Background: Current [F-18]-fluorodeoxyglucose positron emission tomography (FDG-PET) procedures in tumor-bearing mice typically includes fasting, anesthesia, and standardized uptake value (SUV)-based quantification. Such procedures may be inappropriate for prolonged multiscan experiments. We hypothesize that normalization of tumor FDG retention relative to a suitable reference tissue may improve accuracy as this method may be less susceptible to uncontrollable day-to-day changes in blood glucose levels, physical activity, or unnoticed imperfect tail vein injections.
Material and methods: Fed non-anesthetized tumor-bearing mice were administered FDG intravenously (i.v.) or intraperitoneally (i.p.) and PET scanned on consecutive days using a Mediso nanoScan PET/magnetic resonance imaging (MRI). Reproducibility of various PET-deduced measures of tumor FDG retention, including normalization to FDG signal in reference organs and a conventional SUV approach, was evaluated.
Results: Day-to-day variability in i.v. injected mice was lower when tumor FDG retention was normalized to brain signal (T/B), compared to normalization to other tissues or when using SUV-based normalization. Assessment of tissue radioactivity in dissected tissues confirmed the validity of PET-derived T/B ratios. Mean T/B and SUV values were similar in i.v. and i.p. administered animals, but SUV normalization was more robust in the i.p. group than in the i.v. group.
Conclusions: Multimodality scanners allow tissue delineation and normalization of tumor FDG uptake relative to reference tissues. Normalization to brain, but not liver or kidney, improved scan reproducibility considerably and was superior to traditional SUV quantification in i.v. tracer-injected animals. Day-to-day variability in SUV's was lower in i.p. than in i.v. injected animals, and i.p. injections may therefore be a valuable alternative in prolonged rodent studies, where repeated vein injections are undesirable.