Technical and biological validation of hypoxia PET imaging using [18F]fluroazomycin (FAZA) in NSCLC

Purpose or Objective There is an unmet need to validate hypoxia PET to select patients for future hypoxia-targeted therapy trials. This study aimed to define optimal [18F]FAZA PET acquisition and analysis in NSCLC patients and assess repeatability of hypoxic volumes (HV) and fractions (HF) using fixed and image-derived thresholds. As an exploratory objective, we compared tumor HFs from [18F]FAZA PET with tissue hypoxia, quantified using the exogenous hypoxia marker pimonidazole. Material and Methods Twelve NSCLC patients underwent one (n=6) or two (n=6) [18F]FAZA PET- CT at 0-1 h and 1.5-2.5 h post injection (pi); figure 1A. Seventeen tumor lesions, reference tissue (muscle) and blood (aorta) were manually contoured on CT (figure 1B). Maximum and mean standardized uptake values (SUVmax and SUVmean) were calculated. Tumor SUVmax was divided by muscle or aorta SUVmean to derive tumor-to-muscle (TMRmax) and tumor-to-aorta (TARmax) ratios. HVs and HFs were defined using a TMR ratio >1.2, >1.4 or >1.96 standard deviation (SD) above muscle SUVmean (image-derived threshold); figure 1C. Tumor pimonidazole immunostaining was quantified in a surgical patient subset who received oral pimonidazole 24 hours preoperatively (n=6). Results Relative to the aorta, tumors and muscle reached equilibrium at 7.3±4.1 and 71±24 min, respectively. There was differential [18F]FAZA distribution and/ or clearance in muscle compared to aorta, highlighting the importance of reference region standardization. Improved and stable tumor-to-reference region contrast was seen at 2-2.5 h, compared to 1.5-2 h pi. TMRmax was significantly higher than TARmax at 2-2.5 h pi (p=0.0331). HVs and HFs based on the 3 thresholds were significantly different with fixed 1.2 > image-derived > fixed 1.4 (figure 2A). An image-derived threshold adapts to the image quality by quantifying the variability of [18F]FAZA uptake in normoxic reference tissue, compared to fixed >1.2 threshold which overestimates hypoxia. HVs and HFs based on an image-derived and fixed >1.2 thresholds showed good repeatability, compared to fixed >1.4 threshold which exhibited moderate to poor scan-scan repeatability (figure 2B). Shortening [18F]FAZA PET acquisition duration (from 30 min to 20 min and 10 min) lowers the sensitivity to detect hypoxic voxels. Nonspecific pimonidazole immunostaining was seen in tumor stroma (figure 2C), inflammatory cells (figure 2D) and necrotic regions (figure 2E). All examined tumour specimens demonstrated positive pimonidazole immunostaining, with different pimonidazole immunostaining patterns in adenocarcinoma compared to squamous cell carcinoma (SCC). There was concordance in the hypoxic status classification between [18F]FAZA PET and pimonidazole immunostaining in all 4 SCC patients but not in the 2 adenocarcinoma patients with tissue data. Conclusion Important new [18F]FAZA PET validation data are presented that are necessary to permit optimal application of this modality to derive potential NSCLC hypoxia biomarkers.