It has already been indicated that the question of sensitivity and specificity of cell response is inseparably bound to the problems of dose response, virus-strain homogeneity, and genetic character of the test animal. Historically, it will be recalled (19-21) that the variety of tumors observed in mice, after inoculation with direct extracts of leukemic tissues, was much smaller than that seen subsequent to passage of the agent through tissue cultures, either of monkey kidney (83), normal mouse embryo (5), or of mouse lymphoma cells (18). Only parotid tumors, subcutaneous fibrosarcomas, adrenal medullary tumors, and occasional thymic tumors were observed before tissue-culture passage. After receiving culture-passaged material, mice developed a much wider variety of tumors and other lesions, reported by Stewart, Eddy, and Borgese (5) to be as high as 23. That this change of response was due to a dose-response effect of a single virus is supported by Buffet and associates (57). The production of a similar effect, through partial purification of virus by differential centrifugation, and by viral cloning studies by plaque methods, was reported by Eddy and Stewart (51), Stewart et al. (52), and Sachs et al. (53). Possible objections to the plaque method for cloning this agent in mouse-embryo cultures have been raised (39), but at present there is no compelling positive evidence that more than one agent is concerned in the induction of tumors of salivary glands, lacrimal glands, submucosal glands of the upper respiratory tract, thymus gland, mammary gland, hair follicles, thyroid gland, tooth-germ epithelium, adrenal gland, renal medulla, and bone. Others (13, 42, 48) have confirmed this finding, by and large, but it has also been stressed (18, 14) that the polyoma virus by no means possesses "cancer pan-tropism" (37). This point deserves re-emphasis, since it has been the author's experience that many who are not actively working in this field seem to have developed the impression that polyoma virus can induce tumors in any organ, and conversely, that any tumor of the laboratory mouse should be considered a "polyoma tumor" until proved otherwise. This attitude is not in accord with the known facts. The majority of the tumors induced by this agent have special histologic and biologic characteristics in which they differ from previously known murine tumors, even of the same organ. Dunn (see 21) noted the unique histologic character of the salivary-gland tumors, and also delineated certain features of mammary tumors associated with the polyoma virus in which they differed from mammary tumors arising in mice with the milk agent (IS). It is noteworthy that the tumors induced or accelerated in mammary tissue by 2 different viruses (Bittner milk agent and polyoma virus) have distinct histologic patterns, as well as differences in distribution between sexes. This suggests that the two agents operate through different pathways, or enter the same pathway at different points. A similar situation may perhaps exist for virus-induced leukemias and leukemia-like conditions, though it is not certain that these agents all act initially on the same cell type.

It may be remembered that only a decade ago it was often said by investigators, wearied of histologic classifications, that when causative agents of tumors were found, these old confusing classifications could be dispensed with. It is reassuring to microscopists that, now that some causative agents are known, there seems to be more need than ever to classify histologically the corresponding tumors. It is imperative to be aware, however, that most of our evidence today indicates that in the polyoma virus we are dealing not with a new-found cause of a previously familiar disease, but with a newly found disease and a newly found agent causally related to it. This awareness in no way detracts from the system as a research tool, but is indispensable for proper use of the tool and understanding its products.

Differences in classification already exist with respect to many of the tumors associated with this agent. Some cases in point will be discussed here because they are important to the interpretation of past reports as well as of future investigations.

The tumors of the thymus gland have now been described by many investigators (5, 13, 14, 48)- They were designated thymic epitheliomas by Stewart (5) who was first to recognize them as part of the polyoma pattern. Later, Buffet et al. (57) described similar tumors in mice inoculated with polyoma virus, but classified them as reticulum-cell sarcomas. McCulloch et al. (58), working with an agent apparently closely related to, if not identical with polyoma virus, also observed thymic tumors and designated them as thymic lymphomas. It will readily be seen that baffling complications are brought into the picture if the thymic tumors are either lymphomas or reticulum-cell sarcomas. For, in the early stages of recognition of the polyoma virus, there was considerable question as to whether this agent and the leukemia agent of Gross were one and the same (59). If the thymic tumors following polyoma-virus inoculation are neoplasms of hemic cells, the possibility of either a genetic relationship to, or an admixture with the leukemia agent of Gross would have to be considered. Morphologic evidence for the epithelial nature of the tumors has been put forth by Stewart et al. (5) and by others (13, 14)-There is also biologic evidence for this view. Transplantation attempts carried out by Law (49) have resulted in successful transfer of thymic tumors to only 2 of approximately 200 recipients of compatible strain. These 2 successful transfers occurred in 2 animals of the same litter, inoculated with a cell suspension when less than 24 hours old. The transplanted tumors appeared after a latent period of nearly 3 months at the site of subcutaneous inoculation. No tumors of salivary glands or other organs were found, and the histology of the transplants was identical with that of the primary tumor, including the presence of large numbers of lymphoid cells adjacent to the proliferating epithelial masses. There was no evidence of leukemia or lymphoma in any organ of these animals. The extreme difficulty of transplantation and the failure to yield a morphologic picture consistent with a lymphoma or leukemia are strong evidence that these tumors are not hemic in nature.

Perhaps closely related to the problem of the thymic tumors is a situation that is found only occasionally in the parotid-gland tumors. Among the multicentric tumors of the parotid gland there are some examples that resemble, but are not identical with, the thymic tumors, and are also not tumors of ectopic thymus, such as have been described (13). Yet they contain relatively large numbers of lymphoid cells, and relatively small numbers of epithelial cells (figs. 1 and 2). The epithelium is confined to a few tubular structures, or may be distributed perivascularly, as in the thymic epithelial tumors. These tumors might be mistaken for lymph nodes or lymphocytic neoplasms, since they are composed predominantly of lymphoid tissue. Recognition of the epithelial component reveals their compound nature. On close examination, it becomes evident fig. 2) that the lymphocytes in these tumors are aggregated in groups in which all stages of maturation are represented. This is unlike the situation in lymphocytic neoplasms, where uniformity in the degree of maturation, or lack of it, is the rule. Dr. M. H. Salaman, on a visit to our laboratory, informed us that he has been investigating this type of tumor in transplants and was kind enough to show and give us sections. Two tumors transplanted through several generations repeatedly showed the compound "lymphoepithelial" pattern described previously, and none of them resembled usual lymphocytic neoplasms.