ADVANCING VACCINE VALUE ASSESSMENT THROUGH CROSS-STUDY ANTIBODY KINETICS MODELING: PREDICTING LONG-TERM MENABCWY SEROPROTECTION
Author(s)
Thao M.P. Tran, PhD1, Robin Bruyndonckx, PhD1, Raffaella Lantomasi, PhD, CMD2, Worku Ewnetu, PhD1, Robbert Vandermost, PhD1, Jamie Findlow, PhD3.
1P95, Leuven, Belgium, 2Pfizer, Milan, Italy, 3Pfizer, Tadworth, United Kingdom.
1P95, Leuven, Belgium, 2Pfizer, Milan, Italy, 3Pfizer, Tadworth, United Kingdom.
OBJECTIVES: To develop a Bayesian non-linear mixed-effects (NLMM) model that integrates sparse multi-study data to characterize post-vaccination antibody kinetics and extrapolate long-term persistence.
METHODS: Antibody titres (Ab) were modelled using a piecewise power-law NLMM model reflecting biologically plausible kinetics: post-dose increases, peak responses, and decays consisting of fast and slow components. We analysed data from trials with 0-6, 0-12, and 0-36 month vaccine schedules of MenABCWY (Penbraya™; Pfizer Inc, New York, NY, USA) vaccine using a shared structural model with study-specific parameterization. A Bayesian framework was used to overcome identifiability constraints arising from limited blood sampling. Pre-second-dose parameters were based on estimates from the 0-6 month trial (information borrowing) and informative priors were applied where necessary.
The 95% credible intervals (CIs) were obtained directly, allowing the estimates of uncertainty without the use of additional techniques (bootstrapping, etc). Model adequacy was assessed via observed-predicted plots and residual diagnostics. Long-term projections of Ab were generated after applying a baseline standardization strategy.
RESULTS: The Bayesian NLME model achieved stable convergence and adequately described Ab kinetics across dosing schedules. Incorporation of prior information resolved identifiability challenges and enabled estimation of post-second-dose kinetics in sparse datasets. Projected results showcase the benefit of extended interval schedule on persistence of seroprotective titres strain B44 used in the human serum bactericidal assay. Seroprotection (% of subjects with an hSBA titer ≥8) at 5 years post-dose 2 (PD2) was higher for the 0-,12 schedule (23.3%) compared to the 0-,6 schedule (13.%). For the 0-,36 schedule, seroprotection rates PD2 were the highest (54%) among three schedules.
CONCLUSIONS: The proposed approach with shared structural assumptions and cross-study information borrowing enables robust estimation and extrapolation of antibody kinetics. The data application showed greater antibody persistence up to 5 years after 2 doses of Pfizer MenABCWY when administered in an extended interval.
METHODS: Antibody titres (Ab) were modelled using a piecewise power-law NLMM model reflecting biologically plausible kinetics: post-dose increases, peak responses, and decays consisting of fast and slow components. We analysed data from trials with 0-6, 0-12, and 0-36 month vaccine schedules of MenABCWY (Penbraya™; Pfizer Inc, New York, NY, USA) vaccine using a shared structural model with study-specific parameterization. A Bayesian framework was used to overcome identifiability constraints arising from limited blood sampling. Pre-second-dose parameters were based on estimates from the 0-6 month trial (information borrowing) and informative priors were applied where necessary.
The 95% credible intervals (CIs) were obtained directly, allowing the estimates of uncertainty without the use of additional techniques (bootstrapping, etc). Model adequacy was assessed via observed-predicted plots and residual diagnostics. Long-term projections of Ab were generated after applying a baseline standardization strategy.
RESULTS: The Bayesian NLME model achieved stable convergence and adequately described Ab kinetics across dosing schedules. Incorporation of prior information resolved identifiability challenges and enabled estimation of post-second-dose kinetics in sparse datasets. Projected results showcase the benefit of extended interval schedule on persistence of seroprotective titres strain B44 used in the human serum bactericidal assay. Seroprotection (% of subjects with an hSBA titer ≥8) at 5 years post-dose 2 (PD2) was higher for the 0-,12 schedule (23.3%) compared to the 0-,6 schedule (13.%). For the 0-,36 schedule, seroprotection rates PD2 were the highest (54%) among three schedules.
CONCLUSIONS: The proposed approach with shared structural assumptions and cross-study information borrowing enables robust estimation and extrapolation of antibody kinetics. The data application showed greater antibody persistence up to 5 years after 2 doses of Pfizer MenABCWY when administered in an extended interval.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
CO228
Topic
Clinical Outcomes, Methodological & Statistical Research, Study Approaches
Topic Subcategory
Clinical Outcomes Assessment, Relating Intermediate to Long-term Outcomes
Disease
Vaccines