SEARCH

SEARCH BY CITATION

Keywords:

  • Cancer immunotherapy;
  • Cancer vaccine;
  • DC;
  • Mutated antigens;
  • Oncoantigens

Abstract

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

DC initiate and regulate T-cell immunity and are thus the key to optimization of all types of vaccines. Insights into DC biology offer many opportunities to enhance immunogenicity. In this Viewpoint, I discuss some recent developments and findings that are of immediate relevance for the clinical development of cancer vaccines. In addition, I emphasize my personal view that we should explore the potential of adoptively transferred DC (i.e. DC vaccination) as cancer vaccines by performing two-armed trials that address critical variables and by delivering antigens via mRNA-transfected DC.

In the past decade, new developments in cancer treatment have been dominated by targeted therapies using kinase inhibitors and monoclonal antibodies, which have become part of clinical routine to treat hematological as well as solid tumors. In contrast, cancer vaccines, which are active immunization approaches to induce tumor-specific T cells in patients, i.e. harnessing the power of the immune system against cancer, have proven more difficult to develop, although T cells are clearly a unique and effective means of attacking tumor cells and regressing tumors. Given the apparent success of other targeted therapies, some have questioned whether it makes sense to invest in cancer vaccines. This view is about to change as indicated by the increasing interest of large pharmaceutical companies such as GlaxoSmithKline to develop cancer vaccines. In addition, Dendreon's Provenge™ vaccine has scored positive in phase III trials, further suggesting that cancer vaccines are valid therapeutic approaches. The approval of Provenge™ by the FDA on April 29th, 2010, for the treatment of asymptomatic or minimally symptomatic, hormone-resistant metastatic prostate cancer heralds a new exciting era.

Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

T cells are a suitable tool to successfully attack tumors in vivo in patients 1 as exemplified by the graft-versus-leukemia effect in bone marrow transplantation, and the success of adoptive T-cell therapy in treating metastatic melanoma, a solid tumor. The latter approach requires not only large numbers of long-lived high-quality CTL but also preconditioning of the host by non-myeloablative lymphodepletion. It is, therefore, not surprising that the moderate induction of tumor-specific T cells by current cancer vaccines is usually not sufficient for inducing regressions.

A roadmap to effective cancer vaccination is, however, emerging. Current vaccination strategies must be improved to achieve higher T-cell frequencies and most importantly, the quality of these T cells must be comparable to the protective T-cell response observed during acute or chronic viral infections. This is expected to enhance clinical efficacy, as already small numbers of high-quality vaccine T cells appear to be able to induce regressions in a minority of patients by inducing a second wave of T cells (the so-called “spark” hypothesis) 2, 3. In addition, somatically mutated antigens, which are associated with regression and long-term survival 3, 4, should be tested, as well as antigens relevant for the oncogenic phenotype (mutated and viral oncogenes, certain non-mutated antigens that tumors over-express) to diminish antigen loss and escape 5, 6. Side effects observed recently with adoptive T-cell therapy 7 suggest that tumor-specific antigens (such as cancer testis or mutated antigens, or Muc-1) 5, 6, 8 should be prioritized to avoid similar toxicity with highly immunogenic cancer vaccines of the future. In addition, it appears mandatory to block some of the immunosuppressive circuits and to enhance migration of T cells into tumor sites in order to make cancer vaccines more clinically effective 9. Identification of patients who can respond to vaccines is also very important, although this requires reliable biomarkers yet to be identified. Currently, tumor burden is considered an important response marker and it is expected that in the setting of minimal residual disease, optimized vaccines might even be clinically effective alone.

Tumors silence DC and exploit their tolerogenic function

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

T cells are the natural way to attack cells harboring non-self proteins as exemplified by the elimination of virally infected cells by virus-specific CTL. Tumor cells also express mutated and thus foreign proteins, and if exposed to immune pressure, they also tend to escape immune control. In contrast to viruses, which deliver strong “danger” signals resulting in DC maturation, naturally growing tumors do not, so that the cross-presentation of tumor antigens by DC exposed to endogenous maturation signals is unlikely to result in vigorous activation and expansion of high-quality T cells, even if the tumor does not block DC migration 10. As soon as the tumor has induced – to a large extent via STAT-3 activation – an immunosuppressive microenvironment (containing abnormal macrophages, myeloid-derived suppressor cells, and Treg), the situation is exacerbated 9. Once this immunosuppressive environment is established, tumor-specific effector T cells can no longer readily enter the tumor site (also because of the abnormal tumor vasculature and expression of chemokines that preferentially attract Treg), and those T cells that do migrate into the tumor are suppressed. DC at the tumor site as well as the draining lymph nodes become increasingly suppressed and contribute to T-cell anergy/deletion or to T-cell suppression via Treg.

Tumor vaccines must exploit the immunogenic function of DC

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

Cancer vaccines must induce high-quality effector T cells, which as CTL or Th cells can eliminate tumor cells (including tumor-initiating cells) and/or produce advantageous cytokines. It is, however, also important to generate long-lived tumor-specific memory T cells in order to maintain anti-tumor responses as well as to prevent relapse, or as in the case of prophylactic vaccines, avoid disease.

DC control T cells and guide their differentiation 11. The DC system is astonishingly complex and composed of different subsets with considerable plasticity 12. It is important to note that tolerance induction is the major function of DC in the steady state as also outlined in the accompanying article by Maria Rescigno 13. Only presentation of antigen by sufficiently matured DC expressing costimulatory molecules (such as membrane-bound CD86 and CD70 and/or soluble IL-12) will result in robust T-cell proliferation and differentiation, and thus effective T-cell immunity. DC maturation is therefore a complex program; depending on both the DC subset and the type of maturation stimulus, variable set of genes and pathways are activated, leading to diverse differentiation programs in the stimulated T cells.

DC are, therefore, key to successful vaccination 14, 15. An optimal vaccine requires both antigen-targeting to DC and adequate DC maturation by adjuvants to induce a maturation program that is suitable for T-cell immunity. These crucial aspects of DC biology have until recently been largely underestimated, while the choice/format of antigen and vaccination schedule were prioritized.

The three ways of exploiting DC as vectors for vaccination (Fig. 1)

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

In vivo DC targeting” is considered by many as the vaccine strategy of the future and will be tested in clinical trials in the coming years. In mouse models, injection of anti-DC (e.g. anti C-type lectin) antibodies fused to antigens induces tolerance by several mechanisms, but vigorous immunity can also be induced if a suitable maturation stimulus is co-delivered 16, 17. Our knowledge on the type of stimulus required in such a setting for optimal immunogenicity is expanding 18. Differential expression of receptors might allow targeting of select DC subsets and “tailor-made” immunization 19, 20. Effective immunization was also observed following delivery of TLR ligand–antigen conjugates, which contain the maturation stimulus but are not as precise in cell targeting 21. Although the in situ targeting strategy is clearly a promising approach, one caveat is that in diseased patients, DC subsets and precursors as well as their receptor expression may be variable and even abnormal so that translation into the clinic might hold unexpected surprises. Studies on DC subsets in blood and organs in man are vital and likely to be demanding.

thumbnail image

Figure 1. Phenotype and function of mouse skin DC subsets. Based on the expression of CD207, CD11b and of CD103, five distinct DC subsets have been identified in mouse steady-state dermis. Due to the fact that these subsets are less well defined and less extensively studied than the other subsets, the CD207+CD103 and CD207CD11b DDC have not been represented. CLN contain a migratory counterpart of each of the DC subsets identified in the skin. Upon skin inflammation, monocytes are rapidly recruited and convert into inf-DC. The functional specialization existing among skin DC subsets is specified in the boxes. All skin DC subsets are capable of inducing delayed type hypersensitivity reactions provided that they have been exposed to the inducing agent 1, 5, 56. In addition, CLN contain LT-DC that correspond to the CD8α+, the CD11b+ and the pDC subsets.

Download figure to PowerPoint

“Conventional vaccination” strategies in contrast do not allow precise targeting. Many strategies have been attempted to make such vaccines nevertheless work, for example, through the use of antigen in various formats in conjunction with novel adjuvants. Recombinant vectors (which may preferentially reach DC) and prime-boost regimens are also explored as approaches to optimization (for review see 22). I would like to address a few items that are of general relevance:

  • (i)
    A recent alarming finding is that patients treated with either Canvaxin™ (made up of three melanoma cell line lysates+BCG) or a GM-2+QS21 vaccine (composed of a ganglioside antigen+saponin as adjuvant) experienced worse clinical outcome than the control arm in large phase III melanoma trials 23. Immunomonitoring data are currently not available in order to rationalize the negative results (e.g. tolerance induction because of insufficient DC targeting and suboptimal maturation?). These results, however, caution that in case of vaccines, clinically promising phase II data (particularly if based on comparison with historical controls) have to be supplemented by solid immunomonitoring and demonstration of T-cell (not only B-cell) immunogenicity before going on to phase III studies, in order to avoid exacerbating patients' disease.
  • (ii)
    The importance of using an adequate adjuvant to mature DC is dramatically highlighted by a study in which Mage-A3 protein booster vaccination of cancer patients revealed long-term memory if priming occurred with adjuvant, but led to tolerance upon priming with antigen alone 24.
  • (iii)
    Short peptides, usually administered in IFA (incomplete Freund's adjuvant), have often been used in cancer vaccines, with some peptides resulting in significant T-cell responses. The MelanA peptide EAAGIGILTV is particularly potent, and upon delivery in IFA+QS21 (=saponin)+CpG Oligo (=TLR9 ligand), yielded significant CTL and even Th responses, albeit without clinical responses 25, 26. The immunogenicity of injected short peptides that fit exactly into MHC peptide grooves is, however, complex and variable as elucidated over the years by C. Melief's group. The minimal CTL epitope E1A (derived from human adenovirus type 5 E1A), for example, induces tolerance if given s.c. in PBS or IFA in mice. Co-delivery of a DC maturation stimulus (PolyIC or agonistic anti-CD40 antibody) resulted in CTL activation 27 as does injection of peptide-loaded DC. The problem is that some peptides, depending on their pharmacokinetics, spread through the body and are presented by tolerizing non-professional antigen-presenting cells. To avoid this complication, Melief has systematically developed the concept of using synthetic long peptides (SLP) of about 30 amino acids in length for immunization. These peptides are taken up solely by DC, which process the SLP for prolonged presentation by the various class I and II molecules to induce CD4+ as well as CD8+ T cells 28, 29, while proteins are not efficiently cross-presented and, therefore, induce primarily CD4+ T cells. The SLP concept has already convincingly passed the “in vivo veritas” test in mice and recently also in humans as shown by immunization with overlapping SLP covering the whole sequence of HPV E6 and E7 oncoproteins, which resulted in regression of HPV-induced high-grade cervical intraepithelial neoplasias 30. Regressions were slow, correlated well with the induced HPV-specific immune responses, but occurred only in the smaller (and thus probably newer) lesions. This once more illustrates that significant induction of even high-quality T cells without combination therapy is not automatically leading to regressions of larger lesions, and vaccination is logically perhaps best suited for prevention of metastasis.

DC vaccination,” i.e. active immunization by adoptive transfer of DC, either enriched/isolated from peripheral blood or generated ex vivo from (CD34+ or more often CD14+) precursors offers the possibility to monitor and address variables crucial for an optimized T-cell immunogenicity, notably aspects of DC biology most relevant for immunogenicity (e.g. DC type, maturation status, migratory capacity, and antigen loading) as well as important vaccination logistics such as DC number, route, and frequency of vaccination.

Adoptive transfer of antigen-loaded DC into patients (DC vaccination)

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

The first DC vaccination study was published in 1996 and used rare DC isolated directly ex vivo from peripheral blood to immunize B-cell lymphoma patients against their tumor-specific idiotype protein 31. By March 2010, almost 300 papers have been published reporting on 4422 patients treated (not all under Good Clinical Practice (GCP) conditions: primarily melanoma 32, 33 and prostate cancer 34 patients – 1301 and 510 patients, respectively). Most trials employed monocyte-derived DC (MoDC), which were most often generated from monocytes by culture in GM-CSF+IL-4 over approximately 6 days to obtain immature DC, followed by exposure to monocyte-conditioned medium or its mimic (cocktail composed of TNF-α+IL-1β+IL-6+PGE2) for 1–2 days, to yield “cocktail”-matured DC. Short-term culture methods have also been described 35, but appears to be more variable in inducing stably differentiated DC and probably for this reason, have not been explored extensively in trials. CD34+ cell-derived DC contain a mixture of interstitial DC and DC of the Langerhans cell type, the latter being more potent than the standard MoDC in CTL priming in vitro. Only select initial trials used DC derived from CD34+ cells, perhaps a result of a more difficult production process 36. All the various trials are difficult to compare due to a range of differences, but the immunogenicity of mature DC was obvious, and hints for clinical efficacy have been observed.

Phase III trial in prostate cancer with DC-enriched antigen-presenting cells

The first vaccine with proven clinical benefit (Dendreon's Sipuleucel-T, Provenge™) is, however, not based on highly enriched ex vivo isolated or in vitro generated DC, but is a cellular vaccine prepared by isolating via gradient centrifugation a DC-precursor-enriched fraction from apheresis products, which is exposed for 2 days to a fusion protein consisting of GM-CSF and prostatic acid phosphatase antigen. This results in targeting to the GM-CSF receptor-expressing cells, including DC precursors, which then undergo activation/maturation in vitro (leading to CD54 upregulation which serves as a potency marker 37).

Dendreon has recently confirmed in its pivotal phase III study (IMPACT trial, n=512) that its first-generation cellular vaccine product Sipuleucel-T (Provenge™) significantly improved overall survival (by a median of 4.1 months, reducing the risk of death by 22.5% compared to the control group) in asymptomatic or minimally symptomatic metastatic, hormone-refractory prostate cancer even though classical regressions did not occur and time to progression was not prolonged 38. The result of the IMPACT trial led to the approval by the FDA of the first therapeutic cancer vaccine ever on April 29th, 2010. The positive outcome has already fostered interest in the development of cancer vaccines in general and DC in particular. The Dendreon story is interesting in several aspects. It reflects, for example, that in the cancer research community until recently the classical acute response criteria developed for chemotherapy were considered indispensable in judging vaccine effectiveness, even though it had been pointed out early in the 1990s that vaccines require time but not necessarily classical regressions to produce clinical benefit 39. In 2007, the Cancer Vaccine Clinical Trial Working Group came up with an important position paper proposing a new clinical development paradigm for cancer vaccines 40, and phase II trials employing anti-CTLA-4 antibodies also unraveled distinct response patterns associated with favorable survival 41. The Sipuleucel-T product development – a nerve-wracking roller coaster – also shows that one may succeed with a product that for practical reasons has not been extensively optimized (e.g. to better enrich DC) as long as it can be reproducibly manufactured, a potency marker is available, and one has chosen a tumor amenable to the cancer vaccine.

A phase III DC trial in melanoma highlights the importance of standardized vaccine production

In this multicenter trial, cocktail-matured MoDC loaded with select MHC class I and class II restricted tumor peptides were compared with dacarbazine standard chemotherapy as first-line treatment in stage IV metastatic cutaneous melanoma. The trial was stopped following interim analysis, as there was no significant difference between the two arms 42. We have reported that the mean number of DC injected into the skin was low (2.8×106per class I peptide) and highly variable (SD 1.1×106), and that in addition, the DC were of inferior quality (48% of applied vaccines contained more than 25% immature DC) 42. We have now performed immunomonitoring in a cohort of the patients and found that the vaccine responses were negligible when compared to the robust immunogenicity observed in our more recent 62-patient monocenter multi-peptide trial, in which peptide-loaded DC of high quality were injected at a higher dose of 10 million DC/class I peptide (our unpublished data). In retrospect, the multicenter trial was premature because product development, standardization, and validation had not reached the level required to obtain a GMP manufacturing license. In Europe, an EU directive dictates that GMP products have to be used in clinical trials of all phases 43. This implies that in all member states, only products of GMP quality can be used for the production of DC vaccines. The securing of the GMP quality of the end product, i.e. the DC vaccine, is, however, left to the national authorities and is guaranteed by the requirement for a GMP manufacturing license, which imposes substantial validation requirements, only in some European countries such as Germany. In contrast, in the USA, there is not a strict need for full GMP quality of products (e.g. cytokines) in early phase I/II investigator-initiated trials.

Urgent need for two-armed trials in order to optimize DC vaccination

After more than 10 years of DC vaccination, it is now imperative to systematically address, in small two-armed, science-driven immunogenicity trials (which so far have been a rare exception 44–46) the important variables and opportunities to identify an optimized DC vaccine for later testing in randomized phase II and III trials. At this point, many factors remain to be systemically tested, including the dose, frequency, and route of DC vaccine administration, let alone the many ideas and possibilities arising from DC biology. DC, depending on their subset and maturation status, can induce and activate all kinds of T cells (including Treg), B cells, and antibodies 36, NKT 47, 48 and NK cells 49–52, in principle allowing a broad “coordinated anti-tumor response” 53. With respect to clinical testing, one priority is the induction of strong T-cell responses, which in my view has yet to be achieved. It will also be valuable to compare DC directly to other vaccine strategies, e.g. in case of HPV E6/E7 antigens to synthetic long peptide (SLP) vaccination, or in case of the prostatic acid phosphatase antigen to Dendreon's Provenge™ that requires one apheresis for preparing a single vaccine.

MoDC as a focus for comparison

I think the monocyte as a source of DC has value for several reasons:

  • (i)
    “GM-CSF+IL-4” MoDC matured by the “cocktail” (TNF-α+IL-1β+IL-6+PGE2) represent – if properly produced under GMP rules – a highly reproducible standard for comparison;
  • (ii)
    “GM-CSF-IL-4” MoDC can be matured in other, possibly better ways for immunization;
  • (iii)
    the combination of GM-CSF+other cytokines but IL-4 (e.g. IL-15) can produce unique MoDC types; and finally
  • (iv)
    “GM-CSF+IL-4” DC can be modified by mRNA transfection to generate “designer” DC. Such trials will become easier as commercial kits to generate MoDC in a fully closed system become available.

We have used “GM-CSF+IL-4” MoDC matured by the standard maturation cocktail intentionally in all of our trials using peptide-loaded DC so far to obtain solid insight into their immunogenic properties by serial immunomonitoring. In our 62-patient trial (our unpublished data), in metastatic melanoma, tumor-peptide-specific ex vivo detectable (Elispot, tetramer staining) IFN-γ-producing CD4+ T cells were regularly detectable. This is surprising given the low amounts of IL-12p70 released from cocktail-matured DC in vitro but may be explained by their expression of CD70 20, 54. The vaccine-specific Th cells were FOXP3-negative, indicating that vaccine-specific natural Treg were not induced, which is counterintuitive to a previous report 55, yet compatible with recent Treg cloning data 56. Vaccine-specific CTL were, however, only weakly detectable ex vivo, but the CTLp frequency was markedly increased, and many CTL were of high affinity. Interestingly, prolonged survival in stage IV melanoma patients beyond 24 months appeared to require both a “strong” induction of immunity in the first 3 months and a “friendly” transcriptome pattern (e.g. upregulation of T-cell markers, chemokines, and innate immune factors) in pre-vaccination metastases 57. Similar findings by others 58, 59 indicate that transcriptional profiling should be tested prospectively as a stratification factor. Figdor and de Vries have made the remarkable observation that the detection of functional peptide-specific T cells in DTH lesions induced by injection of peptide-loaded DC positively correlates with time to progression and overall survival 60. Another recent notable finding is that strong expression of tumor endothelial marker-8 (TEM-8) on mature DC correlates negatively with overall survival in DC-vaccinated melanoma patients 61. This observation may reflect abnormalities of myeloid cells in advanced cancer patients, and calls for a more thorough investigation of the quality of autologous DC preparations beyond the usual release criteria, e.g. by transcriptome analysis.

It is imperative to compare the widely used standard maturation cocktail to other maturation stimuli. Several combinations of stimuli (notably TLR ligands and CD40L) have been described to generate superior T-cell stimulatory/differentiation capacity in vitro (where migratory DC and their released products are forced against their nature to stay together with T cells in culture vessels), but too little is known whether this will translate in vivo into enhanced effectiveness 62–68, 69, 70. One obstacle to clinical testing is that these reagents are often not available to the scientific community in GMP quality. Apart from the maturation stimuli, it will also be interesting to determine the impact of better antigen loading, e.g. by testing SLP.

It will be important that one can rapidly and realiably identify the preferred DC vaccine by successively addressing variables in small, two-armed trials. Because patients and their tumors are so variable, one should include a mix of memory antigens (usually protein for CD4+ recall responses 30) to evaluate immune competence and changes throughout vaccination. This will also help to more objectively categorize an immune response as “strong” or “weak,” e.g. by comparing a vaccine-specific CTL response to an endogenous CMV response if class I control peptides also are used. In addition, it is my personal opinion that we should also load a separate batch of control DC with relevant antigens for priming (e.g. HIV or other viral epitopes) to identify a superior DC vaccine in a small number of patients. With increased vaccine efficacy, T cells will become more often detectable ex vivo, so that one can also sort tetramer-positive T cells for easier testing of mono- versus polyclonality (the latter observed to occur with cocktail-matured DC 71, 72), polyfunctionality (which correlates – at least in viral disease – with clinical benefit 73, proliferative capacity (relevant as it reflects one memory T-cell feature), and transcriptome analysis (which appears to reflect priming by different vaccines, and in case of DC vaccines might reflect the DC transcriptome 74). With enhanced DC vaccines, one should then even see characteristic cellular and/or humoral signatures in whole blood as observed in case of the strong yellow fever vaccine 75–78. Immunomonitoring requires standardization and reproducibility, an important component of which are discussion groups (e.g. the MIATA project, www.miataproject.org) and proficiency panels, which are already offered for tetramer staining, intracellular cytokine labeling, and Elispot assays by the CVC and the CIMT (see www.cancerresearch.org and www.cimt.eu, respectively). I strongly support participation in such intercomparison programs to facilitate accurate and transparent data presentation.

MoDC generated using GM-CSF+IL-15

These are also high on the list of priorities, as MoDC cultured in GM-CSF+IL-15 are superior in vitro in inducing high-affinity CTL 79, 80. It remains difficult for us to generate a sufficient number of highly standardized DC under these conditions, which is a prerequisite for true GMP production. I suspect that similar problems have occurred to others, which explains why there are no data available yet on their immunogenicity in vivo in humans.

RNA-transfection of DC

First reported by E. Gilboa in 1996 (for review see 81), RNA-transfection of DC offers distinct opportunities, particularly since the unreliable “simple” addition of mRNA to DC has been substituted by electroporation 82, which allows strong protein expression and intracellular staining in the majority of DC, a prerequisite for reproducibility, validation, and thus GMP production 83. A crucial regulatory advantage is that mRNA transfection does not constitute gene therapy as mRNA is not integrated into the genome. It is possible to transfect mRNA coding for defined antigens, which similar to loading with SLP allows the production of multiple HLA class I and – if, e.g. DC-LAMP-modified mRNA is used – also class II epitopes. In addition, there is the potential to include functional molecules to program a next generation of “designer” DC. We are, for example, currently testing in a comparative trial “GM-CSF-IL-4” MoDC transfected with mRNA (but after rather than before maturation) coding for three antigenswith or without an E/L selectin fusion molecule, designed to bring about migration of DC upon i.v. injection from the blood to the lymph nodes and, thereby, achieve stronger T-cell responses with a more diversified homing pattern 84. This would be a major advantage because limited migration even of mature DC from skin injection sites to draining lymph nodes remains a major limitation, notably as intranodal injection has proven unreliable 85 and pre-conditioning of the injection site in contrast to mice does not enhance DC migration in man (de Vries, personal communication). Interestingly, in our current trial intravenous (but not intracutaneous) injection of DC led to some cases of clinical regressions, and should thus be explored despite a previous comparative trial pointing to the inferiority of the i.v. route 45. We are also exploring DC transfected after maturation with an optimized CD40L mRNA, which results in DC that induce highly proliferative, inflammatory CTL in vitro63, 64. Within the DC-THERA Network of Excellence (www.dc–thera.org), another novel “designer” DC type is currently being compared to other DC, the so-called Tri-Mix DC (generated by transfecting immature GM-CSF+IL-4 DC with mRNA coding for CD70, CD40L, and a constitutively active TLR4) 86. There are many other possibilities to enhance the stimulatory capacity of DC for T or also NK cells, either by introducing other advantageous molecules via mRNA or silencing inhibitory ones by siRNA transfection (e.g. SOCS1) 87.

Loading DC with dying tumor cells has proven promising in clinical trials 88, particularly with autologous tumor cells and “only” cocktail-matured DC 89, 90. The workup of the patients treated by C.W. Schmidt's group 89, 90 using a laborious yet highly informative strategy 4 has shown that the vaccine-induced immune responses are dominated by highly individualized responses to shared and neoantigens generated by somatic point mutations (Thomas Wölfel, personal communication) in congruence with previous observations in select melanoma patients 3, 4. The mRNA transfection approach allows for exploring the total antigenic repertoire of tumors without limitations imposed by availability of tumor tissue, as even a few cells can provide sufficient amounts of mRNA for PCR amplification 81. An alternative approach yet to be tested is to take advantage of the increasing knowledge on the cancer genomes, and to use mRNA-transfected DC to specifically target oncogenic driver mutations 91. T cells recognizing such mutated epitopes have rarely been observed, probably for simple statistical reasons given the handful of patients studied and the much larger number of other somatically mutated proteins in cancer cells 92. We had been the first to use DC to generate Bcr-abl-specific CTL capable of killing CML cells 93, but to test the mRNA approach, we will now vaccinate to the V600E mutated B-RAF and check for specific T cells for proof of principle in melanoma 94, 95. Immunizing against multiple driver mutations in succession would be appealing because some will also be present in the cancer-initiating cells. Following an approach recently developed to target a rapidly mutating and escaping HIV virus by mRNA-transfected DC would even permit exploitation of the changes in oncogene mutations over time 96. In addition, the T-cell-based approach should allow an attack on the entire tumor cell in a natural way, and to prevent its escape by hitting multiple immune targets. This is not easily possible by blocking mutated signaling pathways with small molecules as it appears relatively easy for a cancer cell to find a way around a single block, and combinations might be too toxic even with advanced drugs. The highly selective PLX4032 inhibitor of B-RAF (V600E) rapidly induces impressive shrinkage of melanoma metastases 97, but many tumors evade later on, and other complications may arise if there are concurrent N-RAS mutations 98. Blocking tumor growth even transiently, e.g. by such highly specific kinase inhibitors that do not impede DC or T-cell function, opens up the possibility to allow a gradually evolving vaccine response directed to somatically mutated or other, preferably functionally relevant and tumor-restricted or stromal antigens 6, to produce clinical benefit.

Combination therapies

There are thus many opportunities to make DC vaccines better, but combination therapies will likely still be required to achieve higher clinical efficacy in patients with higher tumor load. Because much needs to be researched, we have to concentrate on testing in the clinic both what makes sense and what is available right now, without complicated negotiations to obtain access to proprietary experimental drugs. Combination with chemotherapy or local irradiation 99, for example, is attractive. Anti-CTLA-4 antibodies will hopefully be approved soon 100, and can then be systematically tested also in the context of DC vaccines, which will be very interesting given promising observations in previously vaccinated patients 101, 102. Another possibility for “off label” use is Sunitinib, which appears to inhibit STAT3 9, and could be combined with DC vaccination as it does not appear to block DC or anti-tumor T cells 103, 104.

Concluding remarks

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information

The domain of tumor vaccines in the future is likely therapy in the adjuvant setting (“minimal residual disease”), or even the prophylactic treatment of high-risk patients. While virus-associated cancers can be prevented by prophylactic vaccines (e.g. Hepatitis B or HPV) that induce anti-viral neutralizing antibodies, tumor vaccines have to prevent the appearance of spontaneous, non-microbially triggered malignancies via tumor-specific T cells and are, therefore, more difficult to design. Also we need to know more about how to attack cancer-initiating and dormant tumor cells.

The step-wise rational development of effective cancer vaccines requires coordinated networks, new procedures to get access to drugs under development to test promising combinations, and a much better task management as currently also discussed in the USA (see www.nap.edu/catalog/12879.html). Clinical trials are both costly and demanding because of the ethical, logistical, and increasingly stringent regulatory requirements. As the number of trials possible is therefore limited, it is crucial to develop consensus strategies to pick the right ideas and critical variables. In the DC-THERA network (www.dc–thera.org) and the CIMT integrated project (www.cancerimmunotherapy.eu), we have been quite successful in reaching a consensus on such priorities regarding DC vaccination trials but in spite of this, obtaining sufficient financial support for such consensus trials remains a major hurdle. We as scientists will have to put much more effort into convincing politicians as well as the public that it is crucial to invest in this field so that discoveries can be efficiently and promptly translated into therapies that are of help to the patients. We also have to point out the crucial role of academic research as a think tank where many ideas are promoted to finally trigger the interest of investors or pharmaceutical companies.

References

  1. Top of page
  2. Abstract
  3. Why do cancer vaccines induce T cells but so far exert only limited clinical efficacy
  4. Tumors silence DC and exploit their tolerogenic function
  5. Tumor vaccines must exploit the immunogenic function of DC
  6. The three ways of exploiting DC as vectors for vaccination (Fig. )
  7. Adoptive transfer of antigen-loaded DC into patients (DC vaccination)
  8. Concluding remarks
  9. Acknowledgements
  10. References
  11. Supporting Information