Transcript Multi-omics data integration - Cancer Imaging Archive Wiki

Radiogenomics in glioblastoma
multiforme
Olivier Gevaert, PhD
Stanford University
Cancer Center for Systems Biology
Information Sciences in Imaging Program
Overview
Multi-omics data integration
Quantitative image features in GBM
Integrating 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
• 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)
5
Method
• Two step algorithm
1. Generating the List of Candidate Drivers
2. Associating Candidate Drivers with their
Downstream Targets
Step 1
Generating the List of Candidate Drivers
• If gene expression can
be explained by genomic
events
candidate driver gene
Rationale:
Genes driven by multiple genomic
events in a significant subset of
samples are unlikely to be randomly
deregulated.
7
Copy Nr Methylation
Mutation
Expression
Gene
Step 1
Generating the List of Candidate Drivers
• Model gene expression as a
function of
Gene
Gene
expression
expression
– DNA
– Copy
methylation
number
correlation
– Copy number
– DNA methylation
– Mutation data
Ei = F ( b1CNi + b2 Methyli + b3 Muti )
• Incorporating prior
knowledge
– β1 has to be positive
– β2 has to be negative
Pairwise Spearman correlation
8
Step 2: Associating Candidate Drivers
with their Downstream Targets
Driver
genes from
Step 1
Known
transcription
factors
genes
patients
clustering
clusters
Potential
Regulators Ri
Expressioncluster-i = F (a1R1 + a 2R2 + ...+ a n Rn )
E2F8
FOXM1
Module
Cluster
37
Linear Regression
+
Lasso regularization
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
Results Step 2: Module network
GBM
12
Gevaert O and Plevritis S. PSB 2013
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
13
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 features
Tissue 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
2; < 8=>?=>355'
23'/ $, -'+) &$(4%41. '
25'6789: '
B "#*$, 4, '
B "#*$, 4, '
B "#*$, 4, '
@%A) %#". "%-'
@&". ) '
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 project
Gevaert et al. Radiology 2012
iPAD
Computational features
Quantitative image features
Texture Features
Shape Features
Edge Sharpness Features
Scale (S)
Gabor Filter Bank
150 Computational features
Window (W)
Jiajing Xu, Sandy Napel
Quantitative image features
iPAD
Computational features
computational features
Texture
Feature
Shape
Feature
Edge
Sharpness
150-element feature vector
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 analysis
Quantitative image feature
ROI type
Wald Test
HR
HR lower
HR upper
RDS std
Edge sharpness window kurtosis
Enhancement
Necrosis
0.0030232
0.010502
1.6809
1.4889
1.1925
1.0976
2.3692
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
RDS mean
Compactness
LAII std-8R
Edema
Edema
Necrosis
Edema
0.026997
0.034709
0.037068
0.037589
1.4695
1.4878
1.4827
1.4493
1.0448
1.029
1.0239
1.0215
2.0669
2.1514
2.147
2.0562
Creating a radiogenomic map
Radiogenomic map
Radiogenomics map
Necrosis
Radiogenomics map
Necrosis
• Compactness of
Necrosis ROI
– high = irregular shape
– low = spherical shape
• Correlated with
Module 64
– P-value 0.0021
(Spearman rho, FDR
4%)
– Inverse correlation
Low
compactness
High
compactness
Radiogenomics map
Necrosis
• Edge sharpness
window necrosis
– high = blurry edge
– low = sharp edge
• Correlated with
Module 10
– P-value 0.0178
– Inversely correlated
Sharpe
edge
Blurry
edge
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