VaxGen Trial Yields Trove of Behavioral and Social Science Findings
By Emily Bass
When data from VaxGen’s completed AIDSVAX® trial were released in February, the media flurry focused on the efficacy results, which were disappointing. Nearly lost in the shuffle were the trial’s unqualified successes with recruitment and retention. Here, data refuted pre-trial concerns about the feasibility of following thousands of high-risk volunteers over three years and seven immunizations. Similarly, early fears about major changes in risk behavior during the trial also proved to be unfounded. As the efficacy data continue to be analyzed and debated (see VaxGen: Are There Hints of Race-Based Effects?), examination of how the trial cohort was managed, and of participants’ motivations and experiences, offers some encouraging lessons—and clear prescriptions—for future Phase III studies.
Incidence rates were higher than anticipated
One concern facing the VaxGen team at the outset was whether they could establish a North American and European cohort with high enough HIV incidence rates for an efficacy trial. Working with two study populations—men who have sex with men (MSM) and high-risk women—the trial’s planners established inclusion criteria intended to select for volunteers at high risk for sexual infection. In the MSM group, this meant using a detailed questionnaire to gather data about sexual behaviors and other risk factors, and selecting individuals who reported anal sex with a male partner within the past year. Men in monogamous relationships with HIV-negative partners for at least 12 months were excluded. For women, enrollment criteria were smoking crack, exchanging sex for money or drugs, having five or more male sex partners within the past year, or having sex with an HIV-positive man within the past month.
Using these criteria, the study population was found at the trial’s end to have an annual incidence rate of 2.7% (2.8% in men; 1.5% in women)—higher than the 1.5% rate used for designing the study protocol. This is not unusual: trial designs generally use conservative incidence estimates, so that the study will retain its statistical power to measure efficacy even if infection rates decrease, for example due to ongoing risk-reduction counseling.
In an effort to recruit a high-incidence cohort, at least one site applied more than one risk criteria to its cohort. “We knew from prior experience that, to recruit high-risk women, trials have to require at least two risk factors,” said Rick Novak, principal investigator at a Chicago site that enrolled about one-third of the female participants, and who helped VaxGen develop the screening protocol for women.
It is possible to recruit and retain cohorts of high- risk women
“Small but mighty” could be the rallying cry for the 308 high-risk women recruited for the trial. Despite their being only about 6% of the cohort, studies of this group yielded a wealth of information about recruitment and retention of a population traditionally considered difficult to retain in long-term studies.
Sites in Chicago, the South Bronx and Boston proved that, with the right strategies, it can be done. “You need a full time person from the community working on retention,” says Pamela Brown-Peterside, principal investigator at the New York Blood Center’s South Bronx site. Extensive contact information is also critical: The Bronx group developed a six-page locator form, while in Chicago, the volunteers—who moved roughly once every six months—provided contact information for a female relative most likely to know their whereabouts.
Tracking women also meant dealing frequently with the police and prison systems. “The police gave us a back telephone number, which we used on many occasions,” says Parrie Graham, the Chicago women site’s study coordinator. Since Illinois Institutional Review Board regulations prohibited visits to incarcerated women, the team stayed in contact by mail, sending magazines, Christmas cards and trial updates, until the women were released.
Retention success also lay in going beyond the study requirements. In the South Bronx, the site made contact with participants every three months, rather than the required six; the Chicago team photographed participants to help its outreach workers locate them for follow-up visits. The results: Follow-up rates comparable to those for the MSM group. Looking beyond trials, Novak points out that women like those involved in the study will also be important groups to reach with a licensed vaccine.
The trial did not eliminate—or encourage—risk behaviors
Alongside VaxGen’s clinical monitoring of the volunteers, researchers from the US Centers for Disease Control and Prevention (CDC) studied participants’ perceptions and behaviors over the course of the trial. Their data—which only include individuals who received all scheduled immunizations—contradict worries that high-risk acts would skyrocket in the context of an efficacy trial. As reported by Brad Bartholow at the Barcelona AIDS conference (abs. #WePdD2104; see figure), participants’ reported risk behavior remained at or below baseline levels 24 months after the trial began.
But looking more closely, there are subtleties within these findings. Rates of any high-risk behavior in both men and women dipped below baseline at intermediate trial visits and then began to climb again, remaining below baseline in men. But in women, the drop-off ended by 12 months, at which point reported risk behaviors began to climb again until they were back to baseline at 24 months. A different pattern emerged for reports of unprotected sex with an HIV-positive partner. Here, rates decreased slightly from baseline and then remained stable out to 24 months in both genders. (Three year findings from the CDC have not yet been made public.) VaxGen’s newly-released behavioral data, which has a 36-month timepoint, is based on all study participants, including those who dropped out or missed visits.
The CDC’s study was designed to answer questions on whether rates of risk behavior were influenced by volunteers’ assumptions about whether they received the vaccine or placebo. Statistical analysis showed that male participants who believed they had received the vaccine reported significantly more frequent unprotected anal intercourse (UA) than those who thought they received a placebo or were unsure; they also showed an upward trend in UA between 12 and 24 months.
These data provide a clear take-home message: Ongoing assessment of beliefs about vaccines, plus appropriate counseling, is crucial during efficacy trials. They also reveal some perplexing trends. For example, women who assumed they received the placebo had higher rates of risky behavior at 24 months than those who thought they were given vaccine, although these trends disappeared at 36 months. The findings also raise thought-provoking questions about whether similar (or more severe) trends might be seen in the context of a partially effective, licensed vaccine—when vaccinees will not be receiving continuous counseling to counteract a false sense of protection.
Could vaccine trial participation actually increase risk behavior? To tackle this question, the CDC is now comparing a group of VaxGen volunteers with a comparable cohort of about 800 people (not in a vaccine trial) meeting similar risk criteria and given the same counseling messages, minus vaccine information.
The trial data also underscore the need for tailored messages. Women were significantly more likely than men to believe that the vaccine was highly effective (defined as 76-100% effective), and to cite protection from HIV infection as a motivation for their participation. Volunteer assumptions or motivations were not analyzed by demographics; however, the women had significantly lower income, education and literacy levels than the male volunteers, factors that may have contributed to specific beliefs. (Men and women volunteers used the same tools for informed consent and knowledge assessment.)
As the IAVI Report went to press, the CDC was preparing an article summarizing the full 36-month data on behavior change and reported rates of risk behavior.