Molecular predictors of response to selinexor in advanced unresectable de-differentiated liposarcoma (DDLS).

Abstract

11509 Background: Patients (pts) with recurrent inoperative DDLS have a poor prognosis and limited treatment options. Selinexor is an oral, selective inhibitor of nuclear export (SINE) compound approved for previously treated pts with myeloma and diffuse large B-cell lymphoma. SEAL was a Phase 2-3 randomized, double-blind, study of selinexor versus placebo in pts with progressive DDLS and 2-5 prior systemic therapies. SEAL showed significantly prolonged progression-free survival (PFS, HR = 0.70, p = 0.0228) with well managed toxicity. A biomarker predictive of clinical activity could be used to optimize selection of pts with DDLS for selinexor. Methods: Pts were randomized 2:1 for Phase 3: 188 received twice weekly selinexor (60mg) and 97 received placebo. Three exploratory biomarker analyses (RNA sequencing of biopsies) from selinexor-treated pts were performed: discovery set of sensitive (n = 8) or resistant (n = 9) tumors; a validation set of pts with favorable (n = 19) or poor (n = 14) tumor control based on PFS, and paired lesions from a pt who harbored both a responsive and resistant lesion. Tumor biopsies from 24 pts on placebo with short ( < 5 months, n = 18) and long ( > 6 months, n = 6) PFS were RNA sequenced. Gene expressions were compared using a negative binomial distribution with DeSeq2. Pathway analyses were performed using Gene Set Enrichment Analysis (GSEA) with MSigDB Cancer Gene Neighborhoods. Results: RNA sequencing analysis comparing 17 sensitive and resistant tumors identified 114 differentially expressed genes (adjusted p-values < 0.05). Expression of CALB1, which encodes the calcium-binding protein calbindin, was significantly lower in sensitive tumors (adjusted P [Padj] = 7.5x10-20), and expression of GRM1, which encodes a metabotropic glutamate receptor that activates phospholipase C, was higher in selinexor sensitive tumors (Padj= 0.003). These findings were confirmed in an independent validation set (Padj = 0.01 – 0.02). In the pt with paired sensitive and resistant lesions, CALB1 expression was 52-fold lower in the sensitive tumor. In a comparison of placebo-treated pts, neither CALB1 or GRM1 was differentially expressed between pts with short or long PFS, indicating they are markers of response to selinexor treatment, rather than general markers of disease aggressiveness. Gene set enrichment analyses revealed that selinexor sensitive tumors in the discovery and validation sets showed upregulation of cancer genes related to SNRK and the netrin 1 receptor tumor suppressor DCC. The resistant tumors showed upregulated EIF3S2 translation initiator-related genes. Conclusions: Selinexor sensitive DDLS tumors showed low expression of CALB1 and high GRM1. If validated, pts with DDLS whose tumors match this expression profile are especially likely to benefit from selinexor. Clinical trial information: NCT02606461.