Whole tumor cell vaccines engineered to secrete GM‐CSF (GVAX)

Generation of immunity against cancer through vaccination has long been an elusive goal for tumor immunologists. Putative candidates for vaccination targets include oncofetal antigens, viral antigens, neoantigens, and differentiation antigens. The first attempts at cancer vaccination used injections of whole autologous tumor cells. However, these unmodified tumor cells did not engender a robust immune response. Subsequent efforts were focused at enhancing the immunogenicity of whole autologous tumor cell vaccines through genetic modification, often through virally mediated transduction of genes encoding immunostimulatory molecules. Of many immunostimulatory cytokines evaluated in the context of gene‐modified tumor cell vaccines, granulocyte–macrophage colony‐stimulating factor (GM‐CSF) emerged as the most potent in generating protective antitumor immunity. Vaccination using irradiated, GM‐CSF producing tumor cells (GVAX) consistently induced antitumor immunity across several experimental tumor models. The term GVAX can connote GM‐CSF secreting cell vaccines prepared with different vectors as well as vector targets including autologous tumor cells, allogeneic tumor cell lines, and bystander third party tumor cells lines. GVAX has been evaluated against solid tumors, hematologic malignancies, and in the context of hematopoietic stem cell transplantation. GVAX has been extensively studied in clinical trials, both alone and in conjunction with lymphodepleting chemotherapy, immune checkpoint inhibitors, and other vaccines.

presentation, upregulate B7 family members, migrate to regional lymph nodes, and stimulate the priming and/or expansion of cancerspecific CD4 + and CD8 + T cells. 3,4 From there, CD4 + T cells are able to foster an anticancer antibody response, provide help to CD8 + T cells to attack the malignancy through perforin-granzyme effector molecules, and directly contribute to tumor clearance through the secretion of proinflammatory cytokines.
A long studied approach to boost antitumor immune responses has been the generation of cancer vaccines, an antigenic payload that stimulates the priming and expansion of cancer-reactive T-cell clones. Putative candidates for vaccination targets include oncofetal antigens that are re-expressed in malignant tumors through epigenetic modification, viral antigens, neoantigens, and differentiation antigens, the latter which may also be expressed on normal tissues. Defined tumor antigen vaccines are derived from specific gene products and can include peptides, full-length proteins, or genetically encoded vectors. 5 Delivery of defined antigen or whole-cell material can also be facilitated by direct loading onto autologous dendritic cells ex vivo, followed by inoculation into patients. Tumor cell cancer vaccines employ the entire tumor cell and, in the cases of autologous tumors, the associated tumor stroma that can potentiate immune activation.
The first attempts at cancer vaccination in the early 20th century used injections of whole autologous tumor cells, irradiated so that they did not cycle. However, these unmodified tumor cells did not engender a robust immune response. 5

BIOLOGY OF GM-CSF
GM-CSF protein was first isolated from conditioned media from the lungs of lipopolysaccharide (LPS)-treated mice and subsequently cloned. 10,11 GM-CSF is produced by natural killer (NK) cells, invariant natural killer (iNKT) cells, innate lymphoid cell (ILC)3 cells, and T helper 17 (Th17) cells. 12,13 Nonhematopoietic cells such as fibroblasts and epithelial cells can also secrete GMCSF during inflammation. 14,15 The receptor for GM-CSF is expressed on granulocyte-monocyte progenitors, granulocytes, monocytes, macrophages, dendritic cells (DCs), and microglia. GM-CSF increases the number of circulating neu-trophils, monocytes, and eosinophils and also increases granulocyte activation. 16 GM-CSF is key to the rapid expansion and differentiation of myeloid cells in inflammation. The combined lymphoid and nonhematopoietic production of GM-CSF in inflammation causes GM-CSF levels to rise in circulation, exerting a hormone-like influence on the bone marrow, and increasing neutrophil and monocyte production. 17 The binding of GM-CSF to target cells that express its receptor enhances their inflammatory capability. GM-CSF provides antiapoptotic signals to neutrophils and upregulates integrin CD11b to increase neutrophil trafficking into tissues. GM-CSF induces monocytes to differentiate into CD11b+ monocytic DCs (moDCs), which phagocytose peptides and cross-present antigens to prime CD8 + T cells. 1,16 It also increases the activation of macrophages, DCs, and NKT cells to promote cancer antigen presentation, including the induction of DCs to migrate to regional lymph nodes to prime T and B cells.
GM-CSF can have an immunomodulatory as well as a proinflammatory effect. In GM-CSF-deficient mice, GM-CSF can induce myeloid immunosuppression to resolve inflammation for wound healing and tissue repair. 16,17 Treg-deficient mice demonstrate defective inflammatory responses. 18 Though generally considered as a proinflammatory cytokine, GM-CSF is now known to exert dual controls on immunity in a context-dependent manner. Under certain conditions, GM-CSF can also stimulate Treg expansion in vivo. 19,20 This function of GM-CSF is in part related to the induction of milk-fat globule epidermal growth factor-8 (MFG-E8) in DCs and macrophages, which mediates uptake of apoptotic cells and downstream antigen degradation pathways that result in Treg induction and tolerance. 20 In the B16 tumor model, combinatorial treatment with GVAX and a dominant-negative form of MFG-E8 attenuated Treg induction and potentiated GM-CSF efficacy, providing strong evidence for Treg targeting as a mean to increase the therapeutic potential of GVAX. 21 In mice, priming of Treg is observed with DCs differentiated from bone marrow in the presence of GM-CSF and linked to DC expression of OX40L and Jagged1. 22,23 Human DCs generated in the presence of GM-CSF without other cytokines can be tolerogenic. 24

CLINICAL APPLICATIONS OF GM-CSF WITH AND WITHOUT VACCINATION IN CANCER THERAPY
GM-CSF has been incorporated into two clinically approved therapeutic vaccines. Sipuleucel-T is an autologous myeloid cell vaccine preparation in which dendritic cells are differentiated ex vivo in the presence of a protein conjugate of GM-CSF and the antigen prostatic acid phosphatase. In a randomized clinic trial, Sipuleucel-T prolonged survival in patients with asymptomatic or minimally symptomatic metastatic castrate-resistant (hormone refractory) prostate cancer. 25 Talimogene laherparepvec, or T-VEC, is a modified oncolytic herpes virus that produces human GM-CSF and is injected intra-tumorally for the treatment of discrete inoperable metastatic melanoma lesions. 26,27 T-VEC improved durable response rates and overall survival (OS) compared to GM-CSF therapy alone. T-VEC has been evaluated in combination with checkpoint inhibitors including ipilimumab in melanoma and pembrolizumab for sarcoma and has shown preliminary evidence of therapeutic synergy. 28,29 A phase 2 randomized study demonstrated superiority of ipilimumab plus T-VEC to ipilimumab alone. Thirty-eight patients (39%) in the combination arm and 18 patients (18%) in the ipilimumab arm had an objective response (odds ratio, 2.9; 95% CI, 1.5-5.5; p = 0.002). 30 T-VEC has also been effective in patients with locoregional disease resistance to checkpoint inhibition. 31 GM-CSF (sargramostim) has been incorporated into immunomodulatory cancer therapeutic regimens without vaccination in hopes of stimulating antitumor inflammatory responses. 32 A regimen combining GM-CSF with anti-GD2 antibody, IL-2, and isotretinoin has been approved for the treatment recurrent neuroblastoma. 33

GVAX IN SOLID TUMORS
A number of clinical trials have been conducted with GVAX, predominantly in solid tumors (Table 1). While vaccination appears safe and  In the initial study of patients with advanced pancreatic cancer, there were no objective responses among 14 vaccinated patients although several did experience prolonged survival associated with evidence of immune activation in vivo. 54 Allogeneic GVAX has been administered following pancreaticoduodenectomy to be followed by systemic 5-FU chemotherapy and booster vaccination. In this single-arm trial, median disease-free survival was 17 months and OS 24 months. 55  Experimental models had suggested a synergy between vaccineinduced immune responses and PD-1/PD-L1 pathway blockade. 75 However, when nivolumab was added to patients receiving Cy/GVAX with CRS-207, there was no significant improvement in response or median survival. 76 [82][83][84] Regulatory T cells are also relatively deficient early after HCT, setting the stage for what be optimal conditions for GVAX. 85 GVAX has been tested in the autologous transplant setting in humans. 86  a TLR-independent pathway can also be harnessed to enhance immune activation. GVAX formulated with a STING agonist (STINGVAX) generated enhanced antitumor immunity compared to GVAX alone, and PD-1 blockade further potentiated tumor control. 93 Linking GVAX with tetanus toxoid is being tested in a cervical cancer model. 94 A neoantigen-based GVAX combination is being explored in colorectal cancer. 95 New checkpoint combinatorial strategies incorporating agonist OX-40 antibodies appear effective in murine glioma. 96,97 The proand anti-inflammatory properties of GM-CSF are being exploited by a codon-modified GM-CSF tumor cell vaccine to maximize inflammatory and suppress inhibitory effects. 98 Although GVAX clinical studies to date have been disappointing, these studies have taught the field much about the subtleties of tumor immunity. A deeper understanding of the checks and balances regulating GVAX-mediated immune responses will be necessary if it is to play a pivotal role in cancer therapy.