Introduction

Since the identification of tumor-associated antigens (TAA) in different tumor histological types, many cancer vaccination strategies have been investigated, including peptide-based vaccines, recombinant DNA- or protein-based vaccines, and cell-based vaccines. Results from early trials, although demonstrating the feasibility and the good toxicity profile of this approach, provided evidence of clinical activity in only a minority of patients [1]. However, there have recently been noteworthy advances in the clinical application of immunotherapy. In 2010, sipuleucel-T (Provenge; Dendreon Corporation, Seattle, WA, USA), an autologous cellular immunotherapy product designed to stimulate T-cell immune responses against human prostatic acid phosphatase (PAP), was first approved for patients with castration-resistant prostate cancer (CRPC) by the U.S. Food and Drug Administration (FDA) [2]. In addition, another immunotherapeutic agent, ipilimumab, an anti-cytotoxic T-lymphocyte antigen (CTLA)-4 monoclonal antibody, was also approved for melanoma patients by the FDA in 2011 [3]. Despite these significant advances, however, most other randomized clinical trials of immunotherapies, including peptide vaccines, recombinant DNA- or protein-based vaccines, and cell-based vaccines, have so far failed to show beneficial therapeutic effects in patients compared to existing treatments [4]. The failure of recent clinical trials has raised several issues that need to be addressed for the successful development of cancer vaccines. We describe here a novel immunotherapeutic approach, “personalized peptide vaccination (PPV),” in which a maximum of four human leukocyte antigen (HLA) class IA-matched peptides are selected for vaccination from a pool of peptides on the basis of both HLA class IA type and the preexisting host immunity before vaccination. This strategy may confer several advantages, such as the possibility of bypassing both immunological diversity and tumor heterogeneity. For example, “personalized” antigens with preexisting immunity, which are designed to stimulate antigen-specific memory T cells, could be expected to induce rapid and strong secondary immune responses. For example, we previously reported that PPV quickly induced infiltration of CD45RO+ memory T cells, rather than naïve T cells or B cells, into cancer tissues [5]. In addition, selection of multiple epitopes for PPV could reduce the risk of tumor escape through existence and/or induction of antigen-negative clones escaping peptide-specific immune responses. Indeed, it would be relatively rare that tumor cells escape from peptide-specific immune responses by simultaneously losing all of multiple antigens selected for vaccination.

Characteristics of candidate peptides for PPV

A large number of tumor-associated antigens (TAA) have been identified by several different approaches, including complementary DNA (cDNA) expression cloning [6], serologic analysis of recombinant cDNA expression libraries (SEREX) [7], and a reverse immunological approach. We have identified a number of TAA genes and their peptides, some of which have been used as vaccine antigens for PPV, by cDNA expression cloning techniques. For example, a series of the squamous cell carcinoma antigens recognized by T cells (SART), including SART1, SART2, and SART3, were identified from a cDNA library of a squamous cell carcinoma cell line for the first time as TAA derived from epithelial cancers except for melanoma, by using a CTL line established from a patient with esophageal cancer [8–11]. Similarly, other TAAs with interesting characteristics, such as p56lck and multidrug resistance-associated protein 3 (MRP3), have also been identified as vaccine antigen candidates for PPV. p56lck, the src family tyrosine kinase essential for T-cell development and function is reported to be aberrantly expressed in colon, small cell lung carcinoma, and prostatic cancer cells with a trend toward preferential expression in metastatic cancer cells [12, 13]. This molecule encodes epitopes, which can frequently induce cytotoxic T lymphocytes (CTLs) in the peripheral blood lymphocytes of HLA-A2+, HLA-A24+, or HLA-A3 supertype+ cancer patients with distant metastases [14–16]. Peptides derived from MRP3, which are recognized by CTLs in an HLA-A2402-restricted manner [17], often show positive immune responses in advanced cancer patients in whom standard chemotherapy had failed.

One of the notable characteristics of PPV is to screen CTL epitope candidates for therapeutic cancer vaccines on the basis of their ability to induce CTL and/or humoral responses in pre-vaccination samples, since all of the CTL epitopes currently employed for PPV have B-cell epitopes as well. This is based on the hypothesis that a CTL peptide possessing a B-cell epitope could provide more effective clinical benefits than a CTL peptide without it. Although this hypothesis has not been confirmed by randomized clinical trials yet, it has been well recognized that both cellular and humoral immune responses are crucial to induce potent anti-tumor immunity in animal models [18, 19]. As a result of basic and clinical studies, we have focused on 31 HLA class I-restricted peptide epitopes with minimal optimal length for PPV, [2 peptides for HLA-A2, 14 peptides for HLA-A24, 9 peptides for HLA-A3 supertype (A3, A11, A31, or A33), and 4 peptides for HLA-A26] (Table 1). These peptides were identified from 15 different TAAs, including SART2, SART3, p56lck, MRP3, prostate-specific antigen (PSA), prostatic acid phosphatase (PAP), prostate-specific membrane antigen (PSMA), and a variety of other epithelial tumor antigens [8–11, 14–17, 20–28]. The safety and potential immunological effects of these vaccine candidates have been shown in previously conducted clinical studies.

Table 1 Information of peptide candidates used for personalized peptide vaccination
Full size table

Although short peptide epitopes with minimal optimal length have been reported to bear the potential to induce tolerance rather than effective immune responses [29], PPV using short epitopes has been reported to efficiently induce antigen-specific IFN-γ-producing CD8+ T cells with cytotoxic activity, but not tolerance to them, possibly because only immunogenic epitopes are selected in each patient by pre-vaccination screening. Although long peptides have shown excellent immune and clinical responses in some of clinical trials [30], we do not currently employ long peptides for PPV, since it may be possible that they contain undesirable T-cell epitopes that stimulate immune suppressive cells, such as regulatory T cells or T helper-2 cells [31], which may negatively regulate beneficial immune and clinical responses.

For PPV, a maximum of 4 peptides, selected on the basis of the results of HLA typing and the preexisting immune responses specific to each of the 31 different vaccine candidates, are subcutaneously administered in complex with incomplete Freund’s adjuvant (Montanide ISA51; Seppic, Paris, France) weekly or biweekly. To prevent interaction/competition among peptides at the vaccinated sites, each of vaccine peptides is injected separately at different sites, but not in a mixture at a single site. Since more than five peptides per vaccination seemed intolerable due to adverse skin reactions, which sometimes may cause unpleasant symptoms, such as itching and pain, in our previous feasibility studies (unpublished data), four peptides per vaccination have currently been employed. Regarding the vaccination schedule, selected peptides are administered at the weekly schedule at least for the first cycle of six vaccinations, since a clear trend toward better immune responses was observed among the patients who underwent the weekly administration protocol, compared to the biweekly protocol in previous clinical trials [32].

Rationale for PPV

Although the number of cancer vaccine candidates is becoming almost limitless, antigen peptides employed for vaccination against individual patients might not always be appropriate. In general, anti-tumor immunity is known to be dependent on both the immunological characteristics of tumor cells and the host immune cell repertoires. Since immune cell repertoires of the hosts are quite diverse and heterogeneous, anti-tumor immunity might differ substantially among individuals. Therefore, it is likely that vaccine antigens that are selected and administered without considering the host immune cell repertoires would not efficiently induce beneficial anti-tumor immune responses. To increase the clinical benefits from cancer vaccines, particular attention should be paid to the immunological status of each patient by characterizing the preexisting immune responses to vaccine antigens before vaccination. However, in most of the current clinical trials of therapeutic cancer vaccines, common antigens are employed for vaccination independently of the immunological status of patients.

Patients who have an immunological memory to vaccine antigens are expected to show quick and strong immune responses to them. In contrast, patients with no immunological memory against vaccine antigens would take more time to develop effective anti-tumor immune responses because several rounds of repeated vaccinations might be required to prime antigen-specific naive T cells to functional effector cells (Fig. 1). In such situations, vaccinations could not easily provide clinical benefits, especially in advanced cancer patients who show a relatively quick disease progression. Moreover, immune responses induced by inadequate vaccines that are non-specific to tumor cells may not only be ineffective for tumor control, but also may erode preexisting immunity. On the basis of the current paradigm that the size and composition of the adaptive immune system are limited and that individual immune cells are constantly competing with each other in a limited space, inadequate vaccination may have negative consequences for the host by suppressing preexisting beneficial memory cells specific to tumors and/or infections, which might result in acceleration of cancer progression or early death in vaccinated patients. In addition, the approach, which is designed to stimulate antigen-specific memory T cells, but not to prime naïve T cells, might not need additional immune boosting, such as the blocker of checkpoint molecule, CTLA-4, since it has been known that memory T cells are less dependent on costimulatory molecules for recall responses [33, 34]. Considering these issues, it would be quite reasonable for vaccine antigens to be selected on the basis of the preexisting immunological status in each patient.

Fig. 1
figure 1

Concept of personalized peptide vaccination. Patients who have an immunological memory to vaccine antigens are expected to show quick and strong immune responses to them. In contrast, patients with no immunological memory against vaccine antigens would take more time to develop effective anti-tumor immune responses because several rounds of repeated vaccinations might be required to prime antigen-specific naïve T cells to functional effector cells

Full size image

In addition, it should be noted that cancer cells possess or develop a variety of mechanisms to maintain their malignant behavior. For example, it has been well recognized that cancer cells escape from host immunological surveillance [35]. Through the interaction between the host immune system and tumor cells at the equilibrium phase, immunological pressure often produces tumor cell variants that decrease or lose tumor-associated antigens. Therefore, for better control of cancer cells, it would be recommended to administer multiple tumor-associated antigens to reduce the risk of outgrowth of antigen-loss variants.

In view of the complexity and diversity of the immune cell repertoires of hosts and the immunological characteristics of tumors, we have developed the new concept of PPV. In this “personalized” cancer vaccine formulation, multiple peptide antigens appropriate for vaccination are screened and selected from a list of pooled vaccine candidates in each patient, based on preexisting host immunity. In the early-phase translational study of PPV, the preexisting immunity was defined by the frequencies of CTL precursors in pre-vaccination PBMCs by using the peptide-specific IFN-γ production assay with the cutoff level of around 1 of 10,000 cells as reported, since we found that the magnitude of CTL activation could be in part dependent on the frequencies of peptide-specific CTL precursors in circulation, which were determined by this assay [36]. Indeed, when CTL precursors were measured in pre-vaccination PBMC followed by administration of peptides with higher CTL precursor frequencies, rapid and strong activation of CTL with potential clinical benefits was induced in certain patients of a series of clinical trials for advanced cancers [32, 37, 38].

Nevertheless, we are currently evaluating the preexisting immunity to vaccine candidates by peptide-specific immunoglobulin G (IgG) responses in pre-vaccination plasma, which are determined by the multiplex bead-based LUMINEX assay with the cutoff level of 10 FIU [39], rather than by CTL responses, since we have found that the IgG-based selection is useful for predicting CTL boosting after vaccination in our clinical trials, which showed the safety, high immunogenicity, and possible clinical benefits of PPV. For example, in the phase I trial of PPV for recurrent or progressive glioblastoma patients, CTL responses were boosted in 23 of 48 vaccinated peptides (48 %), which were chosen solely by humoral responses [40]. In addition, CTL responses were induced in 16 of 28 vaccinated peptides (57 %), which were chosen solely by humoral responses in another phase I trial of PPV for castration-resistant prostate cancer (CRPC) patients [41]. Based on these results, the prediction power of evaluating the preexisting immunity solely by the humoral responses for the existence of CTL responses could be estimated around 50 % when four peptides were chosen for the vaccination. Figure 2 shows one example of original data for determining the preexisting immune responses before vaccination by the humoral responses to vaccine candidates. As shown in this figure, immune responses to multiple peptides can be detected before vaccination in most of the patients treated with PPV. In such situation, the peptides showing higher IgG responses are selected for vaccination from the list of peptides that match the patient’s HLA types.

Fig. 2
figure 2

An example of boosting immune responses. IgG responses to PSA-248 peptide increased from 20 to 14,000 FIU in the post (6th)-vaccination samples. The similar boosting was observed in the other two peptides. CTL response was also increased in all four vaccinated peptides. Clinical response of this case was PR

Full size image

There are some reasons for assessing IgG responses, instead of CTL responses, to define the preexisting immune responses. The most critical reason is a technical issue that standard methods to measure the CTL activity have not been well established yet. The performance characteristics, such as sensitivity and reproducibility, of currently available CTL assays require more modification/sophistication to detect low frequencies of antigen-specific CTL, and it seems difficult to validate the quality of the assays in clinical trials [42, 43]. In contrast, the multiplex bead-based LUMINEX technology that we have developed for monitoring IgG responses allows simple, quick, and highly reproducible high-throughput screening of IgG levels specific to large numbers of peptide antigens with a tiny amount of plasma [39]. Indeed, we have recently published several papers describing the clear correlations between clinical benefits and antigen-specific B-cell responses measured by IgG antibody production in patient plasma after vaccination [44, 45]. Of course, we believe that cellular immune responses might represent the most important marker if appropriate CTL assay conditions are defined and become available. More sophisticated CTL assays remain to be developed for the further evolution of cancer vaccination.

Clinical trials of PPV for advanced cancer

To date, a series of phase I, I/II, and II clinical trials using PPV have been conducted [5, 32, 37, 38, 40, 41, 46–62]. We have summarized the observed immune and clinical responses in advanced cancer patients induced by the PPV (Table 2). In the following sections, we provide a more detailed account of these studies, categorized by the different cancer types.

Table 2 Immunological and clinical responses of personalized peptide vaccination for advanced cancer
Full size table

Castration-resistant prostate cancer (CRPC)

Most prostate cancer-related deaths occur in patients with advanced CRPC. Chemotherapy plays only a palliative role in the treatment for prostate cancer, although two docetaxel-based randomized clinical trials demonstrated a survival benefit of only 2.4 months compared with those with mitoxantrone and prednisone in CRPC patients. A large number of agents and treatment strategies including immunotherapy are currently under investigation for various stages of CRPC. Indeed, several immunotherapy strategies for advanced CRPC, such as single-peptide-based vaccine, multiple-peptide-based vaccine, cell-based vaccine, viral vaccine, antibody-based therapy, and their combination with other therapies, have been evaluated. In phase I studies of PPV for advanced CRPC, we have reported the increase in cellular and humoral immune responses and decrease in PSA levels in some patients [41, 46, 48]. Phase I dose-escalation study of PPV for CRPC with 1, 3, and 5 mg/peptide injection showed that a dose of 3 mg/peptide injection was better than those of 1 and 5 mg/peptide injections in terms of the induction of cellular immune responses to peptides, although the maximum tolerated dose (MTD) was not estimated [41]. In a phase I/II study, 58 patients with HLA-A2+ or HLA-A24+ with CRPC were treated with a combination of PPV and low-dose estramustine phosphate (EMP) [50]. As a result, the majority (76 %) of patients showed a decreased serum PSA level, along with a median survival time (MST) of 17 months ( 95 % CI, 12–25 months). In addition, this study showed that a small number of lymphocytes, a negative immunological response after PPV, and poor performance status were independent predictors of disease-related death. In this study, long MST with the combination therapy supports the hypothesis that this combination with a low-dose cytotoxic drug produces additional antitumor effects with minimum immunosuppression. Sequentially, we conducted a randomized, cross-over, phase II trial of PPV plus low-dose EMP comparing standard dose EMP in HLA-A2+ or HLA-A24+ patients with CRPC [51]. Median progression-free survival (PFS) was 8.5 months in the PPV group and 2.8 months in the EMP group with a hazard ratio (HR) of 0.28 (95 % CI, 0.14–0.61; log-rank P = 0.0012), and the MST for the PPV plus low-dose EMP group was 22.4 months, while the MST for the standard dose EMP group was 16.1 months (95 % CI, 8.0–13.4 months) (P = 0.0328). The HR for overall survival was 0.3 in favor of the PPV plus low-dose EMP group. These results suggest that PPV is well tolerated and active in CRPC patients. In another phase II study, we compared the MST in docetaxel-based chemotherapy (DBC)-resistant CRPC patients treated by PPV (n = 20) with a historical control (n = 17) [52]. MST from the first day of progressive disease (PD) were 17.8 and 10.5 months in DBC-resistant CRPC patients receiving PPV and those with no PPV, respectively. These encouraging preliminary study results suggested that PPV warrants further study as a novel therapy for CRPC patients with PD after DBC. Now, we are conducting a phase III randomized clinical trial of PPV in DBC-resistant CRPC patients.

Glioblastoma multiforme (GBM)

Although immunotherapy is theoretically attractive due to the discovery of TAAs and peptides capable of inducing specific immunity in patients with GBM, previously conducted immunotherapy trials failed to provide evidence of any definite clinical benefit in patients with GBM. One of the potential hurdles hindering the development of effective immunotherapy for the treatment of GBMs is the blood–brain barrier, but recent studies have shown that it does not always function in cases involving recurrent GBMs. We previously showed the feasibility of vaccination with PPV for advanced GBM patients in a phase I study [32]. Twenty-one patients received more than six vaccinations, and clinical responses were five cases of partial response (PR), eight of stable disease (SD), and eight of PD with MST of 20.7 months in this study. More importantly, significant levels of peptide-specific IgG were detected in the post-vaccination tumor cavity or spinal fluid of all of the tested patients who showed favorable clinical responses. Another clinical study showed the safety and increased immune boosting with potential clinical benefits in cases of recurrent or progressive GBM, even in temozolomide refractory settings [40]. On the basis of these promising results, double-blind randomized phase III trials are currently underway in GBM patients.

Colorectal and gastric cancer

We reported previously that SART3 is expressed in the majority of colorectal cancers and that two to three SART3-derived peptides are present in the majority of cancer patients with HLA-A24+ and HLA-A2+ [8–10, 14]. In a phase I clinical trial of PPV on 10 patients with advanced colorectal cancer, we observed one PR and one SD continuing for more than 6 months [53]. These PR and SD cases were vaccinated with three kinds of SART3- and p56lck-derived peptides, suggesting that the combination of these peptides might constitute a promising vaccine strategy for advanced colorectal carcinomas. In addition, a phase I/II clinical trial of PPV in combination with oral administration of a 5-fluorouracil derivative (TS-1) in advanced gastric or colorectal cancer patients indicated that administration of the standard dose of TS-1 in combination with PPV does not necessarily impede immunological responses in these cancer patients, and actually maintains or augments them [54]. Another phase I clinical trial of PPV in 13 patients with advanced gastric cancer demonstrated prolonged survival and cellular and humoral immune responses to the vaccinated peptides in the post-vaccination samples, including those of all four patients with the scirrhous type [57]. Even though only a small number of selected patients were treated, the encouraging clinical response warrants further studies of PPV in colorectal and gastric cancers.

Pancreatic cancer

For patients with advanced pancreatic cancer (APC), the treatment options are limited, although gemcitabine (GEM) is currently used as the standard therapy. We have conducted a phase I trial of PPV in 13 HLA-A24- or HLA-A2-positive patients with APC, in which patients were treated by PPV at three different dose settings of 1, 2, and 3 mg/peptide with GEM [55]. This combination therapy was well tolerated, and 11 of 13 patients (85 %) showed clinical responses, such as reduction in tumor size and/or the level of tumor markers. Augmentation of peptide-specific CTL activity against pancreatic cancer cells was observed at each dose level, and the increment of peptide-specific IgG antibodies was dependent on peptide dose. These results suggested that GEM did not inhibit the immune responses induced by PPV. Subsequently, we have evaluated the safety, clinical efficacy, and immune response to PPV with GEM as the first therapy in 21 patients with APC [56]. This phase II study showed a longer survival (MST of 9 months with a 1-year survival rate of 38 %) than in previously reported results of GEM alone (MST of 5.7 months with a 1-year survival rate of 18 %). Importantly, MST was 15 months in the patients who showed immunological responses to vaccinated peptides in the early stages of vaccination. In view of these findings, the survival benefit in comparison with GEM alone needs to be confirmed in future clinical studies.

Lung cancer

The prognosis of advanced lung cancer patients remains very poor with a median survival time of around 6–10 months. Phase I and II studies of PPV in a small number of patients with refractory non-small cell lung cancer (NSLC) showed longer survival (MST of 10.1–15.2 months) [37, 59] than in previous reports. A clinical study of advanced small cell lung cancer (SCLC) showed the feasibility of PPV since there were higher rates of peptide-specific immunological boosting after PPV [58]. In order to identify potential biomarkers for predicting overall survival in advanced lung cancer patients, we retrospectively analyzed pre-vaccination clinical findings and laboratory data. In patients with refractory NSLC, a higher C-reactive protein (CRP) level before vaccination and a low frequency of CD3+CD26+ cells after vaccination were significant predictors of unfavorable overall survival [59]. In patients with refractory SCLC, the number of previous chemotherapy treatments and the frequency of CD3+CD26+ cells in PBMCs before vaccination were potential prognostic predictors in patients who received PPV [58]. These findings demonstrate that less inflammation may contribute to better responses to the PPV, suggesting that evaluation of the inflammatory factors before vaccination could be useful for selecting appropriate cancer patients for PPV.

Other cancers

We have also conducted phase I clinical trials for other advanced cancers including metastatic renal cell carcinoma (RCC), malignant melanoma, gynecologic cancers, and bladder cancer [38, 60–62]. All of these studies demonstrated that PPV was safe and well tolerated with no major adverse effects and that more immune responses were observed in the majority of patients after PPV than with the pre-designated peptide vaccination. Some patients treated by PPV showed objective clinical responses evaluated by the response evaluation criteria in solid tumors criteria with boosted immune responses: CR in one patient with chemotherapy-resistant advanced bladder tumor and PR in two patients with cervical cancer [38, 62]. These results indicate that PPV can be applied in further clinical trials aimed at the treatment for these cancers.

Conclusions

The field of immunotherapy has advanced dramatically during the past 20 years, but there have remained several issues to be addressed in order to achieve successful cancer vaccine development. In view of the complexity and diversity of the immunological characteristics of tumors and the immune cell repertoires of hosts, selection of suitable peptide vaccines for individual patients based on the preexisting host immunity before vaccination could induce potent anti-tumor responses that provide clinical benefit to cancer patients. We have shown promising results of PPV in this review article as a new treatment modality for patients with various types of advanced cancer. Further randomized phase III clinical trials are essential to prove the clinical benefits of PPV. In addition, novel biomarkers for selecting patients who would benefit most from PPV remain to be identified.