Interesting article and editorial in today's JCO seeing we were talking about cancer vaccines:
Journal of Clinical Oncology, Vol 22, No 21 (November 1), 2004: pp. 4240-4243
© 2004 American Society of Clinical Oncology
EDITORIALS
Toward a Glioblastoma Vaccine: Promise and Potential Pitfalls
Howard A. Fine
Neuro-Oncology Branch, National Cancer Institute, National Institutes of Neurological Disorder and Stroke, National Institutes of Health, Bethesda, MD
Despite technical improvements in neurosurgery and radiation therapy, and countless clinical trials of various chemotherapeutic regimens, the treatment of malignant gliomas remains largely suboptimal. In particular, there has been little improvement over the last two decades in the prognosis of patients with glioblastomas, the most common and deadly type of glioma. Although recent data suggest that the addition of temozolomide to radiotherapy improves long-term outcome, the overall median survival for patients with glioblastoma treated with combined therapy is still less than 15 months.1 Clearly, novel treatment strategies are desperately needed.
In this issue of the Journal of Clinical Oncology, Steiner et al2 report the preliminary results of one such novel approach. These investigators explored the use of immunotherapy as an adjuvant to standard radiotherapy in patients with glioblastoma. In this trial, selected patients with glioblastoma were treated with radiotherapy followed by a vaccine consisting of autologous cells grown from their resected tumor samples and then infected with an avian paramyxovirus (Newcastle Disease Virus) used as a nonspecific immunostimulant. The vaccines were then administered every 3 to 4 weeks to a maximum of eight treatments. During the vaccination period and until tumor progression, patients received no additional treatment. Steiner et al reports an impressive median progression-free survival of 40 weeks and a median overall survival of 100 weeks, significantly better then their historical control group (26 and 49 weeks, respectively). Additionally, the 2-year survival rate in the vaccine group was 39%, compared to 4% in the control group. The investigators attribute these improved outcomes to a vaccine-mediated antitumor response.
The outcomes of the patients in this trial are clearly better than one would expect from an unselected group of patients with glioblastoma, and the report by Steiner et al2 is therefore of potential importance. Nevertheless, the authors' conclusion that this therapy "appears to be feasible, safe, and seems to improve the prognosis of patients with glioblastoma" is premature. Progress in medicine, and particular in cancer therapy, generally occurs in sequential and measured increments. Claims of extraordinary leaps forward, particularly for strategies that have proved unsuccessful in the past, require a careful assessment by the critical reader of both the scientific rationale for the therapy and for potential confounding variables that may have erroneously accounted for the perceived effect.
Harnessing the potential of the immune system for specifically attacking tumor cells has been the holy grail of cancer therapy for decades. There have been tremendous gains in our understanding of molecular and cellular immunology that have lead to an even greater interest in tumor immunotherapy. A quick PubMed search of the term "tumor immunotherapy" returns more than 33,000 titles, most published over the last 15 years, and many demonstrating impressive results in various animal models of cancer. These exciting reports in preclinical models have unfortunately not translated to clinical successes. Although there have been several dramatic examples of the potential ability of the immune system to induce dramatic tumor regressions through the administration of large numbers of immune effector cells ("passive immunization"), there are few definitive examples of cancer vaccines ("active immunization") having substantial clinical efficacy.3 Thus, the results reported by Steiner et al are unexpected. This is particularly true when one considers that immunotherapeutic approaches for CNS tumors may pose even greater challenges than for their systemic counterparts.
It has long been known that the CNS is a relative immunologic sanctuary; an environment where infections and tumors may progress with relative impunity from an effective immunologic attack. The mechanistic basis for the sanctuary status of the CNS remains largely unknown, but is probably more related to the molecular make-up of the interstitial microenvironment of the CNS rather than the oft-cited blood-brain barrier. Gliomas, in particular, represent difficult immunologic targets in that these tumors tend to express large amounts of transforming growth factor-beta, prostaglandin E2, and interleukin 10, all potent immunosuppressants. There is also the question of the existence of glioma-specific antigens; although several molecules have been suggested as being potential candidate antigens, such as the peptide corresponding to the epidermal growth factor receptor vIII deletion region, no definitive glioma-specific antigens have been identified to date. Finally, the promising results reported in several preclinical studies of brain tumor immunotherapy must be viewed cautiously, as the most common glioma cell lines used in such experiments are not truly syngeneic and therefore elicit allogeneic immunoresponsiveness. Furthermore, such orthotopic models require traumatic introduction of tumors cells into the brain, thereby perturbing the normal microarchitecture that may be contributing to the immunologic sanctuary status of the CNS.
Within this context, one must read the report by Steiner et al, who utilized an older approach of re-administering whole cultured autologous tumor cells that have been infected with a virus as a nonspecific immunogen, cautiously. This general strategy has failed in the past for other types of tumors because of its inability to generate sufficient numbers of in vivo tumor-specific cytotoxic T lymphocytes (CTLs). Although the authors argue that the use of the Newcastle Disease Virus as a nonspecific immunostimulant will result in better immune responses than prior viral vaccine approaches, there remains the issue of using a whole tumor cell vaccine, opening up the potential for presenting unwanted normal versus tumor-specific antigens. An additional concern with this approach is that the vaccine is made up of cells selected for their ability to grow under in vitro conditions rather than the potentially very different clonogenic cells that grow in vivo. Finally, the design of this trial makes the interpretation of radiographic responses difficult. Although it is highly likely that the most beneficial time to administer a vaccine would be at the time of minimal residual tumor burden in the postradiation setting, the use of the vaccine following radiotherapy does not allow determination of true vaccine-mediated objective radiographic responses rather than delayed radiotherapy responses. Thus, one may ask whether Steiner et al truly induced an antiglioma immunologic response. The authors offer three pieces of evidence: the generation of a delayed-type hypersensitivity reaction to nonvirally infected autologous tumor cells in the majority of vaccinated patients; increased gamma interferon production from cocultured memory T-lymphocytes; and an increased number of CD8 positive (CD8+) lymphocytes within the resected tumors of vaccinated patients whose tumors had progressed following treatment.
The presence of delayed-type hypersensitivity, gamma interferon-producing, and intratumor CD8+ lymphocytes do demonstrate that some type of immunologic response was elicited to the injected cells, which is not necessarily a surprising finding given that the immune system may not generate adequate tolerance to certain antigens expressed within the protected environment of the CNS. One could argue, however, that the observed immunologic response was not an effective tumor-specific cytotoxic response. The fact that there were many more CD8+ lymphocytes found in the tumor specimens from vaccinated patients whose tumors progressed around the same time as the control patients (42 v 32 weeks, respectively) argues that the abundance of CD8+ lymphocytes in those tumor specimens were not exerting an effective antitumor effect. It would be optimal to determine whether these lymphocytes have clear antiglioma activity by generating tumor-specific lymphocyte lines from either the tumor-infiltrating lymphocytes within the resected tumor specimen or from the patient's blood. Such cell lines could then be assayed for their ability to lyse the autologous glioma cell lines grown in vitro. Possibly even more important would be the demonstration that the immunologic response was generated against a tumor-specific, or at least tumor-selective, antigen.
If, for argument's sake, an effective antiglioma immunologic response was not generated by the vaccine, then what might have accounted for the excellent clinical outcomes of many of the patients in the Steiner et al study?2 The most likely answer is inadvertent patient selection bias. Although there is little question that glioblastoma is a uniformly lethal tumor, there is still significant individual patient heterogeneity related to overall survival. Numerous clinical trials over the last three decades have defined several prognostic factors that are strongly associated with longer survival of patients with glioblastoma, including younger age, better performance status, minimal postoperative radiographic residual disease, normal mental status as assessed by the mini mental status examination, and lack of certain other neurologic symptoms such as motor dysfunction and headaches.4 The most compelling evidence for the power of these prognostic markers come from the recursive partitioning analyses (RPA) of data from more than 1,500 patients with high-grade gliomas accrued to three different Radiation Therapy Oncology Group trials.5 The Radiation Therapy Oncology Group RPA segregated patients with high grade gliomas into six different prognostic categories, four of which included patients with glioblastomas (RPA class III-VI). The heterogeneity of patient outcome is exemplified by the fact that patients within RPA class VI have an overall survival of 4.6 months and a 2-year survival rate of 4% compared to patients in RPA class III (including those with glioblastoma) who have a median survival of 17.9 months and a 2-year survival rate of 35% (similar to the vaccinated patients in the Steiner et al study).
There are reasons to believe that the selected patients in the Steiner et al vaccinated group may have been a group of patients with particularly favorable prognostic characteristics. First, these patients were highly selected merely by virtue of the small number of patients treated. Over the study period, 155 patients with gliomas underwent surgery at the University of Heidelburg (Heidelburg, Germany), whereas the vaccine was apparently "offered to only 35 patients" (23% of all patients) of which only 23 were ultimately evaluated (15% of all patients). By contrast, the comparative control group consisted of 56% of all patients operated on at the University of Heidelberg. Criteria for the selection of this small subset of patients, other than protocol eligibility, are not clear. These 23 assessable patients did indeed possess a number of positive prognostic factors. For example, the protocol stipulated that patients must not be taking dexamethasone following completion of the postsurgical radiation therapy. Many patients with glioblastoma require long-term glucocorticoids following radiation therapy for control of neurologic symptoms and headaches that occur secondary to cerebral edema from residual tumor. Thus, patients who no longer require dexamethasone are generally those patients with minimal or no neurologic symptoms and minimal residual tumor; two very important positive prognostic factors. The pattern of treatment for patients with tumor recurrence also suggested that the vaccine group may have been a more favorable group of patients, in that 65% of such patients received chemotherapy following recurrence, whereas only 48% of the control group was treated with chemotherapy. Generally, it is the younger patients with better performance status that are offered chemotherapy following tumor recurrence. Likewise, nearly twice as many patients in the vaccine group underwent repeat surgical resection following tumor progression compared to the control group (35% v 18%, respectively). Again, only patients with some hope of prolonged survival and better prognoses are usually considered for resection following recurrence of a glioblastoma. The authors' own data demonstrate the power of patient selection for potentially skewing clinical outcome data in that their "best 23" historical control patients demonstrated a median, 1-, and 2-year survival roughly equivalent to the vaccinated group of patients (100 v 88 weeks; 91% v 100%; and 39% v 43%, respectively).
Despite the reservations discussed above, the data from Steiner et al2 are tantalizing and certainly worthy of further investigation. One could argue that Steiner et al's vaccine approach should be immediately evaluated in a randomized trial given the promising preliminary data and the paucity of effective therapies for glioblastoma. Indeed, Steiner et al suggest just such a trial. An appropriately designed trial for this vaccine approach, however, would be both large and expensive, as it would need to accrue patients with glioblastoma for whom autologous tumor cells are grown under good manufacturing practice conditions with all the appropriate quality checks. Such a trial would also need to selectively accrue the relatively small subset of patients with better prognostic features who do not require glucocorticoids following radiation. Another consideration is that there may be a new standard of care in the treatment of glioblastoma based on the recently completed European Organization for Research and Treatment of Cancer randomized trial showing a significant survival benefit for patients treated with temozolomide both during and after radiation (26% had 2-year survival).1 Given that chemotherapy can not be given during the vaccination period secondary to the associated immunosuppression, one must ask whether it is ethically justifiable to design a treatment arm that does not include temozolomide based purely on the preliminary data from 23 patients.
I believe it is important to generate more correlative laboratory data supporting the biologic rationale behind the vaccine strategy and more clinical efficacy data, such as consistent radiographic evidence of an antitumor response, before subjecting a large number of patients to a treatment arm that may be inferior to the new standard of care. In addition to generating additional data demonstrating a tumor-specific antitumor effect, it would be preferable to conduct a formal phase I trial to establish an optimal dose schedule based on immunologic end points, such as the generation of tumor-specific CTLs. Such a trial should also be designed to objectively and carefully assess vaccine safety relative to both acute and delayed neurotoxicity. Since the Steiner et al vaccine involves intact glioma cells, it is conceivable that a cytotoxic immunologic response will be generated to antigens found both on gliomas and normal glia, such as astrocytes and oligodendrocytes, which could lead to significant and potentially lethal demyelinating encephalopathic syndromes such as multiple sclerosis or acute disseminated encephalomyelitis. If these early translational studies do prove that an autologous tumor-specific CTL response is being safely generated, then a multicenter expanded phase II trial using the optimal vaccine dose schedule should be performed with outcome comparisons to a rigorously defined historical control group. Should the vaccine still appear promising following completion of these studies, then a large randomized trial is clearly justifiable.
Patients diagnosed with glioblastoma and their families face a terrifying prognosis of progressive neurocognitive decline and very short survival with a paucity of effective treatment options. As clinical scientists and as the physicians who care for these desperate individuals, it is our scientific and moral responsibility to bring to them, in a timely fashion, novel therapies that are tested in the most scientifically rigorous way possible.
Author's Disclosures of Potential Conflicts of Interest
The author indicated no potential conflicts of interest.
REFERENCES
Stupp R, Mason WP, Van Den Bent MJ, et al: Concomitant and adjuvant temozolomide (TMZ) and radiotherapy (RT) for newly diagnosed glioblastoma multiforme (GBM). Conclusive results of a randomized phase III trial by the EORTC Brain & RT Groups and NCIC Clinical Trials Group. Proc Am Soc Clin Oncol 23:1, 2004 (suppl; abstr 2)
Steiner HH, Bonsanto MM, Beckhove P, et al: Antitumor vaccination of patients with glioblastoma multiforme: A pilot study to access feasibility, safety, and clinical benefit. J Clin Oncol 22:4272-4281, 2004[Abstract/Free Full Text]
Dudley ME, Wunderlich JR, Robbins PF, et al: Cancer regression and autoimmunity in patients after clonal repopulation with antitumor lymphocytes. Science 298:850-854, 2002[Abstract/Free Full Text]
Buckner JC: Factors influencing survival in high-grade gliomas. Semin Oncol 30:10-14, 2003 (6 suppl 19).
Curran WJ Jr, Scott CB, Horton J, et al: Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst 85:704-710, 1993[Abstract] |
