HIV-RNA VIRAL LOAD REDUCTION OF .5 LOG IS PREDICTIVE OF CLINICAL BENEFIT -- FDA ADVISORY CMTE.; FUTURE OF HIV TREATMENT IS INDIVIDUALIZING THERAPY -- FDA's FEIGAL

A .5 log reduction in HIV-RNA viral load is an acceptable indicator of virologic response in clinical trials of experimental HIV drugs, FDA's Antiviral Drugs Advisory Committee agreed at its July 14-15 meeting.

"I concur with my colleagues that the data shows that you need at least one-half log reduction for [a] period of time...to translate into meaningful clinical response. However, that by no means is an optimal target," committee Chairman Scott Hammer, MD, Harvard Medical School, said. Committee members agreed that the greater the reduction in viral load, the more clinical benefit could be expected.

The committee recommendation followed presentations by several sponsors of antiviral drugs that found that the smallest decrease in HIV viral load studied (.5 log) was associated with a reduction in clinical events.

FDA convened the committee to determine whether the group agrees with the agency and industry "that a durable reduction in plasma HIV-RNA is evidence of clinical benefit." Presenters at the meeting included Merck, Pharmacia & Upjohn, Abbott, Agouron, Boehringer Ingelheim, Roche and Glaxo Wellcome.

Committee members agreed that reduction in viral load is predictive of clinical benefit, but stopped short of declaring that HIV-RNA reduction could be considered evidence of clinical outcomes. "I guess I'm just hung up on a laboratory change equaling clinical benefit," committee consultant Fred Valentine, MD, New York University Medical Center, commented. "It certainly correlates very nicely...I'm convinced of that."

Guest statistician Joel Verter, PhD, George Washington University, suggested that response be viewed as a continuum. Verter said that companies should "try to put together an appropriate analytical program that would look at the data available to see if indeed there is a cut or whether it should be more of a continuum." If a threshold of efficacy is apparent, industry should demonstrate how "that cut relates to specific outcomes of mortality, morbidity and quality of life."

The committee agreed that CD4 cell counts remain an important surrogate marker. The sponsor presentations indicated that CD4 levels correlate with clinical outcomes independent of virus levels.

"It's hard to think of a virologically meaningful endpoint without having some component there that says that it should not be associated with a deleterious effect on the CD4 count," committee member Wafaa El-Sadr, MD, Harlem Hospital Center, commented. "Somehow within that virologic response should be at least a maintenance or sustained CD4 count."

El-Sadr proposed looking at a composite response that could include the duration of suppression of viral load, the rapidity of the response, resistance and maintenance of CD4 cell counts.

The committee recommended that clinicians use the standard PCR assay, Roche's Amplicor, which has the sensitivity to detect as few as 400 copies/mL of HIV. Sponsors of clinical trials should use more sensitive assays which can detect levels as low as 20-50 copies, the committee said.

"The data we saw indicated that we should probably use the more sensitive assays for research purposes, but for clinical management," the PCR assay appears appropriate, FDA guest Mark Harrington, Treatment Action Group, said. "I'm comfortable with using that as the current definition of `undetectable'...although I do think that industry needs to work on the more sensitive assays."

"We should be dealing with the assays that are validated under 400 or 500 copies, but rapidly we will be moving toward a 20-50 copy range," Hammer said. "I don't know of one clinical trial that is going on now where pharmaceutical sponsors are not...looking at the ultra-sensitive assay...along with the standard assay."

Studies should follow patients for at least one year to determine durability of response, committee Chairman Hammer suggested. The definition of loss of response would be dependent on the definition of response.

"Returning to within a half-a-log of your baseline certainly would be a loss of response," Hammer said. As far as following patients to assess durability of response, Hammer said "the absolute minimum is going to have to be 48 weeks, but I would suggest it should be longer than that." He recommended "studies with enough flexibility in them to look at patients 48 weeks after the last subject is enrolled" to "give an average follow-up...of 66-72 weeks."

Committee members also cautioned against switching patients' therapy too quickly. "We have to be careful in the design of the studies and in switching and requirements for getting off the trial that we are reflecting what is accepted as good clinical practice," El-Sadr said.

The durability of viral load suppression correlated with a lower clinical event rate, manufacturers' studies indicated. However, durable responses past 24 weeks were rare and evidence of loss of response in patients led to questions as to what the causes of loss of response might be and what action to take when this happens.

"There are many patients with transient loss of response...[who] without a change in therapy appear to recover that response," FDA Office of Drug Evaluation IV Director David Feigal, MD, noted. "What exactly is going on with all of that?" Clinicians "need to have some sense of how long you expect that response to last and what kinds of things need to be done to evaluate what to do when the response is lost."

FDA "is moving away from a label which simply says this agent is approved to treat HIV infection, to a label that would describe the performance characteristics of the product," Feigal told the committee.

"There needs to be some sense of how to use these products clinically to know how long to wait for a response and to have some sense of what the magnitude of response is so you can individualize the therapy to baseline viral load levels," Feigal said. "One conclusion of the [antiviral] division...is that we don't think it's very satisfactory any longer to treat HIV empirically."

In the move toward individualized treatment, "we still need to have controlled trials," Feigal added. "We'll usually not be able to figure out advances just by comparing people to their own baselines" because "baseline characteristics affect the response and because the individual variability for [the assay]...is a larger magnitude than the effects of some of the weaker drugs."

The idea of "looking for responders...would make most clinical trialists and statisticians a little bit queasy, but there are methodologically sound ways to use this tool," Feigal said. "Randomization can be kept intact by taking a look at time to loss of response and the people who never have a response."

"What we're looking at today is really how are we going to use viral load and CD4 counts in our trials in the next period of time," Feigal said. "It's probably presumptuous for us to assume that things have settled down and that we can predict more than a couple years at a time in this business." It "should be quite obvious that the ability and the very way we measure viral load is a constantly moving target."

"We need to make sure that we don't treat optimal [endpoints] as a simple question, because this is a disease that needs to be treated for a long period of time," Feigal concluded. "What is optimal in the short term may not be optimal in the long term. We need to be able to study the tradeoffs of agents with simpler regimens versus maximal therapy....The details are complex but the goal is simple: we want to create the incentives in drug development to find agents that have the longest durable response as possible and preserve the maximal therapeutic options."