(P 246) THE GENOMIC AND IMMUNE LANDSCAPE OF LIPOSARCOMA AND ITS SUBTYPES

Objective: Liposarcoma (LPS) is one of the most common soft tissue sarcomas, constituting a heterogeneous group of malignant tumors of adipocytic differentiation. It is classified in to four subtypes which include well differentiated (WDLS), dedifferentiated (DDLS), myxoid/round cell (MLS/RCLS), and pleomorphic (PLS), which are each characterized by their own distinct biologic and clinical features.

LPS subtype is one of the most important determinants of clinical outcome, and improved understanding of subtype intrinsic pathogenesis at genetic and transcriptomic levels is necessary for advancements in clinical decision-making and the development of new therapeutics. Further, the role of the tumor immune microenvironment within LPS, especially within each subtype, remains poorly described and understood. This is critical for the application and utilization of novel immune modulators in the treatment of LPS.

We aim to describe the unique genomic and immune landscape of each subtype of LPS utilizing a database of molecularly profiled real-world patient tumor samples.

Methods: Patient samples (N=344) underwent DNA (592- gene or whole exome) and RNA (whole transcriptome) sequencing at CLIA-certified laboratory (Caris Life Sciences, Phoenix, AZ). These included WDLS (N=35), DDLS (N=226), MLS (N=56), PLS (N=31), and a subset not otherwise specified (N=6). Microenvironment Cell Populations-counter (MCP-counter) and quanTIseq bioinformatic tools were used to create immuno-expression profiles. PD-L1+ expression was assessed by immunohistochemistry (IHC; SP142, positive expression defined by 2+ intensity, 5% cell stains), and high tumor mutational burden (TMB-H) was defined by 10 mutations/Mb. Genomic loss-of-heterozygosity (gLOH) was analyzed using 250,000 SNPs to divide chromosomes into 552 segments, with gLOH-High defined by 16% of segments with skewed SNP frequencies (+/-15% from 50%).

Results: Consistent with prior reports, MLS was enriched with DDIT3 fusions (N=21/56, 37.5%; primarily fused to FUS), TERT promoter mutations (N=24/36, 66.7%), and PIK3CA mutations (N=18/56, 32.1%). PLS was enriched with TP53 (N=18/29, 62.1%), RB1 (N=5/19, 26.3%), and ATRX (N=4/26, 15.4%) mutations. WDLS and DDLS were enriched with copy number amplification (6 copies) of MDM2 (N=21/25, 84.0%; N=197/220, 89.5%, respectively), CDK4 (N=17/25, 68.0%; N=189/220, 85.9%), and HMGA2 (N=11/24, 45.8%; N=100/199, 50.3%) amplifications.

PDL1+ IHC was most common in DDLS (N=28/218, 12.8%), but also observed in WDLS (N=2/23, 8.7%) and PLS (N=2/30, 6.7%). PDL1+ status in PLS samples was concurrent with TMB-High (N=2/2) and/or dMMR/MSI-High (N=1/2) status, which were absent in WDLS, DDLS and MLS. gLOH-High was most common in DDLS (N=28/151, 18/5%). T cell-inflamed scores, which are predictive of response to immunotherapy, were lowest in MLS (3.4 arbitrary units vs 4.0-4.3; P< 0.01).

Estimation of immune cell abundance by RNA deconvolution suggests unique tumor immune microenvironments (TMEs) across LPS subtypes. When utilizing MCP-counter, WDLS had the highest median abundance of most cell populations, with notably high B cell abundance (median >10-fold higher than other subtypes). DDLS had the highest median abundance of fibroblasts (1.5 to 2.7-fold higher than other subtypes). MLS had low abundance of monocytic lineage, T cells, and myeloid dendritic cells (0.56, 0.76, and 0.66-fold lower, respectively, compared to the overall median). PLS has generally lower immune cell abundance, with notably lower endothelial cell abundance (0.54 to 0.72-fold compared to other subtypes). In addition to B cells, the abundance of T cells and myeloid dendritic cells was significantly higher in WDLS versus DDLS (1.7 and 1.5-fold, respectively, P< 0.05).

RNA deconvolution by quanTIseq demonstrated that NK cell, and dendritic cell fractions were significantly increased in MLS compared to all other subtypes (>1.3 and 2.6-fold, respectively, P<.05). Neutrophils, CD8 T cells, M1, and M2 macrophages were significantly lower in MLS compared to all other subtypes (< 0.30, 0.26, 0.37, and 0.41-fold, P< 0.05). The cell fraction of neutrophils, CD8 T cells, and B cells was significantly higher in WDLS versus DDLS (1.3, 1.4, and 1.5-fold, P< 0.05).

Conclusion: There is significant heterogeneity in the TME and mutational profiles across LPS subtypes. MLS has a particularly immune-cold TME shown by lower infiltration of lymphocytes, T cell-inflamed scores, and PDL1+ prevalence, while DDLS has a relatively immune-hot profile. These findings are critical to develop immune modulatory strategies for LPS and provide novel insights for future studies.