Quantifying the heterogeneity of tumor microvasculature using pharmacokinetic modeling for [18F]FMISO PET

Objectives Intratumor heterogeneity has been reported to be related to therapy response [1,2]. It is challenging to characterize tumor heterogeneity using macroscopic PET imaging with relatively low spatial resolution. This study aims to quantify the heterogeneity of tumor microvascularization using pharmacokinetic modeling of dynamic PET.

Methods 16 nude mice with xenografted human squamous cell carcinomas (FaDu or CAL-33) were scanned using 2 h dynamic [18F]FMISO PET. Tumors were resected, sliced and the vascular marker CD31 was immunostained followed by H&E staining. The microvessels on 2D staining images were automatically segmented using programs developed on Definiens and the microcirculatory supply units (MCSU) [3] were identified by nearest-neighbour graph. A heterogeneity coefficient (HC) was computed as the standard deviation over the mean of the MCSUs for the quantification of the microvascular heterogeneity. PET images were analyzed using the irreversible two-tissue compartment model. The kinetic parameters and their combinations were compared and correlated with HC.

Results K1 correlated significantly with the mean of MCSU (r=0.54, p=0.03). No obvious correlation was observed between HC and individual K1, k2 or k3. Significant correlation was found between k3/K1 and HC (r=0.60, p=0.01).

Conclusions The combination of kinetic parameters of pharmacokinetic modeling on dynamic [18F]FMISO PET is correlated with the underlying heterogeneity of tumor microvasculature of the investigated tumors. The preliminary results may provide the potential to extend pharmacokinetic modeling for the quantification of intra-voxel heterogeneity which is generally difficult to assess by other in-vivo methods.

Research Support DFG SFB 824