An Explanatory Computational Simulation of Contiguous Capillary Occlusion in Diabetic Retinopathy based on Patient‐derived Vasculature

Early stages of diabetic retinopathy often feature contiguous areas of capillary occlusion in both perifoveal and peripheral areas. While leukocyte‐endothelium interactions appear to cause individual capillary occlusions, the underlying mechanisms by which multiple occlusions create large ischemic areas are not clear. To explain the observed pattern of contiguous capillary blockage seen in diabetics using fluorescein angiography, we developed a computational simulation based on a hypoxia‐driven vascular endothelial growth factor ( VEGF )‐mediated adverse feedback mechanism. We constructed an in silico network of retinal capillaries in the perifovea (with an architecture derived from an actual human perifoveal arteriovenous sector capillary map obtained using adaptive‐optics scanning laser ophthalmoscopy ( AOSLO ) imaging) and in the retinal periphery. A hemodynamic model calculates blood flow rates in the simulated capillary network and submodels simulate advection, diffusion and metabolism of oxygen and synthesis, diffusion and decay of VEGF. Our main hypothesis is that hypoxic retinal tissue produces VEGF, which diffuses to adjacent capillaries and increases their probability of occlusion. Our simulations reproduce the experimentally observed pattern of contiguous areas of capillary occlusion. Furthermore, the simulated temporal evolution of blood flow agrees quantitatively with experiment, and the simulated venous oxygenation agrees qualitatively. We used multiple replica simulations to generate a vulnerability map for the capillary network, which showed that only a few capillary segments have a high probability of triggering large areas of derived occlusions. We then simulated the effects of specific patterns of pan‐retinal photocoagulation therapy using patterned alterations of tissue oxygen consumption to explore alternative therapeutic approaches to inhibiting progressive tissue ischemia. Future versions of this simulation framework may increase our ability to intervene early, using patient‐specific photocoagulation or anti‐VEGF treatments, to prevent large ischemic areas from developing and thereby prevent blindness from diabetic retinopathy. Support or Funding Information This work was sponsored by National Institutes of Health/National Institute of General Medical Sciences grants R01 GM076692, R01 GM077138, and U01 GM111243 and National Institutes of Health/National Eye Institute R01 EY024315. We have received support from the College of Arts and Sciences, the Office of the Vice President for Research under their IU Collaborative Research Grant Program at Indiana University, Bloomington.