Hierarchical Poisson Regression Models for HIV Vaccine Studies

• Activity Number: 517
• Type: Contributed
• Date/Time: Thursday, August 10, 2006 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Xin Huang*+ and W. John Boscardin and Elissa Schwartz
• Companies: University of California, Los Angeles and University of California, Los Angeles and Harvey Mudd College
• Keywords: nonlinear regression ; Markov chain Monte Carlo

Abstract

The safety and immunogenicity of two potential HIV-1 vaccines were evaluated by NIH/NIAID randomized blinded studies (AI 50467) comparing mucosal and blood responses of deltoid versus inguinal vaccination. CD8+ cellular immune (CTL) responses were measured longitudinally in the blood versus mucosal lymphocytes, and assayed for frequencies of cells reacting to 53 different peptide pools of the HIV-1 genome (by measuring the frequency of spot-forming cells in the ELISpot assay). The rate of any potential vaccine activity was low compared to the background noise, and it was thus essential to make efficient use of the data. We present a hierarchical Poisson regression model for assessing the de-noised rate of HIV-1-specific CTL responses by immunization site, peptide pool, and time point. Log linear mean structures are used, but the background is assumed additive on the original scale.