Abstract presented by Ron Dandurand, MD at REG 2016 Summit

1. Ron Dandurand, MD
Respiratory Effectiveness Group Summit
Lyon, France
April 15, 2016

2. Introduction

3. Introduction
 Definition, prevalence, social & economic impact

4. Introduction
 Definition, prevalence, social & economic impact
 Emphysema
 permanent enlargement of the acinus and destruction of
lung parenchyma distal to the terminal bronchioles
 present in up to half of COPD

5. Introduction
 Definition, prevalence, social & economic impact
 Emphysema
 permanent enlargement of the acinus and destruction of
lung parenchyma distal to the terminal bronchioles
 present in up to half of COPD
 Despite > 60 years of quantitative research CTs continue
to be characterized qualitatively by clinicians

6. Gough-Wentworth Sections
Dr. W. Thurlbeck, circa 1968
Courtesy of Dr. R. Fraser
McGill University

7. Gough-Wentworth Sections
Human Pathology 1970;1:215-226

8. Macroscopic Emphysema Quantification
 Limitations of Gough-Wentworth sections
 Autopsy
 Time consuming
 Operator dependent
 Kappa scores 0.43-0.59
 Qualitative CT
 Operator dependent
 Kappa scores 0.45-0.63
Radiology 1999;211:851–858
COPD 2012;9:151–159

9. Principles of CT Scanning

10. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree

11. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received

12. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received
 Moving patient through gantry converts plane/grid to
volume

13. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received
 Moving patient through gantry converts plane/grid to
volume
 Volumated pixels = Voxels

14. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received
 Moving patient through gantry converts plane/grid to
volume
 Volumated pixels = Voxels

15. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received
 Moving patient through gantry converts plane/grid to
volume
 Volumated pixels = Voxels
 Voxels have an attenuation number

16. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received
 Moving patient through gantry converts plane/grid to
volume
 Volumated pixels = Voxels
 Voxels have an attenuation number
 Software creates images from raw quantitative data

17. Principles of CT Scanning
 All materials attenuate (absorb/scatter) radiation to a
varying degree
 Rotating X-ray beam produces a 512 X 512 grid of
attenuation i.e. difference between delivered and
received
 Moving patient through gantry converts plane/grid to
volume
 Volumated pixels = Voxels
 Voxels have an attenuation number
 Software creates images from raw quantitative data
 Each attenuation number is assigned a shade of gray

18. Principles of CT Scanning
 Attenuation number = Hounsfield Unit (HU)
-1000 0 +1000
 Lung tissue threshold < -400 HU
Medical Physics 1998:25;2432-2439

19. CT Data Processing
 OsiriX MD v. 2.6 64-bit
 Decompression
 Inspection
 Reconstruction
 AirwayInspector
www.airwayinspector.acil.bwh.org
 Lung density histogram
 TLV
 LAA<-950 HU
 P15 LD

20. QCT Metrics of Interest
 TLV
 Total lung volume, ideally TLC
 Volume < -400 HU
 LAA
 Low attenuation area
 Fraction or % of volume < a set threshold e.g. -950 HU
 LD
 Lung density
 15th percentile of the lung density histogram + 1000 HU
 g / L

21. QCT Metrics of Interest

22. QCT Metrics of Interest
TLV (ml)

23. QCT Metrics of Interest

24. QCT Metrics of Interest

25. LAA%<-950 HU = 5%
QCT Metrics of Interest

26. LAA%<-950 HU = 5%
QCT Metrics of Interest

27. LAA%<-950 HU = 5%
LD = 15th Percentile + 1000 HU
e.g. -925 HU + 1000 HU = 75 g/L
QCT Metrics of Interest

28. Upper Lobe Centrilobular Emphysema

29. Upper Lobe Centrilobular Emphysema

30. Upper Lobe Centrilobular Emphysema

31. Index Case Motivating Study
 ZZ AATD

32. Index Case Revisited
0.0
0.1
0.2
0.3
0.4
0 5 10 15 20 25
LAA<-950HU
Time (months)
 ZZ AATD

33. Index Case Revisited
0.0
0.1
0.2
0.3
0.4
0 5 10 15 20 25
LAA<-950HU
Time (months)
0
10
20
30
40
50
60
0 5 10 15 20 25
LungDensity(g/L)
Time (months)
 ZZ AATD

34. Index Case Revisited
0.0
0.1
0.2
0.3
0.4
0 5 10 15 20 25
LAA<-950HU
Time (months)
0
10
20
30
40
50
60
0 5 10 15 20 25
LungDensity(g/L)
Time (months)
 ZZ AATD
Proc Am Thorac Soc 2008;5:925–928

35. What is Going On?

36. What is Going On?
 QCT in the research setting has stringent quality control
 Make and model of scanner
 Calibration protocol and frequency
 Acquisition protocol
 Reconstruction protocol
 Lung volume corrected to predicted TLC
 Use of contrast media

37. What is Going On?
 QCT in the research setting has stringent quality control
 Make and model of scanner
 Calibration protocol and frequency
 Acquisition protocol
 Reconstruction protocol
 Lung volume corrected to predicted TLC
 Use of contrast media
 Clinical CT scans cannot match these conditions

38. Hypotheses

39. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.

40. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.

41. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.
 Contrast media alters QCT metrics in a predictable
manner such that metrics from contrast and non
contrast scans may be compared if only correction
factors were available.

42. Methods: Subject Selection
 Population
 1000 CT scans from 600 COPD and 100 non-COPD subjects
 Subject selection
 2 CT scans within 13 months
 PFT within 14 months of at least 1 CT scan
 CT scan free of significant infiltrates
 1 focal > 4 cm
 2 focal > 2 cm
 Diffuse infiltrates
 Evolving infiltrates

43. Methods: CT Scan Selection
 109 CT scan pairs
 56 non-contrast/non-contrast
 53 contrast/non-contrast
 6 subjects had both non-contrast/non-contrast and
contrast/non-contrast pairs
 21 (19%) excluded because of infiltrates
 82 subjects included
 67 COPD
 15 non-COPD (14 asthma, 1 non-fibrotic sarcoidosis)
 RI-MUHC Review Board approval, 14-467-BMB

44. Methods: PFTs
 Spirometry before and after 200 μg salbutamol
 Lung volumes (Jaeger or Vmax22)
 Plethysmography, n=74
 Nitrogen washout, n=8
 Predicted values
 Spirometry, Knudson
 Lung volumes, Goldsman

45. Methods: CT Scanners
 10 different centres
 7 different models
 4 GE Lightspeed VCT
 1 GE Lightspeed Ultra
 1 GE Lightspeed CT750 HD
 2 Siemans Somatom Definition AS+
 1 Siemans Somatom Definition Edge
 1 Simeans Sensation 64
 1 Toshiba Aquillion

46. Methods: Volume Correction
 P15 LD
 P15 LDVC = P15 LD X TLV / Predicted TLC
 Reduces P15 LD in proportion to lung under-inflation
 Dirksen Proc Am Thorac Soc 2008;5:925–928
 LAA<-950 HU
 LAAVC-950 HU = LAA<-950 HU X Predicted TLC / TLV
 Increases LAA<-950 HU in proportion to lung under-inflation
 Both P15 LD and LAA<-950 HU corrected as above but using the
measured TLC

47. Biometrics and PFTs

48. GOLD Classification Distribution

50. CT-TLV vs. Predicted or Measured TLC
slope=1.15
r=0.76
P<0.001

51. CT-TLV vs. Predicted or Measured TLC
slope=1.15
r=0.76
P<0.001
slope=0.98
r=0.93
P<0.001

52. CT-TLV vs. Measured TLC-700 ml
slope=0.98
r=0.93
P<0.001
n=82

54. Effect of Contrast on TLV
slope=0.93
r=0.97
p<0.001
n=46

55. Effect of Contrast on TLV
slope=0.93
r=0.97
p<0.001
n=46
slope=0.90
r=0.91
p<0.001
n=42

56. Effect of Contrast on LAA<-950 HU
slope=1.01
r=0.95
p<0.001
n=46

57. Effect of Contrast on LAA<-950 HU
slope=1.01
r=0.95
p<0.001
n=46
slope=0.67
r=0.94
p<0.001
n=42

58. Effect of Contrast on Lung Density
slope=0.98
r=0.96
p<0.001
n=46

59. Effect of Contrast on Lung Density
slope=0.98
r=0.96
p<0.001
n=46
slope=1.21
r=0.96
p<0.001
n=42

61. Contrast Effect on LAAVC & LDVC
slope=0.75
r=0.94
p<0.001
n=46
slope=1.10
r=0.98
p<0.001
n=42

62. Lung Volume and Contrast Effects

63. Lung Volume and Contrast Effects

64. Lung Volume and Contrast Effects

65. Lung Volume and Contrast Effects

66. Lung Volume and Contrast Effects

67. Lung Volume and Contrast Effects

68. Lung Volume and Contrast Effects

69. Lung Volume and Contrast Effects

70. Lung Volume and Contrast Effects

71. Lung Volume and Contrast Effects

72. Lung Volume and Contrast Effects

73. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.
 Contrast media alters QCT metrics in a predictable
manner such that metrics from contrast and non
contrast scans may be compared.

74. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.
 Contrast media alters QCT metrics in a predictable
manner such that metrics from contrast and non
contrast scans may be compared.

75. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.
 Contrast media alters QCT metrics in a predictable
manner such that metrics from contrast and non
contrast scans may be compared.

76. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.
 Contrast media alters QCT metrics in a predictable
manner such that metrics from contrast and non
contrast scans may be compared.

77. Hypotheses
 Selected clinical CT scans performed on differing
scanners from differing centres and without uniform
quality control can provide reproducible QCT metrics.
 QCT metrics are more reproducible when corrected for
measured TLC than predicted TLC.
 Contrast media alters QCT metrics in a predictable
manner such that metrics from contrast and non
contrast scans may be compared. LAA=1.25, LD=0.90

78. Index Case Revisited
 ZZ AATD

79. Index Case Revisited
0.0
0.1
0.2
0.3
0.4
0 5 10 15 20 25
LAA<-950HU
Time (months)
0
10
20
30
40
50
60
0 5 10 15 20 25
LungDensity(g/L)
Time (months)
 ZZ AATD

80. Index Case Revisited
0.0
0.1
0.2
0.3
0.4
0 5 10 15 20 25
LAA<-950HU
Time (months)
0
10
20
30
40
50
60
0 5 10 15 20 25
LungDensity(g/L)
Time (months)
 ZZ AATD

81. Future Directions

82. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort

83. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast

84. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast
 Is it involuntary gas compression?

85. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast
 Is it involuntary gas compression?
 Compare early (CTA) with delayed contrast infusion

86. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast
 Is it involuntary gas compression?
 Compare early (CTA) with delayed contrast infusion
 Application

87. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast
 Is it involuntary gas compression?
 Compare early (CTA) with delayed contrast infusion
 Application
 Longitudinal real-world data

88. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast
 Is it involuntary gas compression?
 Compare early (CTA) with delayed contrast infusion
 Application
 Longitudinal real-world data
 Revisit –ve alpha-1 trials

89. Future Directions
 Validation in other cohorts
 ECLIPSE Cohort
 CanCOLD Cohort
 Elucidate mechanism behind reduced lung volume with
contrast
 Is it involuntary gas compression?
 Compare early (CTA) with delayed contrast infusion
 Application
 Longitudinal real-world data
 Revisit –ve alpha-1 trials
 “Digital structure-function” modelling
 i.e. QCT-OS phenotyping

90. Acknowledgements
 Harvard School of Medicine
 Raúl San José Estépar – Department of Radiology
 University of British Columbia
 Harvey Coxson – Department of Radiology
 McGill University
 Jean Bourbeau – RECRU
 David Eidelman – Meakins-Christie Labs
 Myriam Dandurand – Research Assistant