New Model to Better Understand Data from EBA and BA Trials

Research Resource: EBA and BA trials are a mainstay in the early clinical development of TB treatment regimens. TB Alliance worked with the University of Free State and Quintiles, both located in Bloemfontein, South Africa, to redefine the methodology used to analyze data for EBA and BA trials. This has resulted in a more flexible model that provides important insight into the nature of the EBA and BA of TB treatments. The partners have now made the methodology and program available to other researchers to freely use.

EBA (Early Bactericidal Activity) and BA (Bactericidal Activity) trials assess the decline, during the first few days to months of treatment, in CFU count of M. tb (or a surrogate) in the sputum of people with smear-microscopy-positive pulmonary TB—an important predictor of efficacy of new treatments. Working with partners, the TB Alliance applied a new approach to analyzing the EBA and BA data for its trials and is now making it freely available to use to the research community at large.

A new methodology for analyzing EBA and BA data was motivated by the need for a more flexible regression model for CFU count versus time data. Such data have conventionally been modelled using linear, bi-linear or bi-exponential regression, which are adequate to evaluate data from the older drugs, where CFU counts typically decline at a fast rate in the first few days of treatment, followed by a slower rate of decline during a second phase. These current approaches, however, are inadequate for understanding the bi-phasic decline of CFU counts observed for some of the newer agents (e.g. bedaquiline), where the initial decline in CFU counts is slow, but is then followed by a faster decline. The proposed new methodology also handles missing data and "O" values effectively.

In this paper that is now published with open access and which can be accessed here, the authors propose a new bi-phasic non-linear regression model for CFU data, which has been tested against the databases of four EBA trials. The technical report, which served as the basis for this paper and which contains additional detail, can be accessed here. The open-source software to run the OpenBUGS program can be accessed here, and the code for the program is available here, as well as in the technical report.