Cyclin D1 as a diagnostic immunomarker for endometrial stromal sarcoma with YWHAE-FAM22 rearrangement.

Endometrial stromal sarcoma (ESS) is a histologically and genetically heterogeneous group of uterine sarcoma. The majority, particularly those with a classic low-grade histologic appearance, contain a genetic fusion involving JAZF1 and members of polycomb complex gene ( SUZ12, PHF1, EPC1 ) most frequently resulting from a chromosomal translocation, t(7;17). 6 , 13 , 16 , 26 , 28 , 29 , 31 We recently described and characterized a series of ESS with YWHAE-FAM22A/B ( YWHAE-FAM22 ) genetic fusion resulting from t(10;17)(q22;p13). 22 Clinically, the majority of patients with YWHAE-FAM22 ESS present with evidence of extrauterine spread, in contrast to patients with JAZF1 ESS (encompassing JAZF1/SUZ12/PHF1/EPC1 -rearranged tumors), in which the disease is usually confined to the uterus at initial presentation. Patients with YWHAE-FAM22 ESS also experience more frequent recurrences compared to those with JAZF1 ESS. In contrast to JAZF1 ESS, YWHAE-FAM22 ESS nearly always display a cellular high-grade round cell morphologic appearance with larger nuclei, more irregular nuclear contour, high mitotic activity and frequent necrosis. This high-grade round cell component is accompanied in about half of the cases by a low-grade bland spindle cell component (with low mitotic activity) with a fibrous to fibromyxoid stromal background. 21 The high-grade round cell area of YWHAE-FAM22 ESS exhibits an undifferentiated appearance immunohistochemically, being negative for estrogen receptor (ER), progesterone receptor (PR) and CD10, in contrast to the frequent diffuse ER, PR and CD10 immunoreactivity seen in JAZF1 ESS while the low-grade spindle cell component is consistently immunoreactive for ER, PR and CD10, an immunohistochemical profile that overlaps with that seen in JAZF1 ESS. The consistent lack of ER and PR expression in the high-grade round cell component implies that hormonal therapy, which has been used as a systemic therapy for ESS, will likely be ineffective against YWHAE-FAM22 ESS. 34 Because of the difference in clinical behavior and its potential therapeutic implications, the distinction of YWHAE-FAM22 ESS from JAZF1 ESS is important. However, while YWHAE-FAM22 ESS exhibits higher grade histologic features compared to JAZF1 ESS, it can in some instances be challenging to distinguish these different ESS subsets on histologic grounds alone, particularly given their rarity. Furthermore, YWHAE-FAM22 ESS also needs to be distinguished from other high-grade typically pleomorphic uterine sarcomas including undifferentiated endometrial sarcoma (UES) and leiomyosarcoma, which in contrast to YWHAE-FAM22 ESS, are highly aggressive with a five year survival of less than 50%. 1 , 18 While molecular confirmation (RT-PCR or FISH analysis) currently represents the standard to establish a definitive diagnosis of YWHAE-FAM22 ESS, these tests are presently offered only at a few centers. Thus, there is a need to identify diagnostic immunomarker(s) that would be more accessible to practicing pathologists in general. In order to potentially identify such a marker, we analyzed published gene expression data of YWHAE-FAM22 ESS and JAZF1 ESS. Following identification of a potential marker, we then rigorously evaluated its diagnostic utility by examining its sensitivity and specificity for YWHAE-FAM22 ESS in the context of uterine mesenchymal and mixed epithelial-mesenchymal tumors.

Formalin-fixed paraffin-embedded tumor tissues were obtained from ESS in which FISH studies had demonstrated either YWHAE-FAM22 rearrangement (n=12), JAZF1/SUZ12/PHF1/EPC1 rearrangement (n=34) or no demonstrable genetic rearrangement (n=21); these were from the pathology archives at Brigham and Women’s Hospital, Massachusetts General Hospital, Vancouver General Hospital and Toronto General Hospital. The histologic features of all tumors were reviewed by two authors (C.H. Lee and M. R. Nucci) and previously described. 21 A comparison group consisted of tissue microarrays of 8 undifferentiated endometrial sarcomas with nuclear uniformity (UES-U), 18 13 undifferentiated endometrial sarcoma with nuclear pleomorphism (UES-P), 119 uterine leiomyosarcomas (LMS, 80 primary and 39 metastatic from uterine primary), 26 uterine adenosarcomas (8 with sarcomatous overgrowth), 13 uterine carcinosarcomas, 4 uterine PEComas, 49 uterine leiomyomas, 7 polypoid endometriosis, 3 endometrial stromal nodules and 2 uterine tumors resembling ovarian sex cord tumor (UTROSCT). 6 , 20 , 23 All the tumors in the comparison group lacked YWHAE rearrangement by FISH study. The study was approved by the Institutional Review Boards at the respective institutions. 3′ end sequencing-based gene expression profile data on a series of 3 YWHAE-FAM22 ESS and 4 JAZF1 ESS was previously published by our group. 22 Significance of microarray analysis was used to identify differentially expressed genes between different tumor groups (37) and a false-discovery rate (FDR) of <1% was considered to be significant. Immunohistochemical analysis was performed on whole sections of 12 YWHAE-FAM22 ESS and on whole sections or tissue microarray sections of 298 other gynecologic tumors. Cyclin D1 (Thermoscientific, clone SP4, rabbit monoclonal, 1:50) immunostaining was performed following heat-induced antigen retrieval using CC1 antigen retrieval buffer (Ventana Medical Systems, AZ, USA). After incubation with the primary antibodies, sections were stained on the Ventana automated slide stainer (NEXES, Tucson, AZ) using the Ventana diamino-benzidine (DAB) detection kit (Ventana Medical Systems, Tucson, AZ). Only nuclear staining was considered positive. Strong cyclin D1 staining was defined as having the same intensity as cyclin D1-positive mantle cell lymphoma control, moderate as definite staining that is visible at low power (4x objective) but weaker than cyclin D1-positive mantle cell lymphoma control, and weak as nuclear staining that require high power (20-40x) examination to recognize ( Figure 2C ). Percentage of nuclei showing moderate to strong nuclear cyclin D1 staining was assessed by evaluating representative high-power fields (for whole tissue sections) or the entire tissue microarray cores. Tumors showing ≥ 70% moderate to strong nuclear staining for cyclin D1 were considered to be positive.

The result of filtered 3′end sequencing performed on a series of 3 YWHAE-FAM22 ESS and 4 JAZF1 ESS was previously described. 22 We performed SAM analyses comparing YWHAE-FAM22 ESS to JAZF1 ESS and identified a number of differentially expressed genes with a false discovery rate of < 0.01% ( Table 1 ). CCND1 (which encodes Cyclin D1) was among the top five most upregulated genes in YWHAE-FAM22 ESS compared to JAZF1 ESS. Given the general availability of cyclin D1 in pathology laboratories, this potential diagnostic biomarker was selected for immunohistochemistry validation studies in YWHAE-FAM22 ESS. Cyclin D1 immunohistochemistry was evaluated in 12 YWHAE-FAM22 ESS, 4 of which also contained a low-grade fibrous/fibromyxoid spindle cell component in the material available for the study, while the other 8 contained pure high-grade round cell morphology. All tumors showed ≥ 70% moderate to strong cyclin D1 nuclear staining of the tumor cells (average 91%) with an homogenous staining in the high-grade round cell components ( Figure 1 ). In the 4 cases where a low-grade spindle cell component was present, cyclin D1 immunostaining demonstrated significant intra-tumoral variability in all cases ( Figure 2 ), which ranged from less than 1% to more than 50% nuclear staining in different areas of the same tumor. The staining intensity in the low-grade spindle cell component was weaker (weak to moderate intensity) in contrast to the strong staining intensity seen in the high-grade round cell component ( Figure 2 ). A series of 34 JAZF1 ESS were also examined and 31 displayed absent to very focal cyclin D1 nuclear staining (≤ 5%), typically in the form of scattered positive single cells ( Figure 3 and 4 ). In histologically classic low-grade ESS with no detectable genetic rearrangements, the pattern of cyclin D1 staining was similar to that seen in JAZF1 ESS, with 19 of 21 showing ≤5% nuclear staining ( Figure 4 ). The cyclin D1 immunostaining results of other gynecologic mesenchymal and mixed epithelial/mesenchymal tumors are summarized in Table 2 and illustrated in Figure 5 . A total of 8 UES with nuclear uniformity (UES-U) lacking YWHAE or JAZF/PHF1 rearrangements were studied and 2 showed diffuse cyclin D1 staining ( Figure 5C ). Both cyclin D1 positive UES-U displayed strong diffuse CD10 immunoreactivity and were negative for ER and PR. All 13 UES with nuclear pleomorphism (UES-P) showed focal (<20%) cyclin D1 staining ( Figure 5D ). Other than UES-U, diffuse strong cyclin D1 was rarely seen in other uterine mesenchymal or mixed epithelial-mesenchymal tumors; only 1 of 80 primary uterine leiomyosarcomas (and 0 of 39 metastatic leiomyosarcomas) examined showed ≥70% cyclin D1 nuclear staining ( Figure 5E-F ); this tumor had a prominent histiocytic and lymphocytic infiltrate ( Figure 5F ). Thus, in this series, cyclin D1 immunostaining showed a sensitivity of 100% and specificity of 99% as a diagnostic marker for YWHAE-FAM22 ESS, with positive and negative predictive values of 80% and 100%, respectively.

We recently described the histologic features of YWHAE-FAM22 ESS. The tumors were characterized by the presence of a high-grade round cell component where the tumor cell nuclei are larger in size, more irregular in contour and demonstrate greater mitotic activity and necrosis compared to the tumor cells in JAZF1 ESS. Even though RT-PCR and/or FISH analysis of YWHAE-FAM22 genetic rearrangement should be considered the gold standard for diagnosis, our studies show that cyclin D1 immunohistochemistry is an informative diagnostic adjunct, which can be used to support a provisional diagnosis of YWHAE-FAM22 ESS and screen-in cases for confirmatory molecular YWHAE-FAM22 analyses. Among uterine tumors, the differential diagnosis for the high-grade round cell area of YWHAE-FAM22 ESS includes other genotypes of ESS with increased mitotic activity and/or necrosis (that correspond to either JAZF1 ESS or ESS without demonstrable genetic rearrangements), undifferentiated endometrial sarcomas (particularly UES-U), leiomyosarcoma (particularly epithelioid variant), PEComa and sarcoma-predominant carcinosarcoma. As shown here, diffuse strong cyclin D1 immunoreactivity is consistently observed in the high-grade round cell component of YWHAE-FAM22 ESS. Although diffuse cyclin D1 immunostaining was seen in a subset of UES-U and rare case of uterine leiomyosarcoma in this study, we have not seen this in JAZF1 ESS, ESS without demonstrable rearrangements, undifferentiated endometrial sarcomas with nuclear pleomorphism (UES-P), PEComa, leiomyosarcoma, leiomyoma, adenosarcoma (including cases with sarcomatous overgrowth) and carcinosarcoma. Two of 8 UES-U in our current series showed diffuse cyclin D1 immunoreactivity comparable to that in YWHAE-FAM22 ESS, but lacked YWHAE (or JAZF1/PHF1) rearrangements by FISH analysis. Importantly, both cyclin D1 positive UES-U showed diffuse strong CD10 staining, in contrast to the high-grade round cell component of YWHAE-FAM22 ESS, which consistently lacked substantial CD10 staining. Therefore, when a combined panel of cyclin D1 and CD10 is applied to a histologically high-grade but non-pleomorphic uterine sarcoma, the finding of diffuse strong cyclin D1 in the absence of CD10 staining appears to be highly sensitive and specific for YWHAE-FAM22 ESS. Cyclin D1 immunoreactivity has been previously evaluated by Kurihara et al in both endometrial stromal tumors and UES, and they observed similar staining patterns to our present study. 19 None of the 8 endometrial stromal nodules (4 positive for JAZF1-SUZ12 genetic fusion by RT-PCR assay), 16 low-grade ESS (5 positive for JAZF1-SUZ12 genetic fusion) and 6 UES-P showed ≥ 70% nuclear cyclin D1 staining. Among 7 UES-U, 3 showed ≥ 70% nuclear cyclin D1 staining. However, it is unknown whether any of the UES-U harboured YWHAE rearrangement or exhibited CD10 immunoreactivity. Cyclin D1 immunoreactivity has also been previously examined in uterine carcinosarcoma. 7 , 14 De Jong et al found 7 of 31 tumors to show cyclin D1 positivity in the mesenchymal component, 7 however, these authors used a much lower nuclear cut-off staining (10%). In our series, 4 of 13 carcinosarcomas would have been classified as positive for cyclin D1 if the same lower cut-off would have been used. Kanthan et al reported no significant cyclin D1 expression of the mesenchymal component of 23 uterine carcinosarcomas, 14 which is in concordance with the present findings. While the low-grade spindle cell component of YWHAE-FAM22 ESS is seen next to the high-grade round cell component in about half of the tumors, it is possible that a biopsy may sample only the low-grade spindle cell area. The histologic differential for the low-grade spindle cell component of YWHAE-FAM22 ESS includes fibrous/fibromyxoid variant of ESS, well-differentiated leiomyosarcoma and the sarcomatous overgrowth of an adenosarcoma. The low-grade spindle cell component of YWHAE-FAM22 ESS showed overall weak and focal cyclin D1 nuclear staining with significant intra-tumoral heterogeneity, ranging from absent to diffuse weak/moderate staining. Cyclin D1 immunostaining is therefore of limited diagnostic value for identifying YWHAE-FAM22 ESS when the tissue sample shows a low-grade spindle cell proliferation in a fibrous/fibromyxoid background. Further studies are necessary to determine whether there is significant difference in cyclin D1 immunostaining pattern between the low-grade spindle cell component of YWHAE-FAM22 ESS and fibrous/fibromyxoid variant of ESS lacking known genetic rearrangement. 9 , 13 , 30 , 40 With these considerations in mind, we propose diagnostic algorithm incorporating morphologic and immunophenotypic features to identify uterine YWHAE-FAM22 ESS. Among non-gynecologic type sarcomas that may present as a pelvic mass, Ewing sarcoma/peripheral neuroectodermal tumor, gastrointestinal stromal tumors (GIST), rhabdomyosarcoma (both alveolar and embryonal type), malignant solitary fibrous tumors, and less frequently epithelioid malignant peripheral nerve sheath tumors (MPNST) and round cell liposarcomas can possess histologically high-grade round cell areas that could potentially resemble YWHAE-FAM22 ESS. Nuclear cyclin D1 immunostaining has been found to be focal (<50% nuclear staining) in malignant solitary fibrous tumors, 3 round cell liposarcomas, 8 MPNST, 17 alveolar and embryonal rhabdomyosarcomas. 36 Nakamura et al observed an average of 12% nuclear immunostaining (standard deviation of 21%) in a series of 88 GIST, but rare cases (number of cases not specified) showed ≥ 70% nuclear staining. 27 Strong membranous/cytoplasmic KIT and DOG1 immunoreactivity would suggest a diagnosis of GIST. 24 In Ewing sarcomas, a significant subset (13 of 31) can show diffuse strong nuclear cyclin D1 immunostaining. 10 Furthermore, YWHAE-FAM22 ESS can also show strong diffuse CD99 immunostaining. 2 This overlap in immunophenotype does present a diagnostic challenge when one encounters a malignant round cell tumor in the biopsy of a pelvic mass. While the presence of a concurrent low-grade fibromyxoid spindle cell component and/or the presence of moderate eosinophilic cytoplasm in the round cell component would favor YWHAE-FAM22 ESS, these features may not be present in the tumor or be represented in the sampled tissue. Moreover, rosette formation may occur in Ewing’s sarcoma (although not described to date in those of the uterus) and YWHAE-FAMM 22 ESS, therefore, this histologic finding is of little utility in this distinction. 2 In this selected scenario, molecular analysis (FISH or RT-PCR) is necessary to confirm the diagnosis of YWHAE-FAM22 ESS versus Ewing sarcoma. In contrast to YWHAE-FAM22 ESS with its t(10;17), the vast majority of Ewing’s sarcoma show a t(11;22)(q24;q12), which results in the fusion of EWS with FLI1 ; in a minority, EWS is involved with variant partners: t(21;22)(q12;q12) and t(7;22)(p22;q12), leading to EWS-ERG and EWS-ETV1 fusions, respectively. In the scenario of uterine tumor consisting purely of monomorphic high-grade round cells with a cyclin D1-positive and CD10/ER/PR-negative immunoprofile, the presence of a myopermeative growth pattern would indicate a YWHAE-FAM22 ESS, as all molecularly confirmed uterine Ewing sarcomas reported to date formed an expansile uterine mass and lacked the extensive and characteristic finger-like growth pattern that is consistently observed in uterine YWHAE-FAM22 ESS. 4 , 33 , 35 Cyclin D1 (also known as BCL1) is a diagnostic immunomarker best known for its utility in the diagnosis of mantle cell lymphoma. 5 , 39 In the case of mantle cell lymphoma, genetic rearrangement between IGH@ and CCND1 results in upregulated cyclin D1 expression at mRNA and protein levels. 32 , 38 Cyclin D1 regulates cell cycle progression in G1/S transition and elevated cyclin D1 is observed in various types of human malignancy including sarcomas. 11 , 12 , 15 , 25 The mechanisms underlying cyclin D1 upregulation in tumors include chromosomal translocation, amplification and increased protein stability. 15 Notably, although cyclin D1 immunostaining extent and intensity in YWHAE-FAM22 ESS is comparable to that in mantle cell lymphoma, the mechanism of cyclin D1 overexpression in YWHAE-FAM22 ESS has not been determined. Further studies are ongoing to address this issue. In conclusion, we have identified and demonstrated in this study the utility of cyclin D1 as a sensitive and specific diagnostic immunomarker for YWHAE-FAM22 ESS. Given the general availability of cyclin D1, this approach is useful in supporting a provisional diagnosis of YWHAE-FAM22 ESS when encountering a uterine tumor in which the differential of a histologically high-grade ESS is considered.