Image Guided Adaptive Brachytherapy (IGABT)
for Cervical Cancer.
Uncertainties in dose reporting
Astrid de Leeuw
University Medical Center Utrecht
The Netherlands
Outline Uncertainties Dose Reporting
Background:
– Procedure in Utrecht
1. Geometrical uncertainties:
– intra/inter application variation
2. Combination External Beam and Brachy
3. Same dose parameters same dose distribution??
Utrecht Treatment Cervical Cancer
MR guided adaptive RT
39.6 Gy EBRT
19.2 Gy
PDR-BT
5.4 Gy
EBRT
19.2 Gy
PDR-BT
10-14 Gy
EBRT
External Beam IMRT
25 x 1.8 Gy (44.3 Gyαβ10)
Cisplatin
Brachytherapy
two applications
Pulsed Dose Rate (PDR)
(or 4xHDR)
32 pulses of 60 cGy
1pulse/hour
MR guided treatment planning optimization
according to GEC-ESTRO recommendations
Utrecht Applicator
Nucletron Veenendaal
1
GEC-ESTRO recommendations:
Target definition
Reporting Dose Volume parameters
DVH analysis: aims and constraints on total dose: EBRT + Brachytherapy
DVH analysis based on EQD2 with :
α/β(target)
=10 Gy
α/β(OAR)
=3 Gy
T1/2
=1.5 h
GTV
HR-CTV
IR-CTV
bladder
rectum
bowel
Dose volume parameters:
target:
D90 and D100
OAR:
D2cc
Haie-Meder et al. Radiother Oncol 2005
Pötter et al. Radiother Oncol 2006
Optimization
standard plan
optimized plan IC
optimized plan
in Utrecht:
aim:
D90 HR-CTV >84 Gyαβ10
constraints:
D2cc bladder <90 Gyαβ3
D2cc rect/sigm/bowel <75 Gyαβ3
optimized plan IC/IS
Uncertainties
•
anatomical
– contouring
•
geometrical
– registration between CT and MR
– Organ motion and deformation
– applicator reconstruction
•
physical
– QA of both EB and BT
•
combining dose distributions and dose parameters
•
biological
– dose rate, α/β, repair half time T1/2
•
clinical
– scoring tumour response, scoring toxicity
2
Geometrical dose uncertainties Organ motion
planning versus irradiation
•
•
Intra application:
– PDR during 32 hours
– HDR: planning and irradiation 2-3 h later
Inter application
Results of some (small) investigations:
– Study in 2008 PDR
second MR on day 2 of PDR
– Data pooling from 6 institutes
– Study PDR patients 2011
– Study HDR patients 2011
Study Intra-application in Utrecht 2008
An example: high decrease rectum
Appl1/day 1
Appl2/day 1
Appl1/day 2
Appl2/day2
D90 hr-ctv
D2 bladder
D2 rectum
+0.4 Gyαβ10
+2.4 Gyαβ3
-10.5 Gy αβ3
D90 hr-ctv
D2 bladder
D2 rectum
RUTRxxx Li
+02 Gyαβ10
+0.1Gyαβ3
+3.0 Gy αβ3
Study Intra-application Utrecht 2008
Change in total dose EQD2 Gy
9 patients /18 applications
MR during PDR on day 2 of application
Average time interval between MR acquisitions 22.0 (SD 2.5) hour
HR-CTV
18 PDR fractions: difference plan on MRday2 versus MRplan
bladder
rectum
5
4
number of fractions
3
On average differences limited,
but variation can be large
2
1
0
-11.0
-10.0
-9.0
-8.0
-7.0
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
incre ase in Gy EQD2
18 PDR fractions: Difference plan on MRday2 versus MRplan
note: sensitive to contouring
15
EQD2
HR-CTV
Bladder
REctum
D90
D2cc
D2cc
Intra
n=18
Mean SD
-0.2 2.0
1
0.6
3
3.9
10
min
-4.0
-4
-10.5
max
4.7
8
4.9
difference in EQD2 Gy
•
•
•
5
5%
max
0
avg
HR-CTV
-5
bladder
rectum
min
95%
-10
-15
3
Multicenter comparison
GYN GEC-ESTRO Network
#
patients
treatment
fractions
time range
4
Image type
variation type
images
1
21
HDR
18-20 hrs
MRI
84
Intra-app.
2
21
HDR
3
5 hrs
MRI
72
Intra-app.
3
9
PDR
2 x 29 / 32
22 hrs
MRI
36
Intra-app.
4
14
HDR
5
1-22 days
CT
69
Inter-app.
5
27
HDR
4
7-10 days
MRI
54
Inter-app.
6
31
PDR
2 x 20
1 week
MRI
62
Inter-app.
123 patients
5 h – 3 weeks
377
3+3
• Different methods used for reporting of variations in single institution publications:
total EQD2 (EBRT+BT), EQD2 for BT, physical dose changes for BT, single plan vs. multiple
plans, etc.
• For direct comparison, raw data were collected and analysed using a common method.
Nesvacil et al. Radiother Oncol 2012 abst ESTRO Barcelona
Multicenter Comparison - Preliminary Results:
relative change in physical dose between images
Center
∆ D90 [%] (fixed plan,
variable anatomy)
∆ D2cm³ between 2 acquisitions [%] (fixed plan, variable anatomy)
bladder
#
mean
median
rectum
SD
mean
median
sigmoid/bowel
SD
mean
median
HR CTV
SD
mean
SD
20-30%
1 Conclusion:
1.5
-2.5
21.9 about
6.1
2.6
21.6 for
-2.7 OAR
-3.9
22.8
-3.1
2
-0.1
2.5
14.7 about
0.6
0.0
17.4 for
1.8 HR-CTV
3.6
24.6
SD
10-15%
3
3.7
2.0
12.0
3.6
11.3
23.4
-1.1
Note:
4
6.4
7.5
24.7
8.6
11.5
24.8
5 translation
0.9
-1.8 to EQD2
13.1
0.6
-1.4
15.1
11.9
3.7
37.5
-0.9
complicated:
6
2.1
-1.6
24.3
5.2
-0.2
26.0
2.3
3.1
21.3
1.6
larger
impact
for
higher
pulse/fraction
dose
(HDR)
total
SD
-5.6
11.6
-0.6
9.0
-5.5
19.4
1.0
10.4
4.1
22.0%
1.6
-0.9
26.8%
-1.1
-1.7
13.1%
Intraaplication
1.3
1.5
17.7
3.8
2.3
20.5
-2.3
-3.7
23.5
-2.5
-4.3
10.8
interapplication
3.9
0.0
22.3
5.8
5.2
23.2
6.8
3.7
30.2
0.4
-0.8
15.1
2.7
1.5
20.3%
median
4.5
Note:
Changes correspond to physical dose change between 2 time points during course of BT
Effect on total EQD2 (EBRT+BT) depends on fractionation schedule (PDR, HDR, …)
Study Intrafraction Utrecht 2011
Results: Change in total dose EQD2 Gy
10 patients / 20 fractions
– 5 patients: MR the same day
– 5 patients: MR the second day
EQD2
HR-CTV
D90
Bladder
Rectum
Sigmoid
Bowel
D2cc
D2cc
D2cc
D2cc
Intra
n=20 All
Mean SD
-0.4 2.7
-0.1
3.1
1.4
-0.9
3.3
5
4.2
6.4
min
-5
max
6.8
-7.9
-4.3
-6.8
-14
5.1
18.1
13
9.2
Intra
n=10 MRafternoon
Mean SD min max
-0.6 2.2 -4.6 2.2
-0.6
0.5
0.2
0.6
2.9
3.3
2.5
4.2
-5.5
-4.3
-2.6
-4.9
4.2
6.2
5.0
7.9
20 PDR fractions
20.0
Intra
n=10 MRday2
Mean SD min
-0.2 3.3 -5.0
0.4
5.8
2.6
-2.4
3.7
5.1
5.4
8.0
-7.9
0.5
-6.8
-14.0
max
6.8
Difference in EQD2 Gy
•
5.1
18.1
13.0
9.2
15.0
10.0
max
avg
min
95%
0.0
-5.0
-10.0
HR-CTV
10 PDR fractions : MR2 in afternoon
bladder
rectum
10 PDR fra ctions: MR2 on day2
20.0
20.0
15.0
10.0
5%
max
5.0
avg
min
95%
0.0
-5.0
-10.0
Difference in EQD2 Gy
Difference in EQD2 Gy
5%
5.0
15.0
10.0
5%
max
5.0
avg
min
95%
0.0
-5.0
-10.0
HR-CTV
bladder
rectum
HR-CTV
bladder
rectum
4
Comparison
Intra-application2008 and Intra-application2011
18 PDR fractions: Difference plan on MRday2 versus MRplan
10 PDR fractions: MR2 on day2
20
20.0
15.0
10
5%
max
5
avg
0
min
HR-CTV
bladder
rectum
95%
-5
Difference in EQD2 Gy
d ifferen ce in EQ D2 G y
15
10.0
max
avg
0.0
min
95%
-5.0
-10
-10.0
-15
-15.0
Message: be aware!
5%
5.0
HR-CTV
bladder
rectum
Message:
Study 2008 Rather big uncertainties
versus in rectum/sigmoid
Study 2011: dose
On avg 31
stable
rectum
dose
versus
increase in rectum dose
during
hours
of PDR
Why?to more HDR
Strong rationale to change
Cause: New workflow:
no X-ray image, no contrast in rectum
MR scanning directly after application
Result: Empty rectum at planning MR, increase of gas during treatment
MR at department
In
Shielded Treatment Room
HDR system secured
Application on MR table
Imaging workflow HDR patients.
HDR : 2 applications, with 2 fractions each
Application1
BT1
Fraction1
Application2
BT2
Fraction2
MRplan
optimised plan BT1
BT4
Fraction2
MRplan
MRpreRad
m atch on applicator
BT3
Fraction1
MRpreRad
MRpreRad
MRpreRad
MRpostRad
MRpostRad
m atch (appl) w ith MRplanBT1
contours of M Rplan on MRpreRad contours of MRplanBT1 on MRBT2
OK?
OK?
irradiation opt planBT1
irradiation opt planBT1
MRpostRad
MRpostRad
≥10 MR scans
5
HDR workflow: Illustration
MRplan BT3
MRpreRad BT4
Visual inspection:
OK? > irradiation
if not OK >
adept contours
match on applicator
adapting contours
contours back to master
DVH analysis in BPS
OK? > irradiation
if not OK > adept plan
D2cc rectum 4.2 5.8 Gy
HDR Intra-application variation
•
Difference DVH parameters
MRplan and MRpreRad (3-4 hour):
– very small for HR-CTV
– on average increase in rectum:
– SD for rectum 3 Gy EQD2, for
bladder 5 Gy EQD2.
5%
max
avg
min
95%
HR-CTV
bladder
rectum
difference MRpostRad-MRpreRad
Difference MRpreRad and MRpostRad
– small
8.0
6.0
4.0
2.0
0.0
-2.0
-4.0
-6.0
-8.0
-10.0
5%
max
avg
min
95%
HR-CTV
bladder
rectum
HDR Intra-application variation
Difference in EQD2 Gy
difference MRpreRad-MRplan
Start using rectumprobe
8.0
6.0
4.0
2.0
5%
max
0.0
-2.0
-4.0
-6.0
-8.0
-10.0
avg
min
95%
HR-CTV
bladder
rectum
learning phase!
MRpreRad - MRplan
the more you see, the more
you can control!
8.0
6.0
Difference in Gy EQD2
•
8.0
6.0
4.0
2.0
0.0
-2.0
-4.0
-6.0
-8.0
-10.0
Difference in EQD2 Gy
5 patients
Difference in EQD2 Gy
difference MRpreRad-MRplan
•
D90 HR-CTV
4.0
D2cc bladder
2.0
0.0
-2.0
D2cc rectum
P1
P2
P3
P4
P5
D2cc sigmoid
-4.0
-6.0
-8.0
-10.0
6
Summary Geometric Uncertainties
•
can be around 20% for OAR
•
can change due to change in workflow
– necessary for each center to investigate for themselves!
•
can be better controlled for HDR, especially with MR scans just
prior to irradiation
– But still remain
– And contouring …..
Outline Uncertainties Dose Reporting
1. Geometrical uncertainties: intra/inter application variation
2. Combination External Beam and Brachy
– EB node boost
3. Same dose parameters same dose distribution??
Dose summing EB and BT
In GEC-ESTRO recommendations:
Underlying assumption is that targets and OAR receive full EB
dose.
HR-CTV: 25 x 1.8 Gy ≡ 44.3 Gyαβ10 EQD2
OAR:
25 x 1.8 Gy ≡ 43.2 Gyαβ3 EQD2
Total dose received from combined EB and BT:
HR-CTV:
D90 = 44.3 + D90(BTfraction1) + D90(BTfraction2)
Gyαβ10 EQD2
OAR: worst case scenario:
D2cc = 43.2 + D2cc(BTfraction1) + D2cc(BTfraction2) Gyαβ3 EQD2
“Adding parameters” Method
Is this a valid approximation?
7
“Adding 3D dose distributions” Method
trans
CT +
3D dose IMRT
Register underlying
CT and MRI datasets
sag
MR +
3D dose Brachy
α/β and T1/2 tumour & OAR,
defined on MRI
sag
Compute 3D EQD2
and add voxel-by-voxel

Combined DVH
Illustration of
combined EQD2 dose distributions
trans
A
100
EQD2
Gyαβ
trans
B
cor
with node boost
0
C
sag
trans
cor
sag
with parametrium boost
Van de Kamer et al. Radiother Oncol 2010
Difference between Parameter Adding and
3D dose distribution Adding
for 5 patients:
differences between methods in %
only with boosts
Van de Kamer et al. Radiother Oncol 2010
8
EBRT + Brachy + EBNodeBoost
Dosimetric interaction
reporting D90 to HR-CTV
dose of EB boost to HR-CTV
Brachy dose
to ‘pathological’ lymph nodes
prescription dose of EBboost
1. what
is dose contribution from node boost to HR-CTV?
2. what is dose from brachy to nodes?
Dosimetric interaction study
•
15 consecutive (EMBRACE) patients receiving an EB boost to
enlarged lymph nodes
•
for these 15 patients
• 1) we estimated dose from Node Boost to HR-CTV
• 2) we determined dose from Brachy to nodes
Part 1: 15 patients with EB node boost:
contribution to D90 HR-CTV
D90 HR-CTV
Patient
HR-CTV
nr
volume (cc)
#1
32.7
#2
15.9
#3
91.3
#4
29.3
#5
19.5
#6
33.0
#7
115.0
#8
26.7
#9
18.6
#10
26.0
#11
22.3
#12
39.7
#13
52.5
#14
24.1
#15
44.3
avg
SD
min
max
39.4
28.1
15.9
115.0
including boost
due to boost
Gyα
αβ
Gyα
αβ
94.1
93.4
86.1
89.9
89.5
96.0
91.6
97.2
89.7
93.3
89.7
92.7
91.1
95.8
90.7
1.0
0.6
0.7
0.3
0.3
0.6
0.2
0.5
0.1
0.3
0.6
0.8
0.3
0.3
92.1
3.0
86.1
97.2
0.6
0.4
0.1
1.6
1.6
conclusion:
at our institute, currently the
contribution is low.
Uncertainty in D90 HR-CTV
small
but…rigid registration
9
Part2: Contribution Brachy to total dose lymph
nodes
•
•
•
15 patients from Utrecht UMCU
• 2 fractions PDR
27 patients from Pittsburgh UPCI
• 5 fractions HDR
For each brachy fraction:
– Contouring of enlarged lymph nodes
– DVH analysis in BPS D90 pLNN
• physical dose D90 in cGy
• physical dose D90 in percentage PD
• biologically weighted D90 in EQD2 Gy
Van den Bosch et al. Radiother Oncol 2012 abst ESTRO Barcelona
Results: Brachy dose to lymph nodes
UPCI
HDR
57 pLNN
Mean
SD
all 40 pLNN
Mean
SD
UMCU
Volume HR-CTV (cm3)
D90HR-CTV (Gy EQD2)
36.9
39.3
10.2
5.2
39.4
47.1
28.1
3
39.4
47.1
28.1
3.1
39.4
41.2
28.1
4.9
D90 pLNN/PD (%)
D90 pLNN (Gy EQD2)
10.8
2.7
5.1
1.4
20.5
7.1
15.3
5.9
16.7
5.6
7.2
2.5
16.7
3.9
7.2
1.9
PDR
selection of 35 pLNN
Mean
SD
recalculated for HDR
35 pLNN
Mean
SD
** PDR data were recalculated as if given by HDR in 4 fractions of 7 Gy and scaled for OAR constraints.
Van den Bosch et al. Radiother Oncol 2012 abst ESTRO Barcelona
Results: Brachy dose to lymph nodes
selection 35 nodes
UPCI
HDR
57 pLNN
Mean
SD
all 40 pLNN
Mean
SD
UMCU
Volume HR-CTV (cm3)
D90HR-CTV (Gy EQD2)
36.9
39.3
10.2
5.2
39.4
47.1
28.1
3
39.4
47.1
28.1
3.1
39.4
41.2
28.1
4.9
D90 pLNN/PD (%)
D90 pLNN (Gy EQD2)
10.8
2.7
5.1
1.4
20.5
7.1
15.3
5.9
16.7
5.6
7.2
2.5
16.7
3.9
7.2
1.9
PDR
selection of 35 pLNN
Mean
SD
recalculated for HDR
35 pLNN
Mean
SD
** PDR data were recalculated as if given by HDR in 4 fractions of 7 Gy and scaled for OAR constraints.
10
Results: D90 nodes
for different nodal regions
D90 pLNN of BT / Prescription Dose (in %)
100.0
UPCI
UMCU
90.0
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
Common
iliac
Internal
iliac
External
iliac
Obturator
Parametrial
Inguinal
Perirectal
Anatomical nodal regions
Dose Nodes :
relation with volume 100% isodose
14.0
UPCI
12.0
D90 brachy to be taken into account for
prescribing EB node boost
UMCU
UMCU HDR
BT D90 pLNN (Gy EQD2)
10.0
8.0
6.0
4.0
2.0
0.0
0.00
2.00
4.00
6.00
8.00
10.00
12.00
TRAK/Dose (cm2)
Conclusion
IGABT:
individualized dosimetry of pelvic nodal
disease highly recommended.
11
Outline Uncertainties Dose Reporting
• Geometrical uncertainties: intra/inter application
variation
• Combination External Beam and Brachy
• Same dose parameters same dose
distribution??
– planning study tandem/ovoid applicator
Planning study Tandem/Ovoid centers
4 Centers
2x PDR, 2x HDR
4 EMBRACE patients from Utrecht database
small (19.5cc), medium( 31.9 and 33.8cc), large (68cc)HR-CTV
with 0, 2, 3, 6 needles
contoured according GEC-ESTRO recommendations
3 Different treatment plans created by each centre
Standard (Std)
Optimized without needles (Opt-IC)
Optimized with needles (Opt-IC/IS)
Aim
Constraints OARs:
(3 centers)
D90 HR-CTV > 85 Gy EQD2
bladder D2cc < 90 Gy EQD2
rectum/ sigmoid/ bowel D2cc < 75 Gy EQD2
Nomden et al. Radiother Oncol 2012
Comparison of…
• Dose parameters
D90 HR-CTV and D2cc OARs
• Dwell time distribution
• Spatial Dose distribution
12
Clinical route
Applicator insertion
Imaging
Contouring
Applicator reconstruction
Treatment planning
Dose delivery
Treatment planning
Applicator insertion
Imaging
Contouring
Applicator reconstruction
Treatment planning
Dose delivery
Results Dose parameters
Two patients
Small volume
Large volume
13
Results Dose parameters
Two patients
Small volume
Large volume
Dwell time distribution
different choices
Dose distribution map
Tandem to
85 Gy isodose
14
Dose distribution map: Standard plan
Example of patient 4 with large HR-CTV
Centre 2 PDR
Centre 1 PDR
Centre 1 – Centre 2
Centre 1: more dose to tip tandem
Centre 2: more dose to ovoids
Dose distribution maps:
Example of patient 4 with large HR-CTV
85 Gyα/β isodose (blue)
and HR-CTV (red)
Distance maps of 85 Gyα/β isodose contour
Dose distribution centre 2
more conformal to HR-CTV
less dose to OAR (incl vagina)
•
MR guided optimised plans, same anatomy, same goals, similar DVH
parameters:
•
different 3D dose distributions
→ other doses in non-reported regions, e.g. vaginal dose
→ soft constraints
high dose regions etc
use of needles
asymmetry
SLP
•
Learned from discussion:
Practise changing studies!
Difficult to compare treatments based on only few parameters
15
General conclusion
• Geometrical uncertainties: intra/inter application
variation
– using HDR and imaging just prior to treatment
provides the best control
• Combination External Beam and Brachy
– compute D90 EQD2 from brachy to nodes
• Same dose parameters same dose distribution?
– keep in mind the 3D dose distribution, not only the
values of DVH parameters
• Uncertainties are here to stay,
but …
• registration studies like EMBRACE help us to
recognize and deal with them.
E.g. now vaginal dose reporting to correlate with
vaginal toxicity
Thanks:
UMC Utrecht:
•
•
•
•
•
•
•
•
Ina Schulz
Christel Nomden
Judith Roesink
Robert Tersteeg
Rien Moerland
Willemien van den Bosch
Kathelijne van de Ende
Marijke van Deursen
Other
•
•
•
•
•
•
•
•
•
•
•
•
Jeroen van de Kamer (NKI Amsterdam)
Laura Velema (Erasmus Rotterdam)
Sushil Beriwal (UPCI Pittsburgh)
Erik van Limbergen (Leuven)
Marisol de Brabandere (Leuven)
An Nulens (Leuven)
Remy Nout (LUMC)
Mirjam Laman (LUMC)
Martijn Ketelaars (LUMC)
Ludy Lutgens (MAASTRO)
Brigitte Reniers (MAASTRO)
Nicole Nesvacil (AKH Vienna)
16