Skip Navigation


Skip Content Marketing
  • Share this:
  • submit to facebook
  • Tweet it
  • submit to reddit
  • submit to StumbleUpon
  • submit to Google +

AIDS Vaccine Research Subcommittee - September 21-22, 2010

Members present
Nancy L. Haigwood, Chair
James A. Bradac, Executive Secretary
Deborah L. Birx
Dennis R. Burton
Satya Dandekar
Jeffrey D. Lifson
Douglas Nixon
Nina D. Russell
William Snow

Ex Officio Members Present
Lawrence Corey
Bonnie Mathieson (for Jack Whitescarver)
Nelson Michael
Gary Nabel

Speakers Geetha Bansal
Susan Barnett
Alan Bernstein
Peter Gilbert
Glenda Gray
Jerome Kim
Katharine Kripke
John R. Mascola
Michael Proschan
Alan Schultz
Stuart Shapiro Anjali Singh

Discussants Wasima Rida
Donald Stablein

Others Carl Dieffenbach
Peggy Johnston
Karen Goldenthal


I. Welcome, Introductions, and Meeting Overview

Dr. Nancy Haigwood, the AIDS Vaccine Research Subcommittee (AVRS) chair, opened the meeting. Dr. James Bradac reminded the Subcommittee that the next meeting is February 8–9, 2010. No agenda has been established yet.

II. Enterprise Scientific Strategic Plan

Dr. Alan Bernstein presented the 2010 Scientific Strategic Plan of the Global HIV Vaccine Enterprise, emphasizing implementation and the related challenges.


The Enterprise Plan is strategic, not tactical, and depends on both funding strategies and implementation by the research community. The Division of AIDS (DAIDS) is well aligned with and supportive of the Plan. The Plan has two priorities:

  • Optimize the clinical trials endeavor by fully integrating iterative scientific inquiry with product development
  • Harness the full potential of preclinical models and the ongoing revolution in biomedical science

Dr. Bernstein outlined associated issues under each priority. The Plan recognizes the value of fundamental research and reinforces the need to establish a bridge with clinical science. In order to address the slow pace of research, the Plan recommends better integration of the steps of the research process; a comparison of vaccine trials with the drug discovery process could be helpful in addressing this issue.

Implementation opportunities are of concern, and bridging the gaps between basic, preclinical, and clinical research can be difficult. In addition, industry is not as engaged as some would like. Industry can come up with ideas, offer funds, and provide expertise, all of which are essential in developing a vaccine.


Dr. Haigwood asked Dr. Bernstein to address how the various outcomes of the first 5 years will inform the Plan going forward. He replied that collaboration is a factor; his sense is that the field has grown increasingly collaborative. The analysis of RV144 is a model of what the Enterprise stands for: collaboration, rapid sharing of data, and openness. The 2005 Plan also reviewed regulatory issues, which have not been addressed sufficiently.

It was suggested that there will be a rich debate about the balance between vaccine trials that go forward and are informed by the clinical trial community but are not very deep, versus deeper efforts that motivate and engage the basic science community. To capture all of the concepts and attract young investigators, it would be helpful to know the balance between science-laden clinical trials and more streamlined trials. The research plan reflected the views of academic researchers almost exclusively. However, the private sector has a different perspective. Some companies look at the development focus and emphasize getting an answer. There is a tension between what might be a development focus and a research focus.

A Subcommittee member countered that the reality is that private industry wants this, but is not putting up its own money; everything is being funded by the government. Clinical trials need about 2,000 subjects per arm, and industry will not commit to that. There is no biomarker predicting efficacy right now and the mechanism is unknown. Using public money to go into product-development mode in hopes of hitting the target is not right. Many investigators are thinking long-term, and it is not right and a possible waste of funds to try to circumvent the process. The only thing to do is test of concept trials. Industry may put its own funds into another mode, but public funds should go into the standard research process.

There was some disagreement about one Subcommittee member’s contention that few, if any, vaccines have been developed knowing correlates in advance. Another member said that it is indeed necessary to know correlates in advance. Yet another member maintained that the best way to make an HIV vaccine remains unknown. In order to understand what occurs with the immune system when researchers inoculate, there must be science-rich trials.

The Enterprise working groups agreed on the need for science-driven trials and the need to learn about the immune system. The first priority in the Plan sees efficacy trials as part of the discovery process. This does not preclude R01 awards or NHPs. In order to learn as much as possible from the trials, investigators must study the immune system.

A Subcommittee member said that this field has driven investigators’ understanding of the immune system. Yet as they understand more, they see how little they understand. It is important to admit this. Dr. Bernstein agreed. In answer to a question regarding the possible existence of a tactical implementation plan, he said that ARAC would look at implementation by individual funders. Implementation issues are best addressed multilaterally, with industry engagement and focus on how best to accomplish implementation.

Dr. Haigwood observed that NIH did an outstanding job of engaging industry with the early initiatives. She was pleased to hear Dr. Bernstein proposing more industry engagement and participation. There is a need for diversified and increased funding, and it would be a mistake to view industry as participating in only a portion of the process. Industry is science-based, too. It is increasingly difficult to obtain funding, and perhaps additional funding should be channeled across the various levels. Young investigators are choosing to bypass the field due to the lack of funding.

In further discussion, it was suggested that it might be possible to engage industry by providing scientific proof of concept. Industry would never launch a Phase III trial without a clear signal that it was likely to work. Therefore, a Phase II trial providing that proof would engage them. Moving forward, the real issue is that if there is a desire to increase the throughput of Phase II efficacy trials, there is first a need for the efficacy signal. Whether there is an immune correlate is debatable, and that is a discussion point in light of limited resources. A lot of money is spent on storing cells that might not be looked at further. The efficacy signal could spur more intensive collection of samples. Dr. Bernstein noted that that is happening with RV144. It is important to have industry re-engaged. Earlier engagement will lead to more active participation. Industry is watching RV144. In addition, the research community needs to discuss deposition, access, and analysis of data from clinical trials. The Plan talks about rapid access to the data. This should be driven by the scientific community, not by the funders.

Col. Nelson Michael cautioned that because about $100 million has been put into ALVAC by sources outside the Federal Government, it is not accurate to say that this is all funded by public money. One of the other issues is that of what to test. Who is going to make the vaccines if industry is not engaged early? This is not a matter of one or the other. NIH will support science, but there must be more substantial industry engagement. However, it is also important to avoid an unproductive dichotomy. Different types of trials are needed to emphasize different elements of the search for a vaccine.

III. Overview of NIAID Activities in Support of the Plan

Dr. Katharine Kripke discussed current and projected activities to move forward in addressing Plan goals.

Dr. Kripke reviewed the means of integrating discovery into clinical trials, noting that the primary way to accomplish this is through the HIV Vaccine Trials Network (HVTN), along with the Vaccine Research Center (VRC), CHAVI, and the U.S. Military HIV Research Program (MHRP). To promote the diversity of vaccine approaches, NIAID has a number of discovery and preclinical initiatives, some of which accept new applications every year. Community advisory boards are involved at the site, network, and cross-network levels. To foster collaboration across disciplines, there are both grant programs to pull in researchers from related fields and large collaboration programs. NIAID has a history of engaging industry through new mechanisms and continues to do so. Finally, NIAID offers a number of strategies to attract and mentor early-career investigators, such as an early-stage investigator program through HVTN and CHAVI.


When asked about her perceptions of the remaining gaps, Dr. Kripke replied that there is a need for additional resources. With a level budget, NIAID is confronted with more good ideas than can be funded. Additional clinical research will require additional clinical reagents, and the resources for manufacturing are not sufficient.

Dr. Deborah Birx pointed out that there is a gap between prevention and expanding intervention. What seems clear to some regarding biomarkers and measurement is not clear to rest of the field. She would like to see NIAID staff made available to serve as technical consultants and to share their unique knowledge. This can dramatically influence the success of trials.

IV. FY2012 ARAC-Approved Concepts for New Initiatives

Dr. Haigwood introduced the next topic, presentations by DAIDS’ staff of three new vaccine research initiatives that the AIDS Research Advisory Committee (ARAC) had recently approved.

Consortia for AIDS Vaccine Research in Non-Human Primates

Dr. Alan Schultz gave a presentation discussing the consortia initiative, the purpose of which is to elucidate the viral and host events that occur in acute mucosal SIV infection and to determine how those events can be halted or modulated by vaccines. This is a reissuance of an FY11 initiative. A total of $5M  FY12 funds were requested to fund one or two additional cooperative agreement awards.  Dr. Schultz reviewed the history of the initiative and noted that the strategic plan recommends dissecting the earliest events in NHPs following infection at the mucosa. This initiative meets that criterion.  This initiative will reveal what a vaccine must do in order to be successful, so that researchers can begin more rational vaccine design. Applicants will be expected to propose a research plan defining the overall structure of the consortium, establish a scientific committee, and propose an external scientific advisory board.

Innovation for HIV Vaccine Discovery

Dr. Geetha Bansal presented the Innovation for HIV Vaccine Discovery (IHVD) initiative. The objective of this initiative is to stimulate investigator-initiated research on innovative high-risk/high-impact vaccine approaches that may provide long-term, safe protection from acquisition of HIV. This new initiative will support research via the R01 mechanism for up to 4 years/award, each awarded funded at a maximum of $350,000/year or up to $500,000/yr if NHP studies are proposed. The first year total funding for the initiative will be $4M.  Dr. Bansal explained that new paradigms and discovery are still needed. The initiative will attract new investigators and new perspectives to the field and will not require preliminary data. The review criteria will include assessment of the novelty of the hypothesis, with additional consideration given to both potential gain and biological plausibility. Cross-disciplinary collaboration will be encouraged, and the work will be reviewed by a special panel of senior researchers qualified to identify innovation.

When asked how this initiative would be marketed to investigators not already engaged, Dr. Bansal replied that there would be much communication and networking, including at meetings. Dr. Johnston suggested that the issue is the quality of applications rather than the number, along with the mindset of not having preliminary data. Dr. Bernstein interjected that it might take more effort than anticipated to reach the right people for this award.

Center for HIV/AIDS Vaccine Immunology and Immunology Discovery (CHAVI-ID)

Dr. Stuart Shapiro presented the third concept, for CHAVI-ID. The objective of this award is to accelerate vaccine development by supporting research into immune responses that prevent or contain the virus, and generation of immunogen constructs that will induce broad protection. There will be one or two 7-year U19 awards funded at a level of up to $31.5 million/yr. The effort addresses the lack of an immunogen that induces durable, highly effective, broadly protective immune responses. In addition, the required immune responses to prevent of contain HIV infection remain unknown. A large, multi-disciplinary, integrated approach could help to solve this problem. Elements of the investigations may include basic research into how HIV infection might be prevented, rapid translation of new knowledge into truly novel immunization approaches, and proof-of-concept active and passive protection studies in NHPs. It will require significant resources to attract and maintain investigators, and to support fixed costs. If more than one award is made there will be a shared external scientific advisory board to facilitate coordination. The award is not meant to address manufacturing issues, support the conduct of clinical trials, or fund the study of acute infection in NHPs.

The decision to make one or two awards will be made on the basis of peer review. This will likely be a 2012 award, with the FOA advertised within 6–8 months, giving potential applicants sufficient time to respond.

V. Discussion

Dr. Johnston introduced two questions for discussion:

  • Are there portions of the strategic plan that NIAID should tackle beyond what is ongoing or what was presented at the meeting? If so, at what priority?
  • How can NIAID help coordinate/integrate basic/NHP and clinical researchers? Is there a role for AVRS in helping to address this need?

The first suggestion was to discuss what AVRS members thought was not working. One thought was that the programmatic overlap between clinical research and NHP programs has been less than optimal. There have been efforts in some collaborative programs to foster more interaction and collaboration between the NHP and clinical investigators. This has now become more of an operational issue than a matter of the people involved, and it is not clear what needs to be done beyond the various groups monitoring the research that is being done in each area.  It was advised that more effort be made for “cross-fertilization” of staff from the various research groups; to support staff attendance to meetings and workshops dedicated to the spectrum of research areas; basic, preclinical and clinical.
Dr. Satya Dandakar thought targeted workshops could help bring NHP and clinical researchers together in defined areas.

In discussion of the use of NHPs, it was noted that at least one investigator maintains that models should not be used to predict what would be learned in a clinical trial. NHPs provide a number of unique investigative opportunities. One is studying vaccine-relevant basic pathogenesis. A second is the idea of concept validation of a vaccine approach. There may be theoretical reasons to believe that a given path might produce a response, and NHP models can test that. Trying to use NHPs as predictive of efficacy in clinical studies is a murkier area, both conceptually and practically. Use of NHPs as gatekeepers is a separate issue. In regard to clinical efficacy, investigators are only going to be able to validate predictive utility in retrospect.

Dr. Bernstein said he was struck that the Strategic Plan working groups had broad agreement on the valuable role of NHPs in vaccine development. Working Group 5, the early career investigator group, had broad disciplinary and geographic diversity, and had little difficulty in just talking science. He was pleased to see CHAVI and HVTN funding young investigators, because that is funding for the future.

VI. Introduction to Adaptive Trial Designs

The HVTN has been investigating the use of adaptive trial designs as they consider how to move forward most efficiently in the post-RV144 era.  Dr. Johnston explained that most of the AVRS members are not experts when it comes to the intricacies of trial design and statistical analysis. Therefore, the meeting agenda included an introductory session on the topic of adaptive design, with several presentations.

Dr. Michael Proschan, a biostatistician at NIAID, spoke first, providing a brief overview of the topic. Dr. Proschan discussed a few examples of trial design adaptations, such as dropping an inferior arm or adjusting the sample size. It is crucial to be cautious about making such changes, as investigators may find themselves making an adaptation that seems right but does not control the chance of a false positive, resulting in data that could be a matter of chance. Interpretation of results from a trial employing adaptive design can be difficult, depending on the nature of the adaptation.

Many trials are already adaptive, as investigators make changes in the face of low event rates, futility, and efficacy issues. These adaptations seem standard, but other types of adaptations are new to some people. Dr. Proschan gave an example of why investigators might drop an arm, and described the possible problems that could arise. He also gave an example of dosage testing using adaptive design, and explained both valid and invalid ways to change a study population. Again, interpretation of results can be an issue if the adaptations are not designed carefully.

Adaptive design can involve changing the primary endpoint, which was done in an investigation of a treatment for lung lesions. The investigators planned on counting the individual lesions, but instead found that the lesions ran together and were difficult to distinguish. Therefore, the primary endpoint was changed to the volume of lesions. This was a legitimate change because the study remained blinded between the treatment and control groups. The general principle is that investigators can conduct a valid test even after making design changes, provided the changes were made before breaking the blind.


Discussant Dr. Wasima Rida questioned the need to adjust the alpha for multiple comparisons in multi-arm trials. The investigator is interested in each regimen on its own. In a three-arm vaccine trial in which one or more arms are dropped, that does not necessarily affect other arms in regard to the placebo. Dr. Proschan replied that the hypothetical trial is using the same control group, and a bad control group affects all of the arms. There is not a single right answer. Dr. Rida agreed that a bad control group creates an issue, but said that this might close the door on a product that allows the roll-in of new vaccine products. Adjusting the alpha creates the potential for trial sizes that surpass a Phase III definitive trial.

VII. HVTN Proposal for Phase IIB Design

Dr. Peter Gilbert presented the HVTN’s proposal for future Phase IIB trial design. Following the positive result for RV144 and the potential for multiple prime-boost HIV vaccine regimens, there is interest in designing rapid adaptive Phase IIB efficacy trials. The proposed trials will test one or more active vaccine regimens against a placebo.

Dr. Gilbert presented potential schema for a four-arm design and discussed how such a trial might be conducted. The primary objective would be to assess vaccine efficacy (VE) during the first 18 months. The current thinking is to adhere to the two FDA criteria for adaptive design:

  1. The overall type I error rate for testing for VE > 0% is controlled at a pre-specified level
  2. The trial provides a valid/unbiased estimate of VE

The most important goal will be to expeditiously evaluate immune correlates of protection. Other issues include detection of waning vaccine efficacy, the role of boosters in restoring vaccine efficacy, and whether to adapt accrual to evolving efficacy data in the subgroups. The study will use group sequential monitoring, which is important for timely and accurate evaluation of VE.

Dr. Gilbert then reviewed monitoring approaches and issues, criteria for termination of an arm, and the course a vaccine might follow to demonstrate VE. The purpose of monitoring for positive VE is to assess immune correlates. To power immune correlates analyses, investigators will need a fairly large number of breakthrough infected vaccine recipients. In addition, there are blinding issues. Monitoring must be done in a way that maintains the double blind, or the trial must dispense with monitoring and initiate immune correlates analysis at 24 months for each vaccine not weeded out.

Next, Dr. Gilbert discussed the current symmetric design, developed to reduce the chances of a good vaccine being weeded out. There will be 3 years of follow-up for every enrolled subject in order to assess durability. He explained accrual and incidence assumptions, adding that it takes a while to make an immune correlate assessment well-powered.

The designers of the study compared two options: initiating the trial with four arms, versus starting with three groups and adding a fourth arm later. Dr. Gilbert detailed considerations in choosing between the two approaches. The roll-in design might be necessary due to resources. Another option is to maintain a pipeline of potential vaccine candidates in a “perpetual vaccine trial” that rolls in vaccine regimens as soon as they are ready.

It takes about 2 years to detect waning VE. The question is whether to add boosters if waning efficacy is detected. The study team could roll in a new vaccine regimen with three boosters, but it will likely take longer to get results. There is also the risk of culling the highest risk subjects late in the study, and concern about subjects’ availability for a multi-year study.


Dr. Gilbert was asked how the design would be affected if there were, hypothetically, information from another study that might have some bearing on the investigation. He replied that if the investigators learn something external to the study, there is no need to break the blind. They could switch to co-primary endpoints or burden of illness endpoints, for example. If they are completely caught by surprise, however, that is difficult.

Dr. Gary Nabel added that it is important to think carefully about the numbers. It is critical to make sure the study is properly sized. In regard to immune correlates, he wanted to know the extent to which the design relies on placebos versus negatives in vaccinees, and whether it is possible to increase the number of vaccinees in an arm relative to placebos. Dr. Gilbert replied that because only the vaccinees would show an immune response, they would be the only ones for whom the investigators could measure the response directly and the only ones that could be used to predict the subsequent HIV infection. Statistically, it might not be straightforward but would lend itself to progression models. With the surrogate endpoint, the placebo group would give an indication of how the vaccine varies in terms of the resulting response. The use of placebo groups in this manner has been the subject of quite a few papers. It would be important to see what the placebo responses would be. That latter analysis is more relevant and provides more direct guidance. Dr. Nabel asked if it might be possible to increase enrollment in the vaccine arm to improve power in finding a new correlate. Dr. Gilbert responded that that would not work, because by the time the investigation would be showing good efficacy, all of the subjects would have been accrued and immunized. Dr. Nabel then asked if, in a scenario in which the number of participants was not a factor, the study would have more power to get an immune correlate in more vaccinees, relative to the placebo. Dr. Gilbert agreed that that would be the case, explaining that the scenario would lend itself to testing multiple arms instead of a one-to-one vaccine versus placebo study.

Dr. Gilbert was asked about accrual for multiple arms, in that it appeared the study would need more subjects, which would take longer or require more sites. It was not clear what was being gained. Dr. Gilbert said that the assumption is that accrual can be done in 12 months. Dr. Haigwood was concerned about the effort to eliminate an ineffective arm and the timing, given the rolling accrual. Dr. Gilbert explained that that happens after all vaccinations, so there are few if any injections spared. He did not want to allow the weeding out to happen that way.

Dr. Goldenthal asked about the perpetually rolling Phase III trial, and whether an alpha adjustment would be needed due to the rolling placebo group. Dr. Gilbert said that he would prefer to not adjust the alpha and is more concerned about missing promising candidates. However, the alpha would be adjusted, hypothetically, in this scenario, which might not happen in a new study. If the benchmark is 60 percent for 1 year, adaptive design seems good at getting rid of bad candidates, but it is not clear how it helps good candidates. Dr. Gilbert added that he is quite concerned about how quickly the study can pick up waning efficacy. He would like more power calculations on that.

VIII. The South African Perspective

Dr. Glenda Gray, of the University of Witwatersand Perinatal HIV Research Unit, spoke about research in South Africa. She explained that South Africa is an important place for HIV trials, and the need requires that researchers move quickly and consider innovation. Adaptive designs can be challenging to implement. They must be designed carefully and require full scenario planning that has been approved by regulatory authorities, while also employing real-time monitoring of accruing efficacy data and reliance on nimble regulatory and ethical processes. It is important to have a full understanding of what could go wrong, then be ready to address it.

Dr. Gray discussed likely adaptations for South African vaccine trials. The adaptive design element most likely to be used is group sequential design, with premature stopping or re-estimation of sample size. Dr. Gray then noted adaptive proof-of-concept study designs that have been proposed or used, including “drop-the-loser” designs, adaptive dose finding, and an adaptive seamless Phase II/III design. The combined Phase II/III design reduces time by eliminating the gap between the two phases. Drawbacks of adaptive design are more complex statistics; the trial not being flexible, faster, or realistic; and the sample size.

Dr. Gray next discussed challenges in South Africa, including patient recruitment and the long duration of studies. She noted concerns, including various aspects of blinding, and benefits, such as greater speed in making critical decisions and enrollment efficiencies. South Africa meets many of the criteria for multi-group Phase III testing. Enrollment pace can be an issue, but the country has successfully addressed other recruitment issues.

Community advisory boards are well-organized and pivotal for community engagement. Operational challenges include systems for obtaining data and real-time processing of biological samples. Issues of co-enrollment are being addressed through a fingerprint identification program. Dr. Gray concluded by saying that she believes a novel approach will move the field further and more quickly.


Dr. Gray was asked how 50 percent efficacy, in a hypothetical trial with this type of design, would affect the ability of investigators to do a subsequent licensure trial. Dr. Gray said that the issue of the placebo has come up. The concern is about timing. The analysis team must be nimble. There is no point in having an effective product that cannot be used due to licensure issues.

Dr. Johnston said that in the Thai trial, the team had extensive discussions with regulators that gave them a plan and a roadmap. Col. Michael added that guarantees of technology transfer ensured that there was a true partnership. Dr. Gray said there is a need to think more strategically. Investigators are seldom prepared for success, and the production pipeline is slow. Dr. Bernstein observed that preparing for success involves conversations with industry.

Dr. Johnston noted that the team has fully engaged the companies. A comparison was made to the Centre for the AIDS Programme of Research in South Africa (CAPRISA), where a simple standardized trial was conducted quickly and at low cost. It might be easier to have one protocol, but it cannot be used repeatedly; it must be improved. Investigators should know how to build from a correlate point of view. Another factor is the number of people who will become infected if there are no changes made or risks taken. Industry has a risk in moving quickly, but it can also benefit. There is no downside to obtaining new knowledge sooner.

VIII. Discussant Comments

Dr. Stablein observed that there is a tendency to get stuck on words. Currently, all designs have some form of adaptation. Investigators expect to have stopping boundaries for efficacy or harm. The details provide topics for discussion but seldom result in controversy. Over time, these approaches and their acceptability to stakeholders are improving, and their use is to some degree informative about trial outcome (a negative feature) offset by other values. If investigators have a known adaptive design structure, they might learn something about the trial that they ideally would not, but other significant values may override this.

An interesting topic in the statistical world is how to incorporate external information, like using Bayesian design. Bayesian strategies to incorporate external information can be controversial. There can be excellent acceptable clinical applications in adaptive design, but that is not what AVRS is discussing. When it comes to AIDS vaccines, there is not much external information.

The good thing about a high event-rate population is that there is a short accrual period, which leads to a short follow-up period that minimizes temporal trends in infection rates. It is important to appreciate that any strategy to initiate a correlate search during the study risks diminishing the blind. Dr. Stablein cautioned against underestimating the complications from a multi-arm trial.
Other issues include information related to the number of infections. With fixed-duration trials, investigators cannot predetermine information. Unlike with cancer trials, vaccine trials face practical limits on participant commitment, and fixed duration presents issues of confidence level for this reason. Investigators cannot estimate the control arm loss. The RV144 loss to follow-up had the number of infections affected by censoring, and there were 15.6 percent more observation years than designed. The inability to fix the number of infections introduces approximations to the monitoring board.

One recommendation is to move to continuous stopping boundaries. If an extra look is taken due to being near the boundary, there is an invalidation of the discrete boundary spending function approach. Often there is little practical difference between the discrete and continuous bounds. The difference narrows with the increase in inspections. A continuous boundary allows unlimited looks without harm. Dr. Stablein gave a boundary comparison example from Betensky 1998, following the AIDS Clinical Trial Group (ACTG).

In summary, adaptive design can save time when inadequate vaccine candidates are tested. Investigators should plan to stop along the proposed lines and should be willing to delay termination for lack of efficacy until a reasonable percentage of planned infections occur. They should beware of declining infection rates. Per arm costs preclude general approval of k-arm (three- and four-arm) or rolling k-arm trials.

Dr. Rida spoke next. She was impressed with the complexity and ambition of the proposal, and thought that much careful thinking had gone into the design as Dr. Gilbert presented it. She offered two main comments.

The first issue involves when to start interim monitoring for non-efficacy of vaccine. An NCI employee recently wrote a paper on this. He provides an algorithm of when to start interim monitoring for non-efficacy and how to decide whether to stop. The approach recommends starting interim monitoring when 25 percent of expected infections have occurred. If there is a delay expected in the treatment effect, however, that is premature.

She was pleased to see that Dr. Gilbert suggested it start after 59 percent of the expected events occurred. A hazard to starting interim monitoring prematurely is that there could be a situation where if there is substantial heterogeneity and the vaccine is less effective for those at highest risk, the first infections will be from the highest risk populations. If the vaccine is less effective with those at the highest risk, it affects the VE data. It can be argued that high-risk people have more unknowns, so one has to be cautious about weeding out a vaccine too early. Starting at 59 percent is probably reasonable and will also take about 18 months.

Dr. Rida was concerned about what would happen after dropping the vaccine regimen in the middle of a trial. Would investigators tell the participants with the unlucky regimen good-bye, unblind them, inform them about the results, or what? The individuals remaining in the study might have questions that would affect their continued participation.

Her second area of comment had to do with assessing waning vaccine efficacy. VE can vary over time if there is substantial heterogeneity in a population with substantial risk. If a man identifies himself as engaging in receptive anal sex, he might be several times more at risk than a person who is an insertive person. Once an efficacy signal has been seen, it would be valuable to do an analysis that adjusts for baseline risk behavior or covariate. In assessment of waning VE, there should be some measure of risk behavior. This would be a resource issue of some importance. Overall, she was very impressed with the level of thought put into this effort.


There was a dialogue between Dr. Birx and Dr. Gilbert concerning the views that include people who have not been fully immunized. Dr. Birx commented that it seems unfair for investigators not to complete the entire immunization structure. Dr. Gilbert replied that the analysis would be intent to treat or modified intent to treat, and Dr. Birx asked why they would not wait until full immunization. Dr. Gilbert explained that that all endpoints are counted, and efficacy would include all subjects. Dr. Birx suggested if he were to pick the most immunogenic schedule, he may want to wait until subjects have had full immunization before counting endpoints. Dr. Gilbert thought that could open the door to bias.

Dr. Corey added that most efficacy trials use both approaches. Correlates are chosen as a function of what is chosen for evaluation. Dr. Stablein thought Dr. Birx was looking at the early times, and the study plan is to measure at VE = 18 months. If peak immunity is at 6 months, and it drops later, there is not a long-term vaccine. Dr. Corey said that the Kaplan-Meier curves show no evidence of fluctuation over time on HIV vaccines. The RV144 trials have a rather straight line. Similarly, when looking at 0–18 months, the separation is more or less clear before 6 months. It is important to take the available data and make as rational a decision as possible. A frailty factor for heterogeneity does not preclude unbalanced assessment.

Dr. Haigwood noted that the discussion was about eliminating an arm as used in that particular trial, not eliminating products. Dr. Corey wanted clarification on what was meant by stopping a trial. There has been precedent there, which taught investigators the importance of following up on subjects. The implications of what happens when researchers find something, and what the Data Safety Monitoring Board (DSMB) recommends, is an interesting issue that has not been discussed. There are questions about the ethics of the group versus the ethics of the study, for example. It is impossible to anticipate everything.

Regarding trial designs with inadequate power for an additional arm, differences in infection rates can be seen, so that the comparison is of small effect rates in the vaccines. The study design will need a designation of measurement thresholds. The most likely positive outcome is to come up with a product to take into another trial. The decision is then whether to continue the current trial or proceed to a comparative licensure trial. There is significant concern in moving from high-risk to low-risk scenarios, in that it is not clear that the study will reveal any efficacy. This may provide a direction for the field to move, as those are the issues researchers want to examine. There are other issues to be considered. For example, if two arms had similar efficacy, there could be a licensure trial. There should be agreement up front about what the various steps would be. Dr. Haigwood said that she looks forward to problems surrounding observed efficacy in multiple arms of a trial.

IX. Discussion

A suggestion was made to map out specific scenarios of how a design might play out and impact future decision making. It would also be interesting to probe more deeply into some of the operational challenges. The Subcommittee could flesh out a straw-man, multi-arm, adaptive trial to compare against a traditional design study, and have an industry representative in that discussion so the members could have a better idea of industry thinking. This could be a simulation, like a war game. Dr. Johnston noted that while study cost is an issue, AVRS provides technical assessments, and discussion of costs would be a distraction. She wondered if they might have a workshop on some of these issues, separate from a regular AVRS meeting. They could also use a workshop to obtain broader, international input.

Dr. Haigwood thanked the presenters and adjourned the meeting at 5 p.m.


Day Two, September 22

I. Recap of Day 1

Dr. Haigwood welcomed the speakers for the second day and reviewed the highlights of the previous day.

II. Recap of Prior AVRS Discussion on RV144

Dr. Bradac provided a follow-up of the February 2, 2010, discussion of RV144. Much of the discussion at that meeting was about upcoming work. In particular, Dr. Phil Gomez talked about the Product Development Advisory Group (PDAG) recommendations. In considering efficacy from the Thai trial, investigators found that protection was seen in low-risk individuals but not high-risk subjects. PDAG discussed improving the vaccine, but without knowing the correlate, researchers do not know how exactly to improve it. PDAG’s strategic considerations include understanding the immune response, advancing the regimen, and improving the vaccine.

The meeting included representatives from a number of groups, who talked about their ongoing and planned activities. Dr. Corey discussed test of concept trials and sequential adaptive design, and gave his thoughts on ALVAC versus NYVAC plus the protein boost. Dr. Giuseppi Panteleo talked about strategies to improve poxvirus and protein vaccine. Col. Michael discussed proposed clinical trials for RV305, RV306, and the Phase IIB repeat in Thailand with the correct samples. Finally, Dr. Jose Esparza presented the Bill and Melinda Gates Foundation plan to follow up on poxvirus prime/boost vaccines.

After considerable discussion at that previous meeting, AVRS recommended the following:

  • Support for further immunogenicity clinical trials in Thailand
  • General support for reproducing RV144, but limited enthusiasm for a large-scale trial in Thailand, with no details fixed and no consensus
  • Support for sequential adaptive design trials in high-incidence sites where the trials keep changes from RV144 to a minimum
  • Not tipping the balance of research away from basic/discovery science.

Preparations for follow-up trials include:

  • ALVAC, where remanufacture will be required
  • Alternative/improved poxvirus vector
  • Protein components

III. Report from gp120-gp140 Workshop

DAIDS recently held a workshop on gp120 versus gp140 immunogens.  Dr. Anjali Singh, organizer of the workshop, presented a summary.. The goal of the workshop was to evaluate the recent developments of HIV env candidate immunogens, specifically to determine whether the available gp140 trimers are sufficiently better than gp120 monomers in eliciting broadly protective antibodies to warrant addressing the manufacturing hurdles.

Among workshop conclusions were the following:

  • gp140 trimers have provided an incremental improvement in immunogenicity over gp120 monomers
  • monomeric gp120 is the best currently available env protein boost option for the next Phase IIB clinical trial
  • Continued support to design, evaluate, and eventually manufacture HIV env immunogens that induce broadly protective antibodies remains a high priority

IV. Protein for the Next Phase IIB Trial

Dr. Susan Barnett of Novartis discussed the company’s HIV vaccine program. She explained that Novartis has an HIV vaccine design and development team (HVDDT) contract to advance the concept. They have entered into a collaboration with Sanofi Pasteur to confirm and extend the results of the RV144 trial.  The plan is to test Sanofi’s NYVAC vector in conjuction with Novartis subtype C env proteins formulated in in MF59 adjuvant.   

Dr. Barnett reviewed requirements for env selection. She described the process of selecting the clade C isolates and the production of protein. Most of the clade C gp120 envs proteins bind to a battery of neutralizing monoclonal antibodies. Novartis has been making a number of cell lines that will be available in 2011. The goal is to make 50,000 doses of each protein, with stable cell lines, for five to six clones. Novartis will make the final selection of the clade C envs to move into clinical testing by early 2011.

V. Discussion and Next Steps

Dr. David Montefiori, who spoke via conference call, provided an update on his activities. He has organized an effort to sequence the early infection and transmitted founder virus from at least 200 individuals in South Africa, Malawi, and Zambia. This work is in progress. They are finding primarily clade C, with very few non-Cs. There has been one clade Aand two or three recombinants.
Dr. Barnett said that researchers should watch for whether the As are entering the population. Dr. Gray added that subtype C is responsible for most infections in South Africa. Consideration of an A boost was suggested, because of diversity and the desire to increase breadth. Dr. Nabel stated that there is evidence that individuals infected with clade A viruses show the broadest neutralizting activity.  It was also pointed out that the vaccine tested in RV144 contained envelope proteins from two different clades.  Also mentioned was the consideration of the possibility of trials being carried out at a later date at sites where both clade A and C are circulating in the population.  Dr. Haigwood suggested that various envelopes and combinations of envelopes could be tested in small animals to see what types of neutralizing antibodies are induced.  There was discussion concerning where clade A envelopes might be obtained and the impact of including clade A on the timeline.  There was also the suggestion that the the animal immunogenicity testing protocol should include not only the envelope protein immunogens but also the poxvirus vectors that will be used in the planned clinical trial.

VI. Update on RV144 Correlates Analysis

Dr. Jerome Kim gave an update on the RV144 correlates studies. Investigators have found that RV144 vaccinees have CD4+ responses as measured by ELISPOT to 2 distinct peptides (44 and 49) that map to the V2 region of Env that are not found in HIV+ Thais or in HIV infected individuals from the RV144 trial cohort.  Peptide 44 includes the α4β7 binding motif.  RV144 vaccinees show a high T-cell response rate (80%) that is predominantly Env specific CD4+T-cells. Preliminary analysis shows no differences in NK cell function before or after vaccination, while mDCs show significantly higher phagocytosis after vaccination than at baseline.  RV144 vaccinees show Env V2 antibody responses not seen in individuals receiving DNA/Ad5 vaccines.  All IgG subclasses were induced by vaccination, with anti-Env antibodies mapping to V2, V3, C1 and C5 of gp120.  IgG and IgA antibodies to the HSV gD peptide (the leader sequence in the gp120 proteins) were elicited in the majority of subjects.  Neutralizing antibody responses from RV144 against tier 1 and tier 2 viruses was relatively weak compared to Vax003 and Vax004 vaccinees.
Finally, Dr. Kim addressed sequencing of breakthrough infection viruses. There has not been much analysis in this area yet. Preliminary sieve analyses did not show evidence of distinction between vaccine and placebo sequences with respect to peptides 44 and 49.

VII. Report from VRC Passive Transfer Discussion

Dr. John Mascola of NIAID reported on isolation of broadly neutralizing antibodies and potential applications of such antibodies for prevention of HIV infection.  He discussed three prevention strategies that could utilize neutralizing antibodies: topical microbicides, regulated gene expression, and passive infusion of antibodies.

In addition he summarized a workshop held in July, 2010 on clinical applications for HIV-1 neutralizing monoclonal antibodies.  In this workshop, the overall discussions focused on the rationale and impact of testing a broadly neutralizing antibody to prevent HIV-1 infections in humans, and on the clinical, epidemiological, logistic and ethical issues that relate to the conduct of such trials.  There was strong consensus that demonstration of antibody-mediated protection against HIV infection in humans would have positive impact on the HIV vaccine field.  It was felt that manufacturing of monoclonal antibodies and clinical efficacy studies are feasible at this time, but adult and infant studies have unique issue to address that will require prospective consideration.  Dr. Mascola reported that the VRC had initiated production of the VRC01 mAb with the plans of initiating phase I safety and pharmacokinetic trials in early 2012.

VIII. General Discussion

Dr. Haigwood asked that the Subcommittee consider topics for follow-up. She suggested passive transfer. Col. Michael thought that AVRS had gotten a good background on that topic from Dr. Mascola. Mr. William Snow wanted to learn more about various types of dosing trials, especially those with multiple arms. Another suggestion was to examine more tactical and logistical issues related to trial design, product manufacturing, and the like. It might be productive to look at protocol designs. Dr. Corey recommended bringing in industry representatives to discuss where the technologies are headed.  

Dr. Haigwood asked what members thought of having workshops in conjunction with the regular meetings, or having workshops separately with summaries presented at the meetings. Other ideas recommended by Subcommittee members included a presentation on what industry has in the pipeline; more about specific vaccines; other prevention modalities; and the ways in which prevention is having an impact, with the possible influence on vaccine trials.

It was suggested that NIAID should be generating one test of concept trial per year for the next several years, which leads to the question of the Subcommittee’s role in that process. There has been discussion about the clinical trial side and design, but it has not gone into depth. There was a question about whether that was something that fell into the AVRS purview of advising. Dr. Johnston said that NIAID could bring in a cohort of investigators in the NIAID portfolio who are working on designs, so that the grantees themselves could provide more detail. It was noted that not everyone on AVRS had a good sense of the timelines, although the staff does an annual presentation.

The meeting was adjourned at 12:03 p.m.

Last Updated June 27, 2011

Last Reviewed June 21, 2011