BIOM-26. MOLECULAR PREDICTORS OF RESPONSE TO SELINEXOR IN RECURRENT GLIOBLASTOMA (GBM). (9th November 2020)

• Record Type: Journal Article
• Authors: Walker, Christopher; Shen, Yao; Mau-Soerensen, Morten; Wen, Patrick; van den Bent, Martin; Plotkin, Scott; Walenkamp, Annemiek; Green, Adam; Califano, Andrea; Chang, Hua; Tamir, Sharon; Henegar, Leah; Shacham, Sharon; Alvarez, Mariano; Landesman, Yosef; Lassman, Andrew
• Is Part Of: Neuro-oncology. Volume 22(2020)Supplement 2
• Page Start: ii7
• Page End: ii7
• Publication Date: 2020-11-09
• Subjects: Brain Neoplasms -- Periodicals; Brain -- Tumors -- Periodicals; Brain -- Cancer -- Periodicals; Nervous system -- Cancer -- Periodicals; 616.99481
• DOI: 10.1093/neuonc/noaa215.026
• Languages: English
• ISSNs: 1522-8517

Abstract

Abstract: The nuclear export protein exportin 1 (XPO1) is overexpressed in many cancers, including GBM. Selinexor is an inhibitor of XPO1 that crosses the blood-brain-barrier and targets cancer cells by sequestering tumor suppressor proteins and oncoprotein mRNAs in the nucleus, inducing cancer cell apoptosis. We previously reported encouraging results from a phase II trial of selinexor for molecularly unselected patients with recurrent GBM (ASCO 2019). Pre-treatment tumors from 57 patients were exome and RNA sequenced to explore molecular correlates of response, in a hypothesis generating, post-hoc, exploratory analysis. We compared overall survival (OS) and progression-free survival (PFS) between mutated and wild-type patients. Two mutated genes were associated with longer survival in selinexor treated patients: DOCK8 (n=7; PFS, P=0.013, hazard ratio [HR]=3.75 [1.32–10.62]; overall survival, P=0.009, HR=15.39 [2.00–118.34]) and PDX1 (n=5, PFS, P=0.014, HR=4.468 [1.361–14.670]). Other commonly mutated genes in glioma, including IDH1 (n=9) were observed but not associated with survival. RNAseq data were used to infer differential protein activities, which were input into a machine learning model that compared patients with selinexor sensitive disease (best response of partial or complete response, n=7) and resistant disease (best response of progressive disease, n=23). We found the inferred activities of three proteins emerged as the most associated with response and could be combined in a model to accurately predict benefit from selinexor treatment (area under the ROC curve from leave one out cross validation = 0.89 permutation test P< 0.04). The three proteins were ZC3H12A (also called MCPIP-1), a negative regulator of inflammation; RAB43, a member of the RAS family that binds GTP and regulates vesicle trafficking, and SOCS3, a suppressor of cytokine signaling that can antagonize JAK/STAT signaling. Together these data identified mutations and proteins activities for identifying patients most likely to benefit from selinexor treatment. Further studies are required for validation. ClinicalTrials.gov:NCT01986348.