How are tumour antigens related to normal antigens?

Tumour cells share with normal cells antigens characteristic of defined states of differentiation. Is there anything else? An example of tumour antigens not expressed anywhere in normal tissue is the set of tumour-specific transplantation antigens (TSTA) of murine chemically induced sarcomas. There is evidence that at least one TSTA specificity is retrovirus-derived, is carried on the envelope protein gp70, and probably arises by the recombination events that yield the diverse gp70s of the MCF strains of murine leukaemia viruses. Whether a similar mechanism can generate human tumour antigens depends on the yet unanswered question of whether human cells have retroviruses in their genomes capable of recombination. Aside from this, the only other mechanism known for antigen expression on tumours is via their oncogenes, which seem to make normal cell products. Such products, or secondary consequences of their production, were they normally expressed only at an early stage of development, would be candidates for 'fetal antigens'. While only the two mechanisms mentioned above seem the ready sources of 'tumour-associated antigens', it would be too early--in the face of ever more startling information about gene mobility and rearrangements--to think we have exhausted possible mechanisms for generating tumour antigens.