the extracellular matrix plays an important, if not defining, role. Between the ribbons or nests of cells, an abundant matrix is present, which acts to entrap these cells, resulting in a loss of the typical spindle shape and the gain of an epithelioid appearance. Studies of leiomyoma cells have suggested that extracellular matrix constituents are an important part of their repertoire of expressed genes. In tumors with plexiform differentiation, it would seem that their capacity to synthesize extracellular matrix, particularly type I collagen (unpublished data), is particularly accentuated. In this context, we think it is important to distinguish plexiform leiomyomata with a pseudoepithelioid appearance secondary to matrix deposition from smooth muscle tumors with true epithelioid differentiation (discussed later). Molecular studies have indicated that their gene expression profile distinguishes them from typical leiomyomata, and their cytogenetic abnormalities are distinct from leiomyoblastomas. Frequent, often prominent HMGA-2 overexpression, usually as a result of chromosomal rearrangement other than the typical t(12;14), typifies this morphologic subgroup. Finally, it has been suggested that the single filing of smooth muscle cells in a plexiform leiomyoma may be confused with metastatic breast cancer. Multiple plexiform tumors also may have an infiltrative pattern and consequently mimic ESS. Thus, despite being a morphologic mimic, this variant of epithelioid leiomyoma deserves recognition as a specific entity in routine practice.
In contrast to plexiform leiomyoma, a true epithelioid leiomyoma, also known as leiomyoblastoma ( Fig. 20.36 ), and clear cell leiomyoma ( Fig. 20.37 ) are uncommon. Based on immunohistochemical and ultrastructural studies, it has been suggested, perhaps incorrectly, that this tumor mimics fetal myocytes. Both epithelioid and clear cell leiomyomata are characterized by rounded or polygonal cells, rather than by spindle cells. Epithelioid leiomyoma (leiomyoblastoma) is composed of rounded cells with abundant eosinophilic cytoplasm (see Fig. 20.36 ), whereas, as its name denotes, the clear cell variant is composed of cells with clear cytoplasm in routinely stained tissue sections (see Fig. 20.37 ). This clear appearance to the cytoplasm is due to vacuolization of mitochondria or lysosomes.
Women with epithelioid leiomyomata have clinical characteristics similar to those of leiomyomata, with the typical spindle cell morphology. Specifically, these tumors present in the later reproductive years but can occur at any age, beginning in the third decade. Epithelioid leiomyomata may have an unusual macroscopic appearance, which would not be specific but should prompt more histologic sampling. The unusual features noted on inspection of the incised surface might include a softer texture or yellow or tan color.
The rarity of this tumor has hindered study of its natural history and, by extension, its prognostication. Its unpredictability has led some to regard epithelioid smooth muscle neoplasms as tumors of low malignant potential. In a large series ( N = 18), Prayson et al. retrospectively correlated pathologic features with clinical outcome. Similar to nonepithelioid smooth muscle tumors, no one histologic feature was predictive of metastatic potential. In general, benign epithelioid smooth muscle tumors were smaller (<6 cm), had low mitotic rates (up to 3 mitoses/10 hpf), lacked severe nuclear atypia, and lacked tumor necrosis. In contrast, Atkins et al. have emphasized the strong association of either tumor cell necrosis or mitotic activity in excess of 5 mitoses/10 hpf with poor outcome. Even with these studies, the classification and prognostication of epithelioid smooth muscle tumors remains problematic. Recognizing the limited experience with uterine epithelioid smooth muscle tumors, we have adopted a conservative approach to their classification reflecting published criteria, summarized in Table 20.1 .
|Diagnosis||Geographic Tumor Necrosis||Mitotic Rate: Mitoses/10 hpf||Atypia|
|Epithelioid leiomyoma, leiomyoblastoma, or cell subtypes||Absent||<5||None or minimal|
|Epithelioid leiomyosarcoma||Present||Any rate||Present or absent|
|Absent||≥5||Present or absent|
|Epithelioid smooth muscle tumor of uncertain malignancy||Absent||Present|
A small number of epithelioid leiomyomata have been analyzed cytogenetically to date. In a series of five tumors, four had simple karyotypic abnormalities similar to those seen in leiomyomata of the usual histologic type. Interestingly, two tumors had del(7)(q21.1q31.2), which includes the critical 7q22 region in typical leiomyoma. This chromosomal deletion, however, was reported to be a secondary change in one case. Karaiskos et al. have also found a balanced translocation between 12q15 and 10q22, which raises the possibility of concurrent rearrangements of the HMGA-2 and MORF loci. The similarity between the cytogenetics of typical and epithelioid leiomyomata suggests that they share pathobiologic mechanisms. The greater complexity of cytogenetics aberrations and rearrangements involving 17q21 in two epithelioid tumors, however, may distinguish epithelioid and nonepithelioid smooth muscle tumors.
Lipoleiomyomata are mixed tumors in which both smooth muscle and adipose cells are present. In general, smooth muscle cells outnumber adipocytes ( Fig. 20.38 ). Rarely, the adipocytes may be so numerous as to replicate a lipoma. The incised surface of a lipoleiomyoma is often bright yellow ( Fig. 20.39 ) and, as the fat content increases, it takes on an increasingly yellow and soft appearance, mimicking the gross appearance of a lipoma ( Fig. 20.40 ). Lipoleiomyomata are widely regarded as tumors of older women compared with typical leiomyomata, suggesting that the adipocytic differentiation is degenerative. Cytogenetic analyses of three tumors to date, however, have suggested an alternative explanation. Abnormalities of chromosome 5 have been noted in two tumors, raising the possibility of a pathogenetically relevant gene. More interestingly, two tumors have rearrangements involving 12q15. HMGA-2 , which is located at 12q15, was shown to be aberrantly expressed in one of these tumors. Of note, HMGA-2 is also involved in lipomas, as well as 10% of leiomyomata of the usual type; therefore, aberrant expression of HMGA-2 may account for the occasional adipocytic differentiation in lipoleiomyomata.
Vascular leiomyomata or angiomyomata are an uncommon morphologic pattern in leiomyoma. Thick-walled vessels are a characteristic of leiomyoma in general; however, in vascular leiomyomata, this component is particularly prominent ( Fig. 20.41 ). These tumors are benign, and their diagnosis carries no particular clinical implication. To date, only one tumor has been analyzed by cytogenetics, revealing the presence of t(X;11) (p11.4;p15) in the mainline and inv(2)(p15q13) and t(5;20)(q13;q13.2) in two stemlines.
Miscellaneous Patterns of Differentiation
Other patterns of differentiation may rarely be noted and, in most cases, the alternative differentiation is focal. Some leiomyomas may have patches of cells with nuclear palisading that mimic a schwannoma ( Fig. 20.42 ). In others, sex cord–like elements or tubules may be seen. Finally, uncommon tumors may harbor impressive numbers of inflammatory or hematopoietic cells. Although the presence of these admixed cells or aberrant differentiation may pose an interesting scientific puzzle, such tumors behave in a benign fashion.
In contrast to benign uterine smooth muscle tumors, leiomyosarcomas are fortunately rare, accounting for 1% of all uterine malignancies. Nevertheless, they are the most frequent malignant mesenchymal tumor of the uterus, accounting for 25% of all uterine mesenchymal neoplasms. Their annual incidence has been estimated to be 0.64 cases/100,000 women. In addition, it has been estimated that the incidence of leiomyosarcoma in women with the preoperative diagnosis of fibroid uterus is between 0.13% and 0.29%. The incidence of leiomyosarcomas may be increased in populations of black women compared with those of other ethnic backgrounds, but the magnitude of this increase is modest relative to that noted for benign leiomyomata. In comparison to benign smooth muscle tumors, leiomyosarcomas tend to present later in life, usually around or after menopause. Their incidence rises steadily from 0.2% in the fourth decade to 1.7% in the seventh decade. Consequently, the clinical appreciation of a large or rapidly growing fibroid after menopause or during GnRH agonist (e.g., leuprolide) therapy is a worrisome sign. Most leiomyosarcomas, however, are unsuspected or are presumed to be leiomyomata before pathologic examination of a surgical specimen. A small fraction of malignant or suspicious tumors are diagnosed when fragments are obtained by curettage or hysteroscopic myomectomy.
Uterine leiomyosarcomas are highly malignant neoplasms, notable for aggressive growth, local recurrence, and frequent distant metastasis. The lung and liver are the most frequent sites of metastasis, but unusual sites such as skeletal muscle have been reported. Lymph node and ovarian metastases are uncommon, especially in the absence of extensive extrauterine disease. The morcellation of leiomyosarcoma during the course of a minimally invasive procedure for fibroids may be a more common occurrence than once thought by gynecologists, and the morcellation of leiomyosarcoma may incur substantial risk of dissemination and mortality.
The cytogenetic changes found in uterine leiomyosarcoma are far more complex than those found in leiomyomata of the usual type. These changes include both numeric and structural abnormalities ( Fig. 20.43 ). In addition, specific cytogenetic aberrations may vary from metaphase to metaphase. Such variations suggest that there is a high level of genomic instability in uterine leiomyosarcoma. The high frequency of chromosomal aberrations in leiomyosarcoma correlates with nuclear atypia and accounts for variations from diploidy. These changes also hint at the underlying pathobiologic processes that drive malignant smooth muscle neoplasia. Some uterine and extrauterine leiomyosarcomas, however, have been reported to have far simpler karyotypes. The overall complexity of genomic alteration has precluded a positional approach to leiomyosarcoma gene discovery, but tumors with simpler alterations may be targets for positional cloning. With a few exceptions (noted in subsequent sections), only a few smooth tumors with features between those of typical benign and malignant smooth muscle tumors have been characterized by cytogenetic analysis.
The genomic instability found in leiomyosarcoma by cytogenetic analysis is also reflected by other techniques that measure allelic imbalance. In general, the results of comparative genomic hybridization, a technique for measuring gross chromosomal gains and losses, have shown multiple and complex changes in leiomyosarcoma and little, if any, changes in leiomyoma. In particular, loss of heterozygosity for the long arms of chromosomes 10 and 13 are found in more than half of leiomyosarcomas, but not in leiomyomata. Although not as frequent, gains of 17p, Xp, and especially 1q also have been noted. Although these studies have not pointed directly to new diagnostic or therapeutic targets, they reinforce the notion that genomic instability is an important distinction between benign and malignant uterine smooth muscle tumors. Furthermore, the lack of a common pattern of allelic imbalance in uterine smooth muscle tumors suggests that uterine leiomyosarcomas and leiomyomata—at least tumors of the usual type—have different pathogeneses.
Given the complexity of the genomic alteration in leiomyosarcoma, several groups have attempted to understand uterine smooth muscle neoplasia by studying their transcriptional profiles. As might be expected, the differences in gene expression between myometrium and leiomyomata are small. Larger differences can be found between the expression profiles of leiomyosarcomas and leiomyomata or myometrium. Downregulation of gene expression is more frequent in malignant tumors. The distinct clustering of benign and malignant samples provides additional evidence that the pathways to leiomyoma are distinct from the pathway(s) to leiomyosarcoma and are the targets for the development of new clinical biomarkers.
Given the large difference in frequency between leiomyoma and leiomyosarcoma, many have argued that malignant leiomyosarcomas do not spring forth from benign leiomyomata. Despite this conventional wisdom, cases of uterine leiomyosarcoma apparently arising from fibroids have been reported from occasionally. There are also several lines of molecular and cytogenetic evidence suggesting a neoplastic progression. First, although the overall cytogenetic differences between benign and malignant uterine smooth muscle tumors are large, t(10;17) has been observed as the sole chromosomal abnormality in both leiomyosarcomata and cellular leiomyomata. Next, histologically unusual leiomyomata with deletions of 1p have gene expression profiles more similar to leiomyosarcoma than to leiomyoma of the usual type. Furthermore, a number of pathologists, including ourselves, have seen cases of atypical leiomyomata that recurred and evolved into leiomyosarcoma following conservative myomectomy. Mittal et al. have found the same DNA copy number aberrations in symplastic leiomyoma–like areas and adjacent leiomyosarcoma in six cases, suggesting a common origin. Based on this mounting evidence, one can reasonably hypothesize that at least a subset of leiomyomata, probably atypical leiomyomata and tumors classified as having an uncertain malignant potential, have the capacity to transform into fully malignant leiomyosarcomas. Consequently, one would predict that it could be difficult to diagnose histologically intermediates, especially without understanding the underlying molecular genetic pathways. Not surprisingly, this is our experience and our challenge, particularly when managing patients with incompletely excised tumors or following surgical morcellation.
Macroscopic and Microscopic Features
Most leiomyosarcoma present as large dominant masses, often more than 10 cm in their greatest dimension ( Fig. 20.44 ). Similar to leiomyoma, they may arise throughout the female genital tract. Uterine leiomyosarcomas may be present in the subserosal, intramural, and submucosal compartments and are frequently large enough to span more than one compartment. Most leiomyosarcomata are distinctly different macroscopically from typical leiomyomata ( Table 20.2 ). Their cut surfaces are variegated because of the presence of hemorrhage or necrosis ( Figs. 20.45 to 20.47 ), and the hemorrhagic and necrotic zones are often grossly patchy, involving irregularly shaped areas in an apparently random distribution. The necrotic areas may appear as shades of green and yellow, whereas the non-necrotic areas have a prototypical appearance that some have compared to fish flesh. Leiomyosarcomata have earned this description for their grayer color, softer (i.e., nonrubbery) consistency, indistinct bundling, and decreased bulging of the cut surface as compared with leiomyomata ( Fig. 20.48 ). Recognition of these gross features cannot be emphasized enough because they are the trigger for more extensive sampling by the prosector.
|Multiplicity||Solitary, often dominant mass||Multiple masses|
|Gross circumscription||Poorly defined or grossly invasive||Well-defined, usually bulging and compressing adjacent tissues|
|Variegation of incised surface||Common, often prominent; multifocal hemorrhage or necrosis||Uncommon, often focal or central|
|Color of incised surface||Gray, yellow, or tan||White or tan|
|Consistency of incised surface||Soft, “fish flesh”–like||Firm, whorled|
Although gross features affect the level of suspicion for malignancy, malignancy is solely determined by histologic examination ( Table 20.3 ). Leiomyosarcomas often have severe nuclear and cytologic atypia, obvious proliferative activity, and cellular instability manifested as a particular pattern of necrosis. These three features are the principal factors in the determination of malignancy. In addition, these tumors may be hypercellular and infiltrate into surrounding myometrium. Diagnosis of malignancy is not difficult when all these features are present, particularly when florid. Clinical prognostication becomes more difficult when fewer features are present or are less pronounced. It has been long recognized that any attempt to classify these tumors must assess several histologic features concurrently. Over time, and based on a landmark study by Bell et al., the weight of each recognized feature has been evaluated and the classification schema adapted to reflect the relative importance of the various features studied. Variants of benign leiomyoma defined by each of these histologic features (see next section) must also be taken into account. Before these conceptual approaches can be applied, however, we must first make careful microscopic observations, guided by thorough gross examination.
|Proliferative activity||Usually elevated, often with abnormal forms||Variable, but usually low|
|Nuclear and cytologic atypia||Notable at low magnification; may be diffuse or focal; occasionally extreme||Rarely present|
|Pattern of degeneration||Coagulative, geographic tumor cell necrosis||Infarctive, hyaline, hydropic, or mucinous patterns|
|Cellularity||Usually hypercellular, sometimes prominent||Variable, but usually only moderately elevated|
|Microscopic border||Infiltrating adjacent myometrial fascicles||Sharp or interdigitating with adjacent myometrial fascicles|
Evaluation of Mitotic Activity
The first feature to be evaluated is proliferative activity ( Fig. 20.49 ). As noted, leiomyosarcomas have significant derangements affecting cell cycle control. Mitotic activity may be extreme and is amenable to a more quantitative evaluation. Typically, mitotic figures are counted in adjacent high-power fields (i.e., with the 40× objective) and averaged over 10 hpf. Although there is no specific protocol, it is reasonable to count at least 30 and perhaps more fields when calculating this average. It also may be helpful to note the mitotic rate of proliferative hot spots when found. These counts are probably more clinically meaningful if counting is biased to these areas by starting the count in a proliferative area found at a lower magnification. The area to be counted must be selected carefully; we have seen focal proliferation in benign submucous tumors as they ulcerate into the endometrial or endocervical cavities. A more important caveat is that each potential mitotic figure must be carefully scrutinized to ensure that it is not a degenerating cell. Such mimics, with pyknotic nuclei, are common in benign and malignant tumors, and casual counting may drastically overestimate proliferation in a given tumor ( Fig. 20.50 ).
It should be recognized that area selection and rigor in counting add a subjective component to an otherwise quantitative parameter. To overcome these difficulties, application of immunohistochemistry targeting a mitosis-specific form of histone H3 (by posttranslational phosphorylation of serine 10) has been advocated for the discrimination of true mitotic figures from pyknotic nuclei. Interestingly, Veras et al. have found that this method generally yields results comparable to those of the conventional hematoxylin and eosin (H&E)–based method for smooth muscle tumors of uncertain malignant potential, but actually yield higher mitotic indices in 50% of leiomyosarcomas studied. Their explanation for this increase is that some so-called pyknotic nuclei actually may represent cells early in mitosis (i.e., preprophase) after histone H3 modification but before full chromosomal condensation. Consequently, we prefer to use conventional mitosis enumeration by examination of H&E-stained section, except for special circumstances, such as when fragments of atypical smooth muscle tumor are found in curettings. In such cases, the standard count might be unreliable because too few fields are available for counting or the most proliferative area might be difficult to recognize. Immunohistochemistry for standard proliferation markers (e.g., Ki-67) is helpful in these cases.
Evaluation of Atypical Mitotic Figures
A feature related to mitotic activity is the presence of atypical mitotic figures. Unlike the mitotic rate, atypical mitotic figures are not common and consequently are not amenable to quantification. Atypical mitotic figures, however, reflect genomic instability, a pathologic feature of leiomyosarcoma and other malignant tumors that may be helpful when found. Declaration of atypia in mitotic figures also has a subjective component. To avoid this pitfall, we look for metaphases with more than one spindle axis ( Fig. 20.51 ) or anaphases in which individual chromosomes lag far behind their companions as they are drawn to the pole of their spindle apparatus ( Fig. 20.52 ). Such aberrations are believed to account for aberrant chromosome segregation or cycles of breakage and fusion, respectively. Mitotic atypia also may be observed in the form of metaphases with far too many chromosomes to be diploid. This criterion may be difficult to apply when it is subtle and should be used conservatively.
Evaluation of Cytologic Atypia
Another important feature of malignancy in smooth muscle neoplasia is cytologic atypia ( Figs. 20.53 and 20.54 ; see Fig. 20.49 ). Atypia may be manifested in the nucleus and cytoplasm of the neoplastic smooth muscle cell. The nuclei of benign smooth muscle cells of myometrium and leiomyoma have a fairly uniform spindle or corkscrew shape and are filled with bland dispersed chromatin. Nuclear atypia consists of prominent hyperchromasia and coarsening of chromatin texture, nuclear enlargement, multinucleation, multilobation, and other variations in nuclear shape and uniformity. The degree of nuclear atypia found in leiomyosarcomas spans a wide spectrum. In the extreme, it may be evocative of a trophoblastic malignancy (see Fig. 20.53 ). Pleomorphic tumor cells also may resemble inflammatory giant cells or osteoclasts (see Fig. 20.54 ). By wide agreement, the atypia should be notable at low (i.e., 10× objective) power for it to warrant classification as significant nuclear atypia. Mild nuclear changes, usually first noticed at higher magnification, do not predict clinical or biologic behavior. Cytologic atypia also may be manifested in leiomyosarcoma by divergent (heterologous) patterns of differentiation, which may include osteosarcomatous, rhabdomyosarcomatous, and chondrosarcomatous foci ( Fig. 20.55 ).
Evaluation of Tumor Cell Necrosis
Of the three principal factors, tumor necrosis is considered by some to carry the greatest weight in the determination of malignancy ( Figs. 20.56 to 20.58 ). Malignancy-associated tumor necrosis has to be carefully delineated and distinguished from benign degenerative changes and other therapeutic effects. The elements that should be considered in evaluating nonviable tissue in a uterine smooth muscle tumor are listed in Table 20.4 and are compared with degenerating leiomyoma (see later, “ Histologic Variants of Leiomyoma ”). Tumor necrosis associated with malignancy is described as being geographic because the contour of the interface with viable tumor is often irregular and undulating or angulated, like islands on a map (see Fig. 20.56 ). The transition from viable to necrotic tumor is sharp and on the scale of a few cells (see Fig. 20.57 ). In contrast to benign degeneration resulting from ischemia, malignant tumor necrosis is not associated with any inflammatory or repair reaction. In addition, atypical, so-called ghost cells—necrotic cells in which cell borders and cytologic atypia are still apparent—may be seen in malignant tumor necrosis (see Fig. 20.58 ). Any of these features may herald the diagnosis of malignancy.
|Malignant Geographic Tumor Necrosis||Benign|
|Multifocal||Single, often central|
|Irregular, maplike or island-like contour||Smooth, rounded contour|
|Sharp interface||Ill-defined interface|
|Atypical “ghost” cells||Bland eosinophilic cells without sharp outlines|
|Inflammation uncommon||Inflammatory response at the interface|
|Fibroblastic repair at the interface uncommon||Peripheral fibrosis or central mummifications|
Practical Diagnostic Strategy
The diagnosis of leiomyosarcoma requires the evaluation of multiple individual criteria, as described earlier. The following approach, by integrating the aforementioned microscopic observations, may be helpful in diagnostic deliberations ( Table 20.5 ). The diagnosis of leiomyosarcoma is made when either of the following conditions is satisfied: (1) geographic tumor necrosis is definitely present; or (2) both proliferative activity equal to 10 or more mitotic figures/10 hpf and moderate to severe, diffuse, or multifocal atypia are present.
|Diagnosis||Geographic Tumor Necrosis||Mitotic Rate: Mitoses/10 hpf||Atypia|
|Leiomyosarcoma||Present||Any rate||Present or absent|
|Absent||≥10||Diffuse or multifocal; moderate to severe|
|Smooth muscle tumor of uncertain malignant potential (STUMP)||Questionable||Any rate||Present or absent|
|Absent||Approaching, but less than 10||Diffuse or multifocal; moderate to severe|
|Atypical leiomyoma||Absent||≤10||Diffuse or multifocal; moderate to severe|
|Leiomyoma with increased mitotic activity||Absent||≤15||Absent|
Epithelioid differentiation in uterine leiomyosarcomas, characterized as rounded, nonspindled tumor cells with abundant eosinophilic cytoplasm, may be focal or uniform ( Fig. 20.59 ). When prominent, specific recognition as an epithelioid leiomyosarcoma is appropriate. These tumors have the other features expected in nonepithelioid leiomyosarcomas. In particular, mitotic activity in excess of five mitotic figures/10 hpf or geographic tumor necrosis is indicative of malignancy (see Table 20.1 ). Other features of malignancy in epithelioid smooth muscle tumors include nuclear atypia (particularly if severe), larger tumor size, and vascular invasion. The prediction of aggressive behavior may be improved if a combination of histologic features is considered. Tumors with a histologic phenotype intermediate between clearly benign and malignant epithelioid smooth muscle tumors, or atypical tumors in which the epithelioid phenotype is suspected but cannot be established with confidence, should be classified as tumors of uncertain malignant potential.
Epithelioid smooth muscle tumors may have reduced expression of smooth muscle markers (e.g., desmin, h-caldesmon). De Leval et al. have found that the application of another marker of smooth muscle differentiation, histone deacetylase 8 (HDAC8), may be helpful in the proper histophenotypic classification of such cases.
Myxoid leiomyosarcoma is a rare variant of uterine leiomyosarcoma. Its macroscopic appearance is that of a gelatinous, often large tumor. The criteria for diagnosis of this variant must be adjusted relative to nonmyxoid leiomyosarcoma because the increased extracellular matrix reduces cellularity and consequently counts of mitotic activity. Once the myxoid nature of the extracellular matrix has been appreciated, a malignancy can be diagnosed by recognizing significant (moderate to severe) atypia or necrosis in combination with a seemingly low proliferative rate ( Fig. 20.60 ). Atkins et al. have suggested that a rate in excess of 2 mitoses/10 hpf, regardless of the presence or absence of atypia or necrosis, will separate benign and malignant myxoid tumors. Another important diagnostic feature is infiltration into the adjacent myometrium ( Fig. 20.61 ). For practical purposes, we diagnose myxoid leiomyosarcoma in patients with any one of the following features: (1) more than 2 mitoses/10 hpf; (2) significant cytologic atypia (as outlined earlier); (3) tumor cell necrosis; and (4) destructive infiltration of adjacent myometrium. Observation of vascular invasion also may be helpful, but it must be distinguished from myxoid IV leiomyomatosis. Mitotic activity more than 10 mitoses/10 hpf may portend a fatal outcome. An inflammatory myofibroblastic tumor (IMT) should be considered in the differential diagnosis of myxoid uterine sarcomas and may be excluded by the absence of immunohistochemical anaplastic lymphoma kinase (ALK) staining or ALK gene rearrangement by FISH.
Prognostic Factors and Therapeutic Decision Making
Leiomyosarcoma has a poor prognosis. The 5-year survival has been estimated to be only 30% to 40%. Even in early-stage tumors (stages I and II), the recurrence rate for leiomyosarcoma may be as high as 71%. Extrauterine disease (i.e., high stage) at the time of diagnosis is the most potent predictor of survival. In stages II to IV patients, 5-year survival falls to 8%. The other clinical prognostic factor that has been noted is age. Consequently, some believe that the goal of therapy for patients with advanced or recurrent disease should be palliative, although others have optimistically noted that the outlook for such patients has improved since the late 1990s.
The International Federation of Gynecology and Obstetrics (FIGO) system for uterine leiomyosarcoma staging is provided in Table 20.6 , a modification of that used for endometrial adenocarcinomas. Recognizing that this extrapolation might not be optimal, two groups of investigators have compared the FIGO system to the American Joint Committee on Cancer (AJCC) soft tissue sarcoma staging system. However, neither the FIGO nor AJCC systems was entirely satisfactory, and both groups called for a new system based on a better understanding of leiomyosarcoma pathobiology.
|Leiomyosarcomas and Endometrial Stromal Sarcomas a|
|I||Tumor limited to uterus|
|II||Tumor extends beyond the uterus, within the pelvis|
|IIB||Involvement of other pelvic tissues|
|III||Tumor invades abdominal tissues (not just protruding into the abdomen)|
|IIIB||More than one site|
|IIIC||Metastasis to pelvic or para-aortic nodes|
|IV||Tumor extends beyond the abdomen|
|IVA||Tumor invades bladder or rectum|
|I||Tumor limited to uterus|
|IA||Tumor limited to endometrium and endocervix, with no myometrial invasion|
|IB||Less than or equal to half of the myometrial invasion|
|IC||More than half of the myometrial invasion|
|II||Tumor extends beyond the uterus, within the pelvis|
|IIB||Involvement of other pelvic tissues|
|III||Tumor invades abdominal tissues (not just protruding into the abdomen)|
|IIIB||More than one site|
|IIIC||Metastasis to pelvic or para-aortic nodes|
|IV||Tumor extends beyond the abdomen|
|IVA||Tumor invades bladder or rectum|
Prognostic Importance of Mitotic Activity
A number of histologic and immunohistochemical features have been investigated as prognostic factors in uterine leiomyosarcoma. Of these, the most important is mitotic count because the mitotic count is the prognostic factor identified by the greatest number of studies and is accessible to all pathologists. Related to mitotic activity, Peters et al. have shown that an increasing S phase fraction predicts an adverse outcome. The picture for the Ki-67 antigen, a widely used immunohistochemical surrogate for counting mitotic figures, is far less clear. In one study, Shpitz et al. found that all intermediate- or high-grade leiomyosarcomas highly expressed Ki-67 antigen ; despite this observation, Ki-67 expression did not clearly predict outcome. In another study, Layfield et al. found that Ki-67 expression dichotomized the tumors into prognostic groups but performed no better than a histologic diagnosis. Mayerhofer et al. have suggested that Ki-67 correlates with vascular space involvement and shorter, disease-free survival. High proliferative rates also may be found in benign leiomyomata (see the next section), and Ki-67 immunohistochemistry may play a limited role in the diagnosis of malignancy, particularly in small fragmented samples (e.g., uterine currettings), in which there are too few fields for a proper count. Veras et al. have tested another proliferation marker, phosphorylated histone H3; they found that this marker produces counts comparable to the H&E gold standard for leiomyoma but tends to inflate counts for leiomyosarcoma. In summary, mitotic activity remains the most important histology-based prognostic factor for uterine smooth muscle tumors and should be clearly indicated in pathologic reports diagnosing leiomyosarcoma.
Prognostic Importance of Histologic Grade, Necrosis, and Size
The significance of histologic grading or tumor necrosis as a prognostic factor has been somewhat controversial. Some studies have shown evidence that tumor grading or necrosis is predictive, but most studies have not. Although we include a description of the tumor in our reports, which includes characterization of the atypia and necrosis, we do not routinely grade the histology based on nuclear atypia or necrosis. Tumor size smaller than 5 cm also may be associated with longer survival.
Role of Biomarkers in Prognostication
The next immunohistochemical marker most commonly considered in uterine smooth muscle neoplasia after Ki-67 is the tumor suppressor p53. p53 is a transcription factor that plays a role in the response to DNA damage and is found mutated or overexpressed in a wide variety of tumors; it is not found by immunohistochemistry in myometrium or leiomyoma but frequently is found in leiomyosarcoma. Other studies have reported p53 expression in only about 50% of leiomyosarcomas. In contrast to epitope abundance, the p53 transcript level was not significantly altered in leiomyosarcoma compared with leiomyoma or myometrium. Blom et al. have suggested that p53 overexpression detected by immunohistochemistry may predict a high risk of recurrence in early-stage uterine leiomyosarcomas. Layfield et al. have reported that p53, like Ki-67, divides uterine smooth muscle tumors into prognostic groups, but does not achieve the level of significance achieved by the mitotic rate. Although it is likely that p53 contributes to the pathobiology of leiomyosarcoma, the utility of this prognostic marker has not clearly been established.
p16, also known as cyclin-dependent kinase inhibitor 2A, INK-4, and multiple tumor suppressor 1, has drawn interest as a candidate marker for smooth muscle malignancy. This gene regulates two important cell cycle control pathways—namely, the p53 and retinoblastoma (RB1) pathways. Because both pathways appear to contribute to leiomyosarcoma tumor biology, consideration of p16 was reasonable. Skubitz and Skubitz found p16 overexpression in leiomyosarcoma when they profiled gene expression in myometrium, leiomyoma, and leiomyosarcoma. Such overexpression of a tumor suppressor appears paradoxic at first glance, but most likely represents a compensatory (homeostatic) response to promitogenic signaling within the cell, much like the relationship between p16 and oncogenic papillovirus in cervical epithelium. Bodner-Adler et al. and other groups have suggested that detection of p16 overexpression could be used to aid in the diagnosis of problematic uterine smooth muscle tumors. The results, however, have been mixed; Gannon et al. and O’Neill et al. have found that p16 expression could be used to distinguish leiomyoma variants and smooth muscle tumors of uncertain malignant potential (STUMPs). In contrast, Bodner-Adler et al. have found that p16 is expressed in 21% of STUMPs and 57% of leiomyosarcomata. Finally, Atkins et al. have found strong diffuse expression in 37.5% of STUMPs overall, but the frequency increases to 66% when counting only STUMPs that have metastasized. From these data and the pathobiologic context—that is, p16 has not yet been shown to be required for malignant smooth muscle neoplasia—we have concluded that p16 may be helpful in adjusting the level of concern about a particular atypical smooth muscle tumor, but the diagnosis of malignancy should not rest solely on this biomarker.
A small number of other genes have been investigated as prognostic markers, including c-Kit, CD-34 (a marker of microvessel density), MDM-2 (a p53 binding protein), BCL-2 (an antiapoptosis factor), estrogen receptor 1 (ESR1), progesterone receptor, vascular endothelial growth factor (VEGF), telomerase, NM-23, fascin, and colony-stimulating factor 1 (CSF1). In our opinion, none of these markers is currently suitable for widespread clinical application.
Immunohistochemistry for FH may be helpful in the evaluation of atypical smooth muscle tumors (see next section for additional details).
The treatment of choice for leiomyosarcoma is widely considered to be total abdominal hysterectomy and bilateral salpingo-oophorectomy. The likelihood of nodal disease is low in the absence of more obvious disease, suggesting that lymph node dissection has little prognostic or therapeutic role in the management of leiomyosarcoma. Ovarian preservation may be considered in premenopausal patients with early-stage leiomyosarcoma, but has yielded little benefit in most studies. Surgical treatment of isolated pulmonary metastasis may prolong disease-free survival in patients for whom adequate local control of the primary tumor has been achieved. Adjuvant chemotherapy or radiotherapy has a minimal, if any, effect on survival. a
a References .
Aromatase inhibitors may prolong progression-free survival for some patients with advanced, estrogen receptor–positive leiomyosarcomas. On an encouraging note, gemcitabine-docetaxel adjuvant chemotherapy has been shown to have some modestly positive results in several studies. As a result, patients with smooth muscle tumors of uncertain malignant potential are not currently candidates for adjuvant therapies. The advent of a highly efficacious nonsurgical therapy for leiomyosarcoma, however, will have a major impact on the pathologic diagnosis of uterine smooth muscle tumors.
Histologic Variants of Leiomyoma
Each of the pathologic parameters used in the determination of malignancy in uterine smooth muscle tumors has a benign counterpart. To diagnose a histologic variant, one must be sure that none of the other features of malignancy are present. These variants are discussed in the following sections and listed in Table 20.7 . Finally, the approach to tumors with more than one feature of malignancy, yet not satisfying the criteria for malignancy, will be considered.