Benign and Malignant Intracranial Tumours in Adults

Structural Magnetic Resonance Imaging

A standard clinical MRI protocol for brain tumour imaging should include T ₂ weighted, fluid-attenuated recovery (FLAIR) sequences, DWI, T ₁ weighted images before and after the intravenous (IV) administration of gadolinium contrast medium. Brain tumour patients generally require long-term follow-up, with intervals scheduled according to the expected tumour growth rate and whether recent changes have been observed.

To ensure accurate serial imaging comparisons, MR image parameters (e.g. contrast dosage and timing) should be kept as consistent as possible. Where available, 3D imaging increases small lesion visibility ( Fig. 55.1 ) and can support volumetric tumour measurements.

T ₁ and T ₂ signals vary amongst brain tumour types dependent on tissue density, solid versus cystic components, haemorrhage and/or mineralisation. On FLAIR images, intrinsic lesions often stand out from surrounding brain parenchyma. This additional contrast is valuable but non-specific, and in some cases, FLAIR demonstrates grey-white matter detail less well than T ₂ . Intratumoural blood product and calcification are most conspicuous and hypointense on T ₂ * or susceptibility-weighted images (SWI), where magnetic susceptibility effects are stronger. Hyperintensities on T ₁ images can be due to haemorrhage, atypical calcification, melanin (in metastatic melanomas) or fat. A typical combination of brain tumour site, morphology and signal composition will often substantially reduce the list of differential possibilities, or even generate a single diagnosis.

The presence and pattern of gadolinium uptake can give additional information about the nature of brain tumours, but in many cases this is not specific. Assessing for contrast uptake is also important in post therapy assessment to identify recurrent enhancing disease, and whenever infection is a possible differential. For repeated examinations, potential risks of gadolinium deposition need to be considered; however, there is currently no evidence against the serial use of macrocyclic agents in this situation.

In challenging cases, the visibility of contrast enhancement on MR can be improved by doubling or tripling the gadolinium dose, or by using high-relaxivity gadolinium compounds. In addition, post-contrast FLAIR sequences are excellent for identifying subtle leptomeningeal disease.

MRI can be integrated into neurosurgical navigation systems; furthermore, some institutions possess intraoperative MRI facilities to monitor the resection extent in near real time while preserving eloquent regions of the brain.

Advanced Physiological and Molecular Imaging Methods

Advanced MRI modalities aim to detect specific pathophysiological or metabolic processes that provide complementary information about brain masses, either to predict tumour type ± malignant potential or to assess treated lesions when there is a question regarding tissue viability. To maximise the value of such additional imaging time, the choice of test(s) must be tailored to the clinical situation.

Magnetic Resonance Perfusion-Weighted Imaging

A number of primary intracranial tumours and some metastases exhibit increased vascularity. This does not always equal malignancy, but it can be associated with rapid proliferation if this induces new vessel formation.

There are three methods of PWI:

• 1.

  Dynamic susceptibility-weighted contrast-enhanced (DSC) imaging exploits the susceptibility effects of gadolinium, which (because of its paramagnetic properties) causes a transient signal reduction on T ₂ * weighted images during the passage of a gadolinium bolus. With DSC, a volume of T ₂ * images is repeatedly acquired at rapid (1 to 2 seconds) intervals before and during the contrast injection.

• 2.

  Dynamic contrast-enhanced (DCE) imaging measures the increase of signal intensity on a series of T ₁ weighted images before and during gadolinium administration.

• 3.

  Arterial spin labelling (ASL) uses magnetically labelled blood as endogenous tracer to assess blood flow and does not require an injection of a contrast medium.

DSC is currently the most widely used and best validated perfusion technique in brain tumours. DSC-derived cerebral blood volume (CBV) measurements have been shown to consistently correlate with angiographic and histological markers of tumour neovascularity ( Fig. 55.2 ). For example, DSC is useful to distinguish glioblastoma from slow growing gliomas. To avoid erroneously low measurements due to gadolinium leakage in tumours with blood-brain barrier breakdown, CBV is mathematically normalised (relative cerebral blood volume [‘rCBV’]) relative to normal brain tissue. In addition to this mathematical leakage correction, the administration of a ‘preload’ bolus and the choice of a suitable flip angle can help minimise leakage errors. Underestimation of rCBV may also occur due to susceptibility effects; therefore rCBV measurement should be avoided in such areas. If susceptibility is present in the area of concern (e.g. after surgical intervention), an alternative perfusion technique should be considered in addition.

In DCE PWI, T ₁ weighted images are repeatedly acquired beyond the duration of the first pass of a gadolinium bolus, typically for about 5 minutes. The shape of the time–signal intensity curve is influenced by tissue perfusion, vascular permeability and the extravascular-extracellular space. Several mathematical models (e.g. ‘extended Tofts’) can be used to quantify contrast leakage into the extravascular space as a measure of microvascular permeability. The most frequently used parameter is the transfer coefficient K ^trans , which is influenced by endothelial permeability, vascular surface area and flow. DCE images also can be analysed using a model-free approach by looking at the slope of the time–signal intensity curve. DCE quantification is more challenging and therefore it is less widely used than DSC, but DCE has greater spatial resolution and is not hampered by susceptibility effects.

ASL uses labelling of blood hydrogen and requires no contrast injection; therefore it is suitable for children and patients in whom gadolinium is contraindicated. It is the only perfusion technique that directly measures regional cerebral blood flow (rCBF). As hydrogen is freely diffusible and crosses the blood–brain barrier immediately, it is not possible to measure tissue blood volume in the same way as with DSC PWI where the tracer (gadolinium) stays (or is assumed to stay) predominantly in the intravascular compartment. A good correlation between rCBF measurements obtained with ASL and rCBV measurements derived from DSC has been demonstrated so that the use of ASL in neuro-oncology is likely to increase.

Magnetic Resonance Spectroscopy

MRS analyses the metabolite distribution within a chosen volume of brain tissue, thereby providing semiquantitative information. The commonest pattern in brain tumours is a decrease in N -acetylaspartate (NAA), a neuron-specific marker, and creatine, accompanied by elevation of choline, without the presence of lactate, or lipids ( Fig. 55.3 ). The concentration of choline reflects the turnover of cell membranes (due to accelerated synthesis and destruction) and is amplified in tumour regions with high mitotic activity. Lactate is the end product of non-oxidative glycolysis, which is a hallmark of cancerous metabolism that can be exacerbated by hypoxia in fast growing tumours. Tumour hypoxia has been recognised as a major promoter of tumour angiogenesis and invasion. Lactate is probably associated with viable but hypoxic tissue, whereas mobile lipids are thought to indicate tissue necrosis with the breakdown of cell membranes.

The choice of echo time (TE) is an important technical consideration for performing MRS. TE can be short (20 to 40 ms), intermediate (135 to 144 ms) or long (270 to 288 ms). MRS with a short TE has the advantage of demonstrating additional metabolites, which may improve tumour characterisation, such as myo -inositol, glutamate/glutamine and lipids, but is susceptible to baseline distortion and artefactual NAA peaks. Intermediate echo times produce a better baseline, quantification of NAA and choline is more accurate and reproducible and lactate appears (inconsistently) inverted from the baseline. Long echo times are less commonly used for brain tumour imaging and are inferior at demonstrating some of the minor metabolites.

MRS can be valuable in distinguishing brain tumours from non-neoplastic conditions. It has a less well-defined role in distinguishing tumour types and posttreatment effects. Single voxel spectroscopy is technically easier to achieve good-quality spectra, but it assumes an average of metabolites within the measured area and cannot appreciate tumour spatial heterogeneity. Multivoxel techniques, including chemical shift imaging (CSI), are more challenging but provide 2D or 3D spectra with larger area coverage . With modern 3T MRI systems it is possible to obtain a 16 × 16 × 16 array of spectra with a voxel size of 1 cm ³ in 5 to 10 min.

Magnetic Resonance Diffusion Imaging

DWI measures the movement of water molecules within brain tumours, which is influenced by tissue microstructure, particularly cell density and matrix composition. Standard DWI uses three diffusion directions with two or three b values, typically b0, (± b500) and b1000 mm/s ² . Through mathematical subtraction of T ₂ effects, the apparent diffusion coefficient (ADC) map is calculated that describes the overall water diffusivity in the brain. ADC measurements have been shown to correlate inversely with proliferative indices (Ki-67) in gliomas. There is evidence that lower ADC values (without restricted diffusion) may help identify malignant gliomas, and markedly reduced diffusivity may define certain neoplasms such as lymphoma.

Diffusion tensor imaging (DTI) incorporates additional diffusion directions (6+) to gain information about the directionality of water diffusion. The tendency of water to move in some directions more than others is called anisotropy and can be quantified as fractional anisotropy (FA). With intrinsic tumours, normal white matter tracts show a high degree of anisotropy that can be lost if they become infiltrated with cells destroying the ultrastructural boundaries formed by myelin sheaths.

DTI can be post processed to depict key white matter tracts and their connections (tractography) for display in the form of direction-encoded colour images. Tractography is used widely for preoperative planning, as it helps to depict the relations between intrinsic brain tumours and surrounding neural connections. DTI-based advanced diffusion methods are being trialled in research to enhance the microstructural analysis of certain brain tumours.

Functional Magnetic Resonance Imaging

Functional magnetic resonance imaging (fMRI) is performed by detecting blood oxygen level-dependent (BOLD) imaging changes during various forms of brain activity. The most commonly performed technique is task-based fMRI for presurgical assessment (paradigms using motor tasks, language and speech production, and memory produce activation of relevant cortical areas). fMRI is used almost exclusively for preoperative planning, most commonly for language lateralisation. It also supports the localisation of eloquent (e.g. motor) cortical regions that may have been displaced, distorted or compressed by tumour, and to identify evidence of functional reorganisation. fMRI can improve the safety of surgery and allow for a more radical resection. The BOLD effect is an indirect measure of neuronal activity that may be influenced by numerous physiological factors as well as the MR relaxation properties of soft tissue, such that activation within a tumour may partially reflect angiogenesis rather than eloquent function alone, and susceptibility effects from blood products, for instance, may mask areas of brain activation. Other important caveats include the inability of discriminating between indispensable and expendable brain regions, and limitations in predicting the distance between a lesion and an area of functional activity. Non-BOLD techniques, such as ASL, may potentially reduce these pitfalls in the future. Where relevant, fMRI may be combined with MR tractography to minimise intra-operative injury to white matter tracts connected to the eloquent cortical areas ( Fig. 55.4 ).

Positron-Emission Tomography

The positron emission tomography (PET) image signal is generated through the administration of a radioactive tracer that accumulates preferentially within tumour tissue. The most widely used PET tracer in oncological imaging is fluorodeoxyglucose [ ¹⁸ F] (FDG), which provides a measure of cellular glucose uptake. Its principal limitation in neuro-oncology is the high physiological baseline glucose metabolism of the brain, which may result in a poor lesion to background contrast. Newer PET radiopharmaceuticals for brain tumour imaging include amino acid analogues such as [ ¹¹ C] methionine (MET), [ ¹⁸ F]fluoroethyl- l -tyrosine (FET) and, more recently, [ ¹⁸ F]-dihydroxyphenylalanine (FDOPA). Radioactively labelled choline, [ ¹¹ C] choline (CHO) or [ ¹⁸ F] choline, can be used to assess membrane cell turnover and 68Ga-DOTATOC shows a high uptake in meningiomas due to somatostatin receptor expression.

PET imaging is used in combination with computed tomography (PET/CT) and in the form of PET/MRI systems, allowing simultaneous acquisition and registration of high-resolution MR sequences and molecular information.

World Health Organization Classification of Central Nervous System Tumours 2016

Brain tumours can be classified according to location within (‘intra-axial’) or external to (‘extra-axial’) brain tissue, and by tissue origin (e.g. glial, meningeal, pituitary, etc.). Traditionally, histopathological analysis was limited to a visual assessment of individual cell morphology, combined with grading according to the presence of proliferative features, such as high mitotic rate, cellular atypia and new vessel formation.

The recognition that genetic mutations occur frequently in many human cancer types has led to fundamental changes in brain tumour diagnosis. In 2009, a whole-genome sequencing study (The Cancer Genome Atlas) identified a number of common mutations of key prognostic value in different brain neoplasms. For several tumour types, most notably gliomas, this information is reflected in the updated 2016 World Health Organization (WHO) classification, which now mandates molecular testing as part of an integrated diagnosis. Histological grade is still considered important, but has become complementary to genetic tumour characteristics. With this regrouping according to shared molecular features, a number of entities (e.g. ‘primitive neuroectodermal tumour’) have been discontinued in the 2016 WHO update. Moreover, in the WHO 2016 classification, gliomatosis cerebri is no longer a glioma subtype but instead is considered as a growth pattern. Several new adult brain tumour classes have been introduced, which are highlighted in the relevant sections. With this change, radiological work-up has experienced a corresponding paradigm shift, whereby emphasis is placed on predicting tumour molecular status (‘radiogenomics’) in addition to the WHO grade.

An outline of the WHO 2016 integrated classification for central nervous system (CNS) tumours is shown in Table 55.2 . The upcoming chapters will discuss intra-axial and extra-axial brain tumours, highlighting typical and non-specific imaging appearances on conventional and advanced imaging.