Radiographic Detection and Advanced Imaging of Glioblastoma




Acknowledgments


The authors would like to acknowledge our colleagues and teachers in neuroradiology at the University of Pennsylvania for their insights on brain tumor imaging, particularly Drs Ronald Wolf, Linda Bagley, Suyash Mohan, and John Woo.




Introduction


Neuroimaging plays a critical role at each stage of glioblastoma (GBM) diagnosis and therapy. Imaging provides the first definitive evidence of GBM in most cases, facilitates maximal safe surgical resection of enhancing neoplasm, guides radiation therapy, characterizes residual or recurrent disease, determines progression or response to therapy, and identifies complications of tumor or treatment throughout the disease course. Thus, key features, techniques, and considerations in GBM imaging constitute essential knowledge for all members of the neuro-oncology team. The current chapter presents an overview of the practice and principles of neuroradiology in supporting the comprehensive management of patients with GBM by considering conventional imaging features, advanced imaging approaches, multifocal disease, GBM mimics, recurrent disease and treatment effects, and future directions in imaging.




Conventional imaging features of glioblastoma


Although GBM can present with protean morphologies and locations, nearly all lesions have a constellation of typical imaging characteristics that mirror the pathophysiology of this tumor. Important imaging findings include aggressive infiltration, gyral expansion, hypercellularity, blood-brain barrier (BBB) disruption, and central necrosis. The authors find it helpful to divide these key pathologic and radiologic features into 3 categories: infiltration, cellularity, and vascularity ( Table 7.1 ). GBM has variably extensive infiltrative and cellular nonenhancing components, a property unique to glial tumors and shared by World Health Organization (WHO) grade II and III neoplasms. In addition, variable size and number of enhancing and hypervascular components are present, indicating inflammation with BBB breakdown and abnormal tumor vasculature. Irregular nonenhancing areas within cellular enhancing components indicate necrosis, which is a defining feature of GBM and differentiates this WHO grade IV neoplasm from grade III tumors. Although they may infiltrate widely, including across the corpus callosum, GBMs are usually centered in the white matter of the cerebral hemispheres, sometimes in the thalamus, rarely in the brainstem, and essentially never in the cerebellum in adults.



Table 7.1

Key properties of GBM with pathology-radiology correlates and relevant imaging sequences




































Properties Pathology Radiology Imaging
Infiltration Ill-defined margins
Widespread tumor cells
Nonenhancing signal abnormality
Multifocal/multicentric pattern
Cortical involvement
Spread across corpus callosum
T2/FLAIR
Cellularity Mitoses and cell density a Increased tissue density CT and T2
Mild diffusion restriction DWI
Increased choline level MRS
Vascularity Microvascular proliferation
Inflammation
Necrosis b
Increased perfusion (rCBV, Vp) DSC, DCE
BBB breakdown T1+C
Necrosis T1+C

Abbreviations: CT, computed tomography; DCE, dynamic contrast enhanced; DSC, dynamic susceptibility contrast; DWI, diffusion-weighted imaging; FLAIR, fluid attenuation inversion recovery; MRS, magnetic resonance spectroscopy; rCBV, relative cerebral blood volume; T1+C, contrast enhanced T1-weighted; Vp, plasma volume.

a Other related cellular features of anaplasia, atypia, and pleomorphism do not have specific radiology correlates.


b Necrosis is a symptom of cell proliferation with inadequate or abnormal vascular supply and so is also related to cellularity. Various factors in the tumor microenvironment may contribute, including ischemia, inflammation, and excitotoxicity.



Important differential considerations include other primary glial neoplasms, metastases, lymphoma, subacute infarct, abscess, and tumefactive demyelination. However, knowledge of key radiologic features informed by clinical information makes initial diagnosis of GBM straightforward in most cases ( Tables 7.2 and 7.3 ). The same principles apply when assessing for residual or recurrent disease, except that treatment-related changes complicate interpretation, and often treatment-related changes and tumor are intermixed. Although most GBM imaging focuses on MRI, computed tomography (CT) also deserves consideration as the initial imaging obtained in many patients. In our experience, diagnostic errors in these different modalities run in opposite directions; clinicians more readily attribute CT findings to other disorders and enhancing MRI lesions to GBM. Thus, this chapter emphasizes key differentiating features for both CT and MRI.



Table 7.2

Typical imaging patterns of GBM and other intra-axial lesions




























































Number of Lesions Brain Region Epicenter Cortical Expansion Crosses Midline
GBM Often 1 Almost always supratentorial WM Common Yes
Metastases Sometimes 1 Supratentorial or infratentorial GM-WM junction No No
Lymphoma (primary) Variable Classically periventricular WM Rare Yes
Other glial neoplasms Usually 1 unless gliomatosis or syndrome (NF1, VHL) Supratentorial or infratentorial WM Common Yes
Demyelination Usually 1 when tumefactive Usually supratentorial WM No Yes
Abscess Usually 1 unless hematogenous spread Near routes of infection (frontal sinuses/temporal bones) unless hematogenous WM No No
Infarct Variable Follows vascular territories GM and/or WM Common No

Abbreviations: GM, gray matter; NF1, Neurofibromatosis type 1; VHL, von Hippel-Lindau disease.


Table 7.3

Conventional and advanced imaging features of GBM and other intra-axial lesions




































































Enhancing Component (T1+C) Nonenhancing Component (T2/FLAIR) Hemorrhage (GRE/SWI) Diffusion (DWI/ADC) Perfusion (DSC, DCE, ASL) Metabolism (MRS)
GBM Solid, cystic, necrotic Variable tumor and edema Common Variable Increased Neoplastic spectrum a
Metastases Solid, cystic, necrotic Marked edema Often b Variable Increased Neoplastic spectrum
Lymphoma Solid c Variable edema Occasional Uniform restricted Mildly increased
DSC overshoots
Neoplastic spectrum
Radiation necrosis Solid, cystic, feathery, soap bubble Variable edema Absent Variable Decreased to mildly increased Lipid/lactate
Demyelination Partial ring, spares cortex Chronic lesions Absent Peripheral restricted Normal to decreased Lipid/lactate
Pyogenic abscess Ring Marked edema Typically absent Central restricted Increased peripheral Lipid/lactate, amino acids
Subacute infarct Gyriform Cytotoxic edema Variable Variable Decreased Decreased metabolites

Abbreviations: ADC, apparent diffusion coefficient; ASL, arterial spin labeled; GRE, gradient echo; SWI, susceptibility-weighted.

a Characterized by increased choline/n-acetylaspartate and choline/creatine ratios. For GBM this is seen in enhancing components and adds specificity when present in nonenhancing components.


b Melanoma, renal cell carcinoma, choriocarcinoma, and thyroid cancer are known for hemorrhagic metastases. Lung and breast cancer produce more hemorrhagic metastases overall, because of increased prevalence.


c Peripheral enhancement may be seen with immune compromise, as in acquired immunodeficiency syndrome.



Computed Tomography


CT is the most common neuroimaging modality in the acute setting and usually the first study performed on patients with symptoms of acute intracranial disorder. Depending on tumor location and presence or acuity of hemorrhage or mass effect, the clinical presentation of GBM can range from signs of increased intracranial pressure or herniation to focal neurologic deficits, seizures, functional decline, personality changes, or headaches. Note that the clinical history may be incomplete and should not dissuade the clinician from considering a primary brain tumor, particularly in the primary and secondary care settings, in which stroke, metastases, or hypertensive hemorrhage are typically higher on the differential diagnosis and may bias the provided history. In tertiary and quaternary care settings, patients may present with a known diagnosis of brain tumor on outside institution CT and MRI studies. In these cases, the CT should not be ignored because it provides complementary information that may help refine the differential diagnosis.


Although CT lacks the exquisite soft tissue resolution of MRI, significant diagnostic information can be obtained in most cases. A complete discussion of head CT physics and interpretation is beyond the scope of this chapter. Briefly, the CT scanner consists of 1 or more x-ray sources positioned across a ring from an array of detectors. As in conventional radiographs, high-energy photons are variably blocked (attenuated) as they travel through tissue. Passing the patient through the spinning ring enables attenuation to be calculated (reconstructed) at each point within the volume and a set of cross-sectional, quantitative, gray-scale images to be produced with brightness proportional to density. Reconstructions can emphasize different tissue types, such as soft tissue or bone, and the range and center (window level) of the gray-scale display values can be adjusted to bring out different features. Iodinated contrast increases density wherever it accumulates. Images can be reformatted in multiple planes and rendered to visualize bones, vessels, or other structures in 3 dimensions. CT excels relative to MRI in its spatial resolution, assessment of osseous structures, speed, and availability.


Consistent with its pathology, GBM typically appears as a peripherally isodense to slightly hyperdense, centrally hypodense intra-axial mass compared with normal gray matter ( Fig. 7.1 ). These imaging findings correlate with the pathologic findings of areas of dense cellularity and central regions of necrosis. Findings may be subtle and are usually better seen following careful windowing of CT images to maximize gray-white differentiation. Although regions of central necrosis generally appear hypoattenuating, superimposed hemorrhage can result in varied levels of attenuation.




Fig. 7.1


Conventional imaging in an instructive case of GBM. The patient presented with cognitive decline and expressive aphasia. Unenhanced CT at the level of the insula ( top row ) and temporal lobe ( bottom ) shows areas of increased density in the white matter also extending into and expanding the cortex ( arrows ) reflecting infiltrative tumor. Central low density reflects necrosis with surrounding decreased density reflecting edema superimposed on neoplasm. T2-weighted and enhanced T1-weighted (T1+C) images show corresponding infiltrative tumor and edema with contiguous areas of solid peripheral enhancement and central necrosis. Relative cerebral blood volume (rCBV) is increased in the enhancing areas.


Hypercellular and centrally necrotic areas are surrounded by a variable degree of hypoattenuation reflecting a combination of infiltrative tumor and superimposed edema. Expansion of adjacent cortex may be present, indicating infiltrative tumor and increasing the likelihood of a primary neoplasm (see Fig. 7.1 ; Fig. 7.2 ). The presence or absence of calcium should be noted on CT, because parenchymal calcification in GBM is uncommon but is commonly seen with more slowly growing glial neoplasms, particularly oligodendroglioma and ganglioglioma. Associated mass effect can manifest as asymmetric parenchymal expansion; variable effacement of ventricles, sulci, and cisterns; midline shift or herniation; and obstructive hydrocephalus or ventricular trapping. Intratumoral hemorrhage may also be present and is common in patients presenting with acute symptoms. Most important is to recognize when 1 or more signs of mass effect, edema, cellularity, or infiltration indicate that a scan is not showing a typical case of chronic small vessel ischemic disease and merits further evaluation with enhanced MRI.




Fig. 7.2


Another instructive case of GBM. The patient presented with left facial droop and concern for metastases given a prior cancer history. Unenhanced CT more superiorly ( top row ) and inferiorly ( bottom ) shows a large region of right frontal lobe white matter hypodensity associated with sulcal effacement and midline shift. In addition, there is region of central hyperdensity ( long arrow ) and a subtle ring-shaped density ( short arrow ). Multiplicity and marked edema are good features for metastasis, but localization of numerous metastases exclusively to a single lobe is unusual. MRI is required for further characterization. T2-weighted MRI confirms the white matter signal abnormality and multiple isointense masses. Cortical involvement of the anterior frontal lesion and infiltration into the corpus callosum (far right top row of the coronal T2-weighted image) suggest glioma. Restricted diffusion on diffusion-weighted imaging (DWI) indicates cellularity of the multiple masses with corresponding enhancement, more intense peripherally. The dominant anterior enhancing mass follows rather than displaces the sulci ( arrows , far right bottom row coronal enhanced T1-weighted image), suggestive of infiltration and preferential growth along white matter tracts. Considered in isolation, many features are compatible with metastases or lymphoma, but in combination these conventional imaging findings indicate high-grade glial neoplasm.


Conventional MRI


Contrast-enhanced MRI remains the gold standard for imaging of brain tumors and is critical in assessing the full extent of neoplastic involvement, minimizing the risk of incorrect diagnosis, and identifying potential complications of tumor or therapy. Technical principles of MRI are also beyond the scope of this chapter, although the general terminology of different pulse sequences is likely familiar to clinicians. Magnetic resonance (MR) images are created by measuring the current generated when spinning protons (typically hydrogen nuclei in water molecules) are induced by resonant radiofrequency pulses to flip out of and then return to alignment with a strong magnetic field. T1 and T2 are time constants describing different components of this relaxation and are influenced by the type of tissue in which the water molecules are located. Varying the strength of the magnetic field across the imaged area and the timing of pulses enables localization of signals in 3 dimensions and variable signal intensity according to tissue type. In T1-weighted images, white matter is bright and gray matter dark and the opposite is true for T2-weighted images. Cerebrospinal fluid (CSF), essentially water, is bright on T2-weighted images and dark on T1-weighted images. Typical gadolinium-based MRI contrast agents cause increased signal on T1-weighted images.


Routine brain imaging protocols include T2-weighted, fluid-attenuated inversion recovery (FLAIR; which is like T2 but suppresses signal from simple fluid), diffusion-weighted imaging (DWI), and multiplanar T1-weighted pregadolinium and postgadolinium contrast-enhanced sequences. Susceptibility-weighted or gradient echo sequences may be used to increase sensitivity for blood products and mineralization. Each sequence provides complementary information about the fundamental tissue properties of brain lesions and surrounding brain parenchyma (see Figs. 7.1 and 7.2 ). For glial tumors, both the enhancing potentially necrotic tumor components and surrounding nonenhancing components must be assessed. So-called advanced imaging techniques, which require additional postacquisition processing, can further characterize tumor tissue. These techniques include perfusion and permeability imaging as well as MR spectroscopy (MRS) and are discussed later. DWI could be considered advanced imaging but has become part of routine scans. High-resolution three-dimensional T1-weighted images are often added when the need for operative guidance is anticipated. Similarly, diffusion tensor imaging or functional MRI may be performed to aid surgery depending on tumor location. These topics are discussed in more detail in Chapter 14. Intraoperative MRI is only available at select centers and also discussed separately in Chapter 14.


Nonenhancing Components


Most primary glial tumors have a typical appearance on T2 and FLAIR imaging, with areas of hyperintense infiltrative signal abnormality following white matter tracts and variably associated with focal cortical or gyral expansion. White matter abnormalities are occasionally subtle but are usually readily identifiable when comparing affected regions with more typical-appearing white matter farther from the epicenter of the dominant lesion. FLAIR images generally increase the conspicuity of abnormal signal but the precise extent, especially of cortical involvement, may be better depicted on T2. Therefore, both sequences should be carefully scrutinized to identify the visible margins of the tumor, keeping in mind that tumor cells infiltrate much more widely. In our experience, cortical involvement, when present, is highly specific for glial neoplasm. This finding is in contrast with the increased signal caused by reactive vasogenic edema, which may compress the cortex because of mass effect but remains confined to the white matter. In glial tumors there is usually a combination of infiltrative signal and vasogenic edema causing signal abnormality. Surrounding edema is generally less extensive relative to neoplastic burden than that seen with secondary neoplasms, but can be striking in some cases (as in Fig. 7.2 ).


DWI assesses the flow of water molecules within tissues, producing images that are brighter where diffusion is limited by the microenvironment. Because T2 weighting is inherent in this technique, vasogenic edema and overall increased water content also cause increased signal (T2 shine-through). To account for this, quantitative apparent diffusion coefficient (ADC) maps are calculated from DWI images, showing restricted diffusion as areas of decreased signal intensity. DWI images and ADC maps are interpreted in tandem, looking for areas of increased signal on DWI and corresponding low signal on ADC. For simplicity, some figures in this chapter may show DWI alone but restricted diffusion was confirmed on ADC maps.


Clinicians are most familiar with DWI from its striking sensitivity for acute cerebral infarcts, caused by increased intracellular water in areas where ischemia disrupts cell homeostasis. However, DWI has been used extensively to characterize GBM and other tumors, because the ratio of intracellular to extracellular water is also proportional to cell density and tumor grade. Areas of diffusion restriction usually correspond with enhancing regions, although the presence of restricted diffusion in regions of nonenhancing signal abnormality can help differentiate areas of glial neoplasm from vasogenic edema. Diffusion restriction is also a well-known property of infected (and thus proteinaceous and viscous) fluid collections, including pyogenic cerebral abscess and empyema. The necrotic components of GBM can show restricted diffusion and are sometimes difficult to distinguish from abscess, but associated peripheral enhancement is usually thicker with GBM. In general, the degree of restricted diffusion is less in tumor tissue (less intense on DWI and less dark on ADC) compared with that seen in infarct or abscess.


DWI is affected by magnetic susceptibility, so apparent alterations in peritumoral signal may be caused by paramagnetic substances such as blood products, as seen in hemorrhagic tumors and postoperative studies. In the immediate postoperative period, small areas of restricted diffusion reflecting cytotoxic edema (infarct) can be expected at resection cavity margins, but should be distinguished from neoplasm and artifact from blood products. The purpose of identifying small areas of infarct induced by surgery is to recognize that these areas may enhance in the subacute phase when they might be confused for progressive neoplasm.


Blood products cause characteristic signal changes on T1-weighted and T2-weighted images depending on their age. More sensitive sequences tailored to assess for blood products are helpful and complementary in preoperative and postoperative assessment of GBM. Gradient echo and susceptibility-weighted imaging sequences exploit local magnetic field inhomogeneity produced by blood, mineralization, metal, and gas to better identify these materials. These substances show blooming; that is, conspicuous dark areas on the image extending beyond their physical boundaries, caused by perturbation of the local magnetic field. Note that gradient echo along with spin echo is a fundamental technique for acquiring MR images used in many different sequences, but in this case refers to a particular sequence attuned to hemorrhage.


Contrast Enhancement


Normal cerebral vessels possess specialized tight junctions between endothelial cells that form the BBB, excluding contrast and other molecules from the interstitium. Thus, contrast is particularly useful for brain MRI in general, because parenchymal accumulation highlights areas of BBB breakdown due to pathology, with low background signal. In GBM, contrast enhancement reflects a combination of inflammation causing disruption of the BBB and abnormal, leaky, tumoral vasculature. Contrast-enhancing areas may be solid, cystic with thin walls, or obviously necrotic. The presence of thick, irregular, nodular peripheral enhancement with central nonenhancement indicates necrosis (see Fig. 7.1 ) and is a defining pathologic and radiologic feature of GBM compared with lower grade glial tumors. Regions of enhancement typically correspond with areas of hypercellularity (again seen as mildly restricted diffusion) and perfusional abnormalities on advanced imaging caused by the abnormal vasculature. Increasing patient age, tumor size, cellularity, necrosis, and perfusion are all predictive of higher grade.


Enhancing lesions in GBM can vary from 1 or more small foci, to a dominant lesion with satellite nodules, to a large lobar or transcallosal (butterfly) mass ( Fig. 7.3 ). Although larger masses in general indicate more aggressive tumors, there is interplay between size and location causing symptomatic lesions that prompt imaging. Larger lesions with little nonenhancing signal abnormality are more typical of the mesenchymal GBM subtype. Small enhancing lesions in a background of more extensive nonenhancing tumor may suggest secondary GBMs (proneural subtype) in younger patients. However, most lesions are primary GBMs in older patients and have variable-sized enhancing and nonenhancing components.




Fig. 7.3


Six patients with GBM. Representative postcontrast FLAIR ( left ) and T1-weighted images ( right ). ( A ) Right temporal lesion with cortical involvement. ( B ) Left thalamic multicentric GBM with multiple small rim-enhancing components and extension of nonenhancing neoplasm into the right thalamus through the massa intermedia. ( C ) Classic bifrontal butterfly GBM with extension through the anterior corpus callosum. ( D ) Posterior butterfly GBM extending through the splenium. ( E ) Partially intraventricular GBM protruding into the atrium of the right lateral ventricle. ( F ) Multifocal bifrontal GBM.


GBM may extend to the ventricular margin and show associated ependymal or subependymal enhancement. Sometimes tumoral enhancement is inseparable from choroid plexus, which precludes complete resection. More rarely, GBM shows extra-axial spread, leptomeningeal enhancement, involvement of adjacent vessels or dura, or dissemination in the CSF. Proximity to major vessels or evidence of vascular encasement or invasion should be noted preoperatively. Evidence of leptomeningeal spread or CSF dissemination usually mandates imaging of the spinal cord to assess for drop metastases.




Advanced imaging of glioblastoma


Advanced imaging techniques have become part of the standard evaluation of brain tumors at most academic centers and can be extremely valuable in patient management. The information gleaned from these methods complements conventional imaging in identifying and grading glial tumors, differentiating between other potential causes ( Table 7.4 ), and in general increases diagnostic confidence. In certain cases, advanced imaging findings can dramatically alter the interpretation of a study. Perhaps the best example of this is the use of perfusion imaging to diagnose pseudoprogression in patients treated with chemoradiation when conventional contrast-enhanced images suggest tumor recurrence (see Fig. 7.9 ).


Tags:
Mar 19, 2019 | Posted by in NEUROSURGERY | Comments Off on Radiographic Detection and Advanced Imaging of Glioblastoma

Full access? Get Clinical Tree

Get Clinical Tree app for offline access