Genomic Imaging Gliomas and Perfusion Imaging A TCGA

Genomic Imaging: Gliomas and Perfusion Imaging A TCGA Glioma Phenotype Research Group Project Rajan Jain, MD Division of Neuroradiology, Department of Radiology Department of Neurosurgery Henry Ford Health System Assistant Professor, WSU School of Medicine Detroit, MI, US ASNR 50 th Annual Meeting e. Ed. E-17

Disclosures § Radio-genomics work funded by NCI Contract No. HHSN 261200800001 E. § Research support: Lantheus Medical Imaging.

TCGA Glioma Phenotype Research Group Project Rajan Jain, 1, 2 Laila Poisson, 3 Jayant Narang, 1 David Gutman, 4 Adam Flanders, 5 Lisa Scarpace, 2 Scott N Hwang, 4 Chad Holder, 4 Max Wintermark, 6 Rivka R Colen, 7 Justin Kirby, 8 John Freymann, 8 Brat Daniel, 4 Carl Jaffe, 9 Tom Mikkelsen 2 1 Division of Neuroradiology, Department of Radiology, 2 Department of Neurosurgery and 3 Department of Public Health Sciences, Henry Ford Health System, Detroit, MI 4 Emory University, Atlanta, GA; 5 Thomas Jefferson University Hospital, Philadelphia, PA; 6 University of Virginia, Charlottesville, VA; 7 Brigham & Womens Hospital, Boston, MA; 8 SAIC-Frederick, Inc. ; 9 Boston University, Boston, MA https: //wiki. cancerimagingarchive. net/display/Public/TCGA+Glio ma+Phenotype+Research+Group

Learning Objectives § Genomic mapping of High-grade Gliomas: Sub-classifications based on molecular markers – TCGA (The Cancer Genome Atlas) – TCIA (The Cancer Imaging Archive) § Radio-genomics: Correlation of imaging markers with gene expression pathways § Brain Tumor Perfusion Imaging: Vascular Parameters § Tumor Blood Volume § Tumor Permeability § Correlation of Perfusion Parameters with Genomic Markers of Angiogenesis § Correlation of Perfusion Parameters with Genomic Mapping and Patient Survival in GBM

Genomic Mapping of Gliomas

High-Grade Gliomas: Genomic Mapping § Recently, there has been progress in understanding the molecular basis of the tumor aggressiveness and heterogeneity. § Various molecular sub-classifications have been proposed based on the genetic makeup of high-grade gliomas with the hope that a better understanding of origin of tumor cells and molecular pathogenesis may predict response to targeted therapies. § Phillips HS, et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell 2006; 9(3): 157 -173. § Verhaak RG, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH 1, VEGFR, and NF 1. Cancer Cell 2010; 17(1): 98 -110. §.

High-Grade Gliomas: Molecular Sub-classes Phillips HS, et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell 2006; 9(3): 157 -173.

GBM: Molecular Sub-classes § Verhaak RG, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH 1, VEGFR, and NF 1. Cancer Cell 2010; 17(1): 98 -110.

TCGA and TCIA § The Cancer Genome Atlas (TCGA) researchers have recently cataloged recurrent genomic abnormalities in GBM providing a platform for better understanding of the molecular basis of these aggressive but heterogeneous tumors. § In parallel the Cancer Imaging Program is retrospectively obtaining radiological imaging data for TCGA patients and making it available via The Cancer Imaging Archive (TCIA). § Integration of this vast genomic information with imaging data (radio-genomics) may not only strengthen this understanding but also provide an opportunity to use some of the non-invasive imaging features or parameters as biomarkers. http: //cancerimagingarchive. net

§ TCIA is a large and growing archive service providing DICOM images for use in research § TCGA-GBM collection currently includes over 170 deidentified glioblastoma subjects with accrual still ongoing § TCGA-LGG collection to be formed for collection of lower grade gliomas in coming months § Additional brain related image sets include REMBRANDT (130 subjects) and RIDER Neuro MRI (19 subjects) § Registration is free and new data is being added regularly. Sign up at: http: //cancerimagingarchive. net

Clinical Features Radio-genomics Imaging Features Genomic Features Pathologic Features

Radio-genomics: Morphologic Features § Integration of this vast genomic information with imaging data (radio-genomics) may not only strengthen this understanding but also provide an opportunity to use some of the non-invasive imaging features or parameters as biomarkers. § A limited number of publications on this topic have correlated morphologic imaging features (presence or absence of contrast enhancement) with various gene expression pathways affecting tumor cell mitosis, migration, angiogenesis, hypoxia, edema and apoptosis. Proc Natl Acad Sci U S A 2008; 105(13): 5213 -5218. Diagn Mol Pathol 2006; 15(4): 195 -205. Radiology 2008; 249(1): 268 -277.

Radio-genomics: Functional Imaging Markers § Barajas et al correlated histologic features with ADC and r. CBV estimates, but the authors did not directly correlate physiologic measures with gene expression. Radiology 2010; 254(2): 564 -576. § Pope et al correlated ADC histogram analysis with differential gene expression. AJNR published online § Jain et al correlated perfusion parameters (tumor blood volume and permeability) with angiogenesis related gene expression in GBM. AJNR published online

Perfusion Surrogate Markers of Angiogenesis § Tumor Blood Volume § Tumor Vascular Leakiness or Permeability

Perfusion Surrogate Markers of Angiogenesis: Tumor Blood Volume § CBV is the total volume of blood traversing a given region of brain, ml/100 g – CBV estimated from integral of signal intensity versus time curve § Deconvolution, allows calculation of MTT § Central volume principle CBF = CBV/MTT

Measuring Tumor Blood Volume: Why is it important? § It represents total tumor vascularity. – Thus, in vivo measurement of tumor blood volume could be important for several reasons: § Grading of tumors § Assessing response of tumors to various therapies § Treatment effects – radiation injury or necrosis § Prognosis

Perfusion Surrogate Markers of Angiogenesis: Tumor Vascular Leakiness/Permeability § Physiological terms – leakage of contrast agent from intravascular to extravascular compartment through a deficient BBB § Various imaging techniques and various parameters § Permeability surface area-product (PS), ml/100 g/min § Forward Transfer Constant (Ktrans), min-1 § No standardization of technical parameters of acquisition and post-processing

Measuring Tumor Vascular Permeability: Why is it important? § In vivo measurement of vessel permeability could be important for several reasons: § Grading of tumors - Increased permeability is associated with immature blood vessels, which is seen with angiogenesis and higher grade tumors. § Studying response of tumors to various therapies, especially anti-angiogenic therapy. § Understanding the mechanism of entry of therapeutic agents into the central nervous system. § Development of methods to selectively alter the blood-brainbarrier to enhance drug delivery.

Understanding Angiogenesis Hypoxia leads to expression of hypoxiainducible factor (HIF-1α) and formation of Tumor cell growth outstrips its blood supply leading to hypoxia angiogenic mediators such as VEGF and stromal derived factor-1 (SDF-1) Tumor angiogenesis involves a multitude of controlled signaling cascades and structural changes. The effects of VEGF and SDF-1 lead to formation of immature and leaky blood vessels Leaky blood vessels allow extravasation of plasma, plasma proteins, and deposition of proangiogenic matrix proteins and microvascular cellular proliferation (MVCP) Griffith, B. Jain, R. , et al. Blood-Brain-Barrier Imaging in Brain Tumors: Concepts and Methods. Presented at ASNR 2011, Neurographics (in press)

Understanding Angiogenesis ↑ MVD ↑ TVA ↑ Permeability As pericyte-poor new blood vessels (mother vessels) enlarge and give rise to daughter vessels through a complex series of endothelial rearrangements --- microvascular density (MVD) and total vascular area (TVA) increase, which in turn leads to increased microvascular cellular proliferation (MVCP) and increase in permeability Griffith, B. Jain, R. , et al. Blood-Brain-Barrier Imaging in Brain Tumors: Concepts and Methods. Presented at ASNR 2011, Neurographics (in press)

Understanding Angiogenesis: Correlation with Perfusion Parameters § In the initial phase, vessel leakiness (measured by permeability PS or Ktrans) increases more than the total number and area of blood vessels (measured by blood volume). § Finally, as vessels mature, the total number and area of blood vessels increases more than vessel leakiness, evolving into a very heterogeneous tumor with regions showing different mixtures of vessel characteristics and angiogenesis.

Immuno-histochemical Basis for Tumor Vascular Surrogate Markers

Jain, R. et al. Am J Neuroradiology 2011; 32: 388 - 394.

CBV 2. 41 PS 3. 43 Jain, R. et al. Am J Neuroradiology 2011; 32: 388 - 394.

Multiple Regression Analysis PCT Parameter CBV PS p-value* 0. 777 0. 001 MVD CBV PS <0. 001 0. 364 Tumor Cellularity CBV PS 0. 104 0. 115 WHO grade CBV PS 0. 098 0. 060 Histopathology MVCP Jain, R. et al. Am J Neuroradiology 2011; 32: 388 - 394.

Grade IV Glioblastoma Cellularity=3, MVCP score=2, MVD score=3 (263 vessels/20 x), VEGFR-2+ CBV 6. 98 CBV 1. 22 PS 10. 04 PS 0. 50 Cellularity=2, MVCP score=0, MVD score=1 (55 vessels/20 x), VEGFR-2 - Grade II Astrocytoma

Grade IV Glioblastoma Cellularity=3, MVCP score=2, MVD score=3 (263 vessels/20 x), VEGFR-2+ H & E 20 x CD 34 20 x VEGFR-2 20 x Cellularity=2, MVCP score=0, MVD score=1 (55 vessels/20 x), VEGFR-2 - Grade II Astrocytoma

Radio-genomics: Perfusion Surrogate Markers

Jain, R. et al. Am J Neuroradiology Published March 15, 2012 as 10. 3174/ajnr. A 2956.

Correlation of Perfusion Parameters with Genes Related to Angiogenesis Regulation in GBM § Given our interest in CBV and PS estimates we considered the Gene Ontology biological process pathways for angiogenesis regulation only. § We observed expression levels for 92 genes (332 probes) in the regulation of angiogenesis (pathways GO: 0045766, GO: 0005923) Jain, R. et al. Am J Neuroradiology Published March 15, 2012 as 10. 3174/ajnr. A 2956.

Correlation of Perfusion Parameters with Genes Related to Angiogenesis Regulation in GBM § 19 genes had significant correlation with PS (p<0. 05) § 9 genes had significant correlation with CBV (p<0. 05) § 5 genes showed significant correlation with both CBV and PS. Jain, R. et al. Am J Neuroradiology Published March 15, 2012 as 10. 3174/ajnr. A 2956.

Pro-angiogenic Genes • KDR VEGFR-2 (CBV correlation co-efficient 0. 60, p=0. 0084; PS 0. 59, p=0. 0097) • HIF 1 a (Hypoxia inducible factor 1 alpha) (CBV 0. 29, p=0. 29; PS 0. 66, p=0. 008) • TNFRSF-1 A (Tumor necrosis factor receptor superfamily, member 1 A) (CBV -0. 23, p=0. 3673; PS 0. 53, p=0. 0239) • TIE 1 (CBV 0. 54, P = 0. 0217; PS 0. 49, P = 0. 0389) • TIE 2/TEK (CBV 0. 58, P = 0. 0119; PS 0. 46, P = 0. 0550 ) Significant Correlation with CBV Significant Correlation with PS Significant correlation with both CBV and PS Jain, R. et al. Am J Neuroradiology Published March 15, 2012 as 10. 3174/ajnr. A 2956.

Anti-angiogenic Genes § VASH 2 Vasohibin 2 (CBV correlation co-efficient -0. 35, P = 0. 1568, PS -0. 71, P = 0. 0011) § CX 3 CR 1 (CBV -0. 66, P = 0. 0028; PS -0. 49, P = 0. 0375) • WNT 5 A (CBV -0. 10, P = 0. 6833; PS -0. 52, P = 0. 0284) • C 3 (CBV -0. 63, P = 0. 0051; PS -0. 41, P = 0. 0953) Jain, R. et al. Am J Neuroradiology Published March 15, 2012 as 10. 3174/ajnr. A 2956.

Correlation of Perfusion Parameters with Genes Related to Angiogenesis Regulation in GBM § CBV and PS estimates in GBMs can correlate positively with pro-angiogenic genes § and inversely with anti-angiogenic genes. § The results of this preliminary analysis can help establish a genomic/molecular basis for these commonly used imaging biomarkers and potentially add to our knowledge of their immuno-histological bases. Jain, R. et al. Am J Neuroradiology Published March 15, 2012 as 10. 3174/ajnr. A 2956.

Genomic mapping and survival prediction in GBM: Molecular sub-classification as an adjunct to hemodynamic imaging markers - A TCGA Glioma Phenotype Research Group Project Purpose The purpose of this study was to correlate tumor blood volume, measured using DSC T 2* magnetic resonance (MR) perfusion with patient survival and also correlate it with molecular sub-classes of GBM. Session: 34 a - Adult Brain Neoplasms III , abstract 876 Presentation #: Paper O-285, Date/Start Time: 4/24/2012 3: 00 PM

Material and Methods: Patient Population • 57 patients with treatment naïve GBM underwent DSC T 2* MR perfusion studies at 2 different institutions and were selected from TCIA’s TCGA-GBM collection. • 50 patients had gene expression data available from TCGA. • 35 patients at Institute 1 HFH • 3. 0 T scanner, n=14 • 1. 5 T scanner, n=21 • 15 patients at Institute 2 UCSF • 1. 5 T scanner, n=15 • According to those TCGA requirements, the pathology was confirmed as GBM using adequate frozen tissue ≥ 0. 5 g consisting of ≥ 70% tumor nuclei and < 50% necrosis. https: //wiki. cancerimagingarchive. net/display/Public/DSC+T 2*+M R+Perfusion+Analysis

Material and Methods: MRI Image acquisition Institute 1 Institute 2 0. 1 mmol/kg Gd-DTPA, 5 ml/sec 95 phases of GE T 2* 60 phases of GE T 2* TR = 1900 msec, TE =40 msec, and flip angle =90° TR = 2000 msec, TE =54 msec, and flip angle =30° Temporal resolution 2. 0 sec Matrix size 128 x 128, 26 -cm FOV Slice thickness 5 mm Slice thickness 3 -6 mm

Material and Methods: MRP Post-processing • Studies from both institutions were processed using Nordic. ICE software (Nordic. Imaging. Lab AS) using the FDA approved DSCT 2* perfusion module. • The module corrects for contrast agent leakage from the intravascular to extracellular space using the method published by Boxerman et al AJNR Am J Neuroradiol 2006; 27(4): 859 -867. • Normalized relative cerebral blood volume (r. CBV) maps with leakage correction were produced by the software, which normalizes the CBV relative to a globally determined mean value.

Material and Methods: Image Analysis • ROIs were drawn on the r. CBV maps fused with post-contrast T 1 W images and FLAIR images. • r. CBV mean the contrast enhancing portion of the tumor (excluding any areas of necrosis and vessels) • r. CBV max 10 voxel ROI was placed on the hottest appearing part of the tumor • r. CBVNEL 3, 10 x 10 voxel ROIs on non-enhancing FLAIR abnormality within 1 cm of the edge of the enhancing lesion

Material and Methods: Statistical Analysis § Comparison of average r. CBV measures between groups was done using two-sample t-tests or one-way ANOVA. § Kaplan-Meier estimation was used to calculate median survival and for some univariate testing. § For the Kaplan-Meier curves, the log-rank statistic assesses group differences equally across the full observation time whereas the Wilcoxon statistic weights the early events more heavily thus identifying early separation in the curves. § Survival analysis with Cox proportional hazards models was employed primarily to estimate hazard ratios and for testing multivariable models.

Material and Methods: Statistical Analysis § Potential covariates in the multivariable models – Patient age at diagnosis (years, continuous), – MR scanner type (1. 5 T, 3 T) – molecular classification (Verhaak or Phillips) Huse J et al. Glia 2011; 59(8): 1190 -1199. – Karnofsky performance score (KPS, continuous) – level of resection (gross-total, sub-total) § Age and scanner were not significant predictors and did not enhance the models so they were excluded from the presented models for the sake of parsimony given the sample size. § KPS and resection data were only available for samples from institution 1.

Results: r. CBV analysis using molecular sub-classification r. CBVmean r. CBVmax r. CBVNEL p=0. 66 p=0. 95 p=0. 43 2. 66 (0. 78) 4. 55 (0. 76) 0. 66 (0. 24) 2. 61 (1. 26) 4. 80 (1. 49) 0. 88 (0. 45) Neural (n=11) 2. 30 (0. 84) 4. 68 (0. 95) 0. 81 (0. 27) Proneural (n=12) 2. 27 (0. 68) 5. 06 (3. 61) 0. 84 (0. 26) p=0. 32 p=0. 57 p=0. 70 Mesenchymal (n=24) 2. 68 (1. 16) 4. 76 (1. 33) 0. 83 (0. 40) Proneural (n=20) 2. 32 (0. 72) 5. 03 (2. 79) 0. 83 (0. 25) Proliferative (n=6) 2. 15 (0. 59) 4. 04 (0. 65) 0. 70 (0. 30) Verhaak Classical (n=10) Mesenchymal (n=17) Phillips

Results: Survival analysis using Verhaak sub-classification Present study • • Verhaak et al Cancer Cell 2010; 17(1): 98 -110 Median overall survival (OS) was 1. 14 years (IQR: 0. 49, 2. 11). When the Verhaak classification scheme was applied to these samples, the classical sub -class had the best survival, with median of 2. 13 years (IQR: 1. 53, 2. 59) and the proneural sub-class had the worst survival with median 0. 41 years (IQR: 0. 65, 1. 19).

Results: Survival analysis using Verhaak sub-classification Present study § § Verhaak et al Cancer Cell 2010; 17(1): 98 -110 The difference in survival by Verhaak sub-classification was significant between groups with the difference being more prominent earlier during follow-up (Wilcoxon P=0. 0445, log-rank P=0. 0696). However, the proneural subclass also had the worst median survival (0. 94 years, IQR: 0. 78, 1. 23) in the publication by Verhaak et al.

Results: Survival analysis using Phillips sub-classification Present study • • Phillips et al Cancer Cell 2006; 9(3): 157 -173 There was no evidence that the Phillips classification was associated with survival in our sample (log-rank P=0. 6432, Wilcoxon P=0. 4548). The best median survival is attributed to the mesenchymal sub-class with 1. 28 years (IQR: 0. 61, 2. 22), followed closely by the proneural subclass with 1. 12 years (IQR: 0. 33, 1. 86).

Results: Survival analysis using Phillips sub-classification Present study • Phillips et al Cancer Cell 2006; 9(3): 157 -173 The proliferative sub-class had the worst median survival at only 0. 54 years (IQR: 0. 34, 3. 96) but this class was only represented by six patients (five deaths), one of whom was still surviving at 3. 96 years

Results: Survival analysis using only r. CBV measures Parameters Mean Max NEL 1. 25 (0. 1918) 1. 24 (0. 0131) 2. 45 (0. 0555) 1. 46 (0. 0212) 1. 24 (0. 0062) 2. 56 (0. 0704) (0. 0250) (0. 0476) (0. 0917) Classical 0. 21 0. 26 0. 30 Mesenchymal 0. 43 0. 48 Neural 0. 44 0. 41 0. 55 Proneural 1. 00 1. 27 (0. 1670) 1. 24 (0. 0152) 2. 51 (0. 0566) (0. 5892) (0. 6888) (0. 6533) Mesenchymal 0. 72 0. 79 0. 74 Proliferative 0. 98 1. 02 0. 87 Proneural 1. 00 Model 1: r. CBV Model 2: r. CBV + Verhaak Model 3: r. CBV + Phillips

Results: Survival analysis using r. CBV and molecular sub-classification Parameters Mean Max NEL 1. 25 (0. 1918) 1. 24 (0. 0131) 2. 45 (0. 0555) 1. 46 (0. 0212) 1. 24 (0. 0062) 2. 56 (0. 0704) (0. 0250) (0. 0476) (0. 0917) Classical 0. 21 0. 26 0. 30 Mesenchymal 0. 43 0. 48 Neural 0. 44 0. 41 0. 55 Proneural 1. 00 1. 27 (0. 1670) 1. 24 (0. 0152) 2. 51 (0. 0566) (0. 5892) (0. 6888) (0. 6533) Mesenchymal 0. 72 0. 79 0. 74 Proliferative 0. 98 1. 02 0. 87 Proneural 1. 00 Model 1: r. CBV Model 2: r. CBV + Verhaak Model 3: r. CBV + Phillips

Results: Survival analysis using r. CBV and molecular sub-classification Parameters Mean Max NEL Model 1: r. CBV 1. 25 (0. 1918) 1. 24 (0. 0131) 2. 45 (0. 0555) Model 2: r. CBV + 1. 46 (0. 0212) 1. 24 (0. 0062) 2. 56 (0. 0704) (0. 0250) (0. 0476) (0. 0917) Classical 0. 21 0. 26 0. 30 Mesenchymal 0. 43 0. 48 Neural 0. 44 0. 41 0. 55 Proneural 1. 00 1. 27 (0. 1670) 1. 24 (0. 0152) 2. 51 (0. 0566) (0. 5892) (0. 6888) (0. 6533) Mesenchymal 0. 72 0. 79 0. 74 Proliferative 0. 98 1. 02 0. 87 Proneural 1. 00 Verhaak Model 3: r. CBV + Phillips

Genomic mapping and survival prediction in GBM: Molecular sub-classification as an adjunct to hemodynamic imaging markers Ø Hemodynamic tumor measures (r. CBV) using MR perfusion did not have any significant correlation with the various subclasses using the two most commonly accepted molecular sub-classification systems of GBM. Ø But r. CBV measures did provide important prognostic information, as patients in whom the tumor had higher r. CBV showed worse prognosis and poor survival.

Genomic mapping and survival prediction in GBM: Molecular sub-classification as an adjunct to hemodynamic imaging markers Ø Verhaak sub-classification schema remained significant in the survival models providing additional survival information about r. CBVmean measurements. Ø Even though this needs to be confirmed with a larger sample size, it highlights the importance of non-invasive in vivo imaging biomarkers for patient prognosis and survival and their future role as an adjunct with molecular markers.

Conclusions Ø Molecular mapping of GBM can provide important therapy targets by providing insight into the molecular basis for tumor cell origin. Ø Correlation of imaging (morphologic or functional) markers with gene expression (Radio-genomics) will help in better understanding, standardization and improved use of the Imaging Biomarkers. Ø Integration of imaging markers with molecular or genomic data (Radio -genomics) can provide important prognostic information that may be used as an adjunct to genomic markers in future.

rajanj@rad. hfh. edu