Abstract

In silico trials, utilizing mathematical models calibrated with clinical data, present a transformative approach to expedite drug development. We propose a virtual trial framework for chronic Hepatitis B, accurately simulating clinical protocols, patient characteristics, and endpoints using a mechanistic mathematical model. Clinical trial simulations with this model successfully captured functional cure with standard-of-care therapies (nucleos(t)ide analogs and pegylated interferon) as well as complex clinical observations, facilitating mechanistic hypothesis generation and suggesting biomarker signatures that may predict treatment outcomes. In silico trials revealed that responders exhibited enhanced cytotoxic immunity and significant serum-alanine transaminase increases, suggesting a potential response biomarker. However, a higher baseline Hepatitis B surface antigen did not proportionately increase cytotoxic antiviral immune responses, indicating a potential immune ceiling but in the face of increasing systemic antigen burden, therefore culminating in a lower treatment response. Virtual patients enabled the generation of large virology biomarker synthetic datasets, which empowered a machine learning model to predict functional cure in virtual patients with similar to 95% accuracy. This underscores the potential of in silico trials in enhancing clinical trials, generating mechanistic hypotheses, and accelerating chronic Hepatitis B drug development.