Biopolitics now addresses human existence at the molecular level: it is waged about molecules, amongst molecules, and where the molecules themselves are at stake.

Nikolas Rose (2001)

We are reminded of Louis Pasteur's insight. He said that in dealing with disease, the terrain is as important as the germ.

President Yoweri Museveni of Uganda, (Prepared remarks to the Pharmaceutical Research and Manufacturer's Association, 12 June 2003)

Introduction

The year 2003 was a landmark year in the fight for global access to HIV drugs, as the first free antiretrovirals (ARVs) funded by the Global Fund to Fight AIDS, TB and Malaria began to hit the ground in low-income countries, and the announcement of a US-funded initiative promised billions more dollars for HIV treatment in 15 ‘target’ countries, mostly in Africa.Footnote 1 However, the years leading up to these developments had been fraught with contention in international health. In wealthy countries, these drugs – available since early 1996 – had turned HIV into a chronic, manageable disease for many patients. Yet patients in the global South remained largely untouched by this pharmaceutical breakthrough, as the high cost of ARVs proved to be prohibitive for most individuals and governments. This disparity was most apparent in sub-Saharan Africa, where an estimated two-thirds of the world's HIV-positive individuals were living, the vast majority untreated.

The consequences of this inequality and what should be done about it were the subject of much debate in the late 1990s and early 2000s (Behrman, 2004; Iliffe, 2006). In these arguments over global treatment, the subject of drug resistance – the mutation of the virus such that it no longer responds to medications – took on a particularly powerful political valence. Activists and clinicians advocating for treatment in Africa found themselves thwarted by the conventional wisdom in HIV medicine at the time. This held that less than perfect medication adherence, that is, missing even a few doses, would lead rapidly to drug resistanceFootnote 2 (Chesney, 2003) and that patients in Africa were more likely to miss doses than patients in industrialized nations (Stevens et al, 2004). Indeed, some researchers cautioned that widespread drug access in Africa could lead to ‘antiretroviral anarchy’ and turn poor countries into ‘a veritable “petri dish” for new, treatment-resistant HIV strains’ (Harries et al, 2001; Popp and Fisher, 2002). Other experts and opinion leaders variously described African patients as unlikely to take ARVs properly owing to poverty, lack of education, poor health infrastructure, government corruption and ‘the many other problems that poor people in developing countries face’ (World Bank, 1999, p. 181; see also Rauch, 2000; Sullivan, 2001). Perhaps most infamously, Andrew Natsios, the Chief Administrator of the US Agency for International Development, and an unnamed senior US Treasury official each described African patients as lacking the requisite ‘concept of time’ – more specifically, ‘Western time’ – necessary to comply with ARV regimensFootnote 3 (Donnelly, 2001; Kahn, 2001).

On the basis of this reasoning, Natsios and others made the argument that the Global Fund efforts should be devoted to HIV prevention in Africa, rather than ARV treatment, as it was feared that the provision of HIV drugs might result in the development of a ‘doomsday’ strain of drug-resistant virus. This assertion was undergirded by the long tradition of portraying Africa as a place of social disorganization, illiteracy, esoteric sexual practices, ‘traditional’ and/or ‘exotic’ cultural beliefs and rampant disease in the Western scientific and popular presses (see Farmer, 1999; Treichler, 1999; and Patton, 2002 for critiques). Yet, it would be unfair and inaccurate to blame these voices alone for the slowness with which Western countries responded to the AIDS epidemic in Africa. Fundamentally, the lack of treatment on the continent was a reflection of pharmaceutical pricing and political will, as the governments of wealthy donor nations were reluctant to challenge the patent laws that made their pharmaceutical industries so lucrative (and ARVs so costly). It was only after cheap, generic ‘copycat’ ARV drugs from IndiaFootnote 4 began to be imported into Africa that major policy shifts facilitating free ARV access occurred, such as the founding of the Global Fund and the US President's Emergency Plan for AIDS Relief (PEPFAR). Nonetheless, by allowing the lack of HIV treatment in Africa to be framed as protective of public health, rather than the public health disaster it is now widely held to be, claims about drug resistance provided scientific legitimacy for a reluctance to support treatment that was largely economic and political in nature.

In addition, these scientific claims about HIV treatment and drug resistance were remarkable for their speculative nature. At the time, there were virtually no data available on ARV resistance in Africa due to the scarcity of the drugs on the continent. Furthermore, as I argue below, ‘basic science’ knowledge of HIV resistance – in other words, molecular knowledge about the mutations that cause the virus to become resistant – was (and to a large extent still is) based on genetic forms of HIV rarely found on the African continent. To elaborate: virological and molecular biology research on HIV treatment have traditionally focused almost entirely on the genetic ‘strain’ or subtype of HIV found predominantly in North America, Europe and Australia. It is this virus, known as ‘subtype B’, that has been used to establish nearly all of what is known about ARV treatment and HIV drug resistance (Jülg and Goebel, 2005; Brenner, 2007).

In fact, there are at least 10 different genetic subtypes of HIV, and those responsible for the lion's share of the African epidemic (as well as the global epidemic as a whole) have received little scientific attention until very recently. Instead, scientists have generally relied upon a handful of viruses isolated in American and French laboratories during the first years of the epidemic as the ‘reference strains’ upon which their work is based. The geographic and genetic specificity of these strains is important because the development of ARV drugs, the definition and measurement of drug resistance, and key diagnostic technologies used in patient care are all contingent upon genetic mapping and molecular modeling of the HIV virus. In other words, medical knowledge gained from familiarity with these viruses may not always accurately reflect the ‘non-B’ HIV strains that constitute nearly 90 per cent of the world's infections (Spira et al, 2003).

Using fieldwork conducted among North American and Ugandan HIV scientists in 2004 and 2005, this article aims to describe how these particular viruses from the United States and Europe came to be so central to HIV science, and argues that the implications of this fact are at once clinical, epistemological and political. Though current ARV regimens thankfully appear to work well against all viral subtypes, some scientists view their universal efficacy as ‘lucky’ and fear that future drugs, diagnostic technologies and vaccines designed using subtype B virus might work less well on other strains. Clinically, this presents the possibility of an ARV pharmacopoeia that could be prejudicial not only in pricing, but also in efficacy. In addition, the ability to measure and treat drug resistance in viruses other than subtype B may be compromised by the comparative scarcity of molecular knowledge about how resistance emerges and evolves in the global South.

Epistemologically and politically, the story I tell here illustrates an important but perhaps little explored link between science and technology studies and the political economy of health. By tracking the evolution of HIV science, I argue that the geopolitics of the AIDS epidemic is present at the molecular level, in the laboratories where our knowledge about the molecular biology of HIV and ARVs is produced. The result is a kind of ‘molecular politics’ (Rose, 2001) in which the global inequalities of the AIDS epidemic are manifest at the most minute scale, embedded within the very materials and tools scientists use to study HIV. This article is an examination of those molecular politics, and their relationship to the politics of ARV treatment access.

Mapping Molecular Medicine

HIV medicine is a heavily molecularized field, in which many diagnostics (viral load tests, drug resistance assays) and therapeutics (protease inhibitors) are engineered based on very detailed, codon-by-codonFootnote 5 knowledge of the nine genes that constitute the HIV virus. The use of the term ‘mapping’ to describe this kind of genetic knowledge and practice is more than metaphorical, as the knowledge that is generated by genetic sequencing is essentially spatial, telling scientists the order and location of the molecules that make up the viral genome. In addition, mapping genes and mapping territory serve many of the same purposes: both provide a means of orienting one's self, a way of generating coherence, and a way to establish relationships between things (Rheinberger and Gaudilliere, 2004). Critical geographers have long argued against ‘representationalism’, in which maps are accepted as objective and straightforward depictions of space. Rather, maps should be understood as socially and historically contingent, and, as David Turnbull argues, as expressions of power (Turnbull, 2004). In this article, I aim to show that this argument holds for genetic maps as well.

For example, in his study of the colonial mapping of India, geographer Matthew Edney describes how map-making allowed the British to transform a disparate collection of empires and territories into the single, coherent entity of ‘British India’ (Edney, 1997). These maps then became rapidly naturalized, rendering the exclusions involved in constructing British India invisible. The representation of the territory as it appeared on colonial maps thus became taken-for-granted, and unquestioned – the map and the territory became synonymous. Matthew Sparke has described how alternative, ‘contrapuntal’ cartographies may, in turn, upset the coherence of such naturalized maps (Sparke, 1998). And Turnbull has made a direct link between cartographic and scientific knowledge, arguing that maps are ‘an apt metaphor for scientific discourse. Scientific representations of the phenomenal world are, like maps, laden with conventions, which are kept as transparent, as inconspicuous as possible’ (1989, p. 9).

If we interrogate how a map is constructed, we are able to understand the partiality and contingency of its representation. We are able to understand how the map is productive of certain possibilities – certain forms of understanding – and better able see what was necessarily included and excluded in order to produce a coherent entity. In Edney's analysis, this coherent entity was British India. In my analysis, I aim to show what was excluded in order to create a coherent map of the HIV virus and, by extension, drug resistance. My point is that the generation of coherence has resulted in a situation in which the viral sequences (the genetic maps) of a particular strain of HIV, a strain found mainly in the United States and Europe, now serves as the common template for understanding and studying HIV worldwide. This fact has important implications for addressing the global epidemic.

HIV Subtyping and the Production of Coherence

HIV is highly error-prone in its replication process, causing the virus to mutate rapidly and constantly. Each viral offspring differs slightly from its parent by several mutations. This means that any given individual infected with HIV is carrying a population of genetically different but closely related viruses, rather than genetically identical copies of a single virus. For this reason the scientific literature refers to HIV as a ‘quasispecies’, meaning a mixture of genetic variants of a virus as opposed to a single virus with a consistent genome. This extreme diversity means that generating coherence is one of the key challenges involved in working with HIV in the laboratory. The result is that the need for a genetically consistent, workable virus often takes precedence over approximating HIV as it exists ‘in nature’.

However, despite its diversity, some HIV viruses are more similar than others. The relatedness or ‘phylogeny’ of HIV viruses is based on genetics. Viruses are mapped and grouped according to the similarity of their genetic material, which is understood to reflect their evolutionary proximity. The basic phylogeny of HIV is depicted in Figure 1, showing both HIV-1 and HIV-2 (found only in West Africa), as well as various groups and subtypes (also called ‘clades’) of HIV-1. In common parlance, general references to ‘HIV’ are more accurately describing HIV-1 group M, which accounts for 99 per cent of the world's infections.

Figure 1
figure 1

HIV genetic variation.Source: Haeusser (2001).

Full size image

There are currently at lease 10 identified genetic subtypes of HIV-1 group M. The first nine subtypes are each identified by a different letter of the alphabet. The tenth subtype includes several ‘recombinant’ viruses that are actually mixtures of more than one subtype, such as A/E. (Sometimes these ‘circulating recombinant form’ or CRFs are each counted as a separate subtype, making the total number of subtypes somewhat variable.) The prevalence of HIV subtypes varies geographically, as shown in Figure 2.

Figure 2
figure 2

Global distribution of HIV-1 subtypes.Source: Kahn (2003).

Full size image

In the United States, Western Europe, Australia and parts of Latin America, the vast majority of infections are subtype B infections, represented on the map in purple. In part because it is the site of the oldest HIV epidemic, sub-Saharan Africa encompasses a much greater diversity of subtypes, with the most common being C and A (Kantor and Katzenstein, 2004; Iliffe, 2006). Worldwide, the most prevalent subtype is type C (aqua green on the map), which accounts for 47 per cent of infections globally. This is because type C predominates in the areas of the world that bear the greatest burden of infection – particularly southern and eastern Africa.

The map below shows another way of looking at the same phenomenon. Here the grayscale shows HIV prevalence, with the darkest areas being those with the highest numbers of infections. These hardest-hit regions are also the geographic areas where subtype C predominates.

Like all maps, those shown here are notable for what they include and exclude. While they are, on the one hand, a useful tool for demonstrating the diversity of the HIV virus in relation to geography, they also – by necessity – simplify reality. For example, they complicate the notion of ‘HIV’ as a single virus, but naturalize the differences between subtypes – differences that could easily be deconstructed in their own right. These maps also reify the geography of HIV subtypes, linking particular subtypes to clearly bounded regions. The map in Figure 2, published by the International AIDS Vaccine Initiative, comments on this in the text that accompanies the graphic, cautioning the reader that, ‘this map … demarcates boundaries more distinctly than they exist in reality’ (Kahn, 2003). The map in Figure 3, it might be argued, goes in the other direction by including a list of ‘African clades’ at the bottom of the graphic, thus explicitly linking geography and subtype. Thus, while these maps open the ‘black box’ of HIV diversity, they also depend upon other black boxes that remain closed. Nonetheless, they are instructive in demonstrating one of the key challenges involved in the scientific study of HIV: the fact that it is not one virus, but many.

Figure 3
figure 3

Subtype diversity of HIV-1 infections prevalent worldwide.Source: Spira et al (2003).

Full size image

The ability to understand HIV at the molecular level – to literally make the genomics of the virus legible – is what has allowed scientists to comprehend the breadth of the virus's diversity. However, this exquisite familiarity with the details of HIV genomics also creates certain tensions in relation to generating knowledge about the virus. Specifically, scientists confront the fact that there is no single viral sequence that represents HIV. There is no unity to HIV, no coherence. This presents problems for scientists who must choose a virus to use when conducting basic research, developing drugs and vaccines, and designing diagnostic technologies (such as viral load tests). Furthermore, this viral diversity presents a problem of comparison – a particularly important element in the assessment of drug resistance. A patient's virus is tested for resistance by comparison to a virus known to be drug-sensitive. This drug-sensitive virus then serves as a ‘reference strain’ against which the patient's virus can be assessed. But given the incredible diversity of HIV, and thus the virtually infinite number of drug-sensitive viruses in existence, how is a single virus chosen to serve as a reference strain?

The Contingency of the Arbitrary

When I asked David Capelli,Footnote 6 a researcher who studies the molecular biology of HIV, about how the reference strains came to be selected, he told me the choice was ‘somewhat arbitrary’. Capelli is a prominent junior faculty member at one of the several California universities where I conducted a number of my interviews. ‘In fact’, he continued, ‘the idea that there was a normal strain of HIV is sort of strange to begin with. There really is not. It exists as a population. It's sort of like saying, “what is the representative American?” Well, I don’t know. It's a highly diverse country’. In choosing a reference strain to work with, scientists selected from viruses already available to them in their laboratories; in the United States and Europe these viruses were all of the subtype B variety, as this was the strain infecting the vast majority of local patients. Many HIV researchers are physicians who first encountered the virus in their medical practice. As such, the choice to focus their research on subtype B virus reflected both its convenient availability and the desire to work on the strain that was infecting the patients under their care. However, it is important to recognize this convenient and arbitrary choice as both historically specific and socially contingent. The uptake of subtype B viruses as the basis for HIV laboratory research and technology development was not random, but reflects the fact that the great majority of both research funding and infrastructure are located squarely in the United States and Western Europe, where subtype B predominates. The result was the establishment of a reference strain that represents only 12 per cent of worldwide infections (Kantor and Katzenstein, 2004).

Interestingly, convenience shaped not only the choice of subtype B virus as the reference strain, but also a very specific virus within this subtype. The most commonly used reference strains are closely related and go by a number of names including NL4-3, HXB2 and LAI. This proliferation of names is a product of the complex and contentious history of the virus's discovery. Capelli explained it to me as follows:

So its full name is pNL4-3. And you mention that to basically any lab scientist who works with HIV-1 and they go, ‘Oh, L4-3.’ It probably is the basic reference virus used in North America. It's a well characterized strain and people understand it … . The history on this – this would be I think a good thing to look into. Basically, these are some of the earliest isolates that were grown in the 1980s. And they were some of the earliest variants. So as you know, in the very early stages of the epidemic there was (A) some confusion over what was the causative agent and then, (B) once it was determined that it was HIV-1, there was a great deal of energy put into determining how to appropriately grow and sustain these viruses. And some viruses grow better in culture than others. And NL4-3 was one that did. They also called it HXB2 or LAI…And my understanding is that these viruses are all highly related and came from a handful of labs in the 1980s.

The labs that Capelli refers to are those of Luc Montagnier and Robert Gallo. In the early 1980s, these two scientists emerged at the forefront of the search for a virus that could be the cause of AIDS: Montagnier at the Pasteur Institute in Paris, and Gallo at the National Cancer Institute in Bethesda, Maryland.Footnote 7 Both scientists were specialists in the study of retroviruses – a family of viruses that carry their genetic material in the form of RNA rather than DNA – and both thought that a retrovirus could be the cause of AIDS, a hypothesis that turned out to be correct. In the early 1980s, both labs worked to isolate a retrovirus from patients who were suffering from AIDS. The relationship between the labs was competitive, but they nonetheless exchanged samples according to common scientific etiquette.

In 1983, Montagnier's lab isolated a previously undocumented virus from a patient with lymphadenopathy, the swollen lymph nodes that are one of the hallmarks of AIDS. He named the virus ‘lymphadenopathy-associated virus’ or LAV. Soon thereafter, the lab isolated similar viruses from patients with more advanced disease. One of these viruses – from a patient identified only by his initials ‘LAI’ – was particularly fast-growing and aggressive. However, the Montagnier group's attempt to describe the virus in greater detail was stymied by the difficulty of culturing it in the lab. The virus was difficult to grow because it killed all the cells in which it was cultured within a matter of days and once the cells were dead, the viruses died, too. After much trial-and-error, the French scientists developed a technique of transferring the viral cultures to fresh cells every 3 days over the course of several weeks, a laborious process that would eventually yield them enough virus for further laboratory studies (Garret, 1994).

Shortly after Montagnier's discovery, Gallo's lab also isolated a virus from a patient with AIDS. They named the virus HTLV-III, believing it to be related to a group of human T-cell lymphotropic viruses (HTLVs) that Gallo had discovered in the late 1970s. The American and French groups agreed to compare their viruses and, if they were found to be the same, to hold a joint press conference in which they would co-announce the discovery of the virus that caused AIDS (Gallo, 2002; Rainey, 2006).

What happened next initiated a controversy that would drag on for nearly a decade. Before the viruses could be compared, Margaret Heckler, the US Secretary of Health and Human Services, held a press conference to announce that the AIDS virus had been discovered. At the conference, Gallo was heralded as the discoverer of the AIDS virus, a title he embraced. Montagnier's team was not invited to the press conference, nor was their work cited. Gallo did not dispute that Montagnier had isolated a virus earlier than he had isolated HTLV-III. Rather, he defended himself as the discoverer of the AIDS virus by arguing that it was on the basis of his HTLV-III research that the definitive causal link between the virus and the syndrome was established, and that a blood test could be developed. Gallo's claim was boosted by his team's development of an ‘immortalized’ cell line that did not die when cultured with the virus, eliminating the tedious culturing process used at the Pasteur Institute and providing a technology key to the development of the AIDS antibody test (Garret, 1994). On the same day that the press conference was held, Gallo filed a US patent application for a blood test that would identify infection with the virus. The US government granted him the patent – worth US$100 million annually in sales and $100 000 to Gallo personally – and denied a patent to the French (Rainey, 2006).

Montagnier and the Pasteur Institute challenged the patent, beginning a protracted struggle between the French and the Americans that would eventually involve both heads of state. Gallo continued to assert that HTLV-III was the virus that caused AIDS, and opposed the 1986 renaming of the virus ‘HIV’ (human immunodeficiency virus) by the International Committee on the Taxonomy of Viruses (Epstein, 1996). However, a genetic analysis of both viruses later revealed that they were essentially identical. This confirmed long-held suspicions on the French side that the isolate that Gallo had ‘discovered’ was actually derived from Montagnier's aggressive LAV/LAI virus, which is now believed to have contaminated Gallo's samples as well as those in a number of other labs with which Montagnier had shared cultures (Gallo, 2002; Montagnier, 2002). Eventually, US President Ronald Reagan and French Prime Minister Jacques Chirac declared the two scientists co-discoverers and agreed to split the patent proceeds. Today, Montagnier is generally recognized as the first to identify the virus, while often Gallo is credited with solidifying the link between the virus and AIDS and developing the technology that made the HIV blood test possible (Stine, 2004; Rainey, 2006). Nonetheless, it was Montagnier and his colleague Francoise Barre-Sinoussi who were awarded the 2008 Nobel Prize in Medicine for the discovery, with Gallo conspicuously excluded.

It was the isolate of the virus taken from the French patient with the initials LAI and its derivative copies or ‘clones’ that would go on to become one of the most commonly used viruses in HIV research. Interestingly, LAI was not selected on the basis of its representativeness. In fact, most scientists I spoke with readily agreed that the reference strain they used was not very similar to the type B viruses found in patients (much less to the other non-B subtypes). Rather, these strains were used because they grew well under laboratory conditions. Having undergone genetic changes over the course of numerous manipulations in Paris and Bethesda, these viruses were now what scientists call ‘lab-adapted’. Paula Leigh, the Associate Director of a non-profit California virology laboratory, explained it as follows:

Whatever the first virus that Gallo or Montagnier isolated, that's a lab-adapted strain. And it was grown out in the laboratory in vitro and propagated. And maybe even cloned out. And those are viruses that generally replicate very, very easily. You can grow them easily, that's how they found them in the first place. And they might actually be quite different from what is actually growing in people.

Leigh went on to give her understanding of the complex nomenclature behind the reference strains:

LAV is the original virus that they isolated. And they call it different things depending where in the world you are – LAV, LAI, BRU, HTLV-III. And then there's another reference strain called NL4-3 which is actually a hybrid virus from two patients [that somebody] isolated and they spliced together and it's just used as a reference virus because it grows very well in tissue culture … .And there's another one, HXB2. It's often used as a reference strain and that is just like LAI I think but there's a slight difference from it.

Likewise, Ralph Ernst, a Swiss scientist working at a university-affiliated blood bank in California, echoed Leigh's assessment that the reference strains were ‘different from what is actually growing in people’. He told me, ‘Not only did people use subtype B, they probably used the wrong subtype B. People basically used what they had. And the first thing they had was the cloned viruses – the one that Gallo/Montagnier isolated, HXB2. So everybody kind of uses a very limited set of the oldest virus. Why?’ He then answered his own question, ‘Because they’re convenient. Everybody's got it. You can compare data across labs’.

It was this ease of use, rather than the virus's representativeness, that made the LAI virus ‘the right tool for the job’ for these scientists (Clarke and Fujimura, 1992; see also M’charek, 2005, for a similar case in population genetics). Representativeness, in this context, was less important than availability and adaptivity to laboratory conditions. A virus that was genetically more similar to the viruses in patients could have been chosen, but might have been difficult to grow in the lab.Footnote 8 In addition, once the LAI strain and its cousins had become the common currency of lab work, switching to a different reference strain was impractical because it would impede the comparison of data between laboratories. Jim Greene, the head of a prominent university-based virology lab in California, put it most succinctly when he told me that ‘this is an example where consistency is more important than being right’. After all, he continued, ‘there is no way to be right’. In his view, it was more important for scientists to be explicit about which reference strain they were using and to be consistent in this choice than to use a reference strain that more closely resembled those in patients.

To be sure, this is not a new issue in science. In fact, this twenty-first century story carries many echoes of laboratory work a century earlier, when Drosophila melanogaster, or the common fruit fly, established itself as the first standardized laboratory creature during the early years of experimental genetics at Columbia University in New York. Much like the HXB2 and NL4-3 viruses, D. melanogaster became an integral part of laboratory practice not through calculated consideration but because it was easily accessible to urban US biologists and it was ‘hardy’, meaning it was able to survive and flourish in the laboratory environment (Kohler, 1993). Over time D. melanogaster – like the HIV reference strains – became so commonly used that it established itself as the ‘standard fly’ for use in genetic research, even though by then it had evolved to be quite different from flies found in the wild. As Robert Kohler has argued in his account of the D. melanogaster story, once established, a standard organism becomes very difficult to replace or alter due to ‘what evolutionary biologists call a founder effect: the more elaborate the machinery of experimental production that was built around it, the more costly it would be to replace it with other species’ (Kohler, 1993, p. 309). This was most certainly the case with HIV reference strains as well.

Molecular Economies

In its role as reference strain, subtype B virus (and specifically the LAI-related strains) has come to serve as a proxy for the otherwise highly diverse HIV genome. As such, it provides a standardized ‘map’ of the virus in which a variable genetic code is momentarily stabilized, with particular genetic sequences affixed to a specific locations on the HIV genome.Footnote 9 This is an extremely useful scientific tool for navigating a microbe notorious for its powers of rapid mutation, and one that has allowed scientists to develop a common language for discussing and describing HIV genetics. Consequently, subtype B became the template upon which nearly all the laboratory and much of the clinical knowledge about HIV has been based. This includes everything from the molecular models scientists have developed to understand the virus's structure to the ARV drugs that have transformed the lives of patients able to access them. It also includes knowledge about drug resistance, which is defined according to mutations found in subtype B viruses.

I first learned about the significance of subtype in a 2004 discussion with Eileen Jacobs, an industry scientist who was working on developing a new test for HIV drug resistance for her employer, a multi-national molecular diagnostics company. The company was already a dominant manufacturer of viral load tests, which measure the levels of virus in a patient's blood, and was looking to expand into the market for genotype resistance assays – a test that determines whether a patient's virus has developed genetic mutations that render HIV drugs ineffective. Her interview was one of the first I conducted during my fieldwork, and my understanding of the science behind HIV genotype testing was still very limited at the time. Assuming that the technical challenge in genotyping had to do with the large number of ARVs in use, I asked her if it was difficult to develop a test that could detect resistance to numerous different drugs. Her answer surprised me. The challenge lay not in tailoring the test to the drugs, she told me, but in tailoring it to the virus. A key element of the genotype assay, Jacobs explained, is the use of polymerase chain reaction (PCR) technology to ‘amplify’, or make many copies of, the virus's genes. PCR works through the use of what are called ‘primers’. These are genetic sequences that attach to the beginning and end of a targeted region of a gene, marking the portion to be copied. These primers, Jacobs told me, were all initially designed based on a subtype B reference strain. As a result, they were sometimes less effective in attaching to and amplifying the genetic material of viruses of other subtypes. In the past, this had caused problems for the company's viral load test, which also uses PCR technology and was initially not very good at measuring the presence of non-B virus in patients’ blood. The company had successfully reworked the primers used in the viral load assay to make it work with multiple subtypes (clades) and was now trying to do the same for a genotype test. Jacobs explained:

E. Jacobs: So clade B is most common in this country. And so originally we knew most about that clade because that's where we could get our sequence information, that's where we could do the testing. So that's where the primers have been designed towards. So it's easy to amplify clade B. It's more difficult to amplify every clade.

So is it–are the primers that are used generally the ones that were developed for clade B?

E. Jacobs: Yes. Right. So all the commercial tests have been developed for clade B.

In order to develop a genotype test that could accurately assess viruses from different subtypes, Jacobs's company needed to provide its scientists with non-B viruses to work with. Because the corporation was not involved in any significant international collaborations, these samples had to be purchased from a biological supply company based in Miami. However, the fact that non-B viruses have achieved commodity status seems counterintuitive for a number of reasons. First, these viruses are hardly a scarcity, as they account for nearly 90 per cent of the world's infections. Moreover, the company's willingness to pay for such viruses seems ironic given the multinational pharmaceutical and biotech industries’ early reluctance towards making treatment accessible in the global South (Cooper et al, 2001). The growth in industry interest in re-tooling HIV diagnostic technologies for use in ‘resource-limited settings’ is likely an offshoot of the sudden availability of funding for global HIV care via PEPFAR and the Global Fund. It may also reflect the recent turn towards Africa among American AIDS scientists, many of whom began to take an interest in working on the continent as the epidemic stabilized in the United States following the discovery of effective drugs. In part, this shift reflected a humanitarian desire to use the expertise gained in fighting the US epidemic to assist the world region hardest hit by AIDS. In addition, conducting research in Africa also meshed well with scientific and career ambitions as the field of ‘global health’ rose in prominence and popularity in the first decade of the new millennium (Crane, 2007; Merson and Page, 2009). These US-based scientists rely on accurate molecular diagnostic technologies for their research and, unlike many clinics in sub-Saharan Africa, have resources available to pay for them.

The AIDS researchers I spoke with described the field's reliance on subtype B reference strains as problematic but essentially benign in motivation – a ‘historical fluke’ and an issue of convenience, not favoritism. Nonetheless, these and other HIV scientists are also very much aware of the limitations posed by relying so heavily on one strain of the virus, and emphasized the importance of conducting more research on non-B viruses. In addition, some also acknowledged that at a higher level, market forces were at play. James Briswell, a university-based expert in HIV drug resistance and non-B viruses, described the situation in terms reflecting the political economy of the global pharmaceutical market:

Well I think there are two reasons [why subtype B has been used]. I think one is just that it was the most convenient. People had all those subtype B samples available. And all the studies were done with B because it's in the United States and Europe. So that's more of a benign explanation. But there's no question that a lot of drug development is targeted to parts of the world where you know people are more likely to be able to pay for drugs. There could have been a lot of work done with tuberculosis and malaria, and many new drugs developed during the same time period. But they weren’t because it just wasn’t considered a high enough priority on the part of some of the pharmaceutical companies. You know, people who are looking at their bottom line.

So I think it's a combination. It's just that all the strains that people had in the labs were subtype B, and I think a lot of researchers just tend to work with the same strains. They’re not always aware of the variability. But I think there's also an element of companies targeting the strains that infect the people in the parts of the world that have the most money.

However, whether or not the initial choice of these subtype B viruses for laboratory research was a matter of convenience, they are now chosen by necessity. These lab-adapted strains have become established as the common referent or template for HIV research, and to change the template (by, for example, choosing a subtype C reference strain) would make it impossible to communicate with other laboratories or to compare data. It would make science ‘too chaotic’ in the words of Briswell, who told me that even researchers in the developing world – where subtype B is rare – speak in terms of subtype B:

[Y]ou almost need a common frame of reference to describe things. Even if it's arbitrary…just to facilitate communication. So it has nothing to do with favoring one subtype over another. It's really just a matter of convenience that a lot of the people in the field–you have to pick something as a reference. And in the future you know maybe things will change. But people are used to speaking in terms of subtype B. No matter where you go in the world, even if they have all subtype C, they’re used to speaking in terms of subtype B, from the literature, you know…. I don’t think that's detrimental in any way. Because if you had people using different reference strains it would just be too chaotic. (Emphasis added)

Thus, subtype B has become not only the molecular template upon which all HIV medications and diagnostic technologies are based, but also the lingua franca of AIDS research globally – even though other strains are much more prevalent. In this way subtype B operates almost like a colonial language, allowing communication across different groups and across geography, but also reflecting a very specific and unequal arrangement of power. The power embedded in a standardized ‘language’ is by no means limited to this instance, and can also be seen in examples as diverse as musical notation (Revuluri, 2007) or the ‘ASCII imperialism’ inherent within computer character set standards (Pargman and Palme, 2009). However, the case of HIV science is notable in the potential implications it carries for health around the globe.

Consequences: Clinical and Political

Clinical researchers and clinically applied anthropologists will rightly be concerned with the relevance of the story told above to patients seeking and undergoing HIV care, and may wonder to what extent this reliance on subtype B impacts ARV treatment and drug resistance in patients infected with ‘non-B’ viruses (that is, most patients in the world). This is not a question that can be answered definitively, as HIV scientists have only recently begun turning their attention towards subtype differences, and the field is rapidly evolving. Nonetheless, there are some important findings thus far (see Taylor et al, 2008 for a summary). For example, research has shown that some subtypes may be more aggressive than others, leading more rapidly to death when untreated. One 2006 study of subtypes A and D, the two most common clades in Uganda, found that those infected with subtype D died an average of 2 years earlier than those with subtype A virus (Laeyendecker et al, 2006). This finding was reported in Uganda's leading daily newspaper under the blunt headline, ‘HIV Type Determines How Fast You Die’ (New Vision, 2006). Other research has suggested that some subtypes may be more easily transmissible than others.

Historically, the clinical research arena in which viral subtype has drawn the most attention is in vaccine development. When they announced the ‘discovery’ of the AIDS virus in 1984, virologist Robert Gallo and HSS chief Margaret Heckler made the optimistic prediction that a vaccine would be available within 2–5 years. This prediction proved to be dramatically inaccurate, and 30 years into the epidemic the prospect of an effective vaccine is pessimistic. HIV clades have been a source of controversy in vaccine science because it is likely that preventative vaccines designed using one subtype may not work (or will work less well) against other subtypes.Footnote 10 This is because these vaccines are made using proteins from the exterior ‘envelope’ that encases the HIV virus, which is the portion of viral anatomy with the greatest genetic variation across clades. The earliest attempts at vaccines were indeed subtype-B specific, leading to politically and ethically volatile situations when they were brought to low-income, non-B countries for clinical trials. Ralph Ernst, a blood bank virologist, warned me about this after I posed several questions to him regarding subtype:

[Y]ou’ve got to be careful [about] the political aspect of the whole trend of your research. Because there was a big concern from Africans that ‘oh, you guys in the developed world, you’re going to make a vaccine against subtype B and it's not going to work for us. So you’re not thinking about us.’ You know, they have a point. Now people will make a vaccine against subtype B because that's where the money is. Sadly enough.

Uganda was one site where early vaccine research caused such a controversy. The country hosted the first vaccine trial in Africa in 1999, a Phase I study of a subtype B vaccine. Because Uganda's epidemic is comprised of primarily subtypes A and D, this caused some concern that Ugandans were being used as ‘guinea pigs’ for a vaccine that might not benefit them (a concern possibly augmented by earlier controversies over ethically questionable US-funded AZT research in the region (Warigi, 1997)).Footnote 11 However, this controversy was ultimately overshadowed by another: widespread fears (based on misinformation) that exposure to the vaccine would cause study participants to become infected with HIV (Kaleebu, 2005).Footnote 12 In part as a result of these early controversies, there is now a greater emphasis on the design of non-B vaccines and ‘multi-clade’ vaccines designed to work against multiple subtypes, and the number of non-B vaccines in development now greatly exceeds those based on subtype B (Kahn, 2003).

In addition, African countries have become increasingly involved in vaccine trials. In 2000, a group of African AIDS experts convened in Kenya and adopted ‘The Nairobi Declaration: An African Appeal for an AIDS Vaccine’. This document pledged support for increased African involvement in vaccine development and urged industrialized countries and international donor organizations to increase their financial and technical contributions towards vaccine research for Africa, ‘paying particular attention to the variability of HIV strains between different regions of the world’ (African Council of AIDS Service Organizations, 2000). Under the auspices of the World Health Organization, the group established the African AIDS Vaccine Programme to further promote and support the development of African vaccine research.

A few of the Ugandan scientists interviewed for my research were involved in vaccine studies, some for subtype B vaccines and some for multi-clade vaccines. Ronald Wetege, a leading Ugandan researcher responsible for some of the first studies of the epidemic in his country, was circumspect about the issue of subtype when I spoke with him in 2005. ‘The knowledge we have now is that really a vaccine is likely to be successful if it is tailored to the circulating subtype in the population as much as possible’, he told me. At the same time, he said, it was understandable that Western companies working on vaccine development would work on the subtype that prevailed in their countries. ‘One has to go to look at the other side’, he told me, ‘and say look, most of these companies which have invested billions and billions are operating in countries where there's only basically one subtype. So they are in a dilemma’. Interestingly, for Wetege even a subtype B vaccine study could be beneficial for Uganda, as it provided an opportunity to build research infrastructure and train local scientists:

I mean we did a study which was based on subtype B and I know we heard a lot of arguments like that – ‘Why is it a subtype which is not [here]?’ But we told them, ‘Look, there's something we can benefit. We can build infrastructure. We can train people. And as technology moves, we’ll get vaccines based on our subtypes’.

In this way, Wetege points out an interesting and important connection between research and development in Uganda. I refer here not to the ‘R&D’ of the pharmaceutical industry, but rather development as in ‘developing countries’ – ‘development’ that operates in the name of advancing the social and economic lot of poor nations, or, as James Ferguson writes, the ‘dominant problematic or interpretive grid through which the impoverished regions of the world are known to us’ (1994, p. xiii). What Wetege implies is that a vaccine study is not simply a vaccine study, but also a means by which to improve laboratory infrastructure, train researchers, and establish links with colleagues and funding bodies in the global North. In light of these tangible forms of development – or ‘capacity-building’ as it is often referred to in international health – the issue of subtype matching may be less important. Indeed, given the lack of clinical success thus far, the capacity-building aspects of research may actually be the primary benefit that vaccine studies have to offer at the current time.

In contrast to vaccine research, subtype differences have only more recently become a topic of concern within the science of HIV treatment and drug resistance. The growth of interest in the study of subtype differences is likely a result of the fact that large numbers of patients infected with non-B viral subtypes are now being exposed to ARVs for the first time through the Global Fund and PEPFAR programs. In addition, recent years have shown rising numbers of non-B infections in North America and Europe as the epidemic has become increasingly globalized (Thomson and Najera, 2001). These shifts have sparked concern within the scientific community over whether ARVs designed to target subtype B might be less effective against other viral subtypes (Spira et al, 2003; Atlas et al, 2005; Kinomoto et al, 2005; Fleury et al, 2006; Holguín et al, 2006). Fortunately the answer so far seems to be no, with the caveat that relevant data remain limited (Braitstein et al, 2006; Kantor, 2006; Taylor et al, 2008). As treatment is rolled out in low-income countries, people with non-B virus seem to be responding to and benefiting from existing ARV drugs just as well as people in the United States and Western Europe, a result that James Briswell described to me as ‘lucky’. This is good news for patients, but raises the question: should the universal efficacy of HIV drugs be a question of luck? Of course, luck has long been recognized to play an important role in scientific discovery, and in this way the story of HIV drug development is nothing out of the ordinary. Yet, the question of whether the use of more diverse research materials might lead to different kinds of ‘lucky’ findings seems relevant.

Furthermore, as the molecular targets of HIV drugs become more diverse, the question of whether new drugs will work equally well across subtype is an important one. The first decade of effective HIV treatment relied on drugs that targeted two specific areas of the viral genome, interfering with the virus’s ability to replicate itself. Recently, new drugs have emerged that target different areas of the virus, such as an injectable medication called Fuzeon (enfurvitide), which helps block the entry of HIV into human cells, and Isentress (raltegravir), a drug that interferes with the virus's ability to integrate its genetic material into the DNA of host cells. These drugs represent two entirely new classes of ARVs. If the areas of the viral genome targeted by these drugs vary highly between subtypes, it is possible that newer classes of drugs might work better against some subtypes than others. This question came up at a 2004 conference presentation given by Francoise Brun-Vezinet, a prominent French AIDS scientist. Brun-Vezinet delivered a well-attended talk on drug resistance in non-B viruses at the 2004 Interscience Conference on Antimicrobial Agents and Chemotherapies, a major infectious disease conference held annually in the United States. During the question-and-answer period, a member of the audience of physicians and scientists asked her whether anything was known about the efficacy of the newer classes of drugs in treating non-B infections. She told the audience that although there had been initial skepticism that Fuzeon worked against non-B viruses, a recent study had proved that it was indeed effective (see Holguín et al, 2007). But, she added, for the other new drugs (referring to drug classes under development at the time) ‘I’m afraid that subtype will affect it very much’ (Brun-Vézinet, 2004). Little is known about the performance of newer drugs against non-B viruses, as they were designed and approved based on clinical trials in patients who have spent years on HIV treatment and developed resistance to numerous drugs – a category that currently describes very few individuals in the lower-income countries where non-B viruses predominate. Furthermore, the small amount that is known is based on in vitro studies conducted on non-B viruses in the laboratory, which may or may not represent the clinical reality of how these drugs behave in patients (Taylor et al, 2008).

Lastly, I return to the starting point of this article: the science of HIV drug resistance. The question of drug resistance is directly related to the question of drug efficacy, in that ARV drugs will be ineffective (or less effective) against viruses that have mutations causing drug resistance. To date, the vast majority of research on ARV drug resistance has been conducted on patient populations infected with subtype B virus. This reflects not only the (now shifting) geographical bias of AIDS research, which initially focused much more on the epidemic in North America and Western Europe, but also a technological bias in that the tools used to test for drug resistance were designed to do so in subtype B.

Researchers and clinicians use two different methods to assess whether a virus has become resistant to ARVs: genotyping and phenotyping. Genotype testing, the cheaper and most commonly used option, works by comparing the genetic code of a patient's virus to that of a drug-susceptible ‘reference’ virus. In this context, the reference virus serves as a genetic map of what the virus ‘should’ look like if it has not mutated to become drug resistant. However, because the canonical reference strains are all subtype B viruses, scientists are uncertain about whether they serve as an accurate yardstick by which to gauge drug resistance in other viral subtypes. For example, a genetic configuration that is defined as a resistance ‘mutation’ in a clade B virus might actually be normal or ‘natural’ in another clade (Jülg and Goebel, 2005). In addition, because resistance mutations have been defined based on genetic changes observed in subtype B viruses, much less is known about what resistance might look like in viruses with different (that is, non-B) baseline genetics (Martínez-Cajas et al, 2008).

Scientifically, these are open questions with which a number of the scientists interviewed for this project were actively grappling. The research is ongoing, and has evolved in the period between my fieldwork and this writing. Thus far, there is no evidence that the interpretation of any given resistance mutation differs across subtype: in other words, a mutation that commonly causes drug resistance in subtype B virus will have the same result in a subtype C virus. Where differences do exist are in the patterns of mutation across subtype, meaning that some genetic changes are more common in some subtypes than in others, and that different viral subtypes may take different genetic ‘pathways’ to drug resistance. The clinical implications of these differences for patients, if any, are still unknown (Gertti, 2006).

Concerns about subtype have also arisen in relation to the phenotype assay, which is a costlier but more direct test for drug resistance than genotyping. Rather than looking for genetic mutations, phenotyping involves exposing a patient's virus to different drugs in the laboratory in order to see which drugs are able to stop the virus from replicating. When a virus continues to reproduce itself in the presence of an ARV, it is considered resistant to that drug. Importantly, phenotype tests do not use a patient's virus in ‘whole’ form, but rather graft the relevant genes from a patient's virus into a standardized viral vector, which is then exposed to a series of ARVs in vitro. Because these standardized vectors are derived from the classic subtype B reference strains, HXB2 and NL4-3, some researchers have expressed concern that they might limit the accuracy of phenotype testing in assessing resistance in non-B viruses (Gertti, 2006; Martínez-Cajas et al, 2008)

Overall, although some studies have suggested that certain non-B viruses may become resistant to certain ARVs more quickly or easily than HIV subtype B (Holguín et al, 2006; Martínez-Cajas et al, 2008), there is currently no smoking gun showing radical differences in drug resistance across subtype. However, differences are ‘increasingly emerging’ as research evolves, and further studies of non-B subtypes are considered an important priority within the field (Kantor, 2006; Holguín et al, 2006; Martínez-Cajas et al, 2008). It is only very recently that these questions of difference have become a topic of significant clinical inquiry. The reasons for this, as I hope this article has made clear, rest in the interrelationship between the molecular politics of HIV research and the political economy of ARV treatment access. Early on in the epidemic, viruses reflecting the genetics of HIV in the United States and Europe were integrated into laboratory research, and rapidly became the lingua franca of molecular virology. For many years, these reference strains remained an unmarked category – a universal template for HIV genetics. The choice of these strains was both born out of and served to perpetuate the near-exclusive focus of HIV treatment and drug resistance research on clade B virus. Furthermore, the lack of access to ARVs outside wealthy subtype B countries made research beyond these parameters difficult and unlikely, as it meant that there were very few patients with non-B virus undergoing ARV treatment, and thus very few in which the effects of treatment – including drug resistance – could be effectively studied.

The recent push to roll out ARVs in Africa and other low-income regions of the world has changed this picture. Confronted with a global epidemic comprised mainly of subtype C and other non-B viruses, scientists began to confront the fact that their laboratory reference strains were not unmarked or universal but were, in contrast, very specific viruses with particular histories and genetic cartographies. The result has been a growing awareness of and interest in differences in treatment and drug resistance across geographies and subtypes. However, owing to the newness of these inquiries, a great deal remains unknown. For example, comparative studies often tend to lump all non-B viruses together, making it difficult to determine how, for example, the subtype C viruses that predominate in South Africa might act differently than the subtype D viruses prevalent in Uganda (Holguín et al, 2006). ‘The data’, Ralph Ernst reminded me in 2005, is still ‘so slim. People don’t do serious drug resistance studies in non-B countries or they’re done in very few of them’.

It seems important to juxtapose the ‘slimness’ of these data with the broadly stated conjectures about HIV drug resistance in Africa described in the introduction of this article. Those conjectures were made in the absence of any significant data on drug resistance in Africa, yet they played prominently in policy debates over whether or not treatment should be expanded there. And, significantly, these debates continued to haunt the field even after the political tides turned in favor of treatment access, as James Briswell told me in 2005:

J. Briswell: I think a lot of people who are involved in this drug resistance research in Africa and Asia, like I am, are concerned that [the research] will be taken out of context, and people would exaggerate this fear [of resistance] and use it as a reason not to give therapy.

Is that a concern of yours, or in the field?

J. Briswell: Yeah, that's a major concern among experts in drug resistance – that fears of resistance not be blown out of proportion, so that they’re not looked at as a reason not to give therapy.

Scientists are in agreement that the data on non-B drug resistance are scant, and needs to be pursued in greater depth. There are an increasing number of researchers committed to this project. Yet, as Briswell describes, they conduct their inquiries with the fear that the data that they are beginning to collect may be interpreted as a reason not to expand treatment in poor countries – an expansion that they greatly support. This fear seems to indicate that despite the major shift in international will towards support of ARV treatment in Africa, fear of drug resistance has retained a powerful political valence.

In addition, it is important to note that the research into non-B drug resistance is still centered in wealthy industrialized countries, where the technology exists to conduct the molecular analyses involved in drug resistance research. Uganda, despite its prominence in other areas of AIDS research such as epidemiology and prevention, is ‘rather thin’ in the field of biological research, as one Ugandan epidemiologist put it to me. Training more Ugandan molecular biologists, he thought, would lead to more drug resistance research on Ugandan HIV clades:

Molecular biology is rather thin. I was talking to somebody who was trying to pursue a career in the laboratory sciences and I said, ‘Just go for molecular biology.’ [In] the epidemiological sciences, I think we have trained a sufficient number of epidemiologists. Behavioral scientists are being trained. Laboratory scientists are being trained. But molecular biologists and virologists, immunologists, those are in short demand. So, the only way we could develop, [that] we could study drug resistance working on our clades is by developing the capacity ourselves. I think that will be easier.

Like Ronald Wetege, this scientist makes a link between research and capacity-building. If Uganda had greater local research capacity – particularly within fields like molecular biology, he argues – more research on clades relevant to Uganda would be done. This seems a fair prediction, given that biologists often base their work on clinical samples available from local patients. It also raises the interesting question of how molecular knowledge about the virus might have evolved very differently had the global centers of scientific power laid elsewhere.

Conclusion

This article began with a discussion of cartography, in which I argued that maps are socially constructed representations of a territory based on the specific inclusion and exclusion of different types of information. Over time these representations become naturalized, and their social production and historical specificity become obscured. Here I have argued that the genetic mapping of HIV has followed a similar cartographic trajectory: the drugs built to fight the virus and the tools built to study it are based on a very partial and contingent map of the virus, yet, this map is rarely questioned or historicized.

Thus, not only is bioscientific knowledge about HIV in Africa limited, but most of the knowledge that does exist has been gleaned using tools predicated upon molecular maps of an HIV strain rarely found in Africa. It is within these tools that the geographic and economic inequalities of the global epidemic have become embedded at the molecular level, in technologies that always refer back to the ‘West’ – Western viruses, Western research capacity and Western markets. Whether or not these tools are able to accurately monitor non-B (‘African’) viruses in clinical terms is a question that scientists are actively grappling with. Ultimately, my argument is that regardless of the scientific outcomes, it is politically urgent to juxtapose the willingness with which experts have made very consequential knowledge claims about HIV in Africa – and particularly HIV drug resistance – with the near total absence of empirical research on the topic. In this juxtaposition, we can see how the simultaneous deployment and erasure of African bodies worked to support the claims of treatment skeptics. On the one hand, skeptics’ fears regarding ‘antiretroviral anarchy’ and ‘doomsday strains’ of drug-resistant HIV gained purchase from the long tradition in Euro-American discourse of framing Africa and African bodies as diseased, disorganized, and ‘all that is incomplete, mutilated, and unfinished’ (Mbembe, 2001, p. 1; McNeill, 2003). On the other hand, the absence of African bodies from the scientific study of HIV treatment and drug resistance created a void of information into which these discourses could easily step, unchallenged. The positioning of African HIV subtypes as the exceptions – the different or deviant viruses – and the Euro-American subtype as the universal standard fits what anthropologist Stacy Pigg has called ‘the definition of marginality: to be positioned as the exception, the deviate, the parochial, or the merely local in the face of the universal’ (Pigg, 2001, p. 510). For this reason, a key question for social and clinical scientists is not only, ‘What do we know about HIV in Africa?’, but ‘How do we know what we know?’