Controversial Nucleoside Study Awakens To A World Strangely Disaffected By Its Prepackaged Proclamation
Beware the Composite Endpoint
“Endlessly scrutinizing mediocrity” is how the 3,000 patient combination-versus- monotherapy mega-trial ACTG 175 was characterized in a July 29, 1991 issue of ACT UP/New York’s Treatment+Data Digest, with what seemed at the time as fitting hyperbole. “It is a widely held belief that no combination of AZT and ddI or ddC or AzDU or FLT or d4T or any DNA chain terminator will ever be enough to hold HIV in check,” it cautioned. “Has the time come,” it asked, “to write off the NIAID’s AIDS clinical trials network as a potential source of relevant research into promising new agents–and timely answers to treatment questions?” [Hmm… if memory serves, members of the ACTG’s Scientific Advisory Board were posing these very questions just last month.]
More sophisticated concerns about statistical power aside, “timeliness” and “relevance” are the key points here. In the rough and tumble world of AIDS clinical care, the $25 million ACTG 175 is sadly reminiscent of the Woody Allen character in his 1973 hit Sleeper where after a Van Winkle-esque siesta Greenwich Village restaurateur Miles Monroe awakens to a plasticized space-aged world where deep fat and smoking are deemed healthful pursuits and sex proscribes physical contact, involving electromagnetic gadgetry instead. As Project Inform ‘s treatment point man and veteran ACTG watchdog Ben Cheng wryly noted, “In the few instances where an ACTG [antiretroviral] study does possess sufficient statistical power to detect a difference between treatment regimens, it’s been designed in such a way that by the time the results are in the answer is no longer relevant to state-of-the-art clinical practice.” With HIV+ folks scrambling to get into expanded access programs, lotteries and even controlled clinical studies for 3TC, Crixivan, Invirase and Ritonavir, what possible impact could a shamelessly spin-doctored press release about AZT+ddI or AZT+ddC combinations have on the current treatment zeitgeist ?
ACTG 175 is, in fact, a textbook example of a study, ill-conceived from the start, that simply grew more cumbersome, more embarrassing and more wasteful with each passing day. The original 175 protocol chair, Harvard’s Dr. Deborah Cotton, understandably chose to step down rather than lead a trial which she thought ill-designed. As she saw it, the other regimens (ddI monotherapy or AZT+ddI or AZT+ddC combo therapy) would have to be dramatically better or dramatically worse than AZT if 175 was to show a difference. Given what we knew of these drugs even back then, such an outcome was highly unlikely. (See below )
Stacking up the Combinations
(peak Reductions of Plasma HIV-1 RNA)
AZT — .5-.7 log
ddl — .3-.5 log
ddC — .3-.5 log
3TC — .5-.7 log
AZT+ddl — .6-1 log
AZT+ddC — .8-1 log
Neviraphine — 1.5-2 log
AZT+3TC — 1.5-2 log
MK-639 — 1.5-2 log
ABT-538 — 1.5-2
(Source: Aaron Diamond Research Center)
“ACTG 175 was designed in the heyday of optimism about surrogate markers,” writes Mark Harrington in his “The Crisis in AIDS Clinical Research.” After the Concorde results were released, however, the ACTG 175 team was forced to revise the over-optimistic protocol design to include clinical endpoints as primary study objectives (although the hard-headed study team retained as their primary “clinical” endpoint a CD4 fall of 50% from baseline). But this was sort of like deciding to add a basement to your house after you’d already moved in and begun the decorating; in the end, the re-re-re-revised protocol for #175 would muster only enough statistical power (and even then only by relying on rosy-eyed predictions for event rates and retention of study participants) to compare “immediate” versus “deferred” combination therapy. Direct comparisons between the four treatment groups (remember the ACTG #229 fiasco?) will only be possible by inference (e.g., by comparing confidence intervals).
Then again, its no one’s fault, really, that more “promising” drugs came along and dulled the young luster of these 1980s nukes. Yes, the study could have been better designed, but who could have predicted the enormity of 175’s shortcomings: a 3-year drop-off rate of about 45% and a thoroughly emasculating power “hit” that renders it incapable of detecting all but the most improbably gigantic treatment effect.
And this brings up the central issue which will continue to vex the conduct of antiretroviral studies for the foreseeable future: in a disease that can take up to a decade to present overt symptomology, patients need to be followed for enormously long periods of time. This requires extraordinary commitment by both the volunteers and investigators alike. And as new “hot” therapies come and go, it is increasingly difficult (as trials are currently–and have been historically–designed) to keep trial participants on the assigned therapy.
Clearly a revolution in the conduct of antiretroviral clinical trials is called for. The expedient ACTG HIV RAC investigators would argue that clinical surrogates such as plasma HIV RNA should be substituted for the slower-to-arrive-at clinical endpoints in small, quick studies. They have, in fact, openly conceded that (witness #175) they are no longer able to conduct clinical endpoint studies, and the ACTG machine has been retrofitted to move full steam ahead with antiretroviral studies powered solely on changes in HIV RNA. In the absence of convincing validation studies, though, such an approach harbors the potential for becoming a house of cards. Would that 175 were at least powerful enough to shed some light on the clinical utility of monitoring changes in plasma HIV RNA.