+ All Categories
Home > Documents > Radiogenomics in glioblastoma multiforme Olivier Gevaert, PhD Stanford University Cancer Center for...

Radiogenomics in glioblastoma multiforme Olivier Gevaert, PhD Stanford University Cancer Center for...

Date post: 28-Dec-2015
Category:
Upload: chloe-peters
View: 214 times
Download: 0 times
Share this document with a friend
Popular Tags:
38
Radiogenomics in glioblastoma multiforme Olivier Gevaert, PhD Stanford University Cancer Center for Systems Biology Information Sciences in Imaging Program
Transcript

Radiogenomics in glioblastoma multiforme

Olivier Gevaert, PhDStanford University

Cancer Center for Systems BiologyInformation Sciences in Imaging Program

Overview

Multi-omics data integrationQuantitative image features in GBMIntegrating quantitative image features and

module networks

Multi-omics data integration

Discovering cancer driver genes and their targets

Integrative methods

• Need to develop statistical methods that allow to integrate multi-omics cancer data

• Create mechanistic models of cancer– how is gene expression influenced by genomic

events– how to identify cancer drivers and their targets

5

• Public domain data– Gene & miRNA

expression• Agilent & Affy microarray• RNA sequencing

– Copy number• Affy SNP 6.0

– DNA Methylation• Agilent Infinium (27k)

– Mutation• DNA sequencing

– Medical Images (MRI)

Method

• Two step algorithm1. Generating the List of Candidate Drivers2. Associating Candidate Drivers with their

Downstream Targets

7

Step 1Generating the List of Candidate Drivers

• If gene expression can be explained by genomic events

Gene

Copy Nr MethylationMutation

Rationale: Genes driven by multiple genomic

events in a significant subset of samples are unlikely to be randomly

deregulated.

Expressioncandidate driver gene

8

Gene expression – DNA methylation correlation

• Incorporating prior knowledge– β1 has to be positive

– β2 has to be negative

Step 1Generating the List of Candidate Drivers

• Model gene expression as a function of – Copy number– DNA methylation– Mutation data

Gene expression – Copy number correlation

Pairwise Spearman correlation

Step 2: Associating Candidate Drivers with their Downstream Targets

FOXM1

Cluster37

E2F8

Known transcription

factors

Drivergenes from

Step 1

Potential Regulators Ri

Linear Regression+

Lasso regularization

clustering

clus

ters

gene

s

patients

Module

Lee et al. PLoS Genetics, 2009

Results Step 2: Module network

GBM

Gevaert O and Plevritis S. PSB 2013

Results Step 2: Module network

GBM

Gevaert O and Plevritis S. PSB 2013

12

Results Step 2: Module network

GBM

Gevaert O and Plevritis S. PSB 2013

13

GBM Network

Top DNA repair module• DNMT1

– Key DNA methylation driver gene

• PARP1– binds DNMT1– known function in DNA damage

• CHAF1B– putative gene involved in DNA

repair– cross-cancer

• Top regulator in ovary & breast for DNA repair

Gevaert O and Plevritis S. PSB 2013

Summary

• Integration of multi-omics GBM data incorporating– gene expression– DNA methylation– copy number data

• Reduces molecular data dimensions– reduces the multiple testing correction problem

• Rich model going beyond clustering

Gevaert O and Plevritis S. PSB 2013

Quantitative image featuresTissue level data

GBM Medical Image Data

• 152 TCGA GBM patients have image data

• Imaging data is very heterogeneous– different planes– noisy annotation– different pulse sequences• T1 pre gadolinium• T1 post gadolinium • …

GBM Medical Image Data

GBM Medical Image Data

• AXIAL images– T1 pre gadolinium– T1 post gadolinium– T2 FLAIR

• Highlight different parts of the tumor– necrosis– enhancement– edema

• Manual annotation of ROIs for these concepts– 2D– largest slice for that lesion– for multifocal lesions ROI for each

GBM Medical Image Data

• Results– at least one ROI for 55 patients– ~40 patients have all ROI types

• Three ROI types are matched according to location to create a super ROI

• If multiple super ROIs, features are combined• Two readers + some redundancy to estimate

intra-reader variability

Quantitative image features

• Create a set of quantitative image features for each ROI

• Same feature set as Lung Cancer projectGevaert et al. Radiology 2012

Computational features

iPAD

Quantitative image features

Texture Features

Shape Features

Window (W)

Scale (S)

Edge Sharpness FeaturesGabor Filter Bank

150 Computational features

Jiajing Xu, Sandy Napel

Quantitative image features

Texture Feature

Shape Feature

Edge Sharpness

150-element feature vector

computational features

Computational features

iPAD

Computational features

iPAD

Quantitative image features

• Focused on 28 highly interpretable quantitative image features– Compactness of ROI– Edge sharpness– Edge Shape (LAII)

Explorative analysis

Size metrics of necrosis, enhancement and edema

• Created ratios of the size of each ROI vs. larger ROI– necrosis/enhancement– necrosis/edema– enhancement/edema

Size metrics of necrosis, enhancement and edema

• Comparison with the VASARI features

Size metrics of necrosis, enhancement and edema

• Comparison with the VASARI features

Size metrics of necrosis, enhancement and edema

• Comparison with the VASARI features

Size metrics of necrosis, enhancement and edema

• Correlated these with overall & progression free survival– no significant correlation with any survival outcome

• Potential problems– small data set, not enough power (<55 samples)– how to combine multi-focal lesions

• multi-focal necrosis• multi-focal enhancement

– 2D vs. 3D• Interesting

– size of edema is weakly correlated with progression free survival (p-value 0.03)

Univariate survival analysisQuantitative image feature ROI type Wald Test HR HR lower HR upper

RDS std Enhancement 0.0030232 1.6809 1.1925 2.3692Edge sharpness window kurtosis Necrosis 0.010502 1.4889 1.0976 2.0196

Compactness Enhancement 0.018057 1.4306 1.0632 1.925

Edge sharpness window skewness Necrosis 0.023249 1.4155 1.0485 1.9111

LAII std-5R Enhancement 0.023539 1.4792 1.0541 2.0757

LAII std-8R Enhancement 0.024792 1.5828 1.06 2.3636

Edge Sharpness scale max Edema 0.025168 1.49 1.0509 2.1125

Edge sharpness scale mean Edema 0.025448 1.4692 1.0484 2.059

LAII std-5R Edema 0.026997 1.4695 1.0448 2.0669RDS mean Edema 0.034709 1.4878 1.029 2.1514

Compactness Necrosis 0.037068 1.4827 1.0239 2.147LAII std-8R Edema 0.037589 1.4493 1.0215 2.0562

Creating a radiogenomic mapRadiogenomic map

Radiogenomics mapNecrosis

Radiogenomics mapNecrosis

• Compactness of Necrosis ROI– high = irregular shape– low = spherical shape

• Correlated with Module 64– P-value 0.0021

(Spearman rho, FDR 4%)– Inverse correlation High

compactnessLow

compactness

Radiogenomics mapNecrosis

• Edge sharpness window necrosis – high = blurry edge– low = sharp edge

• Correlated with Module 10– P-value 0.0178– Inversely correlated

Blurryedge

Sharpeedge

Overall summary

• Developed module network method that integrates/summarizes multi-omics data

• Gathered quantitative image features from MRI image data

• Correlated quantitative image features with modules

Acknowledgements

• Sylvia Plevritis• Greg Zaharchuk• Sandy Napel

• Lex Mitchell• Caroline Yu• Jiajing Xu• Chris Beaulieu

Questions

• Pulse sequence annotations• T1 Post is missing in many samples• Readers interested in annotating ROIs


Recommended