A stochastic simulation-based approach to inform the relapsing mouse model study design for non-clinical assessment of tuberculosis.

Abstract:

The development of new regimens to treat tuberculosis (TB), the disease caused by , is critical to improving patient outcomes and decreasing global infectious disease mortality. Early evaluation of candidate regimens in non-clinical models of TB, such as the relapsing mouse model (RMM), remains an important step in prioritizing the most efficacious regimens for further clinical evaluation. Although RMM studies may be informative, they are also animal-, labor-, and time-intensive to complete and represent a significant investment in time and resources during non-clinical development. Given the strong pipeline of regimens in development, identification of “leaner” RMM studies may have a significant impact on resource utilization, and hence we compared alternative study designs to identify study attributes that can be modified to improve resource use, particularly animal use. By simulating relapse outcomes from “virtual” studies (i.e., groups of mice treated for selected durations with control and hypothetical anti-TB regimens) followed by model-based analysis of the simulated data, we were able to compare the “true” (input) values with model estimates of time to 95% cure probability (T) and assess bias and precision of competing designs. Using this approach, we demonstrated that 28% fewer mice could be used in RMM studies while maintaining low bias and a precision for T estimation within ±1-2 weeks for most regimens. Therefore, it is expected that RMM studies based upon the alternative designs evaluated herein may be employed to promote improved animal stewardship while generating informative data for decision-making.