Abstract
PURPOSE: Tumour hypoxia and elevated glycolysis (Warburg effect) predict poor prognosis. Each parameter is assessable separately with positron emission tomography, but they are linked through anaerobic glycolysis (Pasteur effect). Here, we compare the oxygenation-dependent retention of fluoroazomycin arabinoside ([(18)F]FAZA), a promising but not well-characterised hypoxia-specific tracer, and fluorodeoxyglucose ([(18)F]FDG) in four carcinoma cell lines. METHODS: Cells seeded on coverslips were positioned in modified Petri dishes that allow physically separated cells to share the same tracer-containing medium pool. Following oxic, hypoxic or anoxic tracer incubation, coverslips were analysed for radioactivity ([(18)F]FDG + [(18)F]FAZA) or re-incubated in tracer-free oxygenated medium and then measured ([(18)F]FAZA). Next, we tested the reliability of [(18)F]FDG as a relative measure of glucose metabolic rate. Finally, from two cell lines, xenografts were established in mice, and the tracer distribution between hypoxic and well-oxygenated areas were deduced from tissue sections. RESULTS: Three hours of anoxia strongly stimulated [(18)F]FAZA retention with anoxic-to-oxic uptake ratios typically above 30. Three out of four cell lines displayed similar selectivity of [(18)F]FDG versus glucose, but oxic uptake and anoxic-to-oxic uptake ratio of [(18)F]FDG varied considerably. Although less pronounced, [(18)F]FAZA also showed superior in vivo hypoxia specificity compared with [(18)F]FDG. CONCLUSIONS: [(18)F]FAZA displays excellent in vitro characteristics for hypoxia imaging including modest cell-to-cell line variability and no binding in oxic cells. In contrast, the usability of [(18)F]FDG as a surrogate marker for hypoxia is questionable due to large variations in baseline (oxic) glucose metabolism and magnitudes of the Pasteur effects.
