Scientific Aspect of GB Imaging Whitney Pope, MD, PhD Director of Neuro Core, MedQIA Director of Brain Tumor Imaging, Assistant Professor of Neuroradiology, University of California, Los Angeles XL184-205 Investigator Meeting June 3rd, 2010

Outline 1,

Astrocytoma: grade IV - GBM T1 + contrast T2 At UCLA GBM: 2 year survival - 24% 5 year survival – 0% If grade III has imaging evidence of necrosis: same survival curve as GBM Low on T1 Bright on T2 Almost always enhances +/- Necrotic Brain slice Pathology

Appearance of GBM on MRI A, Axial post-contrast T1-weighted images of a patient with GBM: necrosis B, Axial T2-weighted images from the same patient. High T2W signal surrounds the tumor: vasogenic edema C, Axial post-contrast T1-weighted images of another GBM patient: necrosis D, Axial T2-weighted images of the same patient in (C): non-enhancing tumor (nCET) A B C D

Examples of “Typical” GBM Enhancing, Necrotic Tumor T1 Post Contrast T2

Example of Enhancing versus Non-Enhancing Tumor Vasogenic edema T1 Post Contrast T2 FLAIR Non-enhancing and enhancing tumor Non-enhancing tumor Enhancing tumor

FLAIR Can Improve Tumor Visualization insert slide to encourage dwi Insert next slide a dwi slide to encourage this is done FLAIR T2 Tumor

T2 versus FLAIR: More examples Tumor only Tumor and edema T2 FLAIR

Examples of non contrast enhancing tumor GBM Non-enhancing GBM showing biopsy site GBM Imaging Intraventricular GBM with enhancing and non-enhancing tumor

Role of Imaging in Clinical Trials Imaging plays an important role in all Clinical trials Phase I Primary goal tolerated dose Imaging exploratory Mechanism of action Pilot efficacy Phase III Primary goal overall survival Imaging used at clinical sites to manage patients Secondary endpoints

Challenges in GBM assessment Phase II goal therapeutic effect Imaging is a primary endpoint ( radiographic response + with clinical status) Substantial challenges in Radiologic evaluation of tumor size during clinical trials Critical Role of IRF to standardize Image acquisition Across sites Across time lines Assessment of tumor burden

Challenges in GBM assessment Challenges include: – Technical imaging considerations (Please add slides, move slide 39) – Selection of lesions (Please add slides) – Measurement approaches (Discuss confluence and splitting lesions, also measuring multi nodular lesions, measuring around surgical cavity…) – Response criteria – Interval between tumor measurements and response confirmation (delete this bullet) – Validity of imaging as a measure of efficacy (delete this bullet)

Technical Considerations Same imaging technique at every time point Measurements in the Axial plane Acquisition 3-mm, skip 0-mm T1-weighted C+ images improve resolution Increase acquisition time from 3 minutes to 5–6 minutes improved resolution is a great benefit Postcontrast axial Same Gd dose each time Standardized time interval post Gd (>5min < 10min) Move to follow slide 15.

Measurement Techniques Two major approaches for evaluation of contrast-enhancing tumor size: Diameter-based measurement on single-axial section containing largest diameter Computer-assisted volumetric analysis all sections containing tumor Slide 16 and 17:  Need to add diameter based measurement examples, goal here is for site measurements to hopefully be in line with IRF measurements (may want to take off volumetrics, not assessable at site level)

homogeneously enhancing >10 mm in diameter ideal for serial measurement by RECIST or 1D 0rMacdonald or 2D, and volumetric. predominantly necrotic ideal for volumetric measurement (because the enhancing and nonenhancing components can be segmented) too small in diameter (8 mm) for accurate serial measurement and should be followed as a nonmeasurable lesion

Macdonald Criteria (2D) To date, most trials for GBM use the WHO-based “Macdonald criteria” Measure maximal enhancing tumor diameter on single axial Post C T1 image, and largest perpendicular diameter same image Calculate product of the 2 diameters Sum Measurements from multiple lesions Macdonald DR Response criteria for phase II studies of supratentorial malignant glioma. J Clin Oncol. 1990;8: 1277-1280

Limitations of Macdonald Key Limitations Necrotic portions of lesions Assumption contrast enhanced tissue = tumor enhancement nonspecific reflects disrupted blood-brain barrier induced by of nontumoral processes inflammation, seizure activity, postsurgical changes, and radiation necrosis. enhancement influenced changes in corticosteroid dose and radiologic technique Changes in the enhancing area cannot be equated with changes in tumor size or tumor growth/activity

Clinical course of pseudoprogression in a 65-year-old patient with glioblastoma multiforme Brandes, A. A. et al. NEURO ONCOL 2008 10:361-367; doi:10.1215/15228517-2008-008 Copyright restrictions may apply.

Journal of Clinical Oncology, Vol 28, No 11 (April 10), 2010: pp Updated Response Assessment Criteria for High-Grade Gliomas: Response Assessment in Neuro-Oncology Working Group Patrick Y. Wen, David R. Macdonald, David A. Reardon, Timothy F. Cloughesy, A. Gregory Sorensen, Evanthia Galanis, John DeGroot, Wolfgang Wick, Mark R. Gilbert, Andrew B. Lassman, Christina Tsien, Tom Mikkelsen, Eric T. Wong, Marc C. Chamberlain, Roger Stupp, Kathleen R. Lamborn, Michael A. Vogelbaum, Martin J. van den Bent, Susan M. Chang

RANO Criteria: Same Limitations In RANO, increasing areas of FLAIR signal abnormality without corresponding changes in enhancement can be used to establish progressive disease. This is not optimal, however, as increasing FLAIR signal from radiation gliosis and other treatment effects are not necessarily associated with progression of non-enhancing tumor. Indeed the RANO group concluded that “although it would be preferable to have an objective measure of progressive non-enhancing recurrent disease similar to contrast-enhancing disease, the RANO working group felt that this was not possible at present given the limitations of current technology.” T1 Post Contrast T1 Post Contrast Slide 46, 47, 48:  Move to response criteria portion of presentation

Multicentric Lesions Approximately one third of malignant gliomas are multicentric at the time of diagnosis, and in half of these cases, there are discrete foci of enhancement. The approach in this situation is to measure and record each separately enhancing lesion that meets inclusion criteria and sum the measurements. Move to response criteria portion of presentation

Nonmeasurable Lesions Important in Clinical Trials Tumor progression may occur in these sites Nonmeasurable lesion includes: Foci of enhancement <8mm Region of T2-weighted hyperintensity surrounding enhancing tumor Discrete foci of non-enhancing T2-weighted hyperintensity (multicentric tumor) Hemorrhagic or predominantly cystic or necrotic lesions Leptomeningeal tumor Move to lesion selection portion of the presentation.

Example of Conversion from Enhancing to Non-enhancing Tumor Following Avastin Treatment 1st f/u: little enhancement 2nd f/u: remote non-enhancing disease T1 Post Contrast Baseline: avid enhancement T2

Example of Conversion from Enhancing to Non-enhancing Tumor Following Avastin Treatment T1 Post Contrast Response? 2nd f/u: remote non-enhancing disease Baseline: avid enhancement 1st f/u: little enhancement Progression? T2

Example of Faintly Enhancing Tumor (Post Treatment) FLAIR FLAIR T1 Post Contrast :  Can you add measurement caliper placements on these images (to be consistent with IRF measurements) Faintly enhancing tumor Faintly enhancing tumor

Example of Conversion from Enhancing to Non-enhancing Tumor Following Treatment T1 Post Contrast T1 Post Contrast T1 Post Contrast 1st Follow-up: only tiny nodular peripheral enhancement 2nd Follow-up: no enhancement. New distant disease (see next slide) Baseline: avid enhancement

Same Case: Development of distant non-enhancing tumor Same Case: Development of distant non-enhancing tumor. Example of conversion from local to diffuse disease following treatment. T2 T2 T2 2nd Follow-up: interval development of non- enhancing, ill-defined, tumor > 3cm from primary site, thus scored as diffuse progression 1st Follow-up: much less edema. Non-enhancing tumor remains. Baseline: enhancing tumor and vasogenic edema

Example of Multifocal Disease T1 Post Contrast T2 Greater than 3 cm from primary site, separated by normal brain Two sites of enhancing tumor

Same Patient: Conversion from Multifocal to Diffuse Disease Following Treatment T1 Post Contrast T2 Distant non-enhancing tumor with ill-defined margins, enhancement at primary site goes away. Non-enhancing tumor – no cortical ribbon Vasogenic Edema – cortical ribbon seen

Another Example of Multifocal Disease FLAIR T1 Post Contrast Multifocal, enhancing tumor

Local Recurrence Single focus of enhancing tumor

GBM – Distant Non-enhancing Tumor 5-12-05 Delete 12-5-05

Corticosteroid-induced Magnetic Resonance Imaging Changes How to add steroid on study to maintain image accuracy (ie, contrast intolerant subjects what steroid prescribed previous to next MRI) Post Steroids 90% patients had decreased enhancement or T2 signal 30% patients had 25%+ reduction in enhancing tumor 50% had 25%+ reduction in edema Maximal effect was achieved at 2 weeks Thus 2 weeks stable dose steroid recommended before baseline imaging. J Clin Oncol. 1994 Sep;12(9):1886-9.

Steroids: Example of Effect on Edema and Enhancement T1 Post Contrast Initial scan 2 weeks after steroid tx Edema – often only temporary reduction in edema

Computer-Aided Volumetric Methods Segmentation Algorithm generates border between the enhancing and non-enhancing regions on all adjacent axial sections Neuroradiologist Reviews contours Edits contour if needed Program Calculates enhancing volume, Non-enhancing volume (i.e., the centrally necrotic or cystic portion) Total or combined lesion volume in cubic Bi dimensional measurements

Move to response criteria portion of presentation RECIST (1D)3 Macdonald (2D)4 Volumetric Extrapolated from RECIST*, Volumetric Extrapolated from Macdonald*, CR Resolution of all enhancing tumor; confirm at 4 weeks PR 30% decrease in sum of maximal diameters; confirm at 4 weeks 50% decrease in product of 2 orthogonal diameters; confirm at 4 weeks 66% decrease in volume; confirm at 4 weeks 65% decrease in volume; confirm at 4 weeks SD All others PD|| 20% increase in sum of maximal diameters; confirm at 4 weeks 25% increase in product of orthogonal diameters; confirm at 4 weeks 73% increase in volume; confirm at 4 weeks 40% increase in volume; confirm at 4 weeks Comment Single longest diameter of the lesion or sum of longest diameters of multiple measurable lesions (see text) Product of orthogonal diameters on section with largest tumor area; sum of products if multiple measurable lesions Computer-assisted volumetrics using a perimeter methodology; sum of volumes if multiple measurable lesions Use of these values would be equally stringent for PR comparing RECIST and Macdonald criteria but would be more stringent for PD compared with RECIST but comparable with Macdonald criteria

Dynamic Contrast Enhanced (DCE)-MRI Physics Gadolinium causes a change in the longitudinal relaxivity (R1 = 1/T1) of surrounding water proportional to concentration Dynamic Contrast Enhanced MRI uses gadolinium-based contrast agents as a tracer for pharmacokinetic analysis by collecting dynamic T1-weighted images during a bolus Artery, Ca(t) Tissue EES Ce(t) Ve Plasma Cp(t) RBC Ktrans Two-Compartment Tofts Model (most common) 1-Hct kep Vein, Cv(t)

Dynamic Contrast Enhanced (DCE)-MRI Utility in Glioblastoma Ktrans ~ 0 in normal brain tissue because of the blood brain barrier (BBB) Ktrans  in Glioblastoma due to BBB compromise during creation of new (abnormal) blood vessels (angiogenesis)

Dynamic Contrast Enhanced (DCE)-MRI Utility in Glioblastoma Voxel-wise Ktrans Calculations = Permeability Maps

Dynamic Contrast Enhanced (DCE)-MRI Utility in Glioblastoma Voxel-wise Ktrans Calculations = Permeability Maps Biomarker for anti-angiogenic drugs targeting abnormal blood vessels Early Treatment Failure

Dynamic Contrast Enhanced (DCE)-MRI Utility in Glioblastoma Voxel-wise Ktrans Calculations = Permeability Maps Biomarker for anti-angiogenic drugs targeting abnormal blood vessels Widely accepted in clinical trials of other cancers O’Connor, Br J Cancer 2007

Dynamic Contrast Enhanced (DCE)-MRI Utility in Glioblastoma Voxel-wise Ktrans Calculations = Permeability Maps Biomarker for anti-angiogenic drugs targeting abnormal blood vessels Widely accepted in clinical trials of other cancers Voxel-wise changes in Ktrans shows spatially heterogeneous response Parametric Response Maps (PRM)

Dynamic Contrast Enhanced (DCE)-MRI Limitations/Challenges Measurement Error in pre-contrast T1 Crucial for accurate concentration estimation Need accurate flip angle measurements QC: T1 fit (R2 > 0.7, P < 0.05) T1 in Normal Tissues 1.5T: Breger, 1989; Steen, 1994; Whittall, 1997; Haacke, 1999 3.0T: Wansapura, 1999; Helms, 2008

Dynamic Contrast Enhanced (DCE)-MRI Limitations/Challenges Measurement Error in pre-contrast T1 Crucial for accurate concentration estimation Need accurate flip angle measurements Issues with Repeatability Random error, biological variation Confidence depends on Choice of model (2-compartment, 3-compartment, etc) ROI definition AIF determination

Dynamic Contrast Enhanced (DCE)-MRI Limitations/Challenges Measurement Error in pre-contrast T1 Crucial for accurate concentration estimation Need accurate flip angle measurements Issues with Repeatability Random error, biological variation Confidence depends on Choice of model (2-compartment, 3-compartment, etc) ROI definition AIF determination QC: Average of both ICAs

Dynamic Contrast Enhanced (DCE)-MRI Limitations/Challenges Measurement Error in pre-contrast T1 Crucial for accurate concentration estimation Need accurate flip angle measurements Issues with Repeatability Random error, biological variation Confidence depends on Choice of model (2-compartment, 3-compartment, etc) ROI definition AIF determination Median Change in Ktrans > 40% Reflects True Response

Dynamic Contrast Enhanced (DCE)-MRI Limitations/Challenges Measurement Error in pre-contrast T1 Crucial for accurate concentration estimation Need accurate flip angle measurements Issues with Repeatability Random error, biological variation Confidence depends on Choice of model (2-compartment, 3-compartment, etc) ROI definition AIF determination Median Change in Ktrans > 40% Reflects True Response Failure to Detect Response Sampling at wrong time point during treatment Averaging effects over ROI/VOI

Dynamic Contrast Enhanced (DCE)-MRI Limitations/Challenges Measurement Error in pre-contrast T1 Crucial for accurate concentration estimation Need accurate flip angle measurements Issues with Repeatability Random error, biological variation Confidence depends on Choice of model (2-compartment, 3-compartment, etc) ROI definition AIF determination Median Change in Ktrans > 40% Reflects True Response Failure to Detect Response Sampling at wrong time point during treatment Averaging effects over ROI/VOI Solution: Histogram and Voxel-wise analyses

Dynamic Susceptibility Contrast MRI Physics Gadolinium also has a transient effect on the magnetic susceptibility on blood and tissue water in high concentrations Magnetic Susceptibility

Dynamic Susceptibility Contrast (DSC) MRI Physics Gadolinium also has a transient effect on the magnetic susceptibility on blood and tissue water in high concentrations Gadolinium causes signal loss on T2*-weighted images Pre-Injection During Bolus Passage Dark = Vessels

Dynamic Susceptibility Contrast (DSC) MRI Physics Gadolinium also has a transient effect on the magnetic susceptibility on blood and tissue water in high concentrations Gadolinium causes signal loss on T2*-weighted images Area under relaxivity (R2* or R2) vs. time is proportional to blood volume (assuming no leakage) Dynamic T2* Image Acquisition Cerebral Blood Volume (CBV) Voxel Signal vs. Time R2* vs. Time

Dynamic Susceptibility Contrast (DSC) MRI Utility in Glioblastoma Glioblastoma has elevated CBV due to angiogenesis Post-Contrast T1-Weighted Image DSC-MRI Estimate of CBV Normal Vessels Abnormal Vascularity

Dynamic Susceptibility Contrast (DSC) MRI Utility in Glioblastoma Change in CBV is associated with successful treatment

Dynamic Susceptibility Contrast (DSC) MRI Utility in Glioblastoma Voxel-wise changes for spatially heterogeneous response Responder Non-Responder Pre-Tx Post-Tx Pre-Tx Post-Tx

Dynamic Susceptibility Contrast (DSC) MRI Limitations/Challenges CBV measurements are dependent on an intact BBB NOT the case in Glioblastoma Boxerman, 2006

Dynamic Susceptibility Contrast (DSC) MRI Limitations/Challenges CBV measurements are dependent on an intact BBB NOT the case in Glioblastoma Solution: Pre-load + Post-hoc Leakage Correction (not standard in most commercial software) Boxerman, 2006

Dynamic Susceptibility Contrast (DSC) MRI Limitations/Challenges CBV measurements are dependent on an intact BBB NOT the case in Glioblastoma Solution: Pre-load + Post-hoc Leakage Correction (not standard in most commercial software) CBV measurements are “relative” Typical solution is to “normalize” to contralateral tissue Our Solution: Image Intensity “Standardization” (piecewise histogram)

Dynamic Susceptibility Contrast (DSC) MRI Limitations/Challenges CBV measurements are dependent on an intact BBB NOT the case in Glioblastoma Solution: Pre-load + Post-hoc Leakage Correction (not standard in most commercial software) CBV measurements are “relative” Typical solution is to “normalize” to contralateral tissue Our Solution: Image Intensity “Standardization” (piecewise histogram) Prone to susceptibility artifacts if patient has surgical hardware Must have a good bolus (AIF) If tight bolus of contrast agent is not achievable, data will be poor

[18F]-fluorodeoxyglucose (FDG) PET Physics 18F-FDG is has uptake like glucose, but is trapped inside cells during metabolism PET data typically collected at a single (static) time point 60 min after injection (collected for 30 min) Miele, 2008

[18F]-fluorodeoxyglucose (FDG) PET Utility in Glioblastoma High energy demand =  FDG uptake Quantified using Standard Uptake Values (SUV)

[18F]-fluorodeoxyglucose (FDG) PET Utility in Glioblastoma Voxel-wise changes in PET show heterogeneous response [18F]-FDOPA shown below

[18F]-fluorodeoxyglucose (FDG) PET Limitations/Challenges High uptake in normal cortex

Multiparametric Imaging Combining voxel-wise changes in Perfusion & Diffusion 3 Mo. Post-Tx 6 Mo. Post-Tx 12 Mo. Post-Tx

Multiparametric Imaging Combining Perfusion, Diffusion, & PET