Predictors for morbidity from planned
versus delivered rectal dose maps in
RT of prostate cancer
Joakim Trane1, O Casares-Magaz1*, Lise Bentzen2, Kia Busch1, Maria Thor3 and Ludvig P. Muren1
1Dept.
of Medical Physics, Aarhus University Hospital, Denmark.
2 Dept. of Oncology, Aarhus University Hospital, Denmark.
3Dept. of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, USA.
B A C K G R O U N D
•
R E S U LT S
Patient-reported gastro-intestinal (GI) symptoms following
radiotherapy (RT) for prostate cancer have recently been
associated with metrics derived from rectal dose surface maps.
•
• In a recent study we developed rectum dose map based normal
•
tissue complication probability (NCTP) models for three common
late GI symptoms (at least 20% prevalence in the cohort used
for modelling).
A I M
MATERIALS AND METHODS
O F
T H E
S T U D Y
10
In the present study we used such dose maps and connected
NTCP models to compare the planned, daily and summed rectal
CHAPTER 2. MATERIALS AND METHODS
11
dose distributions for patients with repeat volumetric imaging
acquired during the course ofrectum.
RT.
•
Dose differences exceeding +/- 10 Gy (scaled to the full
CHAPTER
3.
RESULTS
treatment course) were seen in the dose maps for all patients
and in all scans.
The largest
dose increase in the maps during the
Patientsystematic
2:
course
of therapy
thewho
patient
that experienced
Patient
2 waswas
theseen
onlyinone
experienced
GI mor
Grade
2+ GI
after
(Fig. 2).
area
ofsymptoms
the CBCTs
is treatment
most notable,
having a large diff
all also
the CBCTs,
seeming
almost
off.spatial
Otherwise,
This for
patient
had higher
NTCPs for
all three
dose
metricaverage
based models
for the average
map across
treatment
of the CBCTs
shows very
well the
differen
mosttoofthe
the
CBCTs,
which
only (e.g.
differ
in vs
the6%width
compared
planned
dose
distribution
10%
for
to the
average,
thewere
top and
faecalpattern
leakage),
while
smaller notably
differences
seenthe
for centra
the
three other patients.
Figure 3.2: Patient 2
Figure 2.2: Segments of the 200x200 pixels 2D-maps.
Figure 1. (Left Panel) Axial cut of the planning CT for one of the patients with
To be
able to see
possible differences
in dose
distribution across the of
the
dose
distribution
superimposed.
(Right
Panel)
2D
map
parametrisation
he transverse plane of the pCT of patientfractions
1. Themore
colorbar
easily, the difference between the pCT and each CBCT
dose
delivered
to the
rectum
wall.
and
the averagein
of the
the CBCTs was calculated and depicted along side
ose the
in [ Gy
]. The
prostate area
is seen
in the
middle,
pCT
Standard dev. for each
voxel
Average CBCT
Difference to the pCT
the daily 2D map representations. The standard deviation for each
prostate
in a pink
voxel, across all CBCTs, was calculated as well.
Part of the bladder is seen above the
the rectum is seen underneath the prostate as the purple
M
A
T
E
R
I
A
L
S
A
N
D
r
ed in CERR via MATLAB .
2.3 NTCP
METH OD
•
The patients included in this
study
Metrics in
this study, were
will be usedtreated
to calculate an according
estimated risk based to
on a
a logistic regression NTCP model. This method was previously established by Casares-Magaz
et al [10], withadvanced
a larger patient cohort
treated
national
clinical trial for patients
with locally
prostate
e map
representation
in Aarhus in 2005-2007, with a validated patient reported outcome questionnaire for toxicity
recording.
cancer,
irradiating
concomitantly
the
rom the pCT and CBCTs was digitally unfolded and 2D pelvic lymph nodes and
reated seminal
by interpolating
across 25
equidistant
points
along
vesicles
to
55
Gy
and
the
prostate
to
78
Gy
using
or each 45o -sector, for each patient. Each contour was unvolumetric
modulated
arc therapy.
ng the posterior
point defined
by the sagittal
plane, which is
•
y/mirror plane of the human body, containing the centroid
The
treatment
plans
were
recalculated
on
weekly
repeat
coneour. The 2D maps were then all normalised to 200 by 200
was done
in MATLAB
(Mathworks
, Natick, MA
beam
(CB)r CTs
(6-8 rCBCTs
perUSA).
patient) following Hounsfield
of the final DSMs, which were used for calculating metrics,
Unit
override
to
bone
and
water.
n figure 2.2. The two lateral edges represent the posterior
vertical
central
line
represent
the
most
anterior
part
of
the
Rectal dose maps were created for the planned dose distribution
•
as well as for the dose distributions re-calculated on weekly
CBCTs using a method recently developed by our group.
• The weekly CBCTs were averaged to provide a measure for the
summed/accumulated dose across the course of RT.
• NTCPs were calculated for the planned, weekly and averaged
ESTRO 2017
rectal dose maps using three spatially based response models
(based on areas and extents from the rectal dose maps) for
three patient-reported GI symptoms: faecal leakage, obstruction
and defecation urgency. The study included four prostate cancer
patients, one with and three free from late Grade 2+ GI
symptoms after RT.
EP-1610
Figure 2. 2DDay
dose# map
for (left)
the planning
CT (upper left),
thepCT
averaged
CBCT
and difference
to the
(right
delivered doses across the 8 CBCT evaluated (lower left), pixel-wise dose
standard deviation Day
(upper1:right) and pixel-wise dose difference in the 2D
map between the planning CT and the averaged delivered (lower right).
C O N C L U S I O N
The rectum dose maps and the connected NTCP models used in
this study identified clinically relevant changes in rectum dose
distributions caused by organ motion during the course of therapy.
This model validation study showed that these maps and models
are useful tools to
evaluate
the
risk
of
normal
tissue
reactions
in
Day 6:
the rectum.
Physics track: (Radio)biological modelling
Oscar Casares Magaz
DOI: 10.3252/pso.eu.ESTRO36.2017
Figure 3.2: The DSMs for patient 2. Describtion of th
Poster
caption
to at:figure 3.1.
presented