Amide Proton Transfer Imaging of Human Brain Tumors at 3T

Transcription

1 Magnetic Resonance in Medicine 56: (2006) Amide Proton Transfer Imaging of Human Brain Tumors at 3T Craig K. Jones, 1,2 Michael J. Schlosser, 3 Peter C.M. van Zijl, 1,2 Martin G. Pomper, 2 Xavier Golay, 1,2,4 and Jinyuan Zhou 1,2 * Amide proton transfer (APT) imaging is a technique in which the nuclear magnetization of water-exchangeable amide protons of endogenous mobile proteins and peptides in tissue is saturated, resulting in a signal intensity decrease of the free water. In this work, the first human APT data were acquired from 10 patients with brain tumors on a 3T whole-body clinical scanner and compared with T 1 -(T 1 w) and T 2 -weighted (T 2 w), fluid-attenuated inversion recovery (FLAIR), and diffusion images (fractional anisotropy (FA) and apparent diffusion coefficient (ADC)). The APT-weighted images provided good contrast between tumor and edema. The effect of APT was enhanced by an approximate 4% change in the water signal intensity in tumor regions compared to edema and normal-appearing white matter (NAWM). These preliminary data from patients with brain tumors show that the APT is a unique contrast that can provide complementary information to standard clinical MRI measures. Magn Reson Med 56: , Wiley-Liss, Inc. Key words: magnetization transfer; amide proton exchange; MT; CEST; APT; brain tumor; protein MRI is an important tool for localizing, diagnosing, and characterizing brain tumors, and assessing the effects of treatment (1 3). In addition to the widely used conventional MRI techniques (4), such as T 1 - (T 1 w) and T 2 - weighted (T 2 w) imaging, contrast-enhanced imaging, and diffusion-weighted imaging (DWI), several other techniques, such as perfusion imaging (5,6) and more recently proton spectroscopy (7 9) and diffusion tensor imaging (DTI) (10 12) are finding their way into clinical research protocols to undergo further clinical validation. At present, a major problem in planning treatment for brain tumor patients is the difficulty of defining tumor boundaries using MRI. For instance, increased signal intensity on T 2 w imaging may be due to either tumor infiltration or peritumoral vasogenic edema. MRI cannot easily distinguish between the two. In gadolinium (Gd)-enhanced T 1 w imaging (13), only the portion of the tumor in which the blood brain barrier (BBB) has been disrupted is visible. This area may correspond to highly cellular and/or growing areas of the lesion, but for malignant gliomas this does not represent the complete extent of tumor infiltration. Thus, the tumor boundaries are not delineated accurately. Proton MR spectroscopic imaging (MRSI) may accomplish such tissue delineation (7 9), but it suffers from reduced sensitivity and concomitant limited spatial resolution and increased scanning time. Therefore, it is important to develop complementary MRI methodologies that can visualize tumor properties. Recently, a new MRI methodology called amide proton transfer (APT) imaging was developed. With this method the interaction between protons of free tissue water and the amide groups of endogenous mobile proteins and peptides is imaged (14,15). When applied to rats implanted with 9L gliosarcoma tumors (15), APT was able to distinguish between pathology-confirmed regions of tumor and edema, which could not be accomplished using standard T 1 w/t 2 w imaging and DWI, in which the tumor border appeared diffuse. It is therefore of interest to explore this approach on humans. Because the efficiency of APT increases with the T 1 of water (16,17), and because of the requirement for slow-exchange NMR conditions, APT imaging improves at higher field. However, APT technology has the disadvantage of high power deposition and is very susceptible to inhomogeneities in the B 0 field. The goals of this study were to 1) address the technical problems and implement the APT technology on a high-field clinical scanner, and 2) quantify the APT effect at 3T in patients with brain tumors. MATERIALS AND METHODS Theory APT imaging is a variant of magnetization transfer (MT) imaging (18 20), in which the selective saturation of the magnetization of amide protons is detected indirectly 1 F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, Maryland, USA. through chemical exchange with bulk water protons. Selective irradiation is possible because there is a composite 2 Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA. amide proton resonance at approximately 3.5 ppm downfield from the water resonance (21 23), and in the brain, 3 Department of Neurosurgery, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA. amide protons exchange with water protons at a rate (k) of 4 Department of Neuroradiology, National Neuroscience Institute, Singapore. Grant sponsor: NIH/NIBIB; Grant numbers: EB02634; EB02666; Grant sponsor: NIH/NCRR; Grant number: RR15241; Grant sponsor: NIH/NCI; Grant ference ( ) with the water resonance is generally suffi- about 30 times per second (14). Thus, the frequency dif- number: CA92871; Grant sponsors: Whitaker Foundation; Philips Medical cient to provide conditions of slow exchange on the NMR Systems. *Correspondence to: Jinyuan Zhou, Ph.D., Department of Radiology, Johns time scale ( k). At 3T, 2 *448 Hz 2815 radians. The amide pool is small (mm concentration range), Hopkins University School of Medicine, 217 Traylor Building, 720 Rutland Ave., Baltimore, MD jzhou@mri.jhu.edu but continuous saturation leads to a measurable decrease Received 17 October 2005; revised 10 April 2006; accepted 6 May by a few percent of the large water signal due to a sensitivity enhancement mechanism, analogously to chemical DOI /mrm Published online 4 August 2006 in Wiley InterScience ( com). exchange saturation transfer (CEST) imaging (24 28) Wiley-Liss, Inc. 585

2 586 Jones et al. Table 1 Characteristics of the 10 Brain Tumor Patients Scanned and the Final Postoperative Pathology Report on the Tumor* Number Sex Lesion location Final path Handedness Presentation APT (%) 1 F Right frontal Grade II oligodendroglioma L Seizures 5.5 (0.3) 2 M Left temporal Glioblastoma R Language dysfunction 1.9 (0.06) 3 F Left frontal Grade II astrocytoma R Loss of control of right foot 4 M Left frontal Grade II astrocytoma L Seizures 1.9 (0.07) 5 M Right parietal Meningioma L Incidental 6 M Right temporal Glioblastoma R Headache, lethargy 4.1 (0.2) 7 F Left frontal Meningioma R Hand weakness/numbness 3.5 (0.09) 8 M Left temporal Grade II astrocytoma L Seizures 9 M Right frontal Grade III oligodendroglioma R Seizures 1.9 (0.06) 10 F Left frontal Grade II oligodendroglioma R Right face and arm dysthesia 0.2 (0.05) *The mean and standard error of the APT in percent units are shown for the APT hot tumor region for each patient with a defined hot region. M male, F female, L left, R right. Even though proton exchange effects are much smaller than the baseline MT effect caused by solid-like tissue components, they can be detected when analyzing the asymmetry of the frequency dependence of the MT effect with respect to the water frequency (14,15,24,25). Conventional MT can be detected over a very large frequency range, up to 100 khz. This frequency range is determined by the dipolar dipolar interaction and chemical shift anisotropy for semisolid macromolecules in tissue. The main differences between MT and APT are the substance specificity of the saturation effects, the mechanism of saturation transfer, and the approximately symmetric appearance of conventional MT with respect to the water resonance. APT effects originate from mobile molecules and depend purely on chemical exchange, while conventional MT depends on both chemical exchange and cross-relaxation. The magnitude of the MT effects is usually described by the MT ratio (MTR), defined by the equation MTR 1 S sat /S 0, in which S sat and S 0 are the signal intensities with and without selective irradiation, respectively. The effects can be visualized by studying the frequency dependence of the water saturation as a function of the frequency of the saturation, the socalled z-spectrum (29). It is common for the frequencies in a z-spectrum to be referenced to water, thereby referring to the amide proton resonance of interest as being at an offset of 3.5 ppm. The effect of APT coexists with direct water saturation and, in vivo, conventional MT (18) and even blood oxygen level-dependent (BOLD) effects (14). However, when performing an asymmetry analysis with respect to the water resonance (14,15,24,25), the APT effect can be distinguished by calculating the MTR asymmetry parameter: MTR asym MTR MTR S sat /S 0 S sat /S 0. [1] It was previously demonstrated (14,15) that for an offset of 3.5 ppm, MTR asym (3.5 ppm) MTR asym (3.5 ppm) APTR [2] in which MTR asym (3.5 ppm) is the inherent asymmetry of the solid-phase MT effect associated with immobile macromolecules and membranes, and APTR is the proton transfer ratio for the amide protons associated with mobile cellular proteins and peptides (APTR is PTR( 3.5 ppm)). When field homogeneity is poor and the water resonance is not centered properly at 0 ppm, MTR asym (3.5 ppm) may also include the contributions of B 0 field inhomogeneity. The APTR can be written as: k amide proton] APTR 1 e [water proton]r R1wt sat ) [3] 1w where k is the single-proton exchange rate (amide to water, k 10 ph pkw ), the square brackets ([...]) denote the concentration, R 1w is the spin-lattice relaxation rate of water, and t sat is the saturation time. The tissue water proton concentration and water content, w, are directly related ([water proton] 2 55 M w), whereas R 1w and water content have been reported to be inversely proportional (30): R 1w 1/T 1w 1 w 1, where is a scaling factor (15). Thus, there are partly compensated-for effects for water content, R 1w, and the exponential relaxation term in the equation. Because the exponential term, e R1wtsat, is close to 0, APTR is approximately proportional to 1/(1 w). In order to be distinguishable from conventional image contrast, the effects of amide proton content and/or exchange have to be substantially different from the other terms, especially those for water content. MRI Experiments Ten patients with brain tumors (see Table 1 for information) and three normal volunteers were scanned on a whole-body Philips 3T Intera scanner (Philips Medical Systems, Best, The Netherlands) equipped with secondorder shims. All scanning was approved by the Johns Hopkins Medicine Institutional Review Board (IRB) and the Kennedy Krieger Institute, and all subjects provided signed consent. For all scans, a quadrature head coil was used for RF transmission and reception. For saturation, an

3 APT Imaging at 3T 587 off-resonance RF pulse was applied for3satapower level of 3 T and a turbo spin-echo (TSE) imaging readout (TSE factor [number of refocusing pulses] 32, TR 6s,TE 30 ms, matrix , FOV mm, slice thickness 5 mm, single slice). Two patients were scanned using 33 offsets from 8 to 8 ppm with an interval of 0.5 ppm to verify the offset dependence of the proton transfer effects. The other eight patients were scanned at two offsets ( 3.5 ppm relative to the water frequency) and with eight averages to increase the SNR. A control image without the saturation RF pulse was also acquired. The total scan time for each of these two experiments was approximately 10 min. For reference, T 2 w (TSE factor 8, TR 4s,TE 80 ms, 60 slices, thickness 2.2 mm), T 1 w (3D fast field echo (FFE), TR 8.33 ms, TE 3.9 ms, TI 820 ms, flip angle 8, 120 slices, 1.1 mm isotropic voxels), fluid attenuated inversion recovery (FLAIR, TSE factor 19, inversion delay 2.8 s, TR 11 s, TE 120 ms, 60 slices, thickness 2.2 mm), diffusion tensor (single-shot SE echo-planar imaging (EPI), TR 9699 ms, TE 63 ms, 30 gradient directions with b 700 s/mm 2 and a reference, 60 slices, thickness 2.2 mm), and Gd-enhanced T 1 w (same parameters as T 1 w above) images were also acquired. Data Processing All of the data were transferred to a Sun Fire 880 server (Sun Microsystems, Mountain View, CA) for processing. Unless otherwise noted, all processing was done using software written in-house in Matlab (The Mathworks, Natick, MA, USA). The APT data acquired at 33 frequency offsets were analyzed within regions of interest (ROIs). The mean signal intensity was calculated for each offset and fitted with a 15-order polynomial (14,15). The minimum of the fitted curve was assumed to be the on-resonance water frequency and was shifted to be 0 ppm (relative to water). The asymmetry curve was then calculated from the shifted z-spectrum. The shift was typically found to be less than 0.2 ppm. For APT data acquired at the two offsets of 3.5 ppm (and a control scan without irradiation), the MTR asym (3.5 ppm) was calculated from Eq. [1] for each voxel in the brain. The asymmetry image was thresholded based on the signal intensity of S 0 images to remove voxels outside the brain. This was necessary because the signal intensity outside the brain was near zero in S 0 and created artificially large values in the asymmetry image. When MTR asym (3.5 ppm) images were computed using the twopoint method, it was assumed that minimal asymmetry was present due to the B 0 inhomogeneity because the second-order shims minimized the issue. The DTI data were processed in DTI Studio (developed by Drs. Hangyi Jiang and Susumu Mori, Department of Radiology, Johns Hopkins University, supported by NIH/ NCRR resource grant RR15241; for more information see The average isotropic apparent diffusion coefficient ADC ave Trace(D)/3 and fractional anisotropy (FA) were calculated from the diffusion tensor matrix D. FA was calculated from the eigenvalues i (i 1, 2, 3) using the equation FA 3/ ]/ Directionally encoded color (DEC) maps of the main direction of water diffusion were calculated based on the orientation of the primary eigenvector of the diffusion tensor matrix (where red, green, and blue represent diffusion primarily in the left right, anterior posterior, and superior inferior directions, respectively (31). For each of the multislice or 3D images, a single slice matched with the APT scans was chosen for quantitative analysis. Five ROIs were drawn by a neurosurgeon (M.J.S.): 1) brain tumor where the region had high APT compared to the rest of the tumor region (APT-hot tumor), 2) a region encompassing the whole brain tumor (tumor), 3) contralateral normal-appearing white matter (CNAWM), 4) ipsilateral NAWM near the tumor (INAWM), and 5) edema near the tumor (edema). When the regions were drawn, all images were displayed on the computer screen for comparison to minimize variability. For the APT, FA, and ADC, the mean and standard deviation (SD) of the absolute measures were calculated. For the FLAIR, preand post-gd T 1 w, and T 2 w images, the signal intensity in each of the regions was normalized by the signal intensity in the CNAWM for ease of comparison across subjects, and then the mean and SD were calculated. RESULTS The z-spectra and MTR asym spectra acquired from a patient (patient 3 in Table 1) with a WHO grade II astrocytoma are depicted in Fig. 1. The normalized signal (S sat /S 0 ) is higher in the tumor than in the CNAWM. This decreased MT effect (32 34), is likely due to increased edema within the tumor region. However, the offset region 0 8 ppm is very close to the water resonance and the difference in z-spectra in this region can be attributed largely to the difference in the T 2 -related direct water saturation (i.e., increased T 2 results in narrowed saturation linewidth for the tumor). The resulting MT asymmetry curves (Fig. 1c) show a varying MTR asym that is slightly positive between 0 and 4 ppm and then becomes slightly negative at offsets greater than 6 ppm. This is in agreement with previous animal studies (14,15,35) showing that the solid-like MT effect is slightly asymmetric with respect to the water resonance, with the MT center frequency in the aliphatic range. Indeed, there is a visible increase in the tumor MTR asym compared to the CNAWM. The range of offsets over which this increase occurs corresponds well to the spectral frequency range of amide protons, which typically resonate between 1 and 5 ppm from the water resonance, indicating that this difference originates predominantly from the amide protons of the mobile proteins and peptides. As expected, the MTR asym measure is not sensitive to the proportional S/S 0 of the tumor. Figure 2 compares the acquired APT-weighted images (i.e., the MTR asym (3.5 ppm) images) with standard MR images for a patient (patient 5 in Table 1) with a pathologyconfirmed meningioma. There is a clear intensity increase in the tumor in the APT-weighted image compared to NAWM. It is important to notice the lack of APT enhancement in a hyperintense edema region that occurs in several of the conventional MR images. The fact that this region

4 588 Jones et al. FIG. 1. Experimental results for a patient with an astrocytoma: (a) regions (circles) of the tumor and CNAWM used to create the MT curves, and (b) z-spectra and (c) MTR asym spectra obtained with tumor relative to CNAWM. Saturation power 3 T, duration 3 s. The maximum APT appears in an offset range between 2 and 4 ppm relative to water. Patient 3 in Table 1. did not show Gd enhancement and was confirmed to be edema supports the potential utility of APT as an endogenous contrast to differentiate edema from tumor, as shown in the initial animal studies (14,15). Figure 3 demonstrates APT and other MR contrasts for a patient with a glioblastoma multiforme (patient 6 in Table 1). The area of high intensity in the APT image delineates the main tumor, the border of which is similar to the Gd-enhanced scan. This is in contrast to the meningioma case shown in Fig. 2, in which the APT-enhanced region is a little different from the Gd-enhanced region. The increased APT signal in the gyri recti superior to the sinus is likely due to a susceptibility artifact, as seen in other APT scans (discussed below). As discussed in our previous animal study (15), the APT image contrast of brain tumors is expected to be dominated by mobile protein and peptide content, which would be higher inside than outside the tumor regions (36,37). In eight out of 10 cases studied, there was an increased signal within the region of the APT image that correlated with the region identified as likely to represent tumor (as opposed to edema) on standard imaging sequences. We identified an APT-hot region in seven out of 10 patients. For the APT-weighted image, the APT-hot region was significantly different from the CNAWM in six of the seven patients (P 0.001) and was significantly different from FIG. 2. An example of the APT image compared with several other types of MR images at 3T for a patient with a meningioma (patient 5 in Table 1). The mean and standard error of the APT (in percent units) are: (tumor), (CNAWM), (INAWM), and (edema). APT differentiates an edema region vs. the tumor.

5 APT Imaging at 3T 589 FIG. 3. APT image compared with several other types of MRI contrast for a patient with a glioblastoma (patient 6 in Table 1). The units of APT are percentage points in the water signal intensity, and the values are (APT-hot tumor), (tumor), (CNAWM), (INAWM), and (edema). The Gd-enhanced T 1 w scan demonstrates a heterogeneous mass in the anterior right temporal lobe. Surrounding WM appears bright on the T 2 w image, consistent with peritumoral edema and/or infiltrating tumor. the INAWM in all seven of the patients (P ). Of the three who did not have a defined APT-hot region, one patient (patient 8 in Table 1) with a WHO grade II astrocytoma in the left anterior-inferior temporal lobe revealed a large signal artifact in the inferior frontal lobes (Fig. 4), just superior to the ethmoid sinuses. We attribute this to a large change in the local magnetic field due to the susceptibility difference between brain tissue and air. No increased signal was visually apparent in the region of the tumor. A second patient (patient 3 in Table 1) with an astrocytoma did not have a distinct APT region with high FIG. 4. Example APT image (right) that did not show enhancement in the tumor area as seen on the T 2 w image (left). The large red spot in the upper middle of the APT image is the result of a susceptibility artifact from the ethmoid sinus. The two-point APT technique is dependent on good B 0 homogeneity, which can be problematic near an air tissue interface. signal, but instead had a homogeneous small increase compared to the surrounding NAWM. Similarly to the second patient, the third patient (patient 5 in Table 1) with a meningioma did not have a distinct region of increased APT. Five of the 10 patients had both a well-defined APThot region and edema region, and for all five patients the APT measures within the two regions were significantly different. A quantitative analysis was performed on the images from all of the patients. For all patients an APT image was generated and compared with the results of the other MRI sequences. For the pre- and post-gd T 1 w, T 2 w, and FLAIR sequences, the signal intensity in the ROI was normalized to the CNAWM signal intensity to allow group comparison. The summary is shown in Fig. 5. The enhanced effect of APT is approximately a 4% change in the water signal intensity in APT-hot tumor regions. The statistical differences between APT-hot tumor or tumor and the other regions are shown for each contrast in Fig. 5. The other significant differences not shown in the figure are as follows: CNAWM and edema (P 0.01) on FA, INAWM and edema (P 0.01) on FA, CNAWM and edema (P 0.001) on T 1 w-gd, INAWM and edema (P 0.01) on T 1 w-gd, CNAWM and edema (P 0.01) on T 1 w, CNAWM and edema (P 0.01) on ADC ave, and INAWM and edema (P 0.01) on ADC ave. DISCUSSION Capability of APT Imaging Ten patients with brain tumors were scanned using the APT technique on a clinical MR scanner. The z-spectrum

6 590 Jones et al. FIG. 5. Seven measures quantified for five ROIs from the MR images. The intensities on the T 1 w, Gd-enhanced T 1 w, T 2 w, and FLAIR images were normalized by those in CNAWM for the corresponding images. APT creates a 4% change in the water signal in the APT-hot tumor area compared to CNAWM. Symbols denote significant differences from APT-hot tumor (*) and tumor ( ). A single symbol represents P 0.05, and two symbols represent P in brain tumors showed increased MT asymmetry around the offset of 3.5 ppm relative to the water resonance. This is in agreement with previous animal studies (14,15), in which an asymmetry analysis of z-spectra showed the maximum effects of APT at 3.5 ppm. In all patients with an APT-hot tumor, the APT signal increase was statistically different from the CNAWM, INAWM, and edema. This is in contrast to all other images in which the tumor was difficult to distinguish from edema. Equation [3] indicates that increased APTR in the tumor can potentially be attributed to increased cellular protein and peptide content, and increased amide proton exchange rates with respect to normal brain. In addition, tissue water content and spin-lattice relaxation effects may also contribute. Thus, care has to be taken in interpreting this result. In the previous animal study (15), the APTR increase in tumors was argued to be predominantly due to increased protein/peptide content. For APT to be a successful image contrast, the effect of the increased amide proton concentration must outweigh that of the increased water content. In addition, the exchange rate of amide protons in tumor tissue should be equal to (or preferably a little higher than) that of brain tissue to enhance the effect of increased amide proton concentration. The data indicate that these requirements are, fortunately, fulfilled for human brain tumor imaging. First, the T 1 w images, which generally reflect brain water content, show contrast in the tumor regions, but do not necessarily relate to regions of increased APT. This can be seen in Fig. 2, in which there is contrast within the tumor on the T 1 w image, but it does not correlate well to the increased signal intensity in the APT image. This would suggest that the water change within a tumor is not the largest factor contributing to signal change in the APT image. Second, multiple studies

7 APT Imaging at 3T 591 (38,39) of many tumor types indicate an intracellular ph range that is higher (up to about 0.1 ph unit) than that of normal brain tissue. As the amide proton exchange rate is base-catalyzed in the physiological ph range, the exchange rate increases with ph, thus increasing the APT contrast. The MTR asym (3.5 ppm) image is weighted by the APT contrast and the inherent asymmetry in the conventional MT effect (Eqs. [2] and [3]). Under the assumption of small differences in the inherent asymmetry term of different tissues, the contrast differences in the MTR asym (3.5 ppm) images predominantly reflect the APT effect and are thus denominated as APT-weighted images. Technical Considerations The APT technique is sensitive to inhomogeneities in the B 0 field. A small change in the B 0 field and a concomitant shift in the z-spectrum may cause a relatively large change in the MTR asym. As described above, one approach to reduce this problem is to acquire images over a range of saturation frequency offsets, and then fit the z-spectra and center them appropriately around the water frequency in each voxel. However, this correction method requires that many points along the z-spectrum be acquired for proper fitting, which is not optimal for fast scanning. A second approach would be to acquire a limited number of saturation points at offsets around 3.5 ppm. The shifts between voxels can then be corrected using a field map. The easiest way to reduce asymmetry changes related to B 0 inhomogeneity would be to reduce the main magnetic field inhomogeneity by improving the shim capabilities. APT requires a long RF pulse for saturation on the magnetization of amide protons and transfer of saturation to the water protons. In this work we used the T/R head coil because the long RF pulses result in high power depositions and are limited by the RF amplifier. We used a 3-s, 3- T off-resonance RF pulse for saturation, for an SAR of 3 W/kg, which is within FDA guidelines. This optimized saturation scheme was based on minimal MTR asym contrast between normal brain tissues (gray and white matter) and CSF. Because the MTR asym value of CSF is always small (around zero) for all offsets, we expect that this saturation scheme causes a positive MTR asym (3.5 ppm) value in the tumor region (or a negative value in the stroke area). Such an optimized power to have a maximum APT signal was also confirmed in studies on protein (bovine serum albumin (BSA), histone, poly-l-lysine) solutions and on normal volunteers (data not shown). Because of SAR guidelines and hardware restrictions, we were unable to use a 3-s pulse with the body coil transmission for our 3T scanner. Shorter pulse durations are possible, but they would have a decreased saturation over a broader frequency range. Parallel imaging techniques, such as sensitivity encoding (SENSE), would result in shorter scan times, but on our clinical whole-body scanner such methods would require the use of the body coil for transmission, and therefore a higher SAR. Coincidentally, the frequency difference between the lipid resonances and water is nearly equal to that for the amide proton resonance ( 3.5 ppm), but of opposite sign. Thus, any MTR asym calculation for the amide proton resonance will also involve saturation effects on the lipid resonance (40). The MTR asym images at 3.5 ppm have a very negative signal in the fat around the brain. Therefore attention must be paid to reduce fat effects, including fat shift artifacts, such as those occurring in multi-readout acquisitions like EPI. To reduce this problem, we chose to use a TSE sequence that does not suffer from such large water fat shift artifacts. The TSE factor had to be chosen carefully to minimize the scan time without too much blurring in the phase-encode direction. We found that a TSE factor of 33 resulted in a reasonable trade-off. It is clear that the APT contrast mechanism displays distinct information that complements standard MR images. However, the implementation of this technique and the reproducible measurement of these small changes are not trivial. In addition, for this technique to be more useful in clinical studies, it will be important to achieve whole brain coverage within a clinically feasible time. In the present study, the average scan time was approximately 10 min for a single slice, which is too long. Multislice/3D imaging would require the build-up of a steady-state MT effect to allow sufficient coverage. A parallel imaging technique would be able to improve scan time, but this would require the use of a larger transmit coil, with concomitant power deposition problems that have to be addressed. CONCLUSIONS APT imaging was implemented on a whole-body 3T scanner. The initial results from patients with brain tumors revealed information that is not provided by conventional MRI methods, especially with respect to the separation of regions with tumor tissue and peritumoral edema. The z-spectra from NAWM and tumor were similar to those obtained in previous animal studies. A quantitative regional analysis showed that the APT contrast distinguished tumor regions from suspected edema and was increased by approximately 4% with respect to normal tissue. Technically, the approach is eminently clinically translatable, and the results will be further validated by selective tissue biopsies. APT is a new MRI modality with the potential to aid in diagnosis and surgical planning for patients with brain tumors. ACKNOWLEDGMENTS We are grateful to our MR technologists, Ms. Terri Brawner and Ms. Kathleen Kahl, for their help with the data acquisition; Mr. Joe Gillen for advice on software issues; and Dr. Jaishri Blakeley for helpful discussions. Dr. Jones was supported by a grant from Philips Medical Systems to the Kennedy Krieger Research Institute. Dr. van Zijl is a paid lecturer for Philips Medical Systems. This arrangement has been approved by Johns Hopkins University in accordance with its conflict-of-interest policies. REFERENCES 1. Gillies RJ, Bhujwalla ZM, Evelhoch J, Garwood M, Neeman M, Robinson SP, Sotak CH, Van Der Sanden B. Applications of magnetic resonance in model systems: tumor biology and physiology. Neoplasia 2000;2:

8 592 Jones et al. 2. Evelhoch JL, Gillies RJ, Karczmar GS, Koutcher JA, Maxwell RJ, Nalcioglu O, Raghunand N, Ronen SM, Ross BD, Swartz HM. Applications of magnetic resonance in model systems: cancer therapeutics. Neoplasia 2000;2: Pomper MG, Port JD. New techniques in MR imaging of brain tumors. Magn Reson Imaging Clin N Am 2000;8: Oh J, Cha S, Aiken AH, Han ET, Crane JC, Stainsby JA, Wright GA, Dillon WP, Nelson SJ. Quantitative apparent diffusion coefficients and T2 relaxation times in characterizing contrast enhancing brain tumors and regions of peritumoral edema. J Magn Reson Imaging 2005;21: Lev MH, Hochberg F. Perfusion magnetic resonance imaging to assess brain tumor responses to new therapies. Cancer Control 1998;5: Akella NS, Twieg DB, Mikkelsen T, Hochberg FH, Grossman S, Cloud GA, Nabors LB. Assessment of brain tumor angiogenesis inhibitors using perfusion magnetic resonance imaging: quality and analysis results of a phase I trial. J Magn Reson Imaging 2004;20: Alger JR, Frank JA, Bizzi A, Fulham MJ, DeSouza BX, Duhaney MO, Inscoe SW, Black JL, van Zijl PC, Moonen CT. Metabolism of human gliomas: assessment with H-1 MR spectroscopy and F-18 fluorodeoxyglucose PET. Radiology 1990;177: McKnight TR, von dem Bussche MH, Vigneron DB, Lu Y, Berger MS, McDermott MW, Dillon WP, Graves EE, Pirzkall A, Nelson SJ. Histopathological validation of a three-dimensional magnetic resonance spectroscopy index as a predictor of tumor presence. J Neurosurg 2002;97: Nelson SJ. Multivoxel magnetic resonance spectroscopy of brain tumors. Mol Cancer Ther 2003;2: Gauvain KM, McKinstry RC, Mukherjee P, Perry A, Neil JJ, Kaufman BA, Hayashi RJ. Evaluating pediatric brain tumor cellularity with diffusion-tensor imaging. AJR Am J Roentgenol 2001;177: Mori S, Frederiksen K, van Zijl PC, Stieltjes B, Kraut MA, Solaiyappan M, Pomper MG. Brain white matter anatomy of tumor patients evaluated with diffusion tensor imaging. Ann Neurol 2002;51: Ulmer JL, Salvan CV, Mueller WM, Krouwer HG, Stroe GO, Aralasmak A, Prost RW. The role of diffusion tensor imaging in establishing the proximity of tumor borders to functional brain systems: implications for preoperative risk assessments and postoperative outcomes. Technol Cancer Res Treat 2004;3: Graif M, Bydder GM, Steiner RE, Niendorf P, Thomas DG, Young IR. Contrast-enhanced MR imaging of malignant brain tumors. AJNR Am J Neuroradiol 1985;6: Zhou J, Payen JF, Wilson DA, Traystman RJ, van Zijl PC. Using the amide proton signals of intracellular proteins and peptides to detect ph effects in MRI. Nat Med 2003;9: Zhou J, Lal B, Wilson DA, Laterra J, van Zijl PC. Amide proton transfer (APT) contrast for imaging of brain tumors. Magn Reson Med 2003;50: Zhou J, Wilson DA, Sun PZ, Klaus JA, Van Zijl PC. Quantitative description of proton exchange processes between water and endogenous and exogenous agents for WEX, CEST, and APT experiments. Magn Reson Med 2004;51: Woessner DE, Zhang S, Merritt ME, Sherry AD. Numerical solution of the Bloch equations provides insights into the optimum design of PARACEST agents for MRI. Magn Reson Med 2005;53: Wolff S, Balaban RS. NMR imaging of labile proton exchange. J Magn Reson 1990;86: Henkelman RM, Stanisz GJ, Graham SJ. Magnetization transfer in MRI: a review. NMR Biomed 2001;14: Stanisz GJ, Odrobina EE, Pun J, Escaravage M, Graham SJ, Bronskill MJ, Henkelman RM. T(1), T(2) relaxation and magnetization transfer in tissue at 3T. Magn Reson Med 2005;54: Mori S, Eleff SM, Pilatus U, Mori N, van Zijl PC. Proton NMR spectroscopy of solvent-saturable resonances: a new approach to study ph effects in situ. Magn Reson Med 1998;40: van Zijl PC, Zhou J, Mori N, Payen JF, Wilson D, Mori S. Mechanism of magnetization transfer during on-resonance water saturation. A new approach to detect mobile proteins, peptides, and lipids. Magn Reson Med 2003;49: Chen W, Hu J. Mapping brain metabolites using a double echo-filter metabolite imaging (DEFMI) technique. J Magn Reson 1999;140: Guivel-Scharen V, Sinnwell T, Wolff SD, Balaban RS. Detection of proton chemical exchange between metabolites and water in biological tissues. J Magn Reson 1998;133: Ward KM, Aletras AH, Balaban RS. A new class of contrast agents for MRI based on proton chemical exchange dependent saturation transfer (CEST). J Magn Reson 2000;143: Aime S, Barge A, Delli Castelli D, Fedeli F, Mortillaro A, Nielsen FU, Terreno E. Paramagnetic lanthanide(iii) complexes as ph-sensitive chemical exchange saturation transfer (CEST) contrast agents for MRI applications. Magn Reson Med 2002;47: Zhang S, Merritt M, Woessner DE, Lenkinski RE, Sherry AD. PARAC- EST agents: modulating MRI contrast via water proton exchange. Acc Chem Res 2003;36: Goffeney N, Bulte JW, Duyn J, Bryant Jr LH, van Zijl PC. Sensitive NMR detection of cationic-polymer-based gene delivery systems using saturation transfer via proton exchange. J Am Chem Soc 2001;123: Bryant RG. The dynamics of water-protein interactions. Annu Rev Biophys Biomol Struct 1996;25: Lin W, Venkatesan R, Gurleyik K, He YY, Powers WJ, Hsu CY. An absolute measurement of brain water content using magnetic resonance imaging in two focal cerebral ischemic rat models. J Cereb Blood Flow Metab 2000;20: Pajevic S, Pierpaoli C. Color schemes to represent the orientation of anisotropic tissues from diffusion tensor data: application to white matter fiber tract mapping in the human brain. Magn Reson Med 1999;42: Mehta RC, Pike GB, Haros SP, Enzmann DR. Central nervous system tumor, infection, and infarction: detection with gadolinium-enhanced magnetization transfer MR imaging. Radiology 1995;195: Kurki T, Lundbom N, Kalimo H, Valtonen S. MR classification of brain gliomas: value of magnetization transfer and conventional imaging. Magn Reson Imaging 1995;13: Hawighorst H, Debus J, Schreiber W, Knopp MV, Engenhart-Cabillic R, Essig M, Brix G, van Kaick G. Contrast-enhanced magnetization transfer imaging: improvement of brain tumor conspicuity and delineation for radiosurgical target volume definition. Radiother Oncol 1997;43: Pekar J, Jezzard P, Roberts DA, Leigh Jr JS, Frank JA, McLaughlin AC. Perfusion imaging with compensation for asymmetric magnetization transfer effects. Magn Reson Med 1996;35: Hobbs SK, Shi G, Homer R, Harsh G, Atlas SW, Bednarski MD. Magnetic resonance image-guided proteomics of human glioblastoma multiforme. J Magn Reson Imaging 2003;18: Howe FA, Barton SJ, Cudlip SA, Stubbs M, Saunders DE, Murphy M, Wilkins P, Opstad KS, Doyle VL, McLean MA, Bell BA, Griffiths JR. Metabolic profiles of human brain tumors using quantitative in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 2003;49: Griffiths JR. Are cancer cells acidic? Br J Cancer 1991;64: Ross BD, Higgins RJ, Boggan JE, Knittel B, Garwood M. 31P NMR spectroscopy of the in vivo metabolism of an intracerebral glioma in the rat. Magn Reson Med 1988;6: Sun PZ, Zhou J, Sun W, Huang J, van Zijl PC. Suppression of lipid artifacts in amide proton transfer imaging. Magn Reson Med 2005;54: