by Mark Harrington
After Berlin, no one could credibly believe that AIDS drug development was on track. Indeed, it looked likely that the various initiatives AIDS activists had worked with FDA, NIH and industry to develop were actually impeding the gathering of useful information about safety and efficacy. The first generation of anti-HIV drugs, the nucleoside analogues – AZT, ddI, ddC, d4T and 3TC – just weren’t very good, but the studies to prove their efficacy were designed over-optimistically. As a result, many studies were too small to show anything useful. The Concorde study questioned the use of early AZT and the use of surrogate markers like short-term CD4+ T cell changes to indicate longer-term clinical benefit. ACTG 155 questioned the prevailing, but never proved, dogma that two anti-HIV drugs might be better than one. It was time to take a critical look at the way drug trials were carried out, and to demand changes to resolve the issues raised by the nucleoside analogue trials.
ddI had finally been approved in 1991, after short-term surrogate marker data showed that, in AZT-experienced patients, those assigned to AZT continued to have a CD4+ T cell drop, while, at eight weeks, those switched to ddI had a CD4+ T cell rise of about fifteen per cubic milliliter of blood. This wasn’t a very impressive T cell rise, but it was enough for FDA to approve the drug. One year later, data from the same study, ACTG 116B/117, confirmed that switching to ddI was, in fact, better than staying on AZT. The benefit was only modest, however, and the side effects of ddI included infrequent peripheral neuropathy and rare but fatal pancreatitis. At least we knew in 1992, however, with the clinical benefit data, that the giant experiment we started in 1989, putting 16,000 people with AIDS on open-label ddI through Parallel Track, must have benefitted at least some of them. Of course, an easier way to have learned the same information would be to have done a much larger 116B/117 trial, which would have obviated the need for a Parallel Track, and gotten the answer within six months, rather than three years.
Also in 1992, prodded perhaps by the continuing deregulatory interests of President Bush, FDA Commissioner David Kessler promulgated new regulations formalizing the process by which ddI could be approved. The Accelerated Approval regulations provided that a drug could be approved based on surrogate marker data (such as CD4 changes) if the marker was reasonably likely to predict clinical benefit. Drug manufacturers would have to promise to conduct post-marketing studies to prove their drugs actually provided clinical benefit. The promise of accelerated approval was clear – drugs which were safe and might be effective could be released sooner – but its downsides weren’t as apparent. Long-term side effects may not have yet been observed at the time of approval. With its drug on the market, the sponsor had much less incentive to carry out promised post-marketing studies, and it would be impossible for FDA to remove an approved drug unless it caused truly ghoulish side effects. Ultimately, then, Accelerated Approval, if the company didn’t keep its end of the bargain, threatened to replace clear answers about how to use new drugs with rapid access in a vacuum of relevant information for doctors and patients.
The result, for people with HIV and their doctors, was a confusing and contradictory situation. People were being told that these new drugs were vital and life-saving, but there was no data that they, in fact, saved lives, and they were expensive, inconvenient and toxic. The first drug released officially under accelerated approval was Hoffmann-LaRoche’s ddC, a popular drug at the time because small surrogate marker studies suggested that ddC taken with AZT might be superior to AZT alone. Activists had waged a frustrating and unsuccessful campaign to force Roche to release ddC on Parallel Track, dating back to G’dali Braverman’s zap in San Francisco in 1990. Roche refused, and as a result, a flourishing underground grew up and thousands of PWAs went on underground ddC. To stoke up demand just before its FDA hearing, Roche sent samples of the underground drug to the FDA, which determined that the amount of ddC per pill ranged from none to ten times the recommended dose, which could lead to permanent peripheral neuropathy. Just before its scheduled FDA hearing, Roche managed to shut down the underground and cynically announced a sudden conversion to the idea of Parallel Track. They were equally cynical at the hearing itself. There was no data that ddC helped anyone. The only solid clinical study was one which showed that twice as many patients assigned ddC alone died in the first year than the number of deaths in the AZT arm. Its effects on CD4 cells were, if anything, weaker than even ddI’s. The application made it through the committee based on a ridiculous six-arm study run by Margaret Fischl and Doug Richman comparing various doses of AZT and ddC in combination. There was the suggestion, but hardly proof, that perhaps some combination regimen might be superior to AZT alone. On the basis of this flimsy evidence, generated from a trial designed like a Hindu icon with many arms, all tiny, ddC was approved for combination use with AZT, and Roche promised a raft of postmarketing studies. Roche never carried out its commitment to undertake the follow-up studies and, in 1993, ACTG 155 dashed the hopes of those who believed that AZT/ddC was better than AZT alone in AZT-experienced patients.
If the ddI experience hinted that accelerated approval might work, the ddC one showed that, in the hands of an unscrupulous or opportunistic drug company, its requirements for postmarketing studies could easily be flouted. ddC really was a rotten drug, and Roche seemed like a rotten company.
A small group of activists, most but not all from the east coast – among them Gregg and I, David Barr and Derek Link, Lynda Dee and others – were increasingly upset at the cynicism with which industry manipulated Accelerated Approval, with the FDA’s flaccid response, and with the access-obsession of other community activists, who no longer seemed to think drugs were worth studying once they were on the market, and yet who clamored endlessly for access to drugs in the early stages of testing. They seemed to believe that drugs were a cure in the test-tube, life-saving in phase I, health-prolonging in phase II and then, once licensed after Accelerated Approval, no longer worth studying and perhaps not even worth taking.
We felt, on the other hand, that activists were no longer representing people with AIDS and HIV if they did not insist that drug companies generate information informing how to use the drugs released under accelerated approval. Most people with HIV are not the polypharmacy maniacs who tended to show up at drug company meetings and FDA hearings. Many treatment activists represent a vocal segment of the community which is intoxicated with access, and has abundant access to health care and the latest information and experimental treatments. Most people with HIV, however, lack such access, and many do not want to take drugs which are toxic, expensive, inconvenient, and for which no clear evidence of benefit exists. From a public health standpoint, we felt we had a responsibility to confront the drug industry, the FDA and the community activists with whom we disagreed, to force them to design studies which would prove whether a given new drug was worth taking or not.
After Concorde and ACTG 155, we no longer knew – though, in the face of the evidence, many scientists and activists still maintained – whether early intervention was useful, whether short-term CD4 changes predicted longer-term clinical benefit, or whether combined nucleoside therapy was better than monotherapy.
TAG and our allies set out to fulfill the promise of Accelerated Approval and to rescue it from expedient distortion by drug companies. In December 1993, as I was finally leaving the ACTG and its Community Constituency Group, after a frustrating year of working with Fischl, Merigan et al. on the Primary Infection Committee, I wrote a valedictory report, The Crisis in Clinical AIDS Research, lambasting the ACTG and industry for trials which were too small, too short, naively over-optimistic, and which failed to generate clear answers about new anti-HIV drugs.
In May 1994, Gregg Gonsalves was asked to sit on the FDA panel reviewing data on Bristol-Myers’ new drug d4T. The company, unlike Roche, had done a great job on its Parallel Track, enrolling 27,000 patients and comparing two doses of d4T (the lower dose proved safer) but, like Roche, its trials were unimpressive. Gregg predicted that its ongoing confirmatory study was too small to prove whether d4T was superior to AZT and the data, 18 months later, proved him right. Gregg could not in good conscience vote for accelerated approval for d4T, taking a step which drew TAG into a radically revisionist campaign to reform clinical research.
At this point, FDA approval hearings resembled scenes from Alice in Wonderland. Baffled by the d4T data, Advisory Committee chair Debbie Cotton asked David Feigal, Cooper’s successor at the FDA’s Antiviral Drug Division, “Is evidence required?”
“No,” replied Feigal. In essence, for anti-HIV drugs, the Kefauver amendments were in abeyance.
“Is evidence allowed?” riposted Cotton.
“Yes,” replied Feigal.
Gregg, Spencer and Derek Link were livid when they returned from the hearing. Yet another drug would be on the market, and no one would know how to use it. On behalf of Derek Link, GMHC wrote a letter to the FDA recommending against approval. The same week, d4T was released.
That summer, several of the new protease inhibitors were in phase I trials in the U.S. and Europe. Protease inhibitors attack a different HIV protein than the nucleoside analogues, and potentially they could shut off viral replication in two different ways. They are complex molecules, much bigger than the nucleosides, designed by computer-assisted structural biologists, and they are expensive to manufacture.
In May, Roche announced the conclusion of ACTG 229, a study comparing AZT plus ddC plus its new protease inhibitor, Saquinavir, versus AZT plus ddC or AZT plus Saquinavir. The study showed modest increases in CD4 counts and modest decreases in HIV levels; benefits were greatest in the three-drug regimen. Saquinavir looked to be about equivalent to ddC, which didn’t seem like much of a breakthrough.
Although fewer than 300 patients been on Saquinavir for less than six months, Roche prepared to file an Accelerated Approval request from the FDA.
TAG decided to undertake a campaign to ensure that a much broader discussion take place between activists, researchers, FDA, NIH and drug companies to develop a comprehensive approach to studying the protease inhibitors. We did not want to end up in five years with five new drugs on the market an no clue how, whether, or when to use them.
On June 16, 1994, in a letter which rocked the AIDS research establishment and sparked a nationwide controversy, TAG, GMHC, the AIDS Action Council and AIDS Action Baltimore wrote a letter to FDA Commissioner Kessler objecting to Roche’s premature request for accelerated approval for Saquinavir, saying, “We feel that such an approval would penalize people with AIDS/HIV by setting an inappropriately low standard of evidential requirements that would govern the regulation of [all protease inhibitors]. We urge you not to invite Hoffmann-LaRoche to apply for Accelerated Approval of Saquinavir until we can complete further discussion between FDA, its Advisory Committee, the company, and people with AIDS/HIV.”. Unlike any previous Accelerated Approval, Roche had conducted not even a pretense of a Parallel Track, and there was no long-term safety evidence on Saquinavir in thousands of people, as there had been with ddI, ddC and d4T. We demanded that Roche implement a Parallel Track to gather safety (and possibly dosing) information before approval.
At a heated meeting between TAG, GMHC and FDA officials, including Kessler, we charged that the FDA was allowing industry to flout the requirements of Accelerated Approval and that if they approved Saquinavir now, no other protease inhibitor sponsor would ever bother doing a study showing whether their drugs worked or not. Certainly Roche’s own record with ddC demonstrated that it was unlikely to meet such commitments. We demanded that the FDA sponsor a meeting to discuss how to design studies for the protease inhibitors as an entire class, rather than reacting to Saquinavir in isolation.
Antiquated activist dogmas had crashed in the face of the new situation, which we had helped to create. Nonetheless, the putative unity of the community, industry, FDA and researchers in their common belief in surrogate markers, expanded access and accelerated approval was shattered by our actions. The FDA staff looked shocked to hear such words coming from AIDS activists. They agreed to schedule a special meeting of the Antiviral Drugs Advisory Committee meeting in September to address our issues.
When Roche got a copy of our letter, they furiously faxed it to scores of activist organizations. A nationwide firestorm ensued. TAG became the most unpopular AIDS organization in the country. We became the target of vitriolic invective and personal abuse. Critics charged that we were trying to destroy Accelerated Approval. We were not – we were trying to save it, by forcing industry to live up to its part of the bargain – but the nuances of our position were quickly lost in the rumor mill.
Barron’s magazine poured gasoline on the flames in mid-August when it ran an front-page article about the controversy, with extensive quotes and a picture of Spencer Cox, headlined, “DO WE HAVE TOO MANY DRUGS FOR AIDS?” That was not what Spencer said, of course, but the headline determined people’s reaction to his presence in the article. Time magazine duly weighed in a week later, with a photo of Spencer looking like a grunge singer loitering in a shadow-streaked Manhattan alley. For a few agonizing weeks, Spencer was the most hated AIDS activist in America.
We struggled to compile a clear statement of our policy position and assembled a monograph for the FDA hearing in September, Rescuing Accelerated Approval: Moving Beyond the Status Quo. The FDA hearing was a circus. Scores of activists harangued the committee, and vilified TAG, for two full days. We were clearly in a minority. David Feigal looked rather smug about this, though David Kessler listened intently to everyone.
Once we clarified that we were not opposed to Parallel Track and Accelerated Approval, the temperature dropped a bit. We worked with other community groups to demand that all three protease inhibitor developers, Abbott, Merck and Roche, develop Parallel Track programs for people with advanced AIDS. The programs that resulted were tiny, enrolling only a thousand patients each, but they gathered some useful safety data. The drug companies needed most of their small drug supply for their studies, and we were demanding that their studies be larger and longer.
After the crisis was over, we met with each of the protease sponsors. Derek Link even engineered a face-to-face meeting with Jürgen Drews, Roche’s president of worldwide research. TAG maintained an ongoing critique of all the companies involved in the field, publishing another monograph, Problems with Protease Inhibitor Development Plans, at another FDA hearing in February 1995.
In response to TAG’s demands, Roche doubled the size of one of its pivotal efficacy trials, from 1,500 to 3,000. Abbott conducted a study we designed which randomized patients to Ritonavir or placebo, but allowed them to take any available nucleoside analogue as underlying therapy. This so-called “standard-of-care” control arm allowed patients maximum flexibility to choose among existing agents, while still rigorously measuring the efficacy of Ritonavir. Merck, certain that their drug was the best, was the most arrogant, and the least receptive to our ideas, until the ACTG finally agreed to do a large-scale study of Merck’s Indinavir, ACTG 320, which would enroll 2,000 patients, largely subsidized with public funds.
By 1996, it was clear that, despite all the sound and fury, our campaign had been at least a qualified success. In February, Abbott Laboratories announced that its protease inhibitor Ritonavir plus nucleoside analogues reduced the risk of death in people with advanced AIDS by 50% over six months. Our standard-of-care control arm and our insistence on clinical endpoints had led Abbott to generate clearer clinical data on the efficacy of a new anti-HIV drug than any trial since the original Burroughs-Wellcome AZT study ten years before. The drug was fully approved overnight by the FDA. Abbott had gained a leap over Merck, which had only surrogate marker data and lacked enough drug to sell, though in a few weeks the FDA, which was now actively soliciting drug companies to apply for approval earlier than ever before, granted Merck Accelerated Approval for Indinavir. In May 1996, Roche released data from its study in advanced patients showing that even Saquinavir, the weakest of the protease inhibitors, prolonged survival in advanced patients when used in combination with a nucleoside.
Once the drugs were approved and on the market, thousands of people with failing immune systems started to take them. Many people experienced drops in their HIV levels of 99% or more, and corresponding CD4+ T cell rises. Some people left the danger zone of low T cells and maintained healthy levels for many months. The full duration of the antiviral activity of the protease inhibitors is not yet known, since they are so new, but, thanks to TAG’s advocacy, more is known about how to use them than was known about ddI, ddC or d4T at the time of their accelerated approval. Moreover, these drugs appear to be more powerful than the nucleosides, and attack a different viral target, so, when used in combination, they may provide the best control of HIV we have yet achieved, although resistance is likely to eventually appear.
There are several downsides to the protease inhibitors. Drug companies always say their experimental drugs are non-toxic, but each protease inhibitor has significant side effects, and they cannot be taken with many other common medications. Their rapid development has considerably complicated the medical management of HIV disease, and many further studies are required in order to figure out how best to use them. It remains unclear whether industry, NIH or anyone else will fund these urgently needed studies.
The protease inhibitor are also forcing a reimbursement crisis on public and private third party payers, comparable to the crisis posed by AZT’s original price of $10,000 per year. Each protease inhibitor costs between $5,000-$8,000 retail, and people are taking them alongside one or two nucleoside analogues. Already, many state AIDS Drug Assistance Programs cannot afford to cover these new drugs alongside already available ones, and the problem may only intensify in the coming years.