Immunotherapy and the development of cancer vaccines

Surgery has been the mainstay of cancer treatment since the turn of the century. William Coley, a New York surgeon, pioneered the modern use of immunotherapy with his eponymously named toxins, which were developed following the observation that patients who developed severe postoperative septicemias occasionally had resolution of residual disease. He was also one of the first to use radiotherapy in the treatment of cancer. Coley’s discovery of the use of the toxins and immunotherapy has been covered on many occasions [1]. However, the main lesson learnt from the development of these toxins is that the type of toxins, dose and schedule used are supremely important with regards to efficacy, and are equally as likely to be important in developing cancer vaccines. Chemotherapy and hormonally based treatments have subsequently been the mainstay of standard cancer treatments for nonresectable and metastatic cancer.

Although the use of toxins lapsed following the introduction of radiotherapy (followed by chemotherapy and, subsequently, endocrine treatment), the belief that the immune system could be harnessed to treat cancer has persisted for several decades. The use of nonspecific immunotherapy, such as bacillus Calmette-Guerin (BCG), as well as tumor antigen-specific treatments, such as cell-based vaccines, have provided numerous anecdotal evidence that vaccination might be useful in the prevention of metastatic disease.

The history of cancer vaccines has been very disappointing, with many encouraging Phase II studies failing to be replicated in randomized studies. However, the concept that cancer vaccines could delay the progression of metastatic solid tumors has been underscored by two major studies in colorectal and renal cancer, both of which used autologous, cell- based vaccines as the mainstay of therapy [2,3]. Making autologous cancer vaccine treatments, however, is labour intensive, expensive and has too many variables for the approach to be truly characterized as a therapeutic product. Several studies have suggested that allogeneic cell lines may be able to replace this approach with a reproducible product, and the initial fear that allogeneic cell line approaches would not be as good as autologous has been demonstrated to be unfounded, as allogeneic cells can often produce better responses than autologous cells in preclinical models [4–6]. This only occurs in the presence of shared antigens and has been most clearly demonstrated in melanoma-based models. The fact that allogeneic cells often perform better than autologous may be related to the fact that allogeneic presentation of shared antigens is able to induce a danger signal and break tolerance. Although initially demonstrated in melanoma-based models, this has also been reported to extrapolate to other tumor cell types, such as prostate cancer.

The major model system for immunotherapy has for a long time been melanoma, with many shared antigens and several very encouraging Phase II studies, both for autologous-based, as well as allogeneic systems. Allogeneic-based systems have been based on either whole cell lines or tumor lysate preparations and, whereas both have had very encouraging Phase II data, all have failed in big randomized studies. The most recent results have involved a variety of approaches in both autologous and allogeneic systems, with no evidence of significant survival benefit in Phase III studies. Some of these approaches have evolved into major commercially backed candidates, by companies such as Antigenics, Inc. (NY, USA), Maxim Pharmaceuticals, Inc. (CA, USA) and CancerVax™ (Ca, USA), to name but a few examples. CancerVax, which used BCG and an allogeneic cell-based vaccine, has been the most recent study candidate to report failure to achieve the primary end point in a Phase III study [101].

However, the focus of many immunotherapies and cancer vaccines has been melanoma, and the number of failures has been extremely disappointing. Although melanoma may be an excellent example for research-based immunotherapies, due to the shared antigens and demonstration of effective clinical responses in the presence of strong immune responses, the failure to demonstrate clinical efficacy in large randomized studies may well be due to the fact that melanoma is a very variable and chaotic disease with regards to its presentation. It is therefore pertinent to note that the survival benefits that have been demonstrated to be significant have occurred in nonmelanoma cancers, such as colorectal and renal. More recently, the author and others have proposed that prostate cancer would be an ideal candidate for immunotherapy and cancer vaccines. The first rationalization is that prostate cancer has several prostate-specific antigens (PSAs), which are eponymously named. Second, prostate cancer progresses in a much less chaotic fashion than melanoma, with most disease progression occurring through the lymph nodes and then the bone, whereas melanoma is characterized by the ability to spread to any organ following lymph node involvement. Moreover, melanoma can progress and express a disease doubling time far in excess of any other tumor. In retrospect, it would make more sense to target cancer vaccines at tumor types that progress in a much more predictable fashion, namely prostate cancer. Recently, Dendreon (WA, USA) planned a cancer vaccine approach for targeting prostate cancer using an autologous approach, whereby dendritic cells are harvested from the patient and then pulsed with a prostate antigen-based vector in the hope that it would prevent time-to-disease progression. Although the primary end points of these studies were not reached in randomized studies, a secondary end point of survival (probably more important) has recently been achieved in these studies, specifically combining two randomized studies, and an application for FDA approval has been made [102]. Other approaches have also produced very encouraging data, including the use of whole-cell vaccines from Onyvax (London, UK) in a Phase II study [7], whereby the reduction in the rate of rise of PSA velocity correlated with time-to-disease progression, as well as PSA responses, using a virus-based vector incorporating PSA antigens from Therion (MA, USA) [8]. These data strongly suggest that prostate cancer may be a much better target than melanoma.

Lessons learned from developing cancer vaccines in the treatment of solid tumors are that autologous approaches to colorectal, renal and prostate cancer have produced a survival benefit, whereas the more obvious target of melanoma has not.

Autologous-based approaches are very much proof-of-principle, and unlikely to produce long-term, viable products. For cancer vaccines to be successful, it is more likely that prostate, colorectal and renal cancer will produce registerable products compared with melanoma for the aforementioned reasons.

However, in the current regulatory environment, alternatives to autologous-based vaccines are likely to be more attractive. Lessons from both preclinical and clinical models strongly favor the use of a multiantigen approach, as peptide-specific approaches often lead to disease escape with downregulation of the targeted tumor antigen. There are, therefore, two main approaches in this regard: either to use allogeneic cell-based vaccines with multiple antigens, several of them probably unknown, or to select several antigens and put them together as a polyepitope vaccine. The advantage of a cell-based vaccine is that one can select antigens that might be pertinent, and that have not yet been discovered. The recent ability to detect hundreds of tumor antigens is not necessarily helpful, as it is not known which are the most effective in preventing disease progression. An allogeneic cell-based approach allows for selection of cells from future metastatic sites, such as lymph nodes and bone metastases, in the case of prostate cancer, as opposed to having to select specific antigens for a limited epitope vaccine.

Nevertheless, there are many other factors that need to be taken into consideration. Vaccine-induced immune responses can be either cell-mediated dominant (T helper [Th]1) or humoural-mediated dominant responses (Th2 dominant). Whereas tumor antigen-specific high-affinity antibody responses have been demosntrated to be beneficial to the point of registration for monocolonal antibodies directed at CD20 (Rituximab^®), HER2-neu (Herceptin^®) and vascular endothelial growth factor (Avastin^®), the protective response of several vaccines have been demonstrated to be more likely to be cell-mediated or Th1-mediated. These do not necessarily have to be tumor antigen-specific in their response. A good example of this is the Mycobacterium vaccae vaccine, which is closely related to the BCG tuberculosis vaccines, and has been demonstrated to be able to convert a humoral-dominant response to a cell-mediated response, and this can translate into survival benefit in melanoma [9]. This approach has also been demonstrated to be useful in other tumor types, such as lung cancer, whereas single-center studies have reported a survival benefit when M. vaccae vaccine has been added to chemotherapy [10]. Unfortunately, a multicenter randomized study failed to confirm the survival benefit reported, and the vaccine was dropped for cancer treatment as a result [11]. However, there were many differences between the randomized Phase II and the multicenter Phase III study, which could explain many of the differences observed in the results. In the case of M. vaccae in the randomized lung cancer study, there were many differences in the protocol and content of the study with regards to dose, schedule and site of injection, which again could explain the differences observed. However, a more important point is that the primary end point taken from the Phase II studies was survival, yet the secondary end points of quality of life and reduction in the side effects of chemotherapy were demonstrated to be significantly different, in favor of the vaccine arm, in the multicenter study [11]. Had these been the primary end points, it is possible that the vaccine could have been registered; minor differences in the dose schedule and protocol could have made a major difference in survival, should they have been adhered to.

The current focus of vaccines is to use dendritic cells as the presentation, á la the Dendreon approach. Whereas the Dendreon approach may be difficult to commercialize on a large scale, it has invaluable applications as a research tool to optimize cancer vaccine protocols. It is clear that the details of antigen loading, maturation, presentation and memory cell expansion are crucial in the development of cancer vaccines, and that dendritic cell approaches lend themselves to the dissection of the details of these aspects. Analyses of these approaches suggests that immunotherapeutics is not just a case of ‘performing’ or ‘utilizing’ immunotherapy, but that it may be incorporated as a sequential managed therapy in the treatment of cancer. There is increasing evidence that patients who have some form of immunotherapy are more susceptible to responses from radiotherapy, endocrine therapy and chemotherapy. In addition, there is recent evidence to suggest that the sequentiality of these approaches is very important: a study by Arlen and colleagues demonstrated that administering a vaccine for prostate cancer followed by endocrine treatment is significantly more effective than the reverse (endocrine treatment followed by vaccine treatment) [8]. This has major implications for analysis of many ongoing vaccine trials.

The analyses of many ongoing cancer vaccine studies, as well as those recently completed, suggest that, first of all, the tumor type is crucially important. Perhaps the first lesson is to avoid melanoma as it is too diverse a disease and too chaotic in its progression to be able to give a statistically significant survival advantage with current vaccine approaches. The fact that it is very sensitive to tumor antigen adjuvant combinations has been clearly demonstrated in the ganglioside vaccine studies in which QS21 was substituted as an adjuvant for BCG. It is highly likely that the studies that failed were over-adjuvanted, whereas BCG stimulated the cell-mediated immune system in a more appropriate manner. Although BCG has been very controversial and universally assumed to be nonsignificant in providing any survival advantage to melanoma, it may be effective enough to have interfered with the CancerVax vaccine trials in which it was used as a placebo; it is possible, had it been a randomized study with no treatment arm at all, that the CancerVax would have been effective. In this regard, the failure of a large multicenter study to confirm very impressive results from a Phase II study should not be totally disregarded, as Herceptin, if used on all advanced breast cancer patients, would have produced a similar result. Thus, up to one-third of the tumors may well have benefited from the vaccine if a suitable biomarker could have been identified to select the vaccine-susceptible patients. The need for suitable biomarkers in future vaccine studies is crucial, as it is clear that vaccines perform much better in the adjuvant setting, and that the duration of disease, for colorectal cancer in particular, is too long for most commercially funded studies to continue [12].

In conclusion, autologous vaccines have provided the sound infrastructure to allow the Roman arch to be built. However, they are too impracticable to provide a marketable product. Paradoxically, the autologous dendritic cell approach will allow us to identify the signals that need to be provided to the immune system to allow the process and maturation of self-antigen presentation and expansion to the immune system. Already, several studies are reporting that chemotherapy and dendritic cell therapy are probably synergistic, which may be due reasons other than the already documented ability of chemotherapy to remove the T regulatory suppressor cells. In addition, the synergy observed with other modalities, such as radiotherapy and endocrine therapy, provide many stimulating possibilities as to how to incorporate immunotherapy into the mainstream of cancer treatment.

In light of the above, it should be remembered that we now have two extremely effective cancer vaccines in a prophylactic situation. The hepatitis B vaccine and the more recent human papilloma-based vaccines are capable of reducing the ‘world cancer’ incidence of vaccines by up to 20%, as hepatitis B and human papilloma viruses are major causes of liver and cervical cancer in the developing world. The challenge for the future is to be able to harness this incredible effectiveness into inducing responses against nonviral tumor antigens of the common solid tumors, with a realistic goal of delaying the progression of these tumors, if not halting them, in association with other modalities such as chemotherapy, endocrine therapy and radiotherapy.

It is only a few years ago that several monoclonal antibody Phase III studies failed in the presence of encouraging Phase II studies. The efficacy of cancer vaccines selected for Phase II studies suggests that it is only a matter of time before they are accepted as a mainstream modality in the treatment of cancer and, moreover, that they will be used in synergy with other accepted modalities.

In summary, the author believes that it is reasonable to conclude that it is not ‘if’ but ‘when’ cancer vaccines are going to be accepted as a major part of the therapeutic arm armamentarium against the common solid tumor.