Coupled Cell-Intrinsic and Microenvironmental Heterogeneity Drives Divergent Trajectories in Castration-Resistant...

Area of Science:

• Oncology
• Computational Biology
• Genomics

Background:

• Prostate cancer progression to castration-resistant prostate cancer (CRPC) is complex.
• Understanding the interplay between cell-intrinsic factors and microenvironmental constraints is crucial for dissecting CRPC.
• Current models often analyze these factors in isolation, limiting predictive power.

Purpose of the Study:

• To mechanistically dissect the coupling between cell-intrinsic heterogeneity and microenvironmental constraints in CRPC.
• To develop an integrated multiscale framework for analyzing spatial and molecular axes of disease.
• To enable personalized prediction of androgen deprivation therapy (ADT) resistance risk.

Main Methods:

• Developed an integrated multiscale framework combining a cellular signaling model (MHS) with a spatial agent-based model (ABM).
• Parameterized MHS with TCGA genomic data and used machine learning (SHAP) to identify intrinsic drivers (PTEN, MDM4, AR).
• Validated findings with clinical data (Kaplan-Meier, Cox regression) and simulated tissue-scale progression using ABM with varying microenvironmental factors.

Main Results:

• Identified PTEN, MDM4, and AR as dominant intrinsic drivers of CRPC, significantly impacting overall survival.
• Demonstrated that microenvironmental factors (confinement, androgen uptake, adhesion) modulate disease selection outcomes for identical genetic alterations.
• Developed patient-specific androgen sensitivity ratios to stratify patients by predicted androgen dependence and ADT resistance risk.

Conclusions:

• CRPC emergence is an emergent property of the intrinsic-extrinsic coupling, not predictable by molecular or spatial analyses alone.
• Mechanistic integration of both intrinsic and extrinsic factors is essential for accurate patient stratification and predicting treatment response.
• This integrated approach provides a model-based route from genomic data to personalized CRPC risk prediction.