A slew of mainstream media articles have now picked up on Jon Cohen’s Science Insider piece from last week. As of right now, only one thing can be said with certainty: it would have been a good idea for the RV144 investigators to have mentioned the per protocol analysis in their initial announcement, even if only to state that it would be described in detail when the full results are presented (they have now added such a statement to the US Military HIV Research Program website). The lack of disclosure has understandably raised suspicions, but none of the additional information that has trickled out (from unnamed sources) really sheds light on the extent to which the per protocol analysis undermines the overall conclusion of marginal efficacy.
The article with the most additional information appeared in the Wall Street Journal on Monday. It reveals that an analysis described as “close to per protocol” reduces the number of HIV-infected trial participants that can be included from 125 to 96. A strict per protocol analysis reduces he number further, to 86. Statistical significance evaporates in both of these analyses, and efficacy is reduced to 30.9% and 26.2% respectively (the latter figure is now being extensively cited in news stories).
In the first instance, the similar efficacy estimate suggests that the result simply reflects the erosion of statistical power that inevitably comes when the number of endpoints (in this case, individuals who acquired HIV infection during the trial) is reduced. The number people eliminated from the per protocol analysis does seem quite large; although the subset of participants who became HIV-infected may not reflect the overall population, at worst it suggests close to a third of trial participants did not adhere entirely to the protocol. What appears to be most troubling observers is the apparent reduction in efficacy seen in the strict analysis; as articulated by Paul Sax in an article on Medscape: "the study subjects who followed the protocol most closely had less protection from the vaccine strategy."
However, the Wall Street Journal article lacks potentially important information, including whether the HIV-infected participants censored from these analyses are equally distributed between vaccine and placebo groups and the overall reduction in numbers in both arms. This information is critical for evaluating the impact of the per protocol analyses on the interpretation of the publicly reported intent-to-treat result.
The presentation of the data is scheduled to occur at the AIDS Vaccine 2009 conference in Paris next Tuesday, October 20 at 10am. The press conference discussing the results will occur at 1pm on the same day (schedule appended below), and live streaming will be available on the conference website. Press conference presentations will also be available for download an hour prior. The webcast of the conference session goes online at 10:50am (5:50am EST) on Wednesday October 21. It has also been reported that the paper describing the results may be published next week, possibly in the New England Journal of Medicine, but as far as I know this has not been confirmed.
PRESS CONFERENCE SCHEDULE
AIDS Vaccine 2009
All press conferences will be held in the St. German des Pres room, Marriott Rive Gauche Conference Center
Embargoes on research presented in AIDS Vaccine 2009 press conferences lift at the end of the press conference or at the end of the conference session in which the study is presented, whichever is earlier
Tuesday, 20 October, 1:00 – 2:00 pm (13.00 – 14.00)
AIDS Vaccine Clinical Trials Update
CHAIRED BY: Alan Bernstein, Global HIV Vaccine Enterprise New York, NY, USA & Jean‐François Delfraissy, French National Agency for Research on AIDS and Viral Hepatitis (ANRS), Paris, France
Phase III trial of HIV prime‐boost vaccine combination in Thailand: result of final analysis (Abstract SS01‐05)
Supachai Rerks‐Ngarm, Department of Disease Control, Ministry of Public Health Nonthaburi, Thailand
Nelson Michael, U.S. Military HIV Research Program, Rockville, MD, USA
Post‐infection cellular immune responses in recipients following ALVAC‐HIV® + AIDSVAX® B/E prime‐boost vaccination in the Thai phase III trial (Abstract OA04‐06 LB)
Jerome Kim, U.S. Military HIV Research Program, Rockville, MD, USA
Immunogenicity of ALVAC‐HIV® (vCP1521) and AIDSVAX® B/E prime boost vaccination in RV144, the Thai phase III HIV vaccine trial (Abstract OA07‐04 LB)
Mark de Souza, U.S. Military HIV Research Program/AFRIMS, Bangkok, Thailand
Recent immunologic findings from the Step and related HIV vaccine clinical trials (Abstract SS01‐03)
Nicole Frahm, HIV Vaccine Trials Network, Fred Hutchinson Cancer Research Center and University of Washington, Seattle, WA, USA
Comments