CMStatistics 2021: Start Registration
View Submission - CMStatistics
B0585
Title: Viral dynamics as an outcome in HIV therapeutic vaccine trials: From AUC to dynamical modelling Authors:  Rodolphe Thiebaut - Bordeaux University U 1219 INSERM (France) [presenting]
Marie Alexandre - Bordeaux University (France)
Melanie Prague - ISPED - Universite de Bordeaux (France)
Abstract: There is no HIV therapeutic vaccine available yet. The goal of such a vaccine is to allow patients immune systems for controlling viral replication without any antiretroviral therapy. Therefore, the evaluation of the candidates is performed through supervised antiretroviral treatment interruption (STI) where HIV infected patients are vaccinated while treated by antiretroviral treatment and then the HIV viral dynamics is studied during STI. However, antiretroviral treatments can be resumed quickly in case of viral replication for safety issues. The question is how to define and analyze the virological endpoint in such context. Viral load data are incomplete because of censoring due to a limit of detection or the censoring of the follow-up due to treatment resumption (drop-out). We recently proposed to compare the area under the outcome time curve (AUC) through a statistical test based on splines-based mixed-model accounting for both censoring and missingness mechanisms in the AUC estimation. The inferential properties were studied through a simulation study of a two-armed trial and compared to traditional methods. However, the limitation of the method in the context of a high level of drop-out underlines the relevance of replacing a descriptive model with a dynamical model based on ordinary differential equations defined by biological mechanisms that are potentially performing better predictions in such context.