Improving vaccine descriptions in model-based impact prognosis of new tuberculosis vaccines: removing arbitrariness and reducing bias.
J Sanz(2,4) M Tovar(2,4) Y Moreno(2,3,4)
1:Universidad de Zaragoza; 2:Universidad de Zaragoza - Instituto BIFI; 3:ISI Foundation; 4:Universidad de Zaragoza - Dept. Física Teórica
The development of vaccines against tuberculosis (TB) poses a series of unique challenges when compared to other diseases. Among these, evaluating their efficacy through randomized control trials (RCTs), and mapping it to prospective impact forecasts based on mathematical modeling is one of paramount complexity. Furthermore, it is a task of utmost importance, given the scarcity of resources for vaccine development in the fight against TB.
One reason underlying this difficulty stems from the co-existence of different routes to disease in the natural history of the disease, (primary TB, endogenous reactivation from latent infection, and TB upon exogenous re-infection). This fact makes it challenging to translate RCT-derived efficacy estimates into specific mechanistic interpretations of vaccine behavior needed to inform mathematical models. This problem is especially relevant when comparing impact forecasts for vaccines with different product profile characteristics, and/or tested through RCTs of different architectures.
To address these challenges, here we describe a series of novel analytic approaches to improve the interpretation of the outcomes of clinical trials deployed to estimate the efficacy of novel TB vaccines. The methods discussed combine in-silico tools such as compartmental models and stochastic simulations to disentangle the different possible mechanisms of action underlying vaccine protection effects against TB inferred from trials conducted on either susceptible individuals (IGRA-) or on individuals previously exposed to the pathogen (IGRA+). Our methods unlock he construction of impact forecast metrics that are less subject to bias than previous approaches which typically require the adoption of arbitrary modeling choices about vaccine behavior.