Abstract
Purpose :
In age-related macular degeneration, drusen built up under the retinal pigment epithelium (RPE) leads to overexpression of vascular endothelial growth factor (VEGF). By developing computational behavioral models to analyze the effects of VEGF overexpression due to RPE cells, we can accurately and efficiently predict and simulate blood vessel growth in wet AMD. We hypothesize this model will be able to simulate the extent and probable locations of blood vessel growth based on production rates of VEGF by the RPE.
Methods :
MATLAB is the programming language chosen for this model due to its matrix capabilities, as the physical space within the eye can be directly represented by a matrix. Within the matrix, the left boundary represented the choroid where capillaries stem from and the right boundary represented the RPE, which is the source of VEGF. A function is created to use point source diffusion to calculate the concentration of VEGF at any distance from the source.Our model also includes a function that calculates the chemoatrractance at locations around a tip cell, based on VEGF concentrations. Based on the amount of VEGF bound to the cell surface, this function computes the probability tip cell movement in each direction. Predicting the movement of the tip cell and formation of the capillary is accomplished by employing a biased random walk based on transition probabilities.
Results :
Our two-dimensional computational model predicts the movement of the tip cell and subsequent growth of the capillary.It also produces a graphical representation of vessel growth for easy visualization by the user. Consistent with data in scientific literature, the blood vessel production increases with increased expression of VEGF. The growth of the capillary also biases towards the locations with higher VEGF concentration near the source as expected.
Conclusions :
Given parameter values found through experimentation or in literature, the model can potentially give a reasonably accurate prediction of blood vessel growth over time. This computational model allows parameter sweep and optimization to accurately predict the probabilistic behaviors of tip cell growth, which enables efficient lab experiments by narrowing the parameter search space.
This is a 2020 ARVO Annual Meeting abstract.