Development of Therapeutic Vaccines for Ovarian Cancer

Ovarian cancer remains the deadliest of all gynecologic malignancies. Our expanding knowledge of ovarian cancer immunology has allowed the development of therapies that generate systemic anti-tumor immune responses. Current immunotherapeutic strategies include immune checkpoint blockade, cellular therapies, and cancer vaccines. Vaccine-based therapies are designed to induce both adaptive and innate immune responses directed against ovarian cancer associated antigens. Tumor-specific effector cells, in particular cytotoxic T cells, are activated to recognize and eliminate ovarian cancer cells. Vaccines for ovarian cancer have been studied in various clinical trials over the last three decades. Despite evidence of vaccine-induced humoral and cellular immune responses, the majority of vaccines have not shown significant anti-tumor efficacy. Recently, improved vaccine development using dendritic cells or synthetic platforms for antigen presentation have shown promising clinical benefits in patients with ovarian cancer. In this review, we provide an overview of therapeutic vaccine development in ovarian cancer, discuss proposed mechanisms of action, and summarize the current clinical experience.


Introduction
Ovarian cancer is the deadliest of all gynecologic malignancies, with an estimated incidence of 11.4 per 100,000 women and death rate of 6.9 per 100,000 women [1]. Globally, approximately 295,000 women are diagnosed yearly with mortality reaching almost 185,000 [2]. Effective screening strategies to detect early stages of ovarian cancer are lacking, thus 75% of women are diagnosed at an advanced stage with a 46% survival five years after diagnosis [3].
Ovarian cancer treatment and management is typically comprised of surgery and chemotherapy. Primary treatment involves a hysterectomy with bilateral salpingo-oophorectomy, comprehensive surgical staging, and debulking followed by adjuvant platinum-based chemotherapy. For patients deemed poor surgical candidates or those with a low likelihood of optimal cytoreduction, neoadjuvant chemotherapy with potential interval debulking surgery is an option [3]. Over 80% of patients will respond to initial therapy, however the majority ultimately recur and require additional therapy. The development of chemotherapy-resistant disease over the course of often multiple lines of therapy is one of the major obstacles in the treatment of recurrent ovarian cancer. This highlights the need for new therapeutic interventions, including the development of immunotherapy for treatment of ovarian cancer [4].
Immunotherapy encompasses several interventions including cancer vaccines, immune checkpoint blockade, and adoptive cell therapy with the goal of enhancing tumor recognition by the immune system and immune effector-mediated tumor cell killing [5]. This multistep process involves the priming and activation of immune effector cells, in particular cytotoxic T cells. Tumor-infiltrating lymphocytes (TILs)

Cancer Vaccines
The concept of utilizing effector T cells to recognize antigen targets for cancer treatment has been studied for over a century. The first attempts to stimulate a cancer patient's immune system were performed by Dr. William Coley in 1891. Inactivated Streptococcus pyogenes and Serratia marcescens were injected intratumorally after observing sarcoma regression in a patient with erysipelas [17]. In 1954, Black and colleagues found a correlation between the degree of lymphocytic infiltration and survival in patients with gastric carcinoma [18]. The link between immune cell infiltration and cancer survival provided evidence that cancer cells could be killed by immune cells. In 1957, Burnet suggested that differences in antigens between cancer and normal cells may be utilized to stimulate effective immunological responses [19]. Cancer vaccines have since emerged as an immunotherapy strategy that induces immune responses against tumor cells by presenting tumor specific antigens to the host.
Tumor associated antigens are recognized by the immune system and can generate T cell specific responses. Human tumor antigens are classified into one or more of the following categories: (i) differentiation antigens, (ii) mutational antigens, (iii) amplification antigens, (iv) splice variant antigens, (v) glycolipid antigens, (vi) viral antigens, and (vii) cancer testis antigens (CTAs) [20]. In addition to the antigen classifications, vaccines are also categorized into different types based on their mechanism of action: (i) dendritic cells, (ii) oncolytic viruses, (iii) modified cancer cells that secrete inflammatory cytokines, (iv) DNA encoding tumor associated antigens, and (v) intratumoral attenuated viral vaccines.

Vaccines in Ovarian Cancer
Vaccines for ovarian cancer have been studied in various clinical trials over the last three decades, however the generation of vaccine-induced humoral and cellular immune responses have not shown significant anti-tumor efficacy. Recently, improvements in vaccine development have shown more promising clinical benefits in patients with ovarian cancer (Figure 1). Table 1 summarizes data from clinical trials that have reported on the clinical experience with vaccines in ovarian cancer patients.

Dendritic Cell Vaccines
Dendritic cells (DCs) play a critical role in innate and adaptive immune responses. DCs are potent antigen presenting cells that capture and process antigens. Antigen presentation at local lymph node sites by dendritic cells stimulate antigen-specific cytotoxic T cells [63]. Vaccine development has sought to capitalize on the role DCs play in antitumor immunity. DCs pulsed with tumor-associated antigens have been shown effective as vaccine therapy in various cancer types [64].
Peptide-loaded and tumor lysate-loaded DCs are the two main strategies when using DCs as vaccines. Peptide-loaded DCs are pulsed with recombinant peptides prior to reinfusion. Data from various clinical trials have been published, providing positive efficacy signals (Table 1). Among these trials, Brossart and colleagues administered HER-2/neu or MUC1-derived peptide-pulsed dendritic cells in heavily pretreated metastatic breast and ovarian cancer patients [21]. One patient with ovarian cancer progression had a stable disease for over eight months while on therapy. Their study paved the way for additional peptide-pulsed DC vaccination therapies [22,24,27]. Loveland et al. used DCs pulsed with mannan-MUC1 fusion protein in 11 patients with adenocarcinomas. One ovarian cancer patient showed stable disease over three years of treatment [22]. Peethambaram et al. administered DCs loaded with recombinant HER-2/neu peptide and a granulocyte-macrophage colony-stimulating factor (GM-CSF) domain [24]. Two out of four ovarian cancer patients demonstrated stable disease over 15.7-18.3 months. WT1 peptide vaccines have had modest efficacy as demonstrated in various studies. The addition of low-dose cyclophosphamide prior to vaccination can potentially enhance vaccine potency [25]. In one study by Chu et al. using a HER-2/neu, hTERT, and PADRE peptide pulsed vaccine for maintenance therapy after treatment of recurrent ovarian cancer, 6 of 11 patients had no evidence of disease at 36 months, and the three-year progression-free survival was 80% with cyclophosphamide compared with 40% without. More recently, Gray et al. utilized a DC vaccine as maintenance therapy in epithelial ovarian cancer patients previously treated with one or two lines of conventional chemotherapy in complete remission [28]. CAN-003 was a phase 2b trial utilizing a MUC-1 protein-targeted DC vaccine. The treatment did not result in an increase in PFS or overall survival (OS), however patients in complete remission after second-line therapy were noted to have an improved OS with vaccination compared with controls (median OS 25.5 months with standard therapy vs. OS not yet reached with vaccination; HR 0.17; 95% CI 0.02-1.44; p = 0.07). DCs pulsed with neoantigen peptides have also been applied in the clinical setting [29,65]. Morisaki and colleagues administered a neoantigen peptide-pulsed DC in a case study of a woman with advanced stage ovarian cancer [29]. Following four rounds of vaccination, the patient had a significant decline in CA-125 levels with evidence of neoantigen-specific CTLs induced by vaccination.
DC vaccines electroporated with mRNA that subsequently is translated into protein have also been studied. Hernando and colleagues transfected DCs with mRNA-encoded folate-receptor-alpha (FR-α) [23]. Another study by Coosemans et al. loaded DCs with WT1 mRNA and found a two-month PFS and 64 month OS in their patient with serous epithelial ovarian cancer [26].
Whole tumor lysate-loaded DCs utilize whole tumor cells as a source of antigens, generating a variety of antigens associated with a specific tumor. In theory, using neoepitopes from tumor mutations will allow increased efficacy over single antigen vaccines. Bapsy et al. administered a whole tumor lysate-pulsed DC vaccine to 51 patients with advanced solid malignancies [33]. Of the seven ovarian cancer patients, one had a partial response and two had stable disease while on therapy. Hernando et al. vaccinated patients with advanced gynecologic malignancies with DCs pulsed with keyhole limpet hemocyanin (KLH) and autologous tumor cell lysate [30]. Mean progression-free interval while under vaccination was 25.5 months for patients with progressive or recurrent ovarian cancer.
Other studies have utilized personalized vaccines using autologous tumor lysate-loaded DCs and tumor antigen matched tumor cell lysates [31,32]. Tanyi et al. tested a personalized vaccine generated by autologous DCs pulsed with oxidized autologous whole-tumor cell lysate. The vaccine was injected into accessible lymph nodes in recurrent ovarian cancer patients and either administered alone, in combination with bevacizumab, or with bevacizumab plus low-dose intravenous cyclophosphamide. The treatment induced T cell responses to autologous tumor antigens and amplified T cell responses against mutated neoepitopes previously unrecognized. Overall survival of patients who showed vaccine treatment responses was 100% at 2 years compared with 25% in non-responders. Rob and colleagues provided encouraging evidence of a personalized dendritic cell vaccine (DCVAC) as maintenance therapy after primary debulking surgery and chemotherapy [66]. Interim analysis of his phase 2 trial demonstrated a 5.7-month improvement in PFS in patients receiving DCVAC sequentially after chemotherapy. Ongoing studies are underway to compare autologous oxidized tumor lysate loaded DCs with a ten peptide neoantigen based DC vaccine [65].

CTA Vaccines
Cancer testis antigen (CTA) are a type of differentiation antigen that is highly expressed in adult male germ cells with low expression in normal tissues and variably expression in tumor cells [67]. Among the over 70 cancer testis gene families identified as potential vaccine targets [67], NY-ESO-1 has been studied most extensively. NY-ESO-1 is a highly immunogenic tumor antigen that is expressed in up to 40% of ovarian cancer patients [68]. NY-ESO-1 expression in ovarian cancer is associated with a more aggressive phenotype, correlating with shorter PFS (22.2 vs. 25.0 months, p = 0.009) and OS (42.9 vs. 50.0 months, p = 0.002) [69].
NY-ESO-1 vaccination has been shown to elicit CD4+ and CD8+ T cell responses while demonstrating durable clinical responses [37,70]. Odunsi and colleagues conducted a phase I study of 18 women with NY-ESO-1-expressing ovarian cancers [35]. Patients immunized with the NY-ESO-1 derived peptide ESO157-170 had detectable ESO157-170-reactive CD4+ and CD8+ T cell responses, which correlated with a PFS of 19.0 months. Diefenbach et al. vaccinated "high-risk" ovarian cancer patients (suboptimal tumor debulking, failure of CA-125 to normalize after 3 cycles of chemotherapy, or positive second-look surgery) with NY-ESO-1b peptide and Montanide ISA-51, a vaccine adjuvant [36]. Median PFS was found to be 13 months. Sabbatini and colleagues investigated the use of overlapping long peptides from NY-ESO-1 in combination with two different vaccine adjuvants in ovarian cancer patients in second or third remission [37]. Of the 28 patients enrolled, 6 had no evidence of disease (NED) with a PFS range of 17-46 months. NY-ESO-1 is regulated by DNA methylation, and preclinical studies have demonstrated enhanced NY-ESO-1 expression and NY-ESO-1-specific CTL-mediated responses in ovarian cancer cell lines when treated with decitabine, a DNA methyltransferase inhibitor [71]. This observation provided the rationale for a clinical trial by Odunsi et al. in ovarian cancer patients. NY-ESO-1 vaccine, decitabine, and GM-CSF were administered to determine if epigenetic modulatory drugs improved antitumor response [38]. Of the 10 patients evaluable for clinical response, one had a partial response/disease remission and five had stable disease.

Protein/Peptide-Based Vaccines
Protein or peptide-based vaccines utilize defined tumor-associated antigens in conjunction with adjuvants. Tumor associated antigens are processed and presented to immune effector cells, in particular T cells, by host dendritic cells. Vaccines targeting HER-2/neu, p53, WT1, CA125, Flt3 ligand, and others have been studied in human clinical trials involving ovarian cancer.
One of the first proteins examined for an ovarian cancer vaccine therapeutic was HER-2/neu. Overexpression of the oncogene HER-2/neu is found in 15-30% of human adenocarcinomas [72]. Studies in humans have demonstrated that HER-2/neu MHC class I epitopes can induce interferon-γ-producing CD8+ T cells [39]. HER-2/neu protein immunization promotes native HER-2/neu immunity as well as antibody epitope spreading [72][73][74]. To date, there are various ongoing clinical trials involving HER-2/neu vaccine in ovarian cancer.
The tumor-suppressor protein p53 is overexpressed in almost all high grade serous ovarian cancer [75,76]. Antibodies against mutated p53 have been identified in approximately 25% of ovarian cancer patients [77]. Though induction of p53-specific immunity has been achieved with well-tolerated vaccines, the clinical efficacy has been modest thus far [40,41,43]. The overall lack of clinical benefit with a p53-specific vaccine prompted strategies for combination therapy with immunomodulatory agents. Chemotherapy, specifically cyclophosphamide, has been shown to suppress Treg function [78,79]. Treg cells in ovarian cancer have been shown to be a negative prognostic factor associated with decreased survival [80]. Vermeij et al. combined their p53-SLP vaccine with cyclophosphamide and demonstrated a 20% stable disease rate [78].
The WT1 protein is expressed in various solid cancers and hematologic malignancies, and has been ranked first in pilot prioritization of 75 cancer antigens [81,82]. The Wnt/β-catenin pathway has been implicated in the alteration of the ovarian cancer tumor microenvironment through immune cell modulation by improving DC, T cell, and macrophage function [83,84]. In ovarian cancer, WT1 expression is related to tumor type, grade, and stage, with WT1 expression highly associated with poor overall survival [85]. Ohno and colleagues administered a modified WT1 peptide vaccine to gynecological cancer patients with three out of 12 demonstrating stable disease [48]. In a phase II trial by Miyatake et al., 40 patients with gynecologic malignancies were given a WT1 peptide vaccine with 40% showing stable disease [49].
The CA125 antigen is a mucin-type glycoprotein associated with the cell membrane that has been routinely utilized as a clinical biomarker for screening and response to treatment in ovarian cancer [86]. It is a repeating peptide epitope of MUC16, which promotes malignant cell growth and inhibits anti-tumor immune responses [87]. In a large study of 119 advanced or recurrent ovarian carcinoma patients, Reinartz et al. utilized an anti-idiotypic antibody vaccine (ACA125) which mimics the CA125 antigen [45]. Overall, 68.1% were found to have an immunological response to the vaccine, with median OS of 19.4 months (range, 0.5-56.1 months). The subset of patients with antibodies to ACA125 had significantly longer survival times compared with negative responders (median 23.4 vs. 4.9 months, respectively).
The Flt3 receptor, a member of the receptor tyrosine kinase family, has also been proposed and studied as a potential vaccine antigen. In murine models, the Flt3 ligand enhances antigen-presenting cell function and stimulates natural killer cell precursor growth [44]. In a pilot study by Freedman and colleagues, the Flt3 ligand was administered to patients with ovarian cancer and mesothelioma via intraperitoneal and subcutaneous routes. Unfortunately, no objective responses were found.
Kalli and colleagues vaccinated ovarian and breast cancer patients with peptides based on folate receptor alpha, a tumor antigen expressed in a variety of cancers such as ovarian, breast, and lung [52]. Following vaccination, IFN-γ-producing T cells were enhanced, however no antibody responses were noted. All patients were alive at last follow-up of at least two years with a median relapse-free survival of 528 days in ovarian cancer patients in first remission and median survival was not reached for those in second remission.
The presentation of multiple peptides in a vaccine might theoretically increase the likelihood of generating T cells responses against a heterogenous tumor cell population and hence induce better anti-tumor responses compared to mono-valent vaccines [47,50]. A polyvalent vaccine conjugated with KLH and administered with OPT-821, an immunological adjuvant derived from the soapbark tree, was used in patients with ovarian, tubal, or primary peritoneal carcinoma of any stage [53]. Positive IgM responses were found in less than 50% of patients with median OS of 47 months.
Efforts have been made to customize cancer vaccines based on pre-existing tumor-specific antigens. Tsuda and colleagues reported on two regimens involving peptide vaccination in recurrent gynecologic cancers [46]. In their first study, patients were administered predesignated peptide vaccines, while the second study vaccinated patients with peptides to which preexisting peptide-specific cytotoxic T lymphocyte precursors in peripheral blood were confirmed. No clinical responses were found with the first regimen, however, in the second approach, seven out of 10 patients had enhanced peptide-specific cytotoxic T lymphocytes to additional peptides. Kawano and colleagues used a personalized peptide vaccine where antigens were selected based on pre-existing host immunity [51]. IgG responses were found augmented in 96.7% of patients following the 12th vaccination, however 31 of 37 cases showed disease progression, suggesting delayed tumor progression.
More recently, Hardwick and colleagues evaluated a Modified Vaccinian Ankara vaccine delivering wild-type p53 (p53MVA) in platinum-resistant ovarian cancer [58]. Patients received a combination of p53MVA and gemcitabine. There was one partial response and three with stable disease with a median PFS of three months (range, 0.95-9.2 months). Five of the 11 patients demonstrated increased p53-reactive CD4+ and CD8+ T cells. In a subset analysis, there was a significant difference in median PFS between responders and non-responders (7.0 vs. 2.3 months, respectively).

Conclusions and Future Perspectives
Vaccine therapy for ovarian cancer has been studied in various clinical trials, and the development of new platforms and combinations with chemotherapy and adjuvants show promising clinical benefit ( Table 2). There are still several challenges in creating safe and effective therapeutic cancer vaccines. The immunosuppressive and heterogenous tumor microenvironment in ovarian cancer remains a challenge. More studies are needed to improve vaccine-host interactions and to understand the variable immune responses to vaccine therapy. Other limitations include the labor-intensive protocols required to generate vaccines including surgical resection of tumor and the generation of autologous DCs. In addition, further studies are needed to determine the optimal indication for vaccine therapy. Maintenance therapy using vaccines to stimulate long-lasting immune system-mediated disease control might improve prognosis particularly in patients that do not derive a significant benefit from PARP inhibition. Novel research utilizing clustered regularly interspaced short palindromic repeats (CRISPR)-caspase 9 (Cas9) gene editing is currently underway, and the advent of more precise gene function alteration for therapy is on the horizon [90]. It is conceivable that continuous optimization of tumor antigen identification and presentation will lead to more effective therapeutic vaccines.