HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2231010594v1 223/1/11    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Cha, S. Articles by Zagzag, D. Search for Related Content PubMed PubMed Citation Articles by Cha, S. Articles by Zagzag, D.

Intracranial Mass Lesions: Dynamic Contrast-enhanced Susceptibility-weighted Echo-planar Perfusion MR Imaging¹

¹ From the Departments of Radiology (S.C., E.A.K., G.J., S.G.W., A.W.L.), Neurosurgery (E.A.K., D.Z.), and Pathology (D.Z.), New York University Medical Center, 530 First Ave, HCC-Basement, MRI Center, New York, NY 10016. Received March 12, 2001; revision requested April 27; revision received June 25; accepted July 16. S.C. supported in part by an RSNA Seed Grant 3; S.G.W. supported in part by Swiss National Science Foundation/Karger Stiftung and by Novartis Stiftung. Address correspondence to S.C. (e-mail: cha@mri.med.nyu.edu).

	ABSTRACT

TOP ABSTRACT INTRODUCTION TECHNICAL CONSIDERATIONS CLINICAL APPLICATIONS CONCLUSION REFERENCES

 
Dynamic contrast agent–enhanced perfusion magnetic resonance (MR) imaging provides physiologic information that complements the anatomic information available with conventional MR imaging. Analysis of dynamic data from perfusion MR imaging, based on tracer kinetic theory, yields quantitative estimates of cerebral blood volume that reflect the underlying microvasculature and angiogenesis. Perfusion MR imaging is a fast and robust imaging technique that is increasingly used as a research tool to help evaluate and understand intracranial disease processes and as a clinical tool to help diagnose, manage, and understand intracranial mass lesions. With the increasing number of applications of perfusion MR imaging, it is important to understand the principles underlying the technique. In this review, the essential underlying physics and methods of dynamic contrast-enhanced susceptibility-weighted echo-planar perfusion MR imaging are described. The clinical applications of cerebral blood volume maps obtained with perfusion MR imaging in the differential diagnosis of intracranial mass lesions, as well as the pitfalls and limitations of the technique, are discussed. Emphasis is on the clinical role of perfusion MR imaging in providing insight into the underlying pathophysiology of cerebral microcirculation.

© RSNA, 2002

Index terms: Brain, MR, 13.121411, 13.121412, 13.121416, 13.12143, 13.12144 • Brain neoplasms, 13.343, 13.363, 13.38 • Lymphoma, 13.343 • Magnetic resonance (MR), perfusion study, 13.121416, 13.12143, 13.12144 • State of the Art

	INTRODUCTION

TOP ABSTRACT INTRODUCTION TECHNICAL CONSIDERATIONS CLINICAL APPLICATIONS CONCLUSION REFERENCES

 
Recent advances in neuroimaging allow the acquisition of physiologic maps that complement the anatomic information provided by traditional techniques. In particular, the development of fast magnetic resonance (MR) imaging methods provides a means of obtaining high-spatial-resolution images of water self-diffusion, axonal organization, and blood oxygenation. Of particular clinical use are the maps of cerebral microcirculation that can be constructed from dynamic contrast agent–enhanced perfusion MR imaging. Several parameters (blood volume, transit time, clearance, extraction fraction, blood flow, and permeability–surface area product) can, in principle, be derived by following the passage of a tracer or an indicator such as an MR contrast agent through the cerebrovascular system (1).

Strictly, cerebral perfusion is defined as the steady-state delivery of nutrients and oxygen via blood to brain tissue parenchyma per unit volume or mass and is typically measured in milliliters per 100 g of tissue per minute (2). In perfusion MR imaging, however, the term "perfusion" is broadly applied to include several tissue microcirculatory hemodynamic parameters (cerebral blood volume [CBV], cerebral blood flow, and mean transit time) that can be estimated from the dynamic data. In the evaluation of intracranial mass lesions, however, CBV appears to be the most useful parameter.

Although a detailed description of the fundamentals of tracer kinetics used in dynamic imaging is beyond the scope of this review (readers can refer to several excellent reviews on this topic [1,3,4]), the essential principles will be discussed. Also, we will focus on clinical applications of perfusion MR imaging and how the technique can be helpful in the diagnosis, management, and understanding of various intracranial mass lesions. Limitations and potential pitfalls will be discussed along with appropriate strategies to overcome some of the weaknesses of the technique. When available, imaging findings will be correlated with histopathologic findings, focusing on the vascular morphology and status of angiogenesis of various intracranial mass lesions.

	TECHNICAL CONSIDERATIONS

TOP ABSTRACT INTRODUCTION TECHNICAL CONSIDERATIONS CLINICAL APPLICATIONS CONCLUSION REFERENCES

 
Contrast Agents for Perfusion MR Imaging
Perfusion MR imaging methods exploit signal changes that accompany the passage of tracer through the cerebrovascular system. The tracer can be either endogenous (arterial water) or exogenous (deuterium oxide, gadopentetate dimeglumine) and either freely diffusible (arterial water, deuterium oxide) or nondiffusible (gadopentetate dimeglumine). In each case, the analysis is based on the indicator-dilution methods originally developed for the radioisotope measurements of blood flow and volume (5).

Endogenous or arterial spin-labeling techniques such as echo-planar MR imaging and signal targeting with alternating radio frequency, or EPISTAR, and flow-sensitive alternating inversion-recovery, or FAIR, magnetically label blood spins upstream from the imaging section with either inverting or saturating radio-frequency pulses. In this approach, the regional changes in signal intensity are determined by means of an interaction between blood flow and T1 relaxation. Comparison between images acquired with and those acquired without labeling allow calculation of tissue perfusion (6–10). Arterial spin labeling does not require injection of an exogenous contrast agent, can be performed without additional hardware, and, in principle, yields fully quantitative measurements of cerebral blood flow. However, the necessarily long transit delay between the application of the labeling pulse and the arrival of labeled blood in the imaging section allows T1 recovery, which decreases perfusion contrast. Even with signal averaging (which increases imaging time), arterial spin-labeling cerebral blood flow maps suffer from a poor signal-to-noise ratio and have not been widely used clinically. Techniques in which freely diffusible exogenous tracers such as deuterium oxide (11), ¹⁷H₂O (12) or trifluoromethane (13,14) are used have also been proposed. However, these are either expensive or suffer from poor signal-to-noise ratio and, again, have not been used widely in clinical practice.

One of the oldest methods of contrast-enhanced perfusion MR imaging, originally developed by Villringer, Rosen, and colleagues (15,16), uses conventional MR imaging contrast agents such as gadopentetate dimeglumine. The change in relaxation times during passage of a contrast agent bolus can be converted to tissue contrast agent concentration. The theory of tracer kinetics is then used to derive CBV values from the concentration-time curve. This method has the advantage that CBV can be measured without knowledge of the arterial input function and provides a better signal-to-noise ratio than do spin-labeling techniques. Although quantitative estimates of CBV are relative rather than absolute, CBV measurements have shown to correlate with the vascularity of various intracranial mass lesions. Contrast-enhanced perfusion MR imaging will be further discussed later in this section.

An entirely different approach based on pharmacokinetic modeling was developed by Tofts and Kermode (17) and others (18,19). These techniques were developed primarily to measure vascular permeability and may in the future provide valuable information on brain tumors in which angiogenesis leads to the formation of immature and permeable blood vessels. To date, however, the methods have not been widely used.

Sequence Considerations
For CBV measurements, a series of images are acquired at approximately 1-second intervals before, during, and after the bolus injection of contrast agent. Longer intervals may be feasible, but the signal intensity–time curve would then be measured less accurately. Rapid gradient-echo imaging is capable of generating approximately two T1-weighted sections per second, which is not generally sufficient to cover a large heterogeneous tumor. Echo-planar imaging, however, is capable of generating approximately 10 MR sections every second and is ideal for rapid dynamic imaging.

The passage of gadopentetate dimeglumine causes changes in both T2 and T2* so that both spin-echo and gradient-echo echo-planar sequences provide robust measurements of CBV. Gradient-echo sequences are, however, much more sensitive. When a paramagnetic contrast agent such as gadopentetate dimeglumine passes through the cerebral vascular system, it induces differences in local magnetic susceptibility between vessels and the surrounding tissue. Although the vascular space is a small fraction (4%–5%) of the total tissue blood volume, this compartmentalization of the contrast agent causes targeted paramagnetism within the intravascular spins, as well as in the surrounding spins within a given voxel. Thus, both intra- and extravascular spins undergo a reduction of T2* that leads to a large transient signal loss of approximately 25% in normal white matter with a standard dose of contrast agent (0.1 mmol per kilogram of body weight). T2-weighted spin-echo images are less sensitive and require a double or even quadruple dose of contrast agent in order to yield substantial signal changes during the bolus passage. On the other hand, gradient-echo sequences are more prone to magnetic susceptibility artifacts.

Asymmetric spin-echo echo-planar sequences provide a potentially useful compromise between gradient-echo and spin-echo echo-planar sequences. In asymmetric spin-echo echo-planar sequences, the echo center is displaced from the Hahn echo time, resulting in a mixture of T2 and T2* weighting. The degree of asymmetry can be adjusted to trade off sensitivity against susceptibility to artifacts (20,21). Thus, when imaging lesions near brain-bone-air interfaces, such as the temporal or inferior frontal lobes where these artifacts are more pronounced, spin-echo sequences may be preferable. However, artifacts on gradient-echo images can be overcome, to a large extent, by reducing section thickness (22). Although this reduces the signal-to-noise ratio, we have found that this technique still provides diagnostic images. A second advantage of spin-echo sequences is that results with simulations and phantoms suggest that spin-echo images will be sensitive only to contrast agent within the capillaries, whereas gradient-echo sequences will be sensitive to contrast agent in both capillaries and larger vessels (23). Although contamination by venous signals with gradient-echo sequences will potentially cause overestimates of CBV, it is relatively easy to identify the location of veins and perform measurements of CBV in regions of interest (ROIs) chosen to avoid them.

Data Analysis: Calculating CBV from Image Intensities
It has been shown by Zierler (5) that in the absence of recirculation and contrast agent leakage, CBV is proportional to the area under the contrast agent concentration–time curve. The gadopentetate dimeglumine concentration is proportional to the change in relaxation rate (ie, change in the reciprocal of T2* [R2*]), which can be calculated from the signal by using the following equation (16): R2* = [-ln(SI_t/SI₀)/TE], where SI_t is the pixel signal intensity at time t, SI₀ is the precontrast signal intensity, and TE is the echo time. This equation is valid only if the T1 enhancement associated with blood-brain barrier disruption has a negligible effect on signal intensity, which is ensured by using either a long repetition time, a small flip angle, or both to reduce saturation.

In general, the assumptions of negligible recirculation and contrast agent leakage are violated. The effects of this can be reduced by fitting a gamma-variate function to the measured R2* curve (3). The gamma-variate function approximates the curve that would have been obtained without recirculation or leakage. CBV can then be estimated from the area under this fitted curve rather than from the original data. In our experience, however, the gamma variate fit is unstable: Small variations in the initial parameter estimates yield wide variations in the results. This instability occurs even with data averaged over ROIs in areas of high perfusion and appears to be inherent to the procedure. In practice, satisfactory fits can often be found only by repeating the procedure with multiple different initial estimates until a set is found that causes the fitting algorithm to converge.

This approach can be applied to fit concentration-time curves from multiple ROIs, since each fit takes very little time, but it is not suited for pixel-by-pixel calculations of CBV maps. To obtain CBV maps, alternative corrections for leakage are preferable. The simplest method is to estimate the end of the bolus and calculate the area within the bolus alone. This will, however, result in a systematic overestimate of CBV in areas where the blood-brain barrier is damaged. Alternatively, having estimated the beginning and end of the bolus, a baseline can be subtracted from under the curve. The area under the corrected contrast agent concentration–time curve is proportional to the CBV and does not yield an absolute measurement. It is necessary, therefore, to express the measurement relative to a standard reference, usually contralateral white matter. We refer to this as relative CBV (rCBV).

The steps in the data analysis are, therefore, as follows (Fig 1): (a) Obtain curves of signal intensity changes over time (Fig 1, A). (b) Estimate the mean precontrast signal intensity from 10 data points acquired before arrival of the bolus; it is important to exclude the first one to two images, which are acquired during the time the steady-state MR signal is established. (c) Calculate R2* and apply baseline subtraction to the R2* curve (Fig 1, B). (d) Calculate the area under the fitted curve (Fig 1, C). (e) Calculate rCBV relative to normal white matter (Fig 1, D).

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Diagrams illustrate calculation of rCBV. A, Signal intensity decrease during passage of contrast agent bolus is measured from a series of gradient-echo echo-planar MR images. B, Change in the relaxation rate (R2*) is calculated from signal intensity, and a baseline subtraction method is applied to measured data. C, Corrected R2* curve. D, rCBV is proportional to the area under curve (shaded area).

View larger version (137K): [in this window] [in a new window] [Download PPT slide]   Figure 2a. (a-d) Transverse gradient-echo echo-planar (repetition time msec/echo time msec, 1,000/54) MR images in a 33-year-old woman with a choroid plexus carcinoma of the left lateral ventricle. (a) Image obtained before intravenous injection of gadopentetate dimeglumine. (b) Image obtained at peak arrival time (in this case, 10 seconds after injection of contrast agent) demonstrates marked signal intensity decrease in the tumor (arrow). (c) Image obtained 50 seconds after injection shows return of signal intensity to baseline except in areas of disruption or absence of the blood-brain barrier (arrows), where leakage has occurred. (d) Color overlay of rCBV map shows the intense hypervascularity (red) of the tumor. (e) Transverse contrast-enhanced T1-weighted (600/14) MR image demonstrates marked enhancement of the tumor (arrow) due to lack of a blood-brain barrier.

View larger version (142K): [in this window] [in a new window] [Download PPT slide]   Figure 2b. (a-d) Transverse gradient-echo echo-planar (repetition time msec/echo time msec, 1,000/54) MR images in a 33-year-old woman with a choroid plexus carcinoma of the left lateral ventricle. (a) Image obtained before intravenous injection of gadopentetate dimeglumine. (b) Image obtained at peak arrival time (in this case, 10 seconds after injection of contrast agent) demonstrates marked signal intensity decrease in the tumor (arrow). (c) Image obtained 50 seconds after injection shows return of signal intensity to baseline except in areas of disruption or absence of the blood-brain barrier (arrows), where leakage has occurred. (d) Color overlay of rCBV map shows the intense hypervascularity (red) of the tumor. (e) Transverse contrast-enhanced T1-weighted (600/14) MR image demonstrates marked enhancement of the tumor (arrow) due to lack of a blood-brain barrier.

View larger version (140K): [in this window] [in a new window] [Download PPT slide]   Figure 2c. (a-d) Transverse gradient-echo echo-planar (repetition time msec/echo time msec, 1,000/54) MR images in a 33-year-old woman with a choroid plexus carcinoma of the left lateral ventricle. (a) Image obtained before intravenous injection of gadopentetate dimeglumine. (b) Image obtained at peak arrival time (in this case, 10 seconds after injection of contrast agent) demonstrates marked signal intensity decrease in the tumor (arrow). (c) Image obtained 50 seconds after injection shows return of signal intensity to baseline except in areas of disruption or absence of the blood-brain barrier (arrows), where leakage has occurred. (d) Color overlay of rCBV map shows the intense hypervascularity (red) of the tumor. (e) Transverse contrast-enhanced T1-weighted (600/14) MR image demonstrates marked enhancement of the tumor (arrow) due to lack of a blood-brain barrier.

View larger version (140K): [in this window] [in a new window] [Download PPT slide]   Figure 2d. (a-d) Transverse gradient-echo echo-planar (repetition time msec/echo time msec, 1,000/54) MR images in a 33-year-old woman with a choroid plexus carcinoma of the left lateral ventricle. (a) Image obtained before intravenous injection of gadopentetate dimeglumine. (b) Image obtained at peak arrival time (in this case, 10 seconds after injection of contrast agent) demonstrates marked signal intensity decrease in the tumor (arrow). (c) Image obtained 50 seconds after injection shows return of signal intensity to baseline except in areas of disruption or absence of the blood-brain barrier (arrows), where leakage has occurred. (d) Color overlay of rCBV map shows the intense hypervascularity (red) of the tumor. (e) Transverse contrast-enhanced T1-weighted (600/14) MR image demonstrates marked enhancement of the tumor (arrow) due to lack of a blood-brain barrier.

View larger version (148K): [in this window] [in a new window] [Download PPT slide]   Figure 2e. (a-d) Transverse gradient-echo echo-planar (repetition time msec/echo time msec, 1,000/54) MR images in a 33-year-old woman with a choroid plexus carcinoma of the left lateral ventricle. (a) Image obtained before intravenous injection of gadopentetate dimeglumine. (b) Image obtained at peak arrival time (in this case, 10 seconds after injection of contrast agent) demonstrates marked signal intensity decrease in the tumor (arrow). (c) Image obtained 50 seconds after injection shows return of signal intensity to baseline except in areas of disruption or absence of the blood-brain barrier (arrows), where leakage has occurred. (d) Color overlay of rCBV map shows the intense hypervascularity (red) of the tumor. (e) Transverse contrast-enhanced T1-weighted (600/14) MR image demonstrates marked enhancement of the tumor (arrow) due to lack of a blood-brain barrier.

Although rCBV maps are good indicators of hypervascular regions within intracranial mass lesions, values calculated on a pixel-by-pixel basis suffer from a poor signal-to-noise ratio. It is therefore preferable to calculate rCBV values from ROIs within the region. In our practice, the rCBV map is primarily used as a road map and a qualitative assessment of lesion vascularity. To obtain ROI rCBV values, multiple ROIs are drawn to obtain a qualitative impression of blood volume. The sizes of ROIs are varied, depending on the size of lesion, but usually 1–5 mm-diameter ROIs are used. The degree of signal intensity change correlates well with rCBV (24), so the region with the highest rCBV is easily identified. Furthermore, in lesions that are located close to major vascular structures, careful inspection of the signal intensity–time curve clearly indicates vessels that produce very large signal intensity changes.

CBV measurements of lesions are generally calculated relative to ROI values in contralateral normal white matter. If the lesion is predominantly cortical, the reference ROI is placed over the contralateral cortex rather than over the white matter. Similarly, with deep gray matter lesions in the basal ganglia or thalamus, a contralateral, uninvolved, similar location should preferably be selected as a reference point.

Imaging Protocol
Three critical components are necessary for dynamic contrast-enhanced perfusion MR imaging. First, an imaging technique, such as echo-planar imaging, capable of acquiring images of multiple sections with a temporal resolution of 1 or 2 seconds is needed. Second, a power injector to ensure rapid uniform administration of contrast agent and, therefore, a well-defined and narrow bolus should be used. Third, a software program that will allow data analysis and rCBV measurements is required.

At our institution, all patients undergo placement of an 18- or 20-gauge intravenous catheter prior to imaging, usually in the antecubital area, for the purpose of contrast agent administration. Since perfusion MR imaging complements rather than replaces conventional MR imaging, our imaging protocol includes nonenhanced transverse T2-weighted (3,400/119) or fluid-attenuated inversion recovery (9,000/110/2,500 [repetition time msec/echo time msec/inversion time msec]) and contrast-enhanced T1-weighted (600/14) sequences. The perfusion imaging protocol uses a gradient-echo susceptibility-weighted blipped echo-planar imaging sequence. Unlike the original echo-planar imaging sequence (25), where the phase-encoding gradient is applied continuously, blipped echo-planar imaging involves a brief phase-encoding gradient "blip" between readout intervals (26). With blipped echo-planar imaging, the k-space trajectory is much more amenable to Fourier transformation. A complete discussion of echo-planar imaging and the required hardware modifications is beyond the scope of this report. Interested readers are referred to the literature (27,28). The specific imaging parameters are 1,000/54; field of view, 230 x 230 mm; section thickness, 3–7 mm; matrix, 128 x 128; in-plane voxel size, 1.8 x 1.8 mm; intersection gap, 0%–30%; flip angle, 30°; signal bandwidth, 1,470 Hz/pixel. Five to seven sections are obtained to cover the entire lesion volume as identified on T2-weighted images. A series of 60 multisection acquisitions are acquired at 1-second intervals. The combination of a 1,000-msec repetition time and a 30° flip angle ensures that T1 effects are minimized. The first 10 acquisitions are performed prior to the contrast agent injection to establish a precontrast baseline. At the 10th acquisition, gadopentetate dimeglumine (0.1 mmol/kg) is injected with the power injector at a flow rate of 3–5 mL/sec through the intravenous catheter, the contrast agent injection is immediately followed by a bolus injection of saline (total of 20 mL at the same rate).

Technical Pitfalls and Limitations
There are several important limitations associated with contrast-enhanced perfusion MR imaging. First, because the technique is susceptibility weighted, it is exquisitely sensitive to structures or lesions that induce strong magnetic field inhomogeneity, such as blood products, calcium, melanin, metals, or lesions near a brain-bone-air interface (eg, middle and anterior cranial fossa). One simple way to reduce inhomogeneity and susceptibility artifacts is to decrease section thickness, as shown in Figure 3, although this will also reduce the signal-to-noise ratio and section coverage. If section coverage is insufficient to cover the tumor, the intersection gap can be increased. This does carry the risk of missing small vascular regions, but even with thicker sections such small regions may be missed due to volume averaging.

View larger version (66K): [in this window] [in a new window] [Download PPT slide]   Figure 3a. Transverse gradient-echo echo-planar (1,000/54) MR images show reduction of susceptibility artifacts when a smaller section thickness is used in a 34-year-old man with left temporal lobe radiation necrosis. (a) Images obtained with 6-mm section thickness and 20% intersection gap show extensive susceptibility artifacts (arrows) in the inferior frontal and temporal lobes. (b) Images obtained at same levels as in a but with 3-mm section thickness and 20% gap show the presence of a cavity (arrowheads) in the left middle cranial fossa. Susceptibility artifacts are markedly reduced.

View larger version (73K): [in this window] [in a new window] [Download PPT slide]   Figure 3b. Transverse gradient-echo echo-planar (1,000/54) MR images show reduction of susceptibility artifacts when a smaller section thickness is used in a 34-year-old man with left temporal lobe radiation necrosis. (a) Images obtained with 6-mm section thickness and 20% intersection gap show extensive susceptibility artifacts (arrows) in the inferior frontal and temporal lobes. (b) Images obtained at same levels as in a but with 3-mm section thickness and 20% gap show the presence of a cavity (arrowheads) in the left middle cranial fossa. Susceptibility artifacts are markedly reduced.

Third, the calculation of rCBV can be grossly inaccurate in lesions such as glioblastoma multiforme (GBM) or meningioma, where there is a severe breakdown or absence of the blood-brain barrier (29). It should also be emphasized that rCBV measurements are a relative rather than an absolute quantification of blood volume.

	CLINICAL APPLICATIONS

TOP ABSTRACT INTRODUCTION TECHNICAL CONSIDERATIONS CLINICAL APPLICATIONS CONCLUSION REFERENCES

 
Perfusion MR imaging is increasingly being used as a diagnostic and research tool that provides maps of the regional variations in cerebral microvasculature of normal and diseased brains (30–34). With relatively short imaging and data processing times and the use of a standard dose of contrast agent, perfusion MR imaging is a promising tool that can easily be incorporated as part of the routine clinical evaluation of intracranial mass lesions. Although still investigational, MR imaging CBV measurements can be used as an adjunct to conventional imaging to help assess the degree of neovascularization in brain tumors, evaluate tumor grading and malignancy, identify tumor-mimicking lesions (such as radiation necrosis, cerebral abscess, and tumefactive demyelinating lesion [TDL]) by demonstrating their lack of angiogenesis, and assess the status of viable tissue surrounding an acute infarct. It must be emphasized, however, that perfusion MR imaging is a relatively new and promising imaging tool rather than a standard proven technique for tumor grading and staging. In the future, perfusion MR imaging may become useful in the monitoring of treatment, and its results may also potentially serve as an arbiter when determining the efficacy of novel therapeutic agents, especially antiangiogenic therapy.

Intracranial Neoplasms
Vascular morphology and the degree of angiogenesis are important elements in the evaluation of different tumor types and the determination of the biologic aggressiveness of intracranial neoplasms (35). Perfusion MR imaging provides in vivo maps of CBV that depict the overall tumor vascularity, which allows indirect assessment of tumor angiogenesis (30). MR measurements of rCBV have been shown to correlate with both conventional angiographic assessments of tumor vascular density and histologic measurements of tumor neovascularization (30,31,36). Increased tumor vascularity, however, is not synonymous with malignancy. There are several intracranial neoplasms, especially those that are extraaxial in location (eg, meningioma, choroid plexus papilloma), that can be highly vascular but rather benign in terms of biologic behavior.

With the recent interest in and development of antiangiogenic cancer therapies that directly attack tumor vessels (37–41), perfusion MR imaging can be used for a noninvasive assessment of changes in tumor rCBV during treatment and thus for monitoring the effectiveness of therapy. Conventional MR imaging is limited by its nonspecificity and inability to differentiate between tumor recurrence and therapy-related necrosis. Perfusion MR imaging, on the other hand, has been shown to correlate better with clinical response in patients undergoing antiangiogenic therapy (32).

Proton (hydrogen 1 [¹H]) MR spectroscopy may allow further characterization of intracranial mass lesions by providing information on metabolic changes such as those associated with cellular turnover or proliferative activity (choline), decrease in neuronal density (N-acetylaspartate), and presence of necrosis (lipids, lactate) (42). Although ¹H MR spectroscopy also remains investigational in its role for accurate preoperative grading of gliomas, the presence of lipids and lactate strongly suggests high-grade tumors (43–46). In a recent ¹H MR spectroscopy study of gliomas (47), choline metabolites were positively correlated with histopathologic findings of proliferative activity as determined with the Ki-67 labeling index. ¹H MR spectroscopy is a useful and promising diagnostic modality for the metabolic characterization of brain tumors, as well as of other mass lesions of the brain. A detailed description of the clinical application of ¹H MR spectroscopy is beyond the scope of this article, and interested readers are referred to the literature (48–52).

Primary Gliomas
Malignant gliomas, the most common primary brain tumors, are highly invasive and extremely vascular tumors. Gliomas are histologically heterogeneous, with components that include varying degrees of cellular and nuclear pleomorphism, mitotic activity, vascular proliferation, and necrosis (53,54). GBM, the most malignant glioma, is the most common primary brain tumor, accounting for 12%–15% of all intracranial neoplasms (35). Despite decades of research, GBM remains resistant to therapies that have proved successful in treating other solid tumors. GBM remains one of the most malignant and fatal neoplasms known to humanity.

An important factor in the malignancy of GBM is the ability of the tumor to recruit and synthesize vascular networks for further growth and proliferation. The degree of vascular proliferation is one of the most critical elements in the determination of tumor grade and prognosis for several reasons. First, the degree of vascular proliferation, or angiogenesis, is one of the most important histologic criteria (along with cellularity, mitosis, pleomorphism, necrosis) for determination of the degree of malignancy and grade of the glioma. Second, vascular networks are not only the principal route for delivery of oxygen and nutrients to the neoplastic cells and removal of metabolic waste products, such networks also serve as a path for tumor infiltration along the perivascular spaces. Third, the cerebral capillary endothelium (site of the blood-brain barrier)—composed of a continuous homogeneous basement membrane, numerous astrocytic processes, and tight junctions and an important host defense mechanism responsible for the regulation of movement of molecules—is frequently destroyed by malignant tumor cells. Fourth, a destroyed or altered blood-brain barrier serves as a diagnostic tool at both computed tomography (CT) and MR imaging by allowing for contrast agent extravasation and, therefore, identification of the tumor (55,56). Preoperative assessment of glioma vascularity can be helpful in determining malignant potential and, therefore, potentially in alteration of therapeutic management.

Contrast-enhanced T1-weighted MR images of the brain depict areas of a disrupted or absent blood-brain barrier (57,58). Perfusion MR imaging, on the other hand, demonstrates variations in regional CBV that reflect alterations in tumor vascularity. Figure 4 demonstrates an example of an rCBV map of a highly vascular GBM with corresponding T2-weighted and contrast-enhanced T1-weighted images. Figure 5 displays a signal intensity–time curve of the GBM in Figure 4. Figure 6 illustrates histopathologic findings of the same GBM, demonstrating striking angiogenesis and glomeruloid capillaries.

View larger version (150K): [in this window] [in a new window] [Download PPT slide]   Figure 4a. Left parietal lobe GBM in a 46-year-old man. (a) Transverse T2-weighted (3,400/119) MR image shows left parietal lobe mass with extensive surrounding edema (arrow). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows enhancement of the tumor (arrow). (c) rCBV map demonstrates marked hypervascularity (red) of the tumor.

View larger version (148K): [in this window] [in a new window] [Download PPT slide]   Figure 4b. Left parietal lobe GBM in a 46-year-old man. (a) Transverse T2-weighted (3,400/119) MR image shows left parietal lobe mass with extensive surrounding edema (arrow). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows enhancement of the tumor (arrow). (c) rCBV map demonstrates marked hypervascularity (red) of the tumor.

View larger version (133K): [in this window] [in a new window] [Download PPT slide]   Figure 4c. Left parietal lobe GBM in a 46-year-old man. (a) Transverse T2-weighted (3,400/119) MR image shows left parietal lobe mass with extensive surrounding edema (arrow). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows enhancement of the tumor (arrow). (c) rCBV map demonstrates marked hypervascularity (red) of the tumor.

View larger version (24K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Signal intensity-time curve shows marked signal intensity decrease during peak arrival of the contrast agent, followed by partial recovery of the signal intensity loss (arrowheads). A second decrease in signal intensity (arrow) is due to recirculation. Recovery of signal intensity is incomplete due to leakage of contrast agent through a disrupted blood-brain barrier.

View larger version (129K): [in this window] [in a new window] [Download PPT slide]   Figure 6. Photomicrographs show histopathologic findings of the vascular hyperplasia of GBM. Top left: Early vascular hyperplasia remote from the glioblastoma "border." Note endothelial cell plumping (arrows). (Hematoxylin-eosin stain; original magnification, x200.) Bottom left: Angiogenesis at the edge of an infiltrating glioblastoma. Invading glioma cells (short arrows) can be seen between the hyperplastic vascular complexes (long arrows). (Hematoxylin-eosin stain; original magnification, x100.) Top right: Intratumoral neovascularization showing glomeruloid vessels (long arrows). Mitotic figures (short arrows) in vascular cells are evident. (Hematoxylin-eosin stain; original magnification, x100.) Bottom right: Azocarmine staining underscores the exuberant vascular proliferation (arrow) that is characteristic of GBM. (Original magnification, x100.)

Glioma grading.—As stated previously, in primary high-grade glioma vascular morphology is a critical parameter in helping determine malignant potential and survival. Glioma grading is important for determination of both prognosis and therapy. Authors of several recent studies (30,31,36) have found statistically significant correlations between tumor rCBV and glioma grade, as well as with conventional catheter-based angiographic tumor vascularities. In our experience with 51 preoperative cases of pathologically proved GBM, the maximum rCBV of the tumor varied from 1.95 to 28.59 (mean ± SD, 5.5 ± 4.5) (unpublished data, 2001). This wide variation is expected, since GBM is one of the most heterogeneous tumors in terms of histologic composition, with frequent areas of necrosis. Anaplastic astrocytoma (19 cases in our experience), which is differentiated from GBM by the lack of necrosis, demonstrated less variation in maximum rCBV measurements, with a range of 1.29–10.26 (mean, 4.03 ± 2.43). Low-grade glioma (13 cases) has a maximum rCBV range of 0.89–3.73 (mean, 1.86 ± 0.77). Figures 7 and 8 demonstrate two cases with similar-appearing nonenhancing gliomas and very different CBV maps and histopathologic tumor grades. The tumor with high rCBV (Fig 7) is an anaplastic astrocytoma, whereas the tumor with low rCBV (Fig 8) is a low-grade fibrillary astrocytoma.

View larger version (148K): [in this window] [in a new window] [Download PPT slide]   Figure 7a. Nonenhancing right frontal lobe tumor in a 32-year-old woman. (a) Transverse T2-weighted (3,400/119) MR image shows a hyperintense mass (arrow) with minimal surrounding edema (arrowheads). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows a homogeneously hypointense right frontal lobe tumor (arrow). (c) rCBV map demonstrates an area of increased vascularity (arrow), and results of a biopsy directed to that area showed histologic vascular proliferation. The patient subsequently underwent surgical resection, and the diagnosis after pathologic examination was anaplastic astrocytoma.

View larger version (134K): [in this window] [in a new window] [Download PPT slide]   Figure 7b. Nonenhancing right frontal lobe tumor in a 32-year-old woman. (a) Transverse T2-weighted (3,400/119) MR image shows a hyperintense mass (arrow) with minimal surrounding edema (arrowheads). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows a homogeneously hypointense right frontal lobe tumor (arrow). (c) rCBV map demonstrates an area of increased vascularity (arrow), and results of a biopsy directed to that area showed histologic vascular proliferation. The patient subsequently underwent surgical resection, and the diagnosis after pathologic examination was anaplastic astrocytoma.

View larger version (142K): [in this window] [in a new window] [Download PPT slide]   Figure 7c. Nonenhancing right frontal lobe tumor in a 32-year-old woman. (a) Transverse T2-weighted (3,400/119) MR image shows a hyperintense mass (arrow) with minimal surrounding edema (arrowheads). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows a homogeneously hypointense right frontal lobe tumor (arrow). (c) rCBV map demonstrates an area of increased vascularity (arrow), and results of a biopsy directed to that area showed histologic vascular proliferation. The patient subsequently underwent surgical resection, and the diagnosis after pathologic examination was anaplastic astrocytoma.

View larger version (131K): [in this window] [in a new window] [Download PPT slide]   Figure 8a. Nonenhancing right frontal lobe tumor in a 29-year-old man. (a) Transverse T2-weighted (3,400/119) MR image shows a mild degree of surrounding edema (arrows). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows a nonenhancing right frontal lobe tumor (arrow). (c) rCBV map demonstrates no evidence of increased tumor vascularity. The patient underwent surgical resection, and the final pathologic results revealed low-grade fibrillary astrocytoma.

View larger version (138K): [in this window] [in a new window] [Download PPT slide]   Figure 8b. Nonenhancing right frontal lobe tumor in a 29-year-old man. (a) Transverse T2-weighted (3,400/119) MR image shows a mild degree of surrounding edema (arrows). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows a nonenhancing right frontal lobe tumor (arrow). (c) rCBV map demonstrates no evidence of increased tumor vascularity. The patient underwent surgical resection, and the final pathologic results revealed low-grade fibrillary astrocytoma.

View larger version (125K): [in this window] [in a new window] [Download PPT slide]   Figure 8c. Nonenhancing right frontal lobe tumor in a 29-year-old man. (a) Transverse T2-weighted (3,400/119) MR image shows a mild degree of surrounding edema (arrows). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows a nonenhancing right frontal lobe tumor (arrow). (c) rCBV map demonstrates no evidence of increased tumor vascularity. The patient underwent surgical resection, and the final pathologic results revealed low-grade fibrillary astrocytoma.

Stereotactic biopsy guidance.—Biopsy remains the definitive method to help determine tumor type and grade. The sampling error in biopsies of high-grade gliomas is well known and is partly attributable to extreme geographic heterogeneity within a single tumor (59,60). Ideally, the grading of gliomas should be based on histologic evaluation of specimens from the most malignant area of the tumor. Determination of this region can be difficult, however. Most biopsies are guided with contrast-enhanced T1-weighted MR or CT images (61), which depict areas of blood-brain barrier breakdown that may not indicate the most malignant or vascular portion of the tumor. CBV maps can depict regions of increased vascularity that can serve as additional targets for stereotactic biopsy. At our institution, rCBV maps are routinely used to select biopsy sites for both enhancing and nonenhancing tumors and help reduce sampling error and the number of nondiagnostic biopsies. As shown in Figure 7, the rCBV map is particularly useful in nonenhancing tumors because it can be used to direct the biopsy to the "hot" area or presumed site of increased tumor vascularity on the map. On the basis of the contrast-enhanced T1-weighted image alone, it may be challenging to select a biopsy target.

Assessment of response to therapy.—Recent developments in antiangiogenic drugs that target actively dividing tumor vasculature have added to the armamentarium of chemotherapeutic agents for brain tumors. Although at an early stage of development, antiangiogenic therapy shows promise as a means to destroy tumor vessels, the main route of nutrient and oxygen for tumor cells, without altering the preexisting normal vasculature. Imaging of gliomas after surgery, irradiation, and/or chemotherapy is often performed to assess tumor activity. However, conventional contrast-enhanced MR imaging cannot be used to reliably identify tumor progression, nor can it be used to monitor changes in tumor vasculature. In a study with patients receiving thalidomide (32), an antiangiogenic drug recently approved by the U.S. Food and Drug Administration, combined with carboplatin, we found that serial rCBV measurements during therapy predicted clinical response in the patients. Serial rCBV measurements correlated better with changes in patients’ clinical status than did conventional contrast-enhanced T1-weighted MR images.

Differentiation of Therapy-induced Necrosis and Recurrent Tumor
Differentiation between radiation necrosis and recurrent tumor carries obvious therapeutic implications. Recurrent tumors may benefit from repeat surgery with adjuvant chemotherapy or targeted high-dose radiation therapy, whereas radiation necrosis may be treated conservatively with steroids. Currently, biopsy or resection is the only definitive means of differentiating between recurrent tumor and radiation necrosis. However, surgical manipulation of radiation necrosis can cause further damage to adjacent brain parenchyma.

Delayed radiation necrosis is usually indistinguishable from recurrent tumor on imaging or clinical grounds. Clinically, both entities can manifest with progressive focal neurologic deficits and signs of increased intracranial pressure. At imaging, both can manifest as a mass lesion with surrounding edema (62–66). Conventional contrast-enhanced CT or MR imaging cannot be used to reliably distinguish radiation necrosis from recurrent tumor. Both disease processes can involve varying degrees of blood-brain barrier disruption that result in abnormal contrast enhancement. In addition, the diffuse neoplastic infiltration of recurrent tumor and the demyelinating process of radiation necrosis can both cause extensive white matter abnormalities, as well as extensive vasogenic edema.

Pathologically, recurrent tumor and radiation necrosis are markedly dissimilar. Although the exact pathogenesis of delayed radiation necrosis remains obscure, a constant pathologic feature is extensive vascular injury and tissue hypoxia, in contrast to the features of recurrent tumor, which is characterized by neovascularization (67,68).

Among imaging modalities, fluorodeoxyglucose positron emission tomography (PET) can be useful in differentiating the two entities (69,70). However, authors of a few recent studies (71,72) have suggested that PET may have low sensitivity and specificity. Moreover, PET scanners, unlike MR machines, are not yet widely available, thus limiting the use of PET in routine diagnosis and evaluation of brain tumors. Differentiation between recurrent tumor and radiation necrosis by means of SPECT can also be difficult (73–75). In addition, the inherent low spatial resolution of both PET and SPECT is a particular issue when recurrent tumor and radiation necrosis coexist, which unfortunately is a common occurrence.

MR imaging CBV maps can demonstrate the pathologic differences in vascularity between therapy-induced necrosis and recurrent tumor, and our preliminary results have shown them to be capable of facilitating differentiation between the two entities (76). Figures 9 and 10 demonstrate surgically proved recurrent GBM and radiation necrosis, respectively, where the conventional MR imaging features were nonspecific in both instances but perfusion MR imaging showed increased rCBV in recurrent tumor and low rCBV in radiation necrosis, thus allowing differentiation between the two.

View larger version (138K): [in this window] [in a new window] [Download PPT slide]   Figure 9a. Recurrent tumor in a 72-year-old man with previously resected and irradiated right frontal lobe GBM. (a) Transverse T2-weighted (3,400/119) MR image displays a large area of hyperintensity abnormality involving the right frontal lobe. (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows nonenhancing right frontal lobe lesion (arrowheads) that extends into the genu of the corpus callosum (large arrow). Also note thick dural enhancement (small arrow) adjacent to the craniotomy site. (c) rCBV map demonstrates marked increase in vascularity in the right frontal lobe and genu of the corpus callosum, suggesting recurrent tumor. At repeat surgery, recurrence of GBM was confirmed.

View larger version (137K): [in this window] [in a new window] [Download PPT slide]   Figure 9b. Recurrent tumor in a 72-year-old man with previously resected and irradiated right frontal lobe GBM. (a) Transverse T2-weighted (3,400/119) MR image displays a large area of hyperintensity abnormality involving the right frontal lobe. (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows nonenhancing right frontal lobe lesion (arrowheads) that extends into the genu of the corpus callosum (large arrow). Also note thick dural enhancement (small arrow) adjacent to the craniotomy site. (c) rCBV map demonstrates marked increase in vascularity in the right frontal lobe and genu of the corpus callosum, suggesting recurrent tumor. At repeat surgery, recurrence of GBM was confirmed.

View larger version (141K): [in this window] [in a new window] [Download PPT slide]   Figure 9c. Recurrent tumor in a 72-year-old man with previously resected and irradiated right frontal lobe GBM. (a) Transverse T2-weighted (3,400/119) MR image displays a large area of hyperintensity abnormality involving the right frontal lobe. (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows nonenhancing right frontal lobe lesion (arrowheads) that extends into the genu of the corpus callosum (large arrow). Also note thick dural enhancement (small arrow) adjacent to the craniotomy site. (c) rCBV map demonstrates marked increase in vascularity in the right frontal lobe and genu of the corpus callosum, suggesting recurrent tumor. At repeat surgery, recurrence of GBM was confirmed.

View larger version (160K): [in this window] [in a new window] [Download PPT slide]   Figure 10a. Delayed radiation necrosis in an 82-year-old woman. (a) Transverse fluid-attenuated inversion recovery (9,000/110/2,500) MR image shows extensive hyperintensity abnormality that extends into the genu of the corpus callosum (large arrow). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows heterogeneous, irregularly enhancing left frontal lobe mass (arrows). (c) rCBV map demonstrates no evidence of increased vascularity, suggesting radiation necrosis rather than recurrent tumor. The patient underwent surgical resection, and radiation necrosis without evidence of tumor was found.

View larger version (143K): [in this window] [in a new window] [Download PPT slide]   Figure 10b. Delayed radiation necrosis in an 82-year-old woman. (a) Transverse fluid-attenuated inversion recovery (9,000/110/2,500) MR image shows extensive hyperintensity abnormality that extends into the genu of the corpus callosum (large arrow). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows heterogeneous, irregularly enhancing left frontal lobe mass (arrows). (c) rCBV map demonstrates no evidence of increased vascularity, suggesting radiation necrosis rather than recurrent tumor. The patient underwent surgical resection, and radiation necrosis without evidence of tumor was found.

View larger version (135K): [in this window] [in a new window] [Download PPT slide]   Figure 10c. Delayed radiation necrosis in an 82-year-old woman. (a) Transverse fluid-attenuated inversion recovery (9,000/110/2,500) MR image shows extensive hyperintensity abnormality that extends into the genu of the corpus callosum (large arrow). (b) Transverse contrast-enhanced T1-weighted (600/14) MR image shows heterogeneous, irregularly enhancing left frontal lobe mass (arrows). (c) rCBV map demonstrates no evidence of increased vascularity, suggesting radiation necrosis rather than recurrent tumor. The patient underwent surgical resection, and radiation necrosis without evidence of tumor was found.

On MR images, gliomatosis manifests as a focal or diffuse infiltrative signal intensity abnormality involving the white matter and expansion of the cortex, usually without contrast enhancement. There is minimal, if any, mass effect or extensive vasogenic edema (79–83). In our series of seven patients with gliomatosis cerebri, the perfusion MR imaging appearance reflected the lack of angiogenesis of this tumor found at histopatholgic examination, and rCBV values tended to be lower than those of low-grade gliomas and even lower than those of normal-appearing unaffected white matter.

Metastatic Neoplasms
Intracranial metastases compose almost 50% of all supratentorial brain tumors (35). Metastatic tumors spread into the central nervous system via hematogenous routes and induce neovascularization as they grow and expand. The newly formed capillaries resemble those of the primary systemic tumor, with gap junctions, fenestrated membranes, and open endothelial junctions, all of which differ radically from normal brain capillaries containing a well-developed blood-brain barrier with tight junctions, a continuous basement membrane, and astrocytic foot processes (84). Intracranial metastatic tumors, especially those that cause neurologic symptoms, are associated with a variable degree of vasogenic edema. The capillaries in the area of vasogenic edema are consistently normal in appearance, and no tumor cells are found beyond the macroscopic boundaries of the tumor (85). The imaging diagnosis of intracranial metastases is usually straightforward and uncomplicated. Metastatic tumors tend to be multiple, well-circumscribed, and favor the gray matter–white matter junction and the patients tend to have a known history of systemic malignancy. On the other hand, solitary metastases, which can occur in 30%–50% of cases, can pose a diagnostic challenge with regard to differentiation from a primary glioma, particularly when there is no history of systemic cancer or the medical work-up fails to identify a primary systemic malignancy.

Conventional MR imaging characteristics of the solitary metastasis and primary glioma are nonspecific and cannot be used with confidence to differentiate between the two. Both tumors tend to enhance and may be associated with a variable degree of peritumoral edema, which is defined as the area of hyperintensity on T2-weighted images in immediate contact with the enhancing tumor margin. In our study of 14 solitary metastases and 25 primary gliomas, perfusion MR imaging could not be used to differentiate the two reliably since both are highly vascular tumors demonstrating increased rCBV (P > .05) (unpublished data, 2002). There was, however, a statistically significant difference (P < .001) in rCBV within the peritumoral region, where the primary glioma values were higher than the solitary metastasis values.

Our experience suggests a promising role for perfusion MR imaging in differentiating solitary metastasis from primary glioma on the basis of differences in peritumoral rCBV measurements (86). This difference in peritumoral rCBV can, in part, be explained by differences in pathophysiology: In metastatic tumors, peritumoral edema represents pure vasogenic edema caused by increased interstitial water due to leaky capillaries (85). In other words, in metastatic tumors, there is no histologic evidence of tumor beyond the outer enhancing margin of the tumor, and the peritumoral region represents reaction of the surrounding intrinsically normal but edematous brain parenchyma. In high-grade gliomas, on the other hand, the peritumoral region represents a variable combination of vasogenic edema and tumor cells infiltrating along the perivascular spaces (87). It has been shown that neoplastic cells can be found in some high-grade gliomas not only outside the enhancing margin but also well beyond the outer edge of the peritumoral zone, as seen on T2-weighted MR images (61). Differentiation between low-grade glioma and metastatic neoplasms can be made on the basis of the usual lack of marked contrast enhancement of the former. Also, peritumoral edema, usually quite marked in solitary metastases, tends to be minimal or absent in low-grade gliomas.

Primary Cerebral Lymphoma
Over the past two decades, the incidence of primary cerebral lymphoma (PCL) has substantially increased in both immunocompromised and immune-competent individuals—a phenomenon that cannot be entirely explained by changes in tumor classification, the increased prevalence of acquired immunodeficiency syndrome, or the growing number of organ transplantations. The incidence ratios of PCL and GBM have risen from 1:250 in 1974 to 1:6 for the years 1981–1990, with PCL now accounting for 6.6%–15.4% of all primary brain tumors (88,89). Unlike other high-grade intracranial neoplasms, PCL is treated with combined high-dose chemotherapy and radiation therapy without surgery. Surgical intervention is usually limited to biopsy to obtain tissue for pathologic diagnosis (90–92). It has been shown that surgical resection of PCL does not necessarily alter prognosis and can lead to profound functional deficits and increased postoperative morbidity. More important, when patients with PCL undergo steroid therapy to decrease intracranial pressure before biopsy, the pathologic findings can be difficult to interpret and a definitive diagnosis cannot be established. In addition, owing to the predilection of PCL to occur near eloquent areas of the brain (eg, deep gray matter, corpus callosum, and subependymal locations), gross total resection may in fact be contraindicated and the performance of biopsy may be challenging. A minimally invasive and accurate diagnosis of PCL is, therefore, an important goal that could clearly alter patient treatment and reduce the risk to the patient.

Radiologic diagnosis of PCL remains a challenge, however. Conventional MR imaging findings of PCL can be similar to those of other intracranial tumors or even to those of demyelinating lesions. PCL lesions usually enhance; can be multiple; can favor, as mentioned previously, a deep gray matter or subependymal location; frequently involve the corpus callosum; and can appear similar to GBM.

The ability of perfusion MR imaging to help detect and quantify tumor angiogenesis can potentially be useful in differentiating high-grade glioma from PCL. One of the most striking histopathologic features of PCL, shown in Figure 11, is the angiocentric growth pattern—tumor cells forming multiple, thick layers around the host vessels—and widening of the perivascular space. Neovascularization is not a prominent feature, although vascular abnormalities such as tumor invasion of endothelial cells and even into the vessel lumen can often be seen.

View larger version (125K): [in this window] [in a new window] [Download PPT slide]   Figure 11. Photomicrographs show angiocentric growth pattern of PCL. Top left: Low-power view shows tumor cells organized around a cerebral vessel (long arrows), as well as in the neuropil (short arrows). (Hematoxylin-eosin stain; original magnification, x50.) Bottom left: Medium-power magnification further emphasizes the perivascular growth of tumor cells that results in a narrowing of the vascular lumina (arrows). (Hematoxylin-eosin stain; original magnification, x100.) Top right: High-power view depicts cross sections of vascular channels (arrows) cuffed by lymphoma cells. (Hematoxylin-eosin stain; original magnification, x200.) Bottom right: Reticulin staining highlights angiocentric growth of the lymphoma and demonstrates reduplication of the vascular matrix (arrows). (Original magnification, x400.)

View larger version (157K): [in this window] [in a new window] [Download PPT slide]   Figure 12a. PCL in a 72-year-old man. (a) Transverse contrast-enhanced T1-weighted (600/14) MR image shows homogeneous enhancement and expansion of the splenium of the corpus callosum (arrows). (b) rCBV map demonstrates no evidence of a marked increase in vascularity (rCBV range, 0.89-1.10).

View larger version (118K): [in this window] [in a new window] [Download PPT slide]   Figure 12b. PCL in a 72-year-old man. (a) Transverse contrast-enhanced T1-weighted (600/14) MR image shows homogeneous enhancement and expansion of the splenium of the corpus callosum (arrows). (b) rCBV map demonstrates no evidence of a marked increase in vascularity (rCBV range, 0.89-1.10).