838 / 1082
S822
ESTRO 36 2017
_______________________________________________________________________________________________
was used allowing to better define the achievable mean
dose to organs at risk (OAR) during inverse planning step
(Moore et al., Int. J. Radiation Oncology Biol. 2011). This
model integrates the overlap volume between the OAR
and the PTV. The aim of the present study was to evaluate
application of this model adapted to our own data to
anticipate the lower dose achievable to the parotid glands
(PGs) and its impact in the inter-operator variability in
head and neck (H&N) cancer treatment planning.
Material and Methods
Twenty patients treated for locally advanced H&N cancer
were used to generate the predictive model (PM). Three
(70/63/56 Gy) and two (60/54 Gy) dose levels VMAT
simultaneously integrated boost treatment plans were
generated using Pinnacle v.9.10 (Philips) Treatment
Planning System. To test the PM, 10 additional cases were
planned with and without the PM (8 patients with 3 levels
of prescribed dose 70/63/56 Gy and 2 patients with 2
levels of prescribed dose 60/54 Gy). In a second time, 12
operators with different treatment planning experience
performed a treatment planning on the same patient, with
and without the PM. Doses to PTVs, PGs, spinal cord PRV,
indexes of conformity (CI), homogeneity (HI) and the
number of Monitor Units (MU) were compared.
Results
Table 1 shows the results for 10 treatment plans with and
without the PM. On average, mean doses (Dmean) to PGs
decreased of 5.3 Gy [-15.4 Gy; +2.6 Gy] using the model.
CI and maximal dose to the spinal cord PRV were similar
with both methods. However, plans obtained using the PM
show less dose homogeneity into PTV (for middle dose
PTV, HI increase by 18% with PM) and had more MU: +13%
on average [-3.1%; +32.6%], indicating an increase of plans
complexity. Figure 1 shows DVH for homolateral and
controlateral PGs with and without PM for the treatment
planning generated by 12 operators. With PM use, the
dispersion of the data were lower, demonstrating a
decrease in inter-operator variability: standard deviation
for mean dose delivered to homolateral PG decrease from
2.16 Gy to 1.19 Gy and for contralateral PG from 2.89 Gy
to 0.78 Gy.
Conclusion
This study showed the utility of a PM to reduce the dose
received by the PGs in H&N treatment planning (Dmean
decreased of 5.3 Gy). The suggested model guides to the
lowest achievable Dmean to the PGs at the beginning of
treatment planning step. Integrating this method in the
treatment planning workflow reduces significantly the
inter-operator treatment planning variability and could
potentially allow to a time reduction in treatment
planning.
EP-1548 Dose to risk organs in deep inspiration breath
hold non-coplanar VMAT for lung cancer radiotherapy
M. Josipovic
1
, G. Persson
1
, J. Bangsgaard
1
, L. Specht
1
, M.
Aznar
1
1
The Finsen Center - Rigshospitalet, Dept. of Oncology-
Section of Radiotherapy, Copenhagen, Denmar
Purpose or Objective
Radiotherapy (RT) for locally advanced lung cancer has
a
high burden on dose to risk organs (OAR), such as lung,
heart and oesophagus. Different strategies have been used
to decrease the dose to OAR, such as volumetric
modulated arc therapy (VMAT) and deep inspiration
breath hold (DIBH). In this study we investigated VMAT
combined with DIBH and non-coplanar (NC) treatment
delivery.
Material and Methods
Patients with central lung tumours were selected from a
cohort treated in a DIBH RT trial. VMAT plans were made
clinically in both free breathing (FB) and DIBH and
consisted of two coplanar (C) partial or full arcs. For NC
plans we aimed for RT delivery within 4-6 DIBHs of 20 s, as
for the clinically delivered plans. Therefore an approach
similar to butterfly VMAT was chosen, with two either 120º
or 240º arcs at couch 0º, depending on clinical choice of
partial or full arcs, and two 60º arcs at couch 90º. FB arc
geometry was kept in DIBH, for both C and NC plans. Dose
to OAR was compared between FB C, FB NC, DIBH C and
DIBH NC plans.
Results
Twelve patients were included, five had central right and
seven central left tumours. Total lung volume in DIBH
increased by median 48% (range 20-82%) compared to FB.
As expected DIBH reduced both mean lung dose (MLD) and
lung V20 (median 2.2 Gy and 4.1%). NC VMAT plan
decreased MLD and V20 compared to C VMAT with similar
amount in both FB and DIBH (median ~0.25 Gy and ~1.5%).
Figure shows impact of techniques on dose to ipsi- and
contralateral lung. In right sided tumours, both MLD, lung
V20 & V40 were smaller compared to left sided tumours.
Mean heart dose (MHD) and heart V50 decreased with
DIBH. NC VMAT had the opposite effect, since the two arcs
delivered at couch 90° could often not avoid dose
entrance through the heart. As
anticipated, MHD, heart
V50 and heart D2 (minimum dose to the hottest 2% of the
heart) were largest in left sided tumours. However, in this
small patients group, the observed heart dose parameters
were much lower than clinically applied constraints. Still,
trying to spare the heart may have resulted in larger lung
doses in patients with left sided tumours in all plans
(median MLD differences 0.5-2.5 Gy).
Mean oesophagus dose (MED) increased in DIBH, but was
not affected by NC technique in either FB or DIBH.
However, MED is not a clinically used constraint.
Oesophagus V66 was smallest in DIHB C, but in none of the
plans it reached close to its limit of 1cm3 (national
guidelines’ constraint).
Dose to spinal cord was reduced with both NC and DIBH,
with DIBH NC offering the best sparing. See Table for
details on dose parameters and clinically used constraints.