Retroviruses as carcinogens and pathogens: expectations and reality.

Retroviruses (without transforming genes) are thought to cause leukemias and other cancers in animals and humans because they were originally isolated from those diseases and because experimental infections of newborns may induce leukemias with probabilities of 0 to 90%. According to this hypothesis viral cancers should be contagious, polyclonal, and preventable by immunization. However, retroviruses are rather widespread in healthy animals and humans where they typically cause latent infections and antiviral immunity. The leukemia risk of such infections is less than 0.1% and thus about as low as that of virus-free controls. Indeed retroviruses are not sufficient to initiate transformation because of the low percentage of symptomatic virus carriers and the complete lack of transforming function in vitro; because of the striking discrepancies between the long latent periods of 0.5 to 10 years for carcinogenesis and the short eclipse of days to weeks for virus replication and direct pathogenic and immunogenic effects; because there is no gene with a late transforming function, since all genes are essential for replication; because host genes, which do not inhibit virus, inhibit tumorigenesis up to 100% if intact and determine the nature of the tumor if defective; and above all because of the monoclonal origin of viral leukemias, defined by viral integration sites that are different in each tumor. On these bases the probability that a virus-infected cell will become transformed is estimated to be about 10(-11). The viruses are also not necessary to maintain transformation, since many animal and all bovine and human tumors do not express viral antigens or RNA or contain only incomplete proviruses. Thus as carcinogens retroviruses do not necessarily fulfill Koch's first postulate and do not or only very rarely (10(-11)) fulfill the third. Therefore it has been proposed that retroviruses transform inefficiently by activating latent cellular oncogenes by for example provirus integration. This predicts diploid tumors with great diversity, because integration sites are different in each tumor. However, the uniformity of different viral and even nonviral tumors of the same lineage, their common susceptibility to the same tumor resistance genes, and transformation-specific chromosome abnormalities shared with non-viral tumors each argue for cellular transforming genes. Indeed clonal chromosome abnormalities are the only known transformation-specific determinants of viral tumors. Since tumors originate with these abnormalities, these or associated events, rather than preexisting viruses, must initiate transformation.(ABSTRACT TRUNCATED AT 250 WORDS)