The introduction of viral load assays in the last two years has brought a revolution in monitoring the progress of HIV infection and vastly simplified clinical trials of new drugs. These days, the FDA accepts six months of viral load data as sufficient to support accelerated or conditional approval of a treatment for HIV and 48 weeks for complete approval. The result has been much shorter and narrower trials than ever before. People are forced to minutely analyze the viral load data that have come out to understand these trials' overall significance. They compare such factors as the trials' baseline viral load and final viral load results, the percent of participants with viral loads falling below test limits, the type of viral load assay used and the CD4 counts and other indicators to trial participants' health status.
Reports from Geneva and the immediately preceding Second International Workshop on HIV Drug Resistance and Treatment Strategies indicated, though, that even the most careful analyses have been overly optimistic. (Note that the official name of last year's resistance workshop -- the International Workshop on HIV Drug Resistance, Treatment Strategies and Eradication -- imparted a decidedly more sanguine tone than this year's implied assessment of therapy's limits.)
At the resistance workshop, Andrew Hill (of Glaxo Wellcome, UK) described the statistical smokescreens that obscure trial reality (abstract 76). In the AVANTI-2 trial, for example, those on AZT/3TC/indinavir were found to have a week-28 viral load reduction of 1.6 to 3.1 logs (97.5% to 99.98%), depending on the statistical methods used. Similarly, the percent below the limit of quantification (%BLQ) with a viral load assay ranged from 78% by the least rigorous analysis to 46% by the most conservative one.
A major factor in these differences is whether individuals who stop taking their trial medicines are included. "On-treatment" analyses, which do not include the patients who drop out of the studies, generally yield better numbers than "intent-to-treat" analyses, which do include all patients entered in the study, regardless of their outcome. Counting in the data from dropouts tends to reduce a drug's observed efficacy because a major reason for dropping out is, obviously, poor response to the trial medications.
The FDA is now advising drug companies to analyze their data according to "intent-to treat" methods. One is called "last observation carried forward" (LOCF) and the other is "noncompleter equals failure" (NC=F). LOCF takes the last results before a person drops out of the trial and counts them as if these figures occurred at the end of the trial. Each method combines all the data from the period when participants are following the trial protocol without having to worry about what happens to dropouts after they leave the trial and switch drugs.
Noncompleter equals failure is useful for trials with yes or no answers. In HIV trials, anyone who drops out is counted as a nonresponder, that is, that person failed to go below the level of detection in the viral load assay. This brings up a second major factor in reducing drugs' success rates. Until recently, trials used viral load assays that only went down to 1200 or 400 copies/ml. This threshold was due to the technological limitations and had no biological significance since considerable HIV replication is taking place below 400 copies/ml.
With the introduction of the ultrasensitive PCR assay, which is accurate down to 50 copies/ml, it has been discovered that many, if not all, people with viral loads of between 50 and 400 will develop drug-resistant HIV and see their viral loads rebound. (See Douglas Richman's lecture at Geneva on Monitoring Therapy -- oral presentation 22) But measuring viral load down to 50 copies/ml cuts into the observed percent BLQ. With the ultrasensitive viral load assay only about 50% or 60% of the selected groups recruited for clinical trials are below the level of quantification, by intent-to-treat analysis.
Here are some examples of trial data presented in Geneva that include percent of trial participants with viral loads below 50 based on some sort of intent-to-treat analysis:
If half the people in trials of highly active antiretroviral therapy (HAART) have a risk of early rebound in their HIV, what about the other half that do successfully get their viral loads below 50 copies? It has been obvious for a while that they still have HIV lurking in the background. It comes back any time therapy is interrupted.
A report by Richard Harrigan from Vancouver's Center for Excellence in HIV/AIDS described what happens in detail (abstract 11152). The group observed a set of patients who had had viral loads of less than 50 copies for one to three years when they stopped therapy. All exhibited an immediate resurgence of HIV, with viral loads doubling every day for the first three weeks off therapy and reaching up to the 100,000 range.
The group tried to use the size and slope of viral load rebound in about 20 patients to estimate the original total bloodborne HIV before therapy was stopped. They found the total remaining free virus even under supposed maximal suppression was on the order of 10,000 to 1 million copies, which would indicate a considerable amount of residual HIV replication. Dr. Harrigan also noted that when you look at the number of HIV-infected cells, the therapy-influenced decline is very slow -- the number goes down by only about 50% per year. His conclusion was that the infected cells were unlikely to be eliminated by current therapies.
Many of the infected cells contain defective HIV gene sets that are incapable of producing new virus. It has been known for over a year, though, that there is a sizable pool of latently infected cells, at least 1 million, that harbor replication competent HIV and could start producing new HIV particles when activated. One of the pioneers in exploring this cell population, Robert Siliciano of Johns Hopkins University, noted in Geneva that the pool size is stable over at least four years of therapy (abstract 11150).
There is a possibility that this pool is slowly replenished by continued rounds of viral replication even when plasma viral loads have been minimized by therapy. If such continued rounds of replication do occur then the apparent size of the long-term latent pool is enhanced by cells that actively produce HIV and have quick turnover times. Also, the latently infected cells might be the source of the rebounding HIV that Dr. Harrigan observed, but the speed of that rebound and the estimated size of the residual free virus population makes the latent cells an unlikely source.
A startling presentation by David Ho of New York's Aaron Diamond Institute revealed that this is precisely not the case (oral presentation 167). In six selected patients who started therapy very early in infection, the size of the latent pool decreased by half every six months. This half-life is the expected one if there is no replenishment and most of the latently infected cells die a natural death without ever producing HIV.
And if there is no residual replication, one would expect that the HIV in the infected cells should have a constant genetic makeup, with little sign of change over time, since it represents old, latent virus left over from the days before HAART was commenced. When looking at patients with suppresed HIV, though, Dr. Ho found accumulating mutations in the gene for HIV envelope protein. (Dr. Harrigan also found indications of this phenomenon.)
So far, Dr. Ho has detected no evidence that drug-resistant HIV is emerging from the residual replication. It may be that the cells in which HIV continues to replicate are to some extent off-limits to antiviral medication. The failure of drugs to reach all the cells could be due to metabolic or physical barriers, whether temporary and accidental or some permanent feature of body chemistry. There is yet to be any determination of where residual replication occurs, so all theories are completely speculative at this point.
Clearly, though, the potency of HAART even under the best conditions has been overestimated. If there is HIV replication in the presence of drugs, then the evolution of drug resistance and the resulting viral rebound is a "disaster waiting to happen," to quote Anthony Fauci, director of the Laboratory of Immunoregulation at the National Institutes of Health.
It was widely remarked that the 12th World AIDS Conference raised more questions than it answered. One of the greatest of these was, are all current antiviral regimens doomed to failure? With successful HAART that reduces viral loads below 50 copies, HIV might take years longer to break through than with less suppressive therapies since the residual rate of HIV replication is so low. Even so, many doctors already are seeing patients who have signs of reemerging HIV. Dr. Michael Saag of the University of Alabama noted, "I've seen people with months of unquantifiable virus who suddenly have measurable virus. The initial response of most investigators would be to blame it on adherence, but I know some of these people wouldn't have missed a dose."
These issues are not really new. Dr Fauci's lab reported last year that it could find signs of newly infected cells in patients with suppressed HIV. (Specifically, the researchers found newly minted HIV DNA that had not yet merged with the cell's genetic machinery -- see Proceedings of the National Academy of Science, Nov. 25, 1997, pages 13193-7.) The report faced widespread criticism that the patients it utilized were not properly selected to ensure that their HIV was completely suppressed.
Another Cassandra was Dr. Saag, who met general skepticism when months ago, he calculated that with a viral load of just one copy per ml, there must be 100,000 cells actively producing HIV in the body at any one time. The Geneva conference marked the first time the medical community took these warnings seriously and started examining ways to cope with residual replication.
The most obvious strategy is just to do more of the same: If three drugs aren't enough, why not try four or five? That suggestion has already been put in practice for people who have very high untreated viral loads or who rebound on three-drug regimens. More drugs might be useful in cases of runaway HIV, but there is no proof that four or five drugs bring viral loads down lower than three drugs do. (See the results for nelfinavir/saquinavir plus two nucleoside analogs mentioned above.)
Drug resistance assays might be used to help pick the most effective drugs, and drug level monitoring could detect whether the drugs were actually reaching the bloodstream. The list of possible additions goes on and on, as the side effects and costs mount. There are already serious fears that protease inhibitors will cause heart disease in many people after a few years of use (see pages 9-12 in this issue), and adherence to complicated regimens already is the major life activity of many people with HIV. As the years go by, such adherence will only decline. As Dr. Ho argued in pushing for new therapeutic strategies, "Some people anecdotally have found greater potency with five drugs, but is it practical to do this for years?"
Having given up on eradication for the foreseeable future, Dr. Ho is now searching for the means to achieve remission. Remission is inspired by the example of long-term nonprogressors, who survive for years and perhaps decades without treatment even as they maintain stable CD4 counts and low viral loads of a few hundred copies per ml or less. If one could mimic the nonprogressors' immunity in treated patients, it might be possible to withdraw some or all drugs. The HIV would not be gone, but the body would be able to control the small amounts left without suffering any damage.
The nonprogressors are distinguished by their active anti-HIV immune responses orchestrated by the T-helper CD4 cells (abstracts 21126 by Fabrienne Hadida of the Hôpital Pitié-Salpétrière in Paris and 425 by Bruce Walker of Massachusetts General Hospital). Outside of the nonprogressors, CD4 cells that proliferate and promote the immune attack in the presence of HIV are not observed in people with chronic HIV, though they may exist during the initial or primary phase of infection. (One theory is that HIV kills off the activated CD4 cells responsible for maintaining the anti-HIV defenses.) As CD4 cells come back under HAART, no return has been noted so far in anti-HIV immunity, except in people whose treatment starts during the first six months after contracting HIV.
The lack of new anti-HIV CD4 cells is curious since other immune responses do seem to return when people's HIV levels decrease. Brigitte Autran, of the Hôpital Pitié-Salpétrière in Paris, described what happened in her cohort of 18 extensively studied persons with AIDS. In half of them, CD4 cells sensitive to CMV and TB returned after a year on HAART, but there were no new cells directed at candida, tetanus or HIV (oral presentation 426). The only exceptions were those who received a tetanus vaccine. Perhaps more time is needed for the response against HIV to reappear, although Dr. Ho thinks that people with viral loads below 50 have so little HIV in their bodies that there is insufficient stimulus for anti-HIV CD4 cells to proliferate and manifest themselves, if any exist. One way to get around this would be to use an HIV vaccine, following the tetanus example.
As it happens, the Immune Response Corporation has been trying to develop and promote its HIV therapeutic vaccine (Remune) for the past dozen years without yet producing definitive proof of its effectiveness. (This vaccine consists of killed HIV stripped of its outer envelope and mixed with an immune-stimulating adjuvant. It was first proposed by the late Jonas Salk.) The company had noticed the opportunities HAART presented for the vaccine, and started a small trial over a year ago to see whether immunizations with Remune could improve on HAART alone.
In Geneva, Fred Valentine of New York University reported the preliminary results from the trial. The 43 treatment-naïve participants were all put on AZT/3TC/indinavir for the first four weeks. Their baseline mean CD4 count was 493 and their mean viral load was 8159, so these were people with relatively early stage HIV infection. After four weeks, those with viral loads reduced to below 200 were divided into two groups. One group was immunized with Remune at weeks 4, 16 and 28. The other group was vaccinated with the adjuvant alone at the same time points.
Dr. Valentine summarized the week 20 results, describing what he called "enormous" new proliferative responses in the Remune group, while the adjuvant-only group showed no improvement. The CD4 cells' proliferation was checked in the test tube by culturing a sample of those cells with either Remune, the HIV p24 core protein from the Remune HIV strain, recombinant p24 made by genetically engineered cells and HIVBAL, an HIV lab strain. Cell stimulation indices increased an average of five times in the tests on cells taken from the Remune recipients. But Conference attendees wondered if these newly active CD4 cells had any protective value.
Although Dr. Valentine claims that the Remune plus HAART recipients now have immunity similar to long-term nonprogressors, there was no evidence presented to show that the new immune responses are in fact protective. Confronted by HIV that has escaped therapeutic suppression, they might prove just as inadequate as they usually do during primary infection. One piece of information from the trial that has yet to be analyzed is the trial volunteers' level of proviral DNA, a measure of the number of infected cells. A Remune-associated reduction here would indicate a decline in residual replication, although it might take much longer than the trial's 32 week follow-up time for any reductions in infected cells to be apparent.
Dr. Valentine is now designing trials that would document the time to HIV rebound in people on HAART with and without immunization. Another possible trial would be to take vaccine recipients and a control group off antiviral therapy for a few days and closely monitor them for the reemergence of HIV. Bruce Patterson of Northwestern University described a new rapid and cheap method for assaying for proviral DNA and HIV-infected cells (abstract 41252). Monitoring trial participants with such an assay might be useful before taking such drastic action as interrupting therapy.
Dr. Ho, who is also preparing a vaccine trial, suggested in Geneva that the immunizations be accompanied by therapies to clean out the pool of latently infected cells. Such a treatment might use the natural cell-activating cytokines to stimulate these cells' HIV into production. The cells would then be killed off either by the virus or the immune system.
Successful treatment of this sort would further diminish the amount of HIV that the vaccine-induced immune responses would have to contain but would probably come with a heavy price in terms of flu-like side effects. Dr. Fauci related some experiments that he has done using immune stimulants on lab samples of cells from eight patients on HAART (oral presentation 157). In only one case were the HIV-infected cells eliminated; in the other seven, they were moderately reduced. Just one round of stimulation was administered, but still, the elimination of the latent pool is obviously not a simple task. Dr. Fauci concluded by calling for more work in this area, terming it "the research agenda for the next several years."
In times of confusion, the rationale for slowing down and adopting a wait-and-see approach is very strong. When protease inhibitors came out, many people had poor experiences with them because they had followed aggressive treatment advice for years. Their HIV gradually developed resistance to all the nucleoside analogs that could complement the protease inhibitors. Having "burned through" these older agents, their viral replication frequently remained high when they added protease inhibitors, and many rapidly burned through those drugs as well. Those who put off treatment did better.
Discussions in Geneva made the important point that there is no solid reason to seek maximum viral suppression if a cure is not in sight. The body can tolerate substantial HIV levels with little apparent harm. One presentation in Geneva, by Dr. David Katzenstein of Stanford University (abstract 12124), went back to ACTG 175, an old trial of AZT, ddI and ddC, to remind attendees that people with both a viral load below 10,000 and a CD4 count above 300 nearly always remain free of major, AIDS-defining conditions over a three-year period.
Short of eradication, the big rationale for maximizing HIV suppression is that it prevents or delays the evolution of drug-resistant HIV. Yet there are other strategies for avoiding drug resistance. The most obvious is to put off taking the drugs. You can afford to do that if you are basically healthy and if your viral load and, especially, your CD4 count are stable.
In a plenary address to the Geneva conference (oral presentation 159), activist Mark Harrington described how he started therapy only after his health had deteriorated and his CD4 count had crashed. By then protease inhibitors were on the market. "I'm lucky I waited," he said, showing before-and-after slides of his lymph nodes. By the time he began treatment, HIV had seriously eroded their structure. Now after two years with the lowest possible viral load, that structure seemed essentially restored.
Indeed, the best news at Geneva was likely that the immune system seems capable of slowly repairing much of the damage it has suffered (with the possible exception of the anti-HIV immunity lost early on). According to Dr. Autran, who also described the experience of a large, 300-person group with AIDS on successful therapy, the reestablishment of the CD4 cell population depends mainly on the level of viral suppression. Those with late-stage disease can slowly regenerate lost cells, too, though it may take six or seven years to reach a normal count. The trick is to know how to successfully administer anti-HIV regimens, not rush to take them at the earliest possible moment.
Delaying therapy may buy you a couple of years while the scientists elaborate wiser treatment strategies that more effectively use the available drugs. But what if your viral load already is rising and your CD4 cells declining? If you face present or imminent illness, you can't wait for treatment, but you can still develop conservative strategies that both improve your health and preserve future treatment options. The strategy would be to use as few drug classes per regimen as possible, carefully choosing agents so as to avoid sparking resistance to other drugs you might want to use later on. With 11 anti-HIV drugs now on the U.S. market and another four available through expanded access, many such narrow but potent combinations are conceivable. Geneva provided several illustrations of what can be done.
One possible regimen is a combination of d4T and ddI, with or without hydroxyurea. d4T and ddI are less susceptible to drug resistance than many competitors, and hydroxyurea helps to potentiate their action, even against drug-resistant HIV. In a 117-person Argentinean study presented in Geneva, d4T/ddI/hydroxyurea performed almost as well as many protease combinations in people without prior treatment and low viral loads. d4T/ddI by itself was also quite respectable. (See following article for a full discussion of the various hydroxyurea trials presented at Geneva.)
Other examples are efavirenz plus indinavir or another protease inhibitor; the abacavir/AZT/3TC triple-nucleoside analog combination; and several double protease-inhibitor regimens. With little overlapping resistance, these combinations can be gradually sequenced when HIV starts rebounding, allowing many years of health. (Treatment Issues will describe many of these regimens in greater detail next month.)
It is conceivable that you could also gain extra years of stability by sticking with a combination that used to work better but now allows modest viral loads. Reports from San Francisco General Hospital and elsewhere have indicated that this is so. But you might also be allowing the development of further drug resistance. It is hard to say which way the advantage lies, and the research presented in Geneva was of no help in planning long-term treatment strategy.
Geneva may have been the conference that raised more questions than it had answers, but this high question-to-answer ratio is more than just an historical accident. Plenty of short-term viral load response trials have been done, but the more general trials that take up the larger questions are still a matter of discussion.