Grading Meningioma

The purpose was to compare capability of fluorine-18 fluorodeoxyglucose (FDG)-PET and thallium-201 (Tl)-SPECT for grading meningioma. This retrospective study was conducted as a case-control study under approval by the institutional review board. In the hospital information system, 67 patients (22 men and 45 women) who had both FDG-PET and Tl-SPECT preoperative examinations were found with histopathologic diagnosis of meningioma. The maximum FDG uptake values of the tumors were measured, and they were standardized to the whole body (SUVmax) and normalized as gray matter ratio (SUVRmax). Mean and maximum Tl uptake ratios (TURmean and TURmax, respectively) of the tumors were measured and normalized as ratios to those of the contralateral normal brain. Receiver-operating characteristic curve analyses of the 4 indexes were conducted for differentiation between lowand high-grade meningiomas, and areas under the curves (AUCs) were compared. Correlation coefficients were calculated between these indexes and Ki-67. Fifty-six meningiomas were classified as grade I (low grade), and 11 were grade II or III (high grade). In all 4 indexes, a significant difference was observed between lowand high-grade meningiomas (P< 0.05). AUCs were 0.817 (SUVmax), 0.781 (SUVRmax), 0.810 (TURmean), and 0.831 (TURmax), and no significant difference was observed among the indexes. Their sensitivity and specificity were 72.7% to 90.9% and 71.4% to 87.5%, respectively. Correlation of the 4 indexes to Ki-67 was statistically significant, but coefficients were relatively low (0.273–0.355). Tl-SPECT, which can be used at hospitals without a cyclotron or an FDG distribution network, has high diagnostic capability of meningioma grades comparable to FDG-PET. hi, MD, PhD, Sus , MD, PhD, hi, MD, PhD positron emission tomography, ROC = receiver-operating characteristic, ROI = region of interest, SPECT = single photon emission computed tomography, SUV = standardized uptake value, SUVmax = the maximum value of SUV, SUVRmax = the maximum SUV ratio, Tl = thallium-201, TURmax = the maximum Tl uptake ratio, TURmean = the mean Tl uptake ratio, WHO = World Health Organization. INTRODUCTION M eningioma is the most common benign intracranial tumor, accounting for about 13% to 26% of all primary intracranial tumors. World Health Organization (WHO) criteria classifies meningioma into grades I, II, and III, which had recurrence in 7%, 35%, and 73%, respectively. Some study reported that the elevated Ki-67 proliferation index has been associated with an increased recurrence rate. Once a meningioma recurs, it is more likely to recur again, resulting in a poor prognosis. Therefore, a noninvasive imaging method that can differentiate the histologic grades and evaluate biological aggressiveness, such as Ki-67, is desirable for presurgical planning of meningioma. Previous studies evaluated intracranial meningioma with fluorine-18 fluorodeoxyglucose (FDG) positron emission tomography (PET), showing correlations between FDG uptake and histopathologic grades or biological aggressiveness of meningioma. A potential role of thallium-201 (Tl) single photon emission computed tomography (SPECT) has also been reported in patients with meningioma. FDG-PET examination is increasing, but its availability is still limited due to requirements of a cyclotron or a local FDG distribution network, as well as a relatively expensive PET system itself. On the other hand, Tl-SPECT examination is much widely available, although spatial resolution is lower than FDG-PET. So far as we could find, no comparative study of the 2 examinations has been conducted, and we do not know if one is better than the other or not. Therefore, the purpose of this study was to compare grading capability of the 2 examinations focusing on meningioma. MATERIALS AND METHODS Patients This study was conducted as a case-control study under approval by the institutional review board. Informed consent etrospective nature of this study. The examined from October 2010 to April o had both FDG-PET and Tl-SPECT www.md-journal.com | 1 examinations before biopsy or embolization, and was histopathologically diagnosed to have intracranial meningioma by surgical resection. Two patients who had meningioma with intraosseous or extracranial extension were not included. A total of 67 patients (22 men and 45 women; range 33–86 years old, mean 60 years old) fulfilled the inclusion criteria. Among them, 61 patients were operated for the first time, and the remaining 6 patients had an operation for recurrence. Histopathologic grades of the meningioma were determined from grade I to grade III by the WHO classification 2007. They were sorted into low-grade (WHO grade I) or highgrade (WHO grades II and III) groups for further analysis. Furthermore, tumor proliferation index was evaluated by MIB-1 immunostaining, which is used for the assessment of Ki-67 expression in tumors. In 2 patients, MIB-1 immunostaining was not conducted, and Ki-67 index values were unavailable. Image Acquisition The mean interval between scans and surgical resection of meningioma was 36 days (range 2–155 days) and 37 days (range 2–155 days) for FDG-PET and Tl-SPECT, respectively. The intervals between FDG-PET and Tl-SPECT were within 7 days. FDG-PET scans were conducted using a PET/CT scanner (Discovery ST Elite; GE Healthcare, Waukesha, WI). Patients fasted for at least 4 hours prior to the scan. After intravenous administration of 4 MBq/kg of FDG, patients rested in a waiting room for 30 minutes. Emission scans of the brain were conducted for 15 minutes. Resolutions were 2.0 2.0 4.25 mm (47 slices). Tl-SPECT was acquired at resting state using a 2-head rotating gamma camera (Infinia; GE Medical Systems, Milwaukee, WI) with an extended low-energy general-purpose collimator. After intravenous administration of 74 MBq of Tl-201, the scan was conducted 15 minutes later. Data were acquired through a 3608 rotation at angle intervals of 68, each for 20 seconds. Total imaging time was 20 minutes. Transverse reconstruction was conducted using ordered subset expectation maximization (subsets 10 and iterations 2), and resolutions were 4.42 4.42 4.42 mm (33–47 slices). MR scans were conducted using 3T MR units (Magnetom Trio or Magnetom Skyra; Siemens, Erlangen, Germany) with a 32-channel head coil. Preoperative scans included 3-dimensional T1-weighted imaging in isotropic 0.7–0.9 mm resolution covering the whole brain before and after administration of a Gadolinium contrast agent (0.1 mmol/kg). Image Analysis Maximum values of SUV (SUVmax) were measured by placing regions of interest (ROIs) of the tumor on FDG-PET images using a workstation (Advantage Windows; GE Healthcare, Waukesha, WI) (Figures 1B and 2B). When the tumor was indistinguishable from the brain parenchyma, ROIs were drawn with careful reference to the MR images (Figures 1A and 2A). SUVmax values of the tumors were normalized by those of the reference areas, and maximum SUV ratios (SUVRmax) were calculated. Five to 6 consecutive reference ROIs (10 mm diameter) were placed in the frontoparietal cortex contralateral to the tumor to measure maximum FDG uptake of the reference Okuchi et al region (Figures 1B and 2B). When a tumor was located near the cerebellum, reference ROIs were placed in the ipsilateral frontoparietal cortex. When a tumor was found on the median line, 2 | www.md-journal.com they were placed at either side of the frontoparietal cortex that was considered less affected by the tumor. On Tl-SPECT images, tumor ROIs were drawn on the slice where the tumor showed maximal diameters. Because tumor border could not be defined clearly, a tumor was encircled including the normal-appearing surroundings, but high Tl uptake areas of the skull were carefully excluded (Figure 1C). A threshold algorithm (Otsu method) was applied to select the tumor (Figures 1D and 2C) using ImageJ software ver. 1.48 (http://imagej.nih.gov/ij/), and mean and maximum Tl uptake values were measured. To normalize these values, a reference ROI was placed on the contralateral side of the tumor on the same slice by flipping the tumor ROI horizontally (Figure 1E). When we could not define a reference ROI on the same slice because the tumor was large or found in the median line, we defined it at a normal brain area on another slice. Mean and maximum values of a tumor were normalized by mean reference values, and the mean and maximum Tl uptake ratios (TURmean and TURmax, respectively) were calculated. Finally, tumor sizes were measured on contrast-enhanced MRI images using a workstation (Centricity PACS; GE Healthcare, Waukesha, WI). All the measurements were conducted independently by 2 radiologists (S.O. and M.Y. both with 7 years of experience). Statistical Analysis Intraclass correlation coefficients (ICCs) were used to examine agreement between 2 evaluators. The measured values were averaged and used for further analysis. The Mann–Whitney test was used to examine differences between low and high grades for SUVmax, SUVRmax, TURmean, TURmax, and size. Age difference was compared between the 2 groups with 2tailed 2-sample t test. The receiver-operating characteristic (ROC) curve analyses were conducted for differentiation of low and high grades. Areas under the curves (AUCs) were calculated and compared among the 4 indexes with the method of DeLong et al. The optimal cutoff values were also determined by the values that maximized the Youden index (sensitivityþ specificity 1). To investigate relationship to Ki-67 proliferation index, Spearman correlation coefficients of the 4 indexes were also calculated and compared. A P value less than 0.05 was considered statistically significant. Statistical analyses were conducted using a commercially available software (MedCalc ver. 13.3; MedCalc Software, Ostend, Belgium). RESULTS Histopathologic diagnosis were grade I in 56 tumors (24 meningothelial, 17 fibrous, 10 transitional, 2 microcystic, 1 angiomatous, 1 meningothelial and angiomatous, and 1 psammomatous meningioma), grade II in 10 tumors (all atypical meningioma), and grade III in 1 tumor (anaplastic meningioma). ICCs between 2 evaluators were 0.94 (95% CI 0.91–0.96), 0.90 (0.85–0.94), 0.83 (0.74–0.89), and 0.85 (0.77–0.90) for SUVmax, SUVRmax, TURmean, and TURmax, respectively. Agreements were almost perfect. In all 4 indexes, the values were significantly higher in high-grade meningioma than lowMedicine Volume 94, Number 6, February 2015 grade meningioma (P< 0.05 in all), and there was no statistically significant difference in age of the patients and tumor sizes between low and high grades (Table 1). Copyright # 2015 Wolters Kluwer Health, Inc. All rights reserved. FIGURE 1. A representative analysis of a meningothelial meningioma (grade I), 70-year-old patient. Contrast-enhanced MRI (A) shows an enhanced extra-axial tumor at the right parietal region. FDG-PET (B) and Tl-SPECT (C–E) images show high uptake in the corresponding area. (B) On the FDG-PET image, ROIs were drawn on the tumor and contralateral gray matter. (C) On the Tl-SPECT image, the area corresponding to the tumor was encircled including the normal-appearing surroundings, but high uptake areas of the skull were excluded. (D) The tumor area way applying a threshold algorithm (Otsu method). (E) The tumor ROI was flipped horizontally, which was placed as the reference ROI at the contralateral normal brain of the same slice. By using these analysis procedures, SUVmax, SUVRmax, TURmean, and TURmax were measured as 5.75, 0.54, 6.84, and 9.30, respectively. FDG1⁄4fluorine-18 fluorodeoxyglucose, PET1⁄4positron emission mis m S Medicine Volume 94, Number 6, February 2015 Grading Meningioma: Thallium-SPECT Versus FDG-PET In ROC curve analysis, AUCs were 0.817, 0.781, 0.810, and 0.831 for SUVmax, SUVRmax, TURmean, and TURmax, respectively. No statistically significant difference was tomography, ROI1⁄4 region of interest, SPECT1⁄4single photon e SUVmax1⁄4 the maximum value of SUV, SUVRmax1⁄4 the maximu ratio, TURmean1⁄4 the mean Tl uptake ratio. observed among them (Figure 3). The sensitivity and specificity were 72.7% to 90.9% and 71.4% to 87.5%, respectively, when optimal cutoff values were applied (Table 2). TABLE 1. Comparisons of Low-Grade Meningioma (WHO Grade Low-Grade Meningioma (n1⁄4 56) Age 60.8 (57.9–63.6) Tumor size 31.5 (27.2–36.8) SUVmax 5.4 (4.9–5.5) SUVRmax 0.51 (0.47–0.56) TURmean 5.3 (4.5–6.4) TURmax 7.5 (5.9–8.8) The numbers in the parentheses are 95% confidence intervals. P values w value, SUVmax1⁄4 the maximum value of SUV, SUVRmax1⁄4 the maximum mean Tl uptake ratio. Copyright # 2015 Wolters Kluwer Health, Inc. All rights reserved. Correlation to Ki-67 was statistically significant in all 4 indexes, but the coefficients were 0.273, 0.355, 0.322, and 0.336, for SUVmax, SUVRmax, TURmean, and sion computed tomography, SUV1⁄4 standardized uptake value, UV ratio, Tl1⁄4 thallium-201, TURmax1⁄4 the maximum Tl uptake TURmax, respectively, which were relatively low (Figure 4). No significant difference was observed among the coefficients. I) and High-Grade Meningioma (WHO Grade II and III) High-Grade Meningioma (n1⁄4 11) P value 56.2 (46.1–66.2) 0.23 43.0 (26.6–48.3) 0.25 8.8 (5.5–9.6) 0.0010 0.85 (0.49–1.05) 0.0034 8.5 (6.7–9.9) 0.0012 12.9 (10.0–14.6) 0.0006 ith an asterisk are statistically significant. SUV1⁄4 standardized uptake SUV ratio, TURmax1⁄4 the maximum Tl uptake ratio, TURmean1⁄4 the www.md-journal.com | 3 FIGURE 2. Contrast-enhanced MRI (A), FDG-PET (B), and Tl-SPECT (C) images of a 53-year-old patient with WHO grade II meningioma, atypical type. The FDG-PET image (B) and Tl-SPECT image (C) show high uptake in the tumor at the right frontal region. (B)The circles show examples of drawn ROI. (C)The tumor is selected with a threshold algorithm. The SUVmax, SUVRmax, TURmean, and TURmax were cose tan the Okuchi et al Medicine Volume 94, Number 6, February 2015 DISCUSSION In this study, capability of Tl-SPECT and FDG-PET were compared for grading meningioma using 2 indexes for each method. In all indexes, significant difference was observed between low and high grades, and correlation to Ki-67 was 8.69, 0.89, 11.92, and 16.57. FDG1⁄4fluorine-18 fluorodeoxyglu SPECT1⁄4single photon emission computed tomography, SUV1⁄4 s SUVRmax1⁄4 the maximum SUV ratio, Tl1⁄4 thallium-201, TURmax1⁄4 statistically significant. AUC was the highest in TURmax, although no significant difference was observed among the indexes.