**5.1 Tumor morphology**

256 Advances in the Biology, Imaging and Therapies for Glioblastoma

*al*., 2006, 2009; Masui *et al*., 2010). Intracerebral implantation of PDGFB-green fluorescent protein (GFP)-expressing retroviruses into rodents were found to induce tumors that closely resembled diffuse human malignant gliomas which have been challenging to treat (Assanah *et al*., 2006; Masui *et al*., 2010). This model involves the use of a viral vector that stimulates neuronal stem cells to become glioma cells by expressing PDGF, which is involved in generating tumor cells (Masui *et al*., 2010). These studies demonstrate that both adult white matter and glial progenitors generate gliomas, as well as recruit resident progenitors to proliferate within the mitogenic environment of a tumor, and therefore contributing to the heterogeneous mass of cells that make up a malignant glioma (Assanah *et al*., 2006; Masui *et al*., 2010). It was previously demonstrated that PDGF-B could play a dose-dependent role in glial tumorigenesis, where PDGFR (PDGF receptor) signaling via elevating levels of PDGF-B chain expression quantitatively regulates tumor grade, and that PDGF-B expression is required to sustain high-grade oligodendrogliomas (Shih *et al*., 2004). PDGF-B expression in tumor cells was elevated by removing inhibitory regulatory elements in the *PDGFB* mRNA and a retroviral delivery system (Shih *et al*., 2004). To generate tumors, DF1 cells transfected with RCAS (repeat with splice acceptor) retroviral vectors, generating a culture of virusproducing cells, were injected intracranially into N-tva transgenic mice (Shih *et al*., 2004). By inhibiting PDGFR activity, it was possible to convert tumors from high to low grade (Shih *et* 

Another recent study involved intracranial injection of lentiviral vectors with GFAP (glial fibrially acidic protein) or CMV (cytomegalovirus) vectors into compound *LoxP*-conditional mice, which resulted in K-Rasv12 expression and loss of p16Ink4a/p19Arf, with or without concomitant loss of p53 or Pten (de Vries *et al*., 2010). Like GFAP, CMV is a promoter (de Vries *et al*., 2010). CMV-Cre injection into *p53;Ink4a/Arf;K-Rasv12* mice was particularly found to result in the formation of high-grade gliomas within 2-3 weeks that had invasiveness and blood-brain barrier functionality characteristics that are found in human high-grade gliomas

Magnetic resonance imaging (MRI) techniques are becoming more commonly used to provide information on brain tumor growth, vasculature, biochemical metabolism, and molecular changes in preclinical models, as MRI is the optimal imaging tool as part of the diagnostic process for human gliomas. Conventional MRI techniques, such as T1- and T2 weighted imaging, contrast-enhanced T1-weighted imaging, dynamic contrast enhanced (DCE) imaging, and diffusion-weighted imaging (DWI) methods can provide useful information on tumor location and extent of growth, blood-brain barrier (BBB) disruption, brain invasiveness, regional blood flow and blood volume, and tumor cellularity, all of which are characteristics associated with glioma grade and prognosis in a clinical setting

Morphological MRI (T2-weighted or T1-weighted contrast-enhanced imaging) is used to provide information on tumor volumes and growth rates, which can be used to distinguish between tumor grades. Contrast-enhanced imaging can be used to assesses BBB disruption, however this feature can be absent in diffuse infiltrative tumor regions or when assessing therapeutic treatment (Waerzeggers *et al*., 2010). DCE imaging can be used to follow tumor angiogenesis by measuring changes in tumor vascular permeability, vascular density and

*al*., 2004).

(de Vries *et al*., 2010).

(Waerzeggers *et al*., 2010).

**5. MRI methods to detect gliomas** 

MRI is obtained on small animal MR imaging systems (7 - 11.7 Tesla), that can accommodate rodents such as mice and rats. MR images are obtained in multiple slices (0.5-1 mm thick) to visualize an entire tumor (Gartesier *et al*., 2010). Examples of rodent tumor models for gliomas (e.g. rat C6 and RG2 models, and mouse GL261 model) are shown in Figure 1, depicting heterogeneous tumors (right cerebral cortex, upper regions) following intracerebral (orthotopic) implantation of rat or mouse glioma cells (Doblas *et al*., 2010).

From the multiple image slices through a tumor, tumor volumes can be measured, and the growth rate can be calculated from multiple imaging sessions over several days, weeks or months (as shown in Figure 2). Tumor areas are traced in multiple slices to calculate tumor volumes, which can be used to determine tumor growth and doubling times (Doblas *et al*., 2008, 2010; Garteiser *et al.*, 2010). Robust tumor volume determinations can be made by using manual or automated segmentation techniques, which can be used to delineate tumor margins on the basis of signal intensity differences from surrounding brain tissue (Waldman *et al*., 2009).

Fig. 1. MR images of rodent gliomas. T2-weighted images of the rat C6 (A; 18 days following intracerebral implantation of cells) and RG2 (B; 13 days following cell implantation), and the mouse GL261 (C; 26 days following cell implantation) glioma models. Tumors appear as heterogeneous regions in the upper right area of the cerebral cortex regions.

Assessment of Rodent Glioma Models Using Magnetic Resonance Imaging Techniques 259

ENU-induced model, indicated that percent necrosis was highest in the ENU model, compared to RG2, and least in the C6 model (Towner *et al*., 2010a). Although the ENUinduced model is used to generate low-grade gliomas, it generates a heterogeneous population of glioma cells ranging from low- to high-grade, which contributes to the high incidence of necrotic lesions (Towner *et al*., 2010a). RG2 gliomas are known to be more aggressive, invasive and infiltrative than C6 gliomas (Groothuis *et al*., 1983). It has also been shown that RG2 gliomas have more diffuse margins at the interface to adjacent brain tissue, whereas C6 gliomas are less infiltrative with a distinct peritumoral region at the margin of

MR angiography can provide information on new blood vessels formed in tumors, a process known as angiogenesis which is required to maintain tumor growth. On small animal imaging MRI systems, the image in-plane resolution is >50 µm, which allows visualization of major blood vessels, arterioles and venules (Doblas *et al*., 2008, 2010). Quantitation of brain and tumor blood vessels can also be obtained, as well as measurements on blood vessel diameters and lengths (Doblas *et al*., 2008, 2010). An increase in total brain tumor blood volume was found to directly correlate with increasing tumor volumes during tumor

Quantification of the Brownian motion of water or diffusion within tissues can be measured through the apparent diffusion coefficient (ADC) which is obtained from diffusion-weighted imaging datasets. DWI yields ultrastructural information on cellular density and the extracellular matrix (Waldman *et al*., 2009; Sadeghi *et al*., 2003). Figure 5 shows an example of an ADC (Fig. 5 b) map in a C6 glioma-bearing rat brain, indicating higher ADC values in

Temporal diffusion spectroscopy based on oscillating gradient spin-echo (OGSE) MRI was used to detect microscopic structural variations at the subcellular scale in C6 rat gliomas

Fig. 4. MR angiography of a GL261 mouse glioma. (A) T2-weighted MR image of a GL261 glioma-bearing mouse brain (23 days following intracerebral implantation of cells.) (B) and (C), 3 dimensional angiograms of mouse brain blood vessels of a GL261 glioma-bearing mouse. Note altered vasculature on left-hand side in image B, as well as increased blood vessels (middle cerebral artery; depicted in left-mid-region of image C) in the tumor region.

tumor tissue compared to contralateral 'normal' brain tissue (Garteiser *et al*., 2010).

the tumor (Doblas *et al*., 2010).

growth (Doblas *et al*., 2010).

**5.2 Tumor vasculature and ultrastructure** 

Fig. 2. MR images used to calculate tumor volumes and tumor growth. T2-weighted images of a mouse GL261 glioma at 14 (A), 23 (B) and 26 days (C) following intracerebral implantation of cells. Tumors are outlined (white ellipses). (D) Calculated GL261 tumor volumes which follow an exponential increase over time.

Fig. 3. Determination of necrotic volumes from MR image-observed tumors. (A) T2 weighted MR image of a GL261 mouse glioma at 23 days following intracerebral implantation of cells. Note dark void regions in the heterogeneous tumor which are necrotic lesions. (B) Necrotic volumes from tumors can be measured (n=5; meanS.D.). (C) Corresponding histological slide depicting necrosis in a GL261 tumor.

Tumor morphology can also provide information on tumor invasiveness and necrotic lesions. Necrotic lesions are depicted as dark void regions in a tumor (as shown in Fig. 3A), of which volumes (e.g. Fig. 3B) can be measured from multiple slices (Towner *et al*., 2010a). A comparison between the orthotopic rat glioma models, C6 and RG2, and the chemical ENU-induced model, indicated that percent necrosis was highest in the ENU model, compared to RG2, and least in the C6 model (Towner *et al*., 2010a). Although the ENUinduced model is used to generate low-grade gliomas, it generates a heterogeneous population of glioma cells ranging from low- to high-grade, which contributes to the high incidence of necrotic lesions (Towner *et al*., 2010a). RG2 gliomas are known to be more aggressive, invasive and infiltrative than C6 gliomas (Groothuis *et al*., 1983). It has also been shown that RG2 gliomas have more diffuse margins at the interface to adjacent brain tissue, whereas C6 gliomas are less infiltrative with a distinct peritumoral region at the margin of the tumor (Doblas *et al*., 2010).
