NeuroGenomics

Here’s a list of TCGA abstracts me and other colleagues have for this year’s ASNR meeting in San Diego 2013

My presentation (aka 6 minutes of glory)

Control #: 1471
Title: Imaging Genomics: Correlation of Invasive Genomic Composition and Patient Survival Using Qualitative and Quantitative MR Imaging Parameters: A TCGA Glioma Phenotype Research Group Project
Session: Adult Brain: Tumors IV: Glioblastoma
Location: Ballroom 6CF
Presentation #: Paper O-539
Presentation Date/Start Time: 5/23/2013 1:15:00 PM
Presentation Length: Presentation – 6 minutes; Discussion – 2 minutes

 

 

 

Control/Tracking Number: 13-O-2079-ASNR
Activity: Scientific Paper (O)
Current Date/Time: 2/14/2013 1:39:52 PM

Association between Volumetric Imaging Features, Genomic Profile, and Patient Outcome in Glioblastoma: A TCGA Glioma Phenotype Research Group Project

Author Block Gutman, D. A.1·Holder, C.1·Cooper, L.1·Dunn, W.1·Mikkelsen, T.2·Jain, R.2·Colen, R.3·Jaffe, C.4·Saltz, J. H.1·Flanders, A.5·Brat, D. J.1·TCGA Glioma Phenotype Research Group
1Emory University, Atlanta, GA, 2Henry Ford, Detroit, MI, 3M. D. Anderson, Houston, TX, 4Boston University, Boston, MA, 5Thomas Jefferson University, Philadelphia, PA.

Abstract:
Purpose
There is growing interest on the integration of gene expression profiles from tumor biopsies to further classify tumors, although gene expression is influenced not only by the underlying molecular abnormalities driving oncogenesis but also micro- and macro-scale properties. For glioblastoma in particular, neuroimaging provides a unique opportunity to characterize macroscale properties which can affect patient survival as well as genomics. We have previously utilized 2-D semiquantitative imaging features extracted from GBM images to explore these associations. We are now integrating 3D volumetric assessments which we hypothesize will provide a more sensitive measure to evaluate radiogenomic-phenotypic associations.
Materials & Methods
Eighty-nine pre-operative MRI image series were obtained from the Cancer Imaging Archive (TCIA). The postgadolinium T1 image and T2 Flair image were annotated using Velocity AI, a radiotherapy treatment planning system. A contrast-enhancing region (any embedded necrosis) was annotated on the T1 series, and a flair “envelope” consisting of adjacent abnormal signal on the T2 flair were marked up. Annotations subsequently were exported and analyzed using a workflow system developed using niPype to perform volume registration, as well as to calculate the volume of necrosis, contrast enhancement, and edema from the image stack. This framework supports extraction of additional 2D and 3D parameters from the imaging volumes. Genetics data was obtained from the MSKCC cBioPortal, and additional clinical data was obtained from the TCGA archives. Correlations were computed using SAM for genetic correlations, or SPSS. Pathway analysis was performed using Ingenuity Pathway Analysis.
Results
We have previously shown high concordance between measurements obtained via velocity (n = 79, p<0.0001) with independent measures using 3D Slicer in a sample of this data set. A number of volumetric features including overall percentage of necrosis (p<0.001), % of edema (p<0.001) as well as the ratio of contrast enhancement to edema (p<0.003) were associated with poor overall survival. Initial genomics analysis indicated a strong gene set enrichment in the hypoxia-inducible factor signaling (HIF, p<0.0001) and anti-apoptosis pathway (p<0.0159) as a function of the percentage of overall necrosis.
Conclusion
This work further establishes a strong association between macroscopic imaging properties and gene expression. Furthermore, the python-based framework supports the integration of additional imaging features for association studies greatly facilitating our radiogenomic work. We currently are exploring gene signatures associated with both contrast enhancement and edema volumes, as well as characterizing whether relative or absolute measures of imaging properties provide greater sensitivity in the identification of gene/imaging features.


 

 

 

ASNR_ColenZinn

Control #: 1471
Title: Imaging Genomics: Correlation of Invasive Genomic Composition and Patient Survival Using Qualitative and Quantitative MR Imaging Parameters: A TCGA Glioma Phenotype Research Group Project
Session: Adult Brain: Tumors IV: Glioblastoma
Location: Ballroom 6CF
Presentation #: Paper O-539
Presentation Date/Start Time: 5/23/2013 1:15:00 PM
Presentation Length: Presentation – 6 minutes; Discussion – 2 minutes