ESTRO 36 Abstract Book

269 / 1096

S256 ESTRO 36
_______________________________________________________________________________________________
involved, baseline shifts of such nodes relative to the
primary tumor can range up to cm’s. Therefore, setup is
often based on bony anatomy using generous planning
margins for the primary tumor and LN. We present an
adaptive strategy to reduce these margins for the primary
tumor.
Material and Methods
In a previous retrospective study in 20 stage III NSCLC
patients we found that a separation into ‘movers’ and
‘non-movers’ is useful. Patients with an average baseline
shift < 3 mm in the first 3 fractions were deemed ‘non-
movers’. Dosimetric analysis of tumor coverage over the
entire treatment (evaluated on 8 CBCTs per patient)
showed that a 6 mm ITV-PTV margin for the primary tumor
was adequate for these non-movers (in contrast to the 1
cm margin applied clinically). In the present study, we
prospectively applied this selective margin reduction
method. Two plans were prepared with 6 respectively 10
mm ITV-PTV margin. All patients started with the 10 mm
plan. Baseline shifts of the primary tumor were
determined with CBCT dual registration (XVI, Elekta) using
a clipbox match on nearby bony anatomy (often vertebrae)
and a simultaneous mask match on the tumor region. This
registration was performed by RTTs in routine clinical
practice. The results of the clipbox matches were used in
an eNAL offline setup correction protocol. If, after 3
fractions, a patient was classified as a non-mover, the 6
mm plan was applied in subsequent fractions. Next, each
3rd or 5th fraction (for 25 respectively 33 fractions) was
imaged to monitor the average baseline shift over the last
3 imaged fractions. If the latter average baseline shift
exceeded 3 mm, a switch back to the 1 cm plan would be
made.
Results
21 stage III NSCLC patients, treated with curative intent,
were included to date. 14 patients (67%, consistent with a
prediction of 70%) were found to be non-movers and
switched to a plan with 6 mm margin for the remainder of
the treatment. Follow-up imaging showed that all these
patients remained non-movers: their average baseline
shift over the entire treatment remained < 3 mm (3D
vectorlength) in all cases. Although highly patient
dependent, the margin reduction decreased OAR dose
significantly. For instance, the average V20Gy was
reduced from 15.0 to 13.4 Gy and the mean heart dose
from 12.2 to 10.0 Gy.
Conclusion
We have developed and clinically applied a practical
adaptive method for planning margin reduction in non-
stereotactic treatment of lung cancer. This method allows
for smaller margins in approximately 70% of patients.
OC-0484 Variability of breathing-induced tumour
motion: 4DCT – a source of misguiding information?
J. Dhont1, D. Verellen2, M. Burghelea1, R. Van Den
Begin1, K. Tournel1, T. Gevaert1, B. Engels1, C. Collen1,
C. Jaudet1, M. Boussaer1, T. Reynders1, G. Storme1, M. De
Ridder1
1Universitair Ziekenhuis Brussel, Radiotherapy Medical
Physics, Brussels, Belgium
2GZA Ziekenhuizen- Sint Augustinus - Iridium
Kankernetwerk Antwerpen, Radiotherapy Medical
Physics, Antwerpen, Belgium
Purpose or Objective
The purpose of this study was to evaluate both the short
and long-term variability of breathing-induced tumour
motion. In addition, it was investigated whether 4DCT is a
reliable source to represent the tumour motion during the
entire course of treatment.
Material and Methods
3D tumour motion was evaluated for 22 patients treated
with SBRT for either primary NSCLC (6/22, 4x12Gy, 2
weeks), metastatic lung lesions (9/22, 10x5Gy, 2 weeks)
or metastatic liver lesions (7/22, 10x5Gy, 2 weeks).
Treatment was delivered with dynamic tracking (DT) on
the Vero SBRT system, requiring a gold fiducial implanted
near the target.
With DT, a 20 s orthogonal fast fluoroscopy (FF) sequence
is acquired before each fraction. Additional sequences are
taken if the breathing motion changes. As such, for each
patient at least one set of X-ray images is available per
fraction from which the 3D tumour motion can be
extracted using the implanted marker. If multiple FF
sequences were available per fraction, the tumour motion
was obtained for each sequence independently.
To evaluate the short-term intra-fractional variability, the
amplitude, tumour position at maximum exhale (r0) and
hysteresis (ie. 3D distance between the tumour position at
mid-inhale and -exhale) were compared between
different FF sequences from the same fraction. To assess
long-term variability, amplitude and hysteresis were
compared between fractions and with the 3D tumour
motion registered by the pre-treatment 4D planning CT
(free-breathing, using RPM (Varian) and amplitude-based
binning in 10 phases, Siemens Somatom Definition AS, 2
mm slices).
Results
Based on r0 , 10 patients showed an intra-fractional
baseline drift of more than 5 mm, in one or more
directions, in 50% or more fractions. For all patients, intra-
fractional differences in amplitude were not statistically
significant (p > 0,05). An example of intra-fractional
variability is shown in Fig 1.
In terms of long-term variability, for 11 patients the
difference in mean amplitude between at least 2 fractions
or a fraction and 4DCT is statistically significant (p < 0,05).
Fig 2 shows the mean amplitude per fraction per patient,
together with the amplitude based on 4DCT.
Both intra- and inter-fractional changes in hysteresis were
smaller than 5 mm for all patients. No correlation could
be found between long or short-term variability and
tumour size or location.