A Phase I/II trial comparing autologous dendritic cell vaccine pulsed either with personalized peptides (PEP-DC) or with tumor lysate (OC-DC) in patients with advanced high-grade ovarian serous carcinoma

Cancer vaccines are intended to “educate” the patient’s own immune system to generate effector T-cells specifically for tumor cells to be detected and destroyed. A tailored cancer vaccine aims to target multiple patient-specific tumor antigens and reduce side-effects by protecting normal tissue and keeping tumors under immune memory regulation for as long as possible [18]. Dendritic cell (DC)-based vaccines are a particularly attractive option for immunotherapy, due to their low toxicity profile, lack of invasive procedures and their potential to induce long-term effects through immunological memory [19]. DCs are unique immune cells responsible for processing and presenting cancer antigens, and are capable of initiating and regulating both innate and adaptive immunity [20]. DCs can present endogenous antigens as human leukocyte antigen (HLA) class I peptides and exogenous antigens as either HLA class II peptides or HLA class I peptides by cross presentation, thus effectively inducing antigen-driven T-cell responses. Monocyte-derived human DCs pulsed with TAAs have been extensively used for clinical therapies against malignancies [21]. Unfortunately, DC vaccines have demonstrated limited efficacy in patients with advanced recurrent disease [22]. Some promising results however suggest a need for further optimization, including combination of different immunotherapy technologies and multiple antigens.

Key factors leading to the poor immune response in ovarian cancer include lack of well-characterized tumor antigens, molecular heterogeneity, selective tumor antigen-loss (immuno-editing) and the immunosuppressive nature of the tumor microenvironment [23]. When vaccines target defined non-mutated self-antigens or shared antigens that are overexpressed in the tumor, vaccine efficacy is often low because T cell reactivity to self-antigens is naturally reduced due to central tolerance [24]. Alternatively, neoantigens (NeoAgs) that arise from somatic DNA alterations as a result of genetic instability are cancer-specific and can be strongly immunogenic [25]. NeoAgs are likely to be effective targets for tumor infiltrating T cells and can lead to successful immunotherapy treatments [26], hence synthetic vaccines targeting patient-specific NeoAgs can display increased efficacy against tumors with moderate or high mutation load. Three recent phase I studies using personalized NeoAg-based vaccines reported immunogenicity and interesting clinical safety and efficacy results [27‐30].

An alternative source of personalized antigens is the whole tumor lysate. In the case of ovarian cancer, tumor cells can be easily recovered by cytoreductive surgery and tumor antigens can be obtained directly from the patient’s own tumor cells by preparing a tumor lysate that contains both the TAAs and the private NeoAg without the hurdles of target identification and preparation [31]. The immunogenicity of whole-tumor antigen vaccines can be enhanced by different tumor lysate preparation methods [31, 32]. During the final culture stage, tumor lysate is loaded onto the DCs and then the DC vaccine is presented to the immune system via intranodal injection(s). The advantage of these whole tumor vaccines is that they target a whole range of antigens thus reducing the chance of tumor escape compared to single epitope vaccines. They are independent of HLA type and consist of epitopes for CD8+ cytotoxic T-cells (CTLs) and CD4+ T helper (Th) cells, leading to a more integral immune response [33]. Previous studies with DC vaccines loaded with whole tumor lysate have already demonstrated that patients with measurable disease, in cases such as non-Hodgkin’s lymphoma [34], melanoma [35], or renal cancer [36], may obtain clinical benefit. We also reported previously on a vaccine generated using autologous monocyte-derived DCs pulsed with autologous tumor lysate produced from tumor tissue dissociated into single cells, oxidized and lysed using freeze–thaw cycles [37]. Hypochlorous acid (HOCl) oxidization approach has previously been shown to be superior to ultraviolet B irradiation or freeze–thaw of tumor lysis, in terms of priming T cell responses against tumor antigens in vitro, and has shown promise in an OC preclinical model [37]. We also showed that it was feasible to produce oxidized lysate-pulsed DCs in five patients with OC and to perform intranodal injections of this vaccine [37], which consisted of ex vivo-cultured, autologous monocyte-derived DCs pulsed with oxidized autologous tumor lysate [38, 39].

In a phase I trial, the OC-DC vaccine was tested in 25 patients with platinum-treated, immunotherapy-naïve, recurrent ovarian cancer in which study the OC-DC treatment injected intranodally was found to be safe, feasible, able to induce T cell responses to autologous tumor antigen, and was associated with significantly prolonged survival [40]. In the cohort receiving the vaccine combined with cyclophosphamide and bevacizumab, the median progression free survival was 11.1 months compared with 4.1 months in the historical control population. Survival at 2 years was 100% for the vaccine plus bevacizumab/cyclophosphamide group, while it was only 40% for patients receiving vaccine and bevacizumab only, as well as for control patients receiving bevacizumab and cyclophosphamide (p = 0.011), confirming that adding cyclophosphamide to the vaccine could increase its effect. The 2-year OS rate for patients responding to treatment was 100%, whereas it was 25% for non-responders. Importantly, only patients with an immune response to whole tumor lysate or autologous tumor showed clinical benefit [40]. Upon analysis of the peripheral blood mononuclear cells (PBMCs) collected pre-vaccination and after five doses of induction vaccination, the frequency of tumor-reactive T cells in the peripheral blood was found to increase gradually over time. Furthermore, six patients displayed CD8+ responses on immunization against one or more neo-epitopes. This study was the first to show that vaccination with whole tumor lysate loaded DCs elicited a CD8 T cell response to antigens derived from private non-synonymous somatic tumor mutations. These results demonstrate that personalized vaccines using whole tumor lysate can enhance pre-existing immune responses to NeoAgs as well as rationalize the potential combinatorial use of whole tumor lysate vaccine followed by NeoAgs targeting to enhance the anti-tumor effect of vaccinations [40].

Early preclinical evidence suggested that ovarian cancer was unsuitable for NeoAg-specific vaccination [41]. Subsequent analysis of human ovarian cancer specimens, however, indicated that ovarian cancer tissues may express NeoAgs [42, 43] and that their expression is associated with better OS [44]. NeoAgs promise high specificity but are hard to identify because they are mostly patient-specific, and they are mainly rare events in a patient cohort. Personalized vaccine design requires the identification of each patient’s NeoAg tumor repertoire (“mutanome”), which has only been possible in recent years due to significant technological advancements such as next-generation sequencing (NGS), analysis of immunopeptidome by mass spectrometry (MS) and development of bioinformatical prediction tools [45‐48]. Limitations concerning the availability of relatively large sample volumes requested (> 1 g of tissue) and the logistics involved in the operation of specialized and sophisticated MS instruments have so far hampered the integration of immunopeptidomics into routine clinical practice. The implementation of an individualized treatment concept based on the mutanome requires both highly interdisciplinary research and an innovative drug development process [49]. It is therefore unclear at this point, whether synthetic vaccines based on private antigens are going to be more effective than whole tumor lysate vaccines and could potentially replace them.

Based on our previous OC-DC study observations, the specific recognition and targeting of tumor specific NeoAgs is a powerful way to further enhance the effectiveness of OC-DC vaccine. We will therefore use a personalized vaccine consisting of autologous monocyte derived DCs pulsed with personalized peptides (PEP-DC) combining tumor-specific NeoAgs and TAAs in patients with advanced ovarian cancer. In order to validate our hypothesis, patients will be randomized to receive either OC-DC or the novel PEP-DC vaccine during the initial part of this study; then the OC-DC arm will switch over to subsequently receive PEP-DC. We are therefore proposing a phase I/II, randomized two-cohort, single-center study to compare immunogenicity and assess the safety of personalized peptide pulsed DC vaccine (PEP-DC) alone, or oxidized tumor cell pulsed DC (OC-DC vaccine) followed by PEP-DC (PEP-DC2). After the comparative phase of vaccination (six cycles), all patients will be vaccinated with PEP-DC. All vaccines will be administered intranodally and applied in combination with cyclophosphamide in patients with OC (Fig. 1).

Based on the patient’s own tumor, epitopes for the PEP-DC vaccine will be determined and will be identified and produced from snapped-frozen tumor specimens and blood samples. The tumor specimens are required for the production of oxidized whole tumor lysate, for DNA and RNA extraction to complete NGS and for elution of the HLA binding peptides for MS based immunopeptidomics [50, 51]. The blood samples will be used for PBMC isolation and HLA-typing. PBMC-derived DNA will serve as the germline reference genome in order to identify personalized NeoAgs. Genomic variants affecting coding genes that are present in the tumor samples and absent from the corresponding blood samples are assumed somatic. The NGS data will be analyzed using our NeoDisc pipeline [52] to generate personalized reference databases for each patient. The MS-based immunopeptidomics data will be searched against the reference database in order to directly identify in vivo presented NeoAg as well as TAAs. In addition, based on these references, NeoAg binding to HLA-I and HLA-II molecules will be predicted in silico. Eventually, 100 personalized targets for each patient will be selected and redesigned as ~ 25mer peptides optimally spanning across the predicted ligands [52].

In order to narrow down the number of target antigens to maximum 10 per patient, we have set-up robust T-cell assays to test patient’s PBMCs, T cells and TILs for their reactivity to private peptides. Using the patients’ isolated CD4 and CD8 T-cells, we will perform functional validation of immune reactivity against the 100 candidate peptides and quantify it by IFNγ ELISpot assay. This additional step allows us to incorporate immunogenic antigens into the PEP-DC design to enhance the patient’s already existing anti-tumor immune response after vaccination, and hence, to increase the vaccine’s potential. Following this immunogenicity analysis, the targets list will be reduced to 10 target sequences and ‘enhanced quality’ peptides for the Top 10 targets will be ordered from Almac (Edinburgh, UK) for vaccine manufacturing. As a few peptides might fail synthesis, up to 10 targets will be eventually used. The Top 10 peptides selected a priori will be used to formulate the PEP-DC vaccine for Arm A patients. In Arm B, patients will first receive OC-DC vaccine then based on the immunogenicity analysis on PBMCs collected a priori and after three OC-DC vaccines (at the end of cycle 3) a personalized PEP-DC vaccine will be designed for them. The top 10 peptides determined at this time point (reflecting new immune responses induced by the preceding OC-DC vaccines) will be used to formulate the PEP-DC2 vaccine to be administered to patients in Arm B after six OC-DC administrations.

OC-DC vaccine will be manufactured using oxidized whole tumor cell lysate derived from autologous tumor harvested during laparoscopy surgery or interval debulking surgery (IDS) pulsed onto autologous dendritic cells. The choice of dendritic cell maturation and the choice of tumor cell preparation are based on data previously published [53]. In this study, the vaccines’ components include agents for which safety has previously been shown to be acceptable [40].

PEP-DC vaccines will be manufactured and formulated at the GMP (good manufacturing practice) facility of the Center of Experimental Therapeutics (CTE, Lausanne) and the Centre Hospitalier Universitaire Vaudois (CHUV, Lausanne). After the Top 10 peptides have been identified and manufactured, patients will undergo leukapheresis for generation of DCs. Autologous monocytes will be enriched in a CliniMACS Prodigy (Miltenyi Biotec) system that is a GMP-compliant closed system, allowing for standardized and reproducible cellular processing across multiple instruments. The leukapheresis bag will be attached using sterile tubing set to the CliniMACS Prodigy device, and with the predefined LP-14 enrichment program and the CD14 reagent (magnetic beads, Miltenyi Biotec), CD14+ monocytes will be purified by positive enrichment. Purified monocytes will be differentiated into immature monocyte-derived DC (iDC) by a 5 days culture in the presence of clinical grade IL-4 and GM-CSF. iDC will then be loaded with a mix of the Top 10 peptides overnight and matured/activated for 6 to 8 h on day 6 using a maturation cocktail consisting of clinical grade monophosphoryl lipid A (MPLA) and IFNγ. Finally, cells will be harvested and cryopreserved as vaccine doses, comprising 5-10 x10⁶ cells per dose. For each injection of the PEP-DC vaccine, one dose will be thawed, washed and resuspended in NaCl 0.9% supplemented with 1% human albumin before being transferred into syringes and stored at 2–8 °C until administration to patients as intranodal injection.

In a study of 104 ovarian cancer patients, regulatory T cells (Treg) have been shown to contribute to growth of human tumors in vivo [15]. It has been demonstrated that low-dose cyclophosphamide, when used in combination with immunotherapy, can reduce Treg numbers and impair their function without eliminating other immune cells [54‐56], thus creating a favorable environment for greater efficacy and immune response. For instance, pre-treatment with intravenous low-dose cyclophosphamide (300 mg/m²) improved immunogenicity of a p53-synthetic long peptide vaccine in patients with recurrent OC [57]. Tanyi et al. [58] assessed OC-DC vaccine either alone (cohort 1, n = 5) or in combination with bevacizumab (cohort 2, n = 10), or bevacizumab plus low-dose intravenous cyclophosphamide (cohort 3, n = 10) in ovarian cancer patients until disease progression or exhaustion of vaccine doses. Low-dose cyclophosphamide (200 mg/m²) was administered about 24 h before each vaccine dose, and bevacizumab (10 mg/kg) was administered on the day of vaccination. The survival observed in the cohort with cyclophosphamide was longer than previously reported in this population with bevacizumab-based biological combinations [59, 60]. Consistently, in our study we will administer intravenous cyclophosphamide at a dose of 200 mg/m² within 24 h prior to each vaccination.

This investigator-initiated trial will be conducted at the Department of Oncology at the CHUV. We propose a Phase I/II, randomized, two-cohort, single-center study in ovarian cancer patients to compare immunogenicity and assess the safety of PEP-DC alone (Arm A) or OC-DC vaccine followed by PEP-DC2 (Arm B); vaccines will be administered intranodally and in combination with cyclophosphamide in all cases. We consider that vaccination with synthetic vaccines (PEP-DC and PEP-DC2) developed on the basis of our integrated tumor antigen discovery pipeline is feasible, and can produce specific immune responses when applied in combination with low dose cyclophosphamide against mutated peptides and other private tumor antigens. The OC-DC vaccine therapy is expected to activate and expand T-cells that recognize both the NeoAg and shared tumor antigens, and to correlate with clinical benefit. Furthermore, we expect OC-DC priming to boost PEP-DC vaccination, as we hypothesize that a priori prediction of relevant immunogenic peptides is currently not optimal, and that upfront vaccination with whole tumor lysates may significantly enhance the detection of immunogenic antigens in in vitro T cell assays, and thus assist in the subsequent development of synthetic personalized vaccines. A randomization of 1:1 to arm A (PEP-DC) or arm B (OC-DC followed by PEP-DC2) will be performed to reduce potential selection bias on outcome.