Translating immunity into cancer therapy: new findings from the first conference of the European Society for Cancer Immunology and Immunotherapy, an ESMO-EIS held in Athens, 15–17 November 2007

The discussion on breast cancer immunotherapy at this first ESMO-supported ESCII conference in Athens was perhaps appropriate given that the very name “cancer” derives from the ancient Greeks’ observations on the external manifestations of breast tumours, with Galen noting that the veins in breast cancer are dilated and similar in appearance to crabs’ legs on both sides of the body, indicating presence of the tumour. Important modern concepts and therapeutic targets of inflammation and angiogenesis were presaged by their observations, and the effect of diet on causation was also perceived. This 2-day meeting, 2000+ years later, considered the current position of immunotherapy from many points of view, including the issue of cancer and inflammation, adoptive therapy by antibodies and cells, active immunotherapy with vaccines and the critical role of immunity versus tolerance. We present here a brief personal view of the highlights of the meeting, organised by H. Mellstedt, Stockholm, Sweden; J. Wagstaff, Swansea, UK; and M. Papamichail, Athens, Greece, with apologies to all those not fully cited. Please refer to the symposium papers from this meeting for more detailed accounts by selected speakers.

As also clearly recognised by the ancient Greeks, prevention is better than cure. Thus, the recent introduction of prophylactic HPV vaccinations for the first time offers real preventive possibilities for this still common disease [1], and hepatocellular carcinoma associated with hepatitis B and C may be amenable to this approach in future. However, many tumours are not known to be virus-induced, and in most cases cancer is already established at presentation, necessitating therapeutic approaches. These are also being explored in cervical cancer, as described in the keynote lecture of C. Melief (Leiden, The Netherlands) who stressed the importance of improved understanding of the mechanisms of successful immunisation for effective triggering of anti-cancer responses. This is particularly important when vaccinating with synthetic peptides to target the T cell immune response to defined antigens expressed by the tumour. The earlier findings of this group, that short peptides may induce tolerance rather than immunity [2] have now been extended to include five different examples of such short peptides being picked up and presented by T and B cells, “non-professional” antigen-presenting cells (APC). To this end, they are using cocktails of 13, 25-residue-long HPV16-E6/E7 peptides in early (CIN) and late (cervical cancer) patients, with better T cell activation when the E6 and E7 multi-epitope vaccines were given separately, at different locations, to minimise competition. This approach resulted in 1/42 CR at 3+ years in terminal cervical cancer, and 5 SD at 1.5–3 years (expected survival, historical controls, <1 year). J. Schneider (Strasbourg, France) presented pre-clinical as well as clinical data from a phase II trial using TG4001/R3484, a viral suspension of recombinant vaccinia virus (MVA) carrying nucleotide sequences also encoding HPV16-E6 and E7 antigens, and including the human immunoregulatory cytokine interleukin-2 (IL-2), as a therapeutic vaccine candidate for CIN2/3 pre-neoplastic lesions. Use of this vaccine was found to be feasible and safe. Its efficacy threshold was significantly above the 20% of spontaneous regression expected; no CIN2/3 relapse, nor HPV16 persistence or re-infection at month 12 was observed in patients who had a clinical response at month 6.

Prevention of recurrence and metastasis by active immunisation of disease-free early stage breast cancer patients was presented by G. Peoples (Washington DC, USA). The data from 188 patients enrolled in one of the largest breast cancer preventive vaccine clinical trials revealed that immunisation with a single peptide, E75, from the HER-2/neu oncoprotein, was effective in prolonging overall survival and decreasing the rate of recurrence at a median of 30 months of follow-up after enrolment. Furthermore, patients appeared to be protected from bone metastasis. Thus far, patients receiving serial booster inoculations have not experienced any recurrences after a median 18 months of follow-up. This single peptide antigen appears to act as a catalyst to the immune system allowing further epitope spreading, with endogenous cancer cells as the substrate [3]. The fact that vaccination with a peptide representing one single epitope resulted in development of T cell clones with other specificities not previously detectable was also emphasised by P. Coulie (Brussels, Belgium) in metastatic melanoma patients. The use of IFN-γ to transiently increase the expression of a decreased (or even lost) antigen before vaccinating for this particular antigen might lead to the generation of effector clones against other antigens, not down-modulated, and thus decrease the importance of antigen loss as a mechanism of tumour escape.

One of the major themes to emerge from this meeting was that combining conventional treatments (chemotherapy, radiotherapy, hormone therapy) with immunotherapy is likely to be the best way forward. The use of chemotherapy for lymphodepletion before adoptive transfer of T cells, depleted of Tregs, or alongside active immunisation, is one such approach, as promulgated by R. Hawkins (Manchester, UK). However, even more strikingly, unexpected, and so far unexplained, beneficial results were counter-intuitively obtained when vaccination preceded chemotherapy. This could be crucial for the re-interpretation of the place of immunotherapy in anti-cancer treatment, namely, even when immunotherapy alone seems to deliver no survival benefit, in disease where cancer chemotherapy alone also delivers minimal or no benefit, preceding apparently ineffective chemotherapy with apparently equally ineffective immunotherapy does deliver clinical benefit. Such phenomena were noted many years ago in guinea pigs, but not consequently applied to the clinic [4]. This was really a very encouraging view emerging from the current work presented at the meeting, also by D. Gabrilovich (Tampa, FL, USA) and A. Dalgleish (London, UK) as well as Melief. Thus, Gabrilovich reported a trial in small cell lung cancer using a p53/DC-based vaccine in 50 stage IV patients, documenting evidence of specific T cell responses by IFN-γ ELISPOT and tetramer analysis in 47% of patients. The presence of myeloid suppressor cells (CD33+, HLA-DR-, lineage-) in the peripheral blood was negatively associated with p53-responses, whereas T regulatory cells were not. However, as so often the case in such immunotherapy trials, the presence of T cell reactivity failed to correlate with clinical efficacy, of which there was, as now unfortunately expected, very little (only two PR in the whole cohort of 50 patients). The surprise lies in the finding that the response to salvage chemotherapy in patients progressing after immunotherapy was greatly improved: 53% of vaccinated patients now responded to chemotherapy (even though only 33% of historical controls had done so). Even more remarkably, in platinum-resistant patients, these values were 50 versus 0% responses, and survival was correspondingly increased. The response to salvage chemotherapy did correlate with immune responses to the p53 vaccine (P = 0.025); median survival was 13.7 versus 8.8 months (not significant due to small numbers of patients). Clearly these exciting results need intensive following up. It remains to be seen whether it will be a general finding that the different types of immunotherapy all have a “sensitising” effect on subsequent different forms of chemotherapy.

There was much speculation and discussion as to the mechanism of this effect of immunotherapy on the subsequently improved responses of refractory patients to previously ineffective chemotherapy, especially as the latter would be expected to inhibit the former. Indeed, Gabrilovich found that chemotherapy completely ablated p53-responses in vaccinated patients in a few weeks. Clearly, the “window” in which the effect must occur is quite limited. Full understanding of these phenomena is likely to have an enormous impact on cancer treatment, but so far all explanations are entirely speculative. This should represent a very fruitful area for future investigation.

Regardless of the mechanism, these findings of immunotherapy acting in an adjuvant manner offer encouragement to further improve ongoing vaccination trials, such as by maximising T cell survival, homing to and persistence in the tumour. However, an increasingly well-recognised characteristic of such T cells is their compromised function, as assessed by decreased signal transduction, cytokine production and proliferative capacity. This may be caused by oxidative stress resulting from the production of reactive oxygen species by myeloid suppressor cells, macrophages or other tumour stromal cell types and causes apoptosis of both T and NK cells, especially the more mature effector cells, due to their lower anti-oxidant protection mechanisms (R. Kiessling, Stockholm, Sweden). Thus, interventions to increase resistance to oxidative stress could have far-reaching effects. This was illustrated by retroviral transfer of catalase into T cells, resulting in decreased susceptibility to oxidative stress of both CD4 and CD8 cells (Kiessling). Moreover, inflammation associated with oxidative stress may not only have a negative effect on anti-tumour effector cells, but also a tumour-promoting action, data on which now emerging rather counter-intuitively suggest that antibody may play a crucial role in this context. This was described in a mouse HPV16 model in which oncogenic transformation alone is not sufficient to drive skin carcinogenesis, which fails to occur in the absence of adaptive immunity and which is dependent on soluble mediators derived from B lymphocytes (K. de Visser, Amsterdam, The Netherlands). Although not generally applicable to all mouse cancer models, similar phenomena may occur in humans too; the mechanism remains unclear, but seems not to involve complement (de Visser). The appearance of antibody in a different mouse model, the transgenic SV40LT sporadic tumour model, has been correlated with the development of tolerance to the tumour antigen which occurs very early on, a long time before appearance of the tumour. These and other findings have led T. Blankenstein (Berlin, Germany) to propose that no functional CTL are induced in these animals and hence the mice are unable to reject the tumour. The presence of both cellular and humoural “anti-tumour” effects for a long time before tumour emergence could rather be taken to indicate the failure of the immune system to spontaneously control cancer, even if it expresses a strong tumour antigen. However, there are many other interpretations, which were energetically discussed at the conference. Moreover, these findings do not imply that immunotherapeutic applications of either cellular or humoural agents are doomed to failure, but rather that the co-evolution of tumour and immune system over extended periods is in a delicate balance that could be tipped one way or the other [5]. Therapeutic manipulation would then be possible with either component; indeed, some of the most successful applications of immunotherapy thus far have been achieved using antibodies, as emphasised by J. van de Winkel (Utrecht, The Netherlands). The evolution of sophistication in antibody development, from murine to chimeric (humanised) and currently to fully human monoclonal antibodies, using human immunoglobulin transgenic mice, as well as their excellent specificity and binding characteristics, and their capacity to effectively modulate signalling, has greatly improved their therapeutic efficacy. However, even this tried and tested technology may still be amenable to improvement, e.g. by replacing antibodies with aptamers (synthetic oligonucleotide-based ligands with specificity and avidity similar to or even better than antibodies), which although usually antagonistic can also work in agonistic fashion (E. Gilboa, Miami, FL, USA). Cellular therapies are still relatively recent compared to antibodies and have not yet yielded such positive results; most of this conference was dedicated to the former and to efforts to improve clinical outcomes. Also in the more “conventional” setting, further developments are ongoing, for example, the antibody-based microarrays, representing a novel technology with great promise for high-throughput proteomics. C. Borrebaeck (Lund, Sweden) presented the next generation of a human recombinant scFv antibody microarray platform for protein expression profiling of nonfractionated human plasma and serum proteomes [6]. This platform was found to provide specific, sensitive [subpicomolar (pM) range] and reproducible means for protein profiling. The use of affinity proteomic technology to identify biomarkers for earlier diagnosis, patient stratification, treatment selection, prediction of therapy resistance and tumour relapse, may greatly assist the improvement of immunotherapeutic approaches.

Cancer vaccines for active immunisation, be they whole cells or peptide-pulsed dendritic cells or any other variant, require a functional immune system for full efficacy. On the other hand, adoptive immunotherapy enables already-activated effectors to be delivered to the patient, either tumour-specific or NK cells. The importance of the latter in protection against cancer is illustrated by the findings of Imai et al. [7] in a 11-year follow-up of >8,000 people, in which of the parameters measured, only the level of NK activity correlated with cancer risk. Thus, high NK activity was protective. There is therefore reason to believe that NK cells would be therapeutically active. However, even employing adoptive immunotherapy, the tumour environment is commonly suppressive for effector cells, including NK cells, which also become impaired in a way paralleling the well-known phenomenon of T cell anergy, i.e. they fail to respond via activating receptors but can be stimulated by calcium ionophores and phorbol esters (K. Malmberg, Stockholm, Sweden). However, it may be that anergy can be broken by culture in IL 2, as is the case with Melan A-specific T cells found in melanoma patients (P. Romero, Lausanne, Switzerland). Because NK cells are not MHC-restricted, not only the patient but any healthy unrelated person may act as a source of effector cells for adoptive therapy; S. Slavin (Tel Aviv, Israel) reported success in applying activated allogeneic NK cells to 40 unrelated patients in a phase I trial; the effector cells are present only transiently before rejection by the host but mediate potent anti-tumour effects prior to that. This procedure is easy, cost-effective and with minimal side-effects, and essentially extends earlier therapeutic experience with LAK cells. Interestingly, despite the anticipated graft-versus-tumour effects mediated by donor NK cells, not excluding additional anti-cancer effects mediated by donor T cells also activated by rIL-2, no GVHD was seen, due to rejection of the donor lymphocytes. Transient circulation of donor lymphocytes may be sufficient for eradication of minimal residual disease but certainly ineffective against larger tumour bulk, which may require months of ongoing reactivity of donor lymphocytes, thus requiring induction of transplantation tolerance by engraftment of donor stem cells. Increased understanding of NK cell biology, especially in terms of their activating and inhibitory receptors, is also expected to result in the refinement of such approaches, for example, by matching the unrelated donor appropriately to the patient to maximise activation (H.-G. Ljunggren, Stockholm, Sweden [8]).

There is still very little experience using T cells in adoptive immunotherapy but large numbers of active immunotherapy trials are still being conducted. Nature of the antigen, APC, adjuvant, monitoring and predicting patient-specific outcome are all crucial factors not yet fully understood. There is currently much interest in suppressive cells, especially T regulatory cells, which are poorly defined and difficult to manipulate in humans. Recent clinical trials have suggested that inhibiting such cells may increase anti-cancer efficacy, at the price of autoimmunity [9]. Tregs can be depleted using different approaches, also including immunotherapeutic strategies to target Foxp3 antigens (Gilboa [10]). Dissecting the different types of Tregs may enable separation of activity or at least better control of autoreactivity. To this end, J. Schultze (Cologne, Germany) is searching for better markers of Tregs by assessing their gene signatures and identifying differences between “constitutive” and the many different varieties of “induced” T reg cells. The best marker so far remains low level expression of CD127 regardless of high or low CD25 expression, both also having the same gene signatures (Schultze). Regardless of the targets selected, T regulatory and other suppressive cells will still need to be modulated to achieve clinical success. Thus, higher levels of CD4 + CD25 + Foxp3 + cells in multiple myeloma patients were associated with non-responsiveness to an idiotype vaccine combined with IL 12 and GM-.CSF which yielded an 85% clinical response rate overall (Mellstedt). Appropriate targeting may also circumvent the Treg problem, as reported by A. Griffioen (Maastricht, The Netherlands) using the synthetic anti-angiogenic peptide Anginex [11]. This agent increases ICAM-1 expression and facilitates tumour infiltration by activated T cells but not the preferential accumulation of Tregs (Griffioen). Thus, in addition to the anti-cancer activity of angiogenesis inhibitors, this facilitation of T cell entry may be an important mechanism of action of such agents. These effects need further exploration.

The antigens to target in T cell-based active immunotherapy, and how best to deliver them, remain under intense investigation. Antigen modification is a frequently applied strategy to strengthen immune responses to self-antigens. However, as D. Speiser (Lausanne, Switzerland) pointed out, although higher T cell frequencies can be induced by vaccination with the modified peptides, TCR avidity and anti-tumour recognition, and surprisingly also T cell activation, are superior after vaccination with natural peptides. The use of adjuvant with natural peptide and better monitoring strategies to assess their efficacy may be more important (H. Levitsky, Baltimore, MD, USA). Levitsky presented a method for the quantification of the impact of vaccine adjuvant, through MR imaging, which should facilitate the clinical evaluation of strategies to improve the efficacy of tumour vaccines. In vivo labelling of DC that simultaneously capture tumor antigen and Super paramagnetic iron oxide (SPIO) allows quantifying the migration of DC that have captured antigen in vivo. Directing DC into the lymph node is challenging and prone to error; intradermal injection is easier and although indium labelling shows that most DC do not reach the LN, those that do enter migrate to the T cell areas (C. Figdor, Nijmegen, The Netherlands). Monitoring T cell sensitisation in response to antigen-pulsed DC injection rarely reveals correlates with clinical outcome, although DTH might [12] but the nature of the T cell response, rather than merely the presence of antigen-specific T cells must be taken into account; T cells producing IL 5, for example, are not indicative of a favourable outcome (Figdor). Although harder to monitor specifically (but not by DTH), the early approach of whole tumour cell vaccination may still offer a strategy as good as or better than any other. Several trials with whole cell vaccines are continuing, in a more modern incarnation in combination with other agents, for example, a phase I trial of GM-CSF-transfected prostate cancer cell lines injected into allogeneic patients treated with escalating doses of CTLA-4 antibody (R. Scheper, Amsterdam, The Netherlands). At the highest dose level of antibody, most patients experienced decreases in PSA and signs of clinical improvement—but also signs of autoimmunity. The use of allogeneic immunogens may also contribute an adjuvant effect, as in this study; the use of a DC line [13] may be valuable in this regard as large amounts of cells, usually a limiting factor in autologous DC vaccination, can be obtained and fused with tumour cells, or exposed to tumour cell RNA (Scheper).

Conclusions

Progress has been made, optimism is warranted, but regulatory hurdles must be overcome and the difficulties of conducting clinical trials reduced in order to translate the ever improving state of immunological knowledge into clinical reality. We look forward to next year’s combined ESCII/PIVAC meeting (Progress in Vaccination Against Cancer, see ref. [14] for a summary of the most recent conference) in the autumn of 2008 to determine the agenda for the future.