Self-generated persistent random forces drive phase separation in growing tumors.

Area of Science:

• Biophysics
• Computational Biology
• Cancer Research

Background:

• Solid tumors exhibit complex cellular dynamics, with distinct behaviors observed in the core versus the periphery.
• Traditional methods for quantifying heterogeneity, like mean square displacement, can obscure intricate collective cell movements.

Purpose of the Study:

• To investigate the heterogeneous dynamics within a single solid tumor.
• To explore the underlying mechanisms driving intratumor heterogeneity using advanced computational techniques.

Main Methods:

• Utilized t-distributed stochastic neighbor embedding (t-SNE), an unsupervised machine learning algorithm, for dimensionality reduction.
• Analyzed the phase space structure of evolving cell colonies, considering cell division and apoptosis.
• Quantified differences in the persistence of active forces generated by cell division.

Main Results:

• t-SNE revealed that the cell population partitions into distinct sets, primarily comprising core and periphery cells.
• Demonstrated non-equilibrium phase separation driven by variations in the persistence of cell-generated active forces.
• Highlighted extensive heterogeneity not apparent with traditional averaging methods.

Conclusions:

• The study reveals distinct dynamical phases within solid tumors, correlating with core and periphery cell populations.
• Differences in active force persistence are identified as a key driver of this observed phase separation and intratumor heterogeneity.
• t-SNE analysis of experimental imaging data offers a powerful approach to understanding the origins of intratumor heterogeneity.