First-in-human fractal methodology for modeling the hepatic arterial tree and tumor microvasculature for 90Y-microsphere brachytherapy.

Abstract

248 Background: 90Y-microsphere therapy (radioembolization) is an effective and safe brachytherapy procedure for control of unresectable primary and metastatic hepatic malignancies. To advance this treatment approach, a method is proposed which enables complex modeling of the hepatic arterial route, and the tumor microvascular bed in which the radioactive particles will become permanently embedded. Methods: Prior work from our group included hepatic vessel measurements from patients undergoing 90Y-microsphere therapy including geometry, pressure, velocity and waveforms prior to and during microsphere deposition in real time. A computation particle-fluid dynamics model was built for proximal hepatic arteries of 900 microns, which is the limit of current imaging technology. Fractal methods were programmed in C++ for simulations of both normal liver and tumor arterial trees. Our goal was to produce sufficient generations to achieve daughter arterioles of 50 micron diameter in tumors where microspheres typically lodge in clusters. Results: Normal liver and tumor artery trees were created, with malignant vessels employing a random generator at each node resulting in corkscrew, bifurcation and/or trifurcation daughter vessel pattern. Three key variables were: asymmetry ratio of radii of daughter vessels; power law of the type of flow present in vessels (turbulent or laminar); and area ratio, which is the area between daughters and parent vessel. Simulations of up to 250 generations resulted in successful modeling of 50 micron diameter normal and tumor arterioles. Impedance in each vessel (modulus and phase shift) was tested with a large variety of vessel geometries. Conclusions: Fractal methods were used to develop a software tool that can represent the microvasculature of human liver and different organs, and can account for disease states, such as liver tumors. This is accomplished using corkscrew, bifurcating, trifurcating and randomized structured trees. Predictive modeling may now be possible for radioactive or non-radioactive microspheres exiting from a catheter in the hepatic artery to its final position in a tumor end arteriole, or for systemic therapies.