Myxoid/round cell liposarcoma misdiagnosis incidence, etiology, and importance of NY-ESO-1 testing: Retrospective analysis of the Tempus database.

11542 Background: Liposarcoma is composed of multiple unique sarcoma subtypes, specifically myxoid/round cell liposarcoma (MRCLS), well differentiated and dedifferentiated liposarcoma (WD LPS and DD LPS), and pleomorphic liposarcoma. These are genetically distinct entities with different treatment paradigms. MRCLS is driven by a translocation between DDIT3 and FUS or EWSR1, detectable in 100% of MRCLS. WD and DD LPS are characterized by amplification of MDM2 and CKD4, and pleomorphic LPS is genetically complex. Recently, new treatments for liposarcoma are specific to the LPS type (CDK4/6 inhibitors; CTA directed engineered T Cells) which highlights the importance of an accurate liposarcoma subtype diagnosis. Methods: A retrospective analysis of the tempus RNA database for DDIT3 translocations and NY-ESO-1 (CTAG1B) expression in MRCLS was conducted. 101 patients (pts) with MRCLS were identified; 27 pts lacked Next Generation Sequence data for tumor samples leaving 74 pts that were selected for analysis. RNA was sequenced with either the Tempus RS or the RS.v2 assays. Patients' tumor samples with tumor purity <30% were excluded from assessment of CTAG1B expression. Gene expression values in this analysis were normalized by transcripts-per-million (TPM) transformed to log2(TPM+1) units. The threshold of 2 (log2(TPM+1)) was set as the cut off of true positive for CTAG1B expression. Results: The DDIT3 fusion was identified in 65/74 (88%) patients. In 9/74 (12%) patients, there was no DDIT3 fusion detected. Out of the 65 patients that were positive for the MRCLS translocation 60/65 (92%) were DDIT3-FUS, and 5/65 (8%) were DDIT3-EWSR1. The MRCLS DDIT3 fusion-negative samples were 5/9 (56%) positive (pos) for MDM2 amplification and CDK4 amplification. The Tempus tumor of origin algorithm predicted all DDIT3 fusion-pos MRCLS samples to be soft tissue sarcoma (where predictions were made; 53 of 65 pts) and the DDIT3 fusion-neg samples to be fibrosarcoma in one case and liposarcoma in 7 cases. The last fusion-neg case did not have a tumor of origin prediction. The median expression of CTAG1B was 7.14 TPM in all patients, 7.2 TPM (range 3.1 to 8.7) in DDIT3-FUS samples, 7.5 TPM (range 5.7 to 7.7) in DDIT3-EWSR1 samples, and 0.04 (range 0.0 to 3.7) in DDIT3 fusion-neg samples. There was no difference in CTAG1B expression by biopsy site, metastatic status, stage, tumor purity, or center of diagnosis. Conclusions: It is of critical importance to accurately determine the liposarcoma subtype which influences the selection of new targeted treatments. This analysis highlights a 12% rate of misdiagnosis of MRCLS and suggests that WD/DD LPS can be misdiagnosed as MRCLS. Additionally, this study predicts that 100% of MRCLS with confirmed DDIT3 translocation will be NY-ESO-1 (CTAG1B) pos which highlights the utility of NY-ESO-1 as a diagnostic marker for NY-ESO-1 directed therapies.