In organizing this brief, but informative review of human melanoma vaccines, Haigh et al have provided an important service to the readers of oncology and are to be commended for their efforts. Their descriptions of the variety of vaccine technologies currently under development and their assessment of the strengths and weakness of each are, for the most part, fair and conservative. What concerns me is that their depiction of the melanoma vaccine landscape in shades of gray has resulted in a loss of contrast and a blurring of focus.
Haigh et al emphasize our ignorance in dealing with the complexities of the antitumor immune response. It is all too true that 30 years of research in human immunotherapy have generated a good deal more heat than light. However, clinical approaches to immunotherapy have their origins in animal models, and it therefore seems appropriate to base an assessment of current vaccine technologies on the extensive body of information derived from these experimental systems.
Lessons Learned From Murine Tumor Models
Murine tumor models have quite clearly established that tumor immunity is mediated predominantly by T-lymphocytes. Other immunologic cells, such as natural killer cells and macrophages, are of secondary importance. There is little experimental evidence to indicate that antibodies destroy tumors. Thus, vaccines that do not induce a tumor-directed T-cell response are very unlikely to be clinically effective.
A Medline search reveals thousands of publications describing protection against experimental tumor challenge by prior administration of a tumor vaccine. In the overwhelming majority of these articles, the vaccines were derived from autologous tumor cells. Successful immunization with allogeneic tumor cells has been reported only rarely. Moreover, it has proved exceedingly difficult to induce antitumor effects by immunizing mice with chemically characterized generic vaccines. In fact, the classic work of Prehn, one of the founders of modern tumor immunology, has shown that multiple tumors induced by the same carcinogen in the same strain of mice are antigenically distinct and fail to provide cross-protection from one another.
Although these results do not preclude the possibility that generic vaccines may induce a therapeutically useful antitumor immune response against certain cancers, they certainly reduce the probability of obtaining such a response. We clinical tumor immunologists must all confront the skepticism of our fellow cancer researchers; however, proponents of allogeneic or generic vaccines have an additional theoretical burden to bear.
Haigh et al may be overly concerned about the logistic problems associated with producing autologous human tumor vaccines. An ineffective vaccine is not rendered more appealing by the ease with which it can be manufactured. On the other hand, once a vaccine has been proven useful, efficient methods for its production and distribution are likely to be rapidly developed.
Autologous Dinitrophenyl-Modified Vaccine
The melanoma vaccine developed in our laboratory consists of irradiated, autologous tumor cells that have been modified with the hapten, dinitrophenyl (DNP). This technology is based on the long-established finding that immunization with a hapten-modified protein, even one that is a normal body constituent, can induce immunity to the native protein. Thus, administration of DNP vaccine induces the development of a delayed-type hypersensitivity to autologous melanoma cellsboth DNP-modified and DNP-unmodified.
In melanoma patients with measurable metastases, the administration of a DNP vaccine is often followed by the development of inflammatory responses in metastatic sitesan effect that had not been previously demonstrated for other human tumor vaccines. The inflammatory responses are characterized by tumor infiltration with T-lymphocytes that represent novel T-cell clones not detectable in patients prior to immunization. In some patients, the DNP vaccine treatment induces regression of established metastases, particularly small-volume lung metastases. As noted by Haigh et al, a series of phase II studies that we have conducted in patients with resected, bulky lymph node metastases have produced 5-year survival rates that appear to be much higher than rates obtained with surgery alone.
Working out of a small research laboratory, we have been able to enter over 350 melanoma patients into studies of the autologous DNP vaccine. AVAX Technologies, Inc., a biotechnology company that has licensed our vaccine, has recently completed a good manufacturing processlevel facility in Philadelphia with the capacity to produce thousands of vaccines yearly. This quantity of vaccine is sufficient not only for conducting large clinical trials but also for commercial production of the vaccine should it be approved for marketing by the FDA.
The new facility has allowed AVAX to undertake a randomized trial comparing the autologous DNP vaccine with high-dose interferon-alfa (Intron A, Roferon-A), the standard approved postsurgical (adjuvant) therapy for patients with bulky, stage III melanoma. This trial is in progress, with over 20 clinical sites participating. Table 3 in the review of Haigh et al would be rendered more complete by its inclusion.
Other Autologous Vaccines
The article by Haigh et al also omits two other vaccine technologies that are based on autologous tumor cells. Tamura, Srivastava, and colleagues have demonstrated that immunization of mice with heat shock proteins derived from autologous, but not from allogeneic, tumors induces protective immunity; clinical trials are in progress.
Dranoffs group has produced provocative results with a vaccine consisting of autologous melanoma cells transfected with the gene for the cytokine, granulocyte-macrophage colony-stimulating factor (GM-CSF). The results of a phase I trial of this vaccine have been reported.
Randomized Trials May Not Provide All the Answers
Haigh et al have adopted the conventional view that the randomized trials of melanoma vaccines currently in progress will determine which vaccines are effective and which are not. That conclusion may be overly optimistic. After all, two large randomized trials of interferon-alfa as postsurgical therapy for high-risk melanoma conducted over a 10-year period have not resolved questions about its usefulness. Randomized trials will never relieve us of our responsibility to understand the biology of the treatments that we are testing and to assess these treatments against the background of established scientific principles.