Purpose: In mouse brain tumor models, little uptake of high molecular weight (MW) MR contrast agents is detected due to blood-brain barrier (BBB) and blood-brain tumor barrier (BTB). We have tested i) a combination of bolus tracking and dynamic contrast enhanced (DCE) MRI following injection of low MW contrast agent, GdDTPA, to extract vascular parameters of the tumor, and ii) indirect detection 13C magnetic resonance spectroscopic imaging (MRSI) in brain tumor-bearing mice following injection of 13C-labeled anticancer agent, temozolomide [13C]TMZ.

Methods: For DCE and bolus tracking experiments, Saturation-recovery T1 (TE/TR=1.245/250ms, TD=0.25, 0.5, 1s, slice thickness=2mm) and FLASH T2* (TE/TR=3.5/500ms, slice thickness=2mm) sequences were acquired. For inverse-detection MRSI experiments, heteronuclear multiple-quantum coherence (HMQC) technique (TE/TR=15/1500ms, 8×8 voxels, slice thickness=4mm) was used with WALTZ-16 decoupling in 13C channel. Prior to MRI/MRSI experiments with mice bearing orthotopic U87MG human glioblastoma xenografts, the mouse tail vein was catheterized for injection of GdDTPA (1:2 dilution of Magnevist®) and an intraperitoneal catheter was inserted for i.p. administration of [13C]TMZ. Experimental parameters were optimized with a 50% methanol phantom. All MR studies were carried out on a horizontal bore Bruker Biospec 9.4T spectrometer. MR data were analyzed using in-house software written in the IDL™ and MatLab™ programming environment.

Results/Discussion:Bolus tracking experiments demonstrated that the initial drop in MR signal quickly recovered in the normal brain while slow recovery was detected in brain tumors. Vascular volume and perfusion was uniform in the normal brain while in the brain tumor heterogeneous distribution was typically observed.

A phantom study demonstrated a pronounced signal detection of 50% methanol and consequently high quality images with a spatial resolution of 2×2×4mm were acquired using HMQC technique. No [13C]TMZ was detected in normal mouse brain probably due to either low drug concentration in the brain or no transfer of [13C]TMZ through BBB in normal mouse brain. An estimation of detection sensitivity suggests that MRSI can be feasible for non-invasive detection of 13C-labeled anticancer agents in brain tumors for compounds that can penetrate BTB and reach sub-millimolar concentrations in the tumor. Perfusion, vascular volume, and vascular permeability parameters of the tumor obtained from GdDTPA enhanced dynamic MRI can reveal the role of tumor vascularization in drug delivery. Non-invasive measurements of the spatial distribution of anticancer agents in brain tumors by 1H/13C MRSI may lead to more effective strategies for cancer chemotherapy.

Acknowledgment: Supported in part by NIH grant R01 CA097310 (DA). Special thanks to Drs Zaver M. Bhujwalla and Arvind P. Pathak for helpful discussions.

98th AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA