J Neurooncol (2015) 121:141–150 DOI 10.1007/s11060-014-1614-z
CLINICAL STUDY
Evaluation of the microenvironmental heterogeneity in high-grade gliomas with IDH1/2 gene mutation using histogram analysis of diffusion-weighted imaging and dynamic-susceptibility contrast perfusion imaging Seunghyun Lee • Seung Hong Choi • Inseon Ryoo • Tae Jin Yoon Tae Min Kim • Se-Hoon Lee • Chul-Kee Park • Ji-Hoon Kim • Chul-Ho Sohn • Sung-Hye Park • Il Han Kim
•
Received: 11 February 2014 / Accepted: 30 August 2014 / Published online: 10 September 2014 Ó Springer Science+Business Media New York 2014
Abstract The purpose of our study was to explore the difference between isocitrate dehydrogenase (IDH)-1/2 gene mutation-positive and -negative high-grade gliomas (HGGs) using histogram analysis of apparent diffusion coefficient (ADC) and normalized cerebral blood volume (nCBV) maps. We enrolled 52 patients with histopathologically confirmed HGGs with IDH1/2P (n = 16) or IDH1/2N (n = 36). Histogram parameters of ADC and nCBV maps were correlated with gene mutations by using the unpaired student’s t test and multivariable stepwise logistic regression analysis. The mean ADC value was higher in the IDH1P group than IDH1N (1,282.8 vs. 1,159.6 mm2/s, P = .0113). In terms of the cumulative ADC histograms, the 10th and 50th percentile values were also higher in the IDH1P than IDH1N (P = .0104 and .0183, respectively). We observed a higher 90th percentile value (3.121 vs. 2.397, P = .0208) and a steeper slope between the 10th (C10) and 90th (C90) of cumulative Electronic supplementary material The online version of this article (doi:10.1007/s11060-014-1614-z) contains supplementary material, which is available to authorized users. S. Lee S. H. Choi (&) I. Ryoo T. J. Yoon J.-H. Kim C.-H. Sohn Department of Radiology, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea e-mail: [email protected] S. H. Choi Center for Nanoparticle Research, Institute for Basic Science, and School of Chemical and Biological Engineering, Seoul National University, Seoul 151-742, Korea T. M. Kim S.-H. Lee Department of Internal Medicine, Cancer Research Institute, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea
nCBV histograms (0.03386 vs. 0.02425/%, P = .0067) in the IDH1N group. Multivariate analysis showed that the mean ADC mean value (P = .0048), the C90 value (P = .0113), and the slope between C10 and C90 (P = .0049) were the significant variables in the differentiation of IDH1P from IDH1N. In conclusion, histogram analysis of ADC and nCBV maps based on entire tumor volume can be a useful tool for distinguishing IDH1P and IDH1N, and it predicts that IDHP tumors have a more heterogeneous microenvironment than IDHN ones. Keywords Isocitrate dehydrogenase IDH gene mutation High-grade gliomas ADC nCBV DWI DSC-PWI
Introduction Gliomas are the most common primary neoplasm of the brain, varying histopathologically from low grade to high
C.-K. Park Department of Neurosurgery, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea S.-H. Park Department of Pathology, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea I. H. Kim Department of Radiation Oncology, Cancer Research Institute, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea
123
142
grade [1]. High-grade gliomas (HGGs) include World Health Organization (WHO) grade III and IV brain tumors such as anaplastic astrocytoma (AA), and primary or secondary glioblastoma multiforme (GBM). Primary GBMs arise de novo in older patients, while secondary GBMs develop from WHO grade II or III gliomas, have a longer duration of gliomagenesis, and arise from an early event such as isocitrate dehydrogenase (IDH) gene mutation [2]. According to a recent genomic analysis of HGGs, IDH gene mutations occur in *60–90 % of diffuse and anaplastic gliomas and secondary GBMs [2]. The recent identification of IDH gene mutation has directed attention to the role of abnormal metabolism in the pathogenesis and progression of these primary brain tumors [3, 4]. In fact, GBM patients with this mutation have significantly better survival, and IDH gene mutation testing is relevant for clinical patient management and for stratification in clinical trials [5–7]. However, the current histopathologic grading method by stereotactic biopsy has major limitations associated with its inherent sampling error or its inability to predict the prognosis of the disease before surgical resection [1]. Currently, conventional magnetic resonance (MR) imaging along with advanced imaging techniques, such as diffusion-weighted imaging (DWI), dynamic-susceptibility contrast perfusionweighted imaging (DSC-PWI), and MR spectroscopy (MRS), helps in grading of gliomas and may complement the histopathologic grade [8]. In particular, DWI and DSC-PWI MRI techniques provide important in vivo physiologic information about brain tumors and have been of increasing utility including tumor grading [9–14]. To our knowledge, however, there have been no previous reports concerning the use of volume-based histogram analysis of DWI and DSC-PWI to differentiate IDHP from IDHN gliomas. Our hypotheses were that (a) IDHP tumors have a more heterogeneous microenvironment, because of their stepwise tumorigenesis, and (b) DWI and DSC-PWI can reflect the microenvironmental heterogeneity of the gliomas. This study aimed to explore the difference between IDH1/2P and IDH1/2N HGGs using histogram analysis of apparent diffusion coefficient (ADC) maps and normalized cerebral blood volume (nCBV) based on the entire tumor volume and to evaluate the characteristics of IDH gene mutants.
Materials and methods Our institutional review board approved this study and informed consent was waived. Patient selection One hundred thirty-four patients who underwent surgical resection or stereotactic biopsy at our institution between
123
J Neurooncol (2015) 121:141–150
May 2005 and December 2012 were selected from our radiology report database. The inclusion criteria were as follows: (a) had a histopathological diagnosis of AA or GBM based on the WHO criteria, with or without oligodendroglial components; (b) had undergone surgery or stereotactic biopsy; and (c) had baseline three-tesla (3-T) MR imaging performed with DWI (b = 1,000 s/mm2) and DSC-PWI prior to surgery or chemoradiotherapy. We excluded 82 patients due to the following reasons: (a) poor quality of the MR images; (b) only 1.5-T MR images; and (c) had undergone baseline MR imaging after surgery. As a result, a total of 52 patients with AA or GBM (32 men and 20 women; age range: 22–72 years; mean age: 50 years) were analyzed. Image acquisition For each patient, the first or follow-up MR imaging before surgical biopsy was performed with one of the following 3-T MR imaging scanners: a Signa Excite (GE Medical Systems, Milwaukee, WI, USA) with an 8-channel head coil (n = 6) and a Verio (Siemens Medical Solutions, Erlangen, Germany) with a 32-channel head coil (n = 46). The imaging sequences of the brain included axial fast/ turbo spin-echo T2-weighted images (T2WI), DWI and DSC-PWI with gadobutrol (Gadovist, Bayer Schering Pharma, Berlin, Germany). The MR imaging parameters were as follows: 4,500–5,160/91–106.3 ms/90–130°/ 448–640 9 220 for fast spin-echo T2WI. The other parameters for the three images were as follows: section thickness, 5 mm with a 1 mm gap; and field of view (FOV), 240 9 240 mm. DWI was performed with a single-shot spin-echo echoplanar imaging sequence in the axial plane before the injection of contrast material with a TR/TE of 6,900–10,000/55–70 ms at b = 0 and 1,000 s/mm2, 35–38 sections, a 3-mm section thickness, a 1-mm intersection gap, an FOV of 240 9 240 mm, a matrix of 160 9 160, three signal averages, and a voxel resolution of 1.5 9 1.5 9 3 mm. DWI data were acquired in three orthogonal directions. Using these data, the averaged ADC maps in the three orthogonal directions were calculated on a voxel-by-voxel basis with the software incorporated into the MR imaging unit. DSC-PWI was performed with a single-shot gradientecho echo-planar imaging sequence during the intravenous injection of the contrast agent. The imaging parameters of DSC-PWI were as follows: TR/TE, 1,500/ 30–40 ms; FA, 35–90°; FOV, 240 9 240 mm; 15–20 sections; matrix, 128 9 128; section thickness, 5 mm; intersection gap, 1 mm; and voxel resolution of 1.86 9 1.86 9 5 mm. For each section, 60 images were obtained at intervals equal to the repetition time. After
J Neurooncol (2015) 121:141–150
four to five time points, a bolus of gadobutrol at a dose of 0.1 mmol/kg of body weight and at a rate of 4 mL/sec was injected with an MR-compatible power injector (Spectris; Medrad, Pittsburgh, PA, USA). After injecting a bolus of the contrast material, a 30 mL bolus of saline was administered at the same injection rate.
Quantitative image analysis The MR data for the ADC maps and perfusion raw data were digitally transferred from the picture archiving and communication system workstation to a personal computer and processed with a dedicated software package (nordicICE; NordicImagingLab, Bergen, Norway). One author (S.H.C., 9 years of experience with neuroradiology imaging) manually drew the tumor borders in each section of the co-registered T2WI; however, the areas of necrosis, cysts, and nontumor macrovessels were drawn in each section of the T2WI and copied to the ADC map and relative CBV (rCBV) map after semiautomatic coregistration. The relative CBV (rCBV) was acquired using an established tracer kinetic model applied to the first-pass data [15, 16]. The dynamic curves were mathematically corrected to reduce contrast-agent leakage effects [17]. To minimize variances in the rCBV value in an individual patient, the pixel-based rCBV maps were normalized by dividing every rCBV value in a specific section by the rCBV value in the unaffected white matter as defined by a neuroradiologist (S.H.C.) [18]. Coregistrations between the T2WI and the ADC maps or the nCBV map were performed based on geometric information stored in the respective data sets by using a dedicated software package (nordicICE). The ADC and nCBV maps were displayed as color overlays on the T2WI (Fig. 1). The data acquired from each section were summated to derive the voxel-by-voxel ADCs and nCBVs for the entire tumor extent of the image by using nordicICE. The ADC and nCBV histograms were plotted with ADC and nCBV on the respective x-axis, with a bin size of 3 9 10-5 mm2/s and 0.1, respectively, whereas the y-axis was expressed as a percentage of the total lesion volume by dividing the frequency in each bin by the total number of analyzed voxels. For further quantitative analysis, the cumulative number of observations in all the bins up to the specified bin was mapped on the y-axis as a percentage in the cumulative histograms. The 10th, 50th and 90th percentile points for ADC and nCBV (A10, A50 and A90, respectively, for ADC; C10, C50 and C90, respectively, for nCBV) were derived (the Xth percentile point is the point at which X% of the voxel values that form the histogram are found to the left of the histogram) [13, 19]. In addition, the slopes between the 10th and 90th percentiles of cumulative ADC and nCBV histograms were also
143
calculated from the following two equations, respectively: (A90–A10)/0.8 and (C90–C10)/0.8. Statistical analysis All statistical analyses were performed with MedCalc software (version 12.1.0 for Microsoft Windows 2000/XP/ Vista/7; MedCalc Software, Mariakerke, Belgium). The results with a P value of less than .05 were considered statistically significant. The clinical characteristics were compared between the IDHP and IDHN groups by using Fisher’s exact test for categorical variables and the unpaired Student t test for noncategorical data. The unpaired Student t test was used to compare the histogram parameters of IDHP and IDHN. The area under the receiver operating characteristic curve (AUC) was used to determine the best cutoff values for the histogram parameters that proved to be substantial predictors in differentiating IDHP from IDHN by using the method of DeLong et al. [20]. Subsequently, a multivariable stepwise logistic regression analysis was used to determine the significant parameters between IDHP and IDHN. Variables with a P value of \.05 according to the univariate analysis were used as input variables for multivariable stepwise logistic regression analysis, with iterative entry of variables on the basis of test results (P values of \.05). The removal of variables was based on likelihood ratio statistics with a probability of 0.10.
Results Differences of clinical characteristics between IDH1/2P and IDH1/2N groups We determined whether any of the clinical characteristics, including age, sex, histology, and other gene mutations, were predictors of IDH1/2P (n = 16) or IDH1/2N (n = 36) before surgery or CCRT. The 52 patients enrolled in the present study included 16 IDH1Pand no IDH2P patients. The IDH1P group had a significantly higher proportion of AA than IDH1N (P = .007). Only patient age was significantly lower in the IDH1P group than in the IDH1N group (Table 1). Differences of histogram parameters derived from ADC maps between the IDH1/2P and IDH1/2N groups The mean ADC value was higher in the IDH1P group than in the IDH1N group (mean ± SD, 1,282.8 ± 184.8 9 10-6 vs. 1,159.6 ± 141.9 9 10-6 mm2/s, P = .0113). In terms of the cumulative histograms, the A10 (788.2 ± 158.8 9 10-6 vs. 699.9 ± 81.4 9 10-6 mm2/s) and A50 (1,229.6 ± 200.2 9 10-6 vs. 1,096.3 ± 173.5 9 10-6 mm2/ s) for ADC values also showed a significant difference
123
144
between the IDH1P and IDH1N groups (P = .0104 and P = .0183, respectively). However, we did not find a significant difference between groups in terms of A90 and slope between A10 and A90 (P [ .05). The histogram parameters from ADC maps are summarized in Table 2. Figure 2a, b shows ADC (b = 1,000 s/mm2) and cumulative ADC histogram for the IDH1P and IDH1N groups. ADC histograms of IDH1N showed a higher relative frequency at the low ADC values compared with IDH1P. The cumulative histogram for the IDH1P and IDH1N groups showed a left shift of the cumulative histogram of the IDH1N group. In terms of each tumor grade, there was no significant difference of ADC value between IDH1P and IDH1N groups in grade III (P [ .05). With regards to the cumulative histograms of grade IV, the A10 (805.3 ± 104.9 9 10-6 vs. 700.7 ± 82.7 9 10-6 mm2/s), A90 (2,120.0 ± 498.2 9 10-6 vs. 1,740.1 ± 268.5 9 10-6 mm2/s), and slope between A10 and A90 (1,643.4 ± 683.3 9 10-6 vs. 1,299.3 ± 317.0 9 10-6 mm2/s) for ADC values also showed a significant difference between the IDH1P and IDH1N groups (P = .007, P = .008, and P = .049, respectively). The histogram parameters from ADC maps of grade III and IV are summarized in Supplementary Table 2 and 3. Differences of histogram parameters derived from nCBV maps between the IDH1/2P and IDH1/2N groups With regard to nCBV values, the IDH1N tumors had a higher C90 value (3.121 ± 0.9,102 vs. 2.397 ± 1.212, P = .0208) and a steeper slope of cumulative nCBV histograms (3.386 ± 1.066 vs. 2.425 ± 1.271, P = .0067) than IDH1P tumors. However, other parameters, including C10, C50 and mean nCBV, were not significantly different between groups (P C .05). The histogram parameters from nCBV maps are also summarized in Table 2. Figure 2c, d shows the nCBV histogram and cumulative histogram for the IDH1P and IDH1N groups. In terms of each tumor grade, there was no significant difference between IDH1P and IDH1N groups in grade III (P [ .05). With regard to nCBV values, there was no significant difference between the IDH1P and IDH1N groups in grade III (P [ .05) and grade IV (P [ .05). The histogram parameters from nCBV maps of grade III and IV are summarized in Supplementary Table 2 and 3. The representative IDH1P and IDH1N GBMs are demonstrated in Fig. 1b, c, respectively.
J Neurooncol (2015) 121:141–150 Fig. 1 Flowchart of histogram analysis, and representative IDH1P c and IDH1N histograms. a Flowchart of histogram analysis. The upper, middle, and bottom rows show the defining region of interest (ROI) and the acquisition of ADC and nCBV maps. Based on T2WI (upper middle), including all the T2 high-signal portions of the lesion, coregistrations between the T2WI (upper middle) and the ADC maps (upper right) or the nCBV maps (upper left) were performed based on geometric information stored in the respective data sets by using a dedicated software package (nordicICE). After acquiring co-registered images on a voxel-by-voxel basis, ROIs were drawn (bottom). Finally, the overall values for each tumor were obtained by summing the histogram parameter values from every plane. b Representative IDH1P histograms in a 49-year-old male with GBM. This figure was acquired from a 49-year-old IDH1P man with GBM. Axial T2weighted MR image (upper left) obtained before surgical resection or CCRT shows a newly developed T2 high-signal-intensity lesion in the right temporal lobe. The ADC map (upper middle) and rCBV map (upper right) with regions of interest corresponding to axial T2weighted MR image (upper left) show the heterogeneous ADC values and vascularities of the lesion. The ADC histogram (middle left) and cumulative histograms (middle right) show a normal distribution of the relative frequency of ADC values and show a lower slope and homogenous relative frequency compared with (c). The nCBV histograms (lower left) and cumulative histograms (lower right) had a lower relative frequency at high nCBV values and lower slope compared with (c). c Representative IDH1N histograms in a 29-yearold female with GBM. This figure was acquired from a 29-year-old IDH1P woman with GBM. Axial T2-weighted MR image (upper left) obtained before surgical resection or CCRT shows a newly developed T2 high-signal-intensity lesion in the left thalamus. The ADC map (upper middle) and rCBV map (upper right) with regions of interest corresponding to axial T2-weighted MR image (upper left) show the heterogeneous ADC values and vascularities of the lesion. The ADC histogram (middle left) and cumulative histograms (middle right) show a higher relative frequency at the low ADC values, higher slope and heterogenous relative frequency compared with (b). The nCBV histograms (lower left) and cumulative histograms (lower right) had a higher relative frequency at high nCBV values, higher slope and at left shift compared with (b)
parameters for the differentiation of IDH1P (n = 16) from IDH1N (n = 36). In the comparison of ROC curves, no AUC value could distinguish IDH1P from IDH1N. The ADC mean value, A10, A50, C90 and the slope between C10 and C90, all of which showed significant differences between the IDH1P and IDH1N groups, were used as input variables for a multivariable stepwise logistic regression analysis. In the multivariate analysis, there was no independently differentiating variable in histogram parameters. However, we identified ADC mean value (P = .0048), C90 (P = .0113), and the slope between C10 and C90 (P = .0049) as significant variables in the differentiation of IDH1P from IDH1N, whereas A10 and A50 were excluded from the logistic regression equation.
Diagnostic performance and the best histogram predictors for the differentiation of IDH1/2P from IDH1/2N
Discussion
Table 3 summarizes the results of the receiver operating characteristic curve (ROC) analyses of the histogram
In this study assessing whether the microenvironmental heterogeneity could discriminate IDH1/2P from IDH1/
123
J Neurooncol (2015) 121:141–150
2N patients with AA or GBM, we found that the mean ADC value, A10, A50, C90 and the slope between C10 and C90 derived from ADC and DSC-PWI had potential differentiating power between these groups. Among these histogram parameters, multivariate logistic regression analysis showed that the ADC mean value, C90 and CBV slope were more useful for differentiating between the IDH1P and IDH1N groups.
145
IDH mutations are a highly selective molecular marker of secondary GBMs that complement the clinical criteria for distinguishing them from primary GBMs [6]. Although several previous studies [3, 4, 6, 21–23] have shown IDH mutations can occur in approximately 60–80 % of diffusely infiltrating gliomas and secondary GBMs but only in approximately 5 % of primary GBMs, the microenvironmental heterogeneity of these lesions has not been
123
146 Table 1 Clinical characteristics of the patients
J Neurooncol (2015) 121:141–150
Characteristic Total number of patients
Total
IDH1/2P
IDH1/2N n = 36
n = 52
n = 16
IDH1
52
16
36
IDH2
0
0
0
Age, years
49.8 ± 14.5
41.6 ± 10.2
53.4 ± 14.7
.005*
Diagnosis
52
16
36
.007
AA, grade III
15
9
6
GBM, grade IV
37
7
30 .549
Sex GBM glioblastoma multiforme, AA anaplastic astrocytoma, MGMT methylguanine methyltransferase * The difference between the two groups was evaluated using the unpaired Student’s t test
The difference between the two groups was evaluated using Fisher’s exact test
Male
32
11
21
Female
20
5
15 [.99
MGMT Unmethylation
26
6
20
Methylation
26
10
16 [.99
BRAF mutation No mutation Mutation
elucidated. Previous studies using MRS demonstrated that the accumulation of 2-hydroxyglutarate (2-HG) in the tumor was associated with IDH mutations in 30 patients with gliomas [8, 24, 25]. In several previous studies [13, 19], HGGs have shown significantly lower ADC values than low-grade gliomas; this finding is possibly due to the high cellularity of HGGs. Such high cellularity is associated with relative reductions in extracellular space, resulting in decreased diffusivity of water molecules. Thus, ADC value has been considered an inverse index of tumor cellularity. Kang et al. [13] reported that the degree of ADC decrease tended to increase with increasing tumor grade using the histogram method. The potential advantage of the histogram method is that obtaining the total voxel values of a tumor would provide data that are more objective in a single tumor [13, 26]. Tan et al. [27] reported that minimal ADC and relative minimal ADC could be used for the differentiation of the HGGs with IDH1 R132H mutation and those without the mutation. In our study, the ADC mean value, A10 and A50 of the IDH1N group were significantly lower than the IDH1P group, which means that IDH1P tumors had a more heterogeneous microenvironment. However, only the ADC mean value had statistical significance in multivariate logistic regression models, because of the high cellularity of the portion with low ADC values. Although the weak difference of ADC value between IDH1P and IDH1N groups (mean ± SD, 1,282.8 ± 184.8 9 10-6 vs. 1,159.6 ± 141.9 9 10-6 mm2/s, P = .0113) derived from and the present study requires further validation, we believe that IDH1P tumors have tumor components of broad grades,
123
P value
48 4
15 1
33 3
Table 2 Histogram parameters from ADC and nCBV Maps in IDH1P and IDH1N IDH1P ADC (910
-6
IDH1N
P value
2
mm /s)
10th percentile (A10)
788.2 ± 158.8
699.9 ± 81.4
.0104
50th percentile (A50)
1,229.6 ± 200.2
1,096.3 ± 173.5
.0183
90th percentile (A90)
1,869.0 ± 551.2
1,717.8 ± 319.0
.1275
Slope [(A90A10)/0.8]
1,350.9 ± 551.2
1,272.3 ± 319.0
.5191
1,282.8 ± 184.8
1,159.6 ± 141.9
.0113
10th percentile (C10)
0.4538 ± 0.330
0.4131 ± 0.252
.628
50th percentile (C50)
1.248 ± 0.732
1.242 ± 0.5073
.9761
90th percentile (C90)
2.397 ± 1.212
3.121 ± 0.9102
.0208
Slope [(C90C10)/0.8]
2.425 ± 1.271
3.386 ± 1.066
.0067
Mean
1.363 ± 0.723
1.544 ± 0.4840
.2925
Mean nCBV
Data are shown as the mean ± SD
which results in higher mean ADC value than IDH1N tumors. However, this difference was not that high, because the HGGs have mainly high cellular components. Thus, we believe that the diagnostic value of the single ADC measurement has limitation for the detection of IDH
J Neurooncol (2015) 121:141–150
147
Fig. 2 Differences in histograms and cumulative histograms between IDH1P and IDH1N. a ADC histogram for the IDH1P and IDH1N groups obtained at a b value of 1,000 s/mm2. ADC histograms of IDH1N showed a higher relative frequency at the low ADC values compared with IDH1P, resulting in substantial divergence between the IDH1P and IDH1N groups at the low end of cumulative histograms. b The cumulative histogram for the IDH1P and IDH1N groups obtained at a b value of 1,000 s/mm2. The left shift of the cumulative histogram of the IDH1N group correlates well with the significant difference in the
Table 3 Diagnostic accuracy of histogram parameters for differentiating IDH1P from IDH1N
10th and 50th percentiles of ADC between the two groups. This left shift in the IDH1N group means that the values in IDH1N are more distributed within the lower ADC values compared to those in IDH1P group. c, d The nCBV histogram and cumulative histogram for the IDH1P and IDH1N groups. The nCBV histograms of the IDH1N group had a higher relative frequency at high nCBV values than did the nCBV histograms of the IDH1P group, resulting in substantial divergence between the IDH1P and IDH1N groups in the slope of the cumulative histograms
AUC, median (95 % CI)
Sensitivity (%)
Specificity (%)
Cutoff value
P value
0.707 (0.564–0.825)
50
91.7
[1,333.42
.0178
ADC Mean
(910-6 mm2/s) 10 % (A10)
0.707 (0.564–0.825)
50
97.2
[797
.0250
(910-6 mm2/s) Unless otherwise indicated, data in parentheses are 95 % confidence interval AUC area under the receiver operator characteristic curve, CI confidence interval
50 % (A50)
0.690 (0.547–0.825)
43.7
91.7
[1,299 (910
-6
.0256 2
mm /s)
nCBV 90 % (C90)
0.713 (0.570–0.830)
56.2
88.9
B2.04
.0195
Slope [(C90-C10)/0.8]
0.732 (0.591–0.845)
56.2
91.7
B1.99
.0069
mutation in HGGs, which can be helpful for the evaluation of the tumor heterogeneity. There have been many reports of glioma grading using perfusion-weighted imaging, although only a few studies have dealt with the histogram method to [14, 26, 28, 29]. Emblem et al. [26] reported that the maximum peak height of the nCBV histogram was lower and exhibited a wider distribution of pixel values in HGGs, which represented the
tumor heterogeneity in HGGs. The 99th percentile value on their cumulative nCBV histogram displayed higher diagnostic accuracy for differentiating high- from low-grade than the peak height [9]. Our results suggest that the C90 value can be used to differentiate IDH1P from IDH1N. In addition, the slope between the C10 and C90 values had a better diagnostic performance than other slope values. These results suggest that IDH1P tumors are composed of
123
148
more heterogeneous microenvironment. However, in terms of number of pixels from the whole tumor segmentation, ADC tends to have more pixels than DSC-PWI imaging, which we believe makes ADC reflect tumor heterogeneity better than DSC-PWI. In addition, we used gradient-echo DSC-PWI imaging for evaluation of nCBV. Gradient-echo DSC may not be sensitive/specific to IDH1P as it is biased more towards larger vessels and spin-echo DSC may be sensitive to smaller vessels to show more difference between IDH1P and IDH1N. In the previous report [12], the conventional MR imaging with the GBM with IDH1P showed the larger tumor size, the presence of large nonenhancing regions, and more frequent involvement of frontal lobe. We believe that the histogram method based on the whole tumor segmentation method in the present study can be used for the objective evaluation of the microenviroment of tumors; thus, our study shows IDH1P tumors tend to be more heterogeneous that IDH1N ones from the analysis of ADC and DSC-PWI. In the present study, we used nCBV maps, which were normalized by dividing every rCBV value in a specific section by the rCBV value in the unaffected white matter, to minimize variances in the rCBV value in an individual patient. In previous reports [17, 30], a disparity in rCBV estimates was found among and between acquisition and postprocessing methods, because of the presence of contrast leakage effect, especially high-grade tumors. However, in our institute, nCBV measurement is more feasible and well-established for the brain tumor imaging. Even though rCBV maps corrected for contrast agent extravasation has several benefits for the evaluation of the brain tumors, nCBV measurement is also one of popular methods in that purpose [9, 10]. Additionally, in previous reports [6, 31], there is a striking difference in the age distribution of patients with primary and secondary glioblastoma, and similarly with or without IDH1/2 gene mutations. The patient with IDH1P had a very high frequency of TP53 mutation and a very low frequency of mutations in other commonly altered GBM genes. This association between IDH1P and improved survival was noted in young patients with TP53 mutations. Apart from the intrinsic limits of any retrospective study, several other limitations of our study should be mentioned. First, we used two different 3-T MR units, which differed in terms of their imaging parameters. However, we optimized the sequences to minimize the differences between the two units. We believe that there might have been a small bias in terms of image analysis of the ADC maps; the IDH1P and IDH1N groups presented similar MR imaging in the two units. Second, the present study included a relatively small patient population. In our study, the incidence of GBM with IDHP is relatively
123
J Neurooncol (2015) 121:141–150
higher than other study. Because of retrospectively designed study, there might be selection bias due to inclusion and exclusion criteria. Further investigation of a larger population is warranted. Third, tumor boundary was defined with reference to high signal intensity on T2WI, and tumor infiltration and peritumoral edema were included in the regions of interests. However, the differentiation between these two components is impossible in imaging studies. In addition, we tried to evaluate the tumor heterogeneity related with IDH mutation, and we believe that T2 signal intensity is better segmentation reference that other imaging methods such as contrastenhanced imaging for the present study, because T2 signal intensities include various components such as tumorous condition and edema, which are also results of tumor heterogeneity. Fourth, the normal-appearing white matter as a reference region for the calculation of nCBV may be affected by, for example, radiation therapy and tumor invasion [11]. However, we excluded the patient who underwent CCRT prior to the imaging, and tumor invasion was evaluated on the conventional MR sequences as best as we could manage. Finally, for the detection of IDH gene mutation, MRS is getting more and more popular to detect 2-HG, the product of the mutated IDH, which is specific biomarker for IDH gene mutation. However, as mentioned above, the aim of our study was to investigate whether the IDHP tumors have more heterogeneity than negative ones, which was shown by DWI and DSC-PWI. In addition, our study results based on these physiologic MR imaging show that MR imaging can also predict GBM genetic and cellular biologic features and imply physiologic MR imaging may be useful in targeting specific areas within heterogeneous tumor tissues. In conclusion, our results suggest that histogram analysis of ADC and nCBV maps based on the entire tumor volume can be a useful tool for distinguishing IDH1P and IDH1N. The cumulative ADC and nCBV histograms can predict that IDHP tumors have a more heterogeneous microenvironment, because of their stepwise tumorigenesis. Those modalities also can reflect the microenvironmental heterogeneity of the gliomas. Application of advanced imaging modalities in HGGs can predict the prognosis of IDH1P HGG patients and is recommended as part of the MR study of suspected HGGs. Acknowledgments This study was supported by a grant from the National R&D Program for Cancer Control, Ministry of Health & Welfare, Republic of Korea (1120300), the Korea Healthcare technology R&D Projects, Ministry for Health, Welfare & Family Affairs (A112028 and HI13C0015), and the Research Center Program of IBS (Institute for Basic Science) in Korea. Conflict of interest interest.
None of the authors have any conflict of
J Neurooncol (2015) 121:141–150
References 1. Daumas-Duport C, Scheithauer B, O’Fallon J, Kelly P (1988) Grading of astrocytomas. A simple and reproducible method. Cancer 62:2152–2165 2. Nikiforova MN, Hamilton RL (2011) Molecular diagnostics of gliomas. Arch Pathol Lab Med 135:558–568 3. Balss J, Meyer J, Mueller W, Korshunov A, Hartmann C, von Deimling A (2008) Analysis of the IDH1 codon 132 mutation in brain tumors. Acta Neuropathol 116:597–602 4. Yan H, Parsons DW, Jin GL, McLendon R, Rasheed BA, Yuan WS, Kos I, Batinic-Haberle I, Jones S, Riggins GJ, Friedman H, Friedman A, Reardon D, Herndon J, Kinzler KW, Velculescu VE, Vogelstein B, Bigner DD (2009) IDH1 and IDH2 mutations in gliomas. New Engl J Med 360:765–773 5. Hartmann C, Hentschel B, Wick W, Capper D, Felsberg J, Simon M, Westphal M, Schackert G, Meyermann R, Pietsch T, Reifenberger G, Weller M, Loeffler M, von Deimling A (2010) Patients with IDH1 wild type anaplastic astrocytomas exhibit worse prognosis than IDH1-mutated glioblastomas, and IDH1 mutation status accounts for the unfavorable prognostic effect of higher age: implications for classification of gliomas. Acta Neuropathol 120:707–718 6. Parsons DW, Jones S, Zhang X, Lin JC, Leary RJ, Angenendt P, Mankoo P, Carter H, Siu IM, Gallia GL, Olivi A, McLendon R, Rasheed BA, Keir S, Nikolskaya T, Nikolsky Y, Busam DA, Tekleab H, Diaz LA Jr, Hartigan J, Smith DR, Strausberg RL, Marie SK, Shinjo SM, Yan H, Riggins GJ, Bigner DD, Karchin R, Papadopoulos N, Parmigiani G, Vogelstein B, Velculescu VE, Kinzler KW (2008) An integrated genomic analysis of human glioblastoma multiforme. Science 321:1807–1812 7. Sanson M, Marie Y, Paris S, Idbaih A, Laffaire J, Ducray F, El Hallani S, Boisselier B, Mokhtari K, Hoang-Xuan K, Delattre JY (2009) Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol 27:4150–4154 8. Andronesi OC, Rapalino O, Gerstner E, Chi A, Batchelor TT, Cahil DP, Sorensen AG, Rosen BR (2013) Detection of oncogenic IDH1 mutations using magnetic resonance spectroscopy of 2-hydroxyglutarate. J Clin Invest 123:3659–3663 9. Kim H, Choi SH, Kim JH, Ryoo I, Kim SC, Yeom JA, Shin H, Jung SC, Lee AL, Yun TJ, Park CK, Sohn CH, Park SH (2013) Gliomas: application of cumulative histogram analysis of normalized cerebral blood volume on 3 T MRI to tumor grading. PLoS ONE 8(5):e63462. doi:10.1371/journal.pone.0063462 10. Jung SC, Choi SH, Yeom JA, Kim JH, Ryoo I, Kim SC, Shin H, Lee AL, Yun TJ, Park CK, Sohn CH, Park SH (2013) Cerebral blood volume analysis in glioblastomas using dynamic susceptibility contrast-enhanced perfusion MRI: a comparison of manual and semiautomatic segmentation methods. PLoS ONE 8(8):e69323. doi:10.1371/journal.pone.0069323 11. Thomsen H, Steffensen E, Larsson EM (2012) Perfusion MRI (dynamic susceptibility contrast imaging) with different measurement approaches for the evaluation of blood flow and blood volume in human gliomas. Acta Radiol 53:95–101 12. Carrillo JA, Lai A, Nghiemphu PL, Kim HJ, Phillips HS, Kharbanda S, Moftakhar P, Lalaezari S, Yong W, Ellingson BM, Cloughesy TF, Pope WB (2012) Relationship between tumor enhancement, edema, IDH1 mutational status, MGMT promoter methylation, and survival in glioblastoma. Am J Neuroradiol 33:1349–1355 13. Kang Y, Choi SH, Kim YJ, Kim KG, Sohn CH, Kim JH, Yun TJ, Chang KH (2011) Gliomas: histogram analysis of apparent diffusion coefficient maps with standard- or high-b-value diffusionweighted MR imaging—correlation with tumor grade. Radiology 261:882–890
149 14. Garzon B, Emblem KE, Mouridsen K, Nedregaard B, DueTonnessen P, Nome T, Hald JK, Bjornerud A, Haberg AK, Kvinnsland Y (2011) Multiparametric analysis of magnetic resonance images for glioma grading and patient survival time prediction. Acta Radiol 52:1052–1060 15. Rosen BR, Belliveau JW, Vevea JM, Brady TJ (1990) Perfusion imaging with NMR contrast agents. Magn Reson Med 14:249–265 16. Ostergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR (1996) High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. 1. Mathematical approach and statistical analysis. Magnet Reson Med 36:715–725 17. Boxerman JL, Schmainda KM, Weisskoff RM (2006) Relative cerebral blood volume maps corrected for contrast agent extravasation significantly correlate with glioma tumor grade, whereas uncorrected maps do not. Am J Neuroradiol 27:859–867 18. Wetzel SG, Cha S, Johnson G, Lee P, Law M, Kasow DL, Pierce SD, Xue X (2002) Relative cerebral blood volume measurements in intracranial mass lesions: interobserver and intraobserver reproducibility study. Radiology 224:797–803 19. Tozer DJ, Jager HR, Danchaivijitr N, Benton CE, Tofts PS, Rees JH, Waldman AD (2007) Apparent diffusion coefficient histograms may predict low-grade glioma subtype. NMR Biomed 20:49–57 20. Delong ER, Delong DM, Clarkepearson DI (1988) Comparing the areas under 2 or more correlated receiver operating characteristic curves—a nonparametric approach. Biometrics 44:837–845 21. Hartmann C, Meyer J, Balss J, Capper D, Mueller W, Christians A, Felsberg J, Wolter M, Mawrin C, Wick W, Weller M, HeroldMende C, Unterberg A, Jeuken JWM, Wesseling P, Reifenberger G, von Deimling A (2009) Type and frequency of IDH1 and IDH2 mutations are related to astrocytic and oligodendroglial differentiation and age: a study of 1,010 diffuse gliomas. Acta Neuropathol 118:469–474 22. Ichimura K, Pearson DM, Kocialkowski S, Backlund LM, Chan R, Jones DTW, Collins VP (2009) IDH1 mutations are present in the majority of common adult gliomas but rare in primary glioblastomas. Neuro-Oncology 11:341–347 23. Watanabe T, Nobusawa S, Kleihues P, Ohgaki H (2009) IDH1 mutations are early events in the development of astrocytomas and oligodendrogliomas. Am J Pathol 174:1149–1153 24. Pope WB, Prins RM, Albert Thomas M, Nagarajan R, Yen KE, Bittinger MA, Salamon N, Chou AP, Yong WH, Soto H, Wilson N, Driggers E, Jang HG, Su SM, Schenkein DP, Lai A, Cloughesy TF, Kornblum HI, Wu H, Fantin VR, Liau LM (2012) Non-invasive detection of 2-hydroxyglutarate and other metabolites in IDH1 mutant glioma patients using magnetic resonance spectroscopy. J Neurooncol 107:197–205 25. Choi C, Ganji SK, DeBerardinis RJ, Hatanpaa KJ, Rakheja D, Kovacs Z, Yang XL, Mashimo T, Raisanen JM, Marin-Valencia I, Pascual JM, Madden CJ, Mickey BE, Malloy CR, Bachoo RM, Maher EA (2012) 2-hydroxyglutarate detection by magnetic resonance spectroscopy in IDH-mutated patients with gliomas. Nat Med 18:624–629 26. Emblem KE, Nedregaard B, Nome T, Due-Tonnessen P, Hald JK, Scheie D, Borota OC, Cvancarova M, Bjornerud A (2008) Glioma grading by using histogram analysis of blood volume heterogeneity from MR-derived cerebral blood volume maps. Radiology 247:808–817 27. Tan WL, Huang WY, Yin B, Xiong J, Wu JS, Geng DY (2014) Can diffusion tensor imaging noninvasively detect IDH1 gene mutations in astrogliomas? A retrospective study of 112 cases. AJNR Am J Neuroradiol 35:920–927. doi:10.3174/ajnr.A3803 28. Emblem KE, Scheie D, Due-Tonnessen P, Nedregaard B, Nome T, Hald JK, Beiske K, Meling TR, Bjornerud A (2008) Histogram
123
150 analysis of mr imaging-derived cerebral blood volume maps: combined glioma grading and identification of low-grade oligodendroglial subtypes. Am J Neuroradiol 29:1664–1670 29. Young R, Babb J, Law M, Pollack E, Johnson G (2007) Comparison of region-of-interest analysis with three different histogram analysis methods in the determination of perfusion metrics in patients with brain gliomas. J Magn Reson Imaging 26:1053–1063 30. Paulson ES, Schmainda KM (2008) Comparison of Dynamic susceptibility-weighted contrast-enhanced mr methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 249:601–613
123
J Neurooncol (2015) 121:141–150 31. Jiao YC, Killela PJ, Reitman ZJ, Rasheed BA, Heaphy CM, de Wilde RF, Rodriguez FJ, Rosemberg S, Oba-Shinjo SM, Marie SKN, Bettegowda C, Agrawal N, Lipp E, Pirozzi CJ, Lopez GY, He YP, Friedman HS, Friedman AH, Riggins GJ, Holdhoff M, Burger P, McLendon RE, Bigner DD, Vogelstein B, Meeker AK, Kinzler KW, Papadopoulos N, Diaz LA, Yan H (2012) Frequent ATRX, CIC, FUBP1 and IDH1 mutations refine the classification of malignant gliomas. Oncotarget 3:709–722
CLINICAL STUDY
Evaluation of the microenvironmental heterogeneity in high-grade gliomas with IDH1/2 gene mutation using histogram analysis of diffusion-weighted imaging and dynamic-susceptibility contrast perfusion imaging Seunghyun Lee • Seung Hong Choi • Inseon Ryoo • Tae Jin Yoon Tae Min Kim • Se-Hoon Lee • Chul-Kee Park • Ji-Hoon Kim • Chul-Ho Sohn • Sung-Hye Park • Il Han Kim
•
Received: 11 February 2014 / Accepted: 30 August 2014 / Published online: 10 September 2014 Ó Springer Science+Business Media New York 2014
Abstract The purpose of our study was to explore the difference between isocitrate dehydrogenase (IDH)-1/2 gene mutation-positive and -negative high-grade gliomas (HGGs) using histogram analysis of apparent diffusion coefficient (ADC) and normalized cerebral blood volume (nCBV) maps. We enrolled 52 patients with histopathologically confirmed HGGs with IDH1/2P (n = 16) or IDH1/2N (n = 36). Histogram parameters of ADC and nCBV maps were correlated with gene mutations by using the unpaired student’s t test and multivariable stepwise logistic regression analysis. The mean ADC value was higher in the IDH1P group than IDH1N (1,282.8 vs. 1,159.6 mm2/s, P = .0113). In terms of the cumulative ADC histograms, the 10th and 50th percentile values were also higher in the IDH1P than IDH1N (P = .0104 and .0183, respectively). We observed a higher 90th percentile value (3.121 vs. 2.397, P = .0208) and a steeper slope between the 10th (C10) and 90th (C90) of cumulative Electronic supplementary material The online version of this article (doi:10.1007/s11060-014-1614-z) contains supplementary material, which is available to authorized users. S. Lee S. H. Choi (&) I. Ryoo T. J. Yoon J.-H. Kim C.-H. Sohn Department of Radiology, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea e-mail: [email protected] S. H. Choi Center for Nanoparticle Research, Institute for Basic Science, and School of Chemical and Biological Engineering, Seoul National University, Seoul 151-742, Korea T. M. Kim S.-H. Lee Department of Internal Medicine, Cancer Research Institute, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea
nCBV histograms (0.03386 vs. 0.02425/%, P = .0067) in the IDH1N group. Multivariate analysis showed that the mean ADC mean value (P = .0048), the C90 value (P = .0113), and the slope between C10 and C90 (P = .0049) were the significant variables in the differentiation of IDH1P from IDH1N. In conclusion, histogram analysis of ADC and nCBV maps based on entire tumor volume can be a useful tool for distinguishing IDH1P and IDH1N, and it predicts that IDHP tumors have a more heterogeneous microenvironment than IDHN ones. Keywords Isocitrate dehydrogenase IDH gene mutation High-grade gliomas ADC nCBV DWI DSC-PWI
Introduction Gliomas are the most common primary neoplasm of the brain, varying histopathologically from low grade to high
C.-K. Park Department of Neurosurgery, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea S.-H. Park Department of Pathology, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea I. H. Kim Department of Radiation Oncology, Cancer Research Institute, Seoul National University College of Medicine, 28, Yongon-dong, Chongno-gu, Seoul 110-744, Korea
123
142
grade [1]. High-grade gliomas (HGGs) include World Health Organization (WHO) grade III and IV brain tumors such as anaplastic astrocytoma (AA), and primary or secondary glioblastoma multiforme (GBM). Primary GBMs arise de novo in older patients, while secondary GBMs develop from WHO grade II or III gliomas, have a longer duration of gliomagenesis, and arise from an early event such as isocitrate dehydrogenase (IDH) gene mutation [2]. According to a recent genomic analysis of HGGs, IDH gene mutations occur in *60–90 % of diffuse and anaplastic gliomas and secondary GBMs [2]. The recent identification of IDH gene mutation has directed attention to the role of abnormal metabolism in the pathogenesis and progression of these primary brain tumors [3, 4]. In fact, GBM patients with this mutation have significantly better survival, and IDH gene mutation testing is relevant for clinical patient management and for stratification in clinical trials [5–7]. However, the current histopathologic grading method by stereotactic biopsy has major limitations associated with its inherent sampling error or its inability to predict the prognosis of the disease before surgical resection [1]. Currently, conventional magnetic resonance (MR) imaging along with advanced imaging techniques, such as diffusion-weighted imaging (DWI), dynamic-susceptibility contrast perfusionweighted imaging (DSC-PWI), and MR spectroscopy (MRS), helps in grading of gliomas and may complement the histopathologic grade [8]. In particular, DWI and DSC-PWI MRI techniques provide important in vivo physiologic information about brain tumors and have been of increasing utility including tumor grading [9–14]. To our knowledge, however, there have been no previous reports concerning the use of volume-based histogram analysis of DWI and DSC-PWI to differentiate IDHP from IDHN gliomas. Our hypotheses were that (a) IDHP tumors have a more heterogeneous microenvironment, because of their stepwise tumorigenesis, and (b) DWI and DSC-PWI can reflect the microenvironmental heterogeneity of the gliomas. This study aimed to explore the difference between IDH1/2P and IDH1/2N HGGs using histogram analysis of apparent diffusion coefficient (ADC) maps and normalized cerebral blood volume (nCBV) based on the entire tumor volume and to evaluate the characteristics of IDH gene mutants.
Materials and methods Our institutional review board approved this study and informed consent was waived. Patient selection One hundred thirty-four patients who underwent surgical resection or stereotactic biopsy at our institution between
123
J Neurooncol (2015) 121:141–150
May 2005 and December 2012 were selected from our radiology report database. The inclusion criteria were as follows: (a) had a histopathological diagnosis of AA or GBM based on the WHO criteria, with or without oligodendroglial components; (b) had undergone surgery or stereotactic biopsy; and (c) had baseline three-tesla (3-T) MR imaging performed with DWI (b = 1,000 s/mm2) and DSC-PWI prior to surgery or chemoradiotherapy. We excluded 82 patients due to the following reasons: (a) poor quality of the MR images; (b) only 1.5-T MR images; and (c) had undergone baseline MR imaging after surgery. As a result, a total of 52 patients with AA or GBM (32 men and 20 women; age range: 22–72 years; mean age: 50 years) were analyzed. Image acquisition For each patient, the first or follow-up MR imaging before surgical biopsy was performed with one of the following 3-T MR imaging scanners: a Signa Excite (GE Medical Systems, Milwaukee, WI, USA) with an 8-channel head coil (n = 6) and a Verio (Siemens Medical Solutions, Erlangen, Germany) with a 32-channel head coil (n = 46). The imaging sequences of the brain included axial fast/ turbo spin-echo T2-weighted images (T2WI), DWI and DSC-PWI with gadobutrol (Gadovist, Bayer Schering Pharma, Berlin, Germany). The MR imaging parameters were as follows: 4,500–5,160/91–106.3 ms/90–130°/ 448–640 9 220 for fast spin-echo T2WI. The other parameters for the three images were as follows: section thickness, 5 mm with a 1 mm gap; and field of view (FOV), 240 9 240 mm. DWI was performed with a single-shot spin-echo echoplanar imaging sequence in the axial plane before the injection of contrast material with a TR/TE of 6,900–10,000/55–70 ms at b = 0 and 1,000 s/mm2, 35–38 sections, a 3-mm section thickness, a 1-mm intersection gap, an FOV of 240 9 240 mm, a matrix of 160 9 160, three signal averages, and a voxel resolution of 1.5 9 1.5 9 3 mm. DWI data were acquired in three orthogonal directions. Using these data, the averaged ADC maps in the three orthogonal directions were calculated on a voxel-by-voxel basis with the software incorporated into the MR imaging unit. DSC-PWI was performed with a single-shot gradientecho echo-planar imaging sequence during the intravenous injection of the contrast agent. The imaging parameters of DSC-PWI were as follows: TR/TE, 1,500/ 30–40 ms; FA, 35–90°; FOV, 240 9 240 mm; 15–20 sections; matrix, 128 9 128; section thickness, 5 mm; intersection gap, 1 mm; and voxel resolution of 1.86 9 1.86 9 5 mm. For each section, 60 images were obtained at intervals equal to the repetition time. After
J Neurooncol (2015) 121:141–150
four to five time points, a bolus of gadobutrol at a dose of 0.1 mmol/kg of body weight and at a rate of 4 mL/sec was injected with an MR-compatible power injector (Spectris; Medrad, Pittsburgh, PA, USA). After injecting a bolus of the contrast material, a 30 mL bolus of saline was administered at the same injection rate.
Quantitative image analysis The MR data for the ADC maps and perfusion raw data were digitally transferred from the picture archiving and communication system workstation to a personal computer and processed with a dedicated software package (nordicICE; NordicImagingLab, Bergen, Norway). One author (S.H.C., 9 years of experience with neuroradiology imaging) manually drew the tumor borders in each section of the co-registered T2WI; however, the areas of necrosis, cysts, and nontumor macrovessels were drawn in each section of the T2WI and copied to the ADC map and relative CBV (rCBV) map after semiautomatic coregistration. The relative CBV (rCBV) was acquired using an established tracer kinetic model applied to the first-pass data [15, 16]. The dynamic curves were mathematically corrected to reduce contrast-agent leakage effects [17]. To minimize variances in the rCBV value in an individual patient, the pixel-based rCBV maps were normalized by dividing every rCBV value in a specific section by the rCBV value in the unaffected white matter as defined by a neuroradiologist (S.H.C.) [18]. Coregistrations between the T2WI and the ADC maps or the nCBV map were performed based on geometric information stored in the respective data sets by using a dedicated software package (nordicICE). The ADC and nCBV maps were displayed as color overlays on the T2WI (Fig. 1). The data acquired from each section were summated to derive the voxel-by-voxel ADCs and nCBVs for the entire tumor extent of the image by using nordicICE. The ADC and nCBV histograms were plotted with ADC and nCBV on the respective x-axis, with a bin size of 3 9 10-5 mm2/s and 0.1, respectively, whereas the y-axis was expressed as a percentage of the total lesion volume by dividing the frequency in each bin by the total number of analyzed voxels. For further quantitative analysis, the cumulative number of observations in all the bins up to the specified bin was mapped on the y-axis as a percentage in the cumulative histograms. The 10th, 50th and 90th percentile points for ADC and nCBV (A10, A50 and A90, respectively, for ADC; C10, C50 and C90, respectively, for nCBV) were derived (the Xth percentile point is the point at which X% of the voxel values that form the histogram are found to the left of the histogram) [13, 19]. In addition, the slopes between the 10th and 90th percentiles of cumulative ADC and nCBV histograms were also
143
calculated from the following two equations, respectively: (A90–A10)/0.8 and (C90–C10)/0.8. Statistical analysis All statistical analyses were performed with MedCalc software (version 12.1.0 for Microsoft Windows 2000/XP/ Vista/7; MedCalc Software, Mariakerke, Belgium). The results with a P value of less than .05 were considered statistically significant. The clinical characteristics were compared between the IDHP and IDHN groups by using Fisher’s exact test for categorical variables and the unpaired Student t test for noncategorical data. The unpaired Student t test was used to compare the histogram parameters of IDHP and IDHN. The area under the receiver operating characteristic curve (AUC) was used to determine the best cutoff values for the histogram parameters that proved to be substantial predictors in differentiating IDHP from IDHN by using the method of DeLong et al. [20]. Subsequently, a multivariable stepwise logistic regression analysis was used to determine the significant parameters between IDHP and IDHN. Variables with a P value of \.05 according to the univariate analysis were used as input variables for multivariable stepwise logistic regression analysis, with iterative entry of variables on the basis of test results (P values of \.05). The removal of variables was based on likelihood ratio statistics with a probability of 0.10.
Results Differences of clinical characteristics between IDH1/2P and IDH1/2N groups We determined whether any of the clinical characteristics, including age, sex, histology, and other gene mutations, were predictors of IDH1/2P (n = 16) or IDH1/2N (n = 36) before surgery or CCRT. The 52 patients enrolled in the present study included 16 IDH1Pand no IDH2P patients. The IDH1P group had a significantly higher proportion of AA than IDH1N (P = .007). Only patient age was significantly lower in the IDH1P group than in the IDH1N group (Table 1). Differences of histogram parameters derived from ADC maps between the IDH1/2P and IDH1/2N groups The mean ADC value was higher in the IDH1P group than in the IDH1N group (mean ± SD, 1,282.8 ± 184.8 9 10-6 vs. 1,159.6 ± 141.9 9 10-6 mm2/s, P = .0113). In terms of the cumulative histograms, the A10 (788.2 ± 158.8 9 10-6 vs. 699.9 ± 81.4 9 10-6 mm2/s) and A50 (1,229.6 ± 200.2 9 10-6 vs. 1,096.3 ± 173.5 9 10-6 mm2/ s) for ADC values also showed a significant difference
123
144
between the IDH1P and IDH1N groups (P = .0104 and P = .0183, respectively). However, we did not find a significant difference between groups in terms of A90 and slope between A10 and A90 (P [ .05). The histogram parameters from ADC maps are summarized in Table 2. Figure 2a, b shows ADC (b = 1,000 s/mm2) and cumulative ADC histogram for the IDH1P and IDH1N groups. ADC histograms of IDH1N showed a higher relative frequency at the low ADC values compared with IDH1P. The cumulative histogram for the IDH1P and IDH1N groups showed a left shift of the cumulative histogram of the IDH1N group. In terms of each tumor grade, there was no significant difference of ADC value between IDH1P and IDH1N groups in grade III (P [ .05). With regards to the cumulative histograms of grade IV, the A10 (805.3 ± 104.9 9 10-6 vs. 700.7 ± 82.7 9 10-6 mm2/s), A90 (2,120.0 ± 498.2 9 10-6 vs. 1,740.1 ± 268.5 9 10-6 mm2/s), and slope between A10 and A90 (1,643.4 ± 683.3 9 10-6 vs. 1,299.3 ± 317.0 9 10-6 mm2/s) for ADC values also showed a significant difference between the IDH1P and IDH1N groups (P = .007, P = .008, and P = .049, respectively). The histogram parameters from ADC maps of grade III and IV are summarized in Supplementary Table 2 and 3. Differences of histogram parameters derived from nCBV maps between the IDH1/2P and IDH1/2N groups With regard to nCBV values, the IDH1N tumors had a higher C90 value (3.121 ± 0.9,102 vs. 2.397 ± 1.212, P = .0208) and a steeper slope of cumulative nCBV histograms (3.386 ± 1.066 vs. 2.425 ± 1.271, P = .0067) than IDH1P tumors. However, other parameters, including C10, C50 and mean nCBV, were not significantly different between groups (P C .05). The histogram parameters from nCBV maps are also summarized in Table 2. Figure 2c, d shows the nCBV histogram and cumulative histogram for the IDH1P and IDH1N groups. In terms of each tumor grade, there was no significant difference between IDH1P and IDH1N groups in grade III (P [ .05). With regard to nCBV values, there was no significant difference between the IDH1P and IDH1N groups in grade III (P [ .05) and grade IV (P [ .05). The histogram parameters from nCBV maps of grade III and IV are summarized in Supplementary Table 2 and 3. The representative IDH1P and IDH1N GBMs are demonstrated in Fig. 1b, c, respectively.
J Neurooncol (2015) 121:141–150 Fig. 1 Flowchart of histogram analysis, and representative IDH1P c and IDH1N histograms. a Flowchart of histogram analysis. The upper, middle, and bottom rows show the defining region of interest (ROI) and the acquisition of ADC and nCBV maps. Based on T2WI (upper middle), including all the T2 high-signal portions of the lesion, coregistrations between the T2WI (upper middle) and the ADC maps (upper right) or the nCBV maps (upper left) were performed based on geometric information stored in the respective data sets by using a dedicated software package (nordicICE). After acquiring co-registered images on a voxel-by-voxel basis, ROIs were drawn (bottom). Finally, the overall values for each tumor were obtained by summing the histogram parameter values from every plane. b Representative IDH1P histograms in a 49-year-old male with GBM. This figure was acquired from a 49-year-old IDH1P man with GBM. Axial T2weighted MR image (upper left) obtained before surgical resection or CCRT shows a newly developed T2 high-signal-intensity lesion in the right temporal lobe. The ADC map (upper middle) and rCBV map (upper right) with regions of interest corresponding to axial T2weighted MR image (upper left) show the heterogeneous ADC values and vascularities of the lesion. The ADC histogram (middle left) and cumulative histograms (middle right) show a normal distribution of the relative frequency of ADC values and show a lower slope and homogenous relative frequency compared with (c). The nCBV histograms (lower left) and cumulative histograms (lower right) had a lower relative frequency at high nCBV values and lower slope compared with (c). c Representative IDH1N histograms in a 29-yearold female with GBM. This figure was acquired from a 29-year-old IDH1P woman with GBM. Axial T2-weighted MR image (upper left) obtained before surgical resection or CCRT shows a newly developed T2 high-signal-intensity lesion in the left thalamus. The ADC map (upper middle) and rCBV map (upper right) with regions of interest corresponding to axial T2-weighted MR image (upper left) show the heterogeneous ADC values and vascularities of the lesion. The ADC histogram (middle left) and cumulative histograms (middle right) show a higher relative frequency at the low ADC values, higher slope and heterogenous relative frequency compared with (b). The nCBV histograms (lower left) and cumulative histograms (lower right) had a higher relative frequency at high nCBV values, higher slope and at left shift compared with (b)
parameters for the differentiation of IDH1P (n = 16) from IDH1N (n = 36). In the comparison of ROC curves, no AUC value could distinguish IDH1P from IDH1N. The ADC mean value, A10, A50, C90 and the slope between C10 and C90, all of which showed significant differences between the IDH1P and IDH1N groups, were used as input variables for a multivariable stepwise logistic regression analysis. In the multivariate analysis, there was no independently differentiating variable in histogram parameters. However, we identified ADC mean value (P = .0048), C90 (P = .0113), and the slope between C10 and C90 (P = .0049) as significant variables in the differentiation of IDH1P from IDH1N, whereas A10 and A50 were excluded from the logistic regression equation.
Diagnostic performance and the best histogram predictors for the differentiation of IDH1/2P from IDH1/2N
Discussion
Table 3 summarizes the results of the receiver operating characteristic curve (ROC) analyses of the histogram
In this study assessing whether the microenvironmental heterogeneity could discriminate IDH1/2P from IDH1/
123
J Neurooncol (2015) 121:141–150
2N patients with AA or GBM, we found that the mean ADC value, A10, A50, C90 and the slope between C10 and C90 derived from ADC and DSC-PWI had potential differentiating power between these groups. Among these histogram parameters, multivariate logistic regression analysis showed that the ADC mean value, C90 and CBV slope were more useful for differentiating between the IDH1P and IDH1N groups.
145
IDH mutations are a highly selective molecular marker of secondary GBMs that complement the clinical criteria for distinguishing them from primary GBMs [6]. Although several previous studies [3, 4, 6, 21–23] have shown IDH mutations can occur in approximately 60–80 % of diffusely infiltrating gliomas and secondary GBMs but only in approximately 5 % of primary GBMs, the microenvironmental heterogeneity of these lesions has not been
123
146 Table 1 Clinical characteristics of the patients
J Neurooncol (2015) 121:141–150
Characteristic Total number of patients
Total
IDH1/2P
IDH1/2N n = 36
n = 52
n = 16
IDH1
52
16
36
IDH2
0
0
0
Age, years
49.8 ± 14.5
41.6 ± 10.2
53.4 ± 14.7
.005*
Diagnosis
52
16
36
.007
AA, grade III
15
9
6
GBM, grade IV
37
7
30 .549
Sex GBM glioblastoma multiforme, AA anaplastic astrocytoma, MGMT methylguanine methyltransferase * The difference between the two groups was evaluated using the unpaired Student’s t test
The difference between the two groups was evaluated using Fisher’s exact test
Male
32
11
21
Female
20
5
15 [.99
MGMT Unmethylation
26
6
20
Methylation
26
10
16 [.99
BRAF mutation No mutation Mutation
elucidated. Previous studies using MRS demonstrated that the accumulation of 2-hydroxyglutarate (2-HG) in the tumor was associated with IDH mutations in 30 patients with gliomas [8, 24, 25]. In several previous studies [13, 19], HGGs have shown significantly lower ADC values than low-grade gliomas; this finding is possibly due to the high cellularity of HGGs. Such high cellularity is associated with relative reductions in extracellular space, resulting in decreased diffusivity of water molecules. Thus, ADC value has been considered an inverse index of tumor cellularity. Kang et al. [13] reported that the degree of ADC decrease tended to increase with increasing tumor grade using the histogram method. The potential advantage of the histogram method is that obtaining the total voxel values of a tumor would provide data that are more objective in a single tumor [13, 26]. Tan et al. [27] reported that minimal ADC and relative minimal ADC could be used for the differentiation of the HGGs with IDH1 R132H mutation and those without the mutation. In our study, the ADC mean value, A10 and A50 of the IDH1N group were significantly lower than the IDH1P group, which means that IDH1P tumors had a more heterogeneous microenvironment. However, only the ADC mean value had statistical significance in multivariate logistic regression models, because of the high cellularity of the portion with low ADC values. Although the weak difference of ADC value between IDH1P and IDH1N groups (mean ± SD, 1,282.8 ± 184.8 9 10-6 vs. 1,159.6 ± 141.9 9 10-6 mm2/s, P = .0113) derived from and the present study requires further validation, we believe that IDH1P tumors have tumor components of broad grades,
123
P value
48 4
15 1
33 3
Table 2 Histogram parameters from ADC and nCBV Maps in IDH1P and IDH1N IDH1P ADC (910
-6
IDH1N
P value
2
mm /s)
10th percentile (A10)
788.2 ± 158.8
699.9 ± 81.4
.0104
50th percentile (A50)
1,229.6 ± 200.2
1,096.3 ± 173.5
.0183
90th percentile (A90)
1,869.0 ± 551.2
1,717.8 ± 319.0
.1275
Slope [(A90A10)/0.8]
1,350.9 ± 551.2
1,272.3 ± 319.0
.5191
1,282.8 ± 184.8
1,159.6 ± 141.9
.0113
10th percentile (C10)
0.4538 ± 0.330
0.4131 ± 0.252
.628
50th percentile (C50)
1.248 ± 0.732
1.242 ± 0.5073
.9761
90th percentile (C90)
2.397 ± 1.212
3.121 ± 0.9102
.0208
Slope [(C90C10)/0.8]
2.425 ± 1.271
3.386 ± 1.066
.0067
Mean
1.363 ± 0.723
1.544 ± 0.4840
.2925
Mean nCBV
Data are shown as the mean ± SD
which results in higher mean ADC value than IDH1N tumors. However, this difference was not that high, because the HGGs have mainly high cellular components. Thus, we believe that the diagnostic value of the single ADC measurement has limitation for the detection of IDH
J Neurooncol (2015) 121:141–150
147
Fig. 2 Differences in histograms and cumulative histograms between IDH1P and IDH1N. a ADC histogram for the IDH1P and IDH1N groups obtained at a b value of 1,000 s/mm2. ADC histograms of IDH1N showed a higher relative frequency at the low ADC values compared with IDH1P, resulting in substantial divergence between the IDH1P and IDH1N groups at the low end of cumulative histograms. b The cumulative histogram for the IDH1P and IDH1N groups obtained at a b value of 1,000 s/mm2. The left shift of the cumulative histogram of the IDH1N group correlates well with the significant difference in the
Table 3 Diagnostic accuracy of histogram parameters for differentiating IDH1P from IDH1N
10th and 50th percentiles of ADC between the two groups. This left shift in the IDH1N group means that the values in IDH1N are more distributed within the lower ADC values compared to those in IDH1P group. c, d The nCBV histogram and cumulative histogram for the IDH1P and IDH1N groups. The nCBV histograms of the IDH1N group had a higher relative frequency at high nCBV values than did the nCBV histograms of the IDH1P group, resulting in substantial divergence between the IDH1P and IDH1N groups in the slope of the cumulative histograms
AUC, median (95 % CI)
Sensitivity (%)
Specificity (%)
Cutoff value
P value
0.707 (0.564–0.825)
50
91.7
[1,333.42
.0178
ADC Mean
(910-6 mm2/s) 10 % (A10)
0.707 (0.564–0.825)
50
97.2
[797
.0250
(910-6 mm2/s) Unless otherwise indicated, data in parentheses are 95 % confidence interval AUC area under the receiver operator characteristic curve, CI confidence interval
50 % (A50)
0.690 (0.547–0.825)
43.7
91.7
[1,299 (910
-6
.0256 2
mm /s)
nCBV 90 % (C90)
0.713 (0.570–0.830)
56.2
88.9
B2.04
.0195
Slope [(C90-C10)/0.8]
0.732 (0.591–0.845)
56.2
91.7
B1.99
.0069
mutation in HGGs, which can be helpful for the evaluation of the tumor heterogeneity. There have been many reports of glioma grading using perfusion-weighted imaging, although only a few studies have dealt with the histogram method to [14, 26, 28, 29]. Emblem et al. [26] reported that the maximum peak height of the nCBV histogram was lower and exhibited a wider distribution of pixel values in HGGs, which represented the
tumor heterogeneity in HGGs. The 99th percentile value on their cumulative nCBV histogram displayed higher diagnostic accuracy for differentiating high- from low-grade than the peak height [9]. Our results suggest that the C90 value can be used to differentiate IDH1P from IDH1N. In addition, the slope between the C10 and C90 values had a better diagnostic performance than other slope values. These results suggest that IDH1P tumors are composed of
123
148
more heterogeneous microenvironment. However, in terms of number of pixels from the whole tumor segmentation, ADC tends to have more pixels than DSC-PWI imaging, which we believe makes ADC reflect tumor heterogeneity better than DSC-PWI. In addition, we used gradient-echo DSC-PWI imaging for evaluation of nCBV. Gradient-echo DSC may not be sensitive/specific to IDH1P as it is biased more towards larger vessels and spin-echo DSC may be sensitive to smaller vessels to show more difference between IDH1P and IDH1N. In the previous report [12], the conventional MR imaging with the GBM with IDH1P showed the larger tumor size, the presence of large nonenhancing regions, and more frequent involvement of frontal lobe. We believe that the histogram method based on the whole tumor segmentation method in the present study can be used for the objective evaluation of the microenviroment of tumors; thus, our study shows IDH1P tumors tend to be more heterogeneous that IDH1N ones from the analysis of ADC and DSC-PWI. In the present study, we used nCBV maps, which were normalized by dividing every rCBV value in a specific section by the rCBV value in the unaffected white matter, to minimize variances in the rCBV value in an individual patient. In previous reports [17, 30], a disparity in rCBV estimates was found among and between acquisition and postprocessing methods, because of the presence of contrast leakage effect, especially high-grade tumors. However, in our institute, nCBV measurement is more feasible and well-established for the brain tumor imaging. Even though rCBV maps corrected for contrast agent extravasation has several benefits for the evaluation of the brain tumors, nCBV measurement is also one of popular methods in that purpose [9, 10]. Additionally, in previous reports [6, 31], there is a striking difference in the age distribution of patients with primary and secondary glioblastoma, and similarly with or without IDH1/2 gene mutations. The patient with IDH1P had a very high frequency of TP53 mutation and a very low frequency of mutations in other commonly altered GBM genes. This association between IDH1P and improved survival was noted in young patients with TP53 mutations. Apart from the intrinsic limits of any retrospective study, several other limitations of our study should be mentioned. First, we used two different 3-T MR units, which differed in terms of their imaging parameters. However, we optimized the sequences to minimize the differences between the two units. We believe that there might have been a small bias in terms of image analysis of the ADC maps; the IDH1P and IDH1N groups presented similar MR imaging in the two units. Second, the present study included a relatively small patient population. In our study, the incidence of GBM with IDHP is relatively
123
J Neurooncol (2015) 121:141–150
higher than other study. Because of retrospectively designed study, there might be selection bias due to inclusion and exclusion criteria. Further investigation of a larger population is warranted. Third, tumor boundary was defined with reference to high signal intensity on T2WI, and tumor infiltration and peritumoral edema were included in the regions of interests. However, the differentiation between these two components is impossible in imaging studies. In addition, we tried to evaluate the tumor heterogeneity related with IDH mutation, and we believe that T2 signal intensity is better segmentation reference that other imaging methods such as contrastenhanced imaging for the present study, because T2 signal intensities include various components such as tumorous condition and edema, which are also results of tumor heterogeneity. Fourth, the normal-appearing white matter as a reference region for the calculation of nCBV may be affected by, for example, radiation therapy and tumor invasion [11]. However, we excluded the patient who underwent CCRT prior to the imaging, and tumor invasion was evaluated on the conventional MR sequences as best as we could manage. Finally, for the detection of IDH gene mutation, MRS is getting more and more popular to detect 2-HG, the product of the mutated IDH, which is specific biomarker for IDH gene mutation. However, as mentioned above, the aim of our study was to investigate whether the IDHP tumors have more heterogeneity than negative ones, which was shown by DWI and DSC-PWI. In addition, our study results based on these physiologic MR imaging show that MR imaging can also predict GBM genetic and cellular biologic features and imply physiologic MR imaging may be useful in targeting specific areas within heterogeneous tumor tissues. In conclusion, our results suggest that histogram analysis of ADC and nCBV maps based on the entire tumor volume can be a useful tool for distinguishing IDH1P and IDH1N. The cumulative ADC and nCBV histograms can predict that IDHP tumors have a more heterogeneous microenvironment, because of their stepwise tumorigenesis. Those modalities also can reflect the microenvironmental heterogeneity of the gliomas. Application of advanced imaging modalities in HGGs can predict the prognosis of IDH1P HGG patients and is recommended as part of the MR study of suspected HGGs. Acknowledgments This study was supported by a grant from the National R&D Program for Cancer Control, Ministry of Health & Welfare, Republic of Korea (1120300), the Korea Healthcare technology R&D Projects, Ministry for Health, Welfare & Family Affairs (A112028 and HI13C0015), and the Research Center Program of IBS (Institute for Basic Science) in Korea. Conflict of interest interest.
None of the authors have any conflict of
J Neurooncol (2015) 121:141–150
References 1. Daumas-Duport C, Scheithauer B, O’Fallon J, Kelly P (1988) Grading of astrocytomas. A simple and reproducible method. Cancer 62:2152–2165 2. Nikiforova MN, Hamilton RL (2011) Molecular diagnostics of gliomas. Arch Pathol Lab Med 135:558–568 3. Balss J, Meyer J, Mueller W, Korshunov A, Hartmann C, von Deimling A (2008) Analysis of the IDH1 codon 132 mutation in brain tumors. Acta Neuropathol 116:597–602 4. Yan H, Parsons DW, Jin GL, McLendon R, Rasheed BA, Yuan WS, Kos I, Batinic-Haberle I, Jones S, Riggins GJ, Friedman H, Friedman A, Reardon D, Herndon J, Kinzler KW, Velculescu VE, Vogelstein B, Bigner DD (2009) IDH1 and IDH2 mutations in gliomas. New Engl J Med 360:765–773 5. Hartmann C, Hentschel B, Wick W, Capper D, Felsberg J, Simon M, Westphal M, Schackert G, Meyermann R, Pietsch T, Reifenberger G, Weller M, Loeffler M, von Deimling A (2010) Patients with IDH1 wild type anaplastic astrocytomas exhibit worse prognosis than IDH1-mutated glioblastomas, and IDH1 mutation status accounts for the unfavorable prognostic effect of higher age: implications for classification of gliomas. Acta Neuropathol 120:707–718 6. Parsons DW, Jones S, Zhang X, Lin JC, Leary RJ, Angenendt P, Mankoo P, Carter H, Siu IM, Gallia GL, Olivi A, McLendon R, Rasheed BA, Keir S, Nikolskaya T, Nikolsky Y, Busam DA, Tekleab H, Diaz LA Jr, Hartigan J, Smith DR, Strausberg RL, Marie SK, Shinjo SM, Yan H, Riggins GJ, Bigner DD, Karchin R, Papadopoulos N, Parmigiani G, Vogelstein B, Velculescu VE, Kinzler KW (2008) An integrated genomic analysis of human glioblastoma multiforme. Science 321:1807–1812 7. Sanson M, Marie Y, Paris S, Idbaih A, Laffaire J, Ducray F, El Hallani S, Boisselier B, Mokhtari K, Hoang-Xuan K, Delattre JY (2009) Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol 27:4150–4154 8. Andronesi OC, Rapalino O, Gerstner E, Chi A, Batchelor TT, Cahil DP, Sorensen AG, Rosen BR (2013) Detection of oncogenic IDH1 mutations using magnetic resonance spectroscopy of 2-hydroxyglutarate. J Clin Invest 123:3659–3663 9. Kim H, Choi SH, Kim JH, Ryoo I, Kim SC, Yeom JA, Shin H, Jung SC, Lee AL, Yun TJ, Park CK, Sohn CH, Park SH (2013) Gliomas: application of cumulative histogram analysis of normalized cerebral blood volume on 3 T MRI to tumor grading. PLoS ONE 8(5):e63462. doi:10.1371/journal.pone.0063462 10. Jung SC, Choi SH, Yeom JA, Kim JH, Ryoo I, Kim SC, Shin H, Lee AL, Yun TJ, Park CK, Sohn CH, Park SH (2013) Cerebral blood volume analysis in glioblastomas using dynamic susceptibility contrast-enhanced perfusion MRI: a comparison of manual and semiautomatic segmentation methods. PLoS ONE 8(8):e69323. doi:10.1371/journal.pone.0069323 11. Thomsen H, Steffensen E, Larsson EM (2012) Perfusion MRI (dynamic susceptibility contrast imaging) with different measurement approaches for the evaluation of blood flow and blood volume in human gliomas. Acta Radiol 53:95–101 12. Carrillo JA, Lai A, Nghiemphu PL, Kim HJ, Phillips HS, Kharbanda S, Moftakhar P, Lalaezari S, Yong W, Ellingson BM, Cloughesy TF, Pope WB (2012) Relationship between tumor enhancement, edema, IDH1 mutational status, MGMT promoter methylation, and survival in glioblastoma. Am J Neuroradiol 33:1349–1355 13. Kang Y, Choi SH, Kim YJ, Kim KG, Sohn CH, Kim JH, Yun TJ, Chang KH (2011) Gliomas: histogram analysis of apparent diffusion coefficient maps with standard- or high-b-value diffusionweighted MR imaging—correlation with tumor grade. Radiology 261:882–890
149 14. Garzon B, Emblem KE, Mouridsen K, Nedregaard B, DueTonnessen P, Nome T, Hald JK, Bjornerud A, Haberg AK, Kvinnsland Y (2011) Multiparametric analysis of magnetic resonance images for glioma grading and patient survival time prediction. Acta Radiol 52:1052–1060 15. Rosen BR, Belliveau JW, Vevea JM, Brady TJ (1990) Perfusion imaging with NMR contrast agents. Magn Reson Med 14:249–265 16. Ostergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR (1996) High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. 1. Mathematical approach and statistical analysis. Magnet Reson Med 36:715–725 17. Boxerman JL, Schmainda KM, Weisskoff RM (2006) Relative cerebral blood volume maps corrected for contrast agent extravasation significantly correlate with glioma tumor grade, whereas uncorrected maps do not. Am J Neuroradiol 27:859–867 18. Wetzel SG, Cha S, Johnson G, Lee P, Law M, Kasow DL, Pierce SD, Xue X (2002) Relative cerebral blood volume measurements in intracranial mass lesions: interobserver and intraobserver reproducibility study. Radiology 224:797–803 19. Tozer DJ, Jager HR, Danchaivijitr N, Benton CE, Tofts PS, Rees JH, Waldman AD (2007) Apparent diffusion coefficient histograms may predict low-grade glioma subtype. NMR Biomed 20:49–57 20. Delong ER, Delong DM, Clarkepearson DI (1988) Comparing the areas under 2 or more correlated receiver operating characteristic curves—a nonparametric approach. Biometrics 44:837–845 21. Hartmann C, Meyer J, Balss J, Capper D, Mueller W, Christians A, Felsberg J, Wolter M, Mawrin C, Wick W, Weller M, HeroldMende C, Unterberg A, Jeuken JWM, Wesseling P, Reifenberger G, von Deimling A (2009) Type and frequency of IDH1 and IDH2 mutations are related to astrocytic and oligodendroglial differentiation and age: a study of 1,010 diffuse gliomas. Acta Neuropathol 118:469–474 22. Ichimura K, Pearson DM, Kocialkowski S, Backlund LM, Chan R, Jones DTW, Collins VP (2009) IDH1 mutations are present in the majority of common adult gliomas but rare in primary glioblastomas. Neuro-Oncology 11:341–347 23. Watanabe T, Nobusawa S, Kleihues P, Ohgaki H (2009) IDH1 mutations are early events in the development of astrocytomas and oligodendrogliomas. Am J Pathol 174:1149–1153 24. Pope WB, Prins RM, Albert Thomas M, Nagarajan R, Yen KE, Bittinger MA, Salamon N, Chou AP, Yong WH, Soto H, Wilson N, Driggers E, Jang HG, Su SM, Schenkein DP, Lai A, Cloughesy TF, Kornblum HI, Wu H, Fantin VR, Liau LM (2012) Non-invasive detection of 2-hydroxyglutarate and other metabolites in IDH1 mutant glioma patients using magnetic resonance spectroscopy. J Neurooncol 107:197–205 25. Choi C, Ganji SK, DeBerardinis RJ, Hatanpaa KJ, Rakheja D, Kovacs Z, Yang XL, Mashimo T, Raisanen JM, Marin-Valencia I, Pascual JM, Madden CJ, Mickey BE, Malloy CR, Bachoo RM, Maher EA (2012) 2-hydroxyglutarate detection by magnetic resonance spectroscopy in IDH-mutated patients with gliomas. Nat Med 18:624–629 26. Emblem KE, Nedregaard B, Nome T, Due-Tonnessen P, Hald JK, Scheie D, Borota OC, Cvancarova M, Bjornerud A (2008) Glioma grading by using histogram analysis of blood volume heterogeneity from MR-derived cerebral blood volume maps. Radiology 247:808–817 27. Tan WL, Huang WY, Yin B, Xiong J, Wu JS, Geng DY (2014) Can diffusion tensor imaging noninvasively detect IDH1 gene mutations in astrogliomas? A retrospective study of 112 cases. AJNR Am J Neuroradiol 35:920–927. doi:10.3174/ajnr.A3803 28. Emblem KE, Scheie D, Due-Tonnessen P, Nedregaard B, Nome T, Hald JK, Beiske K, Meling TR, Bjornerud A (2008) Histogram
123
150 analysis of mr imaging-derived cerebral blood volume maps: combined glioma grading and identification of low-grade oligodendroglial subtypes. Am J Neuroradiol 29:1664–1670 29. Young R, Babb J, Law M, Pollack E, Johnson G (2007) Comparison of region-of-interest analysis with three different histogram analysis methods in the determination of perfusion metrics in patients with brain gliomas. J Magn Reson Imaging 26:1053–1063 30. Paulson ES, Schmainda KM (2008) Comparison of Dynamic susceptibility-weighted contrast-enhanced mr methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 249:601–613
123
J Neurooncol (2015) 121:141–150 31. Jiao YC, Killela PJ, Reitman ZJ, Rasheed BA, Heaphy CM, de Wilde RF, Rodriguez FJ, Rosemberg S, Oba-Shinjo SM, Marie SKN, Bettegowda C, Agrawal N, Lipp E, Pirozzi CJ, Lopez GY, He YP, Friedman HS, Friedman AH, Riggins GJ, Holdhoff M, Burger P, McLendon RE, Bigner DD, Vogelstein B, Meeker AK, Kinzler KW, Papadopoulos N, Diaz LA, Yan H (2012) Frequent ATRX, CIC, FUBP1 and IDH1 mutations refine the classification of malignant gliomas. Oncotarget 3:709–722
