Low-grade fibromyxoid sarcomas (LGFMS) bear either the t(7,16) (q32-34;p11) or t(11,16) (p11;p11) translocations, resulting in FUS-CREB3L2 or FUS-CREB3L1 fusions, respectively. Heretofore, fusion transcripts were mainly detected in frozen tissues, using reverse transcription-polymerase chain reaction. In this study, we aimed to develop a reliable method to detect these in paraffin-embedded tissues, and to examine the clinicopathologic characteristics of a series of translocation-positive LGFMS. Sixty-three neoplasms with typical morphologic features of LGFMS and 66 non-LGFMS tumors selected for their resemblance to LGFMS (LGFMS-like tumors) were examined. RNA of sufficient quality could be extracted from 111/129 (86%) cases (59 LGFMS, 52 non-LGFMS). Of all, 48/59 (sensitivity, 81%) LGFMS contained detectable transcripts (45 FUS-CREB3L2, 3 FUS-CREB3L1). Most relevant clinicopathologic features of fusion-positive LGFMS included predominance in lower extremities (22/48; thigh: 13/48), deep situation (46/48), and occasional presence of unusual histologic features, for example, hypercellular areas (16/48), foci of epithelioid cells (13/48), and giant rosettes (6/48). Most tumors expressed EMA (41/45), at least focally, CD99 (38/41) and bcl-2 (36/41) while being essentially negative for CD34 (2/45), mdm2 (1/41), smooth muscle actin (1/45), S100 protein (0/46), desmin (0/44), h-caldesmon (0/42), keratins (0/44), and CD117 (0/40). Eleven presumed LGFMS were fusion negative. Of all, 7/52 non-LGMFS neoplasms contained FUS-CREB3L2 transcripts, of which 4 had been diagnosed as sclerosing epithelioid fibrosarcoma. In conclusion, FUS-CREB3L1/L2 fusion transcripts can be detected in paraffin-embedded LGFMS in a sensitive manner, using reverse transcription-polymerase chain reaction. Most fusion-positive LGFMS are EMA-positive and CD34/S100/smooth muscle actin negative. The presence of epithelioid cells and fusion transcripts in both LGFMS and a subset of sclerosing epithelioid fibrosarcoma suggest that these neoplasms might be related.

Atypical lipomatous tumor/well-differentiated liposarcomas and dedifferentiated liposarcomas are characterized by the amplification of MDM2 and CDK4 genes. To evaluate the accuracy of fluorescence in situ hybridization (FISH) analysis in the differential diagnosis of adipose tissue tumors, we investigated MDM2-CDK4 status by FISH, real-time polymerase chain reaction (PCR) [quantitative PCR (Q-PCR)] and immunohistochemistry (IHC) in a series of 200 adipose tumors. First, we evaluated MDM2-CDK4 amplification and expression in a series of 94 well-defined adipose tissue tumors. Results showed that FISH was interpretable in 45 of 50 cases (90%), and was more specific and sensitive than Q-PCR and IHC. We then used the same techniques as complementary diagnostic tools in a series of 106 adipose and soft tissue tumors of unclear diagnosis to distinguish between (i) lipomas and atypical lipomatous tumor/well-differentiated liposarcomas, (ii) malignant undifferentiated tumors and dedifferentiated liposarcomas, and (iii) a variety of benign tumors and liposarcomas. Our results indicate that although helpful, IHC alone is often insufficient to solve diagnostic problems. FISH and Q-PCR methods gave concordant results and were equally informative in most cases. However, the proportion of noninterpretable cases was slightly higher with FISH than with Q-PCR. When tumor cells represented a minor component of the tumor tissue, such as with inflammatory tumors, FISH was more powerful than Q-PCR by allowing visualization of individual cells. In conclusion, we recommend that the evaluation of MDM2-CDK4 amplification using FISH or Q-PCR be used to supplement IHC analysis when diagnosis of adipose tissue tumors is not possible based on clinical and histologic information alone.

Data concerning the fine structure of the 12q13-15 amplicon which contains MDM2 and CDK4 in well-differentiated and dedifferentiated liposarcomas (WDLPS/DDLPS) are scarce. We investigated a series of 38 WDLPS/DDLPS using fluorescence in situ hybridization analysis with 17 probes encompassing the 12q13-15 region. In addition, using quantitative RT-PCR we studied the expression of MDM2, CDK4, DDIT3 (CHOP/GADD153), DYRK2, HMGA2, TSPAN31 and YEATS4 (GAS41) in 11 cases. We showed that CDK4 (12q14.1) belonged to a distinct amplicon than MDM2 (12q15). There was no continuity in the amplified sequences between MDM2 and CDK4. Moreover, while MDM2 was amplified and overexpressed in all cases, CDK4 was not amplified or overexpressed in 13% of cases. The centromeric border of the CDK4 amplicon was located immediately downstream the 5 0 end of DDIT3, a gene known for being involved in myxoid liposarcoma translocations. DDIT3 was amplified in 3 cases and overexpressed in 9 cases. The overexpression of DDIT3 was correlated to the CDK4 amplification and not to its own amplification status. This suggested that the CDK4 amplicon, as well as the overexpression of DDIT3, might be generated by the disruption of a fragile region in 5 0 DDIT3. HMGA2 was always amplified and rearranged indicating that it plays a central role in WDLPS/DDLPS. HMGA2 rearrangement frequently resulted in a loss of the 3 0 end region that is a binding site for let-7. We also found a frequent amplification and overexpression of YEATS4, an oncogene that inactivates P53, suggesting that YEATS4 might play an important role together with MDM2 in WDLPS/DDLPS oncogenesis.

No abstract