Expansion and uncertainty: cystic fibrosis, classification and genetics

Authors

  • Adam M. Hedgecoe

    Corresponding author
    1. Sociology Subject Group, University of Sussex
      Address for correspondence: Adam Hedgecoe, Sociology Subject Group, School of Social Sciences, University of Sussex, Falmer, Brighton BN1 9SN. e-mail: A.M.Hedgecoe@sussex.ac.uk
    Search for more papers by this author

Address for correspondence: Adam Hedgecoe, Sociology Subject Group, School of Social Sciences, University of Sussex, Falmer, Brighton BN1 9SN. e-mail: A.M.Hedgecoe@sussex.ac.uk

Abstract

This paper presents an empirical examination of geneticization, the process where genetic explanations gain ascendancy in medical and social discourse. By focusing on Cystic Fibrosis (CF), this study shows how genetic explanations play a role in the reclassification of Cystic Fibrosis. One result of this geneticization is a nosological expansion, where the boundaries of the disease expand to include a certain form of male infertility. In addition this paper highlights the uncertainty in the CF classification system that results from the use of genetic explanations.

‘Compared with polygenic diseases, cystic fibrosis is easy and therefore paints the rosiest picture of what can be achieved by genetic research’ (Geddes and Alton 1999: 1052).

1. Introduction

Although a number of sociologists have engaged with the ‘new’ genetics (see Conrad and Gabe 1999 for a review), they have tended either to critique the broader social implications of genetic technologies (e.g. Katz Rothman 1995) or to focus on patients’ responses to individual aspects of medical genetics such as genetic counselling (Bosk 1992) or perceptions of risk (Hallowell 1999). While these approaches are vital additions to the normal bioethics discussion, dominated as it is by philosophical and legal theory (Hoffmaster 1992, Nelson 2000) they tend to take the facts of genetic science as a given.

This paper takes a different approach, addressing the challenge raised by authors such as Bartley (1990) or Casper and Berg (1995) who suggest that researchers should turn towards medical knowledge as a site of sociological interest. Some sociologists have examined medical knowledge in a number of different areas (see Elston 1997 or Atkinson 1995: Chapter 3 for reviews) and there is work on the social construction of genetic disease (Yoxen 1982, Martin 1999, Kerr 2000). This paper contributes to this literature by showing how the disease category of cystic fibrosis has been re-constructed along genetic lines to incorporate previously neighbouring, but separate, conditions. This in turn conflicts with clinicians’ actual diagnostic practice and highlights the confusion and uncertainty that the introduction of genetic explanations can produce.

This analysis of CF is part of a larger research project (Hedgecoe 2000) examining the process called geneticization, defined as when genetic explanations gain ascendancy and people are reduced to their DNA codes (Lippman 1992). My approach is not to assume that geneticization is necessarily an unwelcome (or welcome) process (Hedgecoe 1999, Novas and Rose 2000), but rather to explore the way in which this process takes place and the effect it has on medical practice1.

2. Background

Much sociological interest in the new genetics focuses on the way different groups discuss genetics. For example, research on the public debates around genetics has looked at: public representation of genetic science (Turney 1998, Petersen 2001); attitudes towards genetic screening (Jallinoja et al. 1998, Singer, Corning and Antonucci 1999); and public knowledge about genetics (Kerr, Cunningham-Burley and Amos 1998a and b). But my interest is in scientists’ and medical professionals’ discourse, and thus my work relates to that of those researchers who have looked at the way in which scientists and doctors talk about new genetic technologies. Although there is survey work in this area (Wertz 1998, Fletcher and Wertz 2000), the dominant methodology has been the interview and focus group. The main theme running through this research is the way in which professionals try to maintain control over their work and restrict ‘interference’ from non-professionals such as politicians and members of the public. This may involve distinguishing their work from eugenics, as well as emphasising the public's lack of genetic knowledge, and hence the need for guidance from professionals (Kerr, Cunningham-Burley and Amos 1997, 1998c). Additional tactics involve presenting genetic knowledge as somehow fundamental and advocating the application of bioethics (Ettorre 1999). Other work has examined the variation in attitude towards the idea of a ‘gene for’ a particular condition, and has shown a close concordance in opinion between geneticists and patients (Turney and Turner 2000).

This paper turns attention away from how professionals distinguish and insulate their work from social concerns, and towards internal debates about the role of genetic explanation in the discourse surrounding particular conditions. Rather than use interviews and focus groups as a source of raw data, this approach focuses on the way in which classification systems are represented in medical/scientific texts. This approach comes from Bowker and Star's Sorting Things Out (1999), which explores the way in which classifications are shaped by contingent and social concerns (the number of lines on a census form for example) and the effects such infrastructure has on our lives. In addition to Bowker and Star's text-based approach to classification, I also wish to borrow some of the ideas that underpin their work. For example, focus on the social shaping of classification systems should not obscure their material aspects: ‘it is easy to see classifications as properties of mind. … But they have material force in the world. … [They are] mixture[s] of physical entities … and conventional arrangements’ (Bowker and Star 1999: 39).

In the case of genetic classifications, this has echoes of Edward Yoxen's 20-year-old defence of a constructionist view of classification which:

does not demand, as some might suppose, that we regard medical conditions as mere artefacts of their social context without organic cause, or as phantoms that have no enduring existence outside certain special social conditions. … One can claim … that many of the phenomena of genetic disease are grounded in a material reality, whilst at the same time asking why we isolate or delineate certain phenomena for analysis (Yoxen 1982: 144).

Bowker and Star also note, following Bruno Latour (1984), that there are two ways of writing the history of rule systems (such as classifications, or legal systems): one can either concentrate on the making of a rule, or on the way in which that rule is put into practice. Of course ‘both words and deeds are valid kinds of accounts’ but this paper focuses on the representation of classification systems in the medical literature, rather than their implementation in the clinic (through diagnosis and medical talk) (Bowker and Star 1999: 48). I accept that there is a link between the classification of CF and the way in which CF clinicians behave: ‘even when people take classifications to be purely mental, or purely formal, they also mould their behavior to fit those conceptions’ (Bowker and Star 1999: 53). But the nature of my methodology precludes the drawing of conclusions about the nature of this behaviour.

The gap between the classification system and clinicians’ behaviour has been noted by commentators on other nosological debates. In the case of diabetes in the 1980s it was a tension between the view that said that classification should be determined by and for clinicians2, and the position which accepted that medical researchers contribute, to some extent, to the drawing up of classification systems3.

The most obvious objection to my approach is that my sample (two articles) can in no way be considered as representative of the scientific and clinical discourse surrounding CF. The point is simply to underline the socially-constructed nature of classification systems through displaying their more rhetorical qualities. Additional support for my position comes from the fact that my research has much in common with recent work carried out by Anne Kerr (2000), and supports many of her conclusions. Kerr provides a widespread analysis of 80 papers from different genres (letters, research reports, conference papers) showing how the classification of CF has changed over time to incorporate Congenital Bilateral Absence of the Vas Deferens (CBAVD), a particular form of male infertility. Although our conclusions are similar, the way in which we reach them is different. Kerr's is an extensive review, while mine focuses on two papers. That we both reach largely the same position regarding the expansion of CF classification supports both of our individual approaches as well as the social constructionism underpinning our work.

An outline of the clinical and genetic basis of CF (Section 3) is followed by a summarised analysis of the construction of the CF continuum, which shows how CF has expanded to incorporate a form of male infertility. Section 5 presents debates surrounding this expansion and the subsequent uncertainty in CF classification, with Section 6 concluding.

3. Cystic fibrosis as a clinical and molecular disease

The clinical diagnosis of cystic fibrosis is based on a number of symptoms mainly affecting the pulmonary system, but also includes gastrointestinal problems, defects in the pancreas and infertility in male CF patients. Treatment for CF includes antibiotics and physiotherapy (to treat the lung infections and clear the airway), as well as large calorie intake (due to both increased effort to breathe and reduced absorption of calories/proteins). In the most extreme cases lung and/or liver transplants are required. In addition, most male CF patients are infertile.

The main clinical test used to confirm most suspected cases of CF is a sweat test. This utilises the fact that sodium and chloride levels are abnormally high in the sweat of patients with CF. The identification of the CF gene in 1989 revealed that the underlying problem is the patient's inability to transfer chloride ions across cell membranes in secretory epithelial cells, such as the lining of the respiratory airways and the small intestine. The protein involved, CFTR (Cystic Fibrosis Transmembrane Conductance Regulator), is needed for effective chloride transfer.

The various mutant types of CFTR present in CF fail to operate correctly in a variety of different ways (for example not responding to activation signals, or degrading inside the cell before reaching the site of impact) (Davis et al. 1996: 1232). With the transport of sodium chloride and water across cell membranes disturbed, the quantity and composition of the fluids secreted by the epithelial cells are altered, hence the thickened mucus in the lungs of CF patients. To date over 1000 different mutations of the CFTR gene have been identified, although many of these are limited to small family groups or individuals4.

With the discovery of the CFTR gene in 1989, the possibility arose that ‘genetic screening may eventually supplant … [the sweat test] … for definitive diagnosis’ (Birnkrant and Stern 1991: 97). This would make sense from a geneticized view of disease, with the presence or absence of a particular gene (or gene version – allele) determining whether a person has a disease or not. Yet even in this period of excitement over the genetics of CF, some commentators suggested caution when it came to using genetic tests to draw up classificatory boundaries (Scambler 1989).

4. CBAVD and the CF continuum

This research adopted a methodology, from the sociology of science, originally developed by Gilbert and Mulkay (1984) and expanded by Greg Myers (1990a). This uses the close reading of a small number of texts to show how scientists and medical professionals construct their arguments and persuade readers of their interpretation of the facts, and has been effective in exploring professional discourse around genetics (Cunningham-Burley and Kerr 1999, Kerr, Cunningham-Burley and Amos 1997, Atkinson, Batchelor and Parsons 1997). My research applied discourse analysis to the rich site of review articles, which have already proved a valuable genre for analysis, as shown by the work of Greg Myers (1991) (see also Sinding 1996). Review articles provide a textual space within which knowledge is constructed, allowing certain experimental reports to be seen as ‘key’ papers. Even the idea of a specific event as the ‘discovery’ of a particular fact depends upon review articles to organise the claims and techniques in a particular direction (Myers 1990b: 104). It focuses on the discursive mechanics of knowledge production – how a particular position is made convincing5.

Drawing on review articles from 1996 (Davis, Drumm and Konstan 1996) and 1997 (Stern 1997)6, the following section summarises work showing both how genetic explanations were incorporated into discussions of CF classification and how, as a result, the CF nosology expanded to include a form of infertility, called CBAVD. This is the cause of infertility experienced by most men diagnosed with CF as well as the cause of up to 1.5 per cent of all male infertility and is characterised by the body's reabsorption of the male genital tract (Patrizio and Zielenski 1996). Over time, this CF continuum has become more and more entrenched within the literature, indicating the expansion of the disease classification7.

CFTR as protagonist

What the 1996 review shows is the way in which the protein CFTR, which was identified as a result of the discovery of the CF gene, and which fails to work properly in the case of CF, has become a ‘protagonist’ in debates around CF classification. It provides a focus for the review article through sections such as ‘Structure of CFTR Gene’ and ‘CFTR Is a Chloride Channel’. The clinical discussion is phrased in terms of the function (or lack of it) of CFTR and this role is heightened by what we might call ‘genetic elision’, seemingly using ‘CF Gene’ and ‘CFTR Gene’ as interchangeable terms. For example, the section titled ‘Mutations in the CF Gene and the Genetics of CF’ states that ‘Since the discovery of the CFTR gene and the ΔF508 mutation, more than 400 other mutations in this gene have been found’ (Davis et al. 1996: 1231) implying that the ‘CF gene’ of the section title and the CFTR gene discussed in the text are interchangeable8.

The linking of the gene for CFTR with the gene that causes CF means that it is far more likely that any condition associated with mutations of the CFTR gene will be automatically linked to CF, and becomes a candidate for CF classification. This running together of CFTR-gene with CF-gene, this ‘genetic elision’, is to some extent the result of the naming of CFTR – Cystic Fibrosis Transmembrane Conductance Regulator. Any defect in a protein named after a disease will tend to be associated with that disease, rather than be seen as a separate condition9.

Syndrome

Prior to the discovery of the gene encoding for CFTR, researchers expressed the hope that once the genetic defect responsible for CF was discovered, CF would cease to be a ‘syndrome’ (a series of connected symptoms with an uncertain cause) and would become a ‘disease’, with a clear aetiology (e.g. Wood et al. 1976). Initially, the 1996 article suggests that CF is still a syndrome10 but once onto the section introducing the molecular genetics of CF (‘Cloning the Gene’) there is no mention of the ‘clinical syndrome of CF’, but only ‘CF’. The effect is to confirm expectations; once the gene is discovered (or in this case, presented in the article), CF ceases to be a syndrome and becomes a disease with an identifiable cause. The terminology of the 1996 article mirrors the chronology of the classification of cystic fibrosis11; the CF-syndrome becomes a disease through the introduction of CFTR and the gene that codes for it. This is reminiscent of Woolgar's discussion of the role of sequence in the textual construction of a scientific fact:

practical expression of, or reference to, a phenomenon both recreates and establishes anew the existence of the phenomenon. In describing a phenomenon, participants simultaneously render its out-there-ness (Woolgar 1981: 246).

The structure of this review article acts to make real the transition of CF from syndrome to disease, strengthening the role of CFTR mutations as causal factors, since diseases are associated with specified causes.

Representing the continuum

Figure 1 of the 1996 paper (reproduced below) represents in graphic form the classification system proposed in this review article. The left hand side of the figure lists the tissues affected in CF, the right hand side the different degrees of CFTR activity in percentages, with the genotype believed to cause it in brackets. For example, 100 per cent protein activity is defined as that produced by ‘wild type’ (wt) protein12 and 50 per cent activity by the normal protein of a CF carrier. Terms such as ‘R117H’, ‘G551D’, and ‘ΔF508’ refer to specific CF causing alleles. The central column of the figure lists the conditions that are associated with these tissue variations and CFTR activity: ‘CBAVD’ (Congenital Bilateral Absence of the Vas Deferens), ‘PS CF’ (Pancreatic Sufficiency CF), ‘PI CF’ (Pancreatic Insufficiency CF).

Figure 1.

based on Davis et al. 1996: 1233.

CF's nosological expansion is most apparent in the case of CBAVD, which in this image (titled ‘CFTR activity and tissue manifestations of CF’) is presented as a form of cystic fibrosis. This diagram, Figure 1 (Figure 3 in Davis et al.), is not referenced and is presented as an original contribution by the authors of the 1996 article although a similar diagram appears in Chillón et al. (1995), a research paper which investigated the role of the 5, 7 and 9T variation of Exon 9 and its relationship to the CBAVD phenotype. In addition it seems as if this diagram was used at CF conferences around this time13. This diagram is reproduced as ‘Table 2’ in Stern's 1997New England Journal of Medicine review article, where what Davis et al. referred to in their 1996 article as ‘speculation’ has for Stern become a system ‘helpful in evaluating a patient whose symptoms suggest cystic fibrosis … but whose sweat test is negative’ (Stern 1997: 489). The figure is not unique to these authors, suggesting that this strategy at least is used by the wider CF research community, rather than being limited to this article14.

One might object that rather than reclassifying ‘CBAVD as CF’, this review is simply linking a certain number of cases of CBAVD with CFTR mutations. If this is the case we might expect clear mention of the fact that CBAVD occurs in men with no CFTR mutations, that CFTR, related to CBAVD, is a subset of a larger type of infertility. This does not happen in this review, where it is not made clear and the implication is that all cases of CBAVD are associated with CFTR mutations15.

‘Classic’ CF

The final discursive strategy presented in this brief summary is the term ‘classic’ cystic fibrosis which does not appear in Davis et al. (1996), but in Robert Stern's 1997 review article in the New England Journal of Medicine. This is a short (five-page) review, focusing specifically on the diagnosis of CF, where Stern refers to ‘Classic cystic fibrosis16, defining it on a clinical basis:

The traditional diagnostic criteria for classic cystic fibrosis [ref.] are still valid: persistently elevated concentration of electrolytes in sweat plus characteristic clinical findings (typical gastrointestinal or pulmonary disease and perhaps obstructive azoospermia) (Stern 1997: 487).

It is contrasted with ‘atypical disease’ (1997: 487) or ‘milder variants’ (1997: 490) and this review explicitly classes CBAVD as ‘the presenting symptom of milder variants’ (1997: 490). It seems as if the introduction of ‘classic’ and ‘atypical’ as descriptive terms is another discursive strategy which solidifies the CF classificatory continuum. While CBAVD cannot be classed as ‘normal’ CF, the terms ‘milder variant’ and ‘atypical’ allow it to become a form of CF. Thus, this article supports the nosological expansion present in Davis et al., promoting the position where ‘All patients with the clinical syndrome of CF have mutations in both copies of a gene on chromosome 7’ (Davis et al. 1996: 1230). Incidentally, Stern uses the term classic in an unqualified way (i.e. with no quotation marks around it) while Colin et al. (1996), who are sceptical about the value of classing CBAVD as a mild form of CF, refer to ‘classic’ CF with the quotation marks qualifying the use of a term which is up for debate17.

I am not claiming that these reviews propose a wholly original thesis, but they do pull together various ideas and present them in a discursively powerful format. At least three of these strategies appear to be in use elsewhere in the literature (the naming strategy, the diagram and classic CF). The nosological expansion presented in my analysis is not the idiosyncratic position of a small number of authors. If it were, it is unlikely that there would be reviews called ‘Congenital absence of the vas deferens: a mild form of cystic fibrosis’ (Patrizio and Zielenski 1996) or that papers would claim that ‘It was postulated that CAVD was a mild form of CF over 20 years ago and during the last two or three years it has become clear that this hypothesis is true for the majority of cases’ (Osborne 1996: 218).

5. Expansion and uncertainty

Thus, although CBAVD was recognised as a CF symptom in the early seventies, it was clear that other conditions, not associated with CF, could also cause CBAVD. While some authors represented CBAVD as a ‘non-CF disease’ associated with the CFTR gene (Zielenski and Tsui 1995: 796) the 1996 review presents CBAVD as part of a continuum of CF, ranging in severity and phenotypic expression, but linked through mutations in the CFTR gene. In this view CBAVD ‘represents a mild form of CF’ (Patrizio and Zielenski 1996: 30).

Yet when considering the expansion of the CF classification system, it is important not to give the impression that changes took place without comment. Classification systems are compromises, and compromise entails debate, disagreement and dispute. Even within the Davis et al. 1996 article, there are inconsistencies and ambiguities18. It was not necessarily beyond doubt that CBAVD made up part of the CF spectrum. Stern comments upon the controversy that surrounds the classification of CBAVD as a mild form of CF, as do Rosenstein and Zeitlin in their 1998 review in The Lancet.

Stern accepts that ‘Different diseases can indeed result from different mutations of the same gene’, but justifies the inclusion of CBAVD in CF since:

clinical management, including genetic counselling, requires that physicians regard these patients as being at the extremely mild end of a disease with a very broad spectrum of severity. A similar debate, many years ago, about patients with pulmonary disease typical of cystic fibrosis but normal pancreatic function, was settled by the advent of sweat testing (Stern 1997: 490).

An alternative view is given by Colin et al. who dispute the new classification system on the same grounds stating:

The appropriateness of classifying CBAVD alone within the spectrum of CF remains unclear. … In this widening spectrum of CF, a dichotomy is emerging between the perspective of the scientist and that of the clinician and patient. Although the scientists may consider that CBAVD is theoretically part of the spectrum of CF, this definition is not necessarily clinically helpful … we suggest caution in describing CBAVD alone as equivalent to CF in the clinical setting (Colin et al. 1996: 444)19.

More recently, the debate over the classification of CBAVD as a form of CF has died down, with the Online Mendelian Inheritance in Man (OMIM) website, the standard repository of knowledge about human medical genetics, describing infertile men carrying two copies of CF mutations as suffering from a ‘primarily genital form of cystic fibrosis’, although they may have no other CF symptoms (OMIM)20. In Cystic Fibrosis: Medical Care, Orenstein, Rosenstein and Stern suggest that a diagnosis of Cystic Fibrosis can be given to ‘individuals with congenital bilateral absence of the vas deferens’ with ‘no evidence of respiratory tract or pancreatic abnormalities and … normal, intermediate or elevated sweat chloride concentrations’ upon ‘identification of two CF mutations’ or other evidence of CFTR abnormality (2000: 33). Another recent review declares that ‘a large body of evidence proved that CBAVD and other forms of obstructive azoospermia indeed represent atypical forms of CF’ (Stuhrmann and Dörk 2000: 71)21. While it may be the case that in day-to-day treatment of CF patients, CBAVD is treated as a separate condition, in the literature it has, through geneticization, become part of a CF-continuum.

So the CF classification system has settled down to include certain cases of male infertility where mutations in the CFTR gene have been detected. The continuum does not class all instances of CBAVD as cases of CF. There has to be evidence of changes in the CFTR protein, either through sweat tests, or a genetic test (although this will probably include most cases of CBAVD; Lissens and Liebaers 1997: 807). But it is important to note that researchers in this area have a tradition of proposing ‘unknown’ CFTR mutations when they come across CBAVD men who carry a recognised mutation on one of their chromosomes. And such men were unlikely to be classified as CF sufferers prior to the introduction of genetic testing. There are clear clinical benefits to such a classification since it allows follow-up of these men to check for previously unnoticed CF symptoms. It is also relevant when infertile men go for IVF treatment: it is important for them and their partners to be aware of the risks of having a child with some form of CF (Lissens and Liebaers 1997). But this usefulness should not obscure the socially constructed, contingent nature of this (and any other) classification system. CABVD does not have to be classed as a mild form of Cystic Fibrosis: ‘Typically, the genotype in those patients [CBAVD] consists of at least one very mild mutation uncharacteristic for CF patients’ (Zielenski 2000: 126)22. It would be quite possible to view CBAVD as a separate condition from CF, with its own CFTR genotypes and clinical diagnosis, while at the same time accepting that such patients require observation for CF symptoms23.

Pancreatitis

Perhaps one way of exploring the concept of nosological expansion is to look beyond the single example of CBAVD, at debates concerning other conditions ‘bordering’ CF, pancreatitis for example. This section briefly focuses on two articles investigating the relationship between CFTR mutations and pancreatitis to show how the expansion of the CF classification system to include CBAVD is part of a broader trend.

One of these articles, Cohn et al. (1998), contributes to the expansionist position. In their study of 27 patients with chronic or idiopathic pancreatitis, these authors found that: ‘CFTR mutations were found at 11 times the expected frequency’ (Cohn et al. 1998: 654) echoing the way in which the early studies on the role of CFTR in CBAVD made the case for reclassification (Kerr 2000). The one compound heterozygote (i.e. two different CFTR mutations) and two 5T heterozygotes (i.e. with one CFTR mutation, and a mutation in the regulatory section) are described as having ‘the CFTR genotypes that are most commonly seen in patients with congenital absence of the vas deferens’ (1998: 656). Cohn et al. claim that:

these findings suggest that in this group of patients, abnormal CFTR genotypes cause pancreatitis as one component of an inherited syndrome affecting multiple epithelial tissues and that such patients should be examined for congenital absence of the vas deferens [ref.] and sinusitis [ref.] which are not typical of pancreatitis (Cohn et al. 1998: 656).

When discussing those patients with no CFTR mutations, the authors adopt a version of Kerr's ‘appeal to unknown mutations’, specifically an appeal to untested mutations (Kerr 2000). They point out that the genotyping in their research only looked for 17 of the then 500 known CFTR mutations. Therefore there could be other pancreatitis patients in their study who carried one or even two copies of CFTR mutations, beyond the seven identified. The finding of only seven carriers in their sample is a ‘cautious interpretation’ (Cohn et al. 1998: 656), a position also adopted with regard to whether any of these seven are actually compound heterozygotes, with an undetected second mutation.

The lack of evidence concerning other CFTR mutations means that the authors: ‘could not determine whether having one copy of an abnormal CFTR allele (as is found in carriers of cystic fibrosis) is sufficient to predispose persons to pancreatitis’ (Cohn et al. 1998: 657). If CFTR mutation heterozygotes (i.e. carriers) are deemed to be at increased risk of pancreatitis, then this does not sit well with the conventional view of cystic fibrosis genetics which requires that heterozygote carriers of CF alleles are not negatively affected by their genome. Admitting pancreatitis into the CF continuum would mean accepting that CF can be caused in heterozygotes. A partial solution lies in suggesting varying tissue sensitivity to CFTR mutations with the pancreas and vas deferens being more susceptible to CFTR related damage (Cohn et al. 1998: 657). We can see here the construction of a continuum of tissue sensitivity and CFTR activity level with the possibility of pancreatitis being reclassified as a mild form of cystic fibrosis. These authors then suggest that genetic testing might be needed as a way of identifying patients at risk of pancreatitis and hence susceptibility of damage from drink and drugs (Cohn et al. 1998).

Sharer et al. (1998) offer a less expansionist position in their article, published in the same issue of the New England Journal of Medicine. They report a study of 134 patients with chronic pancreatitis, where ‘No patient had a mutation on both copies of the CFTR gene’ and 18 had a single CFTR mutation (Sharer et al. 1998: 647). The 5T variant was identified in a total of 14 patients, 4 of whom also carried a CFTR mutation. Another 10 were from the non-carrier group of 116 patients. The authors are clear on the relationship between these pancreatic patients and cystic fibrosis:

None of the 18 patients with a CFTR mutation alone or in combination with a 5T allele met the diagnostic criteria for cystic fibrosis when all the evidence was considered [reference: to Stern (1997)] (Sharer et al. 1998: 647).

Sharer et al. are also cautious about comparisons between CFTR-related pancreatitis and CBAVD. Although they add pancreatitis to the list of conditions (which includes CBAVD) in which CFTR mutations are of pathogenic importance, they seem to be aware of the complexities that this might raise: ‘There is a fundamental difference between our findings and those reported in patients with congenital absence of the vas deferens’ (Sharer et al. 1998: 650–1).

This is an important challenge to the CF continuum. While almost every male with ‘classic’ CF also has CBAVD, it is not the case that everyone with CF has pancreatic problems. Therefore pancreatitis is not a necessary symptom of ‘classic’ CF; it is present in only the severest cases. It would therefore be difficult to place ‘just pancreatitis’ on the CF continuum, in the same way ‘just CBAVD’ has been incorporated. These authors make no attempt to suggest that further CF associated alleles might be found on heterozygotes’ other chromosome, if only time and money permitted the necessary research. Sharer et al. (1998) present the concerns of some in the cystic fibrosis medical community when they claim that far from improving diagnosis, the discovery of the CFTR-gene (i.e. the geneticization of CF) has meant ‘The definition of cystic fibrosis has become progressively more hazy’ (Sharer et al. 1999: 238). As in the case of CBAVD, there is a tension between clinical practicalities and the genetic reclassification: ‘Because the phenotypic spectrum that may now legitimately be called cystic fibrosis has become so large, it is unhelpful in the clinical context’ (Sharer et al. 1999). They suggest a new classification system is needed which would include: cystic fibrosis ‘disease’ which is ‘the progressive suppurative respiratory tract condition irrespective of other phenotypic features’; and cystic fibrosis ‘syndrome’ which includes ‘pancreatic manifestations, congenital absence of the vas deferens, and lesser pulmonary manifestations’ (Sharer et al. 1999).

Resolving uncertainty

Classification systems are balancing acts, representing the views of several different constituencies at once (Bowker and Star 1999: 324). It should comes as no surprise then that there is an element of confusion and uncertainty to the CF classification system. But there is only so much ambiguity a classification system can take before people want to sit down formally and agree on what to call things, and where to put things in relation to each other. For example, it happened in the case of diabetes in 1979 and again in 1997, when the American Diabetes Association convened committees to formally settle the classification and diagnosis of diabetes (Hedgecoe 2002).

The 1998 panel of experts convened by the US Cystic Fibrosis Foundation (CFF) produced a consensus statement on the criteria for CF diagnosis (Rosenstein and Cutting 1998: 590). This clarified many problems, pushing strongly for a definition of CF in clinical terms, with the use of laboratory tests to confirm CFTR malfunction. On the one hand, this CFF consensus statement confirms the expansion of CF to include men with CBAVD and two CFTR mutations (Rosenstein 2002: 84), but, on the other hand, various clinical reports (Mekus et al. 1998, Mickle et al. 1998, Crowley and Bush 2002) have muddied the water to such an extent that: ‘We are thus left with the paradoxical situation of having some patients with typical CF in the absence of mutations, and others, with CFTR mutations in the absence of clinical features’ (Rosenstein 2002: 84).

As well as CBAVD and milder forms of male infertility (such as Congenital Unilateral Absence of the Vas Deferens – CUAVD), conditions such as: idiopathic pancreatitis, disseminated bronchiectasis, allergic broncho-pulmonary aspergillosis, atypical sinopulmonary disease, diffuse bronchiectasis associated with rheumatoid arthritis, and sarcoidosis have all been linked to defects in the gene which codes for CFTR (Zielenski 2000: 125). Classification systems which rely on linking CFTR mutations to CF obviously face a few problems.

More recently, Bush and Wallis (2000) proposed a new classification system for cystic fibrosis which attempts to resolve the problems with the CFF consensus statement. Their central theme is that CF should not be seen as an ‘all or none’ disease, stating that ‘few physicians use this model in practice’ (Bush and Wallis 2000: 140). Their alternative is a four category classification covering: ‘Genetic pre CF’ and ‘Electrical and chemical pre CF’ (the members of which may never develop any disease, but who should be observed in case, e.g. CBAVD); obvious clinical CF; and sub-clinical CF (where a single organ may be affected, but it may have no further clinical significance).

Obviously it is too early to say whether Bush and Wallis's suggestion will be taken up by the CF community, either in terms of use in the clinic, or discussion in the literature. It does suggest though that at least some clinicians are not happy with the current classification system(s), and feel they do not adequately represent their practice. While CF has expanded to include CBAVD, the range of conditions which can be linked to mutations in the CFTR gene prevents further such expansion, producing instead the current state of uncertainty.

6. Conclusion

I noted in the introduction to this paper that classification systems are negotiated compromises between different groups, and that in at least one case (diabetes in the 1980s), there was tension between clinicians and researchers over classification. Perhaps, and at this stage it is only speculation, the same can be said of current CF classification. In his work on gene therapy for CF, Alan Stockdale (1999) noted the tension between CF clinicians and researchers, and some have suggested there is tension between clinicians, who doubt the pragmatic usefulness of the new geneticized classification system in CF, and researchers, who seem to be pushing it (Colin et al. 1996).

The consensus statement on CF diagnosis made by the US Cystic Fibrosis Foundation lists four reasons why CF genotyping is desirable: diagnosis, family information, prediction of phenotypic features and ‘categorization of patients for research protocols’ (Rosenstein and Cutting 1998: 592). From a research point of view, it makes sense to have clear categories, and to be able to distinguish between patients on unequivocal, genetic grounds. But the authors of the statement do not make clear why it is desirable for clinicians to adopt categories required for research. The consensus statement assumes that classifications suitable for researchers are also suitable for clinicians. This conflicts with the position of Sharer and colleagues (1999) who suggest a division between CF ‘disease’ (used by clinicians) and CF ‘syndrome’ (for researchers). Recently, a meeting organised by the World Health Organisation drew up a proposed classification of CF for the next International Classification of Diseases (ICD −11). This concluded that forms of male infertility, sino-pulmonary disease and chronic pancreatitis should not be diagnosed as forms of CF, but should be seen as CFTR-related (WHO 2000). Perhaps this shows that the current classificatory uncertainty brought about by the availability of CFTR testing may yet be unwound.

In this paper I have claimed that disease classification systems are socially constructed and have shown how in the wake of genetic research, the CF classification system has been constructed to include the condition CBAVD24. In addition to this expansion, the geneticization of CF has also produced a degree of nosological uncertainty, which is still being worked out in the community of CF researchers and clinicians.

Like Kerr's work, this paper has unpacked the ‘black box’ of CF classification, and shown how the condition is constructed to expand. Far from being an obscure academic issue, the geneticization of CF concerns concrete points of health policy. As Koch and Stemerding show, the decision to implement a CF screening programme is never just based on impartial scientific advice. Rather, it is the result of a complex negotiation between different societal actors and networks (Koch and Stemerding 1994). And sociologists are among those groups who can contribute to debate about how such programmes should be implemented.

Cystic fibrosis was the test case for the technologies and concepts that would open up human illness and allow us to treat previously impervious conditions. It has turned out to be far more complex than was expected. Presently with over 1000 mutations (and counting), the geneticization of CF should serve as a warning to those who assume that the introduction of a genetic test will automatically introduce clarity and constancy into medical classification. The geneticization of CF is a cautionary tale about how we cannot be absolved of our responsibility for social decision-making in medicine. The introduction of the test for the CFTR gene did not remove social decisions from the classification of cystic fibrosis, it highlighted them.

Acknowledgements

Thanks to Jon Turney, Brian Balmer, John Waller, Anne Kerr, Richard Tutton Carlos Novas and Paul Martin who all read and commented upon various drafts of this article. Thanks to one anonymous SHI referee, and especially to Alan Stockdale who improved this article significantly. This research was supported by the Economic and Social Research Council, Postgraduate Award No.: R00429834453.

Notes

  • 1

    For supporters of the term see Hubbard and Wald 1992, Hoedemakers and Ten Have 1998, Van Dijk 1998. For a critique of some of the concepts supporting geneticization see: Condit et al. 1998, Condit 1999a and b, Novas and Rose 2000. For debate about the role of social science in exploring geneticization: Nelkin and Lindee 1998, Ten Have 2000, Hedgecoe 2001a.

  • 2

    ‘Classification schemes and diagnostic criteria should, above all, be utilitarian. They should help physicians select a management program which is of maximum benefit and minimum harm to individual patients in their offices today’ (Genuth 1982: 1191).

  • 3‘.

    a classification appropriate to a clinician whose concern is with diagnosis and treatment may well be inappropriate to a basic scientist whose concern is research strategy and experimental design’ (Keen 1985: 31).

  • 4

    In addition to the mutations in the coding sections of the CFTR gene, there are also variations in non-coding regions which play a role. Labelled 5T, 7T and 9T they refer to a regulatory section for Exon 9 of the CF gene, which varies in length and can come in three sizes: 5, 7 or 9 thymidines long. It is believed that the shorter the length the less the effectiveness there is in translating the gene into CFTR. When a person inherits a CFTR mutation on one chromosome, and the R117H mutation with the 5T variant on the other, then their body is unlikely to make enough CFTR and they will develop the clinical symptoms of CF. If the variant is 9T, then they may well have no symptoms at all, and be classed as a carrier; 7T and R117H leads to male infertility.

  • 5

    For more detail on the method used in this section see Hedgecoe 2000. For other examples Hedgecoe 2001b and 2002.

  • 6

    The 1996 article is the direct successor to a very heavily cited article in this area by Wood et al. (1976) for which the Science Citation Index lists 696 citations, an average of 29 citations a year.

  • 7

    The term CF continuum is borrowed from Kerr (2000) but can also be found in the technical literature, e.g.Zielenski (2000).

  • 8

    Other examples of this slippage can be found in Davis et al. 1996: 1229 (twice), 1231, and 1234.

  • 9

    It is important to note that this is not a deliberate consequence of the naming of the protein (Riordan et al. 1989).

  • 10

    See Davis et al. 1996: 1230.

  • 11

    There is a possibility that these two articles use the term ‘syndrome’ in slightly different ways with Wood et al. (1976) using it in the way described (as a collection of symptoms with no clear cause) and Davis, Drumm and Konstan (1996) as meaning a disease which manifests itself in a variety of different organ systems. But if this were the case, one would expect consistent use of ‘CF syndrome’ throughout the 1996 article, especially in those sections that deal with the manifestation of cystic fibrosis in the various body systems. But there is a clear cessation of the use of the term syndrome after the section called ‘Cloning the Gene’ (p. 1230); following this section and its introduction of CFTR into the discourse, there is no further mention of the CF syndrome.

  • 12

    Wild type refers to the protein present in unaffected individuals.

  • 13

    Thanks to Alan Stockdale for this.

  • 14

    Where a strategy is ‘simply a theme or theory enveloping and reflecting patterns in the formation of objects, subjects and concepts within a discourse’ (Haller 1998: 64).

  • 15

    Kerr notes that elsewhere in the literature even those cases of CBAVD without CFTR mutations are linked to CF on the grounds that the mutations are there, they just have not been identified by the test used.

  • 16

    1997: 487, 489, and 490 (four times).

  • 17

    Anne Kerr has pointed out that discussion in terms of the ‘classic case of CF’ dates back to the 1940s (Kerr, Personal communication, 28.04.2000), but this does not undermine the way in which Stern uses this phrase in his review article. The fact that others use classic with quotation marks suggests that use of this phrase is, at the very least, open to interpretation.

  • 18

    Davis et al. claim that ‘Eventually, all patients with CF manifest lung disease, though it may develop after the time of diagnosis’ (1996: 1229–1230) contradicting the inclusion of CBAVD in the CF clinical continuum (‘CBAVD is also found in the absence of other organ involvement’ (1996: 1232)).

  • 19

    I stress again that I am making no claims about clinicians’ actual diagnostic practices; I am simply outlining the representation of CF classification in the literature.

  • 20

    The description of CBAVD as a ‘primarily genital form of CF’ occurs in a number of different papers, and seems to be uncontroversial: Amos et al. 1992, Anguiano et al. 1992, Cornud et al. 1997, Dörk et al. 1997, Kanvakis et al. 1998.

  • 21

    This same review also classifies other forms of male infertility such as Unilateral congenital absence of the vas deferens, bilateral ejaculatory duct obstruction and bilateral obstructions within the epididymis as ‘primary genital forms of CF’ (2000: 73).

  • 22

    For example, ‘the R117H mutation on a 7T background affects almost exclusively the male reproductive tract and is not sufficient to produce other CF symptoms’ (Zielenski 2000: 126).

  • 23

    This is more than just an intellectual point. Several authors have noted CBAVD patients’ reluctance to be re-classified as CF sufferers, for psychological reasons and more practical ones, such as health insurance (Colin et al. 1996, Rosenstein 2002).

  • 24

    It would also be wrong to assume that the introduction of genetic explanations into disease classification necessarily leads to such expansion; in diabetes it lead to division into numerous subgroups (Hedgecoe 2002).

Ancillary