A model of reoxygenation dynamics of head-and-neck tumors based on serial 18F-fluoromisonidazole positron emission tomography investigations.

PURPOSE To develop a model for reoxygenation dynamic and its relationship to local control after radiotherapy (RT), based on repeated dynamic [18F]-fluoromisonidazole (FMISO) positron emission tomography (PET) examinations in head-and-neck cancer patients. METHODS AND MATERIALS Ten head-and-neck cancer patients were examined with dynamic FMISO PET before RT with 70 Gy and after approximately 20 Gy. Two of these patients had two additional dynamic FMISO scans during treatment. Local recurrence was assessed by computed tomography-based follow-up 8-24 months after RT. Tumor-specific values for the level of FMISO retention R and the vascular perfusion efficiency P were determined with a kinetic compartment model. RESULTS Individual R-P scattergrams before and during therapy were analyzed, and significant therapy-induced changes in the characteristic R-P patterns were observed. A tumor control probability model was derived that involves the tissue parameters R and P and estimates the time to reoxygenation. On the basis of this model, a malignancy value M was introduced and calibrated by a fit to the observed outcome data. Reoxygenation is reflected by the model as a progression to less-malignant tumor types (i.e., smaller values of M). In 4 of 6 patients with severe hypoxia, M had decreased after 20 Gy, whereas 2 patients showed increasing M. Four patients showed no hypoxia in the pretreatment scan. CONCLUSION A tumor control probability model was developed based on repeated FMISO PET scans during RT. The model combines the local perfusion efficiency and the degree of hypoxia to estimate reoxygenation time. It constitutes a key for hypoxia image-guided dose escalation in RT.