COPDGene® : Genetic Epidemiology of
COPD
NA-MIC: All Hand MeetingRaúl San José & James Ross
BWH
Overview COPDGene Data Analysis Future Directions
COPD: Chronic Obstructive Pulmonary Disease
Definition: “Airflow limitation that is not fully reversible. The airflow limitation is usually progressive and associated with an abnormal inflammatory response of the lung to noxious particles or gases.”
Causes:
Overview
Overview COPDGene Data Analysis Future Directions
COPD: a growing disease
Projected to be the third-leading cause of death by 2020
24 million people affected by COPD in US
Overview
Overview COPDGene Data Analysis Future Directions
COPD: The Diseases
What is the underlying process?
Slowly progressive irreversible destruction of the lung tissue
Emphysema: air sacs that exchange gases
Airway disease (Chronic Bronchitis): airways that conduct these gases
End Result: Progressive shortness of breathCoughSputum production
Overview
Overview COPDGene Data Analysis Future Directions
Disease in Images
Airway Diseases
Emphysema Diseases
Overview
Overview COPDGene Data Analysis Future Directions
COPDGene
Only 20% of smokers develop COPD
COPDGene
Genetic factors
Multi-center study funded by the National Heart, Lung and Blood Institute (NHLBI).
Co-PIs: Drs. James Crapo, Edwin Silverman.
21 clinical sites
3 Image analysis centers:
• Denver
• Iowa
• BWH
Overview COPDGene Data Analysis Future Directions
COPDGene: Hypothesis
1) Precise phenotypic characterization of COPD subjects using computed tomography, as well as clinical and physiological measures, will provide data that will enable the broad COPD syndrome to be segregated into clinically significant subtypes.
2) Genome-wide association studies will identify genetic determinants for COPD susceptibility that will provide insight into clinically relevant COPD subtypes.
3) Distinct genetic determinants influence the development of emphysema and airway disease.
COPDGene
Overview COPDGene Data Analysis Future Directions
COPDGene: Goals
• Collect a large population of COPD subjects and smokers without COPD: 10,000
• Two racial/ethnic groups: Non-Hispanic whites and Non-Hispanic African Americans
• Extensive Characterization (Chest CT scans)– Inspiratory, Expiratory– High resolution CT (0.7mm isotropic with 50% overlap)– 2 reconstructions kernels (smooth and sharp)
• Use genome-wide association analysis to find inherited causes of COPD and COPD subtypes
COPDGene
Overview COPDGene Data Analysis Future Directions
COPDGene: Study Design
COPDGene
Figure 7. Genome Wide Association of COPD: Study Design
Phase 1
Phase 2
Phase 3
Genome Wide Screen
Case/Control Cohorts
Non-Hispanic Whiten=3000/3000
African Americann=1500/1500
Boston Early Onset COPD Family Study
n=160 families,1100 individuals
~6000 SNP Candidates
50 regions (flanking SNPs)
Fast track panel*
Case/ControlNon-Hispanic White
n=1000/1000
Case/ControlNon-Hispanic White
n=1000/1000
Case/ControlAfrican American
n=500/500
Case/ControlAfrican American
n=500/500
Phase 4
Case/ControlNon-Hispanic White
n=1000/1000
Case/ControlAfrican American
n=500/500
Candidate Gene Analysis
~6000 SNP Candidates
50 regions (flanking SNPs)
~3-5 Gene Candidates ~3-5 Gene Candidates
International COPD Genetics
Networkn=1150 probands
+ 1950 sibs
Confirmation of SNP Signals
Region Mapping
GOLD 1 Cohort
Non-Hispanic Whiten=1000
African Americann=500
Overview COPDGene Data Analysis Future Directions
COPDGene: Current status
COPDGene
• Since Sept 2007 over 6500 subjects have been recruited.
• First 2500 have been fully analyzed.
Overview COPDGene Data Analysis Future Directions
Challenges
• Data management– Storage: 40,000 scans, 20 TB expected to archive full
cohort– Quality control: Initial and processing– Querying and retrieval
• Image analysis: inspiration and expiration– Lung area and lobes– Airway segmentation– Robust: outlier detection– Minimal user input
• High-throughput analysis– Batch analysis– Grid deployment
Data Analysis
Overview COPDGene Data Analysis Future Directions
Analysis Centers
• National Jewish (Denver): Drs. Lynch and Schroeder– Q&A– Emphysema and Airway analysis using VIDA– Primary archiving center– Result integration
• BWH (Boston): Dr. Washko– High-throughput emphysema analysis– New phenotypes– Data replication
• University of Iowa: Dr. Hoffman– COPDGene Phantom Analysis (Q&A)– VIDA customization
Data Analysis
Overview COPDGene Data Analysis Future Directions
BWH Pipeline
• Lung Imaging Platform– ITK-based filters for lung extraction, lobe
segmentation and parenchymal texture analysis– Emphysema quantification– Air trapping quantification– Tracheal air correction– VTK-based filters for airway measurements– Command Line Tools
• Grid Wizard Enterprise for pipeline deployment
• Iterative Lobe segmentation in Slicer 3
Data Analysis
Overview COPDGene Data Analysis Future Directions
BWH Pipeline
• Filters– itkLungConventions– itkExtractLungLabelMapImageFilter– itkPartialLungLabelMapImageFilter– itkMergeLungLabelMapsImageFilter– itkAutoThresholdAirwaySegmentatio
nImageFilter– itkWholeLungVesselAndAirwaySegm
entationImageFilter– itkSplitLeftAndRightLungsImageFilte
r– itkLabelLungRegionsImageFilter– itkAirwayGraphTraits– itkImageToAirwayGraphFilter– itkImageToAirwayGraphFunctor– itkMinCostPathAirwaySegmentationI
mageToGraphFilter
Data Analysis
• Tools– ConvertDicom– GeneratePartialLungLabelMap– GenerateEmphysemaMeasures– QualityControl– ExtractLungLabelMap– MergeLungLabelMaps– GenerateHUStatistics– SegmentationAssistant– GenerateMedianFilteredImage– SplitLeftAndRightLungs– LabelLungRegions– GenerateAirwayPhantom
Overview COPDGene Data Analysis Future Directions
BWH Pipeline
Data Analysis
Overview COPDGene Data Analysis Future Directions
Data Quality: Dashboard
Data Analysis
Overview COPDGene Data Analysis Future Directions
Future Directions
Future Directions
• Airway Analysis– User-driven path extraction– Fully automatic approach based on Scale-space
particles
• Development of new airway phenotypes– Role of airway density in disease– New approaches for emphysema
• Multi scanner brand correction• Automatic pipeline supervision and reporting• Insp.-Exp. registration for lobar ventilation
assessment