Sex Cord-Stromal, Steroid Cell, and Other Ovarian Tumors with Endocrine, Paraendocrine, and Paraneoplastic Manifestations
The sex cord-stromal tumor category includes all ovarian neoplasms that contain granulosa cells, fibroblasts, theca cells (and their luteinized derivatives), Sertoli cells, Leydig cells, singly or in various combinations and in varying degrees of differentiation. Those who believe that all these cell types are derived from the “specialized stroma” of the genital ridge favor the term gonadal stromal tumors for these neoplasms (Norris and Taylor, Cancer (Phila) 21:255–263, 1968). Others, recognizing that many embryologists favor the participation of coelomic and mesonephric epithelium in the formation of sex cords, which are the proximal precursors of granulosa cells and Sertoli cells, favor the term sex cord-stromal tumors, and it is now the most widely accepted designation (Scully, Young, and Clement, Tumors of the ovary, maldeveloped gonads, fallopian tube, and broad ligament. In: Atlas of tumor pathology, 3rd series, fasc 23. Armed forces Institute of Pathology, Washington, DC, 1998). The chapter also covers relevant aspects of a wide variety of other ovarian tumors associated with endocrine function and paraneoplastic syndromes.
Sex Cord-Stromal Tumors
This category includes all ovarian neoplasms that contain granulosa cells, fibroblasts, theca cells (and their luteinized derivatives), Sertoli cells, Leydig cells, singly or in various combinations and in varying degrees of differentiation. Those who believe that all these cell types are derived from the “specialized stroma” of the genital ridge favor the term gonadal stromal tumors for these neoplasms (Norris and Taylor 1968). Others, recognizing that many embryologists favor the participation of coelomic and mesonephric epithelium in the formation of sex cords, which are the proximal precursors of granulosa cells and Sertoli cells, favor the term sex cord-stromal tumors, and it is now the most widely accepted designation (Scully et al. 1998).
In the developing testis, the sex cords are clearly distinguishable by the fifth week of embryonic life as slender columns of primitive Sertoli cells, but similar cords, at least in the sense of thin columns, are not encountered in the developing ovary; instead, packets of small pregranulosa cells enveloping germ cells become evident later in embryonic life. For that reason, the term sex cords has been criticized as inaccurate to describe the progenitors of granulosa cells. Nevertheless, the long-established usage of this designation by embryologists and the lack of a better term justify its retention. The term sex cord-stromal tumors has the advantage of acknowledging the presence of neoplasms in this general category of derivatives of either or both the sex cords and the stroma. The components derived from the sex cords (granulosa and Sertoli cells) typically are arranged in epithelial configurations, whereas those derived from the stroma have the appearance of cellular gonadal stroma or its specialized derivatives, the theca and Leydig cells.
Most sex cord-stromal tumors (granulosa-stromal cell tumors) are composed of ovarian cell types, but some (Sertoli-stromal cell tumors) contain cells that differentiate toward patterns more typical of the testis. Occasional tumors have significant components which in isolation would be in the granulosa-stromal or Sertoli-stromal categories and the term gynandroblastoma has been historically used for such cases. Although this heading is used in a section here, in our own practice we prefer to categorize such neoplasms as mixed sex cord-stromal tumors and itemize the components present, and their quantity, roughly, in accord with the manner of dealing with mixed germ cell tumors of the gonads. This gives the clinician more cogent information than the gynandroblastoma terminology. When the neoplastic cells are immature and their appearance is intermediate between those of testicular and ovarian cell types, or when the architectural patterns of the tumor are not specific for either the testis or ovary, it may be impossible to determine whether the tumor belongs in the granulosa or Sertoli-stromal cell category; in such cases the term sex cord-stromal tumor, not otherwise specified, is used.
World Health Organization 2014 classification of sex cord-stromal tumors
Pure Stromal Tumors
Luteinized thecoma associated with sclerosing peritonitis
Sclerosing stromal tumor
Signet-ring stromal tumor
Microcystic stromal tumor
Leydig cell tumor
Steroid cell tumor
Steroid cell tumor, malignant
Pure sex cord tumors
Adult granulosa cell tumor
Juvenile granulosa cell tumor
Sertoli cell tumor
Sex cord tumor with annular tubules
Mixed sex cord-stromal tumors
Sertoli-Leydig cell tumors
With heterologous elements
With heterologous elements
With heterologous elements
Sex-cord stromal tumors, NOS
Within the overall category of tumors considered in this chapter, one finds some of the most mundane diagnoses in ovarian tumor pathology, such as a typical fibroma, and on the other hand, a great array of other cases in which diverse patterns can bring almost any other ovarian tumor into the differential diagnosis. Fundamental to the approach to evaluating the more difficult cases is a broad knowledge of the great diversity of cell types and patterns seen in ovarian tumors and the overlap that exists between various categories (Young and Scully 2001). As difficult as some of these cases can be, it should also be noted that in many cases, one of the most crucial aspects of ovarian tumor evaluation, thorough sampling, will uncover foci that point firmly to the correct diagnosis. Accordingly, in an age when there is often a rush to apply immunohistochemistry, the importance of exceedingly thorough sampling of challenging cases cannot be overemphasized. The sampling must be based not only on the size of the tumor but also its overall gross characteristics and differing features thereof, and, of course, the differential diagnosis engendered when the initial sections are evaluated. Although in certain circumstances these should point to judicious application of immunohistochemistry, they should also often cause reflection as to whether even further sections might give clues to the diagnosis. Although immunohistochemistry can be exceedingly helpful, the extent to which it is done in the current era is not matched by the aid actually provided from the viewpoint of arriving at the correct diagnosis. In some cases, it is also confusing, such as when a negative inhibin stain is obtained in a case of granulosa cell tumor, not a rare event.
Apart from thorough sampling, other fundamentals of ovarian tumor evaluation such as knowledge of the clinical background and awareness of the age of the patient should never be overlooked, albeit they are by no means always significant. The diagnostic formulation is impacted if the patient is known to have had a deeply invasive melanoma a few years ago; such tumors metastatic in the ovary can simulate a malignant steroid cell tumor or both adult and juvenile granulosa cell tumors quite closely. From the viewpoint of the age, clearly one approaches an ovarian tumor with a pattern of slit-like spaces and papillae differently in the first 10 years of life compared to in the seventh decade; in the first situation, a retiform pattern of Sertoli–Leydig cell tumor is high on the list, whereas of course in the older patient, one is most likely looking at another routine serous neoplasm. Most of the tumor types discussed in this chapter have a very high rate of unilaterality, which is often helpful in their differential diagnosis, particularly with secondary tumors of the ovary.
The above emphasis on thorough sampling, awareness of age, and clinical history notwithstanding, there certainly is a role for immunohistochemistry in this area, although the extent to which it will be called upon will inevitably vary depending upon the experience of the evaluating pathologist with the spectrum of morphology of these tumors in general. Since Dr. Robert E. Scully first wrote an essay on the application of immunohistochemistry in ovarian tumor pathology in 1985 (Scully 1985), there has been a vast literature on this topic. No attempt is made here to be complete from the viewpoint of citation and many excellent papers are not cited. Periodic reviews of this topic appear and three of those are cited here (Baker and Oliva 2004; McCluggage 2008; McCluggage and Young 2005) as are a selected group of peer-reviewed contributions (McCluggage 2008; Cao et al. 2001; Cathro and Stoler 2005; Costa et al. 1992, 1997; Deavers et al. 2003; Flemming et al. 1995; Guerrieri et al. 1998; Kommoss et al. 1998; Matias-Guiu et al. 1998; Pelkey et al. 1998; McCluggage and Maxwell 1999, 2001; McCluggage et al. 1997, 2007; Movahedi-Lankarani and Kurman 2002; Oliva et al. 2007; Riopel et al. 1998; Rishi et al. 1997; Stewart et al. 1997, 2000; Zhao et al. 2006, 2007a, b, 2008, 2009; Zheng et al. 1997; Al-Agha et al. 2011; He et al. 2008). Some of these deal with now well-known applications such as that of inhibin and calretinin, but a few deal with more recently explored markers, whose benefit remains to be fully characterized with ongoing experience.
The most helpful triad that currently exists in evaluating sex cord-stromal tumors and their mimics is that of inhibin, calretinin, and epithelial membrane antigen (EMA). The first two are typically positive in sex cord tumors and the third, typically negative (Aguirre et al. 1989a). However, routine evaluation still dominates in arriving at the correct diagnosis and an appreciable minority of typical granulosa cell tumors in our experience are negative for inhibin or calretinin. Also, rare carcinomas are positive for inhibin (McCluggage and Maxwell 1999). If a tumor is negative for inhibin and calretinin and positive for EMA, that it is a granulosa cell tumor is close to unheard of. Beyond the adult granulosa cell tumor, it is important to note that the different tumors within the sex cord stromal category express inhibin and calretinin to varying degrees; indeed some tumors, such as microcystic stromal tumor, are characterized by absence of staining for inhibin and calretinin, as well as expression of certain markers not seen in other sex cord-stromal tumors. More recently described sex cord markers such as SF-1, FOXL2, and CD56 may be helpful in some instances in confirming the sex cord-stromal nature of these tumors, although the literature on all of these newer markers is relatively limited (Al-Agha et al. 2011; He et al. 2008; Zhao et al. 2009). It should be remembered that the luteinized stromal cells of many ovarian tumors may be positive for the various sex cord markers, and this occasionally causes confusion if the lack of staining of epithelial cells of the neoplasm is overlooked.
Although immunohistochemistry has largely supplanted electromicroscopy as a diagnostic aid in ovarian tumor interpretation, there is still an occasional role for ultrastructural evaluation in showing, for example, dense core granules in tumors of neuroendocrine type and even Charcot–Bottcher filaments in some Sertoli cell tumors. From the converse viewpoint, ultrastructural evaluation may show, for example, in malignant melanoma, distinctive features of that neoplasm ruling out a sex cord-stromal or steroid cell neoplasm.
Until recently, the molecular pathogenesis of sex cord-stromal tumors has been largely unknown. However, significant recent progress has been made in this regard, with identification of a mutation in FOXL2 in most adult granulosa cell tumors, and DICER1 in a majority of Sertoli-Leydig cell tumors. These findings are discussed in detail in the sections below. These discoveries raise hope for a better molecular understanding of the sex cord-stromal tumors to complement the intricate morphology-based classification that has been developed, and perhaps even the prospect of targeted therapy for the malignancies discussed below.
Pure Sex Cord Tumors
This category includes all ovarian tumors composed of epithelial-type sex cord cells. These may differentiate along female sex cord lineage, resulting in granulosa cell tumors, or along male lineage, forming Sertoli and Sertoli-Leydig cell tumors. The lineage of the sex cord tumor with annular tubules is uncertain. Although the World Health Organization uses the term “pure,” a component of theca cells is a common finding in granulosa cell tumors. Granulosa cell tumors that occur typically in middle-aged and older women differ in several important respects from those that usually arise in children and young adults and these two subtypes, which are referred to as adult and juvenile granulosa cell tumors, are discussed separately. It should be noted that the designations “adult” and “juvenile” are terms of convenience to capture the fact that the morphology seen under those headings is typically seen in adults and juveniles as the case may be. However, there are exceptions, some adult tumors being seen in young patients, even children, and some juvenile tumors being seen in the middle to later years of life. The adult and juvenile designations, however, are appropriate terms of convenience, as the many differences between adult and juvenile granulosa cell tumors would be difficult to encapsulate in other than extremely lengthy descriptive designations. The current classification has the curious result of separating Sertoli cell tumors from Sertoli-Leydig cell tumors, whereas prior classifications grouped them in the Sertoli-Stromal tumor category. However, these tumors have much in common morphologically and presumptively have a close relationship; the finding of the DICER1 mutation in both tumors supports their link (Conlon et al. 2015).
Adult Granulosa Cell Tumor
Adult granulosa cell tumors account for approximately 1% of all ovarian tumors and 95% of all granulosa cell tumors. They occur more often in postmenopausal than in premenopausal women, with a peak incidence between 50 and 55 years of age (Norris and Taylor 1968; Bjorkholm 1980; Bjorkholm and Silfversward 1981; Sjostedt and Wahlen 1961; Stenwig et al. 1979). They are the most common estrogenic ovarian tumors clinically, but the precise proportion of adult granulosa cell tumors that secrete hormones is difficult to establish because a specimen of endometrium to evaluate the effects of estrogenic stimulation often is unavailable. The typical endometrial alteration associated with functioning tumors in this category is simple hyperplasia, usually exhibiting some degree of precancerous atypicality. Carcinoma of the endometrium, which almost always is well differentiated, has been reported in from slightly less than 5% to slightly more than 25% of the cases; the wide variation in these figures is attributable, at least in part, to differing views of the dividing line between complex atypical hyperplasia and grade 1 adenocarcinoma. If strict criteria for the diagnosis of carcinoma are used, and if all patients with a granulosa cell tumor, not only those who have had an endometrial curettage or hysterectomy, are considered, the best estimate for the frequency of associated endometrial carcinoma is under 5%.
The endometrial changes associated with adult granulosa cell tumors are manifested clinically in women in the reproductive age group by irregular, excessive uterine bleeding, but amenorrhea, lasting from months to years, may precede the abnormal bleeding or may be the only hormonal manifestation. Postmenopausal bleeding is the most common endocrine symptom in older women, in whom carcinoma of the endometrium is encountered about twice as often as in younger patients. Occasionally, swelling and tenderness of the breasts are prominent symptoms. Elevated levels of estrogens have been reported in the blood and urine, and vaginal cytologic smears typically show increased maturation of squamous epithelial cells. Alterations resembling those seen in a secretory endometrium have been observed rarely in association with granulosa cell tumors, suggesting the possibility of a significant production of progesterone in these cases (Young et al. 1994a).
Rarely, androgenic changes are the sole endocrine manifestation of an adult granulosa cell tumor (Nakashima et al. 1984; Norris and Taylor 1969). Most of the patients have been frankly virilized, but some have been only hirsute. The tumors may be solid or solid and cystic. The cysts typically are thin walled and may be single or multiple, resembling serous cystadenomas. Because granulosa cell tumors in general are composed exclusively of thin-walled cysts in only about 3% of the cases, the almost 50% frequency of a cystic gross appearance of tumors associated with androgenic manifestations is of interest but remains an enigma (Nakashima et al. 1984).
Hemoperitoneum occurs in about 10% of cases of granulosa cell tumor, and an acute abdominal presentation is accordingly more typical of that tumor than other forms of ovarian cancer.
The trabecular (Fig. 8) and insular (Fig. 11) patterns are characterized by bands and islands of granulosa cells, separated by a fibromatous or thecomatous stroma. In the solid tubular pattern, the tubules may be uniformly cellular or contain peripheral nuclei and central masses of cytoplasm; occasionally, a few hollow tubules or gland-like structures are encountered. The various tubular patterns encountered in granulosa cell tumors are indistinguishable from those of well-differentiated Sertoli cell tumors; their presence is ignored as a diagnostic criterion unless they account for a significant portions of the tumor (10% or more); in such cases, a diagnosis of mixed granulosa cell and Sertoli cell tumor, or gynandroblastoma, is warranted. Other patterns of granulosa cell tumor are watered silk (moiré silk), gyriform (Fig. 13), and pseudopapillary (Irving and Young 2008a). The first two patterns are manifested by undulating or zigzag rows of granulosa cells, generally in single file. The third is likely a degenerative phenomenon.
The microfollicular pattern, which only rarely predominates, is characterized by the presence of numerous small cavities simulating the Call–Exner bodies of the developing graafian follicle (Fig. 14). These cavities may contain eosinophilic fluid and often one or a few degenerating nuclei, hyalinized basement membrane material, or rarely basophilic fluid. The microfollicles are separated typically by well-differentiated granulosa cells that contain scanty cytoplasm and pale, angular, or oval, often grooved nuclei arranged haphazardly in relation to one another and to the follicles. The macrofollicular pattern of the granulosa cell tumor (Fig. 15) is uncommon and characterized by cysts lined by well-differentiated granulosa cells, beneath which theca cells often are present.
The neoplastic granulosa cells typically have scant cytoplasm and pale nuclei, some of which have nuclear grooves (Fig. 18). The pallor and frequency of nuclear grooves is variable from region to region and case to case and can be impacted amongst other things by vagaries of fixation and staining. An exception to the characteristic nuclear features described above is seen in approximately 2% of granulosa cell tumors that contain bizarre, enlarged hyperchromatic nuclei, including multinucleated forms (Young and Scully 1983a) (Figs. 19 and 20). Cells with these nuclear features typically are a focal finding but rarely are conspicuous and may overshadow more characteristic foci, which can be overlooked if not carefully sought.
The mitotic rate in the tumors is variable. It is generally not brisk, but there are exceptions; however, in cases with a brisk mitotic rate, particular care should be made to make sure one is not misinterpreting a mimic of a granulosa cell tumor as that neoplasm. In some granulosa cell tumors, the neoplastic cells contain abundant dense or vacuolated cytoplasm, approaching, to varying degrees, the appearance of the granulosa cells of the corpus luteum; in such cases, the term luteinized granulosa cell tumor is appropriate. Rarely, the neoplastic granulosa cells have a signet-ring cell morphology, more typical of the signet-ring stromal tumor considered later. Exceptionally, a granulosa cell tumor undergoes sarcomatous transformation (Susil and Sumithran 1987) or transforms into an anaplastic carcinoma (Scully et al. 1998). A handful of granulosa cell tumors have been intimately associated with a mucinous cystic tumor (McKenna et al. 2005; Price et al. 1990; Staats et al. 2010).
The prominence and nature of the stromal component in granulosa cell tumors vary considerably. In some cases, usually those with a diffuse pattern, it is essentially absent. In tumors in which the granulosa cells form nests or trabeculae, the stroma is usually conspicuous and composed of fibroblasts and theca cells that contain appreciable to abundant eosinophilic or vacuolated lipid-rich cytoplasm. The stroma is often quite vascular (Fig. 9), and this may be striking on low-power examination. Rarely, crystals of Reinke are present in the lutein-type cells of a granulosa cell tumor, enabling them to be specifically identified as Leydig cells (Ahmed et al. 1999). Another rare component of the stromal compartment of granulosa cell tumors is hepatic-type cells (Ahmed et al. 1999; Nogales et al. 1993) (Fig. 21). These cells are typically larger and have denser eosinophilic cytoplasm than lutein or Leydig cells. Bile pigment has been detected in canaliculi between some of the hepatic-type cells in three cases. The hepatic-type cells have been immunoreactive for epithelial membrane antigen (EMA), carcinoembryonic antigen (CEA), which stained canaliculi between the cells, and CAM 5. In contrast to lutein or Leydig cells, the hepatic-type cells are negative for inhibin (Ahmed et al. 1999).
The presence of theca cells in varying quantities in most granulosa cell tumors has led to the occasional usage of the term granulosa–theca cell tumor. Although this designation accurately describes the cellular content of many of these neoplasms, the term granulosa cell tumor is more widely accepted for tumors containing both cell types. One reason for this preference is the probability that the presence of theca cells in some cases reflects a response of the ovarian stroma to the growth of granulosa cells rather than the coexistence of a second neoplastic cell component. Evidence favoring such an interpretation includes the nonspecific presence of theca-like cells in a variety of ovarian tumors, both benign and malignant and both primary and metastatic, and the observation that theca cells usually are absent in granulosa cell tumors that have extended beyond the ovary (Fathalla 1968). It is possible, however, that some tumors in which the theca cell element is prominent or even greatly preponderant are truly mixed neoplasms. The theca cells in granulosa cell tumors may resemble theca externa or theca interna cells and may be luteinized. In some tumors, particularly those with a diffuse pattern, differentiation of granulosa and theca cells with routine staining may be difficult or impossible. In such cases, a reticulum stain may be helpful. Just as in a developing graafian follicle, so in a granulosa cell tumor, the fibrils typically invest theca cells individually. In contrast, the granulosa cell layer of a follicle contains no fibrils, and in a granulosa cell tumor, the reticulum usually is sparse, being typically confined to perivascular zones.
The presence of blood-filled cysts in many granulosa cell tumors results in the frequent presence in the tumors of related changes that are nonspecific, but in the context of other typical features of granulosa cell tumors are at the same time quite characteristic. For example, the cysts often are lined by fibrous tissue associated with evidence of old and recent hemorrhage that is sometimes conspicuous.
Several histochemical reactions that are characteristic of steroid hormone-producing cells, particularly those that demonstrate various types of lipid content or oxidative enzyme activity, usually are positive in the theca cells and negative or only weakly positive in the granulosa cells of a tumor containing both cell types (Kurman et al. 1978, 1979, 1984; Scully and Cohen 1964). This finding, as well as ultrastructural observations, has led some observers to conclude that the theca cell component of granulosa cell tumors produces the hormones responsible for estrogenic manifestations. Additional evidence in favor of this conclusion is the observation that granulosa cell tumors that recur outside ovarian tissue and lack theca cells typically are not obviously estrogenic. In some cases, however, histochemical and other evidence has suggested a role for the granulosa cells in estrogen secretion (Armstrong and Papkoff 1976; Dorrington et al. 1975). The above histochemical studies are rarely indicated other than for academic interest.
Immunohistochemically, the vast majority of granulosa cell tumors are positive for inhibin and/or calretinin (Cathro and Stoler 2005; Deavers et al. 2003; Zhao et al. 2009). However, a small percentage of adult granulosa cell tumors fail to express one or both of these markers, and in the presence of typical morphology, immunohistochemical confirmation is not necessary to make the diagnosis. Newer sex cord markers such as SF-1, FOXL2, and CD56 appear to be more sensitive and may be helpful in tumors which fail to stain with the more traditional markers (Al-Agha et al. 2011; He et al. 2008; Zhao et al. 2009). However, experience with these newer markers is limited, particularly in regard to their specificity, so these markers should be interpreted carefully in conjunction with morphology and pertinent negative immunohistochemistry for other tumors in the differential diagnosis. CD56 in particular is known to be expressed in a wide range of tumors. In the distinction with surface epithelial tumors, EMA is a more reliable marker than keratins, as focal pancytokeratin and cam5.2 expression are not uncommon in adult granulosa cell tumor, whereas EMA expression is exceptional.
Misidentification of other ovarian tumors as adult granulosa cell tumors, and presumably the converse, is relatively common. A recent study that re-evaluated a group of FOXL2-mutation-negative adult granulosa tumors found that almost 20% of “adult granulosa cell tumors” were in fact other tumors that had been misclassified (McConechy et al. 2016). Given the substantial differences in prognosis and treatment between adult granulosa cell tumor and its mimics, care must be taken to be certain of the correct diagnosis.
Granulosa cell tumor versus undifferentiated carcinoma and poorly differentiated adenocarcinoma
Granulosa cell tumor
Bilateral, less than 5%
Bilateral, more than 25%
Stage I in 90% of cases
Stage III or IV in most cases
Nuclei round to angular, pale and commonly grooveda
Nuclei hyperchromatic, often bizarre with atypical mitoses
Mucin occasionally in follicles (mainly in juvenile type)
Intracellular droplets or extracellular pools of mucin, psammoma bodies, or glands may be present
Indolent course, when clinically malignantb
Sex cord markers +
Sex cord markers −
Granulosa cell tumors (GCTs) versus small cell carcinoma
Small cell carcinoma
Mostly before 30 years
All ages, but mostly postmenopausal
Occasionally malignant, often with protracted course
Thecomatous component common
Fibrothecomatous component common
Stroma scanty and nonspecific
Mucinous epithelium absent
Mucinous epithelium rare
Mucinous epithelium 12% of cases
Cytoplasm usually abundant
Cytoplasm usually scanty
Cytoplasm usually scanty, but may be abundant
Nuclei dark, ungrooved, and pleomorphic 13% of cases
Nuclei pale and often grooved
Nuclei dark, uniform, and ungrooved
Mitoses usually numerous
Sex cord markers +
Sex cord markers +
Inhibin – (but calretinin +)
AGCTs may occasionally be difficult to distinguish from pure stromal tumors such as thecomas, fibromas, and fibrosarcomas. Reticulum stains may show abundant intercellular fibrils in these tumors, unlike the scant reticulum of AGCTs. In some cases, the pattern of fibrils is intermediate between an AGCT and a thecoma, and in such cases, the differential diagnosis may be difficult or impossible. The almost exclusive spindle cell nature of the cells in fibromas and fibrosarcomas is rarely seen in an AGCT. A recently described series of adult granulosa cell tumors demonstrates the overlap that may exist with thecoma in some cases. These tumors typically have a nodular growth pattern, with cytoplasm similar to thecoma and in some cases hyaline plaques, but areas with architecture and cytoplasm more typical of granulosa cell tumor are usually present. Reticulin stains can be particularly helpful in distinguishing these granulosa cell tumors with “thecoma-like” areas from thecoma (Stall and Young 2018).
Occasionally, the distinction of a unilocular or multilocular macrofollicular AGCT from one or more follicular cysts may be troublesome; this is particularly likely if the patient is pregnant, or in the puerperium, because a large solitary luteinized follicle cyst of pregnancy and the puerperium is indistinguishable grossly from a unilocular cystic AGCT. The large luteinized cells of the former, some of which contain large bizarre nuclei, differ from those of a unilocular AGCT, which are rarely uniformly luteinized and rarely contain bizarre nuclei lining the cysts.
Endometrioid carcinomas occasionally are misdiagnosed as AGCT when the former have small acini imparting a microfollicular pattern. Also, some endometrioid carcinomas have an insular pattern that on low power may suggest the diagnosis of an insular AGCT. Another source of confusion is the presence in some endometrioid carcinomas of a focal diffuse pattern that is occasionally difficult to distinguish from the diffuse pattern of an AGCT. In thoroughly sampled tumors, there will almost invariably be various patterns of one or the other tumor which will establish the correct diagnosis. In the case of endometrioid tumors with a diffuse pattern, the focal presence of abortive squamous differentiation is a very major clue to the diagnosis. In most of these cases, the cytologic features in an endometrioid carcinoma differ from those of an AGCT, although there are rare cases in which the former tumors have pale nuclei that on cytologic grounds are consistent with a diagnosis of AGCT. In our experience, a combination of at least focal cytologic differences, and the presence in endometrioid tumors that are well sampled of other foci incompatible with the diagnosis of an AGCT, such as squamous foci, establish the diagnosis. Immunohistochemistry for sex cord markers and EMA will aid if necessary, as noted previously.
The rare AGCT in which the cells are extensively luteinized (Young et al. 1994a) may superficially resemble a steroid cell tumor. The focal presence of areas with the architectural and cytologic features of an AGCT usually facilitates the diagnosis in these cases. While both of these tumors express most of the sex cord immunohistochemical markers, steroid cell tumors are typically positive for MART1/melanA and negative for FOXL2 and WT1, whereas AGCTs typically have the opposite pattern.
Occasionally, the female adnexal tumor of probable Wolffian origin involves the ovary. The solid areas of that tumor may rarely be reminiscent of diffuse foci of adult granulosa cell tumor. The cystic (sieve-like) pattern, often with eosinophilic secretions, of the former may suggest follicles, but the overall admixture of patterns and cytology is inconsistent with granulosa cell tumor. Well-formed tubules are more common in the Wolffian tumor and indeed make Sertoli cell tumor a more realistic issue. The cells have scant cytoplasm and small, pale, monotonous nuclei, but do not have prominent nuclear grooves. The immunoprofile can be confusing, as Wolffian tumors express most sex cord markers, although inhibin is focal and weak. Diffuse positivity for CD10 is the norm for Wolffian tumors, whereas expression is only weak and focal in most granulosa cell tumors.
AGCTs should be distinguished from the small proliferations of granulosa cells within atretic follicles that are typically an incidental finding within the ovaries of pregnant women. Another rare problem in differential diagnosis is the distinction between a granulosa cell tumor, usually a luteinized one, and an epithelioid smooth muscle tumor. The latter diagnosis does not tend to be considered as quickly, not unreasonably, when examining an ovarian tumor as it does when evaluating a uterine one. This is an area in which immunohistochemistry, specifically strong desmin staining and negative staining for inhibin and calretinin, may play a major role.
Granulosa cell tumor versus carcinoid
Granulosa cell tumor
Variety of patterns
Islands, round acini, solid tubules, and ribbons
Call–Exner bodies, ill-defined, with watery to dense eosinophilic content, occasionally pyknotic nuclei
Acini sharply outlined with dense content, sometimes calcified
Nuclei round to angular, pale, often grooved, haphazardly oriented
Nuclei round with coarse chromatin and regular orientation
Thecomatous stroma common, at least focally
Fibromatous or hyalinized stroma, may be focally luteinized
Usually uninodular and almost always unilateral; no teratomatous elements
Often multinodular and almost always bilateral if metastatic; always unilateral and usually associated with other teratomatous elements if primary
Cells nonargentaffin; may contain fine argyrophilic granules
Cells usually argentaffin; almost always argyrophilic
Sex cord markers +
Synaptophysin, chromogranin −
Sex cord markers −
Synaptophysin, chromogranin +
The AGCT may be confused with two types of metastatic tumors: metastatic malignant melanoma and metastatic breast carcinoma. Metastatic melanomas may have cells with scant cytoplasm that grow diffusely, imparting a low-power appearance that may closely mimic that of an AGCT. Patterns of growth incompatible with a diagnosis of AGCT usually are found when a tumor is thoroughly sampled, and the finding of melanin pigment or immunohistochemical stains (S-100 or other melanoma markers) may be helpful in problematic cases. Metastatic breast carcinoma sometimes also has a diffuse growth of cells with scant cytoplasm, particularly in cases of lobular carcinoma. The history often is helpful in these cases because breast carcinoma rarely presents with an ovarian metastasis. In cases in which a breast cancer is not known in the patient, it is only the presence of focal patterns more suggestive of breast cancer than a granulosa cell tumor and a lack of the typical cytologic features of granulosa cell tumor that will alert the pathologist to the possible correct diagnosis. In both this situation and that of metastatic melanoma, the ovarian metastatic tumors are much more frequently bilateral than is the AGCT. Immunohistochemical staining for EMA and breast-specific markers such as mammoglobin, GATA3, or gross cystic disease fluid protein 15 can be helpful in making the diagnosis of metastatic breast carcinoma; however, the latter markers have not been well characterized in adult granulosa cell tumor.
Recently, it has been recognized that somatic mutation of FOXL2, a gene encoding a transcription factor critical for granulosa cell development, is present in the vast majority of adult granulosa cell tumors. The first such study found the mutation in 86 of 89 (97%) of adult granulosa cell tumors (Shah et al. 2009). Interestingly, the mutation was detected in only one of ten juvenile granulosa cell tumors (10%), leading the authors to speculate that the juvenile granulosa cell tumor is a distinct disease from the adult type. Mutant FOXL2 was found in 3 of 14 thecomas (21%) and was absent in 49 sex cord tumors of other types, including Sertoli–Leydig tumors, fibromas, and steroid cell tumors, and in 329 unrelated ovarian and breast tumors. The findings suggest that mutant FOXL2 is a potential driver in the pathogenesis of adult-type granulosa tumors (Shah et al. 2009). Most subsequent studies have found similarly high rates of mutation in AGCT and low rates in other sex cord-stromal tumors. A recent study found that after exclusion of misdiagnosed tumors, the rare true adult granulosa cell tumors that are FOXL2 wild-type (6% in this study) have identical morphology and appear to have a similar prognosis to those with the mutation (McConechy et al. 2016).
It should be noted that nuclear immunohistochemical staining for FOXL2 is seen in most granulosa cell tumors, even those without FOXL2 mutation, and also in a variety of sex cord-stromal tumors which lack the FOXL2 mutation; while it appears to be a useful immunohistochemical marker of sex cord-stromal differentiation, positive staining is not evidence of a FOXL2 mutation.
Clinical Behavior and Treatment
After the removal of a granulosa cell tumor, the manifestations of hyperestrinism typically regress. If the uterus has been conserved in a young woman, estrogen withdrawal bleeding usually occurs in 1 or 2 days, and regular menses ensue shortly thereafter. Granulosa cell tumors of all patterns have a malignant potential, with a capacity to extend beyond the ovary or recur after apparently successful removal. Spread is largely within the pelvis and lower abdomen; distant metastases are rare but have been reported in many sites. Although recurrences may appear within 5 years, they often are not evident until a much longer postoperative interval has elapsed, and numerous cases have been reported in which the tumor has reappeared two or even three or more decades after the initial therapy. The 10-year survival figures that have been recorded in the literature have varied widely from less than 60% to more than 90%, and progressive declines in survival have been documented after longer follow-up periods (Bjorkholm 1980; Bjorkholm and Silfversward 1981; Stenwig et al. 1979; Bjorkholm and Pettersson 1980). Unfortunately, many old studies of granulosa cell tumor are suspect from the viewpoint of the impact of various features on prognosis, and survival figures overall, because of the almost certain inclusion in some of them of tumors that are not granulosa cell tumors, differential features between the granulosa cell tumor and its various mimics having been much more appreciated in the last several decades. In a recent study carefully excluding mimics, the 10-year overall survival was identical to the general population, and median time to recurrence was 7.2 years, rendering standard 5-year follow-up statistics largely meaningless for this tumor (McConechy et al. 2016). The authors found that 72% of disease-specific deaths originally attributed to adult granulosa cell tumor in their overall cohort were in fact due to misdiagnosed mimics. Of true adult granulosa cell tumors, only 2% died within 5 years of diagnosis, and 57% were alive at 5 years after disease recurrence, supporting the indolent nature of properly characterized adult granulosa cell tumors.
The optimal surgical treatment of a granulosa cell tumor in menopausal or postmenopausal women is total hysterectomy with bilateral salpingo-oophorectomy. In younger women in whom the preservation of fertility is an important consideration, however, removal of only the ovary and the adjacent fallopian tube is justifiable if spread beyond the ovary is not demonstrable and examination of the contralateral ovary shows no suggestion of involvement. Overall, there is insufficient evidence to confirm or deny a useful role for adjuvant therapy (Gurumurthy et al. 2014). Some recurrent tumors have been treated successfully by reoperation, radiation therapy, or a combination of these.
At least 90% of granulosa cell tumors are stage I, and these tumors have a considerably better prognosis than the higher-stage tumors, as shown by an 86% versus a 49% relative survival at 10 years, respectively, in one large series and a 96% versus a 26% survival in another (Bjorkholm and Silfversward 1981). Rupture also adversely affects the outlook, with an 86% relative 25-year survival of patients with intact stage I tumors compared with only 60% survival of those with ruptured tumors that are otherwise in the same stage (Bjorkholm and Silfversward 1981). The impact of the stage has been confirmed in a recent study (Auranen et al. 2007).
The size of granulosa cell tumors also has been related to their prognosis. In one series, all the patients with tumors 5 cm or less in diameter survived 10 years, but only 57% of those with tumors 6–15 cm in diameter and 53% of those with even larger tumors survived for that period of time (Fox et al. 1975). Another investigation reported a 73% crude overall survival of patients with tumors less than 5 cm in diameter, a 63% survival of those with tumors between 5 and 15 cm, and a 34% survival of those with a tumor greater than 15 cm in diameter (Stenwig et al. 1979). In a final series, stage I tumors 5 cm or less in diameter were associated with a 100% relative 10-year survival in contrast to a 92% survival of patients with larger stage I tumors (Bjorkholm and Silfversward 1981). The last series is the only one in which the survival rate was corrected for stage, and on that basis, the improvement in prognosis for the smaller stage I tumors was not statistically significant. Therefore, a relationship between tumor size and prognosis independent of stage has not been clearly established.
Attempts to correlate the histologic pattern and the degrees of nuclear atypia and mitotic activity with prognosis have met with varying success; most investigators have been unable to show a prognostic importance for pattern alone in granulosa cell tumors (Norris and Taylor 1968; Sjostedt and Wahlen 1961).
The degree of nuclear atypicality within granulosa cell tumors has been correlated with their prognosis. In one study, the 5-year survival of patients whose tumors showed no atypia was 92% compared with 80% for those with slight atypicality and 30% for those with moderate atypicality (Stenwig et al. 1979). In another study, there was an 80% relative 25-year survival in cases with grade 1 nuclear atypicality in contrast to only a 60% survival in those with grade 2 atypia (Bjorkholm and Silfversward 1981). In both of these studies, nuclear atypicality was the most reliable prognostic index in cases of stage I tumors; for higher-stage tumors, nuclear atypicality and mitotic rate were of similar significance. With regard to the relation of nuclear atypicality to prognosis, it should be noted that assessment of its degree is somewhat subjective. Also, as noted earlier, approximately 2% of granulosa cell tumors contain mononucleate and multinucleate cells with large, bizarre, and hyperchromatic nuclei (Figs. 19 and 20), the presence of which has not been shown to worsen the prognosis. These nuclear changes, which resemble those seen in the uterine leiomyoma with bizarre nuclei, also are encountered in juvenile granulosa cell tumors, microcystic stromal tumors, occasional Sertoli–Leydig cell tumors, and thecomas, and probably are degenerative. In a study of 8 granulosa cell tumors, 7 Sertoli–Leydig cell tumors, and 2 thecomas with bizarre nuclei, follow-up was obtained on 11 patients, all of whom were alive without evidence of disease 3–21 years postoperatively (Young and Scully 1983a).
The mitotic activity of granulosa cell tumors also has been correlated with their prognosis. In one study, there was a 70% 10-year survival associated with tumors that had two or fewer mitotic figures (MF) per ten high-power fields (HPF) compared with only a 37% survival for those with three or more (Stenwig et al. 1979). In another investigation (Norris and Taylor 1968), tumors with many mitotic figures were associated with a worse prognosis than those with few, but most of the tumors with high mitotic rates also were at a higher stage than those with low mitotic rates, and differences in mitotic rate did not have a statistically significant effect on the prognosis of stage I tumors.
Juvenile Granulosa Cell Tumor
Somewhat less than 5% of granulosa cell tumors are diagnosed before the age of normal puberty (Lack et al. 1981). The great majority of these tumors as well as many granulosa cell tumors in young adults differ histologically from adult granulosa cell tumors, and the designation juvenile has been selected for such tumors because 97% of them occur in the first three decades (Young et al. 1984a). Approximately 80% of juvenile granulosa cell tumors (JGCTs) occurring in children result in isosexual precocity, accounting for 10% of cases of that syndrome in the female. More common forms of isosexual precocity are those of central origin, with premature release of gonadotropins from the anterior pituitary gland, and those resulting from apparently autonomous formation of one or more follicle cysts. The precocity caused by granulosa cell tumors is more specifically designated pseudoprecocity because there is no associated ovulation or progesterone production, precluding the possibility of pregnancy, which exists, in contrast, in cases of true sexual precocity. Typically, pseudoprecocity is heralded by the development of the breasts, followed by the appearance of pubic and axillary hair, stimulation and enlargement of the external and internal secondary sex organs, irregular uterine bleeding, and a whitish vaginal discharge, believed to originate in the stimulated endocervical glands. Somatic and skeletal development typically are accelerated as well. Androgenic manifestations such as clitoromegaly occasionally occur (Young et al. 1984a).
When it occurs after puberty, the JGCT usually presents with abdominal pain or swelling, sometimes associated with menstrual irregularities or amenorrhea. Approximately, 6% of all the patients present with acute abdominal symptoms because of rupture of the tumor and hemoperitoneum. An interesting clinical association of the JGCT has been its association with Ollier’s disease (enchondromatosis) in some patients and with Maffucci’s syndrome (enchondromatosis and hemangiomatosis) in a few others. The JGCT is bilateral in only about 2% of the cases. It appears ruptured at operation in approximately 10% of the cases, and ascites is present in a similar percentage. Spread beyond the ovary is unusual; in our series only 2% of the tumors were stage II; rare tumors are stage III (Young et al. 1984a). The diameter of the tumor has ranged from 3 to 32 cm, with an average of 12.5 cm. Because of the usual moderate to large size of the tumor, an adnexal mass is almost always detectable clinically. Rarely, however, a mass has not been palpable preoperatively even on bimanual rectal examination.
The follicles usually vary in size and shape but may be regular and round to oval. They generally do not reach the large size of the follicles in the rare macrofollicular form of AGCT. Their lumens contain eosinophilic or basophilic secretion, which stains with mucicarmine in approximately two thirds of the cases. Granulosa cells of varying layers of thickness line the follicles and occasionally are surrounded by mantles of theca cells. More often, however, the granulosa cells lining the follicles blend into the intervening diffusely cellular areas. Rarely, the lining cells resemble hobnail cells.
The two characteristic cytologic features of the neoplastic granulosa cells that distinguish them from those of the AGCT are their generally rounded, hyperchromatic nuclei, which lack grooves in most cases, and their frequent abundant content of eosinophilic (luteinized) cytoplasm (Fig. 24). The theca cell element of the tumors usually is also luteinized, and lipid stains typically disclose moderate to large amounts of fat within the cytoplasm of both cellular components. The theca cells are more often spindle shaped than the granulosa cells and, like the latter, usually contain hyperchromatic nuclei. In rare JGCTs, small foci more characteristic of the AGCT are encountered.
There are very rare neoplasms that we designate anaplastic juvenile granulosa cell tumor. In contrast to conventional neoplasms with striking nuclear atypia but which still have an orderly architecture, these neoplasms have zones with a sheet-like growth that when viewed in isolation are not recognizable as juvenile granulosa cell tumor and indeed, in some instances, resemble undifferentiated carcinoma. A diagnosis of anaplastic juvenile granulosa cell tumor can only be made when thorough sampling discloses characteristic features of that neoplasm in the form of the typical follicles.
Nuclear atypicality in JGCTs varies from minimal to marked. In approximately 13% of the cases severe degrees are present (Fig. 25) (Young et al. 1984a). The mitotic rate also varies greatly but is generally higher than that seen in AGCTs.
The immunohistochemical profile is essentially identical to that of adult granulosa cell tumors.
Adult versus juvenile granulosa cell tumor (GCT)
Less than 1% prepubertal
Usual after 30 years
Rare after 30 years
Mature follicles and Call–Exner bodies occasional
Immature follicles with mucin content Call–Exner bodies rare
Nuclei pale, angular, commonly grooved
Nuclei darker, round, rarely grooved
The JGCT may be misdiagnosed as a malignant germ cell tumor. The latter may be associated with human chorionic gonadotropin (hCG)-induced isosexual pseudoprecocity. The nuclei of the JGCT are not as primitive appearing as those of either a yolk sac tumor or an embryonal carcinoma, and the follicular pattern of the JGCT is not a feature of either germ cell tumor, although the cysts of the polyvesicular variant of yolk sac tumor can superficially resemble the follicles of a JGCT in rare cases. Immunohistochemical demonstration of SALL4 in most germ cell tumors, OCT 3/4 in embryonal carcinomas, and alpha fetoprotein in yolk sac tumors may be helpful in difficult cases; conversely germ cell tumors are typically negative for inhibin and calretinin.
The JGCT is sometimes misinterpreted as a thecoma because of the occasional absence or rarity of follicles, the typically abundant cytoplasm of the neoplastic cells and the occasional predominance of theca cells. Thorough sampling to demonstrate follicles and the performance of reticulin stains to establish the granulosa cell nature of at least some of the tumor cells are important diagnostically. Also, thecomas rarely exhibit significant mitotic activity, uncommonly occur before 30 years of age, and are exceptionally rare in children. A focally diffuse pattern in a luteinized JGCT may suggest the diagnosis of a steroid cell tumor, but the uniformity of the pattern and cytologic features of the latter tumor would be unusual for a JGCT, which almost always contains more diagnostic areas. Immunohistochemistry for MART1/melanA, FOXL2, and WT1 may be helpful in this distinction, as discussed for adult granulosa cell tumor. The pregnancy luteoma rarely contains rounded follicle-like spaces and may suggest a luteinized JGCT, but like the steroid cell tumor, its cells are uniform in appearance, and it is multiple and bilateral in one half and one third of the cases, respectively.
The common epithelial tumors with which a JGCT may be confused are clear cell, undifferentiated, and transitional cell carcinomas. The tubulocystic variant of the clear cell carcinoma is rarely suggested when follicles in a JGCT are lined by cells resembling hobnail cells, and JGCTs with high-grade nuclear atypia may suggest an undifferentiated carcinoma. Transitional cell carcinoma is mimicked in rare cases in which a cystic JGCT contains pseudopapillae lined by uniform granulosa cells (Irving and Young 2008a). The young age of the patient and the presence of follicles and of other areas typical of a JGCT should help indicate the correct diagnosis in these cases. If necessary, immunohistochemical analysis for EMA and sex cord markers should clarify the diagnosis.
The JGCT may be confused with a small cell carcinoma of hypercalcemic type (see Table 3) because both neoplasms contain follicles, and both are characteristically found in young patients. In the typical case of small cell carcinoma, the presence of cells with scanty cytoplasm is in marked contrast to the JGCT, in which the tumors almost without exception have cells with appreciable to abundant cytoplasm. The follicles in the small cell carcinoma rarely contain the basophilic secretion that is seen in many JGCTs. Although the JGCT characteristically has easily found mitotic figures, mitoses are in general much more numerous in the small cell carcinoma. Particular difficulty may be caused by cases of small cell carcinoma in which the tumor cells have abundant eosinophilic cytoplasm. The tumors usually can be distinguished even in these cases because the small cell carcinomas have a much more disorderly growth than seen in the JGCT. Additionally, the large cells seen in the cases of small cell carcinoma often have a rather distinctive dense, globular cytoplasm, a feature only occasionally seen in the JGCT. Immunohistochemical markers that aid in this distinction are listed in Table 3.
The only metastatic tumor that we have seen confused with a JGCT is metastatic malignant melanoma, because some malignant melanomas, like many metastatic tumors, contain follicle-like spaces and this metastatic tumor frequently has cells with abundant eosinophilic cytoplasm. The result may be a striking simulation of the solid and follicular pattern of a JGCT. It is helpful clinically that metastatic malignant melanoma is very rare in the first two decades, when approximately 80% of JGCTs are encountered. The clinical history, of course, is helpful in many cases but as the history of malignant melanoma may be remote, or the primary tumor may have regressed, the possibility of malignant melanoma should be considered when entertaining the diagnosis of JGCT in a patient over 20 years of age. The likelihood of this diagnosis will be heightened if the ovarian tumor is bilateral. Other features of metastatic melanoma also may be helpful as will immunohistochemistry.
Clinical Behavior and Treatment
Although the JGCT usually appears less well differentiated than the adult type, follow-up to date indicates a high cure rate (Lack et al. 1981; Young et al. 1984a; Zaloudek and Norris 1982). In contrast to AGCTs, which often recur late, all the clinically malignant juvenile tumors have reappeared within 3 years and several have had a rapid course.
In our series, the feature of greatest prognostic significance was the stage of the tumor (Young et al. 1984a). Only 2 of the 80 stage I tumors for which follow-up information was available were clinically malignant. Rupture did not have an adverse effect on prognosis. Two of the ten stage IC tumors were malignant; in one of them, malignant cells were present on cytologic examination of the ascitic fluid. All three stage II tumors were fatal. Although both the mitotic rate and the degree of nuclear atypicality correlated with the prognosis when tumors of all stages were considered, no such correlation was evident when only stage I tumors were evaluated. In conclusion, despite the frequent presence of disquieting features such as severe nuclear atypicality and very high mitotic rates, a JGCT that is confined to the ovary appears to have an excellent prognosis with rare exceptions.
In view of the rarity of bilateral ovarian involvement, and their excellent prognosis, a stage IA JGCT can be treated by a unilateral salpingo-oophorectomy. Little experience has accumulated on the role of radiation therapy and chemotherapy in the management of persistent or recurrent tumor, but isolated examples of the efficacy of these modes of therapy have been recorded.
The comments just made on behavior, prognosis, and therapy do not apply to the rare cases that we designate anaplastic juvenile granulosa cell tumor, which are highly malignant in our experience.
Sertoli Cell Tumor
Sertoli cell tumors are rare; they account for approximately 4% of Sertoli-stromal cell tumors (Oliva et al. 2005; Tavassoli and Norris 1980; Teilum 1958; Young and Scully 1984a). Most tumors are nonfunctioning, but they can secrete estrogen, or less commonly progesterone. Rare Sertoli cell tumors, usually of the lipid-rich type, have resulted in isosexual pseudoprecocity. Occasionally, these tumors are seen in patients with Peutz–Jeghers syndrome, sometimes in association with a sex cord tumor with annular tubules (Young and Scully 1984a; Solh et al. 1983; Ravishankar et al. 2016). One Sertoli cell tumor was associated with progesterone as well as estrogen production (Tracy et al. 1985). Sertoli cell tumors are unilateral, and most have been stage I.
The differential diagnosis of Sertoli cell tumors is broadly similar to that of Sertoli–Leydig cell tumors, considered below. Brief note should be made of the rare tumors with few or no tubules, one of which is illustrated (Figs. 27 and 28). Their distinction from granulosa cell tumor is aided by lack of typical patterns of that tumor and nuclei lacking the distinctive features of it also. Distinction of these rare tumors from other neoplasms reasonably in the differential is aided by staining for inhibin (Fig. 28) or other sex cord markers, all of which are typically expressed in Sertoli cell tumors.
Sex Cord Tumor with Annular Tubules
Sex cord tumor with annular tubules with and without Peutz–Jeghers syndrome
3 cm or less
Several other ovarian tumors have been described in girls with Peutz–Jeghers syndrome, some of which have caused sexual precocity. These are most commonly Sertoli cell tumors, particularly with lipid accumulation (Solh et al. 1983; Ravishankar et al. 2016), but have also included 2 Sertoli-Leydig cell tumors, a granulosa cell tumor, a steroid cell tumor, and two unclassified sex cord tumors which had microscopic findings that were unique in our experience, including diffuse areas, tubular differentiation, microcysts, and papillae, and the presence of two distinctive cell types, one containing abundant eosinophilic cytoplasm and the other scanty cytoplasm (Young et al. 1983). Rare surface epithelial tumors and germ cell tumors have also been reported in patients with Peutz-Jeghers syndrome.
In patients without Peutz–Jeghers syndrome (Fig. 32), in contrast, the tumors are almost always unilateral and usually form palpable masses. Transitions to typical granulosa cell tumor are much more common than in tumorlets associated with Peutz–Jeghers syndrome. A focal solid proliferation of cells with eosinophilic cytoplasm may be seen (Fig. 32). Forty percent of the patients have had manifestations of estrogen secretion; progesterone secretion, as evidenced by a decidual change of the endometrium, is relatively common. At least one fifth of the tumors have been clinically malignant, with a characteristic spread via the lymphatic system. Recurrences often are late. In one remarkable case, multiple recurrences occurred, mostly within regional and distal lymph nodes, over a period of 24 years, with each recurrent tumor removed surgically. In that case, the tumor produced large amounts of Müllerian-inhibiting substance as well as progesterone, both of which were found useful as tumor markers in the serum in monitoring the course of the patient (Gustafson et al. 1992).
Although reported experience is limited, sex cord tumors with annular tubules appear to be positive for inhibin, calretinin, and other sex cord-stromal markers. The differential diagnosis is typically with adult granulosa cell tumor or Sertoli cell tumor, but the characteristic annular tubules permit distinction from them. Immunohistochemistry is not of help in the distinction.
Pure Stromal Cell Tumors
Tumors in the thecoma–fibroma group are composed exclusively or almost exclusively of theca cells, fibroblasts of ovarian stromal origin, or both. Fibroma, cellular fibroma, and fibrosarcoma exist along a morphologic spectrum and are discussed together in this text. The presence of occasional small nests of granulosa cells or occasional tubules lined by Sertoli cells does not exclude tumors from the pure stromal category; such tumors have been referred to as fibromas or thecomas with minor sex cord elements (Young and Scully 1983b). Similarly, lutein cell may be seen in tumors with a background of fibromas or thecomas; while these have historically been classified as luteinized thecomas, the preferred approach is to categorize these based on the background stromal characteristics, as the lutein cells have no prognostic relevance. The rare entity referred to as luteinized thecoma associated with sclerosing peritonitis is the only tumor that retains the designation luteinized thecoma. The rare sclerosing stromal tumor, signet ring stromal tumor, and microcystic stromal tumor belong to this category. In the 2014 World Health Organization classification, steroid cell tumors, including Leydig cell tumors, have been placed into the pure stromal category as well.
This tumor, which is composed of spindle cells forming variable amounts of collagen, accounts for 4% of all ovarian tumors. Fibroma occurs at all ages, but is most frequent during middle age, with an average age of 48 years (Dockerty and Masson 1944); fewer than 10% of the cases are encountered under the age of 30 years. The fibroma is rarely associated with steroid hormone production but may be accompanied by two unusual clinical syndromes, Meigs’ syndrome (Meigs 1954) and the basal cell nevus syndrome (Gorlin’s syndrome) (Gorlin 1987). The former, which complicates about 1% of ovarian fibromas, is defined as ascites and pleural effusion accompanying a fibrous ovarian tumor, usually a fibroma, and disappearing after the removal of the tumor. Ascites alone is present in association with 10–15% of ovarian fibromas larger than 10 cm in diameter (Samanth and Black 1970). The most widely accepted explanation of Meigs’ syndrome is seepage of fluid from the tumor through its serosal surface into the peritoneal cavity, with subsequent passage into one or both pleural cavities either via lymphatics or through a communication between the abdominal and pleural cavity such as the foramen of Bochdalek.
The hereditary basal cell nevus syndrome is characterized by one or more of the following findings: basal cell carcinomas appearing early in life, keratocysts of the jaw, calcification of the dura, mesenteric cysts, and other less common abnormalities (Gorlin 1987) as well as ovarian fibromas, which typically are bilateral, multinodular, and calcified (Fig. 28). A case of fibrosarcoma of the ovary in a child with the basal cell nevus syndrome has been reported (Kraemer et al. 1984).
Fibromas are considered benign tumors. However, they can rarely recur, particularly if associated with adhesions or rupture.
Studies show highly variable rates of expression of inhibin and calretinin in fibromas (Cathro and Stoler 2005; Deavers et al. 2003; Al-Agha et al. 2011; Zhao et al. 2009). Many are negative for one or both of these markers, and when they are positive, they are often only weakly and focally positive. Based on limited data, fibromas appear to express newer markers FoxL2 (Al-Agha et al. 2011) and SF-1 (Zhao et al. 2009) in most cases, usually strongly and diffusely. CD56 (He et al. 2008) and WT-1 (Zhao et al. 2009) are also usually expressed by fibromas. These markers are rarely necessary in making the diagnosis.
The fibroma must be distinguished from several nonneoplastic ovarian processes, specifically massive edema, fibromatosis, and stromal hyperplasia. The first two disorders usually are unilateral but may be bilateral and are characterized by proliferation of ovarian stromal cells with marked intercellular edema and the production of abundant dense collagen, respectively. Unlike fibromas, which almost always displace follicles, corpora lutea, and corpora albicantia, massive edema and fibromatosis encompass these structures. Stromal hyperplasia, in contrast to the ovarian fibroma, is bilateral and is characterized by a multinodular or diffuse proliferation of closely packed, small stromal cells with minimal collagen formation.
Within the sex-cord stromal tumor category, distinction of fibromas from thecomas can be challenging, and in rare cases is impossible; this is discussed in detail in the discussion of thecomas. Differentiation from diffuse adult granulosa cell tumors has been previously discussed. Rare fibromas have varied cellularity and some vascularity, which may cause them to be misinterpreted as sclerosing stromal tumor. The distinct pseudolobulation, lutein cells, and ectatic vessels, in aggregate, set the sclerosing stromal tumor apart from fibromas that may mimic them to a limited degree. Some fibromas undergo prominent cystic degeneration and may be misconstrued as surface epithelial stromal tumors. However, the cysts (pseudocysts) do not have an epithelial lining.
Other spindle cell tumors, either primary or metastatic of the ovary, also enter into the differential diagnosis. Smooth muscle tumors have more abundant eosinophilic cytoplasm, rounded nuclei, and a prominent fascicular architecture; immunohistochemistry for smooth muscle markers is positive. Endometrial or endometrioid stromal sarcoma has a characteristic whorling of cells around small vessels, and may have foamy macrophages; these tumors are characteristically positive for CD10, which has been negative in the fibromas that have been reported (Oliva et al. 2007). Metastatic gastrointestinal stromal tumors to the ovary have been reported; these have the same features as in the gastrointestinal tract, including intracytoplasmic vacuoles, and express c-kit and DOG1.
Finally Krukenberg tumors often have a fibromatous background stroma and can occasionally be mistaken for fibromas if the epithelial component is sparse and subtle. This is particularly likely to present an issue at the time of interoperative consultation, as signet ring cells can be quite inconspicuous on frozen sections. Bilaterality is a helpful clinical clue in this situation. Grossly and microscopically, Krukenberg tumors are more likely to have a nodular growth pattern and infiltration of the normal ovarian parenchyma. Careful sectioning usually identifies an obvious epithelial component, which can be confirmed if necessary with a mucin stain or immunohistochemistry.
This malignant ovarian tumor is characterized by more or less uniform hypercellularity with significant cytologic atypia and conspicuous mitotic activity (Christman and Ballon 1990). Patients are typically postmenopausal. Tumors are usually large and unilateral, with frequent hemorrhage and necrosis. While criteria of moderate to severe cytologic atypia and >4 mitosis per 10 high-power fields have been established for the diagnosis, in some cases the distinction between low-grade fibrosarcomas and cellular fibromas is extremely challenging (Prat and Scully 1981; Irving et al. 2006).
Although different criteria for the microscopic separation of thecoma and fibroma have resulted in varying estimates of the frequency of these tumors, thecomas are approximately one third as common as granulosa cell tumors. The thecoma occurs at an older average age than the granulosa cell tumor, being very rare before puberty and uncommon before the age of 30 years. In one large series, 84% of the patients were postmenopausal, with a mean age of 59 years; only 10% of the patients were under 30 years of age (Bjorkholm and Silfversward 1980). In the same series, 60% of the postmenopausal women presented because of uterine bleeding, and 21% of the patients had endometrial carcinoma. Thecomas have historically been separated into typical and luteinized forms (Bjorkholm and Silfversward 1980; Banner and Dockerty 1945; Geist and Gaines 1938; Hughesdon 1983; Roth and Sternberg 1983; Zhang et al. 1982); however, the most recent World Health Organization classification eliminates the luteinized thecoma category (Kurman et al. 2014).
Tumors that are predominantly fibromatous or thecomatous but also contain collections of steroid-type cells, resembling luteinized theca and luteinized stromal cells, have been called luteinized thecomas in the past. However, the most recent World Health Organization classification instead categorizes these tumors as either fibroma or thecoma based on the stromal component, as the presence of luteinized cells has no particular prognostic significance (Kurman et al. 2014). In the largest series of luteinized thecomas, half of them were estrogenic, 39% were nonfunctioning, and 11% were androgenic (Zhang et al. 1982). This relatively high frequency of masculinization contrasts with its great rarity in association with nonluteinized thecomas. Luteinized thecomas also occur in a younger age group than typical thecomas; although they are most frequent in postmenopausal women, 30% of them have occurred in patients under 30 years of age. When, on rare occasions, crystals of Reinke are identified in the steroid-type cells (Paraskevas and Scully 1989), the term stromal-Leydig cell tumor has been applied (Zhang et al. 1982; Bohm et al. 1991; Paoletti et al. 1987; Scully 1953; Sternberg and Roth 1973). Approximately half of these tumors have been virilizing.
The most common diagnostic difficulty is distinction from fibroma. There exists a morphologic spectrum between thecoma and fibroma; cases at either end are relatively straightforward, whereas rare tumors in the middle may be impossible to resolve. Such tumors are made up of cells having some but not all the features of theca cells, containing small to moderate amounts of lipid, and being associated with equivocal evidence of estrogen secretion. The primary morphologic distinguishing feature is the abundant pale gray or lipid-rich cytoplasm in thecomas. The term “fibrothecoma” has been used commonly for indeterminate lesions. Some use the same term for all tumors in the fibroma and thecoma categories, but we prefer to separate the vast majority of tumors that can be classified morphologically into one category or the other and limit the use of the indeterminate category.
Adult granulosa cell tumors with abundant fibrothecomatous stroma or a diffuse growth pattern may mimic thecomas. In most granulosa cell tumors, obvious epithelial components will be identified with thorough sampling. Immunohistochemistry is of limited use in this distinction. However, a reticulin stain may be helpful – in thecomas, reticulum fibrils typically surround individual tumor cells (although sometimes they surround small clusters of cells), in contrast to granulosa cell tumors, in which they invest entire nests of cells. The presence of a very minor (<10%) sex cord component is acceptable within a thecoma as these minor components appear to have no adverse prognostic implications. The differential diagnosis with sclerosing stromal tumor and microcystic stromal tumor are discussed below; the morphologic overlap with other sex cord stromal tumors is limited.
Thecomas are almost never malignant. A number of tumors have been reported as “malignant thecomas,” but some of these tumors are better interpreted as endocrinologically inactive fibrosarcomas or diffuse granulosa cell tumors (Waxman et al. 1979). In cases in which the preservation of fertility is important, a thecoma can be treated adequately by oophorectomy. Total hysterectomy with bilateral salpingo-oophorectomy is indicated, however, in most patients who are menopausal or postmenopausal.
Luteinized Thecoma (Thecomatosis) Associated with Sclerosing Peritonitis
Grossly, the tumors range from nearly normal in size with expansion of the cortex and an exaggeration of the normal cerebriform appearance to large spherical masses with no identifiable normal ovary. Edema, cyst formation, and hemorrhage are common. Microscopic examination discloses a bland spindle cell proliferation that circumferentially involves the entire cortex with sparing of the medulla (Fig. 40). Cellularity is generally high, with admixed densely cellular areas and less cellular areas in which edema with microcystic change is often conspicuous (Fig. 41). Lutein cells are present but are arranged in small nests or singly and have only small to moderate amounts of cytoplasm (Fig. 42), making them less easy to recognize than those in the usual luteinized thecoma. The proliferation frequently incorporates preexisting ovarian structures such as follicles. Another unusual aspect of this lesion is the brisk mitotic activity present in many cases. While the luteinized cells stain with common sex cord markers such as inhibin and calretinin, the spindle cell component typically does not; however, this component does stain for FOXL2 and SF-1, supporting ovarian stromal differentiation (McCluggage et al. 2013).
The differential diagnosis is fairly limited. This lesion is distinguished from fibromas and thecomas with luteinized cells by its bilaterality, lack of estrogenic manifestations, cortical rind-like growth pattern, and the characteristics of the luteinized cells, which are smaller and occur singly and in small clusters, rather than large nests. The nonneoplastic lesions fibromatosis and massive edema share involvement of normal elements, but are usually unilateral, are much less cellular, and usually lack luteinized cells. Stromal hyperthecosis may enter the differential diagnosis of small lesions, but the stromal proliferation associated with hyperthecosis is typically medullary rather than cortical. Finally, the luteinized cells may in some cases be subtle, leading to consideration of a variety of spindle cell neoplasms, particularly fibroma. The bilaterality and diffuse cortical growth pattern of luteinized thecomas associated with sclerosing peritonitis are clues to search for the luteinized cells, which are invariably present. Immunohistochemistry for sex cord markers can highlight the luteinized cells, although this is rarely necessary.
The lesion is presumed benign, with no reports of metastasis. Several pieces of evidence suggest a nonneoplastic process: the circumferential cortical growth pattern seen in all lesions in which normal ovary can be identified, the growth around preexisting ovarian structures, and a case report of successful medical management without oophorectomy (Schonman et al. 2008). The associated sclerosing peritonitis (seen in nearly all cases although sometimes only histologically) may cause significant morbidity, even for months or years after removal of the ovaries, and has occasionally been fatal (Staats et al. 2008). The pathogenesis of the sclerosing peritonitis in these cases remains a mystery.
Sclerosing Stromal Tumor
This tumor differs from the fibroma and the thecoma both clinically and pathologically (Chalvardjian and Scully 1973). Although the latter tumors are uncommon in the first three decades, more than 80% of sclerosing stromal tumors have been encountered during the second and third decades, with an average age at diagnosis of 27 years. In contrast to the thecoma, the sclerosing stromal tumor is only rarely associated with evidence of estrogen and/or androgen secretion. All sclerosing stromal tumors encountered to date have been benign, although a local recurrence has been reported (Goebel et al. 2016).
Sclerosing stromal tumor versus fibroma, thecoma, and steroid cell tumors
Sclerosing stromal tumor
Steroid cell tumor
80% under 30 years
10% under 30 years
Average age 63 years
25% under 30 years
Almost always absent
Prominent ectatic vessels
Two cell types
Only in luteinized form
Only in luteinized form
Almost always Benign
Almost always benign
Signet-Ring Stromal Tumor
Microcystic Stromal Tumor
Microcystic stromal tumors have a distinctive immunoprofile. Among the sex cord-stromal markers, they are typically negative for inhibin, calretinin, and CD56, but positive for FOXL2, SF-1, and WT-1, supporting their classification in the sex cord-stromal category. They have uniformly expressed nuclear beta-catenin, cyclin D1, CD10, and vimentin. About one third of the tumors examined to date have been cytokeratin positive, but they have been EMA negative. A heterozygous point mutation in the beta-catenin gene CTNNB1 has been identified in more than half of the microcystic stromal tumors that have been evaluated for it (Irving et al. 2015).
The differential diagnosis of this tumor is broad and has been presented in detail elsewhere (Irving and Young 2008b). The hyaline plaques and cellular morphology most commonly raise consideration of thecoma, but the prominent microcysts and distinctive immunohistochemical profile readily distinguish the two. Exceptional solid pseudopapillary tumors of the ovary, with features identical to those that arise more commonly in the pancreas, have been described (Deshpande et al. 2010), and given the shared CTNNB1 mutations between the two tumors, consideration has been raised that these may be variants of the same tumor. However, among other differences, the microcystic stromal tumors lack a pseudopapillary growth pattern or the characteristic nuclear grooves of solid pseudopapillary tumor. In addition, CD56 is usually positive and WT-1 negative in the latter tumor. As a great many other ovarian tumors may have microcysts, the designation of microcystic stromal tumor should only be made after neoplasms have been thoroughly sampled to disclose other foci that may merit separate categorization.
Steroid Cell Tumors
Clinical and pathologic features of steroid cell tumors
Leydig cell tumor
C- hilus cell tumor
Steroid cell tumor NOS
Age range, years (mean)
Duration of androgenic manifestations
Diameter, cm (mean)
All tumors within the steroid cell group share certain morphologic and immunohistochemical features. The cells are round, with moderate to abundant cytoplasm, which may be eosinophilic or lipid-rich, imparting a spongy multivacuolated and nearly clear appearance. The eosinophilic cytoplasm is predominant in stromal luteoma and Leydig cell tumor, whereas both cytoplasmic patterns are seen in steroid cell tumors, not otherwise specified (NOS). The nuclei are round, centrally located, with coarse chromatin texture and a prominent central nucleolus. Intracytoplasmic lipochrome pigment may be conspicuous.
Steroid cell tumors of all subcategories share a similar immunohistochemical profile. They express most sex cord-stromal markers, including inhibin, calretinin, and SF-1. However, they typically do not express FOXL2 or WT-1, and they do express melanA, a marker not expressed in most other sex cord stromal tumors.
Microscopic examination of a stromal luteoma reveals a more or less rounded nodule of cells of lutein type that generally contain relatively little lipid. Mitoses generally are rare. The cells may be arranged diffusely or in small nests or cords and are more or less completely surrounded by ovarian stroma. One confusing feature, seen in about 20% of the cases of all subtypes, is focal degeneration, with the formation of irregular spaces that may simulate glands or vessels. These spaces may contain, or be surrounded by, lipid-laden cells and chronic inflammatory cells and may be associated with fibrosis. In some cases they contain red blood cells.
Eighty percent of stromal luteomas occur in postmenopausal women. The initial symptom in 60% of patients is abnormal vaginal bleeding probably related to hyperestrinism, although whether the tumor secretes estrogen directly or secretes an androgen that is converted peripherally to an estrogen is unknown. Androgenic manifestations are present in only 12% of the cases. This profile of hormonal function is the opposite of that associated with other categories of steroid cell tumor, which usually are androgenic and only occasionally estrogenic. Underlying stromal hyperthecosis may contribute to the clinical picture in some cases, particularly those in which there is a long history of hormonal disturbance.
All reported tumors with these features have had a benign outcome. Therefore, although stromal luteomas are currently classified as steroid cell tumors not otherwise specified by the WHO, a small steroid cell tumor confined to the ovary and accompanied by stromal hyperthecosis and estrogenic manifestations can be designated as consistent with stromal luteoma in order to identify the high likelihood of benign behavior.
Leydig Cell Tumor
A Leydig cell nature of a steroid cell tumor can be proved only by the identification of the more or less specific crystals of Reinke in the cytoplasm of the neoplastic cells on either light microscopic or electron microscopic examination (Sternberg 1949). Because only 35–40% of Leydig cell tumors of the testis contain crystals of Reinke on light microscopic examination and Leydig cells cannot be differentiated from lutein cells or adrenal cortical cells in the absence of these inclusions, it is probable that a number of unclassified steroid cell tumors are Leydig cell tumors that cannot be identified specifically as such.
Ovarian Leydig cell tumors have been divided into two subtypes by Roth and Sternberg (Roth and Sternberg 1973), the hilus cell tumor and the Leydig cell tumor, nonhilar type. The former, which is much more common, originates in the ovarian hilus from hilar Leydig cells, which have been identified in 80–85% of adult ovaries, usually lying in relation to nonmedullated nerve fibers (Sternberg 1949). Hilus cell tumors, which account for approximately 20% of steroid cell tumors (Paraskevas and Scully 1989), occur at an average age of 58 years and cause hirsutism or virilization in three quarters of the cases; they are rarely associated with estrogenic manifestations. The androgenic changes typically have a less abrupt onset and are milder than those associated with Sertoli–Leydig cell tumors. They sometimes have been present for many years. The urinary 17-ketosteroid levels usually are normal or only slightly elevated because these tumors produce predominantly the potent androgen, testosterone, which is not a 17-ketosteroid, instead of the weaker androgens androstenedione and dehydroepiandrosterone, elevations of which are typically associated with high values of urinary 17-ketosteroids. Hilus cell tumors exceptionally are palpable preoperatively. They are rarely bilateral. There is no convincing example of malignant Leydig cell tumor in the literature.
The diagnosis of hilus cell tumor is favored if a crystal-free steroid cell tumor located in the hilus has a background of hilus cell hyperplasia, is associated with nonmedullated nerve fibers, has fibrinoid necrosis of blood vessel walls, or shows nuclear clustering with intervening nucleus-free zones (Paraskevas and Scully 1989). On electron microscopic examination, crystals of Reinke typically are needle shaped when cut longitudinally and hexagonal when cut in cross section. The interior of the crystal has a cross-hatched appearance. Intracytoplasmic eosinophilic spheres, which may be crystal precursors, also are typically present but are not specific for hilus cell tumors. Stromal hyperthecosis, hilus cell hyperplasia, or both are associated findings in some cases. The Leydig cell tumor of nonhilar type is thought to arise directly from ovarian stromal cells. Only four examples of this tumor have been reported (Roth and Sternberg 1973) and, except for their location, their clinical and pathologic features have not differed from those of hilus cell tumors. An ovarian stromal cell derivation of these tumors is supported by the very rare finding of Leydig cells containing crystals in the steroid cell nests of ovaries that otherwise have the typical appearance of stromal hyperthecosis. In some cases in which a Leydig cell tumor is in equal contact with ovarian stroma and hilar stroma, it may be impossible to determine whether it is of the hilar or nonhilar type.
Steroid Cell Tumor, Not Otherwise Specified
These tumors occur at any age but typically at a younger age (mean, 43 years) than other types of steroid cell tumor, and in contrast to the latter, occasionally occur before puberty (Hayes and Scully 1987b). Steroid cell tumors not otherwise specified (NOS) are associated with androgenic changes, which may be of many years duration, in approximately half the cases, estrogenic changes, including rare examples of isosexual pseudoprecocity in approximately 10% of the cases, and occasionally progestagenic changes. A few tumors have secreted cortisol and caused Cushing’s syndrome (Young and Scully 1987) and others have been accompanied by elevated cortisol levels in the absence of clinical manifestations of the syndrome; one secreted aldosterone. Rare tumors have been associated with hypercalcemia, erythrocytosis, or ascites. The remaining cases have not been accompanied by endocrine or paraendocrine manifestations. Hormone studies performed in patients with androgenic changes, Cushing’s syndrome, or both typically show elevated urinary levels of 17-ketosteroids and 17-hydroxycorticosteroids as well as increased serum levels of testosterone and androstenedione. The tumors that resulted in Cushing’s syndrome were associated with elevated levels of free cortisol in the blood or urine.
The tumors typically are solid and well circumscribed, occasionally are lobulated, and have a mean diameter of 8.4 cm; only about 5% of them are bilateral. The sectioned surfaces typically are yellow or orange if large amounts of intracytoplasmic lipid are present, red to brown if the cells are lipid poor, or dark brown to black if large quantities of intracytoplasmic lipochrome pigment are present. Necrosis, hemorrhage, and cystic degeneration occasionally are observed.
Stromal luteomas and Leydig cell tumors usually do not pose great diagnostic difficulty for the pathologist because of their characteristic locations and obvious composition of steroid-type cells, which contain crystals of Reinke in the Leydig cell tumor. The extensive formation of spaces in occasional tumors in these categories, however, may cause confusion with an adenocarcinoma and more often with a vascular tumor. Awareness of this degenerative phenomenon and its association with cellular debris, inflammatory cell infiltration, and fibrosis, as well as the finding of typical areas elsewhere in the specimen, particularly at the periphery, should facilitate the diagnosis.
Steroid cell tumors in the NOS category vary more widely in appearance than the stromal luteoma and Leydig cell tumor, both architecturally and cytologically, and are accordingly the cause of greater diagnostic difficulty. The tumors that may enter the differential diagnosis include extensively luteinized granulosa cell tumors and thecomas, lipid-rich Sertoli cell tumors, clear cell carcinomas, particularly those of the oxyphilic type, rare oxyphilic endometrioid carcinomas, hepatoid yolk sac tumors and hepatoid carcinomas, endocrine tumors such as oxyphilic struma ovarii, pituitary-type tumors, and paragangliomas (pheochromocytoma), metastatic renal cell carcinomas, adrenocortical carcinomas, hepatocellular carcinomas, other metastatic tumors with oxyphilic appearance, and primary and metastatic melanomas.
The presence of characteristic nonluteinized cells in both luteinized granulosa cell tumors and thecomas, as well as the typical cytologic features and patterns of these neoplasms, and the finding of abundant reticulum in thecomas are of help in the identification of these tumors. The recognition of areas with a solid tubular pattern helps distinguish a usually estrogenic lipid-rich Sertoli cell tumor with a predominant diffuse pattern from a typically androgenic steroid cell tumor. Immunohistochemistry can be helpful in distinguishing the group of steroid cell tumors from other sex cord-stromal tumors; steroid cell tumors are typically positive for melanA and negative for WT-1 and FOXL2, while the other sex cord stromal tumors in the differential diagnosis show the opposite pattern.
In the differential diagnosis with tumors with clear cytoplasm, it is helpful to note that the cytoplasm of steroid cell tumors is better described as spongy or vacuolated than truly clear. In contrast to steroid cell tumors, the clear cells of clear cell carcinomas and metastatic renal cell carcinomas have glycogen-rich cytoplasm and eccentric nuclei. Also, the presence of tubular, glandular, and papillary patterns, which are inconsistent with a steroid cell tumor, generally facilitate the differential diagnosis. MelanA and EMA or cytokeratins can distinguish these immunohistochemically if necessary.
Oxyphilic clear cell carcinomas and endometrioid carcinomas and hepatoid yolk sac tumors and hepatoid carcinomas are all characterized by neoplastic cells with abundant eosinophilic cytoplasm. The first two tumors generally exhibit epithelial patterns, may contain glandular lumens, and are almost always accompanied by more easily recognized patterns. The oxyphilic clear cell carcinoma is almost always accompanied by a variable component of clear and hobnail cells not seen in steroid cell tumors. The hepatoid tumors also have epithelial patterns and may contain glandular lumens; they are characterized by immunohistochemical staining for alpha-fetoprotein. Primary and metastatic melanomas can simulate steroid cell tumors if amelanotic, and if they are pigmented, the pigment granules may be confused with the lipochrome granules of a steroid cell tumor. Melanomas generally have more malignant nuclear features than steroid cell tumors. Special staining, including staining for S-100 protein and absence of inhibin, calretinin, or SF-1, may be helpful in difficult cases; melanA/MART1 is not helpful in this differential diagnosis. An association with other teratomatous elements and the presence of colloid and immunohistochemical staining for thyroglobulin should enable one to distinguish an oxyphilic struma from a steroid cell tumor NOS.
A rare pituitary-type tumor containing cells with abundant eosinophilic cytoplasm that arose in a wall of a dermoid cyst secreted ACTH and caused Cushing’s syndrome (Axiotis et al. 1987). Such a tumor might be confused with a steroid cell tumor. In that case, immunohistochemical staining for ACTH and several other pituitary hormones was positive. In one case we have seen of pheochromocytoma of the ovary, immunohistochemical staining of the tumor cells for chromogranin was helpful in establishing that diagnosis over steroid cell tumor. Finally, the presence or absence of endocrine manifestations and their nature may be important clinical clues to the diagnosis.
Pregnancy luteomas, which are hyperplastic nodules composed of lutein cells that develop during pregnancy, may form large masses that resemble steroid cell tumors grossly and microscopically. As with the latter, they also may be virilizing (in about one quarter of the cases). Unlike steroid cell tumors, however, approximately one third of pregnancy luteomas are bilateral and approximately one half are multiple. Microscopic examination reveals masses of cells with abundant eosinophilic cytoplasm containing little or no lipid; mitotic figures may be numerous, sometimes up to 2 or 3/10 HPF. In contrast, a steroid cell tumor with minimal cytologic atypia that resembles a pregnancy luteoma usually contains only rare mitotic figures. Although it may be impossible to distinguish a lipid-poor or lipid-free steroid cell tumor from a solitary pregnancy luteoma, a lesion encountered during the third trimester of pregnancy is presumed to be a solitary pregnancy luteoma unless clear-cut evidence indicates otherwise.
Clinical Behavior and Treatment
While stromal luteomas and Leydig cell tumors have a benign outcome, steroid cell tumors NOS have a malignant potential. In the two largest series in the literature, the proportion of tumors that were clinically malignant was 25% and 43% (Taylor and Norris 1967; Hayes and Scully 1987b); rare tumors have recurred as many as 19 years postoperatively. Extraovarian spread of tumor is present at the time of operation in a small minority of cases; three of the patients with Cushing’s disease had extensive intraabdominal spread of tumor (Young and Scully 1987). Patients with clinically malignant tumors were on average 16 years older than patients with benign tumors in one series (Hayes and Scully 1987b); no malignant steroid cell tumors have been reported in patients in the first two decades.
The best pathologic correlates with malignant behavior in one series (Hayes and Scully 1987b) were 2 or more mitotic figures per 10 HPFs (92% malignant), necrosis (86% malignant), a diameter of 7 cm (78% malignant), hemorrhage (77% malignant), and grade 2 or 3 nuclear atypia (64% malignant), occasional tumors that appear cytologically benign, however, may be malignant. The metastatic tumor appears similar to the primary tumor in some cases but more poorly differentiated in others.
Mixed Sex Cord-Stromal Tumors
This category includes Sertoli-Leydig cell tumors, as well as the sex cord stromal tumors not otherwise specified. The latter category includes tumors commonly referred to as “gynandroblastomas.” Our preferred approach to these tumors is to classify them as sex cord stromal tumors, NOS, with a comment on the components and their relative quantity. Nevertheless, given the historical relevance of the term, it is listed separately below.
Sertoli–Leydig Cell Tumor
Sertoli–Leydig cell tumors account for less than 0.5% of all ovarian tumors. They are divided into three main subtypes: well differentiated, of intermediate differentiation, and poorly differentiated. The latter two categories may additionally contain heterologous elements, a retiform component, or both, complicating an already often complex appearance. Sertoli–Leydig cell tumors occur in all age groups but are encountered most often in young women. The average age is 25 years; 75% of the patients are 30 years of age or younger and only about 10% are over 50 years (Roth et al. 1981; Young and Scully 1985; Zaloudek and Norris 1984). The well-differentiated tumors (Young and Scully 1984b) occur on an average a decade later, and retiform tumors (Young and Scully 1983c) a decade earlier, than other Sertoli–Leydig cell tumors. Tumors with a retiform pattern are more common in the first decade than any other subtype. The Sertoli-Leydig cell tumors in patients with germline DICER1 mutations (see molecular genetics, below) occur at a median age of 13 years (Rio et al. 2011).
It should be emphasized that in some Sertoli–Leydig cell tumors, usually those that are poorly differentiated, recognizable Leydig cells are few, and in even some cases objectively cannot be appreciated. The tumor is still put in the Sertoli–Leydig category when the overall patterns fit best with that designation.
Although the most striking mode of presentation of Sertoli–Leydig cell tumors is virilization, it develops in only about one third of the cases. In such cases, a patient who has been having normal menstrual periods typically begins to have oligomenorrhea, followed within a few months by amenorrhea. There is a concomitant loss of female secondary sex characteristics, with atrophy of the breasts and disappearance of normal bodily contours. Progressive masculinization is heralded by acne, with hirsutism, temporal balding, deepening of the voice, and enlargement of the clitoris following in its wake. Androgenic manifestations are lower in tumors containing heterologous elements and lower again in those having a prominent retiform component. The androgen secretion by the tumor also may result in erythrocytosis.
Plasma levels of testosterone, androstenedione, and other androgens, alone or in combination, may be elevated in patients with Sertoli–Leydig cell tumors. The urinary 17-ketosteroid values usually are normal or only slightly raised, although occasionally a high level has been recorded. These findings are in contrast to those associated with virilizing adrenal tumors, which often are accompanied by high urinary levels of 17-ketosteroids. The values for plasma androgens and urinary 17-ketosteroids are not reliable, however, in the differentiation of ovarian and adrenal virilizing tumors because the latter often are associated with elevated testosterone and normal urinary 17-ketosteroid levels (Anderson and Rees 1975); also, tests involving attempted stimulation by tropic hormones and suppression by gonadal and adrenocortical steroids have not proved decisive in differentiating these tumors. Occasional Sertoli–Leydig cell tumors are associated with elevated plasma levels of alpha-fetoprotein, but values as high as those accompanying yolk sac tumors are rare (Benfield et al. 1982; Gagnon et al. 1989).
Approximately 50% of patients with Sertoli–Leydig cell tumors have no endocrine manifestations and usually complain of abdominal swelling or pain. Occasional tumors have been associated with various estrogenic syndromes similar to those of the granulosa cell tumor.
At laparotomy, almost all Sertoli–Leydig cell tumors are unilateral. The tumors are stage Ia in about 80% of the cases; in 12% the tumor has either ruptured or involved the external surface of the ovary, and in 4% ascites is present. Only about 2.5% of the tumors have spread beyond the ovary, usually within the pelvis and rarely into the upper abdomen. Poorly differentiated tumors more often are ruptured and present at a higher stage than tumors of intermediate differentiation. Well-differentiated tumors are essentially invariably stage Ia.
Tumors of Intermediate and Poor Differentiation
Retiform Sertoli–Leydig Cell Tumor
Heterologous Sertoli–Leydig Cell Tumor
The Sertoli component of Sertoli-Leydig cell tumors stains similarly to other epithelial sex cord tumors, i.e., they are usually positive for inhibin, SF1, and CD56; calretinin is positive in the majority of cases, and FoxL2 in about half (Cathro and Stoler 2005; Deavers et al. 2003; Al-Agha et al. 2011; Zhao et al. 2009). The Leydig cell component stains similarly to steroid cell tumors, i.e., melanA positive but WT-1 and FOXL2 negative. The retiform components also appear to be negative for FOXL2 (Al-Agha et al. 2011). Positive immunohistochemistry for DICER 1 has been reported on 2 cases with DICER1 mutation, but there are no studies of performance of this marker for clinical use (Rio et al. 2011). As with other sex cord-stromal tumors, keratins may be expressed but EMA is not. Heterologous elements stain similarly to the tissue they recapitulate.
Because of their many patterns, Sertoli-stromal cell tumors often are difficult to differentiate from tumors outside the sex cord-stromal category as well as from granulosa cell tumors. The small hollow tubular structures, solid tubular aggregates, and cords that occasionally are seen in endometrioid carcinomas may closely mimic structures characteristically encountered in Sertoli and Sertoli–Leydig cell tumors. Endometrioid carcinomas also may contain luteinized stromal cells that resemble Leydig cells, creating an even greater problem in differentiation. Mucin secretion, areas of squamous differentiation that range from nests of uniform immature spindle-shaped epithelial cells to morules to keratinizing foci, and an adenofibromatous component of common epithelial type are present in most endometrioid carcinomas, facilitating their diagnosis. Clinical features, such as the usual older age of the patient and the absence of androgenic manifestations, support the diagnosis of endometrioid carcinoma, but it must be emphasized that endometrioid carcinomas occasionally have a functioning stroma, which sometimes is manifested clinically by estrogenic changes and rarely by virilization. Immunohistochemical staining for EMA may be helpful in difficult cases because it is almost always positive in cases of endometrioid carcinoma and only rarely focally positive in a few cells within a Sertoli–Leydig cell tumor. Sex cord stromal markers are also helpful. Sertoli–Leydig cell tumors may be simulated by primary or metastatic endometrioid stromal sarcomas. Criteria that are applicable in the differential diagnosis of these tumors with granulosa cell tumors also are helpful in this situation.
Rarely the alveolar pattern of a Sertoli–Leydig cell tumor, or Sertoli cell tumor, may suggest the diagnosis of dysgerminoma. There is not the typical scattering of lymphocytes of the germ cell tumor, and on high-power scrutiny, there are marked differences in the cytologic features of the neoplastic cells. Should it be indicated, immunohistochemistry for germ cell and sex cord markers will readily distinguish between the two.
The tubular Krukenberg tumor may mimic a Sertoli–Leydig cell tumor, especially if luteinization of the stroma is present; further confusion arises if the former tumor is associated with virilization. Tubular Krukenberg tumors are bilateral, however, in most cases, and contain markedly atypical cells, including signet-ring cells that contain mucin, easily demonstrable by special stains. Other general features of metastatic tumors will aid in cases of Krukenberg tumor.
Carcinoid tumors, especially those of the trabecular type, may be confused with Sertoli–Leydig cell tumors of intermediate differentiation. The ribbons of the former, however, are longer, thicker, and more uniformly distributed than the sex cord-like formations of the latter. Also, rare carcinoid tumors with a solid tubular pattern can be difficult to distinguish from well-differentiated Sertoli cell tumors. Examination of the stroma of carcinoid tumors may be helpful in the differential diagnosis. It is typically less cellular and more fibromatous than that of Sertoli–Leydig cell tumors and rarely contains Leydig cells, typically at the periphery. Primary carcinoid tumors are associated with teratomatous elements in 70% of the cases, and metastatic carcinoids are almost always bilateral and usually are associated with an obvious primary tumor of the intestine and metastases elsewhere in the abdomen. Immunohistochemistry should be definitive in the occasional case in which it is needed.
The tubules seen in ovarian Wolffian tumors may be indistinguishable from those seen in Sertoli and Sertoli–Leydig cell tumors but are virtually always accompanied by other patterns that exclude the diagnosis of a sex cord-stromal tumor. The immunoprofile if these two tumors displays some overlap, with Wolffian tumors expressing sex cord markers, although inhibin is typically weak and focal. CD10 is, however, diffusely and strongly expressed in most Wolffian tumors.
Although in the past, Sertoli–Leydig cell tumors with heterologous elements were sometimes considered “teratomatous” (Reddick and Walton 1982), there are marked differences in the overall constituents of Sertoli–Leydig cell tumors and teratomatous neoplasms such that distinction between the two should be straightforward in the great majority of cases. Rare heterologous neoplasms have a dominant mucinous cystic component and the diagnostic Sertoli–Leydig elements can be absent on certain slides, but judicious sampling will invariably show them, by definition.
The retiform variant of Sertoli–Leydig cell tumor causes specific problems in differential diagnosis. A common misdiagnosis is yolk sac tumor, which is suggested clinically by the young age of the patient and pathologically by the presence of papillae within cystic spaces. The occurrence of androgenic manifestations with about 20% of cases of retiform Sertoli–Leydig cell tumor, however, contrasts with the rare occurrence of such changes in cases of yolk sac tumor, attributable to a functioning stroma. On gross examination, the retiform tumors generally appear less malignant than yolk sac tumors, and microscopic examination reveals less primitive-appearing cells. The presence of other distinctive patterns of either tumor and positive immunohistochemical staining for SALL4 or alpha-fetoprotein and absence of staining for sex cord markers in the yolk sac tumor almost always facilitate the diagnosis.
A greater problem in the differential diagnosis of retiform Sertoli–Leydig cell tumors arises because of their characteristic papillary patterns and the frequent presence of cellular stratification particularly if those features predominate. Under such circumstances, a misdiagnosis of a serous tumor of borderline malignancy or a serous or endometrioid carcinoma occasionally is made. A variety of clinical and pathologic features, including the young age of the patient, the association with virilization, and the presence of other more easily recognizable patterns of Sertoli–Leydig cell tumor, are helpful clues to the correct diagnosis. Finally, the juxtaposition of epithelial and immature mesenchymal elements in some retiform tumors, and in some cases additional mesenchymal heterologous elements, has caused confusion with a malignant mesodermal mixed tumor, but the features already outlined also serve to exclude the latter diagnosis. Although rarely necessary, immunohistochemistry for EMA and one or more sex cord markers should resolve challenging cases.
Adult granulosa cell tumor versus Sertoli–Leydig cell tumor
Granulosa cell tumor
Sertoli–Leydig cell tumor
All age groups, mostly postmenopausal
Mainly young women
Usually estrogenic, rarely androgenic
Usually androgenic, occasionally estrogenic
Microfollicular, macrofollicular, trabecular, insular and diffuse patterns
Hollow or solid tubules, cords, diffuse patterns
Granulosa cells usually mature with pale, often grooved nuclei
Sertoli cells often immature
Fibrothecomatous component common
Fibromatous component uncommon; mesenchyme often immature and cellular, or edematous
Steroid-type cells (lutein cells) usually not prominent and uncommonly clustered
Steroid-type cells (Leydig cells) tend to cluster; rarely contain crystals of Reinke
Heterologous elements rare
Heterologous elements in 20% of cases
Retiform elements absent
Retiform elements in 15% of cases
Recently, somatic or germline mutations in the DICER1 gene, which encodes an RNase III endoribonuclease, have been identified in Sertoli-Leydig cell tumors (Conlon et al. 2015; de Kock et al. 2017; Schultz et al. 2017). While the rate of mutation has varied widely between studies, overall more than half of Sertoli-Leydig cell tumors studied had mutations. One study with central expert pathologist review reported near-universal DICER1 mutations (Schultz et al. 2017). While some studies failed to classify their included tumors by grade, the rate seems to be higher among moderately to poorly differentiated tumors, whereas no case of DICER1 mutation has been reported in well-differentiated Sertoli-Leydig cell tumors to date (Conlon et al. 2015; de Kock et al. 2017). Other ovarian tumors with Sertoli differentiation (Sertoli cell tumor, gynandroblastoma) have also displayed this mutation, whereas testicular sex cord-stromal tumors did not (Conlon et al. 2015; Schultz et al. 2017). Germline mutations in DICER1 result in an increased risk for a variety of lesions, particularly thyroid multinodular goiter, Sertoli-Leydig cell tumor, and pleuropulmonary blastoma; differentiated thyroid tumors, cystic nephroma, nasal chondromesenchymal hamartoma, ciliary body medulloepithelioma, and cervical embryonal rhabdomyosarcomas have also been associated with this mutation. In the ovary, this mutation appears to be relatively specific for Sertoli differentiation, although rare other sex cord stromal tumors, and even rare surface epithelial and germ cell tumors, have had such mutations. Due to the relatively high rate of germline mutations of DICER1, referral for genetic counseling is appropriate for patients with a Sertoli-Leydig cell tumor.
Clinical Behavior and Treatment
After the removal of a virilizing Sertoli–Leydig cell tumor, normal menses characteristically resume in about 4 weeks. The excessive hair usually diminishes to some extent. Clitoromegaly and deepening of the voice are less apt to regress. The prognosis in cases of Sertoli–Leydig cell tumor is closely related to their stage and degree of differentiation. The rare tumors that present in an advanced stage have a poor prognosis, with a mortality rate of 100% in our series (Young and Scully 1985). The survival rates of patients with stage I tumors correlate with the degree of differentiation. In our series, none of the well-differentiated tumors, 11% of those of intermediate differentiation, 59% of the poorly differentiated tumors, and 19% of those with heterologous elements were clinically malignant. The homologous component of the tumor was poorly differentiated in all eight clinically malignant tumors in the heterologous category, and in seven of these the heterologous elements included skeletal muscle, cartilage, or both.
Earlier studies in the literature failed to establish a relation between the degree of differentiation of Sertoli–Leydig cell tumors and their prognosis, but later investigations have supported the findings in our series. In our study, there was also evidence that the presence of a retiform pattern had an adverse effect on the prognosis; 25% of stage I tumors of intermediate differentiation with a retiform component were malignant as opposed to 10% of those with no retiform component (Young and Scully 1985). It is noteworthy that the only stage III tumor of intermediate differentiation in our series had an almost completely retiform pattern, and we have seen an additional Sertoli–Leydig cell tumor with a predominantly retiform pattern that was stage III. Rupture also adversely affected the outcome of stage I tumors. Thirty percent of the tumors of intermediate differentiation that had ruptured were clinically malignant, in contrast to only 7% of those that were intact; in the poorly differentiated category, 86% of the ruptured tumors were malignant compared with 45% of those that had not ruptured. Patients with germline DICER1 mutation appear to have a better prognosis than those due to somatic DICER1 mutations (Schultz et al. 2017).
In contrast to granulosa cell tumors, which often recur many years after primary therapy, Sertoli–Leydig cell tumors typically reappear relatively early. Sixty-six percent of the malignant tumors in our series recurred within 1 year and only 6.6% recurred after 5 years. The recurrent tumor usually is confined to the pelvis and abdomen, but distant metastases to the lung, scalp, and supraclavicular lymph nodes have been reported. Three of the patients in our series had parenchymal liver metastases.
The treatment of a patient with Sertoli–Leydig cell tumor depends on her age, the stage of her tumor, the presence or absence of rupture, and the degree of differentiation. In young women, the low frequency of bilaterality justifies the performance of a unilateral salpingo-oophorectomy if the tumor is stage IA and preservation of fertility is desired. Of note, patients with germline DICER1 mutations who undergo unilateral oophorectomy are at risk of developing a metachronous Sertoli-Leydig cell tumor in the other ovary; these have a favorable prognosis and should not be considered metastases (Schultz et al. 2017). More aggressive surgical therapy and adjuvant therapy are indicated for advanced-stage tumors. Adjuvant therapy also may be advisable for stage I tumors that are poorly differentiated, contain mesenchymal heterologous elements, or are ruptured tumors of intermediate differentiation.
Sex Cord-Stromal Tumors, Not Otherwise Specified
Gynandroblastoma, an extremely rare tumor (Anderson and Rees 1975), has been greatly overdiagnosed. Because small foci of ovarian cell types often are encountered in well-sampled, otherwise typical Sertoli–Leydig cell tumors and conversely, testicular cell types are demonstrable focally in occasional granulosa-stromal cell tumors, the diagnosis of gynandroblastoma should be restricted to the very rare tumors that contain significant components of both forms of neoplasia. According to our criteria, the minor component should account for at least 10% of a tumor in the sex cord-stromal category to warrant a diagnosis of gynandroblastoma. As indicated in the introductory remarks, we prefer to avoid the designation gynandroblastoma for reasons noted there. The nature of the hormones secreted by a sex cord-stromal tumor should not determine its morphologic diagnosis in view of the proven capacity of tumors of testicular cell types to secrete estrogens and of those of ovarian cell types to produce androgens. The rare tumors that have been evaluated molecularly sometimes displayed DICER1 or FOXL2 mutations (Conlon et al. 2015; Schultz et al. 2017).
Sex Cord-Stromal Tumors, Unclassified
Sex cord-stromal tumors, unclassified, is a poorly defined group of tumors that accounts for less than 10% of those in the sex cord-stromal category (Seidman 1996; Simpson et al. 1998). This group includes neoplasms in which a predominant pattern of testicular or ovarian differentiation is not clearly recognizable. The boundary lines between these tumors and those of both ovarian and testicular cell types are vague because interpretations of intermediate patterns of growth and closely similar cell types inevitably are subjective.
Talerman and his associates (Talerman et al. 1982) have described a group of sex cord-stromal tumors, containing diffuse fibrothecomatous and/or granulosa cell-like proliferation as well as areas of tubular differentiation in most of the cases. These authors have interpreted these tumors, which differ in appearance from usual forms of Sertoli–Leydig cell tumor, as diffuse nonlobular androblastomas, but in our opinion, it is more appropriate to place them in the unclassified sex cord-stromal category.
Sex Cord-Stromal Tumors During Pregnancy
Other Ovarian Tumors with Endocrine Function
Ovarian Tumors with Functioning Stroma
Ovarian tumors with functioning stroma (Matias-Guiu and Prat 1990) are associated infrequently with overt endocrine manifestations but commonly accompanied by subclinical elevations of steroid hormone values. The stromal cells responsible for the hormone secretion in ovarian tumors with functioning stroma typically resemble lutein or Leydig cells and have been referred to as luteinized stromal cells. These cells almost always lie within the tumor singly, diffusely, or in clusters, but rarely they are mainly distributed just outside the tumor, sometimes forming a peripheral band (Fig. 73) (Rutgers and Scully 1986). Exceptionally, crystals of Reinke can be identified in the lutein-like cells, warranting their interpretation as Leydig cells (Konishi et al. 1986). It must be emphasized, however, that steroid-type cells may be prominent in the absence of clinical evidence of hormone overproduction and, conversely, evidence of function may exist in the absence of fully developed cells of steroid type. Ovarian tumors with functioning stroma can be divided into three major categories. In the first two categories, germ cell tumors that contain syncytiotrophoblast cells and tumors in pregnant patients, the luteinized stromal cells probably develop as a result of stimulation by hCG. The cause of the stromal alteration in the third (idiopathic) group, which accounts for most of the cases, is unclear, but ectopic production of hCG or some other stromal stimulant by the neoplastic cells may be responsible.
Germ Cell Tumors Containing Syncytiotrophoblast Cells
Two dysgerminomas with syncytiotrophoblast cells have been associated with luteinization of the stroma and endocrine manifestations; one was accompanied by isosexual precocity and the other by postpubertal virilization (Ueda et al. 1972).
Germ cell tumors that produce hCG, including dysgerminomas with syncytiotrophoblast giant cells (Zaloudek et al. 1981), choriocarcinomas, embryonal carcinomas, and polyembryomas, and mixed primitive germ cell tumors, also may cause manifestations of steroid hormone secretion as a result of hCG stimulation of the ovary contralateral to the tumor to form luteinized follicles that secrete steroid hormones.
Tumors with Functioning Stroma Occurring During Pregnancy
Although it is logical to speculate that ovarian tumors with functioning stroma in pregnant patients may secrete estrogens, this possibility has not been investigated by hormone assay, and clinical manifestations of estrogen excess are not expected to be present during gestation. In contrast, more than 20 examples of virilization caused by ovarian tumors with functioning stroma during pregnancy have been reported. These tumors were mostly Krukenberg tumors or mucinous cystic tumors (Fig. 72), but a few have been Brenner tumors or isolated other tumor types. The onset of the virilization in these patients has ranged from the third to the ninth month of gestation. Female offspring may be virilized.
Although ovarian tumors with functioning stroma in the first two categories are encountered in young girls, patients with tumors in the idiopathic group usually are postmenopausal, reflecting the higher prevalence of ovarian tumors, both primary and metastatic, and possibly the higher levels of circulating luteinizing hormone in this age group. A wide variety of ovarian tumors has been associated with an idiopathic functioning stroma, but its frequency has varied from one type of neoplasm to another.
Mucinous tumors often contain functioning stroma, occasionally resulting in either estrogenic or androgenic manifestations. Brenner tumors have been accompanied by endometrial hyperplasia in 10–16% of the cases and occasionally are virilizing. Rare cases of endometrioid carcinoma have been reported to be associated with endometrial hyperplasia in postmenopausal women, and in one case, virilization and breast secretion developed. We have seen a well-differentiated endometrioid carcinoma from a patient with an elevated serum testosterone level and the recent development of hirsutism. Serous and clear cell tumors have been accompanied only exceptionally by hormone manifestations. Germ cell tumors of various types lacking trophoblastic cells have been associated rarely with stromal luteinization and evidence of steroid hormone secretion in the absence of pregnancy. The germ cell tumors within the idiopathic category that have been accompanied by androgenic or estrogenic manifestations have included a variety of subtypes such as dermoid cyst, struma ovarii, carcinoid tumors, embryonal carcinoma, and yolk sac tumor. The steroid cells that are stimulated in cases of germ cell tumor are peripheral rather than within the tumor in many, if not most, of the cases. The lesions associated with peripheral steroid cell formation in the series of Rutgers and Scully (Rutgers and Scully 1986) describing this phenomenon were struma ovarii (nine cases), strumal or trabecular carcinoids (four cases; Fig. 73), rete cysts (four cases), mucinous cystadenomas (three cases), dermoid cysts (two cases), and single examples of dysgerminoma with syncytiotrophoblast giant cells and metastatic carcinoid. In three of the cases, all strumas, a yellow color was appreciated grossly at the periphery or on the surface of the tumor.
The steroid cells that develop adjacent to ovarian tumors rather than within them are of three types: lutein cells within adjacent ovarian stroma, Leydig cells within ovarian stroma, and hilus cells, which are present only along the hilar border of the tumor. The number of cases in each of these three categories in the series of Rutgers and Scully (Rutgers and Scully 1986) were 14, 2, and 8, respectively. The tumors with hilus cell hyperplasia were typically large with an average greatest diameter of 18 cm. The lutein cells and stromal Leydig cells were located predominantly or exclusively in the cortex or medulla peripheral to the tumor and were arranged singly and in nests, forming a discontinuous band up to 2 mm in thickness. The hilus cells were arranged singly and in small nests forming discontinuous bands in the walls of the cysts in which they arose. Lutein cell formation is accompanied most often by estrogenic manifestations, whereas stromal Leydig cell formation and hilar Leydig cell hyperplasia are associated most often with androgenic changes.
Metastatic carcinomas that contain mucinous cells, such as primary mucinous tumors of the ovary, frequently are associated with luteinization of the stroma and in a significant proportion of cases with clinical evidence of elevated steroid hormone levels. Scully and Richardson (Scully and Richardson 1961) found clinical evidence of excess estrogens as manifested by irregular premenopausal bleeding or postmenopausal bleeding in one quarter of patients with metastatic adenocarcinoma from the large intestine and stomach. Occasional Krukenberg tumors from nonpregnant patients have been associated with virilization. Other metastatic tumors are associated much less often with stromal luteinization. One postmenopausal woman was virilized as a result of luteinization caused by bilateral metastatic lobular carcinoma of the breast (Caron et al. 1990). One metastatic colonic carcinoid was associated with peripheral stromal luteinization (Rutgers and Scully 1986).
Ovarian Tumors with Thyroid Hyperfunction
Although strumas and strumal carcinoids of the ovary have been demonstrated by immunohistochemical staining to contain thyroglobulin, triiodothyronine, and thyroxine and, therefore, probably produce thyroid hormones at subclinical levels in many cases, clinical evidence suggestive of hyperthyroidism is present in only 25% of the cases, and florid thyrotoxicosis in only about 5%. Factors that make it difficult to determine accurately the frequency of hyperthyroidism in patients with struma ovarii include variable criteria for the amount of thyroid tissue required for a diagnosis of struma, the observation that approximately one sixth of patients with struma ovarii have concomitant enlargement of the thyroid gland, and a lack of confirmation of the hyperthyroidism by modern laboratory tests in most of the reported cases.
In some patients with clinical or laboratory evidence of hyperthyroidism, the preoperative diagnosis of hyperfunctioning struma ovarii has been established by high iodine uptake in the pelvis with low radioiodine uptake in the neck (Brown et al. 1973). Other cases of struma-associated hyperthyroidism have not been recognized until the symptoms regressed after removal of an ovarian tumor. In some of these cases, a prior thyroidectomy had had no effect on the hyperthyroidism. Occasionally, oophorectomy for struma may precipitate compensatory enlargement of the thyroid gland, increased uptake of radioactive iodine by the thyroid gland, or an episode of thyrotoxicosis. Similarly, torsion of an ovary containing a struma precipitated striking hyperthyroxinemia in a pregnant patient. Occasional strumal carcinoids have been accompanied by evidence of hypersecretion of thyroid hormone in the form of postoperative thyroid storm or hypothyroidism, and thyroglobulin has been demonstrated in the colloid within tumors of this type.
Ovarian Tumors Associated with the Carcinoid Syndrome
Of the four major categories of primary carcinoid tumor of the ovary, insular, trabecular, strumal, and mucinous, one third of the insular tumors and a single example of strumal carcinoid have been associated with the carcinoid syndrome. One patient with the syndrome also had elevated serum levels of calcitonin (Sens et al. 1982). Metastatic carcinoids involving the ovary are associated with the carcinoid syndrome in almost half the cases. The volume of the carcinoid is an important factor determining the presence or absence of the syndrome in cases of primary carcinoids. The syndrome is present in about two thirds of the cases when the tumor is large. One 74-year-old woman had the carcinoid syndrome attributable to an ovarian tumor that resembled an atypical carcinoid with areas of neuroendocrine (oat cell) carcinoma (Brown and Lane 1965). That patient also was virilized and had Cushing’s syndrome. Although no immunohistochemical staining was performed, the authors concluded that the tumor was elaborating both serotonin and ACTH. The carcinoid syndrome typically occurs in the absence of hepatic or other metastases in cases of ovarian carcinoid, because the hormonal effluent of the tumor enters the systemic circulation directly, bypassing the portal venous system and avoiding inactivation in the liver. The carcinoid syndrome caused by a primary ovarian carcinoid, therefore, is usually curable if the tumor is confined to the ovary and irreversible damage to cardiac valves has not occurred.
Ovarian Tumors Associated with Zollinger–Ellison Syndrome
The association between ovarian mucinous tumors and Zollinger–Ellison syndrome is consistent with the frequent finding of neuroendocrine intestinal-type cells within mucinous tumors. A number of studies have shown that all categories of mucinous tumors (benign, borderline, and malignant) commonly contain argyrophil and hormone-immunoreactive cells, although in most of the studies, these cells have been found most frequently in mucinous borderline tumors. The argyrophilic cells often are immunoreactive for serotonin and a variety of polypeptide hormones. The most commonly identified of the latter have been corticotropin, gastrin, somatostatin, glucagon, secretin, and pancreatic polypeptide; in many cases, the tumors have been immunoreactive for multiple hormones.
Ovarian Tumors with Paraendocrine Disorders
A variety of paraendocrine disorders have been described in association with numerous types of ovarian tumor, some manifested by signs and symptoms of a well-known endocrine disease and others by subclinical laboratory abnormalities, indicating ectopic production of hormones or hormone-like substances by the tumor cells. In some of these cases, the hormone being produced has been identified, whereas in others, such as in cases of hypercalcemia, the mechanism of the disorder remains unclear. In all the cases included within this category of neoplasms, successful therapy of the tumor has led to disappearance of the paraendocrine state.
Hypercalcemia, Including Small Cell Carcinoma
About 60% of ovarian tumors causing hypercalcemia have been a distinctive type of small cell carcinoma (Young et al. 1994b); almost half the remainder have been clear cell carcinomas with serous carcinomas, squamous cell carcinoma arising in a dermoid cyst, dysgerminoma, and miscellaneous other neoplasms, each accounting for about one quarter of the remainder.
The mechanism of the hypercalcemia associated with ovarian cancers is unknown (Nussbaum et al. 1990). Attempts to demonstrate parathormone (PTH) within the tumor cells have been unsuccessful with rare exceptions, and in several cases in which PTH has been measured in the serum, the level was normal. Recent evidence has implicated PTH-related peptide (PTHRP) which has been elevated in the serum (Hoekman et al. 1991; Tsunematsu et al. 2000) or detected by radioimmunoassay in some cases (Fujino et al. 1992); three tumors (one, a small cell carcinoma) were immunoreactive for PTHRP (Tsunematsu et al. 2000; Burton et al. 1990). Because of the binding of PTHRP to a receptor common for PTH and PTHRP, the secretion of PTHRP by a neoplasm may produce the biochemical features of hyperparathyroidism. In one case, the patient also had abnormally high serum concentrations of 1,25-dihydroxyvitamin D (1,25-DHD) and increased intestinal calcium absorption (Hoekman et al. 1991). Tumor removal was followed by normalization of the serum calcium, PTHRP, and 1,25-DHD levels, suggesting an intestinal contribution to the maintenance of the hypercalcemia in this patient.
Small Cell Carcinoma, Hypercalcemic Type
The small cell carcinoma of hypercalcemic type is the most common form of undifferentiated carcinoma of the ovary in females under 40 years of age and has been accompanied by elevated levels of calcium in 66% of the cases in which it has been measured (Young et al. 1994b). Although the tumor may be seen at any age, the mean age is 25, and the vast majority (over 90%) occur in women under 40 years of age (Witkowski et al. 2016). The presenting symptoms usually are abdominal pain and swelling.
Immunohistochemical loss of SMARCA4 protein expression secondary to SMARCA4 mutations (see below) appears to be highly sensitive and specific for the diagnosis, being seen in 94% of tested small cell carcinomas of hypercalcemic type, and among other ovarian tumors only in about 4% of clear cell carcinomas and a single dedifferentiated endometrioid adenocarcinoma (Clark et al. 2016), neither of which would realistically enter into the differential diagnosis. Interpretation as abnormal requires complete loss of expression in tumor cell nuclei, with retention in the surrounding nontumoral cells. A few of the rare cases reported with retention of SMARCA4 protein have had loss of SMARCB1 (INI1) expression. Other immunohistochemistry is highly nonspecific; focal staining with cytokeratins and EMA is often present, as well as CD10 and calretinin (Aguirre et al. 1989b; McCluggage et al. 2004; Benfield et al. 1982).
The small cell carcinoma often is confused with a granulosa cell tumor of either adult or juvenile type. The features of these three types of tumor are contrasted in Table 3. Diffuse small cell carcinomas also may resemble malignant lymphomas, particularly on low-power examination, but adequate sampling reveals patterns of growth that indicate the epithelial nature of the tumor; also, the cytologic features of the neoplastic cells are incompatible with any form of malignant lymphoma. Immunohistochemistry for lymphoid markers readily distinguishes challenging cases. Exceptionally, the differential diagnosis of a small cell carcinoma includes other small cell malignant tumors of the ovary, including several metastatic tumors such as metastatic melanoma and metastatic small cell sarcomas. Interestingly, one case of metastatic melanoma to the ovary has been reported to show loss of SMARCA4 staining, although other immunohistochemical markers of melanoma would readily distinguish between these possibilities. Primary or metastatic pulmonary type small cell carcinomas, despite similar names, have limited morphologic or clinical overlap and are readily distinguished by neuroendocrine markers and SMARCA4 immunohistochemistry.
Recently, several groups independently identified a recurrent mutation involving SMARCA4, which may be germline or somatic, in the vast majority of small cell carcinomas of hyperkalemic type (Jelinic et al. 2014; Kupryjańczyk et al. 2013; Ramos et al. 2014; Witkowski et al. 2014). The protein encoded by this gene is referred to as SMARCA4 or BRG1 and is part of the SWI/SNF chromatin remodeling complex. Mutation results in loss of function and consequent loss of immunohistochemical expression, as discussed above. Even among patients with no family history, the rate of germline mutations is 35% (including 100% of patients under 15 years of age), so all patients with this diagnosis should be referred for genetic counseling (Witkowski et al. 2016).
Special staining, immunohistochemical (Aguirre et al. 1989b), and ultrastructural (McMahon and Hart 1988) examination have not revealed any features that identify the specific cell type of this tumor; dense-core granules have been absent in most cases. Flow cytometry on paraffin-embedded material typically shows that the cells are diploid (Eichhorn et al. 1992). Based on the age distribution and the characteristic presence of uniform small cells and follicle formation, some have suggested a possible sex cord derivation. Others have proposed a germ cell origin, based on focal staining for SALL4, as well as a few reported cases of the tumors occurring in association with germ cell tumors, specifically 2 mature cystic teratomas and 3 immature teratomas, 2 of which also had yolk sac tumor components (McCluggage et al. 2017). The presence of mucinous epithelium is not indicative of its derivation, as this could be evidence of surface epithelial derivation, teratomatous glandular epithelium, or heterologous mucinous differentiation analogous to that seen in Sertoli-Leydig cell tumors. The close functional relationship between SMARCA4 to SMARCB1 has raised consideration that small cell carcinoma of hypercalcemic type might represent a malignant rhabdoid tumor of the ovary, as has evidence of rare SMARCA4 mutations in malignant rhabdoid tumor and SMARCB1 mutations in small cell carcinoma, and some morphologic overlap between the two tumors. We do not find the evidence for a germ cell origin or rhabdoid nature convincing. The tumor is currently classified as a miscellaneous ovarian neoplasm in the 2014 World Health Organization classification (Kurman et al. 2014).
Small cell carcinoma, hypercalcemic type has a poor prognosis. In a recent study that evaluated all previously reported cases and the authors’ previously unreported cases (293 total cases), only 44% of patients were alive at last follow-up (Witkowski et al. 2016). However, no case has been reported of recurrence after 5 years, so patients who reach that milestone appear to have a good likelihood of long-term survival. Stage was the most important prognostic feature in that study, with 5-year survival of 55% for Stage 1, 40% for stage 2, 29% for Stage 3, and 0% for stage 4. All stage 4 patients died within 13 months. This observational study found significantly better overall survival in patients treated with high-dose chemotherapy with autologous stem cell rescue, with only 25% of such patients relapsing, compared to 65% of patients treated with any other modality. The discovery of SMARCA4 mutations in this tumor has raised the prospect of targeted therapy.
Five cases of clinically typical and biochemically documented Cushing’s syndrome have been caused by cortisol production by a steroid cell tumor (Young and Scully 1987). Most of the tumors occurred in adults and had metastasized within the abdomen at the time of presentation and had atypical cytologic features. Rarely, primary ovarian tumors other than those of steroid cell type have been associated with Cushing’s syndrome, probably in most cases on the basis of ectopic production of corticotropin or corticotropin-releasing factor.
These cases have included bilateral endometrioid adenocarcinoma (Crawford et al. 1994), a poorly differentiated adenocarcinoma (Parsons and Rigby 1958), a malignant Sertoli cell tumor (Nichols et al. 1962), a trabecular carcinoid (in which the tumor cells were immunoreactive for corticotropin) (Scully et al. 1998), and a tumor that resembled an atypical carcinoid and small cell carcinoma of the lung (Brown and Lane 1965). Finally, two cases have been described in which anterior pituitary tissue within a dermoid cyst caused Cushing’s syndrome. In one of these cases, it was not clear whether the pituitary tissue was neoplastic or hyperplastic, but in the other case, there was a chromophobe adenoma in which the neoplastic cells were immunoreactive for corticotropin (Cocconi et al. 1985).
Human Chorionic Gonadotropin Secretion
Ectopic hCG production was reported by Civantos and Rywlin (1972) in three women with serous papillary or mucinous adenocarcinomas of the ovary. All the patients had elevated urinary hCG level. Each tumor contained poorly differentiated areas with cells resembling syncytiotrophoblast cells; these cells were positive for hCG on immunofluorescence. In one case, the contralateral ovary contained numerous lutein cells, and a decidual reaction was present in the endometrium; that patient had vaginal bleeding, but no endocrine effects were present in the others. Two poorly differentiated surface epithelial carcinomas with a choriocarcinomatous component had elevated serum hCG levels (Oliva et al. 1993). The choriocarcinoma appeared grossly as a necrotic, hemorrhagic, circumscribed, brown nodule. Matias-Guiu and Prat (1990) conducted the most extensive immunohistochemical investigation of hCG in ovarian tumors, using single polyclonal antibodies to the whole hormone and its beta subunit and four monoclonal antibodies to the whole hormone, its beta subunit, and two regions of the carboxyl terminal of the beta subunit. Correlating positive staining results with the presence or absence of an “active” stroma of the tumor (luteinization and/or “condensation”), these authors found that the epithelial cells of 41% of the tumors with active stroma reacted with the polyclonal antibodies and 62% with the monoclonal antibodies; the corresponding figures for the epithelial cells of the tumor with an inactive stroma were 14% and 37%, respectively.
Six ovarian neoplasms have been associated with hypoglycemia: a serous cystadenocarcinoma, a dysgerminoma, a fibroma, a malignant schwannoma, a strumal carcinoid, and a carcinoid tumor with a mixed insular and trabecular pattern (Scully et al. 1998; Ashton 1995; Morgello et al. 1988). In the case of the malignant schwannoma, insulin and proinsulin were recovered from the tumor tissue and the cells of the carcinoid tumor were immunoreactive for insulin. The patient with the insular-trabecular carcinoid tumor also had a parathyroid adenoma and pituitary hyperplasia.
Renin and Aldosterone Secretion
Thirteen cases of hypertension related to hormone secretion by an ovarian tumor have been reported; two of the patients also had Gorlin’s syndrome. In eight cases, the hypertension was associated with a renin-secreting tumor, hyperreninism, and secondary hyperaldosteronism (Anderson et al. 1989; Korzets et al. 1986). In three cases, an aldosterone-secreting ovarian tumor resulted in primary hyperaldosteronism associated with low or normal plasma renin levels (Jackson et al. 1986; Kulkarni et al. 1990; Todesco et al. 1975). Elevated aldosterone levels were present in the 12th case (but plasma renin levels were not measured) (Scully et al. 1998) and in the 13th case (neither renin nor aldosterone levels were determined) (Scully et al. 1998), reported in 1966. In four cases, the tumor also elaborated steroid hormones as manifested by isosexual pseudoprecocity in two cases and elevated serum levels of estradiol and testosterone in two. Eight of the ovarian tumors were interpreted as sex cord-stromal tumors and two as steroid cell tumors. Three tumors in the first category were well-differentiated Sertoli cell tumors, whereas the other four had an appearance that was too nonspecific or poorly differentiated to subclassify. One of these last four tumors occurred in a woman with Peutz–Jeghers syndrome and was benign, whereas the other three were clinically malignant and two were fatal. One of the “steroid cell” tumors occurred in a 7-year-old girl and had a prominent follicular pattern, more in keeping with a diagnosis of JGCT. The final two tumors were a leiomyosarcoma and a mucinous adenocarcinoma. Immunohistochemical staining in five of the sex cord-stromal tumors and the leiomyosarcoma showed cells containing immunoreactive renin or prorenin.
Ovarian Tumors Associated with Paraneoplastic Syndromes
Nervous System Disorders
Ovarian cancer of surface epithelial type is one of the malignant tumors most often associated with nervous system disorders. The tumor is occasionally occult (Mason et al. 1997). A variety of lesions affecting both the gray matter and white matter of the cerebrum, cerebellum, spinal cord, the peripheral nerves, and the myoneural junction, accompanied by myasthenia gravis, may occur (Tyler 1974). Paraneoplastic subacute cerebellar degeneration (SCD) is one of the most common lesions, with ovarian cancer accounting for from 16% to 47% of the cases (Peterson et al. 1992). The cerebellar manifestations usually antedate recognition of the cancer, and typically there is no improvement after the removal of the tumor. In one case, manifestations of the cerebellar degeneration partially regressed after plasmapheresis (Cocconi et al. 1985). The pathogenesis of SCD in these cases appears to be related to the presence of circulating anti-Purkinje cell antibodies that have been shown to react with antigens in the tumor. The presence of such antibodies appears to be much more common in patients with SCD and gynecologic or breast cancer than in patients with SCD and other types of carcinoma.
An unusual form of inflammatory limbic encephalitis has been associated with teratomas (Dalmau et al. 2008; Sabin et al. 2008). These cases have been linked to antibodies against the N-Methyl-D aspartate receptor and are commonly referred to as anti-NMDAR encephalitis. Similar antibodies can develop in patients in the absence of a neoplasm or rarely associated with other neoplasms, but among young women with the disorder (the most common demographic), about half have a teratoma, most commonly dermoid cysts but also immature teratomas. In these patients, therapy including surgical resection of the neoplasm and immunotherapy usually leads to eventual recovery.
Connective Tissue Disorders
The connective tissue disorder most commonly associated with ovarian cancer is dermatomyositis. In one study of ten patients with dermatomyositis or polymyositis and a malignant tumor of the female genital tract, five had ovarian cancer (Verducci et al. 1984). The tumors are most commonly high-grade, high-stage serous carcinomas (Mordel et al. 1988), but there is one case of dysgerminoma and one of leiomyoma associated with dermatomyositis. The onset of the dermatomyositis generally precedes recognition of the tumor, which usually becomes evident within 2 years. Medsger and his associates (Medsger et al. 1982) described six patients with ovarian carcinoma in whom polyarthritis and palmar fasciitis preceded the diagnosis of carcinoma by 5–25 months. The arthritic symptoms were similar to those of rheumatoid arthritis. Four of the ovarian tumors were endometrioid carcinomas, one a serous carcinoma, and one an undifferentiated carcinoma. Occasional patients have had hypertrophic pulmonary osteoarthropathy, rheumatoid arthritis, scleroderma, systemic lupus erythematosus, or the shoulder–hand syndrome.
Acanthosis nigricans occurs in some, typically young, women in association with polycystic ovary disease (POD), stromal hyperthecosis, or combinations thereof (Dunaif et al. 1985), representing a component of the so-called HAIR-AN syndrome (hyperandrogenemia, insulin resistance, and acanthosis nigricans). Four cases of so-called malignant acanthosis nigricans have been cases of ovarian carcinoma. The sign of Leser–Trelat, the sudden onset and rapid increase in size of numerous seborrheic keratoses in association with an occult cancer, considered by some a variant of “malignant” acanthosis nigricans, has been associated with ovarian cancer in one case (Holguin et al. 1986). Rarely, ovarian carcinomas have occurred in patients with the Torre–Muir syndrome (Cohen et al. 1991) or Sweet’s syndrome (Nguyen et al. 1983). Cutaneous melanosis has occurred with a strumal carcinoid (Ashton 1995); the tumor cells were immunoreactive for alpha-melanocyte-stimulating hormone. Other uncommon associations of ovarian neoplasms and skin lesions are sporadically reported (Fathizadeh et al. 1982).
In cases of the nephrotic syndrome, 5–10% have a paraneoplastic background, although the causative tumors in such cases are located only rarely in the female genital tract. Hoyt and Hamilton (Hoyt and Hamilton 1987) described a 65-year-old woman who was found to have the nephrotic syndrome 8 months before the detection of a stage IV poorly differentiated serous carcinoma of the ovary. A renal biopsy showed membranous glomerulopathy. The proteinuria markedly diminished after debulking of the tumor, and after 10 months of combination chemotherapy, the proteinuria had disappeared and there was no evidence of tumor at a second-look laparotomy.
Approximately 30 ovarian tumors have been reported to be associated with autoimmune hemolytic anemia, which is usually Coombs positive. Most of these tumors have been dermoid cysts (Payne et al. 1981), but occasional examples of carcinoma and a single case of granulosa cell tumor also have been reported (Payne et al. 1981). In the latter case, the patient also had splenic angiomas. In many cases, corticosteroid therapy, splenectomy, or both have resulted in little or no improvement, but removal of the ovarian tumor has produced a rapid remission of the hemolytic disorder. Payne and coworkers (Payne et al. 1981) have listed several mechanisms proposed to explain the relation of the dermoid cyst to the anemia: (1) liberation by the tumor of a substance that alters the surface of red cells, making them antigenic to the host, (2) stimulation of production of an antibody that cross-reacts with the red cells by an antigen in the wall or lumen of the cyst, and (3) direct production of a red cell antibody by the tumor. Support for the last theory is provided by the finding of immunoglobulin in the cyst fluid in several cases.
Ovarian tumors are commonly associated with laboratory evidence of disseminated intravascular coagulation (DIC), but clinical manifestations of this disorder are uncommon. Ovarian tumors also have been associated with migratory thrombophlebitis (Trousseau’s syndrome). Nonbacterial thrombotic endocarditis also has been recorded as a complication of ovarian cancer, as has microangiopathic hemolytic anemia. Excluding cases of mild erythrocytosis that may accompany androgenic ovarian tumors, paraneoplastic erythrocytosis is associated only rarely with ovarian tumors (Payne et al. 1981). Examples of erythropoietin-secreting ovarian tumors have included a dermoid cyst and a steroid cell tumor (Payne et al. 1981). Other hematologic abnormalities that have been described rarely in association with ovarian tumors include nonthrombocytopenic purpura (mucinous cystadenoma), granulocytosis (clear cell carcinoma), thrombocytosis (serous carcinoma), thrombocytopenia (hemangioma and adenofibroma) (von dem Borne et al. 1990), and pancytopenia (granulosa cell tumor) (Napoli and Wallach 1976).
Miscellaneous Rare Conditions
One ovarian example of inappropriate antidiuresis syndrome has been reported in association with a serous carcinoma that had a component of small cell carcinoma of pulmonary type (Taskin et al. 1996). Electron microscopic examination of the tumor disclosed neuroendocrine granules, and the neoplastic cells were also immunoreactive for antidiuretic hormone.
A small number of patients with ovarian tumors, usually with high-stage serous carcinomas but rarely with low-stage endometrioid carcinomas, have had hyperamylasemia (O’Riordan et al. 1990). Monitoring the serum amylase level may be a marker of tumor progression and response to therapy in such cases. Rarely, patients with ovarian cancer have had a clinical presentation that has mimicked that of acute pancreatitis (Norwood et al. 1981). In one unusual case, a patient with pseudo Meigs’ syndrome had a pleural effusion rich in amylase that disappeared after removal of a serous tumor or borderline malignancy, which had ruptured preoperatively (Cramer and Bruns 1979). Six patients with poorly differentiated surface epithelial carcinomas (Chahud et al. 2001) have had uveal melanocytic lesions as may be seen with other visceral cancers. Microscopic examination of the eyes of these patients who undergo progressive blurring and loss of vision shows bilateral diffuse proliferation of melanocytes throughout the uveal tracts with the involvement of the sclerae in some instances (Margo et al. 1987). Rarely, pyrexia is a presenting manifestation of a patient with ovarian carcinoma (Schofield et al. 1985). One of the two cases in this category was classified as a clear cell carcinoma, and the other was not subclassified.
When the late Dr. Ancel Blaustein prepared the first edition of this book in the mid-1970s, he asked Dr. Robert E. Scully to contribute the chapter on ovarian sex cord-stromal tumors and it duly appeared in 1977. In the 1980s, one of us (RHY) had the pleasure and privilege to coauthor the chapter, beginning with the version that appeared in the 1987 third edition. Dr. Scully, being retired, declined to participate in the 2011 edition of the chapter and although many of the words were initially written, or edited by him, given his customary awareness of the importance of ethics in medicine, he declined to have his name associated with a work which he did not actively participate in at this time. However, the authors would like to acknowledge our debt to Dr. Scully for the initial chapter, which laid the groundwork for everything to follow, the opportunity to learn from his cases, and most important of all, to be educated by him on the remarkable and fascinating microscopic spectrum encountered within the family of tumors reviewed in this chapter. Similar remarks pertain to chapter “Metastatic Tumors of the Ovary”.
- Baker PM, Oliva E (2004) Immunohistochemistry as a tool in the differential diagnosis of ovarian tumors: an update. Int J Gynecol Pathol 24:39–55Google Scholar
- Banner EA, Dockerty MB (1945) Theca cell tumors of the ovary. A clinical and pathological study of twenty-three cases (including thirteen new cases) with a review. Surg Gynecol Obstet 81:234–242Google Scholar
- Bjorkholm E, Silfversward C (1980) Theca-cell tumors. Clinical features and prognosis. Acta Radiol Oncol Radiat Phys Biol 19:241–244Google Scholar
- Clark BA, Witkowski L, Ton Nu TN, Shaw PA, Gilks CB, Huntsman D, Karnezis AN, Sebire N, Lamovec J, Roth LM, Stewart CJ, Hasselblatt M, Foulkes WD, McCluggage WG (2016) Loss of SMARCA4 (BRG1) protein expression as determined by immunohistochemistry in small-cell carcinoma of the overy, hypercalcaemic type distinguishes these tumors from their mimics. Histopathology 69:727–738CrossRefGoogle Scholar
- Conlon N, Silva A, Guerra E, Jelinic P, Schlappe BA, Olvera N, Mueller JJ, Tornos C, Jungbluth AA, Young RH, Oliva E, Levine D, Soslow RA (2016) Loss of SMARCA4 expression is both sensitive and specific for the diagnosis of small cell carcinoma of ovary, hypercalcemic type. Am J Surg Pathol 40:395–403PubMedPubMedCentralCrossRefGoogle Scholar
- Costa MJ, Ames PF, Walls J, Roth LM (1997) Inhibin immunohistochemistry applied to ovarian neoplasms: a novel, effective, diagnostic tool. Hum Pathol 281:247–254Google Scholar
- Garcia-Villaneuva M, Figuerola NB, del Arbol LR, Ortiz MJH (1990) Zollinger–Ellison syndrome due to a borderline mucinous cystadenoma of the ovary. Obstet Gynecol 75:549–551Google Scholar
- Gurumurthy M, Bryant A, Shanbhag S (2014) Effectiveness of different treatment modalitis for the management of adult-onset granulosa cell tumors of the ovary (primary and recurrent). Cochrane Database Syst Rev 21:CD006912Google Scholar
- Irving JA, Young RH (2008a) Granulosa cell tumors of the ovary with a pseudopapillary pattern: a study of 14 cases of an unusual morphologic variant emphasizing their distinction from transitional cell neoplasms and other papillary ovary tumors. Am J Surg Pathol 32:581–586PubMedCrossRefGoogle Scholar
- Irving JA, Lee C, Yip S, Oliva E, McCluggage WG, Young RH (2015) Microcystic stromal tumor: a distinctive ovarian sex cord-stromal neoplasm characterized by FOXL2, SF-1, WT-1, cyclin D1, and beta-catenin nuclear expression and CTNNB1 mutations. Am J Surg Pathol 39:1420–1426PubMedCrossRefGoogle Scholar
- Kupryjańczyk J, Dansonka-Mieszkowska A, Moes-Sosnowska J, Plisiecka-Hałasa J, Szafron L, Podgórska A, Rzepecka IK, Konopka B, Budziłowska A, Rembiszewska A, Grajkowska W, Spiewankiewicz B (2013) Ovarian small cell carcinoma of hypercalcemic type – evidence of germline origin and SMARCA4 gene inactivation. A pilot study. Pol J Pathol 64:238–246PubMedCrossRefPubMedCentralGoogle Scholar
- Kurman RJ, Carcangiu ML, Herrington CS, Young RH (eds) (2014) WHO classification of tumours of female reproductive organs, 4th edn. International Agency for Research on Cancer, LyonGoogle Scholar
- Ramos P, Karnezis AN, Craig DW, Sekulic A, Russell ML, Hendricks WP, Corneveaux JJ, Barrett MT, Shumansky K, Yang Y, Shah SP, Prentice LM, Marra MA, Kiefer J, Zismann VL, McEachron TA, Salhia B, Prat J, D’Angelo E, Clarke BA, Pressey JG, Farley JH, Anthony SP, Roden RB, Cunliffe HE, Huntsman DG, Trent JM (2014) Small cell carcinoma of the ovary, hypercalcemic type, displays frequent inactivating germline and somatic mutations in SMARCA4. Nat Genet 46:427–429PubMedPubMedCentralCrossRefGoogle Scholar
- Ravishankar S, Mangray S, Kurkchubasche A, Yakirevich E, Young RH (2016) Unusual Sertoli cell tumor associated with sex cord tumor with annular tubules in Peutz-Jeghers syndrome: report of a case and review of the literature on ovarian tumors in Peutz-Jeghers syndrome. Int J Surg Pathol 24:269–273PubMedCrossRefPubMedCentralGoogle Scholar
- Riopel MA, Perlman EJ, Seidman JD et al (1998) Inhibin and epithelial membrane antigen immunohistochemistry assist in the diagnosis of sex cord-stromal tumors and provide clues to the histogenesis of hypercalcemic small cell carcinomas. Int J Gynecol Pathol 17:46–53PubMedCrossRefPubMedCentralGoogle Scholar
- Rishi M, Howard LN, Bratthauer GL, Tavassoli FA (1997) Use of monoclonal antibodies against human inhibin as a marker for sex cord-stromal tumors of the ovary. Am J Surg Pathol 19:927–933Google Scholar
- Schultz KAP, Harris AK, Finch M, Dehner LP, Brown JB, Gershenson DM, Young RH, Field A, Yu W, Turner J, Cost NG, Schneider DT, Stewart DR, Frazier AL, Messinger Y, Hill DA (2017) DICER1-related Sertoli-Leydig cell tumor and gynandroblastoma: clinical and genetic findings from the first 107 cases in the international ovarian and testicular stromal tumor registry. Gynecol Oncol 147:521–527Google Scholar
- Scully RE, Young RE, Clement PB (1998) Tumors of the ovary, maldeveloped gonads, fallopian tube, and broad ligament. In: Atlas of tumor pathology, 3rd series, fasc 23. Armed Forces Institute of Pathology, Washington, DCGoogle Scholar
- Sjostedt S, Wahlen T (1961) Prognosis of granulosa cell tumors. Acta Obstet Gynecol Scand 40:1–26Google Scholar
- Staats PN, Coutts MA, Young RH (2010) Primary ovarian mucinous cystic tumor with prominent theca cell proliferation and focal granulosa cell tumor in its stroma: case report, literature review, and comparison with Sertoli-Leydig cell tumor with heterologous elements. Int J Gynecol Pathol 29:228–233PubMedCrossRefPubMedCentralGoogle Scholar
- Stall JN, Young RH (2018) Granulosa cell tumors of the ovary with prominent thecoma-like foci: a report of ten cases emphasizing the ongoing utility of the reticulum stain in the modern era. Int J Gynecol Pathol (in press)Google Scholar
- von dem Borne AEGK, van Oers RHJ, Wiersinga WM et al (1990) Complete remission of autoimmune thrombocytopenia after extirpation of a benign adenofibroma of the ovary. Br J Rheumatol 74:119–120Google Scholar
- Witkowski L, Carrot-Zhang J, Albrecht S, Fahiminiya S, Hamel N, Tomiak E, Grynspan D, Saloustros E, Nadaf J, Rivera B, Gilpin C, Castellsagué E, Silva-Smith R, Plourde F, Wu M, Saskin A, Arseneault M, Karabakhtsian RG, Reilly EA, Ueland FR, Margiolaki A, Pavlakis K, Castellino SM, Lamovec J, Mackay HJ, Roth LM, Ulbright TM, Bender TA, Georgoulias V, Longy M, Berchuck A, Tischkowitz M, Nagel I, Siebert R, Stewart CJ, Arseneau J, McCluggage WG, Clarke BA, Riazalhosseini Y, Hasselblatt M, Majewski J, Foulkes WD (2014) Germline and somatic SMARCA4 mutations characterize small cell carcinoma of the ovary, hypercalcemic type. Nat Genet 46:438–443PubMedCrossRefPubMedCentralGoogle Scholar
- Witkowski L, Goudie C, Ramos P, Boshari T, Brunet JS, Karnezis AN, Longy M, Knost JA, Saloustros E, McCluggage WG, Stewart CJR, Hendricks WPD, Cunliffe H, Huntsman DG, Pautier P, Levine DA, Trent JM, Berchuck A, Hasselblatt M, Foulkes WD (2016) The influence of clinical and genetic factors on patient outcome in small cell carcinoma of the ovary, hypercalemic type. Gynecol Oncol 141:454–460PubMedCrossRefPubMedCentralGoogle Scholar
- Zhao C, Bratthauer GL, Barner R, Vang R (2007b) Comparative analysis of alternative and traditional immunohistochemical markers for the distinction of ovarian Sertoli cell tumor from endometrioid tumors and carcinoid tumor: a study of 160 cases. Am J Surg Pathol 31:255–266PubMedCrossRefGoogle Scholar
- Zhao C, Barner R, Vinh TN, McManus K, Dabbs D, Vang R (2008) SF-1 is a diagnostically useful immunohistochemical marker and comparable to other sex cord-stromal tumor markers for the differential diagnosis of ovarian Sertoli cell tumor. Int J Gynecol Pathol 27:507–514PubMedCrossRefPubMedCentralGoogle Scholar