Beyond static biomarkers: systems biology and AI for decoding cancer dynamics

Area of Science:

• Quantitative Oncology and Systems Biology
• Biomarker Discovery and Development
• Artificial Intelligence in Medicine

Background:

• Precision oncology currently relies on static biomarkers (e.g., mutations, receptor status).
• These static biomarkers are insufficient for understanding cancer's dynamic behaviors like progression, metastasis, and relapse.
• Cancer is better conceptualized as a dynamic, stochastic biological system rather than a fixed molecular entity.

Purpose of the Study:

• To propose a framework for understanding and modeling cancer dynamics.
• To advocate for the development of future biomarkers that predict cancer's future states and risks.
• To integrate diverse biological and clinical data for a dynamic view of cancer.

Main Methods:

• Development of a mathematically grounded framework using stochastic state-space models and layer-specific dynamical motifs.
• Integration of six biological layers: molecular regulation, cellular plasticity, spatial organization, tumor ecosystem, metabolic-epigenetic coupling, and mechanobiology.
• Modeling longitudinal clinical monitoring as an observation layer, accounting for various sources of noise.

Main Results:

• A conceptual shift from static to dynamic cancer understanding is proposed.
• A formal scaffold is introduced to model cancer dynamics and system instability.
• Evidence for dynamic biomarkers from ctDNA, circulating tumor cells, serial imaging, and adaptive therapy is reviewed.

Conclusions:

• Future biomarkers should estimate transition risk, system instability, and trajectory direction, not just current state.
• AI can aid inference in partially observable dynamic systems without replacing mechanistic understanding.
• Translating dynamic oncology into a clinical discipline requires addressing research, regulatory, and health-equity challenges.