The Retrospective Partner Trials (RPT) HIV vaccine study design for the measurement of vaccine effects on susceptibility and infectiousness.

HIV vaccines possessing only modest, or even poor, ability to reduce susceptibility to infection could be capable of disrupting epidemic HIV transmission, if such vaccines also significantly reduce the infectiousness of those individuals who are infected despite vaccination. Standard vaccine trials designs measure only Vaccine Effects on Susceptibility (VES) to infection (or disease) and, therefore, run the risk of rejecting vaccines capable of halting the HIV pandemic if they have unmeasured, but important, Vaccine Effects on Infectiousness (VEI). The Retrospective Partner Trials (RPT) HIV vaccine study design was developed to measure VES and VEI and the combined Vaccine Effects of susceptibility and infectiousness effects on the basic Reproduction number of an epidemic (VER). The RPT vaccine study design measures VEI and VER by collecting two critical types of information by retrospective contact-tracing of the sexual partners of infected vaccine trial participants. These are: (1) partnership timing information and (2) phylogenetic analysis of HIV samples taken from the HIV infected vaccine trial participants and their sexual partners. This dissertation presents the RPT study design, along with a probability model developed to predict the statistical power the RPT design will have to detect VES, VEI and VER effects. The probability model requires a small set of parameters regarding the population in which the RPT study is being considered. The probability model results indicate that when plausible model parameters are used, an RPT trial conducted in a cohort of homosexual men in the United States should have considerable statistical power to detect important effects for VEI and VER. A Monte Carlo simulation experiment was also conducted using HIVSIM, a discrete-event simulation of an RPT vaccine trial conducted in a homosexual mixing population. The Monte Carlo experiment results indicated that the RPT trial design could produce reasonably accurate and precise estimates of VES, VE I and VER using sample sizes comparable to those employed for current Phase III HIV vaccine trials. The Monte Carlo results also confirmed that when sample sizes were sufficient, the RPT power probability model produced sufficiently accurate statistical power estimates, even when some of the simplifying assumptions of the probability model were violated.PhDBiological SciencesBiostatisticsHealth and Environmental SciencesImmunologyPublic healthUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/132075/2/9959705.pd