MR Imaging

All ASL-PI examinations were performed on a clinical 1.5T MR imaging unit (Magnetom Symphony; Siemens, Erlangen, Germany) with a product transmit/receive head coil within 0–40 days before surgery. ASL-PI was performed by using a second version of quantitative imaging of perfusion by means of a single subtraction with the addition of thin-section periodic saturation (Q2TIPS), which is a pulsed arterial spin-labeling method that enables the acquisition of multiple sections.¹⁵ Q2TIPS has 2 important planes and 3 parameters: a labeling plane; an imaging plane; and TI1, TI1S, and TI2.¹⁵ The labeling plane is the region where the proximal arterial blood is labeled by an inversion recovery radio-frequency pulse. The imaging plane is the region where the data acquisition of the perfusion imaging is performed. TI1 is a timing parameter representing the amount of time that passes between the inversion recovery pulse for labeling the blood and the periodic saturation pulse, both of which are applied to the labeling plane. Therefore, TI1 is related to the amount of labeled blood volume. TI1S is a timing parameter and represents the amount of time between the inversion recovery pulse for labeling and the end of the saturation pulse on the labeling plane. TI2 is a timing parameter and represents the amount of time between the inversion recovery pulse for labeling and the beginning of the imaging data acquisition by an echo-planar imaging (EPI) sequence. Under the current Q2TIPS setting, the labeling plane comprised a 100-mm-wide region and was placed 10 mm proximal to the imaging plane. The imaging plane was set so as to coincide with the greatest dimension of the tumor by reference to the findings on T1-weighted images (T1WI), T2- weighted images (T2WI), and fluid-attenuated inversion recovery images (FLAIR).

Five imaging sections were acquired sequentially in a proximal-to-distal direction by using a single-shot gradient-echo EPI technique (section width/gap = 5 /2.5 mm, TR/TE = 2100/26 ms, FOV = 240 × 240 mm, matrix = 64 × 64 pixels, voxel size = 3.75 × 3.75 × 5 mm, band width = 1532 Hz/pixel, crusher gradients = 10 cm/s, phase partial Fourier acquisition rate = 6/8, phase oversampling = 13%, acquisition time per section = 49.5 ms). The signal intensity of the perfusion images was corrected for the delay due to the sequential acquisition. The inversion and saturation pulse parameters of Q2TIPS were set as follows: TI1 = 900 ms, TI1S = 1300 ms, and TI2 = 1400 ms. Fifty pairs of labeled and nonlabeled images were obtained in 3 minutes 30 seconds. The perfusion images were generated by subtracting the labeled image from the nonlabeled image at each section position.

In addition to ASL-PI, conventional images including T1WI, T2WI, FLAIR, and postcontrast T1WI were obtained by using any of 3 different clinical 1.5T MR imaging units (Magnetom Vision and Symphony, Siemens; Achieva, Royal Philips Electronics, Eindhoven, the Netherlands). The parameters for those conventional images were as follows: T1WI: TR/TE = 441–620/9.5–14 ms; T2WI: TR/TE = 2500–4902.5/90–100 ms; FLAIR: TR/TE/ TI = 9000/97–110/2200–2400 ms; postcontrast T1WI: TR/TE = 525–800/17 ms. Postcontrast images were obtained after intravenous administration of gadolinium diethylene triamine pentaacetic acid (0.1 mmol/kg; Magnevist, Nihon Schering, Osaka, Japan). In 1 patient, postcontrast T1WI was not performed because the patient had asthma.

ASL Data Processing

By reference to the findings on T1WI, T2WI, FLAIR, and postcontrast T1WI, a region of interest that had a size of at least 26 pixels was placed within the tumor. Additionally, on the same image, another region of interest was placed within the normal-appearing cortical gray matter (GM) in the contralateral cerebral hemisphere on a simply visual basis. Each region of interest was set by a neuroradiologist (T.N.) who was not blinded to the diagnosis. As an index of tumor perfusion, we used a relative signal intensity (SI) (%Signal intensity), which was defined as the ratio of maximal SI within the tumor (Tmax) to the averaged SI of GM (GMav) (%Signal intensity = Tmax / GMav × 100[%]). The relative perfusion index was used rather than the absolute blood-flow value because the accuracy of the absolute perfusion measurement by using Q2TIPS has not been fully established, especially in pathologic conditions.^8,9 Although the absolute blood flow value could be theoretically determined by using ASL,^15,16 the quantitative CBF by using the ASL sequence can be affected by unstable factors such as the arterial transit time of the labeled blood from the labeling plane to the brain tissue within the imaging plane, the local relaxation times of tissue and blood, the assumption of a constant T1 relaxation time of arterial blood irrespective of vessel size or blood oxygenation,¹⁰ and the equilibrium magnetization and T1 value of the blood.¹⁷ CBF values measured by using ASL are thought to be proportional rather than equivalent to the actual regional CBF values.^8,9

Additionally, considering the large interindividual differences in perfusion,¹⁸ the relative perfusion measurement is the most reliable method at present. The averaged SI of GM instead of that of white matter (WM) was adopted as a reference of relative perfusion because the arterial transit time of WM was much longer than that of GM, resulting in a substantial underestimation of the blood flow for WM.^19,20 The maximal SI rather than the averaged SI of brain tumors was adopted as a parameter for relative perfusion to detect the most vascularized portion of the tumor, which may correspond to the highest grade region.¹¹

Histopathologic Evaluations

We examined the vascular attenuation and the cell proliferation as the histopathologic parameters of the tumors.

To evaluate the vascular attenuation of the brain tumors, we immunostained the tissue sections for CD-34 (Novocastra Laboratories, Newcastle, UK), which identifies vascular endothelial cells. A microscopic field of the most intense vascularization in each specimen under a ×20 objective field (area = 0.33 mm²) was photographed as a JPEG file (1600 × 1200 pixels, 16.7 million colors, 8-bit) with a microscope digital color camera (Microscope Digital Camera DP20; Olympus, Tokyo, Japan). For each tumor, the relative vascular attenuation (%Vessel) was defined as the total area of vessel lumina surrounded by the endothelial cell layer as a percentage of the entire tissue area. The vessel lumina were recognized as an area surrounded by the CD-34–positive endothelial cell layer. On the image, the vessel area was distinguished by color by using Photoshop CS2 (Adobe Systems, San Jose, Calif). The tissue area was extracted by excluding the tissue-free space on the image by using Adobe Photoshop CS2. The image was analyzed with Image ALPHA 4.0.3.2 (Scion, Frederick, Md) to measure %Vessel by counting the pixel number of the total vessel lumina and that of the entire tissue.^9,21–23

To assess the proliferative activities of tumor cells, we immunostained glioma tissue sections with MIB1 (Dako, Glostrup, Denmark), a monoclonal antibody against Ki-67 antigen that is expressed throughout the cell cycle except at the G0 stage.^24–26 The MIB1 index was measured in the microscopic area that showed the greatest number of immunopositive nuclei under a ×20 objective as a JPEG file with the microscope digital color camera by using the same conditions as previously described. MIB1 index was calculated as the percentage of the positive nuclei area divided by the whole nuclei area by using free computer software (Gunma-LI, Version 017; Gunma University, Gunma, Japan; available at http://omt.med.gunma-u.ac.jp/~tgaku/GunmaLI).

We thank Saori Uchida (Department of Neurosurgery, Kyushu University) for her helpful technical assistance.