RADIOLOGY

doi:10.1093/neuonc/not189

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-001. MRI CHANGES ASSOCIATED WITH BEVACIZUMAB DIFFER BETWEEN TUMOR RECURRENCE AND CEREBRAL RADIATION NECROSIS

Yoshiki Arakawa ¹, Ko-ichi Fujimoto ¹, Daiki Murata ¹, Yuji Nakamoto ², Tomohisa Okada ², Susumu Miyamoto ¹

Abstract

OBJECTIVE: Cerebral radiation necrosis occurs after radiotherapy in brain tumors. However, its etiology still remains unclear and the treatment has not been established. Recently, it has been reported that bevacizumab（BEV）is a therapeutic option for cerebral radiation necrosis. In this study, we investigate the changes of MRI after BEV in patients with malignant glioma followed by cerebral radiation necrosis. PATIENTS AND METHODS: 5 patients with malignant glioma had been treated with radiation therapy. After radiation therapy, they were diagnosed as cerebral radiation necrosis with biopsy or MRI and received BEV 5mg/kg every 2 week. The changes of MRI after BEV were analyzed before and after BEV treatment. RESULTS: The use of BEV improved the symptom due to radiation necrosis. Analysis of MRI showed the decreasing volumes of their enhanced lesion and hyper intensity lesion on FLAIR after BEV treatment. However, changes of the apparent diffusion coefficient (ADC) value are different within the lesion. Tumor recurrences were detected in the locus with stable ADC value. CONCLUSION: BEV improves the necrotic lesion and clinically the symptom in the treatment of patients with cerebral radiation necrosis. The BEV-induced changes on MRI are involved in the decreasing neovascularity and interstitial edema. Therefore, MRI changes associated with bevacizumab may distinguish between tumor recurrence and cerebral radiation necrosis.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-002. BASIC IMAGING MARKER DEPICT A REGIONALLY INCREASED BEVACIZUMAB ACTIVITY IN RECURRENT GLIOBLASTOMA ASSOCIATED WITH SUBSTANTIALLY PROLONGED SURVIVAL

Oliver Bähr ¹, Patrick N Harter ², Lutz Weise ³, Se-Jong You ⁴, Michael W Ronellenfitsch ¹, Johannes Rieger ¹, Joachim P Steinbach ¹, Elke Hattingen ⁴

Abstract

BACKGROUND: Several studies have addressed the significance of radiographic changes under bevacizumab treatment. Bevacizumab-induced lesions with restricted diffusion as well as lesions with a hyperintense signal on T1 sequences have been described and both are associated with favorable outcome. But the origin and significance of these changes remain unclear. METHODS: We screened a cohort of 74 bevacizumab (BEV) treated patients with recurrent glioblastoma for (i) lesions with restricted diffusion and low ADC values and/or (ii) lesions with a hyperintense signal on pre-contrast T1 sequences. We evaluated overall survival, histological analyses and radiographic patterns at recurrence. RESULTS: 25 of 74 patients (33.8%) developed new or significantly increased T1 hyperintense lesions in the tumor area. 35 of 74 patients (47.3%) developed lesions with restricted diffusion and low ADC values. In 21 of 74 patients (28.4%) the lesions were both hyperintense on T1 and had restricted diffusion with low ADC values. There were no substantial differences in patient characteristics. Median survival for patients with restricted diffusion was 9.4 months, with T1 hyperintense lesions 10.5 months and reached 13.0 months in those with combined lesions (p < 0.005). Patients with neither lesion had a median survival of 6.6 months. In 4 patients we were able to conduct histological examinations of the described lesions and always found extensive calcified necrosis. Further, the described combined lesions were only rarely involved in tumor progression while a blinded analysis of patterns of progression showed an increase of distant recurrences in these patients. CONCLUSION: This analysis provides robust imaging markers predictive of survival. Histological analyses support the hypothesis of a regionally increased bevacizumab activity with extensive calcified necrosis. Furthermore, lesions with restricted diffusion and hyperintense T1 signal are rarely the origin of progressive tumor.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-003. QUANTITATIVE T2-MAPPING OF RECURRENT GLIOBLASTOMA UNDER BEVACIZUMAB IMPROVES MONITORING FOR NON-ENHANCING TUMOR PROGRESSION AND PREDICTS OVERALL SURVIVAL

Oliver Bähr ¹, Alina Jurcoane ², Keivan Daneshvar ², Ulrich Pilatus ², Michel Mittelbronn ³, Joachim P Steinbach ¹, Elke Hattingen ²

Abstract

BACKGROUND: Anti-angiogenic treatment in recurrent glioblastoma patients suppresses contrast enhancement and reduces vasogenic edema while non-enhancing tumor progression is common. Thus, the importance of T2-weighted imaging is increasing. We therefore quantified T2-relaxation-times which are the basis for the contrast on T2-weighted images. METHODS: Conventional and quantitative MRI was performed in 18 patients with recurrent glioblastoma before treatment with bevacizumab and every 8 weeks thereafter until further tumor progression. We segmented the tumor on conventional MRI into the 3 subvolumes enhancing tumor, non-enhancing tumor and edema. Using co-registered quantitative maps, we followed changes of T2-relaxation-time in each subvolume. Moreover, we generated differential T2-maps by a voxel-wise subtraction using the first T2-map under bevacizumab as reference. RESULTS: Visually segmented areas of tumor and edema did not differ in T2-relaxationtimes. The non-enhancing tumor volume did not decrease after commencing bevacizumab treatment while, strikingly, increased at progression. Differential T2-maps clearly showed non-enhancing tumor progression in previously normal brain. T2-relaxation-times decreased under bevacizumab without re-increasing at tumor progression. A decrease of less than 26 ms in the enhancing tumor following exposure to bevacizumab was associated with longer overall survival. CONCLUSION: Combining quantitative MRI and tumor segmentation improves monitoring of glioblastoma patients under bevacizumab. The degree of the T2-relaxation-time change under bevacizumab may be an early response parameter predictive of overall survival. The sustained decrease of T2-relaxation-times towards values of healthy tissue masks progressive tumor on conventional T2-weighted images. Therefore, quantitative T2-relaxation-times may detect non-enhancing progression better than conventional T2-weighted imaging.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-004. CORRELATION OF ADC MEAN AND ADC MEAN RATIO IN RADIOGRAPHIC PSEUDOPROGRESSION FOLLOWING TREATMENT WITH ERC 1671 IMMUNOTHERAPY IN RECURRENT GLIOBLASTOMA

Jose Carrillo ¹, Daniela Bota ¹, Jason Handwerker ¹, Lydia Min-Ying Su ¹, Thomas Chen ², Apostolos Stathopoulos ³, Hon Yu ¹

Abstract

BACKGROUND: Differences in ADC maps can characterize tumor progression. Lower ADC ratios and mean ADC values correspond with increased tumor cellularity consistent with tumor recurrence, while higher mean ADC ratios are in keeping with non-recurrence. Radiographic pseudoprogression is seen in 20-40% of GBM treated with XRT/TMZ, however, the presence of pseudoprogression has not been well established in response to vaccine therapy. METHODS: This retrospective study analyzed the MRI features of ERC 1671 vaccine treated recurrent GBM at time of second progression, following initial surgical resection with Stupp protocol, and bevacizumab therapy at first progression. MR imaging consisting of axial pre- and post-contrast T1-weighted, T2-weighted, FLAIR, and DWI sequences, and evaluated from day 0 to day 206 post-vaccine. Calculated ADC maps were analyzed and ROIs were drawn manually onto ADC maps in areas of corresponding T1-weighted image enhancement. Normalization was obtained using corresponding contralateral ROIs on ADC maps, and used to calculate mean ADC and ADC ratios. Mean ADC ratio and mean ADC values were compared using a one sample t-test. RESULTS: Progressively increased enhancement on MRI prompted tumor biopsy on day 257, revealing low Ki-67, widespread focal necrosis, collagenization of blood vessels, permeation by macrophages, reactive astrocytosis, and residual glioma of uncertain growth potential. At day 0 of vaccine therapy the ADC mean ratio was 1.01 and steadily increased to 2.42 by day 206 of vaccine therapy, p = 0.043. At day 0 of vaccine therapy the mean ADC was 861.7 and increased to 1712.7 by day 206 of vaccine therapy, p = 0.05. CONCLUSIONS: ERC 1671 vaccine treatment generated MR imaging enhancement changes concerning for tumor progression, which were consistent with treatment effects upon biopsy. After initiation of vaccine treatment the mean ADC ratios increased significantly, consistent with tumor non-recurrence. The progressive enhancement seen on MR imaging in patients treated with immunotherapy may represent vaccine-related pseudoprogression.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-005. PET IMAGING OF BRAIN GLIOMAS WITH C-11 ACETATE: COMPARISON WITH F-18 FDG PET

Jong Hee Chang ¹, Eui Hyun Kim ¹, Se Hoon Kim ¹, Mi ¹, Jin Yun ¹

Abstract

INTRODUCTION: The purpose of this study is to evaluate the clinical value of C-11 acetate(ACE) positron emission tomography(PET), compared with 18F-fluorodeoxyglucose(FDG) PET, in patients with a brain glioma. METHODS: Forty patients with a glioma were prospectively examined by C-11 acetate PET and F-18 FDG PET. Six patients had grade II, 11 had grade III, and 23 had grade IV gliomas. They underwent PET with both FDG and ACE on separate days. Images were analyzed by the tumor-to-cortex ratio (T/C ratio). The tumor uptake was visually scored into 3 grades as compared with that of the contralateral cortex: low (similar to or lower than contralateral cortex), moderate (slightly increased), and intense (markedly increased). RESULT: In grade IV, all lesions showed intense ACE uptake. Seventy four percent (17/23) of grade IV gliomas had intense FDG uptake and 26% (6/23) had moderate FDG uptake. In grade III, all lesions had moderate ACE uptake and there were 27% (3/11) with intense FDG uptake, 64% (7/11) with moderate FDG uptake, and 9%(1/11) with low FDG uptake. In grade II gliomas, all lesions had low ACE and FDG uptake. The mean ACE T/C ratio in grade IV gliomas (2.91 ± 0.89) was significantly higher than that in both grade III gliomas(1.56 ± 0.37; p = 0.001) and grade II gliomas(0.83 ± 0.31; p = 0.002). The mean FDG T/C ratio in grade IV gliomas (1.27 ± 0.42) was significantly higher than that in grade II gliomas(0.57 ± 0.12; p = 0.016), but not in grade III gliomas(1.00 ± 0.59; p = 0.252). FDG PET was not reliable in differentiating grade III gliomas from grade IV gliomas. CONCLUSION: FDG PET and ACE PET seem to be equal in detecting high grade gliomas. However, ACE PET was better in differentiating grade III from IV tumors. Both FDG PET and ACE PET had no positive uptake in grade II gliomas which was well correlated with biological behavior of tumor.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-006. AUTOPSY AND MAGNETIC RESONANCE IMAGING CORRELATES IN TWO GLIOBLASTOMA PATIENTS ON BEVACIZUMAB

Peter Pytel ¹, John Collins ¹, Yoon Choi ¹, Rimas Lukas ¹, Martin Nicholas ¹

Abstract

INTRODUCTION: Glioblastoma (GBM) is an infiltrating tumor but its radiographic and pathologic growth pattern vary widely, ranging from fairly circumscribed to extensive gliomatosis-like dissemination of tumor cells. Bevacizumab is a monoclonal antibody directed against vascular endothelial growth factor that has had a significant impact on the appearance of brain magnetic resonance imaging (MRI), further complicating the interpretation of MRI for clinical purposes. Changes in diffusion weighted images (DWI) and apparent diffusion coefficient (ADC) maps have been suggested to represent areas of non-enhancing tumor growth in Bevacizumab-exposed patients. Here, we report correlations between brain MRI findings and autopsy specimens from two patients with recurrent GBM, both exposed to Bevacizumab prior to death. MATERIALS AND METHODS: Autopsy specimens and MR images were carefully registered. The pathologist sampled areas predicted by the radiologist to contain significant tumor burden based on changes in patterns of enhancement, DWI signal abnormalities, or both. The entire neuraxis was further sampled without regard for radiographic findings and imaging correlates made. RESULTS: In most areas from both autopsies, there was a strong correlation between regions of brain predicted by MRI changes to contain tumor. In one autopsy, tumor was fairly localized to those regions of brain suggested by imaging findings. However, in the other, diffuse infiltration by tumor (gliomatosis) was seen in regions of brain both with and without suggestive MRI findings. CONCLUSIONS: These findings generally support the evolving impression that changes in DWI and ADC are reflective of non-enhancing tumor in GBM patient receiving Bevacizumab. However, concordance is not complete: one DWI-negative newly-enhancing mass that could have represented either tumor or necrosis was tumor-dense. Furthermore, these findings do not support the use of DWI and ADC mapping in predicting more subtle gliomatosis-like changes.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-007. DIFFUSION IMAGING GENOMIC MAPPING IDENTIFIES GENOMIC TARGETS INVOLVED IN INVASION AND POOR PROGNOSIS

Rivka Colen ¹, Ryan Jafrani ¹, Pascal Zinn ¹

Abstract

PURPOSE: To create an imaging genomic map, linking MR imaging traits with gene- and miRNA expression profiles, in patients with GBM to determine genomic correlates of a MR diffusion radiophenotype to possibly find new genomic targets for GBM treatment. Decreased in diffusion in tumors, specifically GBM, is associated with increase in cellular density and high nuclear to cytoplasm (N:C) ratio. Here, we present the first study examining in a quantitative way diffusion imaging genomics in GBM to determine novel and targetable genomic biomarkers in GBM. METHOD AND MATERIALS: We identified 60 treatment-naive GBM patients from The Cancer Genome Atlas (TCGA) who had both gene- and microRNA expression profiles and pretreatment MR-neuroimaging specifically ADC maps. Two neuroradiologists did morphological analysis of the studies using Slicer 3.6 (slicer.org), where regions of FLAIR hyperintensity, contrast enhancement, and necrosis were segmented to obtain accurate tumor volumetric data. Nordic ICE was used for functional diffusion analysis using the region-of-interest (ROI) and volume segmentation methods. ADC values were calculated. VOIs corresponding to the different FLAIR and post-gad T1 segmented volumes were also analyzed for their median ADC values. Biostatistical analysis was performed for gene and miRNA sets, where median ADC value and histogram cutoffs for each morphological region were defined to separate high from low groups and analyzed by Comparative Marker Selection. All the genomic data was also analyzed to determine the most upregulated mRNAs/miRNAs using ingenuity pathway analysis (IPA). RESULTS: IPA identified molecular networks, as well as canonical and functional pathways highly associated with cancer and invasion in those patients with low ADC values (areas of restricted diffusion). CONCLUSION: The diffusion radiophenotype identified genes and miRNAs and corresponding molecular networks that were highly associated with tumor invasion. The uncovered genes and miRNAs represent new insight into tumors with restricted diffusion seen on MRI.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-008. IMAGING GENOMIC BIOMARKER SIGNATURE FOR MGMT PROMOTOR METHYLATION IDENTIFICATION

Rivka Colen ¹, Omar Ashour ¹, Pascal Zinn ¹

Abstract

PURPOSE: To create an imaging biomarker signature in order to identify those Glioblastoma (GBM) patients with MGMT promoter methylation. METHOD AND MATERIALS: Using The Cancer Genome Atlas (TCGA), we identified 78 treatment-naïve GBM patients whom had both gene- and microRNA expression profiles and pretreatment MR-neuroimaging from the Cancer Imaging Archive. The 3D Slicer software 3.6 (http://www.slicer.org) was used for image analysis and image review was done in consensus by 2 neuroradiologists. Fluid-Attenuated Inversion Recovery (FLAIR) was used for segmentation of the edema and post-contrast T1 weighted imaging (T1WI) for segmentation of enhancement (defined as tumor) and necrosis. Quantitative perfusion parameters where obtain using the region of interest (ROI) method (NordicICE). ROIs were placed in the previously segmented regions of contrast enhancement, necrosis, and non-enhancing perilesional FLAIR hyperintensity- the latter reflecting a mixture of edema/tumor infiltration. Imaging parameters were then correlated with the MGMT status and gene expression profiles. Complex biomarker signatures based on profiling and survival were created. RESULTS: An imaging biomarker signature was created using multiple parameters, including the MR perfusion parameter of rCBV. Multiple parameters were associated with overall survival. An increase in rCBV in the non-enhancing FLAIR hyperintense portion was associated with the strongest survival difference (p< 0.03). CONCLUSION: An imaging biomarker signature using conventional MRI parameters and also advance parameters helps predict MGMT methylation status and expression. The identification of a non-invasive imaging biomarker signature as a surrogate for MGMT promotor methylation can help stratify patients in therapy and predict response versus nonresponse to therapy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-009. IMAGING GENOMIC CORRELATION OF INVASIVE GENOMIC COMPOSITION AND PATIENT SURVIVAL DEMONSTRATES METABOLIC DYSFUNCTION: A TCGA GLIOMA PHENOTYPE RESEARCH GROUP PROJECT

Rivka Colen ¹, Mark Vangel ², David Gutman ³, Scott Hwang ³, Max Wintermark ⁴, Rajan Jain ⁵, Manal Jilwan-Nicolas ⁴, James Chen ^6,⁷, Prashant Raghavan ⁴, Chad Holder ³, Daniel Rubin ⁸, Eric Huang ⁹, Justin Kirby ⁹, John Freymann ⁹, Carl Jaffe ¹⁰, Adam Flanders ¹¹, Pascal Zinn ¹²

Abstract

PURPOSE: Invasion of tumor cells in the adjacent parenchyma is a major cause of treatment failure in glioblastoma. Furthermore, invasive tumors are shown to have a different genomic compositions and metabolic abnormalities that allow for a more aggressive GBM phenotype and resistance to therapy. We thus seek to identify those genomic abnormalities associated with an invasion GBM phenotype. Using establish qualitative imaging invasive parameters, we will perform an imaging genomic mapping screen. METHOD AND MATERIALS: We retrospectively identified 104 treatment- naïve glioblastoma patients from The Cancer Genome Atlas whom had gene expression profiles and corresponding MR imaging available in The Cancer Imaging Archive. The standardized VASARI feature-set criteria were used for the qualitative visual assessments of invasion; quantitative tumor volumetry was obtained using 3D slicer. Patients were assigned to classes based on the presence (Class A) or absence (Class B) of statistically significant invasion parameters to create an invasive imaging signature; imaging genomic analysis was subsequently performed. RESULTS: Our results show that patients with specific invasive imaging signatures have a significant decrease in overall survival and an associated distinctive genomic expression profile signature. Our invasive imaging biomarker signature classes, Class A (invasive phenotype) and Class B, demonstrated a significant survival disadvantage in Class A of 9.9 months (8.7 versus 18.6 months, p < 0.001). Importantly, Class A patients reflected tumor compositions which have genes involved in invasion and oxidative stress/mitochondrial biology. CONCLUSION: This study demonstrates for the first time the ability of imaging features to uncover and predict metabolic and mitochondrial dysfunction in GBM. This is important as the effect of the tumor microenvironment and mitochondrial dysfunction on cancer is increasingly being recognized as an important modulator of epigenetics and facilitator of cancer development.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-010. IMAGING GENOMIC IDH-1 BIOMARKER SIGNATURE

Rivka Colen ¹, Omar Ashour ¹, Pascal Zinn ¹

Abstract

PURPOSE: The IDH1(R132H) mutation is one of the strongest prognostic predictors and a diagnostic hallmark of gliomas and thus has major clinical relevance. Given the latter, it will likely be integrated into the new WHO classification for GBM. Thus, we seek to identify a imaging signature associated with IDH-1 mutated tumors. METHOD AND MATERIALS: We identified 60 GBM patients from The Cancer Genome Atlas (TCGA) who had both genetic- expression profiles of IDH and mutation status and neuroimaging. All morphological image analyses were done using slicer 3.6 (slicer.org) and functional analysis using NordicIce, and reviewed in consensus by 2 neuroradiologists. Fluid-Attenuated Inversion Recovery (FLAIR) was used for segmentation of the edema and post-contrast T1 weighted imaging (T1WI) for segmentation of enhancement (defined as tumor) and necrosis. Quantitative perfusion parameters where obtain using the region of interest (ROI) method. ROIs were placed in the previously segmented regions of contrast enhancement, necrosis, and non-enhancing perilesional FLAIR hyperintensity- the latter reflecting a mixture of edema/tumor infiltration. In order to create a imaging biomarker signature that will predict IDH1 mutated tumors, we divided patients into discovery and validation sets. Multiple imaging parameters were combined that predicted the mutation. This was then applied to the validation set. RESULTS: We created a complex imaging biomarker signature using conventional MR imaging and perfusion (rCBV) to predict those patients with IDH1 mutation and furthermore that was predictive of patient survival. Moreover, the addition of genomic markers such as MGMT, further increased prognostic robustness. CONCLUSION: IDH1 tumors hold a specific imaging biomarker signature that can be used as a predictive and prognostic biomarker and non-invasive surrogate for IDH1. MRI can be used as an imaging surrogate to identify those patients with IDH1 mutation. It can be used as a predictive and prognostic imaging biomarker for GBM patients.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-011. PERFUSION IMAGING GENOMIC MAPPING UNCOVERS POTENTIAL GENOMIC TARGETS INVOLVED IN ANGIOGENESIS AND INVASION

Rivka Colen ¹; TCGA Phenotype Group¹, Pascal Zinn ¹

Abstract

PURPOSE: To create an imaging genomic map, linking MR imaging traits with gene- and miRNA expression profiles, in patients with GBM to determine genomic correlates of a MR perfusion radiophenotype to possibly find new genomic targets for GBM treatment. Increases in angiogenesis demonstrate increases on MRI perfusion relative cerebral blood volume (rCBV) maps. Here, we present the first study examining in a quantitative way the perfusion imaging genomics in GBM to determine novel and targetable angiogenic biomarkers in GBM. METHOD AND MATERIALS: We identified 30 GBM patients from The Cancer Genome Atlas (TCGA) who had both genetic- expression profiles and neuroimaging. All morphological image analyses were done using slicer 3.6 and functional analysis using NordicICE, and reviewed in consensus by 2 neuroradiologists. Quantitative perfusion parameters where obtained using the region of interest (ROI) method. ROIs were placed in the previously segmented regions of contrast enhancement, necrosis, and non-enhancing perilesional FLAIR hyperintensity- the latter reflecting a mixture of edema/tumor infiltration. Biostatistics analysis was performed for gene and miRNA sets whereas the median CBV values of each of the segmented regions were taken as the cutoff to define high and low groups. These groups were then analyzed by Comparative Marker Selection (Broad Inst.). Among the whole gene set the most upregulated mRNAs/miRNAs, were analyzed with ingenuity pathway analysis (IPA). RESULTS: IPA identified molecular networks, as well as canonical and functional pathways highly associated with cancer, angiogenesis, and invasion in those patients with high tumor rCBV. The perfusion radiophenotype identified genes and miRNAs and corresponding molecular networks that were highly associated with angiogenesis and invasion. CONCLUSION: The perfusion radiophenotype identified genes and miRNAs and corresponding molecular networks that were highly associated with angiogenesis and invasion.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-012. NEW PROGNOSTIC SCORING SYSTEM FOR SINGLE BRAIN METASTASIS PATIENTS UTILIZING CLINICAL AND RADIOLOGICAL FACTORS

Saurabh Dahiya ¹, Volodymyr Statsevych ¹, Paul Elson ¹, Hao Xie ¹, Samuel Chao ¹, David Peereboom ¹, Glen Stevens ¹, Gene Barnett ¹, Manmeet Ahluwalia ¹

Abstract

BACKGROUND: Cancer patients with single brain metastasis (BM) have relatively favorable outcomes as compared to patients with multiple BM. Neuroimaging factors are currently not included in prognostic considerations in BM patients. METHODS: Cleveland Clinic Brain Tumor and Neuro-Oncology Center's database was used to identify was used to identify patients with single BM and pre-intervention MRI who were treated between 2000-11. MRI (T1contrast, T2 and FLAIR) were used to assess the location, size of BM, and peritumoral edema. Diffusion weighted imaging (DWI) and apparent diffusion coefﬁcient (ADC) maps were used to distinguish whether BM was restricted or not. Clinical data was obtained by chart review. MRI analyses were performed by investigators blinded to clinical data. RESULTS: 180 patients (88 male, 92 female, median age 60 years) were included in analysis. The primary cancer included 111 lung, 25 breast,12 melanoma and 32 renal cancer patients. Median cross sectional area of BM was 2-cm² and median peritumoral edema was 1.2-cms. Thirty seven percent had hypointense, 34% had isointense and 29% had hyperintense lesions on DWI. Sixteen percent had restricted lesions based on DWI and ADC maps. Median OS was 15.5 months (95%C.I.11.3-19.1). Multivariate analysis was used to identify prognostic factors and for scoring. Age <65(4 points), tumor area <1 cm² (4 points), brain as initial site of metastasis(7 points), >18 months interval from BM diagnosis to primary tumor diagnosis(7 points) and controlled primary tumor(10 points) were independent prognostic parameters for OS(Cox regression model, p<0.05). Median OS in favorable group (score>11, n = 42%) was 35.5 months as opposed to 9.9 months in unfavorable group (score < 11, n = 58%), p <.0001. Although peritumoral edema was significant for OS on univariate analysis, it was not prognostic on multivariate regression analysis. Other imaging parameters including DWI were not prognostic for OS. CONCLUSIONS: We propose a new scoring system for patients with single brain metastasis patients, which incorporates both clinical and radiological factors.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-013. DIFFUSION AND PERFUSION MAGNETIC RESONANCE IMAGING IN NEWLY DIAGNOSED GLIOBLASTOMA FOLLOWING TREATMENT WITH BEVACIZUMAB, TEMOZOLOMIDE AND HYPO-FRACTIONATED STEREOTACTIC RADIOTHERAPY (HFSRT)

Mariza Daras ¹, Sasan Karimi ¹, Lauren Abrey ¹, Juan Sanchez ¹, Kathryn Beal ¹, Philip Gutin ¹, Thomas Kaley ¹, Christian Grommes ¹, Denise Correa ¹, Anne Reiner ¹, Samuel Briggs ¹, Antonio Omuro ¹

Abstract

BACKGROUND: Evaluation of response to anti-angiogenic treatments in malignant gliomas has been challenging because of confounding effects on vascular permeability and FLAIR appearance; developing novel imaging biomarkers is of high interest. OBJECTIVE: We prospectively investigated the role of dynamic susceptibility contrast magnetic resonance perfusion weighted imaging (DSC-MRI) and diffusion weighted imaging (DWI) in predicting treatment outcomes in patients with newly diagnosed glioblastoma (GBM) enrolled in a phase II trial testing the combination of temozolomide, bevacizumab, and HFSRT. METHODS: All 40 enrolled GBM patients underwent DSC-MRI and DWI at baseline and every 2 months until progression. Relative cerebral blood volume (rCBV), relative peak height (rPH), peak signal recovery (PSR), and apparent diffusion coefficient (ADC) were measured using commercially available software (FuncTools 2.6.9, GE Healthcare). RESULTS: Median progression-free (mPFS) was 10.9 months and median overall survival (mOS), 18.6 months. A significant and progressive reduction in mean rCBV was noted over time: mean rCBV at baseline: 3.4 vs 2.3 after RT (p < 0.001), 1.5 after 2 months (p < 0.001), and 1.4 after 4 months (p < 0.001). Higher baseline ADC ratios were associated with poorer OS (HR: 4.0, 95% confidence interval [CI] 1.0-15.8, p = 0.05), but not PFS (MacDonald p = 0.84, RANO p = 0.76). This association was maintained even after excluding patients with a complete resection (HR 6.8, 95% CI 1.3-37.0, p = 0.03). Baseline rCBV, change in rCBV following HFSRT, rPH and PSR were not associated with differences in PFS or OS. CONCLUSIONS: DSC-MRI depicted early and nearly universal decreases in tumor perfusion throughout treatment, validating this technique as a powerful biomarker of anti-VEGF modulation. However, perfusion parameters were not predictive of clinical benefit. Conversely, baseline ADC was a robust prognostic marker in this trial; patients with low ADC tumors may be the best candidates for treatment with bevacizumab.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-014. THE DIAGNOSTIC ACCURACY OF NEURO-IMAGING TO DETECT INFILTRATIVE GLIOMA WITHIN THE BRAIN: A META-ANALYSIS BASED ON 1598 PATIENTS IN 58 PUBLICATIONS

Niels Verburg ¹, Friso Hoefnagels ¹, Petra Pouwels ², Ronald Boellaard ³, Frederik Barkhof ³, Otto Hoekstra ³, Pieter Wesseling ⁴, Jaap Reijneveld ⁵, Jan Heimans ⁵, Peter Vandertop ¹, Koos Zwinderman ⁶, hilip De Witt Hamer ¹

Abstract

BACKGROUND AND AIMS: Glioma imaging, used for diagnostics, treatment planning and follow-up, is currently based on T1-weighted MRI after gadolinium (T1+) for high-grade glioma (HGG) and T2/ FLAIR-weighted MRI for low-grade glioma (LGG). The diagnostic accuracy of these techniques for the delineation of gliomas may be less than optimal. In this study, we systematically review and compare the diagnostic accuracy of available imaging techniques, including MRI, CT, PET and ultrasound, for detecting glioma within normal brain. METHODS: A systematic search in MEDLINE and EMBASE identified the publications reporting on correlation of neuro-imaging and histopathology. Study selection for analysis was based on predefined quality criteria, including the STARD criteria. Summary area under the curve, false positive rate (1-specificity) and true positive rate (sensitivity), were determined for each imaging modality in low- and high-grade glioma patients using bayesian summary hierarchical ROC analysis. RESULTS: Included for analysis were 58 studies with 1598 patients and 4320 samples. The diagnostic accuracy between imaging modalities did not differ significantly based on the available data. The highest sensitivity for high-grade glioma (0.85 CI 0.61 – 0.97) and low-grade glioma (0.90 CI 0.05 – 1.0) was obtained by MR spectroscopy imaging; the highest specificity for high-grade glioma (0.90 CI 0.53-0.99) by PET using the FET tracer, and for low-grade glioma (0.94 CI 0.23 – 1.0) by MRI T2. CONCLUSION: Our data indicates that MRI T1+ and MRI T2/FLAIR are not the most accurate imaging modalities. MRSI and PET are promising advanced imaging modalities with possibly superior diagnostic accuracy to delineate glioma.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-015. ANALYSIS OF PERFUSION AND VOLUMETRIC MRI IN PROGRESSIVE GLIOBLASTOMA PATIENTS TREATED WITH CONTINUOUS DOSE-INTENSE TEMOZOLOMIDE

Heinrich Elinzano ¹, Fatemeh Kadivar ¹, Priyanka O Yadav ², Virginia L Breese ¹, Cynthia L Jackson ¹, John E Donahue ¹, Jerrold L Boxerman ¹

Abstract

Continuous dose-intense temozolomide (_cdiTMZ) may be effective for glioblastomas progressing during maintenance temozolomide (_mTMZ) by inhibiting angiogenesis. We assessed the anti-angiogenic effect of _cdiTMZ by measuring relative cerebral blood volume (_rCBV) and correlating changes in _rCBV and enhancing tumor volume with overall survival, MGMT methylation and IDH-1mutation. This prospective, nonrandomized study included 13 patients (7 males, 6 females; median age 65 years [44-83]; median KPS 80% [60-90]; median dexamethasone dose 4mg [2-12]) progressing on _mTMZ (median 4 cycles) by qualitative radiographic assessment at least 3 months (median 5 months) after standard chemoradiation. Volumetric and dynamic susceptibility-contrast perfusion MRI were performed after each _cdiTMZ cycle (50mg/m²/day × 28 days). 9/13 patients (70%) had progressive enhancing tumor volume and 7/13 patients (54%) had increasing _rCBV on _mTMZ. MGMT methylation (4 methylated, 4 unmethylated, 5 pending) and IDH-1 mutation (10 negative, 3 pending) were tested. After one _cdiTMZ cycle, 3/9 patients (33%) with progressive enhancing tumor had <50% decrease in enhancing tumor volume and 3/7 patients (43%) with increased perfusion had decreased _rCBV (2 with >50%, 1 with <50% decrease). Median time to progression after _cdiTMZ was 3 months (all responders). Median overall survival was longer for _rCBV responders than non-responders (17 vs. 13 months). Patients with >50% _rCBV decrease are still alive. MGMT methylation and IDH-1 mutation did not correlate with response. The most common adverse event was mild fatigue. 1/13 patients had severe lymphopenia. 11/13 patients discontinued _cdiTMZ due to further progression, whereas 2/13 patients with radiographically stable disease after >5 _cdiTMZ cycles remained stable despite discontinued treatment. _cdiTMZ appears to have anti-angiogenic effect in glioblastomas progressing on _mTMZ as manifested by decreased _rCBV. Whereas _rCBV tends to correlate with survival, MGMT methylation or IDH-1 mutation did not correlate with response, suggesting that perfusion MRI provides a useful prognostic imaging biomarker.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-016. DEFORMATION FIELD MORPHOMETRY FOR EARLY DETECTION OF MALIGNANT TRANSFORMATION IN LOW GRADE GLIOMAS

Benjamin Ellingson ¹, Whitney Pope ¹, Albert Lai ², Phioanh Nghiemphu ², Timothy Cloughesy ²

Abstract

PURPOSE: Clinical management of low grade gliomas (LGGs) varies widely and there is no consensus as to their treatment. Clinicians rely on subjective assessments of serial MR scans to get a sense of tumor behavior and growth rates. Therefore, there is a clear need for noninvasive imaging biomarkers that can quantify early risk for malignant transformation in order to optimize clinical management of LGGs. We hypothesize that early detection of malignant transformation may be possible by noninvasively quantifying subtle tissue deformations over time using a technique called Deformation Field Morphometry. METHODS: Ten WHO II gliomas that eventually underwent malignant transformation to either WHO III or WHO IV were enrolled in the current retrospective study. Serial defromation field morphometry was performed with respect to an arbitrary baseline time point, quantifying volumetric growth rates over time. RESULT: In patients with malignant transformation, deformation field morphomety maps of volumetric growth rate increased an average of 62 days before traditional radiographic transformation using a threshold of 10 uL/day maximum growth rate in T2 hyperintense regions. CONCLUSION: Deformation field morphometry is a novel method for quantifying subtle changes in LGGs and detecting early malignant transformation.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-017. PROLIFERATION RATE ESTIMATES DERIVED FROM SERIAL DIFFUSION MRI CORRELATES WITH ¹⁸F-FLT PET SUV VALUES IN RECURRENT GLIOBLASTOMA TREATED WITH BEVACIZUMAB

Benjamin Ellingson ¹, Whitney Pope ¹, Wei Chen ³, Johannes Czernin ³, Michael Phelps ³, Albert Lai ², Phioanh Nghiemphu ², Linda Liau ⁴, Timothy Cloughesy ²

Abstract

PURPOSE: Proliferation rate estimates from cell invasion, motility, and proliferation level estimate (CIMPLE) maps derived from fitting a novel spatiotemporal mathematical model to serial diffusion MR data have been shown to correlate with NMR spectroscopy, glioma grade, and PFS/OS during bevacizumab. These maps predict future contrast enhancement in approximately 30% of patients on bevacizumab, as suggested in a recent pilot study. Proliferation rates based on the time-rate-of-change in ADC values within a voxel may reflect proliferative potential. The current study examined the relationship between CIMPLE map estimates of proliferation rate and ¹⁸F-FLT PET SUV, a molecular marker of DNA synthesis, in order to test whether CIMPLE maps could spatially localize regions of proliferative tumor. METHODS: Fourteen patients with malignant gliomas who had at least three MRI scans and one ¹⁸F-FLT PET scan during bevacizumab therapy were enrolled. MR scans consisted of at least a T2w, post-contrast T1w, and diffusion MR scan with b = 0 and 1000s/mm². Diffusion MR scans were distortion corrected, then all follow-up MR scans were registered to the first post-treatment MR scans. CIMPLE maps were generated by fitting a voxel-wise spatiotemporal model of cell invasion and proliferation, assuming ADC a surrogate for cell density. A new matrix solution to the CIMPLE map algorithm was implemented for patients with >3 scans. ¹⁸F-FLT PET scans were acquired during the period of MR evaluation. PET SUV scans were co-registered to MR space for subsequent analyses. RESULTS: Regions with high proliferation rate on CIMPLE maps appeared generally colocalized to regions of ¹⁸F-FLT PET. Average proliferation rate within contrast-enhancing regions was highly correlated with average ¹⁸F-FLT PET SUV (Pearson's correlation coefficient, R² = 0.79, P < 0.001). CONCLUSION: Regions with elevated proliferation rate on CIMPLE maps derived from serial diffusion MR data appear to reflect regions undergoing rapid DNA synthesis as suggested by ¹⁸F-FLT PET.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-018. QUANTITATIVE PARAMETRIC MAPS OF EFFECTIVE INTERVOXEL CONTRAST AGENT DIFFUSIVE FLUX (D^eff) USING DCE-MRI AS A NOVEL IMAGE BIOMARKER IN HUMAN GLIOBLASTOMA

Benjamin Ellingson ¹, Kevin Leu ¹, Anh Tran ¹, Whitney Pope ¹, Albert Lai ², Phioanh Nghiemphu ², Robert Harris ¹, Davis Woodworth ¹, Timothy Cloughesy ²

Abstract

PURPOSE: Dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) is a well-established perfusion MRI technique used for characterizing vascular permeability in brain tumors. This model, however, does not account for contrast agent diffusion between adjacent voxels that is likely to occur during such long observation times. The purpose of the current study was to explore a method for quantifying maps of effective contrast agent diffusive flux (D^eff) in human GBM. METHODS: A total of 20 recurrent GBM patients were retrospectively examined. DCE-MRI data, diffusion, perfusion, ¹⁸F-FDG PET and/or MR spectroscopic data were available for comparison to maps of D^eff. RESULTS: D^eff was elevated and negative, consistent with active absorption of contrast agent into T2/FLAIR hyperintense voxels confirmed to contain non-enhancing tumor via structural MRI (architectural distortion) and MR spectroscopy (elevated Cho/NAA). Elevated T2/FLAIR regions consistent with edema had low D^eff. Contrast enhancing regions demonstrated positive, elevated D^eff, consistent with contrast agent extravasation and movement from vessels to surrounding tissues. CONCLUSION: Maps of D^eff represents a new MRI technique that may isolate non-enhancing tumor from edema within T2/FLAIR hyperintense regions, potentially via active absorption of contrast agent into glioma cells.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-019. SPATIAL CORRELATION BETWEEN APPARENT DIFFUSION COEFFICIENT AND ¹⁸F-FDOPA PET IN DIFFUSE GLIOMAS

Benjamin Ellingson ¹, Whitney Pope ¹, Kevin Leu ¹, Wei Chen ², Johannes Czernin ², Michael Phelps ², Albert Lai ³, Phioanh Nghiemphu ³, Linda Liau ⁴, Timothy Cloughesy ³

Abstract

PURPOSE: Amino acid transport is increased in malignant cells due to proliferation, transamination, transmethylation, and conversion of glutamine for fuel. As such, amino acid PET tracers, including ¹¹C-MET, ¹¹C-TYR, ¹⁸F-FET, and ¹⁸F-FDOPA, have shown to be useful for identifying active tumor in gliomas. The apparent diffusion coefficient (ADC) has been shown to be negatively correlated with tumor cellularity and provide insight into dense tumor regions. We hypothesized that ¹⁸F-FDOPA PET uptake will be negatively correlated with ADC within areas of suspected tumor. METHODS: We retrospectively examined 30 patients (WHO IV, n = 12; WHO III, n = 6; WHO II, n = 12) from our Neuro-Oncology database, including patients that had T2/FLAIR and diffusion MRI within 3 months of obtaining an ¹⁸F-FDOPA PET scan due to suspected tumor progression. ¹⁸F-FDOPA PET scans were normalized to the basal ganglia and registered to MRI space. T2/FLAIR hyperintense regions were segmented. The voxel-wise correlation between normalized 18F-FDOPA PET uptake and ADC values within T2/FLAIR regions were analyzed for each individual patient. Additionally, the median ¹⁸F-FDOPA PET uptake and median ADC in T2/FLAIR regions from each patient were used to evaluate the correlation across all patients. RESULTS: Regions of elevated ¹⁸F-FDOPA PET tended to be spatially localized to regions with low ADC, whereas regions with abnormal T2/FLAIR signal intensity with high ADC tended to have minimum PET uptake. A strong negative linear correlation was observed between ADC and ¹⁸F-FDOPA PET for all patients. Additionally, most patients showed voxel-wise trends suggesting very dense tumor (low ADC) regions had the highest variability in PET uptake. A strong negative linear correlation was also observed between median ¹⁸F-FDOPA PET uptake and median ADC within T2/FLAIR regions. CONCLUSION: ADC and ¹⁸F-FDOPA PET are strongly correlated in T2/FLAIR abnormal regions in patients with diffuse gliomas

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-020. USING A PROBABILITISTIC RADIOGRAPHIC BRAIN TUMOR ATLAS CONSTRUCTED FROM 500 GLIOBLASTOMA PATIENTS TO PREDICT INDIVIDUAL PATIENT "-OMIC" SIGNATURES AND TREATMENT RESPONSES

Benjamin Ellingson ¹, Dieter Enzmann ¹, Whitney Pope ¹, Albert Lai ², Phioanh Nghiemphu ², Linda Liau ³, Timothy Cloughesy ²

Abstract

PURPOSE: We previously constructed a probabilistic radiographic atlas of tumor locations for 500 glioblastoma patients mapped to a common stereotactic image space, then analyzed tumor locations and tumor volumes with respect to a variety of genomic, molecular, and response data. The purpose of the current study was to demonstrate how this atlas can be used to predict "-omic" signatures and treatment responses in individual glioblastoma patients. METHODS: A total of 507 glioblastoma patients were used to construct the Probabilistic Brain Tumor Atlas from our Neuro-Oncology database. T2/FLAIR, contrast enhancement, and necrotic regions were segmented after registration to MNI template space. A total of 20 patients with glioblastoma, not included in the atlas, were used for validation. T2/FLAIR and post-contrast T1w images from validation patients were registered to MNI space and tumors segmented. The distribution of tumor frequency of occurrence from subsets of the atlas reflecting population maps for different genotypes (e.g. MGMT promoter methylated tumors) were extracted for tumor locations in individual patients. The odds ratio and probabilities of a certain phenotype were calculated for each voxel within tumor masks of individual patients. RESULTS: Validation patients demonstrated >80% sensitivity and specificity for predicting MGMT promoter methylation and < 6 month PFS. Atlas results suggested patients lacking MGMT promoter methylation also tended to be IDH1 wild types and have the mesenchymal gene expression signature. CONCLUSION: Results suggest the Probabilistic Radiographic Brain Tumor Atlas may be useful for early prediction of "-omic" signatures and response to therapy in individual glioblastoma patients.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-021. TUMOR PERFUSION DURING BEVACIZUMAB AND IRINOTECAN IN RECURRENT GLIOBLASTOMA: A MULTIMODAL APPROACH

Marica Eoli ¹, Anna Luisa Di Stefano ², Domenico Aquino ¹, Alessandro Scotti ¹, Elena Anghileri ¹, Lucia Cuppini ¹, Elena Prodi ¹, Gaetano Finocchiaro ¹, Maria Grazia Bruzzone ¹

Abstract

Angiogenesis is a requirement for progression of glioblastoma (GBM) and Bevacizumab (Bev), an antibody directed to vascular endothelial growth factor (VEGF), was recently used to treat GBM. However in vivo modifications induced by treatment are under active investigation. To analyze tumor changes induced by Irinotecan (Ir) and Bev, we use two different methodologies: relative CBV variation (rCBV) and Difference Perfusion Maps (DPMs). METHODS: 42 recurrent GBM patients underwent Bev (10 mg/kg) and Ir (125 or 340 mg/m²) every 2 weeks and were followed up with a radiological protocol, including Dynamic Susceptibility Contrast MRI every 8 weeks. Radiological responses were assessed based on RANO criteria (Wen et al. 2012). Two methods were used to assess perfusion changes. In method A, relative CBV variation after 8 weeks of treatment was calculated through semi-automatic ROI placement in the same anatomic region as in baseline. In method B, relative CBV variations with respect to baseline values were calculated into the evolving tumor region by voxel-by-voxel difference. DPMs were created showing where rCBV significantly increased, decreased or remained unchanged. RESULTS: Method A showed a significant decrease of rCBV in patients with stable disease or partial response after 8 weeks of treatment (p = 0.01) while patients with tumor progression maintained elevated levels of rCBV (p = 0.38). Method B, based on DPMs, showed rCBV was increased in 35% (±16%) of tumor volume (8 week-increased blood volume, 8w-IBV) and was decreased in 17% (±14%) (8 week-decreased blood volume, 8w-DBV). Patients presenting 8w-IBV higher than 18% (first quartile) showed a significantly longer PFS (p = 0.045) and OS (p = 0.016). Using DPM we observed that early increase in global perfusion is related to better survival. Further studies will be necessary to confirm the potential of DPMs in predicting response to anti-angiogenic therapy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-022. MRI CHANGES ASSOCIATED WITH BEVACIZUMAB DIFFER BETWEEN TUMOR RECURRENCE AND CEREBRAL RADIATION NECROSIS

Koichi Fujimoto ¹, Yoshiki Arakawa ¹, Daiki Murata ¹, Yuji Nakamoto ², Tomohisa Okada ², Susumu Miyamoto ¹

Abstract

OBJECTIVE: Cerebral radiation necrosis occurs after radiotherapy in brain tumors. However, its etiology still remains unclear and the treatment has not been established. Recently, it has been reported that bevacizumab（BEV）is a therapeutic option for cerebral radiation necrosis. In this study, we investigate the changes of MRI after BEV in patients with malignant glioma followed by cerebral radiation necrosis. PATIENTS AND METHODS: 5 patients with malignant glioma had been treated with radiation therapy. After radiation therapy, they were diagnosed as cerebral radiation necrosis with biopsy or MRI and received BEV 5 mg/kg every 2 week. The changes of MRI after BEV were analyzed before and after BEV treatment. RESULTS: The use of BEV improved the symptom due to radiation necrosis. Analysis of MRI showed the decreasing volumes of their enhanced lesion and hyper intensity lesion on FLAIR after BEV treatment. However, changes of the apparent diffusion coefficient (ADC) value are different within the lesion. Tumor recurrences were detected in the locus with stable ADC value. CONCLUSION: BEV improves the necrotic lesion and clinically the symptom in the treatment of patients with cerebral radiation necrosis. The BEV-induced changes on MRI are involved in the decreasing neovascularity and interstitial edema. Therefore, MRI changes associated with bevacizumab may distinguish between tumor recurrence and cerebral radiation necrosis.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-023. THE USE OF DYNAMIC O-(2-¹⁸F-FLUOROETHYL)-L-TYROSINE PET IN THE DIAGNOSIS OF PATIENTS WITH PROGRESSIVE AND RECURRENT GLIOMA

Norbert Galldiks ^1,², Gabriele Stoffels ¹, Christian Filss ¹, Veronika Dunkl ^1,², Marion Rapp ³, Michael Sabel ³, Maximilian I Ruge ⁴, Roland Goldbrunner ⁵, Nadim J Shah ¹, Gereon R Fink ^1,², Heinz H Coenen ¹, Karl-Josef Langen ^1,⁶

Abstract

BACKGROUND: Since a number of studies have indicated that dynamic O-(2-[¹⁸F]fluoroethyl)-L-tyrosine (¹⁸F-FET) PET may be helpful for the distinction between high- and low-grade glioma, we evaluated the diagnostic performance of dynamic ¹⁸F-FET PET in patients with progressive/recurrent glioma. PATIENTS AND METHODS: 106 dynamic ¹⁸F-FET PET and conventional MRI scans of 100 glioma patients (primary WHO grading: grade II, n = 55; grade III, n = 19; grade IV, n = 26) (mean age, 50 ± 14 y) were retrospectively analyzed. The patients were consecutively sent with MRI findings suggestive of tumor progression/recurrence (i.e., based on RANO criteria) (91%) or inconclusive MRI findings (9%) to rule out tumor progression/recurrence. Maximum and mean tumor/brain ratios (TBR_max, TBR_mean) of ¹⁸F-FET uptake were determined (20-40 min post injection). Time-activity curves (TAC) of ¹⁸F-FET uptake in the tumors were generated and the time-to-peak (TTP) was calculated. Furthermore, TACs of each lesion were assigned to one of the following curve patterns: (I) constantly increasing ¹⁸F-FET-uptake without identifiable peak uptake; (II) ¹⁸F-FET-uptake peaking at a midway point (> 20-40 min) followed by a plateau; and (III) ¹⁸F-FET-uptake peaking early (≤ 20 min) followed by a constant descent. Diagnoses were confirmed histologically (93%) or by clinical follow-up (7%). Diagnostic accuracy of PET and MR imaging parameters for the detection of tumor progression/recurrence was evaluated by receiver-operating-characteristic (ROC) analyses. RESULTS: MRI and ¹⁸F-FET PET findings were concordant in 81%, discordant findings could be observed in 19%. The highest accuracy (95%) to diagnose tumor progression/recurrence was obtained when both a TBR_max ≥ 2.8 and TAC pattern II or III were present (AUC, 0.966 ± 0.02; sensitivity, 96%; specificity, 91%; P < 0.001). In contrast, the accuracy to diagnose tumor progression/recurrence using conventional MRI according to the RANO criteria was 82%. CONCLUSIONS: Dynamic ¹⁸F-FET PET may contribute significantly to the management of patients with tumor progression/recurrence.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-024. ATYPICAL IMAGING APPEARANCE OF AN INTRASPINAL NEUROCYTOMA

Nandita Guha-Thakurta ¹, Lauren Langford ¹

Abstract

INTRODUCTION: Neurocytomas are uncommon, typically well differentiated tumors of the central nervous system (CNS). They predominantly arise intracranially within the lateral ventricles, occur in young and middle-aged adults, and have no sex predilection. Herein we report the first case of a sacral nerve root neurocytoma and its management. REPORT: A 49-year-old man presented with a 7-month history of back pain. He had right S1 radiculopathy diagnosed on EMG. Magnetic resonance imaging (MRI) of the spine demonstrated an extramedullary enhancing lesion at the right S1 level with the primary imaging diagnostic consideration of a nerve sheath tumor. The patient subsequently underwent surgery with a final histopathological diagnosis of a neurocytoma of the S1 nerve root. CONCLUSIONS: Neurocytomas in the spinal cord are extremely rare. They appear to have a propensity for the male population and can present at ages ranging from 8 to 72 years. Majority of spinal cord neurocytomas present as an intramedullary cord lesion. Its occurrence along a sacral nerve root has not been described before and on such occasions it can be mistaken for a schwannoma. The treatment of choice is surgery with adjuvant radiotherapy advocated only in cases with anaplastic features and high MIB-1 labeling index.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-025. PREOPERATIVE GRADING OF DIFFUSE GLIOMAS WIH MULTIPARAMETRIC IMAGING: MRI, MRS AND [¹⁸F]-FLUORO-L-THYMIDINE PET

Solene Collet ¹, Samuel Valable ¹, Jean-Marc Constans ^1,², Emmanuele Lechapt-Zalcman ^1,², Simon Roussel ¹, Nicolas Delcroix ¹, Myriam Bernaudin ¹, Ahmed Abbas ¹, Eziane Ibazizene ¹, Louisa Barre ¹, Jean-Michel Derlon ^1,², Jean-Sébastien Guillamo ^1,²

Abstract

PURPOSE: Conventional MRI has proven to be of interest in differentiating low grade from high grade diffuse gliomas. However, for anaplastic (grade III) gliomas, MRI does not allow an accurate classification: grade III tumors without contrast enhancement (CE) can be misdiagnosed as grade II and those with CE as grade IV. We therefore assessed multiparametric MRI, MR spectroscopy and [¹⁸F]-fluoro-L-thymidine (FLT) PET as tools to overcome these limitations. PATIENTS AND METHODS: Thirty nine patients with histopathologically proven gliomas of grades II (n = 7), III (n = 9) and IV (n = 23) underwent conventional MRI, perfusion, diffusion, proton magnetic resonance spectroscopy (¹H-MRS) and FLT-PET imaging before surgery. Relative cerebral blood volume (rCBV), apparent diffusion coefficient (ADC), contrast enhancement (CE), Cho/Cr, NAA/Cr, Cho/NAA and FLT-SUV were compared between grades and sensibility and specificity were determined using Receiver Operating Characteristic (ROC) curves. RESULTS: Mean and minimum ADC (ADC_mean and ADC_min) showed significant differences between grade II and grade III tumors without CE. ADC_mean showed the highest correlation with the grade (R² = 0.99, p = 0.0002) with 100% sensitivity and specificity. ADC_min, CBV_mean, CBV_max, FLT_mean, FLT_max, CE_mean, CE_max showed significant differences between grade III with CE and grade IV tumors. FLT_max showed the highest correlation with the grade (R² = 0.75, p < 0.0001) with 90% sensitivity and 100% specificity. CONCLUSION: Quantification of parameters from MR and PET imaging can overcome the limitations of conventional MRI. ADC could be of interest for accurate classification of low grade gliomas and FLT -PET for high grade gliomas.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-026. ALTERED FUNCTIONAL CONNECTIVITY OF THE DEFAULT MODE NETWORK IN DIFFUSE GLIOMA PATIENTS MEASURED WITH PSEUDO-RESTING STATE FUNCTIONAL MRI

Robert Harris ^1,², Susan Bookheimer ³, Timothy Cloughesy ⁴, Hyun Kim ¹, Whitney Pope ¹, Katherine Yang ¹, Albert Lai ⁴, Phioanh Nghiemphu ⁴, Benjamin Ellingson ^1,²

Abstract

Little is known about the effects of brain tumors on resting state fMRI networks within the brain. The purpose of the current study was to explore whether brain tumors disrupt the integrity of the default mode network (DMN), a well-characterized resting-state fMRI network. We evaluated whether tumor grade, volume, surgical status, or location decreased the functional connectivity within the DMN in patients with gliomas. Task-based fMRI data was obtained from 69 diffuse glioma patients and 11 healthy volunteers. Data was preprocessed and DMN integrity was compared across WHO grade, tumor volume, and surgical status using univariate and multivariate linear models. WHO grade was the most significant predictor of DMN integrity (P = 0.004). Surgical status was also a significant predictor of DMN integrity (P = 0.015), whereas tumor volume was not (P = 0.105). DMN integrity was lower in high-grade (WHO III-IV) compared with low-grade (WHO II) patients (P = 0.007). Tumors in the left parietal lobe showed a more impaired DMN compared with tumors in the frontal lobe, while tumors within and outside the network nodes did not differ significantly. Results suggest WHO grade and surgery have the largest impact on DMN functional connectivity, likely due to microscopic damage to tissue architecture from tumor infiltration in high-grade gliomas and physical damage to connections between functional regions, respectively. Comparison of network integrity between gliomas in the frontal and left parietal nodes further supports this hypothesis.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-027. VOLUMETRIC ASSESSMENT AND STRATIFICATION OF PATIENT SURVIVAL WITH EARLY POST-TREATMENT MRI IN PATIENTS WITH RECURRENT GLIOBLASTOMA TREATED WITH BEVACIZUMAB

Raymond Huang ^1,³, Rifaquat Rahman ³, Alhafidz Hamdan ², Caroline Kane ², Christina Chen ³, Andrew Norden ^2,³, David Reardon ^2,³, Srinivasan Mukundan ^1,³, Patrick Wen ²

Abstract

BACKGROUND: Despite a high radiographic response rate in patients with recurrent glioblastoma following bevacizumab therapy, survival benefit has been relatively modest. We assess whether tumor volume measurements based on baseline and early post-treatment MRI can stratify patients in terms of progression-free survival (PFS) and overall survival (OS). METHODS: Baseline (−4 +/− 4 days) and post-treatment (30 +/− 6 days) MRI exams of 93 patients with recurrent glioblastoma treated with bevacizumab were retrospectively evaluated for volume of enhancing tumor as well as volume of the T2/FLAIR hyperintensity. Overall survival (OS) and progression-free survival (PFS) were assessed using volume parameters in a Cox regression model adjusted for significant clinical parameters. RESULTS: In univariable analysis, residual tumor volume, percentage change in tumor volume, steroid change from baseline to post-treatment scan, and number of recurrences were associated with both OS and PFS. With dichotomization by sample median of 52% change of enhancing volume can stratify OS (52 weeks vs. 31 weeks, p = 0.013) and PFS (12 weeks vs. 21 weeks, p = 0.009). Residual enhancing volume, dichotomized by sample median of 8cm3, can also stratify for OS (64 weeks vs. 28 weeks, p< 0.001) and PFS (21 weeks vs. 12 weeks, p = 0.036). CONCLUSIONS: Volumetric percentage change and absolute early post-treatment volume of enhancing tumor can stratify survival for patients with recurrent glioblastoma receiving bevacizumab therapy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-028. IMAGE PROCESSING AND MRI BRAIN TUMOR SEGMENTATION TECHNIQUES: A BRIEF REVIEW

Ryan Jafrani ¹, Pascal Zinn ¹, Rivka Colen ¹

Abstract

PURPOSE/AIM: The purpose of this exhibit is: 1. To review the stages of image processing. 2. To review the various MRI brain segmentation techniques. 3. To discuss the future directions in segmentation research. CONTENT ORGANIZATION: Stages of image processing; Introduction to segmentation; MRI brain segmentation techniques; Summary and conclusion; Future directions: SUMMARY: The major points of this exhibit are: There are several medical image segmentation techniques, based on different physical and biological principles, that can be used to analyze brain MRIs. Performance of these techniques depends on the characteristics of the MRI data itself, including artifacts such as random noise, intensity inhomogeneity, etc. For every technique, there are several ways to correct these MRI artifacts, in order to fully exploit the multispectral character of MR imaging.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-029. IMAGING GENOMIC MAPPING IN BRAIN TUMORS: AN INTRODUCTION

Ryan Jafrani ¹, Pascal Zinn ², Rivka Colen ¹

Abstract

PURPOSE/AIM: The purpose of this exhibit is to: Introduce and detail the concept and history of imaging genomics and imaging genomic mapping in brain tumors; Show the ability of imaging genomics to provide biomarkers that reflect underlying molecular cancer compositions; Illustrate examples of an imaging surrogate and a clinical classification derived from the first large scale study in this field, as well as their clinical significance, and their future implications. CONTENT ORGANIZATION: Background and Emergence of Imaging Genomics; Materials and Methods; Results of Imaging-genomic Correlation; MRI-FLAIR as surrogate for Invasion; VAK Classification; Summary and Future Directions. SUMMARY: The major teaching points of this exhibit are: Imaging genomics/Radiogenomics is a relatively new field which links specific imaging traits (radiophenotypes) with gene-expression profiles, and vice versa; MRI- FLAIR is an imaging surrogate to non-invasively detect specific genomic components responsible for migration and invasion in Glioblastoma; The Volume/Age/KPS score (VAK) classification is readily applicable at patient admission and can predict operative outcome; Imaging genomics provide cost-effective biomarkers for leveraging genomic data that help direct therapy and predict prognosis and drug response.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-030. POTENTIAL OF MR SPECTROSCOPY FOR ASSESSMENT OF GLIOMA GRADING

Radim Jancalek ^1,⁴, Martin Bulik ^2,⁴, Tomas Kazda ^3,⁴

Abstract

MR spectroscopy is an imaging diagnostic method based that allows non-invasive measurement of metabolites in tissues. There are a number of metabolites that can be identified by standard brain proton MR spectroscopy but only a few of them has a clinical significance in diagnosis of gliomas including N-acetylaspartate, choline, creatine, myo-inositol, lactate, and lipids. We describe our experience with potential of MR spectroscopy for non-invasive grading of gliomas. Low-grade gliomas are generally characterized by a relatively high concentration of N-acetylaspartate, low level of choline and absence of lactate and lipids. The increase in creatine concentration indicates low-grade gliomas with earlier progression and malignant transformation. Progression in grade of a glioma is reflected in the progressive decrease in the N-acetylaspartate and myo-inositol levels on the one hand and elevation in choline level up to grade 3 on the other. Malignant transformation of the glial tumors is also accompanied by the presence of lactate and lipids in MR spectra of grade 3 but mainly grade 4 gliomas. It follows that MR spectroscopy is a helpful method for detection of glioma regions with aggressive growth or upgrading due to favorable correlation of the choline and N-acetylaspartate levels with histopathological proliferation index Ki-67. Gliomas of each grade have some specific MR spectroscopic features that can be used for improvement of the diagnostic value of conventional magnetic resonance imaging. Thus, MR spectroscopy may be a helpful method for detection of glioma regions with aggressive growth or upgrading and thus, it may be used for targeting of brain biopsies. Supported by grants IGA MZCR NT/14120 and NT/14600 and the European Regional Development Fund Project FNUSA-ICRC (No. CZ.1.05/1.1.00/02.0123).

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-031. INTRAOPERATIVE MRI FLAIR AND T2 TUMOR VOLUMES ARE UNDERESTIMATED WHEN COMPARED TO POST-OPERATIVE CORRESPONDING VOLUMES IN HIGH AND LOW-GRADE GLIOMAS

Randy Jensen ¹, Karen Salzman ¹

Abstract

INTRODUCTION: Intraoperative MRI is a useful tool in the operative management of human gliomas. In non-contrast enhancing gliomas, and to a lesser extent enhancing tumors, the T2 and FLAIR signals are important targets to determine target volumes for resection margins. We hypothesized that intraoperatively (IO) obtained MRI T2 and FLAIR signal volumes might not reflect corresponding post-operative volumes due to tumor edema after resection and brain manipulation. METHODS: We examined 25 low-grade gliomas (LGG, 10 low-grade astrocytomas and 15 oligodendrogliomas) and compared them to 15 high-grade tumors (HGG, 4 anaplastic astrocytomas, 3 anaplastic oligodendrogliomas, and 8 glioblastomas). Volumetric assessment of T2 and FLAIR signals were performed using the OSRIX image analysis program and ratios calculated for preoperative (Preop), intraoperative (IOp), and post-operative (PoOp)MRI images for each patient. RESULTS: Not surprisingly, for both LGG and HGG IOp and PoOp FLAIR and T2 signal volumes were significantly lower than PreOp volumes. T2 volume ratios were IOp:PreOp 0.22 (range 0.06–0.55) and PoOp:PreOp 0:36 (range 0.04–0.75). FLAIR volume ratios were IOp:PreOp 0.33 (range 0.04–0.86) and PoOp:PreOp 0:33 (range 0.03–0.74). Conversely, the PoOp:IOp ratios for T2 were 3.26 (range 0.43-9.89) and for FLAIR 2.34 (0.1–9.38). Similar trends were found for HGG with IOp and PoOp FLAIR and T2 signal volumes significantly lower than PreOp. Once again, PoOp:IO ratios for T2 and FLAIR were greater than 1. CONCLUSIONS: Postoperative FLAIR and T2 signal volumes are significantly significantly higher than intraoperatively obtained corresponding volumes for all gliomas studied. Further work will be required but this study questions the validity of T2 and FLAIR images obtained intraoperatively.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-032. TRYPTOPHAN METABOLISM IN THE NON-TUMORAL HEMISPHERE: A PET STUDY OF 99 GLIOMA PATIENTS

David Kamson ¹, Tiffany Lee ¹, Kaushik Varadarajan ¹, Natasha Robinette ^3,⁵, Otto Muzik ^1,⁵, Pulak Chakraborty ^1,⁵, Geoffrey Barger ^2,³, Sandeep Mittal ^3,⁴, Csaba Juhász ^1,²

Abstract

BACKGROUND: Positron emission tomography (PET) with alpha[¹¹C]methyl-L-tryptophan (AMT) can track serotonin synthesis (in brain) and also the immunosuppressive kynurenine pathway (often activated in gliomas). Tumoral AMT uptake can predict tumor type and distinguish radiation injury from glioma recurrence. Now we analyzed clinical correlates of decreased AMT uptake in the contralateral hemisphere of glioma patients (reported recently: Juhasz et al. Clin Nucl Med, 2012;37:838-42). METHODS: We analyzed 99 (46/53 pre-/post-treatment) AMT-PET scans of patients (mean age: 52 years) with WHO grade II-IV glioma. Standardized uptake values (SUV) and unidirectional AMT uptake (K-complex, a measure of brain serotonin synthesis; available in 93 cases) were quantified in cortical and subcortical regions of the non-tumoral hemisphere. The effect of gender, age, tumor grade, and treatment on cortical/subcortical AMT uptake, and associations with survival (in the post-treatment group), were analyzed. RESULTS: AMT K-complex was higher in females than males in the thalamus (p = .012, t-test) and frontal cortex (p = .047) in the whole group and in post-treatment patients (p = .017 and p = .009, respectively). Females also had higher thalamic SUV than males in the whole group (p = .029). AMT K-complex in the thalamus showed a moderate decline with age in the whole group (Pearson's r = -0.24; p = .009) and in the post-treatment subgroup (r = -0.34; p = .023). There was no difference in AMT-PET parameters between the pre- and post-treatment subgroups, and no correlation was found between cortical/subcortical AMT-PET parameters and tumor grade. Higher thalamic and putaminal SUV predicted shorter survival (log-rank: p = 0.003, p = 0.013, respectively), but not independently from tumoral SUV. CONCLUSIONS: While glioma grade and treatment do not appear to directly affect tryptophan metabolism in the non-tumoral hemisphere, an age-related decline and gender differences, in subcortical structures and/or frontal cortex, may occur, particularly post-treatment. This could indicate remote serotonergic abnormalities whose relation to clinical symptoms (e.g., glioma-associated depression) and clinical outcome warrants further studies.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-033. TRYPTOPHAN PET HAS A STRONG PROGNOSTIC VALUE IN POST-TREATMENT HIGH-GRADE GLIOMAS

David Kamson ¹, Geoffrey Barger ^2,³, Natasha Robinette ^4,⁶, Otto Muzik ^1,⁶, Pulak Chakraborty ^1,⁶, William Kupsky ^3,⁵, Sandeep Mittal ^3,⁴, Csaba Juhasz ^1,²

Abstract

BACKGROUND: MRI has limited prognostic value in patients with previously treated high-grade gliomas. Positron emission tomography (PET) imaging with alpha[¹¹C]methyl-L-tryptophan (AMT), a radiotracer tracking tryptophan transport and metabolism via the immunosuppressive kynurenine pathway, can accurately differentiate glioma recurrence from radiation injury (J Nucl Med 2012;53:1058-64). We evaluated AMT uptake as a prognostic marker for survival in post-treatment patients with high-grade glioma. METHODS: AMT-PET was performed in 39 patients >1 month (median: 325 days) after resection and chemoradiation of a grade III or IV astrocytic glioma. All patients were followed for at least 1 year or until death after PET. Maximum standardized uptake values (AMT-SUV_max) were measured in brain regions suspicious for tumor on MRI. Unidirectional AMT uptake (K) was also quantified from dynamic PET and blood radioactivity data (available in 34 cases). Optimal cutoff thresholds for prediction of 1-year survival were determined using receiver operating characteristic (ROC) analysis, and survival of patients with above vs. below threshold values was evaluated in the whole group, and in two subgroups (glioblastoma [n = 33], no further surgical or radiation treatment after PET [n = 22]). RESULTS: For 1-year survival, sensitivity/specificity was 84%/80% for AMT-SUVmax and 76%/88% for lesion/cortex K-ratio, when using ROC-defined cutoff thresholds. Both AMT-SUV_max and K-ratio were highly prognostic for survival in all patient groups. Estimated median survival after PET was 966 and 876 days in the low, vs. 177 days in both the high AMT-SUVmax and high K-ratio groups, respectively (log-rank: p< .001). Age- and grade-corrected Cox regression analyses showed a robust association between high AMT values and shorter overall survival after the PET scan (SUV_max: OR= 6.2 [2.2-17.5], K-ratio: OR= 6.6 [2.3-18.6]; p≤ .001). CONCLUSIONS: Focal increase of AMT uptake on PET has a strong prognostic value for 1-year and overall survival in patients with a previously treated high-grade glioma.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-034. DIFFERENT SPATIAL DISTRIBUTION BETWEEN GCB AND NON-GCB PCNSL REVEALED BY MR GROUP ANALYSIS OF 100 CASES

Manabu Kinoshita ^1,², Takashi Sasayama ³, Yoshitaka Narita ⁴, Atsushi Kawaguchi ⁵, Fumio Yamashita ⁶, Yasuyoshi Chiba ¹, Naoki Kagawa ¹, Kazuhiro Tanaka ³, Eiji Kohmura ³, Hideyuki Arita ⁴, Yoshiko Okita ⁴, Makoto Ohno ⁴, Yasuji Miyakita ⁴, Soichiro Shibui ⁴, Naoya Hashimoto ¹, Toshiki Yoshimine ¹

Abstract

MR based group analysis is a powerful tool to elucidate pathological conditions in the brain that are challenging to be revealed by single subject analysis. It enables to connect biological characteristics and spatial distribution of the disease in the brain by performing image registration of individual subjects to an averaged standard anatomical brain MRI and adding biological information into those registered images. The presented research aimed to elucidate special distribution characteristics of primary central nervous system lymphoma (PCNSL) within the brain with respect to its molecular marker expression patterns. MR images of 100 treatment naive PCNSL patients were collected and were registered onto Montreal Neurological Institute 152 standard anatomical MR image (MNI152). Gadolinium enhanced lesions were extracted and a lesion frequency map was created. Along with this procedure, molecular marker expression pattern information (BCL-6, MUM-1, CD10) was incorporated into the created frequency map image. Lymphoma subtypes were classified into GCB and non-GCB based on those molecular marker expression patterns. PCNSL frequency map showed that these tumors tended to occur at around the ventricles including lateral, 3rd and 4th ventricles. Moreover, GCB and non-GCB PCNSL frequency maps showed that GCB lymphomas located at around the posterior portion of the white matter surrounding the lateral ventricles and the cerebellum around the 4th ventricle, while non-GCB lymphomas tended to occupy the anterior portion of the white matter surrounding the lateral ventricles. These differences were statistically significant, confirmed by the existence of voxels with p-values less than 0.05 (random permutation analysis with voxel-wise Fisher's exact test). This is the very first report that addresses phenotypical and spatial distributional difference between GCB and non-GCB PCNSL by use of MR group analytical method. The presented data suggests that GCB and non-GCB PCNSL exhibit different anatomical affinities within the brain.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-035. COMPARISON OF MULTIMODAL MAGNETIC RESONANCE IMAGING UTILIZING 1.5 T AND 3.0 T IN GLIOBLASTOMA MULTIFORME AT THE TIME OF FIRST RECURRENCE

Lara Kunschner Ronan ¹, Clifford Eskey ¹, Thomas Hampton ¹, Camilo Fadul ¹

Abstract

BACKGROUND: The standard initial therapy for high grade glioma is concurrent radiation/temozolomide chemotherapy followng the regimen published by Stupp et al. CE MRI isthe gold standard to determine response to treatment. Unfortunately, treatment may result in geographic areas of CE MRI abnormality suggestive for tumor progression, but that instead reveal, on pathologic examination, treatment related necrosis without tumor recurrence. We hypothesized that advanced multimodal MR techniques will provide tissue signatures that differentiate tumor progression from treatment-related necrosis, and that the tissue signatures obtained at 3T will be more accurate than those at 1.5T. MATERIALS AND METHODS: We prospectively acquired multi-parametric MR variables from GBM patients with MRI suggestive of recurrence that showed pathologically confirmed recurrence. We compared the values of signatures acquired at 1.5T and 3T, to evaluate the predictive strength with the differing MR machines. Correlations between MRS ratios Cho/Cr,NAA/Cr,Lip/Cr, total metabolites T-NAA and T-Cho, ADC ratio and CBV were determined. RESULTS: Our intial hypothesis to compare MRI signatures in patients with recurrence or treatement related necrosis was not testable as only one patient enrolled had pathologically confirmed treatment related necrosis and that patient was subsequently excluded from analysis. The confidence intervals for mean ADC, CBV and MRS were similar with both 1.5T and 3.0T machines. There was poor correlation between metabolite variables on either 1.5T or 3.0T MRI. CBV shows a trend towards higher accuracy with the 3.0T MRI. CONCLUSION: We were unable to discern a clear benefit to the use of 3.0 T MRI for tissue characterization using ADC ratio and MRS over the standard 1.5 T machine in common clinical use. A non-significant trend towards better accuracy with the 3.0 T MRI was seen for CBV. A larger sample may be sufficient todetermine if the improved accuracy of the CBV AT 3.0t is statististically significant.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-036. RELIABILITY OF QUANTITATIVE BIOMARKERS OF TUMOR PROGRESSION BASED ON MULTISPECTRAL MRI IN GLIOBLASTOMA PATIENTS

Pamela LaMontagne ¹, Mikhail Milchenko ¹, Peter Sylvester ¹, Tammie Benzinger ¹, Daniel Marcus ¹, Sarah Jost Fouke ²

Abstract

Imaging biomarkers can be promising predictors of prognosis or therapeutic response in Glioblastoma. However, the use of multispectral MRI in clinical practice remains primarily qualitative. Quantifying predictors of tumor progression is challenging, largely because the relation between tissue pathology and MR imaging parameters is not yet fully understood, and because quantitative integration of multiparametric imaging is difficult to apply in clinical practice. We describe a systematic framework used to investigate whether multispectral automated tumor segmentation algorithms can be used as reliable tumor markers of prognosis and time to tumor progression in Glioblastoma. We considered data from the Comprehensive Neuro-oncology Data Repository (CONDR). CONDR is a multisite study that follows patients with primary brain tumors collecting serial MR imaging data, tissue markers, and clinical data. We identified 45 patients with glioblastoma for analysis. Disease progression was noted in 33 patients with a median progression time of 145 (35-469) days. Twelve patients had not yet progressed at six months post-treatment. To evaluate early predictors of progression, we collected multispectral MR data, including pre- and post-contrast T1 and T2-weighted images, diffusion weighted imaging (DWI), and perfusion weighted imaging (PWI). The data were collected for two time points prior to tumor resection, and within 72hrs post-tumor resection. Each patient's MR images were co-registered to their pre-op T1-weighted scan. We then used the GLISTR (Glioma Image SegmenTation and Registration) automated segmentation tool to model tumor, edema, and necrotic tissue types. GLISTR segmentations are used as masks to define tumor boundaries across all image types. Imaging biomarkers, including ADC and CBV and residual enhancing tumor on post-op scans, will be examined in relation to time to progression. In addition, integrated tissue segmentations will be compared with tumor progression statistics, to find whether GLISTR segmentations can be used as predictors for tumor growth.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-037. BENEFITS OF USING HIGH-FIELD MRI IN EVALUATING THE LONG-TERM EFFECTS OF RADIATION IN PATIENTS WITH GLIOMA

Janine Lupo ¹, Wei Bian ¹, Mekhail Anwar ¹, Suchandrima Banerjee ², Christopher Hess ¹, Susan Chang ¹, Sarah Nelson ¹

Abstract

Radiotherapy (RT) is an integral component in managing patients with glioma, but the toxicity to healthy brain tissue and vasculature, and effects on quality-of-life is of concern. The histologic response to radiation initially shows characteristic vascular changes that later result in the formation of hemosiderin-containing microbleeds. Our prior studies using Susceptibility-Weighted Imaging (SWI) at 7T have characterized the extent and evolution of these lesions. In this study, we demonstrate the benefits of using 7T MRI in the early detection of microbleeds, assessment of damage to surrounding vasculature, and evaluation of their relationship to RT dose. In 10 patients scanned at 3T and 7T, 7T SWI detected significantly more microbleeds than 7T T2*-weighted gradient-echo images (p<0.002) and 3T SWI (p<0.02) when not present in locations of susceptibility artifact or iron-rich basal ganglia. For 10 other patients whose RT dosimetry maps were available, microbleed appearance depended on dose of radiation received, with significantly more microbleeds found in high-dose regions (>45Gy, P > 0.03), and microbleeds forming first in higher-dose (>30Gy regions by 2yrs), then lower-dose regions (<30Gy regions by 4yrs). The development of a multi-echo MR-Angiography/Venography sequence at 7T has allowed for the simultaneous visualization of arteries, veins, and microbleeds all on one image, facilitating the assessment of damage to arterioles and venules that surround microbleeds. The additional flexibility in SWI processing with multiple echoes allowed for thinner slices, which not only improved microbleed detection sensitivity, but also the contrast of smaller arteries. The ability to detect microbleeds with heightened sensitivity and characterize their relationship with dose received to normal-appearing brain tissue and damage to surrounding vasculature may be important in determining which parts of the brain are most susceptible to radiotherapy, assessing cognitive outcome, considering therapeutic strategies, and planning target volumes in patients with lower grade tumors, who have longer survival.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-038. RECURRENT PRIMARY CENTRAL NERVOUS SYSTEM LYMPHOMA: APPARENT DIFFUSION COEFFICIENT DERIVED FROM DIFFUSION WEIGHTED IMAGING IS PREDICTIVE OF SURVIVAL AFTER RECURRENCE

Marc Mabray ¹, Lorena Sanchez ¹, Francisco Valles ¹, Ramon Barajas ¹, James Rubenstein ¹, Soonmee Cha ¹

Abstract

BACKGROUND/PURPOSE: Apparent diffusion coefficient (ADC) values derived from clinical magnetic resonance diffusion-weighted imaging (DWI) at presentation have been shown to be predictive of clinical outcome in immunocompetent patients with primary central nervous system lymphoma (PCNSL). The purpose of our study was to determine whether 1) ADC values at the time of tumor recurrence predict survival, 2) ADC values at the initial presentation predict survival after recurrence, and 3) the relationship between ADC values at the initial presentation and at recurrence within a given patient. MATERIALS AND METHODS: Forty immunocompetent patients with recurrent PCNSL were selected for analysis. ADC maps were used to calculate the minimum (ADC_min), average (ADC_avg), and maximum ADC (ADC_max) values for each patient. The patients were separated using thresholds for ADC_min, ADC_avg, and ADC_max and survival probability was plotted for each group with p values calculated for differences in survival at 24 months. Analysis was also performed on the ADC values at initial presentation. ADC_min, ADC_avg, and ADC_max values were calculated, survival probability plotted, and p values calculated for differences in survival at 24 months after recurrence. RESULTS: Survival at 24 months after recurrence was statistically significant (p < 0.05) using an ADC_min threshold of 621 (ADC x10⁻⁶ mm²/s) (p = 0.041) or 604 (p = 0.023), or using an ADC_avg of 883 (p = 0.023). Survival at 24 months after recurrence using ADC_max of 2161 was not significant (p = 0.069). Using ADC values at initial presentation to predict survival 24 months after recurrence was statistically significant using an ADC_min of 376 (p = 0.039), ADC_avg of 506 (p = 0.024), and ADC_max of 1090 (p = 0.039). CONCLUSIONS: ADC values at recurrence were predictive of survival in patients with recurrent PCNSL. Similarly, the ADC values at initial presentation were also predictive of survival at 24 months after recurrence.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-039. PET GUIDANCE FOR SURGERY PLANNING IN GLIOMAS

Keisuke Miyake ¹, Daisuke Ogawa ¹, Tetsuhiro Hatakeyama ¹, Nobuyuki Kawai ¹, Takashi Tamiya ¹

Abstract

OBJECTIVE: Use of the positron imaging agents, FDG, MET, FLT, 4DST and FMISO, is expected to lead the way for novel applications aimed at achieving efficient malignancy grading and treatment of gliomas. In this study, the usefulness of five PET scans was retrospectively reviewed by comparing the histopathological findings. METHODS: PET scans were performed in 18 patients with WHO grade IV gliomas, 14 patients with WHO grade III gliomas and 8 patients with WHO grade II glioma from April 2009 to April 2013. The resulting PET scans were compared by measuring the T/N or T/B ratios. Based on a pre-determined ROI, the T/N ratio of FDG, MET, FLT and 4DST and the T/B ratio of FMISO were used to compare the values of each of the five PET studies. The tumor activity and degree of malignancy were evaluated using the Ki-67 index. The correlations between the T/N or T/B ratio and the Ki-67 index were determined for all subjects using the non-parametric Spearman's rank test. RESULTS: There were differences in the accumulation of each tracer within the same tumor due to the characteristics of each tracer, which may reflect differences in the tissues by site. We can predict that areas with high accumulation of both MET and FLT were seen histologically to be highly malignant. In all of the patients, uptakes of MET, FLT, 4DST and FMISO were significantly correlated with the Ki-67 index. In astrocytoma cases, FLT was the most significantly correlated with the Ki-67 index (FLT:r = 0.59, p<0.01). In oligodendroglioma cases, MET was the most significantly correlated with the Ki-67 index (MET:r =0.67, p<0.01). CONCLUSIONS: PET imaging is useful for determining the tumor area for removal as well as improving the preoperative diagnosis of gliomas.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-040. ANALYSIS OF THE CORRELATION BETWEEN MALIGNANCY OF GLIOMAS AND FINDINGS ON 3D-PRESTO MRI SEQUENCE

Kanji Mori ¹, Reiichi Ishikura ¹, Yusuke Tomogane ¹, Kumiko Ando ¹, Shuichi Izumoto ¹

Abstract

The principle of echo-shifting with a train of observation (PRESTO) magnetic resonance technique employs an MR sequence that has high sensitivity for susceptibility changes in the brain. Three-dimensional (3D) PRESTO-MRI can detect microhemorrhages, hemosiderin deposits and small veins as markedly low signal intensity (spotty signal void). We examined the correlation between the grade of glioma and findings about spotty signal voids and usefulness for diagnosis. Forty-six surgically treated patients (10 low grade 2, 11 grade 3, 25 grade 4) were included in this study. We defined spotty signal voids as small dark spot (SDS) or oval dark spot (ODS) according to its diameter. And we classified all patients into four groups by density of SDS. N group had no SDS, L group showed low density of SDS, M group showed medium density, and H group with most dense SDS. We compared the proportion of these groups among the grade of gliomas, then compared frequency of cases with ODS. We found that the 76% of grade 4 were classified as H, 16% as M, 8% as L, and no case as N. While, 70% of grade 2 was classified as N, 20% was L, and 10% was M. So the higher is glioma grade, the more SDS were observed. Furthermore, ODS existed in 64% in grade 4, 34.4% in grade 3, and 10% in grade2. It was supposed that SDS might reflect the increase of tumor vessels and contrast enhancement after injection of contrast medium, and ODS might reflect necrosis and intratumoral hemorrhage. Finally, 3D-PRESTO MRI is thought to bring us valuable information for evaluating glioma malignancy on MRI without using contrast medium.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-041. FUNCTIONAL BRAIN MRI FOR ASSESSING EARLY ACTIVITY OF RAMUCIRUMAB AND OLARATUMAB IN PATIENTS WITH RECURRENT GLIOBLASTOMA

Sarah Nelson ¹, Frank Lieberman ², Janine Lupo ¹, Sana Viziri ¹, L Burt Nabors ³, Jason Crane ¹, Patrick Wen ⁴, Andre Cote ¹, David Peereboom ⁵, Qiuting Wen ¹, Timothy Cloughesy ⁶, H Ian Robins ⁷, Joy Fisher ⁸, Serena Desideri ⁸, Stuart Grossman ⁸, Xiaobu Ye ⁸, Jaishri Blakeley ⁸

Abstract

VEGF and PDGF receptors are important therapeutic targets in glioblastoma (GBM). We used functional MRI methods to explore biologic markers of VEGFR-2 or PDGFRalpha inhibition. A subset of patients assigned to receive ramucirumab 8 mg/kg IV q2wks (anti-VEGFR-2) or olaratumab 20 mg/kg IV q2 wks (anti- PDGFRalpha) as part of a multicenter phase 2 study in adults with recurrent GBM underwent standardized MRI brain protocols with dynamic contrast enhanced (CE); diffusion tensor; perfusion weighted, MRS and routine sequences at -1d, +1d, +28d and every +56d from treatment initiation. ACRIN performed site qualification and imaging QA. Regions of CE and T2w hyperintensity were manually defined; percent change in volumes and change in normalized image intensities were calculated. Twenty-eight patients had evaluable data from 6 sites. DSC was analyzable in 23 patients; MRS was obtained in selected examinations from 19 patients. Eleven patients received ramucirumab:7 male, median age 54 (43–67), KPS 90 (70–100) and Dex dose 2mg (0-8). Seventeen patients received olaratumab, 14 male, median age of 58 (24–72), KPS 80 (70-100) and Dex dose 2mg (0–4mg). There was an early (+1d) and notable reduction in the relative contrast enhancing (CE) volume from baseline (mean 80%, SD 14%) maintained at +56d (mean 85%, SD 25%) with ramucirumab,. Normalized ADC intensity decreased in both ramucirumab and olaratumab groups between baseline and day 28 (1.7 to 1.6). VEGFR-2 inhibition resulted in expected markers of vessel normalization (decreased CE lesion volume, relative intensity of enhancement and blood volume) 1 day after treatment and decrease in T2 lesion volume and ADC at 28 days. PDGFRalpha inhibition did not have classic anti-angiogenic effects, but did show reduction in relative intensity of enhancement, blood volume and ADC at 28 days. The relationship of spectral data to other imaging parameters and of imaging findings to treatment outcome will be presented.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-042. RELATION BETWEEN METHIONINE UPTAKE AND MOLECULAR MARKERS IN GLIOMA

Masahiro Nonaka ¹, Shin Nakajima ¹, Tomoko Shofuda ¹, Yonehiro Kanemura ¹

Abstract

Although various reports demonstrate the value of f l-[methyl-(11)C]-methionine ((11)C-MET) PET in glioma for evaluation of tumor extension, histology and prognosis, less is known about the relationship between methionine uptake and molecular markers, such as methylation of the O(6)-methylguanine-DNA methyltransferase (MGMT), and isocitrate dehydrogenase 1 and isocitrate dehydrogenase 2 (IDH) genes mutations. In this study, we focused on grade 2 and grade3 glioma patients, and evaluated (11)C-MET PET, methylation status of MGMT and mutations in IDH1/2 gene in glioma patients. METHODS: In 20 patients with histologically-proven gliomas (G2 and G3), we studied (11)C-MET PET, MGMT methylation and IDH1/2 mutation. For methionine uptake, the ratio of the mean standardized uptake value in the tumor to mean standardized uptake value in background normal cortex (T/N ratio) was calculated. Quantitative polymerase chain reaction method was used to assess promotor methylation of MGMT gene. Methylation rate higher than 1% was evaluated as "methylated". IDH mutations were screened by direct sequencing except for one case, which was evaluated by immunostaining. Statistical analysis was performed using chi-square for independent test. RESULTS: Methionine uptake was documented in 13 patients (group A) while no uptake was documented in 7patients (group B). Methylation of MGMT gene was documented in 76.9% from group A, and 16.7% from group B (p < 0.05). IDH mutation was documented in 10 cases from group A (76.9%) and two cases from group B (28.6%) (p = 0.052). CONCLUSION: Methionine uptake in glioma demonstrated a significant correlation with methylation status of MGMT gene. IDH gene mutation also demonstrated a close relationship with methionine uptake.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-043. RADIOLOGICAL PROGRESSION TYPES FOLLOWING ANTI-ANGIOGENIC THERAPY ARE RELATED TO OUTCOME IN RECURRENT GLIOBLASTOMA PATIENTS

Martha Nowosielski ^1,⁵, Benedikt Wiestler ², Georg Göbel ³, Markus Hutterer ^1,⁴, Heinz Schlemmer ⁶, Günther Stockhammer ¹, Wolfgang Wick ², Martin Bendszus ⁵, Alexander Radbruch ^5,⁷

Abstract

OBJECTIVES: Survival after progression on bevacizumab for recurrent glioblastoma (rGBM) is very limited. Here we analysed if different radiological progression types (PT) based on contrast medium behavior in MRI T1 and changes in T2 hyperintense areas are relate to outcome in bevacizumab-treated patients. METHODS: We retrospectively evaluated MRI scans prior to and at progression on bevacizumab in 83 rGBMs. Based on initial decrease and subsequent flare up of contrast enhancement in T1 and two patterns of T2 hyperintense tumor progression, PTs were categorized as cT1-flare up, T2-diffuse, T2-circumscribed and primary non-responders. Time to progression, salvage therapy and survival from start of bevacizumab therapy (PPS), post bevacizumab progression survival (PostBevS), survival from initial diagnosis until start of bevacizumab (StartBevS) and overall survival (OS) were evaluated using Kaplan-Meier curves, Log-Rank test and Cox regression analyses. RESULTS: The time period to develop a T2-diffuse (n = 15) or cT1-flare up (n = 35) progression was significantly longer compared to progression in primary non-responders (n = 16) or in T2-circumscribed (n = 17). The T2-diffuse PT showed a three times longer survival than the T2-circumscribed PT with hazard ratios between 1.5 and 2.4 for PPS, PostBevS, StartBevS and OS. Salvage therapy tended to show a survival benefit only in the T2-circumscribed PT but not in the other PTs. CONCLUSIONS: Radiological progression types following bevacizumab treatment failure show differences in timely development and are related to outcome. We therefore hypothesize that these PTs reflect different resistance mechanisms to anti-vascular endothelial growth factor therapy and may be associated with different responses to salvage therapy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-044. MRI AND MEDULLOBLASTOMA: SUBTYPING AT FIRST SIGHT?

Sebastien Perreault ^1,², Kristen Yeom ¹, Vijay Ramaswamy ³, David Shih ³, Marc Remke ³, Betty Luu ³, Simone Schubert ¹, Paul Fisher ¹, Sonia Partap ¹, Hannes Vogel ¹, Tina Young Poussaint ⁴, Michael Taylor ³, Yoon-Jae Cho ¹

Abstract

Molecular subtyping of medulloblastoma has recently been established and offer possibilities to better appreciate prognosis and to tailor treatments. However, access to molecular analysis remains limited and results might not be available when treatments are initiated. Conversely, magnetic resonance imaging (MRI) studies are routinely performed at initial presentation and during surgical planning. We investigated the value of MRI features at diagnosis in determining molecular subtypes of medulloblastoma. We conducted a historical cohort study of children diagnosed with medulloblastoma from January 1998 to January 2013 in one center. Patients with an initial MRI available for review and with available tissue were included in the study. Molecular subtyping was conducted using a nanoString-based assay and tumors were classified into four subgroups: WNT, SHH, Group 3 and Group 4. MRIs were reviewed by an independent neuro-radiologist. Specific MRI features were assessed for their correlation to molecular subtype. A total of 47 patients were included. Four (8.5%) were classified as WNT, 13 (27.7%) as SHH, 13 (27.7%) as Group 3 and 17 (36.1%) as Group 4. Logistic regression showed that location (p < 0.001), pattern of enhancement (p = 0.001) and definition of tumor margin (p = 0.01) were predictor of medulloblastoma subtype. WNT tumor were usually located at the cerebellar peduncle (75%), a subset of SHH tumor involved the cerebellar hemisphere (30.8%), most Group 4 tumor presented minimal enhancement (58.8%), whereas Group 3 had poorly defined interface between the cerebellar parenchyma and the tumor (38.4%). Using our regression model, 68% of all tumors were appropriately classified. We conclude that specific imaging characteristics found on MRI at diagnosis can be used as surrogate markers for molecular subtype in medulloblastoma. Validation in two independent cohorts will be used to confirm our current observations.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-045. PREDICTIVE VALUE OF CT AND MR PERFUSION IMAGING IN PATIENTS WITH HIGH-GRADE MALIGNANT GLIOMAS TREATED WITH BEVACIZUMAB

Francesca Piludu ¹, Andrea Pace ¹, Alessandra Fabi ¹, Vincenzo Anelli ¹, Veronica Villani ¹, Carmine Carapella ¹, Simona Marzi ¹, Antonello Vidiri ¹

Abstract

OBJECTIVE: To determine whether early monitoring of the effects of Bevacizumab in patients with recurrent high-grade gliomas, by Perfusion CT or MR imaging, may predict the response to treatment. MATERIALS AND METHODS: 20 patients with malignant glioma were enrolled in the study. For each patient two perfusion examinations, before and after the first dose of Bevacizumab, were acquired. Sixteen patients underwent a Perfusion CT (PCT) examination by using a 128-section CT scanner and four patients a dynamic contrast enhancement (DCE) MR examination on 1.5-T system. The areas of abnormal cerebral blood volume (CBV) have been outlined on the two studies, using contrast-enhanced T1-weighted images as a guide to tumor identification. Normalized CBV (nCBV) maps were obtained dividing the original CBV maps by the mean CBV value inside a healthy region in the hemisphere contralateral with respect to the lesion. Specific hypo- and hyper-perfused sub-volumes were calculated, as absolute voxel counts within the region of interest in which nCBV values were lower or greater than fixed thresholds, respectively. The correlations between the early changes in perfusion and volume changes at the first follow-up were investigated. RESULTS: The changes of nCBV maps indicate an effective normalizing effect of the drug on the areas of abnormal vascularity, even if initial nCBV values > 3 suggest a reduced activity of the anti-angiogenic, agent, as if the normalization effect is eventually less efficient. The reductions of mean, median and standard deviation of the nCBV after the first dose are statistically significant (p ≤ 0.0001). An improvement in hypoxia after a single dose of bevacizumab is a predictor of a greater reduction in T1-weighted contrast-enhanced volumes at first follow-up. CONCLUSIONS: These preliminary results show that a quantification of changes in tumoral regions could be proposed as a potential imaging biomarker of tumor response to anti-VEGF therapies.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-046. GLIOMATOSIS CEREBRI: ASSESSMENT OF CLINICAL AND RADIOLOGICAL RESPONSE TO TEMOZOLOMIDE AND RADIATION IN 20 PATIENTS

Sona Pungavkar ¹, Prasad Tanawde ², Sridhar Epari ², Deepak Patkar ¹, Malini Lawande ¹, Aliasgar Moiyadi ², Tejpal Gupta ², Rakesh Jalali ²

Abstract

INTRODUCTION: Gliomatosis cerebri (GC) is a diffuse cerebral infiltration by neoplastic cells. Type I consists of diffuse overgrowth without a focal mass, while a focal mass is additionally seen in yype II, usually a high-grade glioma. Magnetic resonance imaging (MRI) plays a crucial role in diagnosis, mapping tumor spread, directing stereotactic biopsy and follow up. Optimal therapeutic strategy for GC has not yet been established. Radiotherapy (RT) and temozolomide (TMZ) have been used to treat GC. Despite aggressive and often multimodal therapeutic intervention, survival rates are poor. RESULTS: We identified 20 patients at our institution over a 3 year period. They presented with focal motor and sensory deficits, seizures, cognitive dysfunction, headache, and vertigo. All underwent a routine MR exam, including FLAIR, contrast, Diffusion, Perfusion and Spectroscopy. The histology was astrocytoma grade III and IV in 42.8% and Grade II astrocytoma in 57.17% . 7 underwent biopsy and 13 underwent subtotal excision. All patients received postoperative adjuvant radiotherapy with doses varying from 54 to 60 Gy. All but one patient received chemotherapy in form of concurrent plus adjuvant TMZ (depending upon response). Identical follow up MR exams were performed after completion of RT, completion of 6 cycles of TMZ and subsequently to assess for response. 16 patients had clinico-radiological response. 4 patients progressed after therapy. Of these, 2 patients progressed after adjuvant TMZ and received second line of therapy, Bevacizumab. Median follow-up and progression free survival were 16 and 11.3 months, respectively. CONCLUSIONS: Gliomatosis cerebri has a variable clinical course. RT with TMZ provokes good response of high grade and infiltrative component on serial T2 and FLAIR images. In cases with longer survival, type II changes can occur at other sites within the infiltrated parenchyma. Role of re-excision, re-irradiation, TMZ, Bevacizumab etc. needs to be explored in such patients.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-047. BASELINE SPIN-ECHO ECHO-PLANAR PERFUSION nCBV PRIOR TO CHEMORADIATION IS A STRONG INDEPENDENT PRE-THERAPY PREDICTOR OF PROGRESSION-FREE AND OVERALL SURVIVAL IN NEWLY DIAGNOSED GLIOBLASTOMA

Rifaquat Rahman ¹, Ayca Akgoz ², Hui You ³, Alhafidz Hamdan ⁴, Ravi Seethamraju ⁵, Patrick Wen ^4,¹, Geoffrey Young ^2,¹

Abstract

BACKGROUND: A growing literature demonstrates that gradient-echo echo-planar perfusion weighted imaging (GE-PWI) has independent prognostic value for patients with newly diagnosed glioblastoma. Spin-echo echo planar PWI (SE-PWI) is more selective than GE-PWI for blood volume in microvessels that are the size of glioma neocapillaries, but it has not been comprehensively studied in human clinical use. We assessed whether SE-PWI before and after initiating chemoradiation can stratify patients with respect to progression free survival (PFS) and overall survival (OS). METHODS: Sixty-eight glioblastoma patients with interpretable pre and post-treatment SE-PWI were identified. In each study, normalized cerebral blood volume (SE-nCBV) was calculated by hot-spot method from 3 regions of interest (ROI) selected within the areas of maximal cerebral blood volume (CBV) in enhancing and/or non-enhancing tumor and 1 ROI selected within the contralateral normal appearing white matter. SE-nCBV parameters predictive of PFS and OS were identified by univariate and multivariate Cox proportional hazards modeling. Receiver operator curve characteristic analysis was used to identify thresholds optimized for 15-month survival, and Kaplan-Meier estimates of PFS and OS were calculated. RESULTS: Multivariate analysis demonstrated that baseline tumor mean SE-nCBV was predictive of PFS (p = 0.038) and OS (p = 0.004). Baseline tumor mean SE-nCBV < 2.0 predicted longer patient PFS (median 47.0 weeks, p < 0.001) and OS (median 98.6 weeks, p = 0.003) compared with baseline mean SE-nCBV > 2.0 (median PFS 25.3, median OS 56.0 weeks). Exploratory multi-group stratification demonstrated that very high tumor SE-nCBV > 4.0 was associated with worse patient OS than intermediate high SE-nCBV (>2.0, < 4.0; p = 0.025). CONCLUSION: Baseline mean SE-nCBV can stratify patients for survival prior to initiation of chemoradiation, which may help select patients who require closer management. Our exploratory analysis indicates a magnitude-dependent relationship between baseline SE-nCBV and OS.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-048. A COMBINATORIAL RADIOPHENOTYPE STRATIFIES PATIENT SURVIVAL AND ASSOCIATES WITH INVASION/PROLIFERATION CHARACTERISTICS IN GLIOBLASTOMA (GBM)

Arvind Rao ¹, Ganesh Rao ¹, Adam Flanders ²; TCGA Glioma Phenotype Research Group¹

Abstract

OBJECTIVE: To investigate if combinations of radiographic features can stratify survival in the TCGA GBM cohort. Further, to investigate molecular differences between phenotypic classes induced by the combinatorial phenotype. METHODS: mRNA, micro-RNA and radiographic information for 98 GBM cases was extracted from TCGA (Cancer Genome Atlas) data portal and TCGA Glioma Phenotype Research Group annotations. These radiographic annotations include lesion features such as volume-class, enhancement-quality, proportion contrast-enhancing, proportion non-contrast enhancing tumor, proportion-necrosis, proportion-edema, T1FLAIR-envelope, enhancing margin thickness, distribution and hemorrhage. Patient survival data was downloaded from MSKCC cBioPortal. Radiographic features were clustered using Hierarchical Clustering. Further, one, two- and three-way combinations of these radiographic features were interrogated for their ability to stratify overall survival in a statistically significant manner. The cutpoint for statistically significant difference was identified in a randomly chosen "training set" via a recursive-partitioning analysis (RPA). Differential expression analysis(of miRNA and mRNA) between the RPA-induced phenotypic classes was performed using GenePattern software. Pathway analysis was performed using Gene Set Enrichment Analysis (GSEA). RESULTS: Clustering analysis reveals that three variables: volume-class, Hemorrhage and T1:FLAIR ratio drives clustering structure. A three-way combination of these three variables stratifies survival. A cutpoint analysis yields a statistically significant survival difference in the training set (median survival-difference: 8 months, p-value: 0.03) as well as a held-out validation set (p-value: 0.004). Differential expression analysis reveals that several immune-associated (lymphocyte-homing, TNFR2 modulation, antigen processing, interferon-gamma response) and metabolism-associated (TCA cycle, oxidative phosphorylation) pathways underlie the transition of this combinatorial phenotype. Integrating mRNAs with miRNA data reveals the role of two genes (VSNL1, SYT1) and two miRNA (mir-146b, mir124a) implicated in proliferation and invasion in glioma. CONCLUSION: This study presents the first evidence that combining radiographic features can stratify survival. We also demonstrate the superior predictive ability of radiophenotype-combinations rather than individual features. Further, this combinatorial phenotype identifies pathways relevant to tumor biology.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-049. COMPUTATIONAL IMAGING FEATURES DERIVED FROM MRI CAN DISCRIMINATE THE MOLECULAR SUBTYPES OF GLIOBLASTOMA MULTIFORME

Payel Ghosh ¹, Ganesh Rao ¹, Juan Martinez ¹, Arvind Rao ¹

Abstract

OBJECTIVE: To determine which computationally-derived imaging features correlate with the molecular subtypes of GBM. Further, to investigate whether subtype-specifc image features predict survival. METHODS: Pre-surgical contrast enhanced T1-post and T2-FLAIR MRI images from 83 GBM patients from The Cancer Genome Atlas (TCGA) were analyzed. Images were preprocessed via registration of the T1-post and FLAIR images, background segmentation and intensity normalization. The tumors were segmented semi-automatically using the Medical Image Interaction Toolkit (MITK.org). Texture features were extracted for 3D volumes using MatlabTM scripts. Features included three-dimensional statistical, transform-based and model-based features, pertaining to gray-level heterogeneity measures such as energy, entropy, and correlation (∼150,000 descriptors for each subject). Features were pruned using coefficient of variation criterion. Random Forest Classifiers (with class balancing) were trained with pruned features (using Weka) to discriminate subjects by subtype. Receiver Operating Characteristic curves (ROC) were used to assess the performance of the 4-class subtype classifier. Subtype-specific features were examined for their ability to predict survival. The Wilcoxon ranksum test was used to identify the significant features for each subtype which were used to predict overall survival using a random survival forest (RSF). RESULTS: Feature selection via RF analysis indicates that multi-resolution textures are strongly represented among the selected features. ROC analysis reveals that image-features are capable of determining subtype. The Area Under Curve (AUC) for each of the 4 subtypes (classical, mesenchymal, neural, proneural) is 0.56,0.59, 0.62 and 0.37 respectively. This suggests that is challenging to obtain image signatures for the proneural subtype. RSF survival analysis reveals that subtype-specific image features are capable of predicting survival. The accuracy of survival prediction for each of the 4 subtype-associated image features is 63%, 57%.63% and 64% respectively. CONCLUSION: This study presents preliminary evidence that MRI-derived texture features are predictive of GBM subtype. Our results suggest that imaging-derived correlates are prognostic for overall survival.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-050. RADIOLOGICAL CLASSIFICATION OF POSTERIOR FOSSA HEMANGIOBLASTOMAS

Tae Hoon Roh ¹, Eui Hyun Kim ¹, Jong Hee Chang ¹

Abstract

OBJECTIVE: Hemangioblastomas are known to show very typical radiological findings, however, unusual radiological presentation is not uncommon. In this study, we reviewed magnetic resonance imaging (MRI) of hemangioblastomas in the posterior fossa and classified them. METHODS: We retrospectively reviewed preoperative and postoperative MRI of 85 patients with hemangioblastomas in the posterior fossa who were operated in our institution from 2002 until 2010. Based on the preoperative MRI, we classified all tumors into four categories; purely cystic, cystic with mural nodule (macrocystic), solid, solid mass with central cystic area, and others with unusual presentation. The association with Von Hippel-Lindau (VHL) disease and its clinical implication were also investigated. RESULTS: There were 49 males and 36 females. Total numbers of tumor were 215 (mean, 2; range, 1 - 20). Single tumor was found in 70 patients, whereas the other 15 patients had multiple tumors in their posterior fossa. In 17 patients, VHL disease was confirmed according to diagnostic criteria and/or VHL gene study. Based on preoperative MRI findings, 215 tumors were classified into 3 purely cystic tumors, 58 tumors with cyst and mural nodule, 141 solid tumors, 3 tumors with solid mass with central cystic area, and other 10 tumors with unusual presentation. In 24 patients, second surgery or radiosurgery (Gamma-knife surgery) was performed for the recurrent and newly developed lesions. CONCLUSION: Although MRI findings of hemangioblastomas are usually typical, they may present with unusual and atypical radiological findings. Preoperative suspicion is important.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-051. TIME DELAYED CONTRAST ENHANCED MRI IMPROVES DETECTION OF BRAIN METASTASES AND APPARENT TREATMENT VOLUMES

Marina Kushnirsky ¹, Joel Katz ¹, Jonathan Knisely ¹, Michael Schulder ¹, Jared Steinklein ¹, Lisa Rosen ¹, Craig Warshall ¹, Vinh Nguyen ¹

Abstract

PURPOSE: Contrast-enhanced MRI is the preeminent diagnostic exam for brain metastases (BM). Treatment of BM with stereotactic radiosurgery (SRS) requires accurate detection of lesion volume and size, which may improve with a time delay following administration of gadolinium-based contrast. METHODS: 53 volumetric MRIs from 38 patients were evaluated. All studies were performed on 1.5 and 3T magnets using Gadobenate dimeglumin contrast (MultiHance^®). Three axial T1 post-contrast sequences were performed; one immediately after injection, and two at delays of 10 and 15 minutes. The studies were randomized and examined separately by 3 radiologists, who were blinded to their sequence. The number of tumors in each scan was compared with a Wilcoxon signed-rank test. Tumors ≥2 mm in diameter (n = 234) were contoured using SRS planning software and a mixed model approach examined volume changes. RESULTS: The interclass correlations (<0.72) for all three sequences demonstrated high inter-rater reliability. At least one new lesion was detected in the 2^nd scan as compared to the 1^st in 35.3% of subjects (range of new lesions detected 1-10). At least 1 new lesion was detected in the 3^rd scan as compared to the 2^nd in 21.6% of subjects (range 1-9). Between scans 1 and 3 additional tumors were seen on 41.2% of scans (range 1-14). There was a significant increase in number of BM detected from scans 1 to 2 (P <0.0367), and scans 1 to 3 (P <0.0264). We found an average 25.4% increase in BM volume between scan 1 and 2 (P < 0.0001), and a 9% increase between scans 2 and 3 (P < 0.0001). CONCLUSION: In patients who undergo SRS of BM, delayed MRI after contrast injection increases number and volume of BM to be treated. To avoid missing or undertreating tumors, we recommend MRI be acquired between 10 and 15 minutes after contrast injection.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-052. COMPUTERIZED IMAGE ANALYSIS OF TEXTURE DESCRIPTORS IN MULTI-PARAMETRIC MRI TO DISTINGUISH RECURRENT GLIOBLASTOMA MULTIFORME FROM RADIATION NECROSIS

Pallavi Tiwari ¹, Lisa Rogers ², Leo Wolansky ², Andrew Sloan ², Jill Barnholtz-Sloan ², Curtis Tatsauka ², Mark Cohen ², Anant Madabhushi ¹

Abstract

OBJECTIVE: To identify computer derived quantitative texture descriptors on multi-parametric MRI to distinguish radiation necrosis (RN) from recurrent GBM (rGBM) post-chemoradiation. Differentiating these two conditions (RN, rGBM) on standard MRI is clinically challenging, primarily because the morphological appearance of RN closely mimics that of rGBM. The underlying hypothesis of this study is that computer extracted texture features on MRI can capture subtle, multi-scale morphologic attributes for distinguishing RN versus rGBM, features that may not be visually appreciable by a radiologist. METHODS: 3-Tesla MRI studies with Gd-T1C, T2-FLAIR, T2-w, ADC, T1-w protocols in 17 patients obtained from 9 months to 2 years post-chemoradiation were analyzed. Ten patients were histologically confirmed with RN (> 80% RN) and seven with rGBM (>80% tumor). A set of 400 computer extracted texture descriptors (co-occurrence matrix homogeneity, neighboring gray-level dependence matrix, multi-scale Gaussian derivatives, and Laws) for modeling macro- and micro-scale morphologic changes within the treated lesion area for each MRI protocol were extracted. Triplets of CETDs from the set of 400 were chosen via randomized selection over 10000 iterations. Silhouette Index (SI), a measure to quantify separation between clusters (RN, rGBM) by computing distance between cluster centers (normalized between 0 (no separation between clusters) and 1 (perfect separation)) was recorded for every feature triplet. The triplets with highest SI were identified. RESULTS: 3D plot representation of the 17 studies demonstrated clear separation between the two classes (with 1 misclassified RN patient study) with recorded SI of 0.41 for gradient, law, and co-occurrence feature sets. Law and gradient texture descriptors appeared to capture soap-bubble enhancement and spreading wave-front patterns, qualitative observations previously described on MP-MRI to distinguish RN from rGBM. CONCLUSION: Computerized image analysis of multi-parametric MRI texture descriptors provides a quantitative assessment that appears highly accurate in distinguishing RN from rGBM by capturing subtle architectural details.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-053. INCREASED [⁶⁸Ga]-DOTATOC UPTAKE IN PET IMAGING DISCRIMINATES MENINGIOMA AND TUMORFREE TISSUE

Walter Rachinger ¹, Niklas Thon ¹, Alexander Haug ², Ulrich Schüller ³, Christian Schichor ¹, Jörg-Christian Tonn ¹

Abstract

OBJECTIVE: The somatostatin receptor II (SRII) analogue [⁶⁸Ga]-DOTATOC has been used for positron emission tomography (PET) imaging of meningiomas. However, no correlative analysis of increased [⁶⁸Ga]-DOTATOC uptake, SRII expression and histology has been published so far. This prospective study offers a spatially precise histopathological analysis of SRII expression using [⁶⁸Ga]-DOTATOC-PET-guided biopsies in primary and recurrent meningioma patients. METHODS: Adult patients with suspected intracranial meningiomas and without prior radiotherapy were included. PET analysis included maximum standard uptake value (SUVmax) within areas of biopsy inside and outside of the tumor. Magnetic resonance imaging (MRI), [⁶⁸Ga]-DOTATOC-PET integrated into neuronavigation and intraoperative computed tomography (iCT) were used for a neuronavigated tissue sampling procedure. Histopathological analysis included SRII expression. RESULTS: 12 patients (8x de-novo and 4x recurrent meningiomas; median age 49 years) were included. Overall, 59 specimens were collected. 48 samples displayed tumor, 11 samples showed tumorfree scar tissue or dura. A SUVmax >2.3 was significantly associated with meningioma tissue. Sensitivity, specificity and positive predictive value of SUVmax >2.3 to detect meningioma tissue was 95.8%, 81.8%, and 95.8%, respectively. SRII expression was exclusively found in tumor specimens. Less than 25% SRII positive tumor cells were found in 20, 25-50% positive cells in 10, 50-75% positive cells in 13, and more than 75% in 16 samples. A significant correlation was found between SUVmax and quantitative SRII expression (p = 0.001). Results did not significantly differ between primary and recurrent tumors. CONCLUSION: Increased [⁶⁸Ga]-DOTATOC uptake in PET allows a spatially precise discrimination between SRII expressing meningioma and tumorfree tissue, valuable both for surgery and radiation planning.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-054. INCREASED SENSITIVITY TO RADIOCHEMOTHERAPY IN IDH1 MUTANT GLIOBLASTOMA AS DEMONSTRATED BY SERIAL QUANTITATIVE MR VOLUMETRY

Anh Tran ^1,², Albert Lai ³, Sichen Li ³, Whitney Pope ², Stephanie Teixeira ², Robert Harris ^2,⁴, Davis Woodworth ^2,⁴, Phioanh Nghiemphu ³, Timothy Cloughesy ³, Benjamin Ellingson ^2,⁴

Abstract

BACKGROUND: Isocitrate dehydrogenase 1 (IDH1) mutations have been linked to favorable outcomes in patients with glioblastoma multiforme (GBM). Recent in vitro experiments suggest IDH1 mutation sensitizes tumors to radiation damage. We hypothesized radiographic treatment response would be significantly different between IDH1 mutant versus wild-type GBMs after radiotherapy (RT) and concurrent temozolomide (TMZ). METHODS: A total of n = 39 de novo GBM patients with known IDH1 mutational status (n = 10 IDH1 mutants), who followed standard therapy and had regular post-contrast T1W (T1 + C) and T2W/FLAIR follow-up images in the 6-month period after starting RT. The volumes of contrast-enhancing and FLAIR hyperintensity were calculated from each scan. Volumes measured in scans immediately after RT were used as baseline. Linear regression models were used to estimate the rate of change in VT1 + C and VFLAIR after RT for each individual patient. RESULTS: IDH1 mutant GBMs demonstrated a favorable response to RT/TMZ in the study period, as demonstrated by 10/10 mutants showing radiographic response in decreasing VT1 + C rate, compared to 13/29 wild-types (p < 0.001, binomial). During the study period, VT1 + C and VFLAIR changed at -3.6%/week and +0.6%/week in IDH1 mutant tumors, respectively, as compared to +0.8%/week and +5.2%/week in IDH1 wild-type tumors (p = 0.0076 and p = 0.0118). Amongst the radiographic responders, IDH1 mutant GBMs still demonstrated significant progression-free and overall survival benefit (Kaplan-Meier, p = 0.014 and p = 0.045 respectively) . Aggregated tumor kinetics by IDH1 mutational status showed significant lower rate of change in IDH1 mutant GBMs in specific periods: after 105 days for VFLAIR, and 95-120 and after 150 days for VT1 + C, from starting RT/TMZ. CONCLUSIONS: The current study supports the hypothesis that IDH1 mutant GBMs are more sensitive to radiochemotherapy than IDH1 wild-type GBMs, as demonstrated by tumor kinetics from radiographic response. We also showed that survival benefit from IDH1 mutant is independence from early radiographic response.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-055. T1 RHO MAPPING OF BRAIN TUMORS AND TUMOR-RELATED EDEMA

Javier Villanueva-Meyer ¹, Ramon Barajas ¹, Marc Mabray ¹, Igor Barani ¹, Weitian Chen ², Ajit Shankaranarayanan ², Patrick Koon ², Soonmee Cha ¹

Abstract

PURPOSE: Standard MR imaging techniques are limited in differentiating and quantifying peritumoral vasogenic edema often associated with metastatic carcinoma or meningioma from infiltrative edema comprised of both water and tumor associated with infiltrating glioma. The purpose of our study was to determine the ability of quantitative T1 rho imaging and ADC mapping to differentiate the two different types of tumor-related edema. METHODS AND MATERIALS: Nineteen patients with intracranial brain tumors: 9 with grade I-III gliomas, 5 with glioblastoma multiforme (GBM), 3 with metastasis, and 2 with meningioma; as well as five healthy volunteers were scanned using a 3 Tesla scanner. Conventional MR imaging, diffusion-weighted MR imaging, and 3D T1 rho imaging was performed. T1 rho and ADC values were obtained in regions of contrast enhancement, peritumoral white matter T2 abnormality, and normal-appearing white matter. Additionally, T1 rho values in different regions of white matter within the brain were measured in the healthy volunteers. RESULTS: T1 rho values of vasogenic edema were statistically higher than infiltrative edema related to GBM and grade I-III gliomas (191.92 + 32.72 vs 133.62 + 24.51 vs 132.45 + 23.54; P = 0.002). ADC values in vasogenic edema associated with metastases or meningiomas were significantly higher than those in infiltrative edema related to GBM and grade I-III gliomas (1654.19 + 125.74 vs 1263.05 + 206.25 vs 1248.14 + 135.76; P = 0.0008). ADC and T1 rho values amongst grade I-III gliomas, GBM, and metastases/meningioma were not statistically significant. CONCLUSION: T1 rho and ADC values provide quantitative assessment of intracranial tumor-related edema. Quantitative T1 rho and ADC values were significantly different between vasogenic edema and infiltrative edema. T1 rho and ADC maps can aid in preoperative diagnosis and management. Additionally, T1 rho provides the advantage of high-resolution capability without distortion and as such can be used to guide surgical biopsy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-056. ASSOCIATION BETWEEN MR IMAGING MEASUREMENTS AND TUMOR V.S NON-TUMOR IMAGE-GUIDED TISSUE SAMPLES IN PATIENTS WITH RECURRENT GBM

Qiuting Wen ¹, Adam Elkhaled ¹, Emma Essock-Burns ¹, Annette Molinaro ¹, Joanna Phillips ¹, Susan Chang ¹, Soonmee Cha ¹, Sarah Nelson ¹

Abstract

INTRODUCTION: Radiation and chemotherapy may result in gliosis, edema and necrosis, which can mimic tumor recurrence on standard MR images. Differentiating between these effects is a challenge in neuro-oncology. Acquisition of image-guided tissue samples enables the association of pathological properties of the tissue with pre-surgical MR parameters. This study was to evaluate which in vivo and ex vivo MR parameters were able to distinguish between tumor and treatment effect for patients with GBM. METHODS: 152 image-guided tissue samples were collected from 73 recurrent patients with GBM. In vivo MR imaging was obtained pre-operatively including diffusion, perfusion, and proton spectroscopy. Tissue samples were selected from the suspected tumor regions, and were given a tumor score based on the contribution of tumor to total cellularity, with 0 considered to be non-tumor and 2 or 3 to be tumor. The ex vivo HRMAS data were acquired at 11.7 T. A generalized estimated equation was performed to account for the fact that multiple samples were taken from the same patient. RESULTS: 88 samples were identified as tumor and 64 as non-tumor. Perfusion peak height (PH) was found to be higher in tumor. Both in vivo and ex vivo spectra showed significantly lower NAA in tumor, and creatine (Cr) levels from HRMAS were decreased in tumor. ADC was found to be lower but PH was higher in the CEL. DISCUSSION: Our results showed the ability of PH, in vivo NAA, ex vivo NAA and Cr in differentiating tumor from non-tumor/treatment effect samples, which is consistent with the clinical findings that tumor recurrence has elevated angiogenesis and causes more neuronal disruption. Although these differences were significant on a population basis, there was overlap in values for individual samples. Future studies will investigate the use of multi-variate indices to map out regions of recurrent tumor.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-057. AMIDE PROTON TRANSFER MRI OF MALIGNANT GLIOMAS FOR DETECTING ACTIVE VERSUS INACTIVE TUMOR

David Wolf ¹, Xiaobu Ye ¹, Michael Lim ¹, He Zhu ^1,², Meiyun Wang ², Alfredo Quinones-Hinojosa ¹, Jon Weingart ¹, Alessandro Olivi ¹, Peter van Zijl ^1,², John Laterra ¹, Jinyuan Zhou ^1,², Jaishri Blakeley ¹

Abstract

Although promising therapies are being advanced for patients with malignant gliomas, a major limitation in developing these therapies is the lack of specificity of neuro-imaging, particularly in distinguishing between tumor progression and treatment-induced injury. Amide proton transfer (APT) MRI is a protein-based molecular imaging technique that uses the water signal to detect amide protons of endogenous mobile proteins in tissue. Our prior data suggest that high APT is a typical feature of active tumor. We hypothesized here that patients with <2% APT intensity after chemoradiation would have a longer progression free survival compared to patients with high APT. METHODS: Patients with newly diagnosed malignant glioma underwent standard and APT MR imaging at 3T one month after completion of chemoradiation. Patients were followed for time to progression (defined as either pathologically confirmed active tumor post-treatment or a change in therapy based on clinical assessment of progression) as well as overall survival. RESULTS: Nineteen subjects, 14 male, were enrolled. Eleven had glioblastoma and 8 anaplastic astrocytoma. Twelve subjects had high APT intensity (63.1%) and 7 had low/normal APT intensity. Dose of steroids at time of imaging, time from initial imaging, and dose of radiation treatment was similar between groups. All but one patient received temozolomide with radiation. Median time to progression was 122 days (range 19-488) in the high APT group and 472 days (range 47-698) in the low APT group. Median overall survival was 343 days (range 155-649) in the high APT group and 662 days (495-794) in the low APT group. CONCLUSIONS: APT imaging may allow early detection of treatment response or active tumor in patients with malignant glioma after chemoradiation.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-058. PRE-OPERATIVE MRI DIFFUSION CHARACTERISTICS OF THE TUMOUR AND PERITUMORAL REGION MAY PREDICT SURVIVAL AND LOCAL RECURRENCE FOR OPERATED BRAIN METASTASES

Rasheed Zakaria ^1,², Kumar Das ¹, Vanessa Sluming ², Maneesh Bhojak ¹, Carol Walker ¹, Michael D Jenkinson ^1,²

Abstract

AIM: We investigated pre-operative prognostic factors from diffusion weighted MRI as recent studies have suggested this may correlate with overall survival for resected brain metastases. METHOD: Data was gathered retrospectively following resection of brain metastases at a single UK centre. Pre-operative MR imaging was analysed and apparent diffusion coefficient (ADC) maps produced using clinically available and non-proprietary software. A variety of measures were taken including mean/minimum tumour ADC, mean ADC of near and distant peritumoral regions and the ADC gradient across the tumour-brain boundary. RESULTS: 34 patients underwent resection of a single lesion followed by WBRT in 23. Mean age was 61 years with modal RPA class 2, GPA score 2.5 and median overall survival 6.3 months. Final primaries included 11 non-small cell lung, 5 small cell lung, 4 melanoma, 3 breast (all ER +ve, HER2 -ve) and 3 colorectal. Eight mets showed hyperintense DWI signal and had a shorter median overall survival (4.3 months) than those with iso or hypointense signal (6.6 months). Mean tumoral ADC was 0.816 ^x10-3mm²/sec and those with a mean ADC greater than this showed a significantly longer median survival after resection + WBRT (12.5 months, log rank test p = .023) than those with lower ADC and the same treatment (6.15 months). The mean ADC of the immediate peritumoral region was shown to be highly variable between tumours, perhaps reflecting differing cellularity or infiltration and was a significant predictor of survival with a higher value in this region correlating with longer overall survival (9.1 vs. 4.8 months, log rank test p = .038). CONCLUSIONS: ADC may be used to predict survival in brain metastases. Further investigation is needed to determine which metrics are most appropriate for clinical use. Analysis of the resected lesions for cellularity, stromal density and expression of metastasis inducing proteins is being undertaken.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii191–iii205.

RA-059. AN OPEN-LABEL, SINGLE ARM STUDY TO EXPLORE WHETHER POTENTIAL IMAGE BIOMARKERS CORRELATE WITH EFFICACY OF BEVACIZUMAB COMBINED WITH CONVENTIONAL THERAPY IN NEWLY DIAGNOSED GLIOBLASTOMA

Shuanghu （Tiger） Yuan ¹, Rongjie Tao ¹, Guoren Yang ¹, Zhaoqiu Chen ¹, Dianbin Mu ¹, Shuqiang Zhao ¹, Zheng Fu ¹, Wanhu Li ¹, Jinming Yu ¹

Abstract

BACKGROUND: Glioblastoma multiforme (GBM) is one of the most aggressive cancers and associated with high levels of angiogenesis. Bevacizumab, a new type of anti-tumor angiogenesis monoclonal antibody, precisely targets VEGF to inhibit angiogenesis for continuous tumor control. Bevacizumab has been suggested effective and well tolerated with concurrent radiation therapy and temozolomide in newly diagnosed GBM. We hypothesized that, during bevacizumab therapy, the high and low perfusion changes in specific regions of GBM might predict its efficacy in GBM. In current study, we will use DCE-MRI, DSC-MRI and 18F-Galacto-RGD PET to explore the potential image biomarkers correlated with efficacy of bevacizumab combined with conventional therapy in newly diagnosed GBM. METHODS: This is a single-center, open-label, single arm study. 20 patients with newly diagnosed, histopathologically confirmed GBM (WHO Grade IV) and after surgery will be enrolled. All patients will receive conventional concurrent chemoradiation (radiotherapy: 2 Gy/day, 5 days/week, 60 Gy over 6 weeks; temozolomide: 75 mg/m² p.o. during radiation; bevacizumab: 10 mg/kg, days 1 and 15 of each 28-day cycle, starting at the 4th week of radiation) and adjuvant temozolomide plus bevacizumab (temozolomide: 150 mg/m²/day (first cycle) and 200 mg/m²/day (subsequent cycles if possible) p.o., 5 days of a 28-day cycle; bevacizumab: 10 mg/kg, days 1 and 15 of each 28-day cycle; 6 cycles in total). At the initiation of chemoradiation, week 3 and 10 of the study, relative CBV (rCBV) in DCE-MRI, K^trans in DSC-MRI and SUV_mean, SUV_max and T/NT in RGD PET will be collected. The primary aim is to assess the association between above image biomarker change and progression-free survival (PFS). PFS is set as the primary endpoint while secondary endpoints include overall survival (OS), overall response rate (ORR), health-related quality of life (HRQoL) and safety profile. Recruitment will begin in Sep 2013 and the estimated date of final data collection is Oct 2014.

PERMALINK

RADIOLOGY

RA-001. MRI CHANGES ASSOCIATED WITH BEVACIZUMAB DIFFER BETWEEN TUMOR RECURRENCE AND CEREBRAL RADIATION NECROSIS

Yoshiki Arakawa

Ko-ichi Fujimoto

Daiki Murata

Yuji Nakamoto

Tomohisa Okada

Susumu Miyamoto

Abstract

RA-002. BASIC IMAGING MARKER DEPICT A REGIONALLY INCREASED BEVACIZUMAB ACTIVITY IN RECURRENT GLIOBLASTOMA ASSOCIATED WITH SUBSTANTIALLY PROLONGED SURVIVAL

Oliver Bähr

Patrick N Harter

Lutz Weise

Se-Jong You

Michael W Ronellenfitsch

Johannes Rieger

Joachim P Steinbach

Elke Hattingen

Abstract

RA-003. QUANTITATIVE T2-MAPPING OF RECURRENT GLIOBLASTOMA UNDER BEVACIZUMAB IMPROVES MONITORING FOR NON-ENHANCING TUMOR PROGRESSION AND PREDICTS OVERALL SURVIVAL

Oliver Bähr

Alina Jurcoane

Keivan Daneshvar

Ulrich Pilatus

Michel Mittelbronn

Joachim P Steinbach

Elke Hattingen

Abstract

RA-004. CORRELATION OF ADC MEAN AND ADC MEAN RATIO IN RADIOGRAPHIC PSEUDOPROGRESSION FOLLOWING TREATMENT WITH ERC 1671 IMMUNOTHERAPY IN RECURRENT GLIOBLASTOMA

Jose Carrillo

Daniela Bota

Jason Handwerker

Lydia Min-Ying Su

Thomas Chen

Apostolos Stathopoulos

Hon Yu

Abstract

RA-005. PET IMAGING OF BRAIN GLIOMAS WITH C-11 ACETATE: COMPARISON WITH F-18 FDG PET

Jong Hee Chang

Eui Hyun Kim

Se Hoon Kim

Mi

Jin Yun

Abstract

RA-006. AUTOPSY AND MAGNETIC RESONANCE IMAGING CORRELATES IN TWO GLIOBLASTOMA PATIENTS ON BEVACIZUMAB

Peter Pytel

John Collins

Yoon Choi

Rimas Lukas

Martin Nicholas

Abstract

RA-007. DIFFUSION IMAGING GENOMIC MAPPING IDENTIFIES GENOMIC TARGETS INVOLVED IN INVASION AND POOR PROGNOSIS

Rivka Colen

Ryan Jafrani

Pascal Zinn

Abstract

RA-008. IMAGING GENOMIC BIOMARKER SIGNATURE FOR MGMT PROMOTOR METHYLATION IDENTIFICATION

Rivka Colen

Omar Ashour

Pascal Zinn

Abstract

RA-009. IMAGING GENOMIC CORRELATION OF INVASIVE GENOMIC COMPOSITION AND PATIENT SURVIVAL DEMONSTRATES METABOLIC DYSFUNCTION: A TCGA GLIOMA PHENOTYPE RESEARCH GROUP PROJECT

Rivka Colen

Mark Vangel

David Gutman

Scott Hwang

Max Wintermark

Rajan Jain

Manal Jilwan-Nicolas

James Chen

Prashant Raghavan

Chad Holder

Daniel Rubin

Eric Huang

Justin Kirby

John Freymann

Carl Jaffe

Adam Flanders

Pascal Zinn

RA-017. PROLIFERATION RATE ESTIMATES DERIVED FROM SERIAL DIFFUSION MRI CORRELATES WITH ¹⁸F-FLT PET SUV VALUES IN RECURRENT GLIOBLASTOMA TREATED WITH BEVACIZUMAB