Comparison of actual and simulated tumoricidal effects induced by photodynamic therapy

Numerous studies employ mathematical methods, such as Monte Carlo simulation, to predict the tumor killing effects of photodynamic therapy (PDT) by simulating optical propagation, photosensitizer distribution, and oxygen distribution. Whether these models faithfully reflect tumor killing is unknown, and model validation using tumor cross sections in these studies is usually insufficient to answer this question. To fill this gap in our knowledge, we employed a mouse model of breast cancer to determine the spatiotemporal effects of PDT using direct histopathological and biochemical analyses of whole tumors.

We prepared approximately 700 5-μm-thick serial sections of breast tumors of syngeneic mice treated with PDT employing the photosensitizer photocarcinorin (PsD-007, a second-generation photosensitizer developed in China). Three adjoining sections were subjected to hematoxylin and eosin staining to assess necrosis, the TUNEL assay to evaluate apoptosis, and CD31 staining to detect angiogenesis, respectively. We then generated a three-dimensional (3D) reconstruction of the tumor to evaluate these processes. We simultaneously used the Monte Carlo method to develop a model of light distribution throughout the tumor to evaluate the actual and simulated tumor killing effects induced by PDT.

Tumor necrosis decreased exponentially as a function of distance from the source of illumination, while the distributions of apoptosis and neovascularization were independent of light distribution. Most apoptosis occurred in the lower layers (3000-4000 μm) of the tumor where the light intensity was too low to excite the photosensitizer. Neovascularization occurred at depths ranging from 2500 to 3500 μm. These analyses provided a 3D view of how a tumor is destroyed using PDT.

Although the optical distribution model predicted tumor necrosis caused by PDT, it was ineffective in predicting the sites of apoptosis and vascular destruction. Mathematical modeling is limited in its capabilities required to gain a comprehensive understanding of the spatiotemporal events associated with PDT. The mouse model developed here will serve as a platform for detailed direct histopathological, biochemical, and molecular genetic analyses of the effects of PDT, which will facilitate the development of optimized treatment strategies.