Elsevier

Social Science & Medicine

Volume 73, Issue 6, September 2011, Pages 816-824
Social Science & Medicine

Regulating diagnosis in post-genomic medicine: Re-aligning clinical judgment?

This article is dedicated to Harry Marks (1947--2011), colleague, friend, and source of inspiration.
https://doi.org/10.1016/j.socscimed.2011.04.022Get rights and content

Abstract

In recent years, genomic technologies have entered oncology. In particular, so-called tumor signatures are now commercially available for diagnosing breast cancer. These new diagnostic tools have expanded the content and meaning of diagnosis, by adding a distinctive prognostic (will the disease recur?) and predictive (how will the disease react to treatment?) dimension to this activity, and modifying the relations between diagnosis and therapy. In particular, they raise the issue of the locus of clinical judgment and clinical decision-making insofar as they involve a re-alignment of the biological and clinical components of medical activities. Using as a case study a debate over the regulation of tests for genomic signatures by the US FDA, this paper examines how the actors problematize the issues related to the introduction of molecular diagnostics into clinical settings.

Introduction

Diagnosis, prognosis and therapy are the classical components of medical practice. Each component has its own history. Christakis (1999) has argued, for instance, that although prognosis gradually disappeared from medical textbooks at the turn of the 20th century, it has recently regained lost ground due, in part, to the development of new medical technologies. Jutel (2009) has pleaded for the development of an entire sociological subspecialty devoted to the study of diagnosis. While a focus on these individual components of medical practice is certainly worthwhile, it conceals their mutual, shifting relations. Despite inconsistencies between these domains (Mol, 2002), they must in the end be reconciled if clinical judgment and clinical decision-making are to be justifiable and justified in Boltanski and Thévenot’s (2006) sense of these terms. In addition, their interconnection precludes the possibility of their sequential performance. Diagnosis, for instance, does not necessarily precede and dictate therapy: therapeutic success may confirm a tentative diagnosis (Berg, 1992). More generally, these activities easily bleed into one another, both synchronically and diachronically. This paper will argue that the development of post-genomic platforms further conflates diagnostic, prognostic and therapeutic categories and processes, creating anxieties among health practitioners over clinical jurisdictions. One of their main concerns is that non-clinical, laboratory-based results increasingly tend to dictate, rather than simply contribute to clinical decisions, and by the same token encroach upon what was once the uncontested domain of the clinician, namely clinical decision-making. This situation is nicely captured by the following excerpt from an interview with a leading French cancer clinician who, commenting on the transformation of clinical judgment in recent years, noted: “we often call it [clinical expertise] intuition, but it’s a complex mix of things that our training and clinical experience have taught us, and now, suddenly, we are faced with a situation where [genomic tests] tell us the opposite of what we thought. We said ‘white’ and the test tells us ‘black’, or the other way round” (interview, June 2009; our translation).

Post-World War II biomedicine is characterized by the development of biomedical platforms that embody the meteoric rise of medical biology (Keating & Cambrosio, 2003). Quantifiable, biological variables have become ubiquitous within the practice of medicine, and attempts to bring pathology and biology together within a single space of representation have become commonplace. Post-genomic technologies take a further step in this direction. This does not mean that medicine has been reduced to biology (Keating & Cambrosio, 2004). Physicians continue to exercise clinical judgment, just as pathologists, full-fledged members of the clinical community, continue to act as intermediaries between the bedside and the laboratory. Genomic platforms, however, bear the threat or promise (depending on one’s point of view) of de-centering clinical decision-making. Laboratory tests are hardly new, but have heretofore remained largely external to the treating physician’s clinical judgment: the laboratory test provided a result, the clinician diagnosed. This is still mainly the case, as de-centering entails more a realignment of the biological and clinical components of medical judgment, than a displacement of the clinician. Realignment, however, can have important consequences such as a shift from an individual to a collective epistemic subject—the multi-disciplinary bio-clinical team—as the locus of clinical decision-making (Bourret, 2005), and the inclusion of clinical judgments within laboratory tests.

Promoters of genomic approaches claim“many ‘diseases’ are clinical syndromes, defined observationally, that undoubtedly are made up of a collection of distinct pathogenic states” (Woodcock, 2007, p. 166). On this view, molecular diagnostics gives clinicians and researchers access to “more underlying pathogenic processes, without necessarily providing full mechanistic, explanatory data” (Ibid.). As cancer is often cited as a model in this respect, we focus on breast cancer as an instance of a domain transformed by molecular techniques through the identification of different subtypes of the disease, each requiring a specific therapeutic approach. In particular, following a positive biopsy of a suspicious lump in her breast, a patient does not simply receive a generic diagnosis of breast cancer. Rather, in addition to traditional pathological parameters such as staging (how much the cancer has grown and spread) and grading (the state of the cancer cells), present diagnoses include molecular information concerning the biology of her tumor, such as the presence or absence of hormone receptors. On the clinical research front, molecular oncology provides breast cancer clinicians with gene expression profiles of the tumor, i.e. the expression pattern of a set of genes (a.k.a. the tumor’s signature) that allow the oncologist to assign the tumor to a molecular subspecies (Perou et al., 2000) and, most importantly, to determine if the tumor is likely to recur following surgery and to react to chemotherapy. The emerging genomic platform thus works along three different lines: diagnostic (what kind of disease?), prognostic (will the disease recur?), and predictive (will the disease react to treatment?). This distinction between prognosis and prediction is quite recent and can be ascribed to the emergence of molecular technologies. On the new platform diagnosis incorporates prognosis and prediction. While diagnosis, prognosis and therapy were hardly separate in the past (e.g., hormone-positive and hormone-negative breast cancers carry a different prognosis and are treated differently), the conflation of these three activities has reached new heights. It has become difficult to define, in practice, where and with whom diagnosis ends, and prognosis and therapy begin.

The present article is part of a broader, ongoing research project on emerging laboratory-clinical-commercial interfaces in cancer genomics. It will focus on a specific episode, namely the debates concerning the creation of a separate regulatory category—so-called IVDMIAs or In Vitro Diagnostic Multivariate Index Assays—for gene-expression profiling tests by the U.S. Food and Drug Administration (FDA). As we will see, one of the stakes in the controversy over IVDMIAs was which tests should count as members of this category. A number of tests with functions ranging from the early detection of cancer to the prediction of the genetic risk of common diseases were potential candidates for membership and thus subjects of debate. Our discussion centers on two breast cancer tests that are prognostic of the future course of the disease and predictive of treatment response. Not only did these tests figure prominently in the debate, but the issues surrounding them shed light on ongoing interrogations about the shifting locus of clinical judgment. Regulatory debates and practices offer a unique vantage point for the analysis of biomedical innovation, one that is not confined to upstream (the conception of new tools) or downstream (their deployment in clinical routines) events but, rather, begins “right in the middle of the game” where one is in a position to capture the mutually constitutive relations between these two stages, and, in particular, the realignments that articulate novel platforms and the entities they generate with prior diagnostic and prognostic practices (Keating & Cambrosio, 2003, p. 334; Cambrosio et al., 2006).

Regulation per se is not the topic of this paper, nor is the regulatory environment within which debates took place. Moreover, given the complexities of these issues, it is no surprise that discussions surrounding the FDA initiative ranged from the economics of innovation, to ideological statements about the freedom to practice medicine, and to more concrete discussions of specific aspects of the proposed regulation. Despite these ramifications, our analysis focuses on clinical practices and medical judgment. For while policy and economic considerations play a role in clinical activities, the latter cannot be reduced to the former. Finally, we have chosen, as is now commonplace in science and technology studies, to follow the actors and their statements, thereby implementing a “sociology of critical capacity” rather than a critical sociology (Latour, 2005, Boltanski and Thévenot, 1999). In the present case, the FDA initiative and the arguments deployed by its supporters clearly reflect the belief that genomic tests are likely to have profound consequences on the relations between biology and medicine, displacing the locus of clinical judgment, readjusting the biomedical division of labor, modifying doctor–patient relations and creating new relations between public, not-for-profit, and commercial institutions. Opponents propose a different socio-technical scenario, one that privileges the resilience of existing socio-technical arrangements. This clash of visions provides us with an opportunity to consider the shifting role and nature of diagnosis within the onslaught of molecular technologies.

Section snippets

Methodology

As noted above, the present article is part of a broader research project on cancer genomics that began in 2008 and was approved by McGill University’s IRB (#A05-E11-04B and A02-E12-09B). It deploys a three-pronged ethnographic approach that consists of: (a) the systematic collection and analysis of published and unpublished documents (scientific articles, company reports, web newsletters and blogs); (b) interviews with cancer clinicians, researchers, statisticians and the staff of biotech

A short chronology of events

All key indicators (number of papers published, grants and patents awarded, and the like) show that cancer is the primary domain for the application of genomic technologies. Their deployment has resulted in the definition of new subtypes of this family of diseases (Perou et al., 2000, Alizadeh et al., 2000), and led to the clinical implementation of molecular tests for cancer prognosis and prediction. Although many of these technologies remain experimental, a number of commercial tests have

Biological tests or clinical procedures? displacing clinical judgment and clinical decision-making

The questions that closed the previous section bring us to the core of our subject matter. If, as claimed by the Second Draft Guidance, clinicians are unable to “verify the clinical significance of the IVDMIA result”, and if the “ordering physician [cannot] reach the IVDMIA result on his or her own, nor [can] he or she independently interpret that result”, the genomic test embodies (non-human) clinical-diagnostic agency. As a result, the evidentiary basis for clinical judgment no longer lies in

Prognosis and prediction: regulating drugs, regulating tests

As we saw, the new genomic tests are not “mere” diagnostic tests; they provide prognostic and predictive information. The FDA distinguished between prognosis and prediction to assign tests to different risk-based regulatory categories: “a device intended as an indicator of a patient’s risk of cancer recurrence may be a class II device […] while the same device intended to predict which patients should receive chemotherapy might require premarket approval [a class III device]” (Second Draft

Conclusion

Our analysis raised the question of the extent to which genomic tests participate in the realignment of clinical expertise and biological information, in particular with regards to the expansion of the domain of diagnosis to include new kinds of prognostic and predictive information. We have done so by examining how the actors themselves discuss and debate this issue and its ramifications as part of the process of problematization inaugurated by the FDA. As argued elsewhere (Bourret, 2005,

Acknowledgments

Research for this paper was made possible by grants from the Canadian Institutes of Health Research (MOP-93553), the Fonds québécois de la recherche sur la société et la culture (SE-124896), the Social Sciences and Humanities Research Council of Canada (410-2008-1833), the French Institut National du Cancer (0610/3D1418/SHS08), and the Conseil Régional PACA. We would like to thank all the scientists and clinicians who kindly accepted to be interviewed and to comment on early drafts of the text.

References (34)

  • M. Callon et al.

    The growing engagement of emergent concerned groups in political and economic life: lessons from the French association of neuromuscular disease patients

    Science, Technology & Human Values

    (2008)
  • College of American Pathologists Technology Assessment Committee, in vitro diagnostic multivariate index assays (IVDMIA)

    (2009)
  • N. Christakis

    Death foretold: Prophecy and prognosis in medical care

    (1999)
  • B.J. Evans

    What will it take to reap the clinical benefits of pharmacogenomics?

    Food and Drug Law Journal

    (2006)
  • T. Fong

    Agendia Planning to File IPO in 2011 Genome web

    (2011)
  • K. Garber

    Ready or not: personal tumor profiling tests take off

    The Journal of the National Cancer Institute

    (2011)
  • Gibbs, J. N. (2007). ASRs and IVDMIAs: FDA’s new draft guidances. Two new draft guidances could affect how laboratory...
  • Cited by (47)

    • Framing utility: Regulatory reform and genetic tests in the USA, 1989–2000

      2022, Social Science and Medicine
      Citation Excerpt :

      In 2006, FDA issued draft guidance on regulating a specific class of laboratory-developed genetic tests: so-called in vitro diagnostic multivariate index assays (IVDMIAs). However, Pascale Bourret and colleagues have argued that FDA's embrace of clinical utility as a criterion for assessing IVDMIAs was principally due to researchers' need to standardise outcome measures for clinical trials of novel “personalised” therapies (Bourret et al., 2011; Cambrosio et al., 2017), rather than a concern to protect patients from the risks of genetic testing. This is consistent with FDA's reinterpretation of its regulatory remit as being to foster innovation—for which IVDMIAs hold particular promise—as much as to control the market for medical products (Hogarth, 2015).

    • Multi-polar scripts: Techno-regulatory environments and the rise of precision oncology diagnostic tests

      2022, Social Science and Medicine
      Citation Excerpt :

      Initial forays go back to a 2006 draft guidance centered on a very specific kind of LDTs, convolutedly named IVDMIA (“In Vitro Diagnostic Multivariate Index Assays”). Hotly disputed, this new regulatory category referred to assays—epitomized by two breast cancer molecular signatures, Oncotype DX and MammaPrint—that use gene expression in connection with algorithms to report results that defy interpretation by health care practitioners, thus potentially impinging on clinical judgment (Bourret et al., 2011). Critics pointed to ambiguities in defining the exact category of tests being targeted, the regulatory targets (the algorithm, the technology, or the whole process?)

    • Brains exposed: How new imaging technology reconfigures expertise coordination in neurosurgery

      2022, Social Science and Medicine
      Citation Excerpt :

      Research often shows that when experts face the need to coordinate their hyper-specialized knowledge, deep epistemic boundaries surface (Carlile, 2004), contestation may emerge over whose knowledge takes primacy (Oborn and Dawson, 2010a), and rigid professional roles may stand in the way of exploring common ground (Liberati et al., 2016; Valentine and Edmondson, 2015). Novel medical technologies can add to the complexity of such coordination, especially when they have the potential to produce new kinds of information that may affect professionals’ expert work (Bourret et al., 2011; Bourret and Cambrosio, 2019; Timmermans and Buchbinder, 2010) that, by extension, unsettle existing ways of coordinating knowledge across different professional groups. In this paper we offer an analysis of the consequences of the introduction of novel medical technology through the lens of expertise coordination, which is valuable because of its particular focus on the use and exchange of expertise in practice.

    • Diagnostic layering: Patient accounts of breast cancer classification in the molecular era

      2021, Social Science and Medicine
      Citation Excerpt :

      Genomic techniques have further reconfigured the relationship between diagnosis and treatment, and the wider healthcare systems within which these practices take place (e.g. Bourret et al., 2011; Timmermans et al., 2017). Troubling the sociological delineation of diagnosis as a distinct process, Bourret et al. (2011) have shown how novel molecular techniques can conflate diagnostic, prognostic and therapeutic work. Diagnosis and treatment become intimately connected within molecular approaches, as therapeutic decision-making becomes directly related to knowledge of cancer pathogenesis and progression (Keating and Cambrosio, 2013).

    • Accomplishing an adaptive clinical trial for cancer: Valuation practices and care work across the laboratory and the clinic

      2020, Social Science and Medicine
      Citation Excerpt :

      In focussing on valuation and care, we render visible the work of clinical trials staff and laboratory practitioners whose roles have historically been devalued in healthcare practice and biomedical research. We trace the embodied and affective practices through which care was performed in the laboratory (see Kerr and Garforth, 2015; Myers, 2008; Star, 2007; Star and Strauss, 1999) and the clinic (see Allen and Hughes, 2017; 2014), including the backstage practices of practitioners acting as ‘intermediaries between the bedside and the laboratory’ (Bourret et al., 2011: 817). Our focus is on how these labours became ‘generative doings’ in the accomplishment of experimental biomedicine and in negotiating the complexities provoked by these large-scale genomic based trials (Puig de la Bellacasa, 2017: 54).

    • Imperfect biomarkers for adjuvant chemotherapy in early stage breast cancer with good prognosis

      2020, Social Science and Medicine
      Citation Excerpt :

      Yet, the correlated underestimation of uncertainties in favor of widely shared speculative promises (Hjorleifsson et al., 2008) has specific consequences in the clinic. The perspective, for clinical judgment, to be improperly removed from the professional jurisdiction and regulation of clinicians and included in laboratory tests is an important one (Bourret et al., 2011; Rosenberg, 2002). The genomic turn and the corporatization of life sciences (Clarke et al., 2003; Sunder Rajan 2006) that characterizes it, has further accentuated the sensitivity of the issue.

    View all citing articles on Scopus
    View full text