State planning,cancer vaccine infrastructure,and the origins of the oncogene theory

Cancer research meets Cold War management technologies

Fatefully for his future approach to cancer virus research, Baker’s eager and wide-ranging studies of scientific management brought him into contact with the methods of systems analysis. ⁷ Systems analysis – a managerial cousin to cybernetics – exerted a profound impact on federal bureaucracy throughout the Cold War, most notably in budgeting and planning for aerospace engineering projects at the Department of Defense and the National Aeronautics and Space Administration. Although it traced its origins to aerospace production, it represented a different model for planning science than that offered by the industrial mass-production models that had provided inspiration for antibiotic, cancer chemotherapy or polio vaccine development – particularly in its efforts to and manage uncertainty (Carrese and Baker, 1967; Scheffler, 2019: 72–75, 113–116).

Baker’s first opportunity to apply systems planning to cancer research came when Endicott placed him in charge of restructuring the Viruses and Cancer Panel, whose efforts had been stymied by a lack of favorable developments in the laboratory. A special report for Endicott, likely prepared by Baker, promised that the discovery of a human cancer virus was imminent based on an ‘avalanche of evidence from an unprecedented number of scientific disciplines’. ⁸ Despite his confident assessment of the state of viral research, Baker’s planning effort struggled with the persistent absence of laboratory proof that human cancers were caused by an infectious agent.

Baker was convinced that adeptly deploying the social technologies of systems management would make human cancer viruses tractable for the administration of the NCI even as they remained elusive in the laboratory. While moving into planning might seem ‘unique or even strange’ for biomedical researchers, Baker wrote, it was ‘necessary’ to meet the challenges involved in multidisciplinary research devoted to ‘the needs of society and matters related to health’. In this effort, ‘the concepts and philosophies, the terminologies and procedures satisfactory for the description’ of contemporary biological research were ‘not suitable for describing the newer developments proposed’. ⁹ In this light, ‘program plans’ were intentionally agnostic on the question of particular scientific outcomes; plans were not to ‘require or involve … the detailing of specific research efforts, the establishment of priorities, the selection of the mode of operation, or the insertion of a time-frame’. Such outcomes would emerge organically as experimentation advanced toward the goal of vaccine development. The paramount advantage of program planning was that it provided ‘a mode of approach to the planning of research efforts which are oriented and focused to the achievement of the end result or product’. ¹⁰

To establish a culture more amenable to this style of planning, Baker recruited scientific managers fluent in the vocabulary of operations research, as opposed to medical research. In particular, he hired Louis Carrese, an industrial psychologist, to bring his experience as a contract systems analyst for the Department of Defense to bear on cancer research. ¹¹ Unlike the annual cycle of research grants, Carrese explained that ‘the projection period [for contract systems analysis] should present a reasonably workable timespan (5–10 years) suitable for the framing of critical questions in research in a manner that will permit the evaluation of progress’. This was a way of discussing biomedical research that, as Carrese underscored, provided ‘the basis for action decisions – not further discussions’. ¹²

For these evaluations, Baker and Carrese turned to a particular species of systems analysis, the ‘Program Evaluation Review Technique’, or PERT. While systems analysis as a whole offered a budgetary framework for setting goals, it did not necessarily address the question of how to meet these goals quickly. In the late 1950s, administrators of the Polaris Missile Program – a crash effort to build a submarine-launched ballistic missile – had developed PERT in an effort to accelerate the research and development process. The construction of the missile in record time gave PERT immediate cachet in management circles (Johnson, 2000: 96–98). Its advocates praised its focus on attaining goals in a minimum amount of time for ‘new’, ‘untried’ and ‘non-routine’ activities (Evarts, 1964: 1–6).

Baker and Carrese’s study of PERT and other forms of systems analysis provided the basis of their design of a ‘convergence technique’ for planning biomedical research (Carrese and Baker, 1967). In contrast to these other schemas, which involved problems of engineering, their convergence technique was an attempt to adapt systems planning to an enterprise where ongoing, discovery-oriented inquiry formed the basis of everyday operational activities distributed across dispersed laboratories and clinics. The challenge for Baker and Carrese was in developing network-based management techniques that could integrate loosely interrelated research projects with open-ended timelines, indeterminate tasks and uncertain objectives so that through consistent long-term planning they converged upon the development of a human leukemia vaccine. The task of the SVLP was, accordingly, to plan the conditions that would lead to serendipitous findings in pursuit of a vaccine. To remain accountable to bureaucratic planning, ongoing scientific research would now have to articulate with the ambitious organizational projections embedded in the convergence chart.

OPEN IN VIEWER

Figure 1.

Chart comparing projected vaccine development for suspected cancer viruses in animals and humans by extrapolating from previous vaccine innovation timelines. Special Virus Leukemia Program Progress Report #4, 1967.

The emphasis on time characteristic of the PERT planning process soon permeated the documents of the SVLP. One seemingly innocuous chart in an annual report of the SVLP appeared to compare the state of progress in developing a leukemia vaccine with progress in developing vaccines for other human viral diseases and viruses linked to cancer in animals. The stakes at the heart of the Programs’ virus vaccine research emerge in the powerful rhetorical argument about time and progress communicated through these bureaucratic artifacts. This chart posited eight steps through which every viral vaccine passed – from the acquisition of materials to the industrial production of the identified virus and vaccination. Moreover, the progress vector for each virus was labeled with the number of years the viruses had been studied. In mice, the identification of a leukemia virus had been followed by the development of a vaccine in fifteen years. It took 58 years to develop a polio vaccine, but only 34 to develop a measles vaccine. Human leukemia viruses had barely progressed beyond the ‘detection’ and ‘identification’ stages (itself an optimistic assertion). Most beguiling, however, were the dotted lines included with the human leukemia viruses: evidence of discoveries that were anticipated in the future. ¹³ These projections encapsulated not only the state of Program research to date, but also the projections of scientist-managers that vaulted well over the locally circumscribed boundaries of their immediate experimental settings.

The human cancer virus vaccine as a problem of infrastructure

Though Baker’s planning techniques may have focused on the future, their initial uses addressed the difficulties faced by NCI leadership in the present, especially securing funding for cancer vaccine research through the annual Congressional appropriations process. Endicott drew upon the climate of expectation and crisis surrounding leukemia viruses (which the NCI itself had cultivated) to organize an end-run around the normal NIH budget approval process, seeking a special appropriation from the Senate to get the SVLP started. ¹⁴ The following year, Endicott needed to justify the SVLP’s budget again. His defense of the Program revealed the rhetorical power allowed by Baker’s management techniques. The NCI, he promised, was poised to deploy a system of organization suitable for an accelerated push to develop a leukemia vaccine. Congress could not afford to delay this effort:

We believe that the developments in the research areas mentioned are so important and so opportune that we must do everything possible to push forward now… we cannot await full understanding of the nature of cancer …. we have learned to work with mechanisms for a planned approach …. we await with impatience developments in the human virus area which will demonstrate that a virus is a causative factor in at least one type of human cancer. Our organization is being made ready for such a development because we know that when this happens we must exploit this lead with all possible speed. ¹⁵

The public responded to this new presentation of cancer viruses enthusiastically. LIFE magazine elaborated that in the SVLP, ‘the plans called for starting far in advance to work out the specifications, devise instruments, put up buildings, train personnel, breed animals – everything to ensure that all systems would be Go and A-OK as if for a countdown at Cape Kennedy’ (Rosenfeld, 1966: 111). Headlines such as ‘Leukemia cure near?’, ‘Hope raised for cure of leukemia’, and ‘Virus link found: Huge leukemia project pushed’ indicate the success of the NCI narrative in framing the SVLP as a response to an acute crisis as opposed to a long-term search for cancer control (Anonymous, 1964; Toth, 1964a, 1964b). Congress approved $10 million in special appropriations to fund the SVLP at its initiation, but henceforth the Program could rely upon generous support through the NCI’s regular annual appropriations. Moving forward, the task of administrators would be to sustain a climate of expectation among their Congressional patrons while they awaited positive scientific results.

The assumption that ‘at least one virus is an indispensable element … of at least one kind of human leukemia’, animated the SVLP’s operations and galvanized its supporters. ¹⁶ Yet realizing an actionable future around this hypothesis would entail fundamental transformations of the materials and routines of experimentation in the laboratory as well as decision-making and planning practices oriented toward cancer control. NCI administrators cashed in their congressional surplus to fund construction of scientific infrastructure with an aim toward realizing the SVLP’s projections. ‘In devising any experimental attack on a biological problem’, Frank Rauscher, then-SVLP scientist and future Director of the NCI, explained, ‘one is invariably faced with the need not only to design the experiment, but the need for many materials with which to conduct the experiment’. While he had initially been skeptical of the uses of program planning, Rauscher admitted, the creation of the convergence chart had convinced him that dealing with resources first was necessary, least important experiments were forced to wait ‘months, or perhaps years’. ¹⁷

NCI leadership understood the Cancer Virus Programs’ efforts as more than accelerating an inevitable scientific discovery – they were an indispensable means of creating the conditions under which human cancer viruses would emerge in experiments. The budget of the Cancer Virus Programs grew steadily, reaching 36 million dollars annually before the start of the War on Cancer in 1971 – a substantial sum for virology research. ¹⁸ This level of funding gave a select group of NCI scientist-administrators the ability to shape the whole field of cancer virology.

The first few years of Program operation interleaved the infrastructural, epistemological and ontological facets of cancer virus research. It was not reasonable to wait for a candidate human cancer virus to emerge and then focus resources on its study because the resources of the Program were essential to revealing the existence of human cancer viruses to begin with. Baker’s response to critics who questioned the ‘intellectual underpinning’ of the vaccine effort illustrates how the Program implicated infrastructure, epistemology and ontology in the success of vaccine efforts. ¹⁹ The question, Baker maintained, was not if individual investigators would find proof of human cancer viruses, but if they even had the capacity to do so. Human cancer viruses and other carcinogens, Baker explained, could only be identified through ‘greatly expanded’ search efforts. For academic scientists, Baker continued, this kind of screening work represented ‘mere data collection’ too ‘pedestrian’ to merit competitive grant funding. It fell to the NCI to handle the ‘large and complex’ logistical effort of maintaining the ‘great numbers of animals … over long periods of time’ necessary to generate statistically useful findings. These efforts, Baker concluded, would ‘not be done in the laboratories of individual investigators’. The scale of Program operations was not only a question of acceleration, it was an effort to detect experimental events that individual laboratories would invariably miss. Its ‘complex, integrated, systematic’ efforts were not only required, they were ‘overdue’. ²⁰

Where academic investigators were unwilling to work, the NCI would create a new set of research institutions. In this effort, task-oriented contracts allowing coordination between private industry and the government were an essential tool. Testifying before Congress, Endicott explained that the NCI was ‘trying to develop … a new sort of industry to do the types of technical jobs that need to be done here’. The Department of Defense, Endicott explained, ‘reached this scale a long while ago …. a whole series of consulting laboratories … developed around the country’. ²¹ He was more expansive regarding his ambitions in an interview: ‘We’ll build some industries. What the Hell! We’ll build them because we’ve got to have them’. ²²

This emphasis on infrastructure-building shone through in the first round of contracts issued by the SVLP, which succeeded in spending $9.9 million on 48 contracts in eight months – far faster than an equivalent amount ever could have been distributed through grants. These first contracts sought to rapidly scale up virus research in animals. One line of resource development concerned the cultivation of ‘germ-free’ populations of mice, hamsters and quails for further research – which Germfree Products Inc. of Tampa Florida promised to do for $720,335. However, the Program did not wait for infrastructure projects to issue promising results in animals. It sent $328,399 to Baylor University to use electron microscopy and other methods to reveal the presence of viruses in blood samples taken from children with leukemia at the Texas Children’s Hospital Hematology Service. Reflecting its confidence that the identification of a human leukemia virus was imminent, the largest contract the Program awarded in its first year was for $989,000 to Dow Chemical for the design of a ‘state of the art’ leukemia virus research laboratory incorporating the latest biohazard control measures. ²³

The large influx of funding at Program investigators’ fingertips allowed vaccine-oriented infrastructure to flourish. The Program was willing to build research operations from nearly whole cloth. The SVLP’s interest in the epidemiological, etiological and clinical profile of viral lymphoma spurred the development of infrastructure halfway around the world – an enterprise that would have otherwise been unlikely in the late 1960s. Contracts provided a rapid means for the NCI to pursue materials and support international research seen to help its domestic aims. In one instance, the NCI built a lymphoma center and funded large-scale epidemiology studies and clinical trials to investigate the relationship between Epstein-Barr virus and Burkitt’s lymphoma in Uganda (Masnyk, 1988). Additionally, the conversion of laboratories at Fort Detrick, Maryland from studies of biological warfare to Program-directed cancer virus research in 1971 provided federal researchers with the most technologically sophisticated biohazard safety laboratories in the country. The NCI leveraged Fort Detrick’s biocontainment facilities to support the agency’s continuation of recombinant DNA research on cancer viruses during the National Academy of Science’s voluntary moratorium from 1973 to 1976 (Fredrickson, 2001: 124, 135–140). In all these instances, the NCI was able to leverage Cancer Virus Program contracts to sponsor research that could not have been conducted otherwise.

SVCP contractors generate the cellular oncogene theory

With the hope for a cancer vaccine now resting on retroviruses, the NCI focused its infrastructure-building efforts on resources and training for molecular biological experiments that could generate knowledge about viral oncogenes. Although the unusual concentration of power and resources represented by the SVCP drew heated criticism from the academic molecular biology community, which feared that dependence on cancer research funding would distort its scientific creativity and independence, it nonetheless had major influence on the course of molecular virology (Culliton, 1973; Scheffler, 2019: 166–170; Wade, 1971; Watson, 1973). The Program’s tentacular network of contract laboratories throughout the United States provided means for the traffic of information, practices and resources among communities of researchers whose work was directed, in the provisions of their NCI contracts, to produce evidence of viral oncogenes for the eventual purpose of vaccine development.

It was in this environment that Harold Varmus and J Michael Bishop began their study of cancer viruses. Educated as doctors and then trained at the National Institutes of Health, Bishop, and Varmus entered into a robust network of NCI contractors throughout the West Coast upon assuming positions at the University of California, San Francisco (UCSF). As young scientists overwhelmingly dependent upon soft money to fund their research, Bishop and Varmus found SVCP contracts a boon for their studies on the Rous Sarcoma Virus (RSV), a retrovirus at the center of debate over the viral oncogene. By tailoring their research to ask and answer questions of urgent interest to the SVCP’s targeted efforts, Bishop and Varmus were able to rapidly accelerate the progress of their research. Unlike the months-long grant process, the NCI issued contract funds in a matter of weeks, which enabled Bishop and Varmus to quickly scale up their resources to pursue promising new experimental leads. ²⁹ In the early 1970s, NCI contracts funded 85% of the UCSF laboratory’s RSV research. ³⁰ Fellow West Coast SVCP contractors, with Huebner’s encouragement, circulated vital resources for conducting genetic analyzes of RSV, including strains of mutant virus that enabled detailed study of the molecular mechanisms of viral reproduction. ³¹

In maintaining close relationships with other NCI contractors who shared knowledge, skills and resources, Bishop and Varmus’s ever-expanding laboratory was able to respond nimbly to their fellow West Coast contractors’ proposition that src, a gene fragment of RSV, was responsible for cancerous transformations in infected cells. They drew together their prior expertise in studying RNA based viruses and the technique of DNA-DNA and RNA-DNA hybridization to develop new techniques for precisely demonstrating the presence of src in transformed cells (Scheffler, 2019: 192–193). The UCSF laboratory became the epicenter of major NCI efforts to expand the materials, skills and personnel necessary to advance molecular hybridization techniques with the eventual aim of detecting a viral oncogene and potentially arresting the process of cancerous transformation. The SVCP awarded contracts to Bishop and Varmus in the hopes that src research would both substantiate the viral oncogene hypothesis and, in illuminating the process whereby viral genes transformed cells, pave the way for cancer control. ³² It regarded the Bishop-Varmus laboratory as ‘one of the best in the Program’, tripling the laboratory’s funding from 1971 to 1974. ³³

The Bishop-Varmus laboratory grew into a research unit unto itself. By 1975, VCP contracts to the laboratory sustained a veritable platoon of professors, postdoctoral researchers, graduate students and full-time technicians; purchased and maintained expensive equipment like ultracentrifuges and radioactive isotopes; and cultivated enormous quantities of RSV strains and cell cultures. ³⁴ The scale of this work was different in kind from that of many other retrovirus researchers, who did not enjoy the economies of scale associated with studying hybridization in individual laboratories (Scheffler, 2019: 195–198).

However, the scale of work at Bishop and Varmus’s laboratory produced more than the VCP’s planners anticipated. True to the principle of differential reproduction posited by Rheinberger, the experimental systems developed in Bishop and Varmus’s laboratory with the purpose of supporting the viral oncogene hypothesis to advance the NCI’s vaccine mission produced the conditions for the emergence of very different phenomena. The basis of this productive instability lay within the calibration experiments associated with the synthesis of the src hybridization probe. This work had occupied several researchers in the Bishop and Varmus lab, and in 1974 the project passed to Dominique Stehelin, a visiting researcher from France. A successful probe would detect src in infected cells, but of course it would also not register its presence in uninfected cells. Stehelin set to work testing the src probe against both infected and uninfected chicken cells. The probe revealed src in infected cells, as predicted, but the calibration experiments also revealed that src was present in uninfected cells. This finding was deeply puzzling – after all, viral genes should not be present in cells unexposed to RSV. Solving this puzzle consumed another year and a half of the laboratory’s experimental efforts, resulting in the surprising conclusion that src was not viral, but cellular in origin – a gene chickens and other birds inherited from a common ancestor in the distant evolutionary past (Stehelin et al., 1976).

With hindsight, this finding was recognized as the beginning of the ‘oncogene paradigm’ of cancer, and Bishop and Varmus were awarded a Nobel Prize for their laboratory’s observation in 1989. The explanation of the probe’s unexpected results was theoretically exciting, and other STS scholars (Fujimura, 1996) have focused on the incorporation of evolutionary time into the interpretation of this surprising calibration result. However, this does not address the conditions which stabilized the initial observation as a meaningful event and gave it sufficient credibility to draw further study rather than being dismissed as either an error inconsistent with the viral oncogene theory or a novel result of little consequence.

At the time of the first observation in 1975 the Bishop and Varmus probe was so sophisticated that few other labs were fluent in the technique. The virus they studied had been so extensively cultivated for research that leading tumor virologists such as Temin cautioned that the mutant strains of RSV under examination were not ‘natural virus[es]. [They] do not persist in nonlaboratory populations. RSV is a laboratory creature’ (Temin, 1974: 2836). Other retrovirus studies were far slower to yield viral oncogenes, even with considerable support (Scheffler, 2019: 214–215). Like many other ‘discovery’ moments, historians of molecular biology have shown that the emergence and acceptance of the cellular ‘oncogene paradigm’ attributed to this single experiment actually took the better part of a decade (Morange, 1993).

The conditions that allowed this noise to be seen as a signal were deeply tied to the infrastructural basis of the laboratory’s work. Rouse’s insights into how scientists simultaneously articulate technical issues with broader stakes well beyond self-enclosed experimental systems illuminates how Bishop and Varmus gained the time and resources to address these concerns. This reorientation, while it appeared to contradict the organizational mandate of the NCI, could only have taken place in the Bishop and Varmus laboratory at that time due to the infrastructure to support hybridization probes that had been assembled there in the name of vaccine development.

The first reorientation regarded the probe itself. The Cancer Virus Program had thought it was sponsoring work patterned on precision immunological probes in its effort to maximize specificity and detection of a specific form of src. However, further experimental data to support src’s cellular origins came from making the probe less precise. The evolutionary relationships suggested by the 1976 paper hinged on degrees of reactivity as a proxy for degrees of evolutionary divergence. However, the 1976 paper remained rooted in variants of src in bird species because these were the closest to the specific probe created from RSV. After publication of this paper, Deborah Spector, a new postdoctoral researcher in the laboratory, used her prior experience with RNA hybridization to manipulate the concentration and composition of the salt solutions for the reactions of the src probe with cellular DNA. ³⁵ This allowed src probe hybrids with cellular DNA of lower and lower percentages of overlap. While Spector could not say exactly what the probe hybridized with, her adjustment of the system helped fit the presence of src into an evolutionary schema which extended beyond birds to other vertebrates and even to sea urchins (Spector et al., 1978). Given the expense and labor associated with production of the src probe, these observations could only have been conducted in a laboratory with the capacity to produce large amounts of the src probe – a feature of the infrastructure of the NCI’s programs.

The second reorientation regarded the kind of question the laboratory asked using these findings. The question was no longer if viral oncogenes were linked to cancer, but what purpose these genes served in the cell and to where their origins might be traced. Both of these aims minimized the public health obligations of oncogene research. In anchoring the presence of versions of cellular src outside of bird species, the principal data in the 1976 Nature paper, Spector’s observations allowed Bishop and Varmus argue more forcefully that src was a ‘proto-oncogene’ which served some common task in cellular development rather than directly causing cancerous transformation. Writing in the second edition of The Molecular Biology of Tumor Viruses, Bishop and Varmus conceded that ‘the tools used to trace the lineages of c-onc genes [cellular proto-oncogenes] are blunt’. However, based on their data it remained a ‘reasonable guess’ that cellular oncogenes had been around for millions of years (Bishop and Varmus, 1982: 308). ‘What of the role of retroviruses in the etiology of human neoplasia?’ Bishop asked. ‘I have avoided this controversial issue … because the case at present is highly circumstantial and the attendant issues are not germane to my purposes here’ (Bishop, 1978: 88).

It appeared that the NCI’s hope of identifying a human cancer virus had been rendered remote from and incidental to the aims of molecular biologists. In the 1980s retroviral oncogenes provided the basis for identifying cellular oncogenes in many species, further bolstering an experimental research program focused on interior genetic changes rather than external infection (Vecchio, 1993). The irony of this outcome was not lost on the NCI. A group of the Program’s former champions mused that the ground appeared to have shifted beneath their feet:

Scientists have completely redefined the word ‘virus’ and no longer think of a tumor virus only as a particle that enters a host cell and causes disease, a particle that can be inactivated and turned into a protective vaccine. Thus cancer causation no longer conforms to old concepts of infection and disease (Manaker et al., 1977: 631).

Ultimately, the redefinition that the NCI administrators noticed, even if it did not reflect their aims, was a testament to the generative nature of the infrastructure they had created – its accomplishment was not only in accelerating or furthering research that might have occurred by other means, nor its error in enforcing a controversial theory on the scientific community – rather it was that these came together to create the conditions for Bishop and Varmus’s work and the stabilization of their unexpected discoveries.