(P 371) ELUCIDATING THE CLINICAL UTILITY OF TARGETED GENE EXPRESSION PROFILING IN RETROPERITONEAL LIPOSARCOMA

Objective: The utility of molecular risk stratification for liposarcoma remains in its infancy. Recently, a tumor maturity score deducted from the expression profiles of 30 genes underlying adipogenesis has been proposed by Bevill et al. in efforts to reflect the molecular landscapes of these tumors. Importantly, this system relies on MDM2 expression, a P53 suppressor and one of the most commonly overexpressed genes in liposarcoma, for score assignment. At present, prognostication of this tumor maturity score is unclear. Therefore, in efforts to expand upon the clinical utility of molecular stratification, including targeted expression profiling, we investigated the application of tumor maturing scoring in relation to clinical outcomes in a small cohort of retroperitoneal liposarcoma (RPLPS) patients.

Methods: The cohort of RPLPS patients whose tissue was utilized in the Bevill et al. study was identified. Clinical characteristics for each patient including age, gender, medical history, medications, BMI at diagnosis and tissue acquisition, additional histological characterizations, and clinical outcomes (recurrence and or patient death) were collected. The maturity score data for each tissue sample was also collected. Expression profiles of MDM2 were obtained utilizing bulk RNA-sequencing compared with subcutaneous tissue of the same patient. Clinical outcomes, including overall survival (OS) and disease-free survival (DFS), were analyzed. Significance testing included unpaired t-tests to assess difference in maturity scores and Log-Rank or Wilcoxon analysis as appropriate.

Results: 20 patients provided source tissue samples comprising of broad histological categories of 16 (64%) well-differentiated liposarcoma (WDLPS) and 9 (36%) dedifferentiated liposarcoma (DDLPS) that were included for analysis. Neoadjuvant therapies included radiation (n=20, 80%), chemotherapy (n=6, 24%), and intraoperative radiation therapy (n=5, 20%). Mean tumor maturity score was -0.11 for all tumors, and upon stratification, WDLPS demonstrated higher mean tumor maturity scores compared to DDLPS (0.31 [range -0.43 to 0.83] vs. -0.56 [range -0.87 to 0.76], respectively; p = 0.0003). MDM2 expression was associated with lower maturity scores regardless of underlying histology, consistent with prior analysis. Histology was associated with OS (p=0.047), yet suggestive but not significant for DFS (p=0.079). Tumor maturity score was not associated with OS or DFS in WDLPS (OS p=0.670, DFS p=0.537) or DDLPS (OS p=0.115, DFS p=0.473).

Further histologic subset analysis categorized the samples into 4 distinct groups: DDLPS, WDLPS, Well-differentiated/dedifferentiated liposarcoma (WDDD), and WDLPS that later recurred as DDLPS (WD2DD). Patients who had their initial histology upgraded (WDLPS to WDDD, and WDLPS or WDDD to DDLPS) were associated with decreased OS (p=0.039) and suggestive of decreased DFS (p=0.062). Survival analyses of these groups showed significant differences in DFS (p=0.0047, log-rank) and OS (p=0.0025, log-rank), with the WD2DD subgroup having the lowest survival of all groups.

Conclusion: Tumor maturity scores correlate with underlying histology; however, are not predictive of clinical outcomes within our cohort. Within WDLPS specifically, tumor maturity score was highly variable and low scores were not predictive of worse OS or DFS. Histology correlates with clinical outcomes in RPLPS, and those patients whose tumors recurred as a worse histology had inferior outcomes compared to having a stable poor histology on subsequent recurrences. Future analysis of the molecular drivers of liposarcoma would benefit from additional correlation with clinical outcomes to further characterize the prognostic implications of targeted expression profiling.