Pharmacokinetic modeling of [18F]-ES tumor uptake in patients with breast cancer

Objectives To determine a non-invasive metric of estrogen receptor (ER) status in breast tumors by PET imaging of fluoroestradiol (18F-FES). The uptake of the tracer 18F-FES has been correlated to ER status determined from biopsy data in breast tumors (Sundararajan, et al. Semin Nucl Med 37,p470-476,2000).Our objective was to evaluate whether pharmacokinetic modeling of dynamic PET data could provide a more robust non-invasive quantitive metric of ER status.

Methods 15 patients with breast lesions of >1cm diameter were imaged on a GE DSTE PET/CT scanner.Dynamic images were acquired over 25 minutes in 3D mode and reconstructed using standard clinical protocols.Regions of interest(ROI) were drawn encompassing the entire tumor whose margins were defined on the last image of the dynamic PET series.A 2-tissue compartment model was used to obtain kinetic parameters for the average time-activity curve each ROI.The SUVmax of each voxel was compared to the respective flux constant Ki=k1k3/(k2 + k3) and the graphical Patlak Ki value and k3 the irreversible trapping rate constant.Both the trapping coefficient k3 and Ki were correlated with SUVmax at 25 mins post injection with the goal of determining whether compartmental analysis is useful or redundant in FES PET imaging.

Results The correlations (R2) between SUVmax and kinetic parameters for flux and was -0.14. The correlation with k3 was 0.10.

Conclusions There is poor agreement between the SUVmax measured at 25 min p.i. and both the flux constant Ki obtained from full compartmental analysis of the entire dynamic data sets suggesting that SUVmax may not provide a sufficiently accurate surrogate measure for ER status. The correlation with k3, a more direct parameter related to estrogen density was weaker