A Predictive Model of Vaccine Reactogenicity Using Data from an In Vitro Human Innate Immunity Assay System.

Area of Science:

• Immunology
• Computational Biology
• Vaccinology

Background:

• Vaccine safety is paramount, necessitating methods to predict excessive immune reactions.
• Early prediction of vaccine reactogenicity can streamline the development of novel vaccine candidates.

Purpose of the Study:

• To develop and validate computational models for predicting vaccine reactogenicity and adverse events (AEs) using in vitro immune assays.
• To establish a framework for forecasting local and systemic reactogenicity and specific AEs.

Main Methods:

• Utilized a modular in vitro immune construct with peripheral blood mononuclear cells (PBMCs) from 40 healthy donors, exposed to 10 vaccines.
• Analyzed cell culture supernatants using flow cytometry and multichemokine/cytokine assays.
• Developed machine learning models integrating in vitro data with clinical AE datasets and performed forward validation with a novel vaccine.

Main Results:

• Observed differential innate immune activity and cell viability profiles in vitro.
• Identified IL-1B, IL-6, IL-10, and CCL4 as key biomarkers for AE risk.
• Developed predictive models for local and systemic reactogenicity, with forward validation showing clinically observed reactogenicity within predicted ranges.

Conclusions:

• Presents a viable framework for developing predictive models of vaccine reactogenicity.
• Highlights the potential of in vitro assays and machine learning to forecast vaccine safety.
• Suggests further refinement of models for predicting specific adverse events.