S465
ESTRO 36
_______________________________________________________________________________________________
PO-0855 Use of the LKB model to fit urethral
strictures for prostate patients treated with HDRB
V. Panettieri
1
, E. Onjukka
2
, T. Rancati
3
, R. Smith
1
, J.
Millar
1
1
Alfred Hospital, Alfred Health Radiation Oncology,
Melbourne, Australia
2
Karolinska University Hospital, Dept of Hospital Physics,
Stockholm, Sweden
3
Fondazione IRCCS- Istituto Nazionale dei Tumori,
Prostate Cancer Program, Milan, Italy
Purpose or Objective
High-Dose-Rate brachytherapy (HDRB) is widely used in
combination with external beam radiotherapy in the
treatment of prostate cancer. Despite providing
biochemical control similar to other techniques, due to
the variety of fractionation regimes used there is no clear
consensus on the dose limits for the organs-at-risk, in
particular the urethra.
The aim of the work has been to fit the Lyman-Kutcher-
Burman (LKB) Normal Tissue Complication Probability
model to clinical outcome on urethral strictures data
collected at a single institution.
Material and Methods
Dose-volume histograms and clinical records of 262
patients were retrospectively analysed. The patients had
follow-up 6, 12, 18, 24 months and then every year until
10 years after the treatment. Clinical and toxicity data
were collected prospectively. The end-point was the time
of the first urethrotomy, a follow-up cut-off time of 4
years was chosen and the average stricture rate was about
12.6%. The LKB NTCP model was fitted using the maximum
likelihood method and used simulated annealing to find a
stable solution. Since the patients were treated with 3
different fractionation regimes (18 Gy in 3, 19 Gy in 2 and
18 Gy in 2 fractions) doses were converted into EQD2 with
α/β = 5 Gy.
Results
For this cohort of patients the risk of urethral stricture
could be modelled by means of a smooth function of EUD
(see Fig 1). Using the LKB model the risk of complication
could be represented by a TD50 of 220 Gy, a steepness
parameter m of 0.55 and a volume-effect parameter n of
2.7. The fitted model showed good correlation with the
observed toxicity rates with the largest deviation shown
at higher doses. The large value of n could suggest a
parallel behaviour of the urethra, however further
validation is required with an independent dataset.
Conclusion
In this work we have fitted the LKB model to clinical
outcome on urethral strictures data for patients treated
with HDRB collected at a single institution. The results
show that the fitted model provides a good representation
of the observed data, however further analysis and
independent validation are necessary to confirm its
behaviour and parameters.
Poster: Physics track: Intra-fraction motion
management
PO-0856 Systematic baseline shifts of lymph node
targets between setup and treatment of lung cancer
patients
M.L. Schmidt
1
, L. Hoffmann
1
, M.M. Knap
2
, T.R.
Rasmussen
3
, B.H. Folkersen
3
, D.S. Møller
1
, B. Helbo
2
, P.R.
Poulsen
2
1
Aarhus University Hospital, Medical Physics, Aarhus C,
Denmark
2
Aarhus University Hospital, Department of Oncology,
Aarhus C, Denmark
3
Aarhus University Hospital, Department of
Pulmonology, Aarhus C, Denmark
Purpose or Objective
Internal target motion results in geometrical uncertainties
in lung cancer radiotherapy. The lymph node (LN) targets
in the mediastinum are difficult to visualize in cone-beam
computed tomography (CBCT) scans for image-guided
radiotherapy, but implanted fiducial markers enable
visualization on CBCT projections and fluoroscopic kV
images. In this study, we determined the intrafraction
motion of mediastinal LN targets in both the setup CBCT
and fluoroscopic kV images acquired during treatment
delivery, and investigated the baseline shifts and
treatment accuracy of LNs for ten lung cancer patients.
Material and Methods
Ten lung cancer patients had 2-4 fiducial markers
implanted in LN targets by EBUS bronchoscope. A total of
26 markers were evaluated. The patient received IMRT
with daily setup CBCT for online soft tissue match on the
primary tumor. During treatment delivery, 5 Hz
fluoroscopic kV images were acquired orthogonal to the
MV treatment beam. Offline, the marker positions were
segmented in each CBCT projection and fluoroscopic kV
image. From the segmented marker positions, the 3D
marker trajectories were estimated from the
segmentations with sample rate of 11 Hz during CBCT
acquisition and 5 Hz during treatment delivery.
The 3D motion amplitude and mean position of each LN
marker as well as the intrafraction baseline shifts between
setup CBCT and treatment delivery were calculated.
Results
Figure 1 shows the internal motion of one marker at one
fraction. The motion is shown relative to the mean
position during the CBCT scan and corrected for the couch
shift between CBCT and treatment. For this marker, the
baseline shift was 4.8 mm cranially, 0.6 mm posteriorly,
and 0.7 mm towards right. Figure 2a shows the distribution
of intrafraction baseline shifts for all patients and LNs at
all fractions. Systematic LN baseline shifts occurred
between CBCT and treatment delivery in the cranial
direction (mean 2.4 mm (SD 1.9 mm)) and posterior
direction (0.8 mm (1.1 mm)). The frequency of cranial
baseline shifts exceeding 4 mm and 6 mm were 15 % and 4
%. The baseline shifts resulted in systematic mean
geometrical errors during treatment delivery of 2.8 mm
(cranial) and 1.4 mm (posterior)(Figure 2b) for the LNs.
These errors were substantially larger than the sub-
millimeter mean errors expected from the setup CBCT
based soft tissue tumor match when correcting for the
applied couch shifts.
In general, the largest LN motion amplitude was observed
in the cranio-caudal direction both during CBCT and
treatment delivery. The mean motion amplitudes during
CBCT and treatment delivery agreed within 0.2 mm in all
three directions.