Abstract
BACKGROUND AND PURPOSE: Hypoxia for head and neck cancer (HNC) can be imaged with positron emission tomography (PET) using (18)F-Fluoroazomycin-arabinoside (FAZA) but is not used routinely. In contrast, fluorodeoxyglucose (FDG) PET visualizing tumor metabolism is routinely used in radiotherapy (RT) of HNC patients. Dual-energy computed tomography (DECT) can generate an iodine concentration (IC) map visualizing the perfused blood volume. Here, we explore how hypoxia can be classified for HNC using a PET derived FDG and a DECT derived IC metric. MATERIALS AND METHODS: Corresponding DECT, FAZA, and FDG PET/CT for 6 HNC patients before and during RT were acquired. A FAZA tumor-to-muscle (TMR) ratio ≥1.2 was used for hypoxic classification. Within the gross tumor volume (GTV), the IC standard deviation over mean ratio, / , was used to model blood perfusion and the percentage of maximum FDG standard uptake value (%SUV(max)) was used for metabolic activity. Receiver Operating Characteristics (ROC) was performed for the modelled blood perfusion and metabolism individually and combined as / . The perfusion and metabolism metrics were further applied in a consumption and supply-based hypoxia (CSH) model. RESULTS: ROC curves improved with AUC around 0.9 when combining the blood perfusion and metabolism metrics. GTVs with high metabolic activity and and low modelled blood perfusion was dominated by hypoxic fractions >0.75 supporting the CSH model. CONCLUSIONS: Combining blood perfusion and metabolism modelled from DECT and FDG PET derived metrics resulted in a superior predictive power as potential hypoxia biomarkers which might be explained by a CSH model.