Contrast improvement of FDG microPET studies by voxel-wise Patlak analysis

Objectives The usefulness of FDG-PET for studying human tumor models in mice is often limited by unexpectedly low FDG uptake and image contrast. Voxel-wise Patlak analysis may help to improve image contrast of PET studies as it corrects for the contribution of unphosphorylated FDG to the PET signal and also takes into account the rapid blood clearance of FDG in mice.

Methods Three groups of nude mice were subcutaneously implanted with three non-small cell lung cancer tumor xenografts cell lines (LXFE-400, LXFA-1584, LXFE-409). The xenografts grew to a size of at least 8 mm. After i. v. injection of 10-15 MBq FDG a one hour dynamic microPET scan was performed. As previously described (1) the input function for Patlak analysis was derived from a ROI in the liver. The FDG net influx constant Ki was calculated voxel-by-voxel from a slope of the Patlak plot from 5 to 40 min p. i. The images of the last frame of the dynamic PET scan (50-60 min p.i) were compared with the parametric images of Ki.

Results All studied tumors demonstrated only low FDG uptake in the static images (mean SUV±SD: 1.1±0.3, n= 14). The average tumor-to-background ratio was 3.2±1.4. The parametric images of Ki demonstrated a markedly better delineation of the tumors than the static PET scans (Figure 1). The average Ki of the tumors was 1.8±0.43 ml/100g/min (range: 0.9-2.4 ml/100g/min). The average tumor-to-background ratio in the parametric images was 12.5±8.2 and therefore about 4 times higher than for the static PET images.

Conclusions Voxel-wise Patlak analysis improves the image contrast of FDG-PET studies of human xenografts in mice. This may facilitate measurements of changes in tumor metabolic activity during therapeutic interventions and should therefore be considered for treatment monitoring studies