ABBREVIATIONS

• ADC

  apparent diffusion coefficient

• ASL

  arterial spin labeling

• BBB

  blood–brain barrier

• BOLD

  blood oxygen level dependent

• CNS

  central nervous system

• DCE

  dynamic contrast enhanced

• DMN

  default mode network

• DSC

  dynamic susceptibility contrast

• DTI

  diffusion tensor imaging

• DWI

  diffusion-weighted imaging

• FA

  fractional anisotropy

• FLAIR

  fluid attenuated inversion recovery

• fMRI

  functional MRI

• MD

  mean diffusivity

• MGMT

  O⁶-methylguanine DNA methyltransferase

• MRI

  magnetic resonance imaging

• MRS

  MR spectroscopy

• NAA

  N-acetylaspartate

• PNET

  primitive neuroectodermal tumor

• PWI

  perfusion-weighted imaging

• RANO

  Response Assessment in Neuro-Oncology

• rCBV

  relative cerebral blood volume

• rs-fMRI

  resting state functional MRI

• RSN

  resting state network

• SMART

  stroke-like migraine attacks after radiation therapy

• SWI

  susceptibility-weighted imaging

• WHO

  World Health Organization

• 3-D

  3-dimensional

Since the discovery of X-rays more than a century ago, radiology has played an integral role in the diagnosis, monitoring, and treatment planning of intracranial masses. While much progress has been made in both clinical medicine and imaging methodologies since the days of skull radiographs, accurate noninvasive diagnosis and assessment of therapeutic response remain the fundamental goals of imaging in patients with brain tumors. MRI, the mainstay of modern neuroimaging, permits superior structural characterization while also capturing the cellular, vascular, metabolic, and functional properties of brain tumors.

The focus of this review is to provide an overview of the current state of adult brain tumor imaging as it relates to neurosurgical practice. The conventional structural imaging features of brain tumors will be presented and complemented by discussion on advanced imaging methods for surgical planning including perfusion mapping, MR spectroscopy, diffusion tensor imaging, and functional MRI. Lastly, the challenges in post-therapy imaging of brain tumors will be reviewed with specific emphasis on pseudoprogression and pseudoresponse.

CONVENTIONAL MAGNETIC RESONANCE IMAGING FEATURES

Despite the myriad refinements in advanced imaging techniques over the past decades, conventional structural magnetic resonance imaging (MRI) remains the standard of care imaging method for neuro-oncologic practice. Current consensus recommendations for a standardized brain tumor MRI protocol are the following: 3-dimensional (3-D) T1, axial fluid-attenuated inversion recovery (FLAIR), axial diffusion-weighted imaging (DWI), axial gadolinium contrast-enhanced T2, and 3-D gadolinium contrast-enhanced T1, performed on a minimum 1.5 tesla MR system.¹ If 3-D sequences cannot be performed due to time constraints or technical limitations, 2-D sequences can be substituted. The structural sequences (T2-weighted, FLAIR, and pre- and postcontrast T1-weighted) provide the primary foundation of an MRI examination. Specific presurgical sequences such as high-resolution isovolumetric 3-D T2-weighted and postcontrast 3-D T1 spoiled gradient recalled echo imaging can be obtained with fiducial markers for intraoperative navigation or with a head frame for stereotactic radiosurgical planning.^2–6 In addition to conventional structural sequences, DWI and T2*-weighted imaging, such as susceptibility-weighted imaging (SWI), are usually performed as part of the routine brain MRI examination. An overview of the MRI techniques discussed in this review and their clinical utility is presented in the Table.

Structural MRI

The primary roles of structural MRI in the initial brain tumor evaluation include determining the lesion location, extent of tissue involvement, and resultant mass effect upon the brain, ventricular system, and vasculature.⁷While identifying an accurate histological tumor type can be challenging on the basis of imaging alone, the correct diagnosis can often be suggested in a short list of differential considerations, particularly when imaging features are considered in the context of patient age, symptom duration, presence of extracranial primary malignancy, and history of prior radiation therapy to the brain.

MRI offers superior soft tissue contrast over other cross-sectional imaging techniques allowing for better visualization of subtly infiltrated or disrupted parenchymal architecture. Furthermore, intravenous gadolinium-based contrast agents shorten T1 relaxation times and increase tissue contrast by accentuating areas where contrast agents have leaked out of the blood–brain barrier (BBB) into the interstitial tissues, resulting in parenchymal enhancement. This breakdown of the BBB is a key feature seen in tumors as well as non-neoplastic conditions.^8,9 Within diffuse gliomas, contrast enhancement is positively correlated with tumor grade, although a few high-grade gliomas may show no or minimal enhancement and certain lower grade gliomas (World Health Organization [WHO] grade I) such as pilocytic astrocytoma or ganglioglioma can enhance avidly (Figure 1).⁹ The region of T2/FLAIR hyperintense signal abnormality surrounding the enhancing tumor core is typically referred to as peritumoral edema and can be vasogenic or infiltrative in nature. Vasogenic edema represents a reactive increase in extracellular water due to leakage of plasma fluid from altered tumor capillaries in the absence of tumor cells and is seen around intracranial metastases or noninfiltrative extra-axial tumors such as meningiomas. Infiltrative edema in gliomas represents a mixture of vasogenic edema and infiltrating tumor cells invading along, but not necessarily disrupting, white matter tracts and can be considered nonenhancing tumor owing to preserved integrity of the BBB.^10,11 In fact, in many gliomas, the T2/FLAIR hyperintense signal abnormality may be indistinguishable from the primary mass lesion.^12–14

Primary lesion location can help differentiate between tumor types. For example, extra-axial tumors such as meningiomas, schwannomas, and skull base tumors can generally, but not always, be differentiated from intra-axial tumors based on associated interposition of cerebrospinal fluid, vessels, or dura between the mass and cortex.¹⁵ Similarly, the number of lesions is an important consideration as multiple lesions suggest metastatic disease or non-neoplastic processes such as demyelination, inflammation, or infection.^9,16 Finally, several imaging characteristics suggest tumor subtypes. The combination of a cyst and solid nodule within a tumor suggests brain tumors such as ganglioglioma, pilocytic astrocytoma, pleomorphic xanthoastrocytoma, and, in the posterior fossa, hemangioblastoma.¹⁷Calcifications can be seen in oligodendrogliomas, ependymomas, and pineal tumors, among others.¹⁸ Necrosis and hemorrhage are seen with higher grade gliomas, certain metastases, and rarely central nervous system (CNS) lymphoma in immunocompromised patients.^19–21

Historically, brain tumors have been classified based on histology according to the WHO criteria.²² Diffuse gliomas are further subdivided into 4 grades by various histological features such as cellularity, nuclear atypia, mitotic activity, pleomorphism, vascular hyperplasia, and necrosis. Grade I gliomas including pilocytic astrocytoma, pleomorphic xanthoastrocytoma, and subependymal giant cell astrocytoma share a relatively benign biology with an indolent clinical course that is distinct from other diffuse infiltrating glioma grades.^23,24 Grade II-IV gliomas are heterogeneous tumors with variable degrees of infiltration, atypia, and mitotic activity. Microvascular proliferation with endothelial hyperplasia and pseudopallisading necrosis are the defining histological hallmarks of grade IV gliomas, frequently referred to as glioblastomas. Recent insights into tumor biology have led to the identification of several molecular aberrations associated with genetic phenotypic differences in brain tumors.^25–31 The updated WHO classification incorporates molecular markers along with histology and defines specific entities on the basis of IDH mutation and 1p19q chromosomal deletion.³²These, along with other molecular markers including p53, RB1, EGFR, PTEN, MGMT, BRAF, ATRX, TERT, and histone H3, represent a nosological shift where histopathological phenotype is complemented by molecular genetic phenotype to better classify brain tumors and predict their clinical behavior.³³MRI is rapidly catching up with these genetic advances and helping to noninvasively explore the link between the molecular genetic basis of glioma biology and the imaging characteristics of their morphological phenotypes.

Diffusion-Weighted Imaging

While primarily used in the setting of suspected acute stroke, DWI offers significant value in the evaluation of brain tumors. DWI probes the random (Brownian) motion of water molecules allowing for the assessment of tumor cellularity, peritumoral edema, regions of tumor hypoxia, integrity of white matter tracts, and postoperative injury. Corresponding apparent diffusion coefficient (ADC) values, reflecting the magnitude of diffusivity, are derived for each voxel and displayed as a calculated ADC map.^34–36 Apart from characterizing tumor, DWI can be used to detect non-neoplastic processes such as tumefactive demyelination or infection.^37–39

In the pretreatment evaluation of brain tumors, DWI best serves to characterize tumor cellularity on the premise that water diffusivity within the extracellular compartment is inversely correlated to the volume of the intracellular space. Low ADC values, representing decreased water diffusivity, can be used to suggest highly cellular tumors such as lymphoma, medulloblastoma, or primitive neuroectodermal tumor (PNET) (Figure 2).^40–42Additionally, low ADC values can be used as a surrogate for increasing tumor grade or as an independent biomarker signifying poor outcomes both in glioma and lymphoma.^43–45 ADC values have also been used to better localize tumor infiltrated foci among regions of vasogenic edema to better direct tissue sampling and therapy.^46–48 Because of the heterogeneous nature of intracranial tumors, particularly gliomas, histogram analysis can be employed to better assess ADC metrics.^43,49–51 Although some authors have demonstrated good correlation between cell density and ADC values based on stereotactic biopsy, the required postprocessing and overlap in ADC values between tumor grades limit the role of quantitative ADC in clinical practice.

Susceptibility-Weighted Imaging

High-resolution 3-D T2* gradient echo sequences such as SWI are highly sensitive to magnetic susceptibility effects from blood products or mineralization. This technique is useful to depict internal vascular architecture and hemorrhage in tumors, which can be used to suggest grade, as well as calcification, which can be used to narrow the differential diagnosis (Figure 3). Both blood products and mineralization appear dark on magnitude images and can be differentiated on filtered phase images in which paramagnetic blood products appear dark and diamagnetic calcium appears bright (Figure 4).^52,53 Minimum intensity projection images can also be reviewed to more clearly visualize normal venous structures, tumoral vascularity, and parenchymal foci of susceptibility.⁵⁴

PRESURGICAL PLANNING TECHNIQUES

Ongoing challenges in the surgical management of brain tumor patients include selecting an appropriate site for tissue sampling and balancing extent of resection with preservation of eloquent function. In the majority of brain tumors, management focuses on maximal safe resection. However, due to the high variability in eloquent cortex between patients, conventional morphological imaging is not sufficient to predict postsurgical deficits. Advanced MRI techniques, such as perfusion-weighted imaging (PWI), MR spectroscopy (MRS), diffusion tensor imaging (DTI), and functional MRI (fMRI), are used alongside intraoperative cortical mapping to guide the degree of resection for the best clinical outcome.^33,42,55

MR Perfusion Imaging

Several MR perfusion techniques are currently employed: dynamic susceptibility contrast (DSC), dynamic contrast-enhanced (DCE), and arterial spin labeling (ASL). Of these, DSC perfusion is the most studied and widely applied, while ASL, which does not require intravenous contrast, has been the subject of increasing investigation and clinical implementation.^56,57

DSC is based on the detection of susceptibility induced signal loss on T2*-weighted sequences after the administration of an intravenous gadolinium contrast agent. A signal intensity time curve is generated from which relative cerebral blood volume (rCBV) and other perfusion metrics are derived. rCBV is elevated in tumor, where it is seen as a marker of angiogenesis, and has been shown to distinguish tumor from non-neoplastic etiologies with lower rCBV such as demyelinating lesions. A signal intensity time curve that does not return to baseline is seen with leaky capillaries and can suggest metastasis, meningioma, or choroid plexus tumor.⁵⁷ rCBV has been positively correlated to glioma grade, although some lower grade gliomas such as oligodendrogliomas may have elevated rCBV.^44,58 rCBV has been noted to be increased in infiltrative edema of gliomas compared to acellular vasogenic edema surrounding metastases, a characteristic which may be used to better target biopsy.⁵⁹ rCBV may also predict areas of progression in glioma prior to changes on contrast-enhanced MRI as well as survival.⁶⁰

The underlying principle behind DCE is that disordered tumor vasculature permits intravascular contrast diffusion into the interstitial compartment which is then quantifiable over a dynamic MR acquisition. The volume transfer constant or k^trans, a measure of capillary permeability, is the primary metric derived from DCE perfusion. k^trans can be used to grade tumors, particularly gliomas, as gliomas with increased capillary permeability are more likely to be higher grade than lower grade. Another metric quantified by DCE is v_e, an estimate of fractional extracellular extravascular space, which has been shown to be related to tumor cellularity, though a strong relationship has not clearly been established.^61–65 Currently, DCE has not gained widespread clinical use because of challenges in acquisition and analysis techniques.

ASL is a noninvasive perfusion imaging technique which quantitatively measures cerebral blood flow. It uses an inversion pulse to label inflowing blood proximal to the area of imaging with subsequent subtraction of these labeled spins from control static images. ASL is of particular clinical interest due to its noncontrast technique, relative speed, ability to image the whole brain, and minimal postprocessing.^66,67 Several studies have shown a promising role for ASL in quantitative characterization of tumor vascularity in meningioma, metastasis, and high-grade glioma as well as in its ability to differentiate high- from low-grade gliomas based on a degree of microvascular proliferation (Figure 1).^68–72

MR Spectroscopy

MRS provides insight into the biochemical profile of interrogated brain tissue. Proton (¹H) MRS is the most studied technique and can be performed with long (288 or 144 ms) and short (35 ms) echo times. MRS can be obtained using a single-voxel technique to a targeted region of interest or a multivoxel technique to cover a broader area and better evaluate regional biochemical differences. The most recognizable metabolite peaks on long echo ¹H MRS include N-acetylaspartate (NAA) at 2.0 parts per million (ppm), creatine (Cr) at 3.0 ppm, choline (Cho) at 3.2 ppm, and myo-inositol (MI) at 3.5 ppm. NAA is a marker of neuronal viability, Cr reflects normal cellular metabolism, Cho is a marker of cell membrane turnover, and MI reflects astrocyte integrity. Lipid and lactate, which have a broad peak at 1.3 ppm, are not seen in normal tissue and considered markers of necrosis and hypoxia, respectively. A normal spectrum demonstrates upward sloping of peaks from myo-inositol to choline, forming the so-called Hunter’s angle of approximately 45°.^73–75

Brain tumor spectra reflect cellular turnover and loss of normal neuronal metabolites, typically as elevated Cho and decreased NAA resulting in a downward sloping appearance of metabolite peaks or reversal of the usual Hunter’s angle.⁷⁵ Generally, absolute heights of metabolite peaks are not used, and rather the peaks are analyzed as ratios such as Cho/NAA and Cho/Cr.^74,76MRS has been shown to differentiate gliomas by grade on the basis of a positive correlation between Cho/NAA and Cho/Cr ratios and grade.^77–80Additionally, lower grade gliomas have been associated with elevated MI/Cr ratio.⁸¹ Within regions of nonenhancing signal abnormality, elevated Cho/NAA and Cho/Cr ratios have been observed in infiltrative edema compared to vasogenic edema reflecting the increased cellularity underlying the signal abnormality (Figure 5). In this way, MRS can be used to differentiate glioma from noninfiltrative tumor such as metastasis or for biopsy targeting and treatment planning in glioma.^82–86

Widespread adoption of MRS is limited by technical issues such as variability in acquisition techniques, differences in metabolite ratio calculations, and volume averaging due to lesion location or voxel size. Despite these challenges, MRS is able to add specificity to conventional MRI and remains an area of intense investigation that, with further refinements, will see increased clinical adoption.

Diffusion Tensor Imaging

An advanced application of diffusion imaging is DTI, which interrogates the 3-D shape of diffusion using both diffusivity (eigenvalues) and direction (eigenvectors). The principle metrics obtained from DTI include mean diffusivity (MD) and fractional anisotropy (FA). In presurgical planning, DTI-based tractography is used to guide surgical resection by analyzing the integrity of white matter fiber trajectory in order to determine whether there is tumor invasion or tumor displacement of the adjacent white matter tracts (Figures 6 and 7).^87,88

FA represents the degree of directionality of water diffusion and in the normal brain reflects the presence of intact myelinated white matter tracts. In brain tumors, disrupted cellular architecture results in altered FA that correlates to cellularity.⁶⁹ Longer progression-free survival and overall survival were seen in glioblastoma patients in whom more DTI abnormality was resected. Additionally, FA has been reported to be increased in the infiltrative peritumoral edema surrounding high-grade gliomas as compared to the vasogenic edema surrounding metastases.^89,90 Often, tumor boundaries are not clearly delineated by conventional imaging, and DTI tractography may improve border characterization leading to greater resection and improved outcomes.^87,91 The identification and preservation of white matter tracts is also important in preserving the neurological functional integrity of patients undergoing resection of lesions near eloquent cortex.

DTI-based tractography is fundamentally limited because a single tensor can only resolve a single fiber direction within an imaging voxel, while up to 90% of white matter voxels in the brain may contain multiple fibers.^92,93 As a result, many higher-order models are being investigated to address the so-called crossing fiber problem.⁹⁴ However, these developments have not yet translated into improvements in clinical practice.

Functional MRI

fMRI indirectly measures neuronal activity using the ratio of deoxyhemoglobin to oxyhemoglobin as a contrast mechanism, known as blood oxygen level dependent (BOLD) signal. fMRI can be used to for sensory motor, language, and memory mapping—all of which have important implications for presurgical planning and intraoperative navigation.^95–97

In task-based fMRI, the patient alternates between a passive resting state and task performance, usually motor or language function, while relative changes in BOLD signal are measured and used to infer areas of cortical activation (Figure 8). Anatomic areas localized with task-based fMRI have been validated to approximate functional sites identified with cortical stimulation mapping. Apart from localizing eloquent cortex, task-based fMRI can be used to characterize tumors. Decreased BOLD signal is noted in cortex involved by tumor and differences are also seen between high- and low-grade tumor suggesting alterations in cerebral blood volume of the tumor affected area.^98,99fMRI can also be applied to guide DTI by delineating a seed region for fiber tractography.^100–102

Recently, there has been increased interest in resting state functional MRI (rs-fMRI), which does not require patient cooperation with task paradigms and can be performed under anesthesia. rs-fMRI detects spontaneous low-frequency fluctuations in the BOLD signal between regions that are spatially distinct to identify functional networks, so-called resting state networks (RSNs).^103,104 The most fundamental RSN is the default mode network (DMN) and evidence regarding other RSNs including somatosensory, visual, auditory, language, attention, and cognitive control networks is evolving.¹⁰⁵ Compared with task-based fMRI, rs-fMRI has the ability to identify many networks simultaneously, thereby providing more comprehensive information on the functional architecture of the brain while reducing imaging time. Although the bulk of investigation has focused on functional connectivity and cognition, a few small studies have reported the use of rs-fMRI to depict changes in vascular physiology and tumor grade as well as predict postsurgical neurological changes.^106,107 While these studies are promising, further work is needed before rs-fMRI can be used routinely in the clinical setting.

IMAGING OF TREATMENT RESPONSE

Assessing brain tumor treatment response by MRI presents considerable challenges—not the least of which is determining tumor growth—and is fraught with pitfalls such as differentiating progression from treatment-related changes.

Current standard of care for glioblastoma involves maximal safe resection followed by concurrent chemoradiotherapy and adjuvant temozolomide.¹⁰⁸Progressive disease is treated with antiangiogenic agents (eg, bevacizumab) and/or nitrosourea alkylating agents (eg, lomustine).^109,110 Gliomas of lower grades are treated with resection and some combination of chemoradiotherapy as adjuvant therapy or for recurrence.^111,112 Treatment of metastases primarily depends on their number, with solitary lesions amenable to resection whereas multiple lesions are often treated radiosurgically.^113–115 CNS lymphoma is treated with steroids and methotrexate with radiotherapy usually reserved for recurrent, chemotherapy-resistant, disease.^116,117 Moreover, numerous investigational therapies are available for a variety of indications. Any one of these therapies can either mimic or mask disease progression. And despite our state-of-the-art imaging techniques, serial imaging is often the most helpful and reliable noninvasive method to assess disease activity.

Brain tumor follow-up imaging reflects both treatment effect and natural evolution of tumor. Typically, increasing contrast enhancement and increasing nonenhancing signal abnormality represent progressive disease (Figure 9). Increasing contrast enhancement is particularly concerning for progression if it is seen at locations distant from the treatment site. However, it is important to keep in mind that small areas of reduced diffusion surrounding the resection cavity noted on immediate postoperative MRI, representing devitalized tumor or ischemic brain, often develop contrast enhancement at short term follow-up. Progressive nonenhancing disease can be suggested if the abnormality is of intermediate T2/FLAIR signal intensity, exerts mass effect upon adjacent structures, involves the cortex, or is associated with reduced diffusion or elevated perfusion. Furthermore, a spectrum of changes, not infrequently mimicking progression or response to therapy, are seen on MRI in the weeks to months to years following chemoradiation.^118,119

Pseudoprogression

Pseudoprogression, an inflammatory response marked by a transient increase in contrast enhancement and edema upon completion of chemoradiotherapy, is observed in up to 30% of high-grade glioma patients and can also be seen in the setting of low-grade glioma. The hallmark of pseudoprogression is subsequent stabilization or improvement of contrast enhancement at follow-up MRI (Figure 10). Pseudoprogression occurs more frequently in tumors harboring O⁶-methylguanine DNA methyltransferase (MGMT) promoter methylation and is associated with improved survival.^120–126 Pseudoprogression typically occurs within the first 3 mo following therapy and it is thought to represent a milder form of radiation necrosis, which manifests as a mass lesion with an appearance similar to recurrent tumor months to years postirradiation.^127,128 Since both pseudoprogression and true tumor progression share pathophysiology characterized by an underlying disruption of the BBB, it is difficult to differentiate the 2 processes using conventional imaging. In light of this difficulty, the updated Response Assessment in Neuro-Oncology (RANO) criteria stipulate that within the first 12 wk following completion of radiotherapy, progression can only be determined if new enhancement is seen outside of the radiation field or if there is histopathological confirmation of tumor growth.¹⁴ Unfortunately, misdiagnoses can lead to under or overtreatment with potentially devastating clinical consequences.

Due to inherent risks of re-operation, major efforts have been made to better characterize increased contrast enhancement seen on postradiotherapy MRI. Although there are no definitive conventional MRI features with negative predictive value for pseudoprogression, some findings, such as multifocality, signal abnormality extending across the corpus callosum, and subependymal involvement, are suggestive of progression.^128,129 Higher ADC values have been observed in pseudoprogression, perhaps related to vasogenic edema of the inflammatory treatment effect, compared to lower ADC values in progressive, cellular disease.^130–132 While several studies have shown that decreased Cho and the presence of lipids and lactate are correlated with radiation necrosis, MRS of pseudoprogression remains variable.^133–138 Using DSC, DCE, and ASL techniques, lower perfusion parameters have been shown in pseudoprogression compared to progressive disease with varying sensitivity and specificity.^{135,139–142} Multiparametric models incorporating diffusivity, spectroscopy, and perfusion parameters among other imaging and clinical features have been used to better identify and predict progressive disease.^143–146 Despite these promising findings, prospective data and radiological–pathological correlation are lacking. Other advanced techniques currently being investigated include nuclear imaging agents, although their use in broad clinical practice is limited and beyond the scope of this review.^109,147,148

Pseudoresponse

Pseudoresponse represents a marked decrease in contrast enhancement on MRI related to diminished leakiness of the BBB following treatment with antiangiogenic agents, most commonly bevacizumab, in patients with recurrent glioblastoma. The marked decrease in contrast enhancement, and often in peritumoral edema, can be observed as early as 1 day after initiation of antiangiogenic therapy and does not necessarily reflect biological antitumor effect of therapy (Figure 11).^119,149,150 Antiangiogenic agents may select for a hypoxic and invasive tumor phenotype that is capable of co-opting existing vasculature and therefore growing as nonenhancing infiltrative tumor before manifesting as progressive enhancing disease.^151,152 Progressive enhancement following antiangiogenic therapy has been shown to be a poor prognostic marker.¹⁵³ Low ADC values, representing viable cellular tumor, are seen in persistent or progressive nonenhancing tumor.^154,155 However, reduced diffusion may represent cytotoxic treatment effect and accurate interpretation requires serial imaging to assess temporal changes.^119,150,156 MRS has been investigated in the setting of antiangiogenic therapy; however, experience remains limited.^157–159 PWI has been used to characterize changes in tumor vasculature in response to therapy with decreased perfusion seen both in tumor and normal appearing brain.¹⁵⁹ Similar to changes in contrast enhancement, PWI can change rapidly and studies have shown that patients whose tumors showed normalized perfusion parameters after therapy had improved outcomes.¹¹⁹ As with pseudoresponse, the use of multiparametic MRI shows promise in better characterizing regions of signal abnormality.^146,160

Considering the importance of the nonenhancing tumor, the updated RANO criteria incorporate T2/FLAIR imaging characteristics as measures of response and define pseudoresponse as a greater than 50% reduction in contrast enhancement without a significant decrease in nonenhancing tumor. Decreased enhancement should persist for greater than 4 wk to be considered a true response.¹⁴ While the RANO criteria do not account for all of the subtleties and nuances of evaluation of the post-treatment brain, they provide our current best framework to standardize response to treatment in gliomas.^161,162

Long-Term Complications of Therapy

In addition to mimics of tumor progression, several other chronic changes attributable to brain tumor therapy are well cataloged. Symmetric white matter signal abnormality representing gradual demyelination, gliosis, and vascular injury following chemotherapy, radiotherapy, or both is associated with progressive neurocognitive decline and disordered white matter diffusion.^163–166 In extreme cases, a diffuse necrotizing leukoencephalopathy can develop following intrathecal chemotherapy without or with radiotherapy.¹⁶⁷ Rarely, patients with a remote history of intracranial irradiation present with headaches and neurological deficits and are found to have abnormal cortical enhancement. This entity has been termed stroke-like migraine attacks after radiation therapy (SMART) syndrome and is self-limited with resolution of imaging findings and symptoms the course of weeks (Figure 12).¹⁶⁸ Additionally, with increased adoption of SWI in routine neuroimaging scattered foci of susceptibility are readily identified in the years following irradiation. While their pathogenesis is not fully understood, these small microhemorrhages or vascular malformations are thought to represent delayed radiation toxicity on cerebral microvasculature (Figure 13).^127,169,170

Another late adverse effect of radiation therapy is the development of a second neoplasm. Radiation-associated tumors, in decreasing order of frequency, include meningioma, gliomas, and sarcomas.^171,172 The risk of developing a radiation-associated meningioma increases and latency to manifestation decreases with increasing radiation dose. Radiation-associated meningiomas tend to occur at a younger age and are more often multiple when contrasted with sporadic meningiomas (Figure 13).^173,174 Radiation-associated gliomas are usually a high-grade astrocytoma occurring in the radiation field with imaging features indistinguishable from a primary glioma.¹⁷¹

CONCLUSION

The past several decades have seen the widespread adoption of advanced MRI techniques in addition to conventional structural MRI for the routine clinical assessment of brain tumors. The incorporation of these biology-driven MRI methods has been indispensable to the neurosurgeon and contributed to improved diagnosis, surgical and radiosurgical planning, and assessment of treatment response. Neuroimaging will continue to evolve to reflect our growing understanding of brain tumor molecular genetics and targeted therapy with the overarching goal of being the objective and quantitative arbiter of therapy response for patients with brain tumor.