Previous BCG vaccination associated with variation in Mycobacterial-specific immune response: a modelling study
Sophie J Rhodes (1), Gwenan M Knight (2), Jeremie Guedj (3), Helen Fletcher (4), Richard G White (1)
(1) TB Modelling Group, CMMID, TB Centre, London School of Hygiene and Tropical Medicine, (2) National Institute for Health Research Health Protection Research Unit in Healthcare Associated Infection and Antimicrobial Resistance, Imperial College London, (3) INSERM, University Paris Diderot, (4) Immunology and Infection Department, London School of Hygiene and Tropical Medicine
Objectives: Mathematical modelling could give us mechanistic insight into dynamics of immune response following vaccination and the ability to quantify the differences in these responses attributed to population covariates. We use modelling techniques to investigate the immune response to vaccination with Tuberculosis (TB) vaccine, Bacillus Calmette–Guérin (BCG) as a better understanding of the variation in response is required to develop a new vaccine against TB disease. We aimed firstly to develop an immune model of the immune response dynamics after BCG vaccination in humans, and to calibrate the model to data. Secondly, we investigated whether population covariates helped reduce variability in predicted model parameters.
Methods: We use IFN-γ as a marker of BCG vaccination immunogenicity and as such, use available ELISPOT data on IFN-γ emitting CD4+ T cells over time after vaccination in 55 humans. Human population covariates were: BCG vaccination status (previously BCG vaccinated (BCG:Y) or naïve (BCG:N) at enrolment), time since BCG vaccination (including “never”), gender and monocyte to lymphocyte cell count ratio. The model was a two-compartmental, ODE model describing the dynamics of CD4+ effector and memory T cells incorporating a Gaussian “delay” model representing the delay in initiation of CD4+ responses following vaccination. Nonlinear mixed effects modelling implemented in Monolix[1] was used to estimate population parameters. The analyses conducted were: i) calibrate model to the human data and ii) assess the impact of human population covariates on immune model parameter values. Bayesian Information Criteria (BIC) alongside graphical results were used to assess fit.
Results: Preliminary results suggest i) the immune model with a combined residual error model represented the data well. ii) the covariate BCG status was associated with a significant (p<0.05) difference in immune model parameter values; those in the BCG:Y group showed significant increases in parameters associated with increased baseline and peak of response. All other covariates were non-significantly associated.
Conclusions: This analysis suggests that previous BCG vaccination is associated with durable IFN-γ responses. Vaccine trials may need to stratify by BCG vaccination history. Mathematical modelling has provided mechanistic insight into the variation in immune response dynamics and how mathematical modelling could be a vital tool to accelerate vaccine development.
References:
[1] Monolix: Users Guide v. 4.3.3 (2014).