A Bayesian modelling framework to improve antibody titer estimation applied to RSV dilution series data

Area of Science:

• Immunology
• Virology
• Statistical Modeling

Background:

• Accurate measurement of neutralizing antibody (nAb) titers is crucial for understanding population immunity and vaccine development.
• Existing methods like the Kärber formula and four-parameter logistic (4PL) model exhibit batch-level biases and experimental noise in nAb titer estimation.
• Respiratory syncytial virus (RSV) foci reduction neutralization tests (FRNTs) are commonly used but susceptible to these inaccuracies.

Purpose of the Study:

• To develop and evaluate a novel Bayesian hierarchical model (BHM) for estimating nAb titers.
• To correct for batch effects and experimental variation in FRNT data.
• To compare the performance of the BHM against traditional methods (Kärber, 4PL) using simulated and experimental data.

Main Methods:

• Development of a Bayesian hierarchical model (BHM) to estimate nAb titers.
• Evaluation of model performance using simulated FRNT data with known truth.
• Analysis of experimental FRNT data to assess population-level immunity measures.

Main Results:

• The BHM demonstrated superior accuracy in simulated data (Spearman $\rho$ = 0.96, RMSE = 0.41) compared to Kärber ($\rho$ = 0.63, RMSE = 1.64) and 4PL ($\rho$ = 0.87, RMSE = 1.09).
• The BHM significantly reduced false negatives (0.93%) compared to Kärber (9.85%) and 4PL (2.42%).
• Population immunity metrics (GMTs, seroprevalence, seroincidence) derived from experimental data varied substantially based on the estimation method.

Conclusions:

• The developed BHM provides a more accurate and robust estimation of nAb titers, correcting for batch effects and experimental noise.
• The BHM outperforms conventional methods in both accuracy and reduction of false negatives.
• This Bayesian framework is adaptable to other antibody assays with dilution series data, enhancing the reliability of immunological assessments.