Improved inference for vaccine-induced immune responses via shape-constrained methods.

Area of Science:

• Immunology
• Biostatistics
• Vaccine Development

Background:

• Evaluating immune responses is crucial for early-phase vaccine trials.
• Quantifying IgG binding to the V1V2 region of HIV envelope protein is key for vaccine efficacy.
• Previous methods lacked precision and reproducibility in comparing immune profiles.

Purpose of the Study:

• To develop and assess shape-constrained statistical methods for analyzing immune response profiles.
• To compare the performance of unimodal and log-concave shape-constrained approaches.
• To quantify and compare IgG binding immune responses in HIV vaccine trials HVTN 097 and HVTN 100.

Main Methods:

• Development of novel shape-constrained tests for stochastic dominance.
• Creation of shape-constrained plug-in estimators for squared Hellinger distance.
• Application of methods to IgG binding data from early-phase HIV vaccine trials.

Main Results:

• Shape-constrained tests demonstrated superior performance compared to existing non-parametric methods.
• Shape-constrained estimators showed comparable performance to non-parametric methods.
• The developed techniques require minimal or no tuning parameters, enhancing reproducibility.

Conclusions:

• Shape-constrained statistical methods provide a robust and reproducible framework for evaluating vaccine-induced immune responses.
• These methods are particularly valuable for early-phase clinical trials requiring prespecified and reliable analyses.
• The study advances the statistical toolkit for vaccine immunology research.