Quantifying and correcting for tail vein extravasation in small animal PET scans in cancer research: is there an impact on therapy assessment?

Background Tail vein injection under short anesthesia is the most commonly used route for administering radiopharmaceuticals. However, the small caliber of the vein in rodents may lead to tracer extravasation and thereby compromise quantitative accuracy of PET. We aimed to evaluate a method for correction of interstitial radiotracer leakage in the context of pre-clinical therapeutic response assessment. Methods In two separate studies involving 16 nude rats, a model of human ovarian cancer was xenografted and each was treated with a Phosphoinositide 3-kinase/mammalian target of rapamycin inhibitor or used as a control. Tracer injections were performed via the tail vein by a single operator. Two observers qualitatively evaluated the resulting images and if appropriate drew a volume of interest (VOI) over the injection site to record extravasated activities. Uncorrected and corrected tumors’ mean standardized uptake value (SUV) mean was computed (corrected injected activity = calibrated activity − decay corrected residual syringe activity − decay corrected tail extravasated activity). Molecular analyses were taken as a gold standard. The frequency and magnitude of extravasation were analyzed, as well as the inter-observer agreement and the impact of the correction method on tumor uptake quantification. Results Extravasation never exceeded 20 % of the injected dose but occurred in more than 50 % of injections. It was independent of groups of animals and protocol time points with p values of 1.00 and 0.61, respectively, in the first experiment and 0.47 and 0.13, respectively, in the second experiment. There was a good inter-observer agreement for qualitative analysis (kappa = 0.72) and a moderate agreement when using quantitative analysis ( ρ c = 0.94). In both experiments, there was significant difference between uncorrected and corrected SUV mean . Despite this significant difference, mean percent differences between uncorrected and corrected SUVmean in the first and the second experiments were -3.61 and -1.78, respectively. Concerning therapy assessment, in both experiments, significant differences in median %SUV mean between control and treated groups were observed over all time points with either uncorrected and corrected data ( p