Multiparametric classification links tumor microenvironments with tumor cell phenotype.

Area of Science:

• Cancer Biology
• Microenvironment Dynamics
• Tumor Metastasis

Background:

• Tumor microenvironment significantly influences metastasis likelihood.
• The precise link between microenvironment and distinct tumor cell phenotypes remains poorly understood.

Purpose of the Study:

• To investigate how the tumor microenvironment controls two distinct tumor cell motility phenotypes crucial for metastasis.
• To establish the relationship between these phenotypes, extracellular matrix degradation, and metastatic dissemination.

Main Methods:

• High-resolution multiphoton microscopy in mouse mammary carcinoma models.
• Systematic analysis of microenvironmental components and their correlation with tumor cell phenotypes.
• Application of Support Vector Machine (SVM) algorithm for nonlinear classification of microenvironments.
• In vivo photoconversion and monitoring of tumor cell fate and extracellular matrix (ECM) degradation.

Main Results:

• Two distinct motile tumor cell phenotypes (fast and slow locomotion) were identified in spatially segregated tumor microenvironments.
• Only slower tumor cells exhibited invadopodia, which were essential for ECM degradation and dissemination.
• Machine learning successfully classified heterogeneous microenvironments, predicting motility phenotypes and tumor cell fate.
• Inhibition of metalloproteases blocked ECM degradation and lung metastasis, confirming the role of invadopodia and ECM degradation.

Conclusions:

• Distinct tumor cell phenotypes are controlled by specific, spatially heterogeneous microenvironments within primary tumors.
• Machine learning provides a powerful tool for classifying in vivo microenvironments and predicting tumor cell behavior and metastatic potential.
• Understanding these microenvironment-phenotype links is critical for predicting metastasis and heterogeneity in treatment response in breast cancer.