Integration of Molecular Signaling into Multiscale Modeling of Cancer - Multiscale Computer Modeling in Biomechanics and Biomedical Engineering - page 375

TGFb Transforming growth factor b

2D Two-dimensional

3D Three-dimensional

TEM Transendothelial migration

1 Introduction

Cancer growth is a multistage complex process originating from molecular and
genetic cell abnormalities [ 1 , 2 ]. Data-driven mathematical/computational mod-
eling has recently gained recognition for its potential to integrate the large volume
of experimental data currently available, to simulate and analyze the behavior of
complex biological systems, including cancer, and to optimize and predict clinical
therapies and outcome [ 3 - 5 ]. Because cancer growth indeed spans multiple spatial
and temporal biological scales (from genes and proteins to individual biological
cells, and tissues, up to the entire organism) [ 6 , 7 ], modeling of cancer across
different biological scales, i.e., multiscale cancer modeling, that accounts for
intracellular signaling dynamics, individual cell properties, and multicellular
tumor growth environment is potentially more appropriate to predict cancer pro-
gression and development and generate experimental intervention strategies.
Focusing on only one scale, as does the vast majority of current cancer models [ 7 ],
simply neglects the correlative dependence and interplay between different scales.
However, since a multiscale cancer model has to quantify parameters on, and
relationships between biological processes that occur at different scales, the
complexity of model development is significantly increased.
There are three main types of modeling approach currently employed in the
cancer modeling community at large: continuum, discrete, and hybrid, and readers
are referred to [ 8 , 9 ] for a detailed discussion on each modeling approach. Briefly,
continuum models benefit from the knowledge gained in fundamental physical
principles [ 7 ], and are capable of capturing larger-scale volumetric tumor growth
dynamics [ 10 ]. However, it is very difficult to use continuum models to explore
heterogeneity in both the tumor and its surrounding microenvironment [ 11 ].
Discrete models can address these shortcomings, since they can work on the scale
of individual cells or a cluster of cells [ 12 ]. Additionally, they can easily incor-
porate biological rules generated from biomedical data. However, a major draw-
back of discrete models is their compute intense nature due to the detail that each
cell is modeled in, which often limits the model to a relatively small number of
cells. For these reasons, hybrid modeling, i.e., the integration of both continuum
and discrete descriptions, currently appears to be a more appealing approach in the
cancer modeling field [ 6 , 12 ].
Agent-based modeling (ABM) is a discrete-based hybrid modeling approach [ 13 ].
In an ABM, agents (often representing individual cells) interact or communicate
with other immediate agents and their common microenvironment according to a