8/2/2012
Molecular Imaging in
Radiation Oncology:
What and Where?
Robert Jeraj
Associate Professor of Medical Physics, Human Oncology,
Radiology and Biomedical Engineering
Imaging and Radiation Sciences Program
University of Wisconsin Carbone Cancer Center, Madison, WI
rjeraj@wisc.edu
Current state of affairs…
Hong and Harari, 2005
FDG PET/CT or CT?
GTVPET < GTVCT
75%
GTVPET > GTVCT
20%
Paulino et al 2005, Int J Radiat Oncol Biol Phys 61: 1385
1
8/2/2012
FDG PET/CT or CT?
“Physicians B and C used contradictory techniques. Physician B would
contour the suspected GTV on the basis of the CT but often totally
disregard this information when given the PET information and contour
only PET avidity. Physician C, on the other hand, would contour the
suspected GTV on the basis of the CT, and then, for the fusion contour,
draw the union of the CT contour and PET avidity.”
“Physician A’s method tended to be a mixture of the methods of
Physicians B and C. Often, Physician A would “split the difference” and
contour the compromise between the drawn CT contour and PET avidity.”
Riegel et al 2006, Int J Radiat Oncol Biol Phys 65: 726
Whatever, adding PET info HELPS!
CT
PET/CT
50% (30%-70%) decrease of the contouring standard deviation!
Steenbakkers et al 2006, Int J Rad Oncol Biol Phys 64: 435
How to increase reproducibility?
AAPM TG211 - Classification, Advantages and
Limitations of the Auto-Segmentation Approaches
for PET
– Manual segmentation is NOT the way to go!
– Auto segmentation
•
•
•
•
•
Thresholding (Erdi 1997, Paulino 2004)
Gradient-based (Geets 2007)
Region-growing (Drever 2006)
Feature-based (Yu 2009)
…
– Reference benchmark dataset
2
8/2/2012
Auto-segmentation, but which one?
CT
Visual
SUV40%
SUV50%
SUV2.5
SBR
Troost et al 2010, Radiother Oncol, 96(3): 328
There are inherent uncertainties
256, 2 iterations
3 mm filter width
256, 2 iterations
6 mm filter width
128, 2 iterations
3mm filter width
128, 2 iterations
6 mm filter width
128,4 iterations
6 mm filter width
3D ITER
SUV
10
2D OSEM
5
0
256, 2 iterations
3 mm filter width
256, 2 iterations
6 mm filter width
128, 2 iterations
3mm filter width
SUV measure
128, 2 iterations
6 mm filter width
CV (%)
9
5
SUVmax
SUVmean
128,4 iterations
6 mm filter width
min - max (%)
4 - 15
1-8
Impact on volume definition
Volume Variations (%)
Impact of post-reconstruction filter width on target volumes
150
Threshold-based
Gradient-based
Region-growing
100
50
0
-100
128x128 Grid-Size
256x256 Grid-Size
-50
3D Acquisition
2D Acquisition
B
C
D
E
F
G
I
J
K
L
Segmentation Techniques
M
N
3
8/2/2012
Need for imaging margins
30
Max. Margin Axial Plane (mm)
Avg. Margin Axial Plane (mm)
5
4
3
2
1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Patients
25
20
15
10
5
0
1
2
3
4
5
6
7
8
9 10 11 12
Patients
Margin
plane
Mean ±
SD (mm)
Maximum
(mm)
Avg. Max ±
SD (mm)
Axial
1.0 ± 0.4
1.8
8.4 ± 6.0
26.0
Coronal
0.5 ± 0.3
1.2
10.6 ± 6.1
26.7
Sagittal
0.5 ± 0.3
1.2
10.2 ± 5.3
22.1
13 14 15 16 17 18 19 20
Maximum
(mm)
Is imaging itself enough?
CT+MRI
CT
CT+PET
MRI
CT+MRI+PET+Physical examination
PET
CT+MRI+PET+P.E. > CT+PET or CT+MRI
CI (CT+PET, CT+MRI) = 0.62
Thiagarajan et al 2012, Int J Radiat Oncol Biol Phys, 83: 220
Using FDG PET for target definition helps
because:
20%
1. It better defines where the tumor is
20%
2. Increases consistency of target definition
21%
19%
20%
3. It will make hospital administrators happy (more
revenue)
4. Doctors think so
5. It actually doesn’t help
4
8/2/2012
And here comes DOSE PAINTING…
Ling et al 2000, Int J Rad Oncol Biol Phys 47: 551
Dose painting workflow
Tumor
biology
1
Biological
imaging
2
Bio-based
prescription
3
Planning &
delivery
4
Clinical
outcome
What are extra challenges?
Microscopy → Macroscopy
Microscopy
1 mm
PET/CT imaging
5 cm
Courtesy of A. van der Kogel, Nijmegen, NL
Proliferation
Hypoxia
5
8/2/2012
Spatial resolution
40 µm
0.5 mm
1 mm
2 mm
We do not see small heterogeneities
Partial volume effects
Recovery coefficients
Recovery Coefficient
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0
5
10
15
20
25
30
Sphere Diameter (mm)
Extraction of biological information
1 min
15 min
60 min
FLT PET/CT
6
8/2/2012
What to dose paint?
FDG PET/CT
(metabolism)
FLT PET/CT
(proliferation)
Cu-ATSM PET/CT
(hypoxia)
FDG PET vs Time-to-progression
FDG SUV
measure
Pre-treatment
p-val (N=19)
SUVmean
0.94
0.005
0.0002
SUVmax
0.86
0.017
0.003
SUVpeak
0.90
0.046
0.004
SUVtotal
0.51
0.047
300
200
100
400
300
200
100
0
2
3
4
1
Pre FDG SUVmean
6 months post RT
Days to Progression
400
1
0.006
3 months post RT
Days to Progression
Days to Progression
Pre-treatment
3 months post RT 6 months post RT
p-val (N=16)
p-val (N=11)
2
400
300
200
100
3
1
3 mo FDG SUVmean
2
3
6 mo FDG SUVmean
Regression analysis
Sarcomas
3mo FDG Regression, N=7
β-FDGpre β-FLTpre β-CuPre β-FLTmid β-CuMid
mean
0.42
-0.23
0.03
0.21
0.25
P-val
0.01
0.35
0.84
0.29
0.13
Carcinomas
3mo FDG Regression, N=11
β-FDGpre
β-FLTpre
β-CuPre
β-FLTmid
β-CuMid
mean
0.15
-0.25
-0.14
0.21
0.47
P-val
0.11
0.01
0.24
0.45
0.001
7
8/2/2012
Which tracer to assess biology?
Cu IIATSM ⇔
Cu IIATSM
CR
FMISO ⇔
FRNO 2
NR
O2
H3O+
REDOX
compartment
H2O
NR
BOUND
compartment
RSH
O2
FRNO •2-
FRNO
MM
DISSOCIATION
compartment
H 2 ATSM ⇔ H 2 ATSM
FRNH 2
Cu I RS
Bowen et al 2011, Nucl Med Biol, 38:771
pO2 transformation functions
Cu-ATSM
FMISO
60
60
Cu-ATSM model
1 SD CI
Cu-ATSM meas.
Lewis et al. 1999
PO2 (mmHg)
50
40
40
30
30
20
20
10
10
0
0
1
2
3
4 5
SUV
6
7
FMISO model
1 SD CI
FMISO meas.
Lewis et al. 1999
FMISO meas.
Piert et al. 2000
50
0
8
1
2
3
4
5
6
7
8
SUV
Bowen et al 2011, Nucl Med Biol, 38:771
Prescription function
Pmid = 3 mmHg, OERmax = 1.4
Prescribed Dose (Gy)
Cu-ATSM
92
Dose
pH
= 7.2
pH = 7.1
pH = 7.2
pH = 7.3
88
84
80
76
72
68
64
pH = 7.1
60
pH = 7.3
0
1
2
3
4
5
Cu-ATSM SUV
6
8
8/2/2012
Overall uncertainty in a patient
Prescribed Dose (Gy)
120
Upper Limit
Average
Lower Limit
110
100
90
80
70
60
0
1
2
3
4
5
6
Cu-ATSM SUV
Mean uncertainty of 20 % (max 60 %) in prescribed
dose to individual patient
Uncertainties in population
Symbol
Parameter
pH
Intracellular Acidity
HP2.5 Fit
Dose Boost vs. Hypoxic
Proportion Function
Pmid
Half-max Sensitization pO2
Range
Dose Uncertainty
Mean (Max)
7.1 – 7.3
4 % (10 %)
(Gerweck 1998)
95 % CI
5 % (14 %)
2 – 5 mmHg
1 % (2 %)
(Nilsson 2002)
Max Oxygen Enhancement
Ratio
OERmax
1.4 – 3.0
1 % (2 %)
(Chan 2008)
Overall
10 % (17 %)
Patient
20 % (60 %)
How many patients need dose painting?
Imaging patients
Microelectrode patients
95
95
N = 12
90
Dose (Gy)
Dose (Gy)
85
80
75
70
65
60
55
N = 69
90
85
80
75
70
65
60
0
1
2
3
4
5
55
SUV
0
10
20
30
40
50
60
pO2 (mmHg)
Imaging 1/12 or 8.3%
Eppendorf 6/69 or 8.7%
9
8/2/2012
Motion impact on dose painting
Dose painting workflow
Heuristic Modeling and Empirical Data
Tumor
biology
1
Biological
imaging
2
Bio-based
prescription
3
Planning &
delivery
4
Clinical
outcome
Uncertainty Characterization and Validation
Micro→Macro
Tracer
Set-up
Which biology
Extraction
of biology
Motion
Outcome
uncertainties
What phenotype should we dose paint?
21%
20%
19%
19%
21%
1.
2.
3.
4.
5.
Hypoxia (Cu-ATSM PET)
Metabolism (FDG PET)
Proliferation (FLT PET)
It depends on the histology
Whatever we have available in the hospital
10
8/2/2012
WHAT AND WHERE TO TARGET?
Using molecular imaging helps in target definition, but
still many issues to resolve:
– Choice of molecular imaging
– Image quantification
– Automatic segmentation
– Validation clinical trials
Molecular imaging-assisted target definition using
molecular imaging in qualitative way is the way to go
at present!
Dose painting is an extremely exciting concept, but we
are just at the beginning
Response during radiation therapy
Mid-RT CT
Pre-RT CT
FDG PET and radiation therapy
16
14
12
SUVmax
10
Average
8
6
4
RT
2
0
0
10
20
30
40
50
60
70
Time [days]
Baardwijk et al 2007, Radiother Oncol, 82: 145.
11
8/2/2012
FDG PET and radiation therapy
16
14
12
Metabolic non-responders
SUVmax
10
Average
8
6
Metabolic responders
4
2
RT
0
0
10
20
30
40
50
60
70
Time [days]
Baardwijk et al 2007, Radiother Oncol, 82: 145.
FDG PET and radiation therapy
nSUVpost = nSUVduring- 20%
~30 days
~3 months
Kong et al 2007, J Clin Oncol, 25: 3116.
Radiation induced inflammation
Radiation induced inflammation is a known effect –
temporal and spatial dependence
Not known how much it is a confounding factor in
treatment assessment
FDG PET shows increased uptake post therapy
3 months post RT
FDG PET/CT
12
8/2/2012
FDG PET late response
HNSCC: Negative FDG PET results post chemoRT have a high
NPV (95%), but low PPV (50%) (Schöder et al 2009, J Nucl Med,
50:74S)
NSCLC: 80% decrease in FDG PET SUVmax post chemoRT has
90% sensitivity, 100% specificity, and 96% accuracy for predicting
pathologic response (Cerfolio et al 2004, Ann Thorac Surg,
78:1903)
Rectal cancer: 70% decrease in FDG PET SUVmax post
chemoRT has 79% specificity, 81% sensitivity, 77% PPV, 89%
NPV and 80% accuracy for predicting pathological response
(Caprici et al 2007, Eur J Nucl Med Mol Imaging, 34:1583)
Esophageal cancer: Mixed results - in adenocarcinomas
negative FDG PET post chemoRT has a high PPV, elsewhere
inconclusive (Krause et al 2009, J Nucl Med, 50:89S)
FLT PET and radiation therapy
Everitt et al 2009, Int J Rad Oncol Biol Phys, 75: 1098.
Mid-treatment
(1 wk of XRT)
Pre-treatment
Application: Treatment adaptation
FLT-PET/CT
13
8/2/2012
Mid-treatment
(1 wk of XRT)
Pre-treatment
Application: Treatment adaptation
FLT-PET/CT
Pre-treatment
Application: Dose painting
Prescription function
Mid-treatment
Treatment response
Should we use FDG PET for treatment
response assessment?
18%
22%
20%
20%
20%
1.
2.
3.
4.
5.
Absolutely
Yes, for post-treatment assessment
Yes, for early-treatment assessment
If there are enough hospital resources
If the physician requests it, it doesn’t matter
anyway
14
8/2/2012
WHAT AND WHERE TO ASSESS?
PET imaging for response assessment in RT still in
its infancy, but with some encouraging results
– Late FDG PET response assessment has high
predicting value of pathological response in many
tumors
– Early FDG PET response assessment limited
because of radiation-induced inflammation
– Alternatives, especially early FLT PET response
assessment promising for early assessment but
lacks clinical validation
Many applications for assessment (treatment
adaptation, dose painting), but much more appealing
with early treatment assessment
Normal tissue assessment should not be forgotten
Thanks to:
Image-guided therapy group
– Vikram Adhikarla
– Tyler Bradshaw
– Enrique Cuna
– Ngoneh Jallow
– Matt La Fontaine
– Paulina Galavis
– Stephanie Harmon
– Courtney Morrison
– Surendra Prajapati
– Urban Simoncic
– Peter Scully
– Benny Titz
– Natalie Weisse
– Koala Yip
– Stephen Yip
– Former students…
Medical Oncology/Hematology
– Glenn Liu
– George Wilding
– Mark Juckett
– Brad Kahl
– Anne Traynor
Human Oncology
– Søren Bentzen
– Paul Harari
– Mark Ritter
Radiology
– Scott Perlman
– Chris Jaskowiak
Veterinary School
– Lisa Forrest
– David Vail
Funding
– NIH, PCF, UWCCC, Pfizer,
AstraZeneca, Amgen, EntreMed
Medical Physics
– Rock Mackie
– Jerry Nickles
– Onofre DeJesus
Phase I Office
15