Wei, Ellen

Abstract Introduction: Immune checkpoint receptor (ICR) blockade has emerged as an effective anti-tumour modality, but only in a subset of cancer patients. Moreover, in Multiple myeloma (MM), single-agent activity has not been observed, highlighting the need to better understand the mechanism of action of this class of drugs. We recently showed that combinatorial ICR blockade using αLAG3 and αPD-1 delays disease progression and improves survival in the transplantable Vκ*MYC model of MM (Croucher et al. ASH 2018). However, despite this being a controlled study with genetically-homogeneous tumours, anti-tumour immune responses were heterogeneous, with only a subset of mice demonstrating a delay in tumour progression (17/29 mice, response rate = 58.6%). Thus, using this model, we set out to define mechanisms underlying variability in response to ICR blockade. Methods: We established a cohort of mice by engrafting 5-week-old C57BL/6 mice with Vκ12598 cells via tail vein injection. Treatment with αLAG3/αPD-1 or Ig-control was initiated 1-week post-engraftment and bone marrow (BM) samples were collected 3 weeks after the start of treatment. Following FACS-enrichment of T cells and plasma cells (PCs), single cell suspensions were subjected to matched single-cell gene expression (5' scRNA-seq) and T cell receptor (TCR)/B cell receptor (BCR) profiling (10x Genomics). Results: Samples were selected for profiling based on response to treatment, with responders (n=4) defined by significantly lower disease burden compared to non-responders (n=3) and control-treated mice (n=5), as measured by serum M-protein and %PCs in BM/spleen at sacrifice. Unsupervised clustering of scRNA-seq data from PCs (n=3,318 cells) identified no gene expression or BCR repertoire differences between control and treated, or between responder and non-responder samples, supporting that variability in response was not related to malignant Vκ12598 cells themselves. Across all samples, a statistically significant difference was not detected between the total number of unique TCR sequences (clonotypes) comparing control-treated (351-2369), non-responders (1185-2327) and responders (1378-1698), with no overlapping TCR sequences between top clonotypes. Evaluation of TCR repertoire diversity revealed that αLAG3/αPD-1 treatment induces clonal T cell expansion in control versus treated mice, but this was not significantly different between responders and non-responders. Analysis of paired scRNA-seq data (n=21,520 cells) revealed that expanded T cells from αLAG3/αPD-1-treated mice occupy a different cell state in responder vs. non-responder mice. We speculate that underlying differences in the TCR repertoire may dictate the downstream phenotype of expanded, anti-tumour T cells in mice treated with combinatorial αLAG3/αPD-1. Tumour control following treatment was associated with clonal expansion of T cells expressing genes related to cytoxicity and activation (Ccl5, Ifng, Fasl, Gzmb), whereas tumour progression was associated with clonal expansion of proliferative T cells (Cdkn3, Birc5, Ccna2, Aurka, Mki67). Although T cell proliferation is typically a phenotype ascribed to effector T cells, recent studies have similarly observed this proliferative cell state in dysfunctional T cells within melanoma tumours. Moreover, emerging evidence supports suppression of T cell proliferation by CDK4/6 inhibitors as a means to augment anti-tumour activity of ICR-based therapy. Thus, studies exploring whether reversal of the observed proliferative T cell state can restore response to αLAG3/αPD-1 treatment in non-responding Vκ12598 mice are ongoing and will be reported. Conclusions: ICR inhibitors demonstrate significant activity in some cancers, however many patients fail to respond and a similarly promising level of efficacy has not been achieved in MM. Studies aimed at unraveling the mechanisms of response and resistance to ICR inhibitors are therefore needed to improve the utility of this class of drugs for all patients. Our approach of using paired single-cell gene expression and TCR repertoire profiling has enabled identification of molecular cell states specifically in expanded T cells of responder vs. non-responder mice. In turn, our work nominates novel mechanisms that may be used as potential biomarkers for anti-tumour immune responses as well as potential targets to augment responses to ICR blockade therapy. Disclosures Chesi: Abcuro: Patents & Royalties: Genetically engineered mouse model of myeloma; Novartis: Consultancy, Patents & Royalties: human CRBN transgenic mouse; Pfizer: Consultancy; Pi Therapeutics: Patents & Royalties: Genetically engineered mouse model of myeloma; Palleon Pharmaceuticals: Patents & Royalties: Genetically engineered mouse model of myeloma. Bergsagel: GSK: Consultancy, Honoraria; Genetech: Consultancy, Honoraria; Janssen: Consultancy, Honoraria; Oncopeptides: Consultancy, Honoraria; Novartis: Consultancy, Honoraria, Patents & Royalties: human CRBN mouse; Pfizer: Consultancy, Honoraria; Celgene: Consultancy, Honoraria. Sebag: Janssen: Research Funding; Bristol Myers-Squibb: Consultancy, Honoraria; Takeda: Consultancy, Honoraria; Novartis: Consultancy, Honoraria; Amgen: Consultancy, Honoraria; Sanofi: Consultancy, Honoraria; Karyopharm Therapeutics: Consultancy, Honoraria. Trudel: BMS/Celgene: Consultancy, Honoraria, Research Funding; Amgen: Honoraria, Research Funding; Janssen: Honoraria, Research Funding; GlaxoSmithKline: Consultancy, Honoraria, Research Funding; Roche: Consultancy; Sanofi: Honoraria; Pfizer: Honoraria, Research Funding; Genentech: Research Funding.

Abstract The identification of the t(4;14) translocation in Multiple Myeloma (MM) provided the first indication that fibroblast growth factor 3 (FGFR3) could act as an oncogene. Several studies have assessed the potential of FGFR3 as a drug target and have shown that inhibition of FGFRs using tyrosine kinase inhibitors induces cell death of FGFR3 expressing MM cells. To clarify the signaling pathways implicated in FGFR3 dependent myeloma, we assessed gene expression changes associated with treatment of MM cell lines with three known small molecule FGFR inhibitors, PD173074, SU5402 and a third novel compound using Human U133_Plus2 arrays. In addition to describing the unique pharmacogenomic signals of each drug in myeloma, 845 differentially expressed genes, common to all three FGFR3 inhibitors, were identified. Cell cycle, DNA replication and ATP binding proteins were the most common (47%) functional classes of genes effected. Further validation and refinement of this gene list was conducted using FGFR3 siRNA inhibition and FGF ligand induction. 120 genes were altered between FGFR3 RNAi and scrambled control and 892 were induced by FGF ligand. Ten genes were commonly identified as significantly regulated by all 4 FGFR3 inhibitors and induced in the opposing direction by ligand. Of these macrophage inflammatory protein (MIP)-1alpha, and dual specific phosphatase 6 (DUSP6) were positively regulative while ANXA9, CR2, AL531683, ZNF589, AW274468, FRMD3, LTB and WDR42A were negatively regulated by FGFR3 signaling. Validating our findings, a feedback loop between FGFR/ERK/MAP kinase activation and DUSP6 has previously been described in the literature. MIP-1alpha however was the most significantly altered (12 fold) on array, and MIP-1alpha regulation in response to FGFR3 pathway stimulation was confirmed by flow cytometry and Western blot. Mining of publicly available array datasets on 174 MM patients significantly associated MIP1-alpha with FGFR3 expression (p=0.01). Of note down-regulation of MIP-1alpha was not observed following FGFR3 inhibition in MM cells with RAS mutations. Based on these, and previous findings, we hypothesized that MIP-1alpha was regulated by the FGFR3/ERK/MAP kinase pathway. Indeed, inhibition of ERK in FGFR inhibitor resistant cells with RAS mutations also led to down regulation of MIP-1alpha. Because of the recognized role of MIP1-alpha (CCL3) in survival and proliferation of MM cells and in MM bone disease, our observations raise the possibility that pharmacological inhibition of MIP-1 alpha may hold therapeutic promise in t(4;14) MM and may serve as a biomarker for successful FGFR3 or RAS signaling inhibition.

Abstract Abstract 1793 Thalidomide (THAL) and IMID® immunomodulary drugs lenalidomide (LEN) and (POM) have proven beneficial in the treatment of a variety of hematological malignancies. Pre-clinical studies demonstrate multiple direct and indirect anti-tumor activities including anti-angiogeneic, proapoptotic, anti-proliferative and immunomodulatory effects. Recent studies have identified cerebron (CRBN) as a potential direct physical target of THAL and IMiD compounds and CRBN expression is reportedly required for IMiD compound activity. However, the precise link between CRBN and IMiD compound mechanism of action (MOA) have not been clearly defined. We applied Drosophila as a drug discovery platform to assess the MOA of THAL and IMiD compounds in vivo. THAL or POM fed Drosophila demonstrate morphological phenotypes that replicate wingless (wg) mutants indicating that drugs are inhibitors of Wg/Wnt signaling. In this model system, THAL and IMiD compounds disrupt membrane localization of GSK-3 indicating that the bioactivity of IMiD compounds is achieved through spatial regulation and potentiation of Sgg/GSK-3 in Drosophila. Using epistasis analysis we show that Drosophila expressing genetic mutants lacking GSK-3 activity and myeloma cells in which GSK-3a and GSK-3b have been knocked down by siRNA fail to respond to POM. In both Drosophila and myeloma cells therefore it appears that GSK-3 activity is required for biological responses. To test the clinical validity of GSK-3 as a biomarker, we obtained patient tumor samples from a Phase II clinical trial of single agent LEN for previously untreated CLL (Chen CI et al., J Clin Oncol, 2011; 29:1175). Twenty five patients were enrolled onto this study and received LEN at a starting dose of 2.5 mg days 1–21 of a 28 day cycle with monthly escalation to a target dose of 10 mg. The primary clinical endpoint for the trial was objective response to lenalidomide (complete response (CR) and partial response (PR)) evaluated as defined in the revised 1996 NCI Working Group guidelines. Peripheral blood samples for correlative studies were collected on days 1 (pre-dosing) and 8 of cycles 1 and 2. CRBN expression was evaluated by gene expression profiling and Western blot and found to be uniformly expressed in all 19 evaluable day 1 patient samples regardless of LEN response. Thus CRBN expression does not appear to be a useful predictive biomarker of response in this population of previously untreated patients. However, GSK-3 localization was correlated with response. Paired analysis of CD19 selected CLL cells comparing day 1 vs day 8 revealed focal membrane localization of GSK-3 on day 1 and subcellular redistribution on day 8 in 11 out of 12 evaluable responders (PR or better). By contrast, in the CLL cells from all 6 evaluable non-responders, GSK-3 expression appeared diffusely distributed before and after treatment. In preliminary studies using confocal immunofluorescence microscopy we determined that CRBN and GSK-3 co-localize in day 1 CLL samples of responders but not in those of non-responders. In summary, our results indicate that THAL and the IMiD compounds target GSK-3 function by spatial regulation and identify GSK-3 localization as a potential clinical biomarker of IMiD response. Disclosures: Trudel: Celgene: Honoraria; GlaxoSmithKline: Research Funding; Janssen: Honoraria. Mercurio:Celgene: Equity Ownership, Research Funding. Lopez-Girona:Celgene Corp: Employment, Equity Ownership. Gaidarova:Celgene Corp: Employment, Equity Ownership. Webb:Celgene: Employment, Equity Ownership. Chen:Johnson & Johnson: Consultancy, Research Funding; Celgene: Consultancy, Research Funding; GlaxoSmithKline: Research Funding; Lundbeck: Consultancy. Stewart:Millenium: Consultancy, Honoraria, Research Funding; Onyx: Consultancy; Celgene: Consultancy. Manoukian:Celgene: Research Funding.

Abstract Abstract 576 Stat3 appears to play a critical role in sustaining and generating memory B-cells and long-lived plasma cells (PCs) the normal counterparts of myeloma cells. In contrast to the transient nature of Stat3 activation in normal cells, many hematological tumors including multiple myeloma (MM) harbor constitutive Stat3 activity. Aberrant STAT signaling is recognized as a master regulator of direct and indirect tumor processes including proliferation, apoptosis, invasion, angiogenesis and cancer inflammation and thus Stat3 represents an important therapeutic target. Using structure-based design, targeting the phosphotyrosine (pY)-SH2 domain interactions that stabilizes active Stat3 dimers, phosphopeptides have been identified that block Stat3 dimerization and DNA binding activity. Peptidic agents however suffer from limited cell permeability and metabolic lability, properties that have restricted their practical application in vivo at therapeutic doses. Here we describe the dependence of myeloma tumors on Stat3 and characterize the anti-tumor activity of BP-2-047, a novel, non-phosphorylated small molecule with reduced peptidic character. We analyzed the activity of Stat3 using a Stat3 gene expression index based on Stat3 target genes in 60 human myeloma cell lines (HMCLs). Several available myeloma cell lines (XG6, XG7, JJN3 and UTMC2) demonstrated elevated STAT3 gene expression signatures (2.3 to 14.8) with confirmed constitutive Stat3 phosphorylation by Western blot analysis. RNA interference revealed that Stat3 is essential for XG6, JJN3, and UTMC2 survival. We similarly analyzed the activity of Stat3 in publicly available gene expression profile data sets derived from newly diagnosed MM patients. Approximately a third of patients demonstrated a functionally significant level of Stat3 signature, comparable to those in the Stat3-dependent cell lines. We next evaluated the anti-myeloma activity of the BP-2-047 in cell-based assays against a panel of HMCLs with high and low STAT3 activity. BP-2-047 delivered a desirable activity profile for targeted therapy. Consistent with its predicted mechanism of action, we observed dose-dependent inhibition of Tyr705Stat3 phosphorylation in XG6, XG7, JJN3 and UTMC2 cells, presumably through the blockade of Stat3 binding to the pTyr motifs of gp130 and the prevention of de novo phosphorylation by JAKs. Using a luciferase reporter assay we confirmed inhibition of Stat3 transcriptional activity in MM cells stably expressing the Stat3-dependent luciferase reporter (pLucTKS3). Inhibition of transcription was associated with decreased expression of short half-life proteins, c-Myc and Mcl-1. BP-2-047 inhibited viability of myeloma tumors with IC50 values of 3.1–7.4 microM, in cell lines with constitutive Stat3 activity. Inhibition of Stat3 phosphorylation resulted in induction of dose-dependent apoptosis as determined by PI/annexin V staining and PARP cleavage. However, cells were resistant to BP-2-047-induced apoptosis when co-cultured with bone marrow stroma cells. Identification of factors that confer survival advantage in the presence of stroma is in progress. Exposure of MM patient derived bone marrow mononuclear cells to < 15 microM BP-2-047 preferentially induced apoptosis of CD138 + MM cells. In contrast, BP-2-047 at concentrations of 30 microM failed to inhibit normal bone marrow-derived CD34 colony formation. In vitro combination studies in cell lines demonstrate synergy with dexamethasone, bortezomib and lenalidomide. Studies assessing in vivo activity of BP-2-047 as a single agent and in combination with bortezomib against Stat3 activated myeloma tumors in xenograft mice are ongoing and will be reported. These data suggest that Stat3 may represent an important therapeutic target in a subset of myeloma patients and supports innovative drug development with a focus on Stat3. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 718 Introduction: In a previously reported phase 2 study of single agent lenalidomide in 25 untreated CLL patients (pts), we reported an overall response rate (ORR) of 56% (14 pts), 40% SD (10 pts) and no CR at a median follow-up of 20.7 months (Chen et al. JCO 2010;29:1175). Although an amended protocol with conservative lenalidomide dosing was used to mitigate tumor lysis and severe myelosuppression observed in the first 2 accrued pts, we continued to observe frequent toxicities of grade 3–4 neutropenia (72%) and tumor flare (TF 88%). We now report long-term efficacy and toxicity from this study at a median follow-up of 47 months (mos). Methods: Patients were eligible if previously untreated and symptomatic (cytopenias, symptomatic adenopathy/organomegaly, constitutional symptoms, lymphocyte doubling count <12 mos). The amended starting dose for lenalidomide was 2.5mg daily on days 1–21 of a 28 day cycle, with slow monthly dose escalations (2.5mg cycle 1, 5mg cycle 2, 10mg cycle 3 and if required for response, further 5mg increments to a maximum of 25mg daily were allowed). Results: Longterm toxicities: Hematologic toxicities were common: grade 3–4 neutropenia (76%), thrombocytopenia (28%), anemia (20%). With longer term use, neutropenia tended to recur (12% of all cycles) and 10 pts required GCSF support (5 routinely during each cycle). Most common non-hematologic toxicities (all grades) were TF (88%), fatigue (76%), rash (60%), muscle cramping (40%), diarrhea (40%). All non-hematologic toxicities were mild (grade 1–2), except for 1 pt each with grade 3 rash and diarrhea. Although TF was most common during cycle 1, repeat flare symptoms upon resuming lenalidomide after the 7 day rest period of each cycle were noted in 16% of all 898 cycles administered, and as late as at cycle 28. Infections were mild (most respiratory, skin) with only 2 grade 3 events (disseminated zoster, S.pneumoniae bacteremia). Other malignancies: 2 pts developed transformed large cell lymphoma 7 and 18 mos after study discontinuation, 1 pt developed squamous cell carcinoma of skin at cycle 51, and 1 pt developed recurrence of remote non-small cell lung cancer at cycle 34. Dose modifications/discontinuation: The median highest dose achieved for all 25 pts was 15 mg (range 2.5–25 mgs); 8 pts were able to escalate to the maximal 25mg dose. Ten pts (40%) required dose reductions for grade 3 cytopenias [neutropenia (2), thrombocytopenia (2), both (2)], febrile neutropenia (2), and diarrhea (2). Of all 25 pts, the median duration on therapy was 31.1 mos (range 28 days – 60.6 mos). Twelve pts (52%) currently remain on study, receiving a median of 59 cycles of therapy (range 48–66). Causes of discontinuation for 13 pts included: treatment-related toxicity (8), lack of response/progressive disease (4), and recurrence of remote lung cancer (1). Toxicities leading to discontinuation included: prolonged cytopenias (3), recurrent infections (1), atrial fibrillation (1), disseminated herpes zoster (1), persistent grade 2 diarrhea (1), and grade 3 skin rash (1). Efficacy: With extended median follow-up from 20.7 to 47 mos, the ORR improved from 56% (14 pts) to 72% (18 pts), with 3 pts in PR upgrading to CR, and 1 SD to PR. Although the median time to response was 7.7 mos, responses occurred as quickly as 1.8 mos to as late as 27.0 mos of therapy. For the 3 CR pts, prolonged therapy with an additional 14.9, 28.3 and 40.6 mos beyond the time of first response was required to achieve CR. To date, 7 pts have progressed with 3-year PFS 68.8% (95% CI:52–91%) and OS 85.3% (95% CI:71.1–100%). Correlatives: Cereblon (CRBN), recently identified as a direct protein target of lenalidomide, was evaluated by gene expression profiling and Western blot and found to be uniformly expressed in all 19 evaluable day 1 pt samples regardless of lenalidomide response. Thus, baseline CRBN expression does not appear to be a useful predictive biomarker of response in this population. The mechanism by which CRBN is linked to response is reported by Trudel et al, ASH 2012. Conclusions: Long-term followup of this study demonstrates that when using low doses of single agent lenalidomide in CLL, prolonged therapy is feasible and may be required for the achievement of durable, high quality responses. Maximal daily doses of 25mg can be reached and may also be needed for optimal response, though recurrent myelosuppression remains limiting. Disclosures: Chen: Celgene: Honoraria, Research Funding; GlaxoSmithKline: Research Funding; Lundbeck: Consultancy; Janssen: Consultancy, Research Funding. Off Label Use: Lenalidomide is not approved for use in chronic lymphocytic leukemia. Kukreti:Roche: Honoraria; Celgene: Honoraria; Janssen Ortho: Honoraria. Trudel:Celgene: Honoraria, Research Funding, Speakers Bureau.

PurposeLenalidomide is an oral immunomodulatory drug with multiple effects on the immune system and tumor cell microenvironment leading to inhibition of malignant cell growth. Based on encouraging reports of lenalidomide in relapsed and refractory chronic lymphocytic leukemia (CLL), we investigated the first-line use of single-agent lenalidomide in CLL.Patients and MethodsUsing a starting dose of lenalidomide 10 mg/d for 21 days of a 28-day cycle and weekly 5-mg dose escalations to a target of 25 mg, we encountered severe toxicities (tumor lysis, fatal sepsis) in the first two patients enrolled. The study was halted and the protocol amended to a more conservative regimen: starting dose of lenalidomide 2.5 mg with monthly escalations to a target dose of 10 mg, and extended tumor lysis prophylaxis and monitoring. Gene expression profiles from patient samples before and after 7 days of lenalidomide were performed.ResultsTwenty-five patients were enrolled on the amended protocol. No further tumor lysis events were reported. Tumor flare was common (88%) but mild. Grade 3 to 4 neutropenia occurred in 72% of patients, with only five episodes of febrile neutropenia. The overall response rate was 56% (no complete responses). Although rapid peripheral lymphocyte reductions were observed, rebound lymphocytoses during the week off-therapy were common. Lenalidomide-induced molecular changes enriched for cytoskeletal and immune-related genes were identified.ConclusionLenalidomide is clinically active as first-line CLL therapy and is well-tolerated if a conservative approach with slow dose escalation is used. A lenalidomide-induced molecular signature provides insights into its immunomodulatory mechanisms of action in CLL.

Abstract Overexpression of fibroblast growth factor receptor 3 (FGFR3) is a hallmark of t(4;14) multiple myeloma (MM). To dissect the mechanism of FGFR3 oncogenesis in MM, we used 3 FGFR selective kinase inhibitors—CHIR258, PD173074, and SU5402—and FGFR3-specific siRNA to modulate FGFR3 activity. Conversely, the ligand FGF was used to stimulate FGFR3 function in human MM cells. The transcriptional response to FGFR3 modification was recorded, and gene expression changes common to all 5 modifiers were documented. Ten genes were commonly regulated. Macrophage inflammatory protein-1 alpha (MIP-1α) was the single most differentially altered gene. MIP-1 α promoter function, gene expression, and protein secretion were each down-regulated following inhibition of FGFR3 signaling. Down-regulation of MIP-1 α was not, however, observed following FGFR3 inhibition in MM cells with RAS mutations implicating RAS-MAPK in MIP-1 α regulation. As confirmation, inhibition of ERK1 also down-regulated MIP-1 α in FGFR3 inhibitor-resistant cells harboring RAS mutations. MIP-1 α is implicated in the survival and proliferation of MM cells and the pathogenesis of MM bone disease. Our observation is the first to directly link an initiating IgH translocation not only to MM-cell growth and survival but also to the disease-associated bone disease.