Can hypoxia-PET map hypoxic cell density heterogeneity accurately in an animal tumor model at a clinically obtainable image contrast?

Abstract

PET allows non-invasive mapping of tumor hypoxia, but the combination of low resolution, slow tracer adduct-formation and slow clearance of unbound tracer remains problematic. Using a murine tumor with a hypoxic fraction within the clinical range and a tracer post-injection sampling time that results in clinically obtainable tumor-to-reference tissue activity ratios, we have analyzed to what extent inherent limitations actually compromise the validity of PET-generated hypoxia maps. Mice bearing SCCVII tumors were injected with the PET hypoxia-marker fluoroazomycin arabinoside (FAZA), and the immunologically detectable hypoxia marker, pimonidazole. Tumors and reference tissue (muscle, blood) were harvested 0.5, 2 and 4h after FAZA administration. Tumors were analyzed for global (well counter) and regional (autoradiography) tracer distribution and compared to pimonidazole as visualized using immunofluorescence microscopy. Hypoxic fraction as measured by pimonidazole staining ranged from 0.09 to 0.32. FAZA tumor to reference tissue ratios were close to unity 0.5h post-injection but reached values of 2 and 6 when tracer distribution time was prolonged to 2 and 4h, respectively. A fine-scale pixel-by-pixel comparison of autoradiograms and immunofluorescence images revealed a clear spatial link between FAZA and pimonidazole-adduct signal intensities at 2h and later. Furthermore, when using a pixel size that mimics the resolution in PET, an excellent correlation between pixel FAZA mean intensity and density of hypoxic cells was observed already at 2h post-injection. Despite inherent weaknesses, PET-hypoxia imaging is able to generate quantitative tumor maps that accurately reflect the underlying microscopic reality (i.e., hypoxic cell density) in an animal model with a clinical realistic image contrast.