Gilbert PB, Cai T, Self SG, Gurwith M, Francis D; Conference on Retroviruses and Opportunistic Infections (11th : 2004 : San Francisco, Calif.).
Program Abstr Conf Retrovir Oppor Infect 11th 2004 San Franc Calif. 2004 Feb 8-11; 11: abstract no. 283.
Fred Hutchinson Cancer Res. Ctr., Seattle, WA, USA
BACKGROUND: In phase 3 preventive HIV vaccine trials, it is important to assess both vaccine efficacy to prevent infection (VEs) and disease progression (VEp) in subjects who become infected. Because vaccine effects on infection and post-infection endpoints are related, analytic methods are needed that evaluate VEs and VEp jointly.METHODS: A new technique for estimating (VEs,VEp) with a joint 95% confidence region was developed, based on Cox models for the 2 periods: between entry and infection diagnosis (Dx), and between infection Dx and a disease endpoint. The method accounts for correlation in the estimates of VEs and VEp, and was applied to a completed placebo-controlled, double-blind, randomized trial of AIDSVAX B/B. VEs was defined as the percentage reduction (Vx vs Plc) in the risk of HIV infection by 36 months post entry, and VEp was defined as the percent reduction in the risk of viral failure >10,000 copies/mL or treatment initiation by 12 months post infection Dx. The null hypothesis region indicating clinical nonsignificance for (VEs,VEp) was pre-specified (shaded region in figure below); this region reflects a trade-off in which the higher the VEs, the lower the VEp is required to declare efficacy. Simulations were conducted to compare the power of a trial design that assesses both VEs and VEp in the primary analysis vs designs that assess only VEs or VEp.RESULTS: Based on 5403 randomized subjects and 347 evaluable infected subjects, (VEs,VEp) was estimated as (8.5%,-11.3%). The 95% confidence region (see the figure) overlaps with the null region and contains (VEs=0%,VEp=0%), suggesting no evidence for a vaccine effect on infection or disease. The power calculations showed a trade-off between trial designs with primary objective to assess (VEs,VEp) vs designs with primary objective to assess only VEs or VEp: If the vaccine is substantially efficacious in only one efficacy parameter, then the optimal design evaluates VEs or VEp alone in the primary analysis; if VEs and VEp are both substantial (VE >25%), then the optimal design evaluates (VEs,VEp) jointly.CONCLUSIONS: Joint analysis of (VEs,VEp) in the AIDSVAX B/B trial suggested VEs <25% and VEp <20% with high probability. Information on whether a vaccine is more likely to prevent infection, slow progression, or both determines the optimal primary objective of an efficacy trial.
Publication Types:
Keywords:
- AIDS Vaccines
- AIDSVAX
- Clinical Trials as Topic
- Communicable Diseases
- Disease Progression
- Double-Blind Method
- HIV Infections
- Physical Examination
- Placebos
- Research Design
- methods
Other ID:
UI: 102271242
From Meeting Abstracts