**3. From proof of the principle to proof of efficacy**

Tumor vaccination therapy, even in its more basic form did yield some interesting results in terms of benefit in overall survival in several fields of oncology e.g. renal cell carcinoma, prostate carcinoma, non-small cell lung cancer, colon carcinoma and others(11). Historically, most attention has been paid to so called immunogenic tumors like melanomas. Although proof of the principle has been extensively demonstrated in these tumors using different vaccine approaches and the clinical results obtained until

potent and popular immunological adjuvants to install an active anti-tumoral immunity(3;4). An additional advantage of the latter is the activation of both the innate and adaptive immunity arms of the patient. The mechanisms underlying this antitumoral priming capacity of dendritic cells seem to be fully consistent with the established paradigms of antigen uptake, processing and presentation. The exclusive potential of DC to present and cross-present exogenous antigens in the same antigen presenting cell is the critical characteristic for a successful antigen presentation to cytotoxic and helper T cells. DC pick up tumor-associated antigens from diverse sources, process them and present them in both an Major Histocompatibility Complex (MHC) class I and II context to cytotoxic and helper T cells respectively. This has been clearly shown, also for glioma associated antigens from a whole-tumor-cell lysate(5). To date, efficient priming of the patient by dendritic cells implies four sequences of interaction between the dendritic cell and the T cell. Upon binding of the T cell receptor with the MHC complex on the antigen presenting cell displaying the appropriate epitope, a specific activation of the antitumor T cells with high enough avidity for the epitope takes place. Expansion of this T cell clones requires co-stimulatory molecules on the DC. A polarized immune response, preferentially with a Th-1 cytokine profile, should result from the previous interactions. Finally, some indication about the target location should be transferred to the T cells leading to the appropriate T cell homing

Many rodent models have demonstrated that prophylactic vaccination of mice with tumor-antigen loaded DC can protect immune competent animals from tumor challenge and outgrow. The most relevant experiments are being performed with syngeneic micetumor models, to prove the efficacy of the vaccination in an autologous setting underscoring the possibility to break tolerance for self-derived antigens. Moreover, one can raise an anti-tumor immunological memory to make mice survive a re-challenge of cancer cells. More elaborated manipulations of the immune system like e.g. regulatory T cell depletion before vaccination even yield a much stronger protective effect in up to 100% of the exposed animals(9;10). The step towards therapeutic vaccination models, being more relevant for the clinical reality of all day can be made if the crucial timing of inoculation and vaccination is being respected. Probably also due to the aggressive nature of the investigated models, where all mice are dead within 3 weeks after tumor inoculation, it appears crucial to vaccinate the animals not later than one week after the tumor inoculations. After that period, the course of the disease can hardly be influenced, presumably because of the establishment of an efficient immune suppressive, pro-tumor micro-environment around the tumor inocula. This should be understood as the first evidence to use DC vaccinations as an adjuvant therapy in minimal residual disease settings, resembling most the prophylactic setting in which anti-tumor vaccines are so

Tumor vaccination therapy, even in its more basic form did yield some interesting results in terms of benefit in overall survival in several fields of oncology e.g. renal cell carcinoma, prostate carcinoma, non-small cell lung cancer, colon carcinoma and others(11). Historically, most attention has been paid to so called immunogenic tumors like melanomas. Although proof of the principle has been extensively demonstrated in these tumors using different vaccine approaches and the clinical results obtained until

properties(6-8).

efficient.

**3. From proof of the principle to proof of efficacy** 

now are equivalent to standard of care chemotherapy regimens, the final results of immunotherapy in these entities are still modest. Malignant gliomas, the most common primary brain tumors have always been considered to be not suitable for immunotherapy because of their location in the immune privileged central nervous system. To date however, we know that all the obstacles like the blood brain barrier, the lack of lymphatic vessels, the lack of residing antigen presenting cells in the brain, the low MHC expression on the tumor target cells are quite relative. They don't seem to hamper the documented immune responses in patients harboring such tumors(12). To the current understanding the real hurdles in glioma vaccination strategies are the multiple immunosuppressive pathways orchestrated by this type of cancer cells and the lack of universally expressed glioma associated antigens that are really crucial for tumor cell survival. Surprisingly, preliminary data might be pointing to an exploitable lack of spontaneous immunogenicity of malignant gliomas. As such, they develop in a micro-environment protecting them from too extensive immune editing: especially that characteristic could be of major help to try to reset the patients' immunity resulting in a much better immune rejection or control since the original immune-sensitive tumor clones have not yet been eliminated by the natural immune surveillance mechanisms(13).

Immune responses in the brain always elicit some fear for potentially disastrous consequences of an auto-immune attack by the patient's immunity. Unlike vitiligo or even destruction of normal prostate tissue, an auto-immune encephalomyelitis could result in devastating neurological symptoms and deficits. Until now however, no preclinical –other than in a heavily manipulated immune environment- and no clinical data have been published showing any suspicion of serious auto-immunity in the central nervous system, using tumor vaccination strategies. Moreover, numerous phase I and small phase II trials have been published showing the safety and the very attractive low toxicity profile of dendritic cell vaccines in brain cancer. As we're still dealing with a palliative treatment thus far, this perfect patient tolerance profile is highly valuable to build upon for upcoming vaccination strategies. Although today's technology to produce autologous dendritic cell vaccines is still very labour-intensive, it proved to be perfectly feasible to implement it in the daily clinical practice, both in the case of relapsed or newly diagnosed high grade gliomas.

The main criticism against tumor vaccine approaches, often raised till today, comprises the presumed low rate of objective tumor responses. Objective, radiological responses as defined according to RECIST(14) or McDonalds(15) criteria have indeed been developed for radio-and especially chemotherapy regimens during which one aims to reduce the measurable tumor load in the patient. Although they provide valuable information in terms of proof of the principle of the investigated chemotherapy, they should not be considered a synonym for clinically relevant efficacy: in only a few cases, one was really able to demonstrate a clear correlation between objective responses and overall survival benefits(16). More intriguing even is the existence of the so-called 'pseudo-progression' since radiochemotherapy with temozolomid became the standard of care(17). In up to 25% (or even 40%) of cases, one might see an initial increase in radiological tumor volume, rather early after the concomitant radiochemotherapy: strikingly, often especially these patients seem to have a better overall survival chance than the group not displaying these types of misleading radiological changes. Dendritic cell vaccination is known to cause similar radiological images in which transient contrast-enhancements on magnetic

Immunotherapy for Malignant Gliomas: A Roadmap for the Future 417

The clinical outcome measures that really do have an impact on the patients' perspectives are overall survival and quality of life. The well tolerated vaccinations often result in a minimal interference with a good quality of life, that is actually only being threatened by disease progression but not substantially by the therapy itself(24). The ambulatory nature of the vaccination schedules further contribute to this low impact on the treatment burden for the individual patient. Especially this low toxicity profile adds to the merits of this therapy in the subgroup of long-term surviving brain cancer patients. Indeed, even more than the possible statistically important impact on median survival data, the substantial group of patients with malignant glioma, both WHO grade III and IV lesions, surviving for many years after vaccination, is the best advocate of this therapeutic approach. Both newly diagnosed, but even more strikingly, relapsed and multi-relapsed patients with high grade gliomas display survival periods of more than 4 to 9 years (manuscript submitted). Considering the classical definition of long-term survivor in the malignant glioma literature, being patients surviving 24 or more months after diagnosis (of primary disease or relapse), substantial numbers of patients, including up to 25% of the relapsed HGG patients undergoing DC vaccination are actually benefitting this opportunity. These long-term survivors are not only an encouragement for this –even not fully mature immunotherapy approach, but also a source of scientificly important translational knowledge to learn more about the factors predicting this type of outcome after immunotherapy. Final scientific proof of efficacy can of course only be delivered by welldesigned, sound randomized controlled trials, several of which are currently running

**4. Lack of standards: Disadvantage or opportunity for further improvement?**  The 'dendritic cell therapy for cancer'-world today is still characterized by a large variety of similar but not identical approaches. Even the definition of dendritic cell can slightly vary according to their progenitors with different resulting markers on their surface. Although direct harvesting from the peripheral blood is possible, some DC are differentiated out of stem cells or cord blood, but for the vast majority of clinical grade DC today, monocytes are harvested out of the peripheral blood and differentiated into DC. Different culture protocols and conditions result in different phenotypes, but the minimal criteria should be respected before one can claim the cells to be dendritic cells for clinical use: they should display clear cytoplasmic veils, have a high expression of MHC class II molecules as well as costimulatory molecules like CD86 and have lost their 'monocyte' markers like CD14. Growing consensus is being reached about the mature DC being the preferred state of the cells to re-inject into the patients, rather than immature cells being able to tolerize rather than immunize the patient. Several maturation cocktails are being used, none of them however have been proven to be superior to the other variants in clinical use, although some evidence exists for the critical involvement of TNFα in the cocktail(25). As the serum of cancer patients might contain identified and unidentified immune suppressive agents inhibiting a good DC differentiation in ex vivo cultures, some favor the use of serum free

Dendritic cells should be loaded with relevant glioma-associated antigens. The sources of these antigens differ widely from well-defined, possibly acid-eluted peptides, proteins, whole tumor cell lysates and homogenates, total tumor RNA, vector constructs, apoptotic

throughout the world.

culture conditions.

resonance imaging (MRI) might represent inflammatory, vaccine-induced radiological changes that can easily be mistaken for disease progression(18). Although the available literature mentions objective responses in about 13% of vaccinated patients with high grade gliomas, objective responses are not the most appropriate outcome measure for this therapy nor for other experimental therapies. Especially the growing consensus that immunotherapy should be used as an adjuvant treatment for minimal residual disease settings, implies that in many cases there will be no measurable tumor volume at the start of the treatment. Indeed, not all patients today are good candidates to possibly benefit from DC-based immunotherapy. Only tumors amenable to meaningful surgical resections should be considered candidates for adjuvant DC vaccination(19). Several reasons exist for that restriction: first of all one need enough tumor specimen to obtain tumor associated antigens. Secondly, one should be able to stably wean the patient from perioperative steroids, which might dampen an efficient immune response in case of vaccinations under steroids. Thirdly, by reducing the bulky tumor load, one partially corrects the immune suppressive environment, both locally and systemically, that results from the presence of an organized tumor. In that context, one should mention that modern resection techniques like the use of 5-ALA induced fluorescence guided resections, leading to more extensive glioma resections, can be a complement to the postoperative DC vaccination.

In the past, many researchers have focused on the measurable immune responses in the blood of vaccinated patients as a surrogate endpoint and even as a surrogate of the desired objective clinical responses. The many immunological assays like delayed type hypersensitivity (DTH) reactions – even with skin biopsies of the test sites - , ELISA, ELISPOT, tetramer analysis and diverse in vitro cytotoxicity assays have provided us with valuable insights in relevant immunological mechanisms contributing to the proof of the principle and to our understanding of the complex interaction of the immune system and cancer cells. Regardless the assay used, there seem to be a fairly constant rate of about 50% immune responders in vaccinated cancer patients with malignant gliomas. Apart from a rare exception, most assays failed to correlate with clinical results. We should realize that we have to leave the former linear paradigm stating that cancer vaccines induce a detectable immune response that results in a detectable clinical response (tumor rejection), finally leading to improved overall survival. It did teach us however the important lesson that to date, dendritic cells indeed seem to be the best adjuvants available for clinical use to elicit measurable immune responses in cancer patients, even if almost all of them had been heavily pretreated with radio-and chemotherapy. It has to be mentioned that a rational combination of preferentially, non-myelo-ablative chemotherapy leading to 'proinflammatory' immunogenic apoptosis of cancer cells rather synergizes than antagonizes with modern vaccine approaches. Several excellent reviews are dealing with that particular finding(20-22).

Indeed, since recent years a growing consensus on the way to proceed with clinical research in cancer vaccine strategies has arisen. The Cancer Vaccine Clinical Trial Working Group already suggested two parallel tracts of investigation: trials focusing on proof of principle and efficacy trials(23). The former should aim to demonstrate immunological activity, the latter should be designed to show clinical benefit for the patients.

resonance imaging (MRI) might represent inflammatory, vaccine-induced radiological changes that can easily be mistaken for disease progression(18). Although the available literature mentions objective responses in about 13% of vaccinated patients with high grade gliomas, objective responses are not the most appropriate outcome measure for this therapy nor for other experimental therapies. Especially the growing consensus that immunotherapy should be used as an adjuvant treatment for minimal residual disease settings, implies that in many cases there will be no measurable tumor volume at the start of the treatment. Indeed, not all patients today are good candidates to possibly benefit from DC-based immunotherapy. Only tumors amenable to meaningful surgical resections should be considered candidates for adjuvant DC vaccination(19). Several reasons exist for that restriction: first of all one need enough tumor specimen to obtain tumor associated antigens. Secondly, one should be able to stably wean the patient from perioperative steroids, which might dampen an efficient immune response in case of vaccinations under steroids. Thirdly, by reducing the bulky tumor load, one partially corrects the immune suppressive environment, both locally and systemically, that results from the presence of an organized tumor. In that context, one should mention that modern resection techniques like the use of 5-ALA induced fluorescence guided resections, leading to more extensive glioma resections, can be a complement to the

In the past, many researchers have focused on the measurable immune responses in the blood of vaccinated patients as a surrogate endpoint and even as a surrogate of the desired objective clinical responses. The many immunological assays like delayed type hypersensitivity (DTH) reactions – even with skin biopsies of the test sites - , ELISA, ELISPOT, tetramer analysis and diverse in vitro cytotoxicity assays have provided us with valuable insights in relevant immunological mechanisms contributing to the proof of the principle and to our understanding of the complex interaction of the immune system and cancer cells. Regardless the assay used, there seem to be a fairly constant rate of about 50% immune responders in vaccinated cancer patients with malignant gliomas. Apart from a rare exception, most assays failed to correlate with clinical results. We should realize that we have to leave the former linear paradigm stating that cancer vaccines induce a detectable immune response that results in a detectable clinical response (tumor rejection), finally leading to improved overall survival. It did teach us however the important lesson that to date, dendritic cells indeed seem to be the best adjuvants available for clinical use to elicit measurable immune responses in cancer patients, even if almost all of them had been heavily pretreated with radio-and chemotherapy. It has to be mentioned that a rational combination of preferentially, non-myelo-ablative chemotherapy leading to 'proinflammatory' immunogenic apoptosis of cancer cells rather synergizes than antagonizes with modern vaccine approaches. Several excellent reviews are dealing with that particular

Indeed, since recent years a growing consensus on the way to proceed with clinical research in cancer vaccine strategies has arisen. The Cancer Vaccine Clinical Trial Working Group already suggested two parallel tracts of investigation: trials focusing on proof of principle and efficacy trials(23). The former should aim to demonstrate immunological activity, the latter should be designed to show clinical benefit for the

postoperative DC vaccination.

finding(20-22).

patients.

The clinical outcome measures that really do have an impact on the patients' perspectives are overall survival and quality of life. The well tolerated vaccinations often result in a minimal interference with a good quality of life, that is actually only being threatened by disease progression but not substantially by the therapy itself(24). The ambulatory nature of the vaccination schedules further contribute to this low impact on the treatment burden for the individual patient. Especially this low toxicity profile adds to the merits of this therapy in the subgroup of long-term surviving brain cancer patients. Indeed, even more than the possible statistically important impact on median survival data, the substantial group of patients with malignant glioma, both WHO grade III and IV lesions, surviving for many years after vaccination, is the best advocate of this therapeutic approach. Both newly diagnosed, but even more strikingly, relapsed and multi-relapsed patients with high grade gliomas display survival periods of more than 4 to 9 years (manuscript submitted). Considering the classical definition of long-term survivor in the malignant glioma literature, being patients surviving 24 or more months after diagnosis (of primary disease or relapse), substantial numbers of patients, including up to 25% of the relapsed HGG patients undergoing DC vaccination are actually benefitting this opportunity. These long-term survivors are not only an encouragement for this –even not fully mature immunotherapy approach, but also a source of scientificly important translational knowledge to learn more about the factors predicting this type of outcome after immunotherapy. Final scientific proof of efficacy can of course only be delivered by welldesigned, sound randomized controlled trials, several of which are currently running throughout the world.
