Understanding Adaptive Clinincal Trial Designs
Researchers are looking at new methodologies to make late-stage efficacy trials more flexible and faster
Before AIDS vaccine candidates can be approved and licensed for use, their safety and efficacy must be demonstrated in a series of animal and human studies. The process begins with animal studies and then small Phase I clinical trials that are primarily conducted to assess the safety of the vaccine candidate in humans.
The most promising candidates are eventually tested in larger clinical trials that are designed to determine the efficacy of the vaccine candidate. These trials are typically Phase IIb test-of-concept trials or even larger Phase III efficacy trials (see VAX September 2005 Primer on Understanding Test-of-Concept Trials). Only a handful of efficacy trials have been conducted for HIV vaccine candidates so far, and until recently, none of them yielded positive results. This changed in 2009 when the results from the RV144 trial in Thailand, involving 16,000 volunteers, provided the first evidence of protection against HIV infection through vaccination.
Following these results, many AIDS vaccine researchers and advocates are calling for more clinical trials and more efficient ways of conducting them. The Global HIV Vaccine Enterprise, a research alliance formed in 2003 to accelerate development of an AIDS vaccine, called for the exploration of new approaches to conducting clinical trials in its 2010 Scientific Strategic Plan, launched this September. And at the recently held AIDS Vaccine 2010 conference in Atlanta, there was extensive discussion about alternate clinical trial designs. One approach being promoted by the HIV Vaccine Trials Network, a leading sponsor of AIDS vaccine trials around the world, is a so-called adaptive clinical trial design that can test multiple candidates simultaneously, comparing them to the same placebo group in a randomized, blinded, Phase IIb trial to see if they are able to prevent HIV infection (see VAX October-November 2007Primer on Understanding Randomized, Controlled Clinical Trials). Adaptive trials allow investigators to modify the trial while it’s underway, giving them more flexibility to drop candidates that don’t seem to be working. This type of trial design would not allow for a direct comparison of different vaccine candidates, but it would allow investigators to rank the different candidates based on how well they work.
More nimble trials
So how would the methodology used in adaptive trials differ from that used in earlier AIDS vaccine efficacy trials? In the late-stage vaccine trials conducted thus far, such as RV144 or the STEP trial, a Phase IIb trial of Merck’s HIV vaccine candidate MRKAd5, the efficacy of each vaccine regimen was evaluated by comparing its effectiveness among vaccinated volunteers to that of placebo recipients. The trials were blinded—meaning volunteers were not aware during the trial whether they had received the vaccine or the placebo—but Data Safety Monitoring Boards (DSMB) collected and analyzed safety and efficacy data at pre-specified time points during the course of the trials and could then determine whether the trials should continue or be stopped either for safety reasons or for futility if there was no evidence the vaccine candidate was working. An interim analysis conducted during the STEP trial is what led the trial’s DSMB to recommend stopping immunizations because the data suggested the vaccine candidate was not effective.
But aside from halting a study for safety or futility reasons, AIDS vaccine researchers have had limited ability to respond immediately to any of the interim data. This means that every trial has gone to completion, or near completion in the case of the STEP trial. However, with adaptive clinical trials, more frequent interim analyses could allow investigators to identify promising candidates more quickly and weed out those with no apparent benefit.
If the interim data indicates that a vaccine candidate is clearly not meeting pre-determined efficacy levels, researchers have the flexibility to shrink or drop that arm of the study while continuing the others. For instance, in a trial population with a 4% annual HIV incidence rate and 2,000 volunteers per group, it would be possible to reach a decision point on whether a vaccine candidate is working in approximately 20 months, as long as volunteers are rapidly enrolled in the trial. If this type of adaptive trial design was employed in past efficacy trials, RV144 could have been stopped two-and-a-half years earlier and the STEP trial could have been stopped nine months earlier, according to researchers.
One important caveat of adaptive clinical trials is that they are not suitable for licensure. That means that the results from an adaptive clinical trial could not be submitted to a regulatory body to serve as the basis of getting the vaccine licensed for use. The more frequent interim data analyses that are conducted in adaptive trials, and the flexibility that researchers will have to respond to the data, reduce the overall power of the study, making it more difficult to interpret the results. For that reason, adaptive trials are meant to serve more as a research tool that allows investigators to rapidly prioritize candidates for further study. Those that show promise could then be tested in much larger, more stringently designed clinical trials that could serve as the basis for licensure.