The basic question before this conference is: What etiologic role, if any, do viruses play in the malignant neoplasms affecting the human race? Despite almost a half century of study of tumors in lower animals caused by agents with the classical characteristics of viruses, the exact biological mechanisms by which these viruses induce malignancy in cells remain undetermined. Certainly it is worthwhile to continue to investigate viral-tumor model systems in animals until the mechanism of carcinogenesis is illuminated. However, the virologist need not delay his etiologic approach to human cancer while waiting for the biologist to solve carcinogenesis.

During the past few years, there has been developing a ferment of interest in viruses and human neoplasms. There are many reasons for the current activity: Tissue-culture techniques have opened a wider vista for sighting hitherto unrecognized viruses than did the previous expansions of the horizon which accompanied the introduction of the suckling mouse, the embryonated egg, and the ordinary laboratory mouse into common isolation procedures. Morphologic techniques, as exemplified by electron microscopy of thin sections, have given tantalizing results with human tumors, while immunofluorescent microscopy and the more recent ferritin antibody procedures for use in electron microscopy (1) seem to provide potent weapons for studies on viruses and cancer. Immunologic procedures have become more delicately sensitive. The scientific world has been stimulated by the infectivity of naked viral nucleic acids. Smallpox and measles, in which infection is almost invariably associated with a typical full-blown clinical and pathological picture, are no longer regarded as the prototypes of most viral diseases. Instead, poliomyelitis, with its high incidence of subclinical infection and rare paralytic manifestations, is considered typical of a large number of viral diseases. Moreover, salivary-gland viruses and adenoviruses have joined the group exemplified by herpes virus and other microbial agents which persist for long periods in clinically recovered persons.

Translation of such findings to the virus-cancer problem provides support for two ideas: (1) Tumor development may occur in only a small proportion of people who are infected with the hypothetical inciting virus, and (2) the causal agent may persist in the individual for years, eliciting its carcinogenic effect long after initial infection. Another development that has stimulated virologists is the findings with the polyoma agent. These demonstrated, more dramatically than earlier observations with chicken malignancies, that multiple types of tumors may arise as a result of infection with a single virus. Can we hope for a simplification in man such as this in mice and chickens?

While virologists have been persuaded to join the exploration of the viral etiology of human cancers primarily because of one or another of the scientific developments just mentioned, there were other factors which assisted in their conversion. A number of investigators were concerned, during and immediately after World War II and the Korean War, with virologic diseases in foreign lands of importance to Americans stationed there and to the indigenous peoples. Others were engaged in work which culminated in the successful control measures for poliomyelitis. Many such persons were willing to rededicate their efforts to another difficult and important cause, i.e., cancer. Fortunately, the interest of the lay public, reflected through the action of Congress, has made funds available for support of research in this field.

Like most of you, I view the problem of the viral etiology of human cancer with hope and enthusiasm. Moreover, I have confidence that investigators entering this field will attain the success here that they have had in their research on the classical viral infections of man.

Although my assigned task was to present briefly certain of the background information which has excited the interest of virologists, particularly, in working on the problem of the viral etiology of human cancer, I cannot refrain from discussing diagnostic virology, which has for many years provided a central theme for my own work. Most cancer investigators have had little exposure to the subject, hence they are not likely to be impressed with the need for the cautious attitude and the stepwise process of elimination that have become ingrained in the virologists who sort out the etiologic agents of human infection. The principles of diagnostic virology have been learned slowly and painfully over the last half century (2). I hope that cancer workers will use this knowledge to advance rapidly in their own field; it would be regrettable if they fumbled, even for a short time, while rediscovering the information and principles.

The basis of diagnostic virology is to be found in Koch's postulates as revised by Rivers in 1937 (8). The latter's ideas, restated somewhat in the light of experience with a host of subsequently recognized viruses, are essentially as follows: A virus is presumed to be of etiologic significance in a given human disease if (1) the agent is present in the tissues or excretions of most persons in the acute stage of the illness and is absent or infrequently present in normal persons or patients with other diseases, and (2) patients who yield the virus during their acute illness develop with regularity, during convalescence, specific antibodies against the agent. Virologists have practically abandoned the idea that Koch's final postulate, i.e., reproducing the disease in a normal host, must be fulfilled in man if a virus is presumed to be a human pathogen. However, an increasing number of the viruses which have been shown by the criteria of Rivers to be of etiologic importance in nonfatal human disease have been inoculated into human beings. Sometimes such transmission studies have resulted in the reproduction of the classical manifestation of the original disease, viz., dengue, measles, and trachoma. In other instances the experimental disease has occasionally mimicked the original but more frequently has reproduced only part of the clinical picture, viz., lymphocytic choriomeningitis and adenovirus infections. Finally, some laboratory strains of viruses, notably the attenuated polioviruses, induce only subclinical infection.

One would hope that a thorough application of old, new, and yet-to-be-developed isolation techniques would yield viral agents from human cancers with which one might fulfill Rivers' criteria. Encouragement will certainly be given to those who try such approaches. However, along with encouragement should go a word of caution, since these approaches will undoubtedly yield numerous agents that will be unrelated etiologically to human cancer, even though they may have been present in the patient and not indigenous to the systems used for the isolation work. The persistent efforts over the years to propagate in the laboratory the viruses of hepatitis and of epidemic hemorrhagic fever provide documentation of the frequency with which agents may be recovered that are not of etiologic significance in the diseases under study. Application of the serologic diagnostic criteria mentioned revealed the irrelevancy of the extraneous isolates; in brief, either the patients developed no antibody against the isolate or if antibodies were present they did not increase in titer during the disease.

If the near future fails to provide techniques by which a viral agent presumably causing human cancer can be put through the standard procedures to fulfill the diagnostic criteria demanded as proof of etiology, then we might decide a priori what data would partially satisfy the basic demands of diagnostic virology.

Our first compromise with principle might be a temporary relaxation of the demand for the appearance of, or increase in, antibody during the course of the disease. Since cancer generally is of insidious onset and has been present for some time before being recognized by the physician, one would anticipate that any antibody response to the hypothetical virus would already have approached its maximum before the patient is examined. With this relaxation, the criteria would then be (1) consistent recovery of the transmissible agent from tumors of a given type along with absence of, or low incidence of, the agent in patients with other diseases, and (2) consistent presence of specific antibody in patients with the tumor. Consistency is the key word here. The ubiquitous viruses which many persons carry for appreciable periods of time, such as herpes simplex, salivary-gland virus, and members of the adenovirus group, would be irregularly distributed among the isolates. For any of the known carrier type viruses to be of even transient interest in the subject under discussion, it would be necessary to demonstrate a high degree of positive association between it and a given tumor.

One might postulate a series of compromises, each of which would retreat a little further from the established methodology and principles applicable to the diagnosis of acute viral infections. However, it would save time to go immediately to that point beyond which only chaos would exist. The minimum evidence for the presumptive establishment of a viral agent as being causally related to human cancer might be fulfilled without propagating the agent in the laboratory and without demonstrating specific antibody by one of the common test-tube techniques. In such an approach, the procedures of electron microscopy might be substituted for those of isolation and transmission in animals, and fluorescent antibody techniques might replace the older serologic methods.

Earlier I spoke of the tantalizing results obtained by electron microscopic examination of human malignant tissues. Intracellular particles resembling those of certain of the established viruses have been observed in some tumors after much searching, and also in occasional control materials. Investigators have been discouraged by the irregular occurrence and sparsity of such particles in cancers of man. Moreover, they have been justified in their cautious interpretation of the observations. In order for the findings to contribute to the partial fulfillment of minimum requirements, they must be demonstrable with great frequency. This obvious statement, in the face of the laborious efforts which have gone into obtaining the presently available data, is no doubt irritating to those investigators who contributed to the subject. The intention is not to provoke but rather to encourage. I recall the difficulty in finding rickettsiae, which are rather large structures, during electron microscopic examination of thin sections of infected yolk-sac tissue that was known to contain at least 10^9 infectious organisms per gram and to contain innumerable rickettsiae when stained smears were examined by ordinary microscopy. The problems encountered in studying small particles in tumor cells might be expected to be infinitely more difficult. The significance of viruslike submicroscopic intracellular particles would be measurably increased if it could be shown that the patient with tumors possessed specific antibody which reacted with them. Perhaps Singer's ferritin-labeled antibodies (1) prepared from serum of the patient might be applied to thin sections of the patient's tumor and examined by electron microscopy.

The fluorescent antibody techniques have proved so valuable in so many fields of endeavor that they will undoubtedly find increasing application in the studies on cancer. One would hope that cells with masses of the particles occasionally encountered by electron microscopy would fluoresce specifically in the presence of labeled antibodies from the patient.

It goes without saying that each sample of tumor tissue used for either electron microscopic or immunofluorescent studies of the general type discussed should be tested by standard virologic isolation techniques to detect the presence of ordinary human pathogens, particularly the viruses, which commonly induce the carrier state. These, if present, might be detected and shown to have induced antibody in the host by the microscopic techniques mentioned, but the standard diagnostic procedures would provide more efficient means for recognizing them.

In closing, I would like to read the final sentences from Rivers' address in 1937 on "Viruses and Koch's Postulates" (3):

"Changes, notably the more extensive use of tissue culture techniques and serologic reactions, will in the future undoubtedly occur in the methods of establishing the specific relation of viruses to disease; the number of changes will be limited only by the amount of ingenuity of investigators. To obtain the best results, however, this ingenuity must be tempered by the priceless attributes of common sense, proper training, and sound reasoning."