Dynamic contrast-enhanced mri in rat glioblastoma models: vascular parameters and corrections for systematic errors

Abstract

Dynamic contrast-enhanced MRI (DCE-MRI) is widely used in pharmacokinetic modeling to assess biomarkers of treatment response and predictors of tumor progression. Plasma volume fraction (vp), blood-to-tissue forward volumetric transfer constant (Ktrans), and interstitial volume fraction (ve) are the most common biomarkers that explain tumor physiology in embedded cerebral gliomas. DCE-MRI demand stability of receiver sensitivity across the time of study but may suffer from high duty-cycle imaging instability that can bias parameter estimates. In this dissertation, an analysis of DCE-MRI data affected by unstable receiver sensitivity acquired at 7T in a rat model of 9L glioblastoma (GBM) is presented. Phantom studies verified that an unexpected variation in Ktrans in 9L rat model of gliosarcoma was the result of reduction in receiver signal amplitude, indicating a baseline signal drift. Cooling the coil and repeating the experiment pointed to heating of the coil as the root of the problem, a finding confirmed by the vendor. A correction to the amplitude drift based on a heuristic logarithmic function is presented.Two patient-derived orthotopic xenograft (PDOX) models of GBM (HF3016 and HF3177) were characterized using multiparametric MRI. Vascular and tumor microenvironmental (TME) parameters (vp, Ktrans, ve, VD, tumor exudate flux) were estimated with DCE-MRI along with multiparametric MRI. DCE data were analyzed voxel-wise using Patlak, extended Patlak, and Logan methods, with a data-driven model selection approach for defining the tumor and the normal tissue regions. While vascular and diffusion parameters showed no significant differences between the two models (p>0.05), Flux strongly correlated with VD at the tumor rim. These data report physiological properties of untreated GBM that are representative of human disease both geno- and pheno-typically. Tumor interstitial fluid pressure (TIFP) was measured using the invasive technique in treatment naïve and in irradiated 9L gliosarcoma cerebral tumors in rats. The signatures of acute treatment response were investigated using DCE-MRI. Parameters Ktrans, ve ,VD and Flux, were assessed in these cohorts in paired experiments 24 hours apart, with intervening radiotherapy (RT) in the treatment animals. TIFP and tumor blood flow were significantly reduced (p<0.05) post RT. DCE-MRI biomarkers showed a decrease post RT; notably Flux and VD significantly decreased post RT (p<0.05) and were strongly correlated indicating a clear signature of radiation response. The regression fits of Flux versus VD also differed (p<0.05) between the two cohorts.