Radiation oxygen biology with pulse electron paramagnetic resonance imaging in animal tumors.

The reduced oxygen in tumors (hypoxia) generates radiation resistance and limits tumor control probability (TCP) at radiation doses without significant normal tissue complication. Modern radiation therapy delivery with intensity-modulated radiation therapy (IMRT) enables complex, high-dose gradient patterns, which avoid sensitive human tissues and organs. EPR oxygen images may allow selection of more resistant parts of a tumor to which to deliver more radiation dose to enhance TCP. EPR O2 images are obtained using injected narrow-line, low relaxation rate trityl spin probes that enable pulse radiofrequency EPR O2 images of tumors in the legs of mice, rats, and rabbits, the latter exceeding 4 cm in size. Low relaxation rates of trityls have enabled novel T1-, rather than T2-, based oximetry, which provides near absolute pO2 imaging. Tomographic image formation and filtered back projection reconstruction are used to generate these images with fixed, linear stepped gradients. Images obtained both with T2 and T1 oximetric images have demonstrated the complex in vivo mechanism explaining the unexpected efficacy of TNFerade, a radiation-inducible adenoviral construct to locally produce TNF-induced vascular as well as radiation damage [1, 2]. The unexpected efficacy of large-dose radiation fractions is seen to be due to an interaction between host microvasculature and tumor cells producing a prompt (15 min) postradiation hypoxia, paralyzing tumor cell repair, and sensitizing tumors. Finally, cure of tumors treated to a single 50 % control dose shows a significant dependence on EPR O2 image hypoxic fractions, best shown with the fraction of voxels less than 10 Torr (HF10). We show that these O2 images provide a quantitative basis for measuring tumor and normal tissue response to abnormally low O2 levels. Measurements of vascular endothelial growth factor (VEGF) production in a specific syngeneic mouse fibrosarcoma, FSa versus fraction of tissue voxels with pO2 less than 10 Torr, produced a slope of 0.14 pg VEGF protein/mg total protein/% HF10. We argue that this quantification may be diagnostic of tumor versus normal tissue, and it may be etiologic in the development of malignancy.