8/2/2012
1
Molecular Imaging in Radiation Oncology: What and Where?
Robert JerajAssociate Professor of Medical Physics, Human Oncology,
Radiology and Biomedical Engineering
Imaging and Radiation Sciences Program
University of Wisconsin Carbone Cancer Center, Madison, WI
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
8/2/2012
2
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!
Steenbakkers et al 2006, Int J Rad Oncol Biol Phys 64: 435
CT PET/CT
50% (30%-70%) decrease of the contouring standard deviation!
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
8/2/2012
3
Auto-segmentation, but which one?
Troost et al 2010, Radiother Oncol, 96(3): 328
CTVisual
SUV40%
SUV50%
SUV2.5
SBR
There are inherent uncertainties
SUV10
5
0
256, 2 iterations 256, 2 iterations 128, 2 iterations 128, 2 iterations 128,4 iterations
3 mm filter width 6 mm filter width 3mm filter width 6 mm filter width 6 mm filter width
256, 2 iterations 256, 2 iterations 128, 2 iterations 128, 2 iterations 128,4 iterations
3 mm filter width 6 mm filter width 3mm filter width 6 mm filter width 6 mm filter width
2D
OS
EM
3
D IT
ER
SUV measure CV (%) min - max (%)
SUVmax9 4 - 15
SUVmean5 1 - 8
B C D E F G I J K L M N-100
-50
0
50
100
150
2D Acquisition
Threshold-based
Gradient-based
Region-growing
Vo
lum
e V
ari
ati
on
s (
%)
Segmentation Techniques
128x128 Grid-Size256x256 Grid-Size
3D Acquisition
Impact of post-reconstruction filter width on target volumes
Impact on volume definition
8/2/2012
4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200
1
2
3
4
5
Av
g. M
arg
in A
xia
l P
lan
e (
mm
)
Patients 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200
5
10
15
20
25
30
Ma
x.
Marg
in A
xia
l P
lan
e
(mm
)
Patients
Marginplane
Mean ±SD (mm)
Maximum (mm)
Avg. Max ±SD (mm)
Maximum (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
Need for imaging margins
Is imaging itself enough?
Thiagarajan et al 2012, Int J Radiat Oncol Biol Phys, 83: 220
CT MRI PET
CT+PETCT+MRI CT+MRI+PET+Physical examination
CT+MRI+PET+P.E. > CT+PET or CT+MRICI (CT+PET, CT+MRI) = 0.62
Using FDG PET for target definition helps
because:
20%
19%
21%
20%
20% 1. It better defines where the tumor is
2. Increases consistency of target definition
3. It will make hospital administrators happy (more
revenue)
4. Doctors think so
5. It actually doesn’t help
8/2/2012
5
And here comes DOSE PAINTING…
Ling et al 2000, Int J Rad Oncol Biol Phys 47: 551
Dose painting workflow
Tumorbiology
Biological imaging1
Bio-basedprescription2
Planning & delivery3
Clinicaloutcome4
What are extra challenges?
Microscopy → Macroscopy
1 mm 5 cm
Proliferation
Hypoxia
Microscopy PET/CT imaging
Courtesy of A. van der Kogel, Nijmegen, NL
8/2/2012
6
Spatial resolution
40 µm 0.5 mm 1 mm 2 mm
We do not see small heterogeneities
0 5 10 15 20 25 300.0
0.2
0.4
0.6
0.8
1.0
1.2
Reco
ve
ry C
oe
ffic
ien
t
Sphere Diameter (mm)
Partial volume effects Recovery coefficients
Extraction of biological information
1 min
15 min
60 min
FLT PET/CT
8/2/2012
7
FDG PET/CT(metabolism)
What to dose paint?
FLT PET/CT(proliferation)
Cu-ATSM PET/CT(hypoxia)
FDG PET vs Time-to-progression
FDG SUV measure
Pre-treatmentp-val (N=19)
3 months post RTp-val (N=16)
6 months post RTp-val (N=11)
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 0.006
1 2 3 4
100
200
300
400
Days to P
rogre
ssio
n
Pre FDG SUVmean
1 2 3
0
100
200
300
400
Days to
Pro
gre
ssio
n
3 mo FDG SUVmean1 2 3
100
200
300
400
Days to
Pro
gre
ssio
n
6 mo FDG SUVmean
Pre-treatment 3 months post RT 6 months post RT
Regression analysis
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
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
Sarcomas
Carcinomas
8/2/2012
8
Which tracer to assess biology?
ATSMCu ATSMCuIIII
⇔
CR O2
H3O+ H2O
RSH
RSCu
ATSMH
I
2⇔ ATSMH2
NR O2
2FRNO FMISO ⇔
-2FRNO •
FRNO
NR
2FRNH
MM
REDOX
compartment
BOUND
compartment
DISSOCIATION
compartment
Bowen et al 2011, Nucl Med Biol, 38:771
pO2 transformation functions
0 1 2 3 4 5 6 7 80
10
20
30
40
50
60
Cu-ATSM model
1 SD CI
Cu-ATSM meas.
Lewis et al. 1999
SUV
PO
2 (
mm
Hg
)
0 1 2 3 4 5 6 7 8
10
20
30
40
50
60
SUV
FMISO model
1 SD CI
FMISO meas.
Lewis et al. 1999
FMISO meas.
Piert et al. 2000
Bowen et al 2011, Nucl Med Biol, 38:771
Cu-ATSM FMISO
0 1 2 3 4 5 660
64
68
72
76
80
84
88
92
Pmid = 3 mmHg, OERmax = 1.4
pH = 7.1
pH = 7.2
pH = 7.3
Pre
scri
bed
Do
se (
Gy
)
Cu-ATSM SUV0 1 2 3 4 5 6
60
64
68
72
76
80
84
88
92
pH = 7.2
Pre
scri
bed
Dose
(G
y)
Cu-ATSM SUV
Prescription function
Cu-ATSM
pH = 7.1
pH = 7.2
pH = 7.3
Dose
8/2/2012
9
Overall uncertainty in a patient
� Mean uncertainty of 20 % (max 60 %) in prescribed dose to individual patient
0 1 2 3 4 5 660
70
80
90
100
110
120
Pre
scri
bed
Dose
(G
y)
Cu-ATSM SUV
Upper Limit
Average
Lower Limit
Uncertainties in population
Symbol Parameter RangeDose Uncertainty
Mean (Max)
pH Intracellular Acidity 7.1 – 7.3 4 % (10 %)
HP2.5 FitDose Boost vs. Hypoxic
Proportion Function95 % CI 5 % (14 %)
Pmid Half-max Sensitization pO2 2 – 5 mmHg 1 % (2 %)
OERmax
Max Oxygen Enhancement
Ratio1.4 – 3.0 1 % (2 %)
Overall 10 % (17 %)
Patient 20 % (60 %)
(Gerweck 1998)
(Nilsson 2002)
(Chan 2008)
How many patients need dose painting?
� Imaging � 1/12 or 8.3%
� Eppendorf � 6/69 or 8.7%
0 1 2 3 4 555
60
65
70
75
80
85
90
95
Do
se (
Gy)
SUV0 10 20 30 40 50 60
55
60
65
70
75
80
85
90
95
Do
se
(G
y)
pO2 (mmHg)
N = 12 N = 69
Imaging patients Microelectrode patients
8/2/2012
10
Motion impact on dose painting
Dose painting workflow
Tumorbiology
Biological imaging1
Bio-basedprescription2
Planning & delivery3
Clinicaloutcome4
Micro→Macro
Which biology
Tracer
Extraction of biology
Set-up
Motion
Outcome uncertainties
Heuristic Modeling and Empirical Data
Uncertainty Characterization and Validation
What phenotype should we dose paint?
21%
19%
19%
20%
21% 1. Hypoxia (Cu-ATSM PET)
2. Metabolism (FDG PET)
3. Proliferation (FLT PET)
4. It depends on the histology
5. Whatever we have available in the hospital
8/2/2012
11
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 CTPre-RT CT
0 10 20 30 40 50 60 70
0
2
4
6
8
10
12
14
16
SU
Vm
ax
Time [days]
Average
FDG PET and radiation therapy
RT
Baardwijk et al 2007, Radiother Oncol, 82: 145.
8/2/2012
12
FDG PET and radiation therapy
0 10 20 30 40 50 60 70
0
2
4
6
8
10
12
14
16
SU
Vm
ax
Time [days]
Average
Metabolic non-responders
Metabolic responders
RT
Baardwijk et al 2007, Radiother Oncol, 82: 145.
FDG PET and radiation therapy
nSUVpost = nSUVduring- 20%
Kong et al 2007, J Clin Oncol, 25: 3116.
~30 days ~3 months
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
FDG PET/CT
3 months post RT
8/2/2012
13
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 adenocarcinomasnegative 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.
Pre
-tre
atm
en
tM
id-t
rea
tme
nt
(1 w
k o
f X
RT
)
FLT-PET/CT
Application: Treatment adaptation
8/2/2012
14
Pre
-tre
atm
en
tM
id-t
rea
tme
nt
(1 w
k o
f X
RT
)
FLT-PET/CT
Application: Treatment adaptation
Application: Dose painting
Pre
-tre
atm
en
tM
id-t
rea
tme
nt
Treatment response
Prescription function
Should we use FDG PET for treatment
response assessment?
20%
20%
20%
22%
18% 1. Absolutely
2. Yes, for post-treatment assessment
3. Yes, for early-treatment assessment
4. If there are enough hospital resources
5. If the physician requests it, it doesn’t matter
anyway
8/2/2012
15
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…
� Funding– NIH, PCF, UWCCC, Pfizer,
AstraZeneca, Amgen, EntreMed
� 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
� Medical Physics– Rock Mackie– Jerry Nickles– Onofre DeJesus
� Phase I Office