Spatial Aspects of HIV Infection - Mathematical Methods and Models in Biomedicine - page 20

A spatial simulation of T cell motion based on experimental observations was
presented by Beltman et al. [ 10 ]. The authors used a cellular Potts model to describe
T cells and used it to study the activation of naıve T cells by their cognate antigen in
a lymph node. In the cellular Potts model presented by Beltman et al. [ 10 ] T cells are
represented by several sites on a grid. While the volume of the cell is kept constant,
the shape of the cell can vary during a simulation. Interaction of cells in a densely
packed environment causes the movement of cells in certain directions. Beltman et
al. [ 10 ] considered T cells and dendritic cells in their model, as well as the stromal
cell environment within a lymph node. By adjusting the rules with regard to the
movement direction of simulated T cells, the authors were able to reproduce motility
characteristics consistent with those experimentally observed for T cells in vivo [ 73 ,
77 - 80 ]. Without explicit modeling, they were also able to observe the organization
of small T cell streams along fibers of the stromal network in their simulation which
is in line with experimental observations.
Bogle and Dunbar [ 16 ] studied the activation and proliferation of motile T cells
in a lymph node. In contrast to their previous approach [ 15 ] and the other spatial
models presented so far, the authors allow for a dynamic simulation environment:
The size of the simulated deep cortical unit of a lymph node is allowed to expand and
contract to account for the in- and outflow of T cells as well as their proliferation.
The authors show that their modeling platform can reproduce realistic motility
characteristics of T cells and their response to antigen.
One of the most ambitious projects to simulate immune dynamics and interac-
tions is the ImmunoGrid - project funded by the European Commission [ 45 , 99 ].
The project was established in 2006 as part of the virtual physiological
human (VPN) initiative. The aim is to develop an in silico model of the entire
immune system which could be applied to clinical practice in order to make
personalized suggestions for therapy and interventions in different types of diseases
such as cancer. ImmunoGrid is still in its infancy facing challenges that include
the computational infrastructure as well as the complexity of the immune system
itself (intermediate report in [ 45 ]). However, some studies done as part of this
project have already generated results. Baldazzi et al. [ 6 ] and Castiglione et al.
[ 20 ] investigated the effect of HAART on HIV disease dynamics using an agent-
based model. As HAART is associated with side effects and requires a high level of
discipline since patients need to take drugs regularly for years, designing therapeutic
regimes is an area of interest. Baldazzi et al. [ 6 ] use an agent-based model to
determine effective therapeutic regimes for HAART, even allowing interruption of
treatment. The efficacy of therapy is allowed to vary during the simulation due to
the rise of drug-resistant viral strains. In Castiglione et al. [ 20 ], the authors use
this framework to find an optimal regime for HAART with the help of a genetic
algorithm. Therapy was applied to virtual patients for a period of 6 months starting
7 5 years after infection. After therapy was terminated, simulated patients were

challenged with an opportunistic infection. The authors compared their determined
optimal therapy strategy with interruption of treatment to other regimes including
the “ideal” scenario where HAART is applied to the patient each day without
interruption. They found that after opportunistic infection, the survival of patients

.