Quantitative Calibration of a Spatial QSP Model Identifies Fibroblast Impact on HCC Immunotherapy

Area of Science:

• Computational biology
• Systems pharmacology
• Cancer research

Background:

• Quantitative systems pharmacology (QSP) models simulate tumor progression and treatment responses.
• Spatial QSP (spQSP) models integrate tumor microenvironment (TME) spatial organization.
• Parameterizing spQSP models for accurate tumor representation is challenging.

Purpose of the Study:

• Develop and validate a calibration framework for spQSP models using clinical and spatial molecular data.
• Enable accurate prediction of TME states and therapeutic responses.
• Identify novel spatial and non-spatial biomarkers for personalized cancer therapy.

Main Methods:

• Utilized Approximate Bayesian Computation - Sequential Monte Carlo (ABC-SMC) for model calibration.
• Integrated spatial-omics data (e.g., CODEX) with spQSP models.
• Matched spQSP predictions to patient tumor architectures by fitting cellular neighborhood statistics.

Main Results:

• Successfully calibrated spQSP models using spatial molecular data from untreated HCC patients.
• Demonstrated predictive power of the calibrated model for TME states in patients receiving combination therapy (ICI and TKI).
• Identified and assessed the predictive capability of spatial and non-spatial pretreatment biomarkers.

Conclusions:

• Integrating spatial-omics with multiscale mechanistic models allows for quantitative calibration and biological insight.
• The developed workflow facilitates in silico biomarker discovery for personalized cancer therapy.
• This framework offers a pathway for advancing personalized cancer treatment strategies across various tumor types.