Statistical and non-Statistical Challenges in a Meta-Analysis of Ad5-vector HIV Vaccine Efficacy Trials
View Presentation View Presentation
*Yunda Huang, Fred Hutchinson Cancer Research Center 

Keywords: Meta-analysis; Adenovirus vector; HIV vaccine; Subgroup analysis; Efficacy trial.

The negative results of three proof-of-concept preventive HIV vaccine efficacy trials, Step, Phambili and HVTN505, pointed to a potential problem that adenovirus serotype 5(rAd5) vector vaccines might increase susceptibility to HIV infection. To advise such a possibility for the HIV vaccine and the general vaccine research community where rAd vectors have been widely explored as delivery vehicles for vaccine antigens, we conducted a meta-analysis of pooled individual participant-level data from these trials and three other pre-efficacy trials. In this talk, we discuss several challenges encountered in the statistical analyses and interpretation of results: 1) the selected trials tested two different rAd5-vectored vaccine regimens in different geographic regions and study populations; 2) like most meta-analyses, this analysis was not prospectively planned; 3) there was limited power for tests of different vaccine effects due to small numbers of infections in some subgroups or time-intervals of interest; 4) early-unblinding of participants in the efficacy trials may have introduced bias in the analyses; and lastly 5) a diverse group of government/industry/academic stakeholders with different focuses were involved. It is critical for statisticians to address these issues properly and present an objective and balanced view in such team-science collaborative studies.