The little death: Rigoni-Stern and the problem of sex and cancer in 20th-century biomedical research

Originator narratives as cultural symbols: a theoretical approach

STS has long questioned the claim that scientists’ discourse provides an unmediated account of their research practices. Historians and sociologists of science instead analyze scientific writings as rhetorical artifacts designed to refract research practice for the purposes of persuasion (Gilbert, 1977; Gilbert and Mulkay, 1984; Gross, 1990; Latour and Woolgar, 1986; Lynch, 1985; Shapin, 1984a). Work in STS has thus fruitfully demonstrated how originators are invoked in service of pedagogy (Allchin, 2003), boundary work (Forman, 1969), and the construction of scientific facts (Latour, 1987; Myers, 1990). Such work analyzes how scientists marshal the legendary feats of originators to preserve the integrity of the scientific method (Medawar, 1963), ensure the moral status of scientists (Sapp, 1990; Shapin and Barnes, 1979), settle theoretical and methodological disputes (Brannigan, 1979; Collins, 1985; Olby, 1979), and reinforce faith in the inevitability of scientific progress (Kuhn, 1996 [1962]).

These approaches reach beyond a conception of rhetoric as mere persuasive oratory toward a ‘rhetorical hermeneutics’ that approaches texts as discursive artifacts that can be interpreted to reveal insights about scientific cultures (Gross and Keith, 1997). Such studies suggest originator narratives need not be treated merely as post hoc strategic exercises analytically separate from activities STS scholars subsume under the concept of ‘scientific practice’ (e.g. research methodology and design, theoretical modeling, instrumental manipulations, techniques of visualization, experimental protocols). ¹ Although they occupy formalized spaces in the genre of the scientific article, originator narratives can also be analyzed as cultural symbols that scientists produce and perpetuate by interpreting their research practice through available discursive conventions and prevailing cultural mores.

A treatment of scientific publications that interprets the content of originator narratives at the conjuncture of discourse and practice emphasizes both the ongoing, reiterative, and practical nature of interpretation and its embeddedness in concrete social and historical settings. This position is consonant with STS scholarship that takes an explicitly ‘pragmatic’ approach to scientific practice (see Fujimura, 1996; Pickering, 1995; Star, 1983) as well as recent cultural sociological insights inspired by the tradition of American pragmatism (see Biernacki, 2005; Gross, 2009; Haydu, 1998; Joas, 1993). Such pragmatic theoretical orientations, by locating the stability and coherence of discursive conventions in their connections to the immediate context and conduct of action, encourage STS scholars to explain the content of originator narratives as emergent from the relationship between experimentation and interpretation. Substantive explanations of scientists’ invocations of originators should consider interpretations as engagements with broader cultural symbols that are made to seem both meaningful and ‘experientially real’ (Biernacki, 2000: 305) in light of scientists’ research practices.

To demonstrate the theoretical gain of this approach to originator narratives, I apply it to a case study in the history of cancer research. At the heart of this case is an empirical puzzle: despite rapid innovation in the technical and theoretical apparatus of epidemiology and molecular biology throughout the 20th century, prominent researchers from these fields have actively sought to maintain continuity with a crude explanatory framework proposed in the 19th century. To resolve this puzzle, I first discuss how Rigoni-Stern proposes a causal relationship between sex and cancer, and then analyze how interpretations of his formulation as advanced in journal articles influenced research into sex and cancer in successive historical periods. I show how both local interpretations of research practices and more global cultural assumptions about sexual activity are refracted through these narratives. I then situate these successive interpretations alongside a broader historical trajectory of the ‘problem of sex and cancer’ – that is, the problem of constructing a causal account that explains the long-standing association between sexual activity and cervical, penile, and anal cancers. I argue that originator narratives of Rigoni-Stern allowed researchers to understand their current work as meaningful in relation to a constructed historical legacy of attempts to causally connect sex and cancer.

Rigoni-Stern and the problem of sex and cancer

Rigoni-Stern first published his iconic explanation of the relationship between sex and cancer in an 1842 issue of Giornale per Servire ai Progressi della Patologia e della Terapeutica (Journal of Progress in Pathology and Therapeutics). In this treatise, entitled ‘Fatti statistici relativi alle malattie cancerose’ [Statistical facts relative to cancer rates], the Paduan physician surveyed death records in the Veneto to render statistics on the prevalence of different cancers in men and women. The article, adapted from a talk given to the Fourth Congress of Italian Scientists, argued against Modena physician Giovanni Gandolfi’s deductive claim that all cancers were products of the same inherent biological processes. Rigoni-Stern, skeptical of a common origin for all cancers, urged physicians not to accept such deductions in the absence of demonstrable statistical associations among maladies. Nevertheless, Rigoni-Stern did suspect a common origin to certain cancers, such as those of the womb and breast. To demonstrate the potency of medical statistics in constructing strong causal arguments, Rigoni-Stern inferred that the complementarity between cancers of the breast and womb suggested a common cause that could be traced to similar external influences rather than to the inevitable degradation of the organs (Rigoni-Stern, 1842).

Referencing his own statistical analysis, Rigoni-Stern noted that deaths from breast cancer were significantly higher among nuns than married women, but that deaths from uterine cancer (‘cancer of the womb’) were far lower in convents than in the secular sphere. The physician postulated that restrictive clothing and compression from prayer irritated the breasts, thus accounting for higher rates of cancer in those organs; regarding uterine cancer, he suggested a protective role for celibacy. In support of this inference, Rigoni-Stern proposed that higher cancer rates in the married, presumably sexually active female population correspond to ‘the great propensity these organs have to mechanical damage’ (Scotto and Bailar, 1969: 71). ²

Rigoni-Stern’s argument that uterine cancer resulted from external injury to organs that ‘disrupt[ed] the mechanics of their physical functions’ (Scotto and Bailar, 1969: 69) would be reinterpreted in diverse ways throughout the 20th century. These reinterpretations often misrepresent the content of the original treatise in systematic ways that suggest they are best explained in light of shifting sexual mores and changing practices in cancer research. The remainder of this article will demonstrate how, even as the theoretical and experimental techniques used to explore the problem of sex and cancer changed, the story of Rigoni-Stern allowed cancer researchers to construct their work coherently through reference to a broader historical trajectory.

Post-war epidemiology and the sexual etiology of cervical cancer

Studies of the link between sex and cancer coalesce into a coherent analytical approach around the middle of the 20th century. As I will demonstrate, particular interpretations of Rigoni-Stern’s problem influenced the research design and methodology of these studies. In this period, epidemiologists attempted to explain the causal connection between sex and cancer in terms of multiple factors that interacted with relevant tissues and organ systems at particular moments or doses to trigger cancerous growth. Epidemiologists framed etiological explanations in terms of interactions among sexual behaviors or environmental influences and tissues and cells, in their attempts to infer the biologically salient factors that cause cancer via sexual activity. In so doing, they, like Rigoni-Stern, relied upon contemporary assumptions about sexuality to guide their methodology, design their studies, and interpret their findings.

The inaugural studies in this era attempted to locate a causal explanation, as did Rigoni-Stern, in the mechanical damage resultant from the sexual act itself. To this end, they explored tissue damage and irritation as potential mechanisms that triggered cancer. In the early 1950s, epidemiologist Harold F. Dorn still maintained, despite an admitted lack of evidence, that rates of uterine and breast cancer were due to the consistent exposure of these organs to trauma through sexual intercourse and its concomitant factors, such as childbirth (Proctor, 1995: 24). This commitment is hardly trivial, given that Dorn, as the architect of the first National Cancer Survey, is considered by some to be the father of post-war American cancer epidemiology (Lilienfeld, 2008). Over the ensuing quarter century, epidemiologists sought greater empirical support for this etiological position.

American studies of nuns and married women not only were some of the most-cited studies into the connection between sex and cancer but also offer the most compelling evidence that Rigoni-Stern operated as an interpretive resource that framed the structure of epidemiologists’ inquiries. Malcolm Griffiths (1991) has already discussed how a number of questionable interpretations of Rigoni-Stern’s treatise, as comparing nuns and prostitutes, motivated most epidemiological studies of cervical cancer worldwide. However, a review of Griffiths’s cited studies reveals that invocation of Rigoni-Stern was an almost exclusively American practice prior to the 1970s. ³ Rigoni-Stern’s re-introduction and dissemination were tied to a peculiar practice whereby American epidemiologists used marital and religious status as proxies for biological processes related to sexual activity.

Throughout the 1950s and 1960s, Rigoni-Stern’s inheritors explored rates of cervical cancer among nuns and other celibate populations in comparison to married women, categories they presumed to indicate the presence or absence of sexual activity with some degree of certainty. Some of these studies targeted ‘marital clusters’ of genital cancer to explain a number of interesting correlations that had emerged in recent decades. Prominent researchers explored the remarkably low rates of cervical and penile cancer among Jewish women and their husbands, seeking explanation in religious practices like ritual circumcision and abstinence during menstruation (Dunn and Buell, 1959; Wynder, 1955). These practices, the authors posited, protected against penile and cervical cancer by eliminating smegma (then a suspected carcinogen) as a potential irritant, or sparing the cervix from damage at a time when cells are particularly vulnerable.

A positive correlation was represented in marital clusters of cervical and penile cancers (Asano and Kurihara, 1956). When Isidro Martínez reviewed the statistics at Puerto Rico’s Central Cancer Registry, he concluded that penile cancer in husbands frequently preceded cervical cancer in wives. In fact, incidence of cervical cancer was eight times higher among the wives of men with penile cancer than it was among controls (Martínez, 1969: 779). Martínez, not prepared to speculate, proposed only a vague ‘common etiologic factor’ (p. 780) to account for marital clusters of penile and cervical cancer. This interpretation reflected a more general shift in how post-war cancer researchers conceptualized etiology, away from the irritation model of mutation that better supported behavioral studies and toward a theory of cancer as cell proliferation brought about by transformations to hereditary material (see Löwy, 2010).

Contemporaneous with studies of genital cancers in married couples was the continuation of research into cancer rates among nuns, the convenient avatars of celibacy. A 1950 study by gynecologist Fabien Gagnon replicated Rigoni-Stern’s study on a larger scale, using the death records of 13,000 Catholic nuns supplemented with histological samples when available, while modifying the original study to disaggregate cancers of the uterine cervix from those of the uterine corpus (Gagnon, 1950). Although Gagnon’s data roughly corroborate Rigoni-Stern’s, the former also show that there is no significant difference between cervical cancer rates among nuns and (presumably celibate) single women outside the cloistered orders. While Gagnon’s studies were meant to support his theory that chronic irritation from cervicitis caused cervical cancer, gynecologist Janet Towne (1955) studied nuns to test a hypothesis that mechanical damage from childbirth was most responsible. Towne’s methodology and findings were similar to Gagnon’s, but unlike Gagnon she interpreted the rare presence of cervical cancer in her sample of nuns as indicating celibacy was not in itself a guarantee against genital cancers. Nevertheless, Towne agreed that some factor related to sexual intercourse must account for the extreme differences in incidence among celibate and sexually active populations.

The era of epidemiological explorations in close analogue to Rigoni-Stern’s treatise came to a close after the 1969 publication of an exhaustive study led by Joseph Fraumeni of the National Cancer Institute (NCI). The Fraumeni study analyzed mortality statistics spanning the period 1900–1954 to compare cancer deaths of women in convents with those of the general female population. The study once again found that cervical cancer mortality rates among nuns were similar to those of single women in general. Alongside Martínez, the authors argued that it was not the presumed celibacy of these populations that offered a protective effect against cervical cancer but rather ‘an environmental factor related to coital experience’ that explained the comparatively higher rates of cervical cancer among married women (Fraumeni et al., 1969: 466).

By the beginning of the 1970s, studies exploring the protective role of celibacy in purportedly virginal populations appeared increasingly unproductive, as epidemiolgists (despite Towne’s conclusions) had largely settled upon the idea that sexual intercourse between men and women was a necessary precondition for cervical and some penile cancers (Rotkin, 1973: 1354). As sexual mores shifted, it also became apparent that men and women engaged in premarital sex at a frequency that made marital status a questionable proxy for sexual behavior (see Prescott, 2010). The work of I.D. Rotkin, who imagined himself as continuing the line of investigation that Rigoni-Stern ‘precociously’ instigated over 100 years prior (Pearre, 1973), is illustrative of how epidemiologists integrated both revised assumptions about premarital sexual activity and new laboratory practices for studying infectious disease into their etiological claims. Beginning in the 1970s, Rotkin attempted to resolve epidemiological findings with suggestive insights from serological studies into herpes simplex virus–2 (HSV-2). In the process, he developed a profile of the typical cervical cancer case:

She was born in deprivation and exists in a low socioeconomic stratum. There is a higher likelihood that she is Spanish, black, Puerto Rican, or East Indian in the United States and elsewhere, but the prospect is also high that she is white if poor and uneducated. The cultural commonality is a low socioeconomic status, but women from higher stations in life are not excluded. She is exposed at an early age to male sexual aggressiveness as an environmental hazard. Although she is too young to be highly arousable, early sexuality is a folkway solution for companionship and an antidote for loneliness. Onset of coitus is during adolescence or earlier. Intromission sets up risk of cervical cancer. Exposure to a plurality of sexual consorts, each with a discrete probability of conveying a carcinogenic influence to her, increases risk. The speculation now is that one of these carcinogenic influences may be herpesvirus type 2. (Rotkin, 1973: 1355) ⁴

The task of laboratory science was to shed additional light on how such a confluence of factors leads to the development of cervical cancer.

Thanks in part to Rotkin’s efforts, by the 1970s, etiological claims had shifted away from a focus on sexual activity as itself cancerous and toward a model that implicated sexual transmission of cancerous elements through risky (i.e. ‘early’ and ‘promiscuous’) intercourse. Although epidemiologists agreed that the age of first intercourse and number of sexual partners were important risk factors, a pathogenic agent was shaping up to be the most likely explanation for the link between genital cancers and sexual intercourse. In the absence of compelling evidence implicating syphilis and gonorrhea in cancer, ⁵ American researchers focused their energies on increasingly sophisticated seroepidemiological studies of the more recently identified HSV-2.

Seroepidemiological studies (epidemiological projects that also collect blood samples for immunological analysis) into HSV-2 and cervical cancer began in earnest in the early 1970s (Rawls et al., 1973). While these studies established that women with cervical cancer and its precursors were both significantly more likely to harbor antibodies against HSV-2 and to show elevated antibody titers in comparison to controls, they could not show that infection with HSV-2 was not merely covariable with a more fundamental cause like ‘promiscuity’. In other words, even when researchers detected HSV-2 DNA in the cell DNA of cervical tumors, they could not demonstrate that HSV-2 infection pre-dated cellular transformation (Frenkel et al., 1972). Epidemiologists also argued that, despite laboratory scientists’ enthusiasm for HSV-2, the virus failed to align with the ‘universally accepted’ etiological profile for genital cancers established through decades of epidemiological inquiry (Alexander, 1973: 1485).

Yet researchers continued to build the case for HSV-2, drawing upon evidence from earlier epidemiological data, laboratory investigations using animal models, and seroepidemiological studies (see Aurelian, 1984). Their persistence in the HSV-2 hypothesis triggered debates about time order effects among epidemiologists and laboratory scientists. Just as these debates fomented into crisis, emerging molecular technologies promised the opportunity to make new causal claims based on the interactions among different genes in human cell lines. The direction of research shifted dramatically when one German virologist used these molecular techniques to argue for a different viral candidate, reinterpreting Rigoni-Stern anew in the process.

Establishing etiology: the shift from epidemiology to molecular biology

While epidemiology’s ability to formulate explanations that consider multiple etiological candidates is often considered a disciplinary strength, explanations based upon statistical associations between variables often struggle to meet the standards of direct causality. Opinions had long differed as to whether epidemiologists could make causal claims in the first place, but the problem of causality became particularly salient in the 1970s, following confrontations between epidemiology and the legal system around the association between smoking and cancer (Proctor, 1995: 110) and exposure to environmental toxins and personal injury (Golan, 2012). The changing relationship between cancer epidemiology and laboratory-based cancer research provides important context for understanding the shifting interpretations of originator narratives in the final decades of the 20th century.

Most of the causal claims epidemiologists made about the relationship between sex and cancer rested upon retrospective case-control methods, which had come under heavy scrutiny by the late 1970s (Horwitz and Feinstein, 1979). As previously noted, these studies could not establish viral infection antecedent to cellular transformation, and could not entirely rule out a confounding variable (such as ‘promiscuity’) that explained the relationship between both outcomes. Researchers were able to manage their concerns about causality by supplementing epidemiological evidence with laboratory analysis or clinical observation. However, two changes took place in the 1970s that made it difficult for epidemiologists to make strong claims about cancer etiology: first, support for traditional epidemiological studies from the largest funding agency in the country was made contingent upon a diminished role for epidemiologists in cancer research, and second, an alternative viral candidate was proposed based upon insights from new molecular genetic laboratory techniques. The latter trend is covered in the next section; however, understanding why epidemiologists lost their ability to make strong causal claims about cancer etiology will lead to a more complete picture of the dominance of molecular explanations after the 1980s, and problematize later attempts to accommodate epidemiological findings to molecular evidence through originator narratives.

Ironically, epidemiologists’ etiological claims were devalorized just as epidemiology was establishing itself as an independent discipline. While most earlier epidemiologists had been trained as MDs, beginning in the early 1970s, students could obtain a dedicated graduate degree in epidemiology. For the most part, the graduate curriculum in epidemiology programs was committed to the risk paradigm and its attendant statistical method (Susser, 1985). While decoupling from medical education buttressed epidemiology’s professional identity by standardizing education and increasing disciplinary autonomy, it hobbled epidemiologists’ ability to benefit from funding through the NCI.

The National Cancer Act of 1971 poured millions of dollars into cancer research, primarily by way of the NCI. The lion’s share was allocated to fundamental research, a boon for laboratory studies but not for the kind of field, survey, and clinical studies typical of epidemiology. Beyond marginalization incidental to the new cancer legislation, leadership at NCI and its parent National Institutes of Health (NIH) had become reluctant to support traditional epidemiological studies, despite their already modest share of federal research moneys. Citing the languid development of the discipline, its ‘low status’ in medicine, and a purported lack of scientific standards, in 1977, NIH director Donald Frederickson organized an institutes-wide committee that would ‘guide the growth of epidemiology’ toward the aims of the Institutes (Frederickson, 1977).

While NIH leadership allowed that epidemiology could make causal claims, citing the link between smoking and lung cancer as a notable example, they saw the future role of chronic disease epidemiology as offering exploratory, analytical, and evaluative support to the explanatory enterprise of laboratory science (Gordon, 1976: 3). Biometric studies such as the Surveillance, Epidemiology, and End Results (SEER) Program became the primary focus of the NCI’s epidemiology branch in the decade between the mid-1970s and mid-1980s. As support was withdrawn or redirected, the remaining epidemiological studies investigating the links among environment, behavior, and cancer failed to ‘keep pace’ with advances in their fields (NCI, 1980: 4). Funding for traditional epidemiological studies investigating the etiology of anogenital cancers picked up after the mid-1980s due in part to the HIV/AIDS epidemic, but these funds explicitly targeted multidisciplinary projects that sought to verify epidemiological findings through clinical and molecular studies (NCI, 1989). The net effect of these policies (as will become apparent in the following section) was to make epidemiology subordinate to laboratory studies that were seen as offering more fundamental and rigorous causal claims.

The organizational gulf that relegated epidemiologists to the fringes of cancer research, coupled with an inability to mount strong causal arguments on retrospective evidence, tempered epidemiologists’ claims. Nevertheless, many prominent epidemiologists remained stalwart in their position that their studies could yield strong etiological arguments. As late as 1974, Ernst L. Wynder, whose research helped establish a causal link between smoking and lung cancer, continued to assume the primacy of epidemiology’s causal claims. In recommending the NCI assemble multidisciplinary ‘working groups’ of virologists, analytical chemists, public health authorities, and other experts to explore the causes of penile and cervical cancers, he resolutely maintained that ‘[t]he starting point and, ultimately, the final measure of success of these working groups should be the epidemiological evidence’ (Wynder, 1974: 1516).

The problem was that many of the causal variables first enumerated in epidemiological mechanisms explaining the connection between sex and cancer were demoted to risk factors and proximate causes as focus shifted to a specific viral agent amenable to laboratory study. While theories of viral oncogenesis, or the ability of viruses to cause cancer, were first proposed at the turn of the 20th century and proliferated in the 1960s (see Gaudillière, 1998; Morange, 1997), laboratory studies of viral cell transformation were usually restricted to animal models. One of the major obstacles to studying viral oncogenesis in humans was the absence of stable and uncontaminated human cell lines that could be cultured for in vitro studies. Given these limitations, laboratory researchers sought to extend the theory of viral oncogenesis to humans by demonstrating the presence of viral DNA in tumors, rather than showing their direct transformative effects at the cellular level. While these efforts to establish proof of viral oncogenesis would eventually lead to the oncogene hypothesis discussed in the next section, in the interim they encouraged researchers to explore diverse viruses implicated in anogenital cancers.

It was in the spirit of such exploration that German virologist Harald zur Hausen would propose a novel candidate for human cancers. Zur Hausen was a trained physician who transitioned into experimental virology after accepting a postdoctoral position at the laboratory of Werner and Gertrude Henle at the Children’s Hospital of Philadelphia in 1966. In his time in the Henles’ laboratory, he demonstrated the presence of Epstein–Barr virus (EBV) particles in patients with Burkitt’s lymphoma. This work helped establish a causal link between EBV (a member of the herpes virus family) and Burkitt’s lymphoma, the first instance in which a virus was shown to cause cancer in humans (Zur Hausen et al., 1970). Galvanized by these results, Zur Hausen continued investigating the link between herpesviridae and human cancers upon returning to Germany in 1969, this time studying the role of HSV-2 in cervical cancer. Yet the problems of establishing causation continued to plague etiological arguments in favor of HSV-2, and Zur Hausen repeatedly failed to find satisfactory evidence that HSV-2 caused cervical cancer (Zur Hausen, 1975). He endured this line of inquiry only briefly before seeking another viral candidate, human papillomavirus (HPV).

In 1976, Zur Hausen’s laboratory published data suggesting that the family of HPVs, or ‘wart viruses’, was genetically quite diverse (Zur Hausen et al., 1976). Upon further investigation, Zur Hausen noticed that the epidemiological patterns for condylomata acuminata, or genital warts, mimicked the epidemiological patterns for cervical cancer (Zur Hausen, 1976; Zur Hausen et al., 1976). Having already unsuccessfully attempted to detect HSV-2 DNA in cervical tumors, Zur Hausen now argued (based upon the complementarity of their epidemiological profiles) for HPV as an etiological candidate for cervical cancer. Zur Hausen’s laboratory published prolifically on HPV in the ensuing decades, particularly in US journals. What Zur Hausen offered was not only a new etiological hypothesis, but a new interpretation of an originator narrative already familiar to his American audience.

Molecular biology and the oncogenic virus as sexually transmitted infection

While the research practices of molecular biologists appear widely divergent from those of epidemiologists, the new breed of cancer researchers continued to invoke Rigoni-Stern to understand their work as addressing the same historic problem. This time, however, Rigoni-Stern’s argument was interpreted as illuminating the role of sexually transmitted viruses in inducing cancers at the molecular level. New narratives about Rigoni-Stern were possible because the changes in experimental practice that ushered in the molecular era, including genetic theories of cancer and DNA hybridization techniques, offered new ways of interpreting ancient afflictions. At the same time, popular assumptions about sexuality facilitated new interpretations of which salient factors associated with sexual intercourse explained the connection between sex and anogenital cancers. Combined, this schemata allowed laboratory scientists to develop originator narratives that supported their claims that suspected cancer viruses were both sexually transmitted and capable of inducing malignant changes at the genetic level.

As Löwy (2010) and others (De Chadarevian and Kamminga, 1998) note, the move to accommodate observations made at higher levels of analysis (e.g. cytological analysis or population data) to processes at the level of the gene is characteristic of a broader trend of ‘molecularization’ in late-20th century biomedicine. Historians Jean-Paul Gaudillière (1998) and Michel Morange (1997) and sociologist Joan Fujimura (1996) have demonstrated how the oncogene hypothesis developed in close concert with molecular experimentation and imaging techniques to enable a comprehensive and compelling shift in etiological claims about cancer. Also interesting is how these changes made new interpretations of Rigoni-Stern’s legacy seem both meaningful and plausible despite their discordance with previous narratives. Originator narratives after the 1970s indicate that, contrary to the occasional hostility laboratory scientists showed toward epidemiology, they were at pains to reconcile epidemiological findings with their own work at the molecular level.

Fujimura (1996) characterizes much of the early period of molecular cancer research in terms of the ‘oncogene bandwagon’, which entailed rapid uptake of emerging molecular theories and technologies in the field of cancer research. The main focus of this bandwagon was a new mechanism in cancer causation that emerged from studies into viral oncogenesis. Proposed by Harold Varmus and J. Michael Bishop in 1976, the proto-oncogene theory holds that cancer is caused by the alteration of existing genes inside normal human cells. As Fujimura argues, once recombinant DNA techniques were stabilized in the 1980s, American scientists began identifying proto-oncogenes and their triggers, and a ‘standardized package’ of theory and experimental apparatus spread throughout cancer laboratories that made molecular biology the favored tool for studying the causal mechanisms of human cancers.

Fujimura argues that the strength of the standardized package was in its ability to yield ‘doable’ problems, exemplified by the race to find homologs of oncogenes detected in animal models in the human genome. This premise is easily extended to virus-cancer research, as HPV researchers throughout the 1980s occupied themselves by mapping the genomes of different HPV types. By 1984, DNA for the two most likely oncogenic strains of HPV (HPV-16 and HPV-18) were available for laboratory study. While European studies, primarily under the banner of Zur Hausen, relied on DNA hybridization techniques well into the late 1980s, American scientists quickly utilized new recombinant DNA techniques to explore HPV’s potential to act upon known proto-oncogenes.

Support for the HPV hypothesis grew in the 1980s, but investment in HSV-2 as an oncogenic virus clung to life in the United States. Citing the wealth of epidemiological, clinical, and laboratory studies supporting an etiological role for HSV-2 in cervical cancer, epidemiologist Irving Kessler declared in 1981 that ‘the viral agent of the disease is clearly established’ (Kessler, 1981: S20). However, unlike a decade earlier, researchers now sought support for their causal claims in molecular studies that would demonstrate the direct action of HSV-2 on cellular oncogenes. While experiments on animal models were initially promising, these results failed to consistently translate to human cells (Manak et al., 1981; Rapp and Reed, 1976). Not only were viable human cell lines still recent technologies (Fujimura, 1996: 64), but HSV-2 itself is extremely cytopathic, in most cases causing cellular death within 24 hours (Rapp and Jenkins, 1981). After repeated attempts to induce cancer with mixed results, the tide turned against HSV-2 in 1985 when leading herpes virologist William Rawls’ laboratory conducted a series of DNA hybridization studies that concluded the evidence for HPV as an etiological agent was much stronger than HSV-2 (Prakash et al., 1985).

When American researchers finally directed their attention to the relationship between HPV and cancer, they did so in the context of large collaborative laboratory projects that only rarely integrated epidemiological data (see Dyson et al., 1989; Pirisi et al., 1987; Werness et al., 1990). These studies advanced causal explanations of the link between sexual activity and cancer that relied upon gene–gene interaction rather than the more complex environmental or behavioral pathways indicated by epidemiological trends. Harald zur Hausen went so far as to argue in an editorial in the Journal of the National Cancer Institute that epidemiology could tell researchers very little about how HPV causes cancer. The virologist who a dozen years prior had leaned heavily upon epidemiological trends to support his etiological hypothesis now reversed his position:

Today there exists a substantial body of evidence linking specific types of human papillomaviruses (HPVs) to anogenital cancer. Interestingly and in remarkable contrast to other viruses linked to human cancer (Epstein-Barr, hepatitis B, human T cell lymphotropic-1), most of the evidence originated from laboratory studies revealing transforming activity of these viruses, a specific mode of integration and of genetic activity in cancer cells, and the role of viral genes in the maintenance of the proliferative phenotype in cells of HPV-positive cervical carcinoma lines. To a much lesser degree, epidemiologic approaches contributed to our present state of knowledge, whereas seroepidemiology virtually contributed not at all. (Zur Hausen, 1989: 1680)

In fact, Zur Hausen argued, the techniques upon which laboratory insights were built not only benefitted little from epidemiological explanations, they ‘primarily created problems for epidemiologists’ trying to investigate the causal link between HPV and anogenital cancers (Zur Hausen, 1989: 1680). Furthermore, laboratory investigations held greater promise for immediate intervention into HPV-related anogenital cancers (by the mid-1980s, Zur Hausen was a staunch believer in the possibility of an HPV vaccine).

Despite using supposedly antithetical laboratory practices to approach the problem, Zur Hausen credits Rigoni-Stern’s 1842 publication as ‘the first important contribution to etiological factors in human genital cancer’ (Zur Hausen, 1987: 7). Yet this assignation of influence is accompanied by sweeping historical revisions. Zur Hausen now presents Rigoni-Stern’s formulation as based upon a comparison of cancer rates between nuns and prostitutes that implicates venereal disease in cancerous growth. While the addition of prostitutes and sexually transmitted infection certainly buttresses Zur Hausen’s findings, there is no mention of either in Rigoni-Stern’s study. Taken at face value, this is a dramatic oversight, especially since Zur Hausen mentions Rigoni-Stern routinely beginning in 1975 and continuing well into the 21st century (see Zur Hausen, 1975, 1987, 1989, 2009). However, when Rigoni-Stern’s definition of the problem of sex and cancer is viewed as an interpretive resource, even erroneous attributions are meaningful because they create continuity through retrospective influence.

The substance of Zur Hausen’s interpretation of Rigoni-Stern’s work indicates how specific narratives become plausible enough to endure over the decades. The crucial change is in the insistence that Rigoni-Stern was really studying sexually transmitted infections. This reinterpretation, wherein a viral agent is the telos, makes the connection to Rigoni-Stern’s work seem realistic to cancer researchers, because it reads the germ of molecular studies into the original treatise. While Zur Hausen appears to have been the first researcher to advance this new interpretation, the change was the outcome of a gradual process. Over time, Zur Hausen’s (1975) declaration that ‘[s]ince more than 130 years the epidemiology of cervical carcinoma suggested to some investigators an infectious agent as cause of the disease’ (Zur Hausen, 1975: 40) was transformed into an assertion that

Observations made by Rigoni-Stern (published in 1842) on a vastly different incidence of cervical cancer in prostitutes as compared to nuns mark the first important contribution to etiological factors in human genital cancer. These studies, confirmed by many subsequent analyses, stimulated research on the role of sexually transmitted agents in the genesis of this neoplasia. (p. 7)

Here Rigoni-Stern’s claims are not only reinterpreted to accommodate contemporary molecular models of etiology, but also to anticipate nearly the entire literature built up by epidemiologists since mid-century. By interpreting Rigoni-Stern as having suspected a sexually transmitted infection all along, Zur Hausen draws a straight line from past attempts to establish the etiological significance of infection to present efforts to firmly support the role of HPV in causing anogenital cancers – and even to the future, where a vaccine might prevent many of these cancers from occurring.

By the late 1980s, Zur Hausen advanced his interpretation of Rigoni-Stern’s formulation as guiding research into the etiology of most virally induced cancers, including those attributed to HIV/AIDS (Zur Hausen, 1987: 5–7). Indeed, it was in the late 1980s and early 1990s that Zur Hausen’s interpretation grew legs, primarily among those investigating the relationship between HPV and cervical or anal cancer (see Schink and Lurain, 1991; Strigle, 1994). Nevertheless, the majority of American authors who cited Rigoni-Stern (most of whom presented more defensible interpretations of his work) did so to a similar end – to understand their inquiries as addressing a persistent puzzle in modern medicine (see Aurelian, 1984; Cobb, 1990; DiPaolo et al., 1996; Odunsi and Ganesan, 1997; Piper, 1985). Yet the continuity established by invoking Rigoni-Stern coexisted with a fundamental shift in how researchers used epidemiology to establish connections between sex and cancer. Epidemiological data were now used (often in tandem with molecular research) to indicate viral infections that triggered malignant disease processes at the molecular level.

This changing relationship between epidemiology and molecular biology was dramatized in the early years of the HIV/AIDS epidemic. ⁶ While the epidemic parts ways from the problem of sex and cancer relatively quickly, epidemiological trends would also soon implicate homosexual activity in the rising incidence of anal cancer among men. The first publication on the trend came in 1982, when epidemiologist Janet Daling and her colleagues compared medical statistics with marital records to demonstrate that rates of anal cancer were higher among gay men than other populations, and suggested that something about male-to-male sexual activity was responsible (Daling et al., 1982). The initial formulation of the problem is remarkably similar to previous eras, right down to their use of marital status to indicate sexual activity. In the absence of data on actual sexual behaviors, the authors cross-listed the names of never-married men with syphilis records to indicate anal-receptive homosexual activity. While Daling and colleagues justified these ‘correlates of homosexual behavior’ by referencing significant epidemiological trends, in the absence of similar data on anal intercourse the imputation of anal-receptive activity to all gay men and the supposition that gay men never marry instead correspond with cultural assumptions about homosexuality that became particularly striking during the HIV/AIDS epidemic.

As the inclusion of sexually transmitted infection in their research design indicates, the trajectory of Daling and her colleagues’ subsequent research would be markedly different from that of epidemiologists in previous decades. By 1989, Daling had co-authored a paper that provided compelling laboratory evidence that HPV infection was responsible for anal cancer. The NCI-funded study was conducted by a large interdisciplinary team of pathologists, molecular biologists, and epidemiologists who used DNA hybridization techniques to detect viral DNA from the HPV strains associated with malignancy in anal tumors (Beckmann et al., 1989). The epidemiological component of the study was secondary to the causal inferences gleaned from laboratory analysis; etiology was assured by the presence of viral DNA, which, with the development of transfection techniques, could now in principle show a causal relationship between infection with oncogenic viruses and cancer (see Morange, 1997).

While the 1989 study corroborated a growing body of evidence indicating HPV as a cause of many anogenital cancers, the link between ‘homosexual activity’ and anal cancer was just beginning to garner attention. Using Daling’s studies as a starting point, Stephen Strigle (1994) proposed a connection between anal cancer and anal-receptive intercourse among gay and bisexual men through analogy with Rigoni-Stern’s findings, which were once again presented as based upon a comparison of cancer rates between nuns and prostitutes. In this instance, however, Strigle relied upon laboratory investigations of cytological samples taken from patients to corroborate the connection. University of California (UC), San Francisco professor of medicine Joel Palefsky, who became the leading expert on HPV and anal cancer in the United States, continued this line of inquiry. Throughout the 1990s, Palefsky conducted dozens of studies detailing the association between HPV and anal cancer in gay men (and later women), especially those with HIV/AIDS. ⁷ These studies used epidemiological trends as their starting point (and often as their sampling source), and then explored molecular interactions between virus particles and host cells in order to establish a causal connection between HPV infection and anal cancer.

By the close of the 1990s, researchers investigating the link between sexual activity and anogenital cancers had accumulated sufficient evidence to implicate HPV in many penile and anal cancers and most cervical cancers. In 1999, a benchmark study announced that it had established HPV as a ‘necessary cause’ of cervical cancer (Walboomers et al., 1999). This study, which re-analyzed serological and histological samples previously determined to be HPV-negative using a more sensitive probe, showed that HPV DNA was in fact present in over 99 percent of all samples analyzed. The authors proposed the near-ubiquity of HPV DNA in these cancers as evidence that sexually transmitted infection with HPV is the cause of most cervical cancers worldwide. This confirmation rested upon a firm epidemiological foundation that established strong and consistent associations between viral exposure and the development of pre-cancerous or cancerous lesions. Yet it was molecular studies that were cited as proof of the causal role of HPV infection in the development of anogenital cancers. ⁸

Conclusion

Working to achieve continuity with the claims of an amateur statistician from 1½ centuries prior may seem inexplicable given the complex changes in theory and research design that accompanied the shift from epidemiological to molecular explanations of cancer etiology. Yet cancer researchers’ preoccupation with constructing continuity with Rigoni-Stern’s original proclamation continued even after the criteria of his direct inheritors in epidemiology were ruled insufficient to establish the necessary cause of anogenital cancers. In fact, invocations of Rigoni-Stern have continued to increase since the turn of the 21st century. Still considered ‘a pioneer in the perception of the pathogenic mechanism of cervical cancer’, Rigoni-Stern has been referenced ‘in almost 70 papers published from 2000 to 2008’ (Gasparini and Panatto, 2009: 4).

The increase in references to Rigoni-Stern suggests the continued resonance of this originator narrative in light of the manufacture of Gardasil and Cervarix, vaccines that protect against the two HPV strains most strongly associated with cervical and other anogenital cancers. Recent scholarship on the HPV vaccines (Aronowitz, 2010; Mamo and Epstein, 2014; Mamo et al., 2010) shows how sexual behaviors as risk factors continue to play an important role in public health campaigns against cervical cancer decades since HPV was declared a necessary cause of those cancers. ⁹ The popular debates around HPV vaccination, alongside the 2008 Nobel Prize in Physiology honoring Zur Hausen’s role in establishing a causal connection between HPV and cervical cancer, highlight the role originators play in narratives about scientific innovation and change.

While this study was introduced as illustrative of the theoretical gain in thinking about originators as cultural symbols rather than formal strategies for persuasion, the case of Rigoni-Stern also dovetails with the literature on memory in science. As Geoffrey Bowker (2005) notes, memory is itself a practice, one that allows scientists to act effectively in the present by interpreting the past. To date, memory practices in science have largely been analyzed in connection to commemorative events, wherein scientists construct formalized, ritualized, and often politically oriented genealogies of their disciplinary past around important historical figures. These genealogies are for collective consumption – they defend controversial figures, projects, or events to the general public, or they generate a sense of community among scientists through the creation of shared memories and traditions (see Abir-Am, 1999; Barberis, 1999; Goldberg, 1999; Haddad, 1999; Smocovitis, 1999). Yet, as I have argued, the substance of these genealogies should also be studied in light of their relationship to scientists’ work. An appreciation of the content of originator narratives as interpretive events can enhance accounts of collective memory by demonstrating how originator narratives draw upon understandings of research practice and cultural context to become meaningful and ‘real’ enough to gain their rhetorical force.

As many STS scholars have pointed out, scientists’ stories about the past tend to universalize the present, to construct the current state of knowledge as inchoate in the works of past figures (Bowker, 2005; Latour, 1996; Pestre, 1999). Scientists routinely accommodate novel findings to existing bodies of literature (Fahnestock, 1986) in an effort to reconcile ‘the tension between change and continuity in science’ (Sinding, 1999: 77). STS scholars have shown continuity as a rhetorical achievement that enables scientists to construct a shared past in reference to important conceptual resources, such as measurement systems and objects of analysis (see Evans, 2010; Van Helvoort, 1994). The case of Rigoni-Stern suggests that efforts to construct continuity through originators are also closely tied to interpretive understandings that develop in reference to research practices as situated in their cultural context. They encapsulate scientists’ interpretations of their own research practices refracted through accretive cultural symbols transmitted in published technical literature. In this case, the changing interpretations of Rigoni-Stern’s treatise mirror shifts in theory–method packages, measurement techniques, and research protocols that followed epistemological and disciplinary realignments in the field of cancer research; at the same time, they reflect broader shifts in assumptions about sexual activity. This case study suggests that, while the narrative form of the originator may persist because this trope serves a more general strategic or mnemonic purpose, the content of originator narratives cohere because scientists make them meaningful and plausible in relation to their work.