Cancer Vaccines: Quo Vadis?
Contrasting results from cancer vaccine studies versus checkpoint blockade or ACT emphasized two major hurdles upstream and downstream of vaccination: the limited competence of the immune repertoire and the altered functionality of immune effector cells within the tumor microenvironment (Figure 1A). Thus, selecting clinical indications where there is a functional preexisting repertoire (either naïve or primed) and which are devoid of organized, vascular tumors with their plethora of immune-inhibiting molecules would increase the vaccines' likelihood of success (Figure 1B).
Cancer Vaccines in Minimal Residual Disease to Clear Residual Cancer & Prevent Tumor Relapse
Several groups are pursuing this concept utilizing various platform technologies. For example, van Tendeloo et al. advanced a vaccine for acute myelogenous leukemia patients previously treated with chemotherapy utilizing matured DCs transfected with mRNA that expresses WT-1 antigen.[79] Repeat immunization resulted in conversion from a partial response diagnosed as Wilm's tumor suppressor gene 1 (WT-1)-positive MRD to a complete response manifested through a WT-1-negative status. The durability of the clinical response was associated with an increase in WT-1-specific CD8+ T-cell immunity and NK activation. Clearance of residual leukemia through vaccination against WT-1 showcased this antigen as a target cancer vaccination in general.[80] Further, monitoring individual patients by using markers of residual disease could revolutionize this area of investigation by providing fairly rapid proof of principle in small-sized trials.[81]
Another study targeting residual disease was conducted in chronic myelogenous leukemia (CML) utilizing the allogeneic GVAX platform which employs a K562 leukemia cell line expressing GM-CSF.[82] CML patients on chronic treatment with the tyrosine kinase inhibitor imatinib mesylate (Gleevec®), responding to this drug, were vaccinated and monitored using the bcr/abl marker. Interestingly, about one-third of the 19 treated patients demonstrated disease clearance, illustrating the power of using potent vaccination during the minimal residual disease stage, together with molecular marker monitoring.
Another study utilized vaccination with GM-CSF and peptides spanning the bcr/abl breakpoint mixed with montanide adjuvant.[83] Interestingly, within a group of 10 CML patients who achieved stable disease after treatment with tyrosine kinase inhibitor, 50% of these witnessed either a substantial disease reduction or a major molecular response assessed by measuring the presence of the bcr/abl transcript.
Another promising disease indication is that of follicular lymphoma in remission where an individualized idiotype vaccine, based on keyhole limpet hemocyanin-coupled idiotype protein adjuvanted with GM-CSF, showed compelling evidence of clinical benefit in a Phase III randomized trial.[84] Results of this study demonstrated a substantial increase in the disease-free survival of vaccinated patients leading to some renewed efforts aimed at designing personalized vaccines.
Efforts to evaluate cancer vaccines in an adjuvant setting in solid tumors have led to mixed results. In particular, it has been difficult to accurately monitor disease burden, therefore precluding a more objective and rapid assessment of the clinical effect in small trials. Alfaro et al. outlined the importance of pursuing improvements of the vaccine platform technology in conjunction with utilizing methods to objectively assess clinical outcome on a patient-by-patient basis by quantifying the impact on both circulating tumor cells and circulating endothelial cells.[85]
Evaluation of vaccines in melanoma suggested a preferential applicability to lower disease burden. In fact, an earlier report in a sizable randomized trial, in resected advanced stage melanoma patients treated with an allogeneic cell-based vaccine (Canvaxin®), indicated a significantly improved overall survival of 5 years in vaccinated versus unvaccinated patients.[86] Nevertheless, the substantial variability in terms of vaccine characteristics and product potency (batch variation in the case of allogeneic vaccines, and patient-to-patient variability in the case of autologous cell-based vaccines) represented a significant barrier in front of vaccine development. Hence, there has been a keen interest in optimizing synthetic vaccines and exploring them in an MRD setting in solid tumors. Thus, an earlier study utilizing a recombinant NY-ESO-1 protein, combined with a saponin-based adjuvant, was conducted in melanoma patients also with resected tumors.[87] The results showed, in addition to induction of antibody and T-cell immune responses, an imbalance in terms of disease relapse. More specifically, patients assigned to the adjuvanted protein group had a several-fold lower likelihood of relapse (2 of 19 fully vaccinated patients relapsed during 2 years follow up vs 5 of 7 placebo and 9 of 16 patients vaccinated with non-adjuvanted protein). Similarly, a vaccine comprising an EGFRvIII-derived, 13-amino acid peptide that spans this specific neoepitope, conjugated to keyhole limpet hemocyanin, showed promising clinical efficacy in first-line glioblastoma, where most of the tumor bulk was previously eliminated by surgery and chemoradiation.[88] In this Phase II multicenter study, survival of vaccinated patients was directly compared to historical survival data from matched controls. In addition, tumor relapses were found to be EGFRvIII-negative, thus providing proof for the vaccine's effect in this setting.
As prevention of cancer relapse by vaccination of patients with solid tumors in complete remission is an extremely appealing opportunity, NeuVax™ is one of the most promising vaccines in Phase III clinical development. It comprises a Her-2/Neu-derived nonapeptide that stimulates CD8+ T cells in HLA-A2/A3+ patients.[89] In a previous Phase II trial, this peptide vaccine administered to node-positive patients after standard of care treatment achieved a very significant reduction of recurrence rate at 60 months of 5.6 versus 25.9% in the control arm.[201] Based on these results, the current Phase III trial seeks out to confirm the vaccine's capability to prevent recurrences in early stage node-positive breast cancer with low-to-intermediate Her-2/Neu expression, after successful utilization of standard of care leading to clinical remission.
In summary, there are compelling scientific arguments in support of MRD as a preferred indication for cancer vaccines. While there are notable examples of promising cancer vaccine trials in this indication, much more needs to be done in terms of fully tapping into the potential of cancer vaccines to clear residual disease and prevent tumor relapse.
Some Indications Associated With Measurable Tumors can Present Opportunities for Therapeutic Vaccination
Interestingly, more advanced disease indications could also represent opportunities for therapeutic vaccination, if the immune repertoire is competent and the immune environment within tumors is permissive. In particular, in situ carcinoma represents an exciting opportunity to test cancer vaccines at the interface between MRD and local progressive disease. An encouraging approach consists of adjuvanted long peptides corresponding to E6 and E7 antigens of HPV, utilized as therapeutic vaccines for cervical carcinoma with lesions confined to the epithelial layer.[38,90] This program elegantly integrates several parameters that maximize the likelihood of success for therapeutic vaccination and could represent a roadmap for adequately positioning this concept within the therapeutic armamentarium against cancer. First, a non-self-target antigen is more immunogenic and second, the use of adjuvanted peptides of optimal size results in coinduction of CD4+ and CD8+ T cell immunity through cross-presentation. Third, the synthetic nature of the vaccine results in a reduced cost of goods. Forth, an early stage yet measurable disease setting, while permissive to the activity of the T cells within lesions, is amenable to fairly rapid and objective evaluation of clinical response in individual patients. In addition, the HPV oncoprotein targets are intimately associated with the biology of this carcinoma. While a majority of patients in early stage (vulvar in situ carcinoma) showed objective and durable responses following vaccination, patients with later disease stage were fairly refractory. In addition to the association between clinical response and disease stage, there was a notable correlation between the type 1 T-cell response elicited by the vaccine and the clinical outcome.
Advanced Disease Indications Are a Difficult Target for Cancer Vaccines
Apart from early stage disease, reports of objective clinical responses measured in individual patients, have been exceedingly rare for cancer vaccines. A recent study exploring an adjuvanted NY-ESO-1 recombinant protein in melanoma showed that advanced disease patients, with metastases to internal organs, have a diminished T-cell immune response, while the antibody response was comparable to that of patients with MRD.[91] There was also a considerable increase in the percentage of Treg cells in advanced disease patients, directly highlighting one of the major hurdles.
In a Phase I trial that evaluated the outcome of an intra-lymph node prime-boost vaccine in patients with late stage metastatic melanoma with visceral lesions (stage IVc) and earlier stage with cutaneous disease and lymph node mets (stages IIIc/IVa), there was a stark discrepancy uncovered. While roughly 50% (4 of 7) of the patients with lymph node metastases showed an objective tumor reduction that qualified as a partial response under RECIST criteria, none of the 14 patients with visceral metastatic disease showed a clinical response.[92] In addition, while the immune response in both groups was similar, only patients with disease confined to lymph nodes and preexisting immunity against one of the immunizing antigens (Melan A/MART-1) showed tumor reduction.
This illustrates two important aspects. First, in the measurable disease setting, therapeutic vaccination could work through mobilizing preexisting tumor-specific T cells. Second, these findings suggest that the resulting effector T cells operate more effectively within the lymph node environment as opposed to visceral metastatic lesions where they presumably encounter a wider range of inhibitory mechanisms. This is supported by a body of evidence showing considerable variability in the tumor microenvironment. This ranges from a noninflammatory to proinflammatory microenvironment that could be permissive or inhibitory toward antitumor immune responses. Such findings also support current efforts to define tumor gene expression signatures that stratify patient populations in 'responsive' versus 'non-responsive' to active immunotherapy.[93]
Stimuvax®, a vaccine comprising liposomal-formulated Muc peptide, also illustrates the difficulty of addressing later stage, unresectable cancer. A recently completed Phase III trial in stage III non-small-cell lung carcinoma showed no survival advantage in vaccinated patients.[94] However, interestingly, a post hoc analysis showed that patients who received concurrent radiochemotherapy at the start of the vaccination regimen, showed a 10-month survival advantage over unvaccinated patients (with a medium overall survival of about 20 months). This speaks to a rational integration of vaccines with certain standard of care, but much more needs to be done to garner appropriate information to rationally design prospective studies.
Finally, an approach that resulted in objective tumor reduction in an advanced disease setting consisted of combining vaccination with cytokines such as IL-2 and IFN-α. Higher rates of objective response were seen in vaccinated patients with melanoma or RCC who received concurrent cytokine treatment, compared to patients who were being treated with cytokines alone.[95,96] While such studies are exciting, the results need to be confirmed in larger and adequately controlled trials that are quite difficult to conduct, due to the niche nature of the clinical indications and the toxicities associated with cytokine therapy.
In summary, while advanced cancer represents a very difficult indication for cancer vaccines,[97] there could be few opportunities where the immune repertoire is sufficiently preserved and metastatic lesions are still permissive for an immune mediated attack. Vaccine utilization in advanced cancer will require adequate patient stratification methods as only a minor subset of patients could be amenable to such therapeutic modality.
Integrative Immune Interventions That Leverage Vaccination
Beyond MRD or select niches consisting of measurable disease but associated with immune responsiveness, a second category of opportunities exists. This consists of vaccines adjunctive to interventions comprising restoration, amplification or engineering of the immune repertoire. In principle, while the exogenous provision of a T-cell repertoire would yield competent immune cells, vaccination would turn on, amplify or maintain their activity in vivo (Figure 1C). This could offer two advantages: a more specific post-adoptive transfer manipulation of T cells as compared to utilization of high dose cytokines and, second, a prolonged in vivo activity of the transferred T cells that face multiple negative homeostatic mechanisms.
The next objectives in the quest to treat cancer are the increase in the response rate and the durability of clinical response. There are few instances where a durable clinical response on ACT has been observed. One example is anti-leukemic CAR-T cells supra-physiologically engineered with potent costimulatory domains and directed against a renewable source of endogenous antigen such as CD19[98,99] that acts similar to an auto-vaccine.[100] The other situation where very durable clinical responses were reported in context of ACT involved TIL treatment of melanoma patients.[101] In this case, it was not clear whether TILs were capable of eliminating the 'last cancer cell' or if there was a persisting, immune-mediated containment of the disease.
The concept of integrating vaccination with ACT is anchored in earlier findings from preclinical and clinical studies.[102,103] These studies showed that the process of T-cell repertoire recovery, after chemotherapy and bone marrow transplantation, permits repertoire manipulation through vaccination. This concept yielded some encouraging results in several multiple myeloma patients immunized with idiotypic antigen.[104] Subsequently, this approach has been tested and refined in a wider range of clinical protocols some involving donor lymphocyte infusion. Other protocols explored administration of autologous peripheral T cells harvested before chemotherapy and bone marrow transplantation, in combination with subsequent vaccination during the T-cell repertoire recovery phase.[105–107] This integrated approach to immunotherapy – based on the key opportunity to manipulate a recovering T cell repertoire – was articulated in a recent review.[108]
A seminal paper was published in 2003 illustrating for the first time in a preclinical mouse model involving a TCR against the melanoma antigen gp100, the profound synergy between adoptive T-cell transfer and a gp100 vaccine. The vaccination, encompassing a pox virus expressing a gp100 peptide agonist, combined with cytokine treatment, resulted in an impressive regression of established B16 tumors without the requirement for lymphoablative conditioning.[109] Subsequently, it was shown that adoptively transferred T cells stimulated by the cognate pox virus vaccine in vivo have the capability of migrating to various tissues and non-antigen-expressing tumors but display full blown effector functions only within antigen-expressing tumors.[110] The specific T cells acquired a capability to proliferate at a much faster rate than the tumor cells.[111] Further, it was shown that a pox virus-based vaccine could be interchanged with a DC vaccine, work that also uncovered the seminal role of IL-7 during the T-cell response recovery phase.[112] Subsequently, it was also demonstrated in the same pMel preclinical model that T cells could be antigen 'primed' just before adoptive transfer, resulting into an outcome similar to the utilization of vaccination in vivo.[113] Nevertheless, it is likely that maintenance and manipulation of the T-cell repertoire over a longer interval would require additional in vivo interventions post-ACT, with vaccination possibly offering a more targeted and safer approach in contrast to high-dose cytokine treatment.
The proof-of-concept of vaccination as adjunct to ACT was also explored in other preclinical models. For example, vaccination with an adenovirus expressing the antigen 5T4 and autologous DCs, followed by local (intratumoral or peritumoral) infusion of CAR-engineered T cells against 5T4, resulted in tumor control as long as all these components of the therapy were provided.[114] DC vaccines were evaluated in combination with ACT, in models of glioblastoma and melanoma, with the specific T-cell population visualized in vivo by micro-PET analysis.[115,116] Other research outlined the importance of endogenous DCs during the recovery phase post-lymphodepletion in context of adoptive T-cell transfer, advancing the idea that the T-cell-infused recipient would be more receptive to vaccination during that critical interval.[117,118] Interestingly, several lines of evidence shed light on the importance of antigen presentation to the infused T cells, necessary for the activity of the latter. For example, in the anti-CD19 CAR T-cell model, it has been shown that continuous exposure to CD19 borne by the B-cell lineage under continuous renewal process is important in maintaining the activity of the CD8+ T cells directed against CD19.[100] Second, it was recently shown that tumor stromal cells acting as very effective professional APC cross-present tumor-derived antigen toward activating the adoptively transferred T cells to induce and support an overall antitumor effect.[119] Finally, in a model involving IL-12 co-engineered T cells adoptively transferred into mice carrying solid tumors, evidence pointed out to a role of tumor stromal macrophages, myeloid derived suppressor cells and DCs activated by local IL-12 to express Fas, in stimulating the incoming CD8+ FasL+ T cells.[120] Altogether, this evidence indicates that the adoptively transferred T cells benefit from 'auto-vaccination' by virtue of engaging resident APCs and antigen, but it is not clear to what extent this could be optimized through provision of exogenous vaccine. Reprioritization of immune interventions and aspects of the concept of combinatorial immunotherapy were critically illustrated in several reviews.[12,121–125]
Surprisingly, despite initial encouraging results in preclinical models, clinical translation of the concept of integrating ACT with vaccination yielded mixed results, thus far. A small yet randomized trial in transplanted multiple myeloma indicated that T-cell repertoire restoration with peripheral autologous T cells from patients immunized with a flu vaccine resulted in more rapid recovery of immune competency.[126] A study that pioneered this concept using a tumor-associated antigen tested peripheral T cells from melanoma patients immunized against gp100 that were expanded in vitro, followed by adoptive transfer, vaccination and IL-2 treatment.[127] Disappointingly, this trial resulted in an absence of antitumoral responses and only two instances of autoimmunity questioning the relevance of the target antigen.[127] A subsequent report on a single melanoma patient treated with TILs expanded in vitro against the same gp100 antigen and followed by fowlpox vaccination post-adoptive T-cell transfer resulted in a major clinical response.[128] The significance of this observation could be high as the same patient was previously treated unsuccessfully with TILs without vaccination.
Next, an effort to integrate vaccination against human telomerase reverse transcriptase and survivin in context of bone marrow transplantation and adoptive T-cell transfer with polyclonally stimulated autologous T cells showed that only 30% of the multiple myeloma patients treated this way developed an immune response while there was no apparent clinical improvement.[129] A variant of this approach was later tested in ovarian carcinoma patients. Instead of utilizing epitope-specific vaccines, the authors used a tumor lysate-based DC vaccine given before harvesting T cells for ex vivo polyclonal expansion and also administered after chemotherapy and T-cell infusion.[130] While the trial size was quite limited, a few patients showed some exciting signals in terms of immune response and clinical signals and there was one case associated with a durable response.
In summary, despite the excitement and support provided by preclinical evidence, the translation of the concept of integrating vaccination with ACT in humans awaits solid proof of concept. It is not clear whether integrating the two modalities could overcome the homeostatic mechanisms that limit the activity of activated T cells in vivo.
The Next Frontier: Optimization of Integrative Immune Interventions
While vaccination could help tremendously in terms of generation, amplification and/or maintenance of adoptively transferred T cells, there still seem to be substantial downstream hurdles as T cells are still facing a relatively hostile immunologic tumor microenvironment. This is reflected in the results of a preclinical study using a different melanoma antigen Trp-2, which showed that adoptive T-cell transfer combined with potent vaccination led to the accumulation of antitumor T cells within melanoma lesions but without a notable effect on the tumor.[131] In addition, extensive analysis of peripheral T cells from melanoma patients adoptively transferred with TCR-engineered T cells and followed by DC vaccination with cognate antigen, showed that the polyfunctionality of T cells was gradually lost and replaced by a regulatory-like profile in the residual antigen-specific T cells.[75,132] A more recent report showed that TCR-engineered T cells adoptively transferred to melanoma patients acquire an exhausted phenotype encompassing elevated PD-1 and CD160 expression[76] raising the question whether integration of a vaccine within this treatment protocol could counteract this phenomenon.
Altogether, these findings suggest a quite noxious impact of host-related factors, on the adoptively transferred T cells. In a mouse model, vaccination with a long polypeptide encompassing non-self antigen (human gp100), together with the potent CpG adjuvant, triggered a massive expansion of specific T cells and pronounced antitumor effects, even in the absence of a preparative regimen.[133] Another recent preclinical study showed that PD-1 blockade in conjunction with ACT, subsequent vaccination and IL-2 treatment via IL-2/anti-IL-2 antibody complexes resulted in substantial tumor regression at a much higher rate when compared to adoptive therapy or vaccination separately.[134] Finally, in a preclinical model of autovaccination by intratumoral administration of immune-stimulating CpG motifs, the authors showed that co-depletion of Treg cells by anti-CTLA4 + anti-OX40 antibodies within one tumor lesion, vastly increased the immune system's impact on injected as well as remote tumors.[135] These findings suggest that endogenously primed or adoptively transferred T cells could greatly benefit from approaches to lower rather than completely obliterate parts of the immune system such as the Treg population.
While these findings create the perspective of rationally integrating vaccination with ACT, they also suggest that one needs to overcome the inhibitory mechanisms that inherently limit the activity of T cells to fully unleash the potential of immunotherapy.
Expert Rev Vaccines. 2013;12(10):1219-1234. © 2013 Expert Reviews Ltd.
