Sereduk, Chris

Abstract INTRODUCTION Dendritic cells (DCs) are potent antigen presenting cells that can be exploited to initiate an adaptive anti-tumoral immune response. DC vaccine clinical trials for primary glioblastoma (GBM) have reported prolonged progression-free survival without any impact on overall survival (OS). We report a radiomics approach that identifies a subpopulation of patients with prolonged OS in clinical trial MC1272. METHODS Twenty adults with primary GBM undergoing standard-of-care therapy were enrolled in MC1272. Autologous DCs were pulsed with allogenic GBM cell lines to generate vaccines that were administered for up to twelve cycles. Standard brain imaging was obtained at the initiation of treatment and two months afterwards. An independent cohort of image-localized biopsies underwent RNA sequencing followed by cellular deconvolution to estimate T cell abundance. A machine learning model was trained to predict intratumoral T cell abundance from imaging features, and the model was applied to MC1272 patient imaging. RESULTS Voxelwise predictions of T cell abundance were generated for each patient’s pre- and post-treatment images. The difference in total intratumoral T cell abundance between imaging timepoints classified patients into increasing or decreasing T cell groups. Patients whose T cells increased had longer OS (median, 21 months) than those whose T cells decreased (median, 10 months; p=0.0035). The significance remained in a Cox proportional hazards analysis that adjusted for patient age and sex (p=0.011). CONCLUSIONS A spatially-resolved radiomics model detected that an intratumoral T cell influx after DC vaccine therapy was associated with prolonged OS. The “post-treatment” imaging in this study was obtained a mere two months after DC vaccine initiation, suggesting that our radiomics model can provide an early indication of treatment responsiveness and prognosis in these patients.

Abstract Intra-tumor genetic heterogeneity is an important cause of treatment failure of GBM. Using MRI and image-localized biopsies, it is possible to train machine learning (ML) models to predict regional genetic status. However, biopsy samples are limited, making it difficult to train a robust ML model. We proposed a data-inclusive model called Weakly Supervised Ordinal Support Vector Machines (WSO-SVM) which leverages the vast amount of MRI data outside the sparsely sampled biopsy regions to augment the biopsy samples to improve ML accuracy. Our study included a unique dataset of 104 image-localized biopsies with spatially matched multiparametric MRI from 30 untreated Glioblastoma (GBM) patients. Each biopsy sample went through genetic sequencing analysis and our study focused on two GBM hallmark genes, EGFR and PDGFRA. For each gene, a biopsy sample was labeled as “altered” if the copy number of this gene was amplified or a mutation was found, and “non-altered” otherwise. From the localized region of six MRI contrasts from T1gd, T2w, diffusion and perfusion imaging, over 300 texture features were extracted. To account for biopsy sample location uncertainty, six neighboring regions of the biopsy sample including four neighbors two pixels away from the biopsy location and two neighbors on adjacent slices were also included in model training. WSO-SVM achieved 0.83 accuracy, 0.77 sensitivity, and 0.86 specificity for classifying EGFR; 0.77 accuracy, 0.74 sensitivity, and 0.79 specificity for classifying PDGFRA, based on 10-fold cross validation. Furthermore, using the trained models, we generated regional EGFR and PDGFRA alteration maps for each patient within the enhancing and non-enhancing tumoral areas. On average we found a greater proportion of enhancing tumor with co-alteration than non-enhancing tumor, while this trend was reversed when considering only one altered gene. The ratio of EGFR vs PDGFRA alterations was higher in non-enhancing tumor than enhancing tumor.

Abstract Glioblastoma (GBM), the most common primary brain tumor in adults, remains uniformly fatal due to the lack of effective targeted therapies for this aggressive malignancy. Genomic amplification of epidermal growth factor receptor (EGFR) occurs in 40-60% of primary GBM. Half of all EGFR-amplified cases of GBM also harbor EGFRvIII, a constitutively active, truncated variant of EGFR. The incidence of EGFR alterations in GBM makes inhibition of EGFR/EGFRvIII an attractive therapeutic approach. Depatuxizumab mafodotin (ABT-414) is an antibody-drug conjugate comprised of ABT-806, a mAB against EGFR, and a cytotoxic payload (monomethyl auristatin F). ABT-414 therapy initially showed promising results as an EGFRvIII therapy in vivo and in vitro; however, ABT414 failed to provide a survival benefit in phase I/II clinical trials, potentially due to therapeutic resistance. In this study, we seek to investigate the mechanisms of resistance to ABT-414 by performing whole exome, transcriptome and single cell RNA seq on ABT-414 resistant GBM PDX tumors. Our data showed an enrichment of mutations unique to the ABT-414 resistant tumors, including a point mutation (S466I) in TEK/TIE2 transmembrane angiopoietin receptor. In vitro expression of TEK S466I in GBM cells showed an increase in activation of ERK and STAT3 and increased TEK/TIE2 immunoprecipitation with EGFR compared to the wild-type TEK receptor. Furthermore, gene ontology analysis reveals that ABT-414-treated flank tumors exhibit increased activation of extracellular matrix organization and CNS developmental processes compared to flank tumors in the control treatment groups. ABT-414-treated tumors also demonstrate increased expression of inhibitor of differentiation (ID)1 and ID3, associated with stem-like phenotype in glioblastoma. Taken together, our data indicate that resistance to ABT-414 is mediated by both de novo mutations not detected in the parent tumor and adaptive dysregulation of pathways which may lead to dedifferentiation and therapeutic resistance.

Abstract Triptolide, which was initially extracted from a Chinese medicinal herb, demonstrates potent antitumor, anti-inflammatory, and immunosuppressive properties, yet it is clinically limited due to its poor solubility, bioavailability, and toxicity. Minnelide, a water-soluble analog of triptolide, has potent antiproliferative activity. It is currently in Phase I/II clinical trials for Acute Myeloid Leukemia (AML) and 7 solid tumor indications, including advanced gastrointestinal tumors, pancreatic cancer, and colorectal cancer. To overcome the potential drug resistance and to improve the drug potency as well as to minimize the drug toxicity, we conducted a high throughput screening (HTS) for drug synergy, combining Minnelide with compounds from the LOPAC and/or Prestwick libraries in 3 different cancer cell lines: BxPC3 (pancreatic), HT29 (colon) and SNU5 (gastric). LOPAC and Prestwick libraries consist of drugs approved by FDA and/or drugs that have reached clinical trial stages, representing 2,274 unique compounds in total. After quality control evaluation and data normalization, hits were selected if greater than 20% inhibition was achieved when combined compared with Minnelide alone. A total of 43, 54, and 47 hits were identified for each cell line, representing an average hit rate of 2.1%. We further prioritized the hits by selecting those that were common in two or more cell lines, which resulted in 9 compounds from Prestwick library and 20 compounds from LOPAC library (Harmane is in common between the two libraries). Only 17 compounds were available for purchase for hits confirmation, which was conducted in a similar manner as in primary HTS, except that drugs were tested in a dose-dependent manner instead of a single concentration done in HTS, a duplicated biological run was performed, and a total of 9 cell lines were screened, with 3 cell lines for each tumor type (pancreatic, colon and gastric). IC50 and Combination Index (CI) values for each drug combination were calculated for each cell line. A fold change of 2 or more for the IC50 values and CI value < 1.0 were used for hit selection. In general, colon lines had the most synergistic combinations and gastric had the least. A total of 15 drug candidates were confirmed to have synergistic effect, representing a confirmation rate of 88%, among which 12 drugs demonstrated synergy in >50% of the cell lines tested, with Harmane and Harmine demonstrating synergy in all 9 cell lines tested. In addition, CGS-15943, Amiloride, Naphthalimide, and Idazoxan indicated good synergy with Minnelide in at least 6 out of 9 cell lines. Interestingly, Phenamil, an irreversible inhibitor of amiloride-sensitive Na+ channels, showed synergy in all cell lines tested for pancreatic and colon cancers, but not in gastric cancer cell lines. Overall, our study shed light on potential drug combinations with Minnlide for clinical use. Citation Format: Nanyun Tang, Chris Sereduk, Ashok Saluja, Mohana Velagapudi, Haiyong Han, Daniel D Von Hoff, Hongwei Holly Yin. Drug combination screen to identify novel synergy with Minnelide [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr A142. doi:10.1158/1535-7163.TARG-19-A142

Abstract Immunomodulatory drugs (IMiDs) are widely used in the treatment of patients with Multiple Myeloma (MM) however, only 30% of relapsed MM patients respond to single agent therapy and most patients eventually develop drug resistance. The molecular target of IMiDs in MM is cereblon but other parallel pathways or downstream events which enhance or preclude drug responsiveness are unknown. We therefore conducted a genome scale small interfering RNA (siRNA) lethality study in MM in the presence of increasing concentrations of lenalidomide. Primary screening was performed in a single-siRNA-per-well format with the human druggable genome siRNA set V4 comprising four siRNAs targeting each of 6,992 genes (total 27968 siRNAs). Lenalidomide was added 24 hours post transfection and cell viability was measured by ATP-dependent luminescence at 144 hours after transfection. Primary screen data was rigorously evaluated for multiple quality control metrics and found to exceed all expected performance parameters with >98% global transfection efficiency, <0.25 CV values, and minimal plate-to-plate and set-to-set variations observed. Hit selection was performed by analysis of IC50 value shift in the presence of each testing siRNA compared with three different control siRNA oligos. 160 candidate genes that enhance lenalidomide sensitivity upon silencing (sensitizers) were selected and re-screened with four siRNA oligos targeting each gene. 50 genes were identified as reproducible lenalidomide sensitizers including three Peroxisome (PEX) family proteins (PEX1, PEX10 and PEX7) and seven RAB family proteins (RAB17, RAB1A, RAB26, RAB30, RAB36, RAB4A and RAB8A). Four kinase genes were also identified in sensitizer hits and two of these, I-Kappa-B Kinase-Alpha (IKK1 or CHUK) and ribosomal protein S6 kinase (RPS6KA3 or RSK2), encode proteins that associate with significance together with a phosphorylation dependent transcription factor (CREB1) in Toll signaling pathways (p-value 0.0068). RSK2 is a serine/threonine-protein kinase that acts downstream of oncogenic FGFR3 mediated signaling and is phosphorylated by ERK (MAPK1/ERK2 and MAPK3/ERK1 signaling) during hematopoietic transformation. Phosphorylated RSK2 was previously reported to be frequently expressed in myeloma cell lines and primary myeloma cells. Using lentiviral shRNA expression, we demonstrated that knockdown of RSK2 in three genetically variable MM cell lines induced cyctocytoxiticy and consistently sensitized to lenalidomide. Two selective small molecular inhibitors of RSK2 (SL 0101-1 and BI-D1870) were then demonstrated to synergize with lenalidomide to induce myeloma cell cytotoxicixity. To further understand the mechanism underlying sensitization, immunoblotting analysis was performed to look at downstream changes after either RSK2 knockdown or RSK2 inhibition by BI-D1870. We found that both RSK2 knockdown and BI-D1870 treatment, mimicking lenalidomide treatment or cereblon inhibition, induced downregulation of both IRF4 and MYC in MM cells. The combination of lenalidomide and BI-D1870 not only produced a substantial synergistic effect inducing MM cytotoxicity, but also demonstrated a significant enhancement of downregulation of IRF4 and MYC. Forced overexpression of RSK2 attenuated the synergistic effects of lenalidomide and BI-D1870. In summary, our high throughput screen identified multiple gene targets that associate with increasing sensitivity to IMiDs in MM cells, of which, RSK2 was further validated by both shRNA silencing and specific inhibitors as an effective target to cooperate with IMiDs to induce myeloma cytotoxicity. Clinical studies of RSK2 inhibition in concert with IMiD (cereblon inhibitor) therapy would be appropriate. Disclosures: Stewart: Onyx: Consultancy, Research Funding; Millennium: Honoraria, Research Funding; Celgene: Honoraria; BMS: Honoraria.

Abstract Abstract 602 The molecular target(s) which co-operate with proteasome inhibition in inducing drug sensitivity or resistance in Multiple Myeloma (MM) remain unknown. We therefore conducted a genome scale small interfering RNA (siRNA) lethality study in KMS11 MM cells in the presence or absence of bortezomib without regard to pre-conceived mechanistic notions. Primary screening was performed in a single-siRNA-per-well format with the human druggable genome siRNA set V4 comprising 13,984 siRNA targeting 6,992 genes and comprising two RNAi per gene. siRNA were transfected at low concentration (13nM) to minimize off-target effects using conditions that resulted in transfection of >95% cells and <5% background cytotoxicity. Bortezomib was added (at the IC10, 25, 70 and 90) 24 hours post transfection. After 96 hours (72 hours after bortezomib), viability was measured by ATP-dependent luminescence. In primary screening 320 candidate bortezomib sensitizing genes were identified and rescreened with four siRNA oligos each gene. 57 of the top sensitizer hits were selected for which at least two distinct siRNA decreased the EC50 by 2 standard deviations from cells treated with control siRNAs. By cross referencing gene expression profile (GEP) data of KMS11, we furthered pared the list to 37 plausibly expressed targets (0.5% of those genes originally screened) as bortezomib sensitizers. After silencing, 50% of these 37 genes also sensitized the lung cancer cell line A549 to bortezomib. After further parsing of genes which also modulated the sensitivity of MM to Melphalan (non specific chemosensitizers), 34 genes remained: The strongest sensitizers to bortezomib were the proteasome subunits PSMA5, PSMB2, PSMB3, PSMB7 but included less obvious targets such as BAZ1B, CDK5, CDC42SE2, MDM4, NME7, TFE3, TNFAIP3, TNK1, TOP1, VAMP2 and YY1 were also identified. Of these, the most potent synergetic effects were observed with siRNAs against the proteasome subunits and against cyclin dependent kinase 5 (CDK5), which caused the greatest shift in EC50. CDK5 is of particular interest as a therapeutic target as it is expressed at high levels in MM and neural tissues but has low expression in other organs. Using viral shRNA expression, silencing CDK5 consistently increased the sensitivity of genetically variable MM cell lines (n=5) to all of the proteasome inhibitors tested: bortezomib, carfilzomib and PR047, and the effect could be at least partially rescued by overexpression of an RNAi resistant CDK5. To explore therapeutic relevance the small molecule CDK5 inhibitor, Roscovitin, was shown to be synergistic or additive with bortezomib in both MM cell lines and primary patient samples. Gene expression profiling was then performed to seek an explanation for the CDK5 sensitization effect and regulation of a proteasome subunit PSMB5 by CDK5 was identified as a probable route to sensitization. In summary inhibition of the existing proteasome either directly by suppression of proteasome subunits, or indirectly by suppression of modulators such as CDK5 appears to confer the greatest sensitization effect suggesting that combinations of bortezomib with other unique proteasome inhibitor drugs or combinations with inhibitors of CDK5 is a logical avenue for clinical exploration. Disclosures: No relevant conflicts of interest to declare.

Abstract Glioblastoma multiforme (GBM) remains the most common and deadly primary brain tumor with minimal advances in therapeutic development for pharmacological treatment. Tumor resection followed by temozolomide (TMZ) and radiotherapy adds months to survival; however, tumor recurrence usually occurs. The fibroblast growth factor-inducible 14 (Fn14) receptor has been shown to be up-regulated in the GBM cells left behind after surgery and contributes to an invasive and pro-survival phenotype, providing a desirable molecule to target after removal of the primary tumor. Previously, L524-0366 was reported by our lab to be the first antagonist of Fn14 signaling; however, minimal efficacy/potency has limited this drug from progressing from a useful research tool to preclinical studies. Structure activity relationship (SAR) studies using the parent L524-0366 compound gave rise to K78, which was found to bind to Fn14 with enhanced affinity, while showing no binding to the structurally-related TNFα receptor. Furthermore, K78 was able to abate the downstream signaling effects initiated by the Fn14 ligand, tumor necrosis factor-like weak inducer of apoptosis (TWEAK), as well as sensitize GBM cells to the cytotoxic effects of TMZ. TWEAK-induced cellular migration was also mitigated with co-treatment of K78 in GBM cells, compared to a vehicle control. It is anticipated that further development of Fn14 antagonists will be of great importance for the advancement of GBM patient treatment and warrant further investigation as primary and combination therapy, especially for the prevention of recurrent disease. Citation Format: Bryan Harder, Chris Sereduk, Harshil Dhruv, Nghia Millard, Anthony Kim, Jeffrey Winkles, Graeme Woodworth, Holly Yin, Nhan L. Tran. Development of an antagonist to the TWEAK-Fn14 ligand-receptor interaction with enhanced efficacy/potency in glioblastoma cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4910.

Abstract Purpose: Small cell carcinoma of the ovary, hypercalcemic type (SCCOHT) is a rare, aggressive ovarian cancer in young women that is universally driven by loss of the SWI/SNF ATPase subunits SMARCA4 and SMARCA2. A great need exists for effective targeted therapies for SCCOHT. Experimental Design: To identify underlying therapeutic vulnerabilities in SCCOHT, we conducted high-throughput siRNA and drug screens. Complementary proteomics approaches profiled kinases inhibited by ponatinib. Ponatinib was tested for efficacy in two patient-derived xenograft (PDX) models and one cell-line xenograft model of SCCOHT. Results: The receptor tyrosine kinase (RTK) family was enriched in siRNA screen hits, with FGFRs and PDGFRs being overlapping hits between drug and siRNA screens. Of multiple potent drug classes in SCCOHT cell lines, RTK inhibitors were only one of two classes with selectivity in SCCOHT relative to three SWI/SNF wild-type ovarian cancer cell lines. We further identified ponatinib as the most effective clinically approved RTK inhibitor. Reexpression of SMARCA4 was shown to confer a 1.7-fold increase in resistance to ponatinib. Subsequent proteomic assessment of ponatinib target modulation in SCCOHT cell models confirmed inhibition of nine known ponatinib target kinases alongside 77 noncanonical ponatinib targets in SCCOHT. Finally, ponatinib delayed tumor doubling time 4-fold in SCCOHT-1 xenografts while reducing final tumor volumes in SCCOHT PDX models by 58.6% and 42.5%. Conclusions: Ponatinib is an effective agent for SMARCA4-mutant SCCOHT in both in vitro and in vivo preclinical models through its inhibition of multiple kinases. Clinical investigation of this FDA-approved oncology drug in SCCOHT is warranted. Clin Cancer Res; 24(8); 1932–43. ©2018 AACR.