AZT: More "Pieces of the Elephant"
Later that month, the NIH would ratify uncertainty as the new wisdom, dressing it in the language of bureaucratese. After a three-day meeting, an expert advisory panel offered up recommendations: Asymptomatic HIV-infected patients with three hundred to five hundred T cells should begin taking AZT. But not taking AZT, and continuing to monitor the immune system, was "an equally valid option." Patients in stable condition with more than three hundred T cells who were already taking AZT should stay on it, the panel recommended. However, the chair, Dr. Merle Sande of the University of California at San Francisco, told the New York Times that "stopping AZT for such people would be a medically sound and logical conclusion from the recommendations." In short, the NIH recommended that patients do X , but if they wanted to do not-X , that was "equally valid."
Sande blamed the current predicament on the competing conceptions of the purpose of clinical trials. He complained to Lawrence Altman that the various trials of the nucleoside analogues had been designed with FDA regulatory questions in mind, rather than the clinical
questions that doctors wanted answered: "We were in a hurry four to six years ago to get drugs approved, so the studies were designed in that way and not necessarily to answer the clinical medical questions that we face in making decisions today. … We are a prisoner of those studies." Altman's article suggested that much of the blame also lay with the activists for having pressed so hard to get drugs approved quickly. In fact, though, it had been activists who had insisted all along that clinical trials were asking the wrong questions—that they should be oriented to the real-life issues confronting patients and doctors rather than the thumbs-up/thumbs-down logic of FDA approval.
To complicate the picture further, barely a month after the new guidelines were issued the New England Journal published the results of the European-Australian Collaborative Group study, a three-year, double-blind, placebo-controlled study of AZT use in 993 asymptomatic patients with CD4 counts above four hundred per cubic millimeter upon entry. The study found that early administration of AZT did delay the onset of symptoms, supporting the view "that most patients with HIV infection should be treated," in the words of an accompanying editorial. The study did not address the issue of survival benefit. Merle Sande commented to the New York Times: "I think all the studies are showing the same thing: that there is a little benefit from AZT for a short period of time." Douglas Richman stressed the differences in endpoints and follow-up periods compared to those used in Concorde and said: "This study looks at another piece of the elephant. … But we can't say if five years from now the two curves of the two different patient groups won't come together. It's part of the complexity of a clinical investigation—but that's life in the big city."
"It's really resulted in incredible frustration," acknowledged Paul Volberding, the UCSF researcher, in 1994: "No one feels they know what to do. No one trusts the guidelines, even though the guidelines have been put together as best as people could do. People want an answer and there isn't one." Volberding's own long-term follow-up study of the participants in ACTG 019 (the study that first had suggested benefit from early use of AZT in asymptomatics) was, in fact, consistent with the results of Concorde. First presented in Berlin and published in JAMA in 1994, the follow-up concluded that the benefits of AZT were time-limited and that early administration conferred no additional survival benefit over late administration.
The overall credibility of AZT received a boost in early 1994, when NIAID announced the results of a trial called ACTG 076, which had
studied whether administration of the drug to pregnant women could prevent transmission of the virus to their fetuses. The results were striking: more than a quarter of the babies born to women who received a placebo became HIV positive, but only eight percent of the babies whose mothers received AZT became infected with the virus. The study left many questions unanswered: What, for example, were the long-term health risks to the baby when AZT was administered to a woman during pregnancy? Did the benefit outweigh the risks, given that, even without AZT, three-quarters of the babies would emerge seronegative? And the study ignited new controversy about the policy implications: Should all pregnant women now be required to take an HIV antibody test? Such questions notwithstanding, there was room for exultation. For the first time, in one particular context, AZT appeared to mean the difference between life and death.
Unfortunately for Burroughs Wellcome, news about ACTG 076 coincided with formal publication of the Concorde study. Burroughs Wellcome took the occasion to issue a spirited defense of its product: "Several independently sponsored studies have clearly demonstrated the benefits and advantages of early therapy," commented David Barry, the vice president of research, in a press release. There were indications, however, that doctors and patients were drawing different conclusions about the drug. JAMA reported that a study in the Canadian province of Ontario showed a 45 percent decline in the number of patients beginning therapy with AZT in the months following the preliminary report on Concorde. Even many patients who were symptomatic were apparently forgoing use of the drug. Burroughs Wellcome acknowledged that sales of the drug were down, particularly in the United States.