To recapitulate, then, the evidence indicates that (1) an agent is present in CFF and CFC prepared from initially radiation-induced mouse lymphomas; (2) this agent can induce identical lymphomas in nonirradiated susceptible mice inoculated at birth; (3) the activity of the agent is appreciably stepped up by serial passage; (4) activity is destroyed by mild heating; (5) normal tissues and other types of neoplasms do not contain demonstrable amounts of the agent; and (6) the agent derived from one strain of mouse is tentatively believed to be specific for mice of that strain or their F1 hybrids. It would be difficult to imagine a chemical agent that is extractable from a specific tumor, is then capable of inducing the same type of tumor, is extractable in increasingly greater yield on serial passage, and exhibits host strain specificity. By exclusion, therefore, the agent is best interpreted as being a virus or viruslike self-replicating genetic determinant. Progress toward the isolation, purification, and further characterization of the agent is seriously hampered by the lack of a rapid and sensitive assay method, preferably in an in vitro system, such as has recently been developed for the Rous (16, 17) and polyoma (18-20) viruses.

Although efforts to develop such an assay are the most urgent need at this time, other kinds of experiments can also contribute to our understanding. Among the questions we may ask are: (1) What is the essential lymphomogenic action of X radiation, or put differently, why is it necessary to irradiate mice of certain strains (C57BL, BALB/c, and C3H) and not others (AK and C58)? (2) Is the profound inhibition of lymphoma development in marrow-shielded or marrow-injected C57BL mice (21) due to restoration by the marrow of radiation-injured immune mechanisms? (3) Is the influence on lymphoma incidence of the hormonal environment (22) consistent with what we know of virus growth requirements generally? (4) Are occult viruslike agents also involved in lymphoma induction by other carcinogens, such as methylcholanthrene or estrogen, and if so, how do these filtrable agents relate to those extracted from radiation-induced lymphomas? It now seems possible to answer question 2, and to make some tentative suggestions about question 1; the other questions remain to be explored.

In a recent experiment, C57BL mice were exposed to X irradiation with or without lead shields over one thigh. The radiations were given in 2 courses 1 month apart, each course consisting of 2 weekly exposures of 168 or 200 r. Thigh-shielded irradiation is known to yield a much lower lymphoma incidence than unshielded whole-body irradiation; this effect is apparently mediated by bone marrow cells protected by the lead shield and can be reproduced by injection of compatible marrow cells from other donors into unshielded mice (21). Since whole-body X irradiation, particularly at higher dose levels, is known to render animals more susceptible to bacterial (23) and viral (24) infection, one ready explanation for the thigh-shielding effect is that it promotes early recovery of immunity in general and of immunity to the lymphoma virus as a by-product. If this were true, one would expect that an initial course of thigh-shielded irradiation would reduce the effectiveness of a second course of unshielded radiation given 1 month later. In fact (25), lymphoma incidence in such a group was significantly greater than that of animals given the second unshielded course alone, and essentially identical with that of animals exposed to unshielded radiation in both courses. The nature of the "priming" effect produced by the first shielded course is not clear, but the result cannot be reconciled in terms of the immunity hypothesis. This conclusion is also supported by some published work of Uphoff and Law (26), who found that injection of fetal hematopoietic tissue, which should not be immunologically competent, conferred almost the same degree of protection as did adult marrow cells against lymphoma development.

Microscopic examination of C57BL thymus glands at serial intervals after irradiation reveals an early, vigorous regeneration and differentiation of the cortical lymphoid-cell elements in marrow-shielded animals and an apparent maturation delay, associated with a persistent abundance of large, immature lymphoblastic cells, in the unshielded mice, which, it will be recalled, are destined to develop a high lymphoma incidence. It may therefore be much more than an interesting coincidence that the thymuses of normal newborn mice, which are highly susceptible to CFF-induced lymphoma development, also exhibit almost a "pure culture" of such large lymphoblastic cells in the outer half of the cortex. If one assumes that these two observations are indeed related, they may indicate a preference by the virus for proliferation in host cells that are at a particular stage of lymphocytic differentiation. Such specificity of appetite has been noted in a few instances with non-tumor viruses (27).

If, as a result of insult produced by radiation, transplantation (28, 29), or urethan (80), occult virus were released from a binding site or from some nonsusceptible "carrier" cell, the virus might find a large population of susceptible cells in the thymus of an unshielded, irradiated host, and could initiate the sequence of intracellular events leading to lymphoma development. Conversely, the thymus of marrow-shielded animals, regenerating more quickly, might contain similarly undifferentiated cells for too brief an interval or in such small numbers that virus proliferation could not readily attain a self-sustaining equilibrium, and few lymphomas would result.

Although this hypothetical model explains certain hitherto puzzling phenomena, its experimental validation must unfortunately await the development of the sensitive virus-assay method referred to previously. The exquisitely refined potentialities of in vitro virus-culture systems should not deter us from continued exploitation of whole animal work, however, since the experimental alteration of host response may reveal important parameters of virus-induced neoplasia which are not apparent at the cellular level.

Dr. Lloyd Law (National Cancer Institute): I would like to congratulate Dr. Kaplan on this provocative paper. I have two comments pertaining to the abstract Dr. Kaplan submitted, which concern interpretations of our present knowledge in this field. He says in the first paragraph that in the past several years it has become well established that the lymphosarcomas and certain lymphatic leukemias which arise spontaneously in certain mouse strains, AK and C58, are due to a virus. I wonder if he really means that our present knowledge indicates this. As I understand it, the present knowledge reveals that certain agents or viruses, obtained from AK and C58 leukemias, are leukemogenic under certain circumstances.

Dr. Gross originated several concepts, one of which is "vertical transmission of the virus"; another, which he called "egg-borne" disease. As I understand, these are 6till concepts, but I did not know that these are well established at this time.

Also, Dr. Kaplan has confused the agent which is described by Rudali, Duplan, and Latarjet, if I understand the data correctly, in saying that the Gross agent not only induces leukemia but accelerates leukemia in the strain of origin. I think there is enough evidence to indicate that these are probably different agents. The agent which Rudali, Duplan, and Latarjet have described is sensitive to cold, and is not infective, for example, in C3H mice. They have shown that it is not transmitted to the next generation, and was obtained only from leukemic tissue and not from normal tissue of AK mice (see Latarjet, R.: Carcinogenesis. In Ciba Foundation Symposium, Boston, Little, Brown & Co., 1959, p. 274.) I mention these only because as a dis-cussor I am supposed to try to stimulate the discussion.