Thompson, Jill

Abstract Diffuse intrinsic pontine glioma (DIPG) is a rare pediatric brainstem tumor that is the leading cause of death among pediatric brain malignancies. Despite over 200 clinical trials, no significant progress has been made, leading our lab to consider suicide gene therapy. The brainstem has particularly sensitive anatomy, creating the potential for toxic inflammation. Furthermore, the clinical standard of care for DIPG includes conventional radiation therapy (RT) followed by dexamethasone (DEX), each of which can alter the efficacy of gene therapy. In this study we found that a variety of gliomas, including human DIPG and murine glioma cell lines, are infectable in vitro with a replication-competent Moloney murine leukemia virus (MoMLV) (currently in Phase 2/3 clinical trials) and susceptible to ganciclovir (GCV)-mediated killing via the HSV thymidine kinase (tk) suicide gene. A B16 melanoma cell line expressing HSV tk implanted into the brainstem of mice, extended median survival with GCV therapy from 13 days to 38.5 days (p=.0006) with no signs of toxic inflammation. Pre-treatment with RT and DEX did not affect therapeutic efficacy. We have shown that GCV/TK therapy is mediated by NK and CD8 cells in subcutaneous tumors, and we are investigating whether a similar immunologic mechanism is involved in this brainstem model. These preclinical data will pave the way for novel clinical trials for the treatment of DIPG.

Abstract INTRODUCTION Malignancies of the central nervous system (CNS) have had largely unchanged survival outcomes despite decades of research. Recently, viral-based therapeutics have shown some benefit for patients with CNS malignancies in early clinical trials. Adenovirus has been demonstrated as safe and is currently being examined in several phase I and II clinical trials. We recently demonstrated that adenovirus expressing CD40L is effective in enhancing survival in murine models of diffuse midline glioma. Therefore, to enhance the tumor specificity of this virotherapy, we hypothesized that by using a novel conditionally replicative adenovirus expressing CD40L, CRAd657-CD40L, we would maintain this survival benefit in multiple murine models for high grade glioma while decreasing off-target toxicity. METHODS We examined the utility of conditionally replicative adenovirus expressing CD40L in both in vitro and in vivo studies. Human cell lines from diffuse intrinsic pontine glioma (DIPG) and glioblastoma were used to confirm infectivity and CD40L expression, and syngeneic murine models of glioma were evaluated for toxicity and survival following intratumoral injection of a conditionally replicative adenoviral vector. RESULTS CRAd657-CD40L generated strong expression of CD40L in human in vitro DIPG XIII and U251 cell lines and induced MHCII expression on CD11c+ DC’s in U251/DC co-culture. Further, in syngeneic murine models of glioma, conditionally replicative adenoviral treatment significantly reduced toxicity while retaining survival efficacy. CONCLUSIONS Given these promising results as well as the critical need for novel therapeutics in CNS malignancies, we are now progressing to human trials targeting pediatric HGG, an unmet need in pediatric neuro-oncology. This would be the first-in-human study using CRAd-657-CD40L in pediatric HGG. In this Phase 1 clinical trial, we hypothesize that intratumoral injection of CRAd657-CD40L will cause selective expression of CD40L, increased infiltration of immune cells into the tumor, and safely enhance tumor clearance.

Oncolytic viruses (OVs) encoding a variety of transgenes have been evaluated as therapeutic tools to increase the efficacy of chimeric antigen receptor (CAR)–modified T cells in the solid tumor microenvironment (TME). Here, using systemically delivered OVs and CAR T cells in immunocompetent mouse models, we have defined a mechanism by which OVs can potentiate CAR T cell efficacy against solid tumor models of melanoma and glioma. We show that stimulation of the native T cell receptor (TCR) with viral or virally encoded epitopes gives rise to enhanced proliferation, CAR-directed antitumor function, and distinct memory phenotypes. In vivo expansion of dual-specific (DS) CAR T cells was leveraged by in vitro preloading with oncolytic vesicular stomatitis virus (VSV) or reovirus, allowing for a further in vivo expansion and reactivation of T cells by homologous boosting. This treatment led to prolonged survival of mice with subcutaneous melanoma and intracranial glioma tumors. Human CD19 CAR T cells could also be expanded in vitro with TCR reactivity against viral or virally encoded antigens and was associated with greater CAR-directed cytokine production. Our data highlight the utility of combining OV and CAR T cell therapy and show that stimulation of the native TCR can be exploited to enhance CAR T cell activity and efficacy in mice.

Abstract Background Diffuse midline glioma, formerly DIPG (diffuse intrinsic pontine glioma), is the deadliest pediatric brainstem tumor with median survival of less than one year. Here, we investigated (i) whether direct delivery of adenovirus-expressing cluster of differentiation (CD)40 ligand (Ad-CD40L) to brainstem tumors would induce immune-mediated tumor clearance and (ii) if so, whether therapy would be associated with a manageable toxicity due to immune-mediated inflammation in the brainstem. Methods Syngeneic gliomas in the brainstems of immunocompetent mice were treated with Ad-CD40L and survival, toxicity, and immune profiles determined. A clinically translatable vector, whose replication would be tightly restricted to tumor cells, rAd-Δ24-CD40L, was tested in human patient–derived diffuse midline gliomas and immunocompetent models. Results Expression of Ad-CD40L restricted to brainstem gliomas by pre-infection induced complete rejection, associated with immune cell infiltration, of which CD4+ T cells were critical for therapy. Direct intratumoral injection of Ad-CD40L into established brainstem tumors improved survival and induced some complete cures but with some acute toxicity. RNA-sequencing analysis showed that Ad-CD40L therapy induced neuroinflammatory immune responses associated with interleukin (IL)-6, IL-1β, and tumor necrosis factor α. Therefore, to generate a vector whose replication, and transgene expression, would be tightly restricted to tumor cells, we constructed rAd-Δ24-CD40L, the backbone of which has already entered clinical trials for diffuse midline gliomas. Direct intratumoral injection of rAd-Δ24-CD40L, with systemic blockade of IL-6 and IL-1β, generated significant numbers of cures with readily manageable toxicity. Conclusions Virus-mediated delivery of CD40L has the potential to be effective in treating diffuse midline gliomas without obligatory neuroinflammation-associated toxicity.

Abstract Antitumor T-cell responses raised by first-line therapies such as chemotherapy, radiation, tumor cell vaccines, and viroimmunotherapy tend to be weak, both quantitatively (low frequency) and qualitatively (low affinity). We show here that T cells that recognize tumor-associated antigens can directly kill tumor cells if used at high effector-to-target ratios. However, when these tumor-reactive T cells were present at suboptimal ratios, direct T-cell–mediated tumor cell killing was reduced and the ability of tumor cells to evolve away from a coapplied therapy (oncolytic or suicide gene therapy) was promoted. This T-cell–mediated increase in therapeutic resistance was associated with C to T transition mutations that are characteristic of APOBEC3 cytosine deaminase activity and was induced through a TNFα and protein kinase C–dependent pathway. Short hairpin RNA inhibition of endogenous APOBEC3 reduced rates of tumor escape from oncolytic virus or suicide gene therapy to those seen in the absence of antitumor T-cell coculture. Conversely, overexpression of human APOBEC3B in tumor cells enhanced escape from suicide gene therapy and oncolytic virus therapy both in vitro and in vivo. Our data suggest that weak affinity or low frequency T-cell responses against tumor antigens may contribute to the ability of tumor cells to evolve away from first-line therapies. We conclude that immunotherapies need to be optimized as early as possible so that, if they do not kill the tumor completely, they do not promote treatment resistance.

Abstract We have previously shown that a vesicular stomatitis virus-based cDNA library expressing xenogeneic altered self-epitopes led to T-cell mediated rejection of prostate cancer, glioma and melanoma. The use of a library of altered self-peptides which reflects the transcriptome of the tumor circumvents the necessity for a priori knowledge of which antigens may be immunogenic in a patient, and primes a polyclonal T-cell response that provides strong cumulative antitumor selective pressure. We have expanded this concept of altered-self library vaccination using tumor vaccines modified by the overexpression of APOBEC3B, a cytosine deaminase that generates C to T transition mutations. A freeze-thawed whole tumor cell vaccine prepared from B16 melanoma cells stably overexpressing human APOBEC3B treated established subcutaneous parental B16 tumors and, when combined with PD-1 checkpoint blockade, cured between 75%-100% of mice. T-cells from mice treated with the vaccine and anti-PD1 produced high levels of IFN-γ when restimulated with parental unmodified B16 melanoma cells in vitro. Whole-genome sequencing of B16APOBEC3B overexpressing cells identified 301 C to T or G to A missense mutations unique to the APOBEC3B line. Using an in silico MHC binding affinity algorithm, we identified and experimentally validated a short list of 10 APOBEC3B- induced heteroclitic peptides. We have also shown that this approach can be extended to an orthotopic brainstem model of high grade glioma. GL261 tumors stereotactically implanted into the brainstem were significantly responsive to treatment with an APOBEC3B- modified GL261 vaccine in combination with anti-PD-1 checkpoint blockade. In summary, when overexpressed in tumor cells, APOBEC3B generates a library of heteroclitic sequences that primes both CD4 and CD8 T-cells that have escaped central tolerance and that recognize both newly mutated antigens from the vaccine, and the corresponding unaltered self epitopes expressed on the tumor cells. Citation Format: Richard Vile, Laura Evgin, Timothy Kottke, Matthew Schuelke, Christopher B. Driscoll, Amanda L. Huff, Jill Thompson, Amy Molan, Reuben S. Harris, Jose S. Pulido, Phonphimon Wongthida. Cancer immunotherapy with APOBEC3B-induced heteroclitic library tumor cell vaccines and immune checkpoint blockade [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr B138.

Abstract Although the chimeric antigen receptor (CAR) T-cell platform has experienced clinical success in patients with hematologic malignancies, CAR T-cells specific to solid tumor targets have met with more limited efficacy. The highly inflammatory nature of oncolytic viruses and their ability to remodel the tumor microenvironment suggested to us that they would provide a complementary mechanism of action to both recruit and potentiate the functionality of CAR T-cells. VSVmIFNβ injection into B16EGFRvIII tumors increased the expression of chemokines such as CXCL10 and CCL5, which we would expect to recruit CXCR3+and CCR5+ EGFRvIII specific murine CAR T-cells. However, we did not observe an increase in overall survival or tumor control using the combination strategy compared to monotherapy with CAR T-cells. We recovered fewer viable CD8+ CAR T-cells from tumors injected with VSVmIFNβ early after adoptive transfer, and observed a similar reduction in the number of CD8+ CAR T-cells which persisted long term in the blood. We have shown that type I interferon increases the expression of the CAR from the retroviral LTR and in turn sensitizes these cells to tonic signaling mediated exhaustion and apoptosis. Correspondingly, CAR T-cells prepared from IFNAR KO T-cells were protected from the deleterious effect of type I IFN in vivo. We are currently investigating strategies to overcome the interference between these two modalities and to uncouple the regulation of the expression of the CAR from type I IFN signaling. Citation Format: Laura Evgin, Amanda L. Huff, Phonphimon Wongthida, Jill Thompson, Timothy Kottke, John Sampson, Luis Sanchez Perez, Richard Vile. Unexpected antagonism between oncolytic virus derived type I interferon and EGFRvIII CAR T-cells [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr A029.

Abstract Immunotherapy is showing promise for otherwise incurable cancers. Oncolytic viruses (OVs), developed as direct cytotoxic agents, mediate their antitumor effects via activation of the immune system. However, OVs also stimulate antiviral immune responses, including the induction of OV-neutralizing antibodies. Current dogma suggests that the presence of preexisting antiviral neutralizing antibodies in patients, or their development during viral therapy, is a barrier to systemic OV delivery, rendering repeat systemic treatments ineffective. However, we have found that human monocytes loaded with preformed reovirus–antibody complexes, in which the reovirus is fully neutralized, deliver functional replicative reovirus to tumor cells, resulting in tumor cell infection and lysis. This delivery mechanism is mediated, at least in part, by antibody receptors (in particular FcγRIII) that mediate uptake and internalization of the reovirus/antibody complexes by the monocytes. This finding has implications for oncolytic virotherapy and for the design of clinical OV treatment strategies. Cancer Immunol Res; 6(10); 1161–73. ©2018 AACR.

Abstract Purpose: Virotherapies are maturing in the clinical setting. Adenoviruses (Ad) are excellent vectors for the manipulability and tolerance of transgenes. Poor tumor selectivity, off-target sequestration, and immune inactivation hamper clinical efficacy. We sought to completely redesign Ad5 into a refined, tumor-selective virotherapy targeted to αvβ6 integrin, which is expressed in a range of aggressively transformed epithelial cancers but nondetectable in healthy tissues. Experimental Design: Ad5NULL-A20 harbors mutations in each major capsid protein to preclude uptake via all native pathways. Tumor-tropism via αvβ6 targeting was achieved by genetic insertion of A20 peptide (NAVPNLRGDLQVLAQKVART) within the fiber knob protein. The vector's selectivity in vitro and in vivo was assessed. Results: The tropism-ablating triple mutation completely blocked all native cell entry pathways of Ad5NULL-A20 via coxsackie and adenovirus receptor (CAR), αvβ3/5 integrins, and coagulation factor 10 (FX). Ad5NULL-A20 efficiently and selectively transduced αvβ6+ cell lines and primary clinical ascites-derived EOC ex vivo, including in the presence of preexisting anti-Ad5 immunity. In vivo biodistribution of Ad5NULL-A20 following systemic delivery in non–tumor-bearing mice was significantly reduced in all off-target organs, including a remarkable 107-fold reduced genome accumulation in the liver compared with Ad5. Tumor uptake, transgene expression, and efficacy were confirmed in a peritoneal SKOV3 xenograft model of human EOC, where oncolytic Ad5NULL-A20–treated animals demonstrated significantly improved survival compared with those treated with oncolytic Ad5. Conclusions: Oncolytic Ad5NULL-A20 virotherapies represent an excellent vector for local and systemic targeting of αvβ6-overexpressing cancers and exciting platforms for tumor-selective overexpression of therapeutic anticancer modalities, including immune checkpoint inhibitors. Clin Cancer Res; 24(17); 4215–24. ©2018 AACR.