Reality Check: Planning for Efficacy Rates of 30-50%, Roche Protease Trials Shoot for Distant Stars

Planning for Efficacy Rates of 30-50%, Roche Protease Trials Shoot for Distant Stars

Deja Vu All Over Again

Peter Staley spoke before a meeting of the National Task Force on AIDS Drug Development (of which he is a member) on the 27th of October in Bethesda. While his entire address is too lengthy for full publication, following is an excerpt of his remarks.

There is an oft quoted saying that those who do not learn from history are condemned to repeat it. I believe that we are now in great danger of ignoring the lessons of the past eight years of antiretroviral research on the reverse transcriptase inhibitors as we evaluate a new class of drugs, the protease inhibitors (PIs). Let me be blunt. Fundamental questions about the drugs currently on the market for HIV infection remain unanswered, among them: (1) When to begin therapy; (2) With what agent or combination of agents; and (3) When to switch or discontinue therapy. Except for AZT in advanced disease, we have no idea if any of these drugs extends or shortens the lives of people with HIV. As for the effects of the approved therapies on progression, we know that ddI is better at delaying progression to a new AIDS-defining illness than AZT in AZT-experienced patients. This is not much after eight years of clinical trials. If we are to have better information on the clinical effects of the PIs, we need to change the way we develop drugs for HIV infection. Are we willing to see yet another set of drugs be prescribed to people with HIV while fundamental questions about their clinical utility remain unanswered? I believe that the Task Force should take an active role in shaping the clinical evaluation of this new class of agents.

Roche is conducting two pivotal efficacy trials for its lead PI, saquinavir. The first trial is well underway, principally at US sites. The second, only recently designed, will be an international study with over two thirds of its 3,300 participants enrolling at sites abroad. Both trials were extensively redesigned this summer, partially in response to the concerns raised by TAG and others. The first trial was simplified and shrunk in size; the second, complicated and expanded. The numerous changes in these two protocols are evidence that Roche and FDA are struggling for a resolution to many complex issues: How should PI trials be designed? How should they be controlled? How large should the trials be? What level of treatment effect should the trials seek to detect? Unfortunately, these important questions are being addressed individually and privately by Roche, Merck, Abbott and other PI developers. The time has come for the Task Force to examine PI development as a whole.

Of the two Roche studies, the second, larger, first-line therapy study appears more likely to lead to an answer about clinical benefit. But let me first address the smaller, second-line, ddC-controlled study, which is now well underway. NV14256B enrolls a population similar to that which enrolled in ACTG 155, substituting saquinavir for AZT in its design, and has a similar sample size of around 900. As we all remember, ACTG 155 showed that, overall, there was no difference between AZT alone, ddC alone, and AZT+ddC combination in terms of clinical efficacy, and that the combination was 50% more toxic. The trial apparently lacked the power to make finer distinctions, though unplanned subset analyses suggested that there might be additional benefit (or harm) to combination therapy in, respectively, those with higher or lower CD4 counts at baseline. Surrogate marker changes on saquinavir-containing regimens in ACTG 229 do not suggest that saquinavir will perform much better than AZT did in ACTG 155, so the ability of the NV study to provide evidence of clinical benefit appears slender.

Based on event rates from ACTG 116B, 155 and CPCRA 002, the designers of ACTG 155 assumed that the 12-month progression-free rate on ddC monotherapy would be 75% (that is, that 25% would progress in 12 months). They powered this study to detect an increase in progression-free rate from 75% to 88%. This amounts to a relative reduction in the event rate of 52%. It strikes me as exceedingly naive and over-optimistic to plan for such a dramatic clinical difference when the only second-line study to show clinical benefit to date, ACTG 116B/117, showed a relative reduction of a mere 15% in time-to-AIDS or death. The tragedy here is that if a much smaller benefit (or harm) occurs, the NV study will be unable to measure it, and we will be plunged back into the disputatious world of post hoc subset trend analyses. TAG has asked several statisticians from around the country to look at the Roche phase III protocols for saquinavir and asked them to assess the ability of these studies to answer questions about progression and survival. We believe we have taken the first step by analyzing the Roche protocols. The Task Force should be ready to evaluate the clinical plans of the other protease companies.

Roche’s first-line study, SV14604C (a 4-arm study which will enroll naive patients, CD4 50-350), was designed to have a 90% power to detect an increase in event-free rates over 80 weeks from 75% to 82.5% (a relative reduction of 30%). Differences of this magnitude have never been seen in first-line active-controlled antiretroviral studies! With 750 patients per arm, the SV study is certainly larger than previous antiretroviral efficacy studies, and we salute Roche for at least moving (if not moving far enough) in the right direction towards larger studies. We in TAG, and many statisticians and clinical researchers with whom we have consulted, believe that it would be desirable to design still larger efficacy trials, with longer follow-up, that enroll before accelerated approval – and that are completed afterwards.

Three statisticians who made presentations at the surrogate markers meeting, held just down the road from here a few weeks ago, all agreed that in order to evalute the new virologic markers we are going to have to mount large, prospective ‘clinical difference’ studies. For those of you who may not have been at that meeting, I’d like to quote the conclusion of Dr. Seth Welles of ACTG’s SDAC: ‘Virus load may well prove to explain, in part, the mechanism for drug action; however, current studies of very limited size and moderate treatment effect make it difficult to interpret the impact of adjustment for virus load on the estimated crude treatment effect … larger studies should be planned to produce stable estimates of treatment effects.’ With such an approach, we could reliably elucidate the effects of the PIs on progression and survival and at the same time evaluate the new virologic markers.

My challenge to you is to join in and help to build a future for clinical research on AIDS where people with HIV will get access to new therapies, and will finally know with greater certainty if these drugs will keep them healthier and alive longer. I hope the Task Force will rise to the challenge posed by the development of the PIs: to help foster a research strategy which will answer questions about clinical benefit as well as to evaluate new virologic markers which might one day assist in the expeditious development of future antiretroviral agents for all people living with HIV. Thank you.