A second type of information necessary for production of tumors and the planning of chemotherapeutic tests is the relationship between dose of virus and the time elapsing between virus inoculation and the appearance of tumors, i.e., latent periods. This relationship has two components which must be determined for each dose level: (1) the magnitude of the mean latent period, and (2) the variation of individual latent periods about their mean.

In some experimental systems it is necessary to transform the original time units, e.g., days, to some other function, such as their reciprocals or logarithms, in order to simplify statistical analyses and to facilitate the construction of charts for graphic interpretations. The Rous sarcoma virus in chickens represents such a system.

Text-figure 3 illustrates typical latent period data obtained with successive tenfold dilutions of Rous sarcoma virus. The reciprocal scale along the ordinate at the extreme left was used in the computations and for plotting the data on the chart, but it is the adjacent ordinate scale showing the corresponding points in days that is used for further graphic interpretations. The central solid line represents the computed regression line drawn through the observed means indicated by the open circles. The other diagonal lines were drawn at the respective multiples of the average standard deviation corresponding to the accumulative percentage values shown at the extreme right of the figure. Since the individual latent periods are approximately normally distributed in terms of their reciprocals (2), deviates on a normal curve were used as the multipliers.

Response of homogenous groups of chickens to the Rous sarcoma virus; 0.2 ml of the dilutions

Text-figure 3. Response of homogenous groups of chickens to the Rous sarcoma virus; 0.2 ml. of the dilutions indicated along abscissa inoculated into wing web of 2-week-old New Hampshire Red chickens-see text.

The percentage incidence results are included on the chart for the sake of completion. They are shown as solid triangles plotted in relation to the diagonal accumulative-percentage lines. Since 100 percent values cannot be plotted in terms of normal deviates, they are indicated on the chart by the points with arrows drawn just above the 99 percent line.

Composite charts such as that of text-figure 3 contain most of the information needed for the planning of experiments, including the systematic production of indicator tumors. They take the place of the "preliminary experiment" usually carried out with rapidly acting viruses of other types, for the purpose of quantitative orientation. Such is not possible with the tumor viruses, however, because of the relatively long times required by them to produce their effects, and by the time the preliminary results are obtained, test animals of the same lot are older and may no longer have the same degree of susceptibility to the virus. Since there is some variation in susceptibility among experimental lots of most test-animal populations (3), charts of this type cannot be used directly for quantitative analytical interpretations. However, they are entirely adequate for preliminary planning and are usually no more inaccurate than are the results of preliminary titration experiments with other viruses, involving small numbers of animals.

Similar orientation charts are shown in text-figures 4 and 5 for the Shope papilloma virus (10) and for a mouse leukemia virus isolated by Moloney (11), respectively. In neither of these cases was it necessary to transform the original time units for the purpose of charting, since the original units gave the relationships desired. The dosage scale for the papilloma data (12) is in terms of log grams of purified virus per ml., the average yield of purified virus per gm. of wart tissue being about 0.1 mg. (IS). In contrast to the results of text-figures 3 and 4 for the Rous sarcoma and Shope papilloma viruses, which represent tumor latent periods, the responses of text-figure 5 represent times-to-death with leukemia. Data on the latent period of leukemia development have not yet been published, but it is stated that the average time from palpable spleen and lymph nodes to death with leukemia is about 19 days (11).

Responses of rabbits to 0.1 ml. inoculums of the Shope papilloma virus

Text-figure 4. Responses of rabbits to 0.1 ml. inoculums of the Shope papilloma virus-see text. Data of (IB).

Responses of BALB/c mice inoculated when 1 or 2 days old

Text-figure 5. Responses of BALB/c mice inoculated when 1 or 2 days old with 0.1 ml. volumes of various dilutions of the Moloney leukemia virus-see test. Preliminary analysis of data made at the 5th month of an experiment still in progress. Data kindly supplied by Dr. J. B. Moloney of the National Cancer Institute. [See (11) for final analysis of the results of this experiment.

As may be seen, this leukemia virus can be diluted at least 2,000-fold and will still induce leukemia in 100 percent of the test mice. Since the volume of the diluted virus used for inoculating young mice is only 0.05 to 0.1 ml., the virus derived from 1 gm. of leukemic tissue would be sufficient for the inoculation of 20,000 to 40,000 animals. The production of large numbers of indicator tumors with this virus should therefore be as practical as with the Rous sarcoma virus.

Papillomas produced by the Shope virus have not been used for systematic screening purposes. Possibly the reason for this is that it is a benign wart, and no one is particularly interested in curing benign warts. However, it is not an ordinary wart since it usually persists, rather than undergoing retrogression after a short time as do similar lesions of other species. Most important are the facts that: (1) The virus is of the deoxyribonucleic acid (DNA) type (14); (2) it is found primarily in the nuclei of the papilloma cells (15), and (3) these benign virus lesions frequently undergo transformation to malignant cancers (16). Compounds that would stop the benign lesions might, therefore, have an important bearing on the prevention of cancer, since they would give a clue to the types of substances that are capable of attacking DNA-type viruses in an intranuclear location, and they would remove the precancerous soil, on a basis of which malignancy may later develop.

Since there is evidence that some chemotherapeutic agents are more effective against tumors produced by weaker doses of virus (17), it would seem desirable to use for the induction of indicator tumors the weakest doses that will produce tumors in all, or most, of the test animals inoculated. As already pointed out, this is also the dose desired for the conservation of seed virus. When the responses to such doses are measured in weeks, or months, as with the mouse leukemia virus, preliminary information such as that illustrated in text-figure 5 is essential for the efficient scheduling of the animal inoculations, the times for initiating routine examinations for tumors, and the use of animal caging facilities.

Finally, since it is probable that the variations in individual latent periods among animals inoculated with the same dose of virus are associated to some extent with differences in individual test-animal susceptibilities (6, 18), it may be possible to reduce the variations among animals in autonomous chemotherapeutic tests by selecting cohorts of indicator tumors appearing within successive, relatively short intervals of time, the animals being divided within cohorts, at random, into control and test groups for the independent tests. The accumulative frequencies with time, such as those shown in text-figures 3 to 5, would permit the advanced planning of appropriate cohort intervals for a given virus-host system. It has already been established that the tumors which appear earliest in groups of animals inoculated with strong doses of virus yield the largest quantities of recoverable virus on extraction (6, 19-21). Such selected tumors are now commonly used for increasing the potency of virus strains through successive passages (6, 11, 19-22), and for maintaining high levels of virus in source tumor tissues (0, 19-21). It seems likely that this same relationship may hold also for tumors produced by the weaker doses used in chemotherapeutic tests, and that the precision of antitumor responses may in some instances be increased by the cohort segregation procedure proposed.