Skip directly to content

Caveat Emptor

Yokohama Roundtable Examines Treatment Quandary Facing Patients, Clinicians Alike

Burning Last Lightbulb At Noon

September 1994

Mark Harrington was invited to speak at the Thursday afternoon roundtable discussion in Yokohama, "When To Start Antiretroviral Treatment." The session was moderated by Australia's BW poster-doc, David Cooper, and WHO's recent "it's-never-too-early" convert, Dr. Joep Lange. Joining Mark at the dais were NIH's Dr. Clifford Lane and Denmark's Dr. Lars Mathiesen (who was absolutely brilliant). A mostly unexpurgated recounting of Mark's address follows.

Good afternoon. I would like to thank the conference and panel organizers for the opportunity to participate in this roundtable session. This topic is obviously a timely one, not only for the AIDS field in general, but also for myself in particular. Some of you who were in Amsterdam two years ago may have seen me present slides of my left axillary lymph node, biopsied in April 1992, in situ hybridization which demonstrated, in the words of my doctors, that my lymph node was 'crammed with virus.' At the time, I had somewhat over 600 CD4+ T cells/mm3; they are now, two years later, somewhat over 400, and I remain asymptomatic.

Over the years I awaited with eagerness, and registered with disappointment, the results of a number of studies which I hoped might provide clear guidance in choosing an appropriate time to begin a useful antiretroviral regimen. While it is widely recognized that the drugs and regimens studied in these trials proved disappointing, it is less widely acknowledged that the trials too - in their design, conduct, analysis and interpretation - have been disappointing: their design, over-optimistic; their conduct, flawed; their analysis and interpretation, often relentlessly and falsely upbeat.

In preparation for this panel, I asked several AIDS physicians when - or if - they tended to initiate antiretroviral therapy in their patients. Responses ranged from 'As soon as I can talk them into it' to, 'At 499 T cells' to, 'Whenever they ask - but never if they don't' to, 'Whenever we develop effective antiretroviral treatments' to, 'Sometime between the second and third trimester of pregnancy.' Obviously, the decision usually is ours, not our physicians'. We are forced to make these decisions without clear evidence of when to start or what to take.

As the diversity of these answers indicates, the disappointing fact is that after eight years of study, we still don't know at what stage of disease people with HIV begin to benefit from antiretroviral therapy. We also don't know whether to begin with mono- or combination therapy, or when to switch, add or stop treatment, but those topics aren't on today's menu.

Last year's NIH state-of-the-art guidelines for antiretroviral therapy in adults outlined twelve clinical scenarios. Several are relevant to this discussion. For people with over 500 CD4/mm3, no data warranted starting therapy. For those with between 200 and 500 with no symptoms, patients were given a choice of initiating AZT monotherapy or waiting. For those with between 200-500 and symptoms, AZT was recommended, as it was for previously untreated patients with below 200 CD4 cells. (The choice of 500 as a cutoff is purely arbitrary.)

In developing these guidelines, the NIH expert panel relied, appropriately, I believe, on randomized studies which measured benefit in terms of delayed clinical progression or death. The basis for the recommendation to initiate therapy somewhere between 200-500 CD4 cells, and around the development of symptoms, was based on the results of four placebo-controlled studies: ACTG 016, both strata of ACTG 019, VA 298, and the Concorde study. The basis for recommending AZT monotherapy as initial treatment was based on ACTG 016, 019 and BW-02, and on two active controlled studies, ACTG 114 and ACTG 116A, in which, respectively, AZT showed a pronounced survival benefit, against ddC, and a modest delay in progression against ddI.

Nonetheless, the SOTA panel decried the lack of better data from well-controlled studies to define exactly when, and in whom, to begin antiretroviral treatment. For, as the results of ACTG 076 in pregnant HIV+ women show, AZT can be a very powerful drug when used at the right time. Yet with adults, we still don't know how best to target its use.

The research community, since Berlin, has wallowed in denial about the lessons of the Concorde study. It is important to point out that Concorde contained more clinical endpoints (and more deaths) than all other antiretroviral studies combined. Some complained that it was naive to study monotherapy, as resistance would surely develop and eventually attenuate the efficacy of AZT. Critics of Concorde (and of ACTG 155) also point out that the rationales for early treatment and for combination therapy remain intact. These rationales fly in the face of current clinical evidence. When a rationale is held against the data, reason becomes bias. Neither the data nor the rationale justifies the routine provision of AZT to everyone with CD4+ T cell counts under 500.

Also often forgotten is the fact that, in 019, AZT reduced the rate of progression from 4% in the placebo arm to 2% in the treatment arm - a relative reduction of 50%, but an absolute reduction of only 2%! Yet, on the bias on 019, NIH began recommending the blanket use of AZT in all patients with CD4 < 500, an arbitrary number used in defining entry strata for 019; that is to say, they extrapolated that the delayed progression seen with AZT in rapid pregressors (4%) would also occur in the slower [rpgressprs (96%).

What both Concorde and VA 198, as well as the long-term follow-up of ACTG 019 itself, showed was that this did not, in fact, occur. All three analyses showed that there was, in fact, a short-term benefit of AZT in rapid progressors, and that this benefit apparently disappeared after two years. The problem was that it was difficult to identify the rapid progressors up front. Otherwise, it would be possible to target the use of AZT in asymptomatics more precisely, to maximize the potential therapeutic benefit. For the transient activity of AZT (actually, all the nucleoside analogs, the NNRTIs and the protease inhibitors) - and the magnitude and duration of the reductions in viral load which they can accomplish - suggests that using them too early may mean missing the opportunity to use them when they would really be of use. Moreover, using combinations may be no better. First-line combination AZT/ddI and AZT/ddC in BW-34,225 did not delay the emergence of AZT resistance. Second-line combination AZT/ddC in ACTG 155 provided no clinical benefit and 50% more toxicity. Triple-drug combination at UAB did not delay the emergence of resistance either.

Recent pathogenesis data suggest why antiretrovirals taken too early don't work. In the DATRI 003 study, reported on Monday by Oren Cohen, 32 HIV+ participants with CD4 >250 were randomized to maintain or switch regimens for 8 weeks and underwent serial lymph node biopsies at weeks 0 and 8. Half those patients not on AZT were randomized to AZT, and half to continue untreated. Half of those on AZT were randomized to add ddI, and half to stay on AZT alone. Overall, there were no reductions in viral burden or expression (viral RNA or DNA) in lymphoid tissue. There was a trend toward reduced viral load in those who added ddI; they had lower CD4 levels at baseline. There are several points to be made about this. Lymphocyte levels, and viral infection levels, in peripheral blood account for only a tiny fraction of the whole-body lymphocyte and virus population - just 2% of the CD4 population exists in peripheral blood; therefore, raising CD4 levels or reducing viral levels in peripheral blood gives an exaggerated picture of the magnitude of impact current antiretroviral agents have on the ultimate end organs of HIV disease, the lymphoid tissues throughout the body, the destruction of which is one hallmark of frank AIDS.

Cohen, Pantaleo, Graziosi et al. concluded that 'the effect of antiretroviral treatment on viral replication appears to be more pronounced in late stage versus early stage disease.' If the trend they saw in the lower CD4 cell, ddI-receiving arm is confirmed, this suggests that nucleoside analog antiretrovirals only affect viral load where it counts after destruction of the lymphoid tissue has reached the point where CD4 cells are around 300/mm3. After all, the typical nucleoside reduces HIV levels by one log (10X) in the peripheral blood for 3-6 months. The 'best' short term reduction in virus (with a protease inhibitor) reduces it by 3 logs for a few weeks or months. The immune system, by contrast, reduces it by 3-4 logs for 8-12 years. But this is only in the peripheral blood. Viral burden in lymphoid tissue remains high. Moreover, antiretroviral therapy may not affect lymphoid destruction if that destruction is mediated by immunological factors, rather than direct viral cytopathicity.

In conclusion, using AZT when your immune system is still holding HIV in check, and your lymph nodes are still trapping virus, may be like using your last light bulb at noon: it may be burnt out by dusk, and you'll be left in the dark. So the bitter paradox at the heart of our current understanding is that, while theoretically it will be better to start treatment early, when viral load is low, making it harder for resistance to develop, the truth appears to be that starting antiretrovirals too early wastes their limited usefulness while a much more effective immune response persists. Only when the immune system damage becomes so great that the person is somewhere on the verge of becoming symptomatic do we begin to have good evidence that current antiretroviral treatments improve clinical status. So the short answer to the topic of the day is, 'Sometime before my CD4 cells drop below 200, or before I become symptomatic, whichever comes first.

This is a pretty bleak and vague answer to give after scores of trials enrolling thousands of patients around the world and carried out over eight years. AZT sales were halved after Concorde. Even if researchers want to continue to believe in the discredited dogmas of yore, patients aren't buying them. Moreover, a growing number of clinicians, researchers and statisticians aren't buying them either. Indeed, the field of antiretroviral clinical research is getting a bad reputation, and justifiably so, in my view, for drawing unwarranted conclusions from inadequate data. This perception jeopardizes the field, the funding, the clinical research networks, and most important of all, the lives and hopes of people with AIDS. All the advances in high-tech virology and immunology have yet to be applied or confirmed in well-designed studies which provide clinical evidence of benefit. Even if you've convinced yourselves, you haven't convinced us. So your work is in vain - unless you learn the lessons of nucleoside analog development and do a better job with newer agents.