Kinetic modeling of 3'-deoxy-3'-18F-fluorothymidine for quantitative cell proliferation imaging in subcutaneous tumor models in mice.

UNLABELLED 3'-Deoxy-3'-(18)F-fluorothymidine ((18)F-FLT) is a thymidine analog that was developed for measuring tumor proliferation with PET. The aim of this study was to establish a kinetic modeling analysis method for quantitative (18)F-FLT PET studies in subcutaneous tumor models in mice. METHODS To explore the validity of an image-derived left ventricular input function, we measured equilibrium constants for plasma and whole blood and metabolite fractions in blood after (18)F-FLT injection. In parallel, dynamic (18)F-FLT PET scans were acquired in 24 mice with a small-animal dedicated PET scanner to compare arterial blood activities obtained by PET and blood sampling. We then investigated kinetic models for (18)F-FLT in human epithelial carcinoma (A431) and Lewis lung carcinoma tumor models in mice. Three-compartment models with reversible phosphorylation (k(4) not equal 0, 3C5P) and irreversible phosphorylation (k(4) = 0, 3C4P) and a 2-compartment model (2C3P) were examined. The Akaike information criterion and F statistics were used to select the best model for the dataset. Gjedde-Patlak graphic analysis was performed, and standardized uptake values in the last frame were calculated for comparison purposes. In addition, quantitative PET parameters were compared with Ki-67 immunostaining results. RESULTS (18)F-FLT equilibrated rapidly (within 30 s) between plasma and whole blood, and metabolite fractions were negligible during PET scans. A high correlation between arterial blood sampling and PET data was observed. For 120-min dynamic PET data, the 3C5P model best described tissue time-activity curves for tumor regions. The net influx of (18)F-FLT (K(FLT)) and k(3) obtained with this model showed reasonable intersubject variability and discrimination ability for tumor models with different proliferation properties. The K(FLT) obtained from the 60- or 90-min data correlated well with that obtained from the 120-min data as well as with the Ki-67 results. CONCLUSION The image-derived arterial input function was found to be feasible for kinetic modeling studies of (18)F-FLT PET in mice, and kinetic modeling analysis with an adequate compartment model provided reliable kinetic parameters for measuring tumor proliferation.