AZT and the Politics of Interpretation (1989–1990)
Signs of Rapprochement
By early 1989, it began to appear that AIDS treatment activists had won a partial convert to the cause: Anthony Fauci himself, the head of NIAID and the government's AIDS research program. To a greater extent than his counterparts at the FDA, Fauci had cultivated good relations with treatment activists, opening up channels of communication with people like Martin Delaney of Project Inform and Mark Harrington and Jim Eigo of ACT UP/New York's Treatment & Data Committee. "In the beginning, those people had a blanket disgust with us," Fauci told the Washington Post in 1989: "And it was mutual. Scientists said all trials should be restricted, rigid and slow. The gay groups said we were killing people with red tape. When the smoke cleared we realized that much of their criticism was absolutely valid." When activists complained about the FDA's slowness in approving a drug called ganciclovir that appeared to prevent blindness in people with AIDS suffering from a viral infection of the retina called CMV retinitis, Fauci went to bat for them and helped to put pressure on the FDA. In an article in the journal Academic Medicine published early in 1989, Fauci defended established scientific methods but also acknowledged some of the points that activists had been making. "Clearly, there is a need for greater access to clinical trials of
investigational drugs by a broader spectrum of the infected population," wrote Fauci.
The turning point came at the Fifth International Conference on AIDS, held in Montreal in June. Activists took center stage at the conference—disrupting the opening ceremony, staging protests against pharmaceutical companies that had been identified as profit-hungry, and presenting formal poster sessions with titles like "AIDS Drugs and the Politics of Biomedicine" and "Drug Regulation Gone Wrong: The Saga of Ganciclovir." Behind the scenes, Larry Kramer of ACT UP/New York met with Fauci—a man he had called "an incompetent idiot" and worse in print—and solidified Fauci's support for "parallel track," a new concept that had been developed by Jim Eigo and other New York activists.
The parallel track program was, in effect, the solution to the sort of dilemma that Martin Delaney had described at the meeting of the Infectious Diseases Society a few months before (a meeting that Fauci had attended): when researchers coerced people into trials by giving them no other means of access to experimental treatments, participants likely wouldn't comply with the study protocols, and as a result the data would be unreliable. To avoid such difficulties, as Fauci told the press in June, a parallel track program "would provide promising drugs to some people with AIDS at the same time as the drugs are undergoing rigorous [Phase II] clinical trials." Patients would be eligible to receive free drugs in the parallel track program "if they were unwilling or unable to participate in the normal clinical trial"—for example, because they failed to qualify for the study or because they lived too far from the study centers.
In essence, Fauci adopted the activist line on this issue, as his comments quoted in the front-page New York Times article made evident: "'Previously, there was a great concern that if we did this, then no one would be in the clinical trial,' Dr. Fauci said. But he added that he has changed his mind and now thinks it is unnecessary 'to hold a gun to their heads' to induce people to join clinical trials." Fauci explained that NIAID could pursue parallel track under its own authority without the need for any new, enabling legislation; indeed, he was prepared to start the program soon with the drug ddI, pending support from the manufacturer, Bristol-Myers. Unlike the FDA's more limited compassionate use policies, parallel track promised to provide large numbers of AIDS patients with easy access to drugs that had passed only Phase I trials. "I came out and stuck myself out on a limb … and
everybody here in Washington fell off their seats and said 'What is he doing?'" Fauci later recalled. But "I thought it was the right thing to do, and I figured the only way we could get it done was to just say that I was in favor of it and apologize later. And as it turned out, I didn't have to apologize, because everybody then jumped on the bandwagon…."
Optimism among activists about their successes in changing federal policies was matched by more upbeat attitudes on the part of researchers and clinicians about therapeutic prospects. With AZT, with prophylaxis against PCP, and with better treatment strategies against other opportunistic infections, AIDS patients were living longer. Other antivirals like ddI and ddC were on the horizon. By combining or alternating the use of these and other drugs (as was often done in cancer treatments and for some bacterial infections), doctors might be able to keep the virus in check while preventing the onset of drug resistance. In Montreal, a new conventional wisdom emerged: HIV infection might soon become a "chronic manageable illness," not fully curable but something that a person might live with.
AZT: "The Time Has Come"
This notion that HIV disease could soon become a chronic manageable illness received a sharp boost in August 1989 with the release of the latest news about AZT. "The drug AZT can delay the onset of AIDS in people who are infected with the virus but have no symptoms," began a front-page New York Times article. ACTG 019, as the ACTG-sponsored study of AZT use in asymptomatic HIV-infected patients was dubbed, had just been halted upon the discovery by the Data and Safety Monitoring Board that the participants on placebo had been twice as likely to develop AIDS-related symptoms as those taking AZT. Just a few weeks earlier, ACTG 016 had also been ended, with the conclusion that AZT helped prevent HIV-infected people with mild symptoms from progressing to full AIDS. Out of 713 people in that study, 36 taking a placebo had progressed to AIDS, versus only 14 of those on AZT. Dr. Jerome Groopman had called it "the first clear proof that early intervention makes a difference." Now, with the results from 019, the verdict seemed to be in. "Today we are witnessing a turning point in the battle to change AIDS from a fatal disease to a treatable one," said Dr. Louis Sullivan, the Bush administration's secretary of health and human services. It was
indeed a turning point, though 019 would prove to be one of the most-debated studies in the history of AIDS clinical trials.
The 019 study finally made its way into the New England Journal the following April, with Paul Volberding of San Francisco listed as the first author. About 1,500 patients at thirty-two trial sites around the country had participated in the study for an average time of fifty-five weeks. Though none of them had AIDS symptoms upon entry, all of them were immune deficient, with T cell counts below five hundred per cubic millimeter of blood. One third had received fifteen hundred milligrams a day of AZT—by the time the study ended, this was generally believed to be an unnecessarily high dose—and 14 patients in that group had developed an AIDS-defining illness. Another third had received five hundred milligrams a day, with 11 developing AIDS. And of the final third, the patients on placebo, 33 had developed AIDS. AZT did not stop patients, as an aggregate, from developing AIDS, but it appeared to slow the process down, at least over a fifty-five-week period. Furthermore, the subjects receiving AZT (either high or low doses) showed, on average, statistically significant increases in their T-cell counts as well as decreases in their "p24 antigen levels" (p24 is one of the core proteins that make up HIV, so a measure of p24 is a crude measure of how much virus is present in the bloodstream—also called the "viral load"). These markers provided additional evidence that AZT was boosting immune functioning while curtailing viral replication.
What about the rumors of "cheating" by patients seeking to avoid the placebo? Did this skew the study and make the results untrustworthy? "There were lots of stories circulated in the press about people sharing drugs, analyzing drugs," recalled Volberding. But lab tests to detect the presence of AZT in the blood of the placebo patients suggested that drug sharing in fact "was an incredibly small problem." In any event, noncompliance actually strengthened the results of the study, the authors of the study results argued, since it "would tend to give results that underestimate the true effect of zidovudine" (the generic pharmaceutical name for AZT). Noncompliance effectively blurred the differences between the treatment arm and the placebo arm, so the demonstration of a statistically significant difference became all the more impressive.
Volberding and his coauthors included a series of important disclaimers. Since the trial had been ended early, there was still no definitive evidence about the long-term benefits or safety of AZT: "Thus,
it is possible that the eventual risks of disease progression in the three treatment groups could become similar after a longer time period." This might prove true particularly if AZT were to lose its effectiveness after extended use, as the virus mutated to resist the drug. (Such "resistance-conferring mutations" had already been observed in the test tube.) Finally, said the authors, "it is possible that even if zidovudine persistently delays the onset of AIDS, it may not have an ultimate effect on survival." On first glance, this was a surprising and paradoxical notion: how could a drug delay the progression to AIDS without extending the patient's life span? Volberding and his coauthors didn't elaborate, but the implication was that once progression did occur, the downhill course might then be rapid. Such cases were not unknown in cancer treatment research, where a drug might shrink the size of a tumor yet not confer any survival benefit upon the patient.
The accompanying editorial, by Dr. Gerald Friedland, acknowledged these various uncertainties and also spoke of the psychological implications of putting asymptomatic people on AZT: "The decision to start therapy … converts an apparently healthy person into a patient probably committed to lifelong treatment." The initiation of therapy in effect transformed the infected person's identity; this was not an act to be undertaken lightly. Nor was there any hard evidence from 019 about the best time to take such action: Was it when the patient's T-cell count dropped to seven hundred per cubic millimeter? Five hundred? Three hundred? But despite these words of caution, the overall tone of the editorial was upbeat, and the title conveyed the basic message: "Early Treatment for HIV: The Time Has Come." Friedland wrote: "The results of this study strongly support a recommendation to institute zidovudine therapy at a dose of five hundred mg per day [the lower, less toxic dose used in the study] in persons with asymptomatic HIV infection and CD4+ cell counts [T-cell counts] below five hundred per cubic millimeter."
Treatment recommendations in an editorial in the New England Journal carry considerable authority, but in a certain sense these were moot. Everyone had known about the study for eight months. Moreover, in early 1990 the FDA's new Antiviral Advisory Committee (one of more than forty standing committees of experts charged with giving independent scientific advice to the agency) had already recommended changing the labeling for AZT to include all HIV-positive patients with T-cell counts below five hundred. Despite concerns about the development of drug resistance, the vote this time was unanimous.
The FDA's Ellen Cooper had reservations about the recommendation—after all, the vast majority of people in each arm of the study had done well over the course of fifty-five weeks; the percentage of "events" (progression to AIDS) was small. Did it really make sense to begin mass administration of AZT to hundreds of thousands of relatively healthy people in order to prevent a small fraction of them from progressing to AIDS each year? As Cooper would later acknowledge, this was ultimately less a scientific question than "a matter of judgment and generalization." In the end, the FDA adopted the advisory committee's recommendations on March 2, and the NIH issued new guidelines for physicians concerning when to prescribe AZT. Burroughs Wellcome's potential market thereby increased by a factor of more than ten: In the United States alone, there were fifty thousand reported living AIDS patients, but an estimated six hundred thousand HIV-infected people with no AIDS-defining illnesses and with T-cell counts below five hundred. In England, the stock value of parent company Wellcome PLC increased by 1.4 billion pounds.
In practice, the findings of the 019 study, announced only in preliminary form to the media by NIAID, had provided the basis for regulatory policy making before the full results had even appeared in a peer-reviewed journal, where they could be scrutinized by other experts. Project Inform praised NIAID for its "efforts to release important news without waiting the extra 6 months or more needed for acceptance and printing of a completed journal article." For some time, activist groups had been insisting on the rights of people with AIDS to have access to medical data, and they had criticized the tendencies of some medical journals to monopolize control of information by threatening not to publish studies that had been disclosed to the press. But others, both within the AIDS movement and the scientific establishment, would condemn the practice of "science by press release." As far as Fauci was concerned, however, he really didn't have much of a choice, given the widespread interest in the 019 trial and the extensive monitoring of science by the AIDS movement: "Very, very quickly, everyone would have known the study was terminated. … It would have been unacceptable to everyone to make them guess why."
Poison? Or Just Mediocre?
In fact, neither NIAID's press conference, nor the FDA's approval, nor even the article in the New England Journal and the
accompanying editorial succeeded in bringing closure to debates about AZT. This was true both within the AIDS movement and among the scientific establishment. In gay communities, controversy about AZT had been bubbling away ever since the original licensing of the drug in early 1987 for use by people with full-blown AIDS. The New York Native , the gay newspaper most closely associated with the promotion of heretical views, had been calling AZT a "poison." Its administration "was an act of genocide on the scale of the kinds of 'medical experiments' conducted in Nazi Germany," the newspaper argued. "AZT's alleged benefits are not backed up by hard data, and are not sufficient to compensate for the drug's known toxicities. … Do not take, prescribe, or recommend AZT," read the Native 's cover in June 1987; an article by John Lauritsen, the HIV dissenter and ally of Peter Duesberg's, accompanied it.
Since Lauritsen did not believe that HIV caused AIDS (see part one), it followed that he would not support the use of an anti-HIV agent as a treatment for the syndrome—especially one that was a DNA chain terminator with potentially serious effects on healthy body cells. But Lauritsen, like Sonnabend, Callen, and other HIV dissenters, also argued that the Phase II AZT study had been methodologically flawed in ways that cast doubt on its substantive conclusions. "In practice, the study became unblinded almost immediately," wrote Lauritsen, recapitulating the various rumors about problems in conducting that study. Though, as with 019, it could be maintained that any noncompliance by participants actually strengthened the results, Lauritsen turned the argument around by proposing that the failure to maintain perfect double-blind conditions had pernicious effects on the research process. Since the research staff knew from lab test results which patients were receiving AZT and which were taking the placebo, he argued, they may have provided better overall care to the AZT patients, whether "unconsciously or deliberately"; this difference in care, rather than the administration of AZT, might explain the difference in progression to AIDS.
Lauritsen's was a textbook case of how to deconstruct a scientific study. "Scientists constantly face uncertainty," Susan Leigh Star has emphasized. "Their experimental materials are recalcitrant; their organizational politics precarious; they may not know whether a given technique was correctly applied or interpreted; they must often rely on observations made in haste or by unskilled assistants." Yet precisely because contingency, confusion, misgivings, and indecision tend to be "written out" of scientists' published work as part of their normal
persuasive practice, nonscientists often have mistaken notions about the degree of certainty behind the knowledge that science generates. As Harry Collins has concluded, "There is a relationship between the extent to which science is seen as a producer of certainty and distance from the research front." Thus one strategy for undercutting the credibility of scientific claims is to bring the audience in for a closer look, so as to recapture the contingency and messiness: "Irrespective of whether the critic describes 'truly disqualifying' acts of clumsiness or incompetence, or irrelevant details, the mere act of describing an experiment as a piece of ordinary life reduces its power to convince."
Although dissenters in the AIDS causation controversy universally rejected AZT, criticisms of the drug were not limited to this group. Particularly in New York City—which Martin Delaney characterized as "almost unique in the nation in its anti-AZT hysteria" —there were numerous pockets of suspicion of AZT. Throughout the epidemic, the New York gay community had been—depending on one's perspective—either more radical in its skepticism toward authority or more possessed of a debilitating paranoia than its counterparts in San Francisco and elsewhere around the country. From early on, New Yorkers had seemed to show more interest in conspiratorial theories about the origins of AIDS. With the advent of the HIV antibody test in 1985, even the more mainstream organizations like the Gay Men's Health Crisis had advised against taking the test, on the grounds that those testing positive might be rounded up and quarantined or at least discriminated against; by contrast, San Francisco organizations like the AIDS Foundation had taken a more neutral approach, while Project Inform had advocated in favor of testing as the necessary first step in a program of early intervention. Randy Shilts has suggested that such political and attitudinal differences reflected the relative degrees of comfort of the two gay communities as they evolved in the years before the epidemic, with New Yorkers more "closeted" and concerned about threats to their social privilege and San Franciscans more out-of-the-closet, secure, and influential vis-à-vis their city government.
Whatever the structural or psychosocial roots of these dispositions, they surfaced as well in debates over AZT. Though the "AZT is poison" argument was always a minority view among treatment activists and the communities at large, it was less of a fringe perspective on the East Coast than it was on the West Coast. Indeed, a 1989 gay health conference at Columbia University in New York erupted into a debate between Delaney and Sonnabend over AZT. Interestingly, Delaney agreed that there were "some problems" with the original AZT study
but also "[took] some responsibility for those problems": "We as a community screamed and hollered to move that drug through the system and study it as fast as humanly possible." Counseling pragmatism over a methodological purism, Delaney told the audience that to obsess about any deficiencies in that study "is a little like having study groups on the Council of Trent."
In its treatment newsletter, PI Perspectives , Project Inform expanded on the view that it was willing to accept a certain degree of uncertainty about drugs as a trade-off for more rapid approval: "Patients and their advocates, including Project Inform, pushed the regulatory and research system hard to make AZT available as soon as possible. We should not be surprised that the drug came into common use while our understanding of it was still very crude." The irony is that, when it was first approved, "there was widespread belief that AZT would be quickly replaced by other drugs with similar benefits and fewer side-effects." That hope had proven to be misplaced, so now patients and advocacy groups found themselves having to make the best of a not so great situation, forced to depend on a mediocre drug. But in response to this predicament, Project Inform advocated judicious risk taking over what it saw as denial and defeatism.
Having committed itself to an interventionist therapeutic strategy, Project Inform in a sense depended on AZT, the only approved anti-HIV drug and the only such drug with widespread public credibility. AZT was, at the moment, an "obligatory passage point": Project Inform needed the drug to advance the group's mission. With the news about ACTG 019, Project Inform pushed its critique of the AZT dissidents: "This latest information should (but won't) sound the death knell for the views of those who have bitterly opposed AZT for the last 3 years." The only real question now, Project Inform's newsletter proposed, was whether every HIV-positive person shouldn't begin immediate AZT use, even if his or her T-cell count was higher than five hundred per cubic millimeter. "At the very least, it is one rationally supportable course of action, perhaps more so than the opposite view."
Two Committees, Two Conclusions
Critiques by the AZT dissidents would continue after the results of the 019 study were announced. But perhaps more noteworthy (though in practice little noted at the time outside of the scientific press) was the lack of consensus in mainstream science about the merits of early intervention with the drug. All mainstream authorities
agreed that AZT should be prescribed to patients with AIDS. But should it be recommended for use by every HIV-infected person with fewer than five hundred T cells (let alone those with more than five hundred T cells)? Here the gap in perception was not between the two coasts of the United States, but rather between the opposite shores of the Atlantic Ocean. While the New England Journal of Medicine had promoted the message that the "time had come" for early intervention, Lancet , the most important British medical journal, suggested that "clinicians should change their prescribing habits with caution." The 019 study, according to Lancet , had simply shown that AZT could prevent a small number of people from progressing to AIDS over a short time period. Could these results be extrapolated to the majority of patients, who would progress more slowly in any case? Did the drug really halve the progression rate, or did it simply delay progression for a period of months until the onset of drug resistance?
According to Anthony Pinching, a British immunologist and AIDS clinical trials investigator, the difference in judgment "was exacerbated by the 8-month delay between the announcement of some preliminary results and publication [of the study]"—a delay that Europeans, including the British, had found irritating. During this time, while the rest of the world was waiting to see the data, the NIH and the FDA had effectively given their blessing to AZT use in asymptomatics. In the United States, hopes had coalesced into certainty; elsewhere, scientists and patients were in limbo. At root, said Pinching in an editorial about "knowledge and uncertainty" in AZT use, the problem was that U.S. authorities had "extrapolated" beyond the limits of the actual data—even if one considered the full data from the published study. They had been led beyond the limits of legitimate scientific deduction by "the widespread desire to see progress achieved and the wish to be seen to have made such progress." To illustrate his point about extrapolation, Pinching listed fifteen questions about AZT use in asymptomatics that still remained to be answered, even after the conclusion of 019.
The issue of what remained to be answered was very much on the minds of researchers throughout Europe, for some of them were smack in the middle of their own study when the media began trumpeting the news about the termination of 019. A consortium of researchers in Britain, Ireland, and France was conducting a trial dubbed "Concorde" that was substantially similar to 019. As in the United States, the European researchers had dealt with the ethics of placebo-controlled
trials; as in the United States, the Europeans had had to justify to community groups why such trials were necessary in studying AZT use in asymptomatics. But now—even after the results from 019 had been reported—the Concorde researchers were intent on continuing the study. As far as they were concerned, a state of equipoise still held: "The results we have seen," said Jean-Pierre Aboulker, the head of the study on the French side, "do not allow us to give a strict recommendation to give AZT." In deference to those who trusted the results of 019, they would modify the protocol to allow participants to begin "open-label" use of AZT, if they chose not to run the risk of being given the placebo. This change would introduce methodological complications into the interpretation of the data. But it would satisfy ethical objections while allowing Concorde to continue.
"Two committees looking at one set of data have come to radically different conclusions about the anti-AIDS drug AZT," was how a reporter for Science characterized the transatlantic dispute. In practice, everyone agreed on which questions they would like to have answered—questions about long-term risks and benefits, about the development of resistance, about whether early use would squander the limited efficacy of the drug against the virus. The different actions taken had to do with different judgments about the implications of what was known and what wasn't. How and when should a particular balance of certainty and uncertainty about a drug inform clinical practice? And what is the feasibility of continuing a placebo-controlled trial, once a specific (but not conclusive) benefit had been shown?
"I think there is generally a greater skepticism in the U.K. about the value of treatments than in the U.S.," commented biostatistician Susan Ellenberg, reflecting a widespread perception in the U.S. biomedical community. "I think there is much more reluctance [in Britain] to treat unless it is absolutely necessary." Such cultural differences were compounded by differences in the sheer magnitude of the epidemic in the two countries, not to mention the more vociferous character of AIDS treatment activism in the United States. Continuing the 019 trial "was not a conceivable possibility when the data were first seen … by those of us who were on the executive committee of the ACTG," researcher Martin Hirsch later recalled. "There was no choice but to stop the study given what the results were. … If we had withheld that kind of information we would have been strung up on trees. And if we had given out the information and said we were going to continue the placebo control trial anyway, we would have been strung up even
quicker." Similarly, the principal investigator for Concorde in Britain, Dr. Ian Weller, suggested to Science that the large numbers of AIDS patients in the United States had generated greater pressures there for the immediate translation of research into results. And once the formal guidelines had been issued, all the remaining uncertainties, though never denied, had in a practical sense been bracketed, largely to be ignored by the media, most practicing physicians, and most patients. Concorde, by contrast, chose to put uncertainty in the foreground. With the goal of obtaining definitive answers about early intervention with AZT, the Concorde trial would keep running all the way to 1993. Ironically, the results of Concorde, when finally available, would provoke more controversy than any AIDS antiviral study ever.