2HG concentrations were measured from tissue specimens of 47 gliomas with mIDH by 2D Correlation Spectroscopy using solid state NMR. All tumors were categorized into three WHO grades. 1p/19q co-deletion was determined using fluorescence in situ hybridization. Progression free survival (PFS) of patients with mIDH and wild type (WT) IDH was determined. 2HG levels were compared with WHO grades, PFS, 1p19q status and selected metabolites, such as lactate (Lac), glutamate (Glu) and glutamine (Gln).Nonparametric Mann–Whitney test for comparison/Spearman for correlation were used for statistical analysis.

Sixteen patients with rHGGs underwent 3D FSE pcASL imaging 1-2 days before (baseline) and within one month after BEV initiation (post-BEV) were included in this study. Average (aCBF) and maximum (mCBF) cerebral blood flow of the enhanced tumor, their normalized values to contralateral normal appearing white matter (rCBF_wm and mCBF_wm) and cerebellum (rCBF_cb and mCBF_cb), and their changes between baseline and post-BEV were evaluated. ROC curve analysis was utilized to define the optimal cutoff perfusion values for PFS and OS prediction. Kaplan-Meier analysis with log-rank test was applied to assess and compare progression-free survival (PFS) and overall survival (OS) rates.

Fifty-three cases with pathologically confirmed low-grade glioma (WHO gradeⅡ) were enrolled in this prospective study, who performed MR scans with the imaging protocol of T1WI, T2WI, T2-Flair , DWI (0,1000 s/mm2), eDWI (22 b-value DWI of 0-5000 s/mm2) and enhanced-T1WI. In tumor regions of T2-Flair hypertense, excluding remarkable cystic parts, four diffusion parameters of monoexponential -derived apparent diffusion parameter (ADC), biexponential-derived slow diffusion coefficient (Dslow) , stretched exponential-derived distributed diffusion coefficient (DDC) and α were measured to correlate with the status of IDH1 mutation, which was determined by immunohistologic chemistry staining and molecular sequencing. Based on the status of IDH1 mutation, 53 gliomas were divided into two groups of IDH1mut (n=37) and IDH1wild (n=16). Four parameters of ADC, Dslow, DDC and α were employed to make comparisons between two groups of IDH1 mut and IDH1 wild. And ROC curve analyses were used to compare the capability of differentiating gliomas of IDH1 mut from IDH1 wild.

This retrospective study included thirty four histopathologically proven glioblastoma patients who had developed any new enhancing lesion after CCRT, which was defined to be either true (n=14) or pseudoprogression (n=20) following RANO criteria. DCE pharmacokinetic parameters, including the volume transfer constant (Ktrans), the rate transfer constant (Kep), the extravascular extracellular space per unit volume of tissue(Ve), and the blood plasma volume per unit volume of tissue(Vp), for each lesion were calculated twice, using both a fixed T1 of 1000ms and tissue T1 measurement using variable flip angle method. Intraobserver reproducibility was assessed using intraclass correlation coefficient (ICC). The mean of each parameter was compared between two groups using unpaired t test and multivariate analysis. The diagnostic performance was evaluated by receiver operating characteristic (ROC) analysis with Bonferroni correction.

Recent advances in glioma genomics have significantly changed our understanding of tumor biology and hence, affected how these patients are treated. Similarly, integrating imaging data with genomic markers has also helped create better prognostic and predictive biomarkers which offer promising future for personalized medicine. This session will highlight a multi-disciplinary approach with the focus on advanced imaging and genomics markers before and after therapy in gliomas.     

Diffusion and perfusion imaging is commonly used for brain tumor patients. These methods provide unique insights into what is happening in tumors, particularly during and after therapeutic intervention. Understanding how these methods work can substatially impact interpretation of MRI examiantions in these patients. Hwoever, the processing of diffusion and perfusion images is complex, and can often lead to incorrect conclusions if artifacts and other differences are not understood.While the computation of ADC images from DWI acquistions is straightforward, one must be careful when making conclusions based on minimum values seen in an ROI due to filtering that may or may not be applied to the images. Registration and subtraction can also be useful, but can also have errors in them.Perfusion images, particuarly DSC images, involve very complex processing based on assumptions about what is happening in tissues. Those assuptions substantially impact calculations that are used to produce rCBV and other types of perfusion images. Even seemingly simple things like selection of the normal-appearing white matter can substantially change measured values. Understanding basic principles used in processing can help to avoid or minimize errors in interpretation.

284 patients (171 men; ages 19-79 years, median 57; 159 on BEV arm) had post-op (baseline) and 22-week (pre-cycle 4) MRI. Central readers measured unidimensional (1D), bidimensional (2D) and volumetric (3D) lesion post-Gd T1 and FLAIR hyperintensity at both time points, and determined change from baseline at week 22 for all six imaging markers. Hazard ratio (HR) from Cox proportional hazards model either with or without treatment and marker interaction term and Kaplan-Meier survival estimates with log-rank test were used for inference. 

21 patients with suspected malignant glioma recurrence underwent a volumetric APTw imaging sequence at 3T. 64 APTw imaging-guided biopsy specimens were obtained and analyzed for tumor cellularity, necrosis, and proliferation. Specimens were classified as active, quiescent, or mixed by neuropathology. Active and mixed were grouped as recurrence, while quiescent and no tumor were grouped as treatment effect.

We enrolled 220 patients who underwent presurgical fMRI in this IRB approved study. Our inclusion criteria were pathologically proven diagnosis of GBM, full pre and post‐operative neurological examination, pre‐operative task based motor fMRI as well as diffusion tensor imaging (DTI).We obtained multiple fMRI and DTI parameters including Lesion to activation distance (LAD) in T1+C (T1LAD) and T2 FLAIR (FLAIR‐LAD). both from center of activity to center of lesion and from margin of activity to margin of lesion. DTI was done to measure distance between neoplasms margin, edema margin and corticospinal tract (CST) in T1+C and T2 FLAIR respectively. Kruskal‐Wallis test was used to compare distance between patient groups, all tests were 2 sided and p‐values of 0.05 or less we reconsidered statistically significant.

Our Institutional review board has approved this HIPAA-compliant retrospective study. We identified a total
of 69 patients (45 Males: 24 females) (average age =50, median age = 51) with pathologically proven GBM.
All patients underwent advanced MRI studies (DSC, DCE, ASL and MRS). All patients had pathological
proof of either PD or PsP after the advanced MRI studies. For each patient, three board-certified
neuroradiologists, blinded to the pathology report, evaluated all the advanced imaging features using a
designed qualitative questionnaire to determine PD and PsP. The questionnaire included the following
parameters: DCE (PEI 3, 60, Curves), DSC (rCBV, NEI, Curves), ASL (CBF), MRS (Nacetylaspartate,
choline/creatine, lipid). Statistical analysis was performed to evaluate the ability of each imaging feature in
discriminating PD from PsD.