The Cancer Journal - Volume 8, Number 6 (November-December 1995)
On the probability of effective anticancer vaccines
The editor has asked me to express my "gut feelings" concerning the possible future efficacy of anticancer vaccines.
There are, of course, two fundamentally different strategies for vaccine use. Traditionally, vaccines were administered to prevent the occurrence of disease rather than as therapy. Vaccines may well prove successful for the prevention of certain cancers, especially those cases which intimately involve a viral agent and it may be a semantic quibble to consider whether these vaccines are anticancer or antiviral. In fact, prevention of hepatoma is probably already occuring in response to antihepatitis vaccines. Also promising may be the use of embryonic tissues for vaccination (1, 2).
The renewed interest in vaccines as therapy stems from evidence that all tumors, even so-called spontaneous tumors, may indeed possess detectable antigens of at least moderate tumor specificity (23). Although the reaction to these antigens may seem very weak, immunogenicity can often be induced by a variety of means. The important observation is that modified tumor cells may induce immunity against the growth of the unmodified and seemingly nonimmunogenic parental cells (3). Globerson and Feldman, using DMBA-induced sarcomas in C57Bl mice, were probably the first to show that seemingly nonimmunogenic tumor cells may respond to immunization with immunogenic cells (4). In this system, the tumors regularly became, during serial passage in syngeneic animals, overtly nonimmunogenic, but they could, nonetheless, be inhibited by prior immunization with cells of an earlier generation transplant of the same tumor (note that this regular progression, during passage, to apparent nonimmunogenicity is not seen in many tumor-transplant systems).
The above observations provide a rationale for the pursuit of therapeutic vaccines; why then is my own optimism rather subdued? My caution stems from my contrariant view of the role of immune mechanisms in tumor biology. Despite the higher incidence of some tumor types in immunocompromised humans, I believe the evidence shows that the immune reaction may usually stimulate rather than inhibit the growth of untransplanted primary tumors (5, 6, 7).
When tumors were induced in mice by a modest dose of the chemical carcinogen 3-methylcholanthrene, the 154 tumors so induced exhibited, by classical immunization-challenge tests, a wide range of immunogenicities (8). The latency of arisal and the growth rate in the primary host were affected by the immunogenicity of the particular tumor; the tumors of shortest latencies were of intermediate immunogenicities. Most of the tumors, however, were distinctly of either a greater or a lesser immunogenicity than appeared optimal for rapid tumor arisal. In the complete absence of immunity, a similar spectrum of immunogenicities was unrelated to latency.
If most human tumor systems are analogous to this mouse model, within most human systems a portion of the tumors of any given type would induce a higher, while the remainder would induce a lower than a growth-maximizing level of immunity. The tumors of lower than optimal immunogenicity would be expected to arise more slowly in immunocompromised patients and would presumably tend to grow faster if the immune response were intensified by an effective vaccine (unless, of course, the intensification were very great). Conversely, it could be predicted that those tumors of higher than optimal immunogenicity might appear more rapidly in immunocompromised patients and respond to vaccine therapy by growing more slowly, ie., a vaccine could be efficacious for tumors of this category. However, among a patient population of tumors, the patients with tumors of lower than optimal immunogenicity and the patients with tumors of higher than optimal immunogenicity would tend to offset each other and it might, therefore, be difficult to see, epidemiologically, an effect of either vaccination or of immunosuppression on tumor incidence. Thus, successful and safe vaccine therapy may require a priori knowledge of whether a patient's tumor arouses an immune reaction that is more than or less than the level that is optimal for that particular tumor's growth.
Until recently, the incidences of most tumor types appeared to be unaffected by immunodeficiency in the human host, as the above mouse model predicts would be the case. The major exceptions, among organ transplant patients, were certain lymphomas and skin carcinomas and, among AIDS patients, Kaposi's sarcomas, all of which appeared in greater than expected numbers (9). The high incidence of lymphomas is not interpretable since these are tumors of the immune system per se; their high incidence might be unrelated to the failure of a putative immunological surveillance. The other exceptional tumors may be, on average, of higher immunogenicity in normal hosts than is optimal for growth stimulation, owing, perhaps, to unusually strong etiological stimuli (10); alternatively, the relative resistance to the appearance of these tumors in normal subjects could be the result of immunity to etiological viruses, if such exist. The exceptional tumors, whose incidence is elevated in immunodepressed patients, may, in reality, be stimulated rather than inhibited by the immune reactions they induce. Even if the immune reactions were greater than optimal for tumor growth, growth could, nonetheless, have exceeded what might have occurred in the total absence of immunity (5). Thus, it is possible that most primary tumors, at least early in their progression, are actually dependent upon some immune reaction and might fail to grow without it (5, 6, 7).
Other than the few exceptions already noted, the incidence of human tumors appeared, until recently, to be little affected by immunodepression. Now it has been discovered that the incidence of breast cancers, among almost 26,000 immunodepressed women with heart or kidney transplants, was markedly below expectation (11). This finding could have explanations other than diminishment of the postulated stimulatory effect of a normal immune reaction on breast tumor growth, but the result does parallel findings in mouse breast cancer (11). If the lower breast tumor incidence were indeed caused by the partial depletion of a growth-stimulating immune reaction, breast tumors in normal subjects must induce, on average, either a nearly optimal or a lower than optimal reaction for tumor growth. If nearly optimal, augmenting the immune reaction by vaccination would be expected to produce a relative inhibition of growth in the majority of tumors, but if lower than optimal, as may often the be the case, immunoaugmentation with a vaccine might more often accelerate tumor growth.
The same study of almost 26,000 immunodepressed women discovered no tumor type, other than breast, that showed a lower than expected incidence; in fact the trend among the other types, apart from those exceptionally immunogenic types that have already been mentioned, seems to have been toward a slightly elevated incidence (11). If the elevation were not artifactual, perhaps somewhat more than 50% of the tumors of each type produced, in response to antigens such as "oncofetal antigens" (2), an immune reaction that, even if weak, may have been greater than was optimal for growth of that particular tumor type. If so, immunodepression would be expected to produce the slightly elevated tumor incidence that was actually observed. Effective vaccination might, in circumstances, inhibit a majority of these tumors, but accelerate others.
Not only may most tumors be dependent, at least early in their evolution, upon a suitable level of immune response, but the immune response may also favor progression. In Hammond's experiments, the speed and degree of progression among carcinogen-induced tumors was directly correlated with the immune capacities of the animals, ie., the greater the immune capacities of the tumor hosts the more rapid and profound was the biological progression (12). The "Hammond effect" may explain the seemingly paradoxical observation that those thin "radial growth phase" cutaneous human melanomas, that are ordinarily completely curable, may carry a poor prognosis if, and only if, they show partial regression (13, 14, 15, 16).
A practical problem associated with vaccine development could be the common use of transplanted tumors as targets. Transplanted tumors may be more vulnerable to immune attack than are de-novo untransplanted targets (5) and may thus give a false impression of the efficacy of a therapeutic vaccine.
A further problem with vaccination is the notoriously heterogeneous and labile nature of cancer cells (17); they may adapt, in time, to whatever environment vaccination can induce. Even successful vaccine therapy may, unfortunately, be of only temporary benefit.
Thus, my "gut feeling", which I direct to both vaccine enthusiasts and vaccine skeptics alike, was well expressed in the words of Oliver Cromwell, in whose time (1599-1658) the bowels were regarded as the seat of tender emotions such as compassion: "I beseech you, in the bowels of Christ, think it possible you may be mistaken".
Richmond T. Prehn
407 Lake Avenue West, Kirkland, WA 98033, USA,
1. Moroson, H, Ioachim, HL. Protection by grafts of embryonal rat tissues (teratomas) against induction and transplantation of malignant tumors. Cancer Res. 55, 3664-3668, 1995.
2. Coggin, JH Jr. Classification of tumor-associated antigens in rodents and humans. Immunol Today. 15, 246-247, 1994.
3. Boon, T. Toward a genetic analysis of tumor rejection antigens. Adv Cancer Res. 58, 177-210, 1992.
4. Globerson, A, Feldman, M. Antigenic specificity of benzo(a)pyrene-induced sarcomas. J Nat Cancer Inst. 32, 1229-1243, 1964.
5. Prehn, RT. Stimulatory effects of immune reactions upon the growths of untransplanted tumors. Cancer Res. 54, 908-914, 1994.
6. Prehn, RT, Prehn, LM. The flip side of tumor immunity. Arch Surg. 124,102-106, 1989.
7. Prehn, RT, Lappe, MA. An immunostimulation theory of tumor development. Transpl Rev. 7, 26-54, 1971.
8. Prehn, RT, Bartlett, GL. Surveillance, latency, and the two levels of MCA-induced tumor immunogenicity. Int J Cancer. 39, 106-110, 1987.
9. Penn, I. Tumors of the immunocompromised patient. Annu Rev Med. 39, 63-73, 1988.
10. Prehn, RT. Relationship of tumor immunogenicity to concentration of the oncogen. J Nat Cancer Inst. 55, 189-190, 1975.
11. Stewart THM et al Incidence of de novo breast cancer in women chronically immunosuppressed following organ transplantation. Lancet 346, 796-798, 1995.
12. Hammond, WG, Benfield, JR, Tesluk, H et al. Tumor progression by lung cancers growing in hosts of different immunocompetence. Cancer J. 8, 130-138, 1995.
13. Ronan, SG, Eng, AM. Thin malignant melanomas with regression and metastases. Arch Dermatol. 123, 1326-1330, 1987.
14. Gromet, MA, Epstein, WL, Blois, MS. The regressing thin malignant melanoma. A distinctive lesion with metastatic potential. Cancer, 42, 2282-2292, 1978.
15. Tefany, FJ, Barnetson, SCR, Halliday, GM, et al. Immunocytochemical analysis of the cellular infiltrate in primary regressing and non-regressing malignant melanoma. J Investig Dermatol. 97, 197-202, 1991.
16. Cook, MG. The significance of inflammation and regression in melanoma. Virchows Arch [A]. 420, 113-115, 1992.
17. Prehn, RT. Analysis of antigenic heterogeneity within individual 3-methylcholanthrene-induced mouse sarcomas. J Nat Cancer Inst. 45, 1039-1045, 1970.