(P 398) PREDICTORS OF EFFICACY IN PATIENTS WITH SOFT TISSUE SARCOMAS TREATED WITH PAZOPANIB

Objective: Pazopanib (PZP) is a multitarget tyrosine kinase inhibitor with activity against several targets including VEGFR, PDGFR, FGFR, c-Kit and LCK. PZP has been approved for the treatment of advanced soft tissue sarcomas (STS) and has been in use for more than a decade. The objective of this study was to identify predictive biomarkers for efficacy of PZP using next- generation sequencing in three histologic subtypes of STS.

Methods:Patients with a diagnosis of angiosarcoma (AS), leiomyosarcoma (LMS) and synovial sarcoma (SS) seen at MD Anderson Cancer Center and treated with pazopanib between 11/1/2010 and 1/31/2018 were identified using an electronic database system. Inclusion criteria for the study were- histologically confirmed AS, LMS or SS; at least 6 weeks of pazopanib treatment; baseline and follow-up imaging at least 6 weeks apart; and availability of archival FFPE or frozen tumor tissue for molecular studies.
Clinical efficacy of pazopanib was assessed using two methods- objective radiological response to pazopanib and progression- free survival (PFS). Patients with best response of stable disease (SD), partial response (PR) and complete response (CR) were categorized as responders (R) while those with progressive disease (PD) were categorized as non- responders (NR). Patients were also categorized into 2 groups based on PFS- those with a PFS duration < 12 weeks and those with a PFS duration ≥ 12 weeks.
Whole exome sequencing (WES) and RNA sequencing (RNAseq) were performed on all samples that fulfilled quality metrics. Expression data from untreated tumor samples were used for analysis. Expression differences between R and NR and PFS <12W and PFS ≥12W patient groups were analyzed for all the genes using the deSeq2 package in R. Differential expression patterns within each histology group were also compared between the patient groups. Volcano plots were used to visualize the results.


Results: We identified 63 patients who fulfilled the inclusion criteria. Of these, 52 had untreated tumor samples that were analyzed (Table 1). In the comparison of R vs NR and PFS <12W and PFS ≥12W, the common significantly differentially expressed genes were CSAG1, MYH7, MAGEA3, MYBPC1, LPAR3, MYL2, MAGEA12, CACNA1S, GAP43, EPSTI1, CLCA2, TNNT3, TNNC1, DKK1, TNNC2, MYL1, MYOM2, MLIP, NBPF22P and TMC5. Several genes from the MAGEA family of testis antigens (MAGEC1, MAGEA3, MAGEA12) are associated with a negative response to pazopanib treatment, and several muscle function-related genes (MYBPC1, MYL1 and MYL2) are associated with a positive response (Figure 1). In patients with AS, expression of MAGEC1 and CLEC4M are significantly associated with a negative response to pazopanib, while MYC and ZFP57 show significantly higher expression in patients who responded to pazopanib (Figure 1). Among SS patients, mRNA expression of AQP5, PI15, SFN shows significant associations with no response to Pazopanib, and PPP1R3CA and several muscle function-related genes are associated with a positive response to pazopanib (Figure 1). In patients with LMS, there are more significantly expressed genes in R group compared to NR group. Among these genes, KDM5D was the most significantly upregulated gene in the R group while several male fertility-related genes (UTY, ZFY, USP9Y, DDX3Y) were found to be significantly highly expressed in this group. In the LMS group, it was also noted that all the male patients positively responded to PZP, while the female patients with non- uterine LMS responded better to PZP compared to those with uterine LMS.

Conclusion: The significantly associated genes with response to pazopanib identified in our study were not known to be directly associated with pazopanib target pathways. Further studies of the genes identified, particularly in the LMS subgroup, as well as genes related to PZP targets, are needed in order to identify robust predictive biomarkers of PZP efficacy in STS.