Cu-Diacetyl-Bis (N-Methylthiosemicarbazone) PET in Human Gliomas: Comparative Study with [F]Fluorodeoxyglucose and L-Methyl-[C]Methionine PET

BACKGROUND AND PURPOSE: Cu-diacetyl-bis(N-methylthiosemicarbazone) was developed as a hypoxic radiotracer in PET. We compared imaging features amongMR imaging and Cu-diacetyl-bis(N-methylthiosemicarbazone)-PET, FDG-PET, and L-methyl-[C]methionine)-PET in gliomas. MATERIALSANDMETHODS: Weenrolled 23 patientswhounderwent Cu-diacetyl-bis(N-methylthiosemicarbazone)-PET and FDG-PET and 19 (82.6%) who underwent L-methyl-[C]methionine)–PET, with all 23 patients undergoing surgery and their diagnosis being then confirmed by histologic examination as a glioma. Semiquantitative and volumetric analysis were used for the comparison. RESULTS: There were 10 newly diagnosed glioblastoma multiforme and 13 nonglioblastoma multiforme (grades II and III), including 4 recurrences without any adjuvant treatment. The maximum standardized uptake value and tumor/background ratios of Cu-diacetylbis(N-methylthiosemicarbazone), as well as L-methyl-[C]methionine, were significantly higher in glioblastoma multiforme than in nonglioblastomamultiforme (P .03 and P .03, respectively); no significant differenceswere observed on FDG. At a tumor/background ratio cutoff threshold of 1.9, Cu-diacetyl-bis(N-methylthiosemicarbazone) was most predictive of glioblastoma multiforme, with 90.0% sensitivity and 76.9% specificity. The positive and negative predictive values, respectively, for glioblastoma multiforme were 75.0% and 85.7% on Cu-diacetyl-bis(N-methylthiosemicarbazone), 83.3% and 60.0% on L-methyl-[C]methionine, and 72.7% and 75.0% on MR imaging. In glioblastoma multiforme, volumetric analysis demonstrated that Cu-diacetyl-bis(N-methylthiosemicarbazone) uptake had significant correlations with FDG (r 0.68, P .03) and L-methyl-[C]methionine (r 0.87, P .03). However, the Cu-diacetyl-bis(Nmethylthiosemicarbazone)–active region was heterogeneously distributed in 50.0% (5/10) of FDG-active and 0% (0/6) of L-methyl[C]methionine)–active regions. CONCLUSIONS: Cu-diacetyl-bis(N-methylthiosemicarbazone) may be a practical radiotracer in the prediction of glioblastoma multiforme. In addition to FDG-PET, L-methyl-[C]methionine)–PET, andMR imaging, Cu-diacetyl-bis(N-methylthiosemicarbazone)-PETmay provide intratumoral hypoxic information useful in establishing targeted therapeutic strategies for patients with glioblastomamultiforme. ABBREVIATIONS: Cu-ATSM Cu-diacetyl-bis(N-methylthiosemicarbazone); GBM glioblastoma multiforme; MET L-methyl-[C]methionine; non-GBM gliomas World Health Organization grade II and III gliomas; SUVmax maximum standardized uptake value; T/B ratio tumor/background ratio Gliomas have heterogeneously infiltrative and proliferative features, among which glioblastoma multiforme (GBM) is the most common and has the worst prognosis in adults. From a histopathologic standpoint, microvascular proliferation and/or necrosis is essential for the diagnosis of GBM. However, random tissue sampling may not always lead to an accurate diagnosis because of tissue heterogeneity. Therefore, other diagnostic modalities to predict highly malignant regions, such as PET imaging, can provide complementary diagnostic and therapeutic information and guide selective target tissue sampling or resection. Malignant tumor cells display an increased flux of glucose metabolism by increased expression of glucose transporters and hexokinase, as well as an increased rate of amino acid uptake and metabolism. This increased transport and high metabolism ocReceived April 3, 2013; accepted after revision June 2. From the Departments of Neurosurgery (K.T., S.N., M.O., J.S., H.M., N.K.) and Radiology (U.T., T.I.), Graduate School of Medicine, Yokohama City University, Yokohama, Japan; and Division of Nuclear Medicine (R.M., K.K.), Department of Radiology, National Center for Global Health and Medicine, Tokyo, Japan. This work was supported by Japan Advanced Molecular Imaging Program (J-AMP) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan, and was partly funded by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Sciences (No. 24791515). Please send correspondence to Nobutaka Kawahara, MD, PhD, Department of Neurosurgery, Graduate School of Medicine, Yokohama City University, 3–9 Fukuura, Kanazawa-ku, Yokohama, Kanagawa 236-0004, Japan; e-mail: [email protected] Indicates open access to non-subscribers at www.ajnr.org Indicates article with supplemental on-line tables. Indicates article with supplemental on-line figure. http://dx.doi.org/10.3174/ajnr.A3679 278 Tateishi Feb 2014 www.ajnr.org cur commonly in GBM and can be detected by widely used techniques, such as FDG-PET and L-methyl-[C]methionine (MET)-PET. However, the predictive value of these PET imaging techniques has not been adequate for a diagnosis of GBM because uptake of these tracers is not specific in GBM. On the other hand, tissue hypoxia and necrosis are cardinal features of GBM that are often associated with resistance to radiation therapy and chemotherapy. Thus, intratumoral hypoxic information may be useful for an accurate diagnosis and establishment of effective therapeutic strategies for gliomas. Some clinical investigations by use of [F]fluoromisonidazole-PET have been recently undertaken to detect tissue hypoxia noninvasively in gliomas. On the other hand, we previously reported the clinical usefulness of Cu-diacetyl-bis(N-methylthiosemicarbazone) (Cu-ATSM)-PET imaging for gliomas. Our preliminary study revealed a relationship between CuATSM uptake values and hypoxia-inducible factor-1 expression, which is increased under hypoxia, suggesting that CuATSM-PET is a practical hypoxic imaging technique in gliomas. However, whether Cu-ATSM uptake is specific in GBM was not determined, and the correlation of Cu-ATSM findings with FDG and MET remained unclear in GBM. In addition, to assess whether Cu-ATSM uptake is dependent on BBB breakdown and to evaluate an additional value over MR imaging findings, we volumetrically and qualitatively compared Cu-ATSM-PET imaging with T1-weighted MR imaging with Gd-DTPA. MATERIALS AND METHODS Patients This study was approved by the local ethics committee (Institutional Review Board no. B1001111026) after written informed consent was obtained from all patients. Between December 2010 and December 2012, we prospectively performed Cu-ATSMPET in 68 patients with malignant brain tumors. Among them, 23 consecutive patients with pathologically confirmed gliomas (10 men and 13 women; age range, 19 – 81 years; mean age, 54.2 17.5 years) who received FDG-PET and/or MET-PET with CuATSM-PET were retrospectively analyzed. FDG-PET and METPET imaging were performed in 23 patients (100%) and 19 patients (82.6%), respectively. Of the 23 patients, 19 (82.6%) were newly diagnosed, and the remaining 4 patients (17.4%), who previously underwent biopsy but had not received any radiation therapy or chemotherapy, were diagnosed as having tumor recurrence. Histologic diagnosis and tumor grade were classified according to the following 2007 World Health Organization criteria: 8 (34.8%) grade II (3 diffuse astrocytomas, 3 oligoastrocytomas, and 2 oligodendrogliomas); 5 (21.7%) grade III (1 anaplastic oligoastrocytoma and 4 anaplastic oligodendrogliomas); and 10 (43.5%) grade IV (GBM). All 10 patients with GBM were newly diagnosed. Thirteen grade II and III gliomas (56.5%) were classified as non-GBM gliomas. The oligodendroglial component was found in 10 (76.9%) of 13 non-GBM gliomas. Intervals from the MR imaging investigation to Cu-ATSMPET, FDG-PET, and MET-PET were 5.4 4.1, 4.3 3.2, and 5.4 3.4 days, respectively (mean SD). All patients underwent surgery the day after a repeated MR imaging study for neuronavigation, and their diagnosis was confirmed on histologic examination. On-Line Table 1 summarizes the patient characteristics. PET and MR Image Acquisition Preparation of Cu-ATSM, FDG, and MET has been described in previous reports. To acquire Cu-ATSM-PET and FDGPET/CT images, a whole-body PET/CT scanner (Aquiduo PCA7000B; Toshiba, Tokyo, Japan) with a 16-row detector in the CT component was used at the Yokohama City University Hospital (Yokohama, Japan). MET-PET imaging was performed with PET/CT scans (Biograph 16; Siemens, Erlangen, Germany) at the National Center for Global Health and Medicine (Tokyo, Japan). An image quality phantom (NU 2–2001; National Electrical Manufacturers Association) was used for cross-calibration because such phantoms are widely used and allow estimation of optimal acquisition times. For Cu-ATSM-PET/CT and FDG-PET/CT, the following conditions were used for acquisition of low-dose CT data: 120 kVp, an auto-exposure control system, a beam pitch of 0.875 or 1, and a 1.5or 2-mm 16-row mode. No iodinated contrast material was administered. After intravenous injection of 740 MBq of Cu-ATSM, the patients were placed in a supine “arm-up” position. Dynamic data acquisition was carried out for 30 – 40 minutes, and PET/CT images were reconstructed from the data. For studies of FDG, the patients received an intravenous injection of 370 MBq of FDG after at least 6 hours of fasting, followed by an uptake phase of approximately 60 minutes. For MET-PET/CT, the following conditions were used for acquisition of low-dose CT data: 120 kVp, an auto-exposure control system, a beam pitch of 0.875, and a 3-mm 16-row mode. After 6 hours of fasting, 370 MBq of MET was intravenously injected, followed by data acquisition at 20 minutes after the injection. The following acquisition settings were used for Cu-ATSM-PET/CT and FDG-PET/CT: 3D data acquisition mode; 180 seconds/bed; field of view, 500 mm; 4 iterations; 14 subsets; matrix size, 128 128; 8-mm Gaussian filter, full width at half maximum; and reconstruction, ordered subset expectation maximization. For MET-PET/CT, the following acquisition settings were used: 3D data acquisition mode; 180 seconds/bed; field of view, 300 mm; 4 iterations; 14 subsets; matrix size, 256 256; 4-mm Gaussian filter, full width at half maximum; and reconstruction, ordered subset expectation maximization. The estimated internal absorbed doses of CuATSM, FDG, and MET were approximately 10, 2.5, and 1.9 mSv, respectively. MR imaging was performed on a 1.5T system (Magnetom Symphony; Siemens). 3D T1-weighted MR imaging with a MPRAGE sequence was used with the following parameters to acquire axial T1-weighted images after administration of 0.2 mL/kg of Gd-DTPA: field of view, 250 250 mm; matrix size, 512 512; TR, 1960 ms; TE, 3.9 ms; TI, 1100 ms; and flip angle, 15°. In total, 120 contiguous 2-mm images were obtained from each patient. Image Interpretation Four board-certified nuclear medicine specialists who were unaware of the clinical information assessed the PET images semiquantitatively and volumetrically in consensus (Cu-ATSM and AJNR Am J Neuroradiol 35:278–84 Feb 2014 www.ajnr.org 279 FDG, U.T. and T.I.; MET, R.M. and T.I.). MR imaging findings were also assessed by board-certified radiologists (U.T. and T.I.), who interpreted the tumors as either GBM or non-GBM gliomas. A volume of interest was outlined within areas of increased tracer uptake and was measured on each section. In extensively heterogeneous lesions, regions of interest covered all components. For semiquantitative interpretations, the standardized uptake value was determined by a standard formula. The tumor/background ratio (T/B ratio) of Cu-ATSM and MET was calculated relative to the uptake in the contralateral frontal cortex. The FDG T/B ratio was calculated relative to the uptake in the contralateral white matter. The uptake values of the CuATSM, FDG, and MET tracers were determined by assessment of the maximum standardized uptake value (SUVmax) values and T/B ratios. Dr. View version R 2.5 for LINUX (Infocom, Tokyo, Japan) software was used to merge the PET images with the MR images, and each PET and MR image was volumetrically compared. To evaluate volumetric analysis, we extracted the uptake regions of the Cu-ATSM images on the basis of the optimal T/B ratio thresholds of 1.8, a cutoff value for predicting hypoxiainducible factor-1 expression in our previous study. The uptake regions of the FDG and MET images were extracted on the basis of the T/B ratio thresholds of 1.5 and 1.3, respectively, in accordance with previous reports. These uptake regions were rated as metabolically active volumes. For GBM, we extracted the tumor volume by measuring a completely covered contrast-enhanced region with necrotic and cystic components on MR imaging. The contrast-enhanced volume was also separately extracted by measuring a contrast-enhanced region without any necrotic and cystic components. Metabolically active regions shown by each PET tracer were overlaid on the MR images for qualitatively comparing metabolically active regions among the 3 tracers. Tumors with Cu-ATSM–active regions that demonstrated 50% volumetric overlap with the active regions of FDG and MET were rated as heterogeneous with respect to intratumoral oxygenation. Correlations among Cu-ATSM, FDG, and MET were also volumetrically analyzed. On the basis of the optimal cutoff value for prediction of GBM (T/B ratio, 1.9), which was defined by receiver operating characteristic analysis, tumors having Cu-ATSM T/B ratios 1.9 were rated as GBM. The optimal cutoff threshold of the MET T/B ratio was set by receiver operating characteristic analysis, and tumors having MET T/B ratios 3.0 were rated as GBM. To assess the clinical value of Cu-ATSM-PET findings relative to those of MET-PET and MR imaging findings for prediction of GBMs, we evaluated the positive and negative predictive values independently. Statistical Analysis All parameters were expressed as means SDs. Two-way repeated measures ANOVA was used to compare the mean uptake values of each tracer. To determine the optimal radiotracer for prediction of GBM by semiquantitative analysis, we performed receiver operating characteristic analysis. To evaluate volumetric correlations of Cu-ATSM with FDG and MET, we used linear regression analysis. The Wilcoxon signed rank test was used to compare the mean tumor volume, contrast-enhanced volume, and metabolically active volumes determined by the 3 PET tracers. The Fisher exact probability test was used to compare the Cu-ATSM-PET with contrast-enhanced MR imaging and MET-PET imaging features. The level of statistical significance was set at P .05. JMP 10 statistical software (SAS Institute, Cary, North Carolina) was used for statistical analyses. RESULTS Semiquantitative Analysis of PET Studies According to Tumor Classification A summary of the uptake values for each PET tracer is presented in On-Line Table 2. Representative images are shown in Fig 1 and in the On-Line Figure. The average Cu-ATSM SUVmax values in GBM and non-GBM gliomas were 1.68 0.94 and 0.98 0.52, respectively. The Cu-ATSM SUVmax was significantly higher for GBM than for non-GBM gliomas (P .03; Fig 2A). A significant difference in SUVmax was also detected between GBM and nonGBM gliomas for MET (5.23 1.11 and 3.25 1.66, respectively; P .02; Fig 2A) but not for FDG (7.31 3.22 and 5.72 2.25, respectively; P .18; Fig 2A). The mean Cu-ATSM T/B ratios in GBM and non-GBM gliomas were 3.10 2.37 and 1.47 0.57, respectively, which were also significantly different (P .03; Fig 2B). Receiver operating characteristic analysis indicated that a Cu-ATSM T/B ratio cutoff threshold of 1.9 was most predictive of GBM, with 90.0% sensitivity and 76.9% specificity (area under the curve, 0.88). Similar to the SUVmax results, there was a significant difference in the T/B ratio between GBM and non-GBM gliomas for MET (3.53 0.70 and 2.27 0.97, respectively; P .01; Fig 2B) but not for FDG (2.71 1.56 and 2.44 1.28, respectively; P .65; Fig 2B). Receiver operating characteristic analysis showed an optimal T/B ratio of 3.0 for MET (sensitivity, 83.3%; specificity, 76.9%; area under the curve, 0. 83, respectively), which was slightly less than