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) 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 Provenzale, J. M. Articles by Barboriak, D. P. Search for Related Content PubMed PubMed Citation Articles by Provenzale, J. M. Articles by Barboriak, D. P.

Diffusion-weighted and Perfusion MR Imaging for Brain Tumor Characterization and Assessment of Treatment Response¹

¹ From the Department of Radiology, Box 3808, Duke University Medical Center, Durham, NC 27710. Received December 1, 2004; revision requested January 25, 2005; revision received March 15; accepted April 8; final version accepted April 18; final review by J.M.P. January 2, 2006. Address correspondence to J.M.P.

	ABSTRACT

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
Diffusion-weighted magnetic resonance (MR) imaging and perfusion MR imaging are advanced techniques that provide information not available from conventional MR imaging. In particular, these techniques have a number of applications withregard to characterization of tumors and assessment of tumor response to therapy. In this review, the authors describe the fundamental principles of diffusion-weighted and perfusion MR imaging and provide an overview of the ways in which these techniques are being used to characterize tumors by helping distinguish tumor types, assess tumor grade, and attempt to determine tumor margins. In addition, the role of these techniques for evaluating response to tumor therapy is outlined.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
Magnetic resonance (MR) imaging plays an important role in the detection and evaluation of brain tumors. To date, MR imaging has principally served the role of showing the neoplasm, helping distinguish tumors from other pathologic processes, and depicting basic signs of tumor response to therapy, such as change in size and degree of contrast material enhancement. In the past few years, however, a number of advanced MR imaging techniques have been developed that provide new methods for the assessment of brain tumors. One of these techniques, perfusion MR imaging, is now recognized as an important new means for assessing tumors and tumor therapies. Another of these techniques, diffusion-weighted (DW) MR imaging, is also gaining acceptance for the same purposes. The intent of this review is to acquaint the reader with the fundamental features of perfusion and DW MR imaging as applied to tumor assessment and to introduce the reader to new applications of these techniques for the evaluation of therapeutic response.

	DW MR IMAGING OF TUMORS: THEORETICAL BACKGROUND

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
DW Imaging
The image appearance in isotropic DW imaging is based on the principle that molecules in any living tissue routinely undergo random (brownian) motion. Isotropic DW images are typically obtained by measuring loss of signal after a pulse sequence that consists of a series of two sequential gradient pulses added to a 90°–180° spin-echo sequence on either side of the 180° pulse (1). The degree of MR signal loss after application of the second gradient pulse is related to two factors: (a) the duration and strength of the magnetic field gradients and (b) the diffusion coefficient of the substance. The degree of signal loss can be represented by the term S/S₀, in which signal after application of diffusion gradients is represented by S and signal before use of diffusion gradients is represented by S₀. The degree of signal loss S/S₀ is proportional to the exponential of the diffusion coefficient (termed D) and the operator-dependent duration and strength of the encoding gradient (or degree of diffusion weighting, b) as represented by the following formula: S/S₀ e^–bD (1). Mobile molecules experience a substantial loss of signal that is related to the diffusion coefficient and spin-spin relaxation time (T2) of tissue, as well as the pulse sequence parameters chosen by the operator (eg, b value and excitation time).

The apparent diffusion coefficient (ADC) is a value that describes microscopic water diffusibility in the presence of factors that restrict diffusion within tissues (eg, cell membranes, viscosity). The ADC can be derived on a voxel-by-voxel basis and depicted on an ADC map, which allows ADCs in specific regions to be measured by using regions of interest. Measurement of the ADC would be expected to be useful in tumor assessment because variations in water content (and diffusivity), which can be found within tumors for various reasons (eg, necrosis, variations in cellularity) and adjacent to tumors (eg, vasogenic edema), likely provide information that is not readily available from conventional MR imaging.

Diffusion-Tensor Imaging
Diffusion-tensor MR imaging is a technique that has been developed more recently than isotropic (trace-weighted) DW imaging. Typical diffusion-tensor imaging techniques sample water motion in at least six noncollinear directions (rather than in the three directions used in isotropic DW imaging), which provide information about both the rate and the direction of water motion. Diffusion-tensor imaging has shown applicability for a number of disease states owing to the fact that normal-brain white matter is highly structured, and fiber tracts impart a strong orientational bias toward microscopic water diffusion.

The tendency for water molecules to diffuse in some directions rather than equally in all directions is termed "anisotropy." Highly compact white matter fiber tracts exhibit a high degree of anisotropy, and less compact white matter pathways exhibit lesser degrees of anisotropy. All types of white matter typically show greater degrees of anisotropy than are seen in gray matter structures, which have a low degree of anisotropy. Thus, diffusion-tensor imaging provides a sensitive means to detect alterations in the integrity of white matter structures. In fact, in many settings, white matter abnormalities can be seen on diffusion-tensor images that are not evident on routine MR images (2,3). Diffusion-tensor imaging also provides a means of depicting white matter pathways (tractography), which may be useful for providing guidance in neurosurgical procedures by preoperatively depicting important white matter tracts, helping determine infiltration of white matter tracts by tumor, and providing evidence of degeneration of white matter tracts distal to tumor sites (ie, wallerian degeneration).

The mathematical and physical principles underlying diffusion-tensor imaging are complex and not properly the subject of this review. The reader is referred to an article that provides necessary details for understanding topic (4). Diffusion-tensor imaging for preoperative planning will likely be increasingly used. We also refer the reader to recent reviews (5–7) that explain the concepts that underlie tractography and discuss its use in assessing the effects of tumors on white matter tracts.

	PERFUSION MR IMAGING OF TUMORS: THEORETICAL BACKGROUND

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
Angiogenesis
The concept of inhibiting angiogenesis for the purpose of treating tumors was considered revolutionary in 1971 when first proposed by Judah Folkman (8), but it is now widely accepted. In fact, to grow beyond a few millimeters in size, tumors must develop networks of new vascular supply (8).

The development of neoangiogenic networks is promoted by an interaction of various pro- and antiangiogenic tissue factors. One important factor is vascular endothelial growth factor (VEGF), which was first characterized independently in two laboratories (9,10). The expression of VEGF is promoted by nutrient-deficient states, such as tissue hypoxia or hypoglycemia, that occur as tumor cells in rapidly growing pathologic tissues expand beyond the limits of diffusion of nutrients from the native capillaries (11,12). VEGF promotes growth of new endothelial cells by stimulating cell division. However, the new vessels formed in tumors are characteristically abnormal by virtue of having increased tortuosity (13,14), lack of maturity (as evidenced by decreased amounts of perivascular cells) (15), and increased permeability to macromolecules due to the presence of large endothelial cell gaps (16). The result is that the neovessels often have both abnormal flow characteristics and abnormal permeability that can be exploited as potential surrogate markers for the evaluation of tumor growth.

Vascular Dynamics in Vivo
Knowledge of the in vivo circulation of contrast agents, especially through the tumor microenvironment, aids in understanding the parameters used in describing vascular dynamics. Two conceptual schemes have been used in the attempt to model the mechanics of contrast material distribution in vessels and extravasation from vessels. One scheme is the indicator-dilution model, which is based on the Fick principle; this model essentially assumes a simple scheme in which the concentration of an agent is solely a function of the contrast agent volume and the vascular volume. However, this scheme does not adequately take into account the effect of extravasation of the agent.

The second and more recently developed scheme is a pharmacokinetic (multicompartment) model that describes a more complex interaction between two compartments in the tumor: the intravascular space and the extravascular extracellular space (EES). A number of pharmacokinetic models, similar to those used to model drug delivery and of varying complexity, have been developed to explain contrast enhancement in vivo. In one widely cited model proposed by Tofts and Kermode (17), contrast material is assumed to partition between the EES and the intravascular space in a bidirectional manner. Microvascular permeability is expressed as the transendothelial transfer constant, k_PS (in milliliters per 100 cm³ per minute). Changes in signal intensity over time are used to mathematically describe (a) the permeability–surface area product between plasma and EES (influx volume transfer constant [per minute]) and (b) the volume of EES per unit volume of tissue. The efflux rate constant (per minute) is the ratio of the influx volume transfer constant divided by the volume of EES. Increased vessel tortuosity, immaturity, and permeability (seen in high-gradetumors) lead to an increase in the permeability–surface area product. One advantage of this approach is that the analysis does not depend on calculation of T1 values prior to infusion of contrast material, unlike other computational models that use the change in T1 of tissue as an indicator of tissue concentration. Contrast material enters the tumor vasculature and then diffuses to the EES. Thereafter, contrast material may also diffuse from the EES back into the tumor vasculature (18).

Susceptibility-based contrast agents are generally delivered by means of intravenous bolus infusion. If a narrow bolus is administered, the bolus generally remains narrow during the first (and often the second) pass through the tumor bed, with specific enhancement of the vascular bed dependent on blood flow and the amount of contrast material administered. When a dysprosium-based contrast agent is used, the rate of administration was found to affect the shape of the time-enhancement curve but not the area under the signal intensity–time curve, which is proportional to the relative cerebral blood volume (rCBV) (19). Moreover, an increase in the dose of contrast material was found to contribute to improved contrast-to-noise ratio in the tumor bed as a result of improved contrast, not of decreased noise. When gadolinium-based agents are used, improved contrast-to-noise ratios have been reported for a dose of 0.2 mmol/kg, in comparison to a dose of 0.1 mmol/kg (20). In the absence of contrast material extravasation (such as in the setting of an intact blood-brain barrier), enhancement of a specific tissue voxel is dependent on the proportion of functional vessels in the voxel. Thus, regions of high rCBV are thought to reflect areas of high capillary density, which is a reflection of tumor aggressiveness (Fig 1). During a single MR examination, concentration in the vascular bed decreases over time owing to redistribution of the bolus within the vascular space, leakage into the EES at sites outside the central nervous system, and renal or hepatic elimination of the contrast agent.

View larger version (151K): [in this window] [in a new window] [Download PPT slide]   Figure 1a: Conventional MR images and rCBV map in 37-year-old man with biopsy-proved World Health Organization grade III astrocytoma. (a) Transverse contrast-enhanced T1-weighted image shows inhomogeneous mass in right frontal lobe. Anterior portion of the mass (arrowhead) shows contrast enhancement; posterior half of the mass does not. (b) Baseline transverse echo-planar T2*-weighted image from DSC sequence (precontrast) shows homogeneous hyperintense signal throughout the mass. (c) rCBV map shows that anterior portion of the mass (corresponding to contrast-enhanced portion in a) has high rCBV (red and yellow), and posterior portion has low rCBV (blue).

View larger version (90K): [in this window] [in a new window] [Download PPT slide]   Figure 1b: Conventional MR images and rCBV map in 37-year-old man with biopsy-proved World Health Organization grade III astrocytoma. (a) Transverse contrast-enhanced T1-weighted image shows inhomogeneous mass in right frontal lobe. Anterior portion of the mass (arrowhead) shows contrast enhancement; posterior half of the mass does not. (b) Baseline transverse echo-planar T2*-weighted image from DSC sequence (precontrast) shows homogeneous hyperintense signal throughout the mass. (c) rCBV map shows that anterior portion of the mass (corresponding to contrast-enhanced portion in a) has high rCBV (red and yellow), and posterior portion has low rCBV (blue).

View larger version (133K): [in this window] [in a new window] [Download PPT slide]   Figure 1c: Conventional MR images and rCBV map in 37-year-old man with biopsy-proved World Health Organization grade III astrocytoma. (a) Transverse contrast-enhanced T1-weighted image shows inhomogeneous mass in right frontal lobe. Anterior portion of the mass (arrowhead) shows contrast enhancement; posterior half of the mass does not. (b) Baseline transverse echo-planar T2*-weighted image from DSC sequence (precontrast) shows homogeneous hyperintense signal throughout the mass. (c) rCBV map shows that anterior portion of the mass (corresponding to contrast-enhanced portion in a) has high rCBV (red and yellow), and posterior portion has low rCBV (blue).

Contrast Agent Kinetics
Noninvasive dynamic imaging of the transit of exogenous contrast agent through the tumor volume is possible with both computed tomography (CT) and MR imaging. This review will first discuss MR approaches, which are more commonly used. The MR approaches can be divided into three categories that are based on the type of imaging phase that is used to acquire the data.

Dynamic susceptibility-contrast (DSC) approaches are first-pass techniques that are T2- or T2*-weighted. Data obtained with this technique can be analyzed by means of deconvolutionbased on indicator-dilution theory. Although a general term, in the medical literature dynamic contrast-enhanced approaches usually specifically indicate T1-weighted approaches and will be used in that manner here. Combined T1- and T2-weighted approaches are sometimes used as well. The data from dynamic contrast-enhanced imaging can be analyzed by using the pharmacokinetic models outlined earlier. A variety of such models exist. The need for a consensus statement defining functional parameters in several well-known kinetic models was addressed in 1999 (21). Finally, arterial spin labeling is an endogenous contrast approach in which the patient's own blood is magnetically labeled with an MR prepulse prior to the blood's entry into the imaging volume (22,23). The authors of one relatively recent publication (24) indicated that this method compares favorably to exogenous contrast agent methods. The spin-labeling technique will not be discussed further in this review.

DSC MR Techniques
DSC MR approaches, which are based on rapid imaging of the first pass of gadolinium-based contrast material through the tumor vasculature, were first developed in the late 1980s (25–27). At high concentrations, the contrast agent induces substantial T2* shortening, resulting first in loss and then in recovery of signal in the tumor bed as the agent is redistributed or diluted.

DSC MR imaging can be performed by using either a gradient-echo or a spin-echo pulse sequence (28). Gradient-echo DSC sequences tend to be more sensitive to larger vessels, such as veins, in the imaged region. Spin-echo DSC techniques tend to show greater sensitivity to smaller vessels (and therefore are more representative of capillary density) or tumor-specific vessels.

It should be noted that T2* effects extend beyond the borders of the blood vessels into the surrounding tissues; this characteristic is important when there is little leakage of contrast agent into the surrounding tissue, such as with an intact blood-brain barrier (29). These methods can be safely used when the rate of vascular leakage is low. However, when the rate of leakage is high, rCBV mapping results can be underestimated for two reasons: (a) The T2* susceptibility effect is reduced as the gradient of contrast agent is reduced, and (b) there is also signal enhancement due to T1 shortening effects of contrast material in the EES (30). For this purpose, T1-insensitive sequences, small flip angle, or dual-echo approaches (31) are used, as is presaturation of the EES with administration of a preinjection dose of contrast material (32). Postprocessing mathematic corrections are also frequently used (28).

Although at this point a relatively large number of studies in which rCBV was measured have been performed, no single standard technique for rCBV measurement has been established. A number of rCBV measurement methods exist, including placement of a single region of interest and calculation of the mean of repeated rCBV measurements, but few studies have been performed on the reproducibility of rCBV measurements (33). Although reproducibility with some of these techniques appears to be acceptable for present clinical purposes, it remains to be seen how well suited these measurements are for the assessment of moderate changes in rCBV after such interventions as antiangiogenesis therapy—that is, how the biologic variation in these measurements compares with changes due to therapy.

Permeability can also be assessed with the use of DSC images, which would allow one to obtain both permeability and rCBV measurements from the same infusion of contrast material (34). However, the technique is not considered to be valid under conditions in which a very high degree of contrast material leakage is present, which is a limitation in many cases (35).

Dynamic Contrast-enhanced MR Techniques
Dynamic contrast-enhanced MR imaging approaches are based on T1 shortening produced by an infusion of paramagnetic contrast material (17). T1-based changes are primarily a result of contrast material diffusion into the EES. Dynamic imaging is typically performed duringan interval of approximately 5–10 minutes rather than during the first pass of the bolus. However, a T1-based first-pass approach has also been proposed and tested (36–38). In addition, the actual T1 values of the tissues at baseline (before contrast material infusion) are required for most analysis algorithms in order to perform the pharmacokinetic analysis (39). This calculation can be performed by using a series of T1-weighted images obtained at different flip angles (Fig 2). To maintain the required temporal resolution, three-dimensional imaging schemes are usually used (which have the drawback of a longer acquisition time than two-dimensional schemes), and arterial input functions are generally obtained in the center of the acquisition volumes to reduce end-section effects. Because dynamic contrast-enhanced approaches rely on T1 shortening of the EES to develop signal intensity and image contrast, these methods are less optimal than T2*-weighted methods, in which the rate of leakage into the EES is low. As might be expected, data acquisition parameters can influence data analysis and need to be optimized (39). Quantification of absolute cerebral blood volume (rather than rCBV) can be obtained with perfusion CT through use of venous and arterial input functions (40); however, the exact method by which quantification of absolute cerebral blood volume can be obtained using perfusion MR imaging is still a matter of active investigation.

View larger version (158K): [in this window] [in a new window] [Download PPT slide]   Figure 2a: Dynamic contrast-enhanced T1-weighted MR to assess permeability in a patient with high-grade glioma who has residual contrast enhancement after previous resection. Three-dimensional spoiled-gradient-echo sequence was performed every 6.45 seconds for 58 seconds after intravenous infusion of 0.1 mmol/kg gadopentetate dimeglumine. Five unenhanced images were first obtained at five different flip angles to derive T1 values of tissue needed for permeability analysis. (a) Transverse anatomic image of surgical site shows large enhancing region around surgical cavity. (b) Quantitative volume transfer constant (Ktrans) map between blood plasma and EES obtained at same location as in a shows marked increase in Ktrans (arrow) surrounding much of the surgical site. (c) Quantitative map of fractional plasma volume (vp) map obtained at same location as in a shows elevated vp (arrow) in only a small portion of the enhancing region (b and c generated with TOPPCAT; www.radweb.duke.edu/dbplab).

View larger version (142K): [in this window] [in a new window] [Download PPT slide]   Figure 2b: Dynamic contrast-enhanced T1-weighted MR to assess permeability in a patient with high-grade glioma who has residual contrast enhancement after previous resection. Three-dimensional spoiled-gradient-echo sequence was performed every 6.45 seconds for 58 seconds after intravenous infusion of 0.1 mmol/kg gadopentetate dimeglumine. Five unenhanced images were first obtained at five different flip angles to derive T1 values of tissue needed for permeability analysis. (a) Transverse anatomic image of surgical site shows large enhancing region around surgical cavity. (b) Quantitative volume transfer constant (Ktrans) map between blood plasma and EES obtained at same location as in a shows marked increase in Ktrans (arrow) surrounding much of the surgical site. (c) Quantitative map of fractional plasma volume (vp) map obtained at same location as in a shows elevated vp (arrow) in only a small portion of the enhancing region (b and c generated with TOPPCAT; www.radweb.duke.edu/dbplab).

View larger version (146K): [in this window] [in a new window] [Download PPT slide]   Figure 2c: Dynamic contrast-enhanced T1-weighted MR to assess permeability in a patient with high-grade glioma who has residual contrast enhancement after previous resection. Three-dimensional spoiled-gradient-echo sequence was performed every 6.45 seconds for 58 seconds after intravenous infusion of 0.1 mmol/kg gadopentetate dimeglumine. Five unenhanced images were first obtained at five different flip angles to derive T1 values of tissue needed for permeability analysis. (a) Transverse anatomic image of surgical site shows large enhancing region around surgical cavity. (b) Quantitative volume transfer constant (Ktrans) map between blood plasma and EES obtained at same location as in a shows marked increase in Ktrans (arrow) surrounding much of the surgical site. (c) Quantitative map of fractional plasma volume (vp) map obtained at same location as in a shows elevated vp (arrow) in only a small portion of the enhancing region (b and c generated with TOPPCAT; www.radweb.duke.edu/dbplab).

Perfusion CT
A brief discussion of perfusion CT is worthwhile at this point to highlight some advantages relative to perfusion MR imaging. CT approaches to perfusion imaging were first proposed in the early 1980s (46–48). However, clinical uses of perfusion CT were slow to progress because the technique suffered initially from relatively limited imaging volumes and poor temporal resolution. With the advent of multidetector CT, rapid scanning of larger volumes at faster speeds has been possible. Also, because CT uses ionizing radiation while MR imaging does not, MR imaging would seem to offer an advantage. However, CT techniques have been developed that require lower milliampere-second values than those initially employed, so a lower radiation dose is now needed. The use of iodinated contrast agents, with the associated risks of allergic reaction and nephrotoxicity, remains a drawback in some patients. Also, in theory, perfusion CT could suffer when high-attenuating acute hemorrhage is present. However, perfusion CT has some clear advantages over perfusion MR imaging. CT scanners are generally more widely available, and CT does not suffer from magnetic susceptibility artifact, which can compromise perfusion MR images when hemorrhage or other causes of magnetic susceptibility effect are present in the area of interest.

Initial analyses of CT data with deconvolution methods used indicator-dilution methods that suffered from the same problems as DSC MR approaches. However, recent analyses in which kinetic theory was used have allowed better approximation of rCBV and permeability surface values (49–52). The main advantage of CT methods over MR methods is the direct relationship between the CT attenuation coefficient and the contrast material concentration. A thorough review of perfusion CT for stroke imaging has recently been published in Radiology (53). A discussion of the uses of perfusion CT for imaging brain tumors is provided later in this review.

Perfusion Parameters
The most fundamental perfusion parameter used for tumor imaging is rCBV. By using indicator-dilution principles, rCBV can be easily calculated from DSC MR signal intensity data by integrating the area under the concentration-time curve during the first pass of contrast material (25,26). More important, rCBV is a measure of the relative proportion of each tissue voxel that is composed of capillary vessels and can be considered an imaging analog of the microvascular density measurement that is used in histopathologic evaluation. Other perfusion parameters include relative cerebral blood flow and mean transit time, but they are less valuable in the evaluation of brain tumors (54).

A more rigorous pharmacokinetic theory–based determination of rCBV and permeability parameters was first developed by Tofts and Kermode in 1991 (17). As mentioned earlier, a consensus statement by leading investigators that reviewed several pharmacokinetic models and defined standard perfusion and permeability parameters was ultimately published (21).

	DW MR IMAGING FOR DEFINITION OF TUMOR MARGINS

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
Delineation of tumor tissue from normal brain has substantial implications for targeting of focal treatments and for measuring treatment responses. For most brain tumors, particularly high-grade tumors and metastases, a tumor is considered to be a relatively well-marginated region of enhancing tissue (sometimes with central zones of necrosis of variable size) surrounded by peritumoral edema. This model of tumor structure is oversimplistic for primary glial tumors, because tumor cells are known to extend into the unenhancing region of peritumoral edema and even into normal-appearing white matter far from the primary lesion. Tumors may have both enhancing and nonenhancing zones or may be primarily nonenhancing.

Physiologic changes in tumor that may portend a poorer prognosis but do not initially increase the volume of contrast-enhancing tumor (eg, high degree of angiogenesis) can occur, but these changes are typically difficult or impossible to detect. One example of this phenomenon is an increase in rCBV in a tumor: Elevated rCBV has been shown in some studies to be a predictor of poor survival (55). As the following discussion will indicate, detection of such important phenomena could potentially be performed by using perfusion and DW MR imaging. In addition, however, a large number of investigators are examining various molecular imaging techniques in animal models that could provide such information and, it is hoped, might one day be available for imaging human brain tumors (56).

When devising an imaging paradigm to best depict and delineate tumor, one must consider the characteristics of the target. It might be assumed that the ideal MR imaging technique for this purpose would directly depict neoplastic cells. In the case of high-grade gliomas, however, microscopic clusters of such cells have been detected at autopsy at a considerable distance from the primary tumor mass (57). In fact, studies with MR spectroscopy (58) have shown that, in a substantial number of cases, abnormal spectra consistent with tumor can be seen beyond the contrast-enhancing tumor and regions that are abnormal on T2-weighted images. Because the signal-to-noise characteristics of DW and perfusion imaging sequences are rather poor and voxel sizes tend to be large, it seems unlikely that these sequences will be able to depict such small clusters of neoplastic cells, which may occupy a small fraction of a voxel.

Other targets related to the pathophysiology of tumor growth and spread have the disadvantage of being indirectly linked to the actual presence of neoplastic cells but may be more easily detected by using MR imaging. For example, degeneration of a white matter tract disrupted by tumor may be easier to detect than the cells responsible for the disruption. Similarly, breakdown of the blood-brain barrier or abnormal vascularity may be easier to detect than the cells secreting angiogenic or permeability-enhancing factors.

Several investigators have studied the application of DW and diffusion-tensor imaging to delineation of the tumor margin. By and large, the results of these studies do not correlate the location of DW or diffusion-tensor imaging abnormalities with histopathologically defined tumor boundaries; rather, the approach taken has been to demonstrate differences in peritumoral characteristics in high-grade gliomas (which show some degree of tumor infiltration into unenhancing regions adjacent to the enhancing mass), as compared with low-grade gliomas, metastases, and meningiomas (which presumably do not show such tumor infiltration).

A number of studies (59–62) have shown that ADCs in peritumoral edema are higher for metastases than forprimary cerebral tumors. These findings suggest that ADC mapping of enhancing lesions may have some usefulness for predicting the underlying disease, but they do not in and of themselves prove that DW imaging is depicting infiltration. Attempts to provide better definition of tumor boundaries by using DW imaging at 1.5 T, rather than conventional MR imaging, by demonstrating areas of nonenhancing tumor have been largely unsuccessful (63–65).

To date, any role for DW or diffusion-tensor imaging in the evaluation of vasogenic edema surrounding a mass remains speculative (59). In one study (60), when fractional anisotropy and ADCs in edema surrounding meningiomas or metastases were compared with the same measurements in edema surrounding low- or high-grade gliomas, no significant differences were found. However, the authors of that study suggested that high-grade glioma could be differentiated from metastases and low-grade glioma by using measurements in perilesional edema of a so-called tumor infiltration index, which is based on the relationship of fractional anisotropy to mean diffusivity. Similarly, in another study (66) investigators who compared fractional anisotropy values and ADCs in perilesional edema surrounding meningiomas and the same measurements in edema surrounding high-grade gliomas in an attempt to distinguish perilesional edema from infiltrating tumor found that the difference approached but did not achieve statistical significance. Thus, it remains unclear whether diffusion-tensor imaging can help distinguish between these two entities. In only one study (67) has there been an attempt to directly correlate ADCs in specific locations with pathologic examination findings ofneuroimaging-navigated biopsy specimens obtained at those sites. That study, which consisted of 44 biopsies in 22 patients, showed considerable overlap between the ADCs of tumor and peritumoral tissues; the authors found that addition of DW imaging to a conventional tumor imaging protocol, including gadolinium-enhanced T1-weighted sequences, provided no important additional diagnostic information.

On the other hand, two reports (66,68) have shown that diffusion-tensor imaging may help in detection of white matter abnormalities in patients with malignant tumors in areas that appear normal on T2-weighted MR images, which raises the possibility that disruption of white matter tracts by tumor infiltration may be detectable by using diffusion-tensor imaging. Price et al (68) used an anisotropy index similar to fractional anisotropy to show areas of possible white matter disruption in normal-appearing white matter in a majority of patients with high-grade glioma, which was not seen in patients with metastatic disease or low-grade glioma. In one case, an area of white matter abnormality detected only with diffusion-tensor imaging converted to an area of contrast-enhancement and hyperintensity on T2-weighted images 2 years after being detected. In another study (66), our group showed significant decreases in fractional anisotropy in normal-appearing white matter near gliomas, compared with normal values in normal-appearing white matter adjacent to meningiomas (Fig 3).

View larger version (135K): [in this window] [in a new window] [Download PPT slide]   Figure 3a: Measurement of fractional anisotropy values in various tumor and peritumoral regions in 56-year-old woman with biopsy-proved World Health Organization grade IV glioblastoma multiforme. Normalized fractional anisotropy values in solid enhancing tumor are similar to those in adjacent unenhancing regions (which may represent vasogenic edema or unenhancing tumor). (a) Transverse contrast-enhanced T1-weighted image shows enhancing mass in left parietal lobe. Regions of interest (ROIs) have been placed in solid portion of tumor (1), hypointense region (vasogenic edema or unenhancing tumor) adjacent to tumor (3), normal-appearing WM adjacent to region of hypointense signal near tumor (5), and in corresponding contralateral WM areas (2, 4, 6). (b) Transverse echo-planarMR image from diffusion-tensor sequence before application of diffusion gradients (b = 0 sec/mm2) on which ROIs from a have been superimposed confirms that ROI 1 is located in region of hyperintense signal related to enhancing tumor, 3 is in area of peritumoral edema or unenhancing infiltrating tumor, and 5 is in region of normal signal intensity. (c) Color-coded fractional anisotropy map (yellow and red = high anisotropy, blue = low) shows anisotropy values in enhancing tumor that are 42% of those in corresponding contralateral regions.

View larger version (134K): [in this window] [in a new window] [Download PPT slide]   Figure 3b: Measurement of fractional anisotropy values in various tumor and peritumoral regions in 56-year-old woman with biopsy-proved World Health Organization grade IV glioblastoma multiforme. Normalized fractional anisotropy values in solid enhancing tumor are similar to those in adjacent unenhancing regions (which may represent vasogenic edema or unenhancing tumor). (a) Transverse contrast-enhanced T1-weighted image shows enhancing mass in left parietal lobe. Regions of interest (ROIs) have been placed in solid portion of tumor (1), hypointense region (vasogenic edema or unenhancing tumor) adjacent to tumor (3), normal-appearing WM adjacent to region of hypointense signal near tumor (5), and in corresponding contralateral WM areas (2, 4, 6). (b) Transverse echo-planarMR image from diffusion-tensor sequence before application of diffusion gradients (b = 0 sec/mm2) on which ROIs from a have been superimposed confirms that ROI 1 is located in region of hyperintense signal related to enhancing tumor, 3 is in area of peritumoral edema or unenhancing infiltrating tumor, and 5 is in region of normal signal intensity. (c) Color-coded fractional anisotropy map (yellow and red = high anisotropy, blue = low) shows anisotropy values in enhancing tumor that are 42% of those in corresponding contralateral regions.

View larger version (131K): [in this window] [in a new window] [Download PPT slide]   Figure 3c: Measurement of fractional anisotropy values in various tumor and peritumoral regions in 56-year-old woman with biopsy-proved World Health Organization grade IV glioblastoma multiforme. Normalized fractional anisotropy values in solid enhancing tumor are similar to those in adjacent unenhancing regions (which may represent vasogenic edema or unenhancing tumor). (a) Transverse contrast-enhanced T1-weighted image shows enhancing mass in left parietal lobe. Regions of interest (ROIs) have been placed in solid portion of tumor (1), hypointense region (vasogenic edema or unenhancing tumor) adjacent to tumor (3), normal-appearing WM adjacent to region of hypointense signal near tumor (5), and in corresponding contralateral WM areas (2, 4, 6). (b) Transverse echo-planarMR image from diffusion-tensor sequence before application of diffusion gradients (b = 0 sec/mm2) on which ROIs from a have been superimposed confirms that ROI 1 is located in region of hyperintense signal related to enhancing tumor, 3 is in area of peritumoral edema or unenhancing infiltrating tumor, and 5 is in region of normal signal intensity. (c) Color-coded fractional anisotropy map (yellow and red = high anisotropy, blue = low) shows anisotropy values in enhancing tumor that are 42% of those in corresponding contralateral regions.

Future Directions
A major limitation of studies in which DW imaging was used to study tumors is the lack of correlation of pathologic and imaging findings. Such a correlation would likely require animal models (60). Another approach would be to carefully correlate the locations and results of biopsies with imaging findings in humans. However, registration of the biopsy site with a DW or diffusion-tensor imaging location is particularly difficult given the distortion of these images due to underlying magnetic susceptibility artifacts. Registration of the ADC maps to the images obtained with b = 0 sec/mm², which are T2-weighted images, may lessen the differences in distortion and improve the quality of the registration. Also, shifting of the brain during craniotomy prevents close anatomic correlation with preoperative images; this a problem is lessened by the use of stereotactic biopsies (although smaller degrees of brain shifting still occur) and may be better managed through the use of intraoperative MR imaging. DW imaging with the periodically rotated overlapping parallel lines with enhanced reconstruction, or PROPELLER, method, in which data are collected in concentric rectangular strips rotated about the k-space origin and which allows correction of spatial inconsistencies in rotation, position, and phase between strips (thereby correcting for motion and decreasing magnetic susceptibility effects), may best provide the means to correct for the artifacts outlined above and allow more accurate coregistration of images (72). Alternatively, line-scanning DW techniques, which are less affected by magnetic susceptibility artifacts but are much more time consuming because they are performed without echo-planar imaging, may be used.

One potential confounding factor not fully accounted for in most studies is the spatial variation of diffusion anisotropy with location. It is known that the diffusion anisotropy of white matter in the subcortical region is substantially less than that of central white matter and the corpus callosum (73). For this reason, it is important that fractional anisotropy values be compared in a region of interest as close as possible to the same location of the brain in the opposite hemisphere (66). Many of the pathologic entities with which the high-grade gliomas are compared, particularly meningiomas and metastases, are more likely to occur near the white matter periphery than in the central white matter. It is possible that differences in peritumoral edema found in studies may reflect differences in the tendency of white matter in different locations to show increased ADC (or decreased fractional anisotropy values) in response to tumor. For instance, it is possible that ADCs related to metastases located in subcortical white matter tend to differ from ADCs related to metastases in central white matter regions.

	PERFUSION AND DW MR IMAGING FOR TUMOR CHARACTERIZATION

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
Perfusion Imaging: rCBV-based Approaches
Given the correlation of microvessel density and tumor grade and the further correlation between microvessel density and rCBV, higher grade tumors would be expected to have higher rCBV. This hypothesis has been verified in a number of DSC MR experiments (74–79). In those studies, rCBV has typically been calculated by comparing the pathologic region to a corresponding region in the contralateral hemisphere. Most investigators have chosen to use a normal-appearing region in contralateral white matter as a control region, which generally differs in location from patient to patient (24). No specific location in normal brain tissue is usually chosen. Values in high-grade tumor are typically seen to be at or near rCBV values in gray matter. Furthermore, some evidence exists that nonglial neoplasms (eg, lymphoma) appear to have lower rCBV than do high-grade glial neoplasms and that the two types of tumors might be distinguishable by using perfusion MR imaging (80). It is important to note that one of the hallmarks of high-grade lesions is tumor heterogeneity; therefore, rCBV maps often show only small foci of high rCBV.

One potential pitfall of rCBV maps derived from DSC MR imaging occurs in the setting of high capillary permeability when substantial contrast material leakage is present. An rCBV map is generated by analyzing a time–signal intensity curve after infusion of MR contrast material. The passage of the contrast material through blood vessels produces a decrease in signal intensity on T2*-weighted images. Thereafter, the signal intensity of tissue returns to (or near) baseline. However, in lesions that are associated with disruption of the blood-brain barrier (eg, contrast-enhancing tumors), the signal intensity as seen on the time–signal intensity curve can actually increase above baseline owing to the T1-shortening effects of leakage of contrast material. If the limits of integration (which are operator derived) are drawn to include the region of the curve that is above baseline, the rCBV of the region is underestimated. This concept has been reviewed in detail elsewhere (29). It is important that individuals deriving rCBV values keep this factor in mind when assessing rCBV in lesions.

Measurement of rCBV by using dynamic contrast-enhanced MR has also shown good correlation with tumor grade according to the World Health Organization tumor grading scheme (81). The majority of results with this method have focused on permeability measurement and will be described later. Recently, perfusion MR imaging with arterial spin labeling has begun to show promise as a technique for hemodynamic assessment of brain tumors. With this noninvasive technique, which does not require an exogenous tracer, a subject's own blood is used as an intrinsic tracer that is labeled by using an inversion pulse outside the imaging plane of interest (24).

In comparison, perfusion CT studies of rCBV are fewer in number. The CT approach has been demonstrated in animal tumor models (50). In addition, investigators have examined regional blood flow, capillary permeability, and relative compartmental volumes in human neck tumors by using a technique that analyzes not only the initial phase of contrast enhancement but also the time-attenuation curve for the first few minutes after infusion of contrast material (49). In a report of five cases from Australia (82), high-grade tumors had heterogeneous rCBV maps with regions of high rCBV, whereas low-grade lesions had only low rCBV. In another case report of a patient with a cardiac pacemaker and high-grade glioma who could not undergo MR imaging (52), perfusion CT showed moderately elevated rCBV and markedly elevated relative cerebral blood flow and mean permeability surface flow within the tumor, compared with those measurements in normal tissue.

Perfusion Imaging: Permeability-based Approaches
A few initial studies have used dynamic contrast-enhanced MR to help assess tumor grade. The details of pulse sequence design and analysis techniques and their advantages relative to one another are described in detail in numerous reports (18,21,83). The groundwork for this effort can be seen in a study (84) in which investigators found a positive correlation between tumor microvascular characteristics on dynamic MR images and histopathologic grade in mammary soft-tissue tumors. By using both a low-molecule-weight contrast medium (gadopentetate) and a macromolecular gadolinium-based or iron-containing agent with enough protein or albumin binding to increase the relaxivity and prolong the residence time of the agent in the intravascular compartment, permeability surface area product (as calculated from the endothelial transfer coefficient) and fractional plasma volume were determined for each tumor and contrast agent (84). Tumor permeability, characterized by the permeability surface product, was highly correlated with increasing tumor grade for the macromolecular contrast medium, but no such correlation was seen in the arm of the study that used gadopentetate. In a similar manner, a significant correlation was seen between increasing fractional plasma volume and higher tumor grade for the macromolecular contrast agent but not for gadopentetate (84).

Other investigators who used dynamic contrast-enhanced MR to assess K^trans, a measure of permeability, found that typical meningiomas could be distinguished from atypical meningiomas by using this technique (85). Specifically, atypical meningiomas, which are characterized by a higher mitotic index, increased cellularity, and higher frequency of spontaneous necrosis than typical meningiomas were found to have a significantly higher degree of degree of permeability (as measured with K^trans) than typical meningiomas. The authors attributed these findings to a possible increased size of endothelial gap junctions, which may be related to the micronecrosis seen in atypical meningiomas.

Comparison of perfusion findings from CT and MR are uncommon in the medical literature. However, in one report (51) in which perfusion CT and dynamic contrast-enhanced MR techniques were compared in the same patient, a relatively good correlation was seen between the two techniques with regard to the tumoral regions of highest permeability and regions of highest rCBV.

As mentioned earlier, permeability can also be assessed from DSC MR images, within certain limitations. In one study (34), investigators using this technique found a correlation between increasing degree of permeability and increasing tumor grade, in cases in which tumor grade was determined from biopsy specimens.

DW Imaging for Grading Tumors
Theoretically, DW MR imaging may allow the cellularity of tumors to be graded noninvasively; because cells constitute a relative barrier to water diffusion, compared with the extracerebral space, more cellular tumors (ie, tumors with more cells per unit volume) would be expected to show less of an increase in ADCs than would less cellular tumors. Alternatively, some authors (86) have suggested that decreased expression of hydrophilic glyocoamnioglycans (for example, hyaluronan) in the extracellular spaces of high-grade tumors may lead to less of an increase in ADC in these tumors. Some studies (87–89) have indeed shown lower water diffusibility in high-grade gliomas than in lower grade gliomas. However, a considerable overlap between ADCs in high-grade tumors and ADCs in low-grade tumors has been noted (63). It remains doubtful whether tumor grade can be evaluated with enough specificity with DW imaging to be useful in a clinical context. Specifically, no evidence has yet been found to suggest that DW or diffusion-tensor MR can be used to help identify specific sites for biopsy.

Some differences in conflicting findings from different studies may be due to whether areas of necrosis are carefully excluded from the analyses of tumor ADCs. Such necrotic regions are more common in high-grade tumors and would be expected to contribute highly elevated ADCs that would raise the mean values for the tumor. Although it might seem relatively simple to separate areas of necrosis from those of solid tumor mass by comparing ADC maps and T2-weighted MR images, in practice it is difficult to accurately register ADC maps to conventional images because of distortion of the underlying DW images.

The use of diffusion-tensor imaging to help detect differences in tumor grades is less well studied. Mean fractional anisotropy values in high-grade gliomas were higher than those in low-grade tumors in one report (90), but this finding was not confirmed in another study (60).

DW Imaging for Evaluation of Tumor Type
The presence of an inverse correlation between tumor cellularity and tumor ADC measurement (discussed in the previous section) raises the possibility that tumor ADC measurement may be helpful in distinguishing tumors with relatively low cellularity (such as high-grade gliomas) from other lesions. Indeed, ADC measurements have been found to be decreased in lymphomas relative to measurements in other intracranial tumors such as high-grade gliomas (Figs 4, 5), metastases, or meningiomas (64,91,92). Similarly in the pediatric age group, water diffusion has been found to be relatively restricted in cellular brain tumors such asmedulloblastoma and other primitive neuroectodermal tumors, compared with diffusion measured in other lesions (93–96). Although these findings are promising, studies with larger numbers of patients will be needed to determine whether ADC measurements can be useful in clinical practice to narrow the differential diagnosis of intracranial masses.

View larger version (158K): [in this window] [in a new window] [Download PPT slide]   Figure 4a: High mean ADC in 70-year-old man with partially resected glioblastoma multiforme. (a) Transverse contrast-enhanced T1-weighted MR image shows rim-enhancing mass in left temporal lobe. (b) Transverse T2-weighted MR image shows hyperintense resection cavity in central portion of mass, which is surrounded by hyperintense abnormality consistent with vasogenic edema or unenhancing tumor. Regions of interest in solid portion of mass (1) and contralateral normal white matter (2) indicate sites of ADC measurement for (c) transverse ADC map, which shows ADC in tumor is 183% of that in normal brain, corresponding to low cell density and nuclear-to-cytoplasmic ratio and abundant extracellular matrix seen at histologic analysis. (Reprinted, with permission, from reference 79.)

View larger version (151K): [in this window] [in a new window] [Download PPT slide]   Figure 4b: High mean ADC in 70-year-old man with partially resected glioblastoma multiforme. (a) Transverse contrast-enhanced T1-weighted MR image shows rim-enhancing mass in left temporal lobe. (b) Transverse T2-weighted MR image shows hyperintense resection cavity in central portion of mass, which is surrounded by hyperintense abnormality consistent with vasogenic edema or unenhancing tumor. Regions of interest in solid portion of mass (1) and contralateral normal white matter (2) indicate sites of ADC measurement for (c) transverse ADC map, which shows ADC in tumor is 183% of that in normal brain, corresponding to low cell density and nuclear-to-cytoplasmic ratio and abundant extracellular matrix seen at histologic analysis. (Reprinted, with permission, from reference 79.)

View larger version (91K): [in this window] [in a new window] [Download PPT slide]   Figure 4c: High mean ADC in 70-year-old man with partially resected glioblastoma multiforme. (a) Transverse contrast-enhanced T1-weighted MR image shows rim-enhancing mass in left temporal lobe. (b) Transverse T2-weighted MR image shows hyperintense resection cavity in central portion of mass, which is surrounded by hyperintense abnormality consistent with vasogenic edema or unenhancing tumor. Regions of interest in solid portion of mass (1) and contralateral normal white matter (2) indicate sites of ADC measurement for (c) transverse ADC map, which shows ADC in tumor is 183% of that in normal brain, corresponding to low cell density and nuclear-to-cytoplasmic ratio and abundant extracellular matrix seen at histologic analysis. (Reprinted, with permission, from reference 79.)

View larger version (113K): [in this window] [in a new window] [Download PPT slide]   Figure 5a: Low mean ADC representing metastasis from primary lesion outside the central nervous system in a 62-year-old man with large B-cell lymphoma. (a) Transverse contrast-enhanced T1-weighted MR image shows large right frontal lobe enhancing lesion with mass effect. (b) Transverse T2-weighted MR image shows signal intensity of central portion of mass is only slightly higher than that of gray matter, suggesting high cell density. Signal intensity is lower than that shown for glioblastoma in Figure 4. (c) Transverse ADC map shows mass has lower signal intensity than normal tissue. Mean ADC in the mass region of interest (1) was 65% of that of normal tissue, much lower than that for patient in Figure 4.

View larger version (122K): [in this window] [in a new window] [Download PPT slide]   Figure 5b: Low mean ADC representing metastasis from primary lesion outside the central nervous system in a 62-year-old man with large B-cell lymphoma. (a) Transverse contrast-enhanced T1-weighted MR image shows large right frontal lobe enhancing lesion with mass effect. (b) Transverse T2-weighted MR image shows signal intensity of central portion of mass is only slightly higher than that of gray matter, suggesting high cell density. Signal intensity is lower than that shown for glioblastoma in Figure 4. (c) Transverse ADC map shows mass has lower signal intensity than normal tissue. Mean ADC in the mass region of interest (1) was 65% of that of normal tissue, much lower than that for patient in Figure 4.

View larger version (85K): [in this window] [in a new window] [Download PPT slide]   Figure 5c: Low mean ADC representing metastasis from primary lesion outside the central nervous system in a 62-year-old man with large B-cell lymphoma. (a) Transverse contrast-enhanced T1-weighted MR image shows large right frontal lobe enhancing lesion with mass effect. (b) Transverse T2-weighted MR image shows signal intensity of central portion of mass is only slightly higher than that of gray matter, suggesting high cell density. Signal intensity is lower than that shown for glioblastoma in Figure 4. (c) Transverse ADC map shows mass has lower signal intensity than normal tissue. Mean ADC in the mass region of interest (1) was 65% of that of normal tissue, much lower than that for patient in Figure 4.

Stereotactic Biopsy Planning
More accurate preoperative grading of primary brain tumors could have a major effect on the lives of patients with brain tumors, perhaps by leading to fewer biopsies. It remains unclear, however, how MR imaging and surgical biopsies can theoretically "compete" for superior preoperative characterization, because these two diagnostic modalities are quite different in scale and scope. Surgical biopsies have the advantage of allowing a thorough histologic evaluation of very small volumes of tumor but pose the possibility of substantial sampling error. MR imaging theoretically allows better overall assessment of the entire volume of a tumor but cannot adequately account for small regional heterogeneities.

Because tumor grade is assigned on the basis of the region of highest grade anywhere in the tumor, an imaging test that could reliably depict such a site of highest tumor grade would be optimal for increasing the likelihood of correct diagnosis. At present, neurosurgeons make heavy use of contrast-enhanced T1-weighted MR images for this purpose and typically attempt to perform biopsy in an area of strong contrast-enhancement. Perfusion parameters such as rCBV and K^trans may well prove to be more reliable indicators of tumor grade than degree of contrast enhancement and could potentially be used for more accurate determination of the optimal biopsy site (Fig 6). Studies in which biopsy of multiple regions in a tumor that has heterogeneous regions showing various rCBV and permeability measurements would be helpful in establishing the role of perfusion techniques, if the risks of multiple biopsies could be minimized. The value of this approach is described by several authors anecdotally, and many institutions appear to be utilizing this approach (28,75). Controlled studies in which samples from various regions in a tumor are obtained might prove difficult to implement but would be one means of establishing the preferred method of optimization of the biopsy site.

View larger version (129K): [in this window] [in a new window] [Download PPT slide]   Figure 6a: Use of rCBV map to determine optimal biopsy site in a 44-year-old woman with seizures who was suspected of having a brain tumor. Subsequent biopsy showed World Health Organization grade III astrocytoma. (a) Transverse T2-weighted MR image shows hyperintense region (arrow) in left frontal lobe. (b) Transverse contrast-enhanced T1-weighted MR image shows no areas of abnormal contrast enhancement, making determination of region of optimal biopsy site difficult. (c) rCBV map obtained with DSC technique (red and yellow = high rCBV, green and blue = low rCBV). A high rCBV area (arrow) corresponds to region of hyperintense abnormality in a. At biopsy, this region was shown to represent high-grade glioma. (Reprinted, with permission, from reference 30.)

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 6b: Use of rCBV map to determine optimal biopsy site in a 44-year-old woman with seizures who was suspected of having a brain tumor. Subsequent biopsy showed World Health Organization grade III astrocytoma. (a) Transverse T2-weighted MR image shows hyperintense region (arrow) in left frontal lobe. (b) Transverse contrast-enhanced T1-weighted MR image shows no areas of abnormal contrast enhancement, making determination of region of optimal biopsy site difficult. (c) rCBV map obtained with DSC technique (red and yellow = high rCBV, green and blue = low rCBV). A high rCBV area (arrow) corresponds to region of hyperintense abnormality in a. At biopsy, this region was shown to represent high-grade glioma. (Reprinted, with permission, from reference 30.)

View larger version (124K): [in this window] [in a new window] [Download PPT slide]   Figure 6c: Use of rCBV map to determine optimal biopsy site in a 44-year-old woman with seizures who was suspected of having a brain tumor. Subsequent biopsy showed World Health Organization grade III astrocytoma. (a) Transverse T2-weighted MR image shows hyperintense region (arrow) in left frontal lobe. (b) Transverse contrast-enhanced T1-weighted MR image shows no areas of abnormal contrast enhancement, making determination of region of optimal biopsy site difficult. (c) rCBV map obtained with DSC technique (red and yellow = high rCBV, green and blue = low rCBV). A high rCBV area (arrow) corresponds to region of hyperintense abnormality in a. At biopsy, this region was shown to represent high-grade glioma. (Reprinted, with permission, from reference 30.)

	PERFUSION AND DW MR IMAGING FOR ASSESSMENT OF TUMOR THERAPY

TOP ABSTRACT INTRODUCTION DW MR IMAGING OF... PERFUSION MR IMAGING OF... DW MR IMAGING FOR... PERFUSION AND DW MR... PERFUSION AND DW MR... SUMMARY ESSENTIALS References

 
Perfusion Imaging
At present, the predominant method for assessing tumor response to therapy is to measure the size of anenhancing lesion on conventional MR images. However, tumor size is a relatively nonspecific feature of tumor behavior. For instance, internal characteristics of tumors associated with tumor aggressiveness (such as permeability and rCBV) may change in the absence of a change in tumor size. The true strength of perfusion imaging techniques is the ability to depict changes in the internal architecture in the setting of no overall change in tumor size. For instance, because rCBV is thought to correspond to microvessel density, early response to antiangiogenesis therapy might manifest as a decrease in rCBV. Alternatively, because some angiogenesis factors are permeability promoters, a decreased rate of contrast material leakage might be an early indication of response to antiangiogenesis therapy.

The use of MR imaging to monitor tumor angiogenesis and assess the response to antiangiogenic therapies has been the subject of much effort. In fact consensus panels have published recommendations concerning the use of MR in this capacity (97,98). The biologic basis of tumor-mediated angiogenesis is now well established, as described in the Angiogenesis section earlier in this review. Several attributes of angiogenesis are detectable with the aid of MR imaging (99). Of these, rCBV, relative cerebral blood flow, and K^trans are the most widely evaluated.

A few initial studies have used dynamic contrast-enhanced MR to assess for tumor response to therapy, but these have primarily been performed in animal models. In a murine breast tumor model, decreased tumor growth rates and decreases in microvascular permeability were seen on MR images after treatment with anti-VEGF antibody (100,101). These studies used a heavily T1-weighted three-dimensional spoiled gradient-recalled acquisition in the steady-state sequence and a macromolecular contrast medium. In another study (102) that used a dynamic contrast-enhanced technique, a decrease in microvascular permeability was seen after rats that harbored a tumor raised from human glioblastoma multiforme cell lines were treated with a monoclonal antibody against VEGF. In one of the few trials of an antiangiogenesis therapy for human brain tumors (103), early decreases were seen for rCBV (measured on DSC MR images) and degree of contrast enhancement (measured on dynamic contrast-enhanced MR images).

DW Imaging
Perhaps one of the most exciting potential applications of DW MR imaging has been in measurement of the response of solid tumors to therapy. The results of several studies (