Kiss My Access: Community Envoy to Food and Drug Panel Urges Caution in Approving New Anti-HIV Therapueutics and in Providing Early Access

The FDA recently convened hearings regarding its Accelerated Approval program and the future use of changes in plasma HIV-RNA as surrogate endpoints in AIDS clinical trials (September 6-7, 1995). TAG’s Peter Staley, representing the National Task Force on AIDS Drug Development, was one of three community representatives to address the advisory committee. Below is an excerpt of his remarks.

My name is Peter Staley and I am founding director of the Treatment Action Group (TAG), the nation’s only AIDS advocacy group focusing exclusively on improving and enhancing basic and clinical research on AIDS. I was diagnosed with AIDS-Related Complex (ARC) in 1985. I have been on antiretroviral therapy for nine years now, and on combination therapy since 1989. Since January of this year, I have been using a viral load test to assist me in making treatment decisions.

This meeting is long overdue. Indeed, TAG has been calling for over two years for just such a meeting, focusing on how best to prove that anti-HIV drugs actually provide clinical benefit to people with HIV disease. There are several reasons why focusing on clinical validation studies and the use of surrogate markers is now timely.

First of all, experience with the past three antiretrovirals approved on the basis of surrogate marker changes in CD4 levels has been mixed. ddI and d4T were each approved on the basis of CD4 changes in long-term AZT users in a single study. About one year later, evidence emerged of a modest clinical benefit (about 3 months) with regard to delayed progression to a new AIDS-defining event. In neither case was a survival benefit shown. ddC was also approved on the basis of surrogate marker changes. Later, no clinical confirmation of benefit emerged vis-à-vis combination therapy, and only the slenderest indication of “equivalence” to ddI was shown–in a very advanced population which did no better than a natural history population on long-term AZT.

Secondly, last year, Tom Fleming’s paper showed that CD4 is, indeed, an imperfect surrogate marker for antiretroviral drug activity; in 8 of 16 randomized controlled clinical endpoint studies which provided the basis for the 1993 State-of-the-Art (SOTA) guidelines, CD4 changes correlated with clinical benefit–in the other 50% they did not. Thirdly, we now have new, powerful tools by which to measure antiretroviral activity directly in the peripheral blood of infected persons: RNA PCR and bDNA.

Fourthly, a new generation of antiretroviral drugs is rapidly approaching consideration by FDA and its advisory committee for accelerated approval: the protease inhibitors. Roche filed its accelerated NDA for saquinavir on Tuesday (5 September 1995), and applications are expected from Merck for indinavir (Crixivan ) and from Abbott for ritonavir in the first half of 1996. The new viral load measurements will form an important part of these surrogate-marker-based accelerated NDAs. Fifthly, many people with HIV, including myself, are already monitoring these viral measurements to help them make their own treatment decisions, even though it is far from clear how best to use them. Let me make an example of myself.

I have been on combination AZT+ddI since 1989, when ddI became available through an expanded access program. My CD4s went from a low of 105 to a high of 700 within 12 months. They slowly fell to the mid-300s, remaining there until last year. Then, without any change of therapy, I experienced another 12-month rise to 540. Six months ago I went to Roche’s lab and received my first PCR viral load test, reflecting a remarkably low count of 19,000, offering further validation of my success in using this combination for six years now. However, just this summer, I was disappointed to see my CD4s fall back to the low 300s in just three months. Worried that resistance had finally appeared and that I should consider changing therapies, I retested my viral load only to find that it had dropped farther, to 11,000.

I must tell you personally that the theory behind the potential for viral load to be a useful therapeutic marker has convinced me to rely on it as a driving force in making my treatment decisions. I decided not to change therapy at this time.

Now that I’ve told you my personal history, as people with a responsibility to the public health, I ask you to take it all with a grain of salt. I am not representative of the vast majority of people living with HIV, and I am an insignificant voice of those living with HIV world-wide. The same could be said of every person with AIDS in this room, and most AIDS advocates that you meet with on a regular basis.

Since the majority of the people in this room are AIDS researchers with major pharmaceutical companies, I must tell you that I think you are some of the smartest people working in the field of AIDS, including bureaucrats from government, industry, academia, or especially AIDS activists claiming to represent the community. Just as you should be willing to pound your heads against a wall advocating for the public health against your very aggressive marketing staffs and your bottom-line zealots higher up in the corporate office, I encourage you to stand up with equal conviction for the public health against those AIDS activists who are willing to hijack clinical trial design to meet their own personal needs. For instance, many activists have done nothing but badger you about access to the protease inhibitors, even suggesting that clinical endpoint trials are not necessary. Have you heard any of them mention to you lately that PCP (Pneumocystis carinii pneumonia) remains the leading cause of AIDS-related sickness and death in New York City, and, presumably, in the United States?

While many of these same advocates have proclaimed that people living with HIV would never enroll in an antiretroviral study using the standard-of-care plus placebo control, two such trials have recently started and have already fully enrolled in record speed: Abbott’s protease study for late-stage patients, and Glaxo’s 3TC study in Europe. What’s my take home message here? That defending the public health should be the joint responsibility of all of those fighting to end the AIDS epidemic, and that all of us have frequently failed to serve that goal.

So while my story is one thing, my friend and colleague Spencer Cox recently experienced a CD4 drop from 260 to 65 in four months, and almost a 1-log drop in plasma HIV-RNA at the same time. I don’t know what I’d do in that situation. The CD4 drop was frightening, so it–and not the viral load drop–would probably dictate my next treatment decision.

I hope I’ve made it clear how difficult it is to make any sense of the current viral load tests without any data from randomized controlled clinical trials with clinical endpoints. Patients and doctors will be clueless until we have better data from such studies to help guide our treatment decisions. Studies to date have been contradictory. For example, the viral substudy from ACTG 116B/117 indicated that there might be a correlation between viral load changes and clinical outcome. ACTG 241, though, indicated the opposite: the patients on AZT+ddI+Nevirapine had a greater reduction in viral load, but they also experienced more clinical endpoints. No viral load-clinical outcome correlation study published to date has had sufficient power (or narrow enough confidence intervals) to make any conclusive judgements with confidence.

There is a second proposed use for viral load measurements other than making individual treatment decisions, and that is for indicating antiretroviral activity in clinical trials. There was only scant antiviral activity data in the NDAs for ddI, ddC and d4T, much of that based on changes in p24 antigen levels, and much of that inconclusive or difficult to interpret. Yet there is every indication that the protease inhibitor sponsors will rely heavily on quantitative viral markers in their accelerated NDAs. Without question, these agents have in vivo antiretroviral activity. I could get very excited when I hear about “two log reductions in viral load–one log greater than AZT monotherapy,” and go out immediately and add a protease inhibitor to my antiretroviral regimen, once these agents are available. However, there remain many unanswered questions and some compelling reasons not to rush to add protease inhibitors.

For example, if a protease inhibitor lowers my viral load by two logs, but AZT lowers it by one, yet each of them only works for twelve to twenty-four weeks, am I going to experience any longer-term benefit from the greater reduction without greater duration of activity of the protease inhibitor? Another troubling issue is cross-resistance, particularly between the Merck and Abbott compounds. Will these drugs, which appear powerful in the short-term, be a one-shot deal whose benefit I can experience only once? If so, when is the best time to try to obtain that benefit? Only long term, randomized studies using these viral load measurements and correlating them with resistance, CD4 changes and clinical outcome will provide reliable answers here.

Our biggest failure is thinking that there is a clear-cut relationship between any of these markers and the clinical effects of therapy. In the age of actively-controlled trials, pulling apart the relationship between markers and clinical benefit and toxicities is even more complicated than in the days of placebo-controlled studies. Everyone in current trials is on antiretroviral therapy. We are comparing different regimens in which everyone is experiencing a mix of antiretroviral activity, emergence of resistance, immunological and clinical progression, and drug-induced toxicities. Surrogate marker changes do not indicate how many people can tolerate a drug, or what the medium- and long-term side effects are.

You can have a ‘home-run’ surrogate marker drug which can kill people faster than the disease left to its own devices. Famous examples of this include the anti-arrhythmia drugs ecanide and flecinide, which reduced arrhythmia significantly compared with placebo but caused tens of thousands of premature deaths. Another sobering example is the anti-hepatitis drug FIAU, which virtually eradicated hepatitis B antigen in phase I trials, but killed half the patients in the phase II trial by destroying their livers. High-dose ddC looked very active antiretrovirally (as judged by p24 and CD4 changes), but caused chronic peripheral neuropathy.

Just last week, an NIH panel completed a meta-analysis of 16 studies of the calcium channel blocker nifedipine (Pfizer’s Procardia ), finding a 60% increased chance of a heart attack when using this putative anti-heart attack drug. Each treatment was adopted because of compelling surrogate marker data based on a convincing therapeutic rationale, yet each may do more harm than good. Surrogate marker analyses are generally blind to toxicity. Similarly, it is likely that long-term AZT use in some populations does more harm than good.

That is why we still don’t know whether ddI or d4T is better than nothing in long-term AZT users, or whether the patients simply benefitted from stopping long-term AZT, with its toxicity to the bone marrow and muscle mass. If a drug is active short-term against surrogate markers but has long-term toxicities, short-term studies relying on surrogates will exaggerate the real-world potential of these treatments and minimize the longer-term dangers. This is exactly what most previous antiretroviral studies have done, and likely helps to explain the failure of AZT started too early (as in Concorde, or in ACTG 019 when starting with over 500 CD4 cells) to affect progression or survival. Markers predictive of the natural history of a disease don’t necessarily translate into good indicators of clinical outcome on therapy.

We need several studies which are large enough to demonstrate a clear clinical benefit between different treatment regimens or strategies and which attempt to correlate viral load measurements, resistance, CD4 changes and clinical outcomes, in order to make sense out of the new tools and the new treatments which we are concerned with here today. It remains to be seen whether the clinical confirmation studies conducted by protease inhibitor developers will meet this task.

Indeed, there may well be a role for large, public-health type NIH-funded phase IV studies to optimize the use of protease inhibitors and the use of viral load in guiding treatment decisions. NIH has been virtually absent-without-leave from the entire development of protease inhibitors (with the exception of the 302-patient ACTG 229 saquinavir study), and neither the ACTG nor the CPCRA currently has the infrastructure or the resources to carry out large-scale, long-term, competent, well-designed and adequately controlled studies of the kind we need. Until such studies are a routine part of AIDS drug development, we will never know whether we are, inadvertently and with the best intentions, doing more harm than good with antiretroviral treatment regimens. Thank you.