Tobacco Documents Online

VIRUSES AND HUMAN CANCERS: ACHIEVEMENTS, PROBLEMS, PROSPECTS Guy de-Th6, M.D., Ph.D. International Agency for Research on Cancer 150 cours Albert Thomas, 69008 Lyon, France In animals a number of spontaneous tumours, such as lymphomas, leukemias, sarcomas, mammary carcinomas, are causally associated with RNA or DNA viruses. The pathogenesis of viral carcinogenesis in such experimental systems has been extensively studied, stressing the importance of genetic susceptibility, time of infection, diet and environmental factors in the success or failure to induce tumours with such viruses in animal populations. The molecular events leading to cell transformation by viruses has yet to be established but the necessity of integration (or at least close associa- tion) of the viral genome in the cellular genetic material appears a requisite for maintenance of the transformed state of tumorous cells. The role of chemical carcinogens in spontaneous tumours in animals has not been investigated systematically and this prevents us from comparing the respective role of chemical and viral carcinogens in natural.conditions. area to another, the cancer incidence for all sites varies by a factor of 3 at maximum (2). Chances exist, then, that once a given environmental carcinogen is as, if the risk for certain tumours varies from 1 to 100 or even 200 times from one that "the most promising approach to the control of the disease is to identify those factors and eliminate them," (1). Such conclusions, however, may be too optimistic to be caused by exposure to factors in the environment" with the logical conclusion by J. Higginson and recently expressed by J. Cairns, that "almost all cancers appear In Man, the situation is the opposite: epidemiological studies suggest that environmental factors are of critical importance in the relative frequency of many cancers in different parts of the world. If one takes the highest and lowest incidences for each tumour's site, one reaches the conclusion, discussed at this meeting removed, the cancer profile hX site may change in the population under study but thatthe all heartily support it. Since the pioneering work of Sir Percival Pott, we know that chemical carcinogens are of immediate importance in our environment - what about viruses? .cancer incidence of the population will only be brought down slightly. However, in any case, the 'cleaning' of our environment is a must for our society and we should The study of the relationship between viruses and human tumours is of theoretical and practical importance. one would like to know, for example, if tumours known to be virally induced in animals are also virally associated in Man. On the other hand such studies have a practical interest as immuno-prevention has success- fully been achieved in animals and should represent a practical goal in the not too distant future for Man. We shall briefly review: i) the epidemiological human tumours favouring a viral aetiology; ii) the criteria relationship between a viral infection and tumour development in the search for a human oncorna virus in leukemia, sarcomas iv) the problem of cervical carcinoma and viral factors; v) (EBV) and its relationship with two human tumours: Burkitt's nasopharyngeal carcinoma (NPC). characteristics of favouring a causal iii) the situation and breast carcinoma; the Epstein-Barr virus lymphoma (BL) and ~ 1. J. Cairns, The cancer problem, Scientific American, 233: 64-80, 1975 2. C.S. Muir, International variation in high-risk populations, In: Persons at high- risk for cancer, p. 293-305, Academic Press, 1975 82 r H n C 2 t t 4

I. EPIDEMIOLOGICAL CHARACTERISTICS OF HUMAN TITHOURS FAVOURING A VIRAL AETIOLOGY In both slow viral infections (3) and in virally induced tumours in animals, there are certain characteristics which should be borne in mind when approaching the study of human tumours: a) a long and silent time interval may elapse between primary infection and disease development; b) while infections occur in many individuals, only very few will develop the disease, stressing the importance of other factors (or 'co-factors') which are essential for the develop- ment of the disease, such as genetic susceptibility, environmental factors, dietary• and other cultural habits etc. c) the age of primary infection by the virus is of critical importance: usually the younger'the individual is at the time of primary infection, the greater the risk, for both slow viral infections or for virally induced tumours in animals. The epidemiological characteristics of human tumours favouring the possibility of a viral aetiology can be listed as follows:- a) Dependence of tumour occurrence on climatic conditions such as temperature, humidity and altitude as well as the clustering of tumour cases in time and/or space as in BL in equatorial Africa, suggests the intervention of an insect- borne agent and, in such a case, a viral carcinogen was proposed as more likely than a chemical carcinogen. However, the dependence upon climatic conditions could reflect the intervention of a co-factor acting on the host's defense mechanism but not that of a transforming agent. b) The horizontal transmission of a tumour risk with the presence of healthy carriers would also be suggestive of an oncogenic viral infection 'transmitted' from one person to another, only few getting the disease. The cluster of Hodgkin's disease (HD) studied in Alabama County in New York by Vianna et al (4) suggested the intervention of an infectious arid transforming agent with the presence of healthy carrier.s butthese results have not been confirmed. The case of cervical carcinoma falls into the same category, i.e. an oncogenic agent could be transmitted from the male to the female during sexual activity, with the possibility of a large number of healthy carriers. c) An association or a sequence in time of benign viral lesions and of specific tumours could also be suggestive of viral intervention in such tumours. This may be the case for the relationship which seems to exist between infectious mononucleosis (IM) and later risk for HD. d) The fact that genetic factors are associated with the risk for some tumours in Man is compatible with a viral aetiology, since animals are known to depend upon genetic factors for susceptibility or resistance to RNA and/or DNA virus-induced tumours. The genetic factors associated with NPC risk in Man may possibly be related to susceptibility or resistance to EBV. 3. Diseases of latent or slow growth viruses, MSS Information Corporation, NY 10021, USA, 1973 4. N.J. Vianna, Is Hodgkin's disease infectious, Cancer Research, 34: 1149-1155, 1974 83

II. CRITERIA FAVOURING A CAUSAL ASSOCIATION BETWEEN A VIRAL INFECTION AND THE DEVELOPMENT OF A TUMOUR IN MAN The definition of 'criteria' may be more dangerous than helpful in the present state of the art. However, one cannot avoid posing certain questions which should stimulate further thinking and speed up research in human tumour viruses. A human tumour can be suspected of having a viral aetiology, either because similar tumours in animals are known to be caused by an oncogenic virus or because the epidemiological characteristics of this tu.wiour favours the intervention of a viral agent. Once a candidate virus has been discovered, the following questions could be posed: i) Do tumour cells, irrespective of the ethnic group or geographical area from which the patient arose, regularly contain viral markers such as viral genomes, and viral products such as antigens? We must bear in mind that if the answer is negative this does not prevent the tumour from being related to a'hit and ruri oncogenic viral agent but, in the present knowledge of experimental viral oncology in animals, this is most unlikely. ii) Do patients suffering from this tumour have particular immune response (humoral and/or cell mediated) to viral products? Again, the absence of a specific immune response could not rule out the possibility of a viral aetiology. iii) Does the candidate virus have in vitro transforming activities in human or animal cells of the same embryonal origin as the tumour cells? iv) Does the candidate virus have an in y;.yo oncogenic potential in experimental animals leading to tumour growth similar to that in Man? As a number of viruses, found to be non-oncogenic in the naLural host, induce tumours when inoculated into foreign hosts, this criteria must be used with care. v) When the candidate virus represents a common infection what are the special conditions of infection or other interfering factors leading to tumour development? To answer this question prospective or retro-prospective epidemiological studies are usually required. vi) Does intervention against the candidate virus decrease the risk for th:tumour? Such intervention trials could be directed against primary viral infection (vaccines, hygienic measures) or against co-factors known or thought to activate the oncogenic potential of the virus. If successful, such intervention might establish the causal relationship and prevent the tumour. We shall try and see to what extent these questions can start to be answered for some human tumours. 84

The reasons for suspecting the intervention of a viral agent in human leukemias, sarcomas and mammary carcinomas come mainly from the fact that these tumours are associated with RNA tumour (oncorna) viruses in many animal species. From the epidemiological viewpoint there is no clue suggesting the intervention of a viral agent in the human diseases. However, the influence of genetic factors both in leukemogenesis (a number of genetic disorders are associated with increased risk for leukemias) and in breast carcinoma development (where there is familial clustering) is compatible with a viral factor. The IARC studies in Iceland (5) aimed at trying to link cases of breast cancer in successive generations should bring basic epidemiological evidence of the transmission of a risk factor for this tumour and help in defining better further studies at the epidemiological, clinical and laboratory level for search for a virus. Environmental factors and cultural habits (such as diet) are known to influence the risk in breast cancer. Similarly, diet was found to influence the risk of mammary tumours in mice. The epidemiological data in women do not rule out the possibility that an endogenous virus is transmitted vertically, although, in contrast to mice, no 'milk factor' seems to exist in the human species. There is, as yet, no known virus candidate for association with human leukemias, sarcomas or breast carcinomas but there is converging evidence suggesting the existence of human oncorna-like viruses in these tumours. This evidence includes the presence in leukemic cells of: i) an enzyme similar to the reverse trans- criptase associated with animal oncorna viruses, the human enzyme being immunologically related to the primate oncorna virus enzyme; ii) RNA and DNA sequences annealing with murine and primate C-type virus probes; iii) p 30 protein which could represent a group specific viral antigen; iv) physical particles resembling virions but with no detectable viral activity. If such components are markers for an oncorna virus infection, the virus must be defective as it is not released in vivo despite numerous efforts (6). The various claims in recent years that a human oncorma virus was at hand have failed each time as the induced virus was later on found to be of murine or feline origin. Such putative,defective human oncorna virus may lead to obtaining phenotypic mixing. Similarly, the evidence of viral specific components in human mammary tumours (7) include i) the presence of reverse transcriptase in human milk with ~ :properties similar to those of the murine mammary tumour virus (mu MTV); ii) there are RNA sequences which hybridize with the complementary DNA of mu MTV RNA up to 75% for the sequence; iii) physical particles suggestive of B-type virions have been found in human milk; iv) human mammary tumours contain an antigen which cross- reacts with the mu MTV. Although there is still a long way to go before we can tell if a human mammary tumour virus exists, some reasearchers are of the opinion that immuno- prevention by vaccinating human subjects with mu MTV components could already be justi- fied (7). Such immuno-prevention of mammary tumours has been successfully achieved in virus-free but MTV-susceptible strains of mice and even in strains bearing an endogenous mu MTV but was unsuccessful in high virus-producing strains of mice. The former alternative probably represents the human situation. 5. 0. Bjarnason, N. Day, G. Snaedal and H. Tulinius, The effect of year of birth on the breast cancer age-curve in Iceland, Int. J. Cancer, 13: 689-696, 1974 6. R.C. Gallo, R.E. Gallagher, M.G. Sarngadharan, P. Sarin, M. Reitz, N. Miller and D•H. Gillespie, The evidence for involvement of type-C RNA viruses in human acute leukemia. Cancer, 34: 1398-1405, 1974 7• D.H. Moore and J. Charney, Breast cancer: etiology and possible prevention, American Scientist, 63: 160-168, 1975 85

''• : IV. CERVICAL CARCINONA AND VIRAL FACTORS There is ample epidemioloqical evidence supporting the hypothesis that a venereal transmitted factor is associated with cervical carcinoma. Both the age at which women start their sexual life and the number of sexual partners they have are directly related to the level and age of cervical cancer risk. The herpes simplex virus (HSV) type 2 has been a candidate for having an aetiological association with this tumour but the data supporting this association are, up to now, tenuous. Numerous sero-epidemiological studies have been carried out on cervical carcinoma cases and controls as well as on normal populations at different risk levels. The results have not led to a clear-cut pattern which has partly been due to the lack of tests differentiating properly between antigens coded by HSV types 1 and 2. to 60% of the HSV-2 genomes in one cervical carcinoma, a number of later studies gave negative results. However, since theoretically only 12 of the viral genome would be sufficient to transform a cell, the present molecular virological techniques may not be sufficiently sensitive to detect this small amount of viral DNA in the tumour cells. The detection of HSV-2 antigens in cervical tumour cells has recently been claimed but needs confirmation. Since the permanent presence of viral DNA in cells transformed by HSV in vitro does not seem to be essential for the maintenance of the transformed state of the cells, the absence of detectable HSV markers in cells from cervical carcinomas does not rule out the possibility that HSV-2 plays a role in the development of this tumour. The potential oncogenic activity of HSV-2 has been supported by the in vitro transforming activity of irradiated HSV-1 and HSV-2 in hamster, rat and mice cells (9). Furthermore, HSV-2 has been shown to have oncogenic activity in vivo in mice and hamsters. HSV-2 markers in tumour cells have been searched for both at the nucleic acid and antigen levels. Although Frenkel et al (8) detected nucleic acid sequences annealing up The preliminary results of the prospective study that Nahmias et al are carrying out among women having had herpetic cervicitis (10), seem to indicate that such women have a twofold increased risk compared to women without such lesions. An intervention trial would be the best way to establish if a relationship exists between HSV-2 and cervical carcinoma when a proper vaccine has been developed against this virus, to prevent primary infection. Not much has been done to study the possible relationship between cytomegalovirus (CMV) and cervical carcinoma. The prostate is frequently the localisation of a latent infection by CMV which, apart from being a potential factor for prostate carcinoma, should be investigated as a possible oncogenic agent for cervical carcinoma. Furthermore, ultraviolet irradiated CMV has transforming activities in animal and human fibroblasts. There is a third group of viruses, namely the papilloma viruses, which merit being studied in relation to cervical carcinoma. From the Shope papilloma virus model we know that these viruses could be inocuous in the natural host while being tumorogenic in another host. Thus the hypothesis that the penis supports a 'latent (inocuous) papilloma virus infection which may represent a transforming hazard for some cervical cells at a certain stage of differentiation and hormonal influence, merits investigation. The first step here would be to look for traces of papilloma virus DNA and/or antigens in cervical dysplasia, in situ carcinomas.and eventually invading carcinomas. exposure to ultraviolet-irradiated herpes simplex virus type 2, J.Virol., 8: 469-477, 1971 10. A.J. Nahmias, Z.M. Naib, N.E. Josey, Epidemiological studies relating genital herpetic 9. R. Duff, F. Rapp, Properties of hamster embryo fibroblasts transformed in vitro after tr O O N tr m .D 0 tn 8. N. Frenkel, B. Roizman, E. Cassai, A. Nahmias, A DNA fragment of herpes simplex 2 and its transcription in human cervical cancer tissue, Proc. Nat. Acad. Sci, USA, 69: 3784- 3789, 1972 86

V. THE EPSTEIN-BARR VIRUS - ITS ASSOCIATION WITH BURKITT'S LYMPHOMA AND NASOPHARYNGEAL CARCINOMA The EBV is the cause of infectious mononucleosis and is associated with two widely different human tumours, namely BL and NPC. This virus is the most promising candidate for becoming the first proven oncogenic virus in Man. BL, as mentioned above, has epidemiological characteristics suggestive of the intervention of a critical environmental factor leading to the development of the clinical disease (11). In contrast, the epidemiological characteristics of NPC stress the importance of genetic factors (12). EBV viral markers are found in tumour cells of both BL and NPC. EBV DNA has been found in tumour cells of both these diseases (13). While the presence of EBV DNA in BL tumours cells, which are B lymphoblasts, is not surprising since this virus infects and transforms B lymphocytes in vitro, the detection of viral EBV DNA in,epithelial tumour cells of NPC and not in the infiltrating lymphoid cells was unexpected and should be an incentive to abandon the term of lymphotropic virus for EBV. The presence of a nuclear antigen specific for the EBV (EBNA) has also recently been described in both BL and NPC tumour cells. The question whether such viral markers are found regularly in BL- type lymphomas, regardless of ethnic or geographical origin of the patient, has not yet been fully ans-ered. All but two BL cases analyzed from equatorial Africa have been shown to contain both viral DNA and EBNA (14). In contrast, most of the tumours having similar histopathology from temperate climates and arising in high socio-economic groups only occasionally showed the presence of both viral DNA and EBNA. These data support the concept that EBV could represent one aetiological factor for BL in tropical areas but that would not be a necesgary factor for BL-type lymphomas in temperate climates.. The situation with NPC seems different since undifferentiated tumours from different geographical areas and ethnic groups regularly contain EBV DNA (15). The state of the immune status of both BL and NPC patients with regard to EBV antigens has been a matter of extensive study since the early discovery of the virus by Epstein et al in 1964. As seen in Figure 1, both types of patient have much higher EBV serological reactivities than controls (other tumour-bearing patients or normal individuals). The most relevant serological reactivity for both clinical diagnosis and prognosis lies in the antibodies directed against early antigens (EA). In BL patients antibodies directed against the 'restricted' early antigen (EA-R) have prognostic value while in NPC patients antibodies directed against the 'diffused' type of early antigen (EA-D) have diagnostic and prognostic value. Antibodies directed against EBNA are of interest since such viral antigen is present in the tumour cells. Strangely enough, although both NPC and BL patients have high titres of EBNA antibodies, these antibodies do not appear to have the same clinical value as EA antibodies. 11. M.C. Pike, E.H. Williams and D.B. Wright, Burkitt's tumour in the West Nile District of Uganda, Brit.Med.J., 2: 395-399, 1967 12. G. de-Th6, J.H.C. Ho, and C. Muir, Nasopharyngeal carcinoma, In: A.S. Evans (ed) Viral Infections in Man: Epidemiology and Control, Plenum Corp., New York, in press 13. H. zur Hausen, Oncogenic Herpesviruses, Biochimica et Biophysica Acta (Reviews on Cancer), 417: 25-33, 1975 14. G. Klein, The Epstein-Barr virus and Neoplasia, N.Engl.J.Med., 293: 1353-57, 1975 15. C. Desgranges, H. Wolf, G. de-The, K. Shanmugaratnam, R. Ellouz, N. Cammoun, G• Klein and H. zur Hausen, NPC X - Presence of Epstein-Barr genomes in epithelial cells of tumours from high and medium risk areas, Int.J.Cancer, 16: 7-15, 1975 87

Figure 1: 1GEL orma indivii IN PC & BL PATIENTS C F/s 3,8 2,2 265,8 ~ 128,0 '~'~--~ and especially of BL. of the possible role of genetic factors associated with the development of lymphomas lymphoma-derived cell lines but absent from the other lines regardless of their being derived from tumorous or non-tumorous tissue. This raises again the question seems to lie in the presence of a chromosome 14 translocation which is present in all \\ \I 19 951, 2 ---- 683,4 EA VCA The EBV has in vitro transforming activities on human and non-human primate B lymphocytes, leading to the establishment of lymphoblastoid cell lines. Recent studies have shown interesting differences between cell lines obtained either from human lymphomas, regardless of their being infected by EBV or not, and from in vitro transformation of B lymphocytes. The most interesting difference to now there is no experimental evidence that EBV can induce carcinomas in such primatet EBV has also an oncogenic potential in vivo in some non-human primates, such as marmosets, in which lymphomas quite similar to BL in Man can be obtained. Up epidemiological characteristics of BL, especially space and time clustering (see ref.11). As mentioned above, the EBV is the best candidate for becoming the first established oncogenic virus in Man. However, the virus by itself is obviously not sufficient to induce BL or NPC, since infection by EBV is meso to holoendemic all over the world (70-100x of the individuals from temperate to tropical areas are infected by EBV) whereas BL and NPC are restricted to well-defined geographical areas and ethnic groups. Figure 2 gives a schematic representation of the various factors intervening in a population and eventually contributing to the development of BL in a few individ- uals. From our sero-epidemiological surveys of populations at risk for BL, NPC and IM we hypothesized that very early EBV infection (factor number 1 in Figure 2) is a critical risk factor for later BL development (16). The second risk factor may lie in a genetic defect as testified by a chromosome 14 translocation. A third factor, environmental in nature, would be hyper-holoendemic malaria and a fourth factor, alsd environmental, should be added since malaria cannot, in our opinion, explain all the 16. G. de-The, N.E. Day, A. Geser, M.F. Lavou'e,,J.H.C. Ho, M:J. Simons, R. Sohier, P. Tukei, V. Vonka, H. Zavadova, Sero-epidemiology of the Epstein-Barr virus - preliminary analysis of an international study, In: G. de-Th6, M.A. Epstein and H. zur Hausen (eds.) Oncogenesis and Herpesviruses II, IARC Sci.Pub.No.11, 2: 3-16. IARC, Lyon, France, 1975 l~~i : ........... .... . ...._..~_._,_,_._~._.1_._.E

BL RISR FACTORS 1. VERY EARLY EBV INFECTION 2. SPECIFIC GENETIC BACKCROUND 3. HEAVY MALAAIA BURDEN 4. OTHER ENVIRONMENTAL FACTORS t Since experiments cannot be done on humans the only way to establish the oature of the association between EBV and BL lies in the possibility of either follow- tug prospectively children from primary infection to tumour development or intervening ;against EBV or co-factors helping EBV. The prospective BL study that IARC undertook in the West Nile District a case the pre-BL sera should lack EBV antibodies providing that pre-BL sera were taken before the incubating period; b) that BL develops in children heavily and chronically infected with EBV in which case pre-BL sera should exhibit high sero- delayed primary EBV infection (hypothesis being derived from the IM model) - in such (17) was aimed at testing three hypothesis, namely: a) that BL results from a 'of Uganda in 1971 to study the relationship between EBV infection and BL development of the general population. ;oase the serological profile of pre-BL sera should not significantly differ from that •irus in the tunour or that its action involves concurrent specific events in which =logical reactivities vis-a-vis the general population; c) that EBV is a passenger Between February 1972 and September 1974 more than 40 000 children aged The results (18) show that the only EBV reactivity which differs between the pre-BL Gora and the various controls' sera is the level of viral capsid antigen (VCA) 2-S years were registered and bled. A thorough case detection programme was set up !n parallel and so far 12 BL cases have emerged from the cohort. Eight pre-BL sera bave been tested in two laboratories (Dr Henle's in Philadelphia and IARC's in Lyon) 17• C. de-TH and A. Geser, A prospective sero-epidemiological study to investigate the role of EBV in Burkitt's Lymphoma, Proceedings of the Vth Int. Symposium on Coaparative Leukemia Research, Padova, Italy, Bibl. Haemat. 39, 1972 i8' IARC Annual Renerr ]97S ca ~ 89 O ~

antibody - significantly higher in the pre-BL group. No significant difference was noted between pre-BL and controls in the EA, EBNA and complement-fixing soluble antigen (CF/S) antibody reactivities. When pre- and post-BL sera were compared, there was a surprising lack of significant rise in VCA titres. In contrast the EA antibody titre, low or negative in the pre-BL sera, showed a significant rise after the development of the disease. The folowing preliminary conclusions can be drawn: i) humoral immunity, i.e. high VCA titres, does not protect against BL development; ii) on the contrary, high VCA titres seem to be a favouring factor and may reflect a very early (possibly in the post-natal period) and severe EBV infection; iii) the rise in EA antibody titres probably reflects a reactivation of EBV infection. We should point out here that, although such prospective studies are very time-consuming and costly, they bring information which could hardly be obtained otherwise. In parallel to this prospective study we are following representative samples (1 000 individuals) of this child population and rebleeding them at regular intervals to follow the evolution of EBV infection-and reactivation in the population at large. A second Rpidemiological approach to establish the nature of the association between EBV and BL is to propose intervention trials against either the virus or co- factors. The idea of a vaccine is not very practic,al at the moment because of technical, epidemiological and ethical reasons (19). However, we must remember that immuno- prevention has successfully been achieved for tumours induced by two oncogenic herpes- viruses in animals, namely Marek's disease virus in chickens and herpesvirus saimiri and ateles in primates. Apart from the possibility of a vaccine, another alternative for intervening against the virus would be to try, and de1_aY-EBV-infectio_n by one or two years thus preventing post-natal EBV infectioc; thought to be related to BL risk (see above). The mode of transmission of EBV in newborns-is not known but saliva and possibly milk may play a role. If this were the case, one could easily imagine prevention of the disease possibly by simple hygiene measures. Another intervention trial, being planned by IARC, is to enter into a partial malaria suE~ression_scheme - ------------- to see if thereby BL would decrease significantly. The IARC project plans to bring down the level of malaria from hyper-holoendemic status to mesoendemic in areas where BL detection has been carried out for the past 10 years and where the same tribes live either on the lowland at high BL risk and with hyper-holoendemic malaria or in the highland where malaria is mesoendemic and BL absent. This project, however, poses certain feasibility and organizational problems which have yet to be solved. 19. J. Higginson, G. de-The, A. Geser and N. Day, An epidemiological analysis of cancer vaccines, Int.J.Cancer, 7: 565-574, 1971 . ,LA

In conclusion, Table 1 gives an outline of the situation for each human tumour studied for its possible viral aetiology, with regard to the criteria and questions posed in Section II. Table 1: Level of Association between viruses and certain human tumours Leukemias Breast Cervical BL NPC Sarcomas Cancer Carcinoma Epidemiological characteristics + {. .F} + + favouring viral aetiology Candidate virus - - HSV-2'CMV EBV EBV papilloma Viral markers in + + {} .}{+ tumorous cells _ Patient's immune response to viral - - + 4+ ++ products In vitro transform- ing activity of N/a N/a + + + virus In vivo oncogenic potential in exper- N/a N/a ± + - imental animals Prospective studies N/a N/a + + - Intervention trials N/a N/a - in plan - N/a = not applicable; + = positive; -= negative. It is our opinion that the pursual of efforts to try and uncover the role of viruses in human tumours is important both at a fundamental and a practical view- point. Immuno-prevention of animal tumours induced by both onco-RNA and onco-DNA viruses has been successfully achieved and there is no reason why humans should differ basically from other species in this respect. The proposals of Dan Moore (see ref. 7) and others to try and vaccinate humans with murine mammary tumour viral antigens, if ahown to have no risk from the public health point of view, could be a new and extremely interesting approach in human cancer research whereby intervention trials swst be considered as an integral part of the ammunition available to understand aetiology and to prevent human cancers. For such intervention, as in the more formal human cancer research, epidemiology must be considered as a first rate tool, if properly iategrated with laboratory investigations and use of experimental model systems, in aissioned multidisciplinary teams. This work was conducted in part under Contract NO1 CP 4-3296 within the Virus Cancer Program of the National Cancer Institute, NIH, PHS 91