Prediction of angiogenesis based on dynamic PET FDG studies and parametric imaging

Objectives Angiogenesis has an impact on the kinetics of FDG (J Nucl Med 2008; 49:1238-1244). Based on PET and gene expression data, regression functions were calculated and applied to parametric data to obtain images, which are predictive for the angiogenesis related gene expression.

Methods Dynamic PET FDG studies of colorectal tumors (n=22) and gliomas (n=6) were used. Gene array data were also available. The PET data were evaluated with VOIs and include all parameters of the two-tissue compartment model, the fractal dimension, and the SUV for 55-60 min post injection. Regression functions were calculated for angiogenesis related genes using gene array and PET results. Parametric images were calculated for all parameters listed above.

Results The evaluation comprises 28 patients with PET results and gene expression data. The VOI based evaluation demonstrated a correlation of r=0.7442 for the PET parameters and VEGF-A. The correlation coefficients were lower for other angiogenesis related genes. A round robin method was used to assess the variance for predicting VEGF-A expression from PET results and revealed accurate results. We used a multiple regression function for the PET parameters to calculate images of the VEGF-A expression using the parametric images. The resulting images were compared to the measured VEGF-A expression data.

Conclusions Dynamic PET data provide information about the angiogenic activity in colorectal tumors and gliomas. Parametric imaging can be performed to obtain images for the VEGF-A expression in the tumor. This information can be helpful for therapy management, e.g. if an anti-angiogenic treatment is considered.