Image-based biomarkers for the invivo evaluation of human brain gliomas

The World Health Organization classification of gliomas is the primary basis for guiding therapy and assessing overall prognosis in gliomas. This classification system, based on histological features, often falls short of predicting therapeutic response of individual tumors within the same histological grade. Yet, it still remains the grading method for both research and clinical prospects.

Unlike any other organ the brain has multiple protective layers such as the skull that ensure a homeostatic environment. The resulting reduced access to the brain and the absence of plasmatic brain tumor markers bring neuroimaging in a central position in diagnosis and management of brain tumors. Moreover, neuroimaging has evolved from a purely morphologic investigation into a diagnostic tool that allows characterization of particular physical alterations within brain tissue. Understanding the relationship between the physical characteristics of tumor tissue, studied by MR imaging, and biological characteristics of the tumor is therefore important for the appropriate integration of neuroimaging in brain tumor management. The general objective of this work is to define the relationship between physiologybased MR imaging and biological features of glial tumors. Diffusion and perfusionweighted imaging, physiologybased MR techniques provide the data based on physical characteristics of the tissues. Diffusion weighted imaging (DWI) allows the measurement of water molecules diffusivity within the brain tissue by means of apparent diffusion coefficient (ADC) measurements. Perfusion weighted imaging (PWI) is based on changes of MR signal during the passage of contrast material through the intravascular space and allows hemodynamic measurements such as those of cerebral blood volume (CBV)within the brain tissue.

Highgrade diffuse gliomas are currently differentiated from low grade diffuse gliomas by increased cellularity with nuclear atypia, mitotic activity, endothelial proliferation and necrosis. Components of the extracellular matrix and angiogenesis constitute some other features of gliomas, which have established links with oncogenic processes that influence the proliferative and infiltrative potentials of these tumors. We have specifically targeted these features in our comparative studies with the working hypothesis that physiologybased MR measurements, obtained in vivo, might provide information that is pertinent in terms of tumor malignancy.

We chose to approach the biology of brain tumors in two ways: in vivo, by means of metabolic imaging techniques such as positron emission tomography (PET) and ex vivo, by means of histological and immunohistochemical analyses of tumor specimens.

Many studies have investigated the relation between ADC values and cellularity in gliomas. The underlining theory is based on the premise that water diffusivity within the 9 extracellular compartment is inversely related to the content and attenuation of the constituents of the intracellular space. Therefore when cellularity increases, intracellular space volume increases with a relative reduction of the extracellular space, leading to restricted diffusion of water molecules. However other factors may affect the value of ADC in gliomas such as the extracellular matrix which contains various amounts of hydrophilic macromolecules susceptible to change water molecules diffusivity. Hyaluronic acid is one highly hydrophilic component of the extracellular matrix of gliomas and its level of expression changes significantly during the progression to anaplasia in gliomas. Our hypothesis was that hyaluronan may influence ADC values in high and low grade gliomas.

We demonstrated a positive correlation between ADC values and the immunohistochemical level of hyaluronan in glial tumors.

Previous studies have confirmed the utility of positron emission tomography using C11 Methionine (PETMET) as a prognostic tool in patients with gliomas. Higher MET uptake is associated with greater proliferative potential and a higher level of malignancy in gliomas.

The increased aminoacid uptake in gliomas seems to reflect increased transport mediated by aminoacid carriers located in the endothelial cell membrane. Our hypothesis was that CBV measurements, index of tumor vascularity, may be related to tumor aminoacid metabolism.

We found a positive correlation between maximum CBV values and maximum MET uptake values in gliomas.

A limitation to these preliminary studies was lack of direct correlation between MRbased measurements and histologic and metabolic data. Indeed, glial tumors are known for their remarkable tissue heterogeneity across different grades, within the same grade, and even within a single given tumor. Therefore we used image coregistration and stereotactic biopsies to further assess the relationship between MRbased imaging data and both metabolic and histologic analysis.

The second part of our studies was based on measurements at the exact same localization on both MR and PET images where biopsy specimens were performed. We found a local relationship between CBV and MET uptake values. Both measurements correlated with mitotic activity and endothelial proliferation; two features of tumor aggressiveness.

In order to quantify tumor cellularity and tumor angiogenesis, we respectively measured cell density and vessel density using immunohistochemical markers to identify vessels. We found a regional relationship between CBV and cell density, as well as vessel density in gliomas whereas no correlation was found regionally between ADC and cell density.

We concluded that CBV measurements may be used locally as indices of angiogenesis and cellularity in gliomas; whereas local ADC measurements are more variable and may not be used as a marker of cellularity in heterogeneous tumors such as gliomas.