SSA17

Neuroradiology (Brain Tumors)

Sunday, Nov. 27 10:45AM - 12:15PM Room: N230B

NRMR

AMA PRA Category 1 Credits ™: 1.50
ARRT Category A+ Credits: 1.50

FDA Discussions may include off-label uses.

Shinji Naganawa, MD, Nagoya, Japan (Moderator) Nothing to Disclose
Whitney B. Pope, MD, PhD, Los Angeles, CA (Moderator) Research Consultant, F. Hoffmann-La Roche Ltd; Research Consultant, Amgen Inc; Research Consultant, Tocagen Inc; ;
Sub-Events
SSA17-01
Iris Chen, MS, New York, NY (Presenter) Nothing to Disclose
Marco Hefti, MD, New York, NY (Abstract Co-Author) Nothing to Disclose
Amish H. Doshi, MD, New York, NY (Abstract Co-Author) Nothing to Disclose
Nadejda Tsankova, MD, New York, NY (Abstract Co-Author) Nothing to Disclose
Bradley N. Delman, MD, New York, NY (Abstract Co-Author) Consultant, Bayer AG; Speaker, Bayer AG
Adilia Hormigo, MD, New York, NY (Abstract Co-Author) Nothing to Disclose
Kambiz Nael, MD, New York, NY (Abstract Co-Author) Research Consultant, Olea Medical
PURPOSE

Apparent diffusion coefficient (ADC) has been shown to inversely correlate with tumor cellular density. Conventional imaging using serial MRIs is often challenging in evaluating progression of low-grade gliomas (LGG). The purpose of this study was to assess the role of MR diffusion in serial imaging of patients with LGG as a parameter of stability vs. high-grade transformation (HGT).

METHOD AND MATERIALS

Patients with histologically proven LGG were included in this retrospective study if they had consecutive clinical and imaging follow-up from the initial diagnosis until they underwent second surgical biopsy. All available MR exams were coregistered and ADC histogram measures determined for each patient using volume-of-interest from the FLAIR hyperintense tumor volume. Normalized ADC-10th percentile values were obtained for each time point, data were plotted over time for each patient and scored to evaluate whether values fit within the expected pattern: HGT (interval decrease in ADC); Stable (plateau, or interval increase in ADC). 

RESULTS

Twenty patients with histologically proven LGG (Grade II oligodendroglioma, n=11; Grade II oligoastrocytoma n=4; Grade I-II astrocytomas, n=5) were followed up for a median of 3 years (range, 1.2 –5.5 years). A total of 60 MRI studies were evaluated (3 MRIs per patient). Based on surgical pathology on repeat biopsy, histological grade in 7 patients remained stable (grade II), while 13 patients transformed to a high-grade (Grade III n=11, grade IV n=2). Progressive downward trend of ADC was significantly associated with HGT (p=0.01, Odds Ratio:45). Sequential ADC analysis showed progressive downward trend in 10/13 (77%) in patients with HGT and remained stable/plateau in all patients who continued to be histologically stable (7/7). There was significant correlation between downward trend of ADC and HGT (r=0.74, p=0.002).  Increase tumor volume measured from volumetric analysis from FLAIR imaging was not a significant association with HGT (p=0.6, Odds ratio:0.9). 

CONCLUSION

Sequential ADC analysis in patients with LGG can help to identify tumor progression. Downward trend of ADC values can predict HGT despite apparent stability of tumor size and extent on conventional imaging. 

CLINICAL RELEVANCE/APPLICATION

Sequential progressive decrease in ADC values is a helpful imaging parameter for follow up of patients with low-grade glioma and is highly associated with high-grade transformation.

SSA17-02
Stefanie C. Thust, MD,FRCR, London, United Kingdom (Abstract Co-Author) Nothing to Disclose
Tarek Yousry, London, United Kingdom (Abstract Co-Author) Nothing to Disclose
Nuria Bargallo, PhD, Barcelona, Spain (Abstract Co-Author) Nothing to Disclose
Meike W. Vernooij, MD, Rotterdam, Netherlands (Abstract Co-Author) Nothing to Disclose
Marion Smits, MD, PhD, Rotterdam, Netherlands (Presenter) Nothing to Disclose
PURPOSE

To assess current practices of glioma imaging throughout Europe, in particular to identify controversies, uncertainties and potential technical hurdles.

METHOD AND MATERIALS

Survey invitations were emailed to ESNR members (n=1662) and known associates (n=6400), European national radiologists’ societies and posted via social media. To avoid duplicate bias, participants were instructed to supply institution details or confirm they were the only person answering from their center. The questionnaire featured 87 individual items divided into multiple choice, single best choice and free text questions on personal practice and preferred techniques.

RESULTS

224 individuals responded by 1/4/2016. Demographics: 79.9% were neuroradiologists, 9.4% general radiologists, 6.3% trainees and 4.5% other professions. More worked at academic (59.2%) than general (37.7%) hospitals with varied science support (none 39.3%, general physics 36.2%, neurophysics 23.2%). Conventional MRI: 99.1% read DWI, but methods to determine restriction differed (visual 76.8%, ADC measurement 17.4%, advanced analysis 3.1%). 82.1% performed 3D acquisitions, mostly T1 and FLAIR. Advanced MRI: 85.3% used perfusion (PWI), (83.2% DSC, 28.3% DCE, 11.5% ASL), with 48.7% acquiring PWI universally. 81.2% used MRS (single voxel 72.4%, multi-voxel 62.4%), but less routinely (21.5%). fMRI was added by 48.9% and DTI by 63.7% with around 60% results integrated into navigation systems. Among barriers to progress, lack of facility or software, time constraints and no clinician requests outweighed reimbursement issues. Scenarios: PWI was the favoured modality to identify pseudoprogression (56.1%). 22.9% used a report template, 59.9% issued qualitative reports and 27.9% obtained RANO measures. Postoperative imaging was routinely performed by 74.8% within 24-72hrs, and 17.2% reported a percent measure of resection.

CONCLUSION

This survey gathered valuable information on differences of equipment, levels of experience, financial and healthcare economic factors, which may determine the practicality of guidelines.

CLINICAL RELEVANCE/APPLICATION

The results of this analysis will inform and support the publication of practical recommendations to image glioma patients with MRI.

SSA17-03
Yoon Seong Choi, MD, Seoul, Korea, Republic Of (Presenter) Nothing to Disclose
Ho Joon Lee, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Sung Soo Ahn, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Jinna Kim, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Seung-Koo Lee, MD, PhD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Yaewon Park, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Sohi Bae, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Tyler Hyungtaek Rim, Gyeonggi-do,, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
PURPOSE

To investigate the value of initial area under the curve (IAUC) derived from dynamic contrast-enhanced magnetic resonance imaging (DCE MRI) combined with MGMT promoter methylation status for predicting survival outcome in patients with glioblastoma (GBM).

METHOD AND MATERIALS

This retrospective study included 88 patients with GBM who underwent preoperative DCE MRI. The mean and 75th percentile (p75) of IAUC values at 30 (IAUC30) and 60 seconds (IAUC60) were acquired from the entire enhancing tumors. Univariate survival analyses were performed for overall survival (OS) and progression-free survival (PFS) with IAUC, MGMT, other clinical factors, and conventional MRI findings using the Kaplan-Meier method and Cox regression. Subgroup univariate analyses were performed with IAUC according to MGMT status. The multivariate models were built with and without IAUC parameters. The diagnostic accuracy and improvement in 1.5-year OS and 1-year PFS prediction of the models after adding the IAUC parameters were evaluated using receiver operating characteristic (ROC) analyses and net reclassification index (NRI). The IAUC parameters were compared according to MGMT status.

RESULTS

High IAUC parameters were associated with worse OS and PFS in the unmethylated MGMT group, but not in the methylated group and in the entire cohort. In the unmethylated MGMT group, the diagnostic accuracies for 1.5-year OS and 1-year PFS prediction were improved when IAUC parameters were added (OS area under the ROC curve [AUC], 0.87 - 0.90 and PFS AUC, 0.69 – 0.71) to the models with clinical factors and conventional MRI findings. (OS AUC, 0.77 and PFS AUC, 0.66). This improvement was significant for 1.5-year OS when the mean and 75th percentile of IAUC30, and the 75th percentile of IAUC60 were added (P = 0.001, 0.001, and 0.056 by NRI, respectively). IAUC parameters were higher in the methylated MGMT group than the unmethylated group (P < 0.05 for all).

CONCLUSION

IAUC parameters combined with MGMT status can be useful for predicting survival outcome in patients with GBM, allowing more accurate prognostication of GBM with unmethylated MGMT promoter.

CLINICAL RELEVANCE/APPLICATION

IAUC parameters combined with MGMT status can be used postoperatively as a predictor of survival outcome of GBM, allowing more accurate prognostication of GBM with unmethylated MGMT.

SSA17-04
Zhongwei Zhang, MD, PhD, Winston-Salam, NC (Presenter) Nothing to Disclose
Dawen Zhao, Winston Salam, NC (Abstract Co-Author) Nothing to Disclose
Christopher T. Whitlow, MD, PhD, Winston-Salem, NC (Abstract Co-Author) Nothing to Disclose
Mark E. Schweitzer, MD, Stony Brook, NY (Abstract Co-Author) Consultant, MMI Munich Medical International GmbH Data Safety Monitoring Board, Histogenics Corporation
PURPOSE

There is increasing evidence that hypoxia (lack of oxygen) is associated with tumor aggressive, resistance to radiation therapy and increased likelihood of biochemical failure and metastatic spread. Brain tissue oxygen tension (pO2) measurement is essential for characterizing the relationship between O2 and metabolism, and for assessing the consequences resulting from an inadequate supply. The purpose of this study was to non-invasively measure human brain tumors oxygen tension (pO2) using 1H MR Oximetry Imaging (MOXI).

METHOD AND MATERIALS

A total of 16 patients with histologically confirmed brain tumors were included in this study. All subjects underwent 1H MRI at 3.0 T including anatomic MR imaging, quantitative T1, T2 mappings and multiple b-value single-shot EPI diffusion measurements. T1 mapping was acquired using variable flip angle gradient echo sequence with 4 flip angles (2°, 7°, 15°, 25°), TR: 5ms. T2 mapping was acquired with a standard CPMG sequence with parameters: TR = 2500ms, 20 echoes ranging from 10ms to 200ms with 10ms echo spacing. EPI diffusion parameters are: TR/TEeff = 4500/98ms and 10 b values with 0, 50, 100, 150, 200, 300, 450, 600, 900 and 1200 seconds/mm2. The brain tumor pO2 was determined using MOXI algorithm which the accuracy was validated by 19F MRI. MOXI for pO2 estimation is based on IVIM diffusion MRI and the dependence of the blood R2 relaxation rate on the inter-echo spacing measured using a multiple spin-echo CPMG sequence and weak-field diffusion model.

RESULTS

Presentative brain tumor pO2 maps were shown in Figure (Fig.A Glioblastoma III and Fig.B Meningioma). Noticeable heterogeneity in pO2 levels was observed throughout the different tumor types with large pO2 variations. The intratumoral and peritumoral pO2 distribution can be clearly revealed by MOXI. 

CONCLUSION

This study demonstrated the feasibility of determining brain tumor pO2 by the use of MOXI. The non-invasive nature of this approach makes it particularly attractive for longitudinally assessing pO2 in tumor. Findings of this pilot study will support MOXI as a promising tool for monitoring individual tumor oxygenation with potential applications for prognostic indicator of therapeutic response.

CLINICAL RELEVANCE/APPLICATION

MOXI provides a non-invasive promising tool for monitoring brain tumor oxygenation.

SSA17-05
Johann-Martin Hempel, MD, Tubingen, Germany (Presenter) Nothing to Disclose
Sotirios Bisdas, MD, Frankfurt, Germany (Abstract Co-Author) Nothing to Disclose
Jens Schittenhelm, MD,PhD, Tubingen, Germany (Abstract Co-Author) Nothing to Disclose
Cornelia Brendle, MD, Tubingen, Germany (Abstract Co-Author) Nothing to Disclose
Benjamin Bender, Tuebingen, Germany (Abstract Co-Author) Nothing to Disclose
Henk Wassmann, Tubingen, Germany (Abstract Co-Author) Nothing to Disclose
Marco Skardelly, Tubingen, Germany (Abstract Co-Author) Nothing to Disclose
Ghazaleh Tabatabai, Tubingen, Germany (Abstract Co-Author) Nothing to Disclose
Salvador Castaneda Vega, Tubingen, Germany (Abstract Co-Author) Nothing to Disclose
Ulrike Ernemann, Tuebingen, Germany (Abstract Co-Author) Nothing to Disclose
Uwe Klose, PhD, Tuebingen, Germany (Abstract Co-Author) Nothing to Disclose
PURPOSE

To assess the diagnostic performance of diffusion kurtosis imaging (DKI) for in vivo molecular profiling of human glioma.

METHOD AND MATERIALS

Mean kurtosis (MK) and mean diffusivity (MD) metrics from DKI were prospectively assessed in 44 patients with histopathologically confirmed glioma. The results were compared in regard to WHO-based histological findings and molecular characteristics: isocitrate-dehydrogenase (IDH1/2) mutation status, alpha-thalassemia/mental retardation syndrome X-linked (ATRX) expression, chromosome 1p/19q loss of heterozygosity (LOH), and O6-methylguanine DNA methyltransferase (MGMT) promoter methylation status.

RESULTS

MK was significantly lower in patients with IDH1/2 mutation (0.55 ± 0.12, p=0.002) and ATRX loss of expression (0.54 ± 0.15, p=0.036) than in those with IDH1/2 wild type (0.69 ± 0.13) and ATRX retained expression (0.62 ± 0.13), respectively. Regarding the “integrated” molecular diagnosis, MK was significantly higher in primary glioblastoma (0.70 ± 0.13, p=0.002) than in astrocytoma (0.51 ± 0.14).

CONCLUSION

MK may be used to provide insights into the molecular profile of human glioma.

CLINICAL RELEVANCE/APPLICATION

Considering the diagnostic and prognostic significance of IDH1/2 mutation status and ATRX expression, MK appears as a promising in vivo biomarker for glioma. The diagnostic performance of MK seems to fit more with the “integrated” molecular approach than the conventional histological findings of the current WHO 2007 classification.

SSA17-06
Francesco Prada, MD, Milan, Italy (Presenter) Nothing to Disclose
Valerio Vitale, MD, lecco, Italy (Abstract Co-Author) Nothing to Disclose
Massimiliano Del Bene, Legnano, Italy (Abstract Co-Author) Nothing to Disclose
Carlo Boffano, MD, Alba, Italy (Abstract Co-Author) Nothing to Disclose
Giovanni Mauri, MD, Milan, Italy (Abstract Co-Author) Consultant, Esaote SpA
Luca Maria Sconfienza, MD, PhD, Milano, Italy (Abstract Co-Author) Travel support, Bracco Group
Ludovico D'Incerti, Milano, Italy (Abstract Co-Author) Nothing to Disclose
Luigi Solbiati, MD, Rozzano, Italy (Abstract Co-Author) Nothing to Disclose
Georgios Sakas, PhD, Darmstadt, Germany (Abstract Co-Author) Nothing to Disclose
Francesco DiMeco, Milano, Italy (Abstract Co-Author) Nothing to Disclose
PURPOSE

To compare intraoperative contrast enhanced ultrasound (CEUS) images to the correspondent co-planar T1 weighted contrast-enhanced magnetic resonance images (gdMRI) using fusion imaging between CEUS and pre-op MRI in glioblastoma (GBM). 

METHOD AND MATERIALS

Ten patients with GBM diagnosis were retrospectively enrolled. All patients underwent tumor excision guided by navigated intra-operative US (ioUS) based on fusion imaging between ioUS and pre-operative MRI. Navigated CEUS scans were performed after intravenous administration of ultrasound contrast agents (CA), before tumor resection. Using fusion imaging we compared CEUS contrast enhancement (location, morphology, margins, dimensions, and pattern) to that of gdMRI 

RESULTS

Registration between pre-operative gdMRI and ioUS demonstrated an error less the 2mm. In all cases CEUS highlighted the lesion. Contrast enhancement of gdMRI and CEUS was superimposable in all cases for location, margins, dimensions, and morphology while the pattern was the same in 9/10 cases; in one case the pattern was different.

CONCLUSION

CEUS contrast enhancement location, margins, morphology, and dimensions are superimposable to that provided by pre-operative gdMRI in all cases; while the pattern is the same in most of the cases.Taking into account that the goal of GBM resection is to remove all the gdMRI enhanced area, the information obtained with CEUS are of paramount importance in surgical management of GBM. 

CLINICAL RELEVANCE/APPLICATION

CEUS permits to have in intra-operative setting information on location, morphology, margins, and dimensions of the GBM identical to that obtained pre-operatively with gdMRI, and can be used as an intra-operative guidance in GBMs removal

SSA17-07
Ji Eun Park, Seoul, Korea, Republic Of (Presenter) Nothing to Disclose
Ho Sung Kim, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Sang Joon Kim, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Seung Chai Jung, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
Choong Gon Choi, MD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
PURPOSE

To determine if cerebral blood flow (CBF) could have predictive role in patients with glioblastoma with ongoing adjuvant temozolomide (TMZ) at first-line treatment and determine usefulness of perfusion fraction of CBF for an entire contrast enhancing lesion. 

METHOD AND MATERIALS

Forty-seven patients with glioblastoma ongoing adjuvant TMZ cycles underwent arterial spin labeling (ASL) MR immediately after concurrent TMZ-radiation therapy (CCRT), and followed up clinically with MR imaging (median follow up, 250 days). Region-of-interests were drawn on ASL where increased CBF compared to contralateral normal gray matter and volume-based perfusion fraction of increased CBF for an entire contrast material-enhanced lesion was calculated. Then, patients were dichotomized to positive- or negative-CBF according to 5, 25, 75, and 95 percentile perfusion fraction cutoffs. Log-rank tests were used to evaluate the association between dichotomized CBF and time to progression by using Kaplan-Meier curves. 

RESULTS

Patients with positive-CBF group showed significantly longer median TTP compared to negative-CBF group, and using 75 percentile perfusion fraction cutoff provided best prediction to TTP (positive-CBF group = 4 months vs negative-CBF group = 10 months; log-rank test, P <.0001). Longer median TTP in positive- CBF group was observed using 5, 25, 50, and 95 percentile perfusion fraction cutoffs (positive-CBF group = 4-7 months vs negative-CBF group = 10-21 months; log-rank test, all P = <.001).

CONCLUSION

Cerebral blood flow can be used to stratify progression in patients with glioblastoma with ongoing adjuvant TMZ, and perfusion fraction of CBF showed robust result on stratifying positive- and negative-CBF group.

CLINICAL RELEVANCE/APPLICATION

Positive-and negative-CBF group might be used to predict progression in patients with glioblastoma with ongoing adjuvant TMZ at their first line treatment. Not only qualitative evaluation of CBF, semiquantitative CBF measurements using volume-based perfusion fraction showed robust result on stratifying positive- and negative- CBF group. 

SSA17-08

Awards
Student Travel Stipend Award

Rachel L. Delfanti, MD, San Diego, CA (Presenter) Nothing to Disclose
Carrie R. McDonald, PhD, La Jolla, CA (Abstract Co-Author) Consultant, CorTechs Labs, Inc
Kelly Leyden, San Diego, CA (Abstract Co-Author) Nothing to Disclose
Anithapriya Krishnan, PhD, La Jolla, CA (Abstract Co-Author) Nothing to Disclose
David Piccioni, MD,PhD, La Jolla, CA (Abstract Co-Author) Nothing to Disclose
Nikdokht Farid, MD, San Diego, CA (Abstract Co-Author) Nothing to Disclose
Jason Handwerker, MD, Oceanside, CA (Abstract Co-Author) Nothing to Disclose
PURPOSE

Tumor genotyping in neuro-oncology holds promise to differentiate histologically similar tumors and provide prognostic information. Isocitrate dehydrogenase (IDH) status has been validated as a prognostic and predictive marker, with IDH wildtype (WT) lower grade gliomas behaving analogous to glioblastoma. Similarly, codeletion of 1p19q (1p19qcodel) is established as both a prognostic and predictive marker. A novel marker, inactivating mutations of alpha-thalassemia/mental retardation syndrome X-linked (ATRX) gene is associated with IDH mutations and is mutually exclusive with 1p19qcodel. Collectively, these markers comprise three distinct molecular groups of lower grade glioma: 1) IDH wildtype (WT); 2) IDH mutant (MT)-1p19qcodel-ATRX intact; 3) IDH MT-1p19q intact-ATRX loss. Therefore, we aim to determine if these molecular subgroups can be predicted by distinct imaging characteristics.

METHOD AND MATERIALS

Pre-treatment brain MRIs were analyzed for 42 patients with pathologically proven lower grade gliomas (WHO grade II or III) by a neuroradiologist, blinded to the pathologic diagnosis and molecular status. FLAIR, post-contrast, and diffusion-weighted sequences were quantitatively evaluated and ranked. The Fisher exact test was used to evaluate the relationship of these parameters with respect to molecular status.

RESULTS

IDH WT tumors were significantly associated with an infiltrative tumor border pattern on FLAIR, whereas the IDH MT tumors demonstrated either well-defined or ill-defined borders on FLAIR (p < 0.001). There was no significant difference between the two IDH MT groups with regards to FLAIR tumor border pattern; however, there was a trend towards restricted diffusion among tumors with 1p19q intact-ATRX loss (p =0.06). Contrast enhancement had no significant associations.

CONCLUSION

IDH WT lower grade gliomas are more likely to demonstrate an infiltrative pattern on FLAIR compared to IDH MT, corresponding to their worse prognosis. Among IDH MT tumors, there was a trend towards restricted diffusion in tumors with ATRX loss. Further forays into advanced imaging may be able to delineate these three molecular subgroups and provide essential prognostic information at initial MR diagnosis.

CLINICAL RELEVANCE/APPLICATION

With the molecular revolution in neuro-oncology underway, initial diagnostic imaging may be able to predict molecular subgroups of lower grade glioma thereby influencing upfront treatment decisions.

SSA17-09

Awards
Student Travel Stipend Award

Eun Kyoung Hong, MD, Seoul, Korea, Republic Of (Presenter) Nothing to Disclose
Seung Hong Choi, MD, PhD, Seoul, Korea, Republic Of (Abstract Co-Author) Nothing to Disclose
PURPOSE

The purpose of this study is to assess major genomic profiles of glioblastoma and correlate genetic information with radiologic features including volumetrics, normalized cerebral blood volume (nCBV) and normalized apparent diffusion coefficient (nADC), and progression free survival (PFS).

METHOD AND MATERIALS

We retrospectively enrolled total of 219 patients with histopathologically diagnosed with glioblastoma, who performed conventional brain MR images, DSC PWI and DWI before treatment. Major genetic information of the tumor (e.g. IDH mutation, 1p deletion, 19q deletion, EGFR amplification, PTEN loss, ATRX loss, and p53 mutation) was analyzed in all patients. Volume of tumor on FLAIR images and enhancing portion on contrast enhanced T1-weighted (CET1) image, ratio of the two volumes and volume of necrosis within tumors were measured. The nCBV and nADC histogram parameters were calculated based on both FLAIR image and CE-T1 images. Measured parameters and PFS in different genetic profiles were compared by using independent samples t test, Mann-Whitney test and ANOVA.

RESULTS

Of 190 patients with available IDH mutation information, IDH mutation was observed in 28 cases and was absent in 162 cases. IDH mutation positive group showed higher volume ratio between FLAIR and CET1 images (8.27 vs. 3.38, p=0.025), and lower mean nCBV (3.89 vs. 5.21, p=0.02) than IDH mutation negative group.  ATRX loss group revealed higher 5th percentile nADC value (1.10 vs. 1.02, p=0.048) than group without ATRX loss. In comparison between the three groups (IDH mutation positive, IDH mutation negative with and without ATRX loss), the 5th percentile nADC value demonstrated a significant difference (1.12 vs. 1.10 vs. 1.01, p=0.019). Additionally, volume ratio between FLAIR and CET1 images and 5th percentile nADC showed a positive correlation with PFS (p=0.0018, and <0.0001, respectively), which was independent of genetic markers.

CONCLUSION

We found that the major genetic markers of glioblastoma including IDH mutation and ATRX loss could be predicted by using imaging biomarkers. In addition, volumetics and nADC can be used for the prognosis prediction.

CLINICAL RELEVANCE/APPLICATION

We found a definite correlation between radiologic parameters, such as volumetrics, normalized CBV and ADC, and major genomic profiles and some radiologic parameters were feasible predictors of prognosis in glioblastoma patients.