138 / 1082
S125
ESTRO 36 2017
_______________________________________________________________________________________________
oncology in Canada, namely, reduction in waiting times
for radiotherapy, and peer-review of radiation oncology
treatment planning. Both are highly relevant to the
overall strategy for quality improvement in
Canada. Health care system performance on both wait
times and peer-review are key quality indicators for
radiotherapy treatment programs in Canada. Wait times
for radiotherapy - known to be associated with treatment
outcomes - have improved in some dimensions, but remain
a problem in others, particularly when overall wait times
from diagnosis to first treatment are considered. Trends
in wait times and current strategies to reduce wait times
will be discussed. Peer review of treatment planning
consists of a radiation plan being reviewed by at least one
independent radiation oncologist (preferably in a multi-
disciplinary setting). Radiation planning peer review is
endorsed as a critical component of a radiation oncology
quality assurance program. The presentation will
describe a pan-Canadian initiative that was launched in
2012 with the intent to disseminate the uptake and to
improve the conduct of peer review activities. The
components of the initiative include a national base-line
survey of current practice and attitudes, strategies to
promote the uptake of peer review, quantitative
evaluations of peer review findings, qualitative
assessments of peer review activities in radiation oncology
programs, and development of “best practice” guidelines
for practitioners and radiation programs for selected
common radiotherapy treatment scenarios.
SP-0250 Waiting time in radiotherapy, an overlooked
problem
J.Overgaard
9
Aarhus University Hospital, Dept Expt. Clin. Oncology,
Aarhus C, Denmark
Abstract not received
Proffered Papers: Automated and robust treatment
planning
OC-0251 Late toxicity in HYPRO randomized trial
analyzed by automated planning and intrinsic NTCP-
modelling
A.W.M. Sharfo
1
, M.L.P. Dirkx
1
, R.G. Bijman
1
, W.
Schillemans
1
, S. Breedveld
1
, S. Aluwini
1
, F. Pos
2
, L.
Incrocci
1
, B.J.M. Heijmen
1
1
Erasmus MC Cancer Institute, Radiation Oncology,
Rotterdam, The Netherlands
2
Netherlands Cancer Institute-Antoni van Leeuwenhoek
Hospital, Radiation Oncology, Amsterdam, The
Netherlands
Purpose or Objective
To analyze delivered OAR doses and NTCPs in the HYPRO
multicenter randomized hypofractionation trial for
prostate cancer patients (Lancet Oncology 2015, 2016)
using automated VMAT planning (autoVMAT). The applied
multivariate NTCP models were derived by correlating
clinically observed complications in the HYPRO trial with
clinical and dosimetric parameters.
Material and Methods
820 prostate cancer patients were included in the HYPRO
trial, randomly assigned to standard fractionation (39x2
Gy, 5 fr/wk) or hypofractionation (19x3.4 Gy, 3 fr/wk).
Our platform for fully automated multi-criterial treatment
planning was used to generate for each patient an
autoVMAT plan. Achieving adequate PTV coverage had the
highest priority, followed by minimization of the dose to
the rectum, anus, bladder and hips. Plans were compared
with respect to PTV dose coverage, rectum D
mean
, V
65Gy
and
V
75Gy
, mean doses in anus and bladder, and maximum doses
in the femoral heads. Moreover, comparisons were
performed using NTCP models derived from the HYPRO
database for grade ≥ 2 late Gastro Intestinal (GI) toxicity,
stool incontinence, stool frequency, rectal bleeding, and
proctitis. For a subgroup of patients, autoIMRT plans with
the clinically used beam angles were generated as well.
For the analyses, all OAR doses in both fractionation
schemes were converted to EQD
2Gy
assuming α/β=3 Gy.
Results
So far, 430 patients (215 in each arm) were analyzed.
Compared to the clinically applied plans, autoVMAT plans
had similar or higher PTV coverage. Large and highly
significant enhanced OAR sparing was observed with
autoVMAT for both treatment arms (see figure 1).
Compared to the clinical plans, the autoVMAT plans
showed reductions in mean doses in the rectum, anus and
bladder of 6.9±4.4 Gy, 7.2±6.2 Gy and 4.1±2.7 Gy
(p<0.001), respectively. Rectum V
65Gy
and V
75Gy
were
reduced by 3.3% (relative difference 23.4%±19.7%) and
1.3% (relative difference 27.5%±51.9%) (p<0.001),
respectively. Maximum doses in the left and right femoral
heads were also reduced by 29% and 32% on average.
Figure 2 compares clinical and autoVMAT plans regarding
NTCPs for the studied GI symptoms. Significant reductions
in rectal NTCPs with autoVMAT were observed with a
relative reduction of 10.5% in late GI grade ≥2, 16.8% in
stool incontinence, and 18.7% in rectal bleeding
(p<0.001). Plan quality improvements with autoIMRT
relative to clinical plans were similar as those observed
for autoVMAT, showing that enhanced plan quality was not
related to the use of VMAT instead of IMRT.
Conclusion
Automatically generated VMAT and IMRT plans resulted in
large plan quality improvements compared to the
clinically applied IMRT plans with significant NTCP
reductions. The enhanced plan quality results from
improved planning, possibly related to improvements in
the treatment planning system (TPS) and/or automation
of planning.
OC-0252 Acceptance rates of automatically generated
treatment plans for breast cancer
G. Van der Veen
1
, A. Duijn
1
, J. Trinks
1
, A. Scholten
1
, R.
Harmsen
1
, G. Wortel
1
, R. De Graaf
1
, D. Den Boer
1
, E.
Damen
1
1
Netherlands Cancer Institute Antoni van Leeuwenhoek
Hospital, Radiation Oncology, Amsterdam, The
Netherlands