A New Clinical Collective for French Cancer Genetics A Heterogeneous Mapping Analysis

Science, Technology, & Human Values Volume 31 Number 4 July 2006 431-464 © 2006 Sage Publications 10.1177/0162243906287545 http://sth.sagepub.com hosted at http://online.sagepub.com

Pascale Bourret INSERM Research Unit 379 (Social Science Applied to Biomedical Innovation) and Université de la Méditerranée (Aix-Marseille II)

Andrei Mogoutov Aguidel

Claire Julian-Reynier INSERM Research Unit 379 (Social Science Applied to Biomedical Innovation)

Alberto Cambrosio McGill University Collaborative forms of work such as extended networks, expert groups, and consortia increasingly structure biomedical activities. They are particularly prominent in the cancer field, where procedures such as multicenter clinical trials have been instrumental in establishing the specialty of oncology, and subfields such as cancer genetics, where bioclinical activities—for example, testing for breast and ovarian cancer (BRCA) genes and follow-up interventions—are predicated on the articulation of a number of tasks performed by new clinical collectives. In this article, we examine the founding and development of a French bioclinical Authors’ Note: Research for this article was supported on the European side by grants from Institut National de la Santé et de la Recherche Médicale (INSERM) Intercommission 6 (n°4M612C) and Centre National de la Recherche Scientifique (CNRS)-INSERM-MiReDREES, SHS 2002 (n°ASE02038ASA) and on the Canadian side by grants from Canadian Institutes of Health Research (CIHR) (MOP-64372) and Fonds Québécois de la Recherche sur la Société et la Culture (FQRSC) (01-ER-70743 and ER-95786). Collaboration between the French and the Canadian authors was made possible by a 2001 INSERM senior invited researcher position and by a CIHR/INSERM 2004-2005 international exchange award. We would like to thank the Groupe Génétique et Cancer members who welcomed Pascale Bourret and Claire JulianReynier within their group (a special thanks, in this respect, to the group’s executive secretary, Dr. Catherine Nogues) and who kindly responded to our many requests for information about their work. All correspondence concerning this article should be addressed to Pascale Bourret. 431

432

Science, Technology, & Human Values

collective—the Groupe Génétique et Cancer (GGC)—that coordinates and structures the activities of most French actors in cancer genetics and operates simultaneously in the clinical, research, and regulatory domains. To examine the group’s structure and dynamics, the article combines information gathered through traditional fieldwork methods with information elicited from a coauthorship and semantic-network analysis of the publications of GGC members from 1969 to 2001. Keywords: genetic testing; breast cancer; French cancer genetics; coauthorship; semantic networks; Reseau-Lu

A

s argued in a recent article (Bourret 2005), biomedical activities are increasingly structured by the presence of clinical collectives. This is particularly true of fields at the therapeutic frontier, such as oncology, where cancer meets the new genetics. Within these collectives, a heterogeneous set of actors interacts in a number of different ways by establishing flexible collaborative arrangements at the national and international level. These interactions give rise to novel practices, engendering and regulating the human (e.g., family patients, oncohematologists) and nonhuman (e.g., genes and mutations) entities mobilized by bioclinical activities. Several authors have discussed the rise of collective configurations of actors in the biomedical field, and in particular, the increasing role played by research networks and consortia in the production of knowledge (e.g., Callon 1991; Vinck 1992; Cassier 1998; Cambrosio, Keating, and Mogoutov 2004; Gaudillière and Rheinberger 2004). This is not an entirely new phenomenon. As a consequence of advances in medical knowledge and technologies, medical work in modern hospitals cannot be dissociated from a complex web of interdependencies between a number of diagnostic, support, and clinical specialties (e.g., Gosselin 1985). The management of cancer patients, for instance, has been linked from its beginnings in the first half of the twentieth century to the interaction of various specialties, initially surgery and radiotherapy (Pinell 1992; Van Helvoort 2001), and after World War II, chemotherapy (Keating and Cambrosio 2002), to which one now should add nursing oncology, psycho-oncology and a number of other ancillary specialties, including laboratory ones. In the French case, this kind of teamwork has found institutional expression in the early development of a specific network of hospitals called Centers for the Fight Against Cancer (Centres de lutte contre le cancer, henceforth CLCCs) (Ménoret 1999; Pinell 1992), of which twenty presently exist in various key locations throughout France. When we speak of new clinical collectives, however, we refer to more than simply teamwork within hospitals. The development of chemotherapy

Bourret et al. / Cancer Genetics

433

has led, for instance, to the development of medical oncology as a specialty grounded in the performance of multicenter clinical trials (Marks 1997; Löwy 1997), that is, a set of procedures simultaneously carried out in several locations and linked by the establishment of conventions (statistical and otherwise) regulating the production, analysis, and publication of results. Diagnostic, prognostic, and therapeutic practices increasingly refer to evidence-based results (e.g., Timmermans and Berg 2003) that are similarly grounded in extended biomedical networks of laboratory, clinical, and regulatory activities. Thus, as argued by Bourret (2005), in addition to local, multidisciplinary teams, recent years have seen the development of data collectives (groups of experts devoted to the production and circulation of statistical and epidemiological data based on large population studies and meta-analysis) and of new bioclinical collectives of which the cancer genetics group analyzed in this article is an example. Briefly, this latter kind of collective is not concerned solely with the interfacing of skills or the construction of data and tools. Rather, its interventions target the production of medical judgment and medical decision making. In fields of activity such as cancer genetics, self-characterized by the presence of incomplete knowledge and a high level of uncertainty, new bioclinical collectives organize the discussion of clinical cases and produce informal rules and conventions as well as formal practice guidelines to support decision-making activities, thus directly affecting the nature and content of clinical work. If we are correct in maintaining that the organization of biomedical work and judgment is shifting to a new, collective configuration, then the question becomes how such a configuration can be further analyzed and characterized. This question raises both methodological and substantive issues. As far as methodology is concerned, in a related article dealing with a different biomedical area, we argued that confronted with large-scale collaborative endeavors, both local ethnographic methods and the resort to quantitative indicators provide unsatisfactory research avenues insofar as they either miss the figurational dimension of the collaborative network or destroy the very phenomena under investigation. We thus championed the combination of ethnographic methods with a computer-based analysis of heterogeneous relational data (Cambrosio, Keating, and Mogoutov 2004; see also Dodier and Barbot 2000; Callon 2001). On the substantive side, several authors investigating networks and consortia, especially those linking the public and private domains, have focused on the peculiar arrangements and agreements developed by these institutions for managing the knowledge they produce with respect to both data circulation and ownership (e.g., Hilgartner 1998; Cassier 1998). While this is certainly an important research endeavor, our focus is different,

434

Science, Technology, & Human Values

and in a sense, more upstream: we are interested in understanding the development of hybrid biomedical collectives, such as the one analyzed in this article, whose activities bridge the research, clinical, and regulatory domains, thus raising epistemic issues that are intimately connected to the evolving material and organizational arrangements that characterize these collectives.1 To sum up, our goal in the present article is twofold. On the methodological side, we want to exemplify a distinctive, semiquantitative approach to the heterogeneous mapping of research activities. On the substantive side, we want to use this approach to analyze the collective driving force behind the development of French cancer genetics, thus also contributing to a better understanding of the dynamics of collaborative research endeavors within a key domain of contemporary biomedicine.

Material and Methods Material This article examines collaborative activities within a subfield of oncology, cancer genetics, in a country, France, where contrary to other countries (Parthasarathy 2004), cancer genetics clinical consultations are exclusively performed by multidisciplinary medical teams located within hospitals. These hospital teams maintain strong links to extended collectives, such as national and international collaborative networks and expert groups. We will examine a national collaborative network of clinicians and researchers called Cancer and Genetics Group (Groupe Génétique et Cancer, henceforth GGC). The GGC was established to promote and organize clinical and research activities in cancer genetics and to discuss the clinical, laboratory, and regulatory issues (including the production of practices guidelines) raised by these practices. As readers will recall, human cancer genetics investigates and clinically manages hereditary forms of cancer, that is, the presence within families of genetic mutations leading to an increased susceptibility to neoplastic diseases. As compared to more traditional forms of medical practice, human cancer genetics stands out both for its focus on family patients rather than individual patients and for its focus on the risk of developing a disease rather than on diseases as such (Bourret 2005). The field has experienced a recent boom; in France, for instance, approximately five thousand specialized cancer genetics consultations took place in more than fifty locations nationwide in 2000 (Sevilla et al. 2004).

Bourret et al. / Cancer Genetics

435

The GGC came together informally at the end of the 1980s, immediately after the creation of the first cancer genetics consultation services between 1987 and 1990. It acquired official status in December 1991 by integrating the National Federation of CLCCs (Fédération nationale des centres de lutte contre le cancer, henceforth FNCLCC), the previously mentioned network of regional hospitals specialized in the treatment of cancer, in which most of the consultation and laboratory services for cancer genetics were first located.2 At the time of our data collection, the GGC included ninety-three members. A bottom-up organization, it initially assembled all the health professionals in charge of cancer genetics consultation services and laboratory testing in France. Only recently, following the dramatic increase in these activities, have a few hospital-based consultations been established outside the network. Still, even at the time of this writing, the GGC membership almost entirely overlaps the French clinical cancer genetics community. We will further examine the development, structure, activities, and role of the GGC in the main section of this article, where we present the results of our analysis.

Methods Our analysis of the GGC resorts recursively both to qualitative (ethnographic) and semiquantitative (computer-based mapping) methods. In other words, the latter are not self-contained, nor are they an end in themselves; computed-generated maps are interpreted using information elicited by ethnographic means, and in turn, they can be used as a starting point for further ethnographic inquiry, for instance, as a heuristic focus point for follow-up interviews. In the present case, the first author staged a group presentation (followed by discussion) of our mapping results during a regular working meeting of the GGC. She also interviewed eight members of the group to elicit specific comments about the maps. Space limitations prevent us from presenting in this article a fully integrated discussion of ethnographic and mapping results. Ideally, the present text should be read as a companion article to the detailed ethnographic analysis reported in Bourret (2005). Fieldwork The first and third authors conducted qualitative fieldwork in two different settings, a local cancer genetics clinic and the national collaborative network itself, in which they were granted membership. They were thus able to observe the organization and development of local and national

436

Science, Technology, & Human Values

clinical and biological activities thanks to repeated, informal discussions and formal interviews with the main actors. They attended the GGC meetings since the mid-1990s and were able to gain access to a large number of internal GGC documents, such as minutes of general meetings and special interest committees, scientific data, collective statements for internal or external use, and documents concerning the development of collaborative projects and the drafting of recommendations and guidelines. Additional information was provided by the qualitative analysis of the publications of GGC members. As previously noted, a detailed compilation of fieldwork results is presented in Bourret (2005). Computer-Based Mapping For the semiquantitative analysis of the GGC, we focused on the group’s publications as intermediaries between actors rather than on the attributes of individual actors (Callon 1995). Indeed, as detailed further below, publications provide links between heterogeneous actors such as researchers, clinicians, patients, genes, techniques, laboratories, and institutions and allow us to follow the transformation of these relationships through time. They also can be collected relatively easily thanks to the existence of electronic databases such as PubMed/Medline. We began by searching PubMed (from 1969, the earliest date for which we found an article, to 2001) for the publications of the ninety-three GGC members. We then asked each member to revise the list of his or her publications by adding missing contributions and eliminating those that had been wrongly assigned to them because of homonymy. The final Excel database contained 3,596 references featuring all the main Medline fields (authors, titles, the exact references and publication years, keywords, and abstracts when available) and included articles published by present-day GGC members before the creation of the group and before a given author had joined the group. To examine the historical evolution of the GGC, we subdivided the database into four periods: 1969–1986 (N = 550), 1987–1991 (N = 671), 1992–1996 (N = 1,075) and 1997–2001 (N = 1,300). This division, while somewhat arbitrary, is based on the following criteria: 1. Each period covers five years except the first period, which predates the existence of the GGC and during which we found only a relatively small number of publications. 2. The first year of the second period corresponds to the beginning of clinical activities in the field of cancer genetics and to the informal coming together of the GGC.

Bourret et al. / Cancer Genetics

437

3. The first year of the third period corresponds to the official creation of the GGC.

The division into periods thus has allowed us to explore the following stages of the group’s development: 1. The prehistory of the GGC, featuring the initial activities of its future members (Period 1) 2. The beginning of clinical activities in France in the absence of robust information about the location and nature of cancer genes and the informal, bottom-up beginnings of the GGC (Period 2) 3. The official beginnings of the group’s activities coupled with the performance, at the international level, of a number of genetic linkage studies culminating in 1994–95 with the identification of the first two breastcancer genes (BRCA1 and BRCA2; see Dalpé et al. 2003) (Period 3) 4. The blooming of cancer genetics testing and counseling activities and related regulatory issues (Period 4)

We analyzed the database by resorting to the following three strategies: 1. Using the author field, we mapped and explored coauthorship patterns both within the GGC (endogamy) and between GGC members and external collaborators (exogamy). There is a well-established tradition within the bibliometric and scientometric literature of analyzing coauthorship patterns, for instance, to look for connections between these patterns and other features such as citations and citation effect (e.g., Glänzel 2002), to define indicators for science policy schemes (e.g., Melin and Persson 1996), or to investigate different kinds of scientific collaboration as part of science evaluation and technology-assessment studies (Laudel 2002; Peters and Van Raan 1991). Large-scale coauthorship patterns have also been analyzed as part of a more recent approach that combines insights from the physical sciences with social-network analysis to examine the statistical properties of networked systems such as small-world structures and power-law degree distributions (e.g., Newman 2001). In the present article, we resort to the analysis of coauthorship patterns to investigate the nature of the GGC— was it mainly a political lobbying institution or did it lead to actual collaborations between its members?3—as well as the dynamics of its subgroups. 2. Using title words,4 we analyzed coword patterns to examine the content of the group’s publications and activities. 3. Finally, by combining the previous two strategies, we produced heterogeneous maps to analyze the mutually constitutive links between human actors and the nonhuman entities mobilized by their work.

438

Science, Technology, & Human Values

To draw the following maps, we opted for Reseau-Lu 9.33 (http://www .aguidel.com), a network-analysis software program specifically designed for the treatment and mapping of complex, heterogeneous relational data so that they can be visually inspected and interpreted: 1. Coauthorship maps. Given the large number of authors included in the database (more than eight thousand) and given the fact that maps featuring more than 200 nodes quickly become unreadable, we limited our analysis to the top-producing authors. A top-200-producers limit corresponded to a publication threshold of approximately thirteen or more articles per author during the entire period. While the database contains all the articles published by GGC members, many of their coauthors obviously are not members of the group. Accordingly, the list of the top producers contains a majority of GGC members but also several external collaborators.5 2. Thematic (coword) maps. We analyzed the titles of each period’s articles by treating them as a single text with each title counting as a separate sentence within that text. As compared to traditional coword analysis (e.g., Callon, Law, and Rip 1986; Cambrosio et al. 1993), we innovated by selecting the co-occurring terms on the basis of their textual relevance as measured by their semantic weight and not merely by their frequency. Briefly, we ran each document through TextAnalyst 2.3 (http://www .megaputer.com), a text-analysis software program that works by eliminating common words carrying no semantic meaning, identifying word stems (lemmatization), establishing a list of relevant concepts (individual words and word combinations),6 calculating the statistical frequency of individual and joint occurrences, and finally, assigning a semantic weight to the latter by renormalizing the frequency weights through a Hopfieldlike neural-network procedure. In a second departure from traditional coword-analysis methods, co-occurring expressions with a high semantic weight within each period were then mapped using Reseau-Lu. Given the nature of the database (cancer genetics), terms such as cancer and gene were obviously widespread and ranked quite high on the semanticweight scale, threatening to cover the resulting maps with a tight network of uninformative links.7 To simply eliminate those (and similar) terms from the analysis did not amount to a satisfactory solution since, for instance, combined expressions like breast cancer would have been affected. Fortunately, we were able to use a feature provided by TextAnalyst that allows users to define certain terms as common words, that is, as terms that only will be taken into account when associated with semantically meaningful terms, thus also eliminating artifactual clusters resulting from the presence of generic and polysemic terms. 3. Heterogeneous maps. We linked a subset of terms appearing on the thematic maps (selected according to semantic weight) to the coauthors of

Bourret et al. / Cancer Genetics

439

the articles in whose title they occurred and mapped the resulting relations by treating human and nonhuman actors (coauthors and terms) on the same level.8 Heterogeneous maps are, in a sense, a nonpurified version of coauthorship and thematic maps: the latter are produced by first linking authors or terms as connected by publications and by subsequently eliminating the intermediaries (the articles) to produce a purely social or semantic network.9 Heterogeneous maps reestablish the initial complexity of the network.

Reseau-Lu uses a dynamic positioning algorithm simulating the interaction between objects. It does so through a three-step optimization process: (1) global initial positioning of the object vis-à-vis all the other objects in the space, (2) micro-optimization of the positioning of the object vis-à-vis the other objects to which it is directly connected (network neighbors), and (3) meso-optimization of groups of highly connected objects (clusters). The optimization process depends on explicit rules defining symmetry properties, structural equivalence of points inside the structure, centrality, and betweenness of objects. The resulting map has no axes. The graphical conventions adopted in all maps are basically the same. The nodes (circles and squares) correspond to authors or title words and the lines connecting them to the existence of a coauthorship, coword, or authorcoword link. Each node appears only once on each map, and its size is proportional to the relative importance of a given author or title word. The size of the connecting line is also proportional to the number of coauthorship or co-occurrence ties. The location of nodes in the two-dimensional space and the length of the connecting lines have no metrical meaning. Reseau-Lu, however, positions closely related words and/or highly collaborating authors close to each other; clusters of authors and/or words thus can be easily visualized, and so can words or authors acting as a bridge between different clusters. We will provide further technical details when needed in the course of our discussion of the results, to which we now immediately proceed.

Results and Discussion Before turning to the maps, let us examine a few quantitative trends. The average number of coauthors of the articles listed in our database shows an increase from 5.2 during Period 1 to 6.3 (Period 2), 7.1 (Period 3), and 8.1 (Period 4). Figure 1 exhibits this pattern in detail; articles with more than ten coauthors are mostly to be found during Period 3, and especially, Period 4. While this increase in the number of coauthors corresponds to a general

440

Science, Technology, & Human Values

Figure 1 Distribution of GGC Database References by Period and Number of Authors

trend in the biomedical sciences (e.g., Haiqi 1997; Cronin 2001), for our present purpose we will use it as an indicator of change in the GGC collaborative practices. A first hypothesis in this respect, which the maps will allow us to explore further, is that the development of the group has led to an increase in collaborations both inside the group and with outside colleagues. If we now look at the endogamy pattern (i.e., the percentage of collaborations within the GGC vs. collaborations with outside researchers), we find that it shows a decrease from 20 percent during Period 1 to 16 percent during Period 4. Thus, the increase in multiauthored collaborations was not at the expense of collaborations with the outside world (more on this further below). It is now time to analyze the fine structure of these general trends by examining the three kinds of maps (coauthorship, thematic, and heterogeneous) for each period. Period 1 (1969–1986) Figure 2 shows the coauthorship map for Period 1, when the GGC did not yet exist.10 No strong collaborative pattern seems to exist between a majority of future GGC members (in capital letters on the map; external collaborators

Bourret et al. / Cancer Genetics

441

Figure 2 Coauthorship Map of Period 1 (1969–1986)

teyssier g.richard o. damon g. freycon f. bethenod m. fraisse j. turc-carel c. FRAPPAZ D. berthier j.c. najean y. philippe n. douste-blazy l. maret a. le coniat souillet g. dellagi k. marty m. m.vecchione d. negre a. favrot m.c. brunat-mentigny m. gluckman e. salvayre r. philip i. valensi f. schaison g. bertrand biron p. preud'homme j.l. flandrin g. bryon s. p.a. vuillaume m. daniel m.t. john t.j. freese u.k. favrot m. philip t. sigaux f. brouet j.c. venkitaraman caubet j.f. dreyfus dore j.f. a.r. weh h.j. bornkamm g. mauchauffe m. j.c.g. klein larsen c.j. berger r. seigneurin j.m. LENOIR G. JUNIEN C. mathieu-mahul d. murphy gest j. BERNHEIM A. leder w. p. de the g. colbert n. despoisses s. kristoffersson u. theillet c. taub r. potter h. de la chapelle a. griscelli c. LIDEREAU R. moulding c. magaud j.p. metezeau p. mitelman f. escot c. daillie j. battey j. callahan r. goldberg m. virelizier j.l. de grouchy j. spyratos f. ooka t. oglobine j. kaplan martinez m. CALENDER A. fellous m. cohen j.h. MIGNOTTE H. j.c. desplaces a. gessain a. GRANDJOUAN S. lalouel j.m. j. epstein m.a. mikaeloff p. LEROUX D. couturier d. frezal j. rowe m. rooney c. BONAITI C. briard m.l. hoffman t. jalbert p. moss d.j. STOPPA-LYONNET D. clerget-darpoux f. bonvini e. hors j. rickinson a.b. jalbert h. BOUGNOUX P. COHEN-HAGUENAUER O. feingold n. pfister a. delpech b. busson m. hochez j. deschamps i. zeitoun p. da lage c. broyer m. garbe e. schmid m. RICHARD S. feingold j. CHEVRIER A. cattan a. namer m. DEMENAIS F. van cong n. NASCA S. khater r. jezekova d. dubrasquet m. elston r.c. coninx p. bois e. bonfils s. FRENAY M.

2

1

3

legros m.

bonnefond a. douchez j. combes p.f. NGUYEN T. blanchet-bardon c. panis x. DEMANGE L. kress m. duvillard p. rosset r. delmas v. prade m. lissitzky s. FEUNTEUN J. AVRIL M.F. monier CAPODANO A. scherneckr.s. mathieu a. stahl a. jordan b. devictor-vuillet m. luciani j.m. koehl c. zimmermann p. ABECASSIS J. ZUMMER K. methlin g. chavy favre a. kac eber m. a. j. THOMAS G.

froissart d.

fossati p. romon m. wemeau VENNIN P. j.l. lefebvre j. vandewalle b. PEYRAT J.P. demaille a. j. adenis l. beuscart r. bonneterre j. demaille m.c. dalifard i. DAVER A. larra f. djiane j. hecquet b.

kelly p.a.

4

chassevent a. bertrand g.

Note: Only authors belonging to the top-200 layer are represented (GGC members in capital letters; outside authors in lowercase).

are in lowercase) before the actual constitution of the group. At the top of the map, we see two large clusters (1 and 2). A series of small, fragmented clusters sits at its bottom, and a medium-sized cluster (3) is located in the center. These clusters, on the basis both of the visual evidence presented by the maps and of our preexisting knowledge of the field, can be characterized as follows: • Cluster 1 features the strong presence of a prolific author—Gilbert Lenoir, at the time based at the International Agency for Research on Cancer (IARC) in Lyon—a leading molecular geneticist from France and a pioneer of the use of the new genetics in the field of cancer, especially

442

Science, Technology, & Human Values

lymphomas (see Keating and Cambrosio 2001). The maps clearly show that he engaged in multiple collaborations with several groups of authors. • Cluster 2 features two prolific authors—Alain Bernheim and Roland Berger (the latter not a GGC member)—both Paris-based cytogeneticists linked to a similarly Paris-based group of hematologists and pediatricians. Clusters 1 and 2 are interconnected, and this connection thus links two major French institutions in the field of cancer, namely the IARC in Lyon and St. Louis Hospital in Paris. • Cluster 3, more loosely linked to the previous two, consists mainly of Paris-based clinical geneticists who were instrumental in establishing that field in the 1960s and early 1970s (Bourret 1988; Bourret and Huard 1990; Gaudillière 2000). It is important to differentiate their work, which focused on traditional (Mendelian, monogenic) genetic diseases (such as Huntington’s disease and cystic fibrosis), from work in molecular cancer genetics that deals with non-Mendelian patterns and low-penetrance susceptibility genes.11

Finally, the bottom of the map consists of a series of small, isolated clusters corresponding to local institutions, mostly regional CLCCs, within which we find, in each case, one or two future members of the GGC who were at the time at the beginning of their careers. The largest of these small clusters corresponds, for instance, to people based in Lille (cluster 4). Thus, a look at the prehistory of the GGC highlights three elements, namely the multiple roots—cytogenetics, molecular virology, Mendelian human genetics, and hematology-oncology—of its future activities, the structuring role still played at this stage by geographic location, and the leading role played at this stage by a few human actors. Researchers to whom we showed the map as part of follow-up interviews confirmed that it displayed meaningful patterns and that our tentative interpretation was basically sound. Interestingly, these researchers spontaneously began annotating the map, drawing circles around clusters, adding explanatory labels, and engaging in a detailed commentary, thus corroborating its usefulness as an ethnographic tool. Concerning cluster 1, Lenoir argued that his leading position within an international research center (the previously mentioned IARC) accounted for the visual pattern showing the presence of multiple national and international collaborations, with the biggest subcluster within cluster 1 consisting of local collaborators in Lyon, many of them clinicians. These collaborations, generally speaking, focused on somatic (as opposed to hereditary) cancer genetics with a special focus on two model diseases—Ewing’s sarcoma and Burkitt’s lymphoma—and on blood cancers (more on this further below). The latter account for the connection between clusters 1 and 2, since cluster 2 consists mostly of cytogeneticists

Bourret et al. / Cancer Genetics

443

and hematologists working, as previously mentioned, in the mecca of French hematology, St. Louis Hospital in Paris. We now turn to Figure 3, showing the thematic map of Period 1. We again notice a certain fragmentation with the presence of several distinct thematic clusters. While thematic fragmentation is not a necessary consequence of a fragmented coauthorship pattern—as different, unrelated groups could be working on identical topics—in the present case, coauthorship fragmentation translates (albeit to a lesser degree) into thematic fragmentation. Let us have a closer look at the map: • The biggest cluster (1) can be subdivided into two subclusters: on the left side, a subcluster centered on Epstein-Barr virus (EBV), and on the right side, a subcluster centered on Burkitt’s lymphoma. The latter is a form of cancer associated with EBV, and the two subclusters, taken together, refer to research on the viral and genetic etiology of Burkitt’s lymphoma. Among the words included in the Burkitt’s lymphoma subcluster, we find myc, oncogene, and translocation that refer to the cellular myc gene that causes cancer (Burkitt’s lymphoma) when moved (translocated) from its original chromosome to another and thus point to the emerging oncogene paradigm, linking it to its viral and cytogenetic roots (Morange 1997; Keating and Cambrosio 2001). • A small cluster (2) is centered on the term chromosome and includes terms referring to chromosome-analysis techniques (fluorescence) as applied to the study of Burkitt’s lymphoma and to other cancers, such as leukemia (e.g., Philadelphia chromosome, the telltale anomaly of chronic leukemia whose analysis was a turning point in the field). The themes found in clusters 1 and 2 thus correspond to the activities of the authors featured in clusters 1 and 2 of the coauthorship map of the same period. • Finally, we notice a third cluster (3) centered on the clinical aspects of breast cancer (diagnosis and treatment). As shown by related cowords, these aspects do not yet concern the genetic dimension of this disease. We can form the hypothesis that this thematic cluster refers to the activities of a number of the small clusters on the coauthorship map.

As explained by interviewees, who once again confirmed our general interpretation, the period covered by the map is relatively long (1969–1986) and it involves a transition in research focus. For instance, the left side of cluster 1 corresponds to the old approach to Burkitt’s lymphoma centered on the viral pathogenesis hypothesis (linked to immunology and seroepidemiology), whereas the right side of the cluster is related to the emerging cancer genetics approach focused on chromosomes and oncogenes. The coexistence of these two approaches on the thematic map can be accounted for by the fact that Lenoir was a key agent of the transition from one approach

444

Science, Technology, & Human Values

Figure 3 Thematic Map of Period 1 (1969–1986) technics squamous indications radiotherapy esophageal Hyperfractionated radiotherapy

cervix concentration platinum cisplatin cis

system chemotherapy

gastric secretion gastrin

breast cancer patients estrogen localization fluorescence human chromosomes

HLA

heterochromatin

2

Identification chromosomes

estradiol breast cancer human breast cancer human breast receptors prolactin steroid receptor prolactin receptors

anomalies Philadelphia chromosome

method

meiotic prophase

photoprotection Thiouridine

progesterone receptors breast

hybridization

juvenile insulin-dependent diabetes

application

oocyte

primary breast

malignant melanoma

detection nucleolus

breast carcinomas ras

Fanconi

carcinogenesis theory

chromatin

3

treatment

esophageal cancer colon complications Determination transposition anemia microglobulin secretion

hormone

linkage Relation

growth Escherichia coli

Indies

chromosomes abnormalities

xeroderma pigmentosum enzyme

coli

methylation

precursor

epidemiological

Phospholipid

RNA

Ewing sarcoma

seroepidemiology

prevalence

abnormalities immunoblastic

human B lymphocytes polyoma virus ribosomal Inhibition interferon

sis

Hodgkin marrow

Martinique

restriction lymphomas

Establishment

HTLV

DNA

Structural proteins proteins

markers porphyria Isolation mutant

lymphocytes

sequences

Characterization

activity virus

region

Cytogenetic lymphoma

Burkitt relationship

infectious mononucleosis

positive Burkitt Epstein-Barr virus cytotoxic antigens child EBV responses nuclear antigen EBNA Epstein-Barr virus nuclear antigen replication lymphoblastoid infection lymphoid expression serology India genome human lymphoblastoid

monoclonal antibodies alteration oncogene

karyotype

simian virus deletion

malignant lymphomas

DNA polymerase

alpha

origin

virus DNA

properties

carcinoembryonic antigen

antibodies

France

activation

rearrangements proto-oncogene

locus

myc myc oncogene lambda immunoglobulin phenotype translocation Burkitt lymphoma Variant translocation myc gene

1

Correlation lymphoma patients

reciprocal translocation human lymphoid

Note: Semantic weight > 50 (word combinations).

to the other.12 The occurrence of terms such as human lymphoblastoid within the old subcluster refers to an applied spin-off of virological work, namely the use of the Epstein-Barr virus for creating a new sort of research material, immortalized human cell lines. Cluster 2, according to one of our informants, could in fact be considered part of cluster 1, since it refers to

Bourret et al. / Cancer Genetics

445

Figure 4 Heterogeneous Map of Period 1 (1969–1986) KAC J. BONNETERRE J. LEFEBVRE J. VENNIN P. VANDEWALLE B.

CLERGET-DARPOUX F. AVRIL M.F. ZUMMER K. BRIARD M.L. DEMENAIS F. FEINGOLD J. FREZAL J. BONAITI C.

4

PEYRAT J.P.

3

Characterization rna FAVRE A.

precursor

breast

STAHL A.

MONIER R. THOMAS G. photoprotection escherichia ribosomal coli Thiouridine mutant NGUYEN T. growth

CAPODANO A.

responses FLANDRIN G. region

Isolation

chromosomes hybridization VALENSI F.

markers

dna FEUNTEUN J.

phenotype

cytotoxic

deletion relationship

BERGER R. Correlation

genome

DANIEL M.T. abnormalities BERNHEIM A.

ABECASSIS J.

DAVER A.

polymerase

activation

BROUET J.C. Cytogenetic LIDEREAU R.

methylation

antigens activity

France lymphocytes

LARSEN C.J. proto

BOUGNOUX P.

virus

expression

BORNKAMM G. properties

replication India

lymphoblastic MATHIEU-MAHUL D.

Phospholipid

lymphoblastoid

ebna Epstein Barr

rearrangements locus translocation lambda

myc immunoglobulin

lymphoid

SEIGNEURIN J.M. infection DE THE G. ebv mononucleosis

origin

LENOIR G.

Burkitt

2

oncogene LEDER P.

serology

lymphomas

CALENDER A. antibodies

Martinique HTLV

child Hodgkin

lymphoma sarcoma BRUNAT-MENTIGNY M. PHILIP T. PHILIP I.

FREYCON F.

sis

1

Ewing

FAVROT M. marrow

FRAPPAZ D.

Note: Top fifty human actors (black nodes and capital letters) and top (single words, semantic weight > 50) nonhuman actors (grey nodes and lowercase); 50 percent more specific ties selected.

work on chromosomal abnormalities in leukemia carried out by the same researchers working on Burkitt’s chromosomes using the same experimental system. Finally, informants explained that while clusters 1 and 2 refer to work undertaken by a well-defined group of collaborators as featured on the coauthorship map, cluster 3 refers to work on breast cancer carried out by separate regional groups. This work includes both a clinical and a research dimension, the latter focusing on receptors and not yet on issues such as genetic susceptibility that would later dominate the field. The heterogeneous map of Period 1 (Figure 4) introduces a new analytical dimension.13 The two main, overlapping clusters (clusters 1 and 2)

446

Science, Technology, & Human Values

correspond to the previously described work on Burkitt’s lymphoma, Epstein-Barr virus, and cytogenetics performed by the Lyon and Paris teams. The two clusters, however, feature different dynamics in terms of the constitutive articulation between humans and nonhumans. Cluster 1 is structured by a human actor (Lenoir) who established a number of links with nonhuman entities that in turn mediate his links with other human actors. Cluster 2, in contrast, is characterized by a set of tight links between human actors with common relations to a discrete set of biomedical entities. The overlap between the two clusters is provided by a thematic field corresponding to the emerging oncogene paradigm. Cluster 3 displays a structure similar to but even more extreme than that of cluster 2: the group it features is defined mainly by collaborative relations between human actors connected to a single research topic, breast cancer. Cluster 4 consists of a more balanced network of molecular biological topics and researchers. Finally, and in spite of the fragmentation characterizing this period, DNA sits at the center of the map, providing a generic common denominator to all the structured entities appearing on the map. Based on these comments, we see that heterogeneous maps have three main functions: 1. The first is a confirmatory or exploratory one: in the present case, since fieldwork had provided us with information about the main actors in the field, we were able to draw (in a figurative sense) connections between coauthorship and thematic maps, connections that are confirmed by the heterogeneous map. Had we not had this previous knowledge of the field, the heterogeneous map could have provided us with initial indications to this effect. 2. Heterogeneous maps also play a second, more analytical role: they point to the intermediaries connecting human and nonhuman actors and thus also to different collective configurations. Lenoir’s strategy, for instance, has been to diversify his research themes, and this in turn has connected him to different teams, whereas other research groups are characterized first and foremost by the relation between team members. In other words, while all clusters are by definition hybrid, their structuring principle can vary, privileging a configuration based mainly on nonhuman connections or a configuration transiting mainly through human links. 3. The previous remark applies to clusters within the map of a given period, but it can also lead to comparisons between maps of different periods. This third point will be further illustrated below.

Bourret et al. / Cancer Genetics

447

We have provided a detailed commentary on the maps of Period 1 to introduce readers to their many features. In what follows, we will limit our comments to the main patterns of each period. Generally speaking, the coauthorship maps of the next three periods show a clear-cut defragmentation trend: the network’s largest component progressively comes to include (almost) all the coauthors appearing on the maps. It could be objected that this is an obvious outcome given the somewhat anachronistic nature of our database that contains the retrospectively collected publications of presentday GGC members. But, as already noted, the group as we know it today could play a number of different roles—for instance, the role of a political lobbying group, the role of a learned society whose members only meet for discussion, or the role of a network of clinicians engaging solely in the treatment of patients—that do not necessarily include actual scientific collaboration and joint publications. As for the thematic maps, the main trend is the progressive increase of a breast cancer cluster: initially related to clinical and somatic work, this cluster takes a genetic turn, and during the final period, clearly dominates the map. Finally, the evolution of heterogeneous maps shows a transition from a situation characterized by the structuring role played mainly by a few selected individuals to a situation corresponding to a large hybrid collective structured by common nonhuman entities. Period 2 (1987–1991) The second period corresponds to the establishment of the first clinical services in the field of cancer genetics. The GGC does not yet officially exist but is already coming together informally. As a whole, the coauthorship map for this period (not shown), while still somewhat fragmented into several clusters, is no longer dominated, as in Period 1, by two large clusters standing in opposition to several small clusters. Rather, we see several medium-sized, interconnected clusters. The three clinicians who, during this period, created the first clinical services are located within different clusters. Somewhat similarly, the thematic map of the second period (also not shown) features a series of medium-sized clusters that can be related to the clusters found on the previous thematic map. The Epstein-Barr virus/Burkitt’s lymphoma cluster is still present but no longer dominates the map, having been reduced to a size equivalent to that of a now larger breast cancer cluster. The latter still points mainly to somatic topics as opposed to hereditarycancer genetics that will dominate subsequent maps; a few co-occurring terms

448

Science, Technology, & Human Values

already point, however, to the emergent interest in genetic susceptibility. Chromosome still occupies a strategic position with connections to several clusters, thus showing the role still played by traditional cytogenetics during this transition period but also the novel role of chromosome analysis in the search for cancer genes. Indeed, the second period is characterized by the emergence of linkage studies leading to the localization of putative cancer genes on specific chromosome locations; we will further discuss linkage analysis in commenting the maps of Period 3. We finally notice the presence of a number of terms referring to different kinds of cancer, such as colorectal, multiple endocrine, or hepatocellular, that bear witness to a transition from a period dominated by a model disease, Burkitt’s lymphoma, to a period characterized by the search for the genetic roots of different kinds of cancer. Period 3 (1992–1996) The third period is characterized by the official creation of the GGC. In the corresponding coauthorship map (Figure 5), the transition that began in the previous period has almost been completed. A large cluster occupies most of the map, and the more peripheral clusters are all connected to the main cluster. The latter can be further subdivided into subclusters. For instance: 1. Subcluster 1 corresponds largely to people working on breast cancer. The clinicians who established the first cancer genetics clinical services during the previous period are now members of this same subcluster and directly connected to each other. As noted by an interviewee, all the large nodes on the coauthorship map single out the GGC clinicians and researchers who established the first major clinical consultations in cancer genetics. Subcluster 1 also features a set of foreign authors who belong, with their French colleagues, to the international Breast Cancer Linkage Consortium (BCLC), a worldwide cooperative network devoted to the investigation of inherited breast and ovarian cancer. The consortium played an important role in the identification of the BRCA genes and subsequently has been involved in the characterization of the pathologies arising in BRCA gene carriers.14 The BCLC was founded in 1989 by Lenoir, then still a member of Lyon’s international agency, the IARC. Thus, it predates the official creation in 1991 of the GGC. At about the same time, another GGC member, Hagay Sobol, put forward a proposal to establish a French working group to draw up an inventory of families harboring multiple cases of breast cancer (Sobol et al. 1991). In 1993, twenty-five European member groups of the BCLC undertook a concerted action with funding

Bourret et al. / Cancer Genetics

449

Figure 5 Coauthorship Map of Period 3 (1992–1996) peiffert d.

carpentier y. GORISSE M.C. pernot m. desoize b. malissard l. DEMANGE L. marchal c. LUPORSI E. grange f. NGUYEN T. gamelin e. larra f. grisoli f. guillaume j.c. LORTHOLARY A. bey p. CAPODANO A. bognel c. margulis a. hoffstetter prade m. s. g. MOUSSEAU M. bolla m. chaplain GRANDJOUAN S. brice a. duvillard p. campion d. DEMENAIS F. martinez BOUGNOUX P. SPATZ A. DAVER A. clerget-darpoux f. m. martin c. AVRIL M.F. lagrange j. liscia d. friend s. FREBOURG T. duverger a. moreau v. hacene k. gambarelli d. terrier-lacombe m.j. callahan r. charbonnier f. cussenot o. danglot g. champeme m.h. lathrop g.m. PUISIEUX A. bieche i. gentet j.c. CHOMPRET A. latil a. LEROUX D. venuat a.m. cambien f. pouillart p. lemerle j. nguyen v c. rubie h. cropp c. jeunemaitre x. SOMMELET D. boileau c. poisson n. daniel m.t. tournade m.f. collod g. berger r. bonnardeaux a. hartmann o. LIMACHER J.M. SOUBRIER F. LIDEREAU R. moutou c. FRENAY M. bordigoni p. carrie c. charru a. ozturk m. milano g. BRUGIERES L. corvol p. fournier g. brunat-mentigny m. BONAITI C. JONVEAUX P. BRESSAC bouffet e. soussi t. nadaud s. jeanpierre c. B. chauvin f. bonneterre j. JUNIEN C. williams t.a. jacquemier j. henry i. BERNHEIM FRAPPAZ D. preudhomme c. negrier s. CUISENIER J. LASSET C. A. namer m. goguel a.f. ladenstein r. fenaux p. NOGUES C. thiesse p. philip t. PEYRAT J.P. JANIN N. blay j.y. zucker j.m. favrot m. j. freycon f. JULIAN-REYNIER C. EISINGER F. michon j. SCHLUMBERGER M. ayme mosseri v. favrot m.c. birnbaum d. VENNIN P. s. SOBOL H. weissenbach j. combaret v. moatti j.p. mark m. STOPPA-LYONNET D. MAUGARD C. chabal f. chambon p. FEUNTEUN J. LONGY M. magdelenat h. GORRY P. aurran y. peter m. BIGNON Y. LENOIR G. MAZOYER S. aurias a. lalle p. desmaze c. LAURENT-PUIG P. SINILNIKOVA O. westerveld a. DELATTRE O. smith weber b. narod s. plougastel b. THOMAS G. s. easton d. dutrillaux zucman j. buffet c. devilee p. melot t. demczuk ford d. b. rouleau g. stratton lynch h. bijlsma e. s. m. tonin p. hamelin r. merel p. hulsebos t.j. hoang-xuan k. ponder b. ruttledge m. dussaix e. sanson m. remvikos y. watson p. OLSCHWANG S. sylla b. resche f. schuffenecker i. muleris m.salmon r. RICHARD S. thuille b. GIRAUD S. flejou j.f. CALENDER A. aletti p.

1

2

Note: Only authors belonging to the top-200 layer are represented (GGC members in capital letters; outside authors in lowercase).

from the European Commission allowing them to further develop their infrastructure. All these activities were, in an important sense, mutually constitutive: international initiatives also strengthened collaborative ties within national collectives, and the latter, in turn, fueled the activities of the international consortia. This is consistent with our previous comments on the endogamy versus exogamy percentage (see our discussion of Figure 1). There has been an increase in collaborative ties within both national and international collaborative groups: a multiplier effect, rather than a zero-sum game. 2. Subcluster 2 consists of people working on other kinds of cancer, mainly colon cancer but also pediatric cancers and a particular genetic condition

450

Science, Technology, & Human Values

known as neurofibromatosis (NF2) that, together with other rare conditions (such as multiple endocrine neoplasia [MEN] and Ewing’s tumor), played an instrumental role in the development of French research on cancer genetics. Colon cancer raises an interesting question: our maps show the dominance of cancer genetic investigations of breast over colon cancer, yet the incidence (number of new cases per year) of these two pathologies is similar. How to explain, then, the fact that the former plays a structuring role in both the coauthorship map and the thematic map (see below) while the latter is contained in a marginal cluster? A possible explanation points to the paradigmatic role played by BRCA1 and BRCA2 genes and related mutations in the constitution and legitimation of the clinical cancer genetics field.15 This situation does not seem to be exclusive to France: while one should be careful not to conflate clinical with publication activities, a recent survey of the former in six European countries and Israel shows that a large majority of cancer genetics activity concerns breast cancer (Hopwood et al. 2003). Another, concurrent explanatory element points to the fact that the hereditary aspects of certain forms of colon cancer had been recognized long ago, even in the absence of detailed information about the putative genes involved in this process. They thus have been integrated into the activities of gastroenterologists, who are not located within CLCCs and have continued to function as a distinct network even after the establishment of the GGC.

Members of the large central cluster are not housed in a single institution but rather are located in several French cities. Their connections thus are not based on a common institutional affiliation but rather on their participation in collaborative projects both within the GGC and with international research consortia working on the identification of cancer genes. Activities in the cancer genetics field have thus reached a stage where they necessarily involve multicenter endeavors, for instance, for the gathering and sharing of family data. This is consistent with information obtained from archival documents. Immediately after its establishment, the GGC defined its main objectives as the assessment of genetic risk for cancers, the development of screening procedures to manage at-risk patients and families, and the pursuit of clinical and molecular research in cancer genetics, with a focus on the production and validation of clinical expertise (GGC 1992). Clinicians interested in creating a cancer genetics service within their institution thus were able to profit directly from the experience and expertise resulting from the initial establishment of three consultation services in Lyon, Clermont-Ferrand, and Paris. Moreover, GGC meetings were devoted to the collective discussion of clinical cases (e.g., family pedigrees) and related clinical management procedures.

Bourret et al. / Cancer Genetics

451

Finally, from the outset, the GGC insisted on its role in developing research activities aimed at contributing to “the enhancement of knowledge on genetic predisposition to cancer” (GGC 1995). Translated into practice, this general objective led initially to the organization of national cooperative research projects based, as already noted, on the collecting and sharing of family data with a view toward performing both molecular analysis and clinical research. Indeed, one of the first articles published by the group states that cooperation between GGC members went far beyond “the simple sharing of medical information and the collection of biological material necessary for research projects and for diagnostic tests” (Sobol et al. 1991, 410, our translation). Research, from the very outset, loomed large among the stated goals of the GGC. The thematic map of the third period (Figure 6) further specifies several elements emerging from our analysis of the coauthorship map. The map clearly illustrates the cancer genetics turn, as shown by the three main, strongly interconnected nodes (and related cowords) occupying the bottom half of the map: breast/breast cancer, mutations, and chromosome. As in Figure 5, we can distinguish two main clusters. Within cluster 1, the terms surrounding breast/breast cancer refer to the activities that led to the identification of the two breast-cancer genes, BRCA1 and BRCA2; both are featured on the map, as well as the term Breast Cancer Linkage Consortium, the previously mentioned international research network involved in the search for and localization of these two genes. The somewhat cryptic, intermediary terms connecting breast/breast cancer to chromosome, namely q12, q21, and so on, as well as other similar terms surrounding the chromosome node, correspond to the scientific terminology for chromosome locations and thus refer to the search, for instance through linkage analysis, for cancer susceptibility genes by first establishing their approximate location on chromosomes. Similarly, the terms that fall between mutations and breast/breast cancer point to specific genes and gene mutations (p53/BRCA) involved in the etiology of cancer. Cluster 2 is mainly related to the genetic forms of colon cancer as represented by a number of small nodes referring to colorectal cancer and familial adenomatous polyposis (FAP), the colorectal family disease for which a genetic connection was initially established. Mutations acts as the link between the two clusters. The heterogeneous map (Figure 7), which once again features only the most significant ties, displays two different configurations in this period. On the left side (cluster 1), the main focus of the network is provided by breast cancer and mutations, these two themes connecting a majority of actors in that sector of the map. Two mutations in particular (BRCA1 and BRCA2) connect members (both French and foreign) of the BCLC consortium.

452

Science, Technology, & Human Values

Figure 6 Thematic Map of Period 3 (1992–1996) Beckwith

2

Wiedemann children series complications epidemiology histiocytosis brachytherapy radiotherapy irradiation treatment squamous prevention induction Ewing Oncology development surgery chimeric FU Pediatric Oncology EWS transcripts Etoposide French Society Wilms von Hippel-Lindau FLI Presymptomatic Pediatric years involvement genome genomic Multivariate polymerase fusion malignant diagnosis melanoma chemotherapy Drosophila replication Contribution Europe hamster polyomavirus reactionhomologue gene therapy melanoma regulation human immunodeficiency virus Prenatal diagnosis alpha instability molecular genetics administration microsatellite allele interleukin infection myeloid human hypertension interferon trisomy genetics virus infection lymphoproliferative CTG therapy basis insertion virus Epstein-Barr virus Myotonic dystrophy hypertension cultures hepatitis C virus comparison lymphocytes generation anaplastic Prevalence Familial adenomatous polyposis abnormality deletion recombinant molecular biology segregation analysis detection National Federation autologous restriction lymphoma membrane neuroblastoma anomalies marrow Genetic predisposition combination essential hypertension marrow transplantation adenomatous polyposis Federation transplantation gene polymorphisms distribution retinoblastoma colorectal cancer polyposis enzyme gene segregation sequence megatherapy predisposition receptor gene angiotensin adenomatous polyposis coli APC polymorphism implications promoter colorectal lesions angiotensin II enzyme system parameters exons gene expression neurofibromatosis protein renin codon mRNA receptor p53 mutations ACE Burkitt gliomas electrophoresis multiple endocrine neoplasia lung cancer frequency Correlation Evaluation expression Serum method chromosome abnormalities isoforms colorectal tumors lung plasma Ataxia telangiectasia linkage analysis cytokeratin Hereditary predisposition tamoxifen mutations genotype products breakpoints phenotype linkage Cytogenetic hepatocellular carcinoma p53 protein proto-oncogene p53 metastasiscolony q13 translocation Association MEN p53 gene cytogenetic analysis identification Genetic linkage families myc estrogen receptor prostate cancer chromosome translocation status markers ras prostate carriers Absence amplification onset ovarian cancer families prognosis interphase q24 Germline mutations application rearrangements fluorescence mutation carriers susceptibility yeast artificial chromosome Prognostic significance BRCA1 Linkage Consortium hybridization information Breast Cancerindications incidence cosmids BRCA1-mutation cancer susceptibility gene isolation breast chromosome human chromosome p53 antibodies breast cancers BRCA2 beta antibodies Alu YAC Hereditary breast cancer contig q11 Consortium relationships PCR DNA metastases assignment alterationsbreast tumors apoptosis fluorouracil region localization q21 breast cancer patients q22 Marfan syndrome differentiation human primary breast cancer Genetic alterations locus human breast heterozygosity growth q12-q21 activity human breast cancer candidate primary human breast candidate genes loss q12 RNA metastatic breast chromosome 22q12 breast carcinoma allelic exclusion breast-ovarian cancer insulin Genetic heterogeneity metastatic characterization diabetes

1

Note: Semantic weight > 50 (word combinations).

Mutations, as argued in greater detail elsewhere (Bourret 2005), are a key intermediary between clinicians, biologists, diseases, and family patients. On the right side (cluster 2), a human actor (Thomas) plays a role similar to the one played by Lenoir during Period 1, establishing the initial bridges that will allow the development of molecular genetic work on colon cancer.

Bourret et al. / Cancer Genetics

453

Figure 7 Heterogeneous Map of Period 3 (1992–1996)

interval

1

q12

edh17b2

q21 markers

loss

q11 heterozygosity

MAZOYER S. candidate germline

SOBOL H. BIGNON Y. PONDER B.

deletion

beta

locus

STOPPA-LYONNET D. identification susceptibility Consortium carriers NAROD S. LYNCH H. brca1 genotype families phenotype FREBOURG T. Absence FEUNTEUN J. neoplasia linkage oncogene LENOIR G.

nf2 chromosome

amplification

BIECHE I.

localization

myc association

contig

dna

BERNHEIM A.

LIDEREAU R.

BERGER R. yeast

prostate mutations EISINGER F. brca2 incidence

Correlation

region

alterations

isolation breakpoints JONVEAUX P.

JUNIEN C. gliomas men

BONAITI C. status

method

breast application

CALENDER A.

Epstein Barr

MELOT T.

THOMAS G.

clonal polymorphism biology

products colorectal

domain

SOUBRIER F.

coli

ace enzyme angiotensin

JANIN N.

significance colony

MOUSSEAU M. protein receptor virus

plasma indications

insulin

infection liver recombinant hepatitis generation

2

NGUYEN T. apc OLSCHWANG S. prevention epidemiology LAURENT-PUIG P. diagnosis ZUCKER J.M. development

Prevalence chemotherapy LEMERLE J. treatment BRUGIERES L. surgery induction BOUGNOUX P. antibodies marrow

AVRIL M.F. BRESSAC B. PEYRAT J.P. activity

growth

adenomatous polyposis

expression relationships implications

CORVOL P.

anemia

p53

antigen

segregation

DELATTRE O. DESMAZE C. translocation AURIAS A. DiGeorge interphase ZUCMAN J.

electrophoresis detection gradient exons HAMELIN R.

DEMENAIS F.

ebv

cosmids

sequence

proto

onset

fluorescence hybridization

pcr Alu

Evaluation PERNOT M. HOFFSTETTER S. LUPORSI E. neuroblastoma

PHILIP T. lesions BRUNAT-MENTIGNY M. SOMMELET D. survival LASSET C. transplantation megatherapy BOUFFET E. autologous FRAPPAZ D.

involvement

RICHARD S. von Hippel Lindau Pheochromocytoma

Note: Top fifty human actors (black nodes and capital letters) and top (single words, semantic weight > 50) nonhuman actors (grey nodes and lowercase); 30 percent more specific ties selected.

Finally, we notice at the interface between the left and right sides of the map a set of entities (e.g., p53, exons) and techniques (e.g., electrophoresis) common to both the breast- and colon-cancer domains.

454

Science, Technology, & Human Values

To sum up, this period has been characterized by the localization and cloning of cancer-related genes, especially BRCA1 and BRCA2, and the related establishment of clinical services in several French hospitals. BRCA1 and BRCA2 have become paradigmatic examples of the application of genetics techniques to cancer both in the sense of showing the research potential of the genetic approach and in the sense of leading to the establishment of clinical-consult services. As shown by the heterogeneous map, this period is also a transitional one insofar as it harbors a dual structuring principle, one corresponding to an already well-structured, hybrid collective (breast cancer) and one that corresponds to an emerging field of work (colon cancer). Period 4 (1997–2001) Moving now to the fourth and final period, the coauthorship map (Figure 8) displays a single, almost fully integrated network. One can of course distinguish a few regions within this seamless web, but rather than testing our readers’ patience even further by describing these regions, we will simply emphasize the main message conveyed by the map, namely the culmination and consolidation of the group’s amalgamation process. The center of the network is occupied by GGC members, the most prolific of whom include the previously mentioned initiators of the first clinical services. The map therefore shows a strong collaborative pattern between GGC members. But what is the actual focus of these collaborations? The thematic map (Figure 9) of this period is dominated by the breast/breast cancer cluster. The latter obviously is not the only substantive pathology represented within the GGC. For instance, on the map’s left side, just opposite the breast/breast cancer node, we find a smaller cluster centered on colorectal cancer. We have already commented, when discussing Figures 5 and 6, on the unequal position occupied by breast cancer and colon cancer; the situation seems to have persisted during the fourth period. As far as breast cancer is concerned, the map is neatly divided into two sections. Above the breast/breast cancer node we find BRCA1 and BRCA2 that, in turn, are connected to the mutations node at the top of the map. In this respect, if we compare the present map to the map of the previous period, we notice that while the term mutations still plays an important role, the term chromosome has almost disappeared. This is easily explained by the fact that, after the discovery of the BRCA genes, clinical and research activities have focused on the role and nature of these genes and related mutations rather than on their search through chromosome analysis. Below the breast/breast cancer node, we find another set of terms such as SOR (a French acronym for

Bourret et al. / Cancer Genetics

455

Figure 8 Coauthorship Map of Period 4 (1997–2001) saudubray j.m. brunelle f. gaudray p. zhang c.x. robert j.j. fournet j. DUGAST C. teh b. CALENDER A. nihoul-fekete c. de lonlay p. larsson c. fournier g. robert j. murat a. cussenot o. QUILLIEN V. JUNIEN C. vidaud d. GIRAUD S. latil a. legoix p. boileau c. beroud c. vidaud m. hamelin r. moller p. germain e. evans g. billaud m. hodgson s. eccles d. RICHARD S. lhuillery c. filetti s. munnich a. OLSCHWANG S. eng c. haites n. brice a. chang-claude j. bidart j.m. asselain b. lyonnet s. caillou LAURENT-PUIG P. goldgar d. martinez bieche i. stratton m. caligo m. hannequin d. baudin e. cailleux a. lynch h. spyratos f. LENOIR G. dumanchin c. borg a. narod s. zucman j. lazar v. campion d. BOUGNOUX P. SINILNIKOVA O. THOMAS G. martin c. larra f. champeme m.h. SCHLUMBERGER M. LIDEREAU R. DAVER A. charbonnier f. philippi a. pages s. weber b. puget n. flejou j.f. LONGY M. uhrhammer n. SOBOL H. BIGNON Y. SOUBRIER F. LIMACHER J.M. rio p. MAZOYER S. FREBOURG sabourin j.c. froguel p. STOPPA-LYONNET D. kwiatkowski f. FEUNTEUN J. jacquemier j. dizier m.h. DEMENAIS birnbaum bernard-gallon d. bertrand v. BONAITI C. NOGUES C MAUGARD C. noguchiJULIAN-REYNIER C. WANG JANIN N. JONVEAUX chaussade s. Q P. EISINGER F. BAY J.O. maurizis j.c. BRESSAC B. couturier d. favy d. LORTHOLARY A. LEROUX D. ducreux m. chassagne j. feingold j. PEYRAT J.P. duvillard p. GUIMBAUD R. penault-llorca f. vissac c. CHOMPRET A. PUISIEUX A. peter m. pautier p. VENNIN P. dauplat j. cure h. hizel c. aurias a. castaigne d. SPATZ A. LASSET C. DELATTRE O. margulis a. AVRIL BRUGIERES L. gamelin e. DELALOGE S. bouvier r. rougier p. MOUSSEAU M. lhomme c. oberlin o. beylot-barry m. FRAPPAZ D. benard j. hondermarck h. bremond a. hornez l. BERNHEIM A. philippe c. delaunay m. terrier-lacombe m.j. cutuli b. ABECASSIS J. hebbar m. bergeron c. bonneterre j. LUPORSI E. danglot g. michon j. revillion f. chevreau c. MIGNOTTE H. SOMMELET D. meddeb m. hartmann o. negrier s. de lafontan b. philip t. y. etienne m.c. plantaz d.perel leverger g. taieb s. combaret v. NGUYEN T. chauvin f. rubie h. brunat-mentigny ychou m. coze c. ROMIEU G. lebrun c. thiesse p. chastagner p. bey p. FRENAY M. peiffert d. ROSTAGNO P. CANNONI H. CAPODANO A.

bringer j. maudelonde t. PUJOL P.

Note: Only authors belonging to the top-200 layer are represented (GGC members in capital letters; outside authors in lowercase).

standards-options-recommendations) that refer to the regulation of the burgeoning clinical activity related to predictive breast-cancer genetics. These terms are connected to terms such as prophylactic mastectomy that point to the clinical practices (and dilemmas) with which these regulations deal; for instance, should women who harbor a deleterious BRCA mutation and who are thus at an increased risk of developing breast cancer at some future point be offered drastic preventive interventions such as the surgical removal of their breast, or should they simply be subjected to increased surveillance through mammography, a less extreme but also more risky alternative? Clearly, while the GGC had from the very outset raised the issue of practice

456

Science, Technology, & Human Values

Figure 9 Thematic Map of Period 4 (1997–2001) domain length APC hMLH1 association p16 DNA somatic mutations duplication Alu INI1 prevalence affinity amino Germline hSNF5 activation proto-oncogene germline mutations deletion chromatography p15 Rett syndrome exon French families gene mutations BRCA rearrangements phenotype ductal cancer families spectrumorigin hyperplasia transcripts BRCA2 proteins promoter ovarian cancer families dystrophy hyperinsulinism localization France missense mutation genotype population mutant Identification murine protein presenilin mutations chromosomal localization BRCA1 autosomal dominant protein expression Alzheimer carriers BRCA1 gene onset BRCA2 families adenocarcinoma Attitudes Apolipoprotein BRCA2 mutations allele strategies pathology BRCA2 genes promoter polymorphismP53 gene relationship individuals familial breast cancer hereditary breast sporadic breast carcinoma hereditary breast cancer insulin-dependent Bax ATM gene multifactorial diabetes mellitus development BRCA1 mutations sporadic breast cancer p53 mutations sequence p21 ATM Loss antibodies characterisation BRCA1 protein Demonstration CDKN2A polymorphism quantification regulation breast carcinoma agents implications Proteomic susceptibility genes prolactin prognostic significance inactivation method bladder Na sporadic breast heterozygosity overexpression death distribution pathway allelic loss human breast carcinoma detection Immunohistochemical interactions symporter p53 human breast cancer human breast survival fibroblast growth microsatellite polyposis serum Iodide symporter gene series synthase gene Familial adenomatous polyposis tamoxifen significance MCF product microsatellite instability MDM2 status erbB amplification instability sodium expression synthase Cowden MYC Mutational analysis Nitric oxide synthase bladder tumors fibroblast growth endothelial sporadic breast tumors susceptibility sensitivity ras gene amplification q22 breast CTG gene expression allelic breast cancer transgenic mice HLA PCR reaction melanoma node locus polymerase RT-PCR Malignant melanoma imbalance transcription q23 breast cancer susceptibility esophageal p11 chromosome region breast cancer patients glutathione breast tumor human chromosome genetic predisposition squamousMEN1 q21 q13 transcription-polymerase colorectal cancer accumulation comparison predisposition multiple endocrine neoplasia determination application colorectal ataxia-telangiectasia proliferation enzyme WT1 colorectal carcinoma candidate women parameters primary breast cancer patients relationdoxorubicin Wilms LOH prophylactic mastectomy molecules metastasis concentrations plasma escalation recurrence FU primary breast cancer metastatic colorectal cancer conservative treatment assessment pharmacokinetic beta radiotherapy induction liver oxaliplatin progesterone receptors ovary irinotecan estrogen receptor prophylactic surgery inhibition toxicity estrogen clinics leucovorin exclusion NF2 gene progenitor proliferative interleukin fluorouracil Consortium frequency ERBB2 characterization treatment receptor-alpha multicenter Standards SOR infusion oncogenetic interferon metastatic breast NF2 metastatic CSF receptors EGEA Genetics Consortium Federation Recommendations Somatic Genetics isoforms fluorescence anthracycline ER Cutaneous Lymphomas neurofibromatosis mRNA progression cisplatin alterations surgery cytotoxic human prostate cancer response Genetic alterations linkage Committee management alpha therapy NF1 indications lymphomas polyunsaturated fatty acids Action iodine chemotherapy complications radiation therapy variation guidelines Health retinoic acid childhood combination efficacy regimen PTEN variables tumor markers irradiation neoadjuvant activity prostate cancer von Hippel-Lindau prevention prostate central nervous system Genetic heterogeneity markers involvement carboplatin human essential hypertension radiation system receptor gene paclitaxel hybridization renin-angiotensin ulcerative colitis antigen glioblastoma genetics FNCLCC angiotensin asthma Gene therapy aldosterone gliomas INSERM correlation adrenocortical production Segregation colon hypertension Comparative genomic hybridization IgE FLI atopy VHL gene multiple myeloma environment genomic haematology EWS VHL Europe neuroendocrine tumors Atlas Ewing cytogenetic Ewing tumors sarcoma Oncologie Pediatrique lymphoma Pediatric Oncology literature infants Pediatric endometrium neuroblastoma Oncology tyrosine

kinase

French Society CD4 children Societe Francaise SFOP marrow transplantation

Note: Semantic weight > 75 (word combinations).

Bourret et al. / Cancer Genetics

457

guidelines (see GGC 1992), the field now has reached a stage where regulatory interventions can be linked to actual clinical activities (rather than future scenarios) and have thus become an important component of the group’s overall activities. The heterogeneous map (Figure 10) is strikingly different from the corresponding map of Period 1. Human and nonhuman actors are closely intertwined, and the largest nodes now refer to nonhuman entities, namely breast and mutations. It is nonetheless possible to identify a few subfields within this densely knitted mesh. For instance, subfield 1 centers on clinical and regulatory work on breast cancer and BRCA mutations, while subfield 2 represents the more biological pole of that same domain. Interestingly enough, a human actor, Stoppa-Lyonnet, is at the interface of these two domains. Subfield 3 refers to colon cancer, and as compared to Period 3, now also shows a balanced mix of human and nonhuman entities: colon cancer has reached a stage similar to the one reached by breast cancer. Finally, a subfield 4 refers to chemotherapy and thus to the continuing presence within the GGC of bridges linking it to more traditional oncological themes.

Conclusion Scientometric tools have often been applied to the analysis of the development of specialties and research fronts. These kinds of studies generally resort to cocitation analysis (or other bibliometric tools) to define the boundaries of the domain under investigation, focusing on a restricted number of core documents (e.g., White and McCain 1998). In our case, we adopted a deliberately ambiguous approach: we did not focus on a domain per se, but rather on a collaborative group that happens, however, to coincide to a large extent with an emerging specialty, cancer genetics (albeit within a national framework, in part constituted, however, through the establishment of international ties). We did so because, as explained in more detail elsewhere (Bourret 2005), we entertain strong hypotheses concerning the role of collective practices in the development of contemporary biomedicine as both a research and a clinical endeavor. Indeed, we believe that collaborative networks offer a strategic empirical starting point for the investigation of new biomedical developments that avoids the twin pitfalls of focusing only on research, strictly defined, or on clinical work, looking instead at the alignment between these activities and at the role played by regulation in this respect. A combination of fieldwork and computer-based analysis has allowed us to map and account for the development of the GGC.16 The beginnings of the

458

Science, Technology, & Human Values

Figure 10 Heterogeneous Map of Period 4 (1997–2001)

1

LONGY M. brca2

Loss

ATM

BIGNON Y.

exon

2

allelic heterozygosity

RIO P. multifactorial BIRNBAUM D.

deletion

Cowden

susceptibility

JACQUEMIER J. BAY J.O. EISINGER F. quantification brca1 SOBOL H. BERNARD-GALLON D. NOGUES C.

transcripts

JULIAN-REYNIER C.

linkage

carriers

BONAITI C.

families

onset

fluorescence

chromosome

hybridization brca

LIDEREAU R.

application

DEMENAIS F. FEUNTEUN J. autosomal

STOPPA-LYONNET D. MAUGARD C.

protein

prolactin

region

q22 locus

hyperplasia

PTEN pathology LENOIR G. presenilin CAMPION D. regulation BIECHE I. mutations BERNHEIM A. breast duplication rt individuals hyperinsulinism prostate France p15 CHOMPRET A. determination Germline reaction VIDAUD M. SOUBRIER F. BRESSAC B. detection JONVEAUX P. polymerase pcr

transcription

amplification

myc

PEYRAT J.P.

localization

LEROUX D.

expression

DELATTRE O. VENNIN P.

progesterone beta growth

kinase association

mcf

adenomatous sequence GIRAUD S. phenotype melanoma THOMAS G.

rearrangements CALENDER A.

estrogen receptors

LASSET C. evaluation Recommendations fibroblast

implications

alterations RICHARD S. instability polyposis HAMELIN R.

serum relationship

SOMMELET D. pathway MIGNOTTE H.

series

PHILIP T. markers BIDART J.M.

LAURENT-PUIG P.

BAUDIN E. cytotoxic enzyme

oxaliplatin

Identification dna

activity radiation

treatment efficacy cisplatin therapy squamous colorectal metastatic LUPORSI E. response chemotherapy neoadjuvant FRENAY M. multicenter MOUSSEAU M. pharmacokinetic fluorouracil anthracycline combination irinotecan

Standards

p21 microsatellite

DAVER A. interleukin SCHLUMBERGER M.

interferon

OLSCHWANG S.

PUISIEUX A. p53

BOUGNOUX P.

sor

BEROUD C.

AVRIL M.F.

LORTHOLARY A. correlation

FRAPPAZ D.

alpha

SPATZ A. FREBOURG T. JUNIEN C.

frequency

node

status sensitivity

3

4

Note: Top fifty human actors (black nodes and capital letters) and top (single words, semantic weight > 75) nonhuman actors (grey nodes and lowercase); 30 percent more specific ties selected.

group can be related to the presence of three elements: a group of researchers bridging virology and cytogenetics and working on human oncogenes in a model disease, Burkitt’s lymphoma; a group of traditional geneticists working on Mendelian diseases; and a national network of specialized cancer

Bourret et al. / Cancer Genetics

459

hospitals harboring clinicians working on breast cancer who would make the transition from somatic to hereditary approaches and from local to national (and international) collaborative work. We followed the informal coming together of the GGC, its official constitution and its present, full-fledged activities that show an overriding, although far from exclusive, focus on breast cancer. Most importantly, we have seen that the group cannot be equated to a mere lobbying institution or a learned society since it obviously engages in collaborative research activities leading to joint publications, nor can it be reduced to a research network since its activities bridge clinical and laboratory work; the GGC has, moreover, become a policy actor by producing guidelines and regulations that have been officially endorsed. The group, in other words, structures and channels its members’ activities by simultaneously producing the (regulatory) environment within which it acts. As a next step, the analysis presented in this article should be expanded to include the activities of international networks and consortia that, as we have seen, maintain mutually constitutive links with national collectives. Finally, heterogeneous maps have allowed us to inspect the constitutive dynamics of the GGC as a sociotechnical network. From a formal point of view, we can distinguish between the following two dimensions. The first one concerns the relation between human and nonhuman entities. We have here a continuum represented, at one pole, by the presence of a human actor connected to a multitude of themes, and at the other pole, by the mirror situation whereby a single theme links a number of actors. Real situations are, of course, located somewhere between these two ideal types. The second dimension refers to the structural properties of the network. One possibility is that a few large nodes, called hubs, account for a majority of links; they are, in other words, obligatory passage points in the same way that, say, a few major airports account for a majority of the connections to regional cities. The second possibility corresponds to a network whose links are more or less homogeneously distributed among its nodes.17 If, combining these two dimensions, we now consider the empirical case analyzed in this article, we see that the prehistory of the GGC (Period 1) is characterized by the presence of a few human hubs. For instance, Lenoir, the largest human hub during this period, connects different themes through participation in several networks. In contrast, the most recent period is characterized by a more even distribution of links between the various nodes; yet, we can still distinguish a few hubs18 that correspond to nonhuman entities (breast cancer, mutations) and that are well connected to the entire web. We have thus documented the making of a hybrid collective space corresponding to the creation of a densely interconnected collaborative network encompassing laboratory, clinical, and regulatory activities. This kind of analysis offers

460

Science, Technology, & Human Values

large-scale empirical evidence for the sociotechnical analysis of hybrid collectives as championed, for instance, by actor-network theory.

Notes 1. Our notion of a hybrid biomedical collective simultaneously engaging in laboratory, clinical, and regulatory activities is not to be confused with the notion of a hybrid forum developed by Callon and Rip (1992) to account for public controversies. 2. Cancer genetics consultations subsequently were also established within university hospital centers (Centres hospitaliers universitaires). 3. Katz and Martin (1997) argue that coauthorship is no more than a partial indicator of collaboration. In the present article, we triangulate coauthorship data (and semantic-network analysis) with fieldwork information, thus increasing the robustness of our conclusions. 4. We discarded two possible alternatives to the use of title words: Medline keywords, which because of their standard nature are ill-suited to map emerging patterns, and words derived from abstracts, because abstracts were only available for the more recent periods and because they introduced unnecessary noise. 5. An alternative strategy would have been to set coauthorship thresholds for each period. This second strategy leads to similar results, since coauthorship frequency is linked to the number of articles produced by a given author. The first strategy, however, is simpler and has the advantage of producing maps that can be more easily compared. 6. By using word combinations (e.g., breast cancer), we introduce a certain degree of redundancy (i.e., the maps will feature links between the terms breast cancer, breast, and cancer), but the results are nonetheless easier to interpret than when using single words. To work around the redundancy problem, one would need a very sophisticated text-analysis program able to deal with anaphora. 7. Had we opted for keywords, the problem would have been even more serious, since the latter are assigned by professional indexers irrespective of whether that particular term actually occurs in the title or abstract of an article. 8. This is obviously an extremely simplified procedural overview. The production of the maps involved a number of technical steps that cannot be detailed here. Readers interested in technical aspects should contact the authors ([email protected]). 9. In their discussion of affiliation networks, Newman, Watts, and Strogatz (2002, 2570) speak, in this respect, of bipartite networks that can then be projected onto two single-mode networks. Newman and his colleagues, however, emphasize the distinctiveness of social networks (as defined in terms of network properties related, in turn, to social behaviors such as community building: see, e.g., Girvan and Newman 2002). Given our interest in the technosocial dynamics of biomedical activities, we take the opposite methodological stance. 10. The maps reproduced in this article do not offer the same functionalities as the maps that can be inspected on the computer screen that are in color, and most importantly, can be inspected by zooming in and out of interesting regions. We have posted an electronic version of the maps on the following Web site: http://www.aguidel.com/sthv. Readers with access to a PDF version of this article can use Acrobat Reader to similarly zoom in on the maps. 11. In lay terms, this means that the presence of a deleterious mutation does not inevitably lead to the development of the disease.

Bourret et al. / Cancer Genetics

461

12. The (international) role played by Lenoir in relation to the C-myc oncogene specialty has been documented similarly by Stokes and Hartley (1989) as part of their coauthorship analysis of two molecular biological specialties. 13. As previously explained, thematic maps feature word combinations. For readability reasons, on the heterogeneous maps, we displayed individual words (above the same semanticweight threshold used in the thematic maps) and the top-fifty human actors. Moreover, the maps feature only the most significant (in the present map: 50 percent) ties, that is, those connections that differentiate authors in reference to themes and vice versa. 14. Initially characterized by data sharing and collaboration, the activities of the consortium were later marked by controversy as members of the network left to work with private companies and patent the BRCA genes (e.g., Gaudillière and Cassier 2000). Further information on the BCLC can be found on http://www.humgen.nl/lab-devilee/BCLC/history.htm (accessed August 2004). 15. Although the women movement and breast-cancer patient advocates have played a major role in putting breast cancer on the forefront of the biomedical scene in the United States, they have not played a similar role in France. 16. For a discussion of the relation between mapping and traditional fieldwork methods, including the qualitative inspection of documents, see Cambrosio, Keating, and Mogoutov (2004, 357). The authors argue that in addition to generating results that could in principle be obtained though more traditional methods but are in practice only elicited by mapping, maps produce not otherwise available information about figurational patterns. 17. This dichotomy that corresponds mathematically to a distinction between power-law and exponential distributions has been discussed in detail by recent contributions to network analysis; for a general introduction, see Barabási (2002) and Watts (2003). 18. Mathematically speaking, this kind of network is characterized by a power-law distribution with an exponential cutoff.

References Barabási, A.L. 2002. Linked: The new science of networks. Cambridge, MA: Perseus Press. Bourret, P. 1988. Le temps, l’espace en génétique: Intervention médicale et géographie sociale du gène. Sciences Sociales et Santé 6 (3/4): 171–98. ———. 2005. BRCA patients and clinical collectives: New configurations of action in cancer genetics practices. Social Studies of Science 35 (1): 41–68. Bourret, P., and P. Huard. 1990. Création, appropriation, recomposition: Processus innovants dans la production du diagnostic prénatal. Sciences Sociales et Santé 8 (4): 57–89. Callon, M. 1991. Techno-economic networks and irreversibility. In A sociology of monsters: Essays on power, technology and domination, ed. J. Law, 132–64. London: Routledge and Kegan Paul. ———. 1995. Four models of the dynamics of science. In Handbook of science and technology studies, ed. S. Jasanoff, G. E. Markle, J. C. Petersen, and T. Pinch, 29–63. Thousand Oaks, CA: Sage. ———. 2001. Les méthodes d’analyse des grands nombres peuvent-elles contribuer à l’enrichissement de la sociologie du travail? In Sociologies du travail: Quarante ans après, ed. A. Pouchet, 335–54. Paris: Elsevier. Callon, M., J. Law, and A. Rip, eds. 1986. Mapping the dynamics of science and technology. Houndmills, UK: Macmillan.

462

Science, Technology, & Human Values

Callon, M., and A. Rip. 1992. Humains, non-humains: Morale d’une coexistence. In La terre outrageé: Les experts sont formels!, ed. J. Theys and B. Kalaora, 140–56. Paris: Autrement. Cambrosio, A., P. Keating, and A. Mogoutov. 2004. Mapping collaborative work and innovation in biomedicine: A computer-assisted analysis of antibody reagent workshops. Social Studies of Science 34 (3): 325–64. Cambrosio, A., C. Limoges, J. P. Courtial, and F. Laville. 1993. Historical scientometrics? Mapping over 70 years of biological safety research with coword analysis. Scientometrics 27 (2): 119–43. Cassier, M. 1998. Le partage des connaissances dans les réseaux scientifiques: L’invention des règles de “bonne conduite” par les chercheurs. Revue Française de Sociologie 39 (4): 701–20. Cronin, B. 2001. Hyperauthorship: A postmodern perversion or evidence of a structural shift in scholarly communication practices? Journal of the American Society for Information Science 52 (7): 558–69. Dalpé, R., L. Bouchard, A. J. Houle, and L. Bédard. 2003. Watching the race to find the breast cancer genes. Science, Technology, & Human Values 28 (2): 187–216. Dodier, N., and J. Barbot. 2000. Le temps des tensions épistémiques: Le développement des essais thérapeutiques dans le cadre du sida (1982–1996). Revue Française de Sociologie 41 (1): 79–118. Gaudillière, J. P. 2000. Whose work shall we trust? Genetics, pediatrics and hereditary disease in postwar France. In Controlling our destinies: Historical, philosophical, ethical, and theological perspectives on the human genome project, ed. P. R. Sloan, 69–93. Notre Dame, IN: University of Notre Dame Press. Gaudillière, J. P., and M. Cassier. 2000. Recherche, médecine et marché: La génétique du cancer du sein. Sciences Sociales et Santé 18 (4): 29–51. Gaudillière, J. P., and H. J. Rheinberger, eds. 2004. From molecular genetics to genomics: The mapping cultures of twentieth-century genetics. London: Routledge. Girvan, M., and M. E. J. Newman. 2002. Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the United States of America 99 (12): 7821–6. Glänzel, W. 2002. Coauthorship patterns and trends in the sciences (1980–1998): A bibliometric study with implications for database indexing and search strategies. Library Trends 50 (3): 461–73. Gosselin, R. 1985. Probing into task interdependencies: The case of physicians in a teaching hospital. Journal of Management Studies 22 (5): 466–97. Group Génétique et Cancer. 1992. Rapport d'activité annuel du Groupe Génétique et Cancer. Unpublished archival document, August 13. Group Génétique et Cancer. 1995. Proposition de charte du Groupe Génétique et Cancer. Unpublished archival document, May. Haiqi, Z. 1997. More authors, more institutions, and more funding sources: Hot papers in biology from 1991 to 1993. Journal of the American Society for Information Science 48 (7): 662–6. Hilgartner, S. 1998. Data access policy in genome research. In Private science: Biotechnology and the rise of the molecular sciences, ed. A. Thackray, 202–18. Philadelphia, PA: University of Pennsylvania Press. Hopwood, P., C. J. van Asperen, G. Borreani, P. Bourret, M. Decruyeneere, S. Dishon, F. Eisinger, et al. 2003. Cancer Genetics service provision: A comparison of seven European centres. Community Genetics 6 (4): 192–205.

Bourret et al. / Cancer Genetics

463

Katz, J. S., and B. R. Martin. 1997. What is research collaboration? Research Policy 26 (1): 1–18. Keating, P., and A. Cambrosio. 2001. The new genetics and cancer: The contributions of clinical medicine in the era of biomedicine. Journal of the History of Medicine and Allied Sciences 56 (4): 321–52. ———. 2002. From screening to clinical research: The cure of leukemia and the early development of the cooperative oncology groups, 1955–1966. Bulletin of the History of Medicine 76 (2): 299–334. Laudel G. 2002. What do we measure by coauthorships? Research Evaluation 11 (1): 3–15. Löwy, I. 1997. Between bench and bedside: Science, healing and interleukin-2 in a cancer ward. Cambridge, MA: Harvard University Press. Marks, H. M. 1997. The progress of experiments: Science and therapeutic reform in the United States, 1900–1990. Cambridge, UK: Cambridge University Press. Melin, G., and O. Persson. 1996. Studying research collaboration using coauthorship. Scientometrics 36 (3): 363–77 Ménoret, M. 1999. Les temps du cancer. Paris: Éditions du CNRS. Morange, M. 1997. From the regulatory vision of cancer to the oncogene paradigm, 1975–1985. Journal of the History of Biology 30 (1): 1–29. Newman, M. E. J. 2001. The structure of scientific collaboration networks. Proceedings of the National Academy of Sciences of the United States of America 98 (2): 404–9. Newman, M. E. J., D. J. Watts, and S. H. Strogatz. 2002. Random graph models of social networks. Proceedings of the National Academy of Sciences of the United States of America 99 (Suppl. 1): 2566–72. Parthasarathy, S. 2004. Regulating risk. Defining genetic privacy in the United States and Britain. Science, Technology, & Human Values 29 (3): 332–52. Peters, H. P. F., and A. F. J. Van Raan. 1991. Structuring scientific activities by co-author analysis: An exercise on a university faculty level. Scientometrics 20 (1): 235–55. Pinell, P. 1992. Naissance d’un fléau: Histoire de la lutte contre le cancer en France (1890–1940). Paris: Métailié. Sevilla, C., P. Bourret, C. Nogues, J. P. Moatti, H. Sobol, C. Julion-Reynier, and Groupe Génétique et Cancer. 2004. L’offre de tests de prédisposition génétique au cancer du sein ou de l’ovaire en France. Médecine/Sciences 20 (8/9): 788–92. Sobol, H. et al. 1991. Oncologie-génétique: Mise en place d’un réseau collaboratif national, application au cancer du sein et au syndrome de Li et Fraumeni. In Conference proceedings: Euromédecine, Génétique et Cancer, 408–11. Montpellier, 7–9 November 1991. Stokes, T. D., and J. A. Hartley. 1989. Coauthorship, social structure and influence within specialties. Social Studies of Science 19 (1): 101–25. Timmermans, S., and M. Berg. 2003. The gold standard: The challenge of evidence-based medicine and standardization in health care. Philadelphia, PA: Temple University Press. Van Helvoort, T. 2001. Scalpel or rays? Radiotherapy and the struggle for the cancer patient in pre-Second World War Germany. Medical History 45 (1): 33–60. Vinck, D. 1992. Du laboratoire aux réseaux: Le travail scientifique en mutation. Luxembourg: Office des Publications Officielles des Communautés Européennes. Watts, D. J. 2003. Six degrees: The science of a connected age. New York: W. W. Norton. White, H. D., and K. W. McCain. 1998. Visualizing a discipline: An author co-citation analysis of information science, 1972–1995. Journal of the American Society for Information Science 49 (4): 327–55.

464

Science, Technology, & Human Values

Pascale Bourret is maître de conferences at the Université de la Méditerranée (Aix/Marseille II), where she teaches sociology, and a researcher within the Institut National de la Santé et de la Recherche Médicale (INSERM) Research Unit 379 (Social Science Applied to Biomedical Innovation) in Marseilles. Her current research interests focus on innovation in the field of genetics and related biomedical fields and its relationship with the transformation of medical work and medical judgment. Andrei Mogoutov is cofounder of the consulting company Aguidel (http://www.aguidel.com) and the initiator and developer of the Reseau-Lu project, a software system for the analysis of heterogeneous data, including relational data mapping, longitudinal data exploration, and text mining. Originally trained as a physicist, he subsequently specialized in methodology and data analysis for the social sciences. Since 1999, he has worked on several research projects conducted at the Centre de Sociologie des Innovations (CSI) of the École Nationale Supérieure des Mines in Paris. Claire Julian-Reynier, MD, MSc, is a public-health physician and epidemiologist and a senior researcher within the Institut National de la Santé et de la Recherche Médicale (INSERM) Research Unit 379 (Social Science Applied to Biomedical Innovation) in Marseilles. Her present research interests include the evaluation of medical innovation in adult genetics and cancer genetics, and in particular, psychosocial issues in patient-provider relationships and psychosocial issues in cancer treatment and prevention. Alberto Cambrosio is a professor in the Department of Social Studies of Medicine of McGill University, Montreal, Canada. His work focuses on the sociology of biomedical practices and innovations, and in particular, on the relation between laboratory and clinical activities.