Introduction

Endometrial stromal tumors (EST) were first described by Norris and Taylor as a distinct group of mesenchymal neoplasms of the uterus based on their morphological resemblance to endometrial stroma [1]. They classified endometrial stromal sarcomas (ESS) based on mitotic activity into endolymphatic stromal myosis (≤ 10 mitoses/10 HPFs) and stromal sarcoma (> 10 mitoses/10 HPFs) [1]. However, Evans pointed out that marked nuclear pleomorphism but not mitotic activity had an impact on prognosis, labeling tumors with marked cytologic atypia as undifferentiated endometrial sarcomas (UES), typically a diagnosis of exclusion [2]. Numerous attempts to subcategorize ESS based on cytologic features and mitotic index have been attempted with limited success [1,2,3,4]. In the largest study to date, Chang and colleagues reported that in tumors confined to the uterus, mitotic activity had no bearing on prognosis [4]. These findings were reflected in the WHO 2003 classification with only two categories of stromal sarcomas being recognized based on cytologic atypia, namely low-grade endometrial stromal sarcoma (LG-ESS) and UES [5]. Later, Kurihara and colleagues reported a third group of ESS characterized by cells with uniform but definite nuclear atypia, permeative myometrial invasion, and a clinical behavior intermediate between LG-ESS and UES [6]. Increasing knowledge on cytogenetics and molecular biology of these high-grade EES led to their reinstitution by the current WHO classification as a distinct group of ESS, known now to be characterized by a t(10;17) leading to the YWHAE-NUTM2 rearrangement (HG-ESS) [7,8,9]. Therefore, the current WHO classification acknowledges four categories within the endometrial stromal family of tumors: endometrial stromal nodule, LG-ESS, high-grade endometrial stromal sarcoma, and undifferentiated uterine sarcoma [9].

This review intends to briefly summarize the morphologic and immunohistochemical features of the well-known categories of endometrial stromal neoplasms and highlight the most relevant discoveries related to the molecular classification of endometrial stromal neoplasms with a special emphasis on HG-ESS (Table 1).

Table 1 Main morphological, immunohistochemical, and molecular features of endometrial stromal tumors

Endometrial stromal nodule and low-grade endometrial stromal sarcoma

Endometrial stromal nodule and LG-ESS share morphological, immunohistochemical, and molecular features. As both tumors display uniform small cells with scant cytoplasm and oval nuclei reminiscent of proliferative phase endometrial stroma that sometime whorl around arterioles, distinction is based on the finding(s) of myometrial infiltration and/or lymphovascular invasion [1, 9, 10]. Endometrial stromal nodule is typically seen on gross examination as a well circumscribed yellow to tan often polypoid endometrial mass (rarely intramyometrial or subserosal) [10, 11]. On microscopic examination, it is also well circumscribed but may have finger-like projections into the myometrium, typically ≤ 3, each measuring < 3 mm in greatest extent [9, 10]. The presence of lymphovascular invasion excludes this diagnosis [9]. In contrast, LG-ESS often shows yellow to brown coalescent nodules and “worm-like” plugs of tumor infiltrating the myometrium as well as myometrial and parametrial veins (Fig. 1a) [9]. In some instances, distinction between these two entities is not easy and extensive sampling may be required with some tumors showing very limited (albeit more than allowed in ESN) myometrial invasion (EST with limited infiltration) [11]. There is scarce and conflicting information in the literature regarding the behavior of this subgroup of EST. In the two reported cases with available follow-up, one patient was alive and well after 62 months [11]. In the other report, they describe a very small (0.7 cm) tumor with apparent complete sampling of the myometrial interface, where a lymphovascular invasion was detected in the right pelvic soft tissues in the staging surgery [12]. Despite this, the patient was alive and well after 13 months [12]. Thus, more studies on ESTs with limited infiltration are needed, although it may be difficult due to the rarity of this scenario. In hysteroscopic or morcellated specimens, even though careful gross evaluation of fragments should be performed, when myometrial invasion is not overt as often seen in ESS, certainty on the tumor myometrium interface and thus infiltration may be limited. In such scenario, a diagnosis of EST with a comment stating that the distinction between a ESN and a LG-ESS with minimal infiltration cannot be made is reasonable.

Fig. 1
figure 1

Low-grade endometrial stromal sarcoma. a On gross examination, the tumor has an intracavitary tan to yellow polypoid component, extensive myometrial permeation as worm-like plugs of tumor, as well as massive extrauterine extension. b Small, round, monotonous cells with a diffuse growth that whorl around small vessels are characteristic of this tumor and reminiscent of proliferative-phase endometrial stroma. c The tumor cells are typically strongly and extensively CD10 positive. d 1. The chromosomal translocation t(7;17) (most common in these tumors) results in a chimeric transcript containing the first three (red) of five total exons of the JAZF1 gene (chromosome 7) fused with the 15 exons (blue) of the SUZ12 gene (chromosome 17). 2. Integrative Genome Viewer (V2.4.10.03) screenshot of the JAZF1-SUZ12 gene fusion depicts numerous split JAZF1 (exon3) to SUZ12 (exon 2) reads; note that JAZF1 is encoded on the minus strand

Other morphologic features that can be seen in both tumors include arterioles, sometimes hyalinized, resembling those seen in proliferative-phase endometrium (characteristic but infrequently striking), foamy histiocytes, and hyaline plaques (Fig. 1b). Areas of stromal hyalinization may be seen [4, 9, 10]. Smooth muscle metaplasia (most commonly seen as starburst pattern) [11, 13,14,15,16,17,18,19,20,21,22], myxoid/fibromyxoid change [11, 17, 23,24,25,26,27], sex cord-like differentiation (Fig. 2a) [10, 11, 16, 20, 21, 23, 28,29,30,31,32], endometrioid-type glands [22, 33,34,35], adipocytic [18] and skeletal muscle differentiation [18, 21, 36], pseudopapillae [37], rhabdoid cells [20, 23, 38], or cells with granular eosinophilic epithelioid [39] or clear cytoplasm [40], and osteoclast-like cells [19] or cells with bizarre nuclei [18] have also been reported.

Fig. 2
figure 2

Low-grade endometrial stromal tumors with sex-cord areas (a). Tumors with this morphology have been reported to carry more frequently PHF1 rearrangements (b)

Tumor cells characteristically express CD10 (Fig. 1c), ER, PR, and WT1 [41,42,43,44]. Smooth muscle actin is also commonly positive [14, 42]. Desmin and caldesmon can be positive, more commonly the former in both endometrial stromal cells and areas with smooth muscle differentiation [14, 41, 45]. Over 70% of tumors are positive for androgen receptors [46], and some investigators have shown nuclear β-catenin positivity [6, 47]. Variable degree of keratin expression may be seen. [11, 48, 49] Areas of sex cord differentiation may be positive for inhibin, calretinin, CD99, melanA, and WT1 [50,51,52], but also for keratins and smooth muscle markers [29, 30]. IFITM1, a novel endometrial stromal marker reported to be positive in the majority of these tumors, has been stated to better discriminate between low-grade endometrial stromal and smooth muscle tumors when compared to CD10; however, more experience is needed to validate these results [53]. BCOR, cyclin D1, KIT, and DOG1 are typically negative, although focal cyclin D1 and c-kit positivity may be seen [54,55,56,57,58,59,60].

Molecular genetic studies have shown recurrent chromosomal translocations involving most frequently t(7;17)(p15;q21) resulting in a JAZF1-SUZ12 fusion transcript (Fig. 1d) [61,62,63]. Other reported molecular alterations include t(6;7)(p21;q21), t(6;10;10)(p21;q22;p12), t(1;6)(p34;p21), t(5;6)(q31;p21), and t(X;17), corresponding respectively to PHF1-JAZF1, PHF1-EPC1, MEAF6-PHF1, PHF1-BRD8, and MBTD-CXorf67 fusions [64,65,66,67,68]. The t(7;17)(p15;q21) leading to a JAZF1-SUZ12 fusion transcript is the most common reported rearrangement, estimated to be present in ~ 50% of LG-ESS and ~ 65% of endometrial stromal nodules and in a small percentage of undifferentiated uterine sarcomas (likely those that originate from low-grade tumors) [6, 63, 64, 69,70,71,72,73,74,75,76,77]. Although both JAZF1 and SUZ12 mRNA is expressed at low levels in normal endometrium, it appears to derive from RNA trans-splicing and only be present in a transient from during part of the cycle [78]. However, JAZF1-SUZ12 fusion is specific of endometrial stromal nodules and low-grade ESS among mesenchymal tumors of the uterus, but it appears to be less common among variants including smooth muscle, fibromyxoid, sex cord-like and epithelioid morphology [63, 79]. JAZF1 has also been shown to rearrange with the PHD finger 1 protein gene (PHF1) from 6p21 [64]. The PHF1 gene itself has been found to be involved in rearrangements with other partners in LG-ESS including: EPC (10p11), MEAF6 (1p34), and BRD8 (5q31.2) [64,65,66, 68]. Yet, another rearrangement t(X;17) resulting in MBTD1-CXorf67 fusion has been also identified [67].

JAZF1, SUZ12, PHF1, and MBDT1 are members of the polycomb group protein family also involved in transcriptional repression [80]. The oncoproteins that result from JAZF1-SUZ12 and JAZ-PHF1 genetic fusions are believed to cause transcriptional dysregulation [64, 81]. EPC1 is part of the nucleosome acetyltransferase of histone H4 complex whereas MEAF6 is part of histone acetyltransferase multi-subunit complexes. The fusion proteins EPC1-PHF1 and MEAF6-PHF1 alter acetylation patterns of histone proteins, which lead to unraveling of the heterochromatin and aberrant gene expression [66, 80]. It appears that benign and low grade ESTs share genes that are known to be involved in transcriptional regulation. Despite being different, all these gene fusions appear to represent biologically and clinically equivalent oncogenic events in the tumorigenesis of this group of tumors [80]. In general, specific translocations do not correlate with specific morphology except for ESS with sex cord-like differentiation found to harbor more commonly alterations in the PHF1 gene in one study (Fig. 2b) [76]; however, these findings were not confirmed in another study [71]. Tumors with PHF1 alterations not infrequently are also associated with smooth muscle metaplasia [76].

Of interest, PHF1-related fusions involving various partners including EP400, MEAF6, and EPC1 have been reported in ossifying fibromyxoid tumors, which are rare soft tissue neoplasms with an intermediate risk of malignancy [82,83,84,85]. Morphologically, these are multinodular, well-circumscribed neoplasms with uniform to round desmin and S100 protein positive cells arranged in cords and nests set in a fibromyxoid stroma, an appearance that vaguely overlaps with some ESTs [83, 86]. One cardiac sarcoma likely representing an ossifying fibromyxoid tumor has also been reported to carry a JAZF1-PHF1 rearrangement [87]. ZC3H7B-BCOR, CREBBP-BCORL1, and KDM2A-WWTR1 fusions have also been reported in these tumors highlighting a significant genetic overlap with ESTs [83, 88].

Rarely, LG-EES harboring JAZF1 translocations have been described to be associated with high-grade areas with a faster progression [89], while some low-grade tumors, with either conventional or variant (fibromyxoid) morphology and t(10,17), have been reported associated with YWHAE rearrangement [90, 91]. Micci and colleagues have shown by microarray analysis a large number of genes to be differentially expressed between EES harboring JAZF1-SUZ12 fusion and those without it [80]. In the same study, genes known to be implicated in acute myeloid leukemia (CREBBP and MLLT4) were also noted to be present for the first time in a LG-ESS by next generation sequencing [80]. These findings suggest that the number of ESS-associated genes is likely to increase in the future [80].

As ESN are benign tumors, they are appropriately treated by hysterectomy. In hysteroscopic or morcellated specimens, where the complete myometrial interface cannot be evaluated, hysterectomy should ensue. In women of reproductive age who desire to preserve fertility, adjuvant hormonal therapy together with follow-up diagnostic imaging and hysteroscopy is a potential option [92,93,94]. Most patients with LG-ESS present with stage I disease and have an excellent prognosis with an overall five-year survival exceeding 90%; however, it is much lower in patients with stage II-IV tumors [4]. Total hysterectomy and bilateral salpingooophorectomy, with or without adjuvant therapy, are the treatment of choice [8, 95,96,97], although hormone therapy including aromatase inhibitors or radiation therapy may be alternative options [95,96,97]. Relapses which can occur long after the initial diagnosis respond to anti-estrogen therapy [95, 98, 99].

High-grade endometrial stromal sarcomas

In the current WHO classification, high-grade endometrial sarcomas are solely represented by those harboring t(10;17) rearrangement leading to the YWHAE-NUTM2 fusion protein [7,8,9]. Recent evidence has shown that endometrial sarcomas harboring other genetic alterations namely in the BCOR gene including those with BCOR tandem internal duplications are also associated with a prognosis that is worse than LG-ESS but better than UES and thus, should be classified as HG-ESS [100,101,102,103]. However, other subsets of HG-ESS likely exist, including those arising from dedifferentiation in JAZF1-related ESS, potentially some within the category of tumors with NTRK fusions or others yet to be unveiled [103].

YWHAE-NUTM2 ESS

Patients with YWHAE-NUTM2 HG-ESS frequently present with advanced stage disease when compared to LG-ESS and are associated with frequent recurrences [8, 104]. On gross examination, tumors have a fleshy cut surface with common areas of necrosis and/or hemorrhage. Morphologically, they show either a destructive or permeative growth and uniform morphology at low-power magnification. Tumors are composed of round “blue” cells with high nuclear to cytoplasmic ratio, nucleomegaly (4–6 times the size of a stromal lymphocyte nuclei), slightly irregular or angulated nuclear contours without prominent nucleoli but typically with brisk mitotic activity (> 10/10 HPFs) associated with a rich but delicate arborizing capillary network (Fig. 3a) [8]. Focal pseudoglandular or rosette-like morphology as well as sex-cord like differentiation have been occasionally described [6, 8, 105]. In approximately half of the tumors, a low-grade component with fibroblastic/myxoid or much less commonly classic morphology can be found, imparting a biphasic appearance [6, 8, 104]. These low-grade and high-grade areas may or not be well-demarcated from each other [8]. However, there is increasing evidence of the morphologic heterogeneity of these neoplasms as recently it has been reported a YWHAE rearranged ESS with pure low-grade morphology in the primary uterine tumor that when progressed displayed fibroblastic as well as large atypical cells which eventually progressed to HG-ESS [90].

Fig. 3
figure 3

High-grade YWHAE-NUT2M rearranged endometrial stromal sarcoma. a A diffuse or vaguely nested pattern of cells with high nuclear to cytoplasmic ratio with round to angulated nuclei accompanied by brisk mitotic activity is characteristic of these tumors. Notice the delicate sinusoidal vasculature. b The tumor cells are typically diffusely and strongly positive for CyclinD1 but negative for CD10, ER, and PR. c These tumors are also positive for BCOR. d G-banded partial karyotype showing a balanced translocation, t(10;17). Courtesy of Paola dal Cin, PhD, Brigham, and Women’s Hospital, Boston, MA, USA

High- and low-grade components typically have opposite immunohistochemical profiles. The high-grade areas are diffusely (> 70% of cells) and strongly positive for cyclin D1 (Fig. 3b) [54, 106]. It has been stated that in the setting of a uterine mesenchymal neoplasm, sensitivity, and specificity of this antibody for the diagnosis of YWHAE, rearranged ESS is very high (100 and 99%, respectively) [104]. Rare tumors with diffuse cyclin D1 positivity lacking t(10;17) likely represent BCOR-related HG-ESS (see below) or HG-ESS with undiscovered genetic changes, while tumors that may only be weakly cyclin D1 positive or show positivity in < 70% of cells can still have a gene fusion confirmed by FISH [60, 101, 102, 104]. It is also important to keep in mind that rarely other tumors including undifferentiated uterine sarcoma, leiomyosarcoma, Ewing sarcoma, and undifferentiated endometrial carcinoma may be strongly positive for this marker [60, 104, 107]. CD117 is also characteristically positive in the high-grade component and can show focal positivity in the low-grade areas [108]. However, no KIT mutations have been detected in these tumors and DOG1 is typically negative in both components [108]. Recently, diffuse and strong BCOR immunostaining has been detected in this high-grade component of YWHAE-rearranged sarcomas (Fig. 3c) [60]. This marker may be especially useful in tumors with variant morphology or with typical morphology with minimal or absent cyclin D1 expression, and it has been stated that it is more reliable than cyclin D1 in the identification of this high-grade component [60]. Weak BCOR expression may be rarely encountered in endometrial stromal nodules and low-grade ESS and more frequently and of weak to moderate intensity in leiomyosarcomas [60]. CD99 has been reported to be positive in one tumor [105]. CD10, ER, and PR are negative, while p53 immunoreactivity is typically wild type [6, 104]. In contrast, the low-grade component is typically CD10, ER, and PR positive and lacks cyclin D1 expression but may show variable BCOR positivity [6, 54, 60].

The t(10;17) was initially reported in clear cell sarcomas of the kidney and only later discovered in ESS (Fig. 3d) [7, 8, 109]. This translocation leads to the fusion between YWHAE and either of two nearly identical NUTM2 proteins (NUTM2A or NUTM2B) resulting in a YWHAE-NUTM2 fusion oncoprotein [7]. Both FISH and RT-PCR have been used to detect YWHAE-NUT2M rearrangement although FISH is more commonly used [8, 77, 104]. It is important to keep in mind that YWHAE-NUT2M and JAZF1/SUZ12/EPC1/PHF1 rearrangements are mutually exclusive [8]. Of note, YWHAE, NUTM2A, and NUTM2B rearrangements have been shown by FISH in a uterine angiosarcoma, but no YHWAE-NUTM2 fusion transcript was identified by RT-PCR [110].

Prognosis of patients with this HG-ESS seems to be intermediate between LG-ESS and UES [8, 111]. Five-year survival is achieved in nearly 33% of patients with a median survival of 20 months, but disease control and long survivals have also be reported even in metastatic settings [8, 96, 111, 112]. Surgery and adjuvant chemotherapy, especially anthracycline-based regimens, and/or radiation therapy remain the treatment of choice [95, 112].

BCOR-related ESS

Recently, a new group of ESS harboring BCOR alterations have been described comprising both ZC3H7B-BCOR rearranged tumors and those with internal tandem duplications (ITD) in the last exons of the BCOR gene [60, 100,101,102]. These neoplasms often display high-grade morphologic features and available albeit limited clinical data suggests that they behave more aggressively than LG-ESS; thus, they have been placed within the HG-ESS category [60, 100,101,102].

Seventeen patients with ESS carrying the ZC3H7B-BCOR fusion have been reported [100, 101]. Median age at diagnosis was 54 (range 28–71) years. Seven (41%), three (18%), and seven (41%) presented with FIGO stage I, II, and III disease, respectively. Two of the five patients who underwent lymph node sampling had lymph node metastases. Clinical follow-up data was available for five of these patients, two with stage III and three with stage I disease. All patients developed recurrences and four (including all stage I patients) died of disease. Tumors ranged in size from 1.5 to 12 (median, 9.7) cm and five were polypoid. They frequently involved both endometrium and myometrium (11/13) and showed either a broad front pattern of invasion, a tongue-like infiltrative pattern (typical of LG-ESS), or both. They were composed of spindle cells arranged in haphazard fascicles sometimes often embedded in a myxoid stroma that ranged from focal to abundant, sometimes forming lakes (Fig. 4a, b). Spindle cells had ovoid to spindle nuclei with even chromatin and no significant pleomorphism in all but one case. Collagen plaques were observed in ~ 50%. Most tumors (14/17) had a mitotic rate ≥ 10/10 HPF, but it ranged from 1 to 50/10 HPF. Necrosis, mostly of infarct-type, was seen in 10/17 tumors.

Fig. 4
figure 4

BCOR rearranged high-grade endometrial stromal sarcoma. a The tumors often have a fascicular growth of spindle cells with oval nuclei with little pleomorphism and variable amounts of cytoplasm. b Myxoid stroma is common and can be abundant and may form lakes. c Neoplastic cells show diffuse reactivity for BCOR, although often of variable intensity. d Schematic representation of the fusion gene transcripts involving BCOR and ZC3H7B. To account for reciprocal fusions (with either ZC3H7B in 5′ position vs. BCOR in 5′ position), the various reported exon structures are provided along with the relevant publication (note: some prior publications refer to “ossifying fibromyxoid tumors”)

In contrast to HG-ESS with t(10,17), these tumors typically show diffuse positivity for CD10 (although it may be weak/focal) and may display focal staining for one myogenic marker (SMA>>>>desmin or caldesmon). ER and PR are variably expressed. Cyclin D1 is typically strong and diffuse (> 95% cells) in most tumors, while diffuse BCOR expression (> 95 cells) has been noted in half of them (all but one strong) Fig. 4c). All 17 tumors tested in the largest study to date harbored ZC3H7B-BCOR gene fusions by next-generation sequencing or FISH analysis (Fig. 4d) [100, 101]. The ZC3H7B-BCOR gene fusion had been previously reported in ESS by Panagopoulos and Micci in two separate studies [80, 113]. Tumors reported were CD10 positive and occurred in patients who presented with high-stage disease [113]. No morphological or clinical details were available on the case reported by Micci [80]. Furthermore, previous karyotypic data reported in two ESS showed t(X;22) likely to correspond to ZC3H7B-BCOR fusion; however, no further morphological details were available on these cases either [7].

A total of four ESS harboring ITDs involving BCOR (BCOR ITD) have been recently reported by two groups [60, 102]. They were identified through genomic PCR and targeted sequencing. BCOR ITD involved exon 15 (in three) and exon 16 (in one)(Fig. 5a) [60, 102]. Despite the small number of cases reported so far, they typically appear to occur in younger patients (ages 18, 22, 25, and 32) when compared to BCOR-rearranged tumors but seem to share with the latter, a similar morphology (although they may also be epithelioid) (Fig. 5a) and clinical course. In three, both permeative and destructive growth patterns were observed and in the remaining, only a tongue-like pattern of myometrial invasion was noted. In all but one tumor, a distinct myxoid stroma was observed whereas in the latter, stroma was collagenous. All neoplasms had spindle and round cells and three of them displayed cytological atypia. Lymphovascular invasion was noted in three out of four tumors in which this feature was evaluated [60, 102]. Follow-up was available in two patients even though stage was not mentioned in any of them; one died of recurrent disease 8 years and the other had no evidence of disease 22 years after diagnosis and anthracycline-based chemotherapy. All tumors display strong and diffuse cyclin D1 and BCOR positivity, with focal CD10 positivity in three [60, 102]. Focal desmin positivity was noted in one tumor (myogenic markers reported in three) [102]. ER and PR, reported in one tumor, were negative [60].

Fig. 5
figure 5

High-grade BCOR-ITD endometrial stromal sarcoma. a BCOR exon 16 internal tandem duplication in high-grade endometrial stromal sarcoma. (1) Schematic demonstrating wild type BCOR (top) and BCOR internal tandem duplication (bottom) sequences. (2) Duplicated region (highlighted in beige) overlapping with the BCOR wild type sequence with a 4-bp insertion (highlighted in blue) in between duplicated regions. Courtesy of Cristina Antonescu, MD, Memorial Sloan Kettering Cancer Center, New York, NY, USA. b The tumor is composed of round to oval cells in a slightly myxoid background with thin vessels

BCOR-related ESS share morphological and immunohistochemical features with myxoid leiomyosarcomas including spindled cells with myxoid background as well as positivity for CD10 and muscle markers and rarely diffuse cyclin D1 positivity in leiomyosarcomas. In fact, some myxoid leiomyosarcomas reported in the literature may likely represent BCOR-related ESS [60, 100, 114]. BCOR appears to be a sensitive marker in identifying YWHAE-rearranged sarcomas but weak to moderate BCOR positivity may be seen in ~ 20% of leiomyosarcomas and ~ 6% of LG-ESS; thus, it is important to use a panel of antibodies as well as extensive sampling to help to reach the correct diagnosis. [60, 100,101,102] Of note, ~ 25% of myxoid leiomyosarcomas show PLAG1 rearrangements, not reported so far in ESS [115].

Recently, a JAZF1-BCORL1-ESS initially diagnosed as a LG-ESS has also been reported and although no morphological description is provided, available pictures show in our opinion high-grade morphology. The tumor underwent an aggressive clinical course indicating that this rearrangement may also be seen in HG-ESS [116].

BCOR is a transcriptional corepressor involved in suppressing gene expression by either interacting with BCL6 or binding to PCGF1 as part of polycomb repressive complex 1 (PRC1) and inducing gene silencing by histone modification [117,118,119,120,121]. Germline BCOR loss of function mutations result in X-linked oculofaciocardiodental (OFCD) syndrome and Lenz microphthalmia [122]. BCOR rearrangements and inactivating mutations have been reported in various hematological and solid human cancers including acute myeloid leukemia, myelodysplastic syndrome, chronic myelomonocytic leukemia, medulloblastomas, retinoblastoma, hepatocellular carcinoma, soft tissue and bone round cell sarcomas, and CNS neoplasms [88, 123, 124]. ZC3H7B (ZC3H7B-BCOR) is involved in protein-nucleic acid interactions [80]. Of interest, ZC3H7B-BCOR rearrangement has also been described in ossifying fibromyxoid tumors, where rearrangements described in LG-ESS (MEAF6-PHF1, EPC1-PHF1) have also been described while BCOR ITD has been reported in tumors where YWHAE-NUTM2B/E fusions are also frequent including clear cell sarcoma of the kidney, undifferentiated round cell sarcoma of infancy, and primitive myxoid mesenchymal tumor of infancy [83, 109, 117, 125, 126]. These tumors also share morphologic features, upregulation of BCOR mRNA, overlapping gene signatures, and BCOR expression by immunohistochemistry [117, 127].

Little is known about the natural course and optimal treatment of BCOR-related ESS [60, 100,101,102]. The available clinical data suggests that most patients present with advanced stage tumors have local relapses as well as lymph node and distant metastases and respond poorly to adjuvant therapy [60, 101, 102].

Other high-grade ESS (dedifferentiated LG-ESS)

Rarely, conventional LG-ESS may be seen in association with a high-grade/undifferentiated sarcoma [6, 89, 128, 129], but these tumors are classified in the current WHO as undifferentiated uterine sarcomas [9]. In contrast to classical LG-ESS, they have a destructive pattern of invasion although permeative areas more reminiscent of LG-ESS can also be recognized [6, 89]. The high-grade component displays large epithelioid cells with nucleomegaly, prominent nucleoli, high mitotic index, and necrosis [6, 89]. In tumors tested, CD10 was often positive in the low-grade component but variably positive or negative in the high-grade areas. Reports concerning ER and PR have shown inconsistent results [89, 129, 130].

There are very few molecularly confirmed ESS with JAZF1-SUZ12 genetic fusion with high-grade features [6, 63]. One of them was a purely high-grade sarcoma classified as an UES [63]. The other contained classic low-grade ESS juxtaposed to a monomorphic high-grade sarcoma (70 mitoses/10 HPF) [6]. Gene expression studies have identified 514 differentially expressed genes between high-grade not otherwise specified and low-grade ESS [80]. Several investigators have also reported that chromosomal aberrations differ considerably between these two groups [80, 131], suggesting that they correspond indeed to different pathological entities and that tumor progression from one to the other is very unlikely [80].

Undifferentiated uterine sarcomas

Undifferentiated uterine sarcomas are high-grade sarcomas that lack specific lines of mesenchymal differentiation and include tumors arising in the endometrium and myometrium, thus the change in the nomenclature in the most recent WHO classification. All other uterine sarcomas as well as poorly differentiated and undifferentiated carcinomas and malignant mixed Müllerian tumors must be ruled out before a diagnosis of UUS is rendered. No recurrent genetic fusions have been described in this group of tumors [9].

In 2008, Kurihara et al. proposed a classification system for UES based on the degree of nuclear pleomorphism delineating two groups with different prognosis: uniform and pleomorphic UES [6]. Overexpression of cyclin D1 was a frequent event in the uniform UES group and this finding was later found to correlate with the YWHAE-FAM22 translocation, raising strong evidence that this group of tumors represented in fact YWHAE-FAM22 sarcomas [106]. The other group of tumors displayed overt pleomorphism growing in a destructive, sheet-like fashion [6]. These morphological findings were later confirmed and expanded by other groups [89, 132]. These tumors are frequently ER and PR negative [89, 132], while CD10 expression is variable with only a small number of cases tested [87]. Cyclin D1 has been reported to be positive in a few tumors but never showing the diffuse and strong staining seen in YWHAE-FAM22 or BCOR related tumors [89, 132]. Aberrant expression of p53 and TP53 gene missense mutations have been reported in 32–50% [6, 132]. Chromosomal alterations in the UES have been found to be heterogeneous, and complex karyotypes have been reported [133, 134]. There appears to be no accumulation of aberrations from LG-ESS to UES as also observed when comparing LG-ESS to HG-ESS [133].

These neoplasms are typically associated with poor outcomes, although a few patients with long survival have been reported [132, 135, 136]. Recently, two different groups have suggested that a mitotic count > 25/10 HPF is the most reliable prognostic discriminator in these group tumors [132, 136]. Surgical treatment is recommended, but the value and choice of adjuvant therapy are yet to be determined [95].

Very recently, a new uterine sarcoma subtype with features of fibrosarcoma harboring NTRK fusions has been described to arise in the uterus. All four tumors presented in premenopausal women and three arose in the cervix. All tumors shared the finding of monomorphic spindle cells typically with focal and moderate nuclear pleomorphism as well as brisk mitotic activity. Focal SMA and S100 expression was seen in all tumors but desmin, ER, PR, SOX10, and CD34 were consistently negative. NTRK rearrangement with various partners was identified in all tumors correlating with tropomyosin receptor kinase(Trk) pan-Trk expression in all tumors and TrkA in three of them. The origin of these tumors is not clear at this time, but it may be possible that they represent a subset of undifferentiated uterine sarcomas [103].

In summary, we have reviewed the widening spectrum of EST, focusing on the subgroup of HG-ESS. Although the current WHO classification only acknowledges YWHAE-related sarcomas in this group, we believe this will be soon modify as the amount of cytogenetic and molecular available evidence increases. Integration of clinical, morphological, and molecular data is becoming of paramount importance in the study of mesenchymal tumors of the uterus.