Merck Vaccine Moves Forward With Efficacy Trial, While Special Committee Tries to Salvage Thai Study
“Going for Broke in Bangkok”
At the recent meeting of the NIH’s AIDS Vaccine Research Working Group (May 27-28, 2004), there were discussions of two important developments in the HIV vaccine field. Firstly, Merck publicly announced plans for a “phase IIB” efficacy trial of their adenovirus-based HIV vaccine candidate. Secondly, a subcommittee of the AVRWG presented a series of recommendations aimed at improving the highly controversial phase III vaccine efficacy trial that the U.S. military HIV research program recently launched in Thailand.
IIB Is to Be
Details of the Merck trial were presented by Robin Isaacs, who has replaced Emilio Emini as the Executive Director for vaccine research at the company (Emini has joined the International AIDS Vaccine Initiative and is also a new member of the AVRWG). Isaacs outlined the phase IIB trial design, which is intended to answer the question of whether the HIV-specific CD4 and CD8 T cell responses induced by the adenovirus vaccine can provide protection against HIV infection and/or reduce post-infection viral load in vaccinees that subsequently become infected. Isaacs emphasized that this represents a “proof-of-concept” study and is not designed to lead to licensure of the vaccine — an additional efficacy trial would be required for Merck to seek FDA approval. The trial will be a collaborative effort between Merck and the NIH-sponsored HIV Vaccine Trials Network (HVTN) and the slated start date is the end of 2004.
The trial will recruit 1,500 individuals aged 18-45 years at a high risk for sexual exposure to HIV infection; individuals whose primary risk factor is intravenous drug use are excluded. Merck is aiming to enroll at least 350-450 women and study sites will be located in the Caribbean and South America in addition to North America. Immunizations with the adenovirus construct (most likely containing gag, pol and nef antigens from HIV — a final decision on which antigens to include had not been taken by Merck at the time of the meeting) will be given at weeks 0, 4 and 26.
A total of 50 events (HIV infections) are anticipated over the 3.5 year duration of the study; Isaacs reported that 30 events would provide 80% power to detect a >1 log difference in post-infection viral load between vaccinees and placebo recipients. A total of 50 events would provide 80% power to detect 60% efficacy of the vaccine in preventing persistent HIV infection (essentially protection against HIV infection). These assumptions are based on an anticipated incidence of HIV infection of 2.5% per year among gay male trial participants and 1.25-1.5% among heterosexual women enrollees. The predicted drop-out rate is 10% per year for the first year of the study and 5% per year thereafter.
The rationale for the trial is based on the relatively impressive immunogenicity results obtained in Phase I and II studies of the adenovirus construct. There is, however, a significant caveat to these data. Because adenoviruses of the serotype used in the vaccine (serotype 5 or Ad5) are common in nature, many people have been exposed to the virus (which causes severe colds) and therefore have high levels of anti-Ad5 neutralizing antibodies. In the phase I and II studies, the ability to mount a T cell response to the HIV gag antigen contained in the Ad5 construct was severely compromised in individuals with anti-Ad5 neutralizing antibody titers greater than 200. Conversely, 63-75% of individuals with titers below this cut-off were able to mount a sustained gag-specific T cell response, representing the best results seen to date with any vaccine aiming to induce T cell immunity (as a comparator, ALVAC typically induces HIV-specific T cell responses in around 20-30% of recipients and these responses are often not sustained).
These observations have led to an additional entry criterion for the phase IIB trial: individuals with baseline anti-Ad5 neutralizing antibody titers over 200 will be excluded. It is estimated that about a third of the population in North America falls into this category; in the developing world the proportion may be upwards of 90%. While it may seem mystifying that Merck has taken such pains to develop a vaccine candidate that will be essentially useless in most of the world, this is due in part to macaque data that suggested that giving a DNA vaccine as a prime followed by the Ad5 vector as a boost could surmount the problem of anti-Ad5 antibodies.
Unfortunately, as Isaacs showed, studies in humans failed to duplicate these findings: among individuals with anti-Ad5 antibody titers over 200, 7/20 (35%) that received DNA priming followed by Ad5 boosting mounted a gag-specific T cell response versus 5/18 (28%) that received both Ad5 priming and Ad5 boosting. In light of these facts, Isaacs stressed that the purpose of the phase IIB study is to answer the question of whether HIV-specific T cell responses can play a useful role in preventing or ameliorating HIV infection, under conditions that are deliberately designed to maximize the potential for vaccinees to develop such T cell responses.
Given that the vast majority of current HIV vaccine candidates are intended to induce T cell immunity, data from Merck’s phase IIB trial could clearly have a broad impact on the vaccine field as whole. Merck is also developing adenovirus vectors using rare serotypes that are less susceptible to the problem of pre-existing neutralizing antibodies; a successful outcome to the phase IIB trial would provide additional impetus to advance these candidates.
Members of the AVRWG soundly endorsed the concept for the trial, although Jerry Sadoff (head of the Aeras Global TB Vaccine Foundation and formerly at Merck) expressed concern that the study may be statistically underpowered to achieve its goals. Bette Korber (Los Alamos HIV Database) felt that the statistical plan was workable but “marginal,” which suggests that the study could be undermined if assumptions regarding the number of events and drop out rates prove incorrect.
Hammering the Thai Trial Design
At the previous AVRWG meeting in January, committee members recommended that a subcommittee — chaired by Scott Hammer and comprising Larry Corey, Jerry Sadoff and ad hoc statistics adviser Steve Self — review the design of the recently initiated phase III vaccine efficacy trial in Thailand and suggest improvements. The study is comparing a prime-boost combination of Aventis Pasteur’s weakly immunogenic ALVAC vCP1521 canarypox vector and VaxGen’s seemingly inert gp120-containing AIDSVAX construct to placebo in a low-risk rural cohort of 16,000 Thais. The idea came from the U.S. military HIV research program which was recently reintegrated into the Division of AIDS at the NIH; unfortunately part of the politicking that surrounded the reintegration involved DAIDS committing to support this $120 million trial to completion.
At the May AVRWG, Scott Hammer presented the recommendations of his subcommittee to the larger group (the trial’s co-principal investigator Debbie Birx was also in attendance). Hammer broke down the subcommittee’s discussions into four key questions and presented their suggested answers:
- Should the primary (acquisition) and the secondary (amelioration of infection) objectives be switched or should the amelioration of infection objective be made co-primary?Yes, the amelioration of infection objective should be made co-primary — 50 viral load endpoints would provide 90% power to distinguish a 1 log difference in viral load and 80% power to distinguish protective efficacy of 60% (compared to 90% power to distinguish protective efficacy of 50% in current design). This would result in at least a 50% reduction in sample size.
- Should the HIV-1 RNA and CD4 secondary endpoints be included in a composite endpoint?Not necessary for this trial, but the viral load endpoint in infected volunteers is crucial and should be better defined; e.g., the geometric mean of 2-3 HIV-1 RNA values post-infection to define early “set-point.”
- Should immunogenicity data in real time be generated and supplied to Data Safety Monitoring Board (DSMB)?Yes, data from 200-300 vaccinees and 100 controls would provide information on the activity of vaccine during the trial and should concentrate on T cell responses. Such data, however, should be used for background and not be part of any guideline for prematurely stopping the study.
- Should the DSMB take an earlier look at interim efficacy? Should a futility analysis be framed (e.g., for slower accrual, lower HIV incidence, or higher lost to follow-up rate than predicted)?Earlier or multiple looks at interim efficacy will not provide an advantage but it would be reasonable to consider outlining operational futility for the trial (criteria for early termination of the trial if it is not going to meet its goals).
There followed a lively and occasionally heated discussion of what such changes might mean in terms of logistics and finances. Although most AVRWG members supported Hammer’s recommendations, some concern was expressed regarding the need to obtain revised informed consent from the approximately 3,000 individuals that have already enrolled in the trial. Jerry Sadoff countered that such extensive reconsenting had been successfully performed in at least one prior vaccine efficacy trial. John Moore asked about the potential cost savings that would accompany a reduction in samples size from 16,000 to 8,000 volunteers. Both Debbie Birx and DAIDS vaccine director Peggy Johnston suggested that the cost savings would be minimal while Jerry Sadoff argued that they would likely total at least $30 million.
At the time of writing, the principal investigators of the Thai trial are planning to discuss the AVRWG’s recommendations with their Thai colleagues during the upcoming XV International AIDS Conference in Bangkok (July 11-16). A final decision on implementing the suggested changes to the trial design will not be made until these discussions are completed.