Evaluating Chemoradiation Resistance of Head and Neck Cancer Via Multiscale Analysis of Tumor Heterogeneity and Cell Topology
Abstract
Purpose
Understanding chemoradiation resistance remains a major barrier to improving outcomes in head and neck cancer. We developed a multiscale framework linking tumor architecture, cellular network topology, and immune context to therapeutic response. This approach aims to identify topology-driven biomarkers of resistance and survival and to provide interpretable insights into tumor behavior.
Methods
First, mouse oral carcinoma cell lines HPV− MOC1 (n=42), HPV− MOC2 (n=36), and HPV+ MLM1 (n=39) were implanted in the buccal mucosa. Chemoradiation (days 0 and 7: cisplatin 5mg/kg + 8Gy) was initiated when tumors reached >50 mm³, with CT imaging on day 14. Radiomic features were extracted and modeled using LASSO-Cox regression. Second, MOC2 tumors (n=16) received chemoradiation on day 0 and were harvested at days 1, 3, or 7. Third, a macrophage knockout model of MOC2 tumors in Ccr2+/− (n=7) or Ccr2−/− (KO) mice (n=7) was followed to humane endpoint to assess immune heterogeneity. Pathomic features were computed at local (10.5µm) and global (15.75µm) scales using a deep learning algorithm that detects individual tumor cells.
Results
Radiomic analysis identified five CT features indicative of tissue heterogeneity that were significantly associated with survival. Dynamic pathomic analysis showed that slow-growing tumors had higher average degree and k-core values, indicating more organized networks early after treatment. In the macrophage knockout model, fast-growing tumors demonstrated increased local heterogeneity, clustering deviation, and correlation variance, whereas lower eigenvector centrality and correlation mean were linked to improved survival. Highly connected k-core structures characterized aggressive tumors, while increased high-betweenness cells and lymphocytic density predicted better outcomes. Across scales, architectural heterogeneity correlated with therapeutic and immune vulnerability, and topological metrics evolved over time, reflecting dynamic tumor network remodeling.
Conclusion
Chemoradiation-resistant tumors exhibit coordinated multiscale disruptions in tissue heterogeneity, cellular topology, and cell architecture. Topology-informed radiopathomic modeling provides an interpretable framework highlighting spatial organization.